For patients of BC with bone metastasis, the long-term survival and life quality in the later period are still not optimistic, and yet convenient and accurate prognostic predicting tool lacks. Recent studies point that the prediction ability of nomogram may be superior to that of traditional, categorical predictive models for various outcomes associated with cancer [17–19]. To take a step further, we performed the first large-cohort and comprehensive retrospective study based on wide multicenter, where the OS and the CSS of luminal A patients with bone metastasis (n = 3,171) selected from the SEER database were retrospectively analyzed Through univariate and multivariate Cox regression analyses, 12 independent variables associated with the OS and CSS were finally identified. Two nomograms established based on these significant prognosis predicting indicators showed high levels of discrimination and calibration in clinical utility.
Although BC bone metastasis is still incurable, our survival curves showed that the survival probability for patients of low-risk group is significantly higher than those of high-risk group. Therefore, it appears to be crucial to identify the risk factors for facilitating the prognosis predicting. Consistent with previous studies, our study suggests that age is a strong independent prognostic factor and young age is an advantageous factor for good prognosis [20, 21]. Besides, a report focusing on the OS time trends indicates that every incremental year of age is in independent and significant association with a higher risk of death [22]. On the contrary, old age may be a disadvantageous factor with poor status and age-related comorbidities. In addition, some targeted therapy or other intensive systemic treatment may be contraindicated to the old patients who are vulnerable to more frequent causes of death. In line with our conclusion, it has been previously reported that race is a significant survival predictor [23]. Parada H et al. [24] pointed that racial differences in gene expression might lead to the survival disparity of BC patients. Our study shows that, insurance status is also a significant variable. In many states of the USA, health insurance not only compensated patients for surgery but also reduced the cost of systemic adjuvant treatments. And insurance status has shown its impact on stages of diagnosis in previous study [25]. Moreover, Pan et al. [23] developed that in addition to the impact on diagnosis, uninsured status was also demonstrated to be an unfavorable factor of poor OS and CSS.
In our conclusion, tumor size, tumor primary site and histology grade were recognized as risk factors of great importance in affecting the prognosis of BC patients, which were in accordance with previous studies [26–29]. With our regression analyses, brain, liver and lung metastasis were also independent predictors of prognosis, among which brain metastasis was most likely to result in poor prognosis, followed by liver and lung metastasis. When considering all BC patients as an entire population, a cohort study has found that different distant metastatic sites presented similar trends in affecting survival [30]. Moreover, the effective implications of ER and PR status have been demonstrated by some large-scale studies in predicting patients’ prognosis and responding to BC endocrine therapy [31, 32]. Seho et al. [33] also concluded that lack expression of either ER or PR was in association with worse prognosis, especially among patients with node-positive luminal A subtype.
For cancer patients who have metastasized, whether to perform surgery is still controversial. Similar to previous reports, our research showed that non-surgical luminal A patients with bone metastasis had unfavorable prognosis. Generally, surgical treatment for primary lesion is recognized as a palliative therapy for BC patients with metastasis. Gnerlich et al. [34] showed an association between receiving surgery and improved survival for BC patients with metastasis. Xiong et al. [35] pointed that the prognosis of certain stage IV BC patients, especially those with bone- or soft tissue-only metastasis, could be improved by surgical removal of primary lesions. Moreover, for BC patients with bone metastasis, surgery can not only prolong the survival time but also improve life quality to some extent. And it is generally believed that chemotherapy can exert similar effect by reducing cancer-related complications through killing or inhibiting cancer cells, thereby relapse delayed and survival time prolonged. However, chemotherapy failed to be identified as a significant predictor for either OS or CSS in our multivariate analysis. In fact, our conclusion is not an exception with support of other studies, where no benefit of adjuvant chemotherapy was detected in luminal A BC patients [36, 37]. The Panel of the St Gallen International Expert Consensus insisted that was less useful in Luminal A subtype patients for their less responsiveness to chemotherapy [38]. In addition, consistent with our results, a retrospective cohort study suggested that there was no significant effect of radiotherapy in improving survival of BC with metastasis [39].
Our nomograms were based on twelve independent and significant prognosis factors selected from univariate and multivariate Cox regression analyses with satisfied level of discrimination, calibration and clinical utility, which can help predict the survival probability and expected benefits of different treatments, so that the most suitable one can be selected and the prognosis can get improved. In addition, there are many kinds of predictors included in our nomograms, implying that the luminal A with bone metastasis is a complex disease with considerable individual differences. In recent years, against the increasing emphasis on personalization of cancer treatment strategies, our nomograms can make accurate individualized predictions for each luminal A subtype patient with bone metastasis.
However, there are several limitations in the present research. First, our nomograms were based on a retrospective cohort obtained from SEER-base, which inevitably creates bias. Second, the data may lack some potentially important variables and key indicators, such as hormone therapy, targeted therapy, recurrence, and other advanced technologies. Third, some data are missing or not in detail, especially specific locations of bone metastasis and types of surgery. These deficiencies remain to be further improved in future studies.