Association of a history of varicella virus infection with multiple sclerosis

1. Introduction 

Multiple sclerosis (MS) prevalence has increased in Latin America [1], in Mexico the frequency doubled from 1983 to 1993 [2] and in the 1990s MS was the second neurological cause for hospitalization [3]. An epidemiological study stated that there was insufficient data to appreciate the actual MS situation in Mexico since the reported prevalence is 1.5–13 cases per 100,000 [4], [5]. Viruses have been proposed as etiological factors and possible mechanisms have been studied in animal models and clinical research in humans. It is thought that an environmental viral-antigenic event in some MS-carriers could change the trait into the disease. This event could be an infection, which need not be symptomatic, or a vaccination [6], [7], [8].

The Herpesviridae family is most frequently implicated [9], one relevant characteristic of this group being its ability to remain latent. Viruses from this family with a possible MS association include herpes simplex virus (HSV), varicella zoster virus (VZV), human herpes virus 6 (HHV-6), Epstein-Barr virus (EBV) and cytomegalovirus (CMV). This association was first studied in the 1980s. Initially a low interferon production in MS patients was found by stimulating lymphocytes with HSV, CMV and VZV [10] and a higher frequency of HVS, HHV-6 and VZV was seen in active and inactive demyelinating plaques when compared with normal brain tissue [11]. An association between the herpes viruses (HV) and immunological alterations in MS patients has been described. PCR studies have revealed the presence of HSV [12], HHV-6 [13], [14], [15], VZV [16], [17], [18], [19], and EBV [20], [21], [22] DNA in MS patients. Also, epidemiological surveys have shown that VZV and MS are more prevalent in temperate zones compared with countries closer to the equator, while migration studies have also assessed the exposure to a non-identified environmental factor before age 14 that increases the risk of developing MS [16]. In high risk areas for varicella and MS, 95% of the population develops varicella before age 10 [23]. In a six-patient case report the appearance of clinical varicella or zoster was coincidental with the diagnosis of MS or with a relapse [24]. A more recent study investigated the presence of VZV DNA on relapses and remissions reporting its presence on 95% of cases during relapses and only in 17% during remissions [25]. VZV activation may be either an epiphenomenon or an active participant in MS etiopathogenesis [26], [27]. On the other hand, a systematic review of 40 studies from 1965 to 1999 found insufficient evidence to support an important etiological role of VZV infection in the etiopathogenesis of MS, as most of the studies lacked appropriate methodology [28].

The objective of this study is to investigate the relationship between a history of VZV infection and MS focusing on its different subtypes.

2. Material and methods 

We carried out an observational study using a case–control design. Patients included were 15 years or older, with a diagnosis of MS according to McDonald criteria [29] attending the National Institute of Neurology and Neurosurgery in Mexico City. Subjects were divided into relapsing-remitting multiple sclerosis (RRMS), primary progressive multiple sclerosis (PPMS), or secondary progressive multiple sclerosis (SPMS) types [30]. The Expanded Disability Status Scale (EDSS) was used to measure the level of impairment among subjects [31]. The control group was formed by 90 healthy subjects and patients with a neurological illness other than MS, including 25 with tension-type headache, 34 with neurocysticercosis and 8 subjects with epilepsy. A research questionnaire was applied in both MS patients and control subjects and answers were cross-checked with one or both parents particularly in areas such as breastfeeding and childhood exanthematic diseases. Such diseases were assessed by specific questions regarding the rash morphology and evolution. Representative images of exanthematic diseases were shown and a positive lesion identification was required in order to minimize recall bias and reporting errors were inherent to the study design. Subjects and the neurologist applying the questionnaire were blinded to the specific risk factors and outcomes were studied. Randomly selected patients and controls underwent an IgG anti-VZV antibodies detection test.

3. Statistical analysis 

Parametric and non-parametric tests were done according to each variable distribution by Levene test. The associations between MS with a history of varicella, duration of maternal breastfeeding and allergies were calculated using odds ratios (OR). Confidence intervals were set at 95% and χ2-tests performed to determine statistical significance. On post hoc analysis, one-way ANOVA was performed to find the differences among MS subtypes. Analysis was performed using SPSS v15 software for Windows.

4. Results 

126 cases were included and classified as follows: 65 relapsing-remitting, 23 primary progressive, 38 secondary progressive, for study purposes SPMS and RRMS were analyzed as a single group. The control group included 157 subjects. No statistical difference was found between healthy and neurological disease controls for variables of interest.

Demographic characteristics are shown in Table 1; MS patients included 69% females and 31% males while controls were 52% females and 48% males. Mean age for MS patients was 36±10 years and 37±12 for the control group, age among groups was similar. History of varicella infection was positive in 66 (42%) controls and in 84 (67%) MS patients (OR 2.72, CI 95% 1.69–4.38, p≤0.001). RRMS–SPMS group demographics were 71% females and 29% males, mean age of disease onset was 27±9 years; mean EDSS score was 3.12±2.3. Initial symptoms were motor in 29% cases, myelitis in 20% and optic neuritis in 19%, sensitive in 16% and cerebellar in 16%. For the PPMS group 57% were female, 43% male with a mean age of onset of 30±11 years, EDSS score was 6.1±1.2, debut symptoms were cerebellar in 48% and motor in 30%. When analyzing all the data by MS subtype, we found that a history of VZV infection was statistically significant in the RR and SP subtypes but not in the other subtypes (Table 2, Table 3). When patients in the RR and SP groups were analyzed, we found a significant difference between this groups (OR 3.08, CI 95% 1.14–8.27, p=0.023). We performed an IgG anti-VZV antibodies detection test 88 randomly selected subjects and 88 controls. Seroprevalence was 76% for MS subjects and 70% for controls. The positive predictive value in MS subjects was 76% but negative predictive value was poor at 33%. In controls PPV was 70% and NPV was 30%.

Table 1. Comparative analysis between MS patients and controls.

	Variable	MS (n=126)	Control group (n=157)	p
	Female/malea	87 (69%)/39 (31%)	81 (52%)/76 (48%)	0.01
	Ageb	36 (±10)	37 (±12)	0.29
	Prior varicella infectionb	84 (67%)	66 (42%)	0.001
	Age at infectionb	9 (±5)	7 (±5)	0.12

a χ2-test. b Unpaired T-test.

Table 2. Comparative analysis between MS subtypes and controls.

	Variable	RRMS–SPMS (n=103)	PPMS (n=23)	Control group (n=157)	p
	Female/malea	73 (71%)/30 (29%)	13 (57%)/10 (43%)	70 (52.2%)/64 (47.8%)	0.013
	Ageb	34 (±10)	38±10	34 (±9)	0.232
	Age VZ infectionb	9 (±5)	8.3±5	7 (±5)	0.267

a χ2 with Bonferroni correction. b One-way ANOVA.

Table 3. Association between MS subtypes and presence of prior VZ infection.

	MS subtype	Variable	OR	CI 95%	p
	MS cases	Prior VZ infection	2.72	1.69–4.38	<0.001
	RRMS	Prior VZ infection	3.89	2.05–7.36	<0.001
	SPMS	Prior VZ infection	2.98	1.4–6.3	0.003
	PPMS	Prior VZ infection	1.26	0.52–3.03	0.97

5. Discussion 

The association between prior varicella infection and MS was three times higher among MS patients as compared with controls. It should be pointed out that controls were not paired by gender. This is an important drawback since there is increasing evidence that MS female-to-male gender ratio is 3:1 [32], nevertheless our patients had a 2.3:1 ratio while other clinical characteristics remained similar to other MS populations reported. On the other hand, varicella infection prevalence is the same on both genders. Self-reported history of varicella has been proved to have a high positive predictive value (PPV) up to 94.3%, yet a low negative predictive value (NPP) of 67.1%. Thus, a positive VZV infection history is a strong predictor of the presence of VZV antibodies while a negative history is poorly predictive for serological status [33], [34]. In this study the seroprevalence was about 73%, lower than the reported 87% prevalence in our country [35], a fact that we could not explain.

When assessing the magnitude of this association we found that the risk remained equal regardless of age, gender and breastfeeding duration. This points to the stability of this association. The presence of prior VZV infection and MS subtype was more evident with the RR subtype, displaying a fourfold risk while being threefold in the SP subtype. This association was not found in the PPMS, possibly as a result of the small sample size. By dividing the group with prior varicella infection in age subgroups (1–4 years; 5 and 6 years; 7 and 8 years; older than 9 years) we found an association 2.5 times higher among the last two subgroups suggesting that age of onset of VZV infection is relevant. When dichotomized according to the age of infection (onset before and over age 7) the association was two times higher among the “over age 7” group. VZV prophylactic immunization could have an impact on MS incidence. Unfortunately, in our study population only one patient had received the vaccination, so we cannot make any conclusions regarding this.

An increase in incidence in our country of both MS and varicella has been recently reported [36]. Therefore, there appears to be an association between varicella infection in childhood and the development of MS; our results support the possibility that such association could have a role as an etiological factor for MS. Prospective studies are needed to confirm this.