Impact of Dose Intensity of Ponatinib on Selected Adverse Events: Multivariate Analyses from a Pooled Population of Clinical Trial Patients

Background: Ponatinib is an approved potent pan–BCR-ABL tyrosine kinase inhibitor (TKI) with proven efficacy in refractory Ph+ leukemia, particularly CP-CML. Previous analyses suggest significant association between ponatinib dose intensity and adverse events (AEs). We explored this further using pooled data from phase 1, 2, and 3 ponatinib trials in refractory Ph+ leukemia or newly diagnosed CP-CML.

Methods: Patients received ponatinib (range 2-60 mg/d) in all 3 trials; dose modification was allowed to manage AEs. Univariate and multivariate logistic regression analyses were conducted for each outcome (presence/absence of treatment-emergent AEs) and dose intensity measure (average daily dose through day of event) in 671 patients with all baseline covariates (Table 1) measured. AEs analyzed were vascular occlusive events; cardiac failure; myelosuppression; pancreatitis; increased lipase, ALT, AST; rash, arthralgia; hypertriglyceridemia; hypertension. Time to event modeling was performed for vascular occlusive AEs to find the optimal exposure window in proximity to events.

Results: As of Jan 6, 2014, 683 patients received ponatinib. Median treatment duration: 224 days; median dose intensity: 37.2 mg/d. Notable differences were observed in distribution of covariates for vascular occlusive AEs by dose intensity quartile (Table 1): more frequent histories of diabetes and ischemic disease, older age, and longer time since diagnosis in lower 2 quartiles; lowest baseline platelet and neutrophil counts in highest quartile. This suggests the importance of adjusting for these covariates when evaluating impact of dose on AEs. After adjusting for covariates, there were significant associations in logistic regression analyses between dose intensity and most AEs (Table 2). Strongest associations (OR >2) were seen for pancreatitis, rash, and cardiac failure.

Logistic regression analyses do not factor in event timing or follow up duration and use average dose intensity. Thus, time to event analyses using daily doses were performed for vascular occlusive AEs to find the exposure window with highest increase in risk in proximity to these AEs. These models show highly significant associations between dose intensity and event risk for all event types except venous thromboembolism. This also allows exploration of the relationship between timing of dose change and event occurrence. Generally, lags of ~≤30 days were most predictive, although in some models there was a higher risk increase with lags of ~6 months—suggesting that there may be up to a 6-month delay following dose reduction before a decrease in risk manifests for some arterial thrombotic events.

Conclusions: These data in a pooled population with longer follow-up replicate previous findings that ponatinib dose intensity is highly associated with many AEs, and support investigating approaches to lower average dose intensity, such as starting at lower doses and/or reducing dose on response. A dose-ranging trial of ponatinib in refractory CML to evaluate benefit/risk of different dosing schemes is being planned.