Childhood body mass index and multiple sclerosis risk: a long-term cohort study

Introduction

The growing world-wide obesity epidemic has multiple deleterious effects on public health, including cardiovascular and metabolic diseases, but recently, obesity has also been associated with an increased risk of multiple sclerosis (MS), a demyelinating disease of the central nervous system. In a prospective study of over 200,000 US women, being obese in late adolescence/early adulthood (age 18) was associated with a 2-fold increased risk of MS,¹ and being overweight during this time period with a 40% increased risk of MS. A recent Swedish case-control study also reported a 2-fold increased risk of MS among individuals who were obese at age 20, and found that the association was similar in men and women.² These findings lead to the question of whether there is an association between body size in childhood or early adolescence and MS risk, and if this association is similar in girls and boys. Therefore, we conducted a prospective, longitudinal study in a population of Danish schoolchildren to determine whether body mass index in late childhood/early adolescence predicted their risk of developing MS later in life and whether this risk differed between girls and boys.

Methods

Results

Of the 774 MS cases occurring among CSHRR participants during follow-up, 501 were among girls and 273 were among boys. As expected, BMI increased with increasing age in both girls and boys. (Table 1) At each age 7 through 13 inclusive, girls and boys who later developed MS had a higher BMI than girls and boys who did not develop MS. The differences between cases and non-cases were only significant among the girls (Table 1).

Among girls, at each age from 7 to 13, a 1-unit increase in a BMI z-score was associated with significant hazard ratios of MS of 1.17 to 1.21 (Table 2). As compared with girls with BMI <85^th percentile, those in the ≥ 95^th percentile had a significant 1.61 to 1.95-fold increased risk of MS at every age (Table 3), while among those with a BMI between the 85^th-<95^th percentiles a significant increased risk of MS was seen only at ages 12 and 13.

Among boys, there was a significant hazard ratio of 1.14 to 1.15 for MS with every 1-unit increase in BMI z-score at ages 8 through 10, but no significant associations at other ages (Table 2). Boys with a BMI ≥ 95^th percentile at age 7 had a significant 1.81-fold increased risk of MS as compared with boys with BMI <85^th percentile, and while there was a tendency to an increased risk at other ages, none of the associations attained significance (Table 3).

While overall the associations between the age-specific BMI measures and risk of MS were attenuated in the boys as compared to the girls, there was no statistical evidence for heterogeneity between the sex- and age-specific hazard ratios (p value for heterogeneity ranged from 0.29-0.81 for the BMI z score analysis, and 0.27-0.94 for BMI in the ≥ 95^th percentile versus <85^th percentile). We therefore combined the girls and boys in one analysis. Overall, there was a significant 1.15-1.18-fold increased risk of MS per a 1-unit increase in the BMI z-score in girls and boys combined (Figure 2), and a significant 1.56-1.81-fold increased risk of MS among children in with BMI ≥95^th percentile as compared to those with BMI <85^th percentile. (Table 3)

Birth weight was not associated with risk of MS in either girls or boys. In pooled analyses of both girls and boys, using birth weight of 2.751-3.25 kg as the reference, HR_{2.0-2.75 kg}=0.94, 95% CI: 0.72-1.24; HR_>3.25-3.75=1.06, 0.87-1.29; HR_>3.75-4.25=1.19, 0.94-1.51; HR_>4.25=0.96, 0.63-1.45.

All of the above analyses were stratified by birth cohort to account for secular changes in body size. There were no consistent trends or differences over time in the associations between body size and MS risk in either boys or girls (data not shown), either within or between the birth cohorts.

Discussion

In this large, prospective study among Danish school children, higher BMI during childhood and early adolescence was associated with an increased risk of MS. Among boys, the association was weaker than for girls, and overall not significant; however, this may be at least in part a result of the smaller sample size and thus lower power. Heterogeneity tests comparing the effect estimates of boys and girls suggest they were not materially different, and justified pooling of the estimates, though we cannot rule out true biological differences. Additionally, we found no association between birth weight and risk of MS, which is in agreement with other studies.^{9, 10}

The results of this study add to the previous findings in the US and Sweden of a 40% increased risk of MS among individuals who were overweight, and a 2-fold increased risk among those who were obese (BMI ≥ 30 kg/m²) at age 18¹ or 20,² by suggesting that overweight or obesity in childhood and early adolescence is also associated with an increased risk of MS. There are several strengths of the current study including the population-based prospective design, large sample size, and, importantly, objective measurements of height and weight during early life. Further, while the CSHRR only includes children who attended public or private school in the city of Copenhagen, the Danish MS Registry is a nationwide registry and would capture cases of MS occurring in adulthood in the CSHRR cohort throughout the country. Thus, we are capturing cases irrespective of whether they have been already in the medical system due to potential complications of overweight/obesity, which otherwise could create a serious selection bias.

Our results are consistent with BMI in childhood being etiologically important for MS in adulthood, and interestingly, there was essentially no variation in the risk of MS associated with BMI across ages 7 to 13, suggesting that, at least among girls, a high pre-puberty BMI (e.g. at age 7) and high pubertal BMI (e.g. at ages 11 to 13) have the same association with MS risk. One consideration, however, is that body size is correlated over the life course and in the two Nurses' Health Study (NHS) cohorts, the correlation between body size, as self-reported by selecting a best representative pictogram, at age 10 and 20 was 0.59 (NHS II) and 0.87 (NHS I).¹ The positive association observed between having a larger body size at age 5 or 10 and MS risk in the NHS did not persist after adjusting for body size at age 20; however, the high correlations between age-specific BMI values may prevent adequate separation of effects with the statistical methods used in that study.¹¹ One important consideration is that weight and body size over the life course, while correlated, is a more dynamic process that typically thought and obese/overweight children do not necessarily become obese/overweight adults.¹² Thus, the increased risk of MS observed with having a larger body size at each age 7 through 13 may be of biological importance irrespective of the tracking through adult life, and certainly they suggest that prevention of or intervention on childhood overweight or obesity may contribute to reduce MS risk.

Information on other MS risk factors such as infection with Epstein-Barr virus (EBV),¹³ HLA-DRB1 genotype,¹⁴ and vitamin D nutritional status,¹⁵ is not available in the CSHRR, and potential confounding by these factors cannot be evaluated. However, seropositivity to EBV was similar in obese and normal weight individuals¹⁶ and studies of HLA-DRB1 genotype have not addressed a potential association with body size. Vitamin D may be on the causal pathway between body size and MS risk (see below), in which case, adjusting for vitamin D status may not be appropriate.

Why having a large body size in childhood would increase the risk of MS in adulthood is not known, but there are a few possibilities to consider. Blood levels of 25-hydroxyvitamin D (25(OH)D), the main circulating vitamin D metabolite and nutritional biomarker of vitamin D status,¹⁷ are generally lower (∼15 ng/mL or less difference) in overweight/obese adults as compared to those of normal weight,^{18, 19} and high BMI has also been associated with lower 25(OH)D in children (4-13 years old).^{20, 21} In a prospective study among US military personnel, elevated 25(OH)D levels during young adulthood in non-Hispanic whites were associated with a decreased risk of MS.²² Thus, if being overweight or obese during early life is causally related to MS, it may be acting through the vitamin D pathway. Another possible pathway is based on the increased secretion of adipokines from the adipose tissue in obesity. There is evidence that some adipokines, such as leptin, can modulate immune system function, including reducing the proliferation of regulatory T-cells,²³ and promoting an inflammatory T helper 1 cell response.²⁴ Additionally, leptin-deficient mice are protected against experimental autoimmune encephalomyelitis,²⁵ an animal model of MS, and administration of leptin worsens the disease.²⁵ High leptin levels have also been associated with lower regulatory T-cells in MS patients.²⁶

Overweight/obesity is a major public health problem worldwide and trends of increasing body size among children are particularly concerning. This study expands on previous work, suggesting that having a high BMI in early life is a risk factor for MS. The incidence of MS has been increasing in Denmark, particularly among women²⁷, and our study suggests that a parallel increase in overweight/obesity in the Danish population²⁸ may, in part, explain this. Participants in the CSHRR are predominately of European descent, ³ and these findings may not be generalizable to other race/ethnic groups. Confirmation of these findings in other populations, studies of body size over the full course of childhood and adolescence, and exploration of possible pathways are needed.