Abstract 3550


Older patients diagnosed with acute myeloid leukemia (AML) present the clinician with particularly difficult treatment decisions. A number of attempts have been made to develop algorithms to assist clinician decision making but these are not widely used. The United Kingdom Medical Research Council AML14 Trial failed to recruit to a randomization between intensive and non-intensive treatment approaches, suggesting that clinicians preferred to use clinical judgement when selecting patients for intensive chemotherapy despite the lack of evidence on how this should be done.


Several published scoring systems were evaluated and two applied retrospectively to 54 consecutive patients older than 60 years old presenting with AML between January 2009 and December 2010 to three collaborating Hematology departments. In addition, the role of the peripheral blast count (less than versus equal to or more than 0.1×109/L), as a marker of degree of proliferation, in predicting outcome was assessed. Treatment decisions for all patients had been based on clinical assessment by a senior physician and discussion by a multidisciplinary team rather than on application of a scoring system. Patients received palliative care only (2), best supportive care (16), non-intensive (9) and intensive (27) chemotherapy.


Median overall survival was 5.5 months. A significant difference in survival (12.0 versus 3.2 months, p<0.005 log-rank test) was seen, irrespective of treatment received, between those with low (<6.9, equivalent to predicted treatment related mortality (TRM) of 3%) versus high (>6.9, equivalent to predicted TRM of 20%) scores using the AML Treatment-Related Mortality (AML-TRM) score (Walter RB et al, 2011, J.Clin Oncol. 29:4417), which was developed to predict risk of early death following induction chemotherapy in older patients. Despite the use of clinical assessment only, 65% of low risk patients by AML-TRM had received intensive chemotherapy; 70% of high risk patients had not received intensive chemotherapy. By contrast, the Hematopoetic cell transplantation - specific comorbidity index (HCT-CI) (Sorror ML et al, 2005, Blood 106:2912), developed to assess fitness for transplantation in younger patients, did not predict outcome in this patient group.

A peripheral blast count of less than versus equal to or more than 0.1×109/L was found to predict outcome for patients who received best supportive care (Table 1).

Table 1
Intensive Chemotherapy: Median Survival (months)Best Supportive Care: Median Survival (months)P value (log-rank test)
Peripheral blast count <0.1×109/L 6.2 15.9 0.684 
Peripheral blast count >0.1×109/L 13.2 1.3 0.0015 
Intensive Chemotherapy: Median Survival (months)Best Supportive Care: Median Survival (months)P value (log-rank test)
Peripheral blast count <0.1×109/L 6.2 15.9 0.684 
Peripheral blast count >0.1×109/L 13.2 1.3 0.0015 

Four patients with peripheral blast counts equal to or more than 0.1×109/L who received non-intensive chemotherapy also showed poor survival (median 2.0 months).


The previously validated AML-TRM score is a strong predictor of outcome in older patients with AML irrespective of treatment received. However, it is not clearly better than a treatment decision based on clinical assessment by an experienced physician. The peripheral blast count was a single parameter which could be used to influence treatment decisions. In this series, patients with peripheral blast counts less than 0.1×109/L, correlating with less proliferative disease, did at least as well with best supportive care as with intensive chemotherapy. These patients should be offered best supportive care unless a clear curative plan, including for example an allogeneic stem cell transplant after intensive chemotherapy, can be made. However, patients with a peripheral blast count equal to or more than 0.1×109/L did very badly with best supportive care only or with non-intensive chemotherapy and should be offered intensive chemotherapy where possible.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.