Seasonal prevalence of MS disease activity

Abstract

Objective: This observational cohort study investigated the seasonal prevalence of multiple sclerosis (MS) disease activity (likelihood and intensity), as reflected by new lesions from serial T2-weighted MRI, a sensitive marker of subclinical disease activity.

Methods: Disease activity was assessed from the appearance of new T2 lesions on 939 separate brain MRI examinations in 44 untreated patients with MS. Likelihood functions for MS disease activity were derived, accounting for the temporal uncertainty of new lesion occurrence, individual levels of disease activity, and uneven examination intervals. Both likelihood and intensity of disease activity were compared with the time of year (season) and regional climate data (temperature, solar radiation, precipitation) and among relapsing and progressive disease phenotypes. Contrast-enhancing lesions and attack counts were also compared for seasonal effects.

Results: Unlike contrast enhancement or attacks, new T2 activity revealed a likelihood 2-3 times higher in March-August than during the rest of the year, and correlated strongly with regional climate data, in particular solar radiation. In addition to the likelihood or prevalence, disease intensity was also elevated during the summer season. The elevated risk season appears to lessen for progressive MS and occur about 2 months earlier.

Conclusion: This study documents evidence of a strong seasonal pattern in subclinical MS activity based on noncontrast brain MRI. The observed seasonality in MS disease activity has implications for trial design and therapy assessment. The observed activity pattern is suggestive of a modulating role of seasonally changing environmental factors or season-dependent metabolic activity.

Figure 1 Distributions of observations and subjects over the calendar year (A) Seasonal distribution of MRI examinations across the calendar year. In black, the number of individual subjects scanned each month; in white, the total number of examinations. For example, in January, 80 MRIs were acquired from 44 subjects. Contribution of subjects to each calendar month and season was very evenly distributed. See also table 1. (B) The distribution of observations over the year in a polar plot. Scale is examination/day. The rate variables reported in figure 2 are counts normalized against this distribution.

Figure 2 Polar plots of annual multiple sclerosis disease activity distributions, with color background showing climate data (solar radiation, daily peak temperature, and daily mean precipitation) Thick and thin curves show smoothed and raw distributions. All units are per day; rates are counts divided by number of observations (see figure 1B). (A) New lesion rates, showing a distinctly elevated activity in one-half of the year and strong associations with solar radiation; (B) random half of the data in A, i.e., the cohort split randomly into 2 groups. The pattern and associations remain. (C) Active scans are scans with at least 1 new lesion, i.e., normalizing for individual levels of disease activity. Same pattern as (A) remains, and associations with daily temperature are apparent; (D) per subject analysis, i.e., further isolating individual activity levels via a subjects-with-activity count; e.g., more than 30% of subjects had activity in June, vs ∼10% in November. The seasonal difference between spring/summer and autumn/winter remains distinct. No significant associations with precipitation are apparent.

Figure 3 Seasonal differences in disease activity Box and whisker plots of seasonal differences in disease activity. Variables were the fractions of the total lesion count or active scans over a calendar year, concentrated on any particular season; e.g., the 44 subjects included had less than 20% of their disease activity during fall and winter months. The red and blue bands show 2 standard deviations across the 1,000 Monte Carlo simulations, representing the (very low) contribution of temporal uncertainty of lesion occurrence, compared to the (much larger) intersubject variability. Both lesion and active scan metrics show significantly higher rates in the spring and summer months.