Skip links

In Silico in Action: A Case Study in Relapsing Remitting Multiple Sclerosis

Computer modeling and simulation of humans, both healthy and with diseases, is a powerful tool. It can augment preclinical and clinical research through mechanistic and predictive investigations that would otherwise be impossible. In recent years, regulatory agencies have begun to accept evidence obtained through modeling and simulation, and in silico clinical trials have emerged as an important approach for virtual testing of pharmacological therapies and medical devices.  

In this post, we introduce a modeling and simulation tool for relapsing remitting multiple sclerosis (RRMS) and present a case study demonstrating its value in clinical research. 

About MS TreatSim 

A mechanistic-based model called MS TreatSim, developed by InSilicoTrials Technologies, incorporates a computational model which enables simulation of the immune system and RRMS. This model focuses on the formation of oligoclonal bands, a key biomarker of RRMS disease activity. It integrates data from every consecutive stage of development to predict clinical endpoints from dosing and mechanism of action by incorporating a complete understanding of the relationships between these pathways. 

The first step in constructing an in silico trial is to define the population characteristics and selection criteria for the virtual patient population, which may include age of onset, lesion load, oligoclonal band status, and disease history and activity. Virtual patients are then generated who can be randomized to one or more cohorts to allow researchers to select treatment options. Once these inputs are provided, the study simulation is run using an algorithm that has been customized to the compound of interest. The output of the simulation consists of not only individual relapse rates and immune dynamics, but also population trial outcomes such as reduction in relapse rate.    

Using MS TreatSim to recreate a historical trial 

Premier Consulting used MS TreatSim to recreate a Phase 3 trial of natalizumab, which was published in 2006. Natalizumab is a disease-modifying treatment for RRMS and is used as first- or second-line therapy depending on a patient’s clinical history. This trial included both a treatment group and a control, and patients were randomized in a 2:1 ratio, with the treatment group receiving 300 milligrams of natalizumab every four weeks for a total of 104-116 weeks.  

Recreating the trial

Given that Premier Consulting did not have access to the root trial data, the recreation focused on recreating the trial design and the global characteristics of the trial. Our in silico recreation included 120 virtual patients randomized in a 2:1 ratio, with 80 virtual patients in the treatment group and a trial duration of 116 weeks. The defined population characteristics were age of onset with three age ranges, high lesion load, and presence of oligoclonal bands. Selection criteria also included a disease duration of five years, at least one relapse in the year preceding inclusion, and no relapses in the past month.    

Comparing historical trial results to the in silico recreation

In the historical trial, the percentages of relapse-free patients in the placebo and treatment groups at Week 104 were 41% and 67% respectively. Lesion reduction between the treatment and placebo groups at Week 104 was 83-92%, depending on the type of magnetic resonance imaging (MRI) used for lesion assessment.  

In our in silico recreation, 40% of patients in the placebo group were relapse-free at Week 104, concordant with the historical trial. However, 90% of patients in the treatment group were relapse-free, an overestimation compared to the observed result for relapse but in line with observed result for lesion reduction in the historical trial. This result was not surprising since disease activity in terms of lesion formation is an important component of the computational model.   

Key takeaway 

Computational models that combine known biological and physiological data with mechanistic knowledge of chemical and physical phenomena can be extremely valuable for both the preclinical and clinical evaluation of a medical product.