Model Identifies Five Skin-Score Classes to Predict Systemic Sclerosis Progression

Steve Bryson, PhD avatar

by Steve Bryson, PhD |

Share this article:

Share article via email

A statistical model applied to a widely used measure of skin thickening in people newly diagnosed with systemic sclerosis (SSc) identified five classes that may predict organ involvement and survival rates.

Using these predictive classes — based on the modified Rodnan Skin Score (mRSS) — in the early stages of the disease may be important for clinical practice and future studies, the researchers said.

They described their findings in a study, “Early trajectories of skin thickening are associated with severity and mortality in systemic sclerosis,” published in the journal Arthritis Research & Therapy.

Skin thickening is one of the hallmarks of systemic sclerosis. The mRSS, with scores ranging from 0 (normal) to 3 (severe), is the standard method to describe skin thickness. The mRSS is often used as a primary or secondary outcome in clinical trials. 

However, as changes in skin thickness vary over time from patient to patient, it is difficult to accurately predict how mRSS will evolve and how these changes are related to long-term outcomes. This limits how patients are compared in clinical trials. 

Researchers in France applied a statistical model known as the latent class mixed model to mRRS scores of patients in the early stages of the disease. They used the approach to identify different predictive classes of skin thickening and assess how these classes were related to clinical characteristics and survival rates. 

The latent class mixed model is a standard statistical method used to analyze change over time when data are heterogeneous, as is the case with mRSS scores. Patients whose mRSS outcomes (trajectories) were similar fell into a specific “latent class,” while those who were dissimilar belonged to different classes. 

The team used data from medical records compiled in the French systemic sclerosis national database, analyzing patients who had had the disease for fewer than two years. Demographic information was collected, along with clinical data, blood test results, disease duration, and which organs were affected. 

A total of 198 patients had an available baseline mRRS and at least one mRRS assessed within four years of follow-up. 

The analysis identified five predictive classes that best fit the mRSS data. The duration of disease in these patients did not significantly differ between classes.

Class 1 represented those with a low mRSS at baseline with no significant change over time. This class included 117 patients with more common limited systemic sclerosis (82.6%). One-third of the patients had organ involvement including joints, interstitial lung disease (ILD), gastrointestinal tract (GIT), and digital ulcers (DU).

Class 2 patients had a slightly improved average mRSS over time (over one, two, three, and four years) from baseline. This class included 43 patients and almost 98% had diffuse systemic sclerosis. In this class, joint, DU, GIT, and ILD involvements were common. 

Class 3 was characterized by a low baseline mRSS that rapidly increased to a peak after 2.3 years, followed by an improvement by the fourth year. Two patients had limited systemic sclerosis, and 11 had diffuse systemic sclerosis. More than two-thirds of patients in this class had joint, DU, GIT, and ILD involvements.

Patients in class 4 had a moderately high baseline mRSS, followed by a rapid increase to peak at 2.2 years, then improvement at four years. This class was composed of 13 patients, and joint, GIT, and ILD involvements were frequent at baseline. 

Class 5 patients were characterized by a high mRRS at baseline and improved to a moderate mRRS within four years of follow-up. There were 12 patients in this class, all with diffuse systemic sclerosis in which most had affected joints and DUs. One-third of these patients also had ILD.  

Steroids and immunosuppressive therapies were more frequently given to those in classes 2 to 5 compared with class 1 (from 82.5% to 100%, compared with 53.3% in class one). Cellcept (mycophenolate mofetil), developed by Genentech, was used more often in class 5 than in the other classes. 

“We observed a progressive increase in the risk of death from classes 2 to 5 compared to class 1,” the researchers said. The results were similar after adjusting for gender and age. 

“This study identified five distinct mRSS trajectories in early SSc patients,” the researchers said. “The mRSS trajectory classes were associated with organ involvement and survival.” 

“This study suggested that mRSS trajectories characterization might be pivotal for clinical practice and future trial designs,” they said.