Novel clustering approach to understanding distinct phenotypes of patients with systemic sclerosis.
7 affiliations
- 1Department of Internal Medicine, University of Michigan Scleroderma Program, Ann Arbor, Michigan, USA.
- 2Therapeutic Area Inflammation, Boehringer Ingelheim Pharmaceuticals, Inc, Ridgefield, Connecticut, USA.
- 3Therapeutic Area Inflammation, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany.
- 4Therapeutic Area Mental and Eye Health, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany.
- 5Cross Therapeutic Area Marketing and Operations, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany.
- 6Adelphi Real World, Bollington, UK.
- 7University of Leeds, Leeds, UK.
Abstract
Objectives
Despite earlier clustering studies on systemic sclerosis (SSc), how clinical subgroups differ in disease burden, treatment and patient-physician dynamics remains unclear. This study applied a novel clustering method to identify clinical subgroups.Design
Data were analysed from 1906 adults with SSc and their treating rheumatologists across the USA, France, Germany, Italy, Spain, UK and Japan (March-September 2022).Setting
Data were collected by Adelphi Real World as part of an independent survey titled the SSc I Disease Specific Programme, comprising data from the USA, France, Germany, Italy, Spain, UK and Japan.Participants
Physicians completed case record forms; patients completed self-reported questionnaires. Latent class clustering was conducted using five physician-reported variables: affected body areas, symptoms, severity, flares in the past 12 months and remission status.Interventions
This was a non-interventional, observational, retrospective study.Main outcome measures
A cluster solution was selected based on model fit and clinical interpretability. We compared time to diagnosis, symptoms at diagnosis, postdiagnosis trajectories, physician management and patient treatment perspectives across clusters.Results
We identified four clusters representing a gradient of disease severity, with all clusters comprising >75% females. Clusters 1/2 comprised younger patients with fewer concomitant conditions; Clusters 3/4 showed greater internal organ involvement and disease severity. Raynaud's phenomenon (RP) was the most frequent first symptom (physician-reported: 73%-85%; patient-reported: 63%-79%), but those with first symptoms other than RP were diagnosed faster (mean 1.06 vs 1.62 years). Musculoskeletal involvement was most pronounced in Clusters 2/4. Postdiagnosis (3.7-7.2 years), the number of symptoms and affected body areas decreased in Cluster 1, increased in Cluster 4 and remained relatively stable in Clusters 2/3. Improvement in disease progression within the first 2 years postdiagnosis was more often reported in Clusters 1/2 vs 3/4 (33%-34% vs 8%-20%). Most patients across clusters reported that the main benefit they wanted from treatment was to get rid of the disease.Conclusions
This analysis improves understanding of SSc beyond limited/diffuse subtypes. Musculoskeletal involvement emerged as a contributor to SSc burden, supporting its integration into monitoring/stratification strategies for earlier identification of high-risk patients.
Related
- Distinct Peri-pubertal Growth Phenotypes Among Vietnamese Urban Children: A Functional Clustering Analysis of 2018-2025 School Health Data
- Measurement-level clustering of ligamentum flavum thickness identifies distinct phenotypes associated with lumbar lordosis in adult spinal deformity: a proof-of-concept study.
- Multimodal data-driven clustering analysis of heart failure patients identifies two distinct patterns of heart failure with preserved ejection fraction: Results from the PACIFIC-PRESERVED study
- Latent class analysis identifies three biologically distinct phenotypes among patients undergoing endovascular thrombectomy: a derivation and temporal validation study
- Machine learning clustering of multidimensional clinical features identifies four prognostically distinct sepsis subtypes
BibTeX
@article{khanna2026novel,
title = {{Novel clustering approach to understanding distinct phenotypes of patients with systemic sclerosis.}},
author = {Dinesh Khanna and Mario Ehlers and Lizette Moros and Arpit Misra and Hyemin Song and Isabel Truman and Liane Gillespie-Akar and Francesco Del Galdo},
journal = {Journal of scleroderma and related disorders},
year = {2026},
month = oct,
doi = {10.1136/jsrd-2026-000041},
url = {https://europepmc.org/article/MED/42948250},
note = {CC BY-NC},
}
RIS
TY - JOUR
TI - Novel clustering approach to understanding distinct phenotypes of patients with systemic sclerosis.
AU - Dinesh Khanna
AU - Mario Ehlers
AU - Lizette Moros
AU - Arpit Misra
AU - Hyemin Song
AU - Isabel Truman
AU - Liane Gillespie-Akar
AU - Francesco Del Galdo
PY - 2026
DA - 2026/10/07
JO - Journal of scleroderma and related disorders
DO - 10.1136/jsrd-2026-000041
UR - https://europepmc.org/article/MED/42948250
AB - Objectives Despite earlier clustering studies on systemic sclerosis (SSc), how clinical subgroups differ in disease burden, treatment and patient-physician dynamics remains unclear. This study applied a novel clustering method to identify clinical subgroups.Design Data were analysed from 1906 adults with SSc and their treating rheumatologists across the USA, France, Germany, Italy, Spain, UK and Japan (March-September 2022).Setting Data were collected by Adelphi Real World as part of an independent survey titled the SSc I Disease Specific Programme, comprising data from the USA, France, Germany, Italy, Spain, UK and Japan.Participants Physicians completed case record forms; patients completed self-reported questionnaires. Latent class clustering was conducted using five physician-reported variables: affected body areas, symptoms, severity, flares in the past 12 months and remission status.Interventions This was a non-interventional, observational, retrospective study.Main outcome measures A cluster solution was selected based on model fit and clinical interpretability. We compared time to diagnosis, symptoms at diagnosis, postdiagnosis trajectories, physician management and patient treatment perspectives across clusters.Results We identified four clusters representing a gradient of disease severity, with all clusters comprising >75% females. Clusters 1/2 comprised younger patients with fewer concomitant conditions; Clusters 3/4 showed greater internal organ involvement and disease severity. Raynaud's phenomenon (RP) was the most frequent first symptom (physician-reported: 73%-85%; patient-reported: 63%-79%), but those with first symptoms other than RP were diagnosed faster (mean 1.06 vs 1.62 years). Musculoskeletal involvement was most pronounced in Clusters 2/4. Postdiagnosis (3.7-7.2 years), the number of symptoms and affected body areas decreased in Cluster 1, increased in Cluster 4 and remained relatively stable in Clusters 2/3. Improvement in disease progression within the first 2 years postdiagnosis was more often reported in Clusters 1/2 vs 3/4 (33%-34% vs 8%-20%). Most patients across clusters reported that the main benefit they wanted from treatment was to get rid of the disease.Conclusions This analysis improves understanding of SSc beyond limited/diffuse subtypes. Musculoskeletal involvement emerged as a contributor to SSc burden, supporting its integration into monitoring/stratification strategies for earlier identification of high-risk patients.
SN - 2397-1983
SN - 2397-1991
ER -