Chronic Myeloid Leukaemia Patients Treated In Teaching Hospitals Seem To Have a Survival Advantage Compared To Other Settings

Treatment of chronic myeloid leukaemia (CML) has changed considerably due to the advent of tyrosine kinase inhibitors. As a consequence, CML is nowadays increasingly treated in municipal hospitals (MH) and at office-based physicians (OBP) and not only in teaching hospitals (TH). It is unknown how this affects the outcomes of CML patients. Thus we analysed if patients at TH have a survival advantage in comparison to patients that are treated in MH or at OBP.

All patients with chronic phase Ph+ CML and admitted to the German CML Study IV [Hehlmann et al. JCO 2013, in press] were eligible for this analysis. Out of 1551 patients randomised, 13 did not match the inclusion criteria, 2 withdrew informed consent during the first days and 45 could not be considered due to missing covariates. Thus, 1491 patients were available for analysis and had complete covariates. Every study centre was classified into one of three categories: TH, MH and OBP. Additionally, the number of patients enrolled in the CML Study IV was taken as a proxy measure of experience with TKI-treatment.

Survival times were calculated starting with the date of diagnosis. Patients that were still alive were censored at the date of last observation. Cox models were estimated to assess the impact of centre type and experience with CML. The models were adjusted for the following covariates: risk category according to the EUTOS score, year of diagnosis, age at diagnosis, and Karnofsky performance status scale.

Out of the 1491 patients, 532 patients (36%) were from TH, 618 (41%) from MH and 341 (23%) from OBP. Percentages of EUTOS high risk patients were fairly similar in the three groups (13% at TH, 12% at MH and 10% at OBP). Patients of OBP had a significantly better Karnofsky scale, while TH patients were younger than MH and OBP patients (median age at diagnosis: 50, 53 and respectively 54). Five-year overall survival was 92%, 89% and 88% in the groups of TH, MH and OBP patients, the median observation time was 5.6 years. In the multivariate Cox model for survival, TH patients had a lower risk of death than MH patients (Hazard ratio (HR): 0.633, 95%-Confidence Interval (CI): [0.414; 0.966], p=0.034) and OBP patients (HR: 0.609, 95%-CI: [0.363; 1.024], p=0.060). Other important risk factors were EUTOS high risk (HR: 1.854, 95%-CI: [1.190; 2.889]) and age (HR: 1.044, 95%-CI: [1.030; 1.058]), whereas experience with CML and year of diagnosis did not seem to have any influence. When the model was stratified according to treatment, only minor changes were observed. One possible explanation for the advantage of the TH patients may be the more successful treatment of blast crisis. Two years after blast crisis, survival probability for TH patients was 47.7% (95%-CI: 28.4-67.4%), while for the MH and the OBP patients it was 22.3% (8.9-39.7%) and respectively 25.0% (7.6-48.3%). On the one hand, we found a tendency for more frequent adverse events (AE) in TH patients, but on the other hand reporting at TH was slightly better with an average of 3.5 AE forms per year compared to 3.3 (MH) and 2.8 (OBP).

Our data show that there was a survival advantage for CML patients treated initially at a TH compared to those that were treated at MH and OBP. This finding even holds, when adjusting for age, Karnofsky scale, EUTOS score, experience with CML, and year of diagnosis. We could not find any hint that more experience with the treatment of CML patients led to better survival probabilities. As the data set was too small to be divided into a learning and a validation set and this analysis was not pre-specified, our results need to be confirmed by an independent data set.

Hehlmann:Novartis: Research Funding; BMS: Consultancy.