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

Learn More →

Uses and Abuses of Tumor Markers in the Diagnosis, Monitoring, and Treatment of Primary and Metastatic Breast Cancer

Uses and Abuses of Tumor Markers in the Diagnosis, Monitoring, and Treatment of Primary and... Although breast cancer incidence continues to study designs cloud the issue of how the marker might increase, mortality has been decreasing, principally as be used. Reliance on p-values rather than the size of the a result of earlier detection and improvements in adju- differences in outcome between patients who are posi- vant systemic therapy. Nonetheless, because antineo- tive and those who are negative for the factor obscures plastic agents are associated with substantial morbid- the importance. Technical issues result in poor repro- ity and occasional mortality, efforts to individualize ducibility and interpretability of assays. Analytical treatment strategies are desirable. In addition to clas- issues lead to poorly def ined cutoff values for marker sic histopathologic diagnosis, molecular and cellular levels. Poor patient selection leads to diff iculty inter- tumor markers may help in establishing prognosis or preting results because of confounders such as differ- prediction of benefit. ences in treatment regimens. This review focuses on Recommendations for routine use of tumor mark- these issues, with an emphasis on currently accepted ers in breast cancer have been conservative. Although tumor markers. Finally, new tumor marker report- several studies have been reported, few are of suff i- ing recommendations are discussed, the adoption of ciently high level of evidence to permit solid conclu- which may lead to improved design and publication sions. Three key issues in tumor marker evaluation are of tumor marker studies in the future. The Oncologist utility, magnitude, and reliability. Poorly conceived 2006;11:541–552 Correspondence: Daniel F. Hayes, M.D., University of Michigan Comprehensive Cancer Center, 1500 East Medical Center Drive, Ann Arbor, Michigan 48109, USA. Telephone: 734-615-6725; Fax: 734-615-3947; e-mail: [email protected] Received February 22, 2006; accepted for publication April 11, 2006. ©AlphaMed Press 1083-7159/2006/$20.00/0 The Oncologist 2006;11:541–552 www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 542 Tumor Marker Pitfalls surveillance of patients free of detectable disease. Mea- Introduction surement of CA 15-3 or CA 27.29 and/or CEA was recom- Breast cancer is the most common cancer in women in the mended, however, to monitor selected patients with MBC U.S., with an estimated 213,000 new cases diagnosed in the undergoing palliative therapy [7]. U.S. in 2005 [1]. Despite these increasing numbers, mortal- Routine measurement of multiple tissue markers was also ity from breast cancer continues to decline. This decline discussed in the guidelines. The panel recommended routine is felt to be a result of a combination of earlier detection of measurement of estrogen and progesterone receptors (ER disease as a result of screening and improved treatment of and PgR, respectively) to identify patients most likely to ben- disease with adjuvant systemic therapies [2, 3]. A major- efit from endocrine therapy in either the early or metastatic ity of patients are cured with surgery and radiation therapy disease settings. In addition, measurement of Her-2/neu over- alone, and these patients will gain no additional benefit expression and possibly amplification was recommended for from adjuvant systemic therapies. In addition, having a high all patients at the time of initial diagnosis or recurrence, as it risk for recurrence does not imply that systemic therapy will is predictive of response to trastuzumab (Herceptin ; Genen- prevent it. Even for those who recur, overall survival and tech, South San Francisco, CA), a monoclonal antibody palliation of symptoms for patients with metastatic breast directed against the Her-2/neu receptor [8–10]. The panel felt cancer (MBC) has improved with the advent of new thera- that data to support assessment of other tissue-based mark- pies. However, currently available methods are inadequate ers, including p53, cathepsin D, and f low cytometry-derived to help the clinician precisely predict a priori which patients estimates of DNA content or S phase, were insufficient to will benefit from many of the available therapies. recommend usage in routine clinical practice. For patients with early-stage breast cancer, it would be Therefore, despite the large number of research studies helpful to identify which patients will relapse without adju- evaluating the prognostic and predictive ability of numer- vant systemic therapy, so that only patients who receive ben- ous tumor markers in breast cancer, the ASCO panel rec- efit are exposed to the inherent toxicities. Approach to treat- ommended few for routine use in clinical practice. Why ment of metastatic disease is generally with palliative intent were these recommendations so conservative? In the suc- rather than for cure. In this setting, identification of those ceeding sections of this paper, we outline the multiple fac- patients with rapidly progressive disease permits selection tors that underlie this conservative approach. of more rapidly acting but perhaps more toxic therapy. During the past few decades, with the explosion of When Is a Tumor Marker Useful (Use)? molecular technology and understanding of the biology of When evaluating tumor markers for use in clinical practice, breast cancer, numerous studies have been performed to clinicians should consider their utility, the magnitude of identify prognostic and predictive factors in breast cancer, their effects, and their reliability (Table 2). Tumor mark- with mixed success. Multiple expert panels have convened ers can be useful at multiple stages of cancer diagnosis and to analyze available data in order to establish guidelines for treatment (Table 3) [11, 12]. For example, for individuals the use of tumor markers, but their recommendations have who do not have cancer, a marker may be helpful in deter- been very conservative [4, 5]. In this review, we address mining the risk for developing the disease and/or it may be the pitfalls that have led to difficulties establishing tumor beneficial for screening for disease. Once an abnormality markers for routine clinical use, with a specific focus on is found, a tumor marker may be helpful for distinguishing tumor markers in breast cancer. between benign and malignant processes or between dif- ferent malignant processes. After confirmation of a cancer American Society of Clinical diagnosis, tumor markers can help monitor disease status Oncology Guidelines during and after therapy. The American Society of Clinical Oncology (ASCO) Tumor markers can also help determine prognosis inde- convened a panel of experts that first published recom- pendent of therapy and predict response to therapy. Prognos- mendations regarding the use of circulating and tissue- tic factors ref lect the metastatic potential and/or growth rate based tumor markers in breast cancer in 1996 [6] and most of the tumor and are used to select patient outcomes without recently updated these recommendations in 2001(Table 1) consideration of treatment given [13]. Predictive factors, on [4]. The ASCO panel evaluated multiple serum markers for the other hand, ref lect the sensitivity or resistance of a tumor breast cancer, including assays for MUC1 protein (cancer to a therapeutic agent and therefore are used to predict which antigen [CA] 15-3 and CA 27.29), carcinoembryonic anti- patients are likely to respond to a specific treatment [14]. gen (CEA), and the circulating extracellular domain of Her- Pure prognostic and predictive factors are depicted sche- 2/neu. The panel did not recommend monitoring of any of matically in Figures 1A and 1B, respectively. these markers for screening, diagnosis, staging, or routine The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 543 Table 1. Summary of American Society of Clinical Oncology guidelines for use of tumor markers in breast cancer [6, 7] Assess in Assess in Use for monitoring only adjuvant metastatic Prognostic Predictive Marker setting setting factor factor NED Met Tissue-based ER Yes Yes Weak Strong N/A N/A PgR Yes Yes Weak Strong N/A N/A Her-2/neu Yes Yes Weak Strong N/A N/A Circulating CA 15-3 No Yes No No No Yes CA 27.29 No Yes No No No Yes CEA No Yes No No No Yes Her-2 ECD No No No No No No For benefit from endocrine treatment. For benefit from trastuzumab therapy. Abbreviations: CA, cancer antigen; CEA, carcinoembryonic antigen; ECD, circulating extracellular domain; ER, estrogen receptor; NED, no evidence of disease; Met: metastatic; N/A: not applicable; PgR, progesterone receptor. Table 2. Requirements essential for acceptance of a tumor Few tumor markers are purely prognostic or predictive. marker In fact, most tumor markers have mixed prognostic and Determine utility of marker (Table 3) predictive features, and the utility typically depends on the Evaluate magnitude of effect (none, weak, moderate, strong) therapeutic agent in question. For example, ER expression is weakly favorably prognostic but strongly predictive of Analyze reliability of marker response to treatment with endocrine therapy, as illustrated Technical issues (assay) in Figure 1C. Her-2/neu overexpression, on the other hand, Analytical issues (cutoff points, test/validation sets, multivariate analysis) is an unfavorable prognostic factor and is strongly predic- Trial design issues (appropriate patient population) tive of response to therapy with trastuzumab, as shown in Figure 1D. Until appropriate studies have been performed both in vitro and in vivo, it can be difficult to know how to use a tumor marker appropriately in the clinical setting. In breast cancer, tumor markers are currently used in only a Table 3. Clinical uses of tumor markers few of these categories. Determine risk of developing disease How Useful Is the Tumor Marker Screening for disease (Magnitude)? Once a tumor marker use has been identified, it is impor- Establish diagnosis Differentiate benign versus malignant disease tant to determine the magnitude of the difference in out- Determine type of malignancy comes for that particular use between those who are marker positive and those who are not. By evaluating the difference Determine prognosis in outcome, regardless of treatment, between a patient posi- For primary disease, predict relapse For metastatic disease, predict progression tive for a given prognostic factor and one who is negative for Predict survival the factor, the relative strength of a prognostic factor can be determined [15]. This assessment requires the selection of Predict response to therapy an appropriate outcome of interest, such as improvement Hormone therapy Chemotherapy in symptoms or survival, or surrogates of these end points, Novel therapies (e.g., trastuzumab) such as response rates or progression-free survival. For example, a breast cancer patient with disease in the Monitor disease lymph nodes at the time of diagnosis is two to three times For primary disease, predict relapse For metastatic disease, follow detectable disease more likely to have a breast cancer event (local recurrence Adapted from Stearns V, Yamauchi H, Hayes DF. Circulating or distant metastasis) than a patient without lymph node tumor markers in breast cancer: accepted utilities and novel involvement, regardless of treatment. Since lymph node prospects. Breast Cancer Res Treat 1998;52:239–259, with status has classically been used to make clinical decisions, kind permission of Springer Science and Business Media. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 544 Tumor Marker Pitfalls we have arbitrarily designated it as a “strong” prognostic the “relative predictive value” (RPV), the ratio of the likeli- factor, using it as the gold standard to set the criteria for con- hood that a factor-positive patient will respond to treatment sideration of other, putative markers [16]. A strong prognos- to the likelihood that a factor-negative patient will respond tic factor is depicted by Factor 1 in Figure 1A. Alternatively, to treatment. As with prognostic factors, we have proposed untreated patients with ER-positive breast cancer have only arbitrary classes of prediction factors for breast cancer slightly better outcomes than those with ER-negative dis- therapies based on what has been accepted by consensus, ease, and therefore we designated ER status as a weak prog- in this case ER [19]. Adjuvant tamoxifen therapy has been nostic factor, as portrayed by Factor 2 in Figure 1A [17, 18]. shown to decrease recurrence rates for ER-positive patients In a previous publication, we have suggested hazard ratios by 40% –50%, whereas ER-negative patients obtain mini- of <1.5, 1.5–2, and >2, to distinguish weak, moderate, and mal, if any, benefit from hormonal therapy [20]. Therefore, strong prognostic factors, respectively, for breast cancer the RPV is >8. Similarly, the majority of patients with ER- [16]. Such arbitrary designations would need to be estab- positive MBC have a clinical response to hormonal therapy, lished for other uses, as appropriate. whereas patients with ER-negative disease do not respond Predictive factors can also be classified as weak (Factor [21, 22]. ER status is therefore a strong predictive factor for 1), moderate, or strong (Factor 2), depending on their ability response to hormonal therapy. In this framework, we have to predict response to, and therefore benefit from, a given arbitrarily proposed that, in breast cancer, weak, moderate, therapy, as illustrated in Figure 1B. One measure to permit and strong predictive factors correspond to RPVs of 1–2, comparison of the relative strengths has been designated 2– 4, and >4, respectively [15]. It is important to understand Figure 1. Schematic representation of prognostic and predictive factors. (A): Pure prognostic factor. (B): Pure predictive fac- tor. (C): Mixed factor associated with weakly favorable prognosis and strong response to specific therapy. (D): Mixed factor associated with unfavorable prognosis and strong response to specific therapy. Modified from Hayes DF, Trock B, Harris AL. Assessing the clinical impact of prognostic factors: when is “statistically significant” clinically useful? Breast Cancer Res Treat 1998;52:305–319, with kind permission of Springer Science and Business Media. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 545 the clinical implications of the relative strength of a prog- blotting or reverse transcriptase-polymerase chain reaction nostic and/or predictive factor when integrating this infor- [RT-PCR]), and/or measures of DNA amplification (by mation into routine practice, and to determine if the data f luorescence or chromogenic in situ hybridization [FISH support its use in a specific clinical situation. and CISH, respectively]). Furthermore, even within these categories, different reagents (e.g., different antibodies in How Reliable Is the Tumor Marker IHC assays) are used in different tests. The results are not (Precision and Accuracy)? interchangeable, either within or between classes of assays, The preceding discussion illustrates the importance of esti- and therefore researchers must decide which methodology mating the magnitude of the relative tumor marker effect they will employ. Once that decision is made, researchers for a selected use. However, the marker is only useful if the must then decide how to perform the assay. For example, estimate of its magnitude is reliable and reproducible. In this when assessing Her-2/neu overexpression by IHC, tech- regard, many investigators conclude that their marker of nical issues such as antibody concentration and antigen interest has clinical utility if in their study the difference in retrieval methods may cause unacceptably high false-posi- outcomes between marker “positive” and marker “negative” tive or false-negative rates. patients is less than conventional measures of statistical sig- In one study, IHC and FISH resulted in only a 65% nificant (p < .05). This conclusion may be mistaken. Statisti- agreement for Her-2/neu status [23]. In a different study, cal significance only suggests that in the population chosen results obtained from local laboratories were compared for that study, the differences observed are likely not to be with those from a central laboratory for two Her-2/neu a result of chance alone. It does not imply clinical utility, assays, the HercepTest™ IHC assay (Dako North America, nor does a p-value <0.05 document the validity of the tumor Inc., Carpinteria, CA) and the FISH assay, with 79% con- marker. Although it is important to determine that the dif- cordance for HercepTest™ and 85% concordance for FISH ferences in outcome achieve statistical significance, statisti- [24]. Therefore, for the same test at multiple laboratories, cal significance alone does not determine clinical utility. and for different tests for the same marker, there is a signifi- In addition to determining when to use a tumor marker cant degree of discordance for two commonly used tests for and the magnitude of its effect, it is important to ensure that the evaluation of Her-2/neu status. the technical aspects of the marker are reliable and repro- The stakes are high. Recently reported data suggest ducible and that the study design and conduct are appropri- that adjuvant trastuzumab decreases recurrence rates by ate to test the marker for a clinical use of interest. Several 50%. However, up to 5% of patients who receive trastu- problems with tumor marker studies, including technical, zumab develop cardiac dysfunction, and the cost of 1 year analytical, and trial design issues, have limited the intro- of therapy may exceed $100,000. Therefore, it is essential duction of new prognostic and predictive factors into rou- that Her-2/neu, the target for trastuzumab, be assayed accu- tine clinical practice [11]. rately and precisely for every tissue sample. Expert panels are now being convened to establish guidelines for the con- What Technical Factors Inf luence Measurement duct and interpretation of common tumor marker assays, of Markers? including ER and Her-2/neu. These guidelines should lead From a technical standpoint, difficulties arise because of to standardization of the assays, which should allow for poor sensitivity and/or specificity of the assay for the ana- more reliable results both for routine clinical practice and lyte, poorly reproducible assays, and differences between for use of these assays in clinical trials. assays that use different reagents for measurement of the same marker [11]. Even for the two most commonly used What Analytical Issues Are Important and accepted tumor markers, ER expression and Her-2/neu to Consider? overexpression, standard methodologies have not yet been established [4, 5]. Two primary technical considerations Assay Interpretation are critical when measuring a tumor marker. The first is Determination of assay results can also vary, even for a single which type of assay should be used. The second is the repro- type of assay. For example, with visual assays such as IHC ducibility of the chosen assay, from both a technical and an for ER and Her-2/neu, intra- and interobserver variability analytical perspective. leads to differences in interpretation [25, 26]. Some attempts For example, Her-2/neu status can be determined by have been made to standardize interpretation, such as devel- measures of protein expression (by immunohistochemis- opment of the so-called “Allred score” for semiquantitation try [IHC], Western blotting, or enzyme-linked immuno- of ER expression [27], but these have not been universally sorbent assay), measures of RNA expression (by Northern adopted. Automated and semiautomated systems appear to www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 546 Tumor Marker Pitfalls be highly accurate and are likely to be more reproducible. lar patient cohort from the same study [31]. Another com- Examples of automated systems include the ChromaVision mon method to generate a data-derived cutoff point is to ACIS system (ChromaVision Medical Systems, Inc., San construct a receiver operating characteristic curve, which Juan Capistrano, CA) for ER expression measurement, the demonstrates the tradeoff between the sensitivity and spec- CellSearch™ assay (Veridex, LLC, Warren, NJ) for detec- ificity of an assay at different cutoff points. tion of circulating tumor cells [28], and the Oncotype DX™ Recently, a novel data-derived method to select cutoff assay (Genomic Health, Inc., Redwood City, CA), a new points, designated subpopulation treatment effect pattern prognostic tool for patients with hormone receptor (HR) - plot (STEPP) analysis, has been proposed [32]. STEPP positive, lymph node-negative breast cancer [29]. analysis evaluates outcomes to specific treatments in sub- populations of patients within randomized clinical trials or Cutoff Point Determination meta-analyses [32]. For example, it has been proposed that Regardless of the assay, one has to select some value or level recurrence rates after treatment with chemotherapy should that distinguishes positive from negative results. However, be evaluated in the context of the endocrine responsiveness there is no consensus regarding cor rect methods to establish of tumors, since HR-positive and -negative tumors appear cutoff points, and different studies of the same prognostic or to behave differently. Rather than arbitrarily defining cutoff predictive factor can have widely varying “optimal” cutoff points for ER positive and negative, the authors performed points [30]. a STEPP analysis of data from a previously conducted ran- Cutoff points may be defined using either arbitrary or domized clinical trial and were able to demonstrate a ben- data-derived methods (Table 4). One approach is to con- efit from chemotherapy only in the subset of patients with sider any value greater than two standard deviations above very low ER values. the mean for normal subjects to be positive. Cutoff points can also represent arbitrary values within affected patients; Cutoff Point Validation for example, one might decide that 10%, 50%, or 90% of Regardless of whether cutoff points are chosen arbitrarily affected patients will be classified as “positive.” Others or are data-derived, the selected cutoffs require subsequent have defined cutoff points based on technical factors, such validation. The initial evaluation should be performed using as the limit of detection of the assay [21]. Finally, the cutoff a “test set” of patients. In the second part of the study, the point for a new assay can be defined by comparing it with an utility of the cutoff point should then be confirmed using a older assay [27]. separate “validation set” composed of a similar, but com- Deriving cutoff points based on patient outcome data pletely independent, patient population. may provide more accurate values. For example, the cutoff For example, the Oncotype DX™ assay is based on the point for ER expression was first defined by limits of the principle of evaluating expression of multiple candidate assay and later by determining the optimal level that dis- genes using quantitative RT-PCR [29]. The investigators tinguished those patients who respond to hormonal therapy initially screened more than 20 0 candidate genes with from those who do not. In another example, the cutoff point the aim of developing a test that would predict the likeli- for the CellSearch™ assay for circulating tumor cells was hood of recur rence of cancer in patients with HR-positive, initially determined based on differences in time to pro- lymph node-negative breast cancer. Breast cancer tissues gression of a test set of patients with metastatic disease, and from 447 patients with HR-positive, lymph node-negative this cutoff was then validated with an independent but simi- tumors were used retrospectively to generate an algorithm using 16 of these genes that permitted division of patients into subgroups with ver y low, intermediate, or ver y high Table 4. Methods used for selection of cutoff points risk for recur rence. Patients are assigned to these groups Arbitrary based on a “recur rence score” der ived from the algo- Based on limit of detection of the assay r ithm. The major ity of the test samples were obtained Defined as two standard deviations above the mean of normal patients from patients treated with tamoxifen alone in the National Defined based on mean value in affected versus normal Surgical Adjuvant Breast and Bowel Project (NSABP) patients “Appropriate” percentage of positive cells B-20 trial [33], and the data-derived algorithm was then validated using a separate retrospective cohort of patients Data-derived from the NSABP B-14 trial [34], who had simila r clini- Plot p-value versus outcome cal characteristics and were also treated with tamoxifen Plot magnitude of effect of marker versus outcome Receiver operating characteristic curve alone. If the cohor ts had different clinical characteristics Subpopulation treatment effect pattern plot analysis or had been treated differently, the validation would not The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 547 have been legitimate. Validation of the results in a sepa- clinical trial is a properly powered, prospective, random- rate patient population strongly suggests that the test is ized trial designed specifically to evaluate the clinical reliable and that the results a re likely to be meaningful in a utility of a tumor ma rker in question for a specific and pre- larger population, as long as the patients tested are similar designated use. In such a study, diagnostic and /or thera- to the cohor ts included in the original studies. peutic decisions for the study arm are based on the tumor marker in question, whereas the decisions for the control Statistical Analysis arm are made independently [11]. Of course, statistical analysis is necessary to determine that Systematic overviews and/or pooled analyses of well- the observations are not a result of chance alone. However, conducted LOE II studies are equivalent to LOE I stud- once a tumor marker has been identified and validated, it is ies, especially if the correlative studies address a specific important to determine the relative value of the marker in use but are underpowered in a single study. It has been the context of previously identified prognostic and/or pre- estimated that a clinical trial that has been appropriately dictive factors, such as lymph node status and tumor size, powered to determine a clinical end point, such as progres- using some type of a multivariate analysis. If such an evalu- sion-free survival, is underpowered for analysis of tumor ation is not performed, clinicians will be unable to deter- marker-designated subgroups by one fourth, even if tissue mine the usefulness of incorporating the new marker into from 100% of enrollees is available. Moreover, interpreta- routine clinical practice. tion still requires judgment regarding the clinical impor- tance of the finding. Trial Design and Frameworks for For example, from the prospective randomized clinical Generating, Reporting, and Evaluating trials of adjuvant trastuzumab versus placebo, we anticipate Tumor Marker Results combined analyses of the multiple underpowered LOE II One of the key steps in identifying and confir ming the studies evaluating novel markers for benefit from this drug. benefit of a new tumor marker is appropriate study design. These pooled results should help focus trastuzumab therapy The Tumor Ma rker Utility Grading System (TMUGS) in the subgroups of Her-2/neu-positive patients most likely was initially developed by members of the ASCO panel to benefit. The combined analyses of small LOE III studies to provide a framework within which the utility of tumor that contain patients with variable clinical characteristics ma rkers can be graded based on published information and treatments, however, are more likely useful to generate [11]. Ma rkers a re assigned a grade based on the level of new hypotheses than to provide clinically useful and vali- evidence (LOE) available. These LOE ref lect the relative dated results. quality of the studies used to generate an estimate of the For all studies, whether prospective or retrospective, it effect of the ma rker (Table 5). LOE I and II studies a re is important to identify the appropriate patient population the most beneficial for evaluating the utility of a tumor to be investigated. All patients should have a similar profile marker. According to the TMUGS framework, the ideal based on known prognostic factors. Importantly, the effects Table 5. Levels of evidence for determining utility of tumor marker Level Type of evidence I Evidence from a single, high-powered, prospective controlled study that is specifically designed to test marker, or evidence from well-done meta-analysis and/or overview of level I studies. Ideally, the study is a prospective, randomized controlled trial in which diagnostic and/or therapeutic clinical decisions in one arm are determined at least in part on the basis of marker results, and diagnostic and/or therapeutic clinical decisions in the control arm are made independently of marker results, or Evidence from overview of level of evidence II studies addressing specific use. II Evidence from study in which marker data are determined in relationship to prospective therapeutic trial that is performed to test therapeutic hypothesis but not specifically designed to test marker utility. Specimen collection for marker study and statistical analysis are prospectively determined in protocol as secondary objectives. III Evidence from large but retrospective studies from which variable numbers of samples are available or selected. Statistical analysis for tumor marker was not dictated prospectively at time of therapeutic trial design. IV Evidence from small retrospective studies that do not have prospectively dictated therapy, follow-up, specimen selection, or statistical analysis. VEvidence from small pilot studies designed to determine or estimate distribution of marker levels in sample populations. Reprinted from Hayes DF, Bast RC, Desch CE et al. Tumor marker utility grading system: a framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst 1996;88(20):1456 –1466, permission of Oxford University Press. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 548 Tumor Marker Pitfalls of systemic treatment are critical and must be considered. Real-World Clinical Examples If the study is addressing the prognostic value of a marker, In the preceding sections we identified the essential ele- all patients should have been treated uniformly and without ments for establishing the usefulness, strength, and reli- whatever treatment might be considered if the patients have ability of tumor markers. Let us now discuss the data sup- a “poor prognosis.” If the question is addressing the utility porting two currently used tumor markers, Her-2/neu and of adding any treatment, all patients should be untreated. Oncotype DX™. If the question is whether more treatment should be given, then all patients should have received the same treatment. Her-2/neu If the study is addressing predictive factors, a control The first report of Her-2/neu as a prognostic factor in breast group that has been treated identically to the study group, cancer was published in 1987 [36]. Since then, more than with the exception that they did not receive the treatment 200 papers addressing this topic have been published, with in question, is essential. Although the control group might widely mixed and disparate results [37]. Different authors be from a selected historical control, predictive factors are have concluded that Her-2/neu is associated with poor out- ideally studied in the context of prospective, randomized, comes, no difference in outcome, or even favorable out- controlled trials comparing the patients who received the comes. Indeed, a great deal of this confusion could have treatment in question with those who did not receive that been avoided if the investigators would have addressed the treatment. components described above: (a) What is the intended use, Tumor marker studies should be carefully designed, (b) What is the magnitude of difference between positive using the above criteria, to obtain clinically useful infor- and negative for that use, and (c) How reliable is the esti- mation. Researchers frequently have practical difficulties mate of the magnitude? designing such studies, however, because of the need for sig- The potential uses for Her-2/neu are for prognosis and nificant numbers of patients with particular clinical char- prediction, as outlined in Table 6. Issues related to the reli- acteristics in order to address a specific clinical question. ability of the assays have been described in detail above. As discussed above, this can sometimes be overcome by Overall, studies support that Her-2/neu overexpression is pooling the results of several well-done but underpowered a poor prognostic factor, although its magnitude appears studies. Another significant drawback is obtaining fund- weak. For example, in Adjuvant! Online it has only a rela- ing for tumor marker studies, as pharmaceutical companies tive predictive value of 1.5 [38]. Its role as a prognostic fac- and third-party payers derive relatively smaller financial tor thus remains unclear. benefit from the results compared to the enormous payoffs More data exist to support the role of Her-2/neu status for for a “blockbuster” therapeutic agent. Regardless, given the prediction of response to standa rd therapies and trastuzumab. consequences, one has to question why it is acceptable for For selective estrogen receptor modulators (SERMs), such tumor marker studies to be performed with less scientific as tamoxifen, preclinical and clinical studies suggest that rigor than studies of new pharmaceutical agents. Her-2/neu positivity confers a relative resistance, with mod- Repor ting of tumor marker studies has also been his- erate magnitude, although the data are LOE III at best [39]. tor ically haphaza rd. Recently, in order to standa rdize Data for aromatase inhibitors (AIs) are mixed, although in repor ting of tumor ma rker study results, the National Can- one pilot study of neoadjuvant endocrine therapy, Her-2/neu cer Institute-European Organization for Resea rch and overexpression correlated with lower response to tamoxifen Treatment of Cancer (NCI-EORTC) Work ing Group on than to AIs [39]. At present, Her-2/neu status is not used to Cancer Diagnostics developed R Epor ting recommenda- determine which endocrine therapy to use, because of the tions for tumor MARKer prognostic studies (R EMARK) poor level of available evidence and conf licting data. Con- [35]. T he guidelines outline items that should be firmation of these results may lead to preferential use of AIs addressed by resea rchers when repor ting the results of in patients with Her-2/neu-positive disease. tumor marker studies, including prospectively defining Patients with tumors that overexpress Her-2/neu appear the question the study is tr ying to address, identifying the to have relative resistance to some chemotherapy regimens, appropriate patient population and controls, determining such as cyclophosphamide, methotrexate, and 5-f luoroura- the end point, and identifying potential confounding fac- cil (CMF), but not to others, such as anthracycline-contain- tors. Explicit recommendations a re given rega rding which ing regimens [37, 40, 41]. Her-2/neu status is not generally information must be contained in publications of tumor a consideration when choosing a chemotherapy regimen, ma rker studies, including patient and treatment infor- however, because, as is the case with endocrine therapy, the mation, specimen characteristics, assay methods, study level of available evidence does not support using Her-2/neu design, and statistical analysis methods. status to predict response to chemotherapy. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 549 Table 6. Theoretical uses for tissue-based Her-2/neu assessment Use Effect Magnitude LOE Prognosis Negative Moderate III Prediction Hormone therapy SERM Negative Weak-moderate III AI Neutral N/A III Chemotherapy Nonanthracycline Negative Weak-moderate III Anthracycline Positive or Neutral Moderate or N/A III Trastuzumab Positive Strong II Negative indicates worse outcome if Her-2/neu-positive; positive indicates better outcome if Her-2/neu-positive; neutral indicates outcome same regardless of Her-2/neu result. Abbreviations: AI, aromatase inhibitor; LOE, level of evidence; N/A, not applicable; SERM, selective estrogen receptor modulator. In contrast, Her-2/neu status appears to be strongly pre- Oncotype DX™ has also been evaluated as a predic- dictive of response to trastuzumab. A patient with a tumor tive factor, although less rigorously. In the NSABP B-14 that overexpresses Her-2/neu is usually treated with trastu- trial, the patient cohorts were treated with tamoxifen ver- zumab in either the adjuvant or metastatic settings [8, 9, 42, sus observation, and comparison of the cohorts suggested 43] because the benefits outweigh the risks in the majority of that the Oncotype DX™ assay is predictive for tamoxifen cases. A patient with a tumor that fails to overexpress Her- [45]. Similarly, analysis of NSABP B-20, in which patients 2/neu appears not to respond to treatment with trastuzumab were randomized to CMF and tamoxifen versus tamoxi- [10, 44] and therefore would not be treated with trastuzumab fen alone, permitted the investigators to determine that the to avoid both unnecessary toxicity and cost. Thus, use of assay is predictive for response to chemotherapy [45]. Her-2/neu to select trastuzumab is recommended based on Are the available data sufficient to conclude that Onco- “use” and “magnitude.” However, as discussed above, there type DX™ has been validated to the extent that patient are substantial apparent difficulties with the technical reli- treatment decisions should be based on the results? Perhaps, ability of all available assays for the marker. Nonetheless, but because these studies were all proposed using available despite the shortcomings of Her-2/neu studies outlined samples from trials performed many years ago and repre- above, at present there is sufficient LOE II evidence to sup- sented only subsets of the overall population entered into port the routine clinical use of Her-2/neu overexpression for the trials, concerns have been raised about wholesale clini- selection of trastuzumab therapy, as indicated in the most cal adoption of this assay. In that regard, the North Ameri- recent ASCO tumor marker guidelines [4]. can Breast Cancer Intergroup is developing the TailorRx clinical trial to further validate and extend the Oncotype Multigene Expression: Oncotype DX™ DX™ results. The trial design assumes that the assay is A more recent example of development of a new tumor prognostic, and will confirm the ability of the assay to pre- marker is provided by the case of Oncotype DX™. The dict response to chemotherapy. investigators specifically developed the test to determine In the TailorR X trial, tumors of patients with ER-posi- prognosis in ER-positive, lymph node-negative patients tive and lymph node-negative breast cancer will be tested who were treated with tamoxifen [29]. The available results using the Oncotype DX™ assay. Patients with low recur- suggest that, in this group of patients, Oncotype DX™ is rence scores, who have good prognoses without chemo- a strong prognostic factor because the ratio of the hazard therapy, will receive hormonal therapy alone. At the other ratios of the high and low recurrence score cohorts is >2 end of the spectr um, patients with high recur rence scores in both the test and validation cohorts [29]. The Oncotype will receive chemotherapy in addition to hormonal ther- DX™ test is also reliable because it fulfills the criteria out- apy. Those patients whose scores fall in the intermediate lined above for technical, analytical, and trial design issues range will all receive hormonal therapy and be randomly (Table 2). The assay is reproducible and was validated in assigned to chemotherapy or not. This trial design will independent test and validation cohorts of patients, as permit validation of the Oncotype DX™ results in a simi- described above. lar patient population in a large prospective clinical trial, www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 550 Tumor Marker Pitfalls and will allow for generation of new data on which to base studies to date, however, very few markers have been t reatment recom mendations for patients whose recur rence accepted for routine use by groups such as ASCO. When scores are intermediate. designing studies to establish a new tumor marker, or new Given the substantial technical, analytical, and trial use for an old marker, it is important to address the utility, design problems with previously performed tumor marker magnitude, and reliability of the marker (Table 2). studies, it is imperative to address these issues. It is espe- Frameworks such as TMUGS can be useful when cially important to standardize the assays for commonly designing and conducting these studies to ensure that used tumor markers. Otherwise, patients with false-posi- appropriate components are included, thereby leading to tive test results for predictive factors will receive treatments the establishment of new tumor markers for routine clinical that are not beneficial but which may cause significant use [11]. By progressively generating and refining a hypoth- toxicity, and those with false-negative results will not be esis, based on data derived from increasingly well-devel- offered potentially life-saving therapies. In addition, a bet- oped studies, tumor markers with clinical utility can be ter understanding of the potential pitfalls in tumor marker identified (Fig. 2). In addition, the new REMARK guide- study design will allow for the development of new, poten- lines should promote better design and conduct of stud- tially more useful assays. ies specifically focused on tumor marker validation [35]. Implementation of these recommendations when designing Conclusions tumor marker studies will result in the generation and pub- Tumor markers, when well defined, can play a significant lication of appropriate and complete clinical data, leading role in prediction and prognosis for breast cancer patients. to the adoption of new, well-validated tumor markers for Because of the abundance of poorly designed tumor marker routine clinical use. Figure 2. Tumor marker development f low chart. LOE, level of evidence; MVA, multivariate analysis. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 551 funding from Immunicon. D.F.H. has been a consultant / Authors’ Note advisor y panel pa r ticipant or held lecture / honora r ium Suppor ted by National Institutes of Health grant position dur ing past yea r for Dendreon, Im municon, 5R01CA092461-03 and Fashion Footwear Charitable Novar tis, Pfizer, Precision Therapeutics, Inc., Oncotech, Foundation of New York /QVC Presents Shoes-on-Sale™. and Veridex. D.F.H. was the principle or coinvestigator during the past yea r on studies funded by Immunicon, Disclosures Wyeth Ayerst-Genetics Institute, Pfizer, and Novar tis. D.F.H. has served as an unpaid consultant for Genomic Health, Inc. in the past yea r and has received resea rch 14 Gaspa r ini G, Pozza F, Ha r r is A L. Evaluating the potential usefulness of References new prognostic and predictive indicators in node-negative breast cancer 1 American Cancer Society. Cancer Facts and Figures, 2005. Available at patients. J Natl Cancer Inst 1993;85:1206 –1219. http://www.cancer.org/downloads/STT/CAFF2005f4PWSecured.pdf. 15 Hayes DF, Trock B, Har ris AL. Assessing the clinical impact of prognos- Accessed March 10, 2005. tic factors: when is “statistically significant” clinically useful? Breast 2 Peto R, Boreham J, Cla rke M et al. UK and USA breast cancer deaths down Cancer Res Treat 1998;52:305 –319. 25% in year 2000 at ages 20 -69 years. Lancet 2000;355:1822. 16 Isaacs C, Stea r ns V, Hayes DF. New prognostic factors for breast cancer 3 Ber r y DA, Cronin K A, Plevr itis SK et al. Effect of screening and adju- recur rence. Semin Oncol 2001;28:53 – 67. vant therapy on mor tality from breast cancer. N Engl J Med 20 05;353: 17 Fisher B, Costantino JP, Wickerham DL et al. Tamoxifen for prevention of 1784 –1792. breast cancer: repor t of the National Surgical Adjuvant Breast and Bowel 4 Bast RC Jr, Ravdin P, Hayes DF et al. 2000 update of recommendations for Project P-1 Study. J Natl Cancer Inst 1998;90:1371–1388. the use of tumor markers in breast and colorectal cancer: clinical practice 18 K night WA, Livingston R B, Gregor y EJ et al. Estrogen receptor as an guidelines of the Amer ican Society of Clinical Oncology. J Clin Oncol independent prognostic factor for early recur rence in breast cancer. Can- 2001;19:1865–1878. cer Res 1977;37:4669 – 4671. 5 Hammond ME, Fitzgibbons PL, Compton CC et al. College of American 19 Hayes DF. Do we need prognostic factors in nodal-negative breast cancer? Pathologists Conference XXXV: solid tumor prognostic factors-which, Arbiter. Eur J Cancer 2000;36:302–306. how and so what? Summa r y document and recommendations for imple- mentation. Arch Pathol Lab Med 2000;124:958 –965. 20 Ea rly Breast Cancer Tr ialists’ Collaborative Group (EBCTG). Effects of chemotherapy and hor monal therapy for ea rly breast cancer on recur- 6 Clinical practice guidelines for the use of tumor ma rkers in breast and rence and 15-yea r survival: an overview of the randomised tr ials. Lancet colorectal cancer. Adopted on May 17, 1996 by the American Society of 2005;365:1687–1717. Clinical Oncology. J Clin Oncol 1996;14:2843 –2877. 21 McGuire W L, Cham ness GC, Costlow ME et al. Steroids and human 7 Bast RC Jr, Ravdin P, Hayes DF et al. Er rata: 20 0 0 update of recommenda- breast cancer. J Steroid Biochem 1975;6:723 –727. tions for the use of tumor markers in breast and colorectal cancer: clinical practice guidelines of the American Society of Clinical Oncology. J Clin 22 Allegra JC, Lippman ME, Thompson EB et al. Estrogen receptor status: Oncol 2001;19:4185 – 4188. an impor tant variable in predicting response to endocrine therapy in met- astatic breast cancer. Eur J Cancer 1980;16:323 –331. 8 Cobleigh MA, Vogel CL, Tr ipathy D et al. Multinational study of the efficacy and safety of humanized anti-HER2 monoclonal antibody in 23 Dressler LG, Ber r y DA, Broadwater G et al. Compa rison of HER2 status women who have HER2-overexpressing metastatic breast cancer that by f luorescence in situ hybridization and immunohistochemistr y to pre- has progressed after chemotherapy for metastatic disease. J Clin Oncol dict benefit from dose escalation of adjuvant doxor ubicin-based therapy 1999;17:2639–2648. in node-positive breast cancer patients. J Clin Oncol 2005;23:4287– 4297. 9 Slamon DJ, Leyland-Jones B, Sha k S et al. Use of chemotherapy plus a 24 Perez EA, Suman VJ, Davidson N E et al. HER2 testing by local, central, monoclonal antibody against H ER2 for metastatic breast cancer that and reference laborator ies in the NCCTG N9831 Intergroup Adjuvant overexpresses HER2. N Engl J Med 2001;344:783 –792. Trial. J Clin Oncol 2004;22:567a. 10 Mass R D, Press MF, Anderson S et al. Evaluation of clinical outcomes 25 Press MF, Hung G, Godolphin W et al. Sensitivity of HER-2/neu anti- according to H ER2 detection by f luorescence in situ hybr idization in bodies in a rchival tissue samples: potential source of er ror in immuno- women with metastatic breast cancer treated with trastuzumab. Clin histochem ical studies of oncogene expression. Cancer Res 1994;54: Breast Cancer 2005;6:240 –246. 2771–2777. 11 Hayes DF, Bast RC, Desch CE et al. Tumor marker utility grading system: 26 Jacobs T W, Gown A M, Yaziji H et al. Specificity of HercepTest in deter- a framework to evaluate clinical utility of tumor ma rkers. J Natl Cancer m ining H ER-2/neu status of breast cancers using the United States Inst 1996;88:1456 –1466. Food and Dr ug Administration-approved scor ing system. J Clin Oncol 1999;17:1983–1987. 12 Stea r ns V, Yamauchi H, Hayes DF. Circulating tumor ma rkers in breast cancer: accepted utilities and novel prospects. Breast Cancer Res Treat 27 Ha r vey JM, Cla rk GM, Osbor ne CK et al. Estrogen receptor status by 1998;52:239–259. im munohistochem istr y is super ior to the ligand-binding assay for pre- 13 McGuire W L, Cla rk GM. Prognostic factors and treatment decisions in dicting response to adjuvant endocr ine therapy in breast cancer. J Clin axillar y-node-negative breast cancer. N Engl J Med 1992;326:1756 –1761. Oncol 1999;17:1474 –1481. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 552 Tumor Marker Pitfalls 28 Alla rd WJ, Matera J, Miller MC et al. Tumor cells circulate in the periph- estrogen receptor-positive prima r y breast cancer: evidence from a phase eral blood of all major ca rcinomas but not in healthy subjects or patients III randomized trial. J Clin Oncol 2001;19:3808 –3816. with nonmalignant diseases. Clin Cancer Res 2004;10:6897– 6904. 40 Muss HB, Thor AD, Ber r y DA et al. c-erbB-2 expression and response to 29 Paik S, Sha k S, Tang G et al. A multigene assay to predict recur rence adjuvant therapy in women with node-positive early breast cancer. N Engl of tamoxifen-treated, node-negative breast cancer. N Engl J Med J Med 1994;330:1260 –1266. 2004;351:2817–2826. 41 Thor AD, Ber r y DA, Budman DR et al. erbB-2, p53, and efficacy of adju- 30 Altman DG, Lausen B, Sauerbrei W et al. Dangers of using “optimal” vant therapy in lymph node-positive breast cancer. J Natl Cancer Inst cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst 1998;90:1346 –1360. 1994;86:829–835. 42 Piccart-Gebhart MJ, Procter M, Leyland-Jones B et al. Trastuzumab after 31 Cr istofanilli M, Budd GT, Ellis MJ et al. Circulating tumor cells, dis- adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med ease progression, and survival in metastatic breast cancer. N Engl J Med 2005;353:1659–1672. 2004;351:781–791. 43 Romond EH, Perez EA, Br yant J et al. Trastuzumab plus adjuvant che- 32 Gelber R D, Bonetti M, Castiglione-Ger tsch M et al. Tailor ing adjuvant motherapy for operable H ER2-positive breast cancer. N Engl J Med treatments for the individual breast cancer patient. Breast 20 03;12: 2005;353:1673–1684. 558–568. 44 Seidman AD, Ber r y D, Cir rincione C et al. CALGB 9840: phase III study 33 Fisher B, Dignam J, Wolma rk N et al. Tamoxifen and chemotherapy for of week ly (W) paclitaxel (P) via 1-hour(h) infusion versus standa rd (S) lymph node-negative, estrogen receptor-positive breast cancer. J Natl 3h infusion ever y third week in the treatment of metastatic breast cancer Cancer Inst 1997;89:1673 –1682. (MBC), with trastuzumab (T) for HER2 positive MBC and randomized for T in HER2 normal MBC. J Clin Oncol 2004;22:512a. 34 Fisher B, Costantino J, Redmond C et al. A random ized clinical tr ial evaluating tamoxifen in the treatment of patients with node-negative 45 Paik S, Shak S, Tang G et al. Expression of the 21 genes in the Recur rence breast cancer who have estrogen-receptor-positive tumors. N Engl J Med Score assay and prediction of clinical benefit from tamoxifen in NSABP 1989;320:479– 484. study B-14 and chemotherapy in NSABP study B-20. Breast Cancer Res Treat 2004;88:24a. 35 McShane LM, Altman DG, Sauerbrei W et al. R Epor ting recommenda- tions for tumour MARKer prognostic studies (R EMARK). Br J Cancer 2005;93:387–391. Additional Reading 36 Slamon DJ, Cla rk GM, Wong SG et al. Human breast cancer: cor relation Clinical practice guidelines for the use of tumor ma rkers in breast and colorec- of relapse and sur vival with amplification of the HER-2/neu oncogene. tal cancer. Adopted on May 17, 1996 by the American Society of Clinical Science 1987;235:177–182. Oncology. J Clin Oncol 1996;14:2843 –2877. 37 Yamauchi H, Stea r ns V, Hayes DF. The role of c-erbB-2 as a predictive Hayes DF, Bast RC, Desch CE et al. Tumor ma rker utility grading system: a factor in breast cancer. Breast Cancer 2001;8:171–183. framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst 1996;88:1456 –1466. 38 Adjuvant! Online breast cancer help files: prognostic estimates. 20 06. https://www.adjuvantonline.com / breathelp0306/ breastindex.html. McShane LM, Altman DG, Sauerbrei W et al. R Epor ting recom menda- Accessed Febr uar y 17, 2006. tions for tumour MARKer prognostic studies (R EMARK). Br J Cancer 2005;93:387–391. 39 Ellis MJ, Coop A, Singh B et al. Letrozole is more effective neoadjuvant endocr ine therapy than tamoxifen for ErbB-1- and /or ErbB-2-positive, The Oncologist http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Oncologist Oxford University Press

Uses and Abuses of Tumor Markers in the Diagnosis, Monitoring, and Treatment of Primary and Metastatic Breast Cancer

Henry, N. Lynn; Hayes, Daniel F.
The Oncologist , Volume 11 (6) – Jun 1, 2006
Download PDF
12 pages

Loading next page...
 
/lp/oxford-university-press/uses-and-abuses-of-tumor-markers-in-the-diagnosis-monitoring-and-vLWnwd06iy

References (46)

Publisher
Oxford University Press
Copyright
Copyright © 2022 Oxford University Press
ISSN
1083-7159
eISSN
1549-490X
DOI
10.1634/theoncologist.11-6-541
Publisher site
See Article on Publisher Site

Abstract

Although breast cancer incidence continues to study designs cloud the issue of how the marker might increase, mortality has been decreasing, principally as be used. Reliance on p-values rather than the size of the a result of earlier detection and improvements in adju- differences in outcome between patients who are posi- vant systemic therapy. Nonetheless, because antineo- tive and those who are negative for the factor obscures plastic agents are associated with substantial morbid- the importance. Technical issues result in poor repro- ity and occasional mortality, efforts to individualize ducibility and interpretability of assays. Analytical treatment strategies are desirable. In addition to clas- issues lead to poorly def ined cutoff values for marker sic histopathologic diagnosis, molecular and cellular levels. Poor patient selection leads to diff iculty inter- tumor markers may help in establishing prognosis or preting results because of confounders such as differ- prediction of benefit. ences in treatment regimens. This review focuses on Recommendations for routine use of tumor mark- these issues, with an emphasis on currently accepted ers in breast cancer have been conservative. Although tumor markers. Finally, new tumor marker report- several studies have been reported, few are of suff i- ing recommendations are discussed, the adoption of ciently high level of evidence to permit solid conclu- which may lead to improved design and publication sions. Three key issues in tumor marker evaluation are of tumor marker studies in the future. The Oncologist utility, magnitude, and reliability. Poorly conceived 2006;11:541–552 Correspondence: Daniel F. Hayes, M.D., University of Michigan Comprehensive Cancer Center, 1500 East Medical Center Drive, Ann Arbor, Michigan 48109, USA. Telephone: 734-615-6725; Fax: 734-615-3947; e-mail: [email protected] Received February 22, 2006; accepted for publication April 11, 2006. ©AlphaMed Press 1083-7159/2006/$20.00/0 The Oncologist 2006;11:541–552 www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 542 Tumor Marker Pitfalls surveillance of patients free of detectable disease. Mea- Introduction surement of CA 15-3 or CA 27.29 and/or CEA was recom- Breast cancer is the most common cancer in women in the mended, however, to monitor selected patients with MBC U.S., with an estimated 213,000 new cases diagnosed in the undergoing palliative therapy [7]. U.S. in 2005 [1]. Despite these increasing numbers, mortal- Routine measurement of multiple tissue markers was also ity from breast cancer continues to decline. This decline discussed in the guidelines. The panel recommended routine is felt to be a result of a combination of earlier detection of measurement of estrogen and progesterone receptors (ER disease as a result of screening and improved treatment of and PgR, respectively) to identify patients most likely to ben- disease with adjuvant systemic therapies [2, 3]. A major- efit from endocrine therapy in either the early or metastatic ity of patients are cured with surgery and radiation therapy disease settings. In addition, measurement of Her-2/neu over- alone, and these patients will gain no additional benefit expression and possibly amplification was recommended for from adjuvant systemic therapies. In addition, having a high all patients at the time of initial diagnosis or recurrence, as it risk for recurrence does not imply that systemic therapy will is predictive of response to trastuzumab (Herceptin ; Genen- prevent it. Even for those who recur, overall survival and tech, South San Francisco, CA), a monoclonal antibody palliation of symptoms for patients with metastatic breast directed against the Her-2/neu receptor [8–10]. The panel felt cancer (MBC) has improved with the advent of new thera- that data to support assessment of other tissue-based mark- pies. However, currently available methods are inadequate ers, including p53, cathepsin D, and f low cytometry-derived to help the clinician precisely predict a priori which patients estimates of DNA content or S phase, were insufficient to will benefit from many of the available therapies. recommend usage in routine clinical practice. For patients with early-stage breast cancer, it would be Therefore, despite the large number of research studies helpful to identify which patients will relapse without adju- evaluating the prognostic and predictive ability of numer- vant systemic therapy, so that only patients who receive ben- ous tumor markers in breast cancer, the ASCO panel rec- efit are exposed to the inherent toxicities. Approach to treat- ommended few for routine use in clinical practice. Why ment of metastatic disease is generally with palliative intent were these recommendations so conservative? In the suc- rather than for cure. In this setting, identification of those ceeding sections of this paper, we outline the multiple fac- patients with rapidly progressive disease permits selection tors that underlie this conservative approach. of more rapidly acting but perhaps more toxic therapy. During the past few decades, with the explosion of When Is a Tumor Marker Useful (Use)? molecular technology and understanding of the biology of When evaluating tumor markers for use in clinical practice, breast cancer, numerous studies have been performed to clinicians should consider their utility, the magnitude of identify prognostic and predictive factors in breast cancer, their effects, and their reliability (Table 2). Tumor mark- with mixed success. Multiple expert panels have convened ers can be useful at multiple stages of cancer diagnosis and to analyze available data in order to establish guidelines for treatment (Table 3) [11, 12]. For example, for individuals the use of tumor markers, but their recommendations have who do not have cancer, a marker may be helpful in deter- been very conservative [4, 5]. In this review, we address mining the risk for developing the disease and/or it may be the pitfalls that have led to difficulties establishing tumor beneficial for screening for disease. Once an abnormality markers for routine clinical use, with a specific focus on is found, a tumor marker may be helpful for distinguishing tumor markers in breast cancer. between benign and malignant processes or between dif- ferent malignant processes. After confirmation of a cancer American Society of Clinical diagnosis, tumor markers can help monitor disease status Oncology Guidelines during and after therapy. The American Society of Clinical Oncology (ASCO) Tumor markers can also help determine prognosis inde- convened a panel of experts that first published recom- pendent of therapy and predict response to therapy. Prognos- mendations regarding the use of circulating and tissue- tic factors ref lect the metastatic potential and/or growth rate based tumor markers in breast cancer in 1996 [6] and most of the tumor and are used to select patient outcomes without recently updated these recommendations in 2001(Table 1) consideration of treatment given [13]. Predictive factors, on [4]. The ASCO panel evaluated multiple serum markers for the other hand, ref lect the sensitivity or resistance of a tumor breast cancer, including assays for MUC1 protein (cancer to a therapeutic agent and therefore are used to predict which antigen [CA] 15-3 and CA 27.29), carcinoembryonic anti- patients are likely to respond to a specific treatment [14]. gen (CEA), and the circulating extracellular domain of Her- Pure prognostic and predictive factors are depicted sche- 2/neu. The panel did not recommend monitoring of any of matically in Figures 1A and 1B, respectively. these markers for screening, diagnosis, staging, or routine The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 543 Table 1. Summary of American Society of Clinical Oncology guidelines for use of tumor markers in breast cancer [6, 7] Assess in Assess in Use for monitoring only adjuvant metastatic Prognostic Predictive Marker setting setting factor factor NED Met Tissue-based ER Yes Yes Weak Strong N/A N/A PgR Yes Yes Weak Strong N/A N/A Her-2/neu Yes Yes Weak Strong N/A N/A Circulating CA 15-3 No Yes No No No Yes CA 27.29 No Yes No No No Yes CEA No Yes No No No Yes Her-2 ECD No No No No No No For benefit from endocrine treatment. For benefit from trastuzumab therapy. Abbreviations: CA, cancer antigen; CEA, carcinoembryonic antigen; ECD, circulating extracellular domain; ER, estrogen receptor; NED, no evidence of disease; Met: metastatic; N/A: not applicable; PgR, progesterone receptor. Table 2. Requirements essential for acceptance of a tumor Few tumor markers are purely prognostic or predictive. marker In fact, most tumor markers have mixed prognostic and Determine utility of marker (Table 3) predictive features, and the utility typically depends on the Evaluate magnitude of effect (none, weak, moderate, strong) therapeutic agent in question. For example, ER expression is weakly favorably prognostic but strongly predictive of Analyze reliability of marker response to treatment with endocrine therapy, as illustrated Technical issues (assay) in Figure 1C. Her-2/neu overexpression, on the other hand, Analytical issues (cutoff points, test/validation sets, multivariate analysis) is an unfavorable prognostic factor and is strongly predic- Trial design issues (appropriate patient population) tive of response to therapy with trastuzumab, as shown in Figure 1D. Until appropriate studies have been performed both in vitro and in vivo, it can be difficult to know how to use a tumor marker appropriately in the clinical setting. In breast cancer, tumor markers are currently used in only a Table 3. Clinical uses of tumor markers few of these categories. Determine risk of developing disease How Useful Is the Tumor Marker Screening for disease (Magnitude)? Once a tumor marker use has been identified, it is impor- Establish diagnosis Differentiate benign versus malignant disease tant to determine the magnitude of the difference in out- Determine type of malignancy comes for that particular use between those who are marker positive and those who are not. By evaluating the difference Determine prognosis in outcome, regardless of treatment, between a patient posi- For primary disease, predict relapse For metastatic disease, predict progression tive for a given prognostic factor and one who is negative for Predict survival the factor, the relative strength of a prognostic factor can be determined [15]. This assessment requires the selection of Predict response to therapy an appropriate outcome of interest, such as improvement Hormone therapy Chemotherapy in symptoms or survival, or surrogates of these end points, Novel therapies (e.g., trastuzumab) such as response rates or progression-free survival. For example, a breast cancer patient with disease in the Monitor disease lymph nodes at the time of diagnosis is two to three times For primary disease, predict relapse For metastatic disease, follow detectable disease more likely to have a breast cancer event (local recurrence Adapted from Stearns V, Yamauchi H, Hayes DF. Circulating or distant metastasis) than a patient without lymph node tumor markers in breast cancer: accepted utilities and novel involvement, regardless of treatment. Since lymph node prospects. Breast Cancer Res Treat 1998;52:239–259, with status has classically been used to make clinical decisions, kind permission of Springer Science and Business Media. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 544 Tumor Marker Pitfalls we have arbitrarily designated it as a “strong” prognostic the “relative predictive value” (RPV), the ratio of the likeli- factor, using it as the gold standard to set the criteria for con- hood that a factor-positive patient will respond to treatment sideration of other, putative markers [16]. A strong prognos- to the likelihood that a factor-negative patient will respond tic factor is depicted by Factor 1 in Figure 1A. Alternatively, to treatment. As with prognostic factors, we have proposed untreated patients with ER-positive breast cancer have only arbitrary classes of prediction factors for breast cancer slightly better outcomes than those with ER-negative dis- therapies based on what has been accepted by consensus, ease, and therefore we designated ER status as a weak prog- in this case ER [19]. Adjuvant tamoxifen therapy has been nostic factor, as portrayed by Factor 2 in Figure 1A [17, 18]. shown to decrease recurrence rates for ER-positive patients In a previous publication, we have suggested hazard ratios by 40% –50%, whereas ER-negative patients obtain mini- of <1.5, 1.5–2, and >2, to distinguish weak, moderate, and mal, if any, benefit from hormonal therapy [20]. Therefore, strong prognostic factors, respectively, for breast cancer the RPV is >8. Similarly, the majority of patients with ER- [16]. Such arbitrary designations would need to be estab- positive MBC have a clinical response to hormonal therapy, lished for other uses, as appropriate. whereas patients with ER-negative disease do not respond Predictive factors can also be classified as weak (Factor [21, 22]. ER status is therefore a strong predictive factor for 1), moderate, or strong (Factor 2), depending on their ability response to hormonal therapy. In this framework, we have to predict response to, and therefore benefit from, a given arbitrarily proposed that, in breast cancer, weak, moderate, therapy, as illustrated in Figure 1B. One measure to permit and strong predictive factors correspond to RPVs of 1–2, comparison of the relative strengths has been designated 2– 4, and >4, respectively [15]. It is important to understand Figure 1. Schematic representation of prognostic and predictive factors. (A): Pure prognostic factor. (B): Pure predictive fac- tor. (C): Mixed factor associated with weakly favorable prognosis and strong response to specific therapy. (D): Mixed factor associated with unfavorable prognosis and strong response to specific therapy. Modified from Hayes DF, Trock B, Harris AL. Assessing the clinical impact of prognostic factors: when is “statistically significant” clinically useful? Breast Cancer Res Treat 1998;52:305–319, with kind permission of Springer Science and Business Media. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 545 the clinical implications of the relative strength of a prog- blotting or reverse transcriptase-polymerase chain reaction nostic and/or predictive factor when integrating this infor- [RT-PCR]), and/or measures of DNA amplification (by mation into routine practice, and to determine if the data f luorescence or chromogenic in situ hybridization [FISH support its use in a specific clinical situation. and CISH, respectively]). Furthermore, even within these categories, different reagents (e.g., different antibodies in How Reliable Is the Tumor Marker IHC assays) are used in different tests. The results are not (Precision and Accuracy)? interchangeable, either within or between classes of assays, The preceding discussion illustrates the importance of esti- and therefore researchers must decide which methodology mating the magnitude of the relative tumor marker effect they will employ. Once that decision is made, researchers for a selected use. However, the marker is only useful if the must then decide how to perform the assay. For example, estimate of its magnitude is reliable and reproducible. In this when assessing Her-2/neu overexpression by IHC, tech- regard, many investigators conclude that their marker of nical issues such as antibody concentration and antigen interest has clinical utility if in their study the difference in retrieval methods may cause unacceptably high false-posi- outcomes between marker “positive” and marker “negative” tive or false-negative rates. patients is less than conventional measures of statistical sig- In one study, IHC and FISH resulted in only a 65% nificant (p < .05). This conclusion may be mistaken. Statisti- agreement for Her-2/neu status [23]. In a different study, cal significance only suggests that in the population chosen results obtained from local laboratories were compared for that study, the differences observed are likely not to be with those from a central laboratory for two Her-2/neu a result of chance alone. It does not imply clinical utility, assays, the HercepTest™ IHC assay (Dako North America, nor does a p-value <0.05 document the validity of the tumor Inc., Carpinteria, CA) and the FISH assay, with 79% con- marker. Although it is important to determine that the dif- cordance for HercepTest™ and 85% concordance for FISH ferences in outcome achieve statistical significance, statisti- [24]. Therefore, for the same test at multiple laboratories, cal significance alone does not determine clinical utility. and for different tests for the same marker, there is a signifi- In addition to determining when to use a tumor marker cant degree of discordance for two commonly used tests for and the magnitude of its effect, it is important to ensure that the evaluation of Her-2/neu status. the technical aspects of the marker are reliable and repro- The stakes are high. Recently reported data suggest ducible and that the study design and conduct are appropri- that adjuvant trastuzumab decreases recurrence rates by ate to test the marker for a clinical use of interest. Several 50%. However, up to 5% of patients who receive trastu- problems with tumor marker studies, including technical, zumab develop cardiac dysfunction, and the cost of 1 year analytical, and trial design issues, have limited the intro- of therapy may exceed $100,000. Therefore, it is essential duction of new prognostic and predictive factors into rou- that Her-2/neu, the target for trastuzumab, be assayed accu- tine clinical practice [11]. rately and precisely for every tissue sample. Expert panels are now being convened to establish guidelines for the con- What Technical Factors Inf luence Measurement duct and interpretation of common tumor marker assays, of Markers? including ER and Her-2/neu. These guidelines should lead From a technical standpoint, difficulties arise because of to standardization of the assays, which should allow for poor sensitivity and/or specificity of the assay for the ana- more reliable results both for routine clinical practice and lyte, poorly reproducible assays, and differences between for use of these assays in clinical trials. assays that use different reagents for measurement of the same marker [11]. Even for the two most commonly used What Analytical Issues Are Important and accepted tumor markers, ER expression and Her-2/neu to Consider? overexpression, standard methodologies have not yet been established [4, 5]. Two primary technical considerations Assay Interpretation are critical when measuring a tumor marker. The first is Determination of assay results can also vary, even for a single which type of assay should be used. The second is the repro- type of assay. For example, with visual assays such as IHC ducibility of the chosen assay, from both a technical and an for ER and Her-2/neu, intra- and interobserver variability analytical perspective. leads to differences in interpretation [25, 26]. Some attempts For example, Her-2/neu status can be determined by have been made to standardize interpretation, such as devel- measures of protein expression (by immunohistochemis- opment of the so-called “Allred score” for semiquantitation try [IHC], Western blotting, or enzyme-linked immuno- of ER expression [27], but these have not been universally sorbent assay), measures of RNA expression (by Northern adopted. Automated and semiautomated systems appear to www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 546 Tumor Marker Pitfalls be highly accurate and are likely to be more reproducible. lar patient cohort from the same study [31]. Another com- Examples of automated systems include the ChromaVision mon method to generate a data-derived cutoff point is to ACIS system (ChromaVision Medical Systems, Inc., San construct a receiver operating characteristic curve, which Juan Capistrano, CA) for ER expression measurement, the demonstrates the tradeoff between the sensitivity and spec- CellSearch™ assay (Veridex, LLC, Warren, NJ) for detec- ificity of an assay at different cutoff points. tion of circulating tumor cells [28], and the Oncotype DX™ Recently, a novel data-derived method to select cutoff assay (Genomic Health, Inc., Redwood City, CA), a new points, designated subpopulation treatment effect pattern prognostic tool for patients with hormone receptor (HR) - plot (STEPP) analysis, has been proposed [32]. STEPP positive, lymph node-negative breast cancer [29]. analysis evaluates outcomes to specific treatments in sub- populations of patients within randomized clinical trials or Cutoff Point Determination meta-analyses [32]. For example, it has been proposed that Regardless of the assay, one has to select some value or level recurrence rates after treatment with chemotherapy should that distinguishes positive from negative results. However, be evaluated in the context of the endocrine responsiveness there is no consensus regarding cor rect methods to establish of tumors, since HR-positive and -negative tumors appear cutoff points, and different studies of the same prognostic or to behave differently. Rather than arbitrarily defining cutoff predictive factor can have widely varying “optimal” cutoff points for ER positive and negative, the authors performed points [30]. a STEPP analysis of data from a previously conducted ran- Cutoff points may be defined using either arbitrary or domized clinical trial and were able to demonstrate a ben- data-derived methods (Table 4). One approach is to con- efit from chemotherapy only in the subset of patients with sider any value greater than two standard deviations above very low ER values. the mean for normal subjects to be positive. Cutoff points can also represent arbitrary values within affected patients; Cutoff Point Validation for example, one might decide that 10%, 50%, or 90% of Regardless of whether cutoff points are chosen arbitrarily affected patients will be classified as “positive.” Others or are data-derived, the selected cutoffs require subsequent have defined cutoff points based on technical factors, such validation. The initial evaluation should be performed using as the limit of detection of the assay [21]. Finally, the cutoff a “test set” of patients. In the second part of the study, the point for a new assay can be defined by comparing it with an utility of the cutoff point should then be confirmed using a older assay [27]. separate “validation set” composed of a similar, but com- Deriving cutoff points based on patient outcome data pletely independent, patient population. may provide more accurate values. For example, the cutoff For example, the Oncotype DX™ assay is based on the point for ER expression was first defined by limits of the principle of evaluating expression of multiple candidate assay and later by determining the optimal level that dis- genes using quantitative RT-PCR [29]. The investigators tinguished those patients who respond to hormonal therapy initially screened more than 20 0 candidate genes with from those who do not. In another example, the cutoff point the aim of developing a test that would predict the likeli- for the CellSearch™ assay for circulating tumor cells was hood of recur rence of cancer in patients with HR-positive, initially determined based on differences in time to pro- lymph node-negative breast cancer. Breast cancer tissues gression of a test set of patients with metastatic disease, and from 447 patients with HR-positive, lymph node-negative this cutoff was then validated with an independent but simi- tumors were used retrospectively to generate an algorithm using 16 of these genes that permitted division of patients into subgroups with ver y low, intermediate, or ver y high Table 4. Methods used for selection of cutoff points risk for recur rence. Patients are assigned to these groups Arbitrary based on a “recur rence score” der ived from the algo- Based on limit of detection of the assay r ithm. The major ity of the test samples were obtained Defined as two standard deviations above the mean of normal patients from patients treated with tamoxifen alone in the National Defined based on mean value in affected versus normal Surgical Adjuvant Breast and Bowel Project (NSABP) patients “Appropriate” percentage of positive cells B-20 trial [33], and the data-derived algorithm was then validated using a separate retrospective cohort of patients Data-derived from the NSABP B-14 trial [34], who had simila r clini- Plot p-value versus outcome cal characteristics and were also treated with tamoxifen Plot magnitude of effect of marker versus outcome Receiver operating characteristic curve alone. If the cohor ts had different clinical characteristics Subpopulation treatment effect pattern plot analysis or had been treated differently, the validation would not The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 547 have been legitimate. Validation of the results in a sepa- clinical trial is a properly powered, prospective, random- rate patient population strongly suggests that the test is ized trial designed specifically to evaluate the clinical reliable and that the results a re likely to be meaningful in a utility of a tumor ma rker in question for a specific and pre- larger population, as long as the patients tested are similar designated use. In such a study, diagnostic and /or thera- to the cohor ts included in the original studies. peutic decisions for the study arm are based on the tumor marker in question, whereas the decisions for the control Statistical Analysis arm are made independently [11]. Of course, statistical analysis is necessary to determine that Systematic overviews and/or pooled analyses of well- the observations are not a result of chance alone. However, conducted LOE II studies are equivalent to LOE I stud- once a tumor marker has been identified and validated, it is ies, especially if the correlative studies address a specific important to determine the relative value of the marker in use but are underpowered in a single study. It has been the context of previously identified prognostic and/or pre- estimated that a clinical trial that has been appropriately dictive factors, such as lymph node status and tumor size, powered to determine a clinical end point, such as progres- using some type of a multivariate analysis. If such an evalu- sion-free survival, is underpowered for analysis of tumor ation is not performed, clinicians will be unable to deter- marker-designated subgroups by one fourth, even if tissue mine the usefulness of incorporating the new marker into from 100% of enrollees is available. Moreover, interpreta- routine clinical practice. tion still requires judgment regarding the clinical impor- tance of the finding. Trial Design and Frameworks for For example, from the prospective randomized clinical Generating, Reporting, and Evaluating trials of adjuvant trastuzumab versus placebo, we anticipate Tumor Marker Results combined analyses of the multiple underpowered LOE II One of the key steps in identifying and confir ming the studies evaluating novel markers for benefit from this drug. benefit of a new tumor marker is appropriate study design. These pooled results should help focus trastuzumab therapy The Tumor Ma rker Utility Grading System (TMUGS) in the subgroups of Her-2/neu-positive patients most likely was initially developed by members of the ASCO panel to benefit. The combined analyses of small LOE III studies to provide a framework within which the utility of tumor that contain patients with variable clinical characteristics ma rkers can be graded based on published information and treatments, however, are more likely useful to generate [11]. Ma rkers a re assigned a grade based on the level of new hypotheses than to provide clinically useful and vali- evidence (LOE) available. These LOE ref lect the relative dated results. quality of the studies used to generate an estimate of the For all studies, whether prospective or retrospective, it effect of the ma rker (Table 5). LOE I and II studies a re is important to identify the appropriate patient population the most beneficial for evaluating the utility of a tumor to be investigated. All patients should have a similar profile marker. According to the TMUGS framework, the ideal based on known prognostic factors. Importantly, the effects Table 5. Levels of evidence for determining utility of tumor marker Level Type of evidence I Evidence from a single, high-powered, prospective controlled study that is specifically designed to test marker, or evidence from well-done meta-analysis and/or overview of level I studies. Ideally, the study is a prospective, randomized controlled trial in which diagnostic and/or therapeutic clinical decisions in one arm are determined at least in part on the basis of marker results, and diagnostic and/or therapeutic clinical decisions in the control arm are made independently of marker results, or Evidence from overview of level of evidence II studies addressing specific use. II Evidence from study in which marker data are determined in relationship to prospective therapeutic trial that is performed to test therapeutic hypothesis but not specifically designed to test marker utility. Specimen collection for marker study and statistical analysis are prospectively determined in protocol as secondary objectives. III Evidence from large but retrospective studies from which variable numbers of samples are available or selected. Statistical analysis for tumor marker was not dictated prospectively at time of therapeutic trial design. IV Evidence from small retrospective studies that do not have prospectively dictated therapy, follow-up, specimen selection, or statistical analysis. VEvidence from small pilot studies designed to determine or estimate distribution of marker levels in sample populations. Reprinted from Hayes DF, Bast RC, Desch CE et al. Tumor marker utility grading system: a framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst 1996;88(20):1456 –1466, permission of Oxford University Press. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 548 Tumor Marker Pitfalls of systemic treatment are critical and must be considered. Real-World Clinical Examples If the study is addressing the prognostic value of a marker, In the preceding sections we identified the essential ele- all patients should have been treated uniformly and without ments for establishing the usefulness, strength, and reli- whatever treatment might be considered if the patients have ability of tumor markers. Let us now discuss the data sup- a “poor prognosis.” If the question is addressing the utility porting two currently used tumor markers, Her-2/neu and of adding any treatment, all patients should be untreated. Oncotype DX™. If the question is whether more treatment should be given, then all patients should have received the same treatment. Her-2/neu If the study is addressing predictive factors, a control The first report of Her-2/neu as a prognostic factor in breast group that has been treated identically to the study group, cancer was published in 1987 [36]. Since then, more than with the exception that they did not receive the treatment 200 papers addressing this topic have been published, with in question, is essential. Although the control group might widely mixed and disparate results [37]. Different authors be from a selected historical control, predictive factors are have concluded that Her-2/neu is associated with poor out- ideally studied in the context of prospective, randomized, comes, no difference in outcome, or even favorable out- controlled trials comparing the patients who received the comes. Indeed, a great deal of this confusion could have treatment in question with those who did not receive that been avoided if the investigators would have addressed the treatment. components described above: (a) What is the intended use, Tumor marker studies should be carefully designed, (b) What is the magnitude of difference between positive using the above criteria, to obtain clinically useful infor- and negative for that use, and (c) How reliable is the esti- mation. Researchers frequently have practical difficulties mate of the magnitude? designing such studies, however, because of the need for sig- The potential uses for Her-2/neu are for prognosis and nificant numbers of patients with particular clinical char- prediction, as outlined in Table 6. Issues related to the reli- acteristics in order to address a specific clinical question. ability of the assays have been described in detail above. As discussed above, this can sometimes be overcome by Overall, studies support that Her-2/neu overexpression is pooling the results of several well-done but underpowered a poor prognostic factor, although its magnitude appears studies. Another significant drawback is obtaining fund- weak. For example, in Adjuvant! Online it has only a rela- ing for tumor marker studies, as pharmaceutical companies tive predictive value of 1.5 [38]. Its role as a prognostic fac- and third-party payers derive relatively smaller financial tor thus remains unclear. benefit from the results compared to the enormous payoffs More data exist to support the role of Her-2/neu status for for a “blockbuster” therapeutic agent. Regardless, given the prediction of response to standa rd therapies and trastuzumab. consequences, one has to question why it is acceptable for For selective estrogen receptor modulators (SERMs), such tumor marker studies to be performed with less scientific as tamoxifen, preclinical and clinical studies suggest that rigor than studies of new pharmaceutical agents. Her-2/neu positivity confers a relative resistance, with mod- Repor ting of tumor marker studies has also been his- erate magnitude, although the data are LOE III at best [39]. tor ically haphaza rd. Recently, in order to standa rdize Data for aromatase inhibitors (AIs) are mixed, although in repor ting of tumor ma rker study results, the National Can- one pilot study of neoadjuvant endocrine therapy, Her-2/neu cer Institute-European Organization for Resea rch and overexpression correlated with lower response to tamoxifen Treatment of Cancer (NCI-EORTC) Work ing Group on than to AIs [39]. At present, Her-2/neu status is not used to Cancer Diagnostics developed R Epor ting recommenda- determine which endocrine therapy to use, because of the tions for tumor MARKer prognostic studies (R EMARK) poor level of available evidence and conf licting data. Con- [35]. T he guidelines outline items that should be firmation of these results may lead to preferential use of AIs addressed by resea rchers when repor ting the results of in patients with Her-2/neu-positive disease. tumor marker studies, including prospectively defining Patients with tumors that overexpress Her-2/neu appear the question the study is tr ying to address, identifying the to have relative resistance to some chemotherapy regimens, appropriate patient population and controls, determining such as cyclophosphamide, methotrexate, and 5-f luoroura- the end point, and identifying potential confounding fac- cil (CMF), but not to others, such as anthracycline-contain- tors. Explicit recommendations a re given rega rding which ing regimens [37, 40, 41]. Her-2/neu status is not generally information must be contained in publications of tumor a consideration when choosing a chemotherapy regimen, ma rker studies, including patient and treatment infor- however, because, as is the case with endocrine therapy, the mation, specimen characteristics, assay methods, study level of available evidence does not support using Her-2/neu design, and statistical analysis methods. status to predict response to chemotherapy. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 549 Table 6. Theoretical uses for tissue-based Her-2/neu assessment Use Effect Magnitude LOE Prognosis Negative Moderate III Prediction Hormone therapy SERM Negative Weak-moderate III AI Neutral N/A III Chemotherapy Nonanthracycline Negative Weak-moderate III Anthracycline Positive or Neutral Moderate or N/A III Trastuzumab Positive Strong II Negative indicates worse outcome if Her-2/neu-positive; positive indicates better outcome if Her-2/neu-positive; neutral indicates outcome same regardless of Her-2/neu result. Abbreviations: AI, aromatase inhibitor; LOE, level of evidence; N/A, not applicable; SERM, selective estrogen receptor modulator. In contrast, Her-2/neu status appears to be strongly pre- Oncotype DX™ has also been evaluated as a predic- dictive of response to trastuzumab. A patient with a tumor tive factor, although less rigorously. In the NSABP B-14 that overexpresses Her-2/neu is usually treated with trastu- trial, the patient cohorts were treated with tamoxifen ver- zumab in either the adjuvant or metastatic settings [8, 9, 42, sus observation, and comparison of the cohorts suggested 43] because the benefits outweigh the risks in the majority of that the Oncotype DX™ assay is predictive for tamoxifen cases. A patient with a tumor that fails to overexpress Her- [45]. Similarly, analysis of NSABP B-20, in which patients 2/neu appears not to respond to treatment with trastuzumab were randomized to CMF and tamoxifen versus tamoxi- [10, 44] and therefore would not be treated with trastuzumab fen alone, permitted the investigators to determine that the to avoid both unnecessary toxicity and cost. Thus, use of assay is predictive for response to chemotherapy [45]. Her-2/neu to select trastuzumab is recommended based on Are the available data sufficient to conclude that Onco- “use” and “magnitude.” However, as discussed above, there type DX™ has been validated to the extent that patient are substantial apparent difficulties with the technical reli- treatment decisions should be based on the results? Perhaps, ability of all available assays for the marker. Nonetheless, but because these studies were all proposed using available despite the shortcomings of Her-2/neu studies outlined samples from trials performed many years ago and repre- above, at present there is sufficient LOE II evidence to sup- sented only subsets of the overall population entered into port the routine clinical use of Her-2/neu overexpression for the trials, concerns have been raised about wholesale clini- selection of trastuzumab therapy, as indicated in the most cal adoption of this assay. In that regard, the North Ameri- recent ASCO tumor marker guidelines [4]. can Breast Cancer Intergroup is developing the TailorRx clinical trial to further validate and extend the Oncotype Multigene Expression: Oncotype DX™ DX™ results. The trial design assumes that the assay is A more recent example of development of a new tumor prognostic, and will confirm the ability of the assay to pre- marker is provided by the case of Oncotype DX™. The dict response to chemotherapy. investigators specifically developed the test to determine In the TailorR X trial, tumors of patients with ER-posi- prognosis in ER-positive, lymph node-negative patients tive and lymph node-negative breast cancer will be tested who were treated with tamoxifen [29]. The available results using the Oncotype DX™ assay. Patients with low recur- suggest that, in this group of patients, Oncotype DX™ is rence scores, who have good prognoses without chemo- a strong prognostic factor because the ratio of the hazard therapy, will receive hormonal therapy alone. At the other ratios of the high and low recurrence score cohorts is >2 end of the spectr um, patients with high recur rence scores in both the test and validation cohorts [29]. The Oncotype will receive chemotherapy in addition to hormonal ther- DX™ test is also reliable because it fulfills the criteria out- apy. Those patients whose scores fall in the intermediate lined above for technical, analytical, and trial design issues range will all receive hormonal therapy and be randomly (Table 2). The assay is reproducible and was validated in assigned to chemotherapy or not. This trial design will independent test and validation cohorts of patients, as permit validation of the Oncotype DX™ results in a simi- described above. lar patient population in a large prospective clinical trial, www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 550 Tumor Marker Pitfalls and will allow for generation of new data on which to base studies to date, however, very few markers have been t reatment recom mendations for patients whose recur rence accepted for routine use by groups such as ASCO. When scores are intermediate. designing studies to establish a new tumor marker, or new Given the substantial technical, analytical, and trial use for an old marker, it is important to address the utility, design problems with previously performed tumor marker magnitude, and reliability of the marker (Table 2). studies, it is imperative to address these issues. It is espe- Frameworks such as TMUGS can be useful when cially important to standardize the assays for commonly designing and conducting these studies to ensure that used tumor markers. Otherwise, patients with false-posi- appropriate components are included, thereby leading to tive test results for predictive factors will receive treatments the establishment of new tumor markers for routine clinical that are not beneficial but which may cause significant use [11]. By progressively generating and refining a hypoth- toxicity, and those with false-negative results will not be esis, based on data derived from increasingly well-devel- offered potentially life-saving therapies. In addition, a bet- oped studies, tumor markers with clinical utility can be ter understanding of the potential pitfalls in tumor marker identified (Fig. 2). In addition, the new REMARK guide- study design will allow for the development of new, poten- lines should promote better design and conduct of stud- tially more useful assays. ies specifically focused on tumor marker validation [35]. Implementation of these recommendations when designing Conclusions tumor marker studies will result in the generation and pub- Tumor markers, when well defined, can play a significant lication of appropriate and complete clinical data, leading role in prediction and prognosis for breast cancer patients. to the adoption of new, well-validated tumor markers for Because of the abundance of poorly designed tumor marker routine clinical use. Figure 2. Tumor marker development f low chart. LOE, level of evidence; MVA, multivariate analysis. The Oncologist Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 Henry, Hayes 551 funding from Immunicon. D.F.H. has been a consultant / Authors’ Note advisor y panel pa r ticipant or held lecture / honora r ium Suppor ted by National Institutes of Health grant position dur ing past yea r for Dendreon, Im municon, 5R01CA092461-03 and Fashion Footwear Charitable Novar tis, Pfizer, Precision Therapeutics, Inc., Oncotech, Foundation of New York /QVC Presents Shoes-on-Sale™. and Veridex. D.F.H. was the principle or coinvestigator during the past yea r on studies funded by Immunicon, Disclosures Wyeth Ayerst-Genetics Institute, Pfizer, and Novar tis. D.F.H. has served as an unpaid consultant for Genomic Health, Inc. in the past yea r and has received resea rch 14 Gaspa r ini G, Pozza F, Ha r r is A L. Evaluating the potential usefulness of References new prognostic and predictive indicators in node-negative breast cancer 1 American Cancer Society. Cancer Facts and Figures, 2005. Available at patients. J Natl Cancer Inst 1993;85:1206 –1219. http://www.cancer.org/downloads/STT/CAFF2005f4PWSecured.pdf. 15 Hayes DF, Trock B, Har ris AL. Assessing the clinical impact of prognos- Accessed March 10, 2005. tic factors: when is “statistically significant” clinically useful? Breast 2 Peto R, Boreham J, Cla rke M et al. UK and USA breast cancer deaths down Cancer Res Treat 1998;52:305 –319. 25% in year 2000 at ages 20 -69 years. Lancet 2000;355:1822. 16 Isaacs C, Stea r ns V, Hayes DF. New prognostic factors for breast cancer 3 Ber r y DA, Cronin K A, Plevr itis SK et al. Effect of screening and adju- recur rence. Semin Oncol 2001;28:53 – 67. vant therapy on mor tality from breast cancer. N Engl J Med 20 05;353: 17 Fisher B, Costantino JP, Wickerham DL et al. Tamoxifen for prevention of 1784 –1792. breast cancer: repor t of the National Surgical Adjuvant Breast and Bowel 4 Bast RC Jr, Ravdin P, Hayes DF et al. 2000 update of recommendations for Project P-1 Study. J Natl Cancer Inst 1998;90:1371–1388. the use of tumor markers in breast and colorectal cancer: clinical practice 18 K night WA, Livingston R B, Gregor y EJ et al. Estrogen receptor as an guidelines of the Amer ican Society of Clinical Oncology. J Clin Oncol independent prognostic factor for early recur rence in breast cancer. Can- 2001;19:1865–1878. cer Res 1977;37:4669 – 4671. 5 Hammond ME, Fitzgibbons PL, Compton CC et al. College of American 19 Hayes DF. Do we need prognostic factors in nodal-negative breast cancer? Pathologists Conference XXXV: solid tumor prognostic factors-which, Arbiter. Eur J Cancer 2000;36:302–306. how and so what? Summa r y document and recommendations for imple- mentation. Arch Pathol Lab Med 2000;124:958 –965. 20 Ea rly Breast Cancer Tr ialists’ Collaborative Group (EBCTG). Effects of chemotherapy and hor monal therapy for ea rly breast cancer on recur- 6 Clinical practice guidelines for the use of tumor ma rkers in breast and rence and 15-yea r survival: an overview of the randomised tr ials. Lancet colorectal cancer. Adopted on May 17, 1996 by the American Society of 2005;365:1687–1717. Clinical Oncology. J Clin Oncol 1996;14:2843 –2877. 21 McGuire W L, Cham ness GC, Costlow ME et al. Steroids and human 7 Bast RC Jr, Ravdin P, Hayes DF et al. Er rata: 20 0 0 update of recommenda- breast cancer. J Steroid Biochem 1975;6:723 –727. tions for the use of tumor markers in breast and colorectal cancer: clinical practice guidelines of the American Society of Clinical Oncology. J Clin 22 Allegra JC, Lippman ME, Thompson EB et al. Estrogen receptor status: Oncol 2001;19:4185 – 4188. an impor tant variable in predicting response to endocrine therapy in met- astatic breast cancer. Eur J Cancer 1980;16:323 –331. 8 Cobleigh MA, Vogel CL, Tr ipathy D et al. Multinational study of the efficacy and safety of humanized anti-HER2 monoclonal antibody in 23 Dressler LG, Ber r y DA, Broadwater G et al. Compa rison of HER2 status women who have HER2-overexpressing metastatic breast cancer that by f luorescence in situ hybridization and immunohistochemistr y to pre- has progressed after chemotherapy for metastatic disease. J Clin Oncol dict benefit from dose escalation of adjuvant doxor ubicin-based therapy 1999;17:2639–2648. in node-positive breast cancer patients. J Clin Oncol 2005;23:4287– 4297. 9 Slamon DJ, Leyland-Jones B, Sha k S et al. Use of chemotherapy plus a 24 Perez EA, Suman VJ, Davidson N E et al. HER2 testing by local, central, monoclonal antibody against H ER2 for metastatic breast cancer that and reference laborator ies in the NCCTG N9831 Intergroup Adjuvant overexpresses HER2. N Engl J Med 2001;344:783 –792. Trial. J Clin Oncol 2004;22:567a. 10 Mass R D, Press MF, Anderson S et al. Evaluation of clinical outcomes 25 Press MF, Hung G, Godolphin W et al. Sensitivity of HER-2/neu anti- according to H ER2 detection by f luorescence in situ hybr idization in bodies in a rchival tissue samples: potential source of er ror in immuno- women with metastatic breast cancer treated with trastuzumab. Clin histochem ical studies of oncogene expression. Cancer Res 1994;54: Breast Cancer 2005;6:240 –246. 2771–2777. 11 Hayes DF, Bast RC, Desch CE et al. Tumor marker utility grading system: 26 Jacobs T W, Gown A M, Yaziji H et al. Specificity of HercepTest in deter- a framework to evaluate clinical utility of tumor ma rkers. J Natl Cancer m ining H ER-2/neu status of breast cancers using the United States Inst 1996;88:1456 –1466. Food and Dr ug Administration-approved scor ing system. J Clin Oncol 1999;17:1983–1987. 12 Stea r ns V, Yamauchi H, Hayes DF. Circulating tumor ma rkers in breast cancer: accepted utilities and novel prospects. Breast Cancer Res Treat 27 Ha r vey JM, Cla rk GM, Osbor ne CK et al. Estrogen receptor status by 1998;52:239–259. im munohistochem istr y is super ior to the ligand-binding assay for pre- 13 McGuire W L, Cla rk GM. Prognostic factors and treatment decisions in dicting response to adjuvant endocr ine therapy in breast cancer. J Clin axillar y-node-negative breast cancer. N Engl J Med 1992;326:1756 –1761. Oncol 1999;17:1474 –1481. www.TheOncologist.com Downloaded from https://academic.oup.com/oncolo/article/11/6/541/6398199 by DeepDyve user on 31 January 2022 552 Tumor Marker Pitfalls 28 Alla rd WJ, Matera J, Miller MC et al. Tumor cells circulate in the periph- estrogen receptor-positive prima r y breast cancer: evidence from a phase eral blood of all major ca rcinomas but not in healthy subjects or patients III randomized trial. J Clin Oncol 2001;19:3808 –3816. with nonmalignant diseases. Clin Cancer Res 2004;10:6897– 6904. 40 Muss HB, Thor AD, Ber r y DA et al. c-erbB-2 expression and response to 29 Paik S, Sha k S, Tang G et al. A multigene assay to predict recur rence adjuvant therapy in women with node-positive early breast cancer. N Engl of tamoxifen-treated, node-negative breast cancer. N Engl J Med J Med 1994;330:1260 –1266. 2004;351:2817–2826. 41 Thor AD, Ber r y DA, Budman DR et al. erbB-2, p53, and efficacy of adju- 30 Altman DG, Lausen B, Sauerbrei W et al. Dangers of using “optimal” vant therapy in lymph node-positive breast cancer. J Natl Cancer Inst cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst 1998;90:1346 –1360. 1994;86:829–835. 42 Piccart-Gebhart MJ, Procter M, Leyland-Jones B et al. Trastuzumab after 31 Cr istofanilli M, Budd GT, Ellis MJ et al. Circulating tumor cells, dis- adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med ease progression, and survival in metastatic breast cancer. N Engl J Med 2005;353:1659–1672. 2004;351:781–791. 43 Romond EH, Perez EA, Br yant J et al. Trastuzumab plus adjuvant che- 32 Gelber R D, Bonetti M, Castiglione-Ger tsch M et al. Tailor ing adjuvant motherapy for operable H ER2-positive breast cancer. N Engl J Med treatments for the individual breast cancer patient. Breast 20 03;12: 2005;353:1673–1684. 558–568. 44 Seidman AD, Ber r y D, Cir rincione C et al. CALGB 9840: phase III study 33 Fisher B, Dignam J, Wolma rk N et al. Tamoxifen and chemotherapy for of week ly (W) paclitaxel (P) via 1-hour(h) infusion versus standa rd (S) lymph node-negative, estrogen receptor-positive breast cancer. J Natl 3h infusion ever y third week in the treatment of metastatic breast cancer Cancer Inst 1997;89:1673 –1682. (MBC), with trastuzumab (T) for HER2 positive MBC and randomized for T in HER2 normal MBC. J Clin Oncol 2004;22:512a. 34 Fisher B, Costantino J, Redmond C et al. A random ized clinical tr ial evaluating tamoxifen in the treatment of patients with node-negative 45 Paik S, Shak S, Tang G et al. Expression of the 21 genes in the Recur rence breast cancer who have estrogen-receptor-positive tumors. N Engl J Med Score assay and prediction of clinical benefit from tamoxifen in NSABP 1989;320:479– 484. study B-14 and chemotherapy in NSABP study B-20. Breast Cancer Res Treat 2004;88:24a. 35 McShane LM, Altman DG, Sauerbrei W et al. R Epor ting recommenda- tions for tumour MARKer prognostic studies (R EMARK). Br J Cancer 2005;93:387–391. Additional Reading 36 Slamon DJ, Cla rk GM, Wong SG et al. Human breast cancer: cor relation Clinical practice guidelines for the use of tumor ma rkers in breast and colorec- of relapse and sur vival with amplification of the HER-2/neu oncogene. tal cancer. Adopted on May 17, 1996 by the American Society of Clinical Science 1987;235:177–182. Oncology. J Clin Oncol 1996;14:2843 –2877. 37 Yamauchi H, Stea r ns V, Hayes DF. The role of c-erbB-2 as a predictive Hayes DF, Bast RC, Desch CE et al. Tumor ma rker utility grading system: a factor in breast cancer. Breast Cancer 2001;8:171–183. framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst 1996;88:1456 –1466. 38 Adjuvant! Online breast cancer help files: prognostic estimates. 20 06. https://www.adjuvantonline.com / breathelp0306/ breastindex.html. McShane LM, Altman DG, Sauerbrei W et al. R Epor ting recom menda- Accessed Febr uar y 17, 2006. tions for tumour MARKer prognostic studies (R EMARK). Br J Cancer 2005;93:387–391. 39 Ellis MJ, Coop A, Singh B et al. Letrozole is more effective neoadjuvant endocr ine therapy than tamoxifen for ErbB-1- and /or ErbB-2-positive, The Oncologist

Journal

The Oncologist – Oxford University Press

Published: Jun 1, 2006

There are no references for this article.