Using machine learning classification of gene expression from peripheral blood to identify intrinsic molecular subsets may help in identifying patients with systemic sclerosis (SSc) who receive significant benefits from hematopoietic stem cell transplant (HSCT), according to study results published in Annals of the Rheumatic Diseases.
Previously, data from the Scleroderma: Cyclophosphamide or Transplantation trial (SCOT; ClinicalTrials.gov Identifier: NCT00114530) supported the clinical benefit of HSCT compared with cyclophosphamide in patients with SSc, with a significant increase in event-free survival following HSCT.
The objective of the current analysis was to determine whether SSc intrinsic subsets can help in predicting long-term responses to HSCT.
The study sample included 63 participants (mean age, 45.0 years; 60.3% women) from the SCOT trial, including 33 from the cyclophosphamide group and 30 from the HSCT group. Researchers analyzed gene expression from peripheral blood cells at baseline and follow-up. Using a machine learning classifier based on multinomial elastic net classification, the participants were classified by intrinsic gene expression at baseline; event-free survival and differentially expressed genes were also analyzed.
In the cyclophosphamide arm, 12 patients were assigned to the normal-like subset, 12 to the inflammatory subset, and 9 to the fibroproliferative subset. In the HSCT arm, 10 patients were assigned to the normal-like subset, 8 to the inflammatory subset, and 11 to the fibroproliferative subset.
Of participants in the fibroproliferative intrinsic subsets, there was a significant improvement in event-free survival among those who received HSCT compared with cyclophosphamide (P =.0091). However, among patients in the normal-like or inflammatory subsets, there was no significant difference (P =.77 and .1, respectively) in event-free survival between those receiving HSCT and cyclophosphamide.
There was a significant difference between the HSCT and cyclophosphamide groups in the number of significant differentially expressed genes; at each time point to 54 months, there were considerably more differentially expressed genes in the HSCT compared with the cyclophosphamide arm (P <.001).
Researchers explained that participants in the HSCT arm showed changes in gene expression, but not all attained event-free survival at 54 months. However, patients receiving cyclophosphamide who attained event-free survival did not exhibit significant differentially expressed genes.
The study had several limitations, including the exploratory nature and limited number of samples available for gene expression analysis.
“This study suggests that intrinsic subset stratification of patients may be used to identify patients with SSc who receive significant benefit from HSCT,” the researchers concluded.
Disclosure: Several study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures.
Franks JM, Martyanov V, Wang Y, et al. Machine learning predicts stem cell transplant response in severe scleroderma. Published online September 15, 2020. Ann Rheum Dis. doi:10.1136/annrheumdis-2020-217033
This article originally appeared on Rheumatology Advisor