Major adverse cardiac events and cardiovascular toxicity with PARP inhibitors-based therapy for solid tumors: a systematic review and safety meta-analysis

Abstract

Background

Poly(ADP-ribose) polymerase (PARP) inhibitors (PARPi) provided significant antitumor activity in various tumors, mainly carrying deleterious mutations of BRCA1/BRCA2 genes. Only few data are available regarding the cardiac and vascular safety profile of this drug class. We carried out a meta-analysis for assessing the incidence and relative risk (RR) of major adverse cardiovascular events (MACEs), hypertension, and thromboembolic events in patients with solid tumors treated with PARPi-based therapy.

Methods

Prospective studies were identified by searching the Medline/PubMed, Cochrane Library, and ASCO Meeting abstracts. Data extraction was conducted according to the Preferred Reporting Items for Systematic review and Meta-Analyses (PRISMA) statement. Combined odds ratios (ORs), RRs, and 95% confidence intervals (CIs) were calculated using fixed- or random-effects methods, depending on studies heterogeneity. RevMan software for meta-analysis (v.5.2.3) was used to carry out statistical analyses.

Results

Thirty-two studies were selected for the final analysis. The incidence of PARPi-related MACEs of any and high grade was 5.0% and 0.9%, respectively, compared with 3.6% and 0.9% in the control arms, corresponding to a significant increased risk of MACEs of any grade (Peto OR 1.62; P = 0.0009) but not of high grade (P = 0.49). The incidence of hypertension of any grade and high grade was 17.5% and 6.0% with PARPi, respectively, compared with 12.6% and 4.4% in the controls. Treatment with PARPi significantly increased the risk of hypertension of any grade (random-effects, RR = 1.53; P = 0.03) but not of high grade (random-effects, RR = 1.47; P = 0.09) compared with controls. Finally, PARPi-based therapies significantly increased the risk of thromboembolic events of any grade (Peto OR = 1.49, P = 0.004) and not of high grade (Peto OR = 1.31; P = 0.13) compared with controls.

Conclusions

PARPi-based therapy is associated with a significantly increased risk of MACEs, hypertension, and thromboembolic events of any grade compared with controls. The lack of a significant increased risk of high-grade events together with the absolute low incidence of these adverse events led not to consider routine cardiovascular monitoring as recommended in asymptomatic patients.

Highlights

• •PARPi-based therapy significantly increases the risk of any-grade MACEs.
• •The risks of any-grade hypertension and thromboembolic events are also increased.
• •No significantly increase in high-grade events was observed.

Introduction

Poly(ADP-ribose) polymerase (PARP) inhibitors (PARPi) have provided, over the last 10 years, significant antitumor activity in various tumor types, including ovarian, breast, pancreatic, and prostate cancers.^1,2 The efficacy of PARPi is well defined in tumors carrying deleterious mutations of BRCA1/BRCA2 genes, whereas it is more controversial in tumors harboring inactivating mutations in other homologous recombination (HR)-related genes and in wild-type cancers.^3,4 BRCA1 and BRCA2 are key tumor suppressor genes involved in the repair process of double-strand DNA break (DSB) by HR. DNA repair genes defects are responsible for genomic instability favoring tumor progression; however, from the other side, they make cells selectively vulnerable to the pharmacological inhibition of DNA repairing enzymes. PARP1 and PARP2 enzymes are involved in the identification of single-strand DNA breaks (SSBs), and mediate the recruitment of effectors of the base excision repair (BER) system for DNA repair. The rationale for the pharmacological inhibition of PARP as antitumor therapy is based on the biological principle of ‘synthetic lethality’: two genomic events, relatively innocuous individually, become lethal if they occur concurrently. Indeed, PARP1 inhibition has no lethal effects per se, as these agents impair BER function leading to DNA lesions (SSBs potentially evolving to DSBs) that can be repaired by other DNA repair pathways, precisely HR. Similarly, the unique loss of BRCA1/2, and the consequent defective HR system, can be counterbalanced by BER and other DNA repair processes. On the contrary, the inhibition of PARP in BRCA1/2-defective cells causes an accumulation of DSBs, leading to cell death.3, 4, 5

Actually, PARPi represent a success of precision medicine, with concrete opportunities ahead to increase their use either alone or combined with other therapeutic strategies (i.e. radiotherapy, DNA damaging chemotherapy, immune checkpoint inhibitors, anti-androgenic agents). The peculiar mechanism of action of PARPi, which selectively target cells carrying a dysfunctional HR pathway sparing the normal cell counterparts, should avoid a severe toxicity profile. Generally, however, PARPi are not free from adverse events (AEs), with some peculiar class-specific side-effects.⁶ Indeed, the most common AEs of PARPi observed in randomized, controlled trials (RCTs) are fatigue, and hematological and gastrointestinal toxicities, which generally occurred during the first 3 months of treatment.⁶ Nevertheless, sporadic but alarming signals of cardiac and vascular events were reported, particularly in patients treated with niraparib due to its large spectrum of off-target effects.⁷ We carried out a systematic review and a safety meta-analysis of phase II and III randomized studies to comprehensively evaluate the incidence and relative risk (RR) of developing major adverse cardiovascular events (MACEs) and cardiovascular toxicity (including hypertension and thromboembolic events) in patients treated with PARPi-based therapy for solid tumors.

Patients and methods

Definition of outcomes

The objective of this analysis was to assess the incidence and RR of MACEs, and cardiovascular toxicity including hypertension and thromboembolic events in cancer patients treated with PARPi-based therapy. MACEs considered included acute myocardial infarction, cardiac arrest, cardiac failure, arrhythmia, tachycardia, and bradycardia. Thromboembolic events included arterial and venous thromboembolism, thromboembolic events, pulmonary embolism, ischemic stroke, transient ischemic attack, cerebrovascular accidents, cerebrovascular disorders, thrombophlebitis, and venous thrombosis. For each trial, the PARPi-based treatment was considered as the experimental arm and the comparator as the control. The primary endpoint was to assess the RR of MACEs both of any grade (grades 1-5) and high grade (grades 3-5), and the analysis was conducted in order to identify a significant difference between the two treatment arms. Secondary endpoints included the evaluation of the RR of hypertension, thromboembolic events, and pulmonary embolism. Moreover, we carried out a subgroup analysis for each cardiovascular event according to the type of PARPi-based treatment received.

Selection of the studies

We reviewed Medline/PubMed, the Cochrane Library, and ASCO University Meeting abstracts for citations up to 30 April 2022. The search criteria were limited to articles published in the English language and phase III or phase II RCTs in patients with solid tumors treated with PARPi-based therapy in the experimental arm and with available data about treatment-related cardiovascular events. The following inclusion criteria have been adopted: >10 patients enrolled and randomized trials reporting data on treatments with a PARPi in cancer patients. Principal exclusion criteria were overlapping publications, lack of relevant safety data, and the presence of a PARPi in the treatment control arm.

The Medical Subject Headings (MeSH) terms used for the search of PubMed and the Cochrane Library were ‘PARPi’, ‘PARP inhibitor’, or the name of the drugs (olaparib, niraparib, rucaparib, talazoparib, etc.), ‘cardiotoxicity’, ‘cardiovascular toxicity’, ‘MACE’, and ‘hypertension’. For the search in the ASCO University abstracts, we used the name of the drugs and the terms ‘phase II’ or ‘phase III’. The summaries for the product characteristics were searched for at http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm. If more than one publication was found for the same trial, the most recent, complete, and updated version was included in the final analysis.

The quality of the selected studies was assessed using the Jadad 5-item scale, taking into account randomization, double blinding, and withdrawals. The final score ranged from 0 to 5.⁸

Data extraction

Two authors (CC, AP) conducted the data extraction independently. It was carried out according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement (Supplementary Table 1, available at https://doi.org/10.1016/j.esmoop.2023.101154), and any types of discrepancies were resolved by consensus.⁹ The data extracted for each trial were: name of the study, year of publication, trial phase, number of enrolled patients, type of solid tumor, type of treatment in each arm (experimental and control), and number of any-grade and high-grade MACEs, hypertension, thromboembolic events, and pulmonary embolism in both treatment arms.

Statistical methodology

The calculation of incidence was carried out from the data available in each study. The proportion of patients with MACEs, hypertension, thromboembolic events, and pulmonary embolism and the derived 95% confidence intervals (CIs) were calculated for each study. We also calculated the RR and CIs of events in patients assigned to PARPi-based treatments compared with the controls in the same study. To calculate the 95% CIs, the variance of a log-transformed study-specific RR was derived using the delta method.¹⁰ Of note, given the rarity of the different types of cardiac events reported in each trial, we decided to merge them together into the definition of MACEs as aggregate data and we used the Peto method for calculating the odds ratio (OR); the same statistical method was used for analyzing the thromboembolic events. The Mantel–Haenszel method was used for calculating the RR of hypertension and pulmonary embolism. Statistical heterogeneity between the trials included in the meta-analysis was assessed using Cochrane’s Q statistic, and inconsistency was quantified with an I² statistic (100% × [Q-df/Q]).¹¹ The assumption of homogeneity was considered to be invalid for P values <0.1. Summary incidence and RRs were calculated using random- or fixed-effects models, depending on the heterogeneity of the included studies. When there was no substantial heterogeneity, the pooled estimate that was calculated based on the fixed-effects model was reported using the inverse variance method. When substantial heterogeneity was observed, the pooled estimate that was calculated based on the random-effects model was reported using the DerSimonian and Laird method,¹² which considers both within- and between-study variations.¹¹ An indirect comparison between the groups was carried out using a χ² test. A two-tailed P value <0.05 was considered to be statistically significant. All the data were collated using Microsoft Office Excel 2007. The statistical analyses were carried out using the RevMan software for meta-analysis (v.5.2.3).¹³

Results

Search results

The electronic search revealed 15 148 citations, after screening 167 full text articles were reviewed for further assessment, and 135 citations were excluded because they did not meet the inclusion criteria. This reviewed process (Figure 1) led to the selection of 32 articles, of which 19 studies were considered for final analysis of MACEs, 24 articles were considered for hypertension, 25 for thromboembolic events, and 17 for pulmonary embolism based on their adequate quality and relevance for inclusion in this meta-analysis.14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 The characteristics of each trial analyzed in this study are shown in Table 1.

Figure 1.

Selection of the included studies.

PARP, poly(ADP-ribose) polymerase; RCT, randomized, controlled trial.

Table 1.

Selected studies for final analysis

Trial	Phase	Cancer type	Disease setting	Trial design								Primary endpoint	Jadad
				Exp				Ctr
				Drug		Pts (N)		Drug		Pts (N)
ARIEL3 Coleman et al. (2017)	III	Ovarian	Platinum-sensitive recurrent ovarian cancer	Rucaparib		375		Pbo		189		PFS (BRCA-mutant, HRD+, ITT)	5
ARIEL4 Kristeleit et al. (2022)	III	Ovarian	BRCA-mutated platinum-sensitive or platinum-resistant relapsed ovarian cancer	Rucaparib		233		CHT (TXL for platinum-resistant or partially platinum-sensitive disease; IC of platinum-based CHT for fully platinum-sensitive disease)		116		PFS (efficacy population, ITT population)	3
BAROCCO Colombo et al. (2022)	II	Ovarian	Platinum-resistant recurrent ovarian cancer	Olaparib + cediranib continuous	Olaparib + cediranib intermittent	41	41	TXL		41		PFS	3
BrighTNess Loibl et al. (2018)	III	Breast	Stage II-III triple-negative breast cancer	CBDCA + TXL + veliparib		316		TXL + Pbo + Pbo	CBDCA + TXL + Pbo	158	160	pCR	5
BROCADE3 Diéras et al. (2020)	III	Breast	BRCA-mutated HER2-negative advanced breast cancer	CBDCA + TXL + veliparib		336		CBDCA + TXL + Pbo		171		PFS	5
CLIO Vanderstichele et al. (2022)	II	Ovarian	Platinum-sensitive or platinum-resistant recurrent ovarian cancer	Olaparib		107		CHT (of PC)		53		Assessment of HRD in circulating tumor DNA to predict response to olaparib monotherapy	2
EMBRACA Litton et al. (2018)	III	Breast	Germline BRCA-mutated advanced breast cancer	Talazoparib		286		CHT (single agent of PC)		126		PFS	3
GOLD Bang et al. (2017)	III	Gastric	Advanced gastric cancer	TXL + olaparib		263		TXL + Pbo		262		OS (overall patient population, ATM-negative population)	5
I-SPY-2 Rugo et al. (2016)	II	Breast	Stage II-III HER2-negative breast cancer	CBDCA + TXL + veliparib		72		TXL		44		pCR (TN, HER2−, HR+/HER2−)	3
I-SPY-2 Pusztai et al. (2021)	II	Breast	Stage II-III HER2-negative breast cancer	TXL + durvalumab + olaparib		73		TXL		299		pCR (HER2−, HR+ /HER2−, TN)	3
MAGNITUDE	III	Prostate	mCRPC	Abiraterone + niraparib		212		Abiraterone + Pbo		211		rPFS (BRCA1/2 mut, HRR BM+)	NE
NCI8993 O’Reilly et al. (2020)	II	Pancreatic	Locally advanced or metastatic gBRCA/PALB2 + PDAC	CDDP + GMZ + veliparib		27		CDDP + GMZ		23		RR	3
NCI9012 Hussain et al. (2017)	II	Prostate	mCRPC	Abiraterone + veliparib		76		Abiraterone		72		PSA RR (overall, ETS+, ETS−	3
NCT00753545 Ledermann et al. (2012)	II	Ovarian	Platinum-sensitive relapsed ovarian cancer	Olaparib		136		Pbo		129		PFS	5
NCT01306032 Kummar et al. (2016)	II	Breast	Advanced triple-negative breast cancer	Oral cyclophosphamide + veliparib		21		Oral cyclophosphamide		18		RR	2
NCT01657799 Chabot et al. (2017)	II	Non-small-cell lung cancer	Brain metastases from non-small-cell lung cancer	WBRT + veliparib 50 mg b.i.d.	WBRT + veliparib 200 mg b.i.d.	103	102	WBRT + Pbo		102		OS	3
NCT01972217 Clarke et al. (2018)	II	Prostate	mCRPC	Abiraterone + olaparib		71		Abiraterone + Pbo		71		rPFS	5
NCT02264990 Govindan et al. (2022)	III	Non-small-cell lung cancer	Advanced non-squamous NSCLC	CBDCA + TXL + veliparib		293		CHT (IC of platinum doublet)		288		OS in LP52 + patients	3
NCT02289690 Byers et al. (2021)	II	Small-cell lung cancer	Extensive-stage SCLC	CBDCA + VP-16 + veliparib followed by veliparib maintenance	CBDCA + VP-16 + veliparib followed by Pbo maintenance	61	59	CBDCA + VP-16 + Pbo followed by Pbo maintenance		61		PFS	5
NCT02305758 Gorbunova et al. (2019)	II	Colorectal	mCRC	Veliparib + FOLFIRI ± bevacizumab		65		Pbo + FOLFIRI ± bevacizumab		65		PFS	4
NORA Wu et al. (2021)	III	Ovarian	Platinum-sensitive recurrent ovarian cancer	Niraparib		177		Pbo		88		PFS	5
NOVA Mirza et al. (2016)	III	Ovarian	Platinum-sensitive recurrent ovarian cancer	Niraparib		367		Pbo		179		PFS (gBRCA mut, non-gBRCA HRD+, overall non-gBRCA mut)	5
NRG-GY004 Liu et al. (2022)	III	Ovarian	Platinum-sensitive recurrent ovarian cancer	Olaparib + cediranib	Olaparib	183	187	CHT (platinum-doublet)		167		PFS	2
OLYMPIA Tutt et al. (2021)	III	Breast	High-risk HER2-negative early breast cancer and a germline BRCA mutation	Olaparib		911		Pbo		904		iDFS	5
PAOLA1 Ray-Coquard et al. (2019)	III	Ovarian	Newly diagnosed advanced ovarian cancer	Bevacizumab + olaparib		537		Bevacizumab + Pbo		269		PFS	5
PRIMA González-Martín et al. (2019)	III	Ovarian	Newly diagnosed advanced ovarian cancer	Niraparib		487		Pbo		246		PFS (HRD+, overall population)	5
PROFOUND de Bono et al. (2020)	III	Prostate	mCRPC	Olaparib		256		Abiraterone/enzalutamide		130		Imaging-based PFS	3
PROpel	III	Prostate	mCRPC	Abiraterone + olaparib		399		Abiraterone + Pbo		397		rPFS	NE
SOLO1 Moore et al. (2018)	III	Ovarian	Newly diagnosed BRCA-mutated advanced ovarian cancer	Olaparib		260		Pbo		131		PFS	5
SOLO2 Pujade-Lauraine et al. (2017)	III	Ovarian	BRCA-mutated platinum-sensitive relapsed ovarian cancer	Olaparib		196		Pbo		99		PFS	5
SOLO3 Penson et al. (2020)	III	Ovarian	Germline BRCA-mutated platinum-sensitive relapsed ovarian cancer	Olaparib		178		CHT (single-agent non-platinum of PC)		88		ORR	3
VERTU Sim et al. (2021)	II	Brain	Newly diagnosed MGMT-unmethylated glioblastoma	Veliparib + RT followed by adjuvant veliparib + temozolomide		83		Temozolomide + RT followed by adjuvant temozolomide		40		PFS-6m	3
ZL-2306-005 Ai et al. (2021)	III	Small-cell lung cancer	Extensive-stage SCLC	Niraparib		125		Pbo		60		PFS, OS	5

CBDCA, carboplatin; CDDP, cisplatin; CHT, chemotherapy; GMZ, gemcitabine; HER2, human epidermal growth factor receptor 2; HR, hormone receptor; HRD, homologous recombination deficiency; HRR, homologous recombination repair; IC, investigator’s choice; iDFS, invasive disease-free survival; ITT, intention-to-treat; LP52, lung panel 52; mCRC, metastatic colorectal cancer; mCRPC, metastatic castration-resistant prostate cancer; MGMT, O6-methylguanine-DNA methyltransferase; mut, mutation positive; N, number of patients; NE, not evaluable; NSCLC, non-small-cell lung cancer; ORR, overall response rate; OS, overall survival; Pbo, placebo; PC, physician’s choice; pCR, pathological complete response; PDAC, pancreatic ductal adenocarcinoma; PFS, progression-free survival; PSA, prostate-specific antigen; Pts, patients; rPFS, radiographic progression-free survival; RR, response rate; RT, radiation therapy; SCLC, small-cell lung cancer; TN, triple negative; TXL, paclitaxel; VP-16, etoposide; WBRT, whole-brain radiation therapy.

Incidence and RR of MACEs

For incidence of MACEs of any grade, 14 studies were selected for the analysis, with a total of 5185 patients. Among them, 2895 were treated in the experimental arms with a PARPi-based therapy, whereas 2290 patients received standard therapies in the control arms. In the overall cohort, MACEs of any grade were reported in 146 out of 2895 patients treated with PARPi-based therapies, corresponding to an incidence of 5.0%, compared with 3.6% in the control arms (82 events among 2290 patients) (Table 2). Treatment with PARPi-based therapies significantly increased the risk of MACEs of any grade compared with controls (Peto, fixed, Peto OR = 1.62, 95% CI 1.22-2.15; P = 0.0009). Significant heterogeneity was observed in this analysis (χ² = 28.70, P = 0.02; I² = 48%) (Figure 2A).

Table 2.

Overall incidence of any- and high-grade MACEs, hypertension, and thromboembolic events

	MACEs				Hypertension				Thromboembolic events
	Any-grade		High-grade		Any-grade		High-grade		Any-grade		High-grade
	PARPi-based therapy	Control	PARPi-based therapy	Control	PARPi-based therapy	Control	PARPi-based therapy	Control	PARPi-based therapy	Control	PARPi-based therapy	Control
Number of events/tot pts	146/2895	82/2290	41/4332	29/3294	866/4950	450/3568	332/5503	172/3876	159/3839	79/2992	91/4575	46/3280
Incidence (%)	5.0	3.6	0.9	0.9	17.5	12.6	6.0	4.4	4.1	2.6	2.0	1.4

MACE, major adverse cardiovascular event; PARPi, poly(ADP-ribose) polymerase inhibitor; pts, patients.

Figure 2.

Relative risk for MACEs of any grade (A) and high grade (B) in patients treated with PARPi-based therapies for solid tumors.

CI, confidence interval; df, degrees of freedom; MACE, major adverse cardiovascular event; PARP, poly(ADP-ribose) polymerase; PARPi, PARP inhibitor.

As concerns MACEs of high grade, data were available from 18 studies and 7626 patients. MACEs of high grade were reported in 41 of 4332 patients treated with PARPi-based therapies, corresponding to an incidence of 0.9%, compared with 0.9% in the control arms (29 events among 3294 patients). Treatment with PARPi-based therapies did not significantly increase the risk of high-grade MACEs compared with controls (Peto fixed, Peto OR = 1.19, 95% CI 0.73-1.92; P = 0.49). Significant heterogeneity was observed in this analysis (χ² = 26.63, P = 0.05; I² = 40%) (Figure 2B).

Moreover, we evaluated if the type of PARPi-based therapy could affect the risk of developing a MACE of both low and high grade. We therefore identified three different subgroups based on whether the therapy of the experimental arm was only with a PARPi (PARPi monotherapy; three studies included for any-grade MACEs and seven studies for MACEs of high grade), or the combination of a PARPi with chemotherapy (chemotherapy + PARPi; data available from six studies for MACE of both any and high grade) or with an Androgen Receptor Signaling Inhibitor (ARSi) (ARSi + PARPi; four studies considered for any-grade and three studies for high-grade MACEs). Of note, while chemotherapy + PARPi did not significantly increase the RR of MACEs of any grade (P = 0.24), the two subgroups of PARPi monotherapy (Peto fixed, Peto OR = 3.26, 95% CI 1.33-7.98; P = 0.010) and ARSi + PARPi (Peto fixed, Peto OR = 1.71, 95% CI 1.08-2.70; P = 0.02) were significantly associated with an increased risk of developing a MACE compared with controls. There was no significant difference in the RR of MACEs of any grade for PARPi between the three subgroups (P = 0.21) (Supplementary Figure S1A, available at https://doi.org/10.1016/j.esmoop.2023.101154).

With regards to high-grade MACEs according to the type of PARPi-based therapy, there was not a significant augmented risk of developing MACEs of high grade with chemotherapy + PARPi (P = 0.89) or with PARPi monotherapy (P = 0.90). On the contrary, the subgroup of ARSi + PARPi significantly increased the risk of high-grade MACEs compared with controls (Peto fixed, Peto OR = 2.08, 95% CI 1.04-4.11; P = 0.04), with the caveat that patient numbers are limited in this analysis. No significant difference was observed in the RR of MACEs of high grade for PARPi between the three subgroups (P = 0.43) (Supplementary Figure S1B, available at https://doi.org/10.1016/j.esmoop.2023.101154).

Incidence and RR of hypertension

As concerns hypertension of any grade, 21 studies were selected for the analysis, with a total of 8518 patients. Among them, 4950 were treated in the experimental arms with PARPi-based therapies, whereas 3568 received standard therapies in the control arms. In the overall cohort, hypertension of any grade was reported in 866 of 4950 patients treated with PARPi-based therapies, corresponding to an incidence of 17.5%, compared with 12.6% in the control arms (450 cases of hypertension among 3568 patients). PARPi-based therapies significantly increased the risk of hypertension of any grade compared with controls (random-effects, RR = 1.53, 95% CI 1.04-2.24; P = 0.03). Significant heterogeneity was observed in this analysis (χ² = 181.94, P < 0.00001; I² = 87%) (Figure 3A).

Figure 3.

Relative risk for hypertension of any grade (A) and high grade (B), and of thromboembolic events of any grade (C) and high grade (D) in patients treated with PARPi-based therapies for solid tumors.

CI, confidence interval; df, degrees of freedom; PARPi, poly(ADP-ribose) polymerase inhibitor.

We evaluated the incidence and RR of developing hypertension of high grade; 24 studies were selected with a total of 9379 patients. Hypertension of high grade was reported in 332 of 5503 patients treated with PARPi-based therapy, corresponding to an incidence of 6.0%, compared with 4.4% in the control arms (172 cases of hypertension among 3876 patients). Treatment with PARPi-based therapies did not significantly increase the risk of high-grade hypertension compared with controls (random-effects, RR = 1.47, 95% CI 0.94-2.32; P = 0.09). Significant heterogeneity was observed in this analysis (χ² = 77.38, P < 0.00001; I² = 68%) (Figure 3B).

When the analysis of hypertension was carried out according to the type of PARPi-based therapy, four subgroups were identified, including chemotherapy + PARPi (5 studies included for any- and high-grade hypertension), PARPi monotherapy (data available from 10 studies for hypertension of any grade and from 12 studies for high-grade hypertension), ARSi + PARPi (4 studies considered for hypertension of both any and high grade), and the combination of a PARPi plus an antiangiogenic agent (antiangiogenic + PARPi; data available for 3 studies).

As concerns hypertension of any grade, we did not observe a significant increase in the risk of any-grade hypertension in any subgroup considered (chemotherapy + PARPi P = 0.30; PARPi monotherapy P = 0.07; ARSi + PARPi P = 0.81; antiangiogenic + PARPi P = 0.14). There was no significant difference in the RR of any-grade hypertension for PARPi between the four subgroups (P = 0.21) (Supplementary Figure S2A, available at https://doi.org/10.1016/j.esmoop.2023.101154). Similarly, none of the PARPi-based therapy subgroup significantly increased the risk of developing hypertension of high grade compared with controls (chemotherapy + PARPi P = 0.89; PARPi monotherapy P = 0.12; ARSi + PARPi P = 0.64; antiangiogenic + PARPi P = 0.25). No significant difference in the RR of any-grade hypertension was observed according to the type of PARPi-based treatment (P = 0.50) (Supplementary Figure S2B, available at https://doi.org/10.1016/j.esmoop.2023.101154).

Incidence and RR of thromboembolic events and pulmonary embolism

We analyzed the incidence and RR of developing thromboembolic events of any grade in patients treated with PARPi-based therapies compared with controls. Nineteen studies were selected for the analysis, with a total of 6831 patients. Among them, 3839 were treated in the experimental arms with PARPi-based therapies, whereas 2992 received standard therapies in the control arms. The incidence of thromboembolic events of any grade was 4.1% (159 events reported among 3839 patients treated with PARPi-based therapies), compared with 2.6% in the control arms (79 cases of thromboembolic events among 2992 patients). PARPi-based therapies significantly increased the risk of thromboembolic events of any grade compared with controls (Peto fixed, Peto OR = 1.49, 95% CI 1.14-1.95; P = 0.004). Significant heterogeneity was observed in this analysis (χ² = 28.29, P = 0.08; I² = 33%) (Figure 3C). When analyzing the thromboembolic events of high grade, data were available from 24 studies and a total of 7855 patients. The incidence of thromboembolic events of high grade was 2.0% (91 events reported among 4575 patients treated with PARPi-based therapies), compared with 1.4% in the control arms (46 cases of thromboembolic events among 3280 patients). PARPi-based therapies did not significantly increase the risk of thromboembolic events of high grade compared with controls (Peto fixed, Peto OR = 1.31, 95% CI 0.92-1.87; P = 0.13). No significant heterogeneity was observed in this analysis (χ² = 25.01, P = 0.41; I² = 4%) (Figure 3D).

The increase in risk of developing a thromboembolic event of any grade compared with controls was maintained in the subgroups of PARPi monotherapy (P = 0.03) and ARSi + PARPi (P = 0.01), but not in those with chemotherapy + PARPi (P = 0.52) and antiangiogenic + PARPi (P = 0.06). No significant difference was observed in the RR of thromboembolic event of any grade among the different PARPi-based treatment subgroups (P = 0.06) (Supplementary Figure S3A, available at https://doi.org/10.1016/j.esmoop.2023.101154). The significant augmented risk of thromboembolic events of high grade was confirmed only in the subgroup of patients treated with PARPi monotherapy (P = 0.04), whereas the other subgroups were not associated with a higher risk of high-grade thromboembolic events (chemotherapy + PARPi P = 0.47; ARSi + PARPi P = 0.16; antiangiogenic + PARPi P = 0.37). No significant difference in the RR of high-grade thromboembolic events was observed according to the type of PARPi-based treatment (P = 0.16) (Supplementary Figure S3B, available at https://doi.org/10.1016/j.esmoop.2023.101154).

Moreover, we analyzed the incidence and risk of developing pulmonary embolism (evaluated independently from other thromboembolic events) related to PARPi-based therapies. As concerns pulmonary embolism events of any grade, 13 studies were considered with a total of 4940 patients, and 16 studies for high-grade events (5496 patients). The incidence of pulmonary embolism of any grade was 2.8% (77 events reported among 2773 patients) compared with 1.6% in the controls (34 events reported among 2167 patients); the incidence of pulmonary embolism of high grade was 1.5% (48 events reported among 3194 patients) compared with 1.4% in the controls (32 events reported among 2302 patients). PARPi-based therapies were associated with a higher risk of pulmonary embolism of any grade compared with the controls (fixed-effects, RR = 1.69, 95% CI 1.14-2.52; P = 0.009). No significant heterogeneity was observed in this analysis (χ² = 17.14, P = 0.14; I² = 30%) (Supplementary Figure S4A, available at https://doi.org/10.1016/j.esmoop.2023.101154). On the contrary, treatment with PARPi-based therapies did not significantly increase the risk of pulmonary embolism of high grade compared with controls (fixed-effects, RR = 1.05, 95% CI 0.67-1.63; P = 0.84). No significant heterogeneity was observed in this analysis (χ² = 15.21, P = 0.51; I² = 0%) (Supplementary Figure S4B, available at https://doi.org/10.1016/j.esmoop.2023.101154).

When the subgroup analysis according to the type of PARPi-based therapy was carried out, an increased risk of pulmonary embolism of any grade was observed with ARSi + PARPi (P = 0.002), but not with chemotherapy + PARPi (P = 0.32) nor with PARPi monotherapy (P = 0.06). A significant difference in the RR of any-grade pulmonary embolism was observed according to the type of PARPi-based treatment (P = 0.005) (Supplementary Figure S4C, available at https://doi.org/10.1016/j.esmoop.2023.101154). None of the subgroups considered, however, was associated with a higher risk of pulmonary embolism of high grade compared with controls (chemotherapy + PARPi P = 0.13; PARPi monotherapy P = 0.18). It is important to underline that the small number of any- and high-grade pulmonary embolism events in each subgroup represented an important limitation for this analysis.

Quality of the studies

All the studies were of good quality according to the Jadad scale (scores ≥3) except three that had a Jadad score of 2 for the lack of description of the method of randomization (Table 1). Risk of bias of each study is reported in Supplementary Figure S5, available at https://doi.org/10.1016/j.esmoop.2023.101154.

Discussion

To the best of our knowledge, this large meta-analysis is the first to provide data on the incidence and RR of developing cardiovascular toxicities for cancer patients treated with PARPi. Our results confirm what emerged from RCTs, where no concerning safety signals occurred. PARPi-based therapies appear to be significantly associated with an augmented risk of low-grade MACEs and cardiovascular AEs (including hypertension, thromboembolic events, and pulmonary embolism), whereas high-grade events are not significantly increased in the overall population. In particular, MACEs and thromboembolic events induced by PARPi are rare, with an incidence not exceeding 5% for any-grade and 1%-2% for high-grade events. PARPi-based therapy is associated with a significant increased risk of developing any-grade MACEs (OR 1.62; P = 0.0009) and thromboembolic events (OR 1.49; P = 0.004) compared with controls, even if without an increased risk of high-grade events. Similarly, treatment with PARPi significantly increases the risk of any-grade hypertension compared with controls (RR 1.53; P = 0.03) but not of high-grade (P = 0.09); however, it is important to note that this adverse event is more common (incidence of 17.5% for any-grade and 4.4% for high-grade) than MACEs and thromboembolic events.

Furthermore, we carried out an exploratory analysis to investigate whether the type of PARPi-based treatment could affect the incidence and RR of developing MACEs and cardiovascular toxicity. No significant differences between the different subgroups were observed. Notably, even with the caveat of the small number of events for these analyses, when a PARPi was combined with an ARSi there were a significant increase of risk of any-grade MACEs and thromboembolic events compared with control and also an increased risk of high-grade MACEs (OR 2.08, P = 0.04). Of note, no increased thromboembolic events were registered with PARPi + antiangiogenic combinations.

Based on these results we can confirm the manageability of this drug class, which does not need strict routine cardiac monitoring in asymptomatic patients. A better investigation on cardiac events in patients treated with PARPi + ARSi combinations is highly warranted.^47,48

The pathogenesis of cardiac toxicity is not completely understood, but seems to be mainly associated with the inhibition of functional proteins other than the PARP targets (off-target effects). Therefore, each PARP inhibitor can display a peculiar toxicity profile (that can differ substantially from that of the others) related to the different binding affinity, trapping, and inhibition of both the on-target PARP proteins and the off-target ones.⁷ In the literature, niraparib is the PARPi most frequently associated with cardiovascular toxicity (i.e. tachycardia, palpitations) and hypertension.^35,49 In particular, one of the most important mechanisms might be the off-target interference on dopamine and norepinephrine metabolism caused by PARPi (especially niraparib).^6,7 Niraparib indeed inhibits the intracellular uptake of dopamine and norepinephrine by the off-target block of serotonin transporter (SERT), dopamine transporter (DAT), and norepinephrine transporter (NET), and by the inhibition of dual-specificity tyrosine phosphorylation-regulated kinase 1A (DYRK1A), ultimately resulting in high blood pressure due to cardiac inotropic effects of these neurotransmitters.^6,7,50 Nevertheless, several preclinical studies have hypothesized a cardioprotective role of PARPi due to the detrimental effect of PARP1 activation on myocardial tissue as a response to ischaemia/reperfusion injury, however without any conclusive evidence in clinical practice.51, 52, 53, 54

There are also conflicting data with regard to the pro-arrhythmogenic effects of PARPi. A study found niraparib not to be associated with clinically relevant alterations of ECG parameters, including QTc prolongation.⁵⁵ On the contrary, in another analysis there was a correlation between niraparib and QT prolongation resulting in potential severe arrhythmia due to the drug off-target inhibition of the Kv11.1 (hERG) potassium ion channel.⁵⁶ We have also carried out a subgroup analysis for each cardiovascular event according to the type of PARP inhibitor in order to evaluate whether each different agent (niraparib in particular) could be associated with a peculiar toxicity, but we found no significant differences (data not reported). Similarly, for every toxicity a subgroup analysis based on the type of cancer (including the most represented ones—ovarian, breast, and prostate tumors) was carried out, without any significant differences between the subgroups (data not reported), with the caveat of the diverse types of PARPi-based combinations used in the different trials for each tumor type.

Actually, patients enrolled in clinical trials are usually highly selected and undergo basal cardiac function screening tests, therefore with a potential lower risk of MACEs compared with the ‘real-world’ population, commonly with more comorbidities. Data from FAERS (one of the largest databases of AEs that supports the Food and Drug Administration post-marketing safety surveillance program of approved drugs) about AEs related to PARPi identified several cardiac and vascular disorders (accordingly to the literature, mainly related to niraparib), including increased blood pressure (N = 696; reporting odds ratios, ROR = 17.81), increased heart rate (N = 365; ROR = 13.15), hypertension (N = 343; ROR = 5.49), and nonspecific tachyarrhythmia/palpitations (N = 183; ROR = 4.87) as the most frequent ones.⁷

Several limitations impair the results of our analysis. First of all, the inability to access raw data and the lack of available public data about the single cardiovascular events for individual patients (with the possibility of a concomitant occurrence of two or more cardiovascular AEs in a single patient and therefore with the consequent potential overestimation of the incidence of MACEs and thromboembolic events) represented the main limit; however, given the low incidence of each single event we considered the sum of the AEs and used the Peto method for limiting this potential bias. Moreover, the definition of MACEs included different events that could not be standardized over the trials included. Finally, some important clinical information (i.e. baseline risk factors for cardiovascular events, patient’s comorbidities, metabolic syndrome, age, body mass index) and other possible data that might increase the risk of cardiovascular toxicity were not available, but they have to be taken into account in daily clinical practice. Likewise, the type of cancer and tumor burden, together with the previous treatment lines, could have influenced the risk of developing these toxicities (i.e. prior anthracycline-based chemotherapy for breast cancer patients, the intrinsic elevated thromboembolic risk of pancreatic tumors), and have to be considered in the monitoring and management of these side-effects.

In conclusion, this study confirms that PARPi-based therapies are responsible for an increased risk of any-grade MACEs, hypertension, thromboembolic events, and pulmonary embolism, albeit with a low absolute incidence of events (with the exception of hypertension) mainly mild in severity, with the lack of a significant association with events of high grade. Further research and individual patient data are wanted to improve understanding and identify patient-specific risk factors and susceptibility to cardiac and vascular toxicity.

Funding

None declared.

Disclosure

AP: speakers’ and travels’ fee from AstraZeneca-Merck Sharp & Dohme (MSD), Novartis, Amgen, Gilead.

CC: speakers’ and consultant’s fee from IPSEN, MSD, Astellas, AstraZeneca, Pfizer, Merck.

RI: advisory board member for Pfizer, Janssen, Sanofi, Ipsen, MSD, Novartis.

LS: speakers’ and consultant’s fee from MSD, AstraZeneca, Servier, Bayer, Merck, Amgen, Pierre-Fabre.

EB is supported by the Italian Association for Cancer Research AIRC-IG 20583; EB was supported by the International Association for Lung Cancer (IASLC), the LILT (Lega Italiana per la Lotta contro i Tumori) and Fondazione Cariverona. EB received speakers’ and travels’ fee from MSD, AstraZeneca, Pfizer, Helsinn, Eli-Lilly, Bristol Myers Squibb (BMS), Novartis, Roche. EB received consultant’s fee from Roche, Pfizer. EB received institutional research grants from AstraZeneca, Roche.

GT: advisory board member for BMS and Novartis.

All other authors have declared no conflicts of interest.