A meta-analysis of application of PD-1/PD-L1 inhibitor-based immunotherapy in unresectable locally advanced triple-negative breast cancer

Breast cancer (BC) is a life-threatening disease [1,2] and has the highest incidence in females worldwide [3,4]. Although the overall prognosis of BC is relatively acceptable, the prognosis of triple-negative BC (TNBC) is considerably poor [5–8]. TNBC, accounting for 15–20% of breast cancers, is more aggressive and more likely to metastasize and recur [9–11]. Due to its special molecular phenotype [4,12], TNBC is not sensitive to endocrine therapy or molecular-targeted therapy [8,13]. Standardized TNBC treatment regimens remain up for discussion [14,15]. Cytotoxic chemotherapy is currently the main treatment for TNBC [16–18]. However, studies in the past two decades have shown the objective response rate of systemic chemotherapy in adjuvant and metastatic treatment to be only 30–40% [19,20]. Moreover, even if TNBC patients are treated with systemic chemotherapy, 5-year survival can only be expected in <30% of metastatic patients after diagnosis [21–23]. Since residual metastatic disease can easily result in tumor recurrence, almost all patients with metastasis will die due to organ failure, reflecting unsatisfactory treatment overall [24,25].

Therefore, the development of new treatment options and targets is urgently needed for TNBC patients. Over the past few years, immunotherapy has been a novel treatment option for TNBC [26,27]. TNBC has a higher number of tumor-infiltrating lymphocytes (TILs) and higher PD-L1 protein and/or mRNA expression compared with other BC subtypes [28,29]. The ligand, PD-L1, and its receptor, PD-1, are involved in the regulation of immune tolerance [30–32]. Therefore, researchers have developed a series of PD-1/PD-L1 inhibitors, such as atezolizumab [33], pembrolizumab [34], nivolumab [35], durvalumab [36], avelumab [37] and so on, and conducted experiments to identify biomarkers for PD-1/PD-L1 inhibitor therapy. At present, the most important biomarker is considered to be the expression of PD-L1, and related evaluation methods include tumor cell proportion score, combined positive score and immune cell proportion score. However, there is a discrepancy between the expression of PD-L1 and the response to PD-1/PD-L1 inhibitor immunotherapy, and there remain doubts about PD-L1 expression as a key point for PD-1/PD-L1 inhibitor therapy. Many studies are still exploring reliable biomarkers beyond PD-L1, such as microsatellite instability, mismatch repair, tumor mutation burden and so on. Therefore, many studies have conducted immunotherapy with PD-1/PD-L1 inhibitors for TNBC without limiting the expression of PD-L1.

Given the diversity of survival results among different immunotherapy regimens, the details of newly proposed immunotherapy regimens and immune agents have been hotly discussed issues among healthcare professionals and researchers in related fields. The KEYNOTE-119 trial [38] found that the median overall survival (OS) was 9.9 months in the pembrolizumab group and 10.8 months in the chemotherapy group without a statistically significant difference in OS (hazard ratio [HR] = 0.97; 95% CI = 0.82–1.15), while the SAFIR02-BREAST IMMUNO trial [39] indicated that durvalumab did improve OS in patients with TNBC (HR = 0.54; p = 0.037). Therefore, this meta-analysis was designed to evaluate the efficacy and safety of immunotherapy, mainly focusing on PD-1/PD-L1 inhibitors, in locally advanced or metastatic TNBC that is unresectable, and provide a primary exploration of immunotherapy for TNBC.

Methods

Literature search

A search of relevant studies investigating the efficacy and safety of immunotherapy in TNBC patients published before October 2021 was conducted in PubMed, EBSCO, Web of Science and Cochrane Library databases. The keywords used were: “triple-negative breast cancer,” “programmed cell death ligand 1 inhibitor” and “immunotherapy.” The complete retrieval formula that was used to identify the related studies included: “triple negative breast cancer” OR “TNBC” OR “triple-negative breast carcinoma” OR “triple-negative breast neoplasms” AND “pembrolizumab” OR “nivolumab” OR “atezolizumab” OR “durvalumab” OR “avelumab” OR “cemiplimab” OR “programmed cell death 1” OR “programmed cell death ligand 1” OR “PD-L1” OR “PD-1” OR “immunotherapy” OR “immune checkpoint.” Only the most recent article or the one with the highest quality was included when duplicates were found. Literature in the reference lists of retrieved studies and recent reviews were also thoroughly reviewed to avoid missing possible studies that may need to be included. The population, intervention/exposure, comparison, outcome and setting criteria were used to describe the research aim. This meta-analysis was conducted following the guidelines of the Preferred Reporting Items for Systematic Review and Meta-Analysis 2009 Checklist [40]. The Prospero registration number for this meta-analysis is CRD42021289859.

Selection criteria

The specific eligibility criteria for this study were as follows: all included studies were randomized controlled trials (RCTs) and were limited to clinical studies. The included studies assessed only patients who were diagnosed with unresectable locally advanced or metastatic TNBC or had been categorized as a subgroup with accessible data attached. Patients in the experimental groups in the RCTs were treated with PD-1/PD-L1 inhibitor-based immunotherapy while those in the control groups were treated with other treatments. All studies had full-text articles available. Exclusion criteria: patients with early TNBC or postoperative adjuvant therapy; studies not reporting relevant outcomes or detailed data; preclinical studies or phase I clinical trials; and studies not published in English.

Two authors independently applied the search strategy to the selected studies and independently reviewed the titles and abstracts of these articles to judge whether they met the inclusion criteria. Full texts were retrieved in the case of doubts. When necessary, authors were contacted for additional information. In case of differences of opinion, a discussion was held with a third author. When consensus failed to be reached, the study was excluded.

Data extraction & quality assessment

A jointly agreed upon data collection form was used to extract all data. The extracted information included the author's name, year of publication, trial duration, NCT number, country, age of patients, therapeutic regimen, trial phase, follow-up time, number of patients, primary outcomes and secondary outcomes. To ensure the objectivity and accuracy of the input data, two researchers independently extracted data from each study. Disagreements were resolved by consensus or consultation with a third author. The quality of each RCT was recorded according to the following indicators from the Cochrane Risk-of-Bias assessment tool score: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting and free of other bias.

Objectives & end points

The primary objective of this meta-analysis was to assess the OS and progression-free survival (PFS) of patients treated with PD-1/PD-L1 inhibitor-based immunotherapy for unresectable locally advanced or metastatic TNBC compared with others. OS was defined as the proportion of tumor patients who survived after various comprehensive treatments from random allocation at the beginning of the trial to the latest follow-up time. The PFS was defined as the proportion of patients without disease progression or death for any reason from random allocation at the beginning of the trial to the latest follow-up time. The secondary objectives included safety and efficacy based on subgroup analysis of PD-L1 status.

Statistical analysis

RevMan 5.3.5 software for Windows^® and Stata software version 12 (StataCrop, TX, USA) were used for analysis. Heterogeneity across included studies was tested by Q statistics and the I² statistic. In accordance with the Cochrane manual and experimental characteristics, I² values between 0 and 30% indicate mild or insignificant heterogeneity, 30∼70% indicate moderate heterogeneity and 70∼100% indicate high or significant heterogeneity [41]. The CIs of HRs and risk ratios (RRs) were set at 95% to determine the efficacy of immunotherapy. A random-effects model was applied to incorporate data from different therapeutic regimens to increase the credibility of the results. When more than ten studies [42,43] were included, sensitivity analysis and publication bias tests were performed to evaluate the stability and reliability of the results. Begg's rank correlation test was applied to test publication bias; p-values < 0.05 were considered statistically significant.

Results

Study selection

Based on the previously described search formula, a total of 2346 relevant articles were identified through preliminary searches in PubMed, EBSCO, Web of Science and Cochrane Library databases. No other records were identified from other sources. A total of 393 duplicate articles were recognized and deleted; 1501 articles were excluded based on the content in the title or abstract. The remaining 452 articles were reviewed through full-text reading. Among them, 446 articles were eliminated because they were not RCTs (n = 354), had no TNBC (n = 60) or immunotherapy (n = 27) involved and were not in English (n = 5). Finally, 6 articles [39,44–48] comprising 3183 patients were eligible for this meta-analysis. The detailed search and study selection process is shown in Figure 1.

Study characteristics

Of 3183 patients (1839 in the experimental groups and 1344 in the control groups), 55.2% were diagnosed with PD-L1-positive (PD-L1+) TNBC and 44.2% were diagnosed with PD-L1-negative (PD-L1-) TNBC. The remaining 0.6% of patients were unknown in terms of PD-L1 status. All participants were female. Four trials were in phase III and two were in phase II. Patients in the experimental groups were all treated with PD-1/PD-L1 inhibitor-based immunotherapy such as pembrolizumab, nivolumab, atezolizumab, durvalumab, avelumab and cemiplimab, while those in the control groups were treated with other treatments such as paclitaxel, gemcitabine, vinorelbine, capecitabine, eribulin or carboplatin. All six studies were published between 2020 and 2021, with median follow-up times ranging from 7.7 to 31.5 years. Two trials are still ongoing. The random allocation ratio was 2:1 in three of the included studies and 1:1 in the other three. Three studies were open-label trials, two were double-blind trials (participant, investigator) and one was a quadruple-blind trial (participant, care provider, investigator, outcomes assessor). All studies had accessible data for PFS and five studies had accessible data for OS. All studies defined PD-L1+ as the covering rate of PD-L1-stained tumor-infiltrating immune cells over the total tumor area is equal to more than 1% in the metastatic lesion sample. Most of the PD-L1 protein expressions were assessed by immunohistochemistry using the SP142 antibody. Details of the included clinical trials are listed in Table 1 & Supplementary Table 1.

Bias risk assessment

The Cochrane Collaboration tool was applied to objectively evaluate the quality of RCTs included in this meta-study. The tool employs six targets overall and risk of bias falls into either the ‘low risk’, ‘high risk' or 'unclear risk' categories. All included studies were of higher quality, according to the quality evaluation results (Supplementary Figure 1).

OS of intention-to-treat & PD-L1+/- populations

In five studies (1273 experimental patients and 1063 control patients) OS was selected to evaluate the efficacy of PD-1/PD-L1 inhibitor-based immunotherapy with other chemotherapy combined for the treatment of unresectable locally advanced or metastatic TNBC. Among them, two studies (283 experimental patients and 279 control patients) had available PD-L1- population data and four studies (597 experimental patients and 501 control patients) had available PD-L1+ population data. The analysis of the results showed that immunotherapy did not improve the OS of the intent-to-treat (ITT) population (HR = 0.90; 95% CI = 0.78–1.04; p = 0.144) with mild heterogeneity (I² = 24.0%). Immunotherapy based on PD-1/PD-L1 inhibitors could not prolong the OS of unresectable locally advanced or metastatic TNBC compared with nonimmunotherapy treatment (Figure 2). In addition, a subgroup analysis based on PD-L1 status showed that there was no statistically significant difference in OS between the immunotherapy group and the nonimmunotherapy group no matter the TNBC subtype, whether PD-L1- TNBC (HR = 0.83; p = 0.545) or PD-L1+ TNBC (HR = 0.85; p = 0.22; Figure 3). Analysis of the relevant binary classification data yielded similar results. PD-1/PD-L1 inhibitor-based immunotherapy did not prolong the OS of the ITT population (RR = 1.00; 95% CI = 0.88–1.14; p = 0.98; I² = 75.0%) or the PD-L1+ TNBC population (RR = 0.96; 95% CI = 0.82–1.12; p = 0.59; I² = 62.0%; Table 2). Hence, PD-1/PD-L1 inhibitor-based immunotherapy has no effect on the OS of unresectable locally advanced or metastatic TNBC and has no relationship to PD-L1 status.

PFS of ITT & PD-L1+/- populations

For six studies (1839 experimental patients and 1344 control patients), PFS was determined to evaluate the efficacy of immunotherapy. Among them, three studies (424 experimental patients and 419 control patients) had PD-L1- population data and five studies (1022 experimental patients and 712 control patients) had PD-L1+ population data. The results demonstrated that the PFS of the ITT population showed statistically significant differences compared with others after various kinds of chemotherapy (HR = 0.82; 95% CI = 0.76–1.14; p < 0.001; I² = 0%; Figure 4). That is, PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy can prolong the PFS of patients with unresectable locally advanced or metastatic TNBC and benefit the ITT population. In addition, subgroup analysis further illustrated that immunotherapy has no effect on the PD-L1- TNBC population (HR = 1.17; 95% CI = 0.82–1.66; p = 0.389; I² = 71.8%; Figure 5A), but can prolong the PFS of the PD-L1+ population specifically, with a significant difference and no heterogeneity (HR = 0.68; 95% CI = 0.61–0.76, p < 0.001; Figure 5B). Analysis of relevant binary classification data suggested that PD-1/PD-L1 inhibitor-based immunotherapy has no effect on the PFS of the ITT population (RR = 1.16; p = 0.21; I² = 96.0%; Table 2), but can prolong the PFS of the PD-L1+ TNBC population compared with any other discussed chemotherapies with a statistically significant difference (RR = 0.83; p = 0.02; I² = 88.0%; Table 2) That is, PD-1/PD-L1 inhibitor-based immunotherapy has a beneficial effect on the PFS of unresectable locally advanced or metastatic TNBC and has a greater effect on the PD-L1+ population among all PD-L1 situations in the targeted population.

Safety analysis

The incidence of adverse events (AEs) was used to evaluate the safety of PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy. The results showed that there was no difference between immunotherapy combined with chemotherapy and other chemotherapies in any-grade all-cause AEs (RR = 1.01; p = 0.07), any grade (RR = 1.01; p = 0.07) or higher than grade 3 (RR = 0.92; p = 0.60) treatment-related AEs and AEs that potentially lead to treatment discontinuation (RR = 1.49; p = 0.21). However, statistically significant differences were calculated in higher than grade 3 all-cause AEs (RR = 1.08; p = 0.02), any grade (RR = 2.31; p < 0.001) and higher than grade 3 (RR = 2.48; p < 0.001) immune-mediated AEs. This indicates that, compared with other chemotherapies, PD-1/PD-L1 inhibitor-based immunotherapy is more likely to cause higher than grade 3 all-cause AEs, any grade and higher than grade 3 immune-mediated AEs in involved patients (Supplementary Table 2). PD-1/PD-L1 inhibitor-based immunotherapy can lead to an increase in the risk of cough (RR = 1.39; 95% CI = 1.14–1.69; p = 0.001) and abnormal thyroid function (hyperthyroidism; RR = 15.17; 95% CI = 5.77–39.83; p < 0.001) and hypothyroidism (RR = 3.85; 95% CI = 2.78–5.32; p < 0.001)) and pneumonia (RR = 2.95; 95% CI = 1.67–5.22; p < 0.001), but no change in the frequency and incidence of other AEs (including alopecia, anemia, diarrhea, dyspnea, fatigue, nausea, neutropenia and pyrexia; p > 0.05; Supplementary Table 3). When it comes to grade 3 or higher AEs, PD-1/PD-L1 inhibitor-based immunotherapy mainly increased the risk of pneumonia (RR = 3.74; 95% CI = 1.45–9.65; p = 0.006), with no statistical difference in other AEs (i.e., alopecia, ALT increased, anemia, decreased appetite, diarrhea, dyspnea, fatigue, nausea, neutropenia and pyrexia; p > 0.05; Supplementary Table 4).

Discussion

Based on the data, PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy had significantly improved PFS in the ITT population with unresectable locally advanced or metastatic TNBC but had no contribution to OS compared with chemotherapy. Subgroup analysis, which was stratified according to PD-L1 receptor status, illustrated that PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy had significantly prolonged PFS in the PD-L1+ TNBC population and not in the PD-L1- TNBC population. As for OS, there was no statistically significant difference between PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy and other types of chemotherapies for either the PD-L1- population or the PD-L1+ population. Similar conclusions were also reached by the related dichotomous data analysis. Safety analysis suggests that, compared with chemotherapy, PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy will increase higher than grade 3 all-cause AEs, any grade and higher than grade 3 immune-mediated AEs. More specifically, AEs of any grade caused by immunotherapy combined with chemotherapy were mainly cough, abnormal thyroid function (including hyperthyroidism and hypothyroidism) and pneumonia, while AEs above grade 3 mainly included pneumonia.

TNBC mainly occurs in young women under the age of 40 before menopause [7,8]. Compared with other BC subtypes (luminal A, luminal B or HER2+ BC), TNBC is more aggressive and more likely to metastasize and recur, resulting in poor prognosis [10,49]. Chemotherapy and targeted therapy have shown unsatisfactory curative effects on TNBC in terms of poor improvement in prognosis [21–23]. Therefore, in recent years, immunotherapy has provided novel treatment options for TNBC due to its remarkable activity and great therapeutic potential in a wide range of tumors [27]. It has been widely recognized that the immune system plays an important role in the development and progression of BC in terms of occurrence, progression and treatment [50,51]. In the early stages of BC, acute inflammation activates innate immunity and sequentially triggers tumor-specific T-cell responses [52]. At this stage, the body can show immune-mediated rejection to initial tumors, or produce tumor cell variants that can evade immune response [53,54]. A transition from acute inflammation to chronic inflammation will eventually be accomplished, creating a complex tumor microenvironment that greatly allows immune escape and tumor progression [55,56].

Under normal physiological conditions, T cells can maintain the equilibrium of self-tolerance and tumor immunity by switching activation and inhibition signals [57–60]. Tumors can express immune inhibitory signals, such as CTLA-4 [61,62] and PD-1 [63], leading to a weakened immune response against pathological antigens [64,65]. The CTLA-4 signal pathway is involved in limiting the initiation of T-cell responses in lymph nodes. PD-1 limits the activity of T cells in the tumor microenvironment after initiation and is responsible for the immune escape of cancer [58,66].

The higher genomic instability [67] and mutational burden [68] of TNBC lead to a greater possibility of generating new antigens [28]. Meanwhile, compared with other BC subtypes, TNBC has a higher number of TILs [29] and can produce more PD-L1 protein [69,70] or express more related mRNA [30,31]. It is believed that the expression of PD-L1 is significantly related to the presence of TILs [71]. There is a linear relationship between matrix TIL content and clinical results [71]. Solid tumors that carry TILs and produce PD-L1 are more likely to respond to PD-1/PD-L1 blockade, suggesting the possible mechanism of immunotherapy being active in the management of TNBC [72–74]. The regulatory mechanism of PD-L1 expression in TNBC patients may participate in the regulatory feedback of immunity [32,75,76].

In TNBC cells, high PD-L1 expression can induce the generation of a tumor microenvironment that can inhibit antitumor immune response [69,77]. PD-1/PD-L1 inhibitors inhibit the interaction between PD-1 and PD-L1 on the membrane of tumor cells and immune cells, thereby antagonizing PD-1 signal transduction in inhibiting T-cell activation in the effective stage of the immune response [78–80]. At the same time, PD-1/PD-L1 inhibitors can induce the cytotoxic function of T cells by upregulating the gene expression of proinflammatory cytokines (IL-2, IFN-γ and TNF-α) [81–84], downregulating the gene expression of anti-inflammatory cytokines (IL-10 and TGF-β) [85,86], prevent T-cell lysis and inhibit antigen tolerance. By reactivating the specific immune response against tumors and effectively inhibiting the metastasis of tumor cells [72–74], immunotherapy based on PD-1/PD-L1 inhibitors can significantly prolong the PFS of the ITT population and, to be more specific, effective in the PD-L1+ population but not in PD-L1- population.

Immunotherapy based on PD-1/PD-L1 inhibitors did not improve the OS of the ITT population and had no effect on either the PD-L1+ subtype or the PD-L1- subtype. The following three points may help explain this finding. First, the activity of immune response in the human body (mainly) depends on the duration of interaction between antigen-presenting cells and target cells [87,88]. When T cells are activated by the PD-1/PD-L1 inhibitor, the PD-1/PD-L1 axis is blocked, leading to disturbance of tumor cells evading from the immune response, thus indicating pharmacological effect [89,90]. However, the anticancer immune response usually fails when antigen-specific T cells and cytokines are strongly regulated, called “immune exhaustion” by researchers [91]. Therefore, PD-1/PD-L1 inhibitors can significantly improve the PFS of unresectable locally advanced or metastatic TNBC in the early follow-up time, but effectiveness, in the long run, is subsequently dependent on the immune regulatory response over time. As the occurrence of immune failure does not significantly improve OS, researchers can try to compare intermittent dosing regimens and continuous dosing regimens to explore the best dosing regimen of inhibitor-based immunotherapy. This may also be related to drug holidays, where a conscious decision is made to discontinue a certain medication for a period of time after long-term use. Second, the activity of immune responses in the human body also depends on the exercise capacity of T cells [88]. PD-L1 can reduce the motility of CD8⁺ T cells to prevent CD8⁺ T cells from performing their functions, to keep them in a state of negative immunoregulation for a long time [92–94]. The resulting decline in the body's immunity and the binding obstacles of antigen-presenting cells and target cells (e.g., the lymphocytes' challenge to completely move to the tumor cell area for infiltration) are factors that limit the effect of PD-1/PD-L1 inhibitors [95,96]. Therefore, the obstacle of T-cell motor function may also explain why immunotherapy failed to significantly improve OS in this study. Last, the number of trials included in this study was limited and potentially accidental factors may cause false negative results in the analysis. The improvement with immune therapy in OS in human trials will be further explored with the development of more clinical trials and the release of experimental data.

Immunotherapy based on PD-1/PD-L1 inhibitors combined with other chemotherapies can increase the risk of cough, abnormal thyroid function (including hyperthyroidism and hypothyroidism) and pneumonia in TNBC patients. It is believed that the most likely mechanism of developing those AEs is that the immune system activated by PD-1/PD-L1 inhibitors not only targets tumor cells, but is also inductive to the death of normal cells, tissues and organs [94]. The thyroid gland is more vulnerable to autoimmune attacks than any other organ [97]. PD-1/PD-L1 blockade can induce autoantibodies' production and secretion in T-cell-dependent B cells [98,99]. The presence of thyroid autoantibodies and early elevation of serum thyroglobulin levels may increase the risk of thyroid dysfunction [100,101]. In addition, Krieg et al. [102] found that IFN-γ and IL-2 are increasingly expressed by CD4⁺ Th1 after the anti-PD-1 treatment. Higher levels of IL-2 can not only induce the binding between human major histocompatibility antigen II and thyroid cell self-antigens but also encourage the killing effect of CD8⁺ cytotoxic T lymphocytes on the cells in the thyroid gland [103,104]. This may explain why the anti-PD-1 treatment group has a higher incidence of developing thyroid dysfunction than other treatment groups. Similarly, pneumonia is also suspected to be caused by attack from the immune system, which is mainly manifested as interstitial pneumonia-like changes [105]. Studies have shown that patients with chronic pneumonia, patients who have received lung radiotherapy and patients with a history of chronic obstructive pulmonary disease are more prone to develop interstitial pneumonia after immunotherapy, and cough maybe is one of the symptoms of interstitial pneumonia [106]. Generally speaking, since most of the trials included in this meta-analysis adopted a grouping scheme of immunotherapy combined with chemotherapy versus chemotherapy alone, an increase in the incidence of certain adverse reactions, mainly focusing on thyroid and lung, in the experimental group is inevitable but safe and manageable.

Reviewing similar studies, only Ali et al. [107] and Lazarus et al. [108] conducted a meta-analysis on the efficacy of immunotherapy in patients with advanced TNBC, but only 2 of the 12 trials included in the former were RCTs (KEYNOTE-355, KEYNOTE-119), only one of 7 trials included in the latter was an RCT (IMpassion130) and the rest were single-arm studies. The smaller sample size and the smaller number of RCTs limit the credibility and comparability of their research. Therefore, this meta-analysis is expected to be the first to fully and systematically explore the efficacy and safety of PD-1/PD-L1 inhibitor-based immunotherapy for unresectable locally advanced or metastatic TNBC, as the included studies were all high-quality RCT studies, and all trials were phase II and above, making this research more convincing and meriting this research over others in terms of convincingness and credibility. Also, data was grouped according to PD-L1 receptor status to explore the efficacy of PD-1/PD-L1 inhibitor-based immunotherapy on different detailed TNBC subtypes to increase the applicability of the research conclusions.

The limitations of this study should also be discussed. First, the number of studies included in this meta-analysis was limited, and long-term survival outcomes for some trials were not published or have not been updated. Only two studies involved available OS data for the PD-L1- population, and four studies had OS data of the PD-L1+ population available, resulting in relatively poor effective data that is not constructive enough to identify the advantages of immunotherapy. Second, owing to obstacles to obtaining more information on the clinical trials, this meta-analysis only included one subgroup analysis of PD-L1 receptors. Subgroup analysis based on CD8⁺ or CD4⁺ T-cell levels is expected in the future to reflect the efficacy of immunotherapy. Lastly, this meta-analysis contained a variety of treatment options, including different dosing regimens, resulting in difficulty in clarifying the best treatment option for TNBC. Therefore, a large-scale RCT is still needed to further improve novel treatment options and verify the relevant conclusions. In short, this meta-analysis had drawn some meaningful conclusions, which may provide a new reference for immunotherapy, especially for patients with unresectable locally advanced or metastatic TNBC.

Conclusion

This meta-analysis found that PD-1/PD-L1 inhibitor-based immunotherapy combined with chemotherapy can significantly increase the PFS of unresectable locally advanced or metastatic TNBC population, but had no effect on OS, compared with chemotherapy alone. Immunotherapy had a significantly prolonged effect on the PFS in the PD-L1+ population but had no effect on OS. No effect on the PD-L1- population (PFS and OS) was found. AEs of any grade caused by immunotherapy mainly focused on cough, abnormal thyroid function (including hyperthyroidism and hypothyroidism) and pneumonia, and AEs above grade 3 were mainly pneumonia, which was all generally safe and manageable.