Low vitamin E status is a potential risk factor for insulin-dependent diabetes mellitus


Dr Paul Knekt National Public Health Institute, Mannerheimintie 166, 00300 Helsinki, Finland (fax: +358 9 4744 8760).


Objectives. To study the association of vitamin E status with occurrence of insulin-dependent diabetes mellitus (IDDM).

Design. A case–control study nested within a 21-year follow-up study.

Subjects. Nineteen incident IDDM patients with an average age of 28 years and three individually matched controls per patient.

Main outcome measure. Serum concentrations of alpha-tocopherol.

Results. Serum alpha-tocopherol concentration at the baseline examination was inversely associated with IDDM occurring 4–14 years later. The cholesterol-adjusted relative risk of IDDM between the highest and lowest thirds of the vitamin concentration was 0.12 (95% confidence interval = 0.02– 0.85).

Conclusions. The finding corroborates the hypothesis of a protective effect of vitamin E against development of IDDM. Because of the relatively old age of the patients in the present population, further epidemiological studies on the topic are warranted.


Autoimmune destruction of pancreatic beta cells is the basic process leading to insulin-dependent diabetes mellitus (IDDM). Predisposition to IDDM is genetically determined and several genetic markers enhancing or protecting against the disease have been identified. Although the details in the pathogenesis of the autoimmune process leading to IDDM are relatively well characterized [ 1], the possible role of environmental factors in initiating or modifying the process is poorly understood.

It is apparent that oxygen-derived free radicals enhance the autoimmune destruction of the beta cells [ 2–4]. Dietary nitrites and nitrates are also shown to be potential environmental risk factors for IDDM [ 5, 6] and their effect is thought to be mediated by the formation of nitroso compounds shown to be toxic for beta cells [ 7]. Vitamin E is one of the most important exogenic free radical scavengers [ 8] and also effectively blocks the formation of nitroso compounds [ 9]. Thus it is plausible to hypothesize that a sufficient vitamin E status plays an important protective role in the pathogenesis of IDDM. However, no studies on the effect of vitamin E or any other antioxidant nutrients on the incidence of IDDM in humans are available.

The aim of this study is to examine the predictive value of serum alpha-tocopherol on the incidence of IDDM within a large prospective population study.


The Finnish Mobile Clinic carried out health examinations in 12 municipalities in different parts of Finland in 1973–76 [ 10]. Altogether 7526 men aged 20 years and older and free from IDDM participated. A questionnaire with items concerning smoking habits was completed. As a part of the survey, blood pressure, body height and body weight were measured, and the body mass index was calculated. Venous blood samples were drawn, serum cholesterol and fasting plasma glucose concentration were determined, and the serum samples were stored at −20 °C. The serum cholesterol concentration was determined using an autoanalyser modification (Technicon Autoanalyser Methodology N-77, Tarrytown, NY, USA) of the Liebermann–Burchard reaction [ 11], and the concentration of fasting glucose in the plasma was determined by an autoanalyser modification (Technicon Autoanalyser Methodology N-2b) of the ferricyanide reduction method [ 12].

According to the national Sickness Insurance Act, drugs are provided free of charge for diabetes mellitus patients in Finland [ 13]. To receive reimbursement, the patient submits an application together with a physician's certificate concerning diagnostic details and progress of the disease. A national registry of all recipients of drug allowances is continuously kept by the Social Insurance Institution. The coverage of this registry is nearly complete [ 14, 15]. The definition of IDDM in the present study was based on the details from the physician's certificate. The diabetic patient was considered to have IDDM if the typical symptoms were short-lasting, body weight was in the normal range or under, ketoacidosis was evident at the time of diagnosis and initial insulin treatment could not be discontinued at any time except for the short possible ‘honeymoon’ period. In several, but not all, cases low C-peptide response results confirmed IDDM diagnosis. Islet cell autoantibodies were not routinely measured.

By linking the drug registry with the present population, 19 new IDDM cases occurred during a maximum follow-up of 21 years. The first diagnosis was made 4 years and the last 14 years after the baseline study. The mean age of the diabetic patients was 26 years (range 21–46). A case–control design nested within the cohort was used [ 16]. Three controls per diabetic patient were chosen by individual matching for sex, age and time of baseline examination. Altogether, 57 controls were included in the study.

The serum samples were thawed in 1995 for insulin, alpha-tocopherol, retinol and selenium determinations. The insulin concentration was measured with the Abbot Imx analyser (Abbot Laboratories, USA) and the selenium concentrations by the direct graphite furnace atomic absorption spectrometric method utilizing Zeeman background correction [ 17, 18]. Serum alpha-tocopherol and retinol were simultaneously determined using high-performance liquid chromatography [ 19]. The sequence of laboratory analyses did not depend on case–control status as each case and its matched controls were analysed in random order. The laboratory personnel were also unaware of the case–control status of the samples. The effect of storage of the serum samples on the alpha-tocopherol concentrations was studied by repeated analyses of 144 serum samples with an interval of 10 years. The mean alpha-tocopherol level was reduced by 22% and the reliability coefficient for overall agreement was 0.66. The determinations can thus be regarded as reliable.

The association between serum alpha-tocopherol concentration and occurrence of IDDM was estimated using the conditional logistic model [ 20]. Relative risks (estimated as odds ratios) were computed between tertiles of the vitamin distribution. Serum cholesterol and other potential confounding factors were included in the model. Statistical significances were tested with the likelihood ratio test based on the model.


Individuals developing IDDM during the follow-up had significantly higher systolic blood pressure levels and non-significantly lower serum alpha-tocopherol levels than the controls ( Table 1). Serum selenium and serum retinol were not associated with IDDM. Although there was no difference between insulin levels of cases and controls, the variation in cases was larger than in controls.

Table 1.   Baseline characteristics of the study population by case–control status Thumbnail image of

There was an inverse association between serum alpha-tocopherol concentration and subsequent occurrence of IDDM ( Table 2). The relative risk of the disease was 0.15 (95% confidence interval [CI] = 0.03–0.79) between the highest and lowest tertiles of the serum alpha-tocopherol concentration. Adjustment for serum cholesterol level did not notably alter the association. The adjusted relative risk was 0.12 (95% CI = 0.02–0.85). Different further adjustments for body mass index, smoking, systolic blood pressure, plasma glucose, and serum levels of insulin, ferritin, retinol and selenium did not notably alter the result, either (data not shown).

Table 2.   Relative risk of insulin-dependent diabetes mellitus between tertiles of serum alpha-tocopherol Thumbnail image of


We found an inverse association between serum alpha-tocopherol concentration and subsequent occurrence of IDDM. The finding corroborates the hypothesis that vitamin E is an important exogenic factor for protection against IDDM. It is supported by in vitro [ 21, 22] and animal [ 23–26] experiments, suggesting a beneficial role of antioxidants in the pathogenesis of IDDM. As far as we know, this is the first human study demonstrating that low serum alpha-tocopherol status predicts occurrence of IDDM several years later amongst originally IDDM-free individuals. Although the hypothesis is plausible and the finding is in agreement with experimental evidence, there are several methodological factors which prevent any definite conclusions about a causal connection being drawn.

First, the number of disease cases and controls was small, and thus the association observed may be due to chance. On the other hand, the upper limit of the confidence interval for the highest category of serum alpha-tocopherol was relatively low, suggesting a stable association. Second, the serum samples were stored for a long time at −20 °C, and loss of alpha-tocopherol concentrations during storage may have lowered reliability of the measurements [ 27]. Study of loss of alpha-tocopherol during storage, however, showed a satisfactory agreement between serum alpha-tocopherol measurements repeated at an interval of 10 years in the present population. Third, the low serum alpha-tocopherol level may be a consequence of latent IDDM. As the IDDM cases were diagnosed 4–14 years after the determination of serum alpha-tocopherol, it is improbable that the inverse association is due to a latent phase of IDDM, which usually lasts a considerably shorter time. Fourth, diet is a very complex combination of different substances. In the event that an association between some dietary component and subsequent serum concentration and disease occurrence can be demonstrated, it is always possible that this association is due to some other dietary factor, combination of dietary factors or lifestyle associated with dietary behaviour. In the present study, dietary data were not available and thus the effects of such confounding factors could not be adjusted for. Finally, the age at diagnosis of the disease was exceptionally high in this population and no information exists on whether the findings can be generalized to younger populations.

In conclusion, the present longitudinal study demonstrated that people with a low serum alpha-tocopherol concentration had an elevated subsequent risk of IDDM. Although this finding is plausible and in agreement with the hypothesis that antioxidant vitamins provide protection against IDDM, no conclusions about causal connection can be drawn. Results from larger observational studies are warranted.

Received 23 February 1998; accepted 9 June 1998.