Occurrence of Pancytopenia Among Patients With Cancer Treated With Poly(Adenosine Diphosphate–Ribose) Polymerase Inhibitors

Abstract

This cross-sectional study uses pharmacovigilance data to investigate the occurrence of pancytopenia among patients with cancer treated with poly(adenosine diphosphate–ribose) polymerase inhibitors and examines the clinical features of pancytopenia related to these drugs.

Pre- and postmarketing reports on 2 poly(adenosine diphosphate–ribose) inhibitors (PARPis), niraparib and olaparib, indicate that these drugs induce pancytopenia adverse events (AEs).^1,2 We used World Health Organization (WHO) pharmacovigilance data (1) to investigate the association between PARPi class and the occurrence of pancytopenia among patients with cancer and (2) to determine the clinical features of PARPi-related pancytopenia.

Methods

For this cross-sectional study (ClinicalTrials.gov identifier: NCT04774627), we used VigiBase, the WHO global pharmacovigilance database of individual case safety reports. We used search dates from the first report of a PARPi-related AE in 2009 to March 24, 2021. First, we performed a case-noncase study using the reporting odds ratio (ROR) to compare observed and expected drug-AE associations when using the full database as the comparator.³ In our study, reports of PARPi-related pancytopenia were considered cases and all other reports were considered noncases. The PARPis studied were olaparib, rucaparib, niraparib, talazoparib, and veliparib. Demographic characteristics and data on pancytopenia features (eg, seriousness, latency period, and outcomes) and coreported AEs were also collected.⁴ The median latency period (in months) was considered the interval between a patient with cancer starting PARPi treatment and then being diagnosed with pancytopenia. The proportion of fatalities was calculated as the number of reported deaths divided by the number of total available reports. A significant signal was detected if the lower limit of the Bonferroni-corrected 95% CI of the ROR (ROR₀₂₅) exceeded 1. Sensitivity analyses were performed by recalculating the ROR₀₂₅ value using the protein kinase inhibitor group (L01E) as the comparator.

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The use of confidential, electronically processed patient data was approved by the Caen University Hospital Research Ethics Committee. No informed consent was required because this was an anonymized retrospective analysis of data from the WHO pharmacovigilance database.

Results

Overall, 23 305 AEs were reported in VigiBase for the 5 PARPis, and the total number of AEs with any drug(s) was 20 269 537. Pancytopenia was significantly associated with PARPi class, with 201 reports (ROR, 5.5; 95% CI, 4.6-6.7) in the database. Moreover, this association remained significant for each PARPi studied, with 16 reports for talazoparib (ROR, 17.4; 95% CI, 8.9-34.0), 102 for niraparib (ROR, 6.8; 95% CI, 5.3-8.9), 57 for olaparib (ROR, 5.3; 95% CI, 3.7-7.6), 5 for veliparib (ROR, 8.2; 95% CI, 2.5-27.1), and 21 for rucaparib (ROR, 2.2; 95% CI, 1.2-4.0). All ROR₀₂₅ values were subjected to Bonferroni correction (α = .05 for 6 analyses), and the results of the sensitivity analyses were consistent with the primary finding. There were 194 reports (97%) of serious pancytopenia, and the median latency period was 1.6 months (IQR, 0.7-11.0 months) from the first PARPi exposure (Table). Among frequently coreported AEs, 18 of 146 patients (12%) were diagnosed with myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML). Finally, there were 117 reports with known outcomes: 69 patients (59%) had recovered or were recovering, 30 (26%) had not recovered, and 18 (15%) died.

Table. Characteristics of Patients With Cancer Treated With a PARPi and Diagnosed With Pancytopenia, Using Pharmacovigilance Data From the World Health Organization VigiBase^a.

Characteristic	No. (%) of pancytopenia reports
Serious cases (n = 201)
Yesb	194 (97)
No	7 (3)
Sex (n = 165)
Female	153 (93)
Male	12 (7)
Age at onset, y (n = 119)
Median (IQR)	65 (57-70)
Range	15-86
Indications, cancer (n = 169)
Ovarian	132 (78)
Breast	14 (8)
Othersc	23 (14)
Latency period, mo (n = 48)
Median (IQR)	1.6 (0.7-11.0)
Range	0.2-29.9
Duration of pancytopenia, wk (n = 20)
Median (IQR)	1.5 (0.9-3.1)
Range	0.4-9.0
Treatment management (n = 157)
Discontinuation	113 (72)
Reduction	27 (17)
Dose not changed	17 (11)
Outcomes (n = 117)
Recovered/recovering	69 (59)
Not recoveredd	30 (26)
Deathse	18 (15)
Coreporting of adverse events (n = 146)
Fatigue	43 (29)
Nausea	29 (20)
Constipation	17 (12)
MDS or AMLf	18 (12)
Vomiting	15 (10)

Abbreviations: AML, acute myeloid leukemia; MDS, myelodysplastic syndrome; PARPi, poly(adenosine diphosphate–ribose) polymerase inhibitor.

^a Data were collected on March 24, 2021. The number of reports with total available data appears in parentheses in the first row for each characteristic.

^b Leading to death, hospitalization or prolongation of hospital stay, disability (persistent or significant), or life-threatening condition (from the Cancer Therapy Evaluation Program Adverse Event Reporting System).

^c Includes 7 with prostate cancer, 5 with glioma, 3 with lung cancer, 3 with ovarian and breast cancer, 2 with pancreatic cancer, 2 with sarcoma, and 1 with cervical and endometrial cancer.

^d Includes 8 of 30 reports with secondary MDS or AML.

^e Includes 7 of 18 reports with cause-specific deaths: 2 with neoplasm progression, 2 with AML, 1 with hemorrhage, 1 with kidney and respiratory failure, and 1 with sepsis.

^f Includes 12 with MDS and 6 with AML (includes 2 reports as progressing from MDS).

Discussion

To our knowledge, this work is the first to highlight a substantial association between PARPi class and an increased risk of reported pancytopenia. Our results indicate that half of the patients in this study had an early onset of pancytopenia within the first 2 months after the initiation of PARPi treatment for cancer, suggesting the need for enhanced blood surveillance. In placebo-controlled Study 19, 2 of 136 patients in the olaparib-only arm experienced pancytopenia and subsequently developed MDS or AML.^5,6 In line with our previous cohort,⁴ 8 of 30 patients in the current study who had not recovered from pancytopenia also subsequently developed MDS or AML, suggesting that pancytopenia could be a relevant safety signal for the early detection of potential PARPi-related MDS or AML. Moreover, the product information for niraparib and olaparib advises temporarily interrupting PARPi treatment in the case of severe hematologic toxicities, including pancytopenia.^1,2

We acknowledge limitations to this study. Although VigiBase provides a significant ROR₀₂₅ value, a causal association could not be formally ascertained on a case-by-case basis. Moreover, we could not compute the incidence of PARPi-related pancytopenia, because the exact number of patients exposed to PARPi was unknown and drug-induced pancytopenia AEs are not systematically reported to pharmacovigilance systems.

Clinicians should consider careful monthly blood monitoring for patients with PARPi-treated cancer to check for early pancytopenia onset. This surveillance should be followed by a dose interruption as soon as possible to avoid the permanent discontinuation of these drugs, which could lead to a cancer relapse.