A total of 140 subjects (120 with MS or CIS and 20 controls; 76 from UCSF and 64 from Stony Brook) were included in the analyses after 19 (16 MS/CIS; 3 controls) were excluded because the status of vitamin D, DRB1 or all viruses was not available. Patients with MS/CIS were slightly older than controls but were similar with respect to sex, race and ethnicity (). The median disease duration at blood collection (interquartile range) for the MS/CIS group was 1.2 years (0.1, 8.3). Control diagnoses included neuromyelitis optica, suspected neuromyelitis optica vs. sarcoidosis, acute disseminated encephalomyelitis, scleroderma en coup de sabre, Guillain-Barre syndrome, cerebral vasculitis, low-grade tumor, recurrent optic neuritis (with normal brain MRI and CSF), mitochondrial or metabolic disorders, suspected paraneoplastic syndromes, multiple cranial neuropathies, non-specific white matter lesions on brain magnetic resonance imaging, psychogenic disease, and unknown disorders. Patients were more likely than controls to be antibody-positive for VCA and EBNA-1, although the confidence intervals do not exclude an association of disease state with the other viruses (). The basic assumptions of the multivariate models appeared to be reasonably addressed, although there were some outliers (using DFBETA); sensitivity analyses excluding the outliers are presented with the full model results in .
Characteristics of patients and controls
Predictors of Viral Antibody Levels in Multivariate Models
Among those who were VCA positive, only Hispanic ethnicity was independently associated with higher VCA antibody levels (coefficient= 0.54, 95% CI [0.13, 0.96], p=0.011) in multivariate models in which vitamin D was a binary variable (). A similar association was detected when vitamin D was a continuous variable or when outliers were excluded. MS/CIS status did not appear to be a confounder when it was added to the models, nor did it interact with vitamin D status. Adding the day of the year on which the blood was drawn was not associated with a meaningful change in the estimates.
Among subjects who were positive for EBNA-1 antibodies, DRB1-positive status appeared to be associated with higher antibody levels to EBNA-1, although the association was attenuated when outliers were removed (). There was no meaningful difference when the blood draw date was added to the models. While there was no evidence of MS/CIS status as a negative confounder in the association between vitamin D status and EBNA-1 antibody levels, there was an interaction between the dichotomized vitamin D level and MS/CIS status (p=0.034 for interaction term). Among those with a sufficient vitamin D level, those with MS/CIS had higher EBNA-1 antibody levels (coefficient=0.49, 95% CI [0.02, 0.97], p=0.043) than controls. However, when vitamin D was insufficient, there appeared to be no meaningful difference in EBNA-1 antibody levels between MS/CIS subjects and controls (p=0.52). This interaction was not apparent, however, in models in which vitamin D was a continuous variable.
In those who were positive for CMV, vitamin D sufficiency was associated with a higher CMV antibody level (coefficient 0.60, 95% CI [−0.06, 1.26], p=0.074). A similar association was discovered when vitamin D was modeled as a continuous variable; each 1 ng/mL greater vitamin D level was associated with a higher CMV antibody level (coefficient 0.03, 95% CI [0.00, 0.06], p=0.024) (). However, when outliers were removed, the association was not apparent. The effect of outliers could not be assessed when vitamin D was dichotomized because no patients with CMV antibodies had sufficient vitamin D levels after outliers were dropped.
In the more inclusive models, there was evidence of a strong interaction between vitamin D status and MS status (p=0.007 for interaction term when vitamin D levels dichotomized; p=0.035 when vitamin D was continuous). Among MS/CIS patients, vitamin D sufficiency was associated with a higher CMV antibody level (coefficient 1.04, 95% CI [0.36, 1.73], p=0.004), while sufficient vitamin D levels in controls were associated with lower CMV antibody levels (coefficient −1.10, 95% CI [−2.44, 0.25], p=0.11). When vitamin D was modeled as a continuous variable, greater vitamin D levels in MS patients were associated with higher antibody levels to CMV (coefficient 0.04, 95% CI [0.01, 0.07], p=0.004); however, the association of vitamin D levels and CMV antibody levels in controls appeared to be less meaningful than in the prior model (p=0.38). Regardless of how vitamin D was modeled, MS patients with lower/insufficient vitamin D had lower CMV antibody levels than controls. There was no evidence of MS confounding an association between vitamin D status and CMV antibody levels. Adding the day of the year on which the blood was drawn was not associated with a meaningful change in the estimates.
Herpes Simplex Virus-1 and -2
Only age appeared to be meaningfully associated with HSV-1 antibody levels. In multivariate models (with dichotomized vitamin D, DRB1 status, race, and ethnicity), each one year older age at blood collection was associated with a lower HSV-1 antibody level (beta coefficient −0.16, 95% CI [−0.34, 0.02], p=0.085); the results were similar in models in which vitamin D was a continuous covariate (). There was no meaningful difference when the blood draw date was added to the models. There appeared to be a possible interaction between MS/CIS status and vitamin D level (p=0.18), but analyses stratified by MS/CIS status did not show meaningful differences. Interaction could not be assessed in models in which vitamin D was dichotomous because there were no controls positive for antibodies to HSV-1 who had sufficient vitamin D levels. MS/CIS status did not appear to be a confounder in the association between vitamin D levels and HSV-1 antibody levels.
Among HSV-2-positive patients, vitamin D levels were weakly associated with HSV-2 antibody levels when vitamin D was a continuous, but not a binary, variable (coefficient for 1 ng/mL greater vitamin D level= 0.03, 95% CI [0.00, 0.07], p=0.08); non-white race also tended to be associated with higher antibody levels (). Adding the day of the year on which the blood was drawn was not associated with a meaningful change in the estimates. MS/CIS status did not appear to confound the association of vitamin D and HSV-2 antibody levels, although in the models that had vitamin D as a continuous variable, there was evidence for a possible interaction (p=0.12 for interaction term). Among those with MS/CIS, each 1 ng/mL greater vitamin D level was associated with a higher HSV-2 antibody levels (coefficient 0.05, 95% CI [0.01, 0.09], p=0.030), while vitamin D did not appear to be associated with the antibody levels in controls. Such an interaction was not apparent when vitamin D was treated as a binary covariate.