EXPERT PERSPECTIVES

Pushing the Envelope: Should We ExtendGenomic Profiling to Node-PositiveDisease?

Eleftherios P. Mamounas, MD, MPH, and JessicaF. Partin, MD

Gene-expression profiling has emerged as a useful tool forassessing risk of recurrence in patients with early-stage breastcancer.1-7 Several gene expression signatures have beenfound to significantly predict risk of distant recurrence bothin untreated patients and in those treated with endocrinetherapy and/or chemotherapy.1-8 A number of commerciallyavailable multi-gene assays have been validated in node-negative breast cancer patients,3-5,9,10 and some have alreadybeen incorporated into practice guidelines11,12 and consensuspanel recommendations.13

Early evaluation of one of these multi-gene assays(21-gene recurrence score [RS], Oncotype DX) in the contextof a randomized clinical trial comparing tamoxifen alonewith tamoxifen plus chemotherapy demonstrated that theassay was not only prognostic of outcome but also predictiveof chemotherapy benefit, ie, patients with a low RS (< 18) didnot benefit from the addition of chemotherapy to tamoxifen,whereas a clinically and statistically significant benefit wasdemonstrated in patients with a high RS (> 30). This obser-vation was pivotal for the widespread clinical adoption ofthis assay and provided the rationale for extending the eval-uation of this and other multi-gene assays to patients withnode-positive disease.

LITERATURE REVIEW

Although prognostic discrimination in node-positive diseasewas demonstrated early on during the development and vali-dation of some of the existing multi-gene assays,2,5 the firstsolid demonstration of the potential value of multi-geneassays in the node-positive setting came from the evaluationof the 21-gene RS in node-positive, estrogen receptor (ER)–positive, postmenopausal breast cancer patients who partici-pated in the Southwest Oncology Group (SWOG) 8814trial.14 This randomized phase III trial compared adjuvanttamoxifen with adjuvant tamoxifen plus cyclophosphamide,doxorubicin, and fluorouracil (CAF) chemotherapy (giveneither concurrently or sequentially) and demonstrateda significant improvement in outcomes with tamoxifen plus

chemotherapy versus tamoxifen alone. Furthermore, thesequential administration of chemotherapy and tamoxifenwas superior to the concurrent administration. Investigatorsfrom that trial evaluated whether the RS was prognostic inwomen treated with tamoxifen alone and whether it couldidentify those who might not benefit from CAF, despitea higher risk of recurrence. A total of 367 patients from thetamoxifen-alone and the CAF before tamoxifen (CAF-T)groups had sufficient RNA for evaluation of the RS (tamox-ifen, n¼ 148; CAF-T, n¼ 219). The results of this studyconfirmed previous observations in patients with node-negative disease that the RS was prognostic in thetamoxifen-alone group and that it was predictive of chemo-therapy benefit.

Similar to the findings from the National SurgicalAdjuvant Breast and Bowel Project (NSABP) B-20 trial,15

the evaluation of data from the SWOG 8814 trial showedno benefit from CAF in patients with a low RS (RS < 18;log-rank P¼ 0.97; hazard ratio [HR]¼ 1.02 [95% CI,0.54-1.93]), but there was a significant improvement indisease-free survival with the addition of CAF in thosewith a high RS (RS > 31; log-rank P¼ 0.033; HR¼ 0.59[95% CI, 0.35-1.01]), after adjustment for the number ofpositive nodes. The RS by treatment interaction was signifi-cant in the first 5 years (P¼ 0.029), with no additionalprediction beyond 5 years (P¼ 0.58), although the cumula-tive benefit remained at 10 years. Results were similar foroverall survival and breast cancer–specific survival (BCSS).On the basis of these findings, the authors concluded thatthe RS was prognostic in tamoxifen-treated patients withpositive nodes and also predictive of benefit from CAF intumors with a high RS. In contrast, low RS identifiedwomen who might not benefit from anthracycline-basedchemotherapy, despite having positive nodes.

Given the small number of patients with node-positivedisease who had a low RS in this study, one can questionwhether these data alone support the use of the RS for select-ing which node-positive patients are candidates for adjuvantchemotherapy. However, the SWOG 8814 results are quiteconcordant with those from the NSABP B-20 trial in node-negative patients and support the concept that patients witha low RS do not appear to derive much benefit from chemo-therapy, whether they are node negative or node positive.

Additional information to support the independent prog-nostic contribution of multi-gene assays in node-positivepatients has been obtained in other clinical settings with

Breast Diseases: A Year Book� Quarterly 201Vol 21 No 3 2010

both the 21-gene RS and the 70-gene prognosis signature(MammaPrint). Dowsett and colleagues8 investigated theprognostic contribution of the 21-gene RS on risk of distantrecurrence in tamoxifen- and anastrozole-treated patientsfrom the Arimidex, Tamoxifen, Alone or in Combinationtrial. In 1231 evaluable patients (872 node-negative and306 node-positive), the RS was significantly associatedwith time to distant recurrence in multivariate analyses(P < 0.001 for node-negative and P¼ 0.002 for node-positivepatients). The RS also provided significant prognosticdiscrimination above and beyond that provided by Adjuvant!Online (P < 0.001). Nine-year distant recurrence rates in thenode-positive group were 17%, 28%, and 49% for patientswith low, intermediate, and high RS, respectively. The prog-nostic value of the RS was similar in anastrozole- and tamox-ifen-treated patients. Thus, this study provided additionalconfirmation that the RS is an independent predictor ofdistant recurrence in a large, contemporary population ofpostmenopausal, hormone receptor–positive patients treatedwith tamoxifen or anastrozole.

The prognostic utility of the RS in node-positive patientswas also evaluated by Goldstein and colleagues16 in 465patients with hormone receptor–positive breast cancer and 0-3 positive nodes who participated in the Eastern CooperativeOncology Group E2197 trial, a randomized trial that compared4 cycles of doxorubicin/cyclophosphamide (AC) with 4 cyclesof doxorubicin/docetaxel (AT). The RS was a highly signifi-cant predictor of recurrence in both the node-negative andnode-positive groups (P < 0.001 for both) and when adjustedfor other clinical variables. The RS also predicted recurrencemore accurately than clinical variables when integrated intoan algorithm modeled after Adjuvant! and adjusted to 5-yearoutcomes. Although all patients in this study were treatedwith 4 cycles of adjuvant chemotherapy, the results suggestthat the 21-gene assay may be used to select low-risk patientsfor abbreviated chemotherapy regimens or high-risk patientsfor more aggressive regimens or clinical trials evaluatingnovel treatments.

Mook and colleagues17 investigated whether the 70-geneprognosis signature could accurately identify patients with1-3 positive lymph nodes who had an excellent diseaseoutcome. Frozen tumor samples from 241 patients withoperable T1-3 breast cancer and 1-3 positive axillary lymphnodes were selected from 2 institutions and were analyzedfor the 70-gene tumor expression signature. The 10-yeardistant metastasis–free survival and BCSS probabilitieswere 91% (€ 4%) and 96% (€ 2%), respectively, forthe good prognosis–signature group (99 patients) and 76%(€ 4%) and 76% (€ 4%), respectively, for the poor prog-nosis–signature group (142 patients). The 70-gene signaturewas significantly superior to the traditional prognostic factors

202 Breast Diseases: A Year Book� Quarterly

Vol 21 No 3 2010

in predicting BCSS, with a multivariate HR of 7.17 (95% CI,1.81-28.43; P¼ 0.005). In an exploratory analysis ofoutcomes according to the 70-gene prognosis signature andwhether patients received chemotherapy or not (per the treat-ing physician’s discretion and not by randomization), theauthors reported excellent outcomes in patients carryingthe good signature irrespective of whether they received adju-vant chemotherapy. Thus, this study also suggests—albeitindirectly—that there may be no benefit from adjuvantchemotherapy in patients with the good 70-gene prognosissignature and that adjuvant chemotherapy may be withheldin this group of patients.

In a more comprehensive analysis of the impact of the70-gene prognosis index on chemotherapy benefit, Knauerand colleagues18 evaluated 541 patients who received eitherendocrine therapy alone (ET, n¼ 315) or endocrine therapyplus chemotherapy (ET + CT, n¼ 226). The 70-gene signa-ture classified 252 patients (47%) as low risk and 289 (53%)as high risk. In the 70-gene low-risk group, the BCSS ratewas 97% for the ET group and 99% for the ET + CT groupat 5 years, with a non-significant univariate HR of 0.58 (95%CI, 0.07-4.98; P¼ 0.62). In the 70-gene high-risk group, theBCSS rate was 81% in the ET group and 94% in theET + CT group at 5 years, with a significant HR of 0.21(95% CI, 0.07-0.59; P < 0.01). Distant disease–free survivalrates were 93% (ET) and 99% (ET + CT) in the 70-genelow-risk group (HR¼ 0.26 [95% CI, 0.03-2.02]; P¼ 0.20)and 76% (ET) and 88% (ET + CT) (HR¼ 0.35 [95% CI,0.17-0.71]; P < 0.01) in the high-risk group. Results weresimilar in multivariate analysis, showing a significant survivalbenefit to adding chemotherapy to endocrine therapy in the70-gene high-risk group. Although the benefit from chemo-therapy was not statistically significant in low-risk patients,there was no evidence of an interaction regarding the effectof chemotherapy according to risk group.

COMMENTARY

The data presented above strongly support the notionthat the 2 most commonly used genomic classifiers(21-gene RS and 70-gene prognosis index) are independentpredictors of outcome in patients with node-positive breastcancer in various clinical settings. These genomic classifierssignificantly enhance the prognostic ability of standard clini-copathologic variables such as those incorporated into theAdjuvant! algorithm.

What is less well established—because of limited avail-able information—is the ability of these assays to predictwith certainty benefit from chemotherapy in node-positivepatients. Only 1 study (SWOG 8814) involved a randomizedcomparison of endocrine therapy versus chemoendocrinetherapy according to the 21-gene RS, but it was limited by

the small number of patients included. Thus, the CIs aroundthe HRs for chemotherapy benefit in patients with low RS arewide, leaving us with uncertainty regarding the true relativebenefit from chemotherapy, or lack thereof. Although thisuncertainty in the relative reduction of risk of recurrencemay translate into a potentially small difference in absolutechemotherapy benefit in patients with a low number of posi-tive nodes (1-3), the potential absolute chemotherapy benefitincreases dramatically as the number of positive nodes (and,as a result, the risk of recurrence) increases. Thus, at thispoint, application of the RS to therapeutic decisions inpatients with positive nodes should be limited to patientswith a low number of positive nodes and/or a small tumorburden of nodal involvement (micrometastases). Ideally,a prospective, randomized clinical trial is the only way todefinitively address this question.

As mentioned above, additional support for the lack ofchemotherapy benefit identified by genomic classifiers inlow-risk, node-positive patients is provided by extrapolationfrom the randomized comparison of endocrine therapy versuschemoendocrine therapy in node-negative patients (RS in theNSABP B-20 trial). The Early Breast Cancer Trialists’Collaborative Group overview analysis has consistentlyshown no differences in the effectiveness of adjuvant chemo-therapy and adjuvant endocrine therapy according to nodalstatus. However, it should be noted that while the NSABPB-20 trial included premenopausal and postmenopausalnode-negative patients, SWOG 8814 included only postmen-opausal, node-positive patients, leaving a dearth of informa-tion on the effectiveness of chemotherapy according to theRS in the subgroup of premenopausal, node-positive patients.

Finally, another potential use for the various genomicclassifiers in patients with positive nodes may be realizedby their ability to predict risk for locoregional recurrence(LRR). Recently, we demonstrated that in node-negative,ER–positive, tamoxifen-treated patients (from the NSABPB-14 and B-20 trials) a significant association exists betweenthe 21-gene RS and risk for LRR, similar to that demon-strated between RS and distant recurrence in the samegroup of patients.19 Similar associations have also beenreported in smaller studies with other genomic classi-fiers.20-22 These results have biological implications andmay also have clinical implications for LRR therapy deci-sions for patients with node-negative, ER–positive breastcancer. However, as the evaluation of RS expands to node-positive patients, if an association between multi-gene assaysand risk of LRR is confirmed in this group, it may providevaluable information for the identification of subsets ofpatients with 1-3 or even with 4 or more positive nodes atlow versus high risk for LRR who may or may not needchest wall and/or regional radiotherapy.

In conclusion, we are witnessing an evolution in theunderstanding and clinical management of patients withnode-positive, ER–positive breast cancer. The introductionof genomic classifiers provides additional information thatcan be used to refine risk of recurrence above and beyondthat provided by traditional factors such as tumor size andthe number of positive nodes. It also appears that tumorbiology and benefit from adjuvant chemotherapy can bebetter predicted by the genomic classifiers than by the exist-ing prognostic and predictive factors. However, those find-ings require prospective validation in a clinical trial beforetheir widespread adoption. More challenges and opportuni-ties lie ahead.

References1. Sørlie T, Perou CM, Tibshirani R, et al. Gene expression

patterns of breast carcinomas distinguish tumor subclasseswith clinical implications. Proc Natl Acad Sci U S A. 2001;98:10869-10874.

2. van de Vijver MJ, He YD, van’t Veer LJ, et al. A gene-expression signature as a predictor of survival in breastcancer. N Engl J Med. 2002;347:1999-2009.

3. van’t Veer LJ, Dai H, van de Vijver MJ, et al. Geneexpression profiling predicts clinical outcome of breastcancer. Nature. 2002;415:530-536.

4. Foekens JA, Atkins D, Zhang Y, et al. Multicenter validationof a gene expression-based prognostic signature in lymphnode-negative primary breast cancer. J Clin Oncol. 2006;24:1665-1671.

5. Paik S, Shak S, Tang G, et al. A multigene assay to predictrecurrence of tamoxifen-treated, node-negative breast cancer.N Engl J Med. 2004;351:2817-2826.

6. Goetz MP, Suman VJ, Ingle JN, et al. A two-gene expressionratio of homeobox 13 and interleukin-17B receptor forprediction of recurrence and survival in women receivingadjuvant tamoxifen. Clin Cancer Res. 2006;12:2080-2087.

7. Jansen MP, Sieuwerts AM, Look MP, et al. HOXB13-to-IL17BR expression ratio is related with tumoraggressiveness and response to tamoxifen of recurrent breastcancer: a retrospective study. J Clin Oncol. 2007;25:662-668.

8. Dowsett M, Cuzick J, Wales C, et al. Risk of distantrecurrence using Oncotype DX in postmenopausal primarybreast cancer patients treated with anastrozole ortamoxifen: a TransATAC study (abstract 53). Cancer Res.2009;69:75s.

9. Buyse M, Loi S, van’t Veer L, et al; TRANSBIG Consortium.Validation and clinical utility of a 70-gene prognosticsignature for women with node-negative breast cancer. J NatlCancer Inst. 2006;98:1183-1192.

Breast Diseases: A Year Book� Quarterly 203Vol 21 No 3 2010

2

10. Ma XJ, Salunga R, Dahiya S, et al. A five-gene moleculargrade index and HOXB13:IL17BR are complementaryprognostic factors in early stage breast cancer. Clin CancerRes. 2008;14:2601-2608.

11. Harris L, Fritsche H, Mennel R, et al; American Society ofClinical Oncology. American Society of Clinical Oncology2007 update of recommendations for the use of tumormarkers in breast cancer. J Clin Oncol. 2007;25:5287-5312.

12. Carlson RW, Allred DC, Anderson BO, et al; NCCN BreastCancer Clinical Practice Guidelines Panel. Breast cancer.Clinical practice guidelines in oncology. J Natl Compr CancNetw. 2009;7:122-192.

13. Goldhirsch A, Ingle JN, Gelber RD, Coates AS,Thurlimann B, Senn HJ; Panel members. Thresholds fortherapies: highlights of the St Gallen International ExpertConsensus on the primary therapy of early breast cancer2009. Ann Oncol. 2009;20:1319-1329.

14. Albain KS, Barlow WE, Shak S, et al; Breast CancerIntergroup of North America. Prognostic and predictivevalue of the 21-gene recurrence score assay inpostmenopausal women with node-positive, oestrogen-receptor-positive breast cancer on chemotherapy:a retrospective analysis of a randomised trial. Lancet Oncol.2010;11:55-65.

15. Paik S, Tang G, Shak S, et al. Gene expression and benefit ofchemotherapy in women with node-negative, estrogenreceptor-positive breast cancer. J Clin Oncol. 2006;24:3726-3734.

04 Breast Diseases: A Year Book� Quarterly

Vol 21 No 3 2010

16. Goldstein LJ, Gray R, Badve S, et al. Prognostic utility of the21-gene assay in hormone receptor-positive operable breastcancer compared with classical clinicopathologic features.J Clin Oncol. 2008;26:4063-4071.

17. Mook S, Schmidt MK, Viale G, et al; TRANSBIGConsortium. The 70-gene prognosis-signature predictsdisease outcome in breast cancer patients with 1-3 positivelymph nodes in an independent validation study. BreastCancer Res Treat. 2009;116:295-302.

18. Knauer M, Mook S, Rutgers EJ, et al. The predictive value ofthe 70-gene signature for adjuvant chemotherapy in earlybreast cancer. Breast Cancer Res Treat. 2010;120:655-661.

19. Mamounas EP, Tang G, Fisher B, et al. Association betweenthe 21-gene recurrence score assay and risk of locoregionalrecurrence in node-negative, estrogen receptor-positivebreast cancer: results from NSABP B-14 and NSABP B-20.J Clin Oncol. 2010;28:1677-1683.

20. Cheng SH, Horng CF, West M, et al. Genomic prediction oflocoregional recurrence after mastectomy in breast cancer.J Clin Oncol. 2006;24:4594-4602.

21. Nuyten DSA, Chang HY, Sneddon JB, et al. Reproducibilityof several gene expression signatures in predicting outcomein breast cancer (abstract 36). Breast Cancer Res Treat. 2004;88:S22.

22. Voduc KD, Cheang MC, Tyldesley S, Gelmon K,Nielsen TO, Kennecke H. Breast cancer subtypes and the riskof local and regional relapse. J Clin Oncol. 2010;28:1684-1691.

The Unproven Assumptions of PreoperativeMRI for Patients Considering BreastConservation

Sarah A. McLaughlin, MD, and Tari A. King, MD

Magnetic resonance imaging (MRI) of the breast is being usedwith increasing frequency in patients at risk for or diagnosedwith breast cancer. Clinicians have incorporated breast MRIinto the practice of high-risk screening, staging of the ipsilat-eral breast, monitoring of treatment response to neoadjuvanttherapy, and problem-solving when standard imaging isequivocal. MRI screening guidelines exist for high-risk patientpopulations1 but not for patients with newly diagnosed breastcancer. Furthermore, the benefits of routine preoperativebreast MRI on patients’ outcomes are questionable, and thenegative consequences are potentially significant.

The potential benefit of preoperative MRI for patientsconsidering breast conservation lies in its sensitivity. It iswell documented that the sensitivity of MRI ranges from77% to 100%1 and that MRI can identify additional sites ofoccult cancer in the index breast, as well as synchronouscontralateral breast cancers, that may otherwise be missedon conventional imaging. In a recent meta-analysis of 2610women with newly diagnosed breast cancer, MRI detectedadditional cancers within the affected breast in 16% (inter-quartile range, 11% to 24%) of patients.2 It is for this reasonthat clinicians and patients pursue preoperative MRI, hopingthat it will accurately identify the extent of disease and, there-fore, aid in surgical planning, reduce the need for additionalsurgeries, minimize local recurrence, and improve patientoutcomes. However, an increasing body of data does notsupport this contention.

In theory, MRI should assist in surgical planning, asit accurately measures tumor size within 5 mm of the