Predictors of exposure, adverse and therapeutic effects of BRAF/MEK inhibitors in advanced melanoma.

Covariates associated with time to an adverse effect or survival time will be evaluated using Cox proportional hazards regression and reported primarily as hazard ratios with 95% confidence intervals. Covariates associated with response will be evaluated primarily by logistic regression and will be reported as odds ratios. Continuous variables (e.g. age, LDH) will be evaluated for non-linearity of association with outcomes by using restricted cubic splines. Crude associations will be reported based on univariate analysis, and adjusted associations based on a multivariable analysis. Multivariable analysis will generally adjust for all available baseline variables that are known or plausibly associated with the outcome of interest. Statistical interaction between predictors and specific drug treatment will be evaluated. It is recognised that (1) patients/clinicians will need to re-evaluate decisions on whether to start/stop/continue/recommence therapy at a number of time points over the course of therapy, and (2) the predictors that are available and most useful may change over the course of therapy. Time-dependent covariates will be principally evaluated as a time-dependent covariate with sensitivity analyses using a landmark approach to specifically evaluate whether the variable may serve as an early marker of the outcome. Very little research has been undertaken on covariates that may be predictive of therapeutic and adverse effects of these classes of drugs. Thus, the research proposed will explore a range of covariates that are plausibly associated with adverse and therapeutic outcomes based on studies undertaken in other forms of cancer, or clinical and biological plausibility. As these analyses are primarily hypothesis generating and will require subsequent validation of any findings, no formal adjustment for multiple testing is intended. However, this limitation will be clearly stated in any publications of results along with the importance of validation of any findings in independent cohorts. Covariates to be explored include:•   Baseline values. For time-varying variables, the baseline values are defined as the value closest and prior to the first dose of study treatment. o   Basic demographics - age, sex, race/ethnicity, region, weight/height/BMI, smoking status/historyo   Standard prognostic factors such as performance status, number, size and sites of metastases, serum lactate dehydrogenase (LDH), stage, V600 BRAF mutation status, prior therapy for advanced melanomao   Comorbidities, concomitant use of non-cancer medicines, laboratory values and patient reported outcomes that are plausibly related to the outcome being analyzed based on clinical opinion and literature review•   Post-baseline valueso   Adverse event grade (or underlying laboratory value) as potential predictors of tumor response, survival, exposure or other adverse eventso   Extent of drug exposure (based on measures of drug concentration or a surrogate) as potential predictor of tumor response, survival, or adverse eventso   Extent of tumor response/shrinkage, changes in circulating LDH, and changes in patient reported outcomes as early predictors of therapeutic outcomesMultiple imputation (n=20) will be undertaken for analyses in which there is > 5% missing data. If a sufficient number of useful predictors are identified, clinical prediction models will be developed and evaluated. The performance of any prediction models developed will evaluated in terms of both discrimination, using Harrell's c-index for the time-to-event, and calibration using graphic assessment and the Hosmer-Lemeshow ?2-tests for the time-to-event data. Should models with reasonable prediction performance result from this project, the intention will be to externally validate the prediction performance using a separate (non clinical trial) cohort [outside the scope of this proposal].Sample size calculation Many potential predictors of survival (e.g. early adverse events) occur in approximately 30%-50% of study participants. The number of OS events recorded during follow-up of clinical studies for BRAF inhibitors is approximately 500. The power to detect an association (e.g. early adverse event associated with improved survival) with a hazard ratio of 0.8 and alpha=0.05, is approximately 0.70.Quality control plansPrior to beginning analyses related to this project the individual variables required for the analysis will be extracted/derived based on the raw and analysis datasets provided. To ensure that each variable has been correctly extracted/derived from the data provided, basic analyses and descriptive statistics will be reproduced to check for consistency with pertinent results in published manuscripts or CSRs relating to the specific trial. Where there are insufficient published results to confirm the proper extraction of the variable, the values will be manually checked against a random sample of the original dataset values. All analyses will be coded and documented in an analysis script to facilitate transparency and reproducibility of analyses. All analysis code and results will be reviewed by a second biostatistician involved the project.