Incidence of myelodysplastic syndrome and acute myeloid leukemia in patients receiving poly-ADP ribose polymerase inhibitors for the treatment of solid tumors: A meta-analysis of randomized trials

Abstract

Background: Clinical trials demonstrated that PARPi (poly [adenosine diphosphate-ribose]-ADP polymerase inhibitor) therapy is effective in solid tumors. However, long term effects such as therapy-related myelodysplastic syndrome or acute myeloid leukemia (MDS/AML) are poorly described. We sought to quantify whether PARPi therapy is associated with the development of MDS/AML.

Methods: Medline, Embase, and Cochrane databases were searched (inception to January 6, 2020) and phase 2 and 3 clinical trials that randomized patients with solid tumors to a PARPi or control therapy were included. The PRISMA guidelines were used to extract data independently by multiple authors. We extracted person-time and number of cases of MDS/AML in the PARPi and control arms of each study and pooled results with a random-effects Poisson regression model. The pooled incidence rate ratio (IRR) for MDS/AML among patients randomized to PARPi therapy was compared to those randomized to a control.

Results: We identified 14 studies that included 5739 patients. Accounting for intra-study clustering, the risk of MDS/AML was similar in patients who were randomly assigned to receive PARPi compared to controls (IRR 1.32, 95% confidence interval [CI] 0.78-2.26). In the front-line setting, PARPi therapy was associated with developing MDS/AML (IRR 5.43, 95% CI 1.51-19.60). Among patients treated for recurrence, however, the risk of MDS/AML appeared to be similar among patients randomized to PARPi or control treatment. Among studies that included only patients with a BRCA mutation, the risk of MDS/AML was similar in both treatment groups (IRR 0.83, 95% CI 0.45-1.53), but PARPi therapy was associated with MDS/AML in studies with an unrestricted population (IRR 2.43, 95% CI 1.17-5.06).

Conclusion: The pooled overall effect was not statistically significant. However, treatment with PARPi was associated with a statistically significant increase in the incidence of MDS/AML among patients receiving front-line cancer therapy and those with limited prior exposure to chemotherapy.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Figure 2.

Forest Plot of Study Results Studies with zero-counts in at least one study arm have inestimable study-level effect sizes and cannot be visualized on a forest plot. All studies, however, contributed to the overall estimate. The box size corresponds to the weight of the study in the meta-analysis. As this is a count-based model, weights are defined by number of events per study. The diamond depicts the point estimate (95% CI) of the pooled estimate. The vertical black line is centered at the null, whereas the dashed red line is centered at the pooled IRR estimate. Abbreviations: IRR, Incidence rate ratio; PARPi, poly [adenosine diphosphate–ribose]-ADP polymerase inhibitor; CI, confidence interval.