SG has garnered attention for its manageable safety profile and promising survival outcomes in patients with metastatic TNBC refractory to prior lines of palliative chemotherapy. Drawing from a multicenter national study in Poland, we aimed to explore the safety and efficacy of SG in a real-world population.

Our patient cohort resembled the ASCENT trial population in terms of median age, distribution of visceral metastases to the lungs and liver, percentage of patients with HER2-low TNBC, number of previous anticancer regimens (including lines in the [neo]adjuvant setting), and previous application of taxanes. In our cohort, all patients were white, whereas in the ASCENT trial, 20% of the population was reported as belonging to other racial or ethnic groups. Additionally, the rates of metastasis to lymph nodes and bones were greater in our real-world data. The time from initial diagnosis to diagnosis of metastatic disease was almost twice as long in the ASCENT trial as in our cohort [6]. This interval was shown to have a positive impact on PFS in our study: a longer interval between BC diagnosis and SG initiation and between diagnosis of metastases and SG initiation was linked to improved PFS. This outcome likely reflects the inherent biological aggressiveness of the disease rather than a stronger therapeutic response. Patients with TNBC who experience an early relapse following (neo)adjuvant chemotherapy typically exhibit a more aggressive disease course [21]. In the ASCENT trial, 14% of patients received a single line of therapy in the metastatic setting and had recurrence of disease within 12 months of (neo)adjuvant chemotherapy. SG extended PFS and OS with a manageable safety profile. That was consistent with the overall study population [21]. In our cohort, nearly 50% of patients received SG as a second-line treatment.

Age was not shown in our study to influence the treatment outcome, confirming earlier results [6, 22]. Approximately 30% of the patients in the registration phase III study were not initially diagnosed with TNBC, which was comparable to our study (32%) [6, 23]. Similar or even higher percentages have been registered in other RWSs [14, 16]. This inconsistency between the staining of the primary tumor and that of the metastatic site is a well-known phenomenon [24, 25] and emphasizes the importance of obtaining biopsy samples at the time of metastasis or recurrence. As shown by analysis of patients in the ASCENT study, patients without TNBC at initial diagnosis had better outcomes and a manageable safety profile with SG, supporting its effectiveness regardless of initial subtype [23].

In our cohort, we showed median PFS of 4.4 months and median OS of 10.3 months. These outcomes are slightly lower than those reported in the ASCENT trial, which showed median PFS of 5.6 months and OS of 12.1 months for patients without brain metastases. When including patients with brain metastases, the ASCENT trial reported median PFS of 4.8 months and OS of 11.8 months [6]. Our findings align with other RWSs, which have consistently reported median PFS of 3–5 months and median OS ranging from 8 to 21 months [13,14,15,16,17,18, 26]. This consistency supports the efficacy of SG in treating patients with metastatic TNBC, despite variations in patient populations.

In the ASCENT trial, the benefit of SG was observed across all clinical and prespecified subgroups, including patients who had previously been treated with immune checkpoint inhibitors (ICIs), PD-1 inhibitors, or PD-L1 inhibitors (27% of the population). Although the impact of first-line ICIs on the efficacy of SG in subsequent lines has not been explored, we consider this an interesting topic. This group of patients had OS and PFS of 4.2 months, which is numerically lower than the data obtained in the general population. Our study suggested worse outcomes in these patients in terms of OS; however, the group was very small (see Table 2), and only univariate analysis was applied. Further RWSs could provide more evidence on this topic.

Our data do not confirm poorer OS in patients with central nervous system involvement. In a small study by Singh et al., PFS for patients without and with brain metastases was 4.4 months and 6.0 months, respectively, with identical OS [17]. In other RWSs, these patients present with numerically worse outcomes [14]. According to the ASCENT trial results, the whole population (with and without brain metastases) presented with better outcomes in the SG group than in the standard therapy group [6].

The American Society of Clinical Oncology (ASCO) data presented by Kalinsky et al. revealed differences in OS between patients treated with palliative treatment via the second line or third line and beyond (13.9 vs. 8.4 months) [18]. Although we did not observe such differences in our study, this aspect should be reported in other RWSs.

In our patients with a known radiological response, the ORR was 35%, mirroring data from clinical trials [6] and other RWSs [13, 14]. The median time to achieving a response was 2 months in our study, which was slightly longer than the 1.5 months reported in the ASCENT trial, likely due to differences in the timing of the first radiological assessment [6].

In our study, dose reduction was necessary for 25.3% of patients, primarily due to neutropenia. In other RWSs, 17–54% of patients required dose modifications [13,14,15,16], while in the ASCENT trial, this percentage was 22% [6]. Only 2.5% of patients in our cohort discontinued treatment due to AEs, and this percentage is consistent with other scarce available data (1–7%) [13,14,15,16] and registration trials [6].

There are significant differences in reporting safety data in published RWSs. Not only do the results differ in the frequency of different AEs reported, but the method of reporting these data also varies [13,14,15,16]. Our data regarding the most frequent treatment-related AEs of grade > 2 were consistent with clinical trial results and RWS, including neutropenia in 43% of patients (51% in the ASCENT trial [6], approximately 30% in the RWS trial) [13, 14, 16]) and anemia in 10% (8% in the ASCENT trial). Similarly, a study published by Caputo et al. reported a numerically lower incidence of diarrhea (4%) [13] (10% in the ASCENT trial and as high as 15–19% in some other RWSs)[15, 16]) and febrile neutropenia (3%, 6% in the ASCENT trial) [6]. We also noted a 5% incidence of grade 3 or higher platelet count decrease.

Alopecia of any grade was reported in 34% of patients. There is a wide range in the incidence of this complication: 46% in the ASCENT trial, 91% reported by Reinisch et al., and 67% by Caputo et al. [6, 13, 16]. Some RWSs did not report on this AE [14, 15]. It is possible that the reporting of this AE, which is not treatment-limiting toxicity, could be underrepresented in our cohort.

This study is subject to various constraints that warrant attention. First, the relatively small sample size restricts the depth of subgroup analyses and precludes adjustments based on important patient characteristics. Second, its partially retrospective design hinders the establishment of causal relationships between variables. These are typical limitations of RWSs.

Moreover, the absence of a comparative group prohibits direct comparisons with standard chemotherapy and impedes the assessment of the relative efficacy of SG. Additionally, the lack of data on previous medication, including granulocyte colony-stimulating factor (G-CSF) application, underscores a notable gap in our understanding of its impact on treatment outcomes. We plan to gather these data to explore its influence on treatment outcomes and safety issues. We have not assessed TROP-2 expression in our population, so we cannot elaborate on the relationship between TROP-2 expression and SG efficacy.

Data regarding PD-L1 status were not collected. This is because the majority of patients received their first-line treatment for TNBC before the reimbursement of chemo-immunotherapy with pembrolizumab in Poland, and therefore, CPS assessment was not conducted. In addition, chemo-immunotherapy with atezolizumab has never been reimbursed in Poland.

Furthermore, the study's limited scope, confined to investigational sites in Poland, raises concerns about the generalizability of the findings to broader populations and other healthcare settings. These limitations highlight the need for further research. Addressing these gaps through larger-scale studies, comprehensive data collection, and meticulous control of confounders will provide a more nuanced understanding of the real-world effectiveness and safety profile of SG.