Tumour markers of prognosis in colorectal cancer

H L McLeod; G I Murray

doi:10.1038/sj.bjc.6690033

Tumour markers of prognosis in colorectal cancer

McLeod, H L; Murray, G I 1998-12-18 00:00:00 British Journal of Cancer (1999) 79(2), 191–203 © 1999 Cancer Research Campaign Review 1 2 HL McLeod and GI Murray 1 2 Departments of Medicine & Therapeutics, and Pathology, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, UK Keywords: colorectal cancer; prognosis; immunohistochemistry; tumour markers Colorectal cancer is a common disease in developed nations. In systemic chemotherapy. These data highlight the need for informa- Europe, there are approximately 210 000 new cases per year and tive prognostic markers which will aid identification of patients around 124 500 people die from this disease each year (Parkin et best treated with surgery alone, those requiring chemotherapy and al, 1993). Thus, it is the second most common malignancy in men those who might benefit from more aggressive or experimental and third most common in women. About one-third of patients therapy. have a primary tumour in the rectum with the majority of the remaining tumours occurring in the sigmoid colon. Patients who DEFINITION OF PROGNOSTIC MARKER die from the disease most commonly succumb to the effects of secondary metastasis to the liver. The prognosis for this cancer is A practical definition of a prognostic marker is a variable that influenced by a variety of features present at the time of initial provides prospective information on patient outcome which is diagnosis. These factors include age, gender, duration of symp- complementary to the data obtained by the pathologist from the toms, presence of bowel obstruction, tumour location, need for diagnostic sections and on which therapeutic decisions can be blood transfusion and the quality of surgical intervention guided. As Dukes’ staging is widespread and common practice, (Wolmark et al, 1983; McArdle and Hole 1991; Hermanek et al, there would be no logic in finding a marker that simply gave the 1995). A number of tumour characteristics, such as vascular and same information (Astler and Coller, 1954; Cohen et al, 1991). lymphatic invasion, ploidy, differentiation and preoperative levels Rather, proteins (or other variables) which demonstrate a strong of plasma carcinoembryonic antigen, have also been shown to be correlation with Dukes’ stage (or other staging systems) should be of prognostic significance (Witzig et al, 1991; Hermanek et al, considered as possible novel targets for anti-cancer therapies. 1995). However, the simplest and most widely applied prognostic Dukes’ stage does provide a useful ‘internal control’, as it is the feature is the presence of lymph node metastasis in a surgical gold standard on which prognostic markers should be compared, resection specimen. This forms the basis of all staging systems for and any study including Dukes’ A to D disease that does not this disease and has repeatedly been shown to have more prog- demonstrate a relationship between stage and survival should be nostic power than any other single feature at presentation (Astler viewed with great suspicion. and Coller, 1954; Cohen et al, 1991). A large number of prognostic markers for colorectal cancer are Treatment of colorectal cancer is unsuccessful if the tumours featured in the medical literature. Almost all studies of prognostic have an aggressive phenotype at the time of diagnosis. To date, the markers in colorectal cancer involve the correlation of variables most important variable influencing the long-term outcome from observed at diagnosis with eventual outcome. Most published colorectal cancer is the stage of disease at diagnosis, with depth of studies have used a single marker, while a minority of studies have tumour invasion and infiltration of lymph nodes or distant organs used a small group of prognostic markers from which clinical being used to define patient prognosis (Table 1). This classifica- correlations were derived. Although there are many claims of tion discriminates patients with early-stage disease from those prognostic significance in colorectal cancer the question remains: with very advanced disease, but is less able to predict the prog- which markers are important enough that they should be applied nosis of patients with intermediate levels of tumour invasion. For prospectively in routine pathological evaluation of colorectal example, in the Intergroup study of 929 patients with Dukes’ C cancer? colorectal cancer, 45% of patients treated with surgery alone were alive and disease free after 5 years compared with 65% of patients INCLUSION OF TREATMENT DATA treated with surgery followed by 5-fluorouracil(5-FU)/levamisole (Moertel et al, 1995). This suggests that, in retrospect, nearly half One difficulty with many prognostic marker studies is their applic- of the patients with Dukes’ C colorectal cancer did not require ability to modern practice. A study demonstrating that overexpres- adjuvant chemotherapy and 35% will die even when treated with sion of marker ‘X’ gives a survival advantage in patients with Dukes’ stage C disease may have little relevance if the data are derived from a population treated prior to the routine use of adju- Received 20 March 1998 vant chemotherapy. This puts constraints on the length of clinical Revised 1 June 1998 follow-up information available, as chemotherapy only became Accepted 4 June 1998 standard practice within the last 5–7 years in most centres. The details of patient therapy are often inadequate in published studies Correspondence to: HL McLeod 191 192 HL McLeod and GI Murray Table 1 Comparison of the two most commonly used pathological staging the protein structure of target molecules, affecting the use of many systems for colorectal cancer antibodies. These factors can also alter the integrity of genomic DNA extracted from the tissue sections and thereby interfere with Dukes’ stage AJCC/UICC microsatellite analysis or other DNA-based techniques. The devel- stage opment of national guidelines for the processing of tumour samples Tumour invading the submucosa or muscularis A I (such as the UKCCCR Handbook for the ClinicoÐPathological propria Assessment and Staging of Colorectal Cancer) should lead to a more Tumour invading through the muscularis propria B II uniform handling of sample material. Evidence of tumour in regional lymph nodes, C III regardless of local tumour invasion Distant metastasis, regardless of local tumour D IV THE ENRICHMENT APPROACH invasion or lymph node status It is not practicable to test all potential markers of prognosis in a Stage D was not included in the original Dukes’ staging system, but has large data set of tumours of a particular Dukes’ stage. One poten- become commonly used to represent distant metastasis. tial strategy for providing initial data for putative prognostic marker studies is the use of an ‘enrichment’ approach. With this approach, a data set of patients could be retrospectively selected of prognostic markers. It is not satisfactory for the degree of hetero- based on their clinical phenotype, such as 5-year survival or geneity of treatment regimens within a study to be greater than the response to chemotherapy. By selecting patients at the extremes of degree of tumour heterogeneity for the marker under evaluation. the desired endpoint (i.e. Dukes’ C patients alive 5 years post adju- vant therapy vs patients who developed metastatic disease while receiving adjuvant therapy) an ‘enrichment’ step should take CHOICE OF STUDY POPULATION place. Variables only important for tumorigenesis should be The choice of study population needs to be carefully considered for equally distributed between the two patient groups, while markers studies of markers of potential prognostic importance. Many of tumour invasion, chemosensitivity or other factors regulating published studies include patients with all Dukes’ stages (A to D), the efficacy of adjuvant therapy should be delineated between the which is useful for initial evaluation of putative markers of patient groups. This approach assumes a minimum degree of prognosis. However, subsequent studies of patients with a specific heterogeneity in therapeutic regimens and can only be performed disease stage are then more appropriate for the assessment of the using retrospective analysis. It also relies on the availability of clinical utility of a marker. For example, a study of 200 patients with the relevant pathological specimens, often limiting laboratory Dukes’ stage C disease will have greater statistical power to yield approaches to those that are applicable to formalin-fixed tissues. definitive results than a study which consists of 20 Dukes’ A, 80 There are, however, statistical concerns about the use of a large Dukes’ B, 80 Dukes’ C and 20 Dukes’ D patients. In addition to number of potential tumour markers in what will often be a small diluting the statistical power of the study, the inclusion of all stages number of clinical specimens. However, the goal is to identify of disease is inappropriate for the question asked, as there is little markers with a clear association with outcome, and clinical signif- potential for the clinical use of markers in stage A disease and icance needs to have a greater role than statistical significance. A markers of tumour response to therapy or quality of life are more protein that is expressed in 30% of survivors and 0% of patients applicable to stage D patients. The time scale in which the samples with disease progression will be taken forward to more definitive are collected also has a significant influence on the applicability of analysis, regardless of the statistics. This approach will not prove the results to modern cancer treatment. As mentioned above, the the clinical utility of a marker, but rather should be a fertile ground treatment of colorectal cancer has changed significantly over the for identifying putative markers for further analysis in large past 10 years, and this changes the utility of prognostic findings. studies of specific tumour stage. Specialization of both surgeons and oncologists has improved In this review, the current data on cellular or molecular markers patient outcome (Hoffman et al, 1997). The technical ability of the with the potential for prognostic predictive ability for patients with surgeon is also a relevant variable which should not be ignored in colorectal cancer are summarized. It would not be possible to studies of prognosis (Reinbach et al, 1994). The processing of explore all of the putative prognostic markers in the literature, tumour specimens can alter the application of many techniques. For therefore those for which literature is available from either large example, the type of formalin and length of time in fixative can alter single studies or analysis from multiple laboratories were selected. Table 2 Comparison and applicability of different methodologies for assessment of tumour markers PCR PCR SSCP LOH RT-PCR Northern blotting ISH IHC Cellular constituent examined DNA DNA DNA mRNA mRNA DNA/mRNA Protein Use in formalin-fixed, wax-embedded tissue Y Y Y N N Y Y a a a a a Microdissection needed Y Y Y Y Y NN Cellular localization evaluable N N N N N Y Y b b Application to routine diagnosis Y NN N N Y Y IHC, immunohistochemistry; ISH, in situ hybridization; LOH, loss of hetorozygosity; PCR, polymerase chain reaction; RT, reverse transcriptase, SSCP, single- strand conformational polymorphism. Depending on the cellularity of colorectal tumours, microdissection may be desirable to ensure an enriched population of tumour cells. Academic centres and larger general hospitals. British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 193 Papers were identified from the Institute for Scientific Information (Pricolo et al, 1996); advanced disease (Wadler et al, 1997); database (through the Bath Information and Data Services) using Markowitz et al, 1995], while the studies demonstrating a positive the search terms colon, rectal or colorectal with prognosis, prog- correlation with prognosis included a mix of Dukes’ A–D nostic or survival. The search was conducted from January 1990 to tumours. It is possible that the presence of ras mutations is associ- January 1998 and was restricted to English-language publications. ated with disease invasiveness, independent of pathological stage, Negative studies may be under-represented in the literature but is not of prognostic significance within a given stage of because of the personal and editorial bias against publishing such disease. This may account, in part, for an association with survival work. The markers can be separated into oncogenes, tumour- in the studies including Dukes’ A–D tumours and the lack of suppressor genes, proteins involved in apoptosis, factors involved prognostic importance in the studies including larger numbers of in angiogenesis/metastasis/invasion and cellular targets for the patients with a specific pathological stage. chemotherapy used for colorectal cancer. An emphasis has been The influence of ras mutations on response to 5-FU-based placed on techniques that use immunohistochemistry, thereby chemotherapy has been evaluated in 108 patients (Markowitz et al, potentially enabling the results to be applied in many different 1995; Wadler et al, 1997). An objective response rate of 39% was settings throughout the world. However, a spectrum of methodolo- achieved with the different regimens, but mutant ras was not asso- gies are available at most medical centres to assess tumour ciated with chemosensitivity. markers (Table 2). Immunohistochemical analysis of ras p21 staining and outcome has been performed in four studies (a total of 399 patients) (Miyahara et al, 1991; Sun et al, 1991; Miller et al, 1992; Bennett ONCOGENES et al, 1995). Different monoclonal antibodies were used in all Ras studies and the method of scoring a tumour p21 as ‘positive’ either was not detailed or differed between the studies. Tumour staining The ras family represent one of the most commonly detected for p21 was found in 54% (216/399) of tumours, but varied oncogenes in human cancer. The ras genes (K-ras, H-ras, N-ras) between studies (31–71% positive). Positive staining for p21 was encode for 21-kDa proteins (p21s), which are localized on the associated with a poor prognosis in three out of four studies inner surface of the plasma membrane. These proteins bind (Miller et al, 1992; Sun et al, 1991; Miyahara et al, 1991) and was guanine nucleotides and have GTPase activity. As with other GTP- shown to be an independent prognostic factor in only one study binding proteins, ras may serve as a transducer molecule for (Sun et al, 1991). Although there are concerns about the method- signals affecting cell proliferation and thus have involvement in ology (non-specific ras antibody, criteria for ‘positive’ expression) the cell cycle. The development of ras mutations is thought to be and statistical analyses used in these studies, there is an overall an early event in the current model of colorectal tumorigenesis indication that tumour staining for p21 is associated with a poor (Fearon and Vogelstein, 1990). The prognostic significance of both prognosis and should be considered for inclusion in more defini- ras mutations and p21 staining has been assessed in a number of tive studies of markers of prognostic significance in colorectal reports, with conflicting results. cancer. The majority of ras mutations are found in codon 12, with addi- tional mutations observed in codons 13 and 61. Nested polymerase Transforming growth factor (TGF) a and b-1 chain reaction (PCR) methods for detection of ras mutations in DNA isolated from formalin-fixed, paraffin-embedded colorectal The transforming growth factor (TGF) superfamily consists of tumour have allowed assessment of prognostic significance using several subgroups, including a and b, of which there are a number retrospective analysis of large numbers of patients. Although there of isoforms (Kumar et al, 1995; Lawrence, 1996). TGF-a produces are data in the literature on ras mutations and survival from over a mitogenic effect through interaction with the epidermal growth 1200 patients, the clinical relevance of these alterations is not yet factor receptor and promotes the growth of colon cancer cells in clear. At least four studies with a total of 498 patients demonstrate vitro in an autocrine manner. The role of TGF-a as a marker of that the presence of a ras mutation at either codon 12 or 13 is survival in colorectal cancer was evaluated in 105 patients (stage associated with a poor prognosis, independent of Dukes’ stage not given) who were treated with surgery alone (Younes et al, (Benhattar et al, 1993; Elnatan et al, 1996; Smith et al, 1996a; 1996). Patients with low TGF-a expression (<25% positive cells) Span et al, 1996). In addition, two studies with a total of 227 had a poorer prognosis than those with >25% positive staining (P = patients found a poorer prognosis in patients with mutations in 0.03). Although this brief report did not evaluate the independence both ras and p53, even when either alteration alone was not an of TGF-a as a prognostic marker, the significant difference in independent predictor of survival (Bell et al, 1993; Smith et al, overall survival makes it a candidate for further evaluation. 1996a). A single study found that ras mutations were associated TGF-b-1 is the most abundant member of the TGF-b family. with tumour progression and poor survival in patients with left- TGF-b-1 inhibits the growth of epithelial cells but stimulates sided tumours, but not in patients with right-sided disease (Elnatan mesenchymal cell proliferation and cell migration. Resistance to et al, 1996). There was no substantial difference in study design or TGF-b-1-induced growth inhibition is observed in colorectal cancer patient selection between the studies listed above and the seven and is associated with tumour progression in laboratory models studies (a total of 610 patients) in which the presence of ras muta- (Robson et al, 1996). TGF-b-1 staining was evaluated in formalin- tions was not associated with prognosis (Dix et al, 1994; Bennett et fixed tissue from 72 patients with colorectal cancer (Dukes’ stage A, al, 1995; Markowitz et al, 1995; Elnatan et al, 1996; Pricolo et al, 5; B, 33; C, 16; D, 18) (Robson et al, 1996). Positive expression was 1996; Andersen et al, 1997; Wadler et al, 1997). Most of the observed in 58% of tumours and was associated with Dukes’ classi- studies in which no significance was identified had restricted fication (Dukes’ D, 72% vs Dukes’ A, 40%; P < 0.05). Poorer analysis to a specific pathological stage of disease [Dukes’ B survival was observed in patients with positive staining for TGF- (Bennett et al, 1995); Dukes’ B/C (Dix et al, 1994); Stage III b-1. This was more apparent for patients receiving a palliative © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 194 HL McLeod and GI Murray resection of their primary tumour than for patients receiving a cura- clear need for a further large study, preferably in patients with tive resection (Robson et al, 1996). Unfortunately, evaluation of the Dukes’ B and C disease, to define the utility of EGFR immuno- independence of TGF-b-1 staining from Dukes’ stage was not histochemistry as a marker of prognosis in colorectal cancer. reported in this study. However, the data suggest a role for TGF-b-1 staining in the context of advanced disease. c-myc Amplification of the cellular oncogene c-myc has been described c-erbB2 in both colorectal cell lines and primary tumours. c-myc gene copy The c-erbB2 proto-oncogene (also called Her-2, neu) is a 185-kDa number was determined using a dot-blot technique in tumour glycoprotein with tyrosine kinase activity. Expression of c-erbB2 DNA from 149 patients with Dukes’ B or C colon cancer treated has demonstrated prognostic significance in breast and gastric on Intergroup study EST2284 (surgery vs surgery + levamisole vs cancer. However, it has not been extensively evaluated in surgery + 5-FU + levamisole) (Augenlicht et al, 1997). Using an colorectal tumours. A study of 151 adenocarcinomas (Dukes’ arbitrary cut-off point of 1.5 to define amplification, 32% of stage A, 34; B, 40; C, 77) identified a significant relationship patients had elevated c-myc. Among patients with no c-myc ampli- between c-erbB2 immunohistochemistry and patient survival fication there was no difference in survival between the three treat- (Kapitanovic et al, 1997). c-erbB2 expression was found in all ment arms. However, c-myc-positive patients treated with surgery cases and was strongly positive in 43% of tumours. An inverse + 5-FU + levamisole (n = 12) had a longer time to disease progres- correlation between c-erbB2 staining and survival was noted, with sion (P < 0.05) and a trend towards longer survival (P = 0.09) than median survival of 120 weeks and 28 weeks for tumours with c-myc-positive patients treated with surgery alone (n = 25) or moderate and strong staining for c-erbB2 respectively. Median surgery + levamisole (n = 11). While the small number of patients survival for patients with weak staining was greater than 275 with c-myc-positive tumours in each treatment group and the weeks. Although c-erbB2 expression was significantly correlated failure to assess the independence of c-myc as a marker of prog- with Dukes’ stage, c-erbB2 was an independent marker of prog- nosis limit the interpretation of this study, it does suggest a role for nosis in this study. c-erbB2 staining was also of prognostic impor- c-myc detection in the selection of patients with Dukes’ stage B or tance in a study of 164 Dukes’ B tumours (Kay et al, 1994). C colon cancer for whom adjuvant chemotherapy is indicated. Cytoplasmic staining was observed in 33.5% of tumours and was a RNA-based analysis of the prognostic impact of c-myc expres- poor prognostic indicator (5-year survival 47% compared with sion has also been performed in 119 patients (Dukes’ stage A, 31; 77% in c-erbB2-negative tumours). c-erbB2 staining was also an B, 27; C, 29; D, 32) and 29 patients (Dukes’ stage A, 2; B, 10; C, independent prognostic marker in 293 tumours (stage not stated) 10; D, 7) (Nagai et al, 1992; Smith and Goh, 1996). In addition, (Sun et al, 1995). However, unlike the previous studies, a analysis of c-myc DNA copy number was performed in the smaller favourable prognosis with overexpression was observed. This study (Nagai et al, 1992). Neither c-myc gene copy number nor finding was primarily confined to aneuploid tumours. This RNA expression was of prognostic significance in the smaller discrepancy in the prognostic significance of c-erbB2 overexpres- study (Nagai et al, 1992). In contrast, c-myc overexpression was sion needs to be more definitively addressed before this marker observed in 60% of cases in the larger study and these patients had can be recommended for use in the management of patients with a better prognosis than patients without overexpression (P = 0.02) colorectal cancer. (Smith and Goh, 1996). However, the authors found that the beneficial effect of c-myc overexpression was restricted to those Epidermal growth factor receptor tumours containing wild-type p53 (Smith and Goh, 1996). This highlights the fact that evaluation of a specific prognostic marker Epidermal growth factor receptor (EGFR) is a 170-kDa transmem- cannot be carried out in isolation but needs to be studied in the brane protein with tyrosine kinase activity. The receptor is able to context of a more complex, multigenic environment. bind epidermal growth factor and to transduce the signal into the Immunohistochemical analysis of c-myc protein in colorectal cytoplasm. Expression of EGFR is of prognostic significance for a tumours has been evaluated in two separate studies [n = 118, number of solid tumours, including breast and ovarian carcinoma. Dukes’ A, 7; B, 35; C, 41; D, 35 (Miller et al, 1992) and n = 48, EGFR was evaluated in 82 patients with colorectal cancer (Dukes’ Dukes’ B, 24; C, 24 (Bhatavdekar et al, 1997)]. Positive staining stage A, 4; B, 41; C, 22; D, 15; Mayer et al, 1993). Patients with for c-myc was found in 71% and 65% of patients, respectively, and Dukes’ C disease received adjuvant 5-FU and levamisole, while was not associated with Dukes’ stage or histological grade in Dukes’ D patients received 5-FU and leucovorin. EGFR-positive either study (Miller et al, 1992; Bhatavdekar et al, 1997). A signif- staining was detected in 80/82 (97.6%) colorectal tumours, with the icantly better overall survival was observed in the 17 patients majority (62.2%) having >50% positive cells. Patients with >50% without c-myc staining, when compared with the 31 with positive positively staining cells had a poorer survival (median 22.5 staining in the smaller study (Bhatavdekar et al, 1997). However, months) than those with a lower degree of expression (>60 months; c-myc was not of prognostic importance in the larger study (Miller P < 0.01) (Mayer et al, 1993). The role of EGFR as an independent et al, 1992). prognostic marker was not clearly defined, but Dukes’ stage and lymph vessel invasion were also shown to be of prognostic impor- TUMOUR-SUPPRESSOR GENES tance. A second study evaluated EGFR in 84 patients with Dukes’ B colorectal carcinomas who had 10 years of follow-up informa- p53 tion (Linden et al, 1996). Although 52% of tumours expressed The p53 tumour-suppressor gene is a commonly altered gene in EGFR, no association was found between EGFR and survival human solid tumours. The gene has been localized to chromosome (Linden et al, 1996). These two studies leave a great deal of uncer- 17p13.1, a region of allelic deletion in many tumours. The remaining tainty regarding the role of EGFR in colorectal cancer. There is a British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 195 Table 3 Molecular and cellular evaluation of p53 as a marker of prognosis in colorectal cancer. All immunohistochemistry (IHC) studies were performed on formalin-fixed sections and the antibody used is listed in parenthesis. Studies evaluated nuclear staining for p53 unless otherwise indicated. The studies presented here were selected from the literature based on large sample number or focus on a specific tumour stage Technique Number of Significance Reference tumours Immunohistochemical analysis IHC (DO-1) Dukes’ A = 42 Overexpression Bartik et al (1997) Dukes’ B = 82 associated with poor Dukes’ C = 84 survival in Dukes’ Dukes’ D = 45 B, but not Dukes’ A, C, D IHC (DO-7) Dukes’ A = 54 Not an independent Kressner et al (1996) Dukes’ B = 133 predictor of survival Dukes’ C = 73 Dukes’ D = 34 IHC (CM-1) Dukes’ A = 43 Cytoplasmic, but not Sun et al (1992) Dukes’ B = 101 nuclear, overexpression = Dukes’ C = 94 poor survival Dukes’ D = 50 Independent IHC on endoscopic biopsies (1801) Dukes’ A = 25 Overexpression = Yamaguchi et al Dukes’ B = 68 poor prognosis (1993) Dukes’ C = 55 Independence not Dukes’ D = 55 assessed IHC (CM1) Stage I = 35 Cytoplasmic, but not Bosari et al (1994) Stage II = 78 nuclear, overexpression = Stage III = 70 poor survival Stage IV = 23 Independent factor IHC (DO-7) Stage I = 9 Not an independent Lanza et al (1996) Stage II = 80 predictor of survival Stage III = 69 Stage IV = 46 IHC (DO-7) and PCR–SSCP exons 5–8 Dukes’ A = 14 Overexpression = Leahy et al (1996) Dukes’ B = 21 poor survival Dukes’ C = 31 Independent factor Mutation only important for Dukes’ A and B (P = 0.01) IHC (DO-1) Dukes’ A = 29 Not an independent Poller et al (1997) Dukes’ B = 88 predictor of survival Dukes’ C = 130 IHC (DO-7) and PCR + SSCP Dukes’ B = 263 Overexpression = Soong et al (1997) exons 5–8 Dukes’ C = 278 good prognosis, most significant for distal tumours Independent factor IHC (DO-1) Stage IB = 53 Overexpression = Baretton et al (1996a) Stage II = 48 poor survival Independent factor PCR-RFLP exon 4 plus IHC Dukes’ B = 168 Not associated with Bennett et al (1995) (CM1) survival IHC (1801) Dukes’ C = 107 Overexpression = Zeng et al (1994) poor survival Cytoplasmic p53 not associated with survival IHC (1801) Dukes’ D = 50, Overexpression = Belluco et al all treated with poor survival (1996) intrahepatic chemotherapy IHC (DO-1) Dukes’ D = 59 Overexpression Brett et al (1996) associated with lower response (P < 0.03), but not survival IHC (1801) Dukes’ D = 59 Overexpression not Costa et al (1997) associated with relapse or survival IHC (1801) Dukes’ D = 71 Not associated with Paradiso et al treated with response to therapy (1996) 5-FU modulated or survival with methotrexate © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 196 HL McLeod and GI Murray Table 3 Cont Technique Number of Significance Reference tumours Molecular analysis RT-PCR-SSCP Dukes’ A–D, Mutation = poor Goh et al (1995) n = 192 survival Independent of stage (P = 0.031) IHC (DO-7) and PCR–SSCP Dukes’ A = 14 Mutation = poor Leahy et al (1996) exons 5–8 Dukes’ B = 21 survival for Dukes’ Dukes’ C = 31 A and B PCR–RFLP exon 4 plus IHC Dukes’ B = 168 Not associated with Bennett et al (1995) (CM1) survival IHC (DO-7) and PCR + SSCP Dukes’ B = 263 Not associated with Soong et al (1997) exons 5–8 Dukes’ C = 278 survival allele often contains an inactivating mutation. Wild-type p53 is presence of several larger studies (250–300) has not clarified the clin- thought to be a stress response gene, the protein product of which ical relevance of p53 immunohistochemistry. acts to maintain genetic integrity by inducing cell cycle arrest and Three out of four studies of p53 and survival in advanced apoptosis following damage to the genome (Cox and Lane, 1995). colorectal cancer found no association (Brett et al, 1996; Paradiso According to this model, inactivation of p53 function by allelic loss, et al, 1996; Costa et al, 1997), with the exception being an evalua- gene mutation or protein sequestration would be expected to lead to tion of the impact in the context of intrahepatic chemotherapy for increased genetic instability and to the survival of cells with damaged liver metastasis (Belluco et al, 1996). A single study found no DNA. Most studies of p53 protein in tumours have exploited the association between either p53 staining or ras mutations and observation that wild-type p53 has a short half-life and is usually not survival, but identified a poor prognosis in patients whose tumour detectable by immunohistochemistry, while missense gene mutations contained both alterations (Bell et al, 1993). Additional studies are believed to extend greatly the half-life of the protein such that it identified additive prognostic significance when p53 and bcl-2 becomes detectable. Although there is a good correlation between the alterations were present in the same patient (Barreton et al, 1996a; presence of p53 gene mutations and positive p53 staining, discrepan- Manne et al, 1997). This suggests the need for several large, cies do exist and can occur for a number of reasons, including lack of prospective studies (>500 patients) with long clinical follow-up protein from truncating mutations, overuse of ‘antigen retrieval’ (>5 years) and a uniform treatment policy to provide definitive methods and the criteria for classifying a tumour as ‘positive’ for p53 data on the interplay between p53 and other putative prognostic (reviewed in Wynford-Thomas, 1992; Baas et al, 1994; Hall and markers to identify patients with a poor prognosis for whom more Lane, 1994). aggressive therapy may be appropriate. A large number of studies on the association between p53 Molecular analysis used either direct DNA or cDNA immunohistochemistry or mutational analysis and survival have sequencing or single-strand conformational polymorphism been published, giving data for over 3000 patients. Immuno- (SSCP) and focused on p53 exons 5–8, where 98% of mutations histochemical studies were performed using one of five different have been described. The concordance between immunohisto- monoclonal antibodies against p53, and the criteria for ‘positive’ chemical and molecular analysis is 60–85%, with both false-posi- staining varied from greater than 1% to more than 10% of tumour tive and false-negative staining observed. nuclei. Several studies used two separate p53 antibodies and Studies of mutations in p53 have also been conflicting, with the inconsistencies between antibodies were noted for the same majority showing no association with patient survival (Goh et al, tumour sections. Additional controversy also exists regarding the 1995; Leahy et al, 1996; Pricolo et al, 1996; Soong et al, 1997) localization of p53. Several studies have reported the presence of (Table 3). The studies showing no association had longer patient cytoplasmic staining for p53, which they report to be of greater follow-up, suggesting that the effort of tumour microdissection and prognostic significance than the more commonly observed nuclear the time and expense of p53 sequencing or SSCP is not a worth- staining (Sun et al, 1992; Bosari et al, 1994). It is unclear whether while exercise for identifying patients at risk of poor survival. the cytoplasmic staining represents a methodological artifact or a The largest study of p53 alterations and patient prognosis (541 finding of biological and/or prognostic importance. patients) used both immunohistochemistry and PCR-SSCP and The wide availability of antibodies for immunohistochemistry of had a mean patient follow-up time of over 7 years (Soong et al, p53 in formalin-fixed tissue has led to its use in the majority of 1997). Although no association with prognosis was identified studies assessing the association between p53 and survival (Bosari from the molecular analysis, an improved survival was found in and Viale, 1995; reviewed in Manne et al, 1997). The role of p53 patients with an overaccumulation of p53. This puzzling finding staining as a prognostic tool in colorectal cancer has been clouded by was strongest for distal colonic tumours and patients with Dukes’ a similar number of studies (and patient numbers) that have reported C disease. However, even in the left-sided tumours, the improve- poor survival in p53-positive cases as demonstrating no association ment in patient survival associated with p53 staining was 10–20%. between p53 overexpression and outcome (Table 3). This conflict in This smaller degree of survival advantage may help to explain prognostic importance exists both in studies focusing on specific why many smaller studies of mixed tumour stages have found no stage of disease and in those including Dukes’ stage A–D. Although statistically significant prognostic importance for p53. Only a the sample size of the majority of studies was 100–200 patients, the small number of patients in the study received post-operative British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 197 chemotherapy, restricting the application of these data to current B, 23; C, 47; D, 15), in which normal and tumour DNA was treatment protocols. analysed by restriction fraction length polymorphism (RFLP) Studies in cell lines with a homozygous mutant p53 genotype (Ishimaru et al, 1994). Loss of heterozygosity (LOH) at DCC gave demonstrated a high degree of resistance to radiation and several a 5.5-fold higher risk of death and was an independent marker of chemotherapy agents, including 5-FU (Lowe et al, 1993). This survival. However, Dukes’ stage was not an independent prog- provided the basis for four studies (a total of 222 patients) of the nostic marker in the multivariate analysis. association between p53 staining and response to either systemic Not all studies of 18q have demonstrated a prognostic influence. or intrahepatic 5-FU-based chemotherapy for advanced colorectal Analysis of normal and tumour DNA from 100 colorectal tumours cancer (Belluco et al, 1996; Brett et al, 1996; Paradiso et al, 1996) (Dukes’ B, 56; Dukes’ C, 44) using RFLP found no relationship or combined chemoradiation neoadjuvant therapy in rectal cancer between LOH at DCC and survival (Dix et al, 1994). The median (Spitz et al, 1997). Immunohistochemistry was performed in clinical follow-up was 18.9 months in the Dix et al (1994) study, formalin-fixed tissues using either monoclonal antibody 1801 compared with 35.4 months in the study by Jen et al (1994), and (Belluco et al, 1996; Paradiso et al, 1996) or DO-1 (Brett et al, may be a reason for the discrepancy in results for this marker in 1996; Spitz et al, 1997) and the criteria for a p53-positive tumour patients with Dukes’ B and C disease. The influence of 18q loss on ranged from greater than 1% to greater than 10% of tumour nuclei. prognosis has also been evaluated in metastatic colorectal cancer Tumour staining for p53 was not associated with response to either (Kochhar et al, 1997). LOH on 18q (microsatellite markers D18S34, biochemically modulated (with methotrexate) or intrahepatic 5-FU D18S49, D18S35, D18S58, D18S61) was assessed in 141 patients therapy (Belluco et al, 1996; Paradiso et al, 1996), but a poorer undergoing potentially curative liver resection for colorectal cancer. response rate to 5-FU and folinic acid was observed in patients The majority of tumours were non-informative for 18q markers but with p53 overexpression (Brett et al, 1996). Positive p53 staining were informative for other loci, limiting the interpretation of this in pretherapy biopsy samples was inversely correlated with a study. Among the informative cases for 18q, no relationship with complete pathological response to chemoradiation therapy and either survival or disease-free survival was observed. directly correlated with an increased likelihood of residual tumour The use of DNA-based assessment of 18q or DCC is cumbersome in the lymph nodes of surgical specimens (Spitz et al, 1997). This and requires facilities for both high-quality microdissection and suggests that p53 staining may be a tool for identifying patients quality-controlled molecular biology, and therefore is unlikely to who will benefit most from neoadjuvant chemoradiation therapy provide an approach that will have broad application to the planning or those for whom more aggressive therapy is required. of colorectal cancer treatment. There has been one published study of the prognostic implications of DCC protein staining (Shibata et al, 1996). Using a polyclonal antibody (antibody 721) and a Deleted in colon cancer/allelic variance at 18q microwave antigen retrieval protocol on formalin-fixed, paraffin- The deleted in colon cancer (DCC) gene, which is found on chro- embedded tissue sections, DCC staining was evaluated in 132 mosome 18q21.2, is a putative tumour-suppressor gene with impli- patients with colorectal cancer (stage II, 70; stage III, 62) and a cations for tumour progression and prognosis in colorectal cancer. median follow-up of 92 months. Immunohistochemistry revealed a An important step in the Vogelstein and Fearon model of colon granular cytoplasmic staining pattern for DCC and was scored as the tumorigenesis is loss of parts of chromosome 18q, including the presence of any immunoreactivity or no evidence of staining. The region containing DCC (Fearon and Vogelstein, 1990). Although expression of DCC was a prognostic factor for survival in both stage there is conflicting information on the role of the DCC protein in II and stage III disease. In patients with stage II disease, the 5-year tumour biology, there is evidence that it is of prognostic impor- survival in patients with DCC staining was 94.3% compared with tance in colorectal cancer. The observed allelic loss of chromo- 61.6% in DCC-negative tumours (P < 0.001). In stage III disease, some 18q has been linked to the prognosis of patients with stage II the 5-year survival was 59.3% with DCC staining compared with colorectal cancer (Jen et al, 1994). Analysis of normal and tumour 33.2% in DCC-negative tumours. Expression of DCC was an inde- DNA from microdissection of formalin-fixed, paraffin-embedded pendent prognostic factor which gave a threefold relative risk of sections was performed using the dinucleotide microsatellite death. Only tumour staging gave a similar level of prognostic infor- markers D18S58 and D18S61. No difference in prognosis was mation. The ease of immunohistochemical assessment of DCC will observed between patients with stage III disease and 18q loss (n = allow additional large prognostic studies to determine the role of 54) when compared with those with no allelic loss (n = 16). DCC expression in the treatment planning for colorectal cancer. However, among patients with stage II disease, the 5-year survival was 93% for those with no 18q loss (n = 29) compared with 54% p27 for those with allelic loss (n = 36, P = 0.006). The survival rate for stage II patients with 18q loss was not statistically different from The expanding understanding of cell cycle regulation has identified that of patients with stage III disease. In multiple regression putative oncogenes and tumour-suppressor genes that may be proportional hazards models, 18q was an independent prognostic involved in colorectal tumorigenesis. However, there have been marker and 18q loss gave a 2.5 to 3.0-fold higher risk of death. The few evaluations of the prognostic implications of alterations in cell prognostic importance of 18q allelic loss was verified in a study of cycle proteins. The cell cycle inhibitor p27 (also known as Kip 1) is 126 patients (stage I, 32; II, 50; III, 44), using four microsatellite a putative tumour-suppressor gene, but inactivating mutations have markers on microdissected tumour (Ogunbiyi et al, 1998). Loss of not been described in human tumours. Immunohistochemical 18q was an independent marker of poor prognosis, with a 1.65 analysis of p27 in formalin-fixed, paraffin-embedded sections relative risk of disease-free survival. Loss of 18q was prognosti- detected variable expression in the cytoplasm, nucleus, or both, in cally important in both stage II and III disease. A poor prognosis in 90% of colorectal tumours, when using a microwave-based method patients with loss of a polymorphic allele in DCC was also (Loda et al, 1997). Scoring of p27 staining was expressed as a ratio observed among 95 patients with colorectal cancer (Dukes’ A, 10; of positive cells to total number counted and all immunoreactive © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 198 HL McLeod and GI Murray cells were considered positive. A significant relationship between provides a growth advantage to tumour cells in vitro by blocking p27 staining and prognosis was observed among 149 patients with apoptosis. However, evaluation of bcl-2 protein expression in colorectal adenocarcinoma (stage I, 31; II, 66; III, 51; IV, 1), when lung, thyroid and breast carcinomas identified a favourable prog- median survival was 151 months in patients with p27-positive nosis in patients with overexpression of bcl-2 on tumour staining. tumours (n = 134) and 69 months in patients with p27-negative The association between bcl-2 immunohistochemistry and patient tumours (n = 15). Both tumour stage and grade showed a signifi- survival has been evaluated in at least six studies of colorectal cant association with survival. The prognostic significance of p27 tumours (a total of 932 patients) (Bosari et al, 1995; Ofner et al, staining was independent of tumour stage and grade (P = 0.003). 1995; Sinicrope et al, 1995; Baretton et al, 1996b; Manne et al, The effect of p27 was observed when analysis was confined to 1997; Schneider et al, 1997). The larger studies (n = 205 and 231) patients with stage II disease, with absence of staining giving a 2.9 combined Dukes’ stages A–D and found no association between relative risk of death from colorectal cancer. These encouraging bcl-2 overexpression and relapse or survival (Bosari et al, 1995; data now needs confirmation in a larger independent data set Schneider et al, 1997). In addition, bcl-2 staining was not associ- containing patients with uniform treatment before widespread use ated with response to 5-FU-based chemotherapy (Schneider et al, of p27 as a prognostic marker can be recommended. 1997). Bcl-2-positive tumours were associated with a better prog- nosis in the remaining studies, although not always as an indepen- Allelic loss/genome damage dent marker of overall survival (Ofner et al, 1995; Sinicrope et al, 1995; Baretton et al, 1996b; Manne et al, 1997). An interaction Although allelic instability is a common event in colorectal cancer, between bcl-2 overexpression and p53 staining was identified in the prognostic impact of instability has not been extensively eval- several studies, with bcl-2-positive/p53-negative tumours being uated. Seventy-five patients (stage not reported) were evaluated associated with a better survival than tumours with other staining for the impact of allelic deletion in chromosomes 5q and 17p on patterns (Baretton et al, 1996b; Manne et al, 1997). This supports survival, using Southern blot analysis (Khine et al, 1995). Loss of the need for studies of the interplay between bcl-2 and other puta- chromosome 5q was observed in 32% of cases and 17p loss was tive prognostic markers to identify patients with a poor prognosis found in 69%, with loss of both regions in 27% of patients and for whom alternative therapy may be appropriate. neither region in 25%. Poorer survival in patients with loss of bcl-2 does not act alone but forms heterodimers with other chromosome 5q was apparent (P = 0.014). Although patients with family members, including BAX, to regulate apoptosis in in vitro 17p allelic deletion had a poorer survival, this did not reach systems. However, there are no data on the association between statistical significance (P = 0.16). Multivariate analysis identified BAX staining and prognosis in colorectal cancer. The recent Dukes’ stage and allelic loss of chromosome 5q as the two inde- development of techniques for the use of BAX antibodies in pendent markers of survival (P = 0.0001). An additional study formalin-fixed, paraffin-embedded tumour tissue should allow this evaluated allelic loss on chromosome 1p in 116 colorectal marker to be assessed in the near future (Apolinario et al, 1997; tumours, of which 82 (stage I, 17; II, 36; III, 29) were informative Koshida et al, 1997). and evaluable for survival analysis (Ogunbiyi et al, 1997). Loss of 1p was detected in 22/82 tumours and was associated with a poorer disease-free survival. Loss of 1p was a weak but statisti- CELL PROLIFERATION cally significant independent marker of survival in multivariate Cell proliferation has been extensively studied using a variety of analysis. Although the methodology used in these studies is too direct and surrogate markers. Currently one of the most popular cumbersome for widespread application, the availability of methods of assessing cell proliferation is using antibodies that microsatellite markers of regions throughout the genome may recognize nuclear proteins involved in, or altered during, the cell facilitate further evaluation of this approach for identifying cycle. One of the most frequently studied targets is proliferating patients at risk for poor long-term survival. cell nuclear antigen (PCNA). This is a DNA polymerase d acces- Genomic DNA fingerprinting has recently been used to assess sory protein which can complex with cyclin D and cyclin-depen- the prognostic implications of tumour genome instability (Arribas dent kinases (i.e. p21). Ki-67 is another frequently used antibody, et al, 1997). By comparing banding patterns from arbitrarily detecting an antigen expressed in all phases of the cell cycle primed PCR analysis of 63 colorectal tumours and adjacent except G . Evaluation of PCNA in 79 colorectal tumours (Dukes’ normal tissue, an index of genomic instability was derived and was A, 4; B, 48; C, 27) by immunohistochemistry was performed with an independent marker of prognosis. This technique does not follow-up for a minimum of 10 years (Neoptolemos et al, 1995). A identify specific target genes for development as markers or PCNA score was derived by a relatively complex, subjective four- cellular targets for chemotherapy, but does provide an apparently point scale according to the percentage of positive tumour cells. reproducible method for detection of genome-wide instability for The PCNA index was not associated with patient survival. Sun et high-throughput prognostic studies. The small number of patients al (1996) studied PCNA in a series of 293 colorectal carcinomas at each disease stage limits the conclusions from the initial study, (Dukes’ A, 43; B, 101; C, 94; D, 50; stage not available for 5) but does provide the basis for further analysis of this methodology which had been collected over a 15-year period and followed up as a potential marker of patient survival. for a minimum of 5 years. PCNA was assessed by counting 500 tumour nuclei and the final PCNA score was divided into four APOPTOSIS PATHWAYS groups according to the percentage of positive nuclei. PCNA did not correlate with survival. This group also studied PCNA expres- bcl-2 sion in 56 primary tumours and their corresponding lymph node The bcl-2 proto-oncogene encodes a 25-kDa protein that is metastases with good correlation between the tissues. PCNA involved in the regulation of cell death by inhibiting apoptosis expression was investigated in 71 advanced colorectal cancers through mechanisms which are not yet defined. Its overexpression (metastatic, n = 33, or recurrent disease, n = 38) eligible to receive British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 199 5-FU-based therapy (Paradiso et al, 1996). PCNA was scored by tumour invasion and metastasis. The current concept of the MMPs analysing 1000 nuclei, and a ‘high’ PCNA score (greater than 32% in tumour invasion is that they not only have a direct role in facili- positive cells) was associated with improved survival and was tating extracellular matrix degradation but as a consequence they independently prognostic. also have an important role maintaining the tumour environment and thus promoting tumour growth (Chambers and Matrisian, 1997). There are currently at least 15 known MMPs, which are ANGIOGENESIS/METASTASIS/INVASION broadly classified into collagenases, gelatinases, stromelysins and membrane-type MMPs. The MMPs (except membrane-type Thymidine phosphorylase MMPs) are produced as proenzymes and are activated by cleavage Thymidine phosphorylase, or platelet-derived endothelial cell of N-terminal propeptides. Our group identified MMP-1 (intersti- growth factor, is an enzyme that catalyses the reversible phos- tial collagenase) in 64 patients with colorectal cancers (Dukes’ A, phorylysis of thymidine and deoxyuridine to their bases and de- 1; B, 38; C, 25). The presence of MMP-1 in tumour cells was asso- oxyribose phosphate. It has been shown to be capable of inducing ciated with poorer prognosis and was independent of Dukes’ stage angiogenesis, and expression of thymidine phosphorylase may (Murray et al, 1996). This study used a simple assessment of the also be important for the activation of certain prodrugs, for presence or absence of MMP-1 immunoreactivity and is likely to example deoxyfluorouridine conversion to fluorouracil (Griffiths be more reproducible than some of the more complex scoring and Stratford, 1997). Thymidine phosphorylase was evaluated in schemes used in other immunohistochemistry studies. MMP-9 93 colorectal cancers (Dukes’ stage A, 30; B, 42; C, 21) by mRNA has been identified by in situ hybridization and Northern immunohistochemistry using a monoclonal antibody (Takebayashi blotting in 71 tumours (Dukes’ A, 11; B, 18; C, 16; D, 26), and a et al, 1996). The immunohistochemical assessment used was a high ratio of tumour to normal MMP-9 mRNA was associated straightforward two-point scale, with tumour being classified as with poor prognosis and was independent of Dukes’ stage (Zeng et positive if more than 5% of tumour cells displayed staining. al, 1996). In a small group (n = 31) of colorectal cancers, total type Normal colon mucosa did not stain, whereas staining was IV collagenase activity (MMP-2 and MMP-9) has been found to observed in both tumour and stromal cells. No comment was made be prognostically significant, with high activity associated with regarding the significance of stromal cell staining in the absence of both shorter disease-free survival and shorter overall survival tumour cell staining or how this was scored. The presence of (Ambiru et al, 1997). These findings of the different MMPs thymidine phosphorylase was associated with a significantly emphasize the need to perform a comprehensive series of studies poorer prognosis. Although expression correlated with other of MMPs in colorectal cancer to understand the interactions histopathological criteria (lymph node metastasis, lymphatic and between the individual MMPs and to identify which of the MMPs venous invasion), thymidine phosphorylase also appeared to be an are most appropriate as prognostic markers. Studies are also independent prognostic factor. required to determine whether it is more relevant to assess total enzyme or active enzyme. Clinical trials of MMP inhibitors are now under way and require evaluation in relation to tumour MMP Vascular endothelial growth factor expression to determine if MMP expression is a reliable method of Vascular endothelial growth factor (VEGF) is a glycoprotein that selecting patients for anti-MMP therapy. The MMPs are inhibited shares some homology with platelet-derived growth factor and is in vivo by naturally occurring inhibitors, tissue inhibitors of intimately involved in promoting angiogenesis. Interest in angio- metalloproteinases (TIMPs). There are currently four known genesis has grown because of the possibility of using anti-angio- TIMPs, and studies are also necessary to determine the contribu- genic therapy to prevent tumour progression. There have been tion of TIMPs to prognosis. several recent reports of the analysis of VEGF in colorectal cancer, all of which used immunohistochemistry on formalin-fixed, wax- Urokinase-type plasminogen activator and inhibitor embedded sections and showed that VEGF is primarily localized to tumour cells. In the largest study of 175 colorectal carcinomas Urokinase-type plasminogen activator (uPA) is a serine protease (stage 0, 7; I, 24; II, 48; III, 59; IV, 37), the presence of VEGF which catalyses the activation of plasminogen to plasmin, a broad- correlated with increasing tumour stage and was associated with spectrum protease which in turn can break down a variety of extra- decreased overall survival (P < 0.05) and decreased disease-free cellular matrix components and also promote the activation of survival (P < 0.01) (Kang et al, 1997). However, it was not an MMP-2. In a series of Dukes’ B colon cancers (n = 70), a higher independent prognostic factor. Amaya et al (1997) studied the proportion of uPA-positive tumour cells by immunohistochemistry presence of VEGF in 133 cases of colorectal cancer (stage I, 14; II, was associated with poorer survival (P = 0.01) (Mucalhy et al, 42; III, 45; IV, 29 with three unclassified cases). VEGF was local- 1994). However, most tumours had a high percentage of positive ized to tumour cells, and VEGF-positive tumours had a signifi- tumour cells (57/70) and only 13 cases had a low percentage of cantly poorer prognosis than VEGF-negative tumours (P < 0.02), positive tumour cells. In addition, stromal staining for uPA showed which remained significant in a multivariate analysis. A much an association with survival. This study is one of relatively few to smaller study of VEGF in 27 Dukes’ stage B patients showed study a marker in relation to a single Dukes’ stage. The patients in essentially similar results (Takahashi et al, 1997). this study all had uniform treatment with no post-operative radio- or chemotherapy and follow-up was for between 5 and 10 years. High tumour levels of plasminogen activator inhibitor-2 (PAI-2) in Matrix metalloproteinases and inhibitors Dukes’ B (n = 82) and C (n = 54) colorectal cancers has also been The matrix metalloproteinases (MMPs) are a group of enzymes found to be prognostically significant (P = 0.05), but not indepen- with a central role in matrix degradation, and there is considerable dent of Dukes’ stage, in a group of 136 colorectal cancers (Ganesh evidence to indicate that this group of enzymes have a key role in et al, 1997). © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 200 HL McLeod and GI Murray CD44 colorectal cancer, 5-FU and raltitrexed (tomudex). Elevated expression of the TS protein has been demonstrated in cell lines CD44 is a cell-surface glycoprotein which functions as an adhesion which were made resistant to either 5-FU or tomudex (Johnston et molecule. Multiple isoforms of CD44 are generated by alternative al, 1992; Jackman et al, 1995). A higher degree of tumour TS splicing of up to ten exons that encode for the molecule’s extracel- mRNA and protein was observed in patients with a poor response lular portion. CD44 variants have been identified on the surface of a to 5-FU in a small study of colorectal (n = 9) and gastric (n = 12) variety of tumour cells. CD44v6 was studied by immunohistochem- cancers (Johnston et al, 1995). TS mRNA expression provided a istry in a series of colon cancers (n = 68) with follow-up of between clearer definition of patient response than did protein level in this 6.5 and 9.5 years (Mulder et al, 1994). The report did not indicate study. Quantitative PCR assay of TS mRNA was then applied to the distribution of tumours in different Dukes’ stages or the post- tumour tissue from 46 patients with advanced colorectal cancer operative treatments received by the patients. Tumours that strongly –2 receiving protracted infusion of 5-FU (200 mg m per day for 21 expressed CD44v6 had a poorer prognosis than those tumours that days) and weekly leucovorin (Leichman et al, 1997). No patient showed less intense or no immunostaining. This was stated to be with a ratio of TS to b-actin (used as an intertumour control) independent of Dukes’ stage. CD44 exons 8–10 was studied using –3 greater than 4.1 ´ 10 responded. Using the median TS/b-actin immunohistochemistry in 215 colorectal tumours (stage I, 37; II, 71; ratio as a cut-off point, survival was significantly higher (median III, 85; IV, 22) (Yamaguchi et al, 1996). There was no significant 13.6 months) in patients with less than median TS levels than in immunoreactivity for CD44 in normal mucosa while there was those with greater than median values (8.2 months) (Leichman et positive immunoreactivity in tumour cells. Immunoreactivity was al, 1997). Similar results have recently been reported in patients assessed as either present or absent, although the cut-off point for receiving hepatic artery infusion of fluoropyrimidines, when negative score was less than 25% of tumour cells. The presence of patients with low TS mRNA expression were 4.1-fold more likely CD44 was associated with poorer prognosis, and in multivariate to respond to chemotherapy (Kornmann et al, 1997). TS anti- analysis this factor remained independently prognostic. bodies, which work on formalin-fixed, paraffin-embedded tumour tissue, have provided the opportunity for retrospective analysis of E-cadherin the prognostic significance of TS staining in patients with E-cadherin is a member of the cadherin family of calcium-depen- colorectal cancer. TS was assessed with a semiquantitative visual dent adhesion molecules that function as mediators of cell-to-cell grading system based on the intensity of staining (0 to +++) and adhesion. It is a transmembrane protein with a large extracellular extent of staining (focal, <25% positive cells; diffuse, >25% posi- domain and a smaller cytoplasmic component which interacts with tive cells) (Johnston et al, 1994). An initial study in 294 patients cytoplasmic proteins, particularly b-catenin. Loss of E-cadherin is (Dukes’ stage A, 32; B, 92; C, 146; D, 24) from NSABP study R- associated with increased cell invasiveness. E-cadherin was studied 01 demonstrated that high expression of TS protein in tumour is by in situ hybridization in 49 Dukes’ B colorectal cancers (Dorudi et associated with a poor survival rate in patients receiving 5-FU- al, 1995). In situ hybridization was performed because antibodies based adjuvant chemotherapy for rectal cancer (Johnston et al, which reliably detect E-cadherin in formalin-fixed, wax-embedded 1994). TS staining intensity was an independent prognostic sections do not appear to be currently available. The lowest expres- marker and identified a significant impact of 5-FU-based sion of E-cadherin was associated with poor survival. In a second chemotherapy in patients with high TS levels. However, TS study, E-cadherin expression was investigated by in situ hybridiza- staining did not predict disease response among 134 patients tion in 32 colorectal cancers (Dukes’ A, B and D). None of the receiving 5-FU-based chemotherapy for advanced colorectal patients had received chemotherapy and all had been followed up cancer (Findlay et al, 1997). A significant time interval between for at least 5 years (Kitadai et al, 1996). The profile of expression of surgical resection of the primary tumour and treatment of E-cadherin was also compared with several other genes involved in advanced disease was observed for most patients. As this study tumour invasion, including type IV collagenase. There was no prog- performed TS staining on the primary tumour only, this may point nostic significance of E-cadherin expression, although low levels of to clinically significant differences between TS staining in primary E-cadherin and a high ratio of type IV collagenase to E-cadherin and metastatic tumour. Although TS staining has demonstrated were associated with an increased incidence of metastasis. potential prognostic significance, prospective studies are required to confirm the importance in adjuvant chemotherapy and to deter- Helix pomatia agglutinin mine the role of TS levels in advanced colorectal cancer. Helix pomatia agglutinin (HPA) is a sugar-binding protein, or lectin, which binds specifically to N-acetylgalactosamine residues. GENERAL MARKERS The binding of this lectin has been suggested to be of prognostic Carcinoembryonic antigen significance in breast cancer. HPA binding was studied in 130 colorectal cancers (Dukes’ stage A, 14; B, 49; C, 62; unknown, 5). Carcinoembryonic antigen is often used as a serum marker in Using a histochemical approach to detect binding of this lectin follow-up studies of colorectal cancer to detect recurrences/ (Schumacher et al, 1994). HPA binding correlated with Dukes’ metastasis. However, generally, the detection of CEA staining in stage, but was not independently prognostic. colorectal tumour samples has not been found to be useful in providing prognostic information. INTRACELLULAR TARGETS OF CHEMOTHERAPY Sialyl Lewis antigen Thymidylate synthase The sialyl Lewis antigen (sLe ) is a cell-surface glycoprotein which is part of the blood group system of antigens and may serve as Thymidylate synthase (TS) is the primary intracellular target for a ligand for endothelial leucocyte adhesion molecule. This may two of the more commonly used chemotherapeutic agents for British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 201 human colonic carcinoma cell lines and tumours: a frequent, p53-independent facilitate the binding of tumour cells to endothelial cells, which is a mutation associated with improved outcome in a randomized multi-institutional step in the metastatic cascade. High expression of this antigen is trial. Cancer Res 57: 1769–1775 associated with increased tumour cell invasion. The presence of Baas IO, Mulder JR, Offerhaus GJA, Vogelstein B and Hamilton SR (1994) An sLe in 132 colorectal cancers (stage I, 36; II, 40; III, 38; IV, 18) evaluation of six antibodies for immunohistochemistry of mutant p53 gene product in archival colorectal neoplasms. J Pathol 172: 5–12 was evaluated by immunohistochemistry on formalin-fixed, wax- Baretton GB, Vogt M, Muller C, Diebold J, Schneiderbanger K, Schmidt M and embedded sections with a monoclonal antibody to this antigen Lohrs U (1996a) Prognostic significance of p53 expression, chromosome 17 (Nakamori et al, 1993). Tumours with greater than 5% of positive copy number, and DNA ploidy in non-metastasized colorectal carcinomas cells were assessed as positive and those with less than 5% positive (stages IB and II). Scand J Gastroenterol 31: 481–489 cells were scored as negative. Follow-up was between 5 and 7 Baretton GB, Diebold J, Christoforis G, Vogt M, Müller C, Dopfer K, Schneiderbanger K, Schmidt M and Löhrs U (1996b) Apoptosis and years. The presence of the sLe was associated with depth of immunohistochemical bcl-2 expression in colorectal adenomas and carcinomas. tumour invasion, lymphatic invasion, lymph node metastasis and Aspects of carcinogenesis and prognostic significance. Cancer 77: 255–264 tumour stage. Overall and disease-free 5-year survival was signifi- Bartik Z, Nordenskiöld BO and Sun Z-F (1997) p53 overexpression as a prognostic cantly less in the sLe -positive tumours, and this factor was also factor in patients with Dukes’ B colorectal adenocarcinoma. Int J Oncol 11: 1019–1023 independently prognostic for both overall survival and disease-free Bell SM, Scott N, Cross D, Sagar P, Lewis FA, Blair GE, Taylor GR, Dixon MF and survival. Quirke P (1993) Prognostic value of p53 overexpression and c-Ki-ras gene mutations in colorectal cancer. Gastroenterology 104: 57–64 CONCLUSIONS Belluco C, Guillem JG, Kemeny N, Huang Y, Klimstra D, Berger MF and Cohen AM (1996) p53 Nuclear protein overexpression in colorectal cancer: a Although the difficulty in identifying markers of prognosis has led dominant predictor of survival in patients with advanced hepatic metastases. to much scepticism, there is still a real need for methodologies to J Clin Oncol 14: 2696–2701 Benhattar J, Losi L, Chaubert P, Givel J-C and Costa J (1993) Prognostic guide the choice of therapy and use health care expenditure in a significance of K-ras mutations in colorectal carcinoma. Gastroenterology 104: more rational manner for this common disease. Future studies 1044–1048 must focus on multiple markers in large numbers of patients at a Bennett MA, Kay EW, Mulcahy H, O’Flaherty L, O’Donoghue, DP, Leader M and single disease stage in order to provide definitive evidence on the Croke DT (1995) ras and p53 in the prediction of survival in Dukes’ stage B clinical utility of prognostic markers and their relative contribution colorectal carcinoma. J Clin Pathol 48: M310–M315 Bhatavdekar JM, Patel DD, Ghosh N, Chikhlikar PR, Trivedi TI, Suthar TP, Doctor to the prediction of survival. The recent emphasis by the National SS, Shah NG and Balar DB (1997) Coexpression of Bcl-2, c-Myc, and p53 Cancer Institute on pathology archives linked to large therapeutic oncoproteins as prognostic discriminants in patients with colorectal carcinoma. studies as a resource for collaborative research is an example of Dis Colon Rectum 40: 785–790 the way funding can be used to answer these important questions Bosari S and Viale G (1995) The clinical significance of p53 aberrations in human tumors. Virchows Arch A Pathol Anat Histopathol 427: 229–241 about patient care. Bosari S, Viale G, Bossi P, Maggioni M, Coggi G, Murray JJ and Lee AKC (1994) Cytoplasmic accumulation of p53 protein: an independent prognostic indicator ACKNOWLEDGEMENTS in colorectal adenocarcinomas. J Natl Cancer Inst 86: 681–687 Bosari S, Moneghini L, Graziani D, Lee AKC, Murray JJ, Coggi G and Viale G Research in the authors’ laboratories are supported in part by an (1995) Bcl-2 oncoprotein in colorectal hyperplastic polyps, adenomas, and Aberdeen Colorectal Cancer Initiative grant from the University of adenocarcinomas. Human Pathol 26: 534–540 Aberdeen Development Trust. The helpful comments of Professor Brett MC, Pickard M, Green B, Howel-Evans A, Smith D, Kinsella A and Poston G (1996) p53 protein overexpression and response to biomodulated 5-fluorouracil J Cassidy, and Dr J McKay were greatly appreciated. Many thanks chemotherapy in patients with advanced colorectal cancer. Eur J Sur Oncol 22: to Maria Johnson for her secretarial skills. 182–185 Chambers AF and Matrisian LM (1997) Changing views of the role of matrix metalloproteinases in metastasis. J Natl Cancer Inst 89: 1260–1270 REFERENCES Cohen AM, Tremiterra S, Candela F, Thaler HT and Sigurdson ER (1991) Prognosis of node-positive colon cancer. Cancer 67: 1859–1861 Amaya H, Tanigawa N, Lu C, Matsumura M, Shimomatsuya T, Horiuchi T and Costa A, Doci R, Machen C, Bignami P, Faranda A, Gennari L and Silvestrini R Muroaka R (1997) Association of vascular endothelial growth factor expression (1997) Cell proliferation-related markers in colorectal liver metastases: with tumor angiogenesis, survival and thymidine phosphorylase/platelet- correlation with patients prognosis. J Clin Oncol 15: 2008–2014 derived endothelial cell growth factor expression in human colorectal cancer. Cox LS and Lane DP (1995) Tumor suppressors, kinases and clamps – how p53 Cancer Lett 119: 227–235 regulates the cell-cycle in response to DNA-damage. Bioessays 17: 510–508 Ambiru S, Miyazaki M, Ito H, Nakagawa K, Shimizu H, Nukui Y, Nozawa S, Okuno Dix BR, Robbins P, Soong R, Jenner D, House AK, Lacopetta BJ, and The General A, Yoshitomi H and Nakajima N (1997) A prospective study of prognostic Surgeons at Sir Charles Gairdner Hospital (1994) The common molecular value of type IV collagenase in colorectal cancer tissue. Dig Dis Sci 42: genetic alterations in Dukes’ B and C colorectal carcinomas are not short-term 1660–1665 prognostic indicators of survival. Int J Cancer 59: 747–751 Anderson SN, Lovig T, Breivik J, Lund E, Guadernack G, Meling GI and Rognum Dorudi S, Hanby AM, Poulsom R, Northover J and Hart IR (1995) Level of TO (1996) K-ras mutations and prognosis in large-bowel carcinomas. Scand J expression of E-cadherin mRNA in colorectal cancer correlates with clinical Gastroenterol 32: 62–69 outcome. Br J Cancer 71: 614–616 Apolinario RM, van der Valk P, de Jong JS, Deveille W, Van Arkotte J, Dingemans Elnatan J, Goh H-S and Smith DR (1996) C-KI-RAS activation and the biological A-MC, Van Mourik JC, Postmus PE, Pinedo HM and Giaccone G (1997) behaviour of proximal and distal colonic adenocarcinomas. Eur J Cancer 32A: Prognostic value of the expression of p53, bcl-2, and bax oncoproteins, and 491–497 neovascularization in patients with radically resected non-small-cell lung Fearon ER and Vogelstein BA (1990) The genetic model for colorectal cancer. J Clin Oncol 15: 2456–2466 tumorigenesis. Cell 61: 759–767 Arribas R, Capellà G, Tórtola S, Masramon L, Grizzle WE, Perucho M and Peinado Findlay MPN, Cunningham D, Morgan G, Clinton S, Hardcastle A and Aherne GW MA (1997). Assessment of genomic damage in colorectal cancer by DNA (1997) Lack of correlation between thymidylate synthase levels in primary fingerprinting: prognostic applications. J Clin Oncol 15: 3230–3240 colorectal tumours and subsequent response to chemotherapy. Br J Cancer 75: Astler VB and Coller FA (1954) The prognostic significance of direct extension of 903–909 carcinoma of the colon and rectum. Ann Surg 192: 846–850 Ganesh S, Sier CFM, Heerding MM, Vankrieken JHJM, Griffioen G, Welvaart K, Augenlicht LH, Wadler S, Corner G, Richards C, Ryan L, Multani AS, Pathak S, van de Velde CJ, Verheijen JH, Lamers, CB and Verspaget HW (1996) Benson A, Haller D and Heerdt BG (1997) Low-level c-myc amplifications in Contribution of plasminogen activators and their inhibitors to the survival © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 202 HL McLeod and GI Murray prognosis of patients with Dukes’ stage B and C colorectal cancer. Br J Cancer Leahy DT, Salman R, Mulcahy H, Sheahan K, O’Donoghue DP and Parfrey NA 75: 1793–1801 (1996) Prognostic significance of p53 abnormalities in colorectal carcinoma Goh H-S, Yao J and Smith DR (1995) p53 Point mutation and survival in colorectal detected by PCR-SSCP and immunohistochemical analysis. J Pathol 180: cancer patients. Cancer Res 55: 5217–5221 364–370 Griffiths L and Stratford IJ (1997) Platelet-derived endothelial cell growth factor Leichman CG, Lenz H-J, Leichman L, Danenberg K, Baranda J, Groshen S, Boswell thymidine phosphorylase in tumour growth and response to therapy. Br J W, Metzger R, Tan M and Danenberg PV (1997) Quantitation of intratumoral Cancer 76: 689–693 thymidylate synthase expression predicts for disseminated colorectal cancer Hall PA and Lane DP (1994) p53 in tumour pathology: can we trust response and resistance to protracted-infusion fluorouracil and weekly immunohistochemistry? – Revisited! J Pathol 172: 1–4 leucovorin. J Clin Oncol 15: 3223–3229 Hermanek P, Wiebelt H, Staimner D and Riedl S (1995) Prognostic factors of rectum Linden MD, Nathanson DS and Jacobson G (1996) Lack of prognostic significance carcinoma – experience of the German multicentre study SGCRC. German of epidermal growth factor receptor expression in Dukes’ B colorectal Study Group Colo-Rectal Carcinoma. Tumori 81: 60–64 carcinomas. Lab Invest 74: 61A:343 Hoffmann D, Moore J and Roder D (1997) Trends in survival from colonic cancer: Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta GF, Jessup JM and The impact of subspecialization. Aus N Z J Surgery 67: 842–845 Pagano M (1997) Increased proteasome-dependent degradation of the cyclin Ishimuru G, Ookawa K, Yamaguchi N, Sakamoto M, Hirohashi S, Muto T and dependent kinase inhibitor p27 in aggressive colorectal carcinomas. Nature Yokota J (1994) Allelic losses associated with the metastatic potential of Med 3: 231–234 colorectal carcinoma. Int J Oncol 5: 267–273 Lowe SW, Ruley HE, Jacks T and Housman DE (1993) p53-dependent apoptosis Jackman AL, Kelland LR, Kimbell R, Brown M, Gibson W, Aherne GW, Hardcastle modulates the cytotoxicity of anticancer agents. Cell 74: 957–967 A and Boyle FT (1995) Mechanisms of acquired resistance to the quinazoline Manne U, Myers RB, Moron C, Poczatek RB, Dillard S, Weiss H, Brown D, thymidylate synthase inhibitor ZD1694 (tomudex) in one mouse and three Srivastava S and Grizzle WE (1997) Prognostic significance of Bcl-2 human cell lines. Br J Cancer 71: 914–924 expression and p53 nuclear accumulation in colorectal adenocarcinoma. Int J Jen J, Kim HG, Piantadosi S, Liu ZF, Lecitt RC, Sistonen P, Kinzler KW, Vogelstein Cancer 74: 346–358 B and Hamilton SR (1994) Allelic loss of chromosome 18q and prognosis in Markowitz S, Hines JD, Lutterbaugh J, Myeroff L, Mackay W, Gordon N, Rustum colorectal cancer. N Engl J Med 331: 213–221 Y, Luna E and Kleinerman J (1995) Mutant K-ras oncogenes in colon cancers Johnston PG, Drake JC, Trepel J and Allegra CJ (1992) Immunological quantitation do not predict patient’s chemotherapy response or survival. Clin Cancer Res 1: of thymidylate synthase using the monoclonal antibody TS 106 in 5- 441–445 fluorouracil-sensitive and -resistant human cancer cell lines. Cancer Res 52: Mayer A, Takimoto M, Fritz E, Schellander G, Kofler K and Ludwig H (1993) The 4306–4312 prognostic significance of proliferating cell nuclear antigen, epidermal growth Johnston PG, Fisher ER, Rockette HE, Fisher B, Wolmark N, Drake JC, Chabner factor receptor, and mdr gene expression in colorectal cancer. Cancer 71: BA and Allegra CJ (1994) The role of thymidylate synthase expression in 2454–2460 prognosis and outcome of adjuvant chemotherapy in patients with rectal cancer. McArdle CS and Hole D (1991) Impact of variability among surgeons on post- J Clin Oncol 12: 2640–2647 operative morbidity and mortality and ultimate survival. Br Med J 302: Johnston PG, Lenz H-J, Leichman CG, Danenberg KD, Allegra CJ, Danenberg PV 1501–1505 and Leichman L (1995) Thymidylate synthase gene and protein expression Miller F, Heimann TM, Quish A, Pyo DJ, Szporn A, Martinelli G and Fasy TM correlate and are associated with response to 5-fluorouracil in human colorectal (1992) Ras and c-myc protein expression in colorectal carcinoma. Dis Colon and gastric tumours. Cancer Res 55: 1407–1412 Rectum 35: 430–435 Kang SM, Maeda K, Chung YS, Onodda N, Ogawa Y, Takatsuka S, Ogawa M, Miyahara M, Saito T, Kaketani K, Sato K, Kuwahara A, Shimoda K and Kobayashi Sawada T, Nakata B, Nishiguchi Y, Ikehara T, Okuno M and Sowa M (1997) M (1991) Clinical significance of ras p21 overexpression for patients with an Vascular endothelial growth factor expression correlates with hematogenous advanced colorectal cancer. Dis Colon Rectum 34: 1097–1102 metastasis and prognosis in colorectal carcinoma. Oncol Rep 4: 381–384 Moertel CG, Fleming TR, MacDonald JS, Haller DG, Laurie JA, Tangen CM, Kapitanovic S, Radosevic S, Kapitanovic M, Ajndelinovic S, Ferencic Z, Tavassoli Ungerleider JS, Emerson WA, Tormey DC, Glick JH, Veeder MH, Mailliard JA M, Primorac D, Sonicki Z, Spaventi S, Pavelic K and Spaventi R (1997) The and Graff J (1995) Fluorouracil plus levamisole as effective adjuvant therapy HER-2/neu expression of p185 correlates with the stage of disease and survival in after resection of stage III colon carcinoma: a final report. Ann Intern Med 122: colorectal cancer. Gastroenterology 112: 1103–1113 321–326 Kay EW, Mulcahy H, Walsh CB, Leader M and O’Donoghue D (1994) Cytoplasmic Mulcahy HE, Duffy MJ, Gibbons D, McCarthy P, Parfrey NA, O’Donoghue DP and c-erb B2 protein expression correlates with survival in Dukes’ B colorectal Sheahan K (1994) Urokinase type plasminogen activator and outcome in carcinoma. Histopathology 25: 455–461 Dukes’ B colorectal cancer. Lancet 344: 583–584 Khine K, Goh H-S and Smith DR (1995) Prognostic significance of chromosome Mulder JWR, Kruyt PM, Sewnath M, Oosting J, Seldenrijk CA, Weidema WF, 5q and 17p allelic deletion in colorectal adenocarcinomas. Int J Oncol 7: Offerhaus GJ and Pals ST (1994) Colorectal cancer prognosis and expression 631–635 of exon-v6 containing CD44 proteins. Lancet 344: 1470–1472 Kitadai Y, Ellis LM, Tucker SL, Greene GF, Bucana CD, Cleary KR, Takahashi Y, Murray GI, Duncan ME, O’Neil P, Melvin WT and Fothergill JE (1996) Matrix Tahara E and Fidler IJ (1996) Multiparametric in situ mRNA hybridization metalloproteinase-1 is associated with poor prognosis in colorectal cancer. analysis to predict disease recurrence in patients with colon carcinoma. Am J Nature Med 2: 461–462 Path 149: 1541–1551 Nagai MA, Habr-Gama A, Oshima BS and Bretani MM (1992) Association of Kochhar R, Halling KC, McDonnell S, Schaid DJ, French AJ, O’Connell MJ, genetic alterations of c-myc, c-fos, and c-Ha-ras proto-oncogenes in colorectal Nagorney DM and Thibodeau SN (1997) Allelic imbalance and microsatellite tumors. Frequency and clinical significance. Dis Colon Rectum 35: 444–451 instability in resected Dukes’ D colorectal cancer. Diagn Mol Pathol 6: 78–84 Nakamori S, Kameyama M, Imaoka S, Furukawa H, Ishikawa O, Sasaki Y, Kabuto Kornmann M, Link KH, Lenz H-J, Pillasch J, Metzger R, Butzer U, Leder GH, T, Iwanaga T, Matsushita Y and Irimura T (1993) Increased expression of Weindel M, Safi F, Danenberg KD, Beger HG and Danenberg PV (1997) Sialyl Lewis antigen correlates with poor survival in patients with colorectal Thymidylate synthase is a predictor for response and resistance in hepatic carcinoma: clinicopathological and immunohistochemical study. Cancer Res artery infusion chemotherapy. Cancer Lett 118: 29–35 53: 3632–3637 Koshida Y, Saegusa M and Okayasu I (1997) Apoptosis, cell proliferation and Neoptolemos JP, Oates GD, Newbold KM, Robson AM, McConkey C and Powell J expression of Bcl-2 and Bax in gastric carcinomas: immunohistochemical and (1995) Cyclin/proliferation cell nuclear antigen immunohistochemistry does clinicopathological study. Br J Cancer 75: 367–373 not improve the prognostic power of Dukes’ or Jass’ classifications for Kressner ULF, Lindmark G, Gerdin B, Pahlman L and Glimelius B (1996) colorectal cancer. Br J Surg 82: 184–187 Immunohistochemical p53 staining is of limited value in the staging and Öfner D, Riehemann K, Maier H, Riedmann B, Nehoda H, Tötsch M, Böcker W, prognostic prediction of colorectal cancer. Anticancer Res 16: 951–958 Jasani B and Schmid KW (1995) Immunohistochemically detectable bcl-2 Kumar V, Bustin SA and Mckay IA (1995) Transforming growth factor-alpha. Cell expression in colorectal carcinoma: correlation with tumour stage and patient Biol Int 19: 373–388 survival. Br J Cancer 72: 981–985 Lanza G, Maestri I, Dubini A, Gafa R, Santini A, Ferretti S and Cavazzini L (1996) Ogunbiyi OA, Goodfellow PJ, Gagliardi G, Swanson PE, Birnbaum EH, Fleshman P53 expression in colorectal cancer. Relation to tumor type, DNA ploidy JW, Kodner IJ and Moley JF (1997) Prognostic value of chromosome 1 p pattern, and short-term survival. Am J Clin Pathol 105: 604–612 allelic loss in colon cancer. Gastroenterology 113: 761–766 Lawrence DA (1996) Transforming growth factor-beta – a general review. Eur Ogunbiyi OA, Goodfellow PJ, Herfarth K, Gagliardi G, Swanson PE, Birnbaum EH, Cytokine Network 7: 363–374 Read TE, Fleshman JW, Kodner IJ and Moley JF (1998) Confirmation that British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 203 chromosome 18q allelic loss in colon cancer is a prognostic indicator. J Clin fraction and prognosis in colorectal adenocarcinoma. Eur J Cancer 27: Oncol 16: 427–433 1646–1649 Paradiso A, Rabinovich M, Vallejo C, Machiavelli M, Romero A, Perez J, Lacava J, Sun XF, Carsten JM, Zhang H, Stal O, Wingern S, Hatschek T and Nordenskjold B Cuevas MA, Rodriquez R, Leone B, Sapia MG, Simone G and De Lena M (1992) Prognostic significance of cytoplasmic p53 oncoprotein in colorectal (1996) p53 and PCNA expression in advanced colorectal cancer: response to adenocarcinoma. Lancet 340: 1369–1373 chemotherapy and long-term prognosis. Int J Cancer 69: 437–441 Sun XF, Carstensen JM, Stal O, Zhang H and Nordenskjold B (1995) c-erb B2 Parkin DM, Pisani P and Ferlay J (1993) Estimates of the worldwide incidence of oncoprotein in relation to DNA ploidy and prognosis in colorectal eighteen major cancers in 1985. Int J Cancer 54: 594–606 adenocarcenoma. APMIS 103: 309–315 Poller DN, Baxter KJ and Shepherd NA (1997) p53 and Rb1 protein expression: are Sun XF, Carstensen JM, Stal O, Zhang H and Nordenskjold B (1996) Proliferating they prognostically useful in colorectal cancer? Br J Cancer 75: 87–93 cell nuclear antigen (PCNA) in relation to ras, c-erbB-2, p53, clinico- Pricolo VE, Finkelstein SD, WU T-T, Keller G, Bakker A, Swalsky PA and Bland KI pathological variables and prognosis in colorectal adenocarcinoma. Int J (1996) Prognostic value of TP53 and K-ras-2 mutational analysis in stage III Cancer 69: 5–8 carcinoma of the colon. Am J Surg 171: 41–46 Takahashi Y, Tucher SL, Kitadai Y, Koura AN, Bucana CD, Cleary KR and Ellis LM Reinbach DH, McGregor JR, Murray GD and Odwyer PJ (1994) Effect of the (1997) Vessel counts and expression of vascular endothelial growth factor as surgeons speciality interest on the type of resection performed for colorectal- prognostic factors in node-negative colon cancer. Arch Surg 132: 541–546 cancer. Dis Colon Rectum 37: 1020–1023 Takebayashi Y, Akiyama S-I, Akiba S, Yamada K, Miyadera K, Sumizawa T, Robson H, Anderson E, James RD and Schofield PF (1996) Transforming growth Yamada Y, Murata F and Aikou T (1996) Clinicopathologic and prognostic factor beta 1 expression in human colorectal tumours: an independent significance of an angiogenic factor, thymidine phosphorylase, in human prognostic marker in a subgroup of poor prognosis patients. Br J Cancer 74: colorectal carcinoma. J Natl Cancer Inst 88: 1110–1117 753–758 Wadler S, Bajaj R, Neuberg D, Agarwal V, Haynes H and Benson ALB (1997) Schneider HJ, Sampson SA, Cunningham D, Norman AR, Andreyev HJN, Tilsed Prognostic implications of c-Ki-ras2 mutations in patients with advanced JVT and Clarke PA (1997) Bcl-2 expression and response to chemotherapy in colorectal cancer treated with 5-fluorouracil and interferon: A study of the colorectal adenocarcinomas. Br J Cancer 75: 427–431 Eastern Cooperative Oncology Group (EST 2292). Cancer J Scientific Am 3: Schumacher U, Higgs D, Loizidou M, Pickering R, Leathem A and Taylor I (1994) 284–289 Helix pomatia agglutinin binding is a useful prognostic indicator in colorectal Witzig TE, Lopfinzi CL, Gonchoroff NJ, Reiman HM, Cha SS, Wieand HS, carcinoma. Cancer 74: 3104–3107 Katzmann JA, Paulsen JK and Moertel CG (1991) DNA ploidy and cell kinetic Shibata D, Reale MA, Lavin P, Silverman M, Fearon ER, Steele G Jr, Jessup JM, measurements as predictors of reoccurrence and survival in stages B2 and C Loda M and Summerhayes IC (1996) The DCC protein and prognosis in colorectal adenocarcinoma. Cancer 68: 879–888 colorectal cancer. N Engl J Med 335: 1727–1732 Wolmark N, Wieand HS, Rockette HE, Fisher B, Glass A, Lawrence W, Lerner H, Sinicrope FA, Hart J, Michelassi F and Lee JJ (1995) Prognostic value of bcl-2 Cruz AB, Volk H, Shibata H, Evans J and Prager D (1983) The prognostic oncoprotein expression in stage II colon carcinoma. Clin Cancer Res 1: significance of tumour location and bowel obstruction in Dukes’ B and C 1103–1110 colorectal cancer. Ann Surg 198: 743–750 Smith DR and Goh H-S (1996) Overexpression of the c-myc proto-oncogene in Wynford-Thomas D (1992) p53 in tumour pathology: can we trust colorectal carcinoma is associated with a reduced mortality that is abrogated by immunohistochemistry? J Pathol 166: 329–330 point mutation of the p53 tumor suppressor gene. Clin Cancer Res 2: Yamaguchi A, Nakagawara G, Kurosaka Y, Nishimura G, Yonemura Y and 1049–1053 Miyazaki I (1993) P53 immunoreaction in endoscopic biopsy specimens of Smith DR, Elnatan J, Yao J and Goh H-S (1996) Point mutation of the c-Ki-ras colorectal cancer, and its prognostic significance. Br J Cancer 68: 399–402 proto-oncogene and the p53 tumour suppressor gene in distal colonic Yamaguchi A, Urano T, Goi T, Saito M, Takeuchi K, Hirose K, Nakagawara G, adenocarcinomas. Int J Oncol 8: 1165–1169 Shiku H and Furukawa K (1996) Expression of a CD44 variant containing Soong R, Grieu F, Robbins P, Dix B, Chen D, Parsons R, House A and Lacopetta B exons 8 to 10 is a useful independent factor for the prediction of prognosis in (1997) P53 alterations are associated with improved prognosis in distal colonic colorectal cancer patients. J Clin Oncol 14: 1122–1127 carcinomas. Clin Cancer Res 3: 1405–1411 Younes M, Fernandez L and Lechago J (1996) Transforming growth factor alpha Span M, Moerkerk PT, De Goeij AF and Arends JW (1996) A detailed analysis of (TGF-a) expression in biopsies of colorectal carcinoma is a significant K-ras point mutations in relation to tumor progression and survival in prognostic indicator. Anticancer Res 16: 1999–2004 colorectal cancer patients. Int J Cancer 69: 241–245 Zeng ZS, Sarkis AS, Zhang ZF, Klimstra DS, Charytonowicz E, Guillem JG, Spitz FR, Giacco GG, Hess K, Larry L, Rich TA, Janjan N, Cleary KR and Skibber Cordon-Cardo C and Cohen AM (1994) P53 nuclear overexpression – an JM (1997) P53 immunohistochemical staining predicts residual disease after independent predictor of survival in lymph node positive colorectal cancer chemoradiation in patients with high-risk rectal cancer. Clin Cancer Res 3: patients. J Clin Oncol 12: 2043–2050 1658–1690 Zeng ZS, Huang Y, Cohen AM and Guillem JG (1996) Prediction of colorectal Sun XF, Wingren S, Carstensen JM, Stål O, Hatschek T, Boeryd B, Nordenskjöld B cancer relapse and survival via tissue RNA levels of matrix metalloproteinase- and Zhang H (1991) Ras p21 expression in relation to DNA ploidy, s-phase 9. J Clin Oncol 14: 3133–3140 © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png British Journal of Cancer Springer Journals http://www.deepdyve.com/lp/springer-journals/tumour-markers-of-prognosis-in-colorectal-cancer-h0ZBfK9CQQ

Loading next page...

References (111)

D. Smith, J. Elnatan, J. Yao, H. Goh (1996)
Point mutation of the c-Ki-ras proto-oncogene and the p53 tumour suppressor gene in distal colonic adenocarcinomas.
International journal of oncology, 8 6
T. Witzig, C. Loprinzi, N. Gonchoroff, H. Reiman, S. Cha, H. Wieand, J. Katzmann, J. Paulsen, C. Moertel (1991)
DNA ploidy and cell kinetic measurements as predictors of recurrence and survival in stages B2 and C colorectal adenocarcinoma
Cancer, 68
D. Hoffmann, James Moore, David Roder (1997)
Trends in survival from colonic cancer: the impact of subspecialization.
The Australian and New Zealand journal of surgery, 67 12
M. Nagai, A. Habr-Gama, C. Oshima, M. Brentani (1992)
Association of genetic alterations of c-myc, c-fos, and c-Ha-ras proto-oncogenes in colorectal tumors
Diseases of the Colon & Rectum, 35
Y. Takebayashi, S. Akiyama, S. Akiba, Kazutaka Yamada, K. Miyadera, T. Sumizawa, Yuji Yamada, F. Murata, T. Aikou (1996)
Clinicopathologic and prognostic significance of an angiogenic factor, thymidine phosphorylase, in human colorectal carcinoma.
Journal of the National Cancer Institute, 88 16
S. Dorudi, A. Hanby, Richard Poulsom, J. Northover, Ian Hart (1995)
Level of expression of E-cadherin mRNA in colorectal cancer correlates with clinical outcome.
British Journal of Cancer, 71
G. Lanza, I. Maestri, A. Dubini, R. Gafà, A. Santini, S. Ferretti, L. Cavazzini (1996)
p53 expression in colorectal cancer: relation to tumor type, DNA ploidy pattern and short-term survival.
American journal of clinical pathology, 105 5
S. Bosari, L. Moneghini, D. Graziani, Arthur Lee, J. Murray, G. Coggi, G. Viale (1995)
bcl-2 oncoprotein in colorectal hyperplastic polyps, adenomas, and adenocarcinomas.
Human pathology, 26 5
P. Hermanek, H. Wiebelt, D. Staimmer, S. Riedl (1995)
Prognostic factors of rectum carcinoma--experience of the German Multicentre Study SGCRC. German Study Group Colo-Rectal Carcinoma.
Tumori, 81 3 Suppl
S. Kapitanović, S. Radošević, M. Kapitanović, S. Andelinović, Ž. Ferenčić, M. Tavassoli, D. Primorac, Z. Sonicki, Š. Spaventi, K. Pavelić, R. Spaventi (1997)
The expression of p185(HER-2/neu) correlates with the stage of disease and survival in colorectal cancer.
Gastroenterology, 112 4
C. McArdle, G. Murray, D. Hole (1991)
Impact of variability among surgeons on postoperative morbidity and mortality and ultimate survival.
British Medical Journal, 302
(1997)
A prospective study of prognostic value of type IV collagenase in colorectal cancer tissue
S. Ganesh, Cornelis Sier, M. Heerding, J. Krieken, G. Griffioen, K. Welvaart, C. Velde, J. Verheijen, C. Lamers, H. Verspaget (1997)
Contribution of plasminogen activators and their inhibitors to the survival prognosis of patients with Dukes' stage B and C colorectal cancer.
British Journal of Cancer, 75
J. Bennett, S. Shousha, B. Puddle, N. Athanasou (1995)
Immunohistochemical identification of tumours of adipocytic differentiation using an antibody to aP2 protein.
Journal of Clinical Pathology, 48
Xiao-Feng Sun, J. Carstensen, O. Stål, Hong Zhang, B. Nordenskjöld (1996)
Proliferating cell nuclear antigen (PCNA) in relation to ras, c‐erbB‐2, p53, clinico‐pathological variables and prognosis in colorectal adenocarcinoma
International Journal of Cancer, 69
M. Bennett, W. Kay, H. Mulcahy, L. O’Flaherty, P. O'Donoghue, M. Leader, D. Croke, Dublin Ireland, E. Kay, D. O'Donoghue (1995)
ras and p53 in the prediction of survival in Dukes' stage B colorectal carcinoma
Clinical Molecular Pathology, 48
A. Paradiso, M. Rabinovich, C. Vallejo, M. Machiavelli, A. Romero, J. Pérez, J. Lacava, M. Cuevas, R. Rodriquez, B. Leone, M. Sapia, G. Simone, M. Lena (1996)
p53 and PCNA expression in advanced colorectal cancer: Response to chemotherapy and long‐term prognosis
International Journal of Cancer, 69
H. Schneider, S. Sampson, D. Cunningham, AR Norman, H. Andreyev, J. Tilsed, P. Clarke (1997)
Bcl-2 expression and response to chemotherapy in colorectal adenocarcinomas.
British Journal of Cancer, 75
S. Andersen, T. Løvig, J. Breivik, E. Lund, G. Gaudernack, G. Meling, T. Rognum (1997)
K-ras mutations and prognosis in large-bowel carcinomas.
Scandinavian journal of gastroenterology, 32 1
P. Johnston, J. Drake, J. Trepel, C. Allegra (1992)
Immunological quantitation of thymidylate synthase using the monoclonal antibody TS 106 in 5-fluorouracil-sensitive and -resistant human cancer cell lines.
Cancer research, 52 16
M. Kornmann, K. Link, H. Lenz, J. Pillasch, R. Metzger, U. Butzer, G. Leder, M. Weindel, F. Safi, K. Danenberg, H. Beger, P. Danenberg (1997)
Thymidylate synthase is a predictor for response and resistance in hepatic artery infusion chemotherapy.
Cancer letters, 118 1
Z. Zeng, Á. Sarkis, Zuo‐Feng Zhang, D. Klimstra, E. Charytonowicz, J. Guillem, C. Cordon-Cardo, A. Cohen (1994)
p53 nuclear overexpression: an independent predictor of survival in lymph node--positive colorectal cancer patients.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 12 10
L. Augenlicht, S. Wadler, G. Corner, C. Richards, L. Ryan, A. Multani, S. Pathak, A. Benson, D. Haller, B. Heerdt (1997)
Low-level c-myc amplification in human colonic carcinoma cell lines and tumors: a frequent, p53-independent mutation associated with improved outcome in a randomized multi-institutional trial.
Cancer research, 57 9
R. Soong, F. Grieu, P. Robbins, B. Dix, D. Chen, R. Parsons, A. House, B. Iacopetta (1997)
p53 alterations are associated with improved prognosis in distal colonic carcinomas.
Clinical cancer research : an official journal of the American Association for Cancer Research, 3 8
H. Goh, J. Yao, Duncan Smith (1995)
p53 point mutation and survival in colorectal cancer patients.
Cancer research, 55 22
天谷博一 (1997)
Association of vascular endothelial growth factor expression with tumor angiogenesis,survival,and thymidine phosphorylase/platelet-derived endothelial cell growth factor expression in human colorectal cancer(大腸癌における血管内皮増殖因子発現と血管新生,患者生存,及びチミジンホスホリラーゼ/血小板由来血管内皮増殖因子発現との関連)
S. Wadler, R. Bajaj, D. Neuberg, V. Agarwal, H. Haynes, A. Benson (1997)
Prognostic implications of c-Ki-ras2 mutations in patients with advanced colorectal cancer treated with 5-fluorouracil and interferon: a study of the eastern cooperative oncology group (EST 2292)
The cancer journal from Scientific American, 3 5
S. Kang, K. Maeda, Y. Chung, N. Onoda, Y. Ogawa, S. Takatsuka, M. Ogawa, T. Sawada, B. Nakata, Y. Nishiguchi, T. Ikehara, M. Okuno, M. Sowa (1997)
Vascular endothelial growth factor expression correlates with hematogenous metastasis and prognosis in colorectal carcinoma.
Oncology reports, 4 2
S Bosari, G Viale (1995)
The clinical significance of p53 aberrations in human tumors
Virchows Arch A Pathol Anat Histopathol, 427
E. Kay, H. Mulcahy, C. Walsh, M. Leader, D. O'Donoghue (1994)
Cytoplasmic c‐erbB‐2 protein expression correlates with survival in Dukes' B colorectal carcinoma
Histopathology, 25
J. Elnatan, H. Goh, Duncan Smith (1996)
C-KI-RAS activation and the biological behaviour of proximal and distal colonic adenocarcinomas.
European journal of cancer, 32A 3
O. Ogunbiyi, P. Goodfellow, K. Herfarth, G. Gagliardi, P. Swanson, E. Birnbaum, T. Read, J. Fleshman, I. Kodner, J. Moley (1998)
Confirmation that chromosome 18q allelic loss in colon cancer is a prognostic indicator.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 16 2
V. Pricolo, S. Finkelstein, T. Wu, Gerald Keller, A. Bakker, P. Swalsky, K. Bland (1995)
Prognostic value of TP53 and K-ras-2 mutational analysis in stage III carcinoma of the colon.
American journal of surgery, 171 1
Xiao-Feng Sun, S. Wingren, J. Carstensen, O. Stål, T. Hatschek, B. Boeryd, B. Nordenskjöld, Hong Zhang (1991)
ras p21 expression in relation to DNA ploidy, S-phase fraction and prognosis in colorectal adenocarcinoma.
European journal of cancer, 27 12
N. Wolmark, H. Wieand, H. Rockette, B. Fisher, A. Glass, W. Lawrence, H. Lerner, A. Cruz, Herbert Volk, H. Shibata, James Evans, David Prager (1983)
The prognostic significance of tumor location and bowel obstruction in Dukes B and C colorectal cancer. Findings from the NSABP clinical trials.
Annals of surgery, 198 6
F. Stals, Prof. Steinhoff, S. Wagenaar, J. Vriesman, A. Haverich, P. Dormans, F. Moeller, C. Bruggeman (1996)
Cytomegalovirus induces interstitial lung disease in allogeneic bone marrow transplant recipient rats independent of acute graft-versus-host response.
Laboratory investigation; a journal of technical methods and pathology, 74 2
J. Mulder, M. Sewnath, G. Offerhaus, S. Pals, J. Oosting, P. Kruyt, Wf Weidema, C. Seldenrijk (1994)
Colorectal cancer prognosis and expression of exon-v6-containing CD44 proteins
The Lancet, 344
Z. Zeng, Ying Huang, A. Cohen, J. Guillem (1996)
Prediction of colorectal cancer relapse and survival via tissue RNA levels of matrix metalloproteinase-9.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 14 12
P. Johnston, H. Lenz, C. Leichman, K. Danenberg, C. Allegra, P. Danenberg, L. Leichman (1995)
Thymidylate synthase gene and protein expression correlate and are associated with response to 5-fluorouracil in human colorectal and gastric tumors.
Cancer research, 55 7
J. Benhattar, L. Losi, P. Chaubert, J. Givel, José Costa (1993)
Prognostic significance of K-ras mutations in colorectal carcinoma.
Gastroenterology, 104 4
K. Khine, H. Goh, D. Smith (1995)
Prognostic-significance of chromosome 5q and 17p allelic deletion in colorectal adenocarcinomas.
International journal of oncology, 7 3
R. Apolinario, P. Valk, J. Jong, W. Devillé, J. Ark‐Otte, A. Dingemans, J. Mourik, P. Postmus, H. Pinedo, G. Giaccone (1997)
Prognostic value of the expression of p53, bcl-2, and bax oncoproteins, and neovascularization in patients with radically resected non-small-cell lung cancer.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 15 6
B. Dix, P. Robbins, R. Soong, Deborah Jenner, A. House, B. Iacopetta (1994)
The common molecular genetic alterations in Dukes' B and C colorectal carcinomas are not short‐term prognostic indicators of survival
International Journal of Cancer, 59
G. Ishimaru, K. Ookawa, N. Yamaguchi, M. Sakamoto, S. Hirohashi, T. Muto, J. Yokota (1994)
Allelic losses associated with the metastatic potential of colorectal-carcinoma.
International journal of oncology, 5 2
X-F Sun, J. Carstensen, O. Stål, S. Wingren, T. Hatschek, B. Nordenskjöld, H. Zhang (1992)
Prognostic significance of cytoplasmic p53 oncoprotein in colorectal adenocarcinoma
The Lancet, 340
P. Johnston, Edwin Fisher, H. Rockette, B. Fisher, N. Wolmark, J. Drake, B. Chabner, C. Allegra (1994)
The role of thymidylate synthase expression in prognosis and outcome of adjuvant chemotherapy in patients with rectal cancer.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 12 12
D. Parkin, P. Pisani, J. Ferlay (1993)
Estimates of the worldwide incidence of eighteen major cancers in 1985
International Journal of Cancer, 54
U. Kressner, Gudrun Lindmark, Bengt Gerdin, Lars Påhlman, Bengt Glimelius (1996)
Immunohistological p53 staining is of limited value in the staging and prognostic prediction of colorectal cancer.
Anticancer research, 16 2
M. Younes, L. Fernandez, J. Lechago (1996)
Transforming growth factor alpha (TGF-alpha) expression in biopsies of colorectal carcinoma is a significant prognostic indicator.
Anticancer research, 16 4A
Y. Koshida, M. Saegusa, I. Okayasu (1997)
Apoptosis, cell proliferation and expression of Bcl-2 and Bax in gastric carcinomas: immunohistochemical and clinicopathological study.
British Journal of Cancer, 75
M. Span, P. Moerkerk, Anton Goeli, J. Arends (1996)
A detailed analysis of K‐ras point mutations in relation to tumor progression and survival in colorectal cancer patients
International Journal of Cancer, 69
C. Moertel, T. Fleming, John Macdonald, D. Haller, J. Laurie, C. Tangen, J. Ungerleider, W. Emerson, D. Tormey, J. Glick, M. Veeder, J. Mailliard (1995)
Fluorouracil plus Levamisole as Effective Adjuvant Therapy after Resection of Stage III Colon Carcinoma: A Final Report
Annals of Internal Medicine, 122
Xiao-Feng Sun, J. Carstensen, O. Stål, Hong Zhang, B. Nordenskjöld (1995)
c‐erbB‐2 oncoprotein in relation to DNA ploidy and prognosis in colorectal adenocarcinoma
APMIS, 103
M. Brett, M. Pickard, B. Green, A. Howel-Evans, D. Smith, A. Kinsella, G. Poston (1996)
p53 protein overexpression and response to biomodulated 5-fluorouracil chemotherapy in patients with advanced colorectal cancer.
European journal of surgical oncology : the journal of the European Society of Surgical Oncology and the British Association of Surgical Oncology, 22 2
Gustavo Baretton, M. Vogt, Christian Müller, J. Diebold, K. Schneiderbanger, Michael Schmidt, U. Löhrs (1996)
Prognostic significance of p53 expression, chromosome 17 copy number, and DNA ploidy in non-metastasized colorectal carcinomas (stages IB and II).
Scandinavian journal of gastroenterology, 31 5
C. Belluco, J. Guillem, N. Kemeny, Ying Huang, D. Klimstra, M. Berger, A. Cohen (1996)
p53 nuclear protein overexpression in colorectal cancer: a dominant predictor of survival in patients with advanced hepatic metastases.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 14 10
M. Miyahara, Takao Saito, K. Kaketani, Koichi Sato, A. Kuwahara, K. Shimoda, Michio Kobayashi (1991)
Clinical significance of ras p21 overexpression for patients with an advanced colorectal cancer
Diseases of the Colon & Rectum, 34
J. Jen, Hoguen Kim, S. Piantadosi, Zongang Liu, R. Levitt, P. Sistonen, K. Kinzler, B. Vogelstein, S. Hamilton (1994)
Allelic loss of chromosome 18q and prognosis in colorectal cancer.
The New England journal of medicine, 331 4
A. Chambers, L. Matrisian (1997)
Changing views of the role of matrix metalloproteinases in metastasis.
Journal of the National Cancer Institute, 89 17
A. Yamaguchi, T. Urano, T. Goi, M. Saito, K. Takeuchi, K. Hirose, G. Nakagawara, H. Shiku, K. Furukawa (1996)
Expression of a CD44 variant containing exons 8 to 10 is a useful independent factor for the prediction of prognosis in colorectal cancer patients.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 14 4
S. Nakamori, M. Kameyama, S. Imaoka, H. Furukawa, Osanni Ishikawa, Y. Sasaki, T. Kabuto, T. Iwanaga, Y. Matsushita, T. Irimura (1993)
Increased expression of sialyl Lewisx antigen correlates with poor survival in patients with colorectal carcinoma: clinicopathological and immunohistochemical study.
Cancer research, 53 15
Duncan Smith, H. Goh (1996)
Overexpression of the c-myc proto-oncogene in colorectal carcinoma is associated with a reduced mortality that is abrogated by point mutation of the p53 tumor suppressor gene.
Clinical cancer research : an official journal of the American Association for Cancer Research, 2 6
M. Findlay, D. Cunningham, Gareth Morgan, S. Clinton, A. Hardcastle, G. Aherne (1997)
Lack of correlation between thymidylate synthase levels in primary colorectal tumours and subsequent response to chemotherapy.
British Journal of Cancer, 75
MD Linden, DS Nathanson, G Jacobson (1996)
Lack of prognostic significance of epidermal growth factor receptor expression in Dukes’ B colorectal carcinomas
Lab Invest, 74
G. Murray, M. Duncan, P. O'Neil, W. Melvin, J. Fothergill (1996)
Matrix metalloproteinase–1 is associated with poor prognosis in colorectal cancer
Nature Medicine, 2
D. Reinbach, J. Mcgregor, G. Murray, P. O’Dwyer (1994)
Effect of the surgeon's speciality interest on the type of resection performed for colorectal cancer
Diseases of the Colon & Rectum, 37
A. Mayer, M. Takimoto, E. Fritz, G. Schellander, K. Kofler, H. Ludwig (1993)
The prognostic significance of proliferating cell nuclear antigen, epidermal growth factor receptor, and mdr gene expression in colorectal cancer
Cancer, 71
U. Schumacher, D. Higgs, M. Loizidou, R. Pickering, A. Leathem, I. Taylor (1994)
Helix pomatia agglutinin binding is a useful prognostic indicator in colorectal carcinoma
Cancer, 74
Yutaka Takahashi, S. Tucker, Y. Kitadai, A. Koura, C. Bucana, K. Cleary, L. Ellis (1997)
Vessel counts and expression of vascular endothelial growth factor as prognostic factors in node-negative colon cancer.
Archives of surgery, 132 5
A. Jackman, L. Kelland, R. Kimbell, M. Brown, W. Gibson, G. Aherne, A. Hardcastle, F. Boyle (1995)
Mechanisms of acquired resistance to the quinazoline thymidylate synthase inhibitor ZD1694 (Tomudex) in one mouse and three human cell lines.
British Journal of Cancer, 71
Olagunju Ogunbiyi, Paul Goodfellow, Guiseppe Gagliardi, Paul Swanson, E. Birnbaum, James Fleshman, Ira Kodner, J. Moley (1997)
Prognostic value of chromosome 1p allelic loss in colon cancer.
Gastroenterology, 113 3
L. Cox, D. Lane (1995)
Tumour suppressors, kinases and clamps: How p53 regulates the cell cycle in response to DNA damage
BioEssays, 17
R. Kochhar, Ke Hailing, S. McDonnell, D. Schaid, A. French, M. O’connell, D. Nagorney, S. Thibodeau (1997)
Allelic Imbalance and Microsatellite Instability in Resected Duke's D Colorectal Cancer
Diagnostic Molecular Pathology, 6
S. Bell, S. Bell, N. Scott, N. Scott, D. Cross, D. Cross, P. Sagar, P. Sagar, F. Lewis, F. Lewis, G. Blair, G. Blair, G. Taylor, G. Taylor, M. Dixon, M. Dixon, P. Quirke, P. Quirke (1993)
Prognostic value of p53 overexpression and c-Ki-ras gene mutations in colorectal cancer.
Gastroenterology, 104 1
Y. Kitadai, L. Ellis, S. Tucker, G. Greene, C. Bucana, K. Cleary, Y. Takahashi, E. Tahara, I. Fidler (1996)
Multiparametric in situ mRNA hybridization analysis to predict disease recurrence in patients with colon carcinoma.
The American journal of pathology, 149 5
Akio Yamaguchi, G. Nakagawara, Y. Kurosaka, G. Nishimura, Yutaka Yonemura, I. Miyazaki (1993)
p53 immunoreaction in endoscopic biopsy specimens of colorectal cancer, and its prognostic significance.
British Journal of Cancer, 68
(1999)
British Journal of Cancer Cancer Research Campaign
L. Griffiths, I. Stratford (1997)
Platelet-derived endothelial cell growth factor thymidine phosphorylase in tumour growth and response to therapy.
British Journal of Cancer, 76
I. Baas, J. Mulder, G. Offerhaus, B. Vogelstein, S. Hamilton (1994)
An evaluation of six antibodies for immunohistochemistry of mutant p53 gene product in archival colorectal neoplasms
The Journal of Pathology, 172
E. Fearon, B. Vogelstein (1990)
A genetic model for colorectal tumorigenesis
Cell, 61
M. Loda, Barry Cukor, S. Tam, P. Lavin, Michelangelo Fiorentinc, G. Draetta, J. Jessup, M. Pagano (1997)
Increased proteasome-dependent degradation of the cyclin-dependent kinase inhibitor p27 in aggressive colorectal carcinomas
Nature Medicine, 3
G. Baretton, J. Diebold, Gregor Christoforis, Michael Vogt, Christian Müller, Karin Dopfer, K. Schneiderbanger, Michael Schmidt, U. Löhrs (1996)
Apoptosis and immunohistochemical bcl‐2 expression in colorectal adenomas and carcinomas: Aspects of carcinogenesis and prognostic significance
Cancer, 77
A. Cohen, S. Tremiterra, Frank Candela, H. Thaler, E. Sigurdson (1991)
Prognosis of node‐positive colon cancer
Cancer, 67
MA Nagai, A Habr-Gama, BS Oshima, MM Bretani (1992)
Association of genetic alterations of c-myc, c-fos, and c-Ha-ras proto-oncogenes in colorectal tumors. Frequency and clinical significance
Dis Colon Rectum, 35
Aurora Costa, R. Doci, C. Mochen, P. Bignami, A. Faranda, Leandro Gennari, R. Silvestrini (1997)
Cell proliferation-related markers in colorectal liver metastases: correlation with patient prognosis.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 15 5
V. Astler, F. Coller (1954)
THE PROGNOSTIC SIGNIFICANCE OF DIRECT EXTENSION OF CARCINOMA OF THE COLON AND RECTUM
Annals of Surgery, 139
MA Bennett, EW Kay, H Mulcahy, L O’Flaherty, DP O’Donoghue, M Leader, DT Croke (1995)
ras and p53 in the prediction of survival in Dukes’ stage B colorectal carcinoma
J Clin Pathol, 48
P. Hall, D. Lane (1994)
p53 In tumour pathology: Can we trust immunohistochemistry?—revisited!
The Journal of Pathology, 172
J. Neoptolemos, G. Oates, K. Newbold, M. Robson, C. McConkey, J. Powell (1995)
Cyclin/proliferation cell nuclear antigen immunohistochemistry does not improve the prognostic power of Dukes' or Jass' classifications for colorectal cancer
British Journal of Surgery, 82
V Kumar, SA Bustin, IA Mckay (1995)
Transforming growth factor-alpha
Cell Biol Int, 19
D. Poller, K. Baxter, N. Shepherd (1997)
p53 and Rb1 protein expression: are they prognostically useful in colorectal cancer?
British Journal of Cancer, 75
R. Arribas, G. Capellá, S. Tórtola, L. Masramon, W. Grizzle, M. Perucho, M. Peinado (1997)
Assessment of genomic damage in colorectal cancer by DNA fingerprinting: prognostic applications.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 15 10
D. Wynford‐Thomas (1992)
P53 in tumour pathology: Can we trust immunocytochemistry?
The Journal of Pathology, 166
J. Bhatavdekar, D. Patel, N. Ghosh, P. Chikhlikar, T. Trivedi, T. Suthar, Sonali Doctor, N. Shah, D. Balar (1997)
Coexpression of Bcl-2, c-Myc, and p53 oncoproteins as prognostic discriminants in patients with colorectal carcinoma
Diseases of the Colon & Rectum, 40
Markowitz, Hines, Lutterbaugh, Myeroff, Mackay, Gordon, Rustum, Luna, Kleinerman (1995)
Mutant K-ras oncogenes in colon cancers Do not predict Patient's chemotherapy response or survival
Clinical cancer research : an official journal of the American Association for Cancer Research, 1 4
D. Shibata, M. Reale, P. Lavin, M. Silverman, E. Fearon, G. Steele, J. Jessup, M. Loda, I. Summerhayes (1996)
The DCC protein and prognosis in colorectal cancer.
The New England journal of medicine, 335 23
Qian Dp, Zhou Ar (1988)
[Transforming growth factor alpha].
Sheng li ke xue jin zhan [Progress in physiology], 19 4
Z. Bartik, B. Nordenskjold, X. Sun (1997)
p53 overexpression as a prognostic factor in patients with Dukes B colorectal adenocarcinoma.
International journal of oncology, 11 5
Silvano Bosari, G. Viale, Paola Bossi, Marco Maggioni, Guido Coggi, John Murray, Arthur Lee (1994)
Cytoplasmic accumulation of p53 protein: an independent prognostic indicator in colorectal adenocarcinomas.
Journal of the National Cancer Institute, 86 9
U. Manne, R. Myers, C. Moron, Robert Poczatek, Stephenie Dillard, H. Weiss, David Brown, S. Srivastava, W. Grizzle (1997)
Prognostic significance of Bcl‐2 expression and p53 nuclear accumulation in colorectal adenocarcinoma
International Journal of Cancer, 74
D. Öfner, K. Riehemann, H. Maier, B. Riedmann, H. Nehoda, M. Tötsch, W. Böcker, B. Jasani, K. Schmid (1995)
Immunohistochemically detectable bcl-2 expression in colorectal carcinoma: correlation with tumour stage and patient survival.
British Journal of Cancer, 72
H. Robson, E. Anderson, R. James, P. Schofield (1996)
Transforming growth factor beta 1 expression in human colorectal tumours: an independent prognostic marker in a subgroup of poor prognosis patients.
British Journal of Cancer, 74
Lawrence Da (1996)
Transforming growth factor-beta: a general review.
European cytokine network, 7 3
Dermot Leahy, R. Salman, H. Mulcahy, K. Sheahan, D. O'Donoghue, N. Parfrey (1996)
PROGNOSTIC SIGNIFICANCE OF p53 ABNORMALITIES IN COLORECTAL CARCINOMA DETECTED BY PCR–SSCP AND IMMUNOHISTOCHEMICAL ANALYSIS
The Journal of Pathology, 180
N. Wolmark, H. Wieand, H. Rockette, B. Fisher, A. Glass, W. Lawrence, H. Lerner, A. Cruz, H. Volk, H. Shibata, James Evans, D. Prager (1983)
The Prognostic Significance of Tumor Location and Bowel Obstruction in Dukes B and C Colorectal Cancer
Annals of Surgery, 198
C. Leichman, H. Lenz, L. Leichman, K. Danenberg, J. Baranda, S. Groshen, W. Boswell, R. Metzger, M. Tan, P. Danenberg (1997)
Quantitation of intratumoral thymidylate synthase expression predicts for disseminated colorectal cancer response and resistance to protracted-infusion fluorouracil and weekly leucovorin.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 15 10
F. Miller, T. Heimann, A. Quish, D. Pyo, A. Szporn, G. Martinelli, T. Fasy (1992)
ras and c-myc protein expression in colorectal carcinoma
Diseases of the Colon & Rectum, 35
F. Sinicrope, J. Hart, F. Michelassi, J. Lee (1995)
Prognostic value of bcl-2 oncoprotein expression in stage II colon carcinoma.
Clinical cancer research : an official journal of the American Association for Cancer Research, 1 10
S. Lowe, H. Ruley, T. Jacks, D. Housman (1993)
p53-dependent apoptosis modulates the cytotoxicity of anticancer agents
Cell, 74
H. Mulcahy, D. O'Donoghue, M. Duffy, D. Gibbons, P. McCarthy, N. Parfrey, K. Sheahan (1994)
Urokinase-type plasminogen activator and outcome in Dukes' B colorectal cancer
The Lancet, 344
F. Spitz, G. Giacco, K. Hess, L. Larry, T. Rich, N. Janjan, K. Cleary, J. Skibber (1997)
p53 immunohistochemical staining predicts residual disease after chemoradiation in patients with high-risk rectal cancer.
Clinical cancer research : an official journal of the American Association for Cancer Research, 3 10

Publisher: Springer Journals
Copyright: Copyright © 1999 by The Author(s)
Subject: Biomedicine; Biomedicine, general; Cancer Research; Epidemiology; Molecular Medicine; Oncology; Drug Resistance
ISSN: 0007-0920
eISSN: 1532-1827
DOI: 10.1038/sj.bjc.6690033
Publisher site: See Article on Publisher Site

Abstract

British Journal of Cancer (1999) 79(2), 191–203 © 1999 Cancer Research Campaign Review 1 2 HL McLeod and GI Murray 1 2 Departments of Medicine & Therapeutics, and Pathology, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, UK Keywords: colorectal cancer; prognosis; immunohistochemistry; tumour markers Colorectal cancer is a common disease in developed nations. In systemic chemotherapy. These data highlight the need for informa- Europe, there are approximately 210 000 new cases per year and tive prognostic markers which will aid identification of patients around 124 500 people die from this disease each year (Parkin et best treated with surgery alone, those requiring chemotherapy and al, 1993). Thus, it is the second most common malignancy in men those who might benefit from more aggressive or experimental and third most common in women. About one-third of patients therapy. have a primary tumour in the rectum with the majority of the remaining tumours occurring in the sigmoid colon. Patients who DEFINITION OF PROGNOSTIC MARKER die from the disease most commonly succumb to the effects of secondary metastasis to the liver. The prognosis for this cancer is A practical definition of a prognostic marker is a variable that influenced by a variety of features present at the time of initial provides prospective information on patient outcome which is diagnosis. These factors include age, gender, duration of symp- complementary to the data obtained by the pathologist from the toms, presence of bowel obstruction, tumour location, need for diagnostic sections and on which therapeutic decisions can be blood transfusion and the quality of surgical intervention guided. As Dukes’ staging is widespread and common practice, (Wolmark et al, 1983; McArdle and Hole 1991; Hermanek et al, there would be no logic in finding a marker that simply gave the 1995). A number of tumour characteristics, such as vascular and same information (Astler and Coller, 1954; Cohen et al, 1991). lymphatic invasion, ploidy, differentiation and preoperative levels Rather, proteins (or other variables) which demonstrate a strong of plasma carcinoembryonic antigen, have also been shown to be correlation with Dukes’ stage (or other staging systems) should be of prognostic significance (Witzig et al, 1991; Hermanek et al, considered as possible novel targets for anti-cancer therapies. 1995). However, the simplest and most widely applied prognostic Dukes’ stage does provide a useful ‘internal control’, as it is the feature is the presence of lymph node metastasis in a surgical gold standard on which prognostic markers should be compared, resection specimen. This forms the basis of all staging systems for and any study including Dukes’ A to D disease that does not this disease and has repeatedly been shown to have more prog- demonstrate a relationship between stage and survival should be nostic power than any other single feature at presentation (Astler viewed with great suspicion. and Coller, 1954; Cohen et al, 1991). A large number of prognostic markers for colorectal cancer are Treatment of colorectal cancer is unsuccessful if the tumours featured in the medical literature. Almost all studies of prognostic have an aggressive phenotype at the time of diagnosis. To date, the markers in colorectal cancer involve the correlation of variables most important variable influencing the long-term outcome from observed at diagnosis with eventual outcome. Most published colorectal cancer is the stage of disease at diagnosis, with depth of studies have used a single marker, while a minority of studies have tumour invasion and infiltration of lymph nodes or distant organs used a small group of prognostic markers from which clinical being used to define patient prognosis (Table 1). This classifica- correlations were derived. Although there are many claims of tion discriminates patients with early-stage disease from those prognostic significance in colorectal cancer the question remains: with very advanced disease, but is less able to predict the prog- which markers are important enough that they should be applied nosis of patients with intermediate levels of tumour invasion. For prospectively in routine pathological evaluation of colorectal example, in the Intergroup study of 929 patients with Dukes’ C cancer? colorectal cancer, 45% of patients treated with surgery alone were alive and disease free after 5 years compared with 65% of patients INCLUSION OF TREATMENT DATA treated with surgery followed by 5-fluorouracil(5-FU)/levamisole (Moertel et al, 1995). This suggests that, in retrospect, nearly half One difficulty with many prognostic marker studies is their applic- of the patients with Dukes’ C colorectal cancer did not require ability to modern practice. A study demonstrating that overexpres- adjuvant chemotherapy and 35% will die even when treated with sion of marker ‘X’ gives a survival advantage in patients with Dukes’ stage C disease may have little relevance if the data are derived from a population treated prior to the routine use of adju- Received 20 March 1998 vant chemotherapy. This puts constraints on the length of clinical Revised 1 June 1998 follow-up information available, as chemotherapy only became Accepted 4 June 1998 standard practice within the last 5–7 years in most centres. The details of patient therapy are often inadequate in published studies Correspondence to: HL McLeod 191 192 HL McLeod and GI Murray Table 1 Comparison of the two most commonly used pathological staging the protein structure of target molecules, affecting the use of many systems for colorectal cancer antibodies. These factors can also alter the integrity of genomic DNA extracted from the tissue sections and thereby interfere with Dukes’ stage AJCC/UICC microsatellite analysis or other DNA-based techniques. The devel- stage opment of national guidelines for the processing of tumour samples Tumour invading the submucosa or muscularis A I (such as the UKCCCR Handbook for the ClinicoÐPathological propria Assessment and Staging of Colorectal Cancer) should lead to a more Tumour invading through the muscularis propria B II uniform handling of sample material. Evidence of tumour in regional lymph nodes, C III regardless of local tumour invasion Distant metastasis, regardless of local tumour D IV THE ENRICHMENT APPROACH invasion or lymph node status It is not practicable to test all potential markers of prognosis in a Stage D was not included in the original Dukes’ staging system, but has large data set of tumours of a particular Dukes’ stage. One poten- become commonly used to represent distant metastasis. tial strategy for providing initial data for putative prognostic marker studies is the use of an ‘enrichment’ approach. With this approach, a data set of patients could be retrospectively selected of prognostic markers. It is not satisfactory for the degree of hetero- based on their clinical phenotype, such as 5-year survival or geneity of treatment regimens within a study to be greater than the response to chemotherapy. By selecting patients at the extremes of degree of tumour heterogeneity for the marker under evaluation. the desired endpoint (i.e. Dukes’ C patients alive 5 years post adju- vant therapy vs patients who developed metastatic disease while receiving adjuvant therapy) an ‘enrichment’ step should take CHOICE OF STUDY POPULATION place. Variables only important for tumorigenesis should be The choice of study population needs to be carefully considered for equally distributed between the two patient groups, while markers studies of markers of potential prognostic importance. Many of tumour invasion, chemosensitivity or other factors regulating published studies include patients with all Dukes’ stages (A to D), the efficacy of adjuvant therapy should be delineated between the which is useful for initial evaluation of putative markers of patient groups. This approach assumes a minimum degree of prognosis. However, subsequent studies of patients with a specific heterogeneity in therapeutic regimens and can only be performed disease stage are then more appropriate for the assessment of the using retrospective analysis. It also relies on the availability of clinical utility of a marker. For example, a study of 200 patients with the relevant pathological specimens, often limiting laboratory Dukes’ stage C disease will have greater statistical power to yield approaches to those that are applicable to formalin-fixed tissues. definitive results than a study which consists of 20 Dukes’ A, 80 There are, however, statistical concerns about the use of a large Dukes’ B, 80 Dukes’ C and 20 Dukes’ D patients. In addition to number of potential tumour markers in what will often be a small diluting the statistical power of the study, the inclusion of all stages number of clinical specimens. However, the goal is to identify of disease is inappropriate for the question asked, as there is little markers with a clear association with outcome, and clinical signif- potential for the clinical use of markers in stage A disease and icance needs to have a greater role than statistical significance. A markers of tumour response to therapy or quality of life are more protein that is expressed in 30% of survivors and 0% of patients applicable to stage D patients. The time scale in which the samples with disease progression will be taken forward to more definitive are collected also has a significant influence on the applicability of analysis, regardless of the statistics. This approach will not prove the results to modern cancer treatment. As mentioned above, the the clinical utility of a marker, but rather should be a fertile ground treatment of colorectal cancer has changed significantly over the for identifying putative markers for further analysis in large past 10 years, and this changes the utility of prognostic findings. studies of specific tumour stage. Specialization of both surgeons and oncologists has improved In this review, the current data on cellular or molecular markers patient outcome (Hoffman et al, 1997). The technical ability of the with the potential for prognostic predictive ability for patients with surgeon is also a relevant variable which should not be ignored in colorectal cancer are summarized. It would not be possible to studies of prognosis (Reinbach et al, 1994). The processing of explore all of the putative prognostic markers in the literature, tumour specimens can alter the application of many techniques. For therefore those for which literature is available from either large example, the type of formalin and length of time in fixative can alter single studies or analysis from multiple laboratories were selected. Table 2 Comparison and applicability of different methodologies for assessment of tumour markers PCR PCR SSCP LOH RT-PCR Northern blotting ISH IHC Cellular constituent examined DNA DNA DNA mRNA mRNA DNA/mRNA Protein Use in formalin-fixed, wax-embedded tissue Y Y Y N N Y Y a a a a a Microdissection needed Y Y Y Y Y NN Cellular localization evaluable N N N N N Y Y b b Application to routine diagnosis Y NN N N Y Y IHC, immunohistochemistry; ISH, in situ hybridization; LOH, loss of hetorozygosity; PCR, polymerase chain reaction; RT, reverse transcriptase, SSCP, single- strand conformational polymorphism. Depending on the cellularity of colorectal tumours, microdissection may be desirable to ensure an enriched population of tumour cells. Academic centres and larger general hospitals. British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 193 Papers were identified from the Institute for Scientific Information (Pricolo et al, 1996); advanced disease (Wadler et al, 1997); database (through the Bath Information and Data Services) using Markowitz et al, 1995], while the studies demonstrating a positive the search terms colon, rectal or colorectal with prognosis, prog- correlation with prognosis included a mix of Dukes’ A–D nostic or survival. The search was conducted from January 1990 to tumours. It is possible that the presence of ras mutations is associ- January 1998 and was restricted to English-language publications. ated with disease invasiveness, independent of pathological stage, Negative studies may be under-represented in the literature but is not of prognostic significance within a given stage of because of the personal and editorial bias against publishing such disease. This may account, in part, for an association with survival work. The markers can be separated into oncogenes, tumour- in the studies including Dukes’ A–D tumours and the lack of suppressor genes, proteins involved in apoptosis, factors involved prognostic importance in the studies including larger numbers of in angiogenesis/metastasis/invasion and cellular targets for the patients with a specific pathological stage. chemotherapy used for colorectal cancer. An emphasis has been The influence of ras mutations on response to 5-FU-based placed on techniques that use immunohistochemistry, thereby chemotherapy has been evaluated in 108 patients (Markowitz et al, potentially enabling the results to be applied in many different 1995; Wadler et al, 1997). An objective response rate of 39% was settings throughout the world. However, a spectrum of methodolo- achieved with the different regimens, but mutant ras was not asso- gies are available at most medical centres to assess tumour ciated with chemosensitivity. markers (Table 2). Immunohistochemical analysis of ras p21 staining and outcome has been performed in four studies (a total of 399 patients) (Miyahara et al, 1991; Sun et al, 1991; Miller et al, 1992; Bennett ONCOGENES et al, 1995). Different monoclonal antibodies were used in all Ras studies and the method of scoring a tumour p21 as ‘positive’ either was not detailed or differed between the studies. Tumour staining The ras family represent one of the most commonly detected for p21 was found in 54% (216/399) of tumours, but varied oncogenes in human cancer. The ras genes (K-ras, H-ras, N-ras) between studies (31–71% positive). Positive staining for p21 was encode for 21-kDa proteins (p21s), which are localized on the associated with a poor prognosis in three out of four studies inner surface of the plasma membrane. These proteins bind (Miller et al, 1992; Sun et al, 1991; Miyahara et al, 1991) and was guanine nucleotides and have GTPase activity. As with other GTP- shown to be an independent prognostic factor in only one study binding proteins, ras may serve as a transducer molecule for (Sun et al, 1991). Although there are concerns about the method- signals affecting cell proliferation and thus have involvement in ology (non-specific ras antibody, criteria for ‘positive’ expression) the cell cycle. The development of ras mutations is thought to be and statistical analyses used in these studies, there is an overall an early event in the current model of colorectal tumorigenesis indication that tumour staining for p21 is associated with a poor (Fearon and Vogelstein, 1990). The prognostic significance of both prognosis and should be considered for inclusion in more defini- ras mutations and p21 staining has been assessed in a number of tive studies of markers of prognostic significance in colorectal reports, with conflicting results. cancer. The majority of ras mutations are found in codon 12, with addi- tional mutations observed in codons 13 and 61. Nested polymerase Transforming growth factor (TGF) a and b-1 chain reaction (PCR) methods for detection of ras mutations in DNA isolated from formalin-fixed, paraffin-embedded colorectal The transforming growth factor (TGF) superfamily consists of tumour have allowed assessment of prognostic significance using several subgroups, including a and b, of which there are a number retrospective analysis of large numbers of patients. Although there of isoforms (Kumar et al, 1995; Lawrence, 1996). TGF-a produces are data in the literature on ras mutations and survival from over a mitogenic effect through interaction with the epidermal growth 1200 patients, the clinical relevance of these alterations is not yet factor receptor and promotes the growth of colon cancer cells in clear. At least four studies with a total of 498 patients demonstrate vitro in an autocrine manner. The role of TGF-a as a marker of that the presence of a ras mutation at either codon 12 or 13 is survival in colorectal cancer was evaluated in 105 patients (stage associated with a poor prognosis, independent of Dukes’ stage not given) who were treated with surgery alone (Younes et al, (Benhattar et al, 1993; Elnatan et al, 1996; Smith et al, 1996a; 1996). Patients with low TGF-a expression (<25% positive cells) Span et al, 1996). In addition, two studies with a total of 227 had a poorer prognosis than those with >25% positive staining (P = patients found a poorer prognosis in patients with mutations in 0.03). Although this brief report did not evaluate the independence both ras and p53, even when either alteration alone was not an of TGF-a as a prognostic marker, the significant difference in independent predictor of survival (Bell et al, 1993; Smith et al, overall survival makes it a candidate for further evaluation. 1996a). A single study found that ras mutations were associated TGF-b-1 is the most abundant member of the TGF-b family. with tumour progression and poor survival in patients with left- TGF-b-1 inhibits the growth of epithelial cells but stimulates sided tumours, but not in patients with right-sided disease (Elnatan mesenchymal cell proliferation and cell migration. Resistance to et al, 1996). There was no substantial difference in study design or TGF-b-1-induced growth inhibition is observed in colorectal cancer patient selection between the studies listed above and the seven and is associated with tumour progression in laboratory models studies (a total of 610 patients) in which the presence of ras muta- (Robson et al, 1996). TGF-b-1 staining was evaluated in formalin- tions was not associated with prognosis (Dix et al, 1994; Bennett et fixed tissue from 72 patients with colorectal cancer (Dukes’ stage A, al, 1995; Markowitz et al, 1995; Elnatan et al, 1996; Pricolo et al, 5; B, 33; C, 16; D, 18) (Robson et al, 1996). Positive expression was 1996; Andersen et al, 1997; Wadler et al, 1997). Most of the observed in 58% of tumours and was associated with Dukes’ classi- studies in which no significance was identified had restricted fication (Dukes’ D, 72% vs Dukes’ A, 40%; P < 0.05). Poorer analysis to a specific pathological stage of disease [Dukes’ B survival was observed in patients with positive staining for TGF- (Bennett et al, 1995); Dukes’ B/C (Dix et al, 1994); Stage III b-1. This was more apparent for patients receiving a palliative © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 194 HL McLeod and GI Murray resection of their primary tumour than for patients receiving a cura- clear need for a further large study, preferably in patients with tive resection (Robson et al, 1996). Unfortunately, evaluation of the Dukes’ B and C disease, to define the utility of EGFR immuno- independence of TGF-b-1 staining from Dukes’ stage was not histochemistry as a marker of prognosis in colorectal cancer. reported in this study. However, the data suggest a role for TGF-b-1 staining in the context of advanced disease. c-myc Amplification of the cellular oncogene c-myc has been described c-erbB2 in both colorectal cell lines and primary tumours. c-myc gene copy The c-erbB2 proto-oncogene (also called Her-2, neu) is a 185-kDa number was determined using a dot-blot technique in tumour glycoprotein with tyrosine kinase activity. Expression of c-erbB2 DNA from 149 patients with Dukes’ B or C colon cancer treated has demonstrated prognostic significance in breast and gastric on Intergroup study EST2284 (surgery vs surgery + levamisole vs cancer. However, it has not been extensively evaluated in surgery + 5-FU + levamisole) (Augenlicht et al, 1997). Using an colorectal tumours. A study of 151 adenocarcinomas (Dukes’ arbitrary cut-off point of 1.5 to define amplification, 32% of stage A, 34; B, 40; C, 77) identified a significant relationship patients had elevated c-myc. Among patients with no c-myc ampli- between c-erbB2 immunohistochemistry and patient survival fication there was no difference in survival between the three treat- (Kapitanovic et al, 1997). c-erbB2 expression was found in all ment arms. However, c-myc-positive patients treated with surgery cases and was strongly positive in 43% of tumours. An inverse + 5-FU + levamisole (n = 12) had a longer time to disease progres- correlation between c-erbB2 staining and survival was noted, with sion (P < 0.05) and a trend towards longer survival (P = 0.09) than median survival of 120 weeks and 28 weeks for tumours with c-myc-positive patients treated with surgery alone (n = 25) or moderate and strong staining for c-erbB2 respectively. Median surgery + levamisole (n = 11). While the small number of patients survival for patients with weak staining was greater than 275 with c-myc-positive tumours in each treatment group and the weeks. Although c-erbB2 expression was significantly correlated failure to assess the independence of c-myc as a marker of prog- with Dukes’ stage, c-erbB2 was an independent marker of prog- nosis limit the interpretation of this study, it does suggest a role for nosis in this study. c-erbB2 staining was also of prognostic impor- c-myc detection in the selection of patients with Dukes’ stage B or tance in a study of 164 Dukes’ B tumours (Kay et al, 1994). C colon cancer for whom adjuvant chemotherapy is indicated. Cytoplasmic staining was observed in 33.5% of tumours and was a RNA-based analysis of the prognostic impact of c-myc expres- poor prognostic indicator (5-year survival 47% compared with sion has also been performed in 119 patients (Dukes’ stage A, 31; 77% in c-erbB2-negative tumours). c-erbB2 staining was also an B, 27; C, 29; D, 32) and 29 patients (Dukes’ stage A, 2; B, 10; C, independent prognostic marker in 293 tumours (stage not stated) 10; D, 7) (Nagai et al, 1992; Smith and Goh, 1996). In addition, (Sun et al, 1995). However, unlike the previous studies, a analysis of c-myc DNA copy number was performed in the smaller favourable prognosis with overexpression was observed. This study (Nagai et al, 1992). Neither c-myc gene copy number nor finding was primarily confined to aneuploid tumours. This RNA expression was of prognostic significance in the smaller discrepancy in the prognostic significance of c-erbB2 overexpres- study (Nagai et al, 1992). In contrast, c-myc overexpression was sion needs to be more definitively addressed before this marker observed in 60% of cases in the larger study and these patients had can be recommended for use in the management of patients with a better prognosis than patients without overexpression (P = 0.02) colorectal cancer. (Smith and Goh, 1996). However, the authors found that the beneficial effect of c-myc overexpression was restricted to those Epidermal growth factor receptor tumours containing wild-type p53 (Smith and Goh, 1996). This highlights the fact that evaluation of a specific prognostic marker Epidermal growth factor receptor (EGFR) is a 170-kDa transmem- cannot be carried out in isolation but needs to be studied in the brane protein with tyrosine kinase activity. The receptor is able to context of a more complex, multigenic environment. bind epidermal growth factor and to transduce the signal into the Immunohistochemical analysis of c-myc protein in colorectal cytoplasm. Expression of EGFR is of prognostic significance for a tumours has been evaluated in two separate studies [n = 118, number of solid tumours, including breast and ovarian carcinoma. Dukes’ A, 7; B, 35; C, 41; D, 35 (Miller et al, 1992) and n = 48, EGFR was evaluated in 82 patients with colorectal cancer (Dukes’ Dukes’ B, 24; C, 24 (Bhatavdekar et al, 1997)]. Positive staining stage A, 4; B, 41; C, 22; D, 15; Mayer et al, 1993). Patients with for c-myc was found in 71% and 65% of patients, respectively, and Dukes’ C disease received adjuvant 5-FU and levamisole, while was not associated with Dukes’ stage or histological grade in Dukes’ D patients received 5-FU and leucovorin. EGFR-positive either study (Miller et al, 1992; Bhatavdekar et al, 1997). A signif- staining was detected in 80/82 (97.6%) colorectal tumours, with the icantly better overall survival was observed in the 17 patients majority (62.2%) having >50% positive cells. Patients with >50% without c-myc staining, when compared with the 31 with positive positively staining cells had a poorer survival (median 22.5 staining in the smaller study (Bhatavdekar et al, 1997). However, months) than those with a lower degree of expression (>60 months; c-myc was not of prognostic importance in the larger study (Miller P < 0.01) (Mayer et al, 1993). The role of EGFR as an independent et al, 1992). prognostic marker was not clearly defined, but Dukes’ stage and lymph vessel invasion were also shown to be of prognostic impor- TUMOUR-SUPPRESSOR GENES tance. A second study evaluated EGFR in 84 patients with Dukes’ B colorectal carcinomas who had 10 years of follow-up informa- p53 tion (Linden et al, 1996). Although 52% of tumours expressed The p53 tumour-suppressor gene is a commonly altered gene in EGFR, no association was found between EGFR and survival human solid tumours. The gene has been localized to chromosome (Linden et al, 1996). These two studies leave a great deal of uncer- 17p13.1, a region of allelic deletion in many tumours. The remaining tainty regarding the role of EGFR in colorectal cancer. There is a British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 195 Table 3 Molecular and cellular evaluation of p53 as a marker of prognosis in colorectal cancer. All immunohistochemistry (IHC) studies were performed on formalin-fixed sections and the antibody used is listed in parenthesis. Studies evaluated nuclear staining for p53 unless otherwise indicated. The studies presented here were selected from the literature based on large sample number or focus on a specific tumour stage Technique Number of Significance Reference tumours Immunohistochemical analysis IHC (DO-1) Dukes’ A = 42 Overexpression Bartik et al (1997) Dukes’ B = 82 associated with poor Dukes’ C = 84 survival in Dukes’ Dukes’ D = 45 B, but not Dukes’ A, C, D IHC (DO-7) Dukes’ A = 54 Not an independent Kressner et al (1996) Dukes’ B = 133 predictor of survival Dukes’ C = 73 Dukes’ D = 34 IHC (CM-1) Dukes’ A = 43 Cytoplasmic, but not Sun et al (1992) Dukes’ B = 101 nuclear, overexpression = Dukes’ C = 94 poor survival Dukes’ D = 50 Independent IHC on endoscopic biopsies (1801) Dukes’ A = 25 Overexpression = Yamaguchi et al Dukes’ B = 68 poor prognosis (1993) Dukes’ C = 55 Independence not Dukes’ D = 55 assessed IHC (CM1) Stage I = 35 Cytoplasmic, but not Bosari et al (1994) Stage II = 78 nuclear, overexpression = Stage III = 70 poor survival Stage IV = 23 Independent factor IHC (DO-7) Stage I = 9 Not an independent Lanza et al (1996) Stage II = 80 predictor of survival Stage III = 69 Stage IV = 46 IHC (DO-7) and PCR–SSCP exons 5–8 Dukes’ A = 14 Overexpression = Leahy et al (1996) Dukes’ B = 21 poor survival Dukes’ C = 31 Independent factor Mutation only important for Dukes’ A and B (P = 0.01) IHC (DO-1) Dukes’ A = 29 Not an independent Poller et al (1997) Dukes’ B = 88 predictor of survival Dukes’ C = 130 IHC (DO-7) and PCR + SSCP Dukes’ B = 263 Overexpression = Soong et al (1997) exons 5–8 Dukes’ C = 278 good prognosis, most significant for distal tumours Independent factor IHC (DO-1) Stage IB = 53 Overexpression = Baretton et al (1996a) Stage II = 48 poor survival Independent factor PCR-RFLP exon 4 plus IHC Dukes’ B = 168 Not associated with Bennett et al (1995) (CM1) survival IHC (1801) Dukes’ C = 107 Overexpression = Zeng et al (1994) poor survival Cytoplasmic p53 not associated with survival IHC (1801) Dukes’ D = 50, Overexpression = Belluco et al all treated with poor survival (1996) intrahepatic chemotherapy IHC (DO-1) Dukes’ D = 59 Overexpression Brett et al (1996) associated with lower response (P < 0.03), but not survival IHC (1801) Dukes’ D = 59 Overexpression not Costa et al (1997) associated with relapse or survival IHC (1801) Dukes’ D = 71 Not associated with Paradiso et al treated with response to therapy (1996) 5-FU modulated or survival with methotrexate © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 196 HL McLeod and GI Murray Table 3 Cont Technique Number of Significance Reference tumours Molecular analysis RT-PCR-SSCP Dukes’ A–D, Mutation = poor Goh et al (1995) n = 192 survival Independent of stage (P = 0.031) IHC (DO-7) and PCR–SSCP Dukes’ A = 14 Mutation = poor Leahy et al (1996) exons 5–8 Dukes’ B = 21 survival for Dukes’ Dukes’ C = 31 A and B PCR–RFLP exon 4 plus IHC Dukes’ B = 168 Not associated with Bennett et al (1995) (CM1) survival IHC (DO-7) and PCR + SSCP Dukes’ B = 263 Not associated with Soong et al (1997) exons 5–8 Dukes’ C = 278 survival allele often contains an inactivating mutation. Wild-type p53 is presence of several larger studies (250–300) has not clarified the clin- thought to be a stress response gene, the protein product of which ical relevance of p53 immunohistochemistry. acts to maintain genetic integrity by inducing cell cycle arrest and Three out of four studies of p53 and survival in advanced apoptosis following damage to the genome (Cox and Lane, 1995). colorectal cancer found no association (Brett et al, 1996; Paradiso According to this model, inactivation of p53 function by allelic loss, et al, 1996; Costa et al, 1997), with the exception being an evalua- gene mutation or protein sequestration would be expected to lead to tion of the impact in the context of intrahepatic chemotherapy for increased genetic instability and to the survival of cells with damaged liver metastasis (Belluco et al, 1996). A single study found no DNA. Most studies of p53 protein in tumours have exploited the association between either p53 staining or ras mutations and observation that wild-type p53 has a short half-life and is usually not survival, but identified a poor prognosis in patients whose tumour detectable by immunohistochemistry, while missense gene mutations contained both alterations (Bell et al, 1993). Additional studies are believed to extend greatly the half-life of the protein such that it identified additive prognostic significance when p53 and bcl-2 becomes detectable. Although there is a good correlation between the alterations were present in the same patient (Barreton et al, 1996a; presence of p53 gene mutations and positive p53 staining, discrepan- Manne et al, 1997). This suggests the need for several large, cies do exist and can occur for a number of reasons, including lack of prospective studies (>500 patients) with long clinical follow-up protein from truncating mutations, overuse of ‘antigen retrieval’ (>5 years) and a uniform treatment policy to provide definitive methods and the criteria for classifying a tumour as ‘positive’ for p53 data on the interplay between p53 and other putative prognostic (reviewed in Wynford-Thomas, 1992; Baas et al, 1994; Hall and markers to identify patients with a poor prognosis for whom more Lane, 1994). aggressive therapy may be appropriate. A large number of studies on the association between p53 Molecular analysis used either direct DNA or cDNA immunohistochemistry or mutational analysis and survival have sequencing or single-strand conformational polymorphism been published, giving data for over 3000 patients. Immuno- (SSCP) and focused on p53 exons 5–8, where 98% of mutations histochemical studies were performed using one of five different have been described. The concordance between immunohisto- monoclonal antibodies against p53, and the criteria for ‘positive’ chemical and molecular analysis is 60–85%, with both false-posi- staining varied from greater than 1% to more than 10% of tumour tive and false-negative staining observed. nuclei. Several studies used two separate p53 antibodies and Studies of mutations in p53 have also been conflicting, with the inconsistencies between antibodies were noted for the same majority showing no association with patient survival (Goh et al, tumour sections. Additional controversy also exists regarding the 1995; Leahy et al, 1996; Pricolo et al, 1996; Soong et al, 1997) localization of p53. Several studies have reported the presence of (Table 3). The studies showing no association had longer patient cytoplasmic staining for p53, which they report to be of greater follow-up, suggesting that the effort of tumour microdissection and prognostic significance than the more commonly observed nuclear the time and expense of p53 sequencing or SSCP is not a worth- staining (Sun et al, 1992; Bosari et al, 1994). It is unclear whether while exercise for identifying patients at risk of poor survival. the cytoplasmic staining represents a methodological artifact or a The largest study of p53 alterations and patient prognosis (541 finding of biological and/or prognostic importance. patients) used both immunohistochemistry and PCR-SSCP and The wide availability of antibodies for immunohistochemistry of had a mean patient follow-up time of over 7 years (Soong et al, p53 in formalin-fixed tissue has led to its use in the majority of 1997). Although no association with prognosis was identified studies assessing the association between p53 and survival (Bosari from the molecular analysis, an improved survival was found in and Viale, 1995; reviewed in Manne et al, 1997). The role of p53 patients with an overaccumulation of p53. This puzzling finding staining as a prognostic tool in colorectal cancer has been clouded by was strongest for distal colonic tumours and patients with Dukes’ a similar number of studies (and patient numbers) that have reported C disease. However, even in the left-sided tumours, the improve- poor survival in p53-positive cases as demonstrating no association ment in patient survival associated with p53 staining was 10–20%. between p53 overexpression and outcome (Table 3). This conflict in This smaller degree of survival advantage may help to explain prognostic importance exists both in studies focusing on specific why many smaller studies of mixed tumour stages have found no stage of disease and in those including Dukes’ stage A–D. Although statistically significant prognostic importance for p53. Only a the sample size of the majority of studies was 100–200 patients, the small number of patients in the study received post-operative British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 197 chemotherapy, restricting the application of these data to current B, 23; C, 47; D, 15), in which normal and tumour DNA was treatment protocols. analysed by restriction fraction length polymorphism (RFLP) Studies in cell lines with a homozygous mutant p53 genotype (Ishimaru et al, 1994). Loss of heterozygosity (LOH) at DCC gave demonstrated a high degree of resistance to radiation and several a 5.5-fold higher risk of death and was an independent marker of chemotherapy agents, including 5-FU (Lowe et al, 1993). This survival. However, Dukes’ stage was not an independent prog- provided the basis for four studies (a total of 222 patients) of the nostic marker in the multivariate analysis. association between p53 staining and response to either systemic Not all studies of 18q have demonstrated a prognostic influence. or intrahepatic 5-FU-based chemotherapy for advanced colorectal Analysis of normal and tumour DNA from 100 colorectal tumours cancer (Belluco et al, 1996; Brett et al, 1996; Paradiso et al, 1996) (Dukes’ B, 56; Dukes’ C, 44) using RFLP found no relationship or combined chemoradiation neoadjuvant therapy in rectal cancer between LOH at DCC and survival (Dix et al, 1994). The median (Spitz et al, 1997). Immunohistochemistry was performed in clinical follow-up was 18.9 months in the Dix et al (1994) study, formalin-fixed tissues using either monoclonal antibody 1801 compared with 35.4 months in the study by Jen et al (1994), and (Belluco et al, 1996; Paradiso et al, 1996) or DO-1 (Brett et al, may be a reason for the discrepancy in results for this marker in 1996; Spitz et al, 1997) and the criteria for a p53-positive tumour patients with Dukes’ B and C disease. The influence of 18q loss on ranged from greater than 1% to greater than 10% of tumour nuclei. prognosis has also been evaluated in metastatic colorectal cancer Tumour staining for p53 was not associated with response to either (Kochhar et al, 1997). LOH on 18q (microsatellite markers D18S34, biochemically modulated (with methotrexate) or intrahepatic 5-FU D18S49, D18S35, D18S58, D18S61) was assessed in 141 patients therapy (Belluco et al, 1996; Paradiso et al, 1996), but a poorer undergoing potentially curative liver resection for colorectal cancer. response rate to 5-FU and folinic acid was observed in patients The majority of tumours were non-informative for 18q markers but with p53 overexpression (Brett et al, 1996). Positive p53 staining were informative for other loci, limiting the interpretation of this in pretherapy biopsy samples was inversely correlated with a study. Among the informative cases for 18q, no relationship with complete pathological response to chemoradiation therapy and either survival or disease-free survival was observed. directly correlated with an increased likelihood of residual tumour The use of DNA-based assessment of 18q or DCC is cumbersome in the lymph nodes of surgical specimens (Spitz et al, 1997). This and requires facilities for both high-quality microdissection and suggests that p53 staining may be a tool for identifying patients quality-controlled molecular biology, and therefore is unlikely to who will benefit most from neoadjuvant chemoradiation therapy provide an approach that will have broad application to the planning or those for whom more aggressive therapy is required. of colorectal cancer treatment. There has been one published study of the prognostic implications of DCC protein staining (Shibata et al, 1996). Using a polyclonal antibody (antibody 721) and a Deleted in colon cancer/allelic variance at 18q microwave antigen retrieval protocol on formalin-fixed, paraffin- The deleted in colon cancer (DCC) gene, which is found on chro- embedded tissue sections, DCC staining was evaluated in 132 mosome 18q21.2, is a putative tumour-suppressor gene with impli- patients with colorectal cancer (stage II, 70; stage III, 62) and a cations for tumour progression and prognosis in colorectal cancer. median follow-up of 92 months. Immunohistochemistry revealed a An important step in the Vogelstein and Fearon model of colon granular cytoplasmic staining pattern for DCC and was scored as the tumorigenesis is loss of parts of chromosome 18q, including the presence of any immunoreactivity or no evidence of staining. The region containing DCC (Fearon and Vogelstein, 1990). Although expression of DCC was a prognostic factor for survival in both stage there is conflicting information on the role of the DCC protein in II and stage III disease. In patients with stage II disease, the 5-year tumour biology, there is evidence that it is of prognostic impor- survival in patients with DCC staining was 94.3% compared with tance in colorectal cancer. The observed allelic loss of chromo- 61.6% in DCC-negative tumours (P < 0.001). In stage III disease, some 18q has been linked to the prognosis of patients with stage II the 5-year survival was 59.3% with DCC staining compared with colorectal cancer (Jen et al, 1994). Analysis of normal and tumour 33.2% in DCC-negative tumours. Expression of DCC was an inde- DNA from microdissection of formalin-fixed, paraffin-embedded pendent prognostic factor which gave a threefold relative risk of sections was performed using the dinucleotide microsatellite death. Only tumour staging gave a similar level of prognostic infor- markers D18S58 and D18S61. No difference in prognosis was mation. The ease of immunohistochemical assessment of DCC will observed between patients with stage III disease and 18q loss (n = allow additional large prognostic studies to determine the role of 54) when compared with those with no allelic loss (n = 16). DCC expression in the treatment planning for colorectal cancer. However, among patients with stage II disease, the 5-year survival was 93% for those with no 18q loss (n = 29) compared with 54% p27 for those with allelic loss (n = 36, P = 0.006). The survival rate for stage II patients with 18q loss was not statistically different from The expanding understanding of cell cycle regulation has identified that of patients with stage III disease. In multiple regression putative oncogenes and tumour-suppressor genes that may be proportional hazards models, 18q was an independent prognostic involved in colorectal tumorigenesis. However, there have been marker and 18q loss gave a 2.5 to 3.0-fold higher risk of death. The few evaluations of the prognostic implications of alterations in cell prognostic importance of 18q allelic loss was verified in a study of cycle proteins. The cell cycle inhibitor p27 (also known as Kip 1) is 126 patients (stage I, 32; II, 50; III, 44), using four microsatellite a putative tumour-suppressor gene, but inactivating mutations have markers on microdissected tumour (Ogunbiyi et al, 1998). Loss of not been described in human tumours. Immunohistochemical 18q was an independent marker of poor prognosis, with a 1.65 analysis of p27 in formalin-fixed, paraffin-embedded sections relative risk of disease-free survival. Loss of 18q was prognosti- detected variable expression in the cytoplasm, nucleus, or both, in cally important in both stage II and III disease. A poor prognosis in 90% of colorectal tumours, when using a microwave-based method patients with loss of a polymorphic allele in DCC was also (Loda et al, 1997). Scoring of p27 staining was expressed as a ratio observed among 95 patients with colorectal cancer (Dukes’ A, 10; of positive cells to total number counted and all immunoreactive © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 198 HL McLeod and GI Murray cells were considered positive. A significant relationship between provides a growth advantage to tumour cells in vitro by blocking p27 staining and prognosis was observed among 149 patients with apoptosis. However, evaluation of bcl-2 protein expression in colorectal adenocarcinoma (stage I, 31; II, 66; III, 51; IV, 1), when lung, thyroid and breast carcinomas identified a favourable prog- median survival was 151 months in patients with p27-positive nosis in patients with overexpression of bcl-2 on tumour staining. tumours (n = 134) and 69 months in patients with p27-negative The association between bcl-2 immunohistochemistry and patient tumours (n = 15). Both tumour stage and grade showed a signifi- survival has been evaluated in at least six studies of colorectal cant association with survival. The prognostic significance of p27 tumours (a total of 932 patients) (Bosari et al, 1995; Ofner et al, staining was independent of tumour stage and grade (P = 0.003). 1995; Sinicrope et al, 1995; Baretton et al, 1996b; Manne et al, The effect of p27 was observed when analysis was confined to 1997; Schneider et al, 1997). The larger studies (n = 205 and 231) patients with stage II disease, with absence of staining giving a 2.9 combined Dukes’ stages A–D and found no association between relative risk of death from colorectal cancer. These encouraging bcl-2 overexpression and relapse or survival (Bosari et al, 1995; data now needs confirmation in a larger independent data set Schneider et al, 1997). In addition, bcl-2 staining was not associ- containing patients with uniform treatment before widespread use ated with response to 5-FU-based chemotherapy (Schneider et al, of p27 as a prognostic marker can be recommended. 1997). Bcl-2-positive tumours were associated with a better prog- nosis in the remaining studies, although not always as an indepen- Allelic loss/genome damage dent marker of overall survival (Ofner et al, 1995; Sinicrope et al, 1995; Baretton et al, 1996b; Manne et al, 1997). An interaction Although allelic instability is a common event in colorectal cancer, between bcl-2 overexpression and p53 staining was identified in the prognostic impact of instability has not been extensively eval- several studies, with bcl-2-positive/p53-negative tumours being uated. Seventy-five patients (stage not reported) were evaluated associated with a better survival than tumours with other staining for the impact of allelic deletion in chromosomes 5q and 17p on patterns (Baretton et al, 1996b; Manne et al, 1997). This supports survival, using Southern blot analysis (Khine et al, 1995). Loss of the need for studies of the interplay between bcl-2 and other puta- chromosome 5q was observed in 32% of cases and 17p loss was tive prognostic markers to identify patients with a poor prognosis found in 69%, with loss of both regions in 27% of patients and for whom alternative therapy may be appropriate. neither region in 25%. Poorer survival in patients with loss of bcl-2 does not act alone but forms heterodimers with other chromosome 5q was apparent (P = 0.014). Although patients with family members, including BAX, to regulate apoptosis in in vitro 17p allelic deletion had a poorer survival, this did not reach systems. However, there are no data on the association between statistical significance (P = 0.16). Multivariate analysis identified BAX staining and prognosis in colorectal cancer. The recent Dukes’ stage and allelic loss of chromosome 5q as the two inde- development of techniques for the use of BAX antibodies in pendent markers of survival (P = 0.0001). An additional study formalin-fixed, paraffin-embedded tumour tissue should allow this evaluated allelic loss on chromosome 1p in 116 colorectal marker to be assessed in the near future (Apolinario et al, 1997; tumours, of which 82 (stage I, 17; II, 36; III, 29) were informative Koshida et al, 1997). and evaluable for survival analysis (Ogunbiyi et al, 1997). Loss of 1p was detected in 22/82 tumours and was associated with a poorer disease-free survival. Loss of 1p was a weak but statisti- CELL PROLIFERATION cally significant independent marker of survival in multivariate Cell proliferation has been extensively studied using a variety of analysis. Although the methodology used in these studies is too direct and surrogate markers. Currently one of the most popular cumbersome for widespread application, the availability of methods of assessing cell proliferation is using antibodies that microsatellite markers of regions throughout the genome may recognize nuclear proteins involved in, or altered during, the cell facilitate further evaluation of this approach for identifying cycle. One of the most frequently studied targets is proliferating patients at risk for poor long-term survival. cell nuclear antigen (PCNA). This is a DNA polymerase d acces- Genomic DNA fingerprinting has recently been used to assess sory protein which can complex with cyclin D and cyclin-depen- the prognostic implications of tumour genome instability (Arribas dent kinases (i.e. p21). Ki-67 is another frequently used antibody, et al, 1997). By comparing banding patterns from arbitrarily detecting an antigen expressed in all phases of the cell cycle primed PCR analysis of 63 colorectal tumours and adjacent except G . Evaluation of PCNA in 79 colorectal tumours (Dukes’ normal tissue, an index of genomic instability was derived and was A, 4; B, 48; C, 27) by immunohistochemistry was performed with an independent marker of prognosis. This technique does not follow-up for a minimum of 10 years (Neoptolemos et al, 1995). A identify specific target genes for development as markers or PCNA score was derived by a relatively complex, subjective four- cellular targets for chemotherapy, but does provide an apparently point scale according to the percentage of positive tumour cells. reproducible method for detection of genome-wide instability for The PCNA index was not associated with patient survival. Sun et high-throughput prognostic studies. The small number of patients al (1996) studied PCNA in a series of 293 colorectal carcinomas at each disease stage limits the conclusions from the initial study, (Dukes’ A, 43; B, 101; C, 94; D, 50; stage not available for 5) but does provide the basis for further analysis of this methodology which had been collected over a 15-year period and followed up as a potential marker of patient survival. for a minimum of 5 years. PCNA was assessed by counting 500 tumour nuclei and the final PCNA score was divided into four APOPTOSIS PATHWAYS groups according to the percentage of positive nuclei. PCNA did not correlate with survival. This group also studied PCNA expres- bcl-2 sion in 56 primary tumours and their corresponding lymph node The bcl-2 proto-oncogene encodes a 25-kDa protein that is metastases with good correlation between the tissues. PCNA involved in the regulation of cell death by inhibiting apoptosis expression was investigated in 71 advanced colorectal cancers through mechanisms which are not yet defined. Its overexpression (metastatic, n = 33, or recurrent disease, n = 38) eligible to receive British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 199 5-FU-based therapy (Paradiso et al, 1996). PCNA was scored by tumour invasion and metastasis. The current concept of the MMPs analysing 1000 nuclei, and a ‘high’ PCNA score (greater than 32% in tumour invasion is that they not only have a direct role in facili- positive cells) was associated with improved survival and was tating extracellular matrix degradation but as a consequence they independently prognostic. also have an important role maintaining the tumour environment and thus promoting tumour growth (Chambers and Matrisian, 1997). There are currently at least 15 known MMPs, which are ANGIOGENESIS/METASTASIS/INVASION broadly classified into collagenases, gelatinases, stromelysins and membrane-type MMPs. The MMPs (except membrane-type Thymidine phosphorylase MMPs) are produced as proenzymes and are activated by cleavage Thymidine phosphorylase, or platelet-derived endothelial cell of N-terminal propeptides. Our group identified MMP-1 (intersti- growth factor, is an enzyme that catalyses the reversible phos- tial collagenase) in 64 patients with colorectal cancers (Dukes’ A, phorylysis of thymidine and deoxyuridine to their bases and de- 1; B, 38; C, 25). The presence of MMP-1 in tumour cells was asso- oxyribose phosphate. It has been shown to be capable of inducing ciated with poorer prognosis and was independent of Dukes’ stage angiogenesis, and expression of thymidine phosphorylase may (Murray et al, 1996). This study used a simple assessment of the also be important for the activation of certain prodrugs, for presence or absence of MMP-1 immunoreactivity and is likely to example deoxyfluorouridine conversion to fluorouracil (Griffiths be more reproducible than some of the more complex scoring and Stratford, 1997). Thymidine phosphorylase was evaluated in schemes used in other immunohistochemistry studies. MMP-9 93 colorectal cancers (Dukes’ stage A, 30; B, 42; C, 21) by mRNA has been identified by in situ hybridization and Northern immunohistochemistry using a monoclonal antibody (Takebayashi blotting in 71 tumours (Dukes’ A, 11; B, 18; C, 16; D, 26), and a et al, 1996). The immunohistochemical assessment used was a high ratio of tumour to normal MMP-9 mRNA was associated straightforward two-point scale, with tumour being classified as with poor prognosis and was independent of Dukes’ stage (Zeng et positive if more than 5% of tumour cells displayed staining. al, 1996). In a small group (n = 31) of colorectal cancers, total type Normal colon mucosa did not stain, whereas staining was IV collagenase activity (MMP-2 and MMP-9) has been found to observed in both tumour and stromal cells. No comment was made be prognostically significant, with high activity associated with regarding the significance of stromal cell staining in the absence of both shorter disease-free survival and shorter overall survival tumour cell staining or how this was scored. The presence of (Ambiru et al, 1997). These findings of the different MMPs thymidine phosphorylase was associated with a significantly emphasize the need to perform a comprehensive series of studies poorer prognosis. Although expression correlated with other of MMPs in colorectal cancer to understand the interactions histopathological criteria (lymph node metastasis, lymphatic and between the individual MMPs and to identify which of the MMPs venous invasion), thymidine phosphorylase also appeared to be an are most appropriate as prognostic markers. Studies are also independent prognostic factor. required to determine whether it is more relevant to assess total enzyme or active enzyme. Clinical trials of MMP inhibitors are now under way and require evaluation in relation to tumour MMP Vascular endothelial growth factor expression to determine if MMP expression is a reliable method of Vascular endothelial growth factor (VEGF) is a glycoprotein that selecting patients for anti-MMP therapy. The MMPs are inhibited shares some homology with platelet-derived growth factor and is in vivo by naturally occurring inhibitors, tissue inhibitors of intimately involved in promoting angiogenesis. Interest in angio- metalloproteinases (TIMPs). There are currently four known genesis has grown because of the possibility of using anti-angio- TIMPs, and studies are also necessary to determine the contribu- genic therapy to prevent tumour progression. There have been tion of TIMPs to prognosis. several recent reports of the analysis of VEGF in colorectal cancer, all of which used immunohistochemistry on formalin-fixed, wax- Urokinase-type plasminogen activator and inhibitor embedded sections and showed that VEGF is primarily localized to tumour cells. In the largest study of 175 colorectal carcinomas Urokinase-type plasminogen activator (uPA) is a serine protease (stage 0, 7; I, 24; II, 48; III, 59; IV, 37), the presence of VEGF which catalyses the activation of plasminogen to plasmin, a broad- correlated with increasing tumour stage and was associated with spectrum protease which in turn can break down a variety of extra- decreased overall survival (P < 0.05) and decreased disease-free cellular matrix components and also promote the activation of survival (P < 0.01) (Kang et al, 1997). However, it was not an MMP-2. In a series of Dukes’ B colon cancers (n = 70), a higher independent prognostic factor. Amaya et al (1997) studied the proportion of uPA-positive tumour cells by immunohistochemistry presence of VEGF in 133 cases of colorectal cancer (stage I, 14; II, was associated with poorer survival (P = 0.01) (Mucalhy et al, 42; III, 45; IV, 29 with three unclassified cases). VEGF was local- 1994). However, most tumours had a high percentage of positive ized to tumour cells, and VEGF-positive tumours had a signifi- tumour cells (57/70) and only 13 cases had a low percentage of cantly poorer prognosis than VEGF-negative tumours (P < 0.02), positive tumour cells. In addition, stromal staining for uPA showed which remained significant in a multivariate analysis. A much an association with survival. This study is one of relatively few to smaller study of VEGF in 27 Dukes’ stage B patients showed study a marker in relation to a single Dukes’ stage. The patients in essentially similar results (Takahashi et al, 1997). this study all had uniform treatment with no post-operative radio- or chemotherapy and follow-up was for between 5 and 10 years. High tumour levels of plasminogen activator inhibitor-2 (PAI-2) in Matrix metalloproteinases and inhibitors Dukes’ B (n = 82) and C (n = 54) colorectal cancers has also been The matrix metalloproteinases (MMPs) are a group of enzymes found to be prognostically significant (P = 0.05), but not indepen- with a central role in matrix degradation, and there is considerable dent of Dukes’ stage, in a group of 136 colorectal cancers (Ganesh evidence to indicate that this group of enzymes have a key role in et al, 1997). © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 200 HL McLeod and GI Murray CD44 colorectal cancer, 5-FU and raltitrexed (tomudex). Elevated expression of the TS protein has been demonstrated in cell lines CD44 is a cell-surface glycoprotein which functions as an adhesion which were made resistant to either 5-FU or tomudex (Johnston et molecule. Multiple isoforms of CD44 are generated by alternative al, 1992; Jackman et al, 1995). A higher degree of tumour TS splicing of up to ten exons that encode for the molecule’s extracel- mRNA and protein was observed in patients with a poor response lular portion. CD44 variants have been identified on the surface of a to 5-FU in a small study of colorectal (n = 9) and gastric (n = 12) variety of tumour cells. CD44v6 was studied by immunohistochem- cancers (Johnston et al, 1995). TS mRNA expression provided a istry in a series of colon cancers (n = 68) with follow-up of between clearer definition of patient response than did protein level in this 6.5 and 9.5 years (Mulder et al, 1994). The report did not indicate study. Quantitative PCR assay of TS mRNA was then applied to the distribution of tumours in different Dukes’ stages or the post- tumour tissue from 46 patients with advanced colorectal cancer operative treatments received by the patients. Tumours that strongly –2 receiving protracted infusion of 5-FU (200 mg m per day for 21 expressed CD44v6 had a poorer prognosis than those tumours that days) and weekly leucovorin (Leichman et al, 1997). No patient showed less intense or no immunostaining. This was stated to be with a ratio of TS to b-actin (used as an intertumour control) independent of Dukes’ stage. CD44 exons 8–10 was studied using –3 greater than 4.1 ´ 10 responded. Using the median TS/b-actin immunohistochemistry in 215 colorectal tumours (stage I, 37; II, 71; ratio as a cut-off point, survival was significantly higher (median III, 85; IV, 22) (Yamaguchi et al, 1996). There was no significant 13.6 months) in patients with less than median TS levels than in immunoreactivity for CD44 in normal mucosa while there was those with greater than median values (8.2 months) (Leichman et positive immunoreactivity in tumour cells. Immunoreactivity was al, 1997). Similar results have recently been reported in patients assessed as either present or absent, although the cut-off point for receiving hepatic artery infusion of fluoropyrimidines, when negative score was less than 25% of tumour cells. The presence of patients with low TS mRNA expression were 4.1-fold more likely CD44 was associated with poorer prognosis, and in multivariate to respond to chemotherapy (Kornmann et al, 1997). TS anti- analysis this factor remained independently prognostic. bodies, which work on formalin-fixed, paraffin-embedded tumour tissue, have provided the opportunity for retrospective analysis of E-cadherin the prognostic significance of TS staining in patients with E-cadherin is a member of the cadherin family of calcium-depen- colorectal cancer. TS was assessed with a semiquantitative visual dent adhesion molecules that function as mediators of cell-to-cell grading system based on the intensity of staining (0 to +++) and adhesion. It is a transmembrane protein with a large extracellular extent of staining (focal, <25% positive cells; diffuse, >25% posi- domain and a smaller cytoplasmic component which interacts with tive cells) (Johnston et al, 1994). An initial study in 294 patients cytoplasmic proteins, particularly b-catenin. Loss of E-cadherin is (Dukes’ stage A, 32; B, 92; C, 146; D, 24) from NSABP study R- associated with increased cell invasiveness. E-cadherin was studied 01 demonstrated that high expression of TS protein in tumour is by in situ hybridization in 49 Dukes’ B colorectal cancers (Dorudi et associated with a poor survival rate in patients receiving 5-FU- al, 1995). In situ hybridization was performed because antibodies based adjuvant chemotherapy for rectal cancer (Johnston et al, which reliably detect E-cadherin in formalin-fixed, wax-embedded 1994). TS staining intensity was an independent prognostic sections do not appear to be currently available. The lowest expres- marker and identified a significant impact of 5-FU-based sion of E-cadherin was associated with poor survival. In a second chemotherapy in patients with high TS levels. However, TS study, E-cadherin expression was investigated by in situ hybridiza- staining did not predict disease response among 134 patients tion in 32 colorectal cancers (Dukes’ A, B and D). None of the receiving 5-FU-based chemotherapy for advanced colorectal patients had received chemotherapy and all had been followed up cancer (Findlay et al, 1997). A significant time interval between for at least 5 years (Kitadai et al, 1996). The profile of expression of surgical resection of the primary tumour and treatment of E-cadherin was also compared with several other genes involved in advanced disease was observed for most patients. As this study tumour invasion, including type IV collagenase. There was no prog- performed TS staining on the primary tumour only, this may point nostic significance of E-cadherin expression, although low levels of to clinically significant differences between TS staining in primary E-cadherin and a high ratio of type IV collagenase to E-cadherin and metastatic tumour. Although TS staining has demonstrated were associated with an increased incidence of metastasis. potential prognostic significance, prospective studies are required to confirm the importance in adjuvant chemotherapy and to deter- Helix pomatia agglutinin mine the role of TS levels in advanced colorectal cancer. Helix pomatia agglutinin (HPA) is a sugar-binding protein, or lectin, which binds specifically to N-acetylgalactosamine residues. GENERAL MARKERS The binding of this lectin has been suggested to be of prognostic Carcinoembryonic antigen significance in breast cancer. HPA binding was studied in 130 colorectal cancers (Dukes’ stage A, 14; B, 49; C, 62; unknown, 5). Carcinoembryonic antigen is often used as a serum marker in Using a histochemical approach to detect binding of this lectin follow-up studies of colorectal cancer to detect recurrences/ (Schumacher et al, 1994). HPA binding correlated with Dukes’ metastasis. However, generally, the detection of CEA staining in stage, but was not independently prognostic. colorectal tumour samples has not been found to be useful in providing prognostic information. INTRACELLULAR TARGETS OF CHEMOTHERAPY Sialyl Lewis antigen Thymidylate synthase The sialyl Lewis antigen (sLe ) is a cell-surface glycoprotein which is part of the blood group system of antigens and may serve as Thymidylate synthase (TS) is the primary intracellular target for a ligand for endothelial leucocyte adhesion molecule. This may two of the more commonly used chemotherapeutic agents for British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 201 human colonic carcinoma cell lines and tumours: a frequent, p53-independent facilitate the binding of tumour cells to endothelial cells, which is a mutation associated with improved outcome in a randomized multi-institutional step in the metastatic cascade. High expression of this antigen is trial. Cancer Res 57: 1769–1775 associated with increased tumour cell invasion. The presence of Baas IO, Mulder JR, Offerhaus GJA, Vogelstein B and Hamilton SR (1994) An sLe in 132 colorectal cancers (stage I, 36; II, 40; III, 38; IV, 18) evaluation of six antibodies for immunohistochemistry of mutant p53 gene product in archival colorectal neoplasms. J Pathol 172: 5–12 was evaluated by immunohistochemistry on formalin-fixed, wax- Baretton GB, Vogt M, Muller C, Diebold J, Schneiderbanger K, Schmidt M and embedded sections with a monoclonal antibody to this antigen Lohrs U (1996a) Prognostic significance of p53 expression, chromosome 17 (Nakamori et al, 1993). Tumours with greater than 5% of positive copy number, and DNA ploidy in non-metastasized colorectal carcinomas cells were assessed as positive and those with less than 5% positive (stages IB and II). Scand J Gastroenterol 31: 481–489 cells were scored as negative. Follow-up was between 5 and 7 Baretton GB, Diebold J, Christoforis G, Vogt M, Müller C, Dopfer K, Schneiderbanger K, Schmidt M and Löhrs U (1996b) Apoptosis and years. The presence of the sLe was associated with depth of immunohistochemical bcl-2 expression in colorectal adenomas and carcinomas. tumour invasion, lymphatic invasion, lymph node metastasis and Aspects of carcinogenesis and prognostic significance. Cancer 77: 255–264 tumour stage. Overall and disease-free 5-year survival was signifi- Bartik Z, Nordenskiöld BO and Sun Z-F (1997) p53 overexpression as a prognostic cantly less in the sLe -positive tumours, and this factor was also factor in patients with Dukes’ B colorectal adenocarcinoma. Int J Oncol 11: 1019–1023 independently prognostic for both overall survival and disease-free Bell SM, Scott N, Cross D, Sagar P, Lewis FA, Blair GE, Taylor GR, Dixon MF and survival. Quirke P (1993) Prognostic value of p53 overexpression and c-Ki-ras gene mutations in colorectal cancer. Gastroenterology 104: 57–64 CONCLUSIONS Belluco C, Guillem JG, Kemeny N, Huang Y, Klimstra D, Berger MF and Cohen AM (1996) p53 Nuclear protein overexpression in colorectal cancer: a Although the difficulty in identifying markers of prognosis has led dominant predictor of survival in patients with advanced hepatic metastases. to much scepticism, there is still a real need for methodologies to J Clin Oncol 14: 2696–2701 Benhattar J, Losi L, Chaubert P, Givel J-C and Costa J (1993) Prognostic guide the choice of therapy and use health care expenditure in a significance of K-ras mutations in colorectal carcinoma. Gastroenterology 104: more rational manner for this common disease. Future studies 1044–1048 must focus on multiple markers in large numbers of patients at a Bennett MA, Kay EW, Mulcahy H, O’Flaherty L, O’Donoghue, DP, Leader M and single disease stage in order to provide definitive evidence on the Croke DT (1995) ras and p53 in the prediction of survival in Dukes’ stage B clinical utility of prognostic markers and their relative contribution colorectal carcinoma. J Clin Pathol 48: M310–M315 Bhatavdekar JM, Patel DD, Ghosh N, Chikhlikar PR, Trivedi TI, Suthar TP, Doctor to the prediction of survival. The recent emphasis by the National SS, Shah NG and Balar DB (1997) Coexpression of Bcl-2, c-Myc, and p53 Cancer Institute on pathology archives linked to large therapeutic oncoproteins as prognostic discriminants in patients with colorectal carcinoma. studies as a resource for collaborative research is an example of Dis Colon Rectum 40: 785–790 the way funding can be used to answer these important questions Bosari S and Viale G (1995) The clinical significance of p53 aberrations in human tumors. Virchows Arch A Pathol Anat Histopathol 427: 229–241 about patient care. Bosari S, Viale G, Bossi P, Maggioni M, Coggi G, Murray JJ and Lee AKC (1994) Cytoplasmic accumulation of p53 protein: an independent prognostic indicator ACKNOWLEDGEMENTS in colorectal adenocarcinomas. J Natl Cancer Inst 86: 681–687 Bosari S, Moneghini L, Graziani D, Lee AKC, Murray JJ, Coggi G and Viale G Research in the authors’ laboratories are supported in part by an (1995) Bcl-2 oncoprotein in colorectal hyperplastic polyps, adenomas, and Aberdeen Colorectal Cancer Initiative grant from the University of adenocarcinomas. Human Pathol 26: 534–540 Aberdeen Development Trust. The helpful comments of Professor Brett MC, Pickard M, Green B, Howel-Evans A, Smith D, Kinsella A and Poston G (1996) p53 protein overexpression and response to biomodulated 5-fluorouracil J Cassidy, and Dr J McKay were greatly appreciated. Many thanks chemotherapy in patients with advanced colorectal cancer. Eur J Sur Oncol 22: to Maria Johnson for her secretarial skills. 182–185 Chambers AF and Matrisian LM (1997) Changing views of the role of matrix metalloproteinases in metastasis. J Natl Cancer Inst 89: 1260–1270 REFERENCES Cohen AM, Tremiterra S, Candela F, Thaler HT and Sigurdson ER (1991) Prognosis of node-positive colon cancer. Cancer 67: 1859–1861 Amaya H, Tanigawa N, Lu C, Matsumura M, Shimomatsuya T, Horiuchi T and Costa A, Doci R, Machen C, Bignami P, Faranda A, Gennari L and Silvestrini R Muroaka R (1997) Association of vascular endothelial growth factor expression (1997) Cell proliferation-related markers in colorectal liver metastases: with tumor angiogenesis, survival and thymidine phosphorylase/platelet- correlation with patients prognosis. J Clin Oncol 15: 2008–2014 derived endothelial cell growth factor expression in human colorectal cancer. Cox LS and Lane DP (1995) Tumor suppressors, kinases and clamps – how p53 Cancer Lett 119: 227–235 regulates the cell-cycle in response to DNA-damage. Bioessays 17: 510–508 Ambiru S, Miyazaki M, Ito H, Nakagawa K, Shimizu H, Nukui Y, Nozawa S, Okuno Dix BR, Robbins P, Soong R, Jenner D, House AK, Lacopetta BJ, and The General A, Yoshitomi H and Nakajima N (1997) A prospective study of prognostic Surgeons at Sir Charles Gairdner Hospital (1994) The common molecular value of type IV collagenase in colorectal cancer tissue. Dig Dis Sci 42: genetic alterations in Dukes’ B and C colorectal carcinomas are not short-term 1660–1665 prognostic indicators of survival. Int J Cancer 59: 747–751 Anderson SN, Lovig T, Breivik J, Lund E, Guadernack G, Meling GI and Rognum Dorudi S, Hanby AM, Poulsom R, Northover J and Hart IR (1995) Level of TO (1996) K-ras mutations and prognosis in large-bowel carcinomas. Scand J expression of E-cadherin mRNA in colorectal cancer correlates with clinical Gastroenterol 32: 62–69 outcome. Br J Cancer 71: 614–616 Apolinario RM, van der Valk P, de Jong JS, Deveille W, Van Arkotte J, Dingemans Elnatan J, Goh H-S and Smith DR (1996) C-KI-RAS activation and the biological A-MC, Van Mourik JC, Postmus PE, Pinedo HM and Giaccone G (1997) behaviour of proximal and distal colonic adenocarcinomas. Eur J Cancer 32A: Prognostic value of the expression of p53, bcl-2, and bax oncoproteins, and 491–497 neovascularization in patients with radically resected non-small-cell lung Fearon ER and Vogelstein BA (1990) The genetic model for colorectal cancer. J Clin Oncol 15: 2456–2466 tumorigenesis. Cell 61: 759–767 Arribas R, Capellà G, Tórtola S, Masramon L, Grizzle WE, Perucho M and Peinado Findlay MPN, Cunningham D, Morgan G, Clinton S, Hardcastle A and Aherne GW MA (1997). Assessment of genomic damage in colorectal cancer by DNA (1997) Lack of correlation between thymidylate synthase levels in primary fingerprinting: prognostic applications. J Clin Oncol 15: 3230–3240 colorectal tumours and subsequent response to chemotherapy. Br J Cancer 75: Astler VB and Coller FA (1954) The prognostic significance of direct extension of 903–909 carcinoma of the colon and rectum. Ann Surg 192: 846–850 Ganesh S, Sier CFM, Heerding MM, Vankrieken JHJM, Griffioen G, Welvaart K, Augenlicht LH, Wadler S, Corner G, Richards C, Ryan L, Multani AS, Pathak S, van de Velde CJ, Verheijen JH, Lamers, CB and Verspaget HW (1996) Benson A, Haller D and Heerdt BG (1997) Low-level c-myc amplifications in Contribution of plasminogen activators and their inhibitors to the survival © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203 202 HL McLeod and GI Murray prognosis of patients with Dukes’ stage B and C colorectal cancer. Br J Cancer Leahy DT, Salman R, Mulcahy H, Sheahan K, O’Donoghue DP and Parfrey NA 75: 1793–1801 (1996) Prognostic significance of p53 abnormalities in colorectal carcinoma Goh H-S, Yao J and Smith DR (1995) p53 Point mutation and survival in colorectal detected by PCR-SSCP and immunohistochemical analysis. J Pathol 180: cancer patients. Cancer Res 55: 5217–5221 364–370 Griffiths L and Stratford IJ (1997) Platelet-derived endothelial cell growth factor Leichman CG, Lenz H-J, Leichman L, Danenberg K, Baranda J, Groshen S, Boswell thymidine phosphorylase in tumour growth and response to therapy. Br J W, Metzger R, Tan M and Danenberg PV (1997) Quantitation of intratumoral Cancer 76: 689–693 thymidylate synthase expression predicts for disseminated colorectal cancer Hall PA and Lane DP (1994) p53 in tumour pathology: can we trust response and resistance to protracted-infusion fluorouracil and weekly immunohistochemistry? – Revisited! J Pathol 172: 1–4 leucovorin. J Clin Oncol 15: 3223–3229 Hermanek P, Wiebelt H, Staimner D and Riedl S (1995) Prognostic factors of rectum Linden MD, Nathanson DS and Jacobson G (1996) Lack of prognostic significance carcinoma – experience of the German multicentre study SGCRC. German of epidermal growth factor receptor expression in Dukes’ B colorectal Study Group Colo-Rectal Carcinoma. Tumori 81: 60–64 carcinomas. Lab Invest 74: 61A:343 Hoffmann D, Moore J and Roder D (1997) Trends in survival from colonic cancer: Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta GF, Jessup JM and The impact of subspecialization. Aus N Z J Surgery 67: 842–845 Pagano M (1997) Increased proteasome-dependent degradation of the cyclin Ishimuru G, Ookawa K, Yamaguchi N, Sakamoto M, Hirohashi S, Muto T and dependent kinase inhibitor p27 in aggressive colorectal carcinomas. Nature Yokota J (1994) Allelic losses associated with the metastatic potential of Med 3: 231–234 colorectal carcinoma. Int J Oncol 5: 267–273 Lowe SW, Ruley HE, Jacks T and Housman DE (1993) p53-dependent apoptosis Jackman AL, Kelland LR, Kimbell R, Brown M, Gibson W, Aherne GW, Hardcastle modulates the cytotoxicity of anticancer agents. Cell 74: 957–967 A and Boyle FT (1995) Mechanisms of acquired resistance to the quinazoline Manne U, Myers RB, Moron C, Poczatek RB, Dillard S, Weiss H, Brown D, thymidylate synthase inhibitor ZD1694 (tomudex) in one mouse and three Srivastava S and Grizzle WE (1997) Prognostic significance of Bcl-2 human cell lines. Br J Cancer 71: 914–924 expression and p53 nuclear accumulation in colorectal adenocarcinoma. Int J Jen J, Kim HG, Piantadosi S, Liu ZF, Lecitt RC, Sistonen P, Kinzler KW, Vogelstein Cancer 74: 346–358 B and Hamilton SR (1994) Allelic loss of chromosome 18q and prognosis in Markowitz S, Hines JD, Lutterbaugh J, Myeroff L, Mackay W, Gordon N, Rustum colorectal cancer. N Engl J Med 331: 213–221 Y, Luna E and Kleinerman J (1995) Mutant K-ras oncogenes in colon cancers Johnston PG, Drake JC, Trepel J and Allegra CJ (1992) Immunological quantitation do not predict patient’s chemotherapy response or survival. Clin Cancer Res 1: of thymidylate synthase using the monoclonal antibody TS 106 in 5- 441–445 fluorouracil-sensitive and -resistant human cancer cell lines. Cancer Res 52: Mayer A, Takimoto M, Fritz E, Schellander G, Kofler K and Ludwig H (1993) The 4306–4312 prognostic significance of proliferating cell nuclear antigen, epidermal growth Johnston PG, Fisher ER, Rockette HE, Fisher B, Wolmark N, Drake JC, Chabner factor receptor, and mdr gene expression in colorectal cancer. Cancer 71: BA and Allegra CJ (1994) The role of thymidylate synthase expression in 2454–2460 prognosis and outcome of adjuvant chemotherapy in patients with rectal cancer. McArdle CS and Hole D (1991) Impact of variability among surgeons on post- J Clin Oncol 12: 2640–2647 operative morbidity and mortality and ultimate survival. Br Med J 302: Johnston PG, Lenz H-J, Leichman CG, Danenberg KD, Allegra CJ, Danenberg PV 1501–1505 and Leichman L (1995) Thymidylate synthase gene and protein expression Miller F, Heimann TM, Quish A, Pyo DJ, Szporn A, Martinelli G and Fasy TM correlate and are associated with response to 5-fluorouracil in human colorectal (1992) Ras and c-myc protein expression in colorectal carcinoma. Dis Colon and gastric tumours. Cancer Res 55: 1407–1412 Rectum 35: 430–435 Kang SM, Maeda K, Chung YS, Onodda N, Ogawa Y, Takatsuka S, Ogawa M, Miyahara M, Saito T, Kaketani K, Sato K, Kuwahara A, Shimoda K and Kobayashi Sawada T, Nakata B, Nishiguchi Y, Ikehara T, Okuno M and Sowa M (1997) M (1991) Clinical significance of ras p21 overexpression for patients with an Vascular endothelial growth factor expression correlates with hematogenous advanced colorectal cancer. Dis Colon Rectum 34: 1097–1102 metastasis and prognosis in colorectal carcinoma. Oncol Rep 4: 381–384 Moertel CG, Fleming TR, MacDonald JS, Haller DG, Laurie JA, Tangen CM, Kapitanovic S, Radosevic S, Kapitanovic M, Ajndelinovic S, Ferencic Z, Tavassoli Ungerleider JS, Emerson WA, Tormey DC, Glick JH, Veeder MH, Mailliard JA M, Primorac D, Sonicki Z, Spaventi S, Pavelic K and Spaventi R (1997) The and Graff J (1995) Fluorouracil plus levamisole as effective adjuvant therapy HER-2/neu expression of p185 correlates with the stage of disease and survival in after resection of stage III colon carcinoma: a final report. Ann Intern Med 122: colorectal cancer. Gastroenterology 112: 1103–1113 321–326 Kay EW, Mulcahy H, Walsh CB, Leader M and O’Donoghue D (1994) Cytoplasmic Mulcahy HE, Duffy MJ, Gibbons D, McCarthy P, Parfrey NA, O’Donoghue DP and c-erb B2 protein expression correlates with survival in Dukes’ B colorectal Sheahan K (1994) Urokinase type plasminogen activator and outcome in carcinoma. Histopathology 25: 455–461 Dukes’ B colorectal cancer. Lancet 344: 583–584 Khine K, Goh H-S and Smith DR (1995) Prognostic significance of chromosome Mulder JWR, Kruyt PM, Sewnath M, Oosting J, Seldenrijk CA, Weidema WF, 5q and 17p allelic deletion in colorectal adenocarcinomas. Int J Oncol 7: Offerhaus GJ and Pals ST (1994) Colorectal cancer prognosis and expression 631–635 of exon-v6 containing CD44 proteins. Lancet 344: 1470–1472 Kitadai Y, Ellis LM, Tucker SL, Greene GF, Bucana CD, Cleary KR, Takahashi Y, Murray GI, Duncan ME, O’Neil P, Melvin WT and Fothergill JE (1996) Matrix Tahara E and Fidler IJ (1996) Multiparametric in situ mRNA hybridization metalloproteinase-1 is associated with poor prognosis in colorectal cancer. analysis to predict disease recurrence in patients with colon carcinoma. Am J Nature Med 2: 461–462 Path 149: 1541–1551 Nagai MA, Habr-Gama A, Oshima BS and Bretani MM (1992) Association of Kochhar R, Halling KC, McDonnell S, Schaid DJ, French AJ, O’Connell MJ, genetic alterations of c-myc, c-fos, and c-Ha-ras proto-oncogenes in colorectal Nagorney DM and Thibodeau SN (1997) Allelic imbalance and microsatellite tumors. Frequency and clinical significance. Dis Colon Rectum 35: 444–451 instability in resected Dukes’ D colorectal cancer. Diagn Mol Pathol 6: 78–84 Nakamori S, Kameyama M, Imaoka S, Furukawa H, Ishikawa O, Sasaki Y, Kabuto Kornmann M, Link KH, Lenz H-J, Pillasch J, Metzger R, Butzer U, Leder GH, T, Iwanaga T, Matsushita Y and Irimura T (1993) Increased expression of Weindel M, Safi F, Danenberg KD, Beger HG and Danenberg PV (1997) Sialyl Lewis antigen correlates with poor survival in patients with colorectal Thymidylate synthase is a predictor for response and resistance in hepatic carcinoma: clinicopathological and immunohistochemical study. Cancer Res artery infusion chemotherapy. Cancer Lett 118: 29–35 53: 3632–3637 Koshida Y, Saegusa M and Okayasu I (1997) Apoptosis, cell proliferation and Neoptolemos JP, Oates GD, Newbold KM, Robson AM, McConkey C and Powell J expression of Bcl-2 and Bax in gastric carcinomas: immunohistochemical and (1995) Cyclin/proliferation cell nuclear antigen immunohistochemistry does clinicopathological study. Br J Cancer 75: 367–373 not improve the prognostic power of Dukes’ or Jass’ classifications for Kressner ULF, Lindmark G, Gerdin B, Pahlman L and Glimelius B (1996) colorectal cancer. Br J Surg 82: 184–187 Immunohistochemical p53 staining is of limited value in the staging and Öfner D, Riehemann K, Maier H, Riedmann B, Nehoda H, Tötsch M, Böcker W, prognostic prediction of colorectal cancer. Anticancer Res 16: 951–958 Jasani B and Schmid KW (1995) Immunohistochemically detectable bcl-2 Kumar V, Bustin SA and Mckay IA (1995) Transforming growth factor-alpha. Cell expression in colorectal carcinoma: correlation with tumour stage and patient Biol Int 19: 373–388 survival. Br J Cancer 72: 981–985 Lanza G, Maestri I, Dubini A, Gafa R, Santini A, Ferretti S and Cavazzini L (1996) Ogunbiyi OA, Goodfellow PJ, Gagliardi G, Swanson PE, Birnbaum EH, Fleshman P53 expression in colorectal cancer. Relation to tumor type, DNA ploidy JW, Kodner IJ and Moley JF (1997) Prognostic value of chromosome 1 p pattern, and short-term survival. Am J Clin Pathol 105: 604–612 allelic loss in colon cancer. Gastroenterology 113: 761–766 Lawrence DA (1996) Transforming growth factor-beta – a general review. Eur Ogunbiyi OA, Goodfellow PJ, Herfarth K, Gagliardi G, Swanson PE, Birnbaum EH, Cytokine Network 7: 363–374 Read TE, Fleshman JW, Kodner IJ and Moley JF (1998) Confirmation that British Journal of Cancer (1999) 79(2), 191–203 © Cancer Research Campaign 1999 Prognosis in colorectal cancer 203 chromosome 18q allelic loss in colon cancer is a prognostic indicator. J Clin fraction and prognosis in colorectal adenocarcinoma. Eur J Cancer 27: Oncol 16: 427–433 1646–1649 Paradiso A, Rabinovich M, Vallejo C, Machiavelli M, Romero A, Perez J, Lacava J, Sun XF, Carsten JM, Zhang H, Stal O, Wingern S, Hatschek T and Nordenskjold B Cuevas MA, Rodriquez R, Leone B, Sapia MG, Simone G and De Lena M (1992) Prognostic significance of cytoplasmic p53 oncoprotein in colorectal (1996) p53 and PCNA expression in advanced colorectal cancer: response to adenocarcinoma. Lancet 340: 1369–1373 chemotherapy and long-term prognosis. Int J Cancer 69: 437–441 Sun XF, Carstensen JM, Stal O, Zhang H and Nordenskjold B (1995) c-erb B2 Parkin DM, Pisani P and Ferlay J (1993) Estimates of the worldwide incidence of oncoprotein in relation to DNA ploidy and prognosis in colorectal eighteen major cancers in 1985. Int J Cancer 54: 594–606 adenocarcenoma. APMIS 103: 309–315 Poller DN, Baxter KJ and Shepherd NA (1997) p53 and Rb1 protein expression: are Sun XF, Carstensen JM, Stal O, Zhang H and Nordenskjold B (1996) Proliferating they prognostically useful in colorectal cancer? Br J Cancer 75: 87–93 cell nuclear antigen (PCNA) in relation to ras, c-erbB-2, p53, clinico- Pricolo VE, Finkelstein SD, WU T-T, Keller G, Bakker A, Swalsky PA and Bland KI pathological variables and prognosis in colorectal adenocarcinoma. Int J (1996) Prognostic value of TP53 and K-ras-2 mutational analysis in stage III Cancer 69: 5–8 carcinoma of the colon. Am J Surg 171: 41–46 Takahashi Y, Tucher SL, Kitadai Y, Koura AN, Bucana CD, Cleary KR and Ellis LM Reinbach DH, McGregor JR, Murray GD and Odwyer PJ (1994) Effect of the (1997) Vessel counts and expression of vascular endothelial growth factor as surgeons speciality interest on the type of resection performed for colorectal- prognostic factors in node-negative colon cancer. Arch Surg 132: 541–546 cancer. Dis Colon Rectum 37: 1020–1023 Takebayashi Y, Akiyama S-I, Akiba S, Yamada K, Miyadera K, Sumizawa T, Robson H, Anderson E, James RD and Schofield PF (1996) Transforming growth Yamada Y, Murata F and Aikou T (1996) Clinicopathologic and prognostic factor beta 1 expression in human colorectal tumours: an independent significance of an angiogenic factor, thymidine phosphorylase, in human prognostic marker in a subgroup of poor prognosis patients. Br J Cancer 74: colorectal carcinoma. J Natl Cancer Inst 88: 1110–1117 753–758 Wadler S, Bajaj R, Neuberg D, Agarwal V, Haynes H and Benson ALB (1997) Schneider HJ, Sampson SA, Cunningham D, Norman AR, Andreyev HJN, Tilsed Prognostic implications of c-Ki-ras2 mutations in patients with advanced JVT and Clarke PA (1997) Bcl-2 expression and response to chemotherapy in colorectal cancer treated with 5-fluorouracil and interferon: A study of the colorectal adenocarcinomas. Br J Cancer 75: 427–431 Eastern Cooperative Oncology Group (EST 2292). Cancer J Scientific Am 3: Schumacher U, Higgs D, Loizidou M, Pickering R, Leathem A and Taylor I (1994) 284–289 Helix pomatia agglutinin binding is a useful prognostic indicator in colorectal Witzig TE, Lopfinzi CL, Gonchoroff NJ, Reiman HM, Cha SS, Wieand HS, carcinoma. Cancer 74: 3104–3107 Katzmann JA, Paulsen JK and Moertel CG (1991) DNA ploidy and cell kinetic Shibata D, Reale MA, Lavin P, Silverman M, Fearon ER, Steele G Jr, Jessup JM, measurements as predictors of reoccurrence and survival in stages B2 and C Loda M and Summerhayes IC (1996) The DCC protein and prognosis in colorectal adenocarcinoma. Cancer 68: 879–888 colorectal cancer. N Engl J Med 335: 1727–1732 Wolmark N, Wieand HS, Rockette HE, Fisher B, Glass A, Lawrence W, Lerner H, Sinicrope FA, Hart J, Michelassi F and Lee JJ (1995) Prognostic value of bcl-2 Cruz AB, Volk H, Shibata H, Evans J and Prager D (1983) The prognostic oncoprotein expression in stage II colon carcinoma. Clin Cancer Res 1: significance of tumour location and bowel obstruction in Dukes’ B and C 1103–1110 colorectal cancer. Ann Surg 198: 743–750 Smith DR and Goh H-S (1996) Overexpression of the c-myc proto-oncogene in Wynford-Thomas D (1992) p53 in tumour pathology: can we trust colorectal carcinoma is associated with a reduced mortality that is abrogated by immunohistochemistry? J Pathol 166: 329–330 point mutation of the p53 tumor suppressor gene. Clin Cancer Res 2: Yamaguchi A, Nakagawara G, Kurosaka Y, Nishimura G, Yonemura Y and 1049–1053 Miyazaki I (1993) P53 immunoreaction in endoscopic biopsy specimens of Smith DR, Elnatan J, Yao J and Goh H-S (1996) Point mutation of the c-Ki-ras colorectal cancer, and its prognostic significance. Br J Cancer 68: 399–402 proto-oncogene and the p53 tumour suppressor gene in distal colonic Yamaguchi A, Urano T, Goi T, Saito M, Takeuchi K, Hirose K, Nakagawara G, adenocarcinomas. Int J Oncol 8: 1165–1169 Shiku H and Furukawa K (1996) Expression of a CD44 variant containing Soong R, Grieu F, Robbins P, Dix B, Chen D, Parsons R, House A and Lacopetta B exons 8 to 10 is a useful independent factor for the prediction of prognosis in (1997) P53 alterations are associated with improved prognosis in distal colonic colorectal cancer patients. J Clin Oncol 14: 1122–1127 carcinomas. Clin Cancer Res 3: 1405–1411 Younes M, Fernandez L and Lechago J (1996) Transforming growth factor alpha Span M, Moerkerk PT, De Goeij AF and Arends JW (1996) A detailed analysis of (TGF-a) expression in biopsies of colorectal carcinoma is a significant K-ras point mutations in relation to tumor progression and survival in prognostic indicator. Anticancer Res 16: 1999–2004 colorectal cancer patients. Int J Cancer 69: 241–245 Zeng ZS, Sarkis AS, Zhang ZF, Klimstra DS, Charytonowicz E, Guillem JG, Spitz FR, Giacco GG, Hess K, Larry L, Rich TA, Janjan N, Cleary KR and Skibber Cordon-Cardo C and Cohen AM (1994) P53 nuclear overexpression – an JM (1997) P53 immunohistochemical staining predicts residual disease after independent predictor of survival in lymph node positive colorectal cancer chemoradiation in patients with high-risk rectal cancer. Clin Cancer Res 3: patients. J Clin Oncol 12: 2043–2050 1658–1690 Zeng ZS, Huang Y, Cohen AM and Guillem JG (1996) Prediction of colorectal Sun XF, Wingren S, Carstensen JM, Stål O, Hatschek T, Boeryd B, Nordenskjöld B cancer relapse and survival via tissue RNA levels of matrix metalloproteinase- and Zhang H (1991) Ras p21 expression in relation to DNA ploidy, s-phase 9. J Clin Oncol 14: 3133–3140 © Cancer Research Campaign 1999 British Journal of Cancer (1999) 79(2), 191–203

Journal

British Journal of Cancer – Springer Journals

Published: Dec 18, 1998

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Tumour markers of prognosis in colorectal cancer

Tumour markers of prognosis in colorectal cancer

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Tumour markers of prognosis in colorectal cancer

Tumour markers of prognosis in colorectal cancer

References (111)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies