Literature selection and study characteristics
First, we determined our search formula, and through this search strategy identified 1153 studies, of which 6 were conference abstracts. According to inclusion and exclusion criteria, 6 studies from the randomized controlled trials (RCTs) were eventually included in this meta-analysis. The specific search and study selection process is shown in Fig. 2. Among them, there are two EMBRACA tests, and one of them has updated the OS(25, 26). It is noteworthy that, iniparib was not included as a PARP inhibitor in search formula. There is evidence that it is not a real PARP inhibitor, and it does not rely on PARPis to function(27, 28).
These included RCTs were published in 2016-2021. A total of 1477 BRCA-mutated patients, which from phase II and III clinical trials, were eventually included in this meta-analysis. The characteristics of the included studies and enrolled patients are listed in Table 1. Three of them studied the PARPis in combination with chemotherapy, and three evaluated the efficacy of PARPis alone. BROCADE is a randomized, partially blind phase II clinical trial (NCT01506609), assessed the safety and efficacy of intermittent veliparib with carboplatin/paclitaxel (VCP) or temozolomide (VT)(29). BROCADE3 is a randomized, double-blind, placebo-controlled phase III clinical trial. That is, veliparib or placebo in combination with carboplatin and paclitaxel to evaluate the efficacy of platinum in combination with PARPis(30). Shivaani Kummar et al. conducted a randomized phase II trial of veliparib in combination with cyclophosphamide or cyclophosphamide alone(31). Three other studies respectively assessed the efficacy of veliparib or olaparib as monotherapy. Notably, we included a recently published Phase III randomized trial: the OlympiA trial(32). It was a large international randomized trial that evaluated one year of adjuvant olaparib vs placebo after chemotherapy and local treatment in germline BRCA mutation (gBRCAm) carriers with human epidermal growth factor receptor 2 (HER2)-negative breast cancer. In March 2022, the European Society of Medical Oncology (ESMO) virtual plenary was updated with a significant overall survival benefit (hazard ratio (HR) 0.68, p = 0.009)(33). The risk of bias for each study was assessed according to the Cochrane Manual 5.1.0 assessment criteria (Fig. 3). Randomization was unclear in four studies, and two studies had other unclear risk of bias. Overall, the risk of bias is low.
Table 1
Main study characteristics of the included trials.
|
Trials (first author)
|
Year
|
Phase
|
Identifier
|
Treatment arms
|
Target
|
BRCA
|
Disease phenotype
|
Patients
|
RR Control arm
|
RR Experiment arm
|
OS HR 95%CI
|
PFS HR 95%CI
|
Mark Robson
|
2017
|
Ⅲ
|
NCT02000622
|
Olaparib
vs Cape/eribulin/
VRB
|
PARP1 PARP2
|
Mut
|
HER2-
|
302
|
97
|
205
|
0.90 (0.63-1.29)
|
0.58 (0.43-0.80)
|
Jennifer K Litton
|
2018
|
Ⅲ
|
NCT01945775
|
Talazoparib vs Cape/eribulin/
GEM /VRB
|
PARP1 PARP2
|
Mut
|
HER2-
|
431
|
144
|
287
|
0.848
(0.670-1.073)
|
0.54 (0.41-0.71)
|
Véronique Diéras
|
2020
|
Ⅲ
|
NCT02163694
|
Vel +CP-PTX vs PBO+CP-PTX
|
PARP1 PARP2
|
Mut
|
HER2-
|
509
|
172
|
337
|
0.95 (0.73-1.23)
|
0.71 (0.57-0.88)
|
H.S. Han
|
2017
|
Ⅱ
|
NCT01506609
|
Vel +CP /PTX vs PBO+CP/PTX
|
PARP1 PARP2
|
Mut
|
All
|
196
|
99
|
97
|
0.750 (0.503–1.117)
|
0.789 (0.536–1.162)
|
Shivaani Kummar
|
2016
|
Ⅱ
|
NCT01306032
|
Vel +CTX
vs
CTX
|
PARP1 PARP2
|
Mixed
|
TNBC
|
39
|
18
|
21
|
NR
|
0.35
(0.14-0.85)
|
A.N.J. Tutt
|
2021
|
Ⅲ
|
NCT02032823
|
Olaparib vs
PBO
|
PARP1 PARP2
|
Mut
|
HER2-
|
1836
|
921
|
915
|
0.68
(0.51-0.91)
|
NR
|
Cape= Capecitabine; VRB= vinorelbine; GEM=gemcitabine; CP= carboplatin; PTX= paclitaxel; CTX= cyclophosphamide; PBO= placebo; Vel= Veliparib
Efficacy
Primary endpoint: PFS
Progression-free survival (PFS) is the primary endpoint of most studies but cannot be obtained in one study (OlympiA)(32). Since disease-free survival (DFS) was the primary endpoint of this study, we excluded this study and pooled the results of five other studies. Due to the moderate heterogeneity (I2 = 29%, p = 0.23), we chose the fixed-effects model. From the forest map, we found that PARPis were closely related to the improvement of PFS, with HR 0.63 ([95% CI, 0.55-0.73], p < 0.00001; Fig. 4a). Additionally, PARPis, either single-agent or in combination, significantly prolonged PFS in patients compared with control groups (HR 0.56 [95%CI, 0.46 to 0.68], p < 0.00001; HR 0.71 [95%CI, 0.59 to 0.85], p = 0.00002, respectively) and it was noteworthy that there was no significant difference in the benefit of PFS between monotherapy and combination therapy (p = 0.09; Fig. 4b). In addition, we compared the median PFS data provided in the included study (Table 2). Median survival was significantly longer in the PARPis arms than in the placebo or chemotherapy arms.
In breast cancer, two PARPis, olaparib and talazoparib, have been approved for treatment of gBRCAm carriers with metastatic HER2-negative breast cancer, respectively(25, 34). On March 11, 2022, the FDA approved olaparib for the adjunctive treatment of HER2-negative in adults with gBRCAm who are at high risk for early breast cancer and who have previously received neoadjuvant or adjuvant chemotherapy(35). Veliparib, while not approved by the FDA, has shown promising results in several clinical trials in gBRCAm carriers with metastatic HER2-negative breast cancer patients(36, 37). Therefore, we focused on the PFS benefits of HER2-negative breast cancer in the subgroup analysis. In a subgroup analysis based on hormone receptor status, we found an exciting correlation between PARPis and improved PFS in both triple negative breast cancer (TNBC) and HER2-negative hormone receptor (HR)-positive breast cancer patients (HR 0.59 [95%CI, 0.49 to 0.72], p < 0.00001; HR 0.66 [95%CI, 0.54 to 0.80], p < 0.00001,respectively; Fig. 4c) . Furthermore, there is no significant heterogeneity (I2 = 41%, p = 0.15; I2 = 48%, p = 0.14).
Breast cancer patients with BRCA1/2 mutations are more sensitive to platinum drugs such as cisplatin and carboplatin than the wild type(38, 39). The effect of prior use of platinum therapy was also analyzed. However, PARPis significantly improved PFS in patients not receiving platinum-based therapy with HR 0.64 ([95% CI, 0.55 to 0.75] p < 0.00001; Fig. 4d). For platinum-treated patients, the risk of disease progression was also statistically significant in the PARP inhibitor group (HR 0.70 [95%CI, 0.53 to 0.91]). As for BRCA mutation status, PARP inhibitors benefit uniformly in BRCA mutation 1 or 2 (Supplementary Figure1a).
Table 2
Median progression-free survival of the included trials.
|
Trials
(first author)
|
Year
|
Median progression-free survival (months)
|
Hazard ratio
(95% CI)
|
P
|
NCT
|
PARP inhibitors therapy
|
Placebo or chemotherapy
|
H.S. Han
|
2017
|
14.1
|
12.3
|
0.79(0.55-1.16)
|
0.227
|
01506609
|
Jennifer K Litton
|
2018
|
8.6
|
5.6
|
0.54(0.41-0.71)
|
< 0.001
|
01945775
|
Mark Robson
|
2017
|
7.0
|
4.2
|
0.58(0.43-0.80)
|
< 0.001
|
02000622
|
Véronique Diéras
|
2020
|
14.5
|
12.6
|
0.70(0.57-0.87)
|
0.0016
|
02163694
|
Secondary endpoint: OS and ORR
We included three single-agent studies and two combination chemotherapy studies to analyze overall survival (OS) without significant heterogeneity (I2 = 0%, p = 0.51). The pooled HR with 0.83 ([95% CI, 0.73 to 0.95], p = 0.005; Fig. 5a) indicated significant improvement in OS with PARPis. Interestingly, in the subgroup analysis, the significant improvement in OS with PARPis compared to the control groups was statistically significant only in the monotherapy subgroup (HR 0.80 [95%CI, 0.68 to 0.94], p = 0.008; Fig. 5b). There was no statistical significance in combined chemotherapy subgroup (HR 0.88 [95%CI, 0.71 to 1.10], p = 0.027).
Overall, five studies provided data for objective response rate (ORR) analysis. The pooled results showed a significant correlation between ORR and the experimental arms (RR 1.55 [95% CI, 1.02 to 2.34], p = 0.04, I2 = 90%;), with high heterogeneity (Fig. 5c). Subgroup analysis observed that monotherapy had a higher ORR rate than combination therapy (RR 2.21 [95% CI, 1.73 to 2.84] and RR 1.11 [95% CI, 0.93 to 1.33], respectively; Fig. 5d).
Safety
In addition to efficacy, we also paid attention to the possible adverse reactions of PARPis, that is, safety analysis. The comparative safety profile in terms of the (adverse events) AEs of interest is shown in Table 3 and Supplementary Figure1. Overall, Forest plot results showed there is no difference in the probability of AEs between the PARPis arms and the placebo or chemotherapy arms, regardless of whether it was AEs of any grade with RR 1.03 ([95% CI, 0.96 to 1.11], p = 0.40) or AEs grade≥3 with RR 1.09 ([95% CI, 0.74 to 1.61], p = 0.65; Fig. 6a and Fig. 6b). RE model was used due to high heterogeneity (I2 = 97%, respectively). Notably, in subgroup analyses of AEs of any grade, the incidence of AEs with monotherapy and combination therapy was similar and there was no difference between each and their control groups (RR 1.04 [95% CI, 0.96 to 1.12], p = 0.34 and RR 1.02 [95% CI, 0.92 to 1.14], p = 0.68; Fig. 6c). The results of the subgroup with AE grade≥3 were the same as above(Fig. 6d).
Table 3
RRs of grade 3 or higher AEs comparing PARPis groups with the control groups
Adverse event type
|
PARP inhibitors
|
Placebo/
Chemotherapy
|
RR (95% CI)
|
I2(%)
|
P
|
Anemia
|
385/1852
|
99/1406
|
2.92(0.86-9.84)
|
94
|
0.08
|
Neutropenia
|
452/1852
|
271/1406
|
0.97(0.64-1.47)
|
88
|
0.89
|
Leukopenia
|
169/1852
|
81/1406
|
1.21(0.96-1.53)
|
75
|
0.10
|
Lymphopenia
|
15/307
|
3/144
|
2.94(0.84-10.36)
|
0
|
0.09
|
Fatigue
|
58/1831
|
24/1388
|
1.61(1.00-2.57)
|
43
|
0.05
|
Nausea
|
29/1831
|
12/1388
|
1.38(0.74-2.57)
|
46
|
0.31
|
Vomiting
|
27/1852
|
8/1406
|
1.78(0.87-3.66)
|
21
|
0.12
|
Diarrhea
|
26/1831
|
22/1388
|
0.69(0.27-1.74)
|
49
|
0.43
|
Headache
|
14/1831
|
7/1388
|
1.09(0.46-2.58)
|
0
|
0.84
|
In addition, we also assessed the safety of different PARPis shown in Table 4. Olaparib has a higher risk than the other two drugs for any type of AEs of grade≥3 (RR 1.78 [95% CI, 1.49 to 2.13], p = 0.02). Similarly, for hematologic AEs, olaparib also showed a statistically significant higher risk of anemia compared to talazoparib and veliparib (RR 14.27 [95% CI, 6.68 to 30.47], p < 0.00001), while vilipalib had the lowest risk of anemia of the three (RR 1.02 [95% CI, 0.97 to 1.06], p < 0.00001). Talazoparib had better decreased white cell counts (RR 0.84 [95% CI, 0.42 to 1.96], p = 0.01). Interestingly, the above results were not reported in previous META analyses. Other safety results suggest that PARPis may increase the risk of fatigue, nausea, and headaches.
Table 4.
RRs of grade 3 or higher AEs according to drug type.
Adverse events
|
RR (95% CI)
|
P
|
Olaparib (t=2, n=2111)
|
Talazoparib (t=1, n=412)
|
Veliparib (t=3, n=735)
|
Any type
|
1.78(1.49-2.13)
|
1.05(0.75-1.48)
|
1.02(0.97-1.06)
|
0.02
|
Anemia
|
14.27(6.68-30.47)
|
8.44(3.82-18.67)
|
1.16(0.94-1.45)
|
<0.00001
|
Neutropenia
|
1.81(1.19-2.76)
|
0.64(0.46-0.88)
|
1.05(0.95-1.16)
|
0.34
|
Leukopenia
|
2.53(1.32-4.85)
|
0.84(0.42-1.69)
|
1.24(0.94-1.65)
|
0.01
|
Fatigue
|
3.92(1.49-10.32)
|
0.77(0.23-2.59)
|
1.22(0.67-2.24)
|
0.04
|
Nausea
|
6.24(0.77-50.64)
|
0.22(0.02-2.41)
|
1.48(0.69-3.18)
|
0.16
|
Vomiting
|
5.35(0.65-44.36)
|
1.03(0.27-3.91)
|
2.22(0.74-6.67)
|
0.06
|
Diarrhea
|
1.19(0.27-5.30)
|
0.13(0.03-0.60)
|
1.06(0.52-2.13)
|
0.29
|
Headache
|
1.19(0.27-5.30)
|
2.20(0.26-18.66)
|
1.06(0.25-4.38)
|
0.58
|
Sensitivity analyses and publication bias
Sensitivity analysis was used to assess the impact of individual studies on the overall results. The results showed that except for NCT02163694 for ORR and NCT02032823 for AEs, a single study did not significantly change the overall results of HRs (for PFS and OS) and RRs (for ORR and AEs), which demonstrated the robustness of the analysis (Supplementary Figure 2). We evaluated the publication bias of the included literatures by Begg's Funnel Plot and Egger's Test, and the results showed that there was no publication bias in PFS,OS,ORR and AEs (Begg's funnel plot p = 0.806 for PFS, p = 0.806 for OS, p = 0.806 for ORR, p = 0.452 for AEs; Egger's test p = 0.410 for PFS, p = 0.665 for OS, p = 0.186 for ORR, p = 0.12 for AEs; Supplementary Figure 3).