Originally published as JCO Early Release 10.1200/JCO.2007.14.2364 on October 22 2007

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Harris, L. Articles by Bast, R. C. Search for Related Content PubMed PubMed Citation Articles by Harris, L. Articles by Bast, R. C., Jr Related Articles Related Correspondence

American Society of Clinical Oncology 2007 Update of Recommendations for the Use of Tumor Markers in Breast Cancer

Lyndsay Harris, Herbert Fritsche, Robert Mennel, Larry Norton, Peter Ravdin, Sheila Taube, Mark R. Somerfield, Daniel F. Hayes, Robert C. Bast, Jr

From the Yale Cancer Center, Yale University, New Haven, CT; M.D. Anderson Cancer Center, Houston; Texas Oncology PA, Dallas, TX; Memorial Sloan-Kettering Cancer Center, New York, NY; National Cancer Institute, Bethesda, MD; American Society of Clinical Oncology, Alexandria, VA; University of Michigan Medical Center, Ann Arbor, MI

Address reprint requests to American Society of Clinical Oncology, 1900 Duke St, Suite 200, Alexandria, VA 22314; e-mail: guidelines{at}asco.org

	ABSTRACT

TOP ABSTRACT INTRODUCTION UPDATE METHODOLOGY GUIDELINE RECOMMENDATIONS AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Purpose: To update the recommendations for the use of tumor marker tests in the prevention, screening, treatment, and surveillance of breast cancer.

Methods: For the 2007 update, an Update Committee composed of members from the full Panel was formed to complete the review and analysis of data published since 1999. Computerized literature searches of MEDLINE and the Cochrane Collaboration Library were performed. The Update Committee's literature review focused attention on available systematic reviews and meta-analyses of published tumor marker studies. In general, significant health outcomes (overall survival, disease-free survival, quality of life, lesser toxicity, and cost-effectiveness) were used for making recommendations.

Recommendations and Conclusions: Thirteen categories of breast tumor markers were considered, six of which were new for the guideline. The following categories showed evidence of clinical utility and were recommended for use in practice: CA 15-3, CA 27.29, carcinoembryonic antigen, estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2, urokinase plasminogen activator, plasminogen activator inhibitor 1, and certain multiparameter gene expression assays. Not all applications for these markers were supported, however. The following categories demonstrated insufficient evidence to support routine use in clinical practice: DNA/ploidy by flow cytometry, p53, cathepsin D, cyclin E, proteomics, certain multiparameter assays, detection of bone marrow micrometastases, and circulating tumor cells.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION UPDATE METHODOLOGY GUIDELINE RECOMMENDATIONS AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
The American Society of Clinical Oncology (ASCO) first published evidence-based clinical practice guidelines for the use of tumor markers in breast cancer in 1996. ASCO guidelines are updated at intervals by an Update Committee of the original Expert Panel. The last update of the tumor markers guideline was published in 2000. For the 2007 update, the Panel expanded the scope of the guideline to include a broader range of markers in breast cancer. In addition, the impact of genomic technologies was considered in the Update. While molecular subtyping is still in its infancy, and subgroups are not well defined, the use of multiparameter technologies in clinical practice has considerable potential. The updated recommendations are summarized in Table 1.

	UPDATE METHODOLOGY

TOP ABSTRACT INTRODUCTION UPDATE METHODOLOGY GUIDELINE RECOMMENDATIONS AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

The Update Committee's literature review focused attention on available systematic reviews and meta-analyses of published tumor marker studies, although primary data were also reviewed. By and large, however, the primary literature is characterized by studies that included small patient numbers, that are retrospective, and that commonly perform multiple analyses until one reveals a statistically significant result. Furthermore, many tumor marker studies fail to include descriptions of how patients were treated or analyses of the marker in different treatment subgroups. The Update Committee hopes that adherence to a recently published set of suggested guidelines for reporting of tumor marker results (designated the Reporting Recommendations for Tumor Marker Prognostic Studies [REMARK] criteria) will provide more informative data sets in the future.^1,2

The Update Committee has attempted to review tumor markers in reference to a Levels of Evidence framework, which defines the quality of the data on a given marker.³ Most published studies could be designated as Level of Evidence III (evidence from large but retrospective studies), which may generate hypotheses but are insufficient to change clinical practice. The Update Committee attempted, wherever possible, to base the updated recommendations on studies deemed to be Level of Evidence II (prospective therapeutic trials in which marker utility is a secondary study objective), or, ideally, Level of Evidence I (single, high-powered, prospective, randomized controlled trials specifically designed to test the utility of the marker or meta-analyses of well-designed studies).

The Update Committee had two face-to-face meetings to consider the evidence for each of the 2000 recommendations. The guideline was circulated in draft form to the Update Committee. ASCO's Health Services Committee and the ASCO Board of Directors also reviewed the final document.

It is important to emphasize that guidelines and technology assessments cannot always account for individual variation among patients. They are not intended to supplant physician judgment with respect to particular patients or special clinical situations, and cannot be considered inclusive of all proper methods of care or exclusive of other treatments reasonably directed at obtaining the same result.

Accordingly, ASCO considers adherence to this guideline assessment to be voluntary, with the ultimate determination regarding its application to be made by the physician in light of each patient's individual circumstances. In addition, this guideline describes the use of procedures and therapies in clinical practice; it cannot be assumed to apply to the use of these interventions performed in the context of clinical trials, given that clinical studies are designed to evaluate or validate innovative approaches in a disease for which improved staging and treatment is needed. In that guideline development involves a review and synthesis of the latest literature, a practice guideline also serves to identify important questions and settings for further research.

	GUIDELINE RECOMMENDATIONS

TOP ABSTRACT INTRODUCTION UPDATE METHODOLOGY GUIDELINE RECOMMENDATIONS AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
CA 15-3 AND CA 27.29 AS MARKERS FOR BREAST CANCER
2007 recommendation for CA 15-3 and CA 27.29 as screening, diagnostic, or staging tests. Present data are insufficient to recommend CA 15-3 or CA 27.29 for screening, diagnosis, and staging. There is no change from the original guideline.

Literature update and discussion. CA 15-3 and CA 27.29 are well-characterized assays that allow the detection of circulating MUC-1 antigen in peripheral blood. Several studies have been published since the last ASCO guideline that support the prognostic relevance of this circulating marker in early-stage breast cancer.^4-8 In one study of 1,046 patients, Ebeling et al⁴ reported CA 15-3 to be a predictor of worse outcome in univariate but not multivariate analysis including tumor size, lymph node status, histologic grade, and estrogen receptor (ER) status. Gion et al⁵ further reported a highly significant prognostic contribution for CA 15-3 in a Cox regression model that included age, ER status, and tumor stage in a group of 362 node-negative breast cancers. While it is likely that serum tumor markers CA 15-3 and CA 27.29 have prognostic value, their role in the management of early-stage breast cancer is unclear.^9,10 It has yet to be determined that MUC-1–based serum markers are helpful in making treatment decisions in this setting. Therefore, the Update Committee did not recommend their measurement at diagnosis.

2007 recommendation for CA 15-3 and CA 27.29 to detect recurrence after primary breast cancer therapy. Present data do not support the use of CA 15-3 and CA 27.29 for monitoring patients for recurrence after primary breast cancer therapy. There is no change from the guideline published in 2000.

Literature update and discussion. Several well-designed studies have shown that an increase in CA 15-3 or CA 27.29 after primary and/or adjuvant therapy can predict recurrence an average of 5 to 6 months before other symptoms or tests. While additional studies have been published since the last ASCO guideline that address the value of these serum markers at detecting recurrence,^11-16 there are no prospective randomized clinical trials to demonstrate whether detection and treatment of occult or asymptomatic metastases using tumor markers impact on the most significant outcomes (disease-free survival, overall survival, quality of life, toxicity, or cost-effectiveness). Although the assay was approved by the US Food and Drug Administration, the US Food and Drug Administration does not require tests to show clinical benefit if that is not part of the manufacturer's indication. Given the limited evidence, and until clinical benefit is established, present data are insufficient to recommend routine use of CA 15.3 or CA 27.29 for this application. This recommendation is in line with that of the ASCO guideline for follow-up and management of patients with breast cancer.⁹

2007 recommendation for CA 15-3 and CA 27.29 to contribute to decisions regarding therapy for metastatic breast cancer. For monitoring patients with metastatic disease during active therapy, CA 27.29 or CA 15-3 can be used in conjunction with diagnostic imaging, history, and physical examination. Present data are insufficient to recommend use of CA 15-3 or CA 27.29 alone for monitoring response to treatment. However, in the absence of readily measurable disease, an increasing CA 15-3 or CA 27.29 may be used to indicate treatment failure. Caution should be used when interpreting a rising CA 27.29 or CA 15-3 level during the first 4 to 6 weeks of a new therapy, given that spurious early rises may occur. There is no change from the guideline published in 2000.

Literature update and discussion. No relevant studies were identified from the review of the literature conducted for this topic.

CARCINOEMBRYONIC ANTIGEN AS A MARKER FOR BREAST CANCER
2007 recommendation for carcinoembryonic antigen for screening, diagnosis, staging, or routine surveillance of breast cancer patients after primary therapy. Carcinoembryonic antigen (CEA) is not recommended for screening, diagnosis, staging, or routine surveillance of breast cancer patients after primary therapy. There is no change from the guideline published in 2000.

Literature update. No relevant studies were identified from the review of the review of literature conducted for this topic.

2007 recommendation for CEA to contribute to decisions regarding therapy for metastatic breast cancer. For monitoring patients with metastatic disease during active therapy, CEA can be used in conjunction with diagnostic imaging, history, and physical examination. Present data are insufficient to recommend use of CEA alone for monitoring response to treatment. However, in the absence of readily measurable disease, an increasing CEA may be used to indicate treatment failure. Caution should be used when interpreting a rising CEA level during the first 4 to 6 weeks of a new therapy, given that spurious early rises may occur. There is no change from the guideline published in 2000.

Literature update and discussion. CEA levels are less commonly elevated than are levels of the MUC-1 assays, CA 27.29, or CA 15-3. Only 50% to 60% of patients with metastatic disease will have elevated CEA levels, compared with 75% to 90% who have elevated levels of the MUC-1 antigen.^17-22 CEA levels are minimally complementary with MUC-1 levels. For example, in one study of 53 women with metastatic breast cancer, CA 15-3 and CEA levels were elevated in 94% and 69%, respectively. CEA was elevated in only a single case in which CA 15-3 was not.²³ Nonetheless, in several studies there have been selected cases in which CEA is informative (elevated) and CA 15-3 or CA 27.29 is not.^13,24-31 Older studies suggest that, like the MUC-1 assays, CEA levels appear to track with disease status.^20,32-34 Taken together, these data suggest that it is reasonable to evaluate one of the MUC-1 assays and CEA initially in a patient with metastatic disease. If the MUC-1 assay is elevated, there appears to be no role for monitoring CEA, but if not, then CEA levels may provide supplementary information to the clinician in addition to clinical and radiographic investigations.

ERS AND PROGESTERONE RECEPTORS AS MARKERS FOR BREAST CANCER
2007 recommendation for ERs and progesterone receptors. ER and progesterone receptor (PgR) should be measured on every primary invasive breast cancer and may be measured on metastatic lesions if the results would influence treatment planning. In both pre- and postmenopausal patients, steroid hormone receptor status should be used to identify patients most likely to benefit from endocrine forms of therapy in both the early breast cancer and metastatic disease settings. In patients with ductal carcinoma in situ (DCIS) who are candidates for hormonal therapy, data are insufficient to recommend routine measurement of ER and PgR for therapy recommendations.

Literature update and discussion. ER and probably PgR content are associated with a favorable prognosis, and more importantly, highly predictive of benefit from endocrine treatment in both the adjuvant and metastatic settings.^35-37 These treatments include tamoxifen, ovarian ablation (surgical or chemical), aromatase inhibitors (anastrazole, letrozole, exemestane), and irreversible ER inhibitors (eg, fulvestrant). Endocrine treatments are used for prevention of new cancers and of recurrent distant metastases as well as for the treatment of metastatic disease.³⁸ Fortunately, the majority of contemporary clinical trials have incorporated estrogen and progesterone receptor testing with the evaluation of newer antiestrogens and continue to demonstrate the value of these markers for predicting response to hormonal therapy.³⁹ Nonetheless, the Update Committee acknowledges the deficits in standardization for ER and PgR assays (in particular, immunohistochemistry [IHC]), and further efforts at defining reproducibility and accuracy for particular reagents are an important priority. With those caveats, the previous guideline recommendations regarding the use of ER and PgR for diagnosis and treatment of invasive breast cancer remain unchanged.

A topic that has emerged since the 2000 update is the potential role of hormone receptor determination in the management of DCIS. DCIS is a complex group of diseases that have diverse outcomes and account for approximately 20% to 30% of breast cancer cases.^40-42 Most physicians accept the concept that high nuclear grade and necrosis predict a worse outcome for patients with DCIS.^43-48 Although ER negativity is associated with a worse outcome in patients with DCIS, it is not an independent predictor in the context of high nuclear grade and necrosis.⁴⁹ Therefore the Update Committee does not recommend the use of the ER as a predictor of outcome in patients with DCIS.

The current treatment options for DCIS include mastectomy, lumpectomy followed by breast radiation therapy,^50-53 or lumpectomy alone in selected patients.^54-57 The addition of tamoxifen to the lumpectomy followed by breast radiation therapy is supported by the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-24 trial,^51,58 which showed a significant decrease in the recurrence of both in situ and invasive breast cancer in the tamoxifen group, with no impact on overall survival. A single report, available in abstract form only, suggested the benefits of tamoxifen in regard to reduction of local recurrence, and second primary breast cancers might be confined to those patients whose original DCIS expressed ER.⁵⁹ Another large randomized trial of adjuvant tamoxifen in DCIS, the United Kingdom Coordinating Committee on Cancer Research trial, failed to show an advantage for the tamoxifen-treated group in either the recurrence of breast cancer or overall survival.⁵² Data from the Early Breast Cancer Trialists’ Collaborative Overview are mixed regarding whether the hormone responsiveness of a contralateral breast cancer is related to the ER content of the first primary.³⁵ These data were retrospective in design at best. At present, the Update Committee felt that data were insufficient to support using the ER status of DCIS to elect to treat with or withhold tamoxifen in a patient who undergoes breast preservation.

MARKERS OF PROLIFERATION
2007 recommendation for flow cytometry–based proliferation markers. Present data are insufficient to recommend use of DNA content, S phase, or other flow cytometry–based markers of proliferation to assign patients to prognostic groupings. There is no change from the guideline published in 2000.

Literature update and discussion. DNA flow cytometry determination of S phase is one of several markers of proliferative rate in breast tumor specimens. In general, markers of elevated proliferative rate correlate with a worse prognosis in untreated patients, and may predict benefit from chemotherapy.⁶⁰ The implementation of DNA flow cytometry as a marker of proliferative rate is complicated by the variation in methods of tissue preparation and differences in instrumentation and methods for converting information on the histograms to the S-phase estimate. In addition, interpretation of individual studies is complicated by the fact that many are too small to have statistical power, cut-offs have not been prospectively defined, and study populations have not been controlled for adjuvant systemic treatments.

Table 2 summarizes results published from 1999 to 2004 showing the prognostic value of S phase on outcome of node-negative patients. In studies with more than 200 patients, S phase was a consistent univariate predictor of outcome, whereas smaller studies were generally negative. The prognostic value seen in the larger studies was usually maintained after multivariate analysis. In the one large study where multivariate analysis did not confirm its value, the inclusion of another measure of mitotic index eliminated S phase. Of the five larger studies, one that claimed to use prospectively defined methodologies and cut points was strongly positive.⁶¹

2007 recommendation for immunohistochemically based markers of proliferation in breast cancer. Present data are insufficient to recommend measurement of Ki67, cyclin D, cyclin E, p27, p21, thymidine kinase (TK), topoisomerase II, or other markers of proliferation to assign patients to prognostic groupings.

Marker definition. Additional markers of proliferation have been measured by IHC to determine their prognostic and predictive value in breast cancer. These include but are not limited to Ki 67, TK, cyclin E, cyclin D, cyclin inhibitors p27 and p21, and topoisomerase II. These measures of proliferation are typically enzymes involved in DNA metabolism (eg, TK), cell cycle checkpoint functions (eg, cyclins, p27, p21), and DNA-modifying enzymes (eg, topoisomerase II). Ki67, MIB-1 and PCNA are proliferating cell nuclear antigens of unknown function and are present exclusively in dividing cells.

Literature review and discussion. The prognostic and predictive role of Ki67, cyclin D, cyclin E, p27, p21, TK, and topoisomerase II are discussed by Colozza et al⁶⁰ in an exceptionally thorough review of 132 articles including 159,516 patients. The authors appropriately point out that all studies concerning these markers are level IV or III at best, and demonstrate the difficulty in interpreting the literature due to lack of standardization of assay reagents, procedures, and scoring. In addition, the majority of marker studies address the prognostic role of the marker, whereas studies of the predictive value for efficacy of treatment are either lacking or performed on small sample sizes without a randomized comparison for a particular marker. These issues led the authors to conclude that Ki67, cyclin D, cyclin E, p27, p21, TK, and topoisomerase II are not recommended for clinical practice. The Update Committee concurs with these conclusions and refers the reader to this elegant review for additional details. In addition, cyclin E is discussed further in this guideline.

HER2 AS A MARKER FOR BREAST CANCER
2007 recommendation for HER2 evaluation in breast cancer. HER2 expression and/or amplification should be evaluated in every primary invasive breast cancer either at the time of diagnosis or at the time of recurrence, principally to guide selection of trastuzumab in the adjuvant and/or metastatic setting. Other utilities for HER2 evaluation are also discussed separately below.

Literature update and discussion. HER2 is a member of the epidermal growth factor receptor (EGFR) family.⁷⁰ It is amplified and overexpressed in 15% to 30% of newly diagnosed breast cancers and is associated with more aggressive behavior.⁷¹ Several potential clinical applications have been proposed for determination of HER2 status in breast cancer patients, including (1) determination of prognosis in untreated patients; (2) prediction of resistance to endocrine therapy or of selective resistance to tamoxifen but not aromatase inhibitors; (3) prediction of relative resistance to certain chemotherapies, such as cyclophosphamide, methotrexate, and fluorouracil (CMF) –like regimens; (4) prediction of benefit from anthracycline or paclitaxel; and (5) prediction of benefit from anti-HER2 therapies, in particular trastuzumab and lapatinib. Circulating HER2 extracellular domain (ECD) levels have been proposed as a surrogate for tissue measures of HER2, to monitor patients for early relapse or to monitor response to standard therapies or HER2-targeted therapies. These utilities were considered and commented on in the Guideline. HER2 can be measured in tissue by assays for expression, most commonly by IHC, or for gene amplification, most commonly by fluorescent in situ hybridization (FISH). A separate Expert Panel convened jointly by the College of American Pathologists (CAP) and ASCO has recently published a set of guideline recommendations regarding analysis of tissue HER2 status, in which it was strongly recommended that laboratories offering this service be accredited on an annual basis.⁷² The Update Committee endorses the ASCO-CAP guideline; hence, this topic was not covered further in the present guideline update.

The ECD of HER2 can be detected in serum or plasma, most commonly by a commercially available enzyme-linked immunosorbent assay (ELISA), and is elevated in approximately 30% of patients with metastatic breast cancer.^73-84

2007 recommendations for HER2 to define prognosis for early-stage breast cancer patients in the absence of systemic therapy. HER2 amplification, overexpression, and the presence of HER2 extracellular domain are generally associated with a poorer prognosis. However, the value of this information in clinical practice is questionable and the use of HER2 for determining prognosis is not recommended. There is no change from the guideline published in 2000.

Literature update and discussion. The prognostic significance of HER2 overexpression in tumor tissue has been evaluated in several clinical trials with most, but not all, studies suggesting that HER2 positivity is associated with worse prognosis in untreated patients.^71,85,86 Due to the variability in immunohistochemical assays and scoring systems used, there is insufficient evidence to endorse IHC-based testing for HER2 in determining prognosis for breast cancer patients. The results of HER2 amplification as a prognostic factor are more consistent, with HER2 amplification usually associated with worse prognosis, including node-negative populations.^87-89 As discussed below, most studies of serum HER-2 extracelluar domain have found an association with higher tumor stage and increased tumor burden.^74,90,91 As might be expected, elevated levels of HER2/ECD correlate with worse prognosis.^75,92 However, serum HER2 appears to retain its prognostic effect in multivariate models, suggesting a biologic role beyond its association with HER-2 tissue expression.^83,90-94 While the weight of evidence suggests that HER2 amplification/overexpression and/or shedding of ECD are associated with worse outcome, the role of this marker purely to determine prognosis in clinical practice is unclear, given that outcomes are so heavily influenced by subsequent therapy. Hence, the Update Committee does not recommend the measurement of HER2, by any method, for the sole purpose of determination of patient prognosis.

2007 recommendation for use of HER2 to select patients for anti-HER2–based therapy. High levels of tissue HER2 expression or HER2 gene amplification should be used to identify patients for whom trastuzumab may be of benefit for treatment of breast cancer in the adjuvant or metastatic disease settings. There is no change from the guideline published in 2000.

Literature update and discussion. Trastuzumab is a humanized monoclonal antibody that binds to the extracellular domain of HER2. A prospective randomized clinical trial has demonstrated that trastuzumab improves response rates, time to progression, and overall survival when combined with chemotherapy compared with chemotherapy alone in the metastatic setting.⁹⁵ Phase II monotherapy studies have demonstrated that trastuzumab induces responses in approximately 15% to 25% of selected patients.^96-98 Eligibility for all of these trials was based on HER2 positivity, either by IHC or FISH. It has been assumed that patients without HER2-positive cancers will not benefit from trastuzumab. A single unpublished prospective randomized clinical trial has addressed the value of trastuzumab added to paclitaxel in patients with HER2 low (or "equivocal") metastatic breast cancer, and no statistically significant differences were reported for any outcome.⁹⁹

Five prospective randomized clinical trials have now been reported in the adjuvant setting, as well as a single, small, prospective, randomized neoadjuvant clinical trial. Each has shown a remarkable beneficial effect of trastuzumab on pathologic complete response, disease-free survival, and overall survival.^100-104 As in the metastatic setting, eligibility for these trials depended on some measure of HER-2 positivity (either 3+ staining by IHC or FISH amplification more than 2.0). Therefore, at present, trastuzumab is indicated only for HER2-positive patients, and patients with HER2-negative status (IHC 0-2+ and FISH negative) should not receive trastuzumab. The Update Committee refers the reader to the recently published ASCO-CAP detailed guideline for methodology and accreditation of assays for HER2.⁷²

Recently published data from a prospective randomized clinical trial suggest that the addition of the epidermal growth factor family tyrosine kinase inhibitor, lapatinib, to capecitabine resulted in better outcomes than capecitabine alone in patients with HER2-positive metastatic breast cancer. The Update Committee anticipates that HER2 status may also be used to guide lapatinib therapy in the future.¹⁰⁵

Sensitivity to Chemotherapy
2007 recommendation for the utility of HER2 for predicting response to specific chemotherapeutic agents. Level II evidence (prospective therapeutic trials in which marker utility is a secondary study objective) suggests that overexpression of HER2 (3+ by protein or > 2.0 FISH ratio by gene amplification) identifies patients who have greater benefit from anthracycline-based adjuvant therapy. If a clinician is considering chemotherapy for a patient with HER2-positive breast cancer, it is recommended that an anthracycline be strongly considered, assuming there are no contraindications to anthracycline therapy. In the context of trastuzumab therapy, there is Level I evidence (single, high-powered, prospective, randomized controlled trials specifically designed to test the marker or a meta-analyses of well-designed studies) that a nonanthracycline regimen may produce similar outcomes. At present, the Update Committee does not recommend that HER2 be used to guide use of taxane chemotherapy in the adjuvant setting.

Literature update and discussion. The role of HER2 in both tissue and serum in predicting response to specific agents has been evaluated. Most trials involving CMF-based regimens suggest that patients with HER2-positive tumors benefit less with this therapy than do patients with HER2-negative tumors.^106-109 However, results from randomized phase III trials of CMF versus no chemotherapy and CMF with or without the addition of anthracycline-containing therapy suggest that patients with HER2-positive breast cancers still derive some benefit from CMF, but it appears that the addition of an anthracycline further improves their prognosis.^110,111

It is not clear if HER2 is specific for benefit from anthracyclines, or whether HER2 is associated with benefit from addition of any therapy that is more effective overall.^112-118 Indeed, it is not clear whether HER2 itself is the target of anthracyclines or if HER2 status serves as a surrogate for a different gene product that may be the target of the anthracycline. In this regard, several groups have evaluated the abnormalities (amplification and/or deletion) of topoisomerase II (Topo II), which is located on the same amplicon on chromosome 17 as HER2. Anthracyclines directly bind Topo II and function, at least in part, by inhibiting its activity in DNA replication, therefore making it an attractive marker for anthracycline activity.¹¹⁹ Topo II may increase sensitivity to anthracyclines and also confer relative resistance to alkylating agents in preclinical studies.^120,121 While several clinical cohorts have been evaluated for Topo II amplification and the results generally support this explanation for altered sensitivity to anthracyclines in HER2-amplified breast tumors, other studies do not confirm these findings.^122-125 Although these studies approach Level of Evidence II quality as defined earlier (prospective therapeutic trials in which marker utility is a secondary study objective), the uncertainty regarding the biologic relationship between Topo II protein expression, copy number, proliferation, and benefit from anthracylines makes assessment of Topo II unreliable at this time. In fact, recent trials suggest that the model of a direct relationship between Topo II amplification, overexpression of Topo II protein, and benefit from anthracyclines is overly simplistic.^126,127 The fact that Topo II protein level corresponds to proliferation rate, but not Topo II copy number, suggests that the coamplification of Topo II may not be associated with increased target for anthracycline-containing therapy as predicted. Furthermore, both deletion and amplification of the Topo II region are associated with benefit from anthracycline-containing therapy in HER2-amplified tumors.^124,126,127 Since topoisomerase II protein is essential for chromosome segregation and proliferation, and is more abundant in aneuploid tumors, it seems unlikely that Topo II amplification fully explains benefit from anthracyclines in the setting of HER2 amplification.^128,129

The previous discussion notwithstanding, most correlative studies have suggested that HER2 amplification and/or overexpression identifies those patients in randomized trials who benefit from anthracycline-based chemotherapy compared with CMF, while in HER2-negative patients there appears to be no difference between the two regimens.^115,130,131 Thus, given the weight of the evidence for HER2, it seems prudent to recommend anthracycline-based adjuvant chemotherapy for a patient with HER2-positive breast cancer, assuming adjuvant chemotherapy is indicated, the patient has no contraindication to an anthracycline, and trastuzumab administration is not planned.

The benefit of taxane-based therapy for HER2-positive tumors is controversial. Some studies suggest improved response to docetaxel or paclitaxel, while others suggest relative resistance.^{122,130,132,133} This may relate, in part, to the method for detecting HER2, given that serum HER2 has been used to determine HER2 positivity in some studies and is associated with tumor burden (as discussed below), which confounds the ability to discern the independent predictive value of HER2 in this setting. In a retrospective analysis of a trial comparing three different doses of paclitaxel monotherapy in patients with metastatic breast cancer, tissue HER2 status was not associated with response rate, disease-free survival, or overall survival.¹³⁴ In contrast, another retrospective analysis reported that HER2 amplification was associated with benefit from paclitaxel and doxorubicin compared with cyclophosphamide and doxorubicin, while there was no difference in outcomes for HER2-negative patients with metastatic breast cancer.¹³⁵

A recent study of HER2 by FISH and IHC in Cancer and Leukemia Group B 9344/Intergroup 0148 trial suggests that the benefit from the addition of adjuvant paclitaxel after four cycles of doxorubicin and cyclophosphamide in node-positive breast cancer patients is more pronounced in those with HER2-positive breast cancers.¹³² Indeed, there was no detectable benefit from addition of paclitaxel in HER2-negative, ER-positive patients. This observation may explain the variability in studies looking at taxane benefit in HER2-positive tumors because ER status varies by cohort. Again, this study does not distinguish between a benefit from taxane-based therapy versus the addition of more effective chemotherapy in HER2-positive tumors. Until this study is published and corroborated, these results must be viewed as preliminary.

In summary, the data regarding the predictive value of HER2 and response to chemotherapy generally support the concept that the benefit of adjuvant anthracycline therapy is most marked in the HER2-positive subgroup of patients. However, the benefit of taxane-based therapy in HER2-positive patients remains controversial and definitive conclusions have not been reached.

SENSITIVITIES TO ENDOCRINE THERAPY IN GENERAL OR TO SPECIFIC ENDOCRINE THERAPIES
2007 recommendation for use of HER2 to determine sensitivity to endocrine therapy. HER2 should not be used to withhold endocrine therapy for a patient with hormone-receptor positive breast cancer, nor should it be used to select one specific type of endocrine therapy over another. There is no change from the guideline published in 2000.

Literature update and discussion. Complex interactions exist between the HER2 and ER pathways. HER2 expression in human breast cancer cells is downregulated by estrogens.¹³⁶ Conversely, overexpression of HER2 promotes estrogen-independent growth and is associated with resistance to tamoxifen in vitro and in animal models, possibly by promoting ligand-independent growth. These observations are consistent with the inverse association of estrogen and progesterone receptors with HER2 overexpression and also provide a rationale for the lower response of HER2-overexpressing tumors to endocrine therapy shown in several clinical studies.^107,137-141 However, most of these studies were retrospective and nonrandomized. To date, randomized trials have not led to consensus on this association.^142-145 The interaction of HER2 with endocrine therapy may vary depending on the type of hormonal agent in question. Ellis et al¹⁴⁶ have shown that HER2- and/or EGFR-positive tumors were more likely to respond to neoadjuvant letrozole than tamoxifen in a randomized trial of 324 primary breast cancer patients. In contrast, an analysis (presented in abstract form only) of the Anastrozole versus Tamoxifen versus a Combination of the two (ATAC) trial, failed to show that HER2-overexpressing tumors benefit more from the aromatase inhibitor.^147,148

In summary, there are insufficient data to support the use of HER2 in tissue (or serum, as discussed below) as a predictor of response to endocrine therapy, although the evidence does suggest that in patients with ER-positive tumors, the relative benefit from antiestrogens for those with HER2-positive cancers is likely to be lower than for those with HER2-negative cancers. It is not at all clear that the benefit of aromatase inhibitors in this group is any greater than in the HER2-negative, ER-positive group.

Utility of Measures of Circulating ECD of HER2
2007 recommendation for the utility of circulating extracellular domain of HER2. Measuring circulating extracellular domain of HER2 is not currently recommended for any clinical setting. There is no change from the original guideline.

Literature update and discussion. The HER2 extracellular domain was initially isolated in culture media from an HER2-amplified cell line,¹⁴⁹ and in the serum of nude mice bearing xenografts from HER2-amplified cells.¹⁵⁰ It was subsequently isolated from pleural effusions and serum of advanced breast cancer patients.⁷⁸ Several studies have shown it to be present in roughly 25% of unselected patients. On comparison with tissue expression, it appears that the majority of patients who shed ECD are positive for HER2 at the level of the primary tumor.¹¹² The functional significance of ECD shedding has not been determined, but in vitro data suggest that deletion of the extracellular carboxy terminus of the molecule enhances the signaling activity and transforming ability of the NH-2 terminally truncated receptor, p95 HER2.^151,152

Therefore, the ECD of HER2 might serve as a surrogate marker for tissue HER2 status for any or all of the utilities discussed above, especially prediction of benefit from trastuzumab or anthracyclines. Furthermore, serial HER2 ECD levels might be useful for monitoring, either to detect recurrence in asymptomatic patients who are believed to be free of detectable disease, or to determine disease status in patients with metastatic breast cancer.

As with tissue HER2 status, serum HER2 might be useful to determine prognosis. Studies of serum HER2 more uniformly suggest worse outcome. However, in early-stage disease, as with circulating MUC-1 or CEA, levels of circulating HER2 ECD are directly related to tumor burden in patients with HER2-positive breast cancer, and there are no studies that suggest knowledge of HER2 ECD is of value in this setting.¹⁵³ Likewise, in patients with metastases, elevated levels of circulating HER2 are associated with worse outcomes, but not to the extent that a patient might be treated differently based simply on "prognosis."^154,155

Pretreatment circulating HER2 might be used as a predictive factor for selection of specific therapy, especially in the metastatic setting. Many patients with new or serially progressive metastatic disease may not have had HER2 measured in their primary cancers (although the Update Committee anticipates that this situation will become increasingly less common). Furthermore, several studies have suggested that a small fraction of metastatic HER2 evaluations are discordant from the primary measurements.¹⁵⁶ If HER2 status is important to direct therapy, measurement of the HER2 status may be worthwhile in patients with metastases. A circulating tumor marker that accurately reflects tissue HER2 status has certain advantages over rebiopsy of a metastatic lesion, with less morbidity and ability to monitor changes serially in disease biology. Several publications have attempted to address this utility in the context of both endocrine and trastuzumab-based therapy.^81,157-159

As noted, one possible indication for HER2 would be to direct endocrine therapy. Several studies have suggested that pretreatment circulating HER2 ECD levels in metastatic patients are associated with lower response, shorter time to progression, and worse survival in ER-positive patients about to begin a new endocrine treatment. However, most (if not all) of these studies were confounded by the known association of serum HER2 with greater disease burden.^77,84 In a study of patients with advanced breast cancer randomly assigned to receive tamoxifen or letrozole, the presence of elevated ECD correlated with a lower response to both regimens, with no advantage of letrozole over tamoxifen.⁸¹ However, there was a statistically significant improvement in time to progression in patients with shed ECD treated with letrozole versus tamoxifen, suggesting that the aromatase inhibitors may exhibit some advantage in the HER2-positive population.⁸¹ Patients in this trial were randomly assigned to either therapy, but the correlative analysis of ECD and response to therapy was conducted retrospectively. Perhaps the most promising of use of HER2 ECD would be to predict response to trastuzumab (or other HER2-directed therapies, such as lapatinib) and to monitor disease response and progression once treatment has begun.

Given the association of HER2/ECD with HER2 overexpression, it seems likely that this marker could also predict response to trastuzumab. On the other hand, HER2/ECD is associated with a higher tumor burden, which may lower response rates and decrease the half-life of the antibody, due to the abundance of binding sites. Another concern resides in the formation of immune complexes between HER2/ECD and trastuzumab, with the potential for accelerated clearance and reduction in the efficacy of this therapy. Of note, concerns that circulating trastuzumab might interfere with the measurement of HER2/ECD levels have been refuted by results of in vitro experiments.¹⁵⁹ While high levels of HER2/ECD (500 ng/mL) were shown to decrease the half-life of trastuzumab, high levels of serum HER2/ECD do not preclude response in trastuzumab-treated patients and may, in fact, predict a more favorable response.^155,157,160 Most studies show a more precipitous decline in serum HER2 to be associated with favorable response, suggesting that this marker may be useful for monitoring disease course during trastuzumab-containing therapy.¹⁶⁰ However, the definition of a favorable response by HER2/ECD has not been uniformly defined in published studies.

Serum HER2 has been studied to monitor disease for recurrence response and progression in several trials.¹⁵⁸ Although rising ECD has been associated with recurrence in early-stage disease, serum HER2 tracks with response and progression in some patients being treated for metastatic disease, it is frequently discordant with disease course during either chemotherapy or hormonal therapy.^153,160

In summary, although appealing, use of circulating HER2/ECD is hampered by a lack of high-quality studies and a lack of consistent findings. These are required to understand fully the precise utility of this marker in evaluation or monitoring of patients with breast cancer.

p53 AS A MARKER FOR BREAST CANCER
2007 recommendation for p53. Present data are insufficient to recommend use of p53 measurements for management of patients with breast cancer. There is no change from the original guideline.

Literature update and discussion. The results from recently reported studies are insufficient to change the recommendation from the 1999 version of the guideline. A number of studies suggest that high tissue p53 protein levels measured by IHC or mutations or deletions in the p53 gene measured by single-strand conformational gel electrophoresis, manual sequencing, or allele-specific polymerase chain reaction (PCR) appear to be a univariate predictor of poor outcome (Table 3). A meta-analysis performed in 1999¹⁷⁴ suggests that p53 mutations confer an independent relative risk of 1.7 (95% CI, 1.2 to 2.4) for both disease-free survival and overall survival. However, it seems unlikely that IHC for p53 will provide sufficiently accurate results to be clinically useful, given that it detects both mutated p53 and stabilized wild-type p53, and conversely will miss p53 deletions. Methods to define more precisely and conveniently genetic abnormalities in p53 might permit a more accurate analysis of association of p53 and clinical outcomes, either as a pure prognostic factor or as a predictor of benefit from systemic therapies. However, at present, methodologies to do so are cumbersome, expensive, and not widely available as routine clinical assays, limiting the utility of this marker in clinical practice. Furthermore, there are no prospective or retrospective studies to confirm the clinical utility of these methods, even if they were logistically feasible.

The Update Committee again had difficulty discerning the potential bias introduced into most studies of p53 by the confounding effects of therapy. As with many of the other markers addressed in this guideline update, it is likely p53 abnormalities are associated with either resistance or sensitivity to different therapeutic agents. Most studies analyzing p53 have not taken therapy into consideration, and the results may be strongly biased in one direction or the other, depending on the agents in question.

UROKINASE PLASMINOGEN ACTIVATOR AND PLASMINOGEN ACTIVATOR INHIBITOR 1 AS MARKERS FOR BREAST CANCER (Note. This topic is new to the guideline)
2007 recommendation for urokinase plasminogen activator and plasminogen activator inhibitor 1. Urokinase plasminogen activator (uPA)/plasminogen activator inhibitor (PAI-1) measured by ELISAs on a minimum of 300 mg of fresh or frozen breast cancer tissue may be used for the determination of prognosis in patients with newly diagnosed, node-negative breast cancer. IHC for these markers is not accurate, and the prognostic value of ELISA using smaller tissue specimens has not been validated. Low levels of both markers are associated with a sufficiently low risk of recurrence (especially in hormone receptor–positive women who will receive adjuvant endocrine therapy) that chemotherapy will only contribute minimal additional benefit. Furthermore, CMF-based adjuvant chemotherapy provides substantial benefit, compared with observation alone, in patients with high risk of recurrence as determined by high levels of uPA and PAI-1.

uPA and PAI-1: Marker definition. uPA and PAI-1 are part of the plasminogen activating system, which includes the receptor for uPA and other inhibitors (PAI-2 and PAI-3). This system has been shown experimentally to be associated with invasion, angiogenesis, and metastasis.¹⁷⁵

uPA and PAI-1: Methodology. Several assay formats for these two markers have been evaluated, including IHC, quantitative real-time reverse transcriptase (RT) -PCR, and enzyme-linked immunosorbent assays (ELISA).^176-178 ELISA, performed on fresh or frozen tissue or cytosolic fractions remaining after biochemical hormone-receptor measurement, is the only method that has been determined to be prognostic.¹⁷⁹ Importantly, all the data from a pooled analysis study¹⁷⁹ and from a prospective randomized clinical trial¹⁸⁰ in which uPA and PAI-1 were used to stratify patients were obtained based on analysis of large tissue sections from tumors that had not been previously biopsied. Although ELISA using tissue from core needle biopsies would be clinically useful, the prognostic value of such a strategy remains to be confirmed.¹⁸¹ The effects of a prior core biopsy on uPA and PAI-1 levels, which could conceivably alter expression of these tissue-remodeling enzymes, are unknown.

uPA and PAI-1: Literature Review and Analysis
Risk, screening, and monitoring. Currently available data address the impact of uPA and PAI-1 on prognosis for patients with early-stage breast cancer. A retrospective study suggests that ductal fluid uPA/PAI-1 levels might be of use for screening or risk recategorization of high-risk women, but these data require verification.¹⁸² There are few if any data regarding monitoring patients with serial uPA/PAI-1 levels.^183-185

Prognosis in Early-Stage Breast Cancer
Several studies have suggested that overexpression of uPA and/or PAI-1 have been consistently related to poor prognosis in early-stage breast cancer. These studies suggest that these two factors, combined, are associated with 2- to 8-fold higher risk of recurrence and death.^{176,177,186-190} Importantly, studies of node-negative patients who did not receive adjuvant systemic therapy suggest that these two markers are very strong prognostic factors, independent of size, grade, and hormone receptor status.^179,190,191

A pooled analysis of uPA/PAI-1 data collected from 8,377 breast cancer patients was performed by members of the Receptor and Biomarker Group of the European Organisation for Research and Treatment of Cancer.¹⁷⁹ These results demonstrate the reproducibility of the assay among several sites, and they confirm the strong association of overexpression of uPA and PAI-1 with recurrence and survival during a median follow-up of 79 months. A subset analysis of node-negative, untreated patients also confirmed the potential utility of these markers for identifying a low-risk cohort in this group.

The first interim report of a prospective trial using uPA and PAI-1 levels to stratify node-negative patients has been published.¹⁸⁰ Five hundred fifty-six node-negative patients were accrued. Those patients whose tumors expressed low levels of both markers were followed in a prospective registry and were not treated with adjuvant chemotherapy. Patients whose tumors showed elevated uPA and/or PAI-1 levels were randomly assigned to adjuvant chemotherapy (CMF) or no adjuvant chemotherapy. In this report the estimated 3-year recurrence rate for 241 patients with low levels of both uPA/PAI-1 was 6.7%, with a median follow-up of 32 months. The recurrence rate for patients with elevated uPA and/or PAI-1 levels who did not receive chemotherapy was roughly double that, and the hazard rate for recurrence in the group for patients treated with adjuvant chemotherapy was 0.56 of that for patients who were not treated.

Other reports suggest uPA and/or PAI-1 may serve as predictive factors for hormone therapy and/or specific types of chemotherapy, but these are uncontrolled studies.^182,192

The data support the requirement for both uPA and PAI-1 levels to be performed using ELISAs on whole sections (minimum 300 mg) of fresh or frozen cancer tissue. IHC results do not reliably predict outcomes, and the prognostic value of ELISA using smaller tissue specimens, such as tissue collected by core biopsy, has not been validated.¹⁸¹ Furthermore, in the modern era of frequent pre-excision, diagnostic core needle biopsies, one must interpret uPA and PAI-1 ELISA results with caution.

Future Studies
Studies are underway in Europe to address further the utility of uPA/PAI-1 measurements. In an ongoing prospective clinical trial, patients are randomly assigned to two groups: in one group, they will have clinical decisions regarding adjuvant chemotherapy using uPA/PAI-levels; in the other group, these decisions will be made according to existing guidelines. Carefully designed studies addressing the predictive role of uPA/PAI-1 for specific chemotherapy and endocrine therapy are recommended. Finally, components of the urokinase plasminogen activating system appear to be promising targets for future therapeutic studies.

CATHEPSIN D AS A MARKER FOR BREAST CANCER
2007 recommendation for cathepsin D. Present data are insufficient to recommend use of cathepsin D measurements for management of patients with breast cancer. There is no change from the guideline published in 2000.

Literature update and discussion. The role of cathepsin D in breast cancer pathogenesis and outcome has been studied extensively. A Dutch study of 2,810 patients between 1978 and 1992 provides the largest data set used to evaluate the relevance of this marker in breast cancer.¹⁹³ In this study 1,412 patients were node negative and did not receive systemic adjuvant therapy. Median follow-up was 88 months. Cathepsin D levels were determined in breast tumor cytosols using a radiometric immunoassay (ELSA-CATH-D; CIS Bio International, Gif-sur-Yvette, France). The use of a cut point of 45.2 pmol/mg of protein cathepsin D was modestly predictive (hazard ratio, 1.39) in both node-negative and node-positive populations by multivariate analysis, which included tumor size, number of nodes, and ER status but not tumor grade.

In a subsequent study¹⁹⁴ of 1,851 patients (1,182 node-negative patients) with 59 months of follow-up, high levels of cathepsin D expression were associated with a 1.7-fold higher hazard of relapse both in univariate and multivariate analyses using a cut point of 10 pmol/mg of protein; this cut point was defined retrospectively to optimize the results. Although these results show cathepsin D determinations to be predictive of outcome, the magnitude of this effect would be expected to be relatively small (if a relative risk of 1.4 was used), splitting a population with a 20% risk into populations with a low 17% risk and a high 23% risk. In general, the Committee has found that studies of cathepsin D measured by IHC are variable, with no assay standardization and inconsistent associations with outcome, and, again, with little regard to the confounding effects of systemic therapy.

CYCLIN E AS A MARKER FOR BREAST CANCER (Note. This topic is new to the guideline)
2007 recommendation for cyclin E. Present data are insufficient to recommend use of whole length or fragment measurements of cyclin E for management of patients with breast cancer.

Cyclin E: Marker definition. Cyclin E is a 50-kd protein expressed in the late G₁ phase of the cell cycle. Association of cyclin E with CDK2 stimulates kinase activity and promotes transition of cells to the S phase, ensuring subsequent cell division by phosphorylating the Rb protein that then releases bound E2F transcription factors and promotes DNA synthesis. Activity of the cyclin E-CDK2 enzyme complex is inhibited by the p21 and p27 proteins. Elevated levels of cyclin E have been observed in a number of different cancers.¹⁹⁵

In breast cancers, cyclin E is cleaved to lower molecular weight (LMW) fragments (33 to 45 kd) by elastase¹⁹⁶ and by calpain 2.¹⁹⁷ These LMW fragments have greater affinity for CDK2 and resist inhibition by p21 and p27.¹⁹⁸ In addition, the LMW fragments confer resistance to tamoxifen and increase genomic instability.¹⁹⁹ Consequently, there is a biologic rationale for evaluation of cyclin E protein, and particularly its LMW fragments, as a marker of poor prognosis in breast cancer.

Cyclin E: Methodology. Intact cyclin E protein has been measured by IHC in formalin-fixed paraffin-embedded (FFPE) tissue, and mRNA for cyclin E has been quantitated by RT-PCR in fresh frozen specimens.²⁰⁰ LMW forms of cyclin E have been measured by Western blot analysis of proteins in fresh frozen tissue.²⁰¹ Discordance in the prognostic value of cyclin E between IHC and Western blot analysis may be related to the antibodies used for each assay, given that the reagents that detect intact cyclin E may not react with the LMW fragments. Even when antibodies recognize the intact protein and its fragments, however, discordance between IHC and Western blots analysis has been observed in 37% of cases.²⁰¹ In a single study, dramatic results regarding use of cyclin E and outcome were reported only when the LMW fragments were considered and the assay for these was performed by Western blotting.²⁰¹ However, Western blotting is relatively impractical for routine clinical use, and the antibody used in this study cannot be applied successfully to FFPE issue. Monoclonal antibodies are needed to advance studies of this marker in archived tissue and to make its use in routine clinical practice possible.

Cyclin E: Literature review and analysis. Conclusions regarding the prognostic value of cyclin E in the published literature are mixed, perhaps in part due to methodologic differences in the assays (IHC v Western blotting) and due to lack of high-level studies. In addition, cyclin E is closely linked to proliferation and its independent prognostic significance is less clear. Nonetheless, elevated levels of cyclin E protein have been fairly consistently associated with a poor prognosis in breast cancer. In a recent meta-analysis of cyclin E overexpression of 2,534 patients in 12 published studies, overexpression of cyclin E was associated with a 2.32-fold (95% CI, 1.25- to 4.30-fold) increased risk of recurrence in univariate analysis and a 1.72-fold (95% CI, 0.95- to 3.10-fold) risk of recurrence in multivariate analysis.²⁰² In addition, the combined hazard ratio estimate for overall survival and breast cancer–specific survival was 2.98 (95% CI, 1.85 to 4.78) and 2.86 (95% CI, 1.85 to 4.41) in univariate and multivariate analysis, respectively. In a recently published paper in which all patients received one of two regimens of adjuvant doxorubicin and cyclophosphamide in a prospective Southwest Oncology Group randomized clinical trial (SWOG 9313), cyclin E overexpression, as determined by IHC for the full-length protein, was not associated with a worse outcome.²⁰³ However, the negative results of this study must be considered carefully because all of these patients received chemotherapy and the assay was not specific for cyclin E fragments.

Substantially higher prognostic value has been reported when both the LMW fragments of cyclin E and the intact molecule are considered together.²⁰¹ In a single-institution, retrospective study using archived frozen specimens analyzed by Western blot assay, the hazard ratio for death from breast cancer for patients with high total cyclin E levels, as compared with those with low total cyclin E levels on Western blot analysis, was 13.3—about eight times as high as the hazard ratios associated with other independent clinical and pathological risk factors. Although these data are promising, they are from a retrospective study, and additional properly designed studies are required to ascertain whether this marker has clinical utility, especially in the setting of no adjuvant chemotherapy.

PROTEOMIC ANALYSIS FOR BREAST CANCER (Note. This topic is new to the guideline)
2007 recommendation for proteomic analysis. Present data are insufficient to recommend use of proteomic patterns for management of patients with breast cancer.

Proteomic analysis: Marker definition. The emerging field of proteomics is complex. In theory, different clinical states, including cancer, might be represented by distinct protein patterns, or signatures. These signatures might consist of completely different proteins, of various mixtures of truncated peptide fragments, or of modifications of proteins or peptides, such as glycosylation, cysteinylation, lipidation, and glutathionylation, each of which might be cancer specific. Therefore, one might be able to exploit these differences, either in tissue, in the circulation, or in secreted fluids, for diagnostic purposes. For proteomic pattern analysis, computer-based algorithms have been developed to distinguish breast cancer from benign disease, or to identify individuals at high risk of recurrence based on the pattern of peptide peaks. An alternative method uses proteomic methods to identify a limited number of proteins that can be measured by immunohistochemical or serum-based immunoassays. Markers can then be validated individually or in combination as a profile or signature.

Proteomic pattern analysis: Methodology. There are several different approaches to analyzing multiple proteins or peptide fragments simultaneously, and each has its positive and negative features.²⁰⁴ These methods include multiplex ELISA, phage display, and aptamer arrays.^205-207 However, the most widely studied methods involve identification of proteomic profiles as peaks on mass spectometric analysis with precise charge-to-mass ratios. In some cases, proteins have been designated by their apparent molecular weight and isoelectric point within two-dimensional (2D) gel analysis. Specific peptides can be identified further based on their amino acid sequence identity or homology to known proteins or their fragments. Peptides have been identified in serum from breast cancer patients²⁰⁸; drug-resistant breast cancer cell lines²⁰⁹; cancer cell line membranes²¹⁰; nipple aspirate fluid (NAF)²¹¹; and normal, benign, premalignant, and malignant tumor tissue.^212,213 For analysis of breast cancers, some studies have used whole tumor specimens that include both epithelial cells and stroma, whereas others have used microdissected epithelial cells. If isolation of epithelial cells is not required, fine-needle aspirate has obtained adequate material.²¹⁴ Before mass spectoscopic analysis, preliminary separation of proteins can be performed with 2D gel analysis^211,215 or by binding of proteins to surfaces or matrices using surface-enhanced laser desorption and ionization (SELDI)^207,208,214 and matrix-associated laser desorption and ionization (MALDI),²¹⁵ respectively. After desorption and ionization, the pattern of charged peptides generally has been analyzed by time-of-flight (TOF) mass spectoscopy. Other methodologies to examine multiple proteins at once have used multiplex ELISAs that can detect several different proteins simultaneously.²¹⁶ Similar assays using phage displays or aptamers to detect multiple peptides have also been reported.^205,206

Proteomic pattern analysis: Literature review and analysis. During the period 1996 to December 2007, more than 200 articles have been published addressing proteomics and breast cancer. However, many of these are primarily methods articles, and those that do address clinical utility are retrospective in design at best.

SELDI-TOF has been used to profile proteins in serum or plasma from breast cancer patients. Several studies have addressed the potential of SELDI to provide serum biomarkers that differentiate breast cancer from benign disease and/or healthy individuals.^208,217-219 Enrolling between 133 and 169 patients, these studies have identified diagnostic protein profiles with sensitivities and specificities of 76% to 93% and 90% to 93%, respectively. Protein peaks that distinguished healthy women from those with cancer were found at m/z 4,300 and 8,900 in two studies, respectively. However, no protein identification was provided. It is apparent from studies that perform protein identification that the majority of serum proteins identified that differentiate patient and normal samples are host-specific proteins in high abundance.^220,221 New methods that allow isolation of low abundance serum proteins more likely to represent tumor markers are in development.^222,223

Given that a more concentrated source of protein from breast cancer ducts may be better able to identify tumor-specific markers, attention has been paid to the proteomic analysis of NAFs or ductal lavage fluid. When 2D gel electrophoretic separation and MALDI-TOF analysis of NAF were used, gross cystic disease fluid protein-15 levels were lower (P < .001) and alpha-1-acid glycoprotein levels were higher (P < .001) in 52 breast cancer fluids than in 53 nipple aspirates from benign lesions.²¹⁴ When subset analysis was performed, significant differences in levels for the two markers were observed in premenopausal but not in postmenopausal women. However SELDI-TOF analysis failed to detect differences in NAF from the breast with unlilateral early-stage (I-II) cancer and NAF from the contralateral breast.²²⁴ When fluid from the cancer-bearing breast was compared with NAF from healthy volunteers, 17 peaks were overexpressed in fluid from breast cancer patients (P < .0005). Isotope-coded affinity tag (ICAT) tandem mass spectrometry (MS) permits both qualitative and quantitative analysis of paired protein samples.

In a third study, NAF from tumor bearing and contralateral disease-free breasts of patients with unilateral early-stage breast cancer were analyzed using ICAT labeling, sodium dodecyl sulfate–polyacrylamide gel electrophoresis, liquid chromatography, and MS.²²⁵ Alpha2HS-glycoprotein was underexpressed in NAF from tumor-bearing breasts, whereas lipophilin B, beta globin, hemopexin, and vitamin D–binding protein precursor were overexpressed. Western blot analysis of pooled samples of NAF from healthy volunteers versus NAF from women with breast cancer confirmed the overexpression of vitamin D–binding protein in tumor-bearing breasts. Finally, analysis of NAF obtained preoperatively from 114 women and analyzed by SELDI-TOF indicated that three proteins (5,200-H4, P = .04; 11,880-H4, P = .07; and 13,880 Da-SAX, P = .03) were differentially expressed in women with versus those without breast cancer.²²⁶ Although of interest, these studies are all very preliminary. They are hampered by their retrospective design, and the frequent use of incongruent controls. Currently, none would lead to a clinical change in patient management.

At a tissue level, differences in protein profiles have been found between DCIS and normal ductules.²²⁷ Similarly, protein profiling in small numbers of samples with 2D gel electrophoretic separation and MALDI-TOF demonstrated differential expression of several proteins between a fraction of infiltrating ductal carcinomas and normal breast tissue, including gelsolin, vinculin, lumican, ₁-antitrypsin, heat shock protein-60, cytokeratin-18, transferrin, enolase-1,and -actin.²²⁸ Of this group, only heat shock protein-70 (more abundant) and peroxiredoxin-2 (less abundant) displayed the same trend in all of the infiltrating ductal carcinomas examined.

Few published studies have addressed the prognostic significance of protein profiles from breast cancer tissue. Jacquemier et al²²⁹ used IHC on tissue microarrays to profile the expression of 26 selected proteins in more than 1,600 cancer samples from 552 consecutive patients with early breast cancer. Supervised cluster analysis identified a set of 21 proteins whose combined expression significantly correlated with metastasis-free survival (MFS) in a learning set of 368 patients (P < .0001) and in a validation set of 184 patients (P < .0001). Among the 552 patients, the 5-year MFS was 90% for patients classified in the "good-prognosis class" and 61% for those classified in the "poor-prognosis class" (P < .0001). This difference remained significant when the molecular grouping was applied according to lymph node or ER status, as well as the type of adjuvant systemic therapy. In multivariate analysis, the 21-protein set was the strongest independent predictor of clinical outcome. Other studies using analysis of multiple protein biomarkers on tissue microarray have identified subclasses of breast cancer with clinical implications.^230-232 However, these studies are confounded by differences in populations, reagents and analysis methods, and systemic treatments, and therefore the Update Committee is unable to draw conclusions regarding clinical utility of any of these assays. Nevertheless, these studies illuminate the heterogeneity of breast cancer and bring us closer to understanding the relevant subclasses. In summary, these promising results, for the most part, are derived from retrospective studies and require additional confirmation in larger and well-designed prospective studies. At present, none of the proteomic profiling techniques has been validated sufficiently to be used for patient care.

MULTIPARAMETER GENE EXPRESSION ANALYSIS FOR BREAST CANCER (Note. This topic is new to the guideline)
2007 recommendation for multiparameter gene expression analysis. In newly diagnosed patients with node-negative, estrogen receptor–positive breast cancer, the Oncotype DX assay (Genomic Health Inc, Redwood City, CA) can be used to predict the risk of recurrence in patients treated with tamoxifen. Oncotype DX may be used to identify patients who are predicted to obtain the most therapeutic benefit from adjuvant tamoxifen and may not require adjuvant chemotherapy. In addition, patients with high recurrence scores (RSs) appear to achieve relatively more benefit from adjuvant chemotherapy (specifically [C]MF) than from tamoxifen. There are insufficient data at present to comment on whether these conclusions generalize to hormonal therapies other than tamoxifen, or whether this assay applies to other chemotherapy regimens. The precise clinical utility and appropriate application for other multiparameter assays, such as the MammaPrint assay (Agendia BV, Amsterdam, the Netherlands), the so-called Rotterdam Signature, and the Breast Cancer Gene Expression Ratio are under investigation.

Gene expression array analysis: Definition. Gene expression profiling recently has been introduced into the clinical literature during the last decade as research suggests that assessing the expression of multiple genes in a tumor sample may provide useful information about tumor behavior.^233,234 These molecular signatures hold the promise for improving diagnosis, for the prediction of recurrence, and in aiding selection of therapies for individual patients. Molecular classification has identified subtypes of breast cancer that are known to be present based on clinical experience. Among the categories are ER-positive and/or PR-positive tumors and HER2 gene-amplified tumors, both of which exhibit characteristic transcriptional profiles. In addition, a category of breast cancer termed "basal-like" due to the expression of basal keratins (CK5, CK14, CK15, and CK17) has emerged from these studies.^233-238 These tumors characteristically lack ER, PR, and HER2, although some controversy exists about the HER2 element. Furthermore, basal-like tumors often exhibit p53 mutation and low expression of BRCA1 (breast cancer associated 1) tumor suppressor genes, and this phenotype is common among BRCA1 carriers and sporadic triple-negative tumors.²³⁹ The literature surrounding gene expression profiling continues to debate the existence of such molecular subtypes and, if they do exist, the exact definitions of these subtypes. Nevertheless, many clinical trials are now designed to subdivide patients by ER/PR and HER2 status to validate claims that different groups of tumors may be more homogeneous and therapeutic approaches should address these groups rather than the population of breast cancer patients as a whole. At this time, the following profiling platforms have made their way to clinical practice and will be discussed further.

Gene expression array analysis: Methodology. Several technologies have been developed to generate molecular signatures, including cDNA and oligonucleotide arrays and multiplex PCR technologies. An early