Background: Bone marrow examination (BME) is frequently performed for unexplained cytopenias, with wide variations in clinical threshold across centres. Careful patient assessment and clinical judgment are essential to avoid over-use of BME, which may impose burden on patients, clinicians and the healthcare system. There is scarce data examining the clinical significance of abnormal BME findings, and whether they would change clinical management.

Aims: 1) To examine the diagnostic yield of BMEs for cytopenias, and predictors of positive findings 2) To evaluate the proportion of diagnostic marrows that led to changes in clinical management.

Methods: In this retrospective cohort study, we identified consecutive BMEs evaluated by the Department of Pathology at the University of Alberta Hospital (2022-2023), a tertiary centre in Northern Alberta (population >2 million). Only diagnostic marrows for undifferentiated cases were included. BMEs procured for staging, remission assessment, and confirmation of suspected hematologic malignancy (e.g. circulating blasts, JAK2 or BCR-ABL1 positive) were excluded. We collected clinical and laboratory characteristics, indication for BME, factors associated with cytopenia (e.g. liver cirrhosis, chronic kidney disease [CKD], chemotherapy, immunosuppressive therapy), and clinical management. Two hematologists independently reviewed each case, and adjudicated whether the BME changed management. Logistic regression was used to assess predictors of abnormal BME. This study was approved by the University of Alberta Health Research Ethics Board.

Results:

Of 544 BMEs reviewed, 223 (41%) were first-time BMEs. Indications for first-time marrows included: undifferentiated (107; 48%), suspected hematologic malignancy (106; 48%), staging or remission assessment (11; 5%). Among 107 undifferentiated cases, 94 (88%) BME were performed in outpatient settings. Patients experienced prolonged cytopenias or cytoses prior to BME (median 2.0 years, interquartile range [IQR] 0.3-5.8), with a short interval from hematology consult to BME (median 1 month, IQR 0-4). The most common indications were: bi- or pancytopenia (50; 47%), normocytic or macrocytic anemia (28; 26%), elevated counts with negative or unknown JAK2 or BCR-ABL1 mutation (21; 20%), lymphadenopathy or splenomegaly (14; 13%), monocytosis (11; 10%), constitutional symptoms or fever of unknown origin (6; 5%), isolated neutropenia (5; 5%), and autoimmune cytopenia (5; 5%). Among the 79 cases with cytopenia and/or macrocytosis, concomitant factors were common, including systemic autoimmune disorders (16; 20%), stage 3-5 CKD (9; 11%), active cancer (8; 10%), and immunosuppressive therapy (7; 9%). Hematologic profile showed a median hemoglobin of 100 g/L (IQR 86-122), MCV 99 fL (88-103), neutrophil 2.0 x 109/L (1.0-3.4), platelet 110 x 109/L (51-181), reticulocyte count 67 x 109/L (48-90).

The overall diagnostic yield of BMEs was 56 (52%), including 37 (35%) myeloid neoplasm, 7 (7%) lymphoid neoplasm, 7 (7%) clonal hematopoiesis, and 5 (5%) others. The rest of BMEs were either non-diagnostic (21; 20%) or normal (30; 28%). Factors associated with abnormal BME included macrocytic anemia as the indication for BME (odds ratio [OR] 7.4, 95% CI 1.9-49.1), monocytosis as indication (OR 4.7, 95% 1.1-31.8), whereas BME performed for autoimmune cytopenia were more associated with normal or non-diagnostic findings (p=0.02).

Only 46 (43%) BMEs were felt to have changed clinical management. These included: 15 (14%) started on hypomethylating agents, 8 (7%) hydroxyurea, 5 (5%) erythropoietin stimulating agents and 19 (18%) started on other therapies. The majority had no management changes after BME: 47 (44%) were reassured about absence of hematologic neoplasm, 9 (8%) were continued on supportive transfusions for lower-risk myeloid neoplasm, and 3 (3%) deemed poor candidates for chemotherapy.

Conclusion: BME is frequently performed for single- or multi-lineage cytopenias, including cases with mild cytopenias and those with other contributing factors for cytopenia. Our practice audit in a large Canadian academic centre showed a diagnostic yield of only 52% in first-time marrows performed for undifferentiated cases, and even a smaller proportion resulting in changes in management. Longer follow-up period is warranted to assess the impact of early detection of clonal hematopoiesis and lower-risk myeloid neoplasm on outcomes and health resources utilization.

This content is only available as a PDF.
2025
Sign in via your Institution