Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
About 6% of all breast cancer patients present with an intact primary tumor and synchronous distant disease, a rate largely unchanged over the past 20 years. For these patients, overall survival is dictated by their systemic disease rather than the primary tumor status. Consequently, systemic therapy has been considered first-line treatment, and resection of the intact breast tumor is generally not recommended, as most patients succumb to their disease before developing uncontrolled local disease (ULD). If surgical resection of an intact primary is undertaken, it has traditionally been performed to avoid impending complications of ULD or to palliate chest wall progression once it has occurred.
This approach has been questioned, however, as the clinical course of metastatic breast cancer (MBC) changes with patients living longer. In an analysis of temporal trends in survival, 1033 consecutive patients presenting to the West German Cancer Center with MBC between 1990 and 2009 were divided by 5-year intervals. Median survival increased from 24.2 months between 1990 and 1994 to 37.8 months between 2005 and 2009, with incremental improvements plateauing over time. Similar trends are seen in the US Surveillance, Epidemiology and End Results (SEER) and French UNICANCER registries, among others. While some of this difference may be attributed to lead-time bias from earlier diagnosis of metastatic disease as imaging modalities improve over time, much of the improved survival is related to advances in systemic therapy with targeted agents. Several trials have shown that up to 30% of patients with distant metastases treated with modern multimodality therapy can achieve long-term survival. As diagnostic techniques continue to improve and additional molecular targets are manipulated for therapeutic gain, MBC patients in the 21st century can be expected to be diagnosed with increasingly small disease burden and show continued improvements in survival.
In parallel with these advances, elective resection of the intact breast tumor in de novo MBC (dnMBC) became a consideration as a survival benefit was demonstrated in other malignancies. Level I evidence supporting a benefit of primary tumor resection in the metastatic setting was first established in renal cell carcinoma. Two prospective, randomized trials published by the Southwest Oncology Group (SWOG-8949) and European Organization for Research and Treatment of Cancer (EORTC-30947) compared radical nephrectomy with nonoperative management of the primary tumor in patients treated with systemic therapy (interferon alfa-2β). Both trials demonstrated a statistically significant survival advantage for patients treated with surgery (11.1 vs. 8.1 months, P = 0.05; 17 vs. 7 months, P = 0.03, respectively). Subsequent retrospective studies supported a similar benefit of primary gastrectomy in the setting of advanced or metastatic gastric cancer. SEER analysis of over 8200 patients undergoing treatment for stage IV gastric cancer showed that those who underwent surgery had higher 3-year cancer-specific survival rates than nonsurgery patients (9.4% vs. 2.1%, P < 0.001). Similar benefits have been demonstrated in ovarian and colorectal cancers.
New biological insights may explain these clinical findings, although the role of a primary solid tumor in the metastatic process and treatment response is still not well understood (see Fig. 55.1 ). The identification of cancer stem cells and circulating tumor cells suggests that an intact primary tumor may facilitate the development of new metastases through tumor self-seeding. In this multidirectional process, cancer stem cells with enhanced metastatic potential lead to ongoing dissemination and disease progression. Additionally, an increasing body of evidence suggests that there may be molecular communication between the primary tumor and the premetastatic niche. A specific role for mesenchymal stem cells that endow primary tumor cells with enhanced metastatic capacity provides a possible explanation for a beneficial role of primary tumor resection even in the setting of established distant disease. Other hypotheses relating the presence of the primary tumor to the metastatic process implicate immune suppression caused by the primary tumor that can be mitigated by surgical resection.
While these biological hypotheses support local therapy of the primary tumor in the metastatic setting, opposing evidence argues against primary tumor resection (see Fig. 55.1 ). Laboratory data suggest that the presence of the primary tumor may in fact restrain the growth of metastatic lesions, although this has never been demonstrated in humans. The presence of increased circulating tumor cells intraoperatively during surgical manipulation raises the concern for hematogenous tumor cell dissemination and a potential role in producing new metastatic foci. Finally, elevation of protumorigenic and prometastatic cytokines and growth factors in the perioperative period may lead to accelerated tumor cell adhesion and growth, a concept referred to as “surgical wounding.”
While the complex biological relationship between the primary tumor and metastatic disease is still being elucidated, the results of the SWOG and EORTC metastatic renal cell carcinoma studies led to an appropriate questioning of the role of primary tumor resection in dnMBC. Numerous retrospective studies have since evaluated the possibility of a survival benefit of primary site local therapy (PSLT) for dnMBC and have shown a general association between surgical resection of the intact primary and improved survival. Due to the multiple biases in retrospective studies, these outcomes led to several randomized prospective trials with mixed results that have largely demonstrated no survival benefit for PSLT in dnMBC. Both retrospective and prospective studies are discussed at greater length later.
Prompted by the renal cell carcinoma results, Khan and colleagues analyzed survival data on patients reported to the National Cancer Database (NCDB) of the American College of Surgeons. Among 16,023 patients presenting with dnMBC from 1990 to 1993, surgical resection of the primary tumor was performed in 57%. The vast majority of patients were treated with systemic therapy, but data on primary site radiation therapy were not available in the NCDB at that time. Surgical resection of the primary tumor was associated with a 39% reduction in the hazard of death from any cause. Other characteristics associated with overall survival in multivariate analysis included the use of systemic therapy, the number of organ sites involved, and the presence of visceral disease. In this NCDB analysis, the 3-year survival was 35 months in the surgically resected patients with free margins, compared to 26 months in women undergoing resection with involved margins, and 17 months in the nonsurgical group.
Subsequently, over 40 retrospective analyses have evaluated the possible role of local therapy for the primary tumor, with the majority showing improved overall survival in women receiving PSLT. Most of these studies evaluated surgical resection of the primary tumor, although some have addressed primary radiotherapy, as discussed later. In 2020 Gera and colleagues published an updated meta-analysis of 42 studies (39 retrospective, 3 randomized controlled trials) published between 2002 and 2019 that evaluated the effects of PSLT (surgery, radiotherapy, or both) in dnMBC. These authors did not examine sources of treatment selection bias and acknowledged the substantial heterogeneity between studies. Nevertheless, they reported that 81% (30 of 37) of datasets evaluating surgery alone, 58% (11 of 19) of datasets evaluating radiotherapy alone, and 80% (4 of 5) of datasets examining the use of combined locoregional therapy revealed an association between PSLT and longer survival. Random effects models of the combined effect of all locoregional therapy demonstrated a 31.8% decrease in the hazard of death (HR 0.682, 95% CI 0.637–0.731). For unclear reasons, three prospective randomized trials were included in this summary estimate, and it should be noted that the hazard ratio included from the randomized controlled trial by Fitzal and colleagues is incorrectly interpreted to favor surgical resection; in fact, the study demonstrated worsened outcomes in the surgical arm, as discussed later. Another recent meta-analysis included 30 observational studies and evaluated the biases driving the use of PSLT. These include primary and metastatic disease volume, biologic parameters, and use of radiotherapy. The forest plot from this study is shown in Fig. 55.2 . This analysis, along with other earlier attempts, suffers from similar biases, and favors the use of locoregional therapy in dnMBC with similar hazard ratios.
The contributions of axillary surgery to overall survival have been difficult to evaluate in the published retrospective analyses. A meta-analysis by Hartmann and colleagues identified six retrospective studies that provided information about axillary surgery at the time of primary tumor resection. Of the patients reviewed, 42% underwent surgery, and, of those patients, 527 (69%) had axillary procedures. Only three studies investigated the association of axillary surgery with survival and did not find a significant relationship. Given current concepts regarding the role and value of axillary dissection in non-MBC, this procedure cannot be recommended in patients with metastatic disease.
The use of exclusive radiotherapy (ERT) for the primary tumor appears to provide a survival benefit similar to that seen with surgical resection, although it is difficult to decipher as a sole intervention, as it is typically combined with surgery. As mentioned above, 11 of the 19 (58%) retrospective datasets included in the Gera and colleagues meta-analysis associated ERT with longer survival. The largest retrospective studies of ERT have mainly come from single institutions in France and Canada. In 2009, Le Scodan and colleagues identified 581 patients with dnMBC treated with PSLT between 1984 and 2004. Of these patients, 249 received ERT, 30 received surgery alone, and 41 received both surgery and radiotherapy, with radiotherapy including nodal fields and a boost to the tumor site for most patients. The 3-year overall survival rate was 43% versus 27% in the group receiving any locoregional radiotherapy versus those who did not, with an adjusted hazard ratio of 0.70 (95% CI 0.58–0.85). Similarly, Nguyen and colleagues have reported a series from British Columbia evaluating the use of PSLT and survival in 733 dnMBC patients presenting between 1996 and 2005. Of these, 355 patients who received no locoregional therapy were compared to 378 patients who underwent PSLT, consisting of surgery alone (67%), ERT (22%), or combined surgery and radiotherapy (11%). The 5-year overall survival rates were 21% in those who received PSLT compared to 14% in those who did not ( P < 0.001). The rates of locoregional progression-free survival were higher in those who had PSLT (72% vs. 46 % , P < 0.001). Patients who had both surgery and RT had a better 5-year overall survival of 32.5% compared to those who had surgery or RT alone (21% and 17%, respectively). A second French study of 239 patients comparing PSLT with ERT or surgery alone showed a trend toward increased survival. However, no advantage was noted between either group after adjustment for prognostic factors, although ERT did improve local control.
The rationale for early PSLT in MBC may relate to the anticipated need for palliation of a symptomatic tumor or a fear of ULD in the future. As quality of life is a clear goal in MBC, chest wall outcomes are important; unfortunately, pertinent data are scant in retrospective studies. In one retrospective study, chest wall control was evaluated in relation to primary tumor resection and survival in 111 patients with MBC, of whom 42% underwent surgical resection of asymptomatic primary tumors within 6 months of diagnosis. The nonoperative group included patients who never received surgery or who underwent delayed surgery for palliation. Both groups were well matched, and all received systemic therapy. Local disease control was more often maintained in patients treated surgically (82% vs. 34%, P = 0.002). Surgical resection was associated with longer time to first progression (HR 0.5, 95% CI 0.298–0.838), but not with overall survival. However, when survival was examined as a function of chest wall control, patients who maintained local chest wall control survived significantly longer (HR 0.42, 95% CI 0.26–0.66) than those who developed symptomatic chest wall disease (i.e., skin nodules or ulceration). On the other hand, since maintenance of local control may also indicate tumor sensitivity to systemic therapy, it is not possible to draw definitive conclusions from data such as these.
As mentioned earlier, 7 of 37 (19%) retrospective studies demonstrated no improvement in overall survival in patients undergoing surgical resection of their primary tumor in the setting of dnMBC. One of the largest of these was an analysis of 622 patients with dnMBC that matched surgically treated patients with controls who did not receive PSLT. Matched-pair analysis attenuated the survival benefit associated with PSLT in all subsets and eliminated it in 100 patients with visceral disease. A detailed review of medical records of 100 women within the dataset demonstrated inaccurate categorization of tumor stage and surgical procedure, possibly influencing the results and demonstrating a valid concern with retrospective database analyses. The authors noted benefits for patients who received surgery following induction systemic therapy, supporting the notion that patients who had a favorable response to systemic therapy were more likely to benefit from surgery.
Although retrospective studies in general suggest a benefit and no worsened outcome for patients undergoing surgical intervention, they suffer from substantial biases that challenge the validity of the observed association and leave open the possibility that the apparent survival benefit related to PSLT is attributable to selection bias rather than a cause-and-effect relationship.
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here