The present study evaluated the efficacy and safety of atezolizumab and nab-paclitaxel in a real world-setting in 20 metastatic breast cancer patients treated at an academic center in Austria in the initial phase after treatment approval.
At a median follow-up of 7.1 months (IQR 5.2–9.1) median PFS was 3.0 months (SE = .24; 95% CI [2.5; 3.5]) and median OS was 8.94 months (SE = 2.34, 95%CI [4.35; 13.53]).
In comparison, the pivotal study IMpassion130 reported a final median progression-free survival of 7.5 months (95% CI 6.7–9.2) with atezolizumab plus nab-paclitaxel in the ITT PD-L1-positive population at a median follow-up of 12.9 months and an updated median progression-free survival of 7.5 months (95% CI 6.7–9.2) at a median follow-up of 18.5 months (IQR 9.6–22.8), respectively [1, 2]. The median exploratory overall survival at the second interim analysis was 25.0 months (95% CI 19.6–30.7) with atezolizumab plus nab-paclitaxel in patients with PD-L1 positive tumors.
The considerable deviations in efficacy outcome parameters reported in IMpassion130 compared to the present study may be explained by the usage of atezolizumab in later treatment lines, a study population with higher morbidity at baseline and inclusion of tumors with differing biologic characteristics compared to IMpassion130.
While IMpassion130 solely included previously untreated unresectable, locally advanced or metastatic triple-negative breast cancer patients, the majority of patients in the present study had received chemotherapy for metastatic disease prior to atezolizumab and only 45% of patients received atezolizumab and nab-paclitaxel as first line treatment in the metastatic setting. Concordantly to our findings, previous data on atezolizumab monotherapy [3] and pembrolizumab [4] in triple negative metastatic breast cancer patients showed reduced clinical benefit from the anti-PD-L1/PD-1 agents in later lines of therapy. Therefore, immunotherapy should be introduced in the first treatment line for advanced breast cancer as stated in the EMA regulatory approvals [5].
IMpassion130 excluded patients with rapid disease progression referring to patients who had received previous radiotherapy or chemotherapy in the curative setting less than 12 months before randomisation [1]. Contrary, the present study included patients with more aggressive tumors indicated by the considerably shorter mean time from last-surgery until diagnosis with unresectable locally advanced or metastatic disease (9.0 months (6.2–22.5) vs. 24.5 months (15.9–38.9)). Furthermore, the present study population contained a higher proportion of patients with liver (40% vs. 28%) and brain (20% vs. 7%) metastases and – contrary to IMpassion130 – also included 3 patients with treated, yet symptomatic brain metastases. Previous studies on atezolizumab monotherapy for triple-negative metastatic breast cancer patients have demonstrated worse clinical outcomes in the presence of liver metastases regardless of treatment line [3]. Similar results were shown for pembrolizumab in patients with melanoma and non-small cell lung cancer [6]. These findings might be explained by the low prevalence of PD-L1 IC owing to the intrinsic immunosuppressive microenvironment of liver metastases [7, 8]. In addition, a significantly lower number of TILs was observed in breast cancer brain metastases compared to primary breast cancers, also suggesting an immunosuppressive microenvironment [9].
Notably, the IMpassion130 study did not show worse clinical outcomes in exploratory analyses of the subgroup of patients with liver metastases [10].
IMpassion130 demonstrated particular efficacy of atezolizumab and nab-paclitaxel in patients with triple-negative PD-L1 positive tumors exclusively leading to drug approval in this specific patient population.
Contrary, the present study contained three patients with tissue samples with variable immunohistochemistry profiles – with one of them also only showing a positive PD-L1 expression of tumor not immune cells. Additionally, one patient had a mixed PD-L1 expression in different tissue samples.
The predictive value of PD-L1 for PD-L1-targeted therapy in the early and metastatic breast cancer setting has been demonstrated in various studies [11,12,13,14]. Moreover, the clinical activity of atezolizumab has been shown to increase with rising IC and TC expression level of PD-L1 in various cancer types [3, 15, 16]. Contrary, in the exploratory biomarker evaluation of the IMpassion130 study the PD-L1 IC + expression level (≥ 1%) did not appear to affect atezolizumab and nab-paclitaxel efficacy as improved PFS and OS hazard ratios were comparable in patients with PD-L1 IC + low (≥ 1% and < 5%) or PD-L1 IC + high (≥ 5%) status [17]. Similarly, Hoda et al. [18] reported of a treatment response to atezolizumab and nab-paclitaxel of two patients with PD-L1 positive primary breast tumors but discordant PD-L1 negative metastatic lesions. Treatment responses of PD-L1 negative various cancer types with PD-L1 targeted therapy have been reported likewise, though response rates in PD-L1 positive tumors were higher [19]. Notably, in a biomarker analysis of atezolizumab monotherapy patients with tumors with a positive (at least 1%) baseline PD-L1 expression on tumor cells but < 1% PD-L1 expression on immune cells were nonresponders.
Eventually, the PD-L1 IC negative and solely PD-L1 TC positive metastatic lesions in the present study, respectively, might have contributed to the worse efficacy outcome compared to IMpassion130. Yet, due to the retrospective nature of the present study and the small sample size an organ-site and PD-L1-expression dependent treatment response evaluation was not feasible.
The median time between the date of tissue sampling that was (later) used for PD-L1 assessment and the start date of atezolizumab and nab-paclitaxel treatment was 6.6 months (1.1–12.4). Owing to the long gap between testing for PD-L1 and treatment initiation it is possible that some tumors lost PD-L1 positivity meanwhile. Indeed, in one out of four patients with more than one available tissue sample, discordant PD-L1 results were obtained. Temporal as well as spatial heterogeneity of PD-L1 positivity in breast cancer has been described in various studies with inconsistent results and needs further evaluation [7, 20,21,22,23,24,25,26,27].
In IMpassion130 PD-L1 expression was tested by immunohistochemistry using the SP142 PD-L1 immunohistochemical assay (Ventana Medical Systems, Oro Valley, AZ, USA) [1]. In the present study the majority of samples were analyzed using the same analytical method, yet in 2 samples the Clone BSR90 (Nordic Biosite) was utilized to assess PD-L1 status. It has to be noted, that published data indicates that different PD-L1 tests are not analytically equivalent and that patient outcomes have the potential to vary depending upon the assay used to identify patients for treatment [28]. SP142 should preferably be used to identify patients eligible for atezolizumab and nab-paclitaxel treatment.
Safety results in the present study were similar to IMpassion130. At the second interim analysis of IMpassion130 the most common grade 3–4 adverse events were neutropenia (38 [8%] of 453 patients in the atezolizumab group), peripheral neuropathy (25 patients [6%]), decreased neutrophil count (22 patients [5%]) and fatigue (17 patients [4%]) [2]. In comparison, the most common adverse events in the present study were leucopenia (7 patients (35%)), fatigue (6 patients (30%)) and abdominal pain (4 patients (20%)).
At the second interim analysis of IMpassion130 adverse events of special interest had occurred in 262 of 453 patients (58%). Similarly, in the present study any-grade adverse events of special interest occurred in 8 of 20 patients (40%).
Treatment-related deaths were reported in two (< 1%) patients in the atezolizumab group (autoimmune hepatitis related to atezolizumab [n = 1] and septic shock related to nab-paclitaxel [n = 1] at the second interim analysis of IMpassion130. In the present study treatment-related deaths were reported in 1 of 20 patients (5%) (pneumonitis related to atezolizumab).
Furthermore, similar results regarding treatment adherence were observed in the present study compared to IMpassion130.
In IMpassion130 406 of 453 patients discontinued atezolizumab owing to the following reasons: disease progression (n = 330), adverse event (n = 30), withdrawal by patient (n = 17), symptomatic deterioration (n = 14), death (n = 6), physician decision (n = 5), other reasons (n = 3) and non-compliance (n = 1). Leading cause for treatment discontinuation of atezolizumab and nab-paclitaxel due to toxicity was peripheral neuropathy (20 patients [4%]).
In comparison, in the present study atezolizumab was discontinued owing to the following reasons: progressive disease (11 patients), adverse events (1 patient with thrombocytopenia, 1 patient with hypophysitis, 1 patient with diarrhea), death (in total 3 patients, 1 patient with death owing to adverse event (pneumonitis)) and unknown reason (2 patients).
The following limitations of the present study have to be reported.
The study contained only a small sample size. To support the understanding of the impact of this combination outside of the clinical trial setting the presented findings should be reviewed in consideration of other real-world datasets. Radiologic response evaluation was performed less standardized and frequently compared to IMpassion130 potentially delaying the diagnosis of disease progression and prolonging continuation of treatment. Due to the retrospective nature of the present study and consequentially the inpart lack of clinical data some safety data and secondary efficacy endpoints such as the proportion of patients achieving an objective response, the duration of response and the time to deterioration in global health status evaluated in IMpassion130 could not be assessed in the present study.
In conclusion, the present study—evaluating treatment with atezolizumab and nab-paclitaxel in unresectable locally advanced or metastatic breast cancer patients in a real-world setting—showed a considerably shorter PFS (3.0 vs. 7.5 months) and OS (8.94 vs. 25.0 months) compared to the pivotal study IMpassion130. This might be explained by the inclusion of tumors with differing biologic characteristics compared to IMpassion130, a study population with higher morbidity at baseline and the usage of atezolizumab in later treatment lines. Yet, safety and treatment adherence results were similar between both studies.