Nomograms to predict serious adverse events (SAEs) in patients (pts) enrolled in phase II clinical trials of molecularly targeted agents (MTAs)
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
6601 Background: The likelihood of experiencing a SAE in clinical trials with a MTA is of interest for clinicians discussing treatment options. Adverse event data from clinical trials in the Princess Margaret Hospital Phase II Consortium [PMH2C] database were analyzed to address this question. Methods: All pts in the PMH2C database treated at the phase II dose level with either a MTA alone or in combination regimens since 2001 were included. Generalised estimating equations were used to construct optimal regression models predicting the increased/decreased odds of a SAE of all causalities (defined as a grade 3+ non-hematologic adverse event, or a grade 4+ hematologic adverse event) during the first cycle of treatment relative to a ‘reference’ pt. Nomograms were constructed to ease interpretation and internal validation explored using bootstrapping on trials larger than 35 pts. Results: 576 pts (median age=60, 55% male, ECOG PS 0:1:2=259:284:35) were accrued to 42 studies. In order of statistical significance, higher ECOG PS, increased LDH, decreased albumin, increased Charlson score, increased number of target lesions, not having prior radiotherapy and decreased age were predictive of increased odds of cycle 1 SAE. As an example, a 56-year old patient with ECOG 2, Charlson score=0, 5 target lesions, LDH=1.70x upper limit of normal [ULN], albumin=0.84xULN and no prior radiation would have ∼3 times increased odds of a SAE in cycle 1, compared to a 63-year old with ECOG 1, Charlson score=0, 1 target lesion, LDH=0.76xULN, albumin=0.68xULN and no prior radiation. Internal validation of the 4 largest studies indicated moderate-good accuracy (estimated area under the receiver operating characteristic curve = 0.57–0.86). Conclusions: A nomogram was produced allowing estimation of the increased odds of a SAE during cycle 1 of therapy in a phase II trial setting. Actual risk can then be further estimated by incorporating clinical judgment of risks for an average pt when given a particular MTA. This nomogram can potentially improve patient knowledge, risk estimates and the decision-making process. External validation of the model is still necessary to adequately assess model reliability. No significant financial relationships to disclose.
