Abstract
BACKGROUND:
Findings from previous studies on maternal 25-hydroxyvitamin D [25(OH)D] levels during pregnancy and autism spectrum disorder (ASD) in offspring are inconsistent.
METHODS:
The association between maternal 25(OH)D levels during pregnancy and offspring ASD was examined using data from a nationwide population-based register with a nested case-control study design. The ASD cases (n = 1558) were born between 1987 and 2004 and received a diagnosis of ASD by 2015; cases were matched with an equal number of controls. Maternal 25(OH)D levels during pregnancy were measured using quantitative immunoassay from maternal sera collected during the first and early second trimesters and archived in the national biobank of the Finnish Maternity Cohort. Conditional logistic regression examined the association between maternal 25(OH)D levels and offspring ASD.
RESULTS:
In the adjusted model, there was a significant association between increasing log-transformed maternal 25(OH)D levels and decreasing risk of offspring ASD (adjusted odds ratio [aOR] 0.75, 95% confidence interval [CI] 0.62–0.92, p = .005). Analyses by quintiles of maternal 25(OH)D levels revealed increased odds for ASD in the 2 lowest quintiles, <20 (aOR 1.36, 95% CI 1.03–1.79, p = .02) and 20–39 (aOR 1.31, 95% CI 1.01–1.70, p = .04), compared with the highest quintile. The increased risk of ASD was observed in association with deficient (<30 nmol/L) (aOR 1.44, 95% CI 1.15–1.81, p = .001) and insufficient (30–49.9 nmol/L) maternal 25(OH)D levels (aOR 1.26, 95% CI 1.04–1.52, p = .01) compared with sufficient levels.
CONCLUSIONS:
This finding has implications for understanding the role of maternal vitamin D during fetal brain development and increased risk of ASD.
Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder characterized by disrupted social interactions, impaired language, and stereotypic and repetitive behavior with different degrees of severity. The etiology of ASD is largely unknown, although both genetic and environmental factors have been suggested in the development of this disorder (1). Prior findings suggest that vitamin D deficiency during pregnancy might be associated with increased risk of ASD in offspring (2).
Adequate intake of vitamin D is vital for overall physical and mental health. The vitamin D receptor is present in most tissues and cells in the body and is broadly distributed in brain neurons, peripheral neurons, and nonneuronal brain cells. The highest concentrations are in the regions of the nervous system that are required for critical functions, including the pre-frontal cortex and hippocampus, regions strictly related to learning, memory, and executive control (3,4). In rodents, maternal vitamin D deprivation prior to mating and during pregnancy produced morphological changes in the offspring’s brain at birth (5). Children exposed to persistently deficient maternal vitamin D concentrations in utero had lower cerebral gray matter and white matter volumes (6). Increases in vitamin D receptor expression were also associated with reductions in cellular proliferation and a related increase in programmed cellular elimination (7–9).
Prior studies of the association between maternal serum vitamin D deficiency during pregnancy and ASD or autism symptoms had inconsistent findings (2,10–15). A study based on the Generation R cohort (N = 4334) that examined maternal 25-hydroxyvitamin D [25(OH)D] at midgestation and a continuous measure of autism-related traits at 6 years of age (Social Responsiveness Scale) observed elevated scores on the Social Responsiveness Scale in children with gestational vitamin D deficiency (2). Four studies examining vitamin D deficiency during pregnancy and using diagnosis of ASD as an outcome also have reported inconsistent findings (10,13–15). While three of these studies had fewer than 100 cases, the study by Lee et al. (15), based on a Swedish cohort, included 449 cases. That study found that among Nordic born mothers, maternal vitamin D deficiency was associated with ASD with intellectual disability (ID), but not ASD without ID (15).
Limitations in previous studies need to be addressed. First, as noted above, in most of the previous studies, sample sizes of cases with ASD were small, or the outcome was based on ASD symptoms. Second, most of the previous studies did not adjust for important covariates, such as prematurity, socioeconomic status (SES), paternal age, drug consumption, paternal psychiatric disorders, and immigration. Moreover, the Swedish study did not have information about prematurity, maternal SES, or any paternal information (15).
Information about the role of maternal vitamin D deficiency and the risk of ASD is important for understanding the etiology of autism. In addition, vitamin D deficiency can be effectively remedied by supplementation and is thus a potential modifiable risk factor with global significance for prevention. FiPS-A (Finnish Prenatal Study of Autism and Autism Spectrum Disorder), based on a nested case-control study design, features several strengths, including a large number of ASD cases, prospectively acquired maternal sera drawn during pregnancies with these cases, and comprehensive information from several nationwide registers facilitating adjustment for many potential confounders. Of particular importance for vitamin D exposure, the data were collected from one of the northern-most countries in the world, Finland, with scarce sun exposure during the winter and before the national recommendation for vitamin D supplementation during pregnancy started in 2004. The consequent increased vulnerability to vitamin D deficiency during the study period allows for increased power to investigate the association. The main aim of this nested case-control study was to examine maternal 25(OH)D levels in early pregnancy and diagnosed ASD cases among offspring. In addition, we evaluated the association between maternal 25(OH)D levels in early pregnancy and specific ASD subtypes.
METHODS AND MATERIALS
FiPS-A is a nested case-control study including all singleton live births in Finland from 1987 to 2004, for whom ASD diagnoses were available in the Care Register for Health Care (CRHC) by 2015. The study was approved by the Ethics Committee of the Hospital District of Southwest Finland, by the data protection authorities at the National Institute for Health and Welfare, and by the Institutional Review Board of the New York State Psychiatric Institute.
Finnish Maternity Cohort
All of the offspring in FiPS-A were derived from the Finnish Maternity Cohort (FMC), which consists of about 2 million maternal serum samples collected during the first and early second trimesters of pregnancy (5th to 95th percentile: months 2–4 of pregnancy) from more than 950,000 gravidae. Following informed consent, prenatal serum specimens were collected for routine screening for congenital infections. The remaining serum samples (approximately 1–3 mL of serum from each pregnancy) were stored at −25 °C in a protected biorepository at Biobank Borealis in Oulu, Finland, and are available for scientific research. All FMC samples were linked with other Finnish registers by the unique personal identification code assigned to all residents in Finland since 1971.
Nationwide Registers
The CRHC includes all inpatient diagnoses since 1967 and outpatient diagnoses from specialized services since 1998. All the diagnoses in the CRHC are based on ICD: ICD-8 from 1969 to 1986, ICD-9 from 1987 to 1995, and ICD-10 since 1996. Our previous validation study showed that the specificity of childhood autism in the CRHC is good (96%) (16). The Finnish Maternity Birth Register was established in Finland in 1987 and contains comprehensive data on numerous variables pertaining to pregnancy and the postpartum/neonatal periods (up to 7 days of age) of all pregnant and postpartum mothers and newborns in Finland. The Finnish Population Register Centre was established in 1969 and is responsible for usage and maintenance of basic demographic information of all residents in Finland. It includes name, personal identity code, address, citizenship, native language, family members, and date of birth and death (if applicable). The Finnish Population Register Centre was used to identify the controls and to obtain the information on the subjects’ parents and places of birth.
Information on Cases and Controls
The ASD cases were born between January 1987 and December 2004 and were registered in the CRHC with ICD-10 (F84x) and ICD-9 (299x) diagnoses before 2015. The controls were born in Finland and did not have a diagnosis of ASD or ID. Controls were matched 1:1 to case subjects by sex and date of birth (±30 days). The matched controls were required to be alive and residing in Finland at the date of diagnosis of the respective cases. Sufficient serum specimens were available in the FMC for 1558 cases and 1558 matched controls, which were randomly selected throughout the whole ASD cohort (1987–2004, N = 4698 case-control pairs). The subjects with ID were identified from CRHC with ICD codes: ICD-10 (F70, F71, F72, F73, F78, and F79) and ICD-9 (317, 318.0, 318.1, 318.2, and 319).
Maternal Serum 25(OH)D Assay
Maternal serum 25(OH)D was measured by investigators blinded to case/control status using a chemiluminescence microparticle immunoassay and an Architect i2000SR automatic analyzer (Abbott Diagnostics), which has been previously described in detail (17,18). Coefficients of variation were calculated for the sample pairs. They were derived from repeated quality control samples included in the assay with the study samples. The overall assay coefficient of variation percentage calculated from blinded quality control sample pairs with repeated measures of maternal serum 25(OH)D included in each set of analyses was 2.1%, indicating the reliability of the assay.
Covariates
Potential confounders and mediators suggested as having associations with maternal vitamin D levels and offspring ASD were initially selected (19–29). The Finnish Maternity Birth Register was used to obtain information regarding maternal smoking, number of previous births, maternal SES, gestational age, birth weight, Apgar score at 1 minute, and maternal age. Maternal immigrant status was obtained from the Finnish Population Register Centre. The season of blood collection and gestational week of blood draw was obtained from the FMC. Maternal and paternal psychiatric diagnoses were obtained from the CRHC. A detailed description of covariates is given in Table S1.
Statistical Analysis
Maternal 25(OH)D was initially examined as a continuous variable. The variable was log-transformed before analyses owing to the skewed distribution. We also examined maternal 25(OH)D categorized into quintiles. The cut points for the quintiles were based on the distribution of maternal 25(OH)D levels in the control group, with the highest quintile defined as the reference group. In addition, maternal 25(OH)D was examined as a 3-class categorical variable based on clinical categories—deficient [25(OH)D <30 nmol/L], insufficient [25(OH)D 30–49.9 nmol/L], and sufficient [25(OH)D ≥50 nmol/L] levels—with the sufficient category as the reference group.
Categorical potential confounders were tested with Student t and F tests, and continuous potential confounders were tested with linear regression for the association with log-transformed maternal 25(OH)D among controls. Potential confounders were tested for the association with ASD using conditional logistic regression models for the matched sets. The covariates were selected into the models based on their associations with both exposure and outcome at p < .1, which is in accord with standard texts (30). A χ2 test was used to test for collinearity of maternal SES and maternal smoking.
Point and interval estimates of odds ratios (ORs) were obtained by fitting conditional logistic regression models for matched pairs. Unadjusted ORs and adjusted ORs (aORs) and 95% confidence intervals (CIs) were calculated separately for ASD and by 3 subgroups: childhood autism, Asperger’s syndrome, pervasive developmental disorder (PDD)/PDD-not otherwise specified (NOS). Additional analyses were performed by stratifying ASD by the presence or absence of comorbid ID. Conditional logistic regression was used to test for interactions between continuous maternal 25(OH)D and ASD by subgroups of ASD, ID, sex, and age at diagnosis. Statistical significance was based on p < .05. All statistical analyses were performed with SAS 9.4 software (SAS Institute Inc.).
RESULTS
The study included 1558 ASD case/control pairs with mean (SD) age at first ASD diagnosis for cases of 6.58 (3.21) years (range, 0–18 years). The median maternal 25(OH)D level was 38.19 nmol/L among cases and 39.69 nmol/L among controls. The mean gestational week of maternal blood draw was 10.79 weeks (range, 4–37 weeks) for cases and 10.10 (range, 4–30 weeks) for controls. The overall sample of cases and controls was 80.49% male and 19.51% female.
Among potential covariates in the study, gestational age, season of blood collection, maternal age, and gestational week of blood draw were associated with both maternal 25(OH)D level and offspring ASD and were adjusted in the multivariable analyses (model 1). Moreover, we additionally adjusted for maternal smoking, immigration status, psychopathology, and substance abuse (model 2) (Table S1). There was a significant correlation between maternal SES and smoking (p < .001), and therefore only maternal smoking was included in the model. When maternal SES was included in the model instead of maternal smoking, the results remained the same except for the association between maternal 25(OH)D levels and PDD/PDD-NOS in the first quintile and association between maternal 25(OH)D levels and offspring ASD without ID in clinical categories (Tables S2–S5).
Table 1 shows the log-transformed maternal 25(OH)D levels and offspring ASD. In the adjusted models, a significant association was found between increasing log-transformed maternal 25(OH)D levels and decreasing risk of offspring ASD (aOR 0.75, 95% CI 0.62–0.92, p = .005). In a subsequent analysis of clinical ASD categories, maternal 25(OH)D was associated with PDD/PDD-NOS (aOR 0.67, 95% CI 0.47–0.96, p = .03), was marginally associated with childhood autism (aOR 0.73, 95% CI 0.51–1.04, p = .08), and was not associated with Asperger’s syndrome (aOR 0.86, 95% CI 0.59–1.14, p = .24). Adjustment for maternal smoking, immigration, psychopathology, and substance abuse did not change the findings (Table 1). We found no interaction by ASD subgroups for the association between continuous maternal 25(OH)D and ASD (p = .92).
Table 1.
Odds Ratios and 95% CIs of Association Between Maternal Serum Vitamin D (Continuous) and Offspring ASD and Subtypes
| Offspring ASD and Subtypes | Log-Transformed Maternal Vitamin D Levels, nmol/L | Odds Ratio, Unadjusted (95% CI) | p Value | Odds Ratio, Adjusteda (95% CI) | p Value | Odds Ratio, Adjustedb (95% CI) | p Value | |
|---|---|---|---|---|---|---|---|---|
| Cases, Mean (SD) | Controls, Mean (SD) | |||||||
| ASD (n = 1558) | 3.66 (0.45) | 3.69 (0.46) | 0.81 (0.67–0.98) | .02 | 0.75 (0.62–0.92) | .005 | 0.72 (0.59–0.89) | .002 |
| Childhood Autism (n = 490) | 3.68 (0.43) | 3.71 (0.45) | 0.80 (0.57–1.12) | .19 | 0.73 (0.51–1.04) | .08 | 0.71 (0.49–1.03) | .07 |
| Asperger’s Syndrome (n = 537) | 3.65 (0.46) | 3.67 (0.46) | 0.85 (0.63–1.17) | .33 | 0.86 (0.59–1.14) | .24 | 0.78 (0.56–1.10) | .16 |
| PDD/PDD-NOS (n = 531) | 3.65 (0.45) | 3.69 (0.47) | 0.78 (0.57–1.07) | .12 | 0.67 (0.47–0.96) | .03 | 0.63 (0.43–0.91) | .01 |
ASD, autism spectrum disorder; CI, confidence interval; PDD, pervasive developmental disorder; PDD-NOS, PDD-not otherwise specified.
Adjusted for maternal age, gestational week of blood draw, season of blood collection, and gestational age.
Adjusted for all confounders in listed above and maternal smoking, immigration, psychopathology, and substance abuse.
Table 2 shows the distribution of maternal 25(OH)D in quintiles by case-control status. The association between maternal 25(OH)D and ASD was significant for the 2 lowest quintiles, <20 (aOR 1.36, 95% CI 1.03–1.79, p = .02) and 20–39 (aOR 1.31, 95% CI 1.01–1.70, p = .04), compared with the highest quintile. When the subtypes of ASD were analyzed, the associations were significant only in the PDD/PDD-NOS group for the lowest quintiles, <20 (aOR 1.73, 95% CI 1.05–2.85, p = .03), 20–39 (aOR 1.95, 95% CI 1.22–3.11, p = .005), and 60–79 (aOR 1.58, 95% CI 1.03–2.44, p = .03), compared with the highest quintile. The additional analyses adjusting for maternal smoking, immigration, psychopathology, and substance abuse did not change the findings except for the PDD/PDD-NOS group in the 60–79 quintile.
Table 2.
Odds Ratios and 95% CIs of Association Between Maternal Serum Vitamin D (in Quintiles) and Offspring ASD and Subtypes
| Offspring ASD and Subtypes | Maternal Serum Vitamin D, nmol/L | Cases, n (%) | Controls, n (%) | Odds Ratio, Unadjusted (95% CI) | p Value | Odds Ratio, Adjusteda (95% CI) | p Value | Odds Ratio, Adjustedb (95% CI) | p Value |
|---|---|---|---|---|---|---|---|---|---|
| ASD (n = 1558) | <20 | 335 (21.50%) | 315 (20.22%) | 1.25 (0.96–1.6) | .08 | 1.36 (1.03–1.79) | .02 | 1.40 (1.05–1.87) | .01 |
| 20–39 | 331 (21.25%) | 312 (20.03%) | 1.23 (0.96–1.57) | .10 | 1.31 (1.01–1.70) | .04 | 1.33 (1.02–1.75) | .03 | |
| 40–59 | 321 (20.60%) | 305 (19.58%) | 1.20 (0.94–1.54) | .13 | 1.25 (0.96–1.61) | .08 | 1.22 (0.93–1.59) | .14 | |
| 60–79 | 290 (18.61%) | 318 (20.41%) | 1.02 (0.80–1.29) | .87 | 1.01 (0.79–1.30) | .91 | 0.97 (0.75–1.26) | .86 | |
| ≥80 | 281 (18.04%) | 308 (19.77%) | Reference | Reference | Reference | ||||
| Childhood Autism (n = 490) | <20 | 99 (20.20%) | 91 (18.57%) | 1.12 (0.71–1.77) | .61 | 1.25 (0.76–2.05) | .36 | 1.37 (0.81–2.29) | .23 |
| 20–39 | 104 (21.22%) | 103 (21.02%) | 1.04 (0.66–1.62) | .85 | 1.11 (0.69–1.78) | .65 | 1.10 (0.67–1.79) | .68 | |
| 40–59 | 102 (20.82%) | 93 (18.98%) | 1.10 (0.71–1.71) | .64 | 1.16 (0.73–1.83) | .51 | 1.21 (0.75–1.94) | .42 | |
| 60–79 | 93 (18.98%) | 113 (23.06%) | 0.79 (0.52–1.21) | .29 | 0.78 (1.07–3.40) | .27 | 0.82 (0.52–1.29) | .39 | |
| ≥80 | 92 (18.78%) | 90 (18.37%) | Reference | Reference | Reference | ||||
| Asperger’s Syndrome (n = 537) | <20 | 121 (22.53%) | 113 (21.04%) | 1.16 (0.74–1.79) | .50 | 1.20 (0.76–1.90) | .41 | 1.24 (0.77–2.0) | .35 |
| 20–39 | 112 (20.86%) | 108 (20.11%) | 1.09 (0.71–1.68) | .66 | 1.14 (0.73–1.77) | .56 | 1.09 (0.69–1.74) | .69 | |
| 40–59 | 114 (21.23%) | 104 (19.37%) | 1.17 (0.76–1.79) | .45 | 1.17 (0.76–1.81) | .46 | 1.15 (0.73–1.80) | .54 | |
| 60–79 | 91 (16.95%) | 110 (20.48%) | 0.83 (0.53–1.29) | .41 | 0.85 (0.54–1.33) | .47 | 0.81 (0.51–1.28) | .37 | |
| ≥80 | 99 (18.44%) | 102 (18.99%) | Reference | Reference | Reference | ||||
| PDD/PDD-NOS (n = 531) | <20 | 115 (21.66%) | 111 (20.90%) | 1.47 (0.94–2.31) | .09 | 1.73 (1.05–2.85) | .03 | 1.71 (1.009–2.91) | .04 |
| 20–39 | 115 (21.66%) | 101 (19.02%) | 1.60 (1.04–2.45) | .03 | 1.95 (1.22–3.11) | .005 | 2.13 (1.30–3.49) | .002 | |
| 40–59 | 105 (19.77%) | 108 (20.34%) | 1.33 (0.87–2.03) | .17 | 1.52 (0.96–2.43) | .07 | 1.39 (0.85–2.26) | .18 | |
| 60–79 | 106 (19.96%) | 95 (17.89%) | 1.49 (0.99–2.23) | .05 | 1.58 (1.03–2.44) | .03 | 1.44 (0.91–2.27) | .11 | |
| ≥80 | 90 (16.95%) | 116 (21.85%) | Reference | Reference | Reference |
ASD, autism spectrum disorder; CI, confidence interval; PDD, pervasive developmental disorder; PDD-NOS, PDD-not otherwise specified.
Adjusted for maternal age, gestational week of blood draw, season of blood collection, and gestational age.
Adjusted for all confounders listed above and maternal smoking, immigration, psychopathology, and substance abuse.
When maternal 25(OH)D levels were classified into 3 clinical categories, an increased odds of ASD were observed for the deficient (<30 nmol/L) (aOR 1.44, 95% CI 1.15–1.81, p = .001) and insufficient (30–49.9 nmol/L) maternal 25(OH)D levels (aOR 1.26, 95% CI 1.04–1.52, p = .01) compared with the sufficient maternal 25(OH)D levels. When the subtypes of ASD were analyzed, the increased risk was observed only in the PDD/PDD-NOS group with deficient (aOR 1.68, 95% CI 1.11–2.55, p = .01) and insufficient (aOR 1.44, 95% CI 1.02–2.01, p = .03) maternal 25(OH)D categories compared with the sufficient maternal 25(OH)D category. Adjustment for maternal smoking, immigration, psychopathology, and substance abuse did not change the findings (Table 3).
Table 3.
Odds Ratios and 95% CIs of Association Between Maternal Serum Vitamin D (in Categories) and Offspring ASD and Subtypes
| Offspring ASD and Subtypes | Maternal Serum Vitamin D, nmol/L | Cases, n (%) | Controls, n (%) | Odds Ratio, Unadjusted (95% CI) | p Value | Odds Ratio, Adjusteda (95% CI) | p Value | Odds Ratio, Adjustedb (95% CI) | p Value |
|---|---|---|---|---|---|---|---|---|---|
| ASD (n = 1558) | <30 | 473 (30.36%) | 431 (27.66%) | 1.34 (1.08–1.65) | .006 | 1.44 (1.15–1.81) | .001 | 1.52 (1.20–1.91) | .0004 |
| 30–49.9 | 625 (40.12%) | 606 (38.90%) | 1.21 (1.01–1.45) | .03 | 1.26 (1.04–1.52) | .01 | 1.28 (1.05–1.55) | .01 | |
| ≥50 | 460 (29.53%) | 521 (33.44%) | Reference | Reference | Reference | ||||
| Childhood Autism (n = 490) | <30 | 142 (28.98%) | 130 (26.53%) | 1.28 (0.88–1.87) | .19 | 1.39 (0.93–2.08) | .10 | 1.46 (0.96–2.21) | .07 |
| 30–49.9 | 195 (39.80%) | 192 (39.18%) | 1.16 (0.84–1.61) | .35 | 1.22 (0.87–1.71) | .24 | 1.24 (0.87–1.76) | .22 | |
| ≥50 | 153 (31.22%) | 168 (34.29%) | Reference | Reference | Reference | ||||
| Asperger’s Syndrome (n = 537) | <30 | 170 (31.66%) | 156 (29.05%) | 1.29 (0.91–1.83) | .14 | 1.24 (0.93–1.93) | .24 | 1.40 (0.96–2.05) | .07 |
| 30–49.9 | 214 (39.85%) | 209 (38.92%) | 1.18 (0.87–1.60) | .28 | 1.20 (0.87–1.64) | .10 | 1.17 (0.84–1.62) | .34 | |
| ≥50 | 153 (28.49%) | 172 (32.03%) | Reference | Reference | Reference | ||||
| PDD/PDD-NOS (n = 531) | <30 | 161 (30.32%) | 145 (27.31%) | 1.44 (1.001–2.09) | .04 | 1.68 (1.11–2.55) | .01 | 1.76 (1.13–2.74) | .01 |
| 30–49.9 | 216 (40.68%) | 205 (38.61%) | 1.29 (0.95–1.75) | .09 | 1.44 (1.02–2.01) | .03 | 1.46 (1.02–2.09) | .03 | |
| ≥50 | 154 (29.00%) | 181 (34.09%) | Reference | Reference | Reference |
ASD, autism spectrum disorder; CI, confidence interval; PDD, pervasive developmental disorder; PDD-NOS, PDD-not otherwise specified.
Adjusted for maternal age, gestational week of blood draw, season of blood collection, and gestational age.
Adjusted for all confounders listed above and maternal smoking, immigration, psychopathology, and substance abuse.
Table 4 shows the association between maternal 25(OH)D levels and offspring ASD with and without ID. The association of 25(OH)D levels was significant only among ASD without ID; this was observed for the log-transformed maternal 25(OH)D levels (aOR 0.77, 95% CI 0.62–0.95, p = .02), the first 2 lowest quintiles <20 (aOR 1.34, 95% CI 1.01–1.83, p = .03) and 20–39 (aOR 1.35, 95% CI 1.02–1.78, p = .03), and deficient (<30 nmol/L) maternal 25(OH)D level (aOR 1.42, 95% CI 1.12–1.81, p = .004). The findings remained significant for continuous and clinical categories, but not for quintiles after additional adjustment for maternal smoking, immigration, psychopathology, and substance abuse. The test for interaction by ID found no evidence of effect modification by ID on the relationship between continuous maternal 25(OH)D and ASD (p = .19).
Table 4.
Odds Ratios and 95% CIs of Association Between Maternal Serum Vitamin D (Continuous, Quintiles) and Offspring Autism With and Without ID
| Maternal Serum Vitamin D, nmol/L | Association With Offspring Autism Without ID | |||||||
|---|---|---|---|---|---|---|---|---|
| Cases (n = 1337), Mean (SD) or n (%) | Controls (n = 1337), Mean (SD) or n (%) | Odds Ratio, Unadjusted (95% CI) | p Value | Odds Ratio, Adjusteda (95% CI) | p Value | Odds Ratio, Adjustedb (95% CI) | p Value | |
| Continuous: Log-Transformed Analysis | 3.66 (0.45) | 3.69 (0.46) | 0.81 (0.66–0.98) | .03 | 0.77 (0.62–0.95) | .01 | 0.77 (0.61–0.97) | .02 |
| Quintiles | ||||||||
| <20 | 284 (21.24%) | 265 (19.82%) | 1.29 (0.97–1.70) | .07 | 1.34 (1.01–1.83) | .03 | 1.31 (0.96–1.80) | .09 |
| 20–39 | 289 (21.62%) | 264 (19.75%) | 1.30 (0.99–1.70) | .05 | 1.35 (1.02–1.78) | .03 | 1.28 (0.95–1.73) | .10 |
| 40–59 | 271 (20.27%) | 270 (20.19%) | 1.17 (0.90–1.52) | .23 | 1.18 (0.90–1.56) | .21 | 1.08 (0.81–1.45) | .58 |
| 60–79 | 251 (18.77%) | 268 (20.04%) | 1.06 (0.82–1.38) | .61 | 1.05 (0.81–1.38) | .68 | 0.97 (0.73–1.30) | .84 |
| ≥80 | 242 (18.10%) | 270 (20.19%) | Reference | Reference | Reference | |||
| Categories | ||||||||
| <30 | 404 (30.22%) | 365 (27.30%) | 1.35 (1.07–1.69) | .009 | 1.42 (1.12–1.81) | .004 | 1.44 (1.11–1.87) | .005 |
| 30–49.9 | 539 (40.31%) | 526 (39.34%) | 1.21 (0.99–1.47) | .05 | 1.23 (1.008–1.51) | .04 | 1.21 (0.97–1.51) | .07 |
| ≥50 | 394 (29.47%) | 446 (33.36%) | Reference | Reference | Reference | |||
| Maternal Serum Vitamin D (nmol/L) | Association With Offspring Autism With ID | |||||||
| Cases (n = 221), Mean (SD) or n (%) | Controls (n = 221), Mean (SD) or n (%) | Odds Ratio, Unadjusted (95% CI) | p Value | Odds Ratio, Adjusteda (95% CI) | p Value | Odds Ratio, Adjustedb (95% CI) | p Value | |
| Continuous: Log-Transformed Analysis | 3.65 (0.44) | 3.67 (0.48) | 0.84 (0.51–1.38) | .49 | 0.63 (0.36–1.11) | .11 | 0.61 (0.34–1.1) | .10 |
| Quintiles | ||||||||
| <20 | 51 (23.08%) | 50 (22.62%) | 1.04 (0.50–2.14) | .91 | 1.38 (0.61–3.10) | .43 | 1.39 (0.60–3.22) | .44 |
| 20–39 | 42 (19.00%) | 48 (21.72%) | 0.88 (0.44–1.72) | .73 | 1.26 (0.59–2.68) | .54 | 1.21 (0.55–2.62) | .63 |
| 40–59 | 50 (22.62%) | 35 (15.84%) | 1.46 (0.72–2.96) | .29 | 2.04 (0.93–4.45) | .07 | 2.22 (1.0–4.98) | .05 |
| 60–79 | 39 (17.65%) | 50 (22.62%) | 0.79 (0.41–1.52) | .48 | 0.87 (0.43–1.76) | .71 | 0.91 (0.44–1.88) | .81 |
| ≥80 | 39 (17.65%) | 38 (17.19%) | Reference | Reference | Reference | |||
| Categories | ||||||||
| <30 | 69 (31.22%) | 66 (29.86%) | 1.25 (0.72–2.18) | .42 | 1.63 (0.87–3.07) | .12 | 1.50 (0.78–2.87) | .22 |
| 30–49.9 | 86 (38.91%) | 80 (36.20%) | 1.25 (0.78–2.02) | .34 | 1.49 (0.88–2.53) | .13 | 1.45 (0.85–2.45) | .16 |
| ≥50 | 66 (29.86%) | 75 (33.94%) | Reference | Reference | Reference | |||
CI, confidence interval; ID, intellectual disability.
Adjusted for maternal age, gestational week of blood draw, season of blood collection, and gestational age.
Adjusted for all confounders listed above and maternal smoking, immigration, psychopathology, and substance abuse.
Testing for sex-by-maternal 25(OH)D interaction did not reveal evidence of effect modification by sex on the relationship between continuous maternal 25(OH)D and ASD (p = .72) or any ASD subgroup: childhood autism (p = .65), Asperger’s syndrome (p = .71), or PDD/PDD-NOD (p = .65). We also found no interaction between age at ASD diagnosis and continuous maternal 25(OH)D (p = .96) or ASD subgroups and continuous maternal 25(OH)D: childhood autism (p = .99), Asperger’s syndrome (p = .33) or PDD/PDD-NOD (p = .10).
DISCUSSION
This nationwide register–based study revealed an association between low maternal 25(OH)D level during pregnancy and an elevated risk for diagnosed ASD in offspring. The association was significant when vitamin D was analyzed both as a linear and as a categorical variable. Adjustment for numerous potential confounders did not alter the associations. Given the large sample size in addition to other methodologic strengths, these findings in our view provide the strongest evidence to date of a link between prenatal vitamin D deficiency and ASD in offspring.
Several mechanisms may explain the association between low maternal vitamin D levels and offspring ASD. Fetal brain development is a complex process influenced by an individual’s genotype and environmental factors, including the in utero environment (5,31). Low vitamin D levels result from inadequate sun exposure, inadequate dietary vitamin D intake, and physiological risk factors such as obesity and skin color (32). Vitamin D receptors are broadly distributed in central nervous system neurons, peripheral neurons, and nonneuronal brain cells (3). The presence of the vitamin D receptor in the central nervous system suggests a significant role of vitamin D in structural and functional development and maturity in brain development (7). Vitamin D affects brain function through regulation of calcium signaling, neurotrophic and neuro-protective actions, neuronal differentiation, maturation, and growth (33). The present findings suggest that vitamin D deficiency in utero may adversely affect fetal programming, increasing risk for later development of ASD.
Our findings are consistent with results from the two Dutch Generation R studies, which each included several thousand mother-child pairs and used the Social Responsiveness Scale for autism-related traits. The first Dutch study found that low maternal vitamin D status during pregnancy was associated with elevated scores of Social Responsiveness Scale, measured when the children were about 6 years of age (2). The latter Dutch study used medical records for ASD diagnosis and found a significant association with low maternal vitamin D status during midgestation and ASD. However, in that study, no significant association was reported for a continuous measure of maternal vitamin D status (13). In a smaller Swedish study of clinically diagnosed ASD, low maternal sera vitamin D was associated with ASD with ID among Nordic born mothers (15). However, the present study demonstrated a significant association between low maternal vitamin D level and ASD without ID, after controlling for potential confounders. In contrast to the Swedish study, the present study featured more than 3 times the number of ASD cases, offering sub-stantially greater statistical power to demonstrate an association between maternal vitamin D and ASD, including the subgroup of cases without comorbid ID. Of note, the present study found no significant interactions between maternal vitamin D and ASD by ID status, although a similar definition of ID was used in both studies. Furthermore, maternal vitamin D in the present study was measured during early gestation to midgestation, and it is possible that the window for risk associated with exposure may remain open during the postnatal period. A study based on the Northern Finnish Birth Cohort suggested that the lack of vitamin D supplementation during the first year of life was associated with increased risk of schizophrenia in males (34). Although the same association has not been tested for ASD, it is likely that the exposure may be related to a broader range of neurodevelopmental disorders.
In further analyses with clinical ASD subgroups, low maternal vitamin D levels were significantly associated with the PDD/PDD-NOS group. However, the direction of association was similar in all groups, and the test for interaction did not indicate statistically different associations between subgroups. We note that no previous studies have examined associations between maternal vitamin D levels and ASD subgroups. While the current DSM-5 classification system does not differentiate between these subgroups, we believe that this novel finding is worthy of further study. Of note, previously we reported an association between low maternal vitamin D levels and attention-deficit/hyperactivity disorder in offspring (18). Children with attention-deficit/hyperactivity disorder and ASD might share traits that could potentially explain the association with low maternal vitamin D during pregnancy. Moreover, other researchers have reported that maternal vitamin D levels during pregnancy were associated with other neurodevelopmental disorders in the offspring, including schizophrenia, cognitive impairment, and language impairment (35–37). We did not adjust the significance threshold based on other neurodevelopmental conditions or risk factors that have been examined in previous studies because these are not directly relevant to the null hypothesis that we investigated here (38).
There are several limitations that should be considered. First, the ASD cases in this study included only children who had been referred to specialized services and thus likely represent more severe cases of ASD. Second, maternal 25(OH) D was measured only at one time point during pregnancy. However, maternal 25(OH)D concentration is likely to be similar throughout pregnancy (19). Third, residual confounding by unmeasured factors (e.g., maternal body mass index, prenatal vitamin supplementation, or maternal medications during pregnancy) is possible. Confounding by genetic factors leading to both lower levels of maternal vitamin D (for example, through behavioral pathways) and increased risk of autism is also possible. However, the findings of a recent Mendelian randomization analysis did not support a causal relationship between ASD and 25(OH)D levels (39), potentially contradicting this explanation. Fourth, low maternal 25(OH)D could be related to other prenatal risk factors, including poor nutrition or lack of compliance with recommended prenatal care. While this information was not available, we adjusted for several related maternal factors (including smoking, age, psychiatric history, substance use, and immigrant status). The present study was conducted before the national recommendation for vitamin D supplementation during pregnancy was introduced. Therefore, it would be interesting to see whether the strength of the association would change after introduction of the national recommendation for supplementation.
Conclusions
These findings suggest that maternal vitamin D deficiency is related to an increased risk of ASD in offspring. If future studies confirm the association, this will have public health significance, as vitamin D deficiency is readily preventable. Moreover, we propose that maternal vitamin D deficiency and risk of offspring ASD should be studied in risk groups with higher rates of ASD, e.g., immigrants and children born prematurely. Future studies should examine if maternal vitamin D deficiency is associated with specific ASD symptoms. In addition, further work may help elucidate the interplay between genetic variants and maternal vitamin D in the risk of ASD.
Supplementary Material
ACKNOWLEDGMENTS AND DISCLOSURES
This work was supported by the Academy of Finland Flagship Programme (Grant No. 320162 [to AS]), Academy of Finland Health from Cohorts and Biobanks Programme (Grant No. 308552 [to AS]), National Institute of Environmental Health Sciences of the National Institutes of Health (Grant No. 5R01ES028125 [to ASB]), INVEST (Inequalities, Interventions and a New Welfare State) Research Flagship, and PSYCOHORTS (The Finnish psychiatric birth cohort) consortium.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
We thank the investigators and staff at the medical centers involved in this research.
The authors report no biomedical financial interests or potential conflicts of interest.
Footnotes
Supplementary material cited in this article is available online at https://doi.org/10.1016/j.biopsych.2021.07.012.
REFERENCES
- 1.Lai MC, Lombardo MV, Baron-Cohen S (2014): Autism. Lancet 383:896–910. [DOI] [PubMed] [Google Scholar]
- 2.Vinkhuyzen AAE, Eyles DW, Burne THJ, Blanken LME, Kruithof CJ, Verhulst F, et al. (2017): Gestational vitamin D deficiency and autism spectrum disorder. BJPsych Open 3:85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Eyles DW, Smith S, Kinobe R, Hewison M, McGrath JJ (2005): Distribution of the Vitamin D receptor and 1α-hydroxylase in human brain. J Chem Neuroanat 29:21–30. [DOI] [PubMed] [Google Scholar]
- 4.Langub MC, Herman JP, Malluche HH, Koszewski NJ (2001): Evidence of functional vitamin D receptors in rat hippocampus. Neuroscience 104:49–56. [DOI] [PubMed] [Google Scholar]
- 5.Eyles DW, Burne T, McGrath JJ (2013): Vitamin D, effects on brain development, adult brain function and the links between low levels of vitamin D and neuropsychiatric disease. Front Neuroendocrinol 34:47–64. [DOI] [PubMed] [Google Scholar]
- 6.Zou R, Marroun HE, McGrath JJ, Muetzel RL, Hillegers M, White T, et al. (2020): A prospective population-based study of gestational vitamin D status and brain morphology in preadolescents. Neuroimage 209:116514. [DOI] [PubMed] [Google Scholar]
- 7.Cui X, Pelekanos M, Liu PY, Burne TH, McGrath JJ, Eyles DW (2013): The vitamin D receptor in dopamine neurons; its presence in human substantia nigra and its ontogenesis in rat midbrain. Neuroscience 236:77–87. [DOI] [PubMed] [Google Scholar]
- 8.Kesby J, Cui X, O’Loan J, McGrath JJ, Burne T, Eyles D (2010): Developmental vitamin D deficiency alters dopamine-mediated behaviors and dopamine transporter function in adult female rats. Psychopharmacology 208:159–168. [DOI] [PubMed] [Google Scholar]
- 9.Kesby JP, Cui X, Ko P, McGrath JJ, Burne TH, Eyles DW (2009): Developmental vitamin D deficiency alters dopamine turnover in neonatal rat forebrain. Neurosci Lett 461:155–158. [DOI] [PubMed] [Google Scholar]
- 10.Chen J, Xin K, Wei J, Zhang K, Xiao H (2016): Lower maternal serum 25(OH) D in first trimester associated with higher autism risk in Chinese offspring. J Psychosom Res 89:98–101. [DOI] [PubMed] [Google Scholar]
- 11.López-Vicente M, Sunyer J, Lertxundi N, González L, Rodríguez-Dehli C, Sáenz-Torre ME, et al. (2019): Maternal circulating vitamin D3 levels during pregnancy and behaviour across childhood. Sci Reports 9:14792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Whitehouse AJO, Holt BJ, Serralha M, Holt PG, Hart PH, Kusel MM (2013): Maternal vitamin D Levels and the autism phenotype among offspring. J Autism Dev Disord 43:1495–1504. [DOI] [PubMed] [Google Scholar]
- 13.Vinkhuyzen AaE, Eyles DW, Burne THJ, Blanken LME, Kruithof CJ, Verhulst F, et al. (2018): Gestational vitamin D deficiency and autism-related traits: The Generation R Study. Mol Psychiatry 23:240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Egorova O, Myte R, Schneede J, Hägglöf B, Bölte S, Domellöf E, et al. (2020): Maternal blood folate status during early pregnancy and occurrence of autism spectrum disorder in offspring: A study of 62 serum biomarkers. Mol Autism 11:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lee BK, Eyles DW, Magnusson C, Newschaffer CJ, McGrath JJ, Kvaskoff D, et al. (2021): Developmental vitamin D and autism spectrum disorders: Findings from the Stockholm Youth Cohort. Mol Psychiatry 26:1578–1588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lampi KM, Sourander A, Gissler M, Niemelä S, Rehnström K, Pulkkinen E, et al. (2010): Brief report: Validity of Finnish registry-based diagnoses of autism with the ADI-R. Acta Paediatr 99:1425–1428. [DOI] [PubMed] [Google Scholar]
- 17.Munger KL, Åivo J, Hongell K, Soilu-Hänninen M, Surcel H, Ascherio A (2016): Vitamin D status during pregnancy and risk of multiple sclerosis in offspring of women in the Finnish Maternity Cohort. JAMA Neurol 73:515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sucksdorff M, Brown AS, Chudal R, Surcel HM, Hinkka-Yli-Salomäki S, Cheslack-Postava K, et al. (2021): Maternal vitamin D levels and the risk of offspring attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 60:142–151.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hauta-alus HH, Holmlund-Suila EM, Rita HJ, Enlund-Cerullo M, Rosendahl J, Valkama SM, et al. (2018): Season, dietary factors, and physical activity modify 25-hydroxyvitamin D concentration during pregnancy. Eur J Nutr 57:1369–1379. [DOI] [PubMed] [Google Scholar]
- 20.Miliku K, Vinkhuyzen A, Blanken LM, McGrath JJ, Eyles DW, Burne TH, et al. (2016): Maternal vitamin D concentrations during pregnancy, fetal growth patterns, and risks of adverse birth outcomes. Am J Clin Nutr 103:1514–1522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lokki AI, Heikkinen-Eloranta J, Öhman H, Heinonen S, Surcel H, Nielsen HS (2020): Smoking during pregnancy reduces vitamin D levels in a Finnish birth register cohort. Public Health Nutr 23:1273–1277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhou S, Tao Y, Huang K, Zhu B, Tao F (2017): Vitamin D and risk of preterm birth: Up-to-date meta-analysis of randomized controlled trials and observational studies. J Obstet Gynaecol Res 43:247–256. [DOI] [PubMed] [Google Scholar]
- 23.Polo-Kantola P, Lampi KM, Hinkka-Yli-Salomäki S, Gissler M, Brown AS, Sourander A (2014): Obstetric risk factors and autism spectrum disorders in Finland. J Pediatr 164:358–365. [DOI] [PubMed] [Google Scholar]
- 24.Lehti V, Hinkka-Yli-Salomäki S, Cheslack-Postava K, Gissler M, Brown AS, Sourander A (2015): Maternal socioeconomic status based on occupation and autism spectrum disorders: A national case-control study. Nordic J Psychiatry 69:523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Jokiranta E, Brown AS, Heinimaa M, Cheslack-Postava K, Partanen A, Sourander A (2013): Parental psychiatric disorders and autism spectrum disorders. Psychiatry Res 207:203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Tran PL, Lehti V, Lampi KM, Helenius H, Suominen A, Gissler M, et al. (2013): Smoking during pregnancy and risk of autism spectrum disorder in a Finnish National Birth Cohort. Paediatr Perinat Epidemiol 27:266–274. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Sandin S, Schendel D, Magnusson P, Hultman C, Surén P, Susser E, et al. (2016): Autism risk associated with parental age and with increasing difference in age between the parents. Mol Psychiatry 21:693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lee BK, Gross R, Francis RW, Karlsson H, Schendel DE, Sourander A, et al. (2019): Birth seasonality and risk of autism spectrum disorder. Eur J Epidemiol 34:785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lampi KM, Lehtonen L, Tran PL, Suominen A, Lehti V, Banerjee NP, et al. (2012): Risk of autism spectrum disorders in low birth weight and small for gestational age infants. J Pediatr 161:830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rothman KJ, Greenland S (1998): Modern Epidemiology, 2nd ed. Philadelphia: Lippincott Williams & Wilkins. [Google Scholar]
- 31.Freedman R, Hunter SK, Hoffman MC (2018): Prenatal primary prevention of mental illness by micronutrient supplements in pregnancy. Am J Psychiatry 175:607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Holick MF (2017): The vitamin D deficiency pandemic: Approaches for diagnosis, treatment and prevention. Rev Endocr Metab Disord 18:153–165. [DOI] [PubMed] [Google Scholar]
- 33.Kesby JP, Eyles DW, Burne T, McGrath JJ (2011): The effects of vitamin D on brain development and adult brain function. Mol Cell Endocrinol 347:121–127. [DOI] [PubMed] [Google Scholar]
- 34.McGrath J, Saari K, Hakko H, Jokelainen J, Jones P, Järvelin M, et al. (2004): Vitamin D supplementation during the first year of life and risk of schizophrenia: A Finnish birth cohort study. Schizophr Res 67:237–245. [DOI] [PubMed] [Google Scholar]
- 35.Eyles DW, Trzaskowski M, Vinkhuyzen AAE, Mattheisen M, Meier S, Gooch H, et al. (2018): The association between neonatal vitamin D status and risk of schizophrenia. Sci Rep 8:17692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Whitehouse AJ, Holt BJ, Serralha M, Holt PG, Kusel MM, Hart PH (2012): Maternal serum vitamin D levels during pregnancy and offspring neurocognitive development. Pediatrics 129:485–493. [DOI] [PubMed] [Google Scholar]
- 37.Specht IO, Janbek J, Thorsteinsdottir F, Frederiksen P, Heitmann BL (2019): Neonatal vitamin D levels and cognitive ability in young adulthood. Eur J Nutr 59:1919–1928. [DOI] [PubMed] [Google Scholar]
- 38.Perneger TV (1998): What’s wrong with Bonferroni adjustments. BMJ 316:1236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Revez JA, Lin T, Qiao Z, Xue A, Holtz Y, Zhu Z, et al. (2020): Genome-wide association study identifies 143 loci associated with 25 hydroxyvitamin D concentration. Nat Commun 11:1647. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.