Assessing bone marrow biopsy utility in patients with unexplained cytopenias

Abstract Background : In patients with cytopenias not readily explained by non-invasive tests (unexplained cytopenias), bone marrow biopsies may be done to help identify the cause. In some cases, bone marrow biopsies do not lead to a new diagnosis or change in treatment. Methods : We conducted a single-center retrospective review to identify patients who had unexplained cytopenias who underwent bone marrow biopsies, and collected demographic, clinical and lab data as well as their final clinical diagnosis and management. We excluded patients who had blasts detected on peripheral smear and patients undergoing bone marrow biopsies for workup for metastatic malignancy, monoclonal gammopathy, or amyloidosis as they did not fall into our definition of unexplained cytopenia. A logistic regression model was used to identify which predictors were associated with a bone marrow biopsy leading to a new diagnosis or change in treatment. Data was split into 80% for a training dataset (to train the model) and a 20% dataset to test the model. Results : A total of 170 patients were included in the analysis, with mean age 64.9 years. The most common final diagnoses following bone marrow biopsy were idiopathic cytopenia or unclear etiology (16%), MDS without excess blasts or MDS with ringed sideroblasts (14.28%), cytopenia due to chronic illness, kidney or liver disease (11.61%), drug induced cytopenia (6.25%), acute leukemia or MDS with excess blasts (5.36%), and cytopenia secondary to autoimmune disease (5.36%). Bone marrow biopsy led to a change in diagnosis and a change in treatment in 42% and 30.4% of patients respectively. On multivariable logistic regression analysis, smoking status, cirrhosis, history of chemotherapy, hemoglobin level, abnormal smear, GFR, and LDH were statistically significant predictors associated with bone marrow biopsies leading to new diagnosis or change in treatment. Conclusion: Bone marrow biopsy is a useful but invasive tool for investigating unexplained cytopenias. In some patients, bone marrow biopsies may not be needed and clinical prediction models using routinely available data can be created to identify which patients may not require bone marrow biopsy.