In this northern European population-based study spanning a decade, we have shown a seasonal pattern in MS events, with a peak in late spring/early summer, and a trough in late summer. This pattern is most apparent in younger patients with relapsing–remitting disease and is associated with monthly hours of sunshine. However, this does not account for all variation observed, suggesting that seasonal patterns of MS events are multifactorial.
A seasonal pattern in MS relapses has previously been noted, with most studies reporting a solitary peak in spring and/or summer months in both northern [2, 17, 23, 26, 31] and southern hemispheres [7, 26], a pattern which has also been borne out by meta-analysis . Fewer studies have detected a nadir in relapse rates, but in studies that have been able to detect this, the nadir tends to occur in late summer or autumn [9, 23, 26, 29, 31]. Three studies have not detected any seasonal variation [8, 18, 22]; however, one of these studies was a retrospective review of MS hospital admissions and thus may have overlooked less severe relapses , and the other two were relatively small studies in southern Europe where seasonal variation in climate may be less than in northern Europe or America [8, 22]. Indeed, a recent large study incorporating international data from the MS Base Registry suggests an effect of latitude on seasonal pattern of MS events . These findings are also supported by our study, which is the largest single centre study of its kind and uses high-quality clinical data collected prospectively in a well-described cohort, suggesting that a seasonal pattern is genuine and not due to reporting bias.
There is also objective paraclinical evidence to support seasonal variation in neuroinflammatory activity; in a prospective North American study of 44 patients with parallel sequential MR imaging, new T2 activity was 2–3 times more frequent between March and August, although no similar increase in clinical events was observed . Alterations in the pattern of seasonal cytokine variation have also been observed in MS patients: seasonal variation is seen in healthy controls in IL-4, IL-10, TNF-α, and IFN-γ, but in untreated MS patients, only seasonal variation in IL-10 was preserved, while in treated MS patients, the seasonal variation was lost for all four cytokines .
Since our study was based on prospectively collected data from individual patients, we have been able to assess risk of relapse at an individual level. This is a novel approach, which has allowed a more detailed analysis of monthly risk of an event than has previously been possible and has also allowed us to take into account the contribution of clinical variables to the probability of a relapse.
An association between UV levels and pattern of MS events has previously been observed [26, 29], and it has been proposed that this association occurs as a result of the mechanism of vitamin D production . The best-fitting model in our study included hours of sunshine, although the mechanism by which this might take effect is not clear. We observed highest odds of relapse during early summer when hours of sunshine were high, and lowest odds of relapses during late summer. This implies that if vitamin D has a role, it cannot be the only factor that is relevant in triggering a relapse and may be modified by alternative environmental factors which modulate its effects.
Other factors associated with increased probability of relapse include younger age , shorter disease duration , and viral infections [1, 6], while pregnancy and breastfeeding [5, 14], serum vitamin D levels , and DMTs [3, 12, 19, 21] appear protective. It seems most likely that interaction between a number of these and other factors leads to initiation of CNS inflammation, manifesting clinically as a relapse. However, the relative contribution of individual factors and their contribution to the pathological sequence of events remain elusive. In addition, while some factors associated with probability of an MS event exhibit a seasonal pattern, such as climate variables or infections, others have a very different risk profile, such as the declining risk of relapse with increasing age , or the reduction in risk during pregnancy with a return to the previous levels of risk post-natally . Therefore, it would seem unlikely that MS events would conform to a single pattern such as seasonal variation.
This study has a number of strengths. It is a large, population-based study, with good relapse ascertainment in a patient cohort that has had detailed clinical follow-up for more than 10 years. By confining the study to one geographical area, possible confounding effects of other environmental factors have been minimised. The analysis utilised objectively measured climate variables, and allowed clustering of multiple relapses within a given individual to be taken into account, thus allowing for individual variation in relapse rate.
The relapse rate in this study is considerably lower than that observed in most randomised controlled trials, which tend to select patients with active or highly active relapsing disease. This difference in relapse rate may also occur as the result of a more rigorous surveillance system. However, our data are comparable with rates seen in prospective population-based studies and we feel likely to be a true representation of overall relapse frequency in this cohort. We also recognise additional limitation in the analysis of these data; both seasonal variation and age-specific differences in relapse frequency may have occurred as a result of a systematic reporting bias. Although this is difficult to confidently exclude, it would seem unlikely given the rigorous methodology and results of previous studies.
In addition, relapses in this study have been grouped by month, and some excluded by necessity, because no accurate date could be ascribed. Patients with relapse of uncertain date appear to be more likely to have primary progressive disease, have shorter duration of follow-up, be older, and be male. However, this missing data should only bias conclusions if the relapses of uncertain date had a different seasonal pattern than those with a confirmed date. Only a limited number of climate variables were included here, and there may be other factors (or combinations of factors) which could explain remaining monthly variation in relapse rates. Our model also only adjusts for monthly (rather than daily) weather, and thus, some residual confounding effects could remain. It also adjusts for weather at the area level (with data on sunshine, rainfall, and temperature from south Wales, but UV index data for the UK as a whole) so may not represent geographical location on an individual level. Furthermore, there is no information on the amount of time each individual spent outside in a given month, and thus, individual exposure to climate variables is proxied by the monthly values of these variables, leading to measurement error in individual exposure.
Finally, the geographical limitations of this study, whilst minimising confounding by other, unmeasured environmental factors also minimise variation in exposure in this population. Stronger associations with climate variables might be observed in a population with a more diverse range of exposure, both by geography and across the year. All the climate variables except rainfall were very highly correlated, and thus, the independent associations of relapse rate with all climate variables could not be examined.
In conclusion, we have interrogated patterns of relapse in a large population-based cohort of patients from south east Wales with detailed clinical data, confirming the presence of a seasonal pattern in MS events and demonstrating association with hours of sunshine. However, in exploring the relative contribution of this and other factors to observed seasonal variation, we have recognised that the causes of a relapse are likely to be multifactorial and that a better understanding of the effect of environmental factors on the development of MS and its clinical manifestations remains to be established. This, in turn, may offer opportunities for improved treatments, strategies for relapse prevention, and alteration of risk of disease on a population level.