Novel Classifier Can Predict MDS Diagnosis

A novel two-stage diagnostic classifier based on genetic information can predict diagnoses of myeloid malignancies and myelodysplastic syndromes (MDS) in patients with cytopenias.

Amy DeZern, MD, MHS, of the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins and colleagues conducted the research as part of the National Heart, Lung, and Blood Institute National MDS Natural History Study, which is a prospective cohort study enrolling patients with suspected MDS.

Dr. DeZern and colleagues sequenced 53 genes in bone marrow samples from 1,298 patients who were diagnosed with a myeloid malignancy, including MDS and non-MDS myeloid malignancies.

The patients also included those who had alternative marrow conditions with cytopenia. Their inclusion was based on concordance among independent histopathologic reviews at local, centralized, and tertiary levels to “adjudicate disagreements when needed,” according to the study’s authors.

Dr. DeZern and colleagues developed a novel two-stage diagnostic classifier that was based on mutational profiles in 18 of the 53 sequenced genes. These genes were “sufficient to best” predict a diagnosis of myeloid malignancy, and within those cases of predicted myeloid malignancy, predict if it was MDS, according to the researchers.

The diagnostic classifier achieved a positive predictive value (PPV) of 0.84 and a negative predictive value (NPV) of 0.8 with an area under the receiver operating characteristic (AUROC) of 0.85 when classifying patients as having a myeloid malignancy or not based on variant allele frequencies in 17 genes. The classifier achieved a PPV of 0.71 and NPV of 0.64 with an AUROC of 0.73 when classifying patients as having MDS versus a non-MDS malignancy based on variant allele frequencies in 10 genes.

Dr. DeZern and colleagues then assessed how the approach “could complement histopathology to improve diagnostic accuracy.” They found 139 patients had local and centralized histopathologic disagreement in determining if they had a myeloid malignancy or not. The diagnostic classifier achieved a PPV of 0.83 and an NPV of 0.65. The classifier-predicted diagnosis agreed with the tertiary pathology review—which was “considered the internal gold standard”—in 71% of the patients with localized and centralized histopathologic disagreement.

“In conclusion, the new two-stage classifier based on 18 genes can be applied alone or in combination with morphologic review to predict a diagnosis of myeloid malignancy and MDS for a patient with cytopenias, especially when the bone marrow morphology is less definitive,” Dr. DeZern and colleagues wrote, adding that “integrating genetic information into MDS classification schemes is paramount to help establish the appropriate therapeutic inventions to alter the natural history of MDS.”

An online version of the classifier that can be used with variant allele frequencies or binary mutation profiles is available at https://thenationalmdsstudy.net.

Reference

DeZern AE, Goll JB, Lindsley RC, et al. Utility of targeted gene sequencing to differentiate myeloid malignancies from other cytopenic conditions. Blood Adv. 2023. doi:10.1182/bloodadvances.2022008578