ELABORATION DU SCORE PREDICTIF DE L’ATTEINTEMEDULLAIRE DANS LE LYMPHOME DE HODGKIN

Abstract

ELABORATION OF THE PREDICTIVE SCORE OF BONE MARROW INVOLVEMENT IN HODGKIN LYMPHOMA Introduction: In Hodgkin's lymphoma (HL), bone marrow involvement remains rare, its incidence varies between 4% and 14% in national and international studies. In the absence of baseline position emission tomography/ computed tomography (PET/CT) in the initial staging ; bone marrow biopsy (BMB) is performed to determinate a possible involvement of the bone marrow. The role of a routine BMB in the HL remains open to discussion because of the low probability of marrow involvement in some groups of patients. Objectives: The aim of this work is to analyze the various parameters likely to lead to bone marrow involvement patients with HL and, in accordance with the main objective assigned to it, to develop a predictive score for bone marrow involvement from a cohort of patients consistent with the epidemiological profile of HL in Algeria and to validate its predictive value on an external cohort. Patients and methods: Multicenter diagnostic study carried out on HL patientsdiagnosted and treated during the last decade at the level of the hematology departments of different hospitals in the west of the country that participated in this study. The study endpoint was bone marrow involvement by BMB. The final predictive model was obtained by binary logistic regression in the derivation cohort, its validation from external cohort of incident cases recruited in 2021/2022 in the western centers that initially participated in the study (reproducibility score) and from a geographically different cohort of HL patients recruited at the CHU of Sétif and the CAC of Blida (transportability score). Results: We developed a predictive bone marrow involvement score (PBMIS) from 940 HL patients who constituted the derivation cohort and validated on 756 additional patients). The PBMIS has eight independent dichotomous variables for a maximum of 14 points. The score has a good discrimination (statistic C is 0.84 in the derivation cohort and 0.90 in the validation cohort). At a threshold ≥ 4, the sensitivity, specificity, PPV and NPV are respectively of the order of 82.9%, 71%, 20.12% and 97.9% in the derivation cohort and the 86%, 78.4%, 19.1% and 99% in the validation cohort. Conclusion: We offer practitioners a simple, reproducible and validated score capable of predicting the risk of bone marrow involvement in LH thus identifying the subgroup of low risk patients in whom BMB would probably be omitted. Keywords: Hodgkin's lymphoma, bone marrow biopsy, predictive score of bone marrow involvement