The Prevalence of Fanconi Anemia Among Patients with De Novo Acute Myelogenous Leukemia

Abstract 2232

The diagnosis of acute myeloid leukemia (AML) was the first presentation in approximately one-third of published cases of Fanconi anemia (FA) that led to the diagnosis of FA. Our objective therefore was to estimate the number of undiagnosed patients with Fanconi anemia (FA) who may present with acute myelogenous leukemia (AML), in order to justify a universal screening strategy. The probability of having FA among patients with AML, using Bayes rule, was calculated as: P(FAüAML) = [P(AMLüFA) × P(FA)]/P(AML). Data from the Surveillance, Epidemiology and End Results (SEER) Program, published FA cohort studies, and the Fanconi Anemia Research Fund were used to estimate these probabilities. P(FAüAML) was then applied to data on the time to recovery of an absolute neutrophil count (ANC) above 1000/μL from 892 de novo AML patients enrolled on Children's Oncology Group treatment protocol CCG-2961 between 1997 and 2002. A derived, exact binomial distribution of a one-sided Fisher hypothesis test was used to estimate the number of patients with FA among newly diagnosed AML patients required to reject the null hypothesis stating that the prevalence of FA among patients newly diagnosed with AML is less than an arbitrarily chosen threshold of 1%. From the published FA cohort studies, we estimated P(AMLüFA) in patients less than 19 years old at 0.5%/yr. Assuming an FA allele frequency of 1/300, autosomal recessive inheritance and an 80% FA survival rate until age of 18 years, P(FA) in the US population less than 19 years of age was estimated to be 5 × 10^-5. SEER data were used to estimate the age-adjusted P(AML) in persons 0–19 yrs at 0.73/100,000. The estimated probability that a newly-diagnosed AML patient has FA is therefore ∼3%. In CCG-2961, ANC recovery >1000/ μL after the first round of AML induction chemotherapy occurred in a mean of 40 +/− 9.6 days. Eighteen percent (165/892) of the patients never achieved an ANC >1000 and 22% (193/892) had no ANC recovery at 60 days (2 SD from the mean). Based on our estimate of 3% FA patients among de novo AML cases 27 FA patients (3% of 892) could represent at least 14% (27/193) of patients who had no ANC recovery at 60 days. Using the exact binomial distribution we estimate that a sample size of 375 newly diagnosed AML patients would have 80% power and a type I error rate of <0.05 to reject the null hypothesis that the prevalence of FA is less than 1% among those 375 AML patients. For this test, the critical value is 8 FA cases in 375 AML patients. In conclusion, we suggest that at least 3% of newly-diagnosed AML patients may, in fact, have FA. Identification of 8 or more FA cases among 375 newly diagnosed patients with AML would justify a universal screening strategy. FA screening in the evaluation of patients with newly-diagnosed AML would require the development of rapid FA screening techniques. This would permit timely adjustment of the chemotherapy regimen in order to minimize FA-related chemotoxicities. Finally, the remaining of those patients who never achieved ANC recovery might be accounted for by the toxicity of the AML treatment due to polymorphisms in genes involved in the metabolism of chemotherapy or to other, undiagnosed, AML-prone genetic syndromes.