Background

Historically, neoadjuvant chemotherapy (NAC) has been used to downstage locally advanced breast cancer, which by definition involves the skin, chest wall, or multiple axillary lymph nodes [1]. Subsequent studies have found similar outcomes in recurrence and survival for treatment of operable breast cancer in the neoadjuvant versus the adjuvant setting [2, 3]. These findings paved the way for NAC use in multiple settings: downstaging larger tumors and facilitating breast conservation therapy (BCT) [2, 4], downstaging the axilla to avoid axillary lymph node dissection (ALND) [5,6,7], allowing for response monitoring and potential switch to a non-cross-resistant regimen in non-responders [8,9,10], and expediting approval of new drugs using pathologic complete response (pCR) as an endpoint [11, 12].

Hormone receptor-positive breast cancer is the most common molecular subtype, comprising 70–80% of all breast cancer diagnoses [13]. It is recognized that HR+ HER2− tumors are less likely to achieve pCR with NAC than other biologic subtypes [4, 6]. Triple negative breast cancer (TNBC) and HER2+ breast cancer have higher rates of pCR, on the order of 30–60% with chemotherapy and HER2 receptor blocking agents, versus less than 20% of those with HR-positive disease [14]. With evolution of better targeted treatment regimens, particularly for TNBC and HER2+ cancers, NAC use has expanded in these subtypes [15].

For patients with early-stage HR+ HER2− tumors, primary surgery rather than NAC is more common given that those patients are likely to have overall better prognosis than other phenotypes, significant response to endocrine therapy, and typically poor response to chemotherapy. In treating hormone receptor-positive breast cancer NAC is being offered increasingly with the hopes of improving the rates of tumor downsizing and less extensive surgery. However, there is a paucity of large population data in the national setting as to demographics, practice patterns, and outcomes for this specific population. We aim to clarify the demographic and clinicopathologic factors that predict choice of NAC versus adjuvant chemotherapy (AC) for treatment in HR+ HER2− breast cancer.

Methods

We identified the cohort of women from NCDB diagnosed with HR+ HER2− breast cancer from 2010 to 2015 based on corresponding variables in the Collaborative Stage Coding Manual. The NCDB is a joint project of the Commission on Cancer of the American College of Surgeons and the American Cancer Society. It is a clinical oncology database sourced from hospital registry data that are collected in more than 1500 Commission on Cancer-accredited facilities. NCDB data are used to analyze and track patients with malignant neoplastic diseases, their treatments, and outcomes, and represent more than 70% of newly diagnosed cancer cases nationwide and more than 34 million historical records. 2010 was chosen as the start date for our study as it was the first year that the NCDB collected HER2 status data. We used the surgical procedure of the primary site code to stratify patients who did and did not receive surgical treatment. Only patients with both chemotherapy and surgery completed at the primary site within 8 months of diagnosis were included in the analysis. Patients who received chemotherapy before surgery were identified as NAC patients and those who had chemotherapy within 6 months of surgery were identified as AC patients. We restricted our cohort to only include clinical stage 2A, 2B and 3 patients and excluded missing pathologic T or N stage, path Tx or Nx and DCIS.

Potentially relevant demographic variables included age, race, insurance type, median household income, Charlson-Deyo comorbidity score, and year of diagnosis. Tumor-level variables included size, histology, grade and regional nodal involvement, and facility-level variables included distance from medical facility, facility type, and facility location. Clinical TNM was used for staging prior to treatment and in more modern cases positive node status may have been verified by axillary lymph node core needle biopsy. Pathologic TNM was used for determination of pCR.

Multiple imputation with chained equations, via IVEware software [16], was used to handle missing data in prognostic covariates. The imputation process loops through every variable containing missing values, where missing values were imputed using regression models conditional on all other variables. Ten imputed datasets were generated through ten repetitions. This method is superior to alternatives (complete case or missing data indicator methods) as far as analytic bias is concerned, under the assumption that data are missing at random [17]. The subsequent analyses were performed on each of the ten imputed datasets and resulting effect estimates and their corresponding 95% confidence intervals were appropriately combined using the MIANALYZE procedure in SAS.

Baseline prognostic variables were summarized within each treatment group as N (%).Separate multivariable Poisson regression models with a robust error variance were used to estimate adjusted prevalence ratios (PRs) and 95% confidence intervals (CIs) of factors related to utilization of NAC, in-breast pCR, and axillary downstaging within the NAC cohort, adjusting for variables outlined in the prognostic variables section. We chose PR over odds ratio as the latter tends to overestimate the strength of the association [18, 19].

Results

The baseline characteristics of NAC and AC cohorts are summarized in Table 1. Of the 134,574 eligible patients that received chemotherapy during the study period, 105,324 (78.3%) received AC and 29,339 (21.7%) received NAC. The use of NAC steadily increased over the study period from 19.7% in 2010 to 24.2% in 2015, an absolute increase of 4.5% (PR = 1.34, 95% CI 1.30, 1.38, for 2015 compared to 2010).

Table 1 Associations between baseline characteristics and receipt of neoadjuvant therapy, N = 134,574
Full size table

NAC use increased with clinical T stage three or four and with node positivity. Increasing clinical stage likewise had a positive association with NAC use. Other baseline tumor-related characteristics associated with increased NAC use include poorly differentiated tumors, either estrogen receptor (ER) or progesterone receptor (PR) negativity and invasive ductal histology. Baseline patient-related characteristics associated with increased NAC use include Black or Hispanic race, and Charlson score of 0, while age ≥ 50 years, median household income of < $38,000, and government insurance (i.e., Medicare and Medicaid) indicated decreased NAC use. Facility-related characteristics associated with increased NAC use include treatment at an academic or integrated facility and treatment at a facility in the central or western US.

Factors predicting pCR and axillary downstaging for NAC patients

Of the 29,250 patients that received NAC, 2401 (8.3%) patients achieved a pCR (Table 2). We defined pCR as ypT0 N0. pCR rates likewise increased over the study period from 6.3% in 2010 to 9.7% in 2015, an absolute increase of 3.4% (PR = 1.34, 95% CI 1.17, 1.53, for 2015 compared to 2010). pCR decreased with clinical T stage but increased with clinical N stage. Patients were more likely to achieve pCR if they were Black or Hispanic, resided in an urban location, had a poorly differentiated tumor, had invasive ductal histology, and had ER or PR negativity. Patients ≥ 50 years old or of Asian race were less likely to have a pCR.

Table 2 Factors impacting pathologic complete response (pCR) for patients receiving neoadjuvant therapy,N = 29,250 (pCR defined as ypT0
Full size table

Clinically node-positive patients receiving NAC converted to pathologically node negative 21% of the time (Table 3). Predictors of axillary downstaging include age < 50 years, black race, poorly differentiated grade, invasive ductal histology, and either ER or PR negativity (Table 4).

Table 3 Response to neoadjuvant chemotherapy based on clinical nodal status
Full size table
Table 4 Factors impacting axillary downstaging (ypN0) for patients receiving neoadjuvant therapy with cN+, N = 18,971
Full size table

Overall 39.3% of patients underwent BCT and 60.7% had mastectomy. Of patients that had AC 29.4% had BCT, while patients in the NAC group had a 38.6% BCT rate. 60.4% of AC patients and 45.3% of NAC patients underwent ALND. Patients who underwent mastectomy or ALND were found to be less likely to achieve in-breast or nodal pCR.

Discussion

Breast cancer molecular subtypes have played a vital role in our modern appreciation of the disease and our approach to its treatment. Our understanding of these subtypes has promoted a shift toward personalized breast cancer care as opposed to historic methodology. Knowledge of patient molecular phenotype and subtype improves selection of treatment and prognostication of disease-specific outcome.

Our study shows an increasing trend in the use of NAC for patients with HR+ HER2− breast cancer, despite reported pCR rates of 10–20% [4, 6, 12, 20], and only 8.3% in our study. While pCR is predictive of a favorable prognosis, this relationship is stronger in more aggressive subtypes [12, 21, 20, 22, 23]. This can be largely attributed to the fact that the HR-positive subtype has a favorable prognosis regardless of pCR [24]. With pCR rates of 30–40% in TNBC and over 50% in HER2+ breast cancer, it is not surprising that enthusiasm for NAC in these excellent responders has increased over time [15, 25]. The positive trend for NAC use in HR+ HER2− breast cancer is less an attempt to achieve a pCR but rather the intent to downstage the tumor to avoid mastectomy and/or downstage the axilla to avoid ALND.

The ACOSOG Z1071 trial evaluated the impact of tumor biology on the rate of BCT following NAC offering some comparison [4]. The study included 694 patients of all breast cancer subtypes with clinically node-positive disease receiving NAC followed by surgery. The pCR rate in HR+ HER2− patients was 11.4%, compared to 38% for TNBC and 45% for HER2+ cancer. However, a clinical response (complete or partial) was seen in 80.5% of HR+ HER2− patients and a pathologic response was seen in 71.8%. Only 9.5% of these patients showed disease progression while on neoadjuvant therapy. The HR+ HER2− patients were also more likely to undergo mastectomy compared to their TNBC and HER2+ counterparts, at a rate of 65.5%. The retrospective analysis of this prospective study did not allow for discerning which patients were mastectomy or BCT candidates before NAC receipt, and therefore, it is difficult to draw conclusions from this finding. This study was also unable to address which patients were BCT candidates after NAC but elected mastectomy. In fact, the residual tumor size in patients that did not achieve a pCR was similar across all subtypes. One hypothesis for the higher mastectomy rate in HR+ HER2− patients was the higher proportion of invasive lobular carcinoma seen in this subtype, which required more second procedures than patients with invasive ductal carcinoma.

A prospective, single-center study determined the frequency of avoiding ALND in clinically node-positive breast cancer patients that received NAC [6]. They included 155 patients spanning all tumor subtypes, with HR+ HER2− (56%) and invasive ductal histology (95%). The overall rate of nodal pCR was 49% but varied significantly by HR and HER2− receptor status. The rate of nodal pCR in HR+ HER2− cancer was 21%, exactly as seen in our study and in ACOSOG Z1071 [4]. While other subtypes appreciated higher rates of nodal pCR at 47%, 97%, and 70% for TNBC, HR-negative HER2+, and triple positive cancer, respectively, a nodal pCR seen in about one out of five patients with HR+ HER2− subtype is substantial.

Given the relatively lower rates of pCR and axillary downstaging in HR+ HER2− breast cancer treated with NAC, we continue to seek better therapeutic options for these patients. Neoadjuvant endocrine therapy (NET) offers a less toxic alternative as a potential management strategy, especially for postmenopausal women [26]. The ACOSOG Z1031 trial studied the effect of preoperative aromatase inhibitors (AI) on promoting BCT for 374 postmenopausal women with HR+ HER2− breast cancer [27]. Unlike Z1071, this trial designated patients’ pretreatment by their candidacy for BCT and compared it to the most extensive surgery actually performed. 50.9% of mastectomy-only candidates before treatment underwent BCT as their most extensive surgery, as well as 83.1% of marginal for BCT candidates. Three out of four inoperable candidates likewise were managed with BCT. This suggests marked improvements in surgical outcomes with NET with an overall BCT rate of 68%. There is also the potential to improve response rate with newer approaches such as selecting a subset of patients for NET using genomic profiling. Likewise, the ALTERNATE trial is a phase III randomized controlled trial that is seeking to obtain pCR and recurrence-free survival data in postmenopausal women on neoadjuvant AI [28].

Interestingly, our study found a higher rate of in-breast and axillary pCR for black patients compared to white patients. Several studies addressing the impact of race following NAC found no difference in pCR overall or by subtype [29,30,31,32,33]. They did, however, find that black patients receiving NAC consistently had worse outcomes even when matched for BMI, subtype, and stage.

While use of retrospective data from NCDB has limitations, this cohort included just under 135,000 patients which is the main strength of our study. Lack of data regarding compliance with treatment or duration of NAC and degree of estrogen/progesterone receptor expression are limitations. In addition, HER2 status was not collected by the NCDB before 2010; therefore, some HER2-positive patients are included.

NAC use for HR+ HER2− breast cancer has increased over time and, in spite of lower pCR rates compared to other phenotypes, offers a significant clinical benefit for many patients. Our study found that NAC should be strongly considered in patients with locally advanced HR+ HER2− disease who desire BCT or who are clinically node positive, especially if they are young or have poorly differentiated invasive ductal cancer with either ER or PR negativity. We expect improved outcomes going forward as better patient selection is guided by increasing use of pretreatment genomic assays. Black patients showed an improved rate of pCR compared to white patients despite prior studies finding that race did not influence pCR rate. More studies are needed to explore these findings and better predict patients who would be excellent responders.