Separate analysis will be carried out for the per-protocol (PP) population and intent-to-treat (ITT) population. For the PP population, a repeated measures analysis of variance will be employed. In this model, FACT-G scores are the response variables. Time, patient group and their interaction are the predictors. For the ITT population, a random effect model will be fitted to the data. The model can also adjust for covariates of potential interests. Exploratory assessments may also include effects of other variables of interest, such as demographic and baseline values.