Evaluation of Clinical Judgement at Initiation of AML-Therapy in Elderly Patients - A Prospective Randomized Trial.

Ulrich S. Schuler, Mathias Hanel, Heinrich Bodenstein, Hermann Einsele, Karl-Heinz Pflueger, Norbert Schmitz, Thomas Wagner, Hannes Wandt, Regina Herbst, Ulrich Kaiser and Gerhard Ehninger


Treatment decisions for AML in elderly patients are based on a more individualized approach than in younger patients. Increasing age is associated with both decreasing efficacy of conventional therapy, increasing influence of comorbidity and functional impairment on the overall outcome. In a prospective trial we tried to answer two questions. (A) Comparison of DA and I-MA -Induction (B) prospective evaluation of the clinical judgement of the haematologist concerning the overall prognosis of the patient. Physicians requesting the randomization by FAX had to give their impression on a 5 point Likert-scale, whether they regarded the overall prognosis of the patient as inferior, comparable or superior to the respective age group, prior to the first dose of chemotherapy. In part (A) Patients >60 years were randomized between DA (Daunorubicin 45mg/m2 d3−d5 +Ara-C 100mg/m2 d1–d7) and I-MA- induction therapy. In a single arm-study, it had been claimed by others [ASH 2002 abstract 1337] that I-MA (Mitoxantrone 10 mg/m2 d1-3, Ara-C 1 g/m2 2x/d on d1,d3,d5, d7) in patients >60yrs results in high CR-Rates of 73% in de novo and in 59% in secondary AML. Given the expectation of a considerable superiority, several interim analyses were planned, in order to stop the trial early, if I-MA should be shown to improve CR-rates or survival. The most recent analysis is based on 206 patients (more than 60 days on study and evaluable for CR). Concerning part (A) neither CR-rates nor survival differed significantly (overall CR-rate 50%, trend favouring I-MA, survival trend after 1 year favouring DA). In part (B) physicians avoided extremes of the Likert scale. Only in the minority of cases cytogenetic data were available at the time of randomization, but judgement proved to be of value even if this variable was included in the Cox-model. This data show for the first time that clinical judgement (without formal comorbidity scoring or functional assessment) is of value in predicting the prognosis of elderly AML-patients treated uniformly within a randomized trial.

Physician estimate of prognosisFrequencyAverage Age (median) in yearsMedian Survival (months)p-value (log-rank)
inferior23%69,7 (69)4.9
comparable62.7%68,7 (69)7.40.124
superior14.3%68,5 (68,3)10.10,013