The utility of bone marrow examinations for the investigation of patients with immune thrombocytopenia (ITP) is a matter of debate. We designed an agreement study to evaluate the inter-rater reliability of bone marrow biopsies and aspirates for the diagnosis of ITP.
Histological slides of bone marrow biopsies from patients with ITP and controls were prepared from stored tissue blocks and corresponding aspirate slides were retrieved for blinded, duplicate, independent pathological review. Patients had a diagnosis of primary ITP, were 18 years of age or older with a platelet count less than 100 ×109/L as measured within 4 weeks before the date of bone marrow biopsy procurement. Patients with splenomegaly, myelodysplastic syndrome, lymphoproliferative disease, HIV, hepatitis B or C, drug-induced thrombocytopenia or previous treatment with hematopoietic growth factors or thrombopoietin receptor agonists were excluded. Control bone marrow samples were selected from patients 18 years of age or older with stage I to III lymphoma or monoclonal gammopathy of undetermined significance who had a normal platelet count within 4 weeks before bone marrow biopsy procurement. Prior treatment with anti-neoplastic agents or hematopoietic growth factors were exclusions. Bone marrow slides from ITP patients and controls were coded and arranged in random order. A standardized pathological assessment form was used to capture bone marrow cellularity (in deciles), diagnosis [‘ITP’ or ‘within normal limits (wnl)’], megakaryocyte number (increased, decreased or wnl), morphology (abnormal or wnl), distribution (abnormal or wnl) and megakaryocyte assessment on aspirate (abnormal or wnl). The form was piloted by a third independent hematopathologist (n=10 bone marrow examinations) and revised prior to duplicate review. All reviewers were pathologists with at least 5 years experience reading bone marrows. Chance-corrected agreement between reviewers was calculated using kappa (k) and chance-independent agreement was calculated using phi (φ) with 95% confidence intervals (CI).
Bone marrow slides were prepared from 30 ITP patients and 53 controls. Agreement on marrow cellularity (to within 20%), which was used to ensure calibration of assessors, was good (k= 0.75, 95% CI. 0.65–0.85). Of 81 evaluable bone marrows, pathologists agreed on the diagnosis for 69 cases (85.2%); overall agreement was fair (φ = 0.52; 95% CI: 0.10–0.78). Of the 30 ITP bone marrows, pathologists correctly agreed on the diagnosis in 2 (6.7%) cases, incorrectly agreed on 20 (66.7%) and disagreed on 8 (26.7%). Of the 51 control bone marrows, pathologists correctly agreed on the diagnosis in 46 (90.2%) cases, incorrectly agreed on 1 (2.0%) and disagreed on 4 (7.8%). Agreement on megakaryocyte number, morphology and distribution was fair.
Inter-rater reliability of bone marrow examinations for the diagnosis of ITP was fair. Pathologists often incorrectly agreed that ITP bone marrows were considered to be within normal limits. These data suggest that ITP bone marrows often do not exhibit any distinguishing features.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.