Periconceptional Serum Creatinine and Risk of Childhood Autism Spectrum Disorder: A Research Letter

Abstract

Background:

Autism spectrum disorder (ASD) is a neurodevelopmental condition that manifests in early childhood, in which the maternal metabolic syndrome may be a risk factor. The kidney is a barometer of maternal metabolic syndrome duration and severity.

Objective:

The main objective of this study is to determine whether periconceptional kidney function is associated with ASD in early childhood.

Design, Setting, and Participants:

This retrospective population-based cohort study was completed in Ontario, Canada. Included were singleton children born in an Ontario hospital between April 2007 and March 2021, who were alive at age 48 months and whose mother had a recorded prepregnancy body mass index (BMI) and a measured serum creatinine (SCr) between 120 days preconception and 28 days postconception.

Measurement:

The main study outcome was a diagnosis of ASD between ages 24 and 48 months.

Methods:

Relative risks (RRs) of ASD in association with periconceptional SCr were generated using modified Poisson regression and adjusted for several confounders.

Results:

The cohort comprised 86 054 women, who had 89 677 liveborn children surviving to at least 48 months of age. There was no significant association between periconceptional SCr and ASD (RR: 0.86; 95 % confidence interval: [0.67, 1.10]).

Limitations:

Selection bias may have arisen had SCr been ordered on clinical grounds.

Conclusions:

Further study is warranted to determine whether prepregnancy glomerular hyperfiltration is a marker of ASD and other behavioral conditions in childhood. To do so, a more accurate measure of hyperfiltration is needed than SCr.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental condition that manifests in early childhood, traditionally diagnosed starting at age 24 months. ¹ Maternal metabolic syndrome (MetSyn) may be one factor influencing the observed rise in ASD.^2,3 The kidney is a barometer of MetSyn duration and severity, ⁴ manifested by glomerular hyperfiltration and a relative decline in serum creatinine (SCr) in the early phase of MetSyn-related kidney injury. ⁵ This study determined whether periconceptional kidney function is associated with ASD in early childhood.

Methods

This retrospective population-based cohort study was completed in Ontario, Canada. Hospital birth records are linked to patient-level records across several databases using unique encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (ICES). These databases have been previously validated for sociodemographic data, primary diagnoses, laboratory data, and physician billing claims (Supplementary File S1).

Included were singleton children born in an Ontario hospital between April 2007 and March 2021, who were alive at age 48 months (the typical upper age at which ASD is diagnosed), ¹ and whose mother had a recorded prepregnancy body mass index (BMI) and a measured SCr between 120 days preconception up to 28 days postconception (the latter preceding any measurable decline in SCr in pregnancy). ⁶ Exclusion criteria are in Supplementary File S1.

The main study outcome was a diagnosis of ASD between ages 24 and 48 months (the recommended age interval across most of the study period), using a validated algorithm (Supplementary File S1).

Relative risks (RRs) of ASD in association with periconceptional SCr were generated using modified Poisson regression and adjusted for several confounders (Supplementary eTable 1). Three different approaches were taken for handling SCr as an exposure, as outlined in Supplementary eAppendix 1, including the possibility that a relatively low SCr might be present in young women with early glomerular injury due to hyperfiltration related to MetSyn.^4,5

Analyses were performed using SAS, version 9.4 (SAS Institute Inc).

Results

The cohort comprised 86 054 women, who had 89 677 liveborn children surviving to at least 48 months of age (Table 1). Mean (SD) maternal age was 30.7 (5.0) years, 54% were multiparous, mean (SD) BMI was 25.7 (6.3) kg/m², and 3.8% had diabetes or hypertension before pregnancy.

Table 1.

Demographic and Clinical Characteristics of the 89 677 Pregnancies Resulting in Livebirths Among 86 054 Women.

Characteristic
Maternal, at time of conception
Mean (SD) age, years	30.7 (5.0)
Residential income quintile
1 (lowest)	19 360 (21.6)
2	17 948 (20.0)
3	18 753 (20.9)
4	19 118 (21.3)
5 (highest)	14 498 (16.2)
Rural residence	6135 (6.8)
Median (IQR) parity	1 (0-1)
Parity ≥ 1	48 339 (53.9)
Mean (SD) prepregnancy body mass index, kg/m2	25.7 (6.3)
Pre-pregnancy body mass index category, kg/m2
10 to 18.4	4782 (5.3)
18.5 to 24.9	45 376 (50.6)
25 to 29.9	21 930 (24.5)
30 to 49.9	17 054 (19.0)
50 to ≤60	535 (0.6)
Maternal comorbidities within 1 year prior to conception
Diabetes mellitus	3382 (3.8)
Chronic hypertension	3363 (3.8)
Drug dependence	2439 (2.7)
Systemic lupus erythematosus	1506 (0.7)
Maternal laboratory measures between 120 days prior to conception and up to 28 days after conception
Mean (SD) periconceptional serum creatinine, µmol/L	60.7 (9.7)
Mean (SD) periconceptional eGFR, mL/min	115.1 (12.5)
Newborn measures at birth
Mean (SD) newborn gestational age at birth, weeks	38.8 (1.7)
Mean (SD) newborn weight at birth, grams	335 (544)
Female	43 496 (48.5)

Note. All data are shown as a number (%) unless otherwise indicated. A woman may have contributed more than 1 pregnancy. IQR = interquartile range.

Approach 1 found no significant association between periconceptional SCr and ASD (Table 2). With Approach 2, an SCr < fifth percentile was associated with a higher rate of childhood ASD (2.4%) than SCr between the fifth to 95th percentile (1.8%)—corresponding to an unadjusted RR of 1.31 (95% confidence interval [CI]: [1.06, 1.62]). As expected, within Approach 2, adjusting for prepregnancy diabetes, hypertension and BMI, among other covariates, attenuated that RR (1.23, 95% CI: [1.00, 1.52]) (Table 2). A similar, but nonsignificant trend was seen with Approach 3 (Table 2).

Table 2.

Periconceptional Maternal Serum Creatinine and Associated Risk of Childhood Autism Spectrum Disorder (ASD), Assessed Between Ages 24 and 48 Months.

Serum creatinine measurement a		No. (%) with outcome	Relative risk (95% CI)
			Unadjusted	Adjusted b
Approach 1:	≤95th percentile: 78 µmol/L (N = 85 791)	1592 (1.9)	1.00 (referent)	1.00 (referent)
	>95th percentile: 78 µmol/L (N = 3886)	63 (1.6)	0.87 (0.68-1.12)	0.86 (0.67-1.10)
Approach 2:	<Fifth percentile: 46 µmol/L (N = 3797)	91 (2.4)	1.31 (1.06-1.62)	1.23 (1.00-1.52)
	Fifth to 95th percentile: 46-78 µmol/L (N = 81 994)	1501 (1.8)	1.00 (referent)	1.00 (referent)
	>95th percentile: 78 µmol/L (N = 3886)	63 (1.6)	0.89 (0.69-1.14)	0.86 (0.67-1.11)
Approach 3: a	<65 µmol/L (N = 60 976)	1170 (1.9)	1.12 (1.00-1.25)	1.11 (0.99-1.24)
	65-75 µmol/L (N = 22 417)	384 (1.6)	1.00 (referent)	1.00 (referent)
	>75 µmol/L (N = 6284)	101 (1.7)	0.94 (0.76-1.17)	0.91 (0.74-1.14)

Note. Shown are 3 different modeling approaches, using different serum creatinine cut points. CI = confidence interval.

^aCut-points were derived based on inspection of the fractional polynomial curve, as shown in the Supplementary eAppendix.

^bAdjusted for maternal age, residential income quintile, and rural residence—each at the time of conception—as well as pre to pregnancy body mass index, diabetes and chronic hypertension—each within 1 year before the estimated date of conception.

Discussion

Elevated periconceptional maternal SCr was not associated with the development of ASD in childhood, but a low SCr posed a marginally higher risk.

Study strengths included the use of a large, diverse cohort within a universal health care system and the ability to account for some important confounders, such as BMI. The study also has some limitations. First, although often part of the routine annual health exam, selection bias may have arisen had SCr been ordered on clinical grounds. Specifically, of the 193 205 women in the original cohort, 107 151 (55%) were excluded for not having an SCr measurement. Second, the broad window period in which the SCr was measured in relation to the estimated date of conception (EDC), as well as the single SCr measurement necessary to be included in the final cohort may not have been reflective of changes in kidney function that may have occurred at the time of conception. Third, a family or sibling history of ASD was not accounted for herein, nor were other environmental or toxicological factors. Finally, it is plausible that some of the low SCr values may have been associated with malnutrition or low muscle mass, rather than hyperfiltration.

Hyperfiltration and reduced SCr are typical early responses of renal injury in young adults with MetSyn, including women of reproductive age.^4,5 Only in later adulthood does SCr begin to rise with ongoing renal injury. ⁴ It is not surprising that adjusting for prepregnancy diabetes and hypertension attenuated the RR because both are on the causal pathway to glomerular injury. ⁴

Further study is warranted to determine whether prepregnancy glomerular hyperfiltration is a marker of ASD and other behavioral conditions in childhood. To do so, a more accurate measure of hyperfiltration is needed than SCr.