Sample size estimation using repeated measurements on biomarkers as outcomes