Introduction

Among patients with estrogen receptor (ER)-positive, human epidermal growth factor 2 (HER2)-negative breast cancer, progesterone receptor (PR)-negative tumors were shown to have worse prognosis than PR-positive tumors [1,2,3]. PR-negative tumors have been shown to be less likely to respond to endocrine therapy than PR-positive tumors [1, 4, 5], and treatment intensification with adjuvant chemotherapy may improve the outcome of PR-negative tumors [6].

However, PR-negative tumors were underrepresented in recent trials that led to a routine use of 21-gene recurrence score (RS) in clinical practice, and they were only 10.0% and 5.7% of all tumors included in TAILORx and RxPONDER trials, respectively [7, 8]. Findings from these trials may not be generalizable to PR-negative tumors. In addition, a recent guideline by the American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP) recognized the heterogeneity in the extent of staining for ER and PR status, encouraging them to be further stratified by negative, low positive, and positive based on 1% and 10% cutoffs [9]. The role of PR-negative status in the context of RS and nodal staging remains unclear. We performed an observational cohort study to evaluate the association of PR status with RS and the magnitude of chemotherapy benefit on survival.

Method

This study was approved under the protocol (BDR-131220) by Roswell Park Comprehensive Cancer Center, and follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Patient selection and variable definition

The National Cancer Database (NCDB) was queried for women diagnosed between 2010 and 2017 with ER-positive, HER2-negative, pT1-3N0-1aM0 breast cancer who underwent surgery and adjuvant endocrine therapy with available RS. If 1% or greater cells stained positively, they were considered PR-positive. Otherwise, they were considered PR-negative. Variables of interest were facility type, age at diagnosis, race, medical insurance, income and education level, Charlson-Deyo Comorbidity Score (CDS), year of diagnosis, histology, tumor grade, T and N staging, recurrence score, progesterone receptor status, lymphovascular space invasion (LVSI), surgery, surgical margin, radiation therapy, and chemotherapy. Age was stratified by above versus below 50 years of age. Education and income levels were determined based on the 2016 American Community Survey data between 2012 and 2016. Such levels were the percentages of adults who did not graduate from high school and the median household income adjusted for 2016 inflation, respectively, in each patient’s zip code in the United States. High versus low neighborhood-level income and education were determined by the median value of 10.9% and $50,353, respectively. All missing values were labeled as unknown. Other variables, such as performance status, type and duration of systemic therapy, toxicity profile, breast cancer-specific mortality, tumor recurrence, were not captured in the NCDB.

Statistical analysis

Our primary endpoint was overall survival (OS), the time interval between diagnosis and the last follow up or death from any cause. Baseline characteristics were evaluated using Fisher exact test and Mann–Whitney U test as appropriate. Logistic multivariable analysis (MVA) was performed based on baseline patient and tumor characteristics to identify variables associated with PR-negative tumors. Kaplan–Meier method, log-rank test, and Cox MVA models including all clinically relevant variables were performed for OS. Crude odds ratio and hazards ratio results were reported using logistic and Cox univariable analysis (UVA). Variables included for logistic and Cox MVA are listed in eTable 1 and eTable 2, respectively. For patients diagnosed in 2017, OS were not captured in the NCDB, and these patients were excluded for OS analysis.

Interaction term analysis was performed to assess any heterogeneous association of PR status and chemotherapy receipt with OS. If the interaction term was significant, subgroup analyses were performed to compare the magnitude of the effect of chemotherapy and PR status. To reduce the selection bias and further evaluate the subgroup analysis results, propensity score matching was performed based on all variables of interest. Matching was performed using nearest neighbor method in a 1:1 ratio without replacements. The standardized differences of all variables were less than 0.1, indicating adequate match with negligible differences between arms [10]. To address immortal time bias, Cox MVA analyses were repeated after excluding patients with post-diagnosis OS of less than 6 months. Additional subgroup analysis was performed among those with RS ≤ 25 by repeating logistic and Cox MVA.

All p values were two-sided, and p values less than 0.05 were considered statistically significant. All analyses were performed using R (version 4.0.3, R Project for Statistical Computing, Vienna, Austria).

Results

A total of 143,828 women (median [interquartile range (IQR)] age, 60 [51–67] years) met our criteria (Table 1). Of these, 130,349 (90.6%) and 13,479 (9.4%) patients had PR-positive and PR-negative tumors, respectively. Median (IQR) follow up was 51.5 months (34.8–71.9).

Table 1 Baseline characteristics
Full size table

Logistic and Cox MVA

Logistic MVA showed that PR-negative status was associated with higher RS (> 25: aOR 16.15, 95% CI 15.23–17.13, p < 0.001; eTable 1 in the Supplement). Cox MVA showed that PR-negative status was associated with worse OS (adjusted hazards ratio [aHR] 1.20, 95% CI 1.10–1.31, p < 0.001; eTable 2 in the Supplement). Consistent crude results were observed using logistic (odds ratio [OR] 14.87, 95% CI 14.07–15.73, p < 0.001) and Cox UVA (hazards ratio [HR] 1.52, 95% CI 1.42–1.62, p < 0.001).

Interaction term and subgroup analyses

There was no interaction of PR status with RS (p = 0.42), chemotherapy receipt (p = 0.30), or nodal staging (p = 0.13). While its three-way interaction with RS and chemotherapy was not statistically significant (p = 0.53), there was an interaction with nodal staging and chemotherapy (p = 0.049). Subgroup analyses using Cox MVA showed the magnitude of the benefit of chemotherapy was greater among those with PR-negative node-positive tumors than PR-positive node-positive tumors, while it was comparable among those with node-negative tumors regardless of PR status (Fig. 1).

Fig. 1
figure 1

Forest plot of overall survival associated with chemotherapy stratified by progesterone receptor status and nodal staging using multivariable Cox regression. Dotted vertical line represents a hazards ratio of 0.66 associated with chemotherapy use for the entire cohort. No.: number of patients; aHR: adjusted hazards ratio; CI: confidence interval; PR: progesterone receptor; chemo: chemotherapy

Full size image

Propensity score matching analysis

To further evaluate such subgroups, propensity score matching was performed between those with versus without chemotherapy stratified by PR status and nodal staging. Similar findings were observed after propensity score matching. The magnitude of chemotherapy benefits was greater for those with node-positive tumors (pN1a, PR-positive: HR 0.61, 95% CI 0.50–0.75, p < 0.001; pN1a, PR-negative: HR 0.23, 95% CI 0.12–0.43, p < 0.001; eTable 3 in the Supplement and Fig. 2) compared to those with node-negative tumors (pN0, PR-positive: HR 0.43, 95% CI 0.37–0.50, p < 0.001; pN0, PR-negative: HR 0.53, 95% CI 0.41–0.69, p < 0.001; eTable 4 in the Supplement and Fig. 2).

Fig. 2
figure 2

Kaplan Meier plots for overall survival associated with chemotherapy stratified by progesterone receptor status and nodal staging after propensity score matching, Red: no chemotherapy; blue: chemotherapy; PR: progesterone receptor; OS: overall survival; HR: hazards ratio; 95% CI: 95% confidence interval; chemo: chemotherapy

Full size image

Subgroup analysis to address immortal time bias

After excluding those with post-diagnosis OS of less than 6 months, PR-negative status remained associated with worse OS (aHR 1.20, 95% CI 1.10–1.31, p < 0.001). Interaction with chemotherapy and nodal staging remained to be statistically significant (p = 0.037) with similar findings on subgroup analyses (pN0, PR-positive: aHR 0.74, 95% CI 0.66–0.83, p < 0.001; pN0, PR-negative: aHR 0.64, 95% CI 0.52–0.79, p < 0.001; pN1a, PR-positive: aHR 0.57, 95% CI 0.48–0.68, p < 0.001; pN1a, PR-negative: aHR 0.31, 95% CI 0.20–0.47, p < 0.001).

Subgroup analysis among those with RS ≤ 25

On logistic MVA, PR-negative status remained statistically significant for its association with RS 16–25 compared to RS ≤ 15 (aOR 4.45, 95% CI 4.21–4.71, p < 0.001). On Cox MVA, however, PR-negative status was no longer associated with OS (aHR 1.07, 95% CI 0.97–1.19, p = 0.18), and interaction term analysis among PR status, chemotherapy use, and nodal staging was not statistically significant (interaction p = 0.64).

Discussion

To our knowledge, this is the largest study based on nationwide oncology database to suggest that PR-negative status was independently correlated with higher RS and worse OS. It also suggested that, even after adjusting for age and RS, PR-negative tumors were associated with greater OS benefits from chemotherapy among patients with pN1a, but not pN0, breast cancer.

The proportion of PR-negative tumors in our study (9.4%) was comparable to the United States population-based study (9.1%) [11] as well as prospective trials including TAILORx (10.0%) and RxPONDER (5.7%) trials [7, 8]. Such observation suggests the overall consistency of PR status in patient population among the hospital registry-based database, the population-based database, and prospective trials. However, such proportions were lower than 23% of tumors being PR-negative reported by the study from the United Kingdom and Ireland. This inconsistency may be in part explained by allowing up to 10% of cells stained positively to be considered as PR-negative tumors [12], as opposed to less than 1% of cells as defined in our study.

Our findings on the PR-negative tumors associated with aggressive tumor biology as suggested by high RS and worse OS are consistent with other studies suggesting worse prognosis and survival outcomes [13,14,15]. Although 5-year survival rates for breast cancer have been improving in the United States, ER-positive breast cancers that are PR-negative still have a significantly lower OS compared to ER- and PR-positive tumors [14, 16, 17]. Patients with ER-positive breast cancers with low or no PR expression also have a greater risk of tumor recurrence [15, 17, 18]. Prior population study has shown that when ER-positive breast cancer recurs, 26% of the tumors will convert from PR-positive to PR-negative status, suggesting that a loss of PR expression is indicative of refractory response to hormone therapy responsiveness and further disease progression [14].

In our study, reasons for PR-negative tumors with greater OS benefits from chemotherapy among node-positive tumors compared to node-negative tumors remain unclear. At the absence of systemic therapy, node-positive tumors have been shown to be more aggressive with higher distant recurrence rates [19]. Although not statistically powered for subgroup analysis, KEYNOTE-522 trial also showed the treatment with pembrolizumab was associated with numerically higher survival outcomes among node-positive tumors compared to node-negative counterparts [20]. Patients with ER-positive, PR-negative tumors had comparable, poor outcomes similar to triple negative tumors [2], and given such tumor biology, treatment intensification with chemotherapy is associated with survival benefits for node-positive tumors.

Limitations of our study include inherent biases due to its retrospective nature. In addition, given the nature of OS as an endpoint, a short median follow-up of 51.5 months, and a lack of tumor recurrence data, the number of events was too low for subgroup analysis especially among patients younger than 50 years of age. Although most patients were non-Hispanic White in our cohort, heterogeneous characteristics seen between patients with PR-positive and PR-negative tumors (Table 1) may be in part due to sociodemographic factors and likely inter-related with one another. Additional analyses were not performed to investigate complex interactions among such variables, since they were beyond the scope of this study. In addition, as shown in our subgroup analysis among those with RS ≤ 25 suggesting a lack of association between PR status and OS, our study was likely underpowered to evaluate whether PR-negative status as an adverse prognostic factor would be valid among subgroups established based on menopausal status, age, and RS cutoffs by TAILORx and RxPONDER trials [7, 8].

Conclusion

Our study suggests that PR-negative status is associated with high RS and worse OS. PR-negative tumors were shown to benefit more from chemotherapy than PR-positive tumors in the node-positive setting. Further investigations are warranted to tailor systemic therapies among PR-negative tumors.