The association between ketoacidosis and 25(OH)-vitamin D3 levels at presentation in children with type 1 diabetes mellitus


Dr Tony Huynh
Department of Paediatric Endocrinology and Diabetes
Mater Children’s Hospital
Raymond Terrace
South Brisbane
Queensland 4101
Tel: (07) 3163 8111;
fax: (07) 3163 1744;


Background:  There is considerable evidence supporting the role of vitamin D deficiency in the pathogenesis of type 1 diabetes mellitus (T1DM). Vitamin D deficiency is also associated with impairment of insulin synthesis and secretion. There have been no formal studies looking at the relationship between 25(OH)-vitamin D3 and the severity of diabetic ketoacidosis (DKA) in children at presentation with T1DM.

Objective:  To determine the relationship between measured 25(OH)-vitamin D3 levels and the degree of acidosis in children at diagnosis with T1DM.

Subjects:  Children presenting with new-onset T1DM at a tertiary children’s hospital.

Methods:  25(OH)-vitamin D3 and bicarbonate levels were measured in children at presentation with newly diagnosed T1DM. Those with suboptimal 25(OH)-vitamin D3 levels (<50 nmol/L) had repeat measurements performed without interim vitamin D supplementation.

Results:  Fourteen of the 64 children had low 25(OH)-vitamin D3 levels at presentation, and 12 of these had low bicarbonate levels (<18 mmol/L) (p = 0.001). Bicarbonate explained 20% of the variation in vitamin D level at presentation (partial r2 = 0.20, p < 0.001) and ethnic background a further 10% (partial r2 = 0.10, p = 0.002). The levels of 25(OH)-vitamin D3 increased in 10 of the 11 children with resolution of the acidosis.

Conclusions:  Acid–base status should be considered when interpreting 25(OH)-vitamin D3 levels in patients with recently diagnosed T1DM. Acidosis may alter vitamin D metabolism, or alternatively, low vitamin D may contribute to a child’s risk of presenting with DKA.

The contemporary model of the pathogenesis and natural history of type 1 diabetes incorporates genetic susceptibility, environmental triggers and immune dysregulation in the autoimmune destruction of pancreatic islet cells (1). The prevalence of type 1 diabetes mellitus (T1DM) in childhood is increasing (2) with a worldwide annual increase estimated at 3% (range 2–5%). The greatest rise in the prevalence of T1DM has been reported in the below 4 yr age group (3). It is unlikely that this observed increase in the prevalence of T1DM is because of alterations in genetic susceptibility but, rather, suggests gene–environment interactions that increase a child’s risk of developing T1DM.

The prevalence of vitamin D deficiency in Australian children is also increasing (4); however, the reasons for this are unclear and are likely to be multifactorial. Decreased sun exposure has been suggested as a possible explanation for the increased prevalence of vitamin D deficiency in Australian children. Vitamin D is synthesized in the skin from provitamin D3. In addition to the amount of sun exposure, other factors such as skin pigmentation can modulate the production of vitamin D by the skin (5). Although serum vitamin D levels are also dependant on dietary intake, in a sunny country like Australia, moderate sun exposure alone should prevent vitamin D deficiency.

The parallel rise in incidence of both T1DM and vitamin D deficiency raises the possibility that vitamin D may play a role in the pathogenesis of T1DM (6). Evidence from basic, clinical and epidemiological studies provides a rationale for this hypothesis. Receptors for 1,25(OH)2-vitamin D3 are expressed in antigen-presenting cells and T-cells (7–9) as well as in pancreatic beta cells. Observations in animal models of diabetes and human studies have implicated vitamin D deficiency in the impairment of insulin synthesis and secretion (10–12), while vitamin D supplementation has been demonstrated to attenuate cytokine-mediated pancreatic beta-cell destruction (13). Other studies investigating vitamin D status during pregnancy or various stages of childhood have suggested a role for vitamin D in reducing the risk of developing T1DM (14–16). Vitamin D receptor genotype also appears to be important in determining an individual’s susceptibility to develop T1DM (17–19). Moreover, the reported inverse correlation between sunlight exposure and T1DM incidence is consistent with the hypothesis that vitamin D status modulates disease susceptibility (21–23). In addition, children and adolescents with T1DM were found to be three times more likely to have low levels of 25(OH)-vitamin D3 (with no difference in 1,25-dihydroxyvitamin D levels) when compared with historical controls (20), with this difference being already present at diagnosis (21, 22).

To date, there have been no studies looking at the vitamin D status of children with T1DM at diagnosis with particular reference to the relationship between the degree of metabolic acidosis and the 25(OH)- vitamin D3 level. The effects of acidosis on vitamin D metabolism have been investigated previously but predominantly within the context of metabolic acidosis associated with renal failure (23).

Materials and methods


The study was an observational study conducted between the period of July 2006 and December 2007. The subjects were children and adolescents presenting with new-onset T1DM, confirmed with elevated islet antigen-2 and glutamic acid decarboxylase autoantibodies, at the Mater Children’s Hospital, Brisbane, Australia. Assessment and management of T1DM were according to hospital-based protocols including initial measurement of pH, bicarbonate, electrolytes, thyroid function tests, lipid profile, coeliac screen, islet cell antibody status, C-peptide, haemoglobin A1c, calcium and phosphate. From July 2006, routine assessment of serum 25(OH)-vitamin D3 levels was incorporated into the biochemistry panel of children presenting to Mater Children’s Hospital with newly diagnosed T1DM. The vitamin D level was classified as low if <50 nmol/L. Children with an initial 25(OH)-vitamin D3 level of <50 nmol/L had a repeat 25(OH)-vitamin D3 level measured, and vitamin D supplementation was commenced in those with confirmed vitamin D deficiency. Metabolic acidosis was defined as a bicarbonate concentration <18 mmol/L. Weight (through standard scales, kg), height (through Harpenden stadiometer, cm) and body mass index (BMI) [standard formula: weight (kg)/height (m2)] data were collected at the first new patient clinic within 1 wk of discharge from hospital. The study was approved by the Human Research Ethics Committee at the Mater Children’s Hospital, and informed consent for collection of 25(OH)-vitamin D3 levels was obtained from the study participants.


pH and ionized calcium were measured using the Radiometer-ABL 725 (Radiometer, Copenhagen, Denmark). Phosphate and bicarbonate were measured using Vitros 5, 1 FS Chemistry System (Ortho-Clinical Diagnostics, Raritan, NJ, USA). 25(OH)-vitamin D3 levels were measured using the chemiluminescence assay method by DiaSorin LIASON 25(OH)-vitamin D3 (DiaSorin Corporation, Stillwater, MN, USA).

Statistical analysis

Results are presented as mean ± standard deviation except for the β-coefficients in the multiple regression, which are presented as standard error of the mean. Categorical data were compared using the chi-squared test or Fisher’s exact test if cell frequencies were <5. T-tests were used to compare means for continuous variables. Pearson’s correlation coefficient was used, where appropriate, to examine relationships between continuous variables. Multiple regression analysis was used to evaluate the effect of age, sex, ethnic background and bicarbonate concentration on serum vitamin D level. Sex and ethnic background were coded 0 and 1 for male or female and for Caucasian or non-Caucasian, respectively. The non-Caucasian group included subjects of Asian, Middle Eastern and South Pacific Island origin. A p value of <0.05 was considered statistically significant. All analyses were two tailed and performed with spss 11.5 statistical package (SPS Inc., Chicago, IL, USA).


A total of 110 children were diagnosed with T1DM over the study period, and the 64 children with initial measurements of 25(OH)-vitamin D3 were included in this study. There were no significant differences between the children included in the study and those excluded (because of a lack of 25(OH)-vitamin D3 levels) with regard to age, sex, ethnicity, BMI z-scores and bicarbonate levels (Table 1).

Table 1.  Comparison of children included in study and children excluded from study
 Children included (n = 64)Children excluded (n = 46)p Value
  1. BMI, body mass index; F, female; M, male; SEM, standard error of the mean.

Sex (M/F)33/3123/230.87
Age (yr) (mean ± SEM)9.78 ± 0.559.94 ± 0.650.85
BMI z-scores (mean ± SEM)0.23 ± 0.140.19 ± 0.20.86
Bicarbonate (mmol/L) (mean ± SEM)17.5 ± 1.017.5 ± 1.00.98

Of the 64 children included in the study, 14 children (mean age 8.6 yr, range 7 months–15 yr) had 25(OH)-vitamin D3 levels less than 50 nmol/L. Only 2 of these 14 children did not have clinically significant acidosis at presentation. Twenty-eight children had acidosis on presentation. Twelve of these 28 children (42.8%) had low vitamin D. Only 2 of the 36 children (5.6%) presenting with a bicarbonate greater than 18 mmol/L demonstrated a low serum 25(OH)-vitamin D3 level.

Low serum 25(OH)-vitamin D3 levels were significantly associated with acidosis (Table 2). The 25(OH)-vitamin D3 level at diagnosis was correlated with bicarbonate (r = 0.44, p < 0.001, n = 64), base excess (r = 0.52, p < 0.001, n = 59) and pH (r = 0.47, p < 0.001, n = 61) (Fig. 1A–C) but was not correlated with age (r = 0.07, p = 0.57, n = 64) or BMI z-score (r = −0.11, p = 0.93, n = 60). In a multiple regression analysis incorporating bicarbonate, age, sex, BMI z-score and ethnicity, bicarbonate (β = 1.64 ± 0.37, p = <0.001) and ethnicity (β = −34.1 ± 10.8, p = 0.002) were the only independent explanatory variables for 25(OH)-vitamin D3. Bicarbonate concentration (partial r2 = 0.20) explained 20% of the variation in 25(OH)-vitamin D3 and ethnicity (partial r2 = 0.10) a further 10%; r2 for the final model was 0.30.

Table 2.  Characteristics of children with low 25(OH)-vitamin D3 levels compared with normal 25(OH)-vitamin D3 levels
 25(OH)D3 low (n = 14)25(OH)D3 normal (n = 50)p Value
  • 25(OH)D3, 25(OH)-vitamin D3; BMI, body mass index; SD, standard deviation.

  • *

    Total numbers were 13 in the low 25(OH)D3 group and 47 in the normal 25(OH)D3 group.

Age (yr) (mean ± SD)8.6 ± 5.310.1 ± 4.10.27
Sex (M/F)7/726/240.90
Ethnicity (Caucasian/non-Caucasian)11/348/20.07
BMI z-score*0.25 ± 0.960.23 ± 1.10.96
Low bicarbonate (n)12160.001
Bicarbonate (mmol/L) (mean ± SD)10.5 ± 7.219.5 ± 6.8<0.001
Figure 1.

(A) Bicarbonate vs. 25(OH)-vitamin D3, (B) base excess vs. 25(OH)-vitamin D3 and (C) pH vs. 25(OH)-vitamin D3 in new presentations for T1DM. T1DM, type 1 diabetes mellitus.

Follow-up measurements of 13 of the 14 patients with an initially low 25(OH)-vitamin D3 level showed an increase of the vitamin D level to above 50 nmol/L in 10 of the 13 children. Ten of the 11 children who had both acidosis and a low vitamin D level at presentation had a vitamin D level greater than 50 nmol/L on retesting. The exception was patient 13. Her repeat 25(OH)-vitamin D3 level performed 210 d after diagnosis was 19 nmol/L (Table 3). This particular patient, for religious reasons, received minimal sunlight exposure. It was thought that this patient had vitamin D deficiency, and supplementation was subsequently commenced. Patients 5 and 10, neither of whom had acidosis on presentation, had repeat 25(OH)-vitamin D3 levels performed at 42 d. Although the vitamin D level increased slightly in both patients, the second measurement remained below 50 nmol/L and they were therefore commenced on vitamin D supplementation. Patient 5 had a mixed Asian/Caucasian background, while patient 10 had a Pacific Islander background.

Table 3.  Follow-up of children with initial low 25(OH)-vitamin D3 levels
 Age at diagnosis (yr)EthnicitySexBicarbonate level (mmol/L)Initial 25(OH)D3 level (nmol/L)Time period of repeat 25(OH)D3 level (d)*Repeat 25(OH)D3 level (nmol/L)
  • 25(OH)D3, 25(OH)-vitamin D3; F, female; M, male.

  • *

    Time rounded up to the nearest week (expressed in days).

1013.8Pacific IslanderF24224238


This study describes the relationship between serum 25(OH)-vitamin D3 and acid–base status in children presenting with T1DM. Eighty-five percent of children with a low 25(OH)-vitamin D3 level at diagnosis had a bicarbonate less than 18 mmol/L, and the serum bicarbonate alone explained 20% of the variation in serum 25(OH)-vitamin D3. Moreover, resolution of acidosis was associated with normalization of the vitamin D level in 10 of the 11 children who had both acidosis and a low vitamin D level at presentation.

Measurement of 25(OH)-vitamin D3 levels in young adults at diagnosis of T1DM (21) in one study demonstrated a mean level within the cohort that was lower than control subjects. This observation was particularly obvious among the male patients and was not attributable to seasonal fluctuations. The authors postulated that the lower levels of 25(OH)-vitamin D3 may have contributed to the development of T1DM. The timing of the samples at diagnosis was not specified, and the relationship between 25(OH)-vitamin D3 levels and the degree of acidosis was not reported. In another study conducted on an Italian cohort (22), 25(OH)-vitamin D3 and 1,25(OH)2-vitamin D3 were measured in patients with newly diagnosed T1DM and compared with a group of age-matched and sex-matched controls. These measurements were conducted approximately 1 wk after diagnosis when ketoacidosis had resolved. This study design excluded acute metabolic acidosis as a factor that could influence 25(OH)-vitamin D3 and 1,25(OH)2-vitamin D3 levels. The authors found significantly reduced 25(OH)-vitamin D3 and 1,25(OH)2-vitamin D3 levels in the patients with T1DM when compared with the control group. In a previous study conducted at our centre (20), 25(OH)-vitamin D3 and 1,25(OH)2-vitamin D3 levels were measured in stored frozen samples of serum from children and adolescents with T1DM at various periods after diagnosis. Patients with T1DM had a lower level of 25(OH)-vitamin D3 but not of 1,25(OH)2-vitamin D3 when compared with controls. Forty-three percent of children with T1DM in that study had 25(OH)-vitamin D3 levels <50 nmol/L compared with 22% in this study. This difference may be explained by a number of factors including the timing of the sample in relation to the initial presentation, the relative number of patients presenting with DKA, temporal changes in sun avoidance behaviours, effects of storage at −70°C on 25(OH)-vitamin D3, and differences in assay methodology.

Serum 25(OH)-vitamin D3, levels are dependant largely on sun exposure in countries such as Australia; however, a number of factors may modulate skin vitamin D production during sun exposure. The area of skin exposed, the wavelength of the light, filtering of sunlight by glass, skin pigmentation, clothing and the use of sun screen are all factors that can affect vitamin D production (5). We confirmed in our multiple regression that ethnic background is independently associated with the 25(OH)-vitamin D3 level. In addition, all the three patients in whom 25(OH)-vitamin D3 did not normalize came from a non-Caucasian background. It is possible that the persistently low 25(OH)-vitamin D3 levels in the two children who were not acidotic at diagnosis were in part because of increased skin pigmentation, which limited their skin vitamin D production. The third child was the only patient whose 25(OH)-vitamin D3 level did not normalize after the ketoacidosis resolved. Skin production of vitamin D in this patient was limited by decreased sun exposure because of cultural and religious reasons.

Approximately 40% of the children who presented with TIDM between July 2006 and December 2007 could not be included in the study because a vitamin D level was not obtained at diagnosis. The possibility of a selection bias therefore cannot be excluded. However, the characteristics of the study group and those not included were comparable with regard to age, sex, BMI and bicarbonate. Therefore, the subgroup of 64 children who had vitamin D levels measured appear to have been representative of the total population of children who presented to the hospital with TIDM.

The association between reduced vitamin D levels and acidosis may result from direct effects of acidosis on vitamin D physiology. A number of studies have demonstrated impaired 1-alpha-hydroxylase activity and conversion of 25(OH)-vitamin D3 to 1,25(OH)2-vitamin D3 with induced chronic metabolic acidosis (24–26), while others have shown increased levels of 1,25(OH)2-vitamin D3 with chronic metabolic acidosis (27, 28). It is also possible that the acidotic state results in a reduction of vitamin D-binding proteins and a corresponding reduction in measured 25(OH)-vitamin D3 levels in accordance with the Michaelis–Menten equilibrium. A direct negative effect of acidosis on 25(OH)-vitamin D3 levels is consistent with the normalization of 25(OH) vitamin D3 levels with resolution of the acidosis in 10 of the 11 children who were followed up. This normalization of the 25(OH)-vitamin D3 levels occurred without any supplementation or advice regarding increasing sun exposure.

Alternatively, reduced vitamin D levels may contribute to an individual’s risk of presenting with DKA either directly by impairing insulin secretion and action (29) or indirectly by altering innate immunity. Adequate vitamin D levels protect against viral and bacterial infections (30), which are a common precipitant of DKA. There were not sufficient patient numbers in this study to compare infection rates in children presenting with and without DKA. Although pre-morbid 25(OH)-vitamin D3 levels were not available in these children, the fact that 25(OH)-vitamin D3 levels spontaneously normalized in 10 of the 11 children with resolution of the acidosis argues however against a permanent state of 25(OH)-vitamin D3 deficiency.

The possibility of an association between measured 25(OH)-vitamin D3 and the degree of ketosis was not studied specifically in this study. It has been previously reported that, in children with epilepsy, a ketogenic diet reduces serum 25(OH)-vitamin D3 and calcium concentrations while increasing serum alkaline phosphatase and parathyroid hormone concentrations (31). Similar biochemical changes were found in children treated with anticonvulsants alone, but the combined ketogenic diet/anticonvulsant regime caused greater bone mass decline.

Our study suggests that acid–base status, along with other factors, should be considered when interpreting 25(OH)-vitamin D3 levels in patients with recently diagnosed T1DM. Lower measured 25(OH)-vitamin D3 levels are associated with acidaemia in children with new-onset T1DM and generally increase when the acidosis resolves. A causal relationship cannot be deduced from this study. Acidosis may alter vitamin D metabolism through various mechanisms, or alternatively, low vitamin D may have a contributory role in determining whether a child with T1DM presents with DKA at diagnosis. Further studies are required to explore these possibilities.