In this study, we retrospectively investigated the clinicopathological characteristics and the prognostic factors of OS in patients with dnMBC according to ER status. To our knowledge, this is the first attempt in this field. Based on the results after PSM, we found that clinical T stage and initial metastatic sites were independent prognostic factors for OS in patients with dnMBC, regardless of ER status. In addition, Ki-67 index, endocrine therapy and radiotherapy for primary tumor were also independent prognostic factors in the ER-positive population, and initial chemotherapy cycles and chemotherapy after disease progression were independent prognostic factors in the ER-negative population.
The results showed a 5-year OS rate of 18.9% and a median OS of 29 months for the whole cohort, which was similar to the results reported in previous studies of patients with dnMBC[17, 19–26]. Several recent studies have also reported favorable survival outcomes with upward trend in patients with dnMBC, with a median OS ranging from 39.2 to 40.8 months in the United States[6, 20], 39.5 months in France[24], and 63.4 months in China[27]. A very important fact for explanation is that the patients enrolled in this study were all from two tertiary cancer centers, and the data were from a discontinuously selected population. In addition to the intrinsic characteristics of the ER subtype, the survival differences of patients with dnMBC by ER status might be related to the follow-up time and the development and application of novel treatment technology. In addition, the median OS was 24 months for the ER-negative cohort and 36 months for the ER-positive cohort in this study, which was similar to previous reports in recent years[24]. However, it is not difficult to find that there were substantial differences in OS according to ER status. ER-positive patients have longer OS, and ER-negative patients have a worse prognosis. This might be partly due to the presence of triple-negative disease with intrinsic poor prognosis in the ER-negative cohort. Zhang et al. found that ER status was independently associated with PFS and OS in patients with dnMBC, and hormone receptor-positive subtype had better survival outcomes than triple-negative subtype[15]. These results supported our findings. Thus, exploring the prognostic factors in ER-positive and ER-negative population respectively is necessary.
In the present study, although the underlying cause of the difference in survival due to different ER status remains unknown, we identified the prognostic value of the clinical T stage and initial metastasis sites for OS in dnMBC patients, regardless of ER status. Moreover, the multivariate Cox models also showed the independent contributions of initial chemotherapy cycles and chemotherapy after disease progression in ER-negative cohort, and Ki-67 index and radiotherapy for primary tumor in ER-positive cohort. Chemotherapy is more important for ER-negative patients as they cannot respond to endocrine therapy. Thus, it is understandable that adequate initial chemotherapy and restart of chemotherapy after progression can result in better prognosis. Ki-67 index is an accepted prognostic marker in ER-positive patients, as suggested by the International Ki67 in Breast Cancer Working Group[28]. The prognostic value of Ki-67 was also observed in this study. Indeed, endocrine therapy is also an independent prognostic factor for ER-positive patients, as confirmed in this study. As expected, the clinical prognostic factors for dnMBC by ER status in the present study were consistent with those previously reported[29–31]. Intriguingly, the two survival curves for patients stratified by ER-negative and ER-positive met at approximately 6 years, and kept declining and crossed over at approximately 9 years after diagnosis. This phenomenon could be partly due to the acquired resistance to endocrine therapy during the long treatment course. It should be made clear that once a tumor has acquired resistance, the selection of effective treatment becomes very difficult, and the survival outcome would be poor. Therefore, overcoming drug resistance may be the key to improving survival outcomes in ER-positive patients with dnMBC.
The question of whether locoregional therapy, including surgical resection and radiotherapy, is beneficial or not in patients with dnMBC remains controversial. In a large series published by Iwase et al reviewing 1981 patients treated between 1995 and 2017 at MD Anderson Cancer Center, favorable OS was observed in tumors with low nuclear grade and in patients who received locoregional therapy[6]. Our previous studies and numerous retrospective studies have shown that locoregional therapy may provide survival benefit for such patiens[18]. Nevertheless, recently published prospective studies have yielded conflicting conclusions[9, 32]. This study found that locoregional therapy (radiotherapy) as an independent prognostic factor had positive effect on survival, especially in ER-positive patients. We believe that the major cause for those controversial results would result from a selection bias. It would be highly probable that the patient with limited disease, including oligometastases, dormant tumors, and high sensitivity to systemic therapy may benefit from locoregional therapy. Obviously, the amount of tumor burden and the site of metastasis would impact the effect of locoregional therapy on the survival outcome in the present study. We expect to obtain further insights from large clinical trials in the future, such as the ongoing JCOG1017 clinical trial[33]. Overall, the impact of locoregional therapy on survival appears to be limited at this moment. However, the benefit of locoregional therapy for controlling local disease progression should be investigated to improve patient’s quality of life.
There are some questions about ER subtype that need to be clarified. Indeed, we did not discuss the association of ER phenotype with the prognosis of metastatic breast cancer in isolation from the hormone receptor status and HER2 status. A study by Bae et al reported that single hormone receptor-positive subtype (ER+/PR- and ER-/PR+) was associated with poorer OS than ER+/PR + subtype, but they found that single hormone receptor-positive subtypes are associated with poor survival similar to that of the hormone receptor-negative subtypes (ER-/PR-) in HER2-negative breast cancer while there was no difference in survival among the four subtypes in patients with HER2-positive breast cancer[34]. In contrast, the ER+/PR + subtype showed better OS. These results are consistent with our findings and other studies[35–38]. In this study, the majority (79.3%) of ER-positive patients were ER+/PR+, while most (93.6%) ER-negative patients were ER-/PR-. However, the prognosis of single hormone receptor-positive breast cancer remains unknown.[36, 38, 39]. Fortunately, a recent study has shed some light on this issue[34]. A study by Dauphine C et al.[34] have shown that single hormone receptor-positive breast cancer subtypes are more likely to have unfavorable characteristics and worse survival than the ER+/PR + subtype, and the ER-/PR + subtype has outcomes similar to those for hormone receptor-negative cancers. Thus, the single hormone receptor-positive subtypes, which account for 10% of HER2-negative cancers, should be considered clinically distinct from hormone receptor-positive disease. As a result, according to several studies, patients with ER+/PR + subtype had a better prognosis than patients with hormone receptor-negative subtype. It should be noted here that this study classified patients according to ER status, regardless of PR status. The single hormone receptor-positive subtype (ER+/PR- or ER-/PR+) does not fall into our category of ER subtype. Therefore, the above analysis supports our conclusion. The analysis of ER status, regardless of PR status, aimed to focus on and explore the characteristics of ER subtype and their prognostic relevance to patients with dnMBC, rather than ignoring the importance of PR status.
Several limitations must be acknowledged. First, the selection bias was inevitable due to the retrospective nature. However, the statistical differences in baseline variables were significantly reduced after PSM, which might improve the reliability of the results. Second, single ER- and ER + status was assessed, and the exact expression levels of ER were unavailable. Because it is now recognized that for breast cancer, low-ER status may be an important ER subcategory. Third, HER2 status was not included in this study because many patients had unknown HER2 status, which inevitably affected the results of the analysis. Fourth, we discussed single ER status without investigating the PR/HER2 status and the interaction between them. Fortunately, we performed the interpretation and related analysis in the hope of reducing the impact on the results. These limitations might have contributed to study bias and undermined the power of analysis. To overcome these limitations, a detailed evaluation via a well-designed prospective study would be crucial.
In conclusion, for patients with dnMBC, there were substantial differences in prognosis according to ER status. ER-negative subtype is clinically different from ER-positive subtype, showing unfavorable tumor characteristics and shorter survival. In addition to clinical T stage and initial metastatic sites, other independent prognostic factors of OS were initial chemotherapy cycles and chemotherapy after disease progression for ER-negative patients, and Ki-67 index, endocrine therapy and radiotherapy for ER-positive patients. These findings may help to further understand the disease behaviors and have meaningful significance in treatment decision.