PD-1/PD-L1 immune checkpoint inhibitors in the treatment of unresectable locally advanced or metastatic triple negative breast cancer: a meta-analysis on their efficacy and safety

Abstract

Background

Triple negative breast cancer (TNBC) represents a particularly aggressive and clinically challenging subtype of breast cancer, characterized by its invasive nature and generally poor prognosis. Treatment options for unresectable TNBC are limited. In recent years, the advent of PD-1/PD-L1 immune checkpoint inhibitors has offered a promising new treatment option for unresectable TNBC. The role of PD-1/PD-L1 immune checkpoint inhibitors (ICIs) in unresectable TNBC management remains a subject of debate. This article aims to synthesize evidence from randomized controlled trials (RCTs) through a meta-analysis (MA) to provide a comprehensive evaluation of the efficacy and safety profile of ICIs in the treatment of unresectable TNBC.

Method

We searched PubMed, Embase, Cochrane library, Web of Science, and ClinicalTrials.gov for the eligible RCTs which compared the efficacy and safety of PD-1/PD-L1 ICIs and chemotherapy alone. The outcomes analyzed included overall survival (OS), progression-free survival (PFS), objective response rate (ORR) and treatment-related adverse effects (AEs).

Results

This meta-analysis included 11 trials. Therapy with PD-L1 inhibitors was superior to chemotherapy in terms of OS in both the intention-to-treat (ITT) population and the PD-L1-positive population. (ITT: HR: = 0.90 [0.81, 0.99], P = 0.04, I² = 48%; PD-L1 + : HR = 0.82 [0.70, 0.95], P = 0.01, I² = 64%); In terms of PFS, treatment with PD-L1 inhibitors prolonged PFS in both the ITT and PD-L1-positive populations compared with chemotherapy (ITT: HR: = 0.85 [0.77, 0.93], P = 0.0006, I² = 46%; PD-L1 + : HR = 0.72 [0.62, 0.83], P < 0.00001, I² = 70%, Fig. 5); Compared with chemotherapy alone, treatment with PD-1 inhibitors prolonged OS in both the PD-L1-positive and ITT populations. (ITT: HR = 0.87 [0.78, 0.96], P = 0.007, I² = 71%; CPS ≥ 1: HR = 0.81 [0.71, 0.92], P = 0.001, I² = 39%); In terms of PFS, therapy with PD-1 inhibitors improved PFS in both the ITT population and the PD-L1-positive population compared with chemotherapy. (ITT: HR = 0.79 [0.70, 0.90], P = 0.0004, I² = 0%; CPS ≥ 1: HR = 0.71 [0.61, 0.83], P < 0.0001, I² = 0%; CPS ≥ 10: HR = 0.67 [0.53, 0.84], P = 0.0008, I² = 0%).

Conclusion

Both PD-1 and PD-L1 inhibitors can offer survival benefits to TNBC patients, with the primary beneficiaries being those who are PD-L1-positive. However, immunotherapy can also lead to an increase in treatment-related adverse events. Therefore, it is essential to conduct a risk assessment for each patient before starting treatment to prevent the occurrence of serious adverse reactions.

Trial registration

This systematic review study has been filed with PROSPERO (Registration number: CRD42024571775).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-024-13105-9.

Introduction

The 2022 Global Malignant Tumor Statistics Report indicates that breast cancer has the second-highest incidence worldwide, with the highest mortality rate among female malignant tumors. Triple negative breast cancer (TNBC), a distinct subtype, is defined by the absence of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 (HER2) expression [1]. It constitutes 10% to 20% of all breast cancer cases [2]. TNBC is marked by aggressive invasiveness, poor differentiation, and elevated tumor immunogenicity, making it the subtype with the highest risk of recurrence and the poorest prognosis [3–5]. Current treatment options for TNBC are limited, primarily relying on systemic chemotherapy, which is susceptible to drug resistance. In light of these challenges, the potential of targeted therapies and immunotherapies has gained significant attention in recent years.

At present, anti-PD-1/PD-L1 immunotherapy has demonstrated good success in tumour treatment. Blocking PD-1/PD-L1 immunotherapy targets and increasing T cell immune response mechanisms have emerged as important anti-tumor therapies. The binding of PD-L1 on tumour cell surfaces to PD-1 on active T cell surfaces can block T cell activation, result in immunological suppression, and cause tumour cells to evade the immune system. Therefore, the interaction between PD-1/PD-L1 plays a crucial role in tumor associated immune escape [6]. Currently, the FDA has approved the combination of pabolizumab and chemotherapy for neoadjuvant therapy, and it can be used as a single adjuvant therapy after surgery for high-risk early triple negative breast cancer patients [7]. In addition, in 2019, the FDA approved the combination of Atezolizumab and albumin paclitaxel for first-line treatment of PD-L1 positive, metastatic TNBC. However, there have been differences in the research results of TNBC immunotherapy in recent years. The KEYNOTE-355 study suggests that the combination of pembrolizumab and chemotherapy shows significant clinical benefits in PFS (CPS ≥ 10) of metastatic TNBC patients, but in the KEYNOTE-119 study, pembrolizumab did not demonstrate significant advantages. Similarly, contradictory results were observed between the IMpassion130 and IMpassion131 trials. Therefore, t debate persists regarding the substantial impact of PD-1/PD-L1 inhibitors on survival rates. This meta-analysis evaluated the efficacy and safety of PD-1/PD-L1 inhibitors in the treatment of unresectable locally advanced or metastatic TNBC from multiple perspectives, including different PD-L1 expression levels and the combination of ICIs with chemotherapy. This can help clinicians gain a more comprehensive understanding of the value of ICIs for treating unresectable TNBC.

Methods

The meta-analysis was conducted in accordance with the latest version of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. [8].

Data sources and search strategy

Search PubMed, Embase, Web of Science, Cochrane Library, and ClinicalTrials.gov from the establishment of the database until August 2024. The search equation is: (TNBC OR triple negative breast cancer OR triple-negative breast carcinoma OR triple-negative breast neoplasms) AND (PD-L1 Inhibitor OR PD-1 Inhibitor OR programmed cell death 1 OR PD-L1 OR PD-1 OR programmed cell death ligand 1 OR immunotherapy OR immune checkpoint OR pembrolizumab OR nivolumab OR atezolizumab OR durvalumab OR avelumab OR cemiplimab OR sintilimab). The Prospero registration number for this meta-analysis is CRD42024571775. For detailed search results, please refer to Supplementary 1.

Inclusion and exclusion criteria

The criteria for selecting studies in the PICOS format were:

• (1) Participants: The included studies only evaluated patients aged > 18 years diagnosed with unresectable locally progressed or metastatic TNBC, using PD-L1 evaluation.

• (2) Intervention: PD-1/PD-L1 inhibitor-based immunotherapy;

• (3) Comparisons: The experimental group received immunotherapy based on PD-1/PD-L1 inhibitors, whereas the control group only received chemotherapy;

• (4) Outcomes: Efficacy outcomes are overall survival (OS), which is defined as the time from randomization to death due to any cause, progression-free survival (PFS), which is defined as the time from randomization to cancer progression or death due to any cause (as assessed by iRECIST 1.1), and objective response rate (ORR). The safety outcomes are serious AEs (SAEs), and immune related adverse (irAE).

• (5) The research type is RCT, and the language is English.

Exclusion criteria:

• (1) Patients with early TNBC or postoperative adjuvant therapy;

• (2) Research that has not reported relevant results or detailed data;

• (4) Preclinical studies or Phase I clinical trials.

• (5) Review, case reports, letters, cell experiments, etc.

Quality assessment and data extraction

Two independent researchers (Peimeng You and Xuhuan Li) discussed and evaluated bias risk using Cochrane's bias risk tool. The bias risk included in the literature was evaluated from the following six aspects: random sequence generation (selection bias), allocation concept (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attack bias), and selective reporting (reporting bias). These six aspects were evaluated based on low risk, high risk, and uncertain risk. Data extraction is carried out through mutual agreement, and all potential disagreements are resolved through consensus [9].

Two researchers (Zhanglei Yang and Hanxi Xiao) independently extracted the following information: the extracted data included study title, first author and publication year, trial phase, NCT number, number of patients receiving treatment in the trial and control groups, average age of participants, number of PD-L1 positive individuals, treatment plans for the trial and control groups, primary and secondary outcome measures. The result data includes OS, PFS, ORR, SAE, irAE [Hypothyroidism, Hyperthyrobism, Pancrititis, Rash, Hepatitis (diagnosis)]. When reporting multiple results, only the latest result is used. To ensure the objectivity and accuracy of the input data, two researchers extracted data from each study independently. In cases of disagreement, it can be resolved through mutual agreement or consultation with a third author.

Statistical analysis

In the results, independent statistician Yinan Yu used Revman version 5.4 for statistical analysis. The generic inverse variance was used to calculate and record log [HR] and corresponding SE for PFS and OS hazard ratio (HR) data. The risk ratio (RR) of successful ICIs combined with chemotherapy versus chemotherapy alone was evaluated for other results. A binary data type was selected, and the CI of HR and RR was set to 95% to determine the efficacy of immunotherapy. In statistical analysis, Q-tests and I² statistics are used to evaluate heterogeneity. Statistical heterogeneity was defined as I² > 50%. To make the results of the meta-analysis more credible and to represent the actual effects more accurately, a random effects model is used if the I² statistic exceeds 50%. Conversely, if the I² statistic is less than 50%, a fixed effects model is used. Sensitivity analysis is used to evaluate the stability and reliability of the results. A P-value < 0.05 is considered statistically significant. A P-value < 0.05 is considered statistically significant. MA has been executed using Review Manager (RevMan 5.4) software.

Results

Study and data selection

Following our predefined search strategy, we initially identified 4,015 studies from various databases: 1,009 from PubMed, 636 from the Cochrane Library, 26 from EMBASE, 1,917 from Web of Science, and 427 from ClinicalTrials.gov. After removing duplicates, 2,453 studies remained. Upon reviewing titles, we excluded 2,291 studies that were reviews, meta-analyses, or other non-relevant literature. Subsequently, after a meticulous full-text review, we further eliminated 151 articles based on our strict inclusion and exclusion criteria, which included missing full texts, neoadjuvant therapy studies for early-stage triple-negative breast cancer, non-randomized controlled trials, and those lacking outcome measures. Consequently, this meta-analysis encompassed 11 randomized controlled trials (RCTs), encompassing a total of 4,314 patients. The detailed literature search flowchart is shown in Fig. 1 [10–20].

Fig. 1.

Flow diagram of study selection process

Study characteristics and quality assessment

The basic characteristics of the patients included in the study are shown in Table 1. Among the 11 articles ultimately included, all were randomized controlled trials (RCTs). There were 7 studies related to PD-L1 inhibitors and 4 studies related to PD-1 inhibitors. Among the studies related to PD-L1 inhibitors, there were 6 studies on Atezolizumab, 1 study on Durvalumab. Among the studies on PD-1 inhibitors, there were 3 studies on Pembrolizumab and 1 study on Toripalimab. COLET, ALICE, and SAFIR02-BREAST are Phase II clinical trials, while the remaining 8 are Phase III clinical trials. A total of 4,314 TNBC patients were included across 11 studies, of which 2,528 patients were PD-L1 positive. Among these, the SAFIR02-BREAST [16] and KEYNOTE-119 [18] experimental groups were treated with ICIs as a single agent, while the other 8 experimental groups in the studies were treated with ICIs combined with chemotherapy. Among the included literature, 10 studies reported the OS results for subjects, 8 of which reported the OS results specifically for PD-L1 positive patients. Additionally, 10 studies reported PFS results for subjects, 9 of which included PFS results for PD-L1 positive patients. Furthermore, 10 studies reported ORR results for subjects, 8 of which reported ORR results for PD-L1 positive patients.The included literature used various survival data statistical methods, including Kaplan Meier survival analysis, Cox proportional hazards regression, and log rank test.

Table 1.

Baseline characteristics of included study

Study, Year [Ref]	Characteristics	Trial	Type of trail	Phase of study	PD-L1 positive cases (n)	Mean age	Number of ITT participants		Therapeutic regimen		outcomes
							Treatment	Control	Treatment	Control
PD-L1 inhibitor
Schmid 2020 [10]	IMpassion130	NCT02425891	Prospective RCT	III	369	55	451	451	Atezolizumab + nab-paclitaxel	Placebo + nab-paclitaxel	PFS, OS OR, CR, PR, DOR, TTD, AEs
Iwata 2019 [11]	IMpassion130(Japanese)	NCT02425891	Prospective RCT	III	25	57	34	31	Atezolizumab + nab-paclitaxel	Placebo + nab-paclitaxel	PFS, OR, CR, PR, DOR, TTD, Aes
Røssevold 2022 [12]	ALICE	NCT03164993	Prospective RCT	II	27	56	40	28	Atezolizumab + Chemotherapy	Placebo + Chemotherapy	PFS,OS,ORR,AEs
Miles 2021 [13]	IMpassion131	NCT03125902	Prospective RCT	III	292	54	431	220	Atezolizumab + paclitaxel	Placebo + paclitaxel	PFS OS, OR, CR, PR, DOR, TTD, AEs
Brufsky 2021 [14]	COLET	NCT02322814	Prospective RCT	II	22	51	32	47	Atezolizumab + cobimetinib + paclitaxel	Cobimetinib + paclitaxel	PFS, OR, PR, CR OS, DOR, EOT, AEs
R. Dent 2024 [15]	IMpassion132	NCT03371017	Prospective RCT	III	354	49	192	188	Atezolizumab + Carboplatin + Gemcitabine/Capecitabine	Placebo + Carboplatin + Gemcitabine/Capecitabine	OS,PFS,ORR, DOR, DCR,AEs
Bachelot 2021 [16]	SAFIR02BREAST IMMUNO	NCT02299999	Prospective RCT	II	32	56	47	35	Durvalumab	Maintenance chemotherapy	OS, ORR, AEs
PD-1 inhibitor
Jiang 2024 [17]	TORCHLIGHT	NCT03777579	Prospective RCT	III	300	54	353	178	Toripalimab + nab-paclitaxel	Placebo + nab-paclitaxel	OS PFS, ORR, DOR,DCR, AEs
Winer 2021 [18]	KEYNOTE-119	NCT02555657	Prospective RCT	III	405	52	312	310	Pembrolizumab	Chemotherapy (capecitabine, eribulin, gemcitabine or vinorelbine)	OS PFS, ORR, DOR, DCR, AEs
Cortes 2022 [19]	KEYNOTE-355	NCT02819518	Prospective RCT	III	636	53	566	281	Pembrolizumab + chemotherapy (nab-paclitaxel, paclitaxel or gemcitabine + carboplatin)	Placebo + chemotherapy (nab-paclitaxel, paclitaxel or gemcitabine + carboplatin)	PFS, OS,ORR, DOR, DCR, AEs
Hattori 2023 [20]	KEYNOTE-355(Japanese)	NCT02819518	Prospective RCT	III	66	53	61	26	Pembrolizumab + chemotherapy (nab-paclitaxel, paclitaxel or gemcitabine + carboplatin)	Placebo + chemotherapy (nab-paclitaxel, paclitaxel or gemcitabine + carboplatin)	PFS, OS, ORR, DOR, DCR,AEs

The results of Cochrane risk bias assessment are shown in Figs. 2 and 3. The 11 studies included are of high quality and mostly have a low risk of bias. The COLET study [14] exhibits attrition bias, while the IMpassion131 study [13] exhibits both attrition bias and reporting bias. The ALICE study [12] has attrition bias due to loss of follow-up, and the KEYNOTE-119 study [18] has performance bias.

Fig. 2.

Risk of bias bar graph

Fig. 3.

Risk of bias summary

Efficacy of PD-L1/PD-I for TNBC

PD-L1 inhibitor

Six studies reported OS in the ITT population, and four of them reported OS in the PD-L1-positive population. In both the ITT population and the PD-L1-positive population, the OS of the experimental group was better than that of the control group, and the difference between the two was statistically significant. (ITT: HR: = 0.90 [0.81, 0.99], P = 0.04, I² = 48%; PD-L1 + : HR = 0.82 [0.70, 0.95], P = 0.01, I² = 64%, Fig. 4). Six studies reported PFS in the ITT population, and five of them reported PFS in the PD-L1-positive population. In both the ITT and PD-L1-positive populations, the PFS of the experimental group was better than that of the control group, and the difference was statistically significant.(ITT: HR: = 0.85 [0.77, 0.93], P = 0.0006, I² = 46%; PD-L1 + : HR = 0.72 [0.62, 0.83], P < 0.00001, I² = 70%, Fig. 5). Six studies reported ORR in the ITT population, and four of those reported ORR in the PD-L1-positive population. There was no statistically significant difference between the two groups in the ITT population. In the PD-L1-positive population, the ORR of the experimental group was better than that of the control group. (ITT:HR: = 0.95 [0.86, 1.05], P = 0.30, I² = 62%;PD-L1 + : HR = 1.27 [1.10, 1.46], P = 0.001, I² = 7%, Fig. 6).

Fig. 4.

Forest plot and meta-analysis results for OS(PD-L1)

Fig. 5.

Forest plot and meta-analysis results for PFS(PD-L1)

Fig. 6.

Forest plot and meta-analysis results for ORR(PD-L1)

PD-1 inhibitor

Four studies reported on the overall survival (OS) of the ITT population, including studies with varying CPS thresholds: four studies on the population with CPS ≥ 1, four studies on the population with CPS ≥ 10, and two studies on the population with CPS ≥ 20. In these subgroup analyses, OS in the experimental group was superior to that in the control group, and the difference was statistically significant. (ITT: HR = 0.87 [0.78, 0.96], P = 0.007, I² = 71%; CPS ≥ 1: HR = 0.81 [0.71, 0.92], P = 0.001, I² = 39%; CPS ≥ 10: HR = 0.70 [0.58, 0.84], P = 0.0002, I² = 0%; CPS ≥ 20: HR = 0.66 [0.50, 0.86], P = 0.002, I² = 0%, Fig. 7)Four studies reported PFS in the ITT population, with four reporting PFS in the CPS ≥ 1 population and four reporting PFS in the CPS ≥ 10 population. The heterogeneity test showed high heterogeneity. After excluding the studies one by one, KEYNOTE-119 was identified as the source of heterogeneity. After removing this study, no heterogeneity was found. In the ITT population, as well as in the populations with CPS ≥ 1 and CPS ≥ 10, the PFS of the experimental group was superior to that of the control group. (ITT: HR = 0.79 [0.70, 0.90], P = 0.0004, I² = 0%; CPS ≥ 1: HR = 0.71 [0.61, 0.83], P < 0.0001, I² = 0%; CPS ≥ 10: HR = 0.67 [0.53, 0.84], P = 0.0008, I² = 0%, Fig. 8). Four studies reported the ORR values for the ITT population, with no statistically significant difference observed between the experimental group and the control group. Similarly, no statistically significant difference was found between the experimental and control groups in the PD-L1-positive population. ( ITT:HR: = 1.03 [0.93, 1.14], P = 0.63, I² = 0%; PD-L1 + :HR = 1.05 [0.92, 1.19],P = 0.49, I² = 3%, Fig. 9).

Fig. 7.

Forest plot and meta-analysis results for OS(PD-1)

Fig. 8.

Forest plot and meta-analysis results for PFS(PD-1)

Fig. 9.

Forest plot and meta-analysis results for ORR(PD-1)

PD-L1/ PD-1 inhibitor

In the ITT population, treatments were analyzed separately according to the experimental group's treatment plan: ICIs combined with chemotherapy and ICIs monotherapy. The combination of ICIs with chemotherapy improved OS compared to chemotherapy alone. (HR = 0.83 [0.72, 0.96], P = 0.01, I² = 57%) However, there was no statistically significant difference between the experimental group and the control group in patients receiving ICIs monotherapy. (HR = 0.78 [0.44, 1.36], P = 0.37, I² = 72%). Two studies have available OS data for the PD-L1-positive population, and eight studies have OS data for the PD-L1-positive population. The result analysis showed that immunotherapy did not improve the overall survival of the PD-L1-negative population. (HR = 0.96[0.81, 1.14]; P = 0.67, I² = 0%). (Supplementary 3 FigureS1 and FigureS2).

Safety of PD-L1/PD-I for TNBC

PD-L1 inhibitor

Six studies reported SAEs, with the incidence of SAEs in the experimental group being higher than that in the control group, and the difference was statistically significant. (HR = 1.36 [1.17, 1.59], P = 0.0001, I² = 0%, Fig. 10). Five studies reported the incidence of immune-related adverse events (irAEs), with the experimental group having a higher incidence of irAEs than the control group, and this difference was also statistically significant. (HR = 1.32 [1.22,1.44], P < 0.00001 I² = 44%, Fig. 11).

Fig. 10.

Forest plot and meta-analysis results for SAE(PD-L1)

Fig. 11.

Forest plot and meta-analysis results for irAE (PD-L1)

PD-1 inhibitor

Four studies reported SAEs in the ITT population, with no statistically significant difference between the two groups. (ITT:HR: = 1.07 [0.98, 1.18], P = 0.14, I² = 50%, Fig. 12). The four included studies also reported on irAEs in the ITT population, and the incidence of irAEs in the experimental group was significantly higher than that in the control group (ITT: HR: = 2.69 [2.14, 3.38], P < 0.00001, I² = 82%, Fig. 13). We compared the types of adverse reactions common in irAEs and found that immunotherapy patients had higher rates of hypothyroidism, hyperthyroidism, pneumonia, rash, and hepatitis than in chemotherapy groups, as shown in Table 2. A summary of the results is shown in Table 3.

Fig. 12.

Forest plot and meta-analysis results for SAE (PD-1)

Fig. 13.

Forest plot and meta-analysis results for irAE (PD-1)

Table 2.

Incidence of various irAEs comparing combined PD-LI/PD-1 inhibitors + chemotherapy and chemotherapy alone

		No. of studies	No. of experimental group	No. of control group	RR	95% CI	P	Heterogeneity(I2)
PD-L1	Hypothyroidism	5	1250	1006	1.54	1.30–1.83	< 0.0001	89%
	Hyperthyroidism	5	1250	1010	1.02	0.56–1.85	0.0009	74%
	Pneumonitis	5	1250	1006	3.56	1.78, 7.12	0.003	0%
	Rash	5	1250	1006	1.23	1.08, 1.41	0.002	0%
	Hepatitis(diagnosis)	4	1210	978	1.72	0.87, 3.39	0.12	0%
PD-1	Hypothyroidism	4	1285	774	1.06	0.83–1.35	0.63	96%
	Hyperthyroidism	4	1285	774	6.15	2.82–13.42	0.0001	4%
	Pneumonitis	4	1285	774	9.38	2.28, 38.61	0.002	0%
	Severe skin reactions	4	1285	774	1.74	0.97–3.14	0.06	0%

Table 3.

Primary and secondary study outcomes

	Outcomes	RR	95%CI	I2	P
PD-L1	OS (ITT)	0.9	0.81–0.99	48%	0.04
	OS (PD-L1 +)	0.82	0.70–0.95	64%	0.01
	PFS (ITT)	0.85	0.77–0.93	46%	0.0006
	PFS(PD-L1 +)	0.72	0.62–0.83	70%	< 0.00001
	ORR(ITT)	0.95	0.86–1.05	62%	0.3
	ORR(PD-L1 +)	1.27	1.10–1.46	7%	0.001
	SAEs (ITT)	1.36	1.17–1.59	0%	0.0001
	irAE (ITT)	1.32	1.22–1.44	44%	< 0.00001
PD-1	OS(ITT)	0.87	0.78–0.96	71%	0.007
	OS (CPS ≥ 1)	0.81	0.71–0.92	39%	0.001
	OS (CPS ≥ 10)	0.7	0.58–0.84	0%	0.0002
	OS (CPS ≥ 20)	0.66	0.50–0.86	0%	0.002
	PFS(ITT)	1	0.90–1.11	92%	0.95
	PFS(CPS ≥ 1)	0.88	0.78–1.00	86%	0.05
	PFS(CPS ≥ 10)	0.8	0.66–0.97	58%	0.02
	ORR(ITT)	1.03	0.93–1.14	0%	0.63
	ORR(PD-L1 +)	1.05	0.92–1.19	3%	0.49
	SAEs (ITT)	1.07	0.98–1.18	50%	0.14
	irAE (ITT)	2.69	2.14–3.38	82%	< 0.00001
PD-L1/PD-1	OS(PD-L1-)	0.96	0.81–1.14	0%	0.67
	OS(PD-L1 +)	0.81	0.74–0.90	47%	0.0001
PD-L1/PD-1	ICIs + chemotherapy ITT OS	0.83	0.72–0.96	57%	0.01
	ICIs ITT OS	0.78	0.44–1.36	72%	0.37

Sensitivity analysis

A sensitivity analysis was conducted on the main outcome measures using a one-by-one exclusion method. Among the findings, the PFS of patients using PD-1 inhibitors changed from not being statistically significant to being statistically significant in the ITT population subgroup after excluding the study by Winer 2021 [18]. Therefore, the results of the Winer 2021 [18] study need to be interpreted with caution. The results are shown in Table 4.

Table 4.

Sensitivity analysis of PFS(PD-1 inhibitor)A.ITT;B.CPS ≥ 1;C..CPS ≥ 10

	Summary▪HR▪[95%CI]	P	I2
Omitting▪Cortes 2022 [19]	1.17 [1.01, 1.34]	0.03	93%
Omitting▪Hattori 2023 [20]	1.02 [0.91, 1.13]	0.75	94%
Omitting▪Jiang 2024 [17]	1.06 [0.94, 1.19]	0.36	94%
Omitting▪Winer 2021 [18]	0.79 [0.70, 0.90]	0.0004	0%
	Summary▪HR▪[95%CI]	P	I2
Omitting▪Cortes 2022 [19]	1.01 [0.85, 1.20]	0.91	88%
Omitting▪Hattori 2023 [20]	0.90 [0.79, 1.03]	0.12	90%
Omitting▪Jiang 2024 [17]	0.93 [0.81, 1.07]	0.33	89%
Omitting▪Winer 2021 [18]	0.71 [0.61, 0.83]	< 0.0001	0%
	Summary▪HR▪[95%CI]	P	I2
Omitting▪Cortes 2022 [19]	0.95 [0.73, 1.24]	0.73	45%
Omitting▪Hattori 2023 [20]	0.81 [0.67, 0.99]	0.04	68%
Omitting▪Jiang 2024 [17]	0.74 [0.44, 1.23]	0.05	72%
Omitting▪Winer 2021 [18]	0.67 [0.53, 0.84]	0.0008	0%

Discusion

The latest research found that the expression levels and activity of PD-L1 and tumor-infiltrating lymphocytes (TILs) in TNBC patients were significantly higher than those in other breast cancer subtypes, indicating that TNBC has strong immunogenicity and may be responsive to anti-PD-1 and anti-PD-L1 treatments [21, 22]. Immunotherapy provides a new treatment option for metastatic triple negative breast cancer. In recent years, research results on PD-L1 and PD-1 inhibitors in triple negative breast cancer have been mixed, making it impossible to conclude definitively that immunotherapy can improve the efficacy of TNBC.

Efficacy of PD-L1/PD-I for TNBC

In our study, we found that in the treatment of patients with TNBC, in the ITT population, patients treated with PD-L1 inhibitors or PD-1 inhibitors had better overall survival compared to those receiving chemotherapy alone. In the PD-L1-positive population, both PD-L1 inhibitors and PD-1 inhibitors improved OS compared with chemotherapy. However, immunotherapy failed to improve OS compared to chemotherapy in the PD-L1-negative population. In the ITT population, PD-L1 inhibitors improved PFS compared with chemotherapy alone. Compared with patients who only received chemotherapy, those who received PD-1 inhibitor treatment also showed improvement in PFS. In the PD-L1-positive population, PFS was superior to chemotherapy alone in patients treated with either a PD-L1 inhibitor or a PD-1 inhibitor. In the ITT population, there was no improvement in ORR for patients treated with either type of immunosuppressant. However, PD-L1 inhibitors improved OS in the PD-L1-positive population, whereas patients treated with PD-1 inhibitors showed no improvement in ORR compared with those treated with chemotherapy alone.

In summary, both PD-L1 inhibitors and PD-1 inhibitors improved OS and PFS in both the ITT and PD-L1-positive populations. The reasons for discussing the outcome indicators of the two immunosuppressants separately in this study are as follows: 1. The number of included studies for both types of immunosuppressants is limited, and the results have certain limitations; 2. The specific medication regimens are different, with variations in the type, dose, frequency, and cycle of immunosuppressants and chemotherapy drugs. For example, the SAFIR02BREAST IMMUNO and KEYNOTE-119 trial groups used immuno monotherapies; the chemotherapy drug for IMpassion130 [10] is nab-paclitaxel, while for IMpassion131 [13] it is paclitaxel, leading to heterogeneity in outcomes.

PD-L1 inhibitors divided the subgroups into PD-L1-positive and ITT populations, and the PD-L1-positive population significantly improved OS and PFS in TNBC patients. The subgroup of PD-1 inhibitors divides the population with PD-L1-positive into subgroups such as CPS ≥ 1, CPS ≥ 10, CPS ≥ 20, etc. It can be found that the higher the CPS score, the more it can prove that PD-1 inhibitors improve OS and PFS compared to chemotherapy alone. The reason why immunotherapy based on PD-1/PD-L1 inhibitors can improve patients' overall survival and progression-free survival may be attributed to the higher levels of PD-L1 expression in TNBC on both tumor and immune cells [23, 24], which provides a direct target for ICIs and is also associated with the response to PD-1 therapy in other tumors [25]. Secondly, TNBC has more non-synonymous mutations [26], which generate tumor-specific neoantigens that activate neoantigen-specific T cells to produce anti-tumor immune responses, and ICIs can enhance this immune response [27]. The relationship between CPS score and the improvement of OS and PFS by PD-1 inhibitors may be related to the aforementioned mechanisms.

Safety of PD-L1/PD-I for TNBC

This study found that the incidence of SAEs and irAEs in patients treated with PD-L1 inhibitor immunotherapy was higher than that in the monotherapy group. The incidence of SAEs in patients treated with PD-1 inhibitors was not significantly increased compared to the chemotherapy group, while the incidence of irAEs was higher than that in the chemotherapy group. This article compared the common types of adverse reactions in irAEs and found that the incidence of hypothyroidism, hyperthyroidism, pneumonia, rash, and hepatitis in immunotherapy patients was higher than that in the monotherapy group. The irAEs caused by PD-1 and/or PD-L1 inhibitors can manifest in various organ systems throughout the body. The pathological molecular mechanisms underlying the occurrence of irAEs are not yet clear and may be related to the immune mechanisms mediated by PD-1 and/or PD-L1 inhibitors, such as increased activity of T cells carrying antigens in both tumor and normal tissues, elevated levels of existing autoantibodies, increased levels of inflammatory cytokines, and enhanced complement-mediated inflammatory responses [28]. Therefore, when using immunosuppressants to treat locally advanced or metastatic TNBC in clinical practice, individualized risk and benefit assessments should be conducted to select a reasonable treatment pathway.

In recent years, several studies have conducted meta-analyses on PD-L1/PD-1 inhibitors. For instance, Yu et al.'s research focused on metastatic triple-negative breast cancer patients and encompassed six articles. Distinct from the current work, their study analyzed immunotherapy alone and in combination with chemotherapy as separate entities. This article, on the other hand, examines PD-1 inhibitors and PD-L1 inhibitors as distinct drug classes. Although both are immunosuppressive, their mechanisms of action differ. PD-1 inhibitors attach to PD-1 on T cells, preventing the interaction with PD-L1/PD-L2, thereby alleviating the inhibition of T cell activation and proliferation, and reactivating tumor-specific T cells to restore their cytotoxic capabilities. In contrast, PD-L1 inhibitors, which bind to PD-L1 on both tumor cells and antigen-presenting cells, exhibit a more potent overall immune response. They enhance T cell-mediated anti-tumor immunity without disrupting the physiological role of PD-L2 [29]. Separating the analysis and discussion of these two types of drugs makes the results more convincing and reduces the heterogeneity of outcome indicators. This article incorporates two newly published studies, enriching the literature sources and increasing the research basis. IMpassion132 is the only randomized phase III trial exclusively targeting early-stage recurrent TNBC patients, whose tumors are highly invasive. In addition, IMpassion132 provides for the first time randomized data on the safety and tolerability of atezolizumab and non-paclitaxel combination therapy in TNBC. The TORCHLIGHT study has compensated for the lack of data on Chinese patients, increased the range of ethnicities included in our study, and achieved good therapeutic effects. In terms of outcomes, this article explores the different immunotherapeutic effects of PD-L1 inhibitors at different CPS values.

Among the outcome indicators related to PD-1 inhibitors, there is high heterogeneity in the PFS outcome indicators. After exclusion in a one-by-one manner, it was found that heterogeneity was significantly reduced after removing the Winer 2021 study. The I² statistic in the ITT population decreased from 92 to 0%, changing from non-statistically significant to statistically significant. After excluding the Winer 2021 study in the subgroup with CPS ≥ 1, the I² decreased from 86 to 0%, and in the subgroup with CPS ≥ 10, the I² decreased from 58 to 0%. After excluding the Winer 2021 study, the PFS outcomes of all subgroups were statistically significant. The reason for this may be that the immunotherapy groups in the other three studies used a first-line regimen that included nab-paclitaxel, whereas KEYNOTE-119 used only immune monotherapy, leading to heterogeneity. This sensitivity analysis laterally validated the superiority of immunocombination chemotherapy over immunological monotherapy.

Limitations

There are several limitations in the research process of this article: Firstly, the number of studies included in this article is limited, the sample size of the study populations is small, and there may be potential publication bias. Secondly, although PD-1 and PD-L1 inhibitors are compared separately, their specific medications also differ. For example, PD-L1 inhibitors include atezolizumab and durvalumab, while PD-1 inhibitors include pembrolizumab and toripalimab. There are differences in their specific drug effects, pharmacokinetics, and adverse reactions. Thirdly, even for the same medication, there may be differences in single dose, frequency, and duration of use. Fourthly, the chemotherapy regimens used in each study differ, which may lead to varying levels of immune efficacy. Fifthly, some studies in the immunotherapy group of the included studies used immunotherapy alone, while others combined immunotherapy with chemotherapy. Sixthly, different studies used different techniques to detect PD-L1 expression. Different techniques define positive immunohistochemical staining differently. Staining intensity, staining site, and background cells all affect the proportion of cells with positive PD-L1 expression and thus the correlation between PD-L1 positivity and treatment benefit.

Conclusion

In summary, both PD-1 and PD-L1 inhibitors can provide survival benefits to patients with unresectable TNBC, with the primary beneficiaries being PD-L1-positive patients. However, immunotherapy can lead to an increase in treatment-related adverse events. Therefore, before treatment, it is necessary to conduct a comprehensive risk assessment for the patient and to have the oncologist's input to promptly detect adverse reactions and prevent the occurrence of serious adverse reactions. At the same time, we also look forward to more high-quality RCT literature and better biological markers to identify advantageous populations, in order to further explore the efficacy of PD-1/PD-L1 treatment for locally advanced or metastatic TNBC patients.

Authors’ contributions

Research design: Yongping Pan, Feng Gao; Data collection and extraction: Zuxiu Wang,Xuhuan Li,Peimeng You,Zanglei Yang,Hanxi Xiao, Xinrong Tang; Statistical analysis: Zuxiu Wang,Xuhuan Li,Peimeng You,Zanglei Yang; Drafting manuscript: Zuxiu Wang,Yongping Pan, Feng Gao; Final approval of manuscript: Zuxiu Wang,Yongping Pan, Feng Gao.

Funding

This research work was supported by a Grant from The Fourth Affiliated Hospital of Nanchang University China, Jiangxi (Project ID: SKJP220210373).

Data availability

All data relevant to the study are included in the article or uploaded as supplementary information. The data used to support the findings of this study are included within the manuscript and supplementary files.

Declarations

Ethics approval and consent to participate

Ethical approval: The study did not involve the personal privacy and personal interests of patients, and the medical records and data included in the study were retrospective research data, which were exempted from medical ethics review after approval by the Ethics Committee of our hospital.

Consent to participants: Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.