Efficacy and safety of PD-1/PD-L1 inhibitors in triple-negative breast cancer: a systematic review and meta-analysis

Abstract

Objective

Triple-negative breast cancer (TNBC) is a subtype of breast cancer with a poor prognosis that seriously threatens women’s health. There is still a lack of effective therapeutic targets for TNBC treatment. We conducted a meta-analysis to evaluate the efficacy and safety of programmed cell death protein 1 (PD-1)/programmed death protein ligand 1 (PD-L1) inhibitors in combination with chemotherapy for TNBC patients.

Methods

We searched PubMed, EMBASE, Cochrane Library, and Web of Science for randomized controlled trials (RCTs) related to PD-1/PD-L1 inhibitors combined with chemotherapy. Literature conforming to the research content was identified according to the inclusion and exclusion criteria. The endpoints of efficacy were pathological complete response (pCR), event-free survival (EFS), progression-free survival (PFS), and overall survival (OS). Safety outcomes included adverse events (AEs) of any grade, AEs of grade ≥3, serious AEs, and the incidence of various AEs. We obtained odds ratios (OR), hazard ratio (HR), and 95% confidence interval (CI) for the included studies. Data analysis was performed using Review Manager software (version 5.3).

Results

A total of 4468 patients from eight RCTs were analyzed. PD-1/PD-L1 inhibitors in combination with chemotherapy significantly improved pCR (OR, 1.59; 95% CI, 1.28 − 1.98, p < 0.0001), EFS (HR, 0.66; 95% CI, 0.48 − 0.91, p = 0.01), and OS (HR, 0.72; 95% CI, 0.52 − 0.99, p = 0.05) in patients with TNBC compared to chemotherapy alone or placebo in combination with chemotherapy. Furthermore, we found that the pCR rate was almost identical in the PD-L1 positive group (OR, 1.65; 95% CI, 1.26 − 2.16, p = 0.0002) and the PD-L1 negative group (OR, 1.56; 95% CI, 1.04 − 2.33, p = 0.03). Among patients with advanced-stage TNBC, PFS (HR, 0.82; 95% CI, 0.74 − 0.90, p < 0.0001) was longer in the combination therapy group than in the chemotherapy group. There were no statistically significant differences between the experimental and control groups in OS (HR, 1.03; 95% CI, 0.74 − 1.42, p = 0.87). In terms of safety, we found that the combination therapy group had a significantly higher incidence of hyperthyroidism in patients with early and advanced TNBC (OR, 5.76; 95% CI, 2.38 − 13.95, p = 0.0001) (OR, 7.86; 95% CI, 2.65 − 23.29, p = 0.0002).

Conclusions

The combination of PD-1/PD-L1 inhibitors and chemotherapy could improve the survival and prognosis of patients with early and advanced TNBC. Combination treatment may be harmful to the thyroid; therefore, active surveillance and regular follow-up are necessary during treatment.

Keywords:

• Triple-negative breast cancer
• PD-1/PD-L1 inhibitors
• immunotherapy
• chemotherapy
• meta-analysis

Background

Triple-negative breast cancer (TNBC) accounts for approximately 15 − 20% of all breast cancers and has a poor prognosis. Since the estrogen receptor, progesterone receptor, and human epidermal growth factor receptor-2 (HER-2) are all negative, there is a lack of effective therapeutic targets; therefore, endocrine therapy and targeted therapy are not applicable. TNBC is a malignant tumor with a high rate of recurrence and metastasis. Chemotherapy is still the fundamental therapy for patients with TNBC, while the median survival of patients with TNBC who are not sensitive to chemotherapy is only 10 − 12 months after recurrence or metastasis [1,2]. Therefore, there is an urgent need to identify novel therapeutic targets. The development of immunotherapy in recent years has increased the number of patients with TNBC. Tumor immunotherapy is the process of restoring, rebuilding, and enhancing the immune system so that it can recognize abnormal tumor cells and produce effector cells to kill tumor cells [3,4]. Immune checkpoint inhibitors (ICIs), such as anti-programmed cell death protein 1 (PD-1) and anti-programmed death protein ligand 1 (PD-L1), are the current popular treatment measures for TNBC. PD-L1 binds to the PD-1 protein. Binding of the PD-1 protein to T cells and the PD-L1 ligand to tumor cells inhibits the normal immune activity of T cells, leading to an immune system that is incapable of functioning. Through the combination of anti-PD-1/PD-L1 antibody with PD-1/PD-L1, the signal pathway of PD-1/PD-L1 is blocked, thus reducing the immune escape of tumor cells. Consequently, T cells can play their normal immune role and achieve the effect of killing cancer cells [5,6].

Although PD-1/PD-L1 inhibitors have been shown to achieve better efficacy in nonsmall cell lung cancer, malignant melanoma, and uroepithelial cancer, the efficacy and safety of ICIs in patients with TNBC are controversial [7–9]. Only 10 − 30% of breast cancer patients have been reported to benefit from PD-1/PD-L1 inhibitors, and some patients may develop drug resistance and uncontrollable adverse events (AEs) as a result of the complexity of immune mechanisms [4,10]. Several clinical trials that focus on immunotherapy in combination with chemotherapy have achieved promising outcomes in TNBC; however, the results of these clinical trials have been inconsistent. IMpassion130 [11] and Keynote-355 [12] suggested that the combination of immunotherapy with chemotherapy could increase efficacy in patients with advanced TNBC compared with chemotherapy alone. However, IMpassion131 [13] showed that atezolizumab combined with chemotherapy did not improve survival in patients with advanced TNBC. Therefore, it was essential to perform a meta-analysis of these studies. Previous meta-analyses have focused on TNBC in early stages [14–16]. In this meta-analysis, we included all available randomized controlled trials (RCTs) to investigate early- and advanced-stage TNBC. This meta-analysis aimed to evaluate the efficacy and safety of PD-1/PD-L1 inhibitors combined with chemotherapy and to provide effective clinical diagnosis and treatment strategies.

Materials and methods

Data sources and search strategy

This systematic review and meta-analysis was reported following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Statement and registered with INPLASY (registration number: 202250128) [17]. We searched PubMed, EMBASE, Cochrane Library, and Web of Science in November 2021. The search strategy combines subject words and free words; it included ‘triple-negative breast neoplasms’ and free words, ‘immune checkpoint inhibitors,’ and free words. Furthermore, specific drug names of PD-1/PD-L1 inhibitors were searched simultaneously, such as ‘pembrolizumab,’ ‘atezolizumab,’ ‘durvalumab,’ ‘avelumab,’ and ‘nivolumab.’ According to McMaster University's search formula, the retrieval strategy of study type was ‘randomized controlled trial’ OR ‘randomized’ OR ‘placebo.’ The search was carried out with an ‘or’ between the subject words and free words; ‘and’ was used to connect between the subject words. Moreover, we screened references from reviews and meta-analyses related to ICIs to ensure the inclusion of all currently available RCTs. The literature search process was performed independently by two investigators. If there was disagreement, a third researcher decided until the results were unified.

Inclusion and exclusion criteria

The following inclusion criteria were used in this study: (1) patients with confirmed TNBC based on pathology; (2) the type of study was RCT; (3) the experimental group consisted of PD-1/PD-L1 inhibitors in combination with chemotherapy, and chemotherapy combined with or without placebo was administered to the control group; (4) the endpoints were pathological complete response (pCR), event-free survival (EFS), progression-free survival (PFS), or overall survival (OS); and 5) AEs associated with ICIs plus chemotherapy could be extracted.

Exclusion criteria were as follows: (1) non-RCTs; (2) patients had other subtypes of breast cancer; (3) studies included systematic reviews or animal studies; (4) available data and relevant outcomes could not be extracted; (5) studies were incomplete or were ongoing trials; and (6) full text was not available.

Data extraction

First, general information about the studies was extracted from the included RCTs, including experimental name, trial phase, first author, year of publication, number of experimental and control groups, tumor stage, study arm, control arm, dose and duration of PD-1/PD-L1 inhibitor, and positive rate of PD-L1 (PD-L1 positive was defined as PD-L1-expressing tumor-infiltrating immune cells covering 1% or more of the tumor area). Second, we extracted outcome indicators such as pCR, EFS, PFS, OS, hazard ratio (HR), 95% confidence interval (CI), and occurrence of AEs. Data extraction was performed independently by two investigators. A third researcher resolved the disagreements if necessary.

Literature quality evaluation

We perform a quality evaluation of the RCTs using the Cochrane Collaboration tool [18]. The process was completed using Review Manager, version 5.3. The criteria used to assess the quality of the literature were as follows: (1) random sequence; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcomes; (5) incomplete outcome data; (6) selective reporting, and (7) other biases: ‘low risk,’ ‘high risk,’ and ‘unclear’ were made for each item. Two researchers independently assessed the quality of the literature and resolved differences through discussion until the results were unified.

Statistical analysis

Data analysis was performed using Review Manager software (version 5.3). Odds ratios (OR) and 95% CI were used to compare pCR and AEs between the two groups. We used HR and 95% CI to assess EFS, PFS, and OS. A P-value <0.05 was considered statistically significant. We judged heterogeneity according to the Cochran’s Q and I² values. Cochran Q p < 0.1 or I² ≥ 50% indicated heterogeneity among the included studies. We further performed subgroup or sensitivity analysis; otherwise, a random-effects model was used. Furthermore, the fixed-effects model was adopted when Cochran Q p > 0.1 and I² < 50% [19,20].

Results

Literature search results

We retrieved 54, 283, 400, and 171 articles from PubMed, EMBASE, Cochrane Library, and Web of Science, respectively. A total of 908 studies were obtained. After eliminating the duplicate 364 articles in the literature, we further screened the remaining 544 studies. A total of 82 studies did not meet these requirements (animal studies, reviews, meta-analyses, etc.). Furthermore, another 326 articles were removed according to their irrelevant titles or abstracts, and 136 studies were downloaded in full text to determine whether they could be selected. Finally, we included eight RCTs after a comprehensive analysis and evaluation of the full text (eFigure 1 in the supplementary files).

Study characteristics and quality assessment

The characteristics of the eight studies [11–13,21–25] that were included in this study were shown in Table 1. A total of 4468 patients with TNBC were included, five RCTs used neoadjuvant chemotherapy (NACT), and three RCTs used advanced-stage treatment. Four studies used atezolizumab (PD-L1 inhibitor) and one RCT used durvalumab (PD-L1 inhibitor). The remaining three studies used pembrolizumab (PD-1 inhibitor). Two RCTs were open-label, and the remaining six RCTs were blinded. The pooled efficacy and safety endpoints were listed in Table 2. All studies were of high quality (eFigure 2 in the supplementary files).

Table 1. Characteristics of included studies.

Trial	Author	Year	Phase	Treatment stage	Experimental	Control	TNM	Treatment arm	Control arm	The dose of anti- PD-1/PD-L1	Duration of anti- PD-1/PD-L1	PD-L1 positive
IMpassion131 [13]	Miles	2021	III	Advanced	431	220	Locally advanced/mTNBC	Atezo + Pac	Placebo + Pac	840mg d1,15 q28d	Until disease progression	45%
KEYNOTE-355 [12]	Javier	2020	III	Advanced	566	281	Locally recurrent inoperable or mTNBC	Pem + NabPac/Pac/ GEM + Cb	Placebo + taxane or GEM-Cb	200mg q3w	Until disease progression	75%
NeoTRIPaPDL1 [21]	Gianni	2019	III	NACT	138	142	T1cN1 T2N1 T3N0 + T3N1,T4 anyN,anyT	Atezo + NabPac + Cb	NabPac + Cb	1200mg d1 q3w	8 cycles	56%
IMpassion031 [22]	Mittendorf	2020	III	NACT	165	168	cT2-cT4/cN0-cN3	Atezo + NabPac + AC	Placebo + NabPac+ AC	840mg q2w + 840mg q2w + 1200mg q3w	12w + 8w + 11cycles	46%
I-SPY2 [23]	Nanda	2020	II	NACT	28	79	II/III	Pem + Pac + AC	Pac + AC	200mg q3w	4 cycles	NA
KEYNOTE-522 [24]	Schmid	2020	III	NACT	784	390	T1cN1-2/T2-4 N0-2	Pem + Pac + Cb + AC/EC	Placebo + Pac + Cb+ AC/EC	200mg q3w	4 cycles + 4 cycles	83%
IMpassion130 [11]	Schmid	2020	III	Advanced	451	451	Locally advanced/mTNBC	Atezo + NabPac	Placebo + NabPac	840mg d1,15 q28d	Until disease progression	41%
GeparNuevo [25]	Loibl	2019	II	NACT	88	86	cT2-cT4a-d	Durva + NabPac + EC	Placebo + NabPac+ EC	1.5g q4w + 1.5g q4w	12w + 4cycles	87%

mTNBC: metastatic triple-negative breast cancer; NACT: neoadjuvant chemotherapy; Atezo: atezolizumab; Pem: pembrolizumab; Durva: durvalumab; NabPac: nab-paclitaxel; Pac: paclitaxel; Cb: carboplatin; GEM: gemcitabine; NVB: vinorelbine; CAPE: Capecitabine; EC: epirubicin + cyclophosphamide; AC: adriamycin + cyclophosphamide; NA: not available; w: week.

Table 2. Pooled the endpoints of efficacy and safety.

Trial	Efficacy					Safety
	Endpoints	HR (95%CI)				AEs	Experimental		Control
							Events	Total	Events	Total
IMpassion131 [13]	PFS	0.86(0.7–1.05)				any grade AEs	268	432	116	217
						grade ≥ 3 AEs	49	432	11	217
						hypothyroidism	60	432	12	217
						hyperthyroidism	25	432	0	217
						hepatitis	7	432	2	217
	OS	1.2(0.88–1.43)				infusion reaction	15	432	7	217
						pneumonitis	16	432	3	217
						colitis	3	432	2	217
						Rash	141	432	66	217
						adrenal insufficiency	3	432	0	217
KEYNOTE-355 [12]	PFS	0.82(0.69–0.97)				any grade AEs	554	562	276	281
						grade ≥ 3 AEs	438	562	207	281
						hypothyroidism	87	562	9	281
						hyperthyroidism	27	562	3	281
						anemia	275	562	129	281
						neutropenia	231	562	107	281
						fatigue	160	562	83	281
						colitis	10	562	4	281
						pneumonitis	14	562	0	281
						elevated transaminase	115	562	46	281
						skin reactions	10	562	1	281
NeoTRIPaPDL1 [21]	pCR	Experimental		Control		grade ≥ 3 AEs	107	138	98	140
		Events	Total	Events	Total
		60	138	58	142	serious AEs	25	138	8	140
IMpassion031 [22]	EFS	0.76(0.4–1.44)				grade ≥ 3 AEs	103	164	101	167
						serious AEs	50	164	30	167
						hypothyroidism	11	164	2	167
						hyperthyroidism	5	164	0	167
						anemia	63	164	64	167
						neutropenia	59	164	59	167
						vomiting	58	164	51	167
	OS	0.69(0.25–1.87)				infusion reaction	17	164	11	167
						hepatitis	2	164	1	167
						fatigue	59	164	64	167
						cough	35	164	32	167
						Rash	46	164	42	167
						diarrhea	68	164	74	167
						sensory neuropathy	54	164	45	167
	pCR	Experimental		Control		stomatitis	39	164	29	167
		Events	Total	Events	Total	pneumonitis	2	164	2	167
		95	165	69	168	colitis	1	164	1	167
						adrenal insufficiency	0	164	1	167
						pyrexia	34	164	21	167
						elevated transaminase	33	164	26	167
						headache	41	164	35	167
I-SPY2 [23]	EFS	0.65(0.23-1.85)				–
	pCR	Experimental		Control
		Events	Total	Events	Total
		17	28	17	79
KEYNOTE-522 [24]	EFS	0.63(0.43–0.93)				any grade AEs	777	781	389	389
						grade ≥ 3 AEs	633	781	295	389
						hypothyroidism	107	781	13	389
						hyperthyroidism	36	781	4	389
						anemia	430	781	215	389
	OS	0.72(0.51–1.02)				neutropenia	365	781	183	389
						vomiting	199	781	85	389
						infusion reaction	132	781	43	389
						fatigue	321	781	147	389
						Rash	170	781	59	389
	pCR	Experimental		Control		adrenal insufficiency	18	781	0	389
						skin reactions	34	781	4	389
		Events	Total	Events	Total	sensory neuropathy	154	781	82	389
		260	401	103	201	diarrhea	230	781	92	389
						elevated transaminase	199	781	96	389
IMpassion130 [11]	PFS	0.8(0.69–0.92)				any grade AEs	450	453	428	437
						grade ≥ 3 AEs	230	453	190	437
						hypothyroidism	63	453	16	437
						serious AEs	105	453	81	437
						anemia	126	453	118	437
						neutropenia	97	453	68	437
						vomiting	89	453	75	437
						fatigue	212	453	195	437
	OS	0.86(0.72–1.02)				sensory neuropathy	72	453	52	437
						Rash	78	453	73	437
						pneumonitis	23	453	7	437
						headache	108	453	95	437
						pyrexia	88	453	48	437
						cough	114	453	85	437
						diarrhea	147	453	152	437
						elevated transaminase	45	453	43	437
GeparNuevo [25]	pCR	Experimental		Control		serious AEs	30	92	29	82
		Events	Total	Events	Total	hypothyroidism	7	92	2	82
		47	88	38	86	hyperthyroidism	9	92	1	82
						anemia	87	92	79	82
						neutropenia	71	92	67	82
						vomiting	12	92	11	82
						infusion reaction	3	92	1	82
						fatigue	70	92	68	82
						hepatitis	1	92	0	82
						stomatitis	15	92	14	82
						skin reactions	45	92	39	82
						sensory neuropathy	76	92	69	82
						Rash	21	92	22	82
						pneumonitis	1	92	1	82
						headache	38	92	28	82
						pyrexia	16	92	12	82
						cough	25	92	15	82
						diarrhea	26	92	34	82

AEs: adverse events; pCR: pathological complete response; OS: overall survival; PFS: progress free survival; EFS: event free survival; HR: hazard ratio; CI: confidence interval.

Efficacy in early-stage TNBC

pCR

Pooled frequencies for pCR were shown in eTable 1. We clearly found that the pCR rate of the combination group was higher than that of the control group (eTable 1 in the supplementary files). Five RCTs [21–25] reported pCR rates for combination therapy compared with chemotherapy alone, and the results suggested that there was obvious heterogeneity (I² = 60%, p = 0.04) (Figure 1(A)), followed by sensitivity analysis. There was no heterogeneity (I² = 12%, p = 0.33) after removing the I-SPY2 trial [23]. We chose the fixed-effects model for the analysis. The results showed that the pCR rate was higher in the combination therapy group than in the chemotherapy group (OR, 1.59; 95% CI, 1.28 − 1.98, p < 0.0001) (Figure 1(B)). Furthermore, we performed a subgroup analysis based on the expression status of PD-L1 and the type of ICI. We found that there was no heterogeneity in the PD-L1 positive (I² = 0%, p = 0.44) and PD-L1 negative groups (I² = 0%, p = 0.50), and the fixed effect model was selected. Furthermore, we found that the PCR rate was almost identical in the PD-L1 positive group (OR, 1.65; 95% CI, 1.26 − 2.16, p = 0.0002) and the PD-L1 negative group (OR, 1.56; 95% CI, 1.04 − 2.33, p = 0.03) (Figure 1(C)). Additionally, we chose the random-effects model for statistical analysis in subgroups of different ICIs types. The results suggested that the anti-PD-1 group could significantly improve the pCR of patients with early TNBC (OR, 2.89; 95% CI, 0.93 − 8.98, p = 0.07). In the anti-PD-1 subgroup, the pCR of ICIs combined with chemotherapy was 2.89 times that of the control group. (Figure 1(D)).

Figure 1. Forest plots for outcomes in early triple-negative breast cancer. (A) OR for PCR (before sensitivity analysis), (B) OR for PCR (after sensitivity analysis), (C) Subgroup analysis for PCR according to different PD-L1 status (positive versus negative), (D) Subgroup analysis for PCR according to different types of interventions (anti PD1 versus anti PD-L1), (E) HR for EFS, (F) HR for OS.

EFS and OS

We found that a total of three RCTs [22–24] studied the EFS of NACT without statistical heterogeneity (I² = 0%, p = 0.89). The results showed that PD-1/PD-L1 inhibitors in combination with chemotherapy had better EFS than chemotherapy alone (HR, 0.66; 95% CI, 0.48 − 0.91, p = 0.01) (Figure 1(E)). OS was reported in two studies [22,24] on early-stage TNBC, and there was no heterogeneity (I² = 0%, p = 0.94). We found that the risk of death in the combination treatment group was 72% of that of the chemotherapy alone group (HR, 0.72; 95% CI, 0.52 − 0.99, p = 0.05) (Figure 1(F)).

Efficacy in advanced TNBC

PFS and OS

Three of the included studies [11–13] reported PFS rates in patients with advanced TNBC. There was no heterogeneity among the studies (I² = 0, p = 0.85). We chose the fixed-effects model for the meta-analysis. To some extent, combination therapy can reduce disease progression in advanced TNBC. The risk of death or disease progression in the PD-1/PD-L1 inhibitor plus chemotherapy group was 82% of that in the chemotherapy group (HR, 0.82; 95% CI, 0.74 − 0.90, p < 0.0001) (Figure 2(A)). We found that two studies assessed OS in patients with advanced TNBC [11,13]. The heterogeneity analysis found I² = 79%, p = 0.03, using the random-effects model for the pooled analysis. The results indicated that the combination of PD-1/PD-L1 inhibitors and chemotherapy did not improve OS in patients with advanced TNBC (HR, 1.03; 95% CI, 0.74 − 1.42, p = 0.87). The results were shown in Figure 2(B).

Figure 2. Forest plots for outcomes in advanced triple-negative breast cancer. (A) HR for PFS, (B) HR for OS.

Safety in early-stage TNBC

We analyzed the safety of PD-1/PD-L1 inhibitors in combination with chemotherapy compared with chemotherapy alone in patients with early-stage TNBC. The combination therapy group had a significantly increased incidence of hypothyroidism (OR, 4.59; 95% CI, 2.73 − 7.72, p < 0.00001), hyperthyroidism (OR, 5.76; 95% CI, 2.38 − 13.95, p = 0.0001), infusion reaction (OR, 1.66; 95% CI, 1.19 − 2.30, p = 0.003), and rash (OR, 1.33; 95% CI, 1.03 − 1.70, p = 0.03). Additionally, we found that grade ≥3 AEs in the combination therapy group were 1.31 times higher than those in the chemotherapy group (OR, 1.31; 95% CI, 1.05 − 1.64, p = 0.02) (Table 3).

Table 3. Pooled odds ratio for toxicities in early triple-negative breast cancer.

Toxicity	No. of included studies	Heterogeneity		OR (95%CI)	p Value
		I^2 (%)	p
Grade ≥ 3 AEs	3	0	0.66	1.31 (1.05–1.64)	0.02
Serious AEs	3	74	0.02	1.80 (0.86–3.77)	0.12
Hypothyroidism	3	0	0.87	4.59 (2.73–7.72)	<0.00001
Hyperthyroidism	3	0	0.76	5.76 (2.38–13.95)	0.0001
Anemia	3	0	0.86	0.99 (0.80–1.22)	0.89
Neutropenia	3	0	0.77	0.98 (0.79–1.20)	0.81
Fatigue	3	18	0.29	1.04 (0.85–1.28)	0.69
Vomiting	3	0	0.88	1.21 (0.95–1.53)	0.12
Hepatitis	2	0	0.89	2.28 (0.33–15.59)	0.40
Infusion reaction	3	0	0.91	1.66 (1.19–2.30)	0.003
Stomatitis	2	0	0.36	1.29 (0.83–2.01)	0.26
Diarrhea	3	73	0.02	0.93 (0.58–1.51)	0.77
Pneumonitis	2	0	0.94	0.97 (0.19–4.87)	0.97
Rash	3	38	0.20	1.33 (1.03–1.70)	0.03
Cough	2	0	0.41	1.31 (0.85–2.01)	0.22
Headache	2	0	0.85	1.30 (0.87–1.92)	0.20
Pyrexia	2	0	0.45	1.59 (0.98–2.56)	0.06
Sensory Neuropathy	3	0	0.41	1.01 (0.79–1.29)	0.93
Skin reactions	2	82	0.02	2.02 (0.49–8.34)	0.33
Adrenal insufficiency	2	73	0.05	2.71 (0.04–166.23)	0.64
Elevated transaminase	2	0	0.40	1.10 (0.86–1.42)	0.46

AEs: adverse events; OR: odds ratio; CI: confidence interval.

Safety in advanced TNBC

The incidence of AEs of any grade (OR, 1.48; 95% CI, 1.09 − 2.02, p = 0.01), grade ≥3 AEs (OR, 1.39; 95% CI, 1.15 − 1.70, p = 0.0009), hypothyroidism (OR, 0.05; 95% CI, 2.82 − 5.82, p = 0.00001), hyperthyroidism (OR, 7.86; 95% CI, 2.65 − 23.29, p = 0.0002), neutropenia (OR, 1.27; 95%CI 1.02 − 1.59, p = 0.04), and pneumonitis (OR, 3.77; 95% CI, 1.91 − 7.43, p = 0.0001) was higher in the combination group than in the chemotherapy group in patients with advanced TNBC (Table 4). We summarized the frequencies of toxicities and the results were shown in eTable 2 (eTable 2 in the supplementary files).

Table 4. Pooled odds ratio for toxicities in advanced triple-negative breast cancer.

Toxicity	No. of included studies	Heterogeneity		OR (95%CI)	p Value
		I^2 (%)	p
Any grade AEs	3	0	0.49	1.48 (1.09–2.02)	0.01
Grade ≥ 3 AEs	3	31	0.24	1.39 (1.15–1.70)	0.0009
Hypothyroidism	3	8	0.34	4.05 (2.82–5.82)	<0.00001
Hyperthyroidism	2	32	0.23	7.86 (2.65–23.29)	0.0002
Anemia	2	0	0.70	1.09 (0.88–1.33)	0.43
Neutropenia	2	24	0.25	1.27 (1.02–1.59)	0.04
Fatigue	2	0	0.51	1.03 (0.84–1.26)	0.77
Colitis	2	0	0.64	1.09 (0.41–2.88)	0.87
Pneumonitis	3	0	0.53	3.77 (1.91–7.43)	0.0001
Rash	2	0	0.79	1.07 (0.84–1.37)	0.58
Elevated transaminase	2	0	0.37	1.18 (0.89–1.57)	0.26

AEs: adverse events; OR: odds ratio; CI: confidence interval.

Discussion

Current systemic therapies for TNBC are dominated by chemotherapy, such as paclitaxel, anthracyclines, and platinum. Currently, clinical treatments are limited, with few effective drugs available, and there is a lack of effective therapeutic targets for patients with TNBC who are not sensitive to chemotherapy or for advanced breast cancer patients with recurrence and metastasis. In recent years, with the advent of immunotherapy, more clinical trials of PD-1/PD-L1 inhibitors have been conducted [26,27]. PD-L1 is not expressed in normal epithelial tissues but is expressed in breast cancer cells and tumor-infiltrating lymphocytes (TIL). Furthermore, the expression of PD-L1 in TNBC is positively correlated with TILs [28,29]. TNBC has the highest expression of PD-L1 (approximately 20%) compared to other molecular subtypes. Relevant studies have also reported a correlation between PD-L1 expression and the efficacy of ICIs in various solid tumors, including breast cancer [30]. Our study included all RCTs that met the criteria to investigate the efficacy and safety of PD-1/PD-L1 inhibitors plus chemotherapy compared to chemotherapy alone.

pCR and EFS in patients with early-stage TNBC

The CTNeoBC meta-analysis suggested that patients with TNBC who achieved pCR after NACT had a better prognosis than non-pCR patients using the same regimen, indicating that pCR is an important reference for chemotherapy sensitivity and prognosis in patients with TNBC [31]. Our results showed that the pCR rates of PD-1/PD-L1 inhibitors combined with chemotherapy were higher than those of chemotherapy alone or chemotherapy combined with placebo in TNBC. In our initial meta-analysis of pCR, we found that the results produced strong heterogeneity and sensitivity analysis suggested that I-SPY2 [23] was the main reason for this heterogeneity. We found that the small sample size of patients with TNBC included in this clinical trial might have impacted the results. We also performed a subgroup analysis of the pCR rates, splitting them into positive and negative groups according to the expression status of PD-L1, and the results indicated that the pCR rates did not correlate with the PD-L1 status. We found similar results in the studies by Michal et al. [32] and Xin et al. [14]. However, in a meta-analysis that included five RCTs of early TNBC that required NACT conducted by Paolo et al. [16], the subgroup analysis showed that pCR rates were significantly related to PD-L1 status and PD-L1 positive patients could benefit from immunotherapy in combination with chemotherapy (OR, 1.65; 95% CI, 1.06 − 2.57), while the effect was not significant in the PD-L1 negative group. Similarly, we also performed a subgroup analysis of different types of ICIs, and the results showed that the use of PD-L1 inhibitors could improve pCR in the combined treatment group. No statistically significant differences were found between them in the anti-PD-1 group. Furthermore, subgroup analyses have shown that patients with TNBC with positive axillary lymph nodes would have a higher pCR rate with a regimen of ICIs combined with chemotherapy [14]. From the included literature, the NeoTRIPaPDL1 trial [21] indicated that the combination of atezolizumab, carboplatin, and nab-paclitaxel did not improve pCR rates, while IMpassion031 and KEYNOTE-522 showed positive results [22,24]. NeoTRIPaPDL1 did not add anthracycline to fundamental chemotherapy regimens, which may explain the difference in efficacy. Relevant studies have indicated that EFS has a prognostic value in patients with TNBC. Our meta-analysis included three RCTs and the results emphasized that the EFS was higher in the group treated with ICIs in combination with chemotherapy. Michal et al. [32] and Xin et al. [14] obtained similar findings. Whether pCR or EFS was used as the observed outcome, these results provided a new therapeutic target for NACT in patients with early-stage TNBC and found a new direction for chemotherapy-insensitive patients.

PFS and OS in advanced TNBC patients

Based on previous meta-analyses [14–16,32], none of them have analyzed the drug efficacy of advanced TNBC. The present analysis of three RCTs on advanced TNBC revealed that the efficacy of immunotherapy in combination with chemotherapy was stronger than that of chemotherapy in terms of PFS. Furthermore, it is suggested that ICIs combined with chemotherapy are valuable in the treatment of advanced TNBC and could provide a new therapeutic target for clinicians. Furthermore, in terms of OS, we realized that IMpassion131 [13] was the main reason for the unstable outcomes. IMpassion131 was studied in unresectable locally advanced or metastatic triple-negative breast cancer (mTNBC), and the results showed that, compared to the application of paclitaxel alone, paclitaxel combined with atezolizumab did not affect patients' PFS. Moreover, the OS data showed a negative trend. In contrast, IMpassion130 [11] showed that atezolizumab plus nab-paclitaxel improved PFS and OS in PD-L1 positive patients. The significant difference in the results of these two RCTs was the class of paclitaxel application and the use of conventional paclitaxel in the clinic required corticosteroids, which could be the main reason for the efficacy of immunotherapy. Our study showed that PD-1/PD-L1 inhibitors combined with chemotherapy could effectively improve PFS in patients with advanced TNBC. In patients with early-stage TNBC, the combination group showed improved OS, but in patients with advanced-stage TNBC, there were no statistically significant differences in OS between the two groups. Therefore, more clinical trials are needed to show the effect of immunotherapy on OS.

Safety of immunotherapy

Regarding safety, our results showed that immunotherapy combined with chemotherapy had the greatest effect on thyroid function in patients with early and advanced-stage TNBC. Therefore, it is important to monitor thyroid function when administering immunotherapy in combination with chemotherapy. Patients who develop endocrine disease may be irreversible and require long-term treatment [33,34]. Furthermore, the combination group was more likely to have pneumonitis, which was similar to the results of a meta-analysis by Zhang et al. [35]. In addition, hepatitis, adrenal insufficiency, skin reactions, infusion reaction and pyrexia also deserve our attention in the combination therapy. Therefore, a comprehensive assessment of tumor conditions in cancer patients and the underlying physical fitness is necessary. Under the premise of ensuring safety, we should fully improve the ability of anticipated treatment to prevent withdrawal of patients due to serious AEs in the clinic.

Clinical implications

In this meta-analysis, the efficacy and safety of combination chemotherapy with PD-1/PD-L1 inhibitors were investigated and the results were promising. A combination of multiple ICIs and clinical trials of immunotherapy combined with radiotherapy or targeted therapy is being carried out, and the results are equally worthy of our expectations. With the advent of precision therapy, the classification of breast cancer is a future direction, and we believe that immunotherapy will provide a good prospect for cancer patients. To date, there is no standardized test for PD-L1 expression. Furthermore, the threshold for determining PD-L1 positivity has been inconsistent. It is a question we should consider which chemotherapeutic agents are used in combination with immunotherapy to achieve optimal efficacy. However, there are still many challenges in immunotherapy.

Study limitations

This study had some limitations. First, only eight studies were included, and the number of studies was limited. With the continuous development of clinical trials on ICIs plus chemotherapy, there will be more data for our research in the future. Second, due to the insufficient number of studies, publication bias was not evaluated. Despite these limitations, this meta-analysis provides objective evidence for the clinical treatment of the combination of chemotherapy with PD-1/PD-L1 inhibitors in patients with TNBC.

Conclusions

In conclusion, compared with chemotherapy alone, PD-1/PD-L1 inhibitors plus chemotherapy in NACT improved pCR rates, EFS, and OS in early-stage TNBC. Similarly, PD-1/PD-L1 inhibitors combined with chemotherapy can increase PFS in patients with advanced TNBC. Additionally, more attention should be paid to thyroid function in patients treated with PD-1/PD-L1 inhibitors.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This project was supported by the National Natural Science Foundation of China [82073282], the Natural Science Foundation of Liaoning Province [2021-BS115], and the China Postdoctoral Science Foundation [2020M681018].