Early prediction of unfavorable evolution of Clinically Isolated Syndrome (CIS) patients

In the last years there is a growing interest about patients with clinically isolated syndrome (CIS) suggestive for multi-focal inflammatory demyelinating disease of the central nervous system. These cases can evolve to clinically definite multiple sclerosis (CDMS), before which a disease-free status can last for an unpredictable time. Recently, the disease-free concept has been applied to autoimmune diseases after the introduction of disease-modifying therapies. For multiple sclerosis (MS), the term disease-free status has been replaced by NEDA (no evidence of disease activity), that has become a new outcome measure and a treatment goal [Giovannoni G, et al. Multiple Sclerosis and Related Disorders 2015].
A reliable prediction of the single patient prognosis is fundamental to the patients who want to be informed about their prospects and to the clinician to identify at an early stage of disease those patients who are at high risk. Moreover predictive studies may provide a useful contribution to rationalizing the design of clinical therapeutic trials.
A previous study developed a Bayesian graphical model with Markov chain Monte Carlo simulation, that allowed to calculate the BREMS (Bayesian Risk Estimate for MS) score, strongly related to the risk of unfavourable course [Bergamaschi R, et al. J Neurol Neurosurg Psychiatry 2007]. Another score, related only to the clinical variables at onset (BREMS at onset, BREMSO), demonstrated to be strongly related to a poor evolution of disease [Bergamaschi R, et al. European J Neurol 2015].
Modeling the course and outcome of a disease is a difficult but not impossible endeavour provided a series of conditions that have to be satisfied .
The aim of the study is to define prognostic models for the prediction of losing NEDA status in CIS patients, through the analysis of a large set of clinical and instrumental variables (MRI, evoked potentials, cerebrospinal fluid) collected retrospectively and prospectively in CIS patients.

---

Introduction and aims
The identification of early indicators of multiple sclerosis (MS) prognosis is useful in clinical practice to support early and personalized therapeutic choices.
The aim of the study was to define a prognostic model, which gives in clinically isolated syndrome (CIS) patients an individual risk score (risk estimate for CIS, RECIS), for the early forecast of losing NEDA status (no evidence of disease activity) within 12 months from disease onset, through the analysis of a large set of clinical-demographics (sex, age, type of neurological involvement, EDSS) and instrumental-laboratory (number and site of brain MRI lesions, cerebro-spinal fluid oligoclonal bands - OB, visual and somatosensory evoked potentials – VEPs and SEPs) variables collected at disease onset.

Results
To identify the best prognostic model, we analyzed retrospective data of 221 CIS patients from 4 Italian MS Centres: Gallarate and Pavia (Northern Italy), Florence (Central Italy), Bari (Southern Italy).
After 12 months from CIS onset, 38% of patients displayed EDA.
The variables were included in four different multivariable regression models with 1000 bootstrap replications: 1) a stepwise logistic regression, 2) a Lasso regression with cross-validation approach, 3) a Lasso regression with adaptive approach, and 4) a Lasso regression with plugin approach. According to the best prognostic model, the risk of EDA at 12 months was higher in younger patients and in those patients that at disease onset had MRI infratentorial lesion(s), OBs and abnormal SEPs in lower limbs.
A final score (RECIS score) composed by the variables selected from the best model was reported and the optimal cut-off (0.65) to predict the probability to change the status from NEDA to EDA within one year was estimated. The RECIS score showed good specificity (72%) as a tool for predicting the evolution of CIS: higher RECIS scores were linked to an increased risk of losing NEDA within 12 months. For example, according to RECIS score, a 30-year-old patient with oligoclonal bands, abnormal lower limb SEP and infratentorial lesions at MRI at the onset of the CIS will have 85% of probability to reach EDA status within 12 months.

Conclusions
RECIS score is a simple tool, which can be used at MS onset to predict its evolution within 12 months at a single patient level. This instrument could promote tailored therapeutic decisions in clinical practice setting.


Pubblicazioni e Comunicazioni a Congressi/ Publications and Congress Presentations
Mallucci G, Trivelli L,Colombo E, Trojano M, Amato MP, Zaffaroni M, Iaffaldano P, Razzolini L, Prestipino E, Saraceno L, Montomoli, C, Bergamaschi R. The RECIS (Risk Estimate in CIS) study: a novel model to early predict clinically isolated syndrome evolution. MULTIPLE SCLEROSIS JOURNAL 2019. 35th Congress of the European-Committee-for-Treatment-and-Research-in-Multiple-Sclerosis (ECTRIMS), Stockholm, SWEDEN
Research project of the Italian Multiple Sclerosis Register

---