Predicting Therapy-Related Acute Myeloid Leukemia By Routine Laboratory Values: An Analysis Of 126 Patients With Breast Cancer Referred For Bone Marrow Biopsy At Memorial Sloan-Kettering Cancer Center

The incidence of therapy related AML (t-AML) after adjuvant treatment of breast cancer has been estimated at 0.5%-1% of all patients in retrospective studies. While the absolute incidence of t-AML is low, the high prevalence of breast cancer in the United States – 2.8 million Americans are currently living with a diagnosis of breast cancer (http://seer.cancer.gov/statfacts/html/breast.html) - makes the burden of t-AML significant.

Oncologists and hematologists are often confronted with a breast cancer patient who develops unexplained cytopenias. The differential diagnosis includes therapy-related myelodysplastic syndrome (t-MDS) or t-AML, chemotherapy effect, or breast cancer metastatic to the bone marrow. We evaluated factors that might inform whether a patient had t-MDS or t-AML to determine whether a bone marrow biopsy was indicated to aid diagnosis or whether a patient could be reassured and monitored with close outpatient follow up.

128 patients seen at Memorial Sloan-Kettering Cancer Center (MSKCC) who had a diagnosis of breast cancer, referred for a bone marrow biopsy between 2002 and 2012 were identified using the MSKCC institutional database and 126 patients were included in the analysis (2 excluded because of missing BM biopsy dates). Baseline characteristics of the patients are detailed in table 1. Patients were grouped based on bone marrow biopsy result: those who were found to have t-MDS or t-AML, those with breast cancer metastatic to the bone marrow, and those with a hypocellular marrow with no evidence of MDS, leukemia, or metastatic breast carcinoma. Logistic regression analysis was performed.

In univariate analysis, low absolute neutrophil count (ANC) < 1.5K was significantly associated with increased likelihood of t-MDS/t-AML versus metastatic breast cancer or no disease compared to those who had an ANC ≥1.5k (OR=9.8, 95%CI(3.9, 24.5), p<0.0001). Thrombocytopenia (<100k) was significantly associated with an increased likelihood of t-MDS/t-AML or metastatic breast cancer to bone marrow compared with no malignant disease.

By multivariable analysis, both ANC and thrombocytopenia remained significantly associated with an increased likelihood of t-MDS/t-AML after controlling for hemoglobin and the receipt of any anthracycline or alkylating agent prior to BM biopsy (OR=9.8, 95%CI(3.7, 25.5), p<0.0001 for ANC; OR=3.1, 95%CI(1.2, 8.2), p=0.02 for platelets).

In breast cancer patients with unexplained and persistent cytopenias, neutropenia <1.5K and thrombocytopenia <100K show a statistically significant correlation with subsequent diagnosis of t-AML or t-MDS on bone marrow biopsy. This finding is unique in that it gives the clinician the ability to correlate severity of cytopenias with the likely underlying etiology for these cytopenias. This can be used as a tool to determine the need for and timing of a bone marrow biopsy, help reassure patients and guide the practicing hematologist and breast oncologist.

No relevant conflicts of interest to declare.