After the initiation of the study, the sponsor became aware of an error in the drug dispensing module of the interactive voice response such that most patients who were randomized prior to 07 November 2007 were dispensed the incorrect study drug at least once. Since the integrity of the data from these patients was seriously compromised, these patients were considered unevaluable for the protocol-specified analyses. These participants are referred to as the breached protocol cohort (BPC) and those still on study treatment at the time of the breach were offered the option of receiving open-label erlotinib for up to 2 years (including posttreatment and long-term follow-up assessments), not receiving open-label erlotinib but remaining in the study for posttreatment and long-term follow-up assessment, or withdrawing consent from treatment and further assessments. Participants who had discontinued study treatment prior to the breach were not offered open-label erlotinib and remained in long-term follow-up. Data from the BPC participants were analyzed separately and were not included in the assessments of primary or secondary endpoints in the randomized cohort and were not considered part of the primary analyses.The sample size for the randomized cohort was not changed due to the BPC and the data from RC and BPC were analyzed separately.
