Validation of a novel algorithm with a high specificity in ruling out MDS

Area of Science:

• Hematology
• Medical Diagnostics
• Bioinformatics

Background:

• A web-based app utilizing Gradient-boosted models (GBMs) was previously developed to aid in the diagnosis or exclusion of myelodysplastic syndromes (MDS).
• This tool integrated eight laboratory parameters for its diagnostic predictions.

Purpose of the Study:

• To validate the diagnostic algorithm developed by Oster et al. for myelodysplastic syndromes (MDS).
• To assess the accuracy of the GBM-based app in distinguishing MDS patients from non-MDS anemic individuals.

Main Methods:

• Comparison of 175 anemic patients with MDS and 1378 non-MDS anemic patients from the German MDS Registry and Düsseldorf university hospital.
• Analysis of laboratory parameters including hemoglobin, leukocyte and platelet counts, MCV, absolute neutrophil and monocyte counts, glucose, and creatinine, along with patient gender and age.

Main Results:

• The algorithm demonstrated a specificity of 90.9% and a positive predictive value (PPV) of 77.1%, but did not achieve a high negative predictive value (NPV).
• 1192 out of 1378 controls were correctly classified as "probably not MDS (pnMDS)".
• Inclusion of chronic myelomonocytic leukemia (CMML) cases suggested accurate exclusion of highly proliferative bone marrow disorders.

Conclusions:

• Peripheral blood laboratory work can be useful for excluding MDS from the differential diagnosis.
• Confirmation of MDS diagnosis necessitates a bone marrow biopsy, despite the app's utility in exclusion.