Nomograms Predicting Locoregional Recurrence in the Subtype Era of Breast Cancer

Abstract

TO THE EDITOR: The recently published article by Albert et al 1 and the editorial by Wazer 2 were of significant interest to us. We agree with the utility of nomograms that “account for all of the relevant variables associated with local failure” 2(p2810) in making evidence-based treatment decisions. An ideal nomogram should alsoincorporateclinicallymeaningfulparameters,reflectingaprofound understanding of the disease. The discovery that breast cancer no longer represents a unique disease 3,4 has not been yet applied to locoregional (LR) treatment decisions, although emerging evidence suggests that breast cancer subtypes have unique qualities in terms of LR recurrence (LRR) patterns. 5 In this analysis database, mainly patients with nonhigh-grade (77.3%), estrogen receptor (ER) ‐positive (86.3%) tumors were included. Negative ER status conferred the highest risk (hazard ratio [HR], 2.27) of a subsequent mastectomy in the future. These results could represent an increased possibility of LRR rates in women with triple-negative or human epidermal growth factor receptor 2 (HER2) ‐positive tumor subtypes. Conversely, more than two thirds of the population had tumors that were possibly luminal A or B. (Given that HER2 status is unknown, one can only speculate on the basis of the grade.) In retrospective evaluations of LRR of the luminal A subtype, a 10-year LRR rate of 3% to 8% has been shown 6 and confirmed in other studies (5-year LRR rate, 2.3% 7 ; 10-year LRR rate, 3.6% 8 ; cumulative incidence of 5-year LRR rate, 0.8% [95% CI, 0.3 to 2.2%] 9 ) after breast-conserving surgery (BCS) and adjuvant radiotherapy (ART), 10 where hormone therapy was underused. Therefore, the LRR rates for the luminal A subtype range 0.8% and 8%. Concerning the luminal B subtype, represented in that study by the patients with ER-positive, high-grade, and partly intermediate-grade tumors, 5-year LRR rates of 7%, 6 1.5%, 9 and 4.3% (10-year LRR rate, 8.7%) 8 have been described. Therefore, the LRR rate for the luminal B subtype ranges between 1.5 and 8.7%. The LRR rate of the luminal subtype (luminal A and B) has been approximated at 5% after BCS and adjuvant radiotherapy (ART), but with underuse of appropriate hormone therapy. 11 Results of this retrospective review 1 fit with this model. Mastectomy rates differed between patients who received ART (89.4% of patients) and those who did not (10.6% of patients), with 5-year mastectomy-free-survival rates (MFSR) of 98.4% in those who received ART versus 5-year MFSR of 95.0% in those who did not. (The projected10-yearMFSRwas95.4% v89%forthosewhoreceivedART compared with those who did not, respectively.) If one interprets MFSRasfreedomfromLRR,thenpatientswithmostly(butnotonly) luminal tumors had 5-year freedom from LRR rates of 2.6% and 10-year freedom from LRR rates of 4.6%. HER2 status is a known prognosticator of LR failure (10-year LRR rate,5%to16% 6 ;5-yearLRRrate,8.4% 9 ;5-yearLRRrate,4.6% 7 ;10-year LRR rate, 7.7%, and 10-year LRR rate, 15.3% 8 ). Therefore, LRR rates for theHER2-positivesubtyperangebetween4%and15%.Inthesestudies, trastuzumab,whichconfersanadvantageinLRcontrol,wasnotused becauseithadnotyetbeenintroducedintoclinicalpractice.Thus,the LRR of the HER2-positive subtype, if appropriately treated, remains unknown, but in this cohort we could anticipate raw rates of recurrencecomparablewiththestudiesmentioned,giventhattrastuzumab was likely not used. Finally, a portion of the included patients might have harbored the triple-negative subtype, which is described to represent10-yearLRRrateof14%, 6 5-yearLRRrateof7.1%, 9 5-yearLRR rate of 3.2%, 7 10-year LRR rate of 9.6%, and 10-year LRR rate of 17.3%. 8 Therefore, LRR rates for the triple-negative subtype range between 3% and 17%. It has been argued elsewhere by the same group that HER2 positivityconfersanHRof3.13forLRR,andERnegativityconfersan HRof2.37 12 inpatientswithtumorsupto1cmwhohavenotreceived any adjuvant trastuzumab or chemotherapy. In this cohort, mastectomy (37.7%) or BCS (62.3%) was undertaken with no difference in LRR rates. One could then anticipate that mastectomies in the nomogram cohortmusthavebeenmostlyperformedinthosepatientswithunfavorablesubtypes,suchasHER2-positive,triple-negative,orluminalB tumors. Incorporating this information into nomograms could improve adjuvant treatment decisions. We believe that the effect of radiotherapy after BCS in these elderly women is lost for the most aggressive subtypes and is overestimated for the less aggressive ones. In conclusion, given the impossibility of properly characterizing the exact disease subtype of this enormous retrospective cohort of patients, we advocate that biologic factors should not be merged togetherandshouldbeincludedinthisnomogram(ERstatus,grade).