The results of the retrospective studies have to be reviewed closely as they carry some significant bias that lead to the concern that the demonstrated survival from local therapy may be due to patient selection. The use of surgery was higher in women who have more favourable profile such as young age, smaller tumours, non-visceral disease, fewer comorbidities [27, 39, 40], and better access to care [25, 33]. Additionally, it is likely that the timing of surgery may introduce bias [27, 28, 32, 44]. For example, if women with T1–2 tumours, for whom a preoperative metastatic survey is not standard of care, were staged post-operatively because of a high nodal burden and were then discovered to have asymptomatic or low-volume metastatic disease, their prognosis may be better than women with symptomatic and high burden metastases diagnosed preoperatively. This consideration particularly affects tumour registry data, where the final stage is usually recorded several months after completion of primary therapy, by which time asymptomatic metastases may have been discovered. In addition, in any cohort study subtle differences in disease burden and patient fitness for intervention and disease biology may bias selection for surgery and hence outcomes.

Several randomized trials have been initiated in an effort to validate the results of the retrospective studies, particularly due to the concern that the apparent benefit of PSLT is driven by patient selection. A total of seven trials were initiated; two have completed accrual. Final results have been published by Badwe and colleagues from India, and preliminary results have been reported by Soran and colleagues from Turkey. Trials are ongoing in the USA/Canada (ECOG-ACRIN 2108) and Japan, whereas the Austrian trial (POSYTIVE, ABCSG 28) closed after accrual of 93 patients. Two additional trials were launched; the trial in the Netherlands closed with minimal accrual, and one in Thailand was withdrawn prior to accrual. Three of these trials (India, Japan, the USA/Canada) required induction systemic therapy prior to surgery.

The Translational Breast Cancer Research Group has completed a prospective multi-institutional data registry gathering data on the modern management of stage IV breast cancer (TBCRC 0313) [52]. The goal of the registry is to record information on the incidence of uncontrolled local disease, metastatic disease, rate of surgical intervention for palliation for the intact primary (in patients not undergoing PSLT), the frequency and effect of uncontrolled chest wall disease, and quality of life. Biologic and molecular data has been collected and will be used to determine interactions between the primary and metastatic sites. A total of 127 eligible patients have been enrolled thus far and are categorized into cohorts of those with an intact primary (A) or those with a resected primary and metastases discovered within 3 months of surgery (B). All patients received first-line systemic treatment according to standard institutional guidelines. Preliminary results presented at ASCO in 2016 showed that for patients in cohort A, after a 54-month median follow-up, 3-year OS is 70% and is better for patients who are responders to chemotherapy compared to nonresponders (78% vs 24%, p = 0.001). However, amongst responders no benefit was noted with surgery in 3-year overall survival (p = 0.85) irrespective of tumour subtype, although HER2-positive status and response seem to provide longer survival (93%). In addition, surgery did not improve progression-free survival with the median time to progression at 13 versus 12 months [52]. The data also demonstrated that surgical palliation of the primary tumour is uncommon in the modern eras as the majority of patients are responders to first-line therapy. The results of this registry cohort study show trends in survival in metastatic breast cancer; however the role of surgery in de novo stage IV breast cancer is still an issue for debate. In a preliminary analysis of survival relative to the 21-gene Recurrence Score (RS), King and colleagues observed that a high RS was independently prognostic for 2-year OS (hazard ratio, 1.83; 95% CI, 1.14–2.95; P = 0.013), and this group of women had a shorter 2-year survival if treated with initial endocrine therapy rather than initial chemotherapy [47].

The mainstay of treatment for the patient with metastatic breast cancer is systemic therapy. The optimal role for primary site local therapy in individuals with an intact primary tumour remains to be determined. After a flurry of enthusiasm following multiple retrospective analyses showing an apparent benefit of PSLT, the first mature randomized trial addressing this question yielded null results in terms of overall survival although there appears to be a benefit in terms of local control [45]; while the second complete trial remains unpublished at this time, results presented at ASCO 2016 appear to show a survival benefit for the use of locoregional therapy for the primary tumour [48]. The results from other randomized trials are pending, and until these are available, practitioners should be circumspect in offering PSLT to patients with metastatic disease. The role of radiotherapy in conjunction with surgery or as primary therapy also remains to be determined. Locoregional management of the primary tumour in stage IV breast cancer patients remains appropriate in the presence of symptoms; it may also be appropriate in highly selected patients if the distant disease is well controlled and the primary site shows unequivocal evidence of progression. If the primary site is also responding to systemic therapy however, prospective data from the TBCRC and the Tata Memorial trial showed that very few patients require palliative surgery. If both local and distant sites are well controlled, the rationale for LRT is also weak, since the primary site is likely to remain well controlled for the patient’s life span. The only possible exception may be the patient who has had a complete response to systemic therapy at distant sites and would be rendered stage IV with no evaluable disease by resection of the primary tumour. In this setting, highly selected series show that a third or more patients may achieve long-term survival [50]. If PSLT is decided upon following the above considerations, breast conservation (if feasible) is clearly the least morbid approach, and the lack of expectation of a survival benefit should be clear to the patient. Finally, in making recommendations for primary site locoregional therapy, it is of utmost importance to consider the potential morbidity and the costs of therapy. Notably, guidelines from the European Society for Medical Oncology published in 2014 state that «that surgery of the primary (tumour) should not be offered as a routine practice but can be discussed on a case-by-case basis and offered to selected patients ». The basis for selection is not specified [53].