Methods

We conducted an Institutional Review Board–approved, retrospective, single-institution analysis of women diagnosed with breast cancer at age ≤40 years and treated between 1991 and 2015. A total of 155 patients were identified, 40 and 115 of whom were diagnosed and treated between 1991–2004 and 2005–2015, respectively. All patients received NAC and surgery with or without radiation therapy (RT). Pathologic confirmation of axillary nodal involvement by fine-needle aspiration was not required for inclusion in our analysis. Patients were excluded if surgical or RT records were not available at the time of data analysis, if they had metastatic disease at diagnosis, or if they received chemotherapy for other cancer diagnoses. pCR was defined as no evidence of residual invasive tumor in the breast and sampled axillary LNs (ypT0/Tis ypN0). All other patients were considered to have partial response if there was no tumor in the LNs but residual tumor in the breast (ypT+ ypN0) or no tumor in the breast but residual tumor in the LNs (ypT0/Tis ypN+). Those who had residual tumor in both the breast and LNs were classified as having a limited response (ypT+ ypN+). Residual cancer burden was not routinely assessed in all pathology samples, and therefore was not included in this analysis.

All patients included in this study had detailed clinical and pathologic information available for review, such as clinical stage at diagnosis, hormone receptor (HR) status, HER2 status, Ki67 percentage, BRCA1/2 mutation status, type of NAC, use of HER2-targeted therapy and hormonal therapy, size of residual disease, LN involvement, presence of lymphovascular invasion, and achievement of pCR. Local (LF), regional (RF), and distant (DF) failure were recorded for each patient.

All statistical analysis was conducted using SAS version 9.4 (SAS Institute Inc.). Competing risks analysis was performed using the Fine-Gray method to analyze cumulative incidence of LF, RF, and DF for each site; hazard ratios were reported with a 95% CI, and OS was calculated using Kaplan-Meier and log-rank tests. Multivariable analysis was performed with Cox proportional hazards regression using select prognostic parameters identified as significant on competing risks analysis. Statistical significance was determined as a 2-sided P value <.05.

Results

A total of 155 patients were included in this analysis. Median age was 36 years (range, 20–40 years), and median follow-up time was 52 months (range, 6–293 months). The median OS for the entire cohort was 122 months. Table 1 details the clinical characteristics. In brief, 97 patients (62.5% of the entire cohort) had genetic test results available for BRCA, of which 25.8% (n=25) carried BRCA1 mutations and 7.2% (n=7) carried BRCA2 mutations. Initial clinical stage at diagnosis was stage I in 4%, stage II in 45%, and stage III in 46%. All patients received NAC followed by mastectomy (n=108; 69.7%) or lumpectomy (n=47; 30.3%). A total of 132 (85.2%) patients received adjuvant RT; 42 of 47 (89.4%) after lumpectomy (5 refused) and 90 of 108 (83.3%) after mastectomy.

After NAC, 39 patients (25.2%) achieved pCR (ypT0/Tis ypN0), 52 (33.5%) achieved ypT+ ypN0, and 5 (3.2%) achieved ypT0/Tis ypN+. The remaining

Table 1.

Patient Characteristics (N=155)

View Table

A total of 22 patients (14.2%) experienced LF, 20 (12.9%) experienced RF, and 59 (38.1%) experienced DF. LRR occurred among 22 of 108 patients (20.4%) who underwent mastectomy and 8 of 47 (17.0%) who underwent lumpectomy, and DF occurred in 46 of 108 of patients (43.6%) who underwent mastectomy and 13 of 47 (27.6%) who underwent lumpectomy. No local or regional recurrences occurred in patients who achieved pCR. A total of 3 patients (7.7%) with pCR developed distant metastases, 2 of whom were still alive with disease at time of writing, and 1 who died 96 months after the initial diagnosis. However, patients with residual disease in either the breast or LNs had a higher incidence of LF, RF, and DF than those who achieved pCR (Figure 1), with approximately 50% experiencing DF at 10 years (P<.001; Table 2). Of those who achieved a pCR in the LNs, patients who still had residual disease in the breast did not have better outcomes compared with those who remained LN-positive. Cumulative incidence of LF, RF, and DF at 10 years by response to NAC is depicted in Table 2, with corresponding cumulative incidence curves in Figure 1. Median OS for patients with pCR was not reached (P<.001). No significant difference in OS was seen between ypT+ ypN0 versus ypN+ patients, with median OS of 97 versus 83 months, respectively (P=.25) (Figure 2A).

To evaluate whether modern systemic therapy affected OS, we stratified patients into 2 groups based on treatment date and chose a cutoff of 2005, because this was when the use of trastuzumab for HER2-positive breast cancer had become standard of care. On Kaplan-Meier analysis, no difference in OS was seen among patients treated between 1991–2004 and 2005–2015 (median OS, 122 months vs not reached, respectively; P=.88). We next analyzed patients treated between 2005–2015 (n=116) separately to evaluate outcomes in a patient population with routine HER2 testing and regular use of trastuzumab. Patients with ypT+ ypN0 and ypN+ disease continued to have worse OS compared with those who achieved pCR, with median OS not reached for the pCR group, and 84 versus 77 months, respectively, for patients with ypT+ ypN0 and ypN+ disease (P<.001; Figure 2B). We also evaluated outcomes of patients with an initial diagnosis of clinical T1–3N1 breast cancer (n=115) who achieved pCR in the LNs (ypT0/Tis ypN0 and ypT+ ypN0) following NAC, because this group is included in the ongoing NSABP B-51 trial. We found that outcomes of patients who achieved pCR were excellent, with median OS not reached (P=.03; Figure 2C); however,

View Full Size

Cumulative incidence of (A) local, (B) regional, and (C) distant failure by pathologic nodal response.

Abbreviations: pCR, pathologic complete response; ypN+, residual tumor in the lymph nodes; ypT+ ypN0, no tumor in the lymph nodes but residual tumor in the breast.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 16, 7; 10.6004/jnccn.2018.7022

Lastly, we investigated the impact of breast cancer subtype on outcomes. Patients were divided into 4 groups: HR-positive/HER2-negative (n=60; 39.2%), HR-positive/HER2-positive (n=21; 13.7%), HR-negative/HER2-positive (n=17; 11.1%), and TNBC (n=55; 35.9%); subtype was not available for 2 patients. pCR by breast cancer subtype was 18.3%

Table 2.

Ten-Year Cumulative Incidence of Local, Regional, and Distant Failure

View Table

Discussion

To our knowledge, this is the first study evaluating outcomes by pathologic nodal response in women aged ≤40 years treated with NAC. We focused on young women treated with NAC because this population is infrequently studied on its own, yet has worse survival compared with older women despite higher rates of pCR. In addition, although the secondary analysis of NSABP B-18 and B-27¹⁰ has been used to guide risk assessment following NAC, it does not include a comprehensive analysis of recurrence risk in women aged ≤40 years. Our report presents LF, RF, and DF rates in this high-risk breast cancer population, making this analysis unique with respect to the published literature.

Our analysis reveals that women aged ≤40 years with residual disease in the breast or LNs have increased LRR rates compared with those who achieve pCR in both the breast and LNs. Perhaps the most compelling finding of our analysis is that women with residual disease in the breast, despite having a pCR in the LNs, had similar LRR rates as those with residual nodal disease. This important finding suggests that omitting RT in patients with any residual disease after NAC, whether in the breast alone or in the breast and LNs, should be considered with caution. The LRR rates remained between 15% and 22% in patients with ypT+ ypN0 and ypN+ disease, even though most patients in our study received

View Full Size

Overall survival for (A) entire cohort, (B) patients treated from 2005–2015, and (C) patients with initial diagnosis of clinical T1–3N1 breast cancer.

Abbreviations: pCR, pathologic complete response; ypN+, residual tumor in the lymph nodes; ypT+ ypN0, residual tumor in the breast but no tumor in the lymph nodes.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 16, 7; 10.6004/jnccn.2018.7022

Table 3.

Multivariable Analysis of Prognostic Variables for Overall Survival

View Table

A large percentage of our cohort (38.1%) experienced DF, driven in large part by women who did not achieve pCR following NAC. This high rate of DF is also likely due to most of these women (≈90%) having stage II–III disease at diagnosis, making our patient population particularly high risk. A retrospective analysis of 170 women aged ≤40 years treated with NAC also observed a high rate of recurrence (29.4%) and mortality (22.9%).¹⁵ It is likely that these women would benefit from adjuvant treatment to improve outcomes, and several trials have evaluated escalation of adjuvant therapy.^16–18 The recently published CREATE-X trial¹⁹ evaluated 910 patients with HER2-negative breast cancer who had residual invasive disease or LN metastases after NAC. Women were randomized to either adjuvant capecitabine or observation. At 5 years of follow-up, the addition of adjuvant capecitabine significantly prolonged DFS and OS. Other efforts to improve outcomes center around increasing rates of pCR through the escalation of NAC regimens.^20,21 In the 5-year follow-up of the NeoSphere trial,²² the addition of pertuzumab to trastuzumab + docetaxel improved long-term outcomes in women with HER2-positive breast cancer. This led to the accelerated FDA approval of pertuzumab in the neoadjuvant setting.²³

Our analysis shows that young women with breast cancer who achieve a pCR in both the breast and the LNs experience improved outcomes, whereas those who have residual disease in the breast, despite having a pCR in the LNs, do not have better outcomes compared with those who remain LN-positive after NAC. The strengths of this study include a detailed assessment of pathologic response and patterns of failure, as well as stratification of patients based on the time frame during which they received treatment (1991–2004 vs 2005–2015), clinical stage at diagnosis, and breast cancer subtype. Our analysis differs from previously reported studies because it assesses LRR, as well as DF and OS for all patients included in the analysis. Limitations of our study include its retrospective nature and relatively small sample size. As a single-institution study, our results may not be generalizable to the population at large, but are hypothesis-generating nonetheless.

Conclusions

Our analysis demonstrates that outcomes in young women aged ≤40 years receiving NAC vary by pathologic response in the breast and the LNs. Young women who achieve pCR have excellent outcomes; however, those who achieve pCR in the LNs but have residual disease in the breast continue to have outcomes similar to those who remain LN-positive. Our data suggest that de-escalation of care in this patient population should be approached with caution until the results of NSABP B-51 are available.