# HerMeS: a registry-based evaluation of the HERCULES criteria for identifying nonrelapsing SPMS

## TL;DR

This study evaluates the HERCULES criteria for identifying non-relapsing SPMS in a large MS registry, comparing it to neurologist-defined cases.

## Contribution

The study applies and validates the HERCULES criteria in a real-world registry to identify non-relapsing SPMS patients.

## Key findings

- HERCULES criteria identified 7.9% of patients as non-relapsing SPMS, compared to 4.26% by neurologists.
- HERCULES-defined SPMS patients had fewer post-conversion relapses and earlier progression independent of relapse activity.
- The HERCULES criteria showed high specificity but low sensitivity for identifying SPMS.

## Abstract

Secondary progressive multiple sclerosis (SPMS) encompasses heterogeneous phenotypes that may or may not exhibit disease activity.

To characterize a cohort of non-relapsing SPMS (nrSPMS) identified through a data-driven definition adapted from the HERCULES trial criteria, compared with neurologist-defined (ND) SPMS.

Relapsing MS patients were retrospectively extracted from the Italian Multiple Sclerosis Register (RISM). ND was defined according to clinical criteria for SPMS and the HERCULES cohort adapting the trial inclusion criteria. Progression independent from relapse activity (PIRA) and relapse-associated worsening (RAW) events were assessed. Diagnostic performances of SPMS definitions were evaluated.

Among 20,306 patients, 866 (4.26%) were classified as SPMS by ND and 1603 (7.89%) by HERCULES criteria. The HERCULES group included older patients, less likely to relapse post-conversion. PIRA occurred in 88.8% of ND and 88.5% of HERCULES-defined cases, with a shorter median time to first PIRA in the latter. The HERCULES definition showed high specificity (93.4%) and low sensitivity (37.6%).

HERCULES-adapted criteria identified a subgroup of 7.9% nrSPMS patients in the real-world cohort of RISM, characterized by fewer post-conversion relapses and earlier PIRA onset. These findings reinforce the value of applying standardized algorithmic definitions of SPMS in registry-based studies.

The online version contains supplementary material available at 10.1007/s00415-026-13716-1.