New AI score enhances monitoring of myelodysplastic neoplasms

Weill Cornell Medicine investigators have developed a new AI-based method to assess patients with a form of blood cancer called myelodysplastic neoplasms. The method, published June 30 in Leukemia, compares the physical locations and morphometric (size and shape) characteristics of every hematopoietic cell in the patient's bone marrow sample with those in healthy bone marrow, to generate a score that reflects disease severity.

Myelodysplastic neoplasms (MDS) is a form of blood cancer that usually affects older adults. About one-third of patients progress to a more aggressive cancer called acute myeloid leukemia. MDS patients must undergo frequent biopsies to identify signs of remission or worsening disease.

It's fundamentally a chronic and progressive disease. With the current methods pathologists use to assess MDS samples, there are some clear-cut cases and a lot of grey area."

Dr. Sanjay Patel, clinical chief of hematopathology, and associate professor of pathology and laboratory medicine, Weill Cornell Medicine

More precise diagnosis

Building on their previous work using AI to develop more detailed images of human bone marrow, Dr. Patel and his collaborator Dr. David Redmond, assistant professor of computational biology research in medicine at Weill Cornell Medicine, developed a new method using fully deidentified patient samples that may provide more clarity for patients with MDS and their physicians.

"Using AI to assist us in looking at the spatial architecture of the bone marrow using widely available laboratory assays allows us to improve our capability to assess patients' prognosis and perhaps triage them for precision therapies," Dr. Redmond said.

The MDS-Microarchitectural Perturbation Score (MDS-MAPS) score ranks patient samples based on 82 features associated with normal tissue and different genetic subtypes of MDS. The tool uses routinely collected samples, tissue staining protocols, and imaging technologies that could be implemented in most hospital pathology departments and laboratories.

"We can generate an MDS-MAPS value for a patient at diagnosis and track how it changes over time," said Dr. Patel, who is also a practicing hematopathologist at NewYork-Presbyterian/Weill Cornell Medical Center.

Lower scores indicate the sample is more like healthy tissue, and higher scores indicate more disease-related changes. They then showed using computer modeling that the tool may be better at classifying patients' disease status than current methods.

"It distills something very complex down into a numerical value," Dr. Patel said. "It is easier for a patient to understand whether their score is trending in the right or wrong direction, or whether their disease is more likely stable."

The next step will be to validate the tool in larger cohorts of patient samples and determine whether its use improves patient care. Dr. Patel and Dr. Redmond will team up with Dr. Pinkal Desai, an associate professor of medicine at Weill Cornell Medicine and a hematologist/oncologist at NewYork-Presbyterian/Weill Cornell Medical Center, to study how the tool performs on samples from patients with other forms of MDS, and increasingly recognized precursor conditions.

"Patients with MDS and related precursor conditions have many mutations as part of the disease biology and each patient's molecular signature is different," said Dr. Desai, who is also clinical director of the Englander Institute Precision Medicine Molecular Aging Institute. "We have always wondered if these mutations signal a different spatial pattern and whether these patterns have an impact in predicting how patients progress and respond to treatments. Together we are poised to harness technology to answer real-world clinical questions."

Molecular clues

The research is also yielding new clues about how mutations in a tumor suppressor gene called TP53 may contribute to the development and progression of MDS. Previous studies by Dr. Redmond and Dr. Shahin Rafii, chief of the Division of Regenerative Medicine at Weill Cornell Medicine, mapped the architecture of the healthy bone marrow. The current study showed that the same basic architecture is present in human bone marrow and that as MDS progresses, bone marrow stem cells become increasingly misplaced.

Dr. Redmond explained that in healthy tissue, blood vessels in the bone marrow help nurture the bone marrow stem cells that later develop into blood and immune system cells. But a receptor, CXCR4, that facilitates these supportive interactions is reduced in stem cells from patients with MDS, particularly in those with TP53 mutations.

Learning more about these interactions could lead to targeted therapies designed to shift gene expression or help restore the normal architecture and cell-cell interactions in the bone marrow.

The information could also help further refine the methods pathologists use to monitor patients, with this and other similar bone marrow conditions, Dr. Patel said.