Efficacy and Safety of PARP Inhibitors in Advanced or Metastatic Triple-Negative Breast Cancer: A Systematic Review and Meta-Analysis

Purpose Poly (ADP-ribose) polymerase (PARP) inhibitors have shown promising results in metastatic triple-negative breast cancers (TNBCs). We therefore performed a systematic review and meta-analysis to evaluate the efficacy and safety of this drug in patients with advanced or metastatic TNBC. Methods A systematic literature search of PubMed, Embase, Scopus, Web of Science, and Cochrane Central Register of Controlled Trials for synonyms of “PARP inhibitors” and “breast cancer” was carried out. All published phase II/III clinical studies of PARP inhibitors in patients with advanced/metastatic TNBC were screened. Data were extracted independently by two authors and analyzed using Review Manager software version 5.3. End points include overall response rate (ORR), progression-free survival (PFS), and adverse events. Results Ten clinical trials were identified, with a total of 1,495 patients included. Pooled analyses showed that PARP inhibitors could provide a significant improvement of ORR [risk ratio (RR) = 2.00; 95% confidence interval (CI), 1.14–3.50; p = 0.02) and PFS [hazard ratio (HR) = 0.68; 95%Cl, 0.59–0.77; p < 0.0001) compared to chemotherapy in the whole population. In subgroup analysis, patients with BRCA mutation had a higher objective response to PARP inhibitor, with an RR of 2.85 (95%CI, 1.34–6.06; p = 0.007) compared to BRCA wild-type patients. However, no significant difference in ORR was observed between the homologous recombination deficiency (HRD) positive and non-HRD subgroups (RR = 1.82; 95%CI, 0.81–4.08; p = 0.14). Hematological toxicity is a common adverse event of PARP inhibitors. Conclusions PARP inhibitors are effective options for the treatment of patients with advanced or metastatic TNBC. Compared with patients without germline BRCA mutation, patients with germline BRCA mutation could benefit more from PARP inhibitors. In clinical setting, hematological toxicity associated with PARP inhibitors should be monitored regularly.


INTRODUCTION
Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer, accounting for 15%-20% of all cases of breast cancer (1,2). Cytotoxic chemotherapy is currently the main treatment of patients with TNBC due to the lack of expression of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 (3). Unfortunately, TNBC often develops resistance to chemotherapy and eventually progresses to lethal metastatic disease, with a median survival of approximately 1 year (4,5). Notably, more than one-third of TNBC patients will develop distant metastases (6), and there is no standard of care therapy for patients with metastatic TNBC (mTNBC) (7). Therefore, there is an urgent need to develop new therapeutic approach for patients with mTNBC.
With the advancement of genomics analysis, new druggable targets are being identified, which may contribute to broadening novel therapeutic scenario for mTNBC. Among patients with TNBC, about 10%-30% cases present with germline BRCA1 or BRCA2 gene mutation (8,9). On the other hand, approximately 70% of BRCA1-mutated breast cancers have a triple-negative phenotype (10). BRCA1 and BRCA2 genes are involved in the homologous recombination repair (HRR) pathway and are responsible for repairing DNA double-strand breaks. The HRR pathway also contains many other genes, such as ATM, PALB2, RAD51, CDK12, and CHK1/2. Alterations on these genes can lead to homologous recombination deficiency (HRD). Poly (ADP-ribose) polymerase (PARP) is an important enzyme for repairing DNA single-strand break (11,12). Thus, the inactivation of PARP in tumors with HRD will increase genomic instability and eventually result in cell death. Preclinical studies have shown that cancer cells with functional BRCA1 or BRCA2 mutations are sensitive to PARP inhibitors (13), providing a strong rationale for using PARP inhibitors to treat mTNBC.
There are currently several PARP inhibitors being tested in clinical trials in patients with mTNBC. Based on the promising results observed in clinical trials, olaparib has been approved by the Food and Drug Administration (FDA) for the treatment of patients with germline BRCA mutation (gmBRCA) and HER2negative metastatic breast cancer (14,15). Several clinical trials have shown that PARP inhibitors could also confer a survival benefit in patients with metastatic TNBC irrespective of BRCA status (16,17). Therefore, in this systematic review and metaanalysis, we aimed to comprehensively evaluate the efficacy and safety of PARP inhibitors in patients with advanced or metastatic TNBC based on available clinical trial results. We also explored biomarkers to determine the subgroup of patients who could benefit most from PARP inhibitors.

Search Strategy
On August 2020, a systematic literature search was performed by two independent reviewers through PubMed, Embase, Scopus, Web of Science, and Cochrane Central Register of Controlled Trials according to the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) guidelines (18). The search terms are as follows: [("Poly(ADP-ribose) Polymerase inhibitors" OR "PARP inhibitors"] OR "Olaparib" OR "rucaparib" OR "talazoparib" OR "veliparib" OR "niraparib" OR "iniparib") AND ("breast") AND ("randomized controlled trial" OR "clinical trial").

Inclusion and Exclusion Criteria
The following were the inclusion criteria: (1) phase II and III clinical trials evaluating the efficacy of PARP inhibitor as single agent or in combination with other anticancer drugs in the treatment of patients with advanced or metastatic TNBC were considered for inclusion; (2) the eligible studies mentioned objective response rate (ORR), progression-free survival (PFS), or safety outcomes; and (3) only English-language articles were included.
Exclusion criteria were as follows: (1) phase I clinical trial, case reports, editorials, review articles, and retrospective studies were excluded; (2) single-arm studies that did not report BRCA or HRD status were not included; (3) clinical trials focusing on neoadjuvant therapy; (4) finally, for the continuously updated and published follow-up data, the latest results were considered for analysis. The selected studies were identified based on inclusion and exclusion criteria by two independent reviewers.

Data Extraction
Two authors independently extracted data from eligible studies included in the meta-analysis. The following data were included: first author's information, year of publication, study design, trial phase, ClinicalTrial.gov Number, sample size, BRCA or HRD status, type of intervention/control, efficacy results (ORR and PFS), and numbers of adverse events (AEs) in each arm. If the PFS was only represented by the Kaplan-Meier curve, the Engauge digitizer 4.1 software was used to digitize and extract the data [only in one study (19)].

Risk of Bias Assessment
The potential risk of bias in the selected studies was independently assessed by two reviewers, using Cochrane Risk of bias tool, which included selection bias, performance bias, detection bias, attrition bias, reporting bias, and other possible sources of bias. The risk of bias was graded as high, low, or unclear risk. Disagreements were resolved through consensus or a third reviewer.

Study Objectives
The objective of this study was to compare the antitumor efficacy and safety between the PARP inhibitor group and the chemotherapy group. The primary outcomes of this metaanalysis were ORR and PFS; AEs were the secondary outcomes. We also performed exploratory subgroup analyses to investigate the therapeutic activity of PARP inhibitors in the BRCA-mutated group vs. BRCA wild-type group, and HRD group vs. non-HRD group.

Statistical Analysis
The hazard ratio (HR) with its 95% confidence interval (CI) was calculated to compare the PFS. The risk ratios (RR) and 95% CI were calculated to measure the ORR and AEs. A two-sided pvalue <0.05 was considered statistically significant. The statistical heterogeneity was assessed using the I 2 statistic and chi-squared test. When heterogeneity was observed (I 2 > 50% and p < 0.05), the random effects model was applied; otherwise, the fixed effects model was used. Funnel plot was used to detect potential publication bias. All statistical analyses were performed using Review Manager software version 5.3.

Study Selection and Characteristics
After searching the electronic databases, a total of 2,689 records were initially retrieved ( Figure 1). After removing duplicates and screening titles and abstracts, only 27 full-text articles were further assessed for their eligibility based on inclusion/ exclusion criteria. After full-text review, 17 published articles were excluded for the following reasons: 5 articles did not report relevant outcomes of this study population; 5 studies reported results from the same population; 4 were clinical trials of neoadjuvant therapy for TNBC; 2 single-arm studies did not report HRD or BRCA mutation status; and 1 single-arm study only reported BRCA mutations. Finally, 10 clinical trials were included for final pooled analysis, including 7 randomized controlled trials (14,16,17,(19)(20)(21)(22)) and 3 single-arm studies (23)(24)(25). Of the 10 studies, 8 studies reported BRCA mutation status. Among them, the type of BRCA mutation reported in five studies was germline mutation, two studies were germline or somatic BRCA mutations, and the type of BRCA mutation reported in another study was unclear. For the BROCADE study (21), they randomly set up two comparison groups: veliparib with carboplatin/paclitaxel versus placebo plus carboplatin/paclitaxel, and veliparib plus temozolomide versus placebo plus carboplatin/paclitaxel; we only evaluated veliparib plus carboplatin/paclitaxel versus placebo plus carboplatin/ paclitaxel to avoid statistical influences on research weight.
The main features of selected studies and enrolled patients are summarized in Table 1. All clinical trials reported the antitumor efficacy of PARP inhibitors in patients with advanced or metastatic TNBC, ranging from 21 to 519 patients per study. Globally, a total of 1,495 patients were included in the meta-analysis, of whom 735 patients harbored somatic or germline BRCA1/2 mutations.

PARP Inhibitors vs. Control
To evaluate the effect of PARP inhibitors on patients with advanced/metastatic TNBC, we first conducted a pooled analysis to compare the antitumor efficacy between the PARP inhibitor group and the chemotherapy group. Seven randomized controlled trials were pooled into the analysis of ORR or PFS (14,16,17,(19)(20)(21)(22), including 778 patients with advanced/metastatic TNBC who received PARP inhibitors (olaparib, iniparib, veliparib, or talazoparib) and 568 participants who were administered with chemotherapy. In two studies, PARP inhibitors were investigated as monotherapy against standard chemotherapy. In five studies, PARP inhibitors were investigated in combination with cyclophosphamide, gemcitabinecarboplatin, and carboplatin/paclitaxel and compared with placebo. In the whole population, the pooled RR showed that PARP inhibitor treatment significantly improved the incidence of achieving ORR compared to chemotherapy (RR = 2.00; 95% CI, 1.14-3.50; p = 0.02) (Figure 2A). The pooled analysis for PFS indicated that the PARP inhibitor group had a better PFS (HR = 0.68; 95% CI, 0.59-0.77, p < 0.0001) when compared with the control group whether in combined therapy or monotherapy [HR = 0.74, 95% CI (0.64-0.87); HR = 0.51, 95% CI (0.39-0.67), respectively] ( Figure 3A).
Three clinical trials focused on the efficacy of PARP inhibitors in advanced or metastatic TNBC irrespective of BRCA or HRD status (16,17,19); we performed a pooled analysis in this unselected population. The results showed that no significant difference in ORR was observed between the PARP inhibitor group and the chemotherapy group (RR = 1.22; 95% CI, 0.99-1.52; p = 0.07) ( Figure 2C). However, PARP inhibitors showed a significant improvement in PFS (HR = 0.74; 95% CI, 0.62-0.89; p = 0.001) ( Figure 3C).

BRCA Mutated vs. BRCA Wild Type
To further compare the efficacy of PARP inhibitors in the BRCAmutated and BRCA wild-type populations, we subsequently conducted an exploratory analysis that directly compared these two groups. Three studies mentioned ORR in two subgroups and were incorporated into this analysis (23)(24)(25). Subgroup analysis demonstrated that PARP inhibitors could provide a significant improvement in ORR to patients with BRCA mutation in comparison to patients without BRCA mutation, with an RR of 2.85 (95% CI, 1.34-6.06; p = 0.007) ( Figure 4A).

HRD vs. Non-HRD
It is still unclear whether, in addition to BRCA mutations, HRD status can be used as a biomarker for predicting PARP inhibitors sensitivity in an advanced/metastatic TNBC setting. Therefore, we performed a subgroup analysis of the HRD-positive group vs. the non-HRD group to address this question. Two articles were eligible for this analysis, and only ORR data were available in the two studies with low statistical power (24,25). Interestingly, when comparing the HRD-positive subgroup with the non-HRD subgroup, there was no significant difference in ORR (RR = 1.82; 95% CI, 0.81-4.08; p = 0.14) ( Figure 4B).

Quality of Included Studies
The "risk of bias graph" revealed that this meta-analysis had a moderate risk of selection bias because 3 out of 10 clinical trials were single-arm studies (Supplementary Figure S1). We used the funnel plots to detect publication bias, and the results suggested that there was a relatively low risk of publication bias (Supplementary Figure S2).

DISCUSSION
The results of this study highlight that compared with chemotherapy, PARP inhibitors can safely and significantly improve ORR and PFS in patients with advanced/metastatic TNBC. Furthermore, exploratory analysis showed that the patients with mutation in BRCA could derive more benefits from PARP inhibitors than BRCA wild-type patients. However, we did not observe any difference in tumor control between HRD-positive patients and non-HRD patients. Based on these recent clinical evidence, BRCA mutations, rather than HRD status, can be used as a predictive biomarker of response to PARP inhibitors in the advanced/metastatic TNBC setting. Both BRCA1 and BRCA2 genes are key components of the HRR pathway and are responsible for repairing DNA doublestrand breaks. Alterations on BRCA1/2 or other components may lead to HRD. PARP is an important enzyme in response to DNA single-strand breaks. Inhibition of PARP can cause DNA single-strand breaks, which subsequently results in DNA doublestrand breaks. In normal cells, these breaks can be repaired through the HRR pathway. However, in HR-deficient tumors, the breaks caused by PARP inhibition would not be repaired, eventually leading to tumor cell death (26). Preclinical studies  have shown that PARP inhibitors have greater efficacy in BRCAdeficient cells when compared with wild-type cells (27,28). In a clinical setting, a proof-of-concept trial by Tutt showed that PARP inhibitor treatment has a favorable therapeutic index for patients with germline BRCA mutation and advanced breast cancer (29). Based on these results, over the past years, several clinical trials have been conducted and are currently evaluating the effects of different PARP inhibitors in this population. Specifically, olaparib has been approved for patients with a germline BRCA mutation and HER2-negative metastatic breast cancer based on the results of the OlympiAD study (14). In view of this meta-analysis, our conclusion also confirmed that patients with germline BRCA mutation might be prime candidates for PARP inhibition treatment. However, our analysis also found that PARP inhibitors could provide significant improvement in PFS for unselected patients, regardless of BRCA mutational status. Similarly, in phase II and III clinical trials of patients with metastatic TNBC, irrespective of BRCA status (16,17), the gain in PFS was obtained in the PARP inhibition group compared with the chemotherapy group. In the setting of neoadjuvant treatment, PARP inhibitors could also increase the pathological complete response rate of TNBC population, regardless of BRCA status (30). Moreover, those with BRCA mutation account for only a small proportion of breast cancer patients (31). Hence, using only BRCA status as a predictive biomarker of PARP inhibitors sensitivity is insufficient, and many potential responders may be missed. The main question facing oncologists is how to go about practically selecting patients with advanced/metastatic TNBC who will benefit from PARP inhibitor therapy. With large-scale sequencing analysis, in addition to BRCA1/2 gene, many other HRD genes (ATM, CHK1/2, and PTEN) were found to be correlated with PARP inhibition sensitivity, which could be utilized as alternative biomarkers for identifying susceptible population (32)(33)(34). In clinical situations, HRD status has good predictive power for the benefits of PARP inhibitors in several cancer types, such as ovarian, prostate, and gastric cancer (35). However, in this study, we did not observe any improvement in ORR of HRD-positive patients compared with HRD-negative patients. It should be noted that there were only two relevant studies in this subgroup analysis, with small sample size and low statistical power, making it difficult to draw strong conclusions. In addition, recent studies have shown that low RAD51 score, high TIL, or high PDL1 expression are all associated with the response to PARP inhibitors in TNBC, which indicates that in addition to BRCAness signature, immunological features could be regarded as potential predictive biomarkers of PARP inhibitors (36)(37)(38). In the future, more clinical trials are needed to evaluate the predictive value of HRD in the TNBC setting and to explore new biomarkers for determining optimal patients who are more likely to benefit from PARP inhibitors.
As PARP inhibitors have gradually been approved for clinical application worldwide, it is of great value and indispensable significance to evaluate the safety and tolerance of PARP  inhibitors in patients. Our pooled analysis of 1,346 patients with advanced/metastatic breast cancer treated with PARP inhibitors from seven randomized controlled trials showed that the three most common AEs of grade 3 and above were neutropenia (47.03%), thrombocytopenia (30.32%), and anemia (27.56%), indicating that hematological toxicity events caused by PARP inhibition treatment are more common and serious. Similarly, a meta-analysis of eight clinical trials also found that olaparib could significantly increase the risk of severe neutropenia in cancer patients (39). This may be because PARP inhibitors not only interfere with the DNA repair of cancer cells but may also interfere with rapidly dividing blood cells, thus leading to myelosuppression. Notably, our meta-analysis showed that PARP inhibitors did not increase the incidence of grade ≥3 AEs except for the risk of thrombocytopenia when compared with chemotherapy. Overall, PARP inhibitors seem to be safe and tolerable for patients with advanced/metastatic breast cancer, but the high risk of PARP inhibitor-related hematological toxicity is an issue that cannot be ignored and should be considered in clinical applications.
There are several limitations in the present meta-analysis. First, the potential biases of this study included the heterogeneity of inclusion criteria, patients and treatment schedule in the included trials, for example, age, race, disease status, type of PARP inhibitors, and interventional arm. These confounding variables were not stratified properly and incorporated into the meta-analysis. In addition, there are few comparative studies on the efficacy of PARP inhibitors between HRD-positive patients and HRD-negative patients, which makes it difficult to fully assess the benefit of PARP inhibitors based on the HRD status of patients. Therefore, it is warranted to carry out randomized controlled trials with longer clinical follow-up time in the future.

CONCLUSIONS
Our findings confirm that PARP inhibitor is an effective, welltolerated treatment in patients with advanced/metastatic TNBC. We also support the view that BRCA status can be used as a predictive biomarker of PARP inhibitor sensitivity to guide clinical decision-making. However, the predictive value of HRD status still needs to be further evaluated in future studies. Hematological toxicity is a common adverse event; thus, regular hematological monitoring is warranted for patients receiving PARP inhibitors.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
XL, KW, and XZ conceived and designed the analysis. XL and KW collected the data. XL, KW, and DZ analyzed the data. CL, YF, XL, and HZ interpreted the data. All authors were involved in the drafting, critical review, and approval of the final manuscript and the decision to submit for publication.