Predicting chemosensitivity and resistance in breast cancer.

Introduction

Over the last three decades, significant improvements in outcome for patients with breast cancer have been achieved by a combination of adjuvant chemotherapy, improved surgery and screening [1,2]. This progress has been matched by a rapid expansion of our understanding of the molecular biology of breast cancer and of the mechanisms underlying tumour progression, metastasis and response to therapy. It is now widely accepted that many patients who receive systemic therapy for breast cancer do not require such treatment. It is equally apparent that the wide biological diversity of breast cancer (for example see [3,4]) underpins not only potential differences in the natural history of the disease but also differences in response to current adjuvant chemotherapy regimens [5]. Indeed extensive and growing evidence suggests that the molecular pathology of breast cancer impacts on both prognosis [6,7] and response to chemotherapy [8-12]. This evidence underpins the belief, increasingly widely held, that a 'one size fits all' approach to selection of combination chemotherapy regimens for early breast cancer is no longer in the best interest of patients and may indeed be restricting progress towards optimising treatment of breast cancer. Thus the search for optimal treatments must in future include the expectation that this optimum is different for different patient subgroups and will be defined by molecular predictive markers for response to specific chemotherapeutic and biological agents. As a result, a key objective of research into predictive biomarkers is the identification of markers which identify the optimal type of chemotherapy regimen and linked biological therapies. This research priority was in fact the second most important goal, identified in a recent multinational focus group document aimed at identification of the top ten most important research questions in breast cancer [13].

Predictive biomarkers for breast cancer

The use of molecular predictive markers has already demonstrated significant potential to deliver improvements in outcome for high-risk molecularly defined subgroups of breast cancers. Trastuzumab, targeted to human epidermal growth factor receptor (HER2)-positive breast cancers, has demonstrated one of the most dramatic improvements in patient outcome following treatment seen for many years in breast cancer [14,15]. The challenge is to build upon this platform and to rapidly extend the ability to personalise cancer treatment by linking molecular subgroups of breast cancer to the optimal treatment regimen, particularly in the context of novel molecularly targeted therapies. For such agents, a link between a diagnostic measure of a particular feature of tumour biology and a molecular agent targeting specific pathways is predicted to be the norm.

However, despite the attraction of focusing on target development we should not overlook the potential for molecular predictive markers in the context of 'conventional' chemotherapy. This area is also likely to provide significant potential for improvement in patient outcome in the near future. Two separate approaches are currently applied to the development of predictive biomarkers: the first, based on current understanding of cancer biology and drug action, is hypothesis-led with molecular targets selected from functional biology approaches; the second is based on more exploratory analyses of molecular features of tumours, often using array-based analyses, which may predict for outcome even when the biology of such profiles with respect to the drug in question is unclear. Both approaches appear to offer potential, and for the first time it is now possible to envisage testing this potential in the context of large Phase III clinical trials. This is largely due to the progress made over the last 5 years in improving support for the banking of clinical materials from such trials for either retrospective or prospective analyses of predictive markers. Many trials, including in the UK Taxotere as adjuvant chemotherapy trial (TACT), TACT2, Paclitaxel, Anthracycline, Gemcitabine and Cyclophosphamide (tAnGo), Tamoxifen versus Exemestane Adjuvant Multicentre (TEAM), Selective Use of Postoperative radiotherapy AftER MastectOmy (SUPREMO) and the Breast International Group (BIG)-1-98 and Microarray In Node negative Disease may Avoid ChemoTherapy (MINDACT) trials have included prospective tissue collection while others including Arimidex, Tamoxifen, Alone or in Combination (ATAC) and National Epirubicin Adjuvant Trial (NEAT) studies have made significant efforts to retrospectively access tumour samples. Such biobanks represent a novel and unique opportunity for the robust testing of previously established and novel predictive markers in carefully controlled experimental settings with high-quality outcome data and characterisation of tissue markers.

Biomarker analyses have to date been restricted by an unstructured approach [16,17], but the publication of specific guidelines for the analysis of biomarkers [18,19] is being followed by publication of guidelines for sample collection. Jointly, such guidelines will provide impetus to improvements in biomarker research. Ultimately, the combination of a structured and regulated approach to biomarker development with increasing understanding of the potential role of candidate predictive biomarkers should lead to robust prospective biomarker-driven clinical trials in breast cancer in the relatively near future.

Should biomarkers be included in future Phase III trials?

In the majority of recent trials of adjuvant therapy in breast cancer, relatively modest improvements in disease-free and overall survival have been observed (for example see [20-22]). In the UK NEAT trial, a small proportion of patients experienced improved outcome following the addition of an anthracycline [epirubicin (E)] prior to cyclophosphamide, methotrexate and 5-fluorouracil (CMF) treatment (E-CMF) [22]. The majority, however, experienced sufficient disease control following surgery, radiotherapy and/or endocrine therapy (if indicated) and CMF alone. A significant proportion of tumours remained refractory to either CMF or E-CMF as evidenced by early relapse in these patients. Similar observations apply to other novel adjuvant treatments such as aromatase inhibitors and even trastuzumab [14], where for the majority of patients added benefit, measured by improved outcome, cannot be clearly substantiated. Added treatment may have been either unnecessary, with sufficient disease control being obtained by conventional treatments, or ineffective, as the disease relapsed despite additional treatment. The conclusion drawn from this is that, while the majority of recent trials clearly demonstrate improved outcome in the trial population treated with novel regimens when observed as a whole, benefit from current advances in breast cancer adjuvant therapies are restricted to subgroups of breast cancer patients. The hypothesis arising from this is that improved targeting of current therapies, based on predictive markers, would allow the selection of patients for whom different therapeutic approaches (e.g. taxanes versus anthracyclines) would provide a greater probability of benefit. The key question arising from such a hypothesis is whether such markers are currently available for incorporation into future trial designs? As a brief illustration of the challenges involved, we will explore and review two potential approaches using hypothesis-led and prospective profiling techniques.

Topoisomerase II[alpha] and HER2 as predictive markers for anthracyclines

Over the last decade, evidence has been emerging to suggest that either HER2 gene amplification or amplification/deletion of the topoisomerase II[alpha] (TII[alpha]) gene are predictive of sensitivity to anthracyclines (reviewed by Piccart-Gebhart [10]). The results of these studies are not clear-cut. Whereas data from one study [23] suggest that TII[alpha] gene amplification or deletion is a predictor of anthracycline response, another study provides evidence supporting the role of HER2 in this context [24]. The water is muddied by the close proximity of the TII[alpha] gene to HER2 on chromosome 17q12-21. Amplification of HER2 is associated with TII[alpha] gene aberrations [25], making it difficult to be clear which gene is more closely associated with anthracycline sensitivity. Data from the BR9601 trial [22] suggest that HER1 and HER3 provide additional information when analysed with HER2, building on the hypothesis that these type I receptor tyrosine kinases are associated with pathway activation and a common resistance phenotype [26-28]. The advantage of this hypothesis-driven approach is that there is now sufficient evidence to predict the frequency of TII[alpha], HER2, epidermal growth factor receptor (EGFR)/HER1 and HER3 expression or genetic abnormalities in future trial populations based on significant pre-existing data [23,24,26-29]. Furthermore, using estimates from existing datasets, the hazard ratios can be provided for future trials [23,24]. These form two strands of evidence for the appropriate design of any future biomarker-led clinical trial. However, the disparity between effects observed and the lack of clear data to support biomarker-driven patient stratification suggests that additional data, perhaps from larger trials or meta-analyses of existing trials, are required.

Gene signatures for taxanes/anthracyclines

Recently, Potti et al. described drug-specific gene signatures for a number of common chemotherapy agents including anthracyclines and taxanes [30] using the National Cancer Institute (NCI) 60-cell line panel. Within the same study, they performed computer analysis of expression profiles derived from breast cancers treated with neoadjuvant chemotherapy including taxanes and confirmed the predictive value of these gene expression profiles [30]. Recently, Bonnefoi et al. [31] have taken this approach to the next stage; using breast cancers from a randomised trial of neoadjuvant therapy they confirmed that these gene signatures could predict response to neoadjuvant taxane therapy with 93% sensitivity and to neoadjuvant 5-fluorouracil, epirubicin, cyclophosphamide (FEC) with 96% sensitivity [31]. Most excitingly, they predicted a marked increase in response rate, if these profiles had been applied prospectively, from ~45% (observed) to ~70% (predicted) [31]. This suggests that implementation of predictive profiles, if successful, might provide one of the most marked improvements in outcome following chemotherapy that has been observed in recent years.

While current predictive molecular profiles have been developed in the neoadjuvant setting, due to the relative ease of handling large datasets from smaller sample sizes and the availability of response assessment at surgery, their application can be tested in a post-operative adjuvant-treated cohort once appropriate technology has been developed.

These recent data [30,31] suggest that molecular signatures for anthracylines and taxanes may be available for testing in the adjuvant, neoadjuvant or pre-operative setting, possibly within a future Phase III clinical trial. Currently these profiles have a high sensitivity (>90%) but low overall specificity (<70%). However, they are strongly predictive for non-response with negative predictive values of >90% [31]. What is lacking is information on the frequency of different response profiles in the adjuvant trial population. Therefore, whereas is it currently possible to estimate hazard ratios for response from existing data, information on the potential stratification of patients into different arms of any future trial is lacking.

Steps required to develop biomarkers for future trials

There is growing acceptance that biomarkers targeted at selection of specific chemotherapy agents have moved beyond the 'proof of concept' stage. Clearly, further development is required, but perhaps we are closer to implementation of prospective biomarker analysis in Phase III trials than is currently accepted. For the implementation of HER2, TII[alpha] or similar biomarker panels, further data, which are currently being gathered in large, adequately powered studies, are awaited. Studies embedded within the UK NEAT trial will address the predictive value of HER1-3 and TII[alpha], and could be complemented by future analysis of the control arms of the UK TACT trial. An ongoing meta-analysis, combining data from UK, Canadian, Belgian and Danish [22-24, 32] studies, will clearly add value and may provide information on potential heterogeneity between different treatment regimens used in these studies. These reports should emerge within the next 12 months. Once such data are available, robust, quality-assured assays for the appropriate markers could be applied in future trials. Lessons learnt during the development of diagnostic HER2 testing could be readily applied to these approaches [33,34].

For expression-profiling approaches, novel evidence suggests these may be emerging as potential diagnostic tools. Further data are clearly required on the population distribution of such profiles before trials can be designed. The question remains, however, as to whether expression arrays provide the optimal approach for future molecular diagnostics. Platforms using RT-PCR-based analyses, even for multigene profiling (as with expression arrays) have been reported to perform more consistently and accurately (Oncotype Dx [35]) than expression profiles. Such RT-PCR approaches are also validated and applicable to routinely formalin-fixed, paraffin-embedded tissues, whereas most current expression array approaches are not [35,36]. RT-PCR approaches, therefore, offer an attractive and potentially robust platform for molecular diagnostics within a clinical diagnostic environment.

A number of studies suggest that quantitative RT-PCR-based analysis of formalin-fixed material may provide information regarding chemotherapy response [5,9,37]. The application of existing multigene profiles to paraffin-embedded tissues, even those stored for some time [5,9], suggests that further chemotherapy regimen-specific profiles may equally readily be applied to formalin-fixed tissues. This, coupled with the data from Bonnefoi et al. identifying novel taxane-specific gene predictors [31], suggests that the development of a novel predictive expression profiling tool for specific chemotherapy agents is achievable.

Future trial design: use of biomarkers

Clearly the design of future biomarker-stratified trials will be challenging. Such challenges are, however, being addressed in the context of ongoing studies such as the MINDACT trial. This trial has sought to address a highly complex strategy of prognostic profiling, with the intention of sparing patients unnecessary chemotherapy, and biomarker profiling using expression array analysis at a central facility. Such approaches are complex logistically and require significant support. Future trials can build on these paradigms using more portable assay systems and simplifying trial analysis. For future trials, the application of existing and developing technologies for the analysis of complex biomarker panels, with attention to appropriate quality measures [17,18], should allow such assays to be applied to formalin-fixed, paraffin-embedded tissues. Such biomarker assays must be robust and reproducible and probably decentralised to some extent. Logistics relating to sample transit in a short time window prior to treatment initiation will need to be addressed. Currently it is possible that either 'conventional' histopathology, in situ assays or simplified expression profiles may be the lead candidates for such assays.

For appropriate studies to be designed and correctly interpreted, statistical approaches that are no less rigorous, and potentially more complex, than those currently employed will be required. Treatment interactions with biomarkers, where treatment response may differ between biomarker-positive and biomarker-negative cases, require multivariate analyses which increase the required sample size. Agreed analytical procedures, which address the potential for interactions between biomarkers and treatments and the potential for such interactions to change over time will need to be determined for future biomarker analyses.

Clearly, with the lead time for such trials remaining uncertain, it is difficult to predict the specific question that may be addressed in such a trial. At present there is clear potential for questions relating to either taxane or anthracycline use to be explored. As time passes, pressure will increase for combination of such approaches with novel agents. It is also unclear whether the potential size of such studies will be reduced as a result of increased predicted benefits from targeting therapies, or increased because of the complexity of biomarker-treatment interactions. However, it would appear that the UK is ideally placed to pursue such approaches because of existing experience with translational biomarker analysis within Phase III trials. This process should be greatly supported by the development of a UK Intergroup which is likely to play a key role in facilitating such studies in the future.

Conclusions

Based on novel developments in the field of predictive biomarkers, it appears that such biomarkers are 'on the cusp' of being available for treatment stratification with respect to response to 'conventional' chemotherapies in future Phase III trials. Significant advances in technologies applicable to formalin-fixed tissues are greatly facilitating this process. Further developments in the technology should ensure that robust and clinically relevant biomarker assays are in place in the near future. Some further data gathering is required, particularly focused on key questions (for example, predicting anthracycline or taxane response). The design of future Phase III trials, incorporating biomarker analyses, will require multidisciplinary input, with focus on biostatistics and molecular pathology issues, especially with respect to quality assurance and sample size. However, the strongest drive in this area remains the need to optimise delivery of both conventional and novel therapeutics to improve patient care.

References

[1.] Peto R. Update of the worldwide evidence on the adjuvant treatment of breast cancer. Eur J Cancer, 2002, 38, E11.

[2.] Abe O, Abe R, Enomoto K et al. Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet, 2005, 365, 1687-1717.

[3.] Perou CM, Jeffrey SS, Van de Rijn M et al. Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A, 1999, 96, 9212-9217.

[4.] Perou CM, Sorlie T, Eisen MB et al. Molecular portraits of human breast tumours. Nature, 2000, 406, 747-752.

[5.] Gianni L, Zambetti M, Clark K et al. Gene expression profiles in paraffin-embedded core biopsy tissue predict response to chemotherapy in women with locally advanced breast cancer. J Clin Oncol, 2005, 23, 7265-7277.

[6.] El Rehim DMA, Ball G, Pinder S and Ellis IO. Molecular classification of breast carcinoma based on the protein expression immunoprofiles. J Pathol, 2004, 204, 3A.

[7.] Ellis I. Immunocytochemistry in breast pathology. In: Theory and Practice of Histological Techniques (Bancroft JD and Gamble M, eds), Churchill Livingstone, London, 2002, pp499-516.

[8.] Mina L, Soule S, Badve S et al. Predicting response to primary chemotherapy: gene expression profiling of paraffin-embedded core biopsy tissue. Breast Cancer Res Treat, 2007, 103, 197-208.

[9.] Paik S, Tang G, Shak S et al. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol, 2006, 24, 3726-3734.

[10.] Piccart-Gebhart MJ. Anthracyclines and the tailoring of treatment for early breast cancer. N Engl J Med, 2006, 354, 2177-2179.

[11.] Gianni L, Zambetti M, Clark K et al. Gene expression profiles in paraffin-embedded core biopsy tissue predict response to chemotherapy in women with locally advanced breast cancer. J Clin Oncol, 2005, 23, 7265-7277.

[12.] Di Leo A, Chan S, Paesmans M et al. HER-2/neu as a predictive marker in a population of advanced breast cancer patients randomly treated either with single-agent doxorubicin or single-agent docetaxel. Breast Cancer Res Treat, 2004, 86, 197-206.

[13.] A global consensus. http://www.toptenresearch.org/ (accessed 20 November 2007).

[14.] Piccart-Gebhart MJ, Procter M, Leyland-Jones B et al. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med, 2005, 353, 1659-1672.

[15.] Romond EH, Perez EA, Bryant J et al. Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med, 2005, 353, 1673-1684.

[16.] Bartlett JMS. Translational research and the development of novel biomakers in breast cancer. Advances in Breast Cancer, 2006, 3, 76-81.

[17.] Bartlett JMS. Biological predictors of tumour response: natural selection or intelligent design? Advances in Breast Cancer, 2005, 2, 11-15.

[18.] Hammond MEH and Taube SE. Issues and barriers to development of clinically useful tumor markers: a development pathway proposal. Semin Oncol, 2002, 29, 213-221.

[19.] McShane LM, Altman DG, Sauerbrei W et al. REporting recommendations for tumour MARKer prognostic studies (REMARK). Br J Cancer, 2006, 93, 387-391.

[20.] Howell A, Cuzick J, Baum M et al. Results of the ATAC (Arimidex, Tamoxifen, Alone or in Combination) trial after completion of 5 years' adjuvant treatment for breast cancer. Lancet, 2005, 365, 60-62.

[21.] Coombes RC, Hall E, Gibson LJ et al. A randomized trial of exemestane after two to three years of tamoxifen therapy in postmenopausal women with primary breast cancer. N Engl J Med, 2004, 350, 1081-1092.

[22.] Poole CJ, Earl HM, Hiller L et al. Epirubicin and cyclophosphamide, methotrexate, and fluorouracil as adjuvant therapy for early breast cancer. N Engl J Med, 2006, 355, 1851-1862.

[23.] Knoop AS, Knudsen H, Balslev E et al. Retrospective analysis of topoismerase IIa amplifications and deletions as predictive markers in primary breast cancer patients randomly assigned to cyclophosphamide, methotrexate, and fluorouracil or cyclophosphamide, epirubicin and fluorouracil: Danish breast cancer cooperative group. J Clin Oncol, 2005, 23, 7483-7490.

[24.] Pritchard KI, Shepherd LE, O'Malley FP et al. HER2 and responsiveness of breast cancer to adjuvant chemotherapy. N Engl J Med, 2006, 354, 2103-2111.

[25.] Tanner M, Jarvinen P and Isola J. Amplification of HER-2/neu and topoisomerase II alpha in primary and metastatic breast cancer. Cancer Res, 2001, 61, 5345-5348.

[26.] Tovey SM, Dunne B, Witton CJ et al. Can molecular markers predict when to implement treatment with aromatase inhibitors in invasive breast cancer? Clin Cancer Res, 2005, 11, 4835-4842.

[27.] Tovey SM, Witton CJ, Bartlett JMS et al. Outcome and human epidermal growth factor receptor (HER) 1-4 status in invasive breast carcinomas with proliferation indices evaluated by bromodeoxyuridine labelling. Breast Cancer Res, 2004, 6, R246-R251.

[28.] Witton CJ, Reeves JR, Going JJ et al. Expression of the HERI-4 family of receptor tyrosine kinases in breast cancer. J Pathol, 2003, 200, 290-297.

[29.] Di Leo A, Chan S, Paesmans M et al. HER-2/neu as a predictive marker in a population of advanced breast cancer patients randomly treated either with single-agent doxorubicin or single-agent docetaxel. Breast Cancer Res Treat, 2004, 86, 197-206.

[30.] Potti A, Dressman HK, Bild A et al. Genomic signatures to guide the use of chemotherapeutics. Nat Med, 2006, 12, 1294-1300.

[31.] Bonnefoi H, Potti A, Delorenzi M et al. Validation of gene signatures that predict the response of breast cancer to neoadjuvant chemotherapy: a substudy of the EORTC 10994/BIG 00-01 clinical trial. Lancet Oncol, in press.

[32.] Di Leo A, Gancberg D, Larsimont D et al. HER-2 amplification and topoisomerase II alpha gene aberrations as predictive markers in node-positive breast cancer patients randomly treated either with an anthracycline-based therapy or with cyclophosphamide, methotrexate, and 5-fluorouracil. Clin Cancer Res, 2002, 8, 1107-1116.

[33.] Bartlett JMS, Ibrahim M, Jasani B et al. External quality assurance of HER2 FISH testing: results of a UK NEQAS pilot scheme. J Clin Path, 2007, 60, 816-819.

[34.] Bartlett JMS. Pharmacodiagnostic testing in breast cancer--focus on HER2 and trastuzumab therapy. Am J Pharmacogenomics, 2005, 5, 303-315.

[35.] Paik S, Shak S, Tang G et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med, 2004, 351, 2817-2826.

[36.] van't Veer LJ, Dai HY, van de Vijver MJ et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature, 2002, 415, 530-536.

[37.] Chang J, Makris A, Gutierrez M et al. Gene expression patterns in formalin-fixed, paraffin-embedded core biopsies predict docetaxel chemosensitivity in breast cancer patients. Breast Cancer Res Treat, 2007, epub ahead of print.

John MS Bartlett and Dana Faratian

Endocrine Cancer Group, Edinburgh Cancer Research Centre, Western General Hospital, Edinburgh, UK

Correspondence to: John Bartlett, Endocrine Cancer Group, Edinburgh Cancer Research Centre, Western General Hospital, Crewe Road South, Edinburgh EH4 2XR, UK (email: [e-mail])