An Observational Follow-up Study of 1st-Line Treatment With Herceptin (Trastuzumab) in Patients With Metastatic Breast Cancer (Post-HERMINE)

The Progression-Free Survival (PFS) was defined as the time between the treatment start date (date of the first trastuzumab infusion) and the date of the first disease progression or disease-related death. Participants who had not progressed at the end of the post-HERMINE study were censored at the last date they were known to be alive. Progression-Free Survival was estimated by using Kaplan-Meier method.

The Time to Progression (TTP) was defined as the time between the treatment start date (date of the first trastuzumab infusion) and the date of the first disease progression or disease-related death. Participants who had not progressed at the end of the post-HERMINE study were censored at the last date they were known to be alive. If the cause of death was unknown, the death was considered for this analysis as due to the disease. All participants who did not progress, the death was considered to be due to the disease.

Treatment duration was defined as the time between the first and the last infusion of trastuzumab. Exposure duration was defined only for the participants who continued trastuzumab after HERMINE study, as the sum of treatment duration as part of HERMINE study and of the treatment durations as part of post-HERMINE study taking into account temporary treatment discontinuations. For analyses of treatment and exposure duration , dates of infusion of trastuzumab were missing for 18 participants, so treatment and exposure durations were calculated for only 202 participants.

Search for prognostic factors for OS was performed using Cox regression model. First, all parameters were analyzed in univariate models, and the hypothesis of proportional risks was tested. Significant parameters at 15%-level were retained for the multivariate model. For the multivariate analysis, two models were built for prognostic factors for OS. In the first one (Model 1), a stepwise selection method was used on all parameters that were significant in the univariate analyses, whatever the significance level of the associations between parameters. In the second one (Model 2), a stepwise selection was used on the significant parameters that were not correlated. The significance level for entry was 10% and the significance level for removal was 5%. The variables n°5 and n°6 were found to be significantly associated, thus only the variable n°6 was tested in the Model 2. This variable was not retained by the stepwise selection in the Model 1, contrary to the variable n°5.

Prognostic factors for TTP were searched by using Cox regression model. First, all parameters were analysed in univariate models, and the hypothesis of proportional risks was tested. Significant parameters at 15% level were retained for multivariate model. For multivariate analyses, 2 models were built for prognostic factor of TTP. In Model 1, stepwise selection method was used on all parameters that were significant in the univariate analyses, whatever the significance level of the associations between parameters. In Model 2, stepwise selection was used on the significant parameters that were not correlated. The significance level for entry was 10% and the significance level for exit was 5%. Six parameters remained in the Model 1 to search for prognostic factors of TTP. Once the correlated variables were removed, there were only 3 variables left in Model 2: the age was not kept by the stepwise selection. Results below are for Model 1 and similar results were obtained for Model 2