Breast cancer has the highest incidence of all malignant tumors among women worldwide (Zhang et al., 2021). Triple-negative breast cancer (TNBC) accounts for approximately 15–20% of all breast cancers. Its pathological features are negative for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER-2), characterized by an early age of onset, strong invasiveness and high recurrence and metastasis rate (Li et al., 2016; Sharma et al., 2018). The clinical prognosis of TNBC patients is poor due to the lack of effective targets for endocrine therapy and targeted therapy. The 5-year survival rate of patients with advanced TNBC is less than 15% (Bagegni et al., 2019; Berger et al., 2021). Immunotherapy provides a new treatment strategy for the patients. Pathologically, TNBC has relatively abundant tumor-infiltrating lymphocytes and high expression level of programmed death-ligand 1 (PD-L1), providing a suitable immune microenvironment and target basis for the application of immune checkpoint inhibitors (ICIs) (Disis et al., 2015; García-Teijido et al., 2016; Huang et al., 2021). Clinical trials such as KEYNOTE-012, KEYNOTE-086, and IMpassion 031 have confirmed that the immune checkpoint programmed cell death-1 (PD-1)/PD-L1 inhibitors are effective treatment options for TNBC (Seiwert et al., 2016; Adams et al., 2019; Mittendorf et al., 2020). However, only a small proportion of patients showed a long-term sustained response (Keenan et al., 2020). Therefore, there is an urgent need for reliable biomarkers of immune therapy to screen out the optimal beneficiary population clinically (Emens et al., 2021). PD-L1 expression level is currently the most important and controversial predictor of immunotherapy efficacy. IMpassion130 trial indicated that atezolizumab in combination with nab-paclitaxel significantly increased the overall survival (OS) and the progression-free survival (PFS) in PD-L1 positive patients with metastatic TNBC (Schmid et al., 2020), but, KEYNOTE-119 trial demonstrated that pembrolizumab did not significantly improved the OS in PD-L1 positive patients with previously treated metastatic TNBC versus chemotherapy (Winer et al., 2021).

In recent years, several studies have investigated the association between PD-L1 and immunotherapy clinical outcomes in TNBC, but most of them focused on the early stage of TNBC. Two meta-analysis studied the prognostic role of PD-L1 in advanced or metastatic TNBC (Ji et al., 2021; Latif et al., 2022), they conducted the subgroup analysis stratifying the status of PD-L1 by including two studies, and the results suggested that better PFS or OS could be found in PD-L1 positive group. Notably, these meta-analyses were short of the latest relevant clinical trials and further subgroup analysis such as ICIs alone or ICIs combination with other chemotherapeutic drugs. In this study, we summarized recent and relevant clinical trials and conducted a meta-analysis of randomized controlled trials (RCTs) to assess the relationship between PD-L1 expression and the efficacy of ICIs in locally advanced or metastatic TNBC.

The studies were independently evaluated and extracted according to the inclusion and exclusion criteria by two reviewers, and the disagreements between them were resolved by discussion. The following data from the eligible study were retrieved: name of the trial, year of publication, study phase, study population, therapeutic regimen, rate of PD-L1 positive expression, cutoff value, HRs, and 95% CI of OS and PFS, if available. The latest and most complete study was chosen between the same studies in different periods.

TNBC has become the mainstay in malignant tumor immunotherapy due to its high tumor mutational burden, increased infiltration of immune cells and high expression of PD-L1. Currently, the PD-1/PD-L1 pathway is the most prominen target of immunotherapy in locally advanced or metastatic TNBC (Lipson et al., 2015; Hartkopf et al., 2016; Pusztai et al., 2016). IMpassion130 was the first phase III clinical trial to demonstrate the efficacy of first-line immunotherapy in advanced TNBC, atezolizumab plus nab-paclitaxel improved both PFS and OS compared with placebo plus nab-paclitaxel in the PD-L1 positive TNBC. However, the subsequent study of IMpassion131 found that atezolizumab combined with paclitaxel had no benefit for PD-L1 positive metastatic TNBC. KEYNOTE-355 study explored the clinical efficacy of pembrolizumab plus chemotherapy versus placebo plus chemotherapy in the first-line treatment of metastatic TNBC, preliminary results confirmed that pembrolizumab combined with chemotherapy significantly improved PFS compared with placebo combined with chemotherapy for TNBC patients with a combined positive score ≥10. It is worth thinking deeply about these inconsistent findings. Therefore, searching for reliable biomarkers for ICI therapy to screen out the optimal beneficiary population is an important clinical problem that needs to be solved.

In terms of PFS, the effect of ICIs combination chemotherapy in first-line was better than that with ICIs alone in post-line. The main reasons for this result might be as follows: first, due to delayed onset and the pseudoprogress in immunotherapy (Wang et al., 2018; Failing et al., 2019), chemotherapy could reduce tumor burden in patients in the shortest time and cause the declination of tumor interstitial fluid pressure (TIFP). The difficulty in monoclonal antibodies’ entrance into the tumor is due to an increased TIFP. Consequently, the decrease of TIFP could promote the entry of macromolecular substances such as monoclonal antibodies into the tumor, thereby improving the effect of antitumor immunotherapy. Meanwhile, the hypoxic state of tumor microenvironment was improved by reducing TIFP, and then alleviated the immunosuppression of T-cells (Hofmann et al., 2013; Baronzio et al., 2015; Patel et al., 2021). Second, factors leading to immunosuppression include: PD-1/PD-L1, cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4), lymphocyte activation gene 3 (LAG-3), and overexpression of androgen receptor (Cimino-Mathews et al., 2015; Andrews et al., 2017; Pardoll et al., 2021; Guan et al., 2022). These factors made it impossible to fully activate T-cells activity by the inhibition of only the PD-1/PD-L1 signal pathway and led to insignificant effects of ICIs alone. Thus, a synergistic anti-tumor effect can result from the combination of PD-1/PD-L1 inhibitors and other drugs which could affect the activity of immune, including 1) Chemotherapy drugs. Chemotherapy drugs could directly and indirectly enhance immune activity. Paclitaxel and gemcitabine could enhance antitumor immune responses by eliminating immunosuppressive cells (myeloid-derived suppressor cells, regulatory T cells) and promoting the activation of immune cells such as the maturation of natural killer cells and the activation of cytotoxic lymphocytes. Chemotherapy could fully expose tumor antigens by killing tumor cells and improve anti-tumor immune recognition, which provide a good antigenic basis for the application of ICIs, thus improving the efficacy of immunotherapy (Lesterhuis et al., 2011; Roselli et al., 2013; Bracci et al., 2014; Wang et al., 2017). 2) Costimulatory molecule. 3) Androgen receptor antagonist. In addition, the lines of treatment might also be one of the factors affecting the efficiency of immunotherapy (first-line in the ICIs combination chemotherapy group versus post-line in the ICIs alone). The increased tumor burden and the decreased autoimmune potential of patients with post-line therapy were also a aspect that should be considered.

In terms of OS, with similarities to various baseline characteristics (age, sex, race, tumor stage, the status of PD-L1) among included clinical studies, it was speculated that paclitaxel might be the most likely source of heterogeneity given that the treatment regimen of this study were ICIs combined with paclitaxel, while the other three studies were either ICIs alone or ICIs combined with nab-paclitaxel. The subgroup analyses demonstrated significant OS benefit in PD-L1 positive TNBC with immunotherapy but without paclitaxel. The reasons for the absence of clinical benefits of atezolizumab combined with paclitaxel compared with atezolizumab combined nab-paclitaxel were as shown below: Different chemotherapy drugs for immune combination led to different synergistic effects. Because paclitaxel is highly lipophilic and insoluble in water, the vehicle of paclitaxel was polyoxyethylene castor oil and absolute ethanol, the vehicle system could easily cause severe allergic reactions. Therefore, glucocorticoids pretreatment is clinically required to prevent allergic reactions before administration. However the efficacy of immunotherapy might be impaired due to glucocorticoids pretreatment with paclitaxel. In the IMpassion131 trial, the effect of weekly high-dose glucocorticoids on the efficacy of immunotherapy was an important factor. Additionally, paclitaxel had high systemic toxicity due to the lack of specific distribution in tumor tissue, leading to poor tolerance after long-term application. Albumin-bound paclitaxel belonged to a nano-drug delivery system, which greatly improved the solubility of the drug and avoided the use of polyoxyethylene castor oil, thus the occurrence of allergic reactions was greatly reduced (Kundranda et al., 2015). Moreover, albumin-bound paclitaxel could passively target tumor tissue by utilizing the enhanced permeability and retention (EPR) effect of blood vessels of tumor tissue (Kouchakzadeh et al., 2015), thereby reducing systemic toxicity, improving patient tolerance and the effect of anti-tumor therapy. The differences in the dosage forms of paclitaxel and nab-paclitaxel might result in different therapeutic effects of the two drugs in combination with immunotherapy drugs, which were able to provide some guidance for the choice of clinical chemotherapy regimens.

Finally, the limitations of this study include; 1) Different antibodies, cutoffs, cell types and scoring criteria were used to assess PD-L1 expression. PD-L1 expression was assessed by IHC using SP142 antibody for atezolizumab (IC ≥ 1%), 22C3 antibody for pembrolizumab (CPS≥10%), and SP142 antibody for durvalumab (IC ≥ 1%). In the exploratory analysis of IMpassion130 study, SP142 antibody was used to detect the PD-L1 expression by IHC in advanced TNBC on tumor-infiltrating immune cells, and the IC positive population of PD-L1 accounted for 41% in TNBC, which included most tumor cell PD-L1 positive patients (8.7%). IC-positive population could achieve clinical benefit regardless of the expression of PD-L1 on tumor cell. Therefore, PD-L1 expression on IC was a biomarker for predicting the benefit of atezolizumab combined with nab-paclitaxel in metastatic TNBC. Besides, the exploratory analysis also assessed the conformity of 22C3 and SP142 in detecting PD-L1 expression, about 80% concordance was observed between IC ≥ 1% (SP142) and CPS≥10% (22C3) (Rugo et al., 2020; Schmid et al., 2020). However, it should be noted that these two approaches were not equivalent (Dill et al., 2017; Torlakovic et al., 2020; Noske et al., 2021). 2) Only five RCTs were included in our meta-analysis, and we could not performed a detailed subgroup analysis due to the small number of included studies, which might lead to bias. In addition, some clinical RCTs in TNBC patients are currently ongoing, and further analysis of clinical data was required to draw sound conclusions in the future. 3) The differences between PD-1 and PD-L1 inhibitors were also worth considering.

In our meta-analysis, we suggested OS benefit in PD-L1 positive advanced or metastatic TNBC, and the subgroup analysis showed that ICIs combination nab-paclitaxel or ICIs alone might be a better choice compared with ICIs combination paclitaxel in the PD-L1 positive TNBC group. In terms of PFS, no significant PFS benefit was found in PD-L1 positive patients, but the subgroup analysis indicated that a significant benefit of PFS was observed for ICIs combination chemotherapy compared to ICIs alone in the first-line treatment in PD-L1 positive TNBC. No significant improvement was observed for OS or PFS in PD-L1 negative group. The results of this meta-analysis may be beneficial to clinicians in forming better treatment strategies to manage TNBC patients. However, further research is needed, given the limited number of studies currently available for data analysis.