Maternal Micronutrient Biomarkers and Risk of Non-syndromic Cleft Lip/Palate: A Case-Control Study

Abstract

Objective:

This case-control study investigated the associations between maternal plasma vitamin B12, homocysteine, and red blood cell (RBC) folate levels and the risk of cleft lip with or without cleft palate (CL/P) in offspring.

Subjects and Methods:

The study compared 94 mothers and children with non-syndromic CL/P from a teaching hospital in Thailand to 94 mother-infant controls from local well-baby clinics, frequency-matched by birth date and mother’s education. Data included anthropometric measurements, blood sample analyses, and a questionnaire. Odds ratios (ORs) and 95% confidence intervals (CIs) estimated the associations through multiple logistic regression, adjusting for confounders.

Results:

Mothers with higher plasma vitamin B12 levels had a lower risk of having a child with CL/P compared to those in the lowest quartile. This association was more pronounced among mothers without a family history of orofacial clefts and those who were not underweight. Conversely, elevated homocysteine levels, a marker of impaired B vitamin metabolism, increased the risk of CL/P. No association was found between RBC folate and CL/P.

Conclusion:

Higher maternal vitamin B12 levels are associated with a reduced risk of CL/P, while elevated homocysteine levels may increase the risk.

Introduction

Orofacial clefts (OFCs) are congenital anomalies, characterized by the failure of lip, alveolus, and/or palate to form normally due to altered embryogenesis during the third to twelfth gestational week (Hammond & Dixon, 2022; Sperber, 2002). The prevalence of OFCs varies widely internationally, ranging from 1.3 to 10.4 per 10,000 births, with a pooled average of 6.4 per 10,000 live births (Mc Goldrick et al., 2023). In Asia, the OFC prevalence is 1.59 per 1,000 live births (Cooper, Ratay, & Marazita, 2006), and the population-weighted pooled prevalence of cleft lip and/or cleft palate in Thailand is 2.14 per 1,000 live births (Fuangtharnthip, Chonnapasatid, Thiradilok, Manopatanakul, & Jaruratanasirikul, 2021).

Genetic and environmental factors contribute to OFCs. Environmental risk factor include maternal health, diet and lifestyle choices before and during pregnancy (Mossey, Little, Munger, Dixon, & Shaw, 2009; Pisek, McKinney, Muktabhant, & Pitiphat, 2024; Suazo, 2022; Tolarová et al., 2006). Because cleft palate without cleft lip (CP) is associated with congenital malformations and genetic etiology at a higher rate (up to 50%) compared to cleft lip with or without cleft palate (CL/P), an investigation of environmental factors focused on infants with CL/P would be less heterogenous and offer greater potential for insights on the etiologic contribution of environmental exposures.

Maternal nutrition is an environmental factor that affects fetal development during the periconceptional period (Cetin, Berti, & Calabrese, 2010; McKinney et al., 2016; Wu, Imhoff-Kunsch, & Girard, 2012). Essential vitamins, particularly folate, vitamin B12, and related biomarkers such as homocysteine, are required for one-carbon metabolism, a complex biochemical pathway crucial for synthesizing various molecules essential for cellular functions, including cell growth, migration, differentiation, and apoptosis (Clare, Brassington, Kwong, & Sinclair, 2019; Garland, Reynolds, & Zhou, 2020; Munger et al., 2011). In early embryogenesis, a deficiency in maternal folate and related micronutrients may disrupt one-carbon pathways during the critical period of lip and palate development, potentially leading to CL/P (Alade, Awotoye, & Butali, 2022; Bokor, Vass, Funke, Ertl, & Molnár, 2022).

The human body lacks the ability to synthesize folate, necessitating its acquisition from external sources such as diet or supplements (Bokor et al., 2022; Crider, Yang, Berry, & Bailey, 2012). Evidence supports the protective effects of folic acids and multivitamin supplementation, as well as fortification, in decreasing the prevalence of neural tube defects (NTDs) and, to a lesser extent, CL/P (Dien, McKinney, Pisek, & Pitiphat, 2018; Jahanbin, Shadkam, Miri, Shirazi, & Abtahi, 2018; McKinney et al., 2013; van Gool, Hirche, Lax, & De Schaepdrijver, 2018; Zhou et al., 2020). However, the direct influence of micronutrients on CL/P development remains unclear. While animal studies have demonstrated the crucial role of adequate folate and one-carbon-related micronutrient intake in reducing CL/P risk (Bendahan, Escobar, Castellanos, & González-Carrera, 2020; Zhang, Wang, Dai, Ding, & Li, 2011), their applicability to human pregnancies requires further investigation.

Given the limitations of animal models, investigating the specific role of micronutrients requires closer scrutiny of human data. Maternal nutrition assessments are challenging, and previous research evaluating nutritional levels in relation to CL/P mainly relied on health questionnaires and food frequency questionnaires to estimate dietary intake, vitamin supplements, and fortification (Wu et al., 2012; Zhou et al., 2020). Blood biomarkers offer a direct and accurate approach to measuring maternal nutritional status, reflecting the cumulative effects of long-term factors such as dietary intake, vitamin supplement use, and the interplay between genes and environment (Bezerra et al., 2015; Munger et al., 2004; Munger et al., 2011). However, human research investigating the association between CL/P and folate-related biomarkers involved in one-carbon metabolism is limited and inconsistent. Studies examining OFCs and maternal folate and vitamin B12 biomarkers have produced mixed results, with some studies indicating positive or negative associations, while others revealed no significant link (Blanco, Colombo, Pardo, & Suazo, 2016; Zhou et al., 2020). This inconsistency emphasizes the need for further exploration of this complex relationship.

This case-control study was conducted in Thailand to examine maternal micronutrient biomarkers involved in one-carbon metabolism and their association with non-syndromic CL/P. Specifically, the study focused on plasma vitamin B12, total plasma homocysteine (as an indicator of both B12 and folate status), and red blood cell (RBC) folate (as an indicator of long-term folate status). The hypothesis was that low maternal vitamin B12 and folate levels are each associated with an increased risk of CL/P, while high homocysteine levels are associated with an increased risk.

Methods

Study Design and Participants

A 1:1 case-control study was conducted in Khon Kaen, Thailand. Study participants were recruited and data were collected from June 2017 to February 2020. Cases were recruited from the Center of Cleft Lip-Cleft Palate and Craniofacial Deformities (Cleft Center) at Khon Kaen University (KKU)’s Srinagarind Hospital. The study involved mother-infant pairs. Infants born in Northeast Thailand with CL/P were eligible if they were between 2 months and 5 years old and did not have a genetic syndrome. Mothers were eligible if they were at least 18 years of age and spoke Thai or Isaan dialect. Infants with atypical clefts (midline or mandibular), or CP alone were excluded.

To determine the cleft type, a craniofacial clinician reviewed clinical images from the Cleft Center’s comprehensive medical records, which included pre-surgical digital imaging, radiographic, and photographic records for all patients with OFCs and craniofacial anomalies. This approach was employed because each mother-infant pair visited the Cleft Center at different ages and for various treatments or procedures. As a result, some infants had already undergone cleft lip or palate repair by the time they were identified as potential cases for this study making phenotypic assessment at the time of their visit infeasible.

Controls were recruited from well-baby clinics within the district (Srinagarind Hospital, the Mother and Child Health Center, and three Primary Health Centers). Frequency matching with cases based on birth date (±4 months) and maternal education was employed to account for potential influence on knowledge, health concerns, and access to prenatal care, which could bias study results. Mothers who were pregnant or within two months postpartum were excluded to avoid potential bias due to the influence of recent pregnancy on their blood biomarkers.

Sample size was calculated based on the results of an earlier study (van Rooij, Swinkels, Blom, Merkus, & Steegers-Theunissen, 2003), which observed a 10.2% prevalence of low plasma vitamin B12 among controls and an odds ratio (OR) of 3.11 for the association with OFCs. With a 1:1 ratio of cases to controls, 80% power, and a two-sided alpha level of 5%, this study determined the need for 92 cases and 92 controls. A total of 94 case and 94 control mother-infant pairs met study criteria and completed data collection.

The study protocol was approved by the internal review boards of the KKU (HE591336) and the University of Washington (IRB#52241). All mothers received a detailed explanation of the study before providing consent for participation.

Data Collection

An interviewer-administered health questionnaire was used to obtain data on maternal exposures and potential confounding factors. The questionnaire was based on the Happy Smiles Study (McKinney et al., 2016), which was itself a modified version of the National Birth Defects Prevention Study (Centers for Disease Control and Prevention, 2010). This ensured that the questionnaire was built upon a foundation of well-established and respected research. Referent dates were generally the 3 months prior to conception for long-term biological effects exposures, such as dietary supplements, food insecurity, or smoking. For short-term biological effects exposures, like fever or infection, the focus was on the first 3 months of pregnancy. Mothers who reported taking any supplements, including multivitamins, folic acid, vitamin B complex, calcium, iron, or any combination of vitamin supplements around the time of conception, were classified as ‘having any multivitamin supplement’. Those who took supplements containing folic acid, folic acid plus iron, vitamin B complex, or multivitamins were further categorized as ‘having folic-acid-containing supplements. The mothers were divided into three groups based on their supplement intake during pregnancy: those who started before the second month, those who started in the second or third month, and those who started later or not at all. To measure food insecurity, we employed the Household Food Insecurity Access Scale (HFIAS), a globally recognized and validated tool for assessing food availability concerns (Coates J, 2007; Swindale & Bilinsky, 2006). Mothers self-reported their own smoking or passive smoking exposure from husbands, family members, coworkers, or a school setting.

Maternal anthropometric data were obtained by two well-trained research assistants. Mothers self-reported their pre-pregnancy and current weight, which was then used to calculate their pre-pregnancy body mass index (BMI). The mothers’ BMI classifications were determined based on the World Health Organization (WHO) standards specifically designed for Asian populations (WHO expert consultant, 2004).

Blood Sample Analyses

Maternal non-fasting venous blood samples were collected by a nurse at the Cleft Center and well-baby clinics and then immediately transported to the Srinagarind Hospital laboratory for processing. RBC and plasma were extracted into aliquots for subsequent micronutrient testing. The aliquots were stored in a -80°C freezer until all samples had been collected. All samples were then analyzed simultaneously to minimize variation, with laboratory technicians being blinded to the case-control status of the samples. RBC folate and plasma vitamin B12 were measured by electrochemiluminescence immunoassay (ECLIA) (Cobas c801 for folate and Cobas c602 for vitamin B12, Roche Diagnostics GmbH, Mannheim, Germany). Total plasma homocysteine was determined by homocysteine enzymatic assay (Cobas c502, Roche Diagnostics GmbH, Mannheim, Germany). The Srinagarind Hospital laboratory meets the requirements for quality and competence in medical laboratories (ISO 15189/ISO15190) and received the National Laboratories Quality System Certification from the Department of Medical Sciences, Ministry of Public Health, Thailand.

This study focused on RBC folate levels due to their role as a sensitive indicator of long-term folate status, while plasma folate provides information on the short-term folate status of the individuals (Bailey et al., 2015). Maternal RBC folate levels below 400 ng/mL were classified as ‘folate deficiency’ according to the WHO guideline, indicating folate insufficiency in women of reproductive age (World Health Organization, 2015). Vitamin B12 deficiency was defined as a plasma concentration below 200 pg/mL (Carmel & Sarrai, 2006; Rosenthal et al., 2017).

Statistical Analysis

Characteristics and blood biomarker levels in case and control groups were compared using chi-square tests for categorical variables, t-tests for normally distributed continuous variables, and Mann-Whitney U tests for skewed continuous variables. Micronutrient intakes were categorized into quartiles based on their distributions among all subjects (Hsieh, Maisonneuve, Boyle, Macfarlane, & Roberston, 1991)

Multivariable logistic regression was used to estimate ORs and their 95% confidence intervals (CIs) to determine the associations between blood biomarkers and CL/P, adjusting for matching variables and potential confounding factors. Infant age was adjusted for as a continuous variable and mother’s education was divided into four categories (primary/secondary/high school/any college). Potential confounders considered in the analysis included father’s education (4 categories), household income (4 categories), family history of oral clefts (yes/no). Behavioral and health risk factors during first trimester considered included: smoking (yes/no), passive smoking exposure (yes/no), alcohol consumption (yes/no), herb use for menstrual regulation (yes/no), pre-pregnancy BMI (4 categories: underweight/normal/overweight/obese), multivitamin supplements and folic-acid-containing supplements (3 categories), having a fever or any infection (yes/no), breastfeeding status (yes/no), and postpartum period (>6 months or not). Each potential confounder was assessed separately and in combination with other confounders. The modeling strategy involved removing confounders that did not change the estimates to a meaningful degree (less than 10%) from the initial and combined models.

To minimize the influence of genetics on the study findings, a subgroup analysis was conducted, excluding mothers with a family history of OFCs. Subgroup analyses were also conducted to isolate the effects of maternal metabolism during pregnancy by focusing on mothers more than 6 months postpartum, and to minimize the effect of malnutrition by analyzing only mothers who were not underweight. All analyses were performed using Stata 13.0 (StataCorp LP, College Station, Texas, USA).

Results

Ninety-four cases and 94 matched controls were enrolled in the study. The case and control groups were similar in demographic characteristics; however, infants in the case group were slightly younger (Table 1). A significantly higher proportion of cases (18.1%) had a family history of OFCs compared to controls (4.3%). Case mothers as compared to control mothers were more likely to be exposed to secondhand smoke (44.7% versus 19.2%), use herbal menstrual regulators (15.9% versus 5.3%), and experience fever in the first trimester (30.8% versus 18.1%), respectively. Case and control mothers were similar in terms of other medical history and behaviors in the first trimester. Both groups showed comparable rates of non-supplementation with multivitamins and folic acid in the first trimester.

Table 1.

Demographic characteristics of mothers and infants by case and control status.

Characteristics	Control (n=94)	Case (n=94)	p-valuea	Characteristics	Control (n =94)	Case (n=94)	p-valuea
	%	%			%	%
Infant characteristics				Mothers’ medical history and behaviors during first trimester
Gender: Male	55.3	50.0	0.47	Had a plan for pregnancy	48.9	42.5	0.38
Preterm (<37gestational weeks)	3.2	5.3	0.47	First prenatal care visit
Low birth weight (<2500 grams)	6.4	10.6	0.30	<4 weeks gestation	7.4	8.5	0.10
Matching variables				4–8 weeks gestation	56.4	45.7
Infant’s age (months)				>8–12 weeks gestation	12.8	26.6
Mean (standard deviation)	20.2 (17.9)	16.0 (15.9)	0.04 b	>12 weeks gestation	23.4	19.2
Median (Min–Max)	12.2	7.8		Smoking during 1st trimester	1.1	2.1	0.56
	(2.2–63.9)	(2.6–58.5)		Passive smoke exposure	19.2	44.7	<0.001
Mother’s education				Alcohol consumption	11.7	20.2	0.11
Primary school	14.9	14.9	1.00	Herb use for menstrual regulation	5.3	15.9	0.02
Junior high school	27.7	27.7		Contraception use	10.6	10.6	1.00
High school	29.7	29.7		Gestational diabetes mellitus	1.1	1.1	1.00
Any college	27.7	27.7		Pre-pregnancy BMI
Mother and family Characteristics				Normal	56.4	41.5	0.14
Maternal age at index child’s birth				Underweight	19.2	19.2
<25 years	36.2	31.9	0.82	Overweight	15.9	24.4
25–34.9 years	43.6	45.7		Obese	8.5	14.9
≥35 years	20.2	22.3		Multivitamin supplements during pregnancy
Father’s educationc				Did not take	81.9	79.8	0.15
Primary school	20.9	13.6	0.20	Started 2nd or 3rd month	3.2	9.6
Junior high school	28.5	23.9		Started prior to 2nd month	14.9	10.6
High school	20.9	34.1		Folic acid-containing supplements during pregnancy
Any college	29.7	28.4		Did not take	25.5	25.5	0.09
Monthly household income (Baht)c				Started 2nd or 3rd month	11.7	23.4
≤ 15,000	30.6	32.6	0.84	Started prior to 2nd month	62.8	51.1
15,001–25,000	23.5	25.9		Trying to lose weight	6.4	14.9	0.06
25,001–35,000	23.5	17.9		Having a fever	18.1	30.8	0.04
>35,000	22.4	23.6		Having any infection	5.3	7.5	0.55
Married	96.7	93.5	0.32	Food insecurity	11.7	21.3	0.08
Family history of oral clefts	4.3	18.1	0.003	Stressful life event	38.3	41.5	0.66

^aχ²-test,

^bMann-Whitney test,

^cMissing value 5–10%.

Boldface indicates statistical significance at p < 0.05.

BMI (Asian population criteria): Normal (18.5–22.9 kg/m²), Underweight (<18.5 kg/m²), Overweight (23.0–27.5 kg/m²), Obese (>27.5 kg/m²); GDM: Gestational Diabetes Mellitus.

Maternal blood levels of key micronutrients are presented in Table 2. The mean levels of hemoglobin, homocysteine, and RBC folate were similar in both groups. The proportions of mothers with anemia and with folate deficiency were also comparable. Mothers of infants with CL/P had slightly lower levels of plasma vitamin B12 than controls, with two documented cases of deficiency.

Table 2.

Maternal blood micronutrients by cases and control status.

Maternal blood micronutrients	Control (n=94)	Case (n=94)	p-value
Hemoglobin (g/ml), mean (sd)	12.6 (0.92)	12.7 (1.07)	0.57a
Anemia, n (%)	19 (20.2)	18 (19.2)	0.85
Plasma Vitamin B12 (pg/mL)
Mean (SD)	674.4 (288.5)	601.2 (274.4)
Median (Min–Max)	599.4 (204.1–1781.8)	598.0 (119.8–1331.0)	0.12b
Vitamin B12 deficiency, n (%)	0 (0.0)	2 (2.1)	0.49c
RBC Folate (ng/mL)
Mean (SD)	893.1 (323.8)	870.9 (313.5)	0.63a
Median (Min–Max)	825.4 (340.9–1992.6)	814.3 (256.4–1930.5)
Folate deficiency, n (%)	3 (3.2)	3 (3.2)	1.00c
Homocysteine (μmol/L)
Mean (SD)	8.30 (2.5)	8.65 (2.0)	0.29a
Median (Min–Max)	8.0 (4.0–15.0)	9.0 (4.0–13.0)

^at-test,

^bMann-Whitney test,

^cFisher’s Exact test.

Anemia: Hb<12 g/ml, Vitamin B12 deficiency: <200 pg/mL, Folate deficiency: <400 ng/mL.

When comparing plasma vitamin B12 levels by quartiles, a decreasing odds of having a child with CL/P was observed for the second quartile (OR, 0.20; 95% CI, 0.06–0.60; p = 0.004) compared to the lowest quartile, adjusted for matching variables and confounding factors (Table 3). Mothers with plasma homocysteine levels in the third quartile had an increased risk of having an affected child (OR, 3.77; 95% CI, 1.04–13.63; p = 0.04) compared to those in the lowest quartile. No significant association was found between RBC folate and CL/P.

Table 3.

Maternal blood micronutrients and their associations with CL/P in all mothers and subgroups of mothers and infants with no family history of oral clefts.

Maternal blood micronutrient	All mothers (n=188)						Mothers and infants with no family history of oral clefts (n=167)
	Control (n=94)	Case (n=94)	Model 1 OR (95% CI)	P-value	Model 2 OR (95% CI)	p-value	Control (n=90)	Case (n=77)	Model 1 OR (95% CI)	p-value	Model 2 OR (95% CI)	p-value
	%	%					%	%
Plasma Vitamin B12 (pg/mL)
Quartile 1	14.9	35.1	Referent		Referent		14.4	36.3	Referent		Referent
Quartile 2	35.1	14.9	0.16 (0.12–0.73) a	<0.001	0.20 (0.06–0.60) b	0.004	35.6	14.3	0.15 (0.05–0.40) a	<0.001	0.20 (0.06–0.65) e	0.008
Quartile 3	22.3	27.7	0.48 (0.20–1.14)a	0.09	0.53 (0.18–1.60)b	0.26	22.2	28.6	0.48 (0.19–1.19)a	0.11	0.50 (0.15–1.65)e	0.25
Quartile 4	27.7	22.3	0.29 (0.12–0.71) a	0.007	0.38 (0.13–1.11)b	0.07	27.8	20.8	0.26 (0.10–0.68) a	0.006	0.25 (0.07–0.80) e	0.02
p-trend				0.07		0.31				0.05		0.09
RBC Folate (ng/mL)
Quartile 1	21.3	28.7	Referent		Referent		22.2	31.2	Referent		Referent
Quartile 2	28.7	21.3	0.58 (0.26–1.34)a	0.20	0.31 (0.09–1.99)c	0.05	26.7	22.1	0.64 (0.26–1.54)a	0.32	0.50 (0.16–1.56)f	0.23
Quartile 3	22.3	27.7	1.06 (0.46–2.47)a	0.89	1.01 (0.29–3.47)c	0.99	22.2	26.0	0.94 (0.37–2.30)a	0.89	0.94 (0.29–3.01)f	0.91
Quartile 4	27.7	22.3	0.57 (0.25–1.31)a	0.18	0.53 (0.16–1.80)c	0.31	28.9	20.7	0.49 (0.20–1.19)a	0.11	0.52 (0.15–1.71)f	0.28
p-trend				0.40		0.65				0.20		0.44
Homocysteine (μmol/L)
Quartile 1	41.5	30.9	Referent		Referent		41.1	32.5	Referent		Referent
Quartile 2	26.6	31.9	1.53 (0.73–3.23)a	0.25	1.19 (0.40–3.58)d	0.75	27.8	26.0	1.09 (0.48–2.47)a	0.83	0.98 (0.31–3.06)g	0.97
Quartile 3	11.7	20.2	2.69 (1.08–6.73) a	0.03	3.77 (1.04–13.63) d	0.04	12.2	22.0	2.76 (1.06–7.17) a	0.03	3.66 (1.03–12.96) g	0.04
Quartile 4	20.2	17.0	1.23 (0.52–2.89)a	0.63	0.62 (0.17–2.18)d	0.45	18.9	19.5	1.42 (0.58–3.49)a	0.44	1.70 (0.34–3.64)g	0.86
p-trend				0.29		0.90				0.17		0.42

Model 1: Adjusted for matching variables (infant’s age and mother’s education). Model 2: Adjusted for matching variable and confounding factors.

^aAdjusted for matching variables (infant’s age and mother’s education).

^bAdjusted for matching variables, paternal education, household incomes, pre-pregnancy BMI, breastfeeding status, and post-partum period.

^cAdjusted for matching variables, paternal education, household incomes, folic-acid containing supplements, family history of clefts, fever in 1st trimester, pre-pregnancy BMI, passive smoke exposure, breastfeeding status, and postpartum period.

^dAdjusted for matching variables, paternal education, household income, family history of clefts, pre-pregnancy BMI, passive smoke exposure, breastfeeding status, and postpartum period.

^eAdjusted for matching variables, paternal education, pre-pregnancy BMI, passive smoke exposure, breastfeeding status, and postpartum period.

^fAdjusted for matching variables, paternal education, pre-pregnancy BMI, herb use for menstrual regulation, fever at 1^st trimester, alcohol consumption, passive smoke exposure, breastfeeding status, and postpartum period.

^gAdjusted for matching variables, paternal education, household incomes, passive smoke exposure, alcohol consumption, breastfeeding status, and postpartum period.

Boldface indicates statistical significance at p < 0.05.

Plasma Vitamin B12 (pg/mL): Quartile 1 (119.8–412.7), Quartile 2 (412.8–598.4), Quartile 3 (598.5–808.0), Quartile 4 (808.1–1781.8); RBC Folate (ng/mL): Quartile 1 (256.3– 655.2), Quartile 2 (655.3–823.8), Quartile 3 (823.9–1023.2), Quartile 4 (1023.3–1992.6); Homocysteine (μmol/L): Quartile 1 (4.0–7.0), Quartile 2 (7.1–9.0), Quartile 3 (9.1–10.0), Quartile 4 (10.1–15.0).

When restricting the analyses to participants with no family history of oral clefts (n = 167), mothers who were >6 months postpartum (n = 131), and mothers who were not underweight (n = 152), the overall results were similar to the main analyses for all three micronutrient biomarkers (Table 4). The inverse association between vitamin B12 and CL/P appeared to be stronger among the subgroup of mothers who were not underweight, although the linear trend was marginally significant (p for trend = 0.05). The association between homocysteine and increased odds of having a child with CL/P was consistently observed in all the analyses. There was no association between RBC folate and CL/P in any subgroup analyses.

Table 4.

Maternal micronutrients and their associations with CL/P in subgroups of mothers who were more than 6 months postpartum and non-underweight mothers.

Maternal blood micronutrient	Mothers who were >6 months postpartum (n =131)						Non-underweight mothers (n =152)
	Control (n=74	Case (n=57)	Model 1 OR (95% CI)	P-value	Model 2 OR (95% CI)	p-value	Control (n=76)	Case (n=76)	Model 1a OR (95% CI)	p-value	Model 2 OR (95% CI)	p-value
	%	%					%	%
Plasma Vitamin B12 (pg/mL)
Quartile 1	13.5	35.1	Referent		Referent		13.2	36.8	Referent		Referent
Quartile 2	33.8	17.5	0.17 (0.06–0.52) a	0.002	0.29 (0.07–1.16)b	0.08	35.5	10.5	0.08 (0.03–0.26) a	<0.001	0.07 (0.02–0.30 e	<0.001
Quartile 3	25.7	24.6	0.32 (0.11–0.95) a	0.04	0.43 (0.11–1.64)b	0.21	25.0	30.3	0.39 (0.14–1.04)a	0.06	0.21 (0.06–0.74) e	0.01
Quartile 4	27.0	22.8	0.28 (0.09–0.84) a	0.02	0.32 (0.08–1.30)b	0.11	26.3	22.4	0.24 (0.08–0.68) a	0.007	0.17 (0.04–0.62) e	0.007
p-trend				0.05		0.20				0.10		0.05
RBC Folate (ng/mL)
Quartile 1	20.3	35.1	Referent		Referent		19.7	26.3	Referent		Referent
Quartile 2	32.4	24.5	0.45 (0.17–1.15)a	0.09	0.30 (0.08–1.08)c	0.06	29.0	18.4	0.51 (0.19–1.35)a	0.17	0.25 (0.06–1.08)f	0.06
Quartile 3	27.0	28.1	0.59(0.22–1.57)a	0.29	0.60 (0.15–2.48)c	0.48	22.4	32.9	1.26 (0.49–3.28)a	0.62	0.91 (0.22–3.74f	0.90
Quartile 4	20.3	12.3	0.34 (0.11–1.09)a	0.07	0.33 (0.07–1.50)c	0.15	28.9	22.4	0.56 (0.21–1.43)a	0.22	0.34(0.14–1.45)f	0.14
p-trend				0.10		0.29				0.56		0.06
Homocysteine (μmol/L)
Quartile 1	39.2	26.3	Referent		Referent		36.8	26.3	Referent		Referent
Quartile 2	24.3	31.6	1.95 (0.77–4.94)a	0.15	1.82 (0.43–7.63)d	0.41	26.3	30.3	1.47 (0.62–3.49)a	0.15	2.21 (0.61–7.99)g	0.22
Quartile 3	14.9	28.1	3.06 (1.11–8.43) a	0.03	4.49 (1.01–20.08) d	0.04	14.5	22.4	2.72 (1.00–7.37) a	0.04	4.69 (1.08–20.22) g	0.03
Quartile 4	21.6	14.0	1.07 (0.35–3.25)a	0.90	0.64 (0.12–3.32)d	0.59	22.4	21.0	1.56 (0.60–4.01)a	0.35	1.10 (0.27–4.56)g	0.89
p-trend				0.43		0.99				0.19		0.67

Model 1: Adjusted for matching variables (infant’s age and mother’s education). Model 2: Adjusted for matching variable and confounding factors.

^aAdjusted for matching variables (infant’s age and mother’s education).

^bAdjusted for matching variables, paternal education, household incomes, family history of clefts, passive smoke exposure, and breastfeeding status.

^cAdjusted for matching variables, paternal education, household incomes, family history of clefts, folic acid-containing supplements, fever in 1^st trimester, passive smoke exposure, and breastfeeding status.

^dAdjusted for matching variables, paternal education, household incomes, pre-pregnancy BMI, folic acid-containing supplements and passive smoke exposure, and breastfeeding status.

^eAdjusted for matching variables, household incomes, folic acid-containing supplements, pre-pregnancy BMI, and passive smoke exposure.

^fAdjusted for matching variables, paternal education, household incomes, pre-pregnancy BMI, folic acid-containing supplements, fever in 1^st trimester, passive smoke exposure, fever during 1^st trimester, breastfeeding status, and postpartum period.

^gAdjusted for matching variables, paternal education, household incomes, pre-pregnancy BMI, passive smoke exposure, and breastfeeding status

Boldface indicates statistical significance at p < 0.05.

Plasma Vitamin B12 (pg/mL): Quartile 1 (119.8–412.7), Quartile 2 (412.8–598.4), Quartile 3 (598.5–808.0), Quartile 4 (808.1–1781.8); RBC Folate (ng/mL): Quartile 1 (256.3– 655.2), Quartile 2 (655.3–823.8), Quartile 3 (823.9–1023.2), Quartile 4 (1023.3–1992.6); Homocysteine (μmol/L): Quartile 1 (4.0–7.0), Quartile 2 (7.1–9.0), Quartile 3 (9.1–10.0), Quartile 4 (10.1–15.0).

Discussion

This study is among the few in Asia to investigate the association between maternal blood micronutrient biomarkers and OFCs (Munger et al., 2021; Munger et al., 2004; Tamura et al., 2005; Tamura et al., 2007). The study focused on a population in northeast Thailand, characterized by a high prevalence of OFCs with diverse prenatal vitamin use. The local diet, relies largely on locally-available, unfortified foods, and represents a typical setting in middle-income countries like Thailand. Biomarkers were used to evaluate the micronutrient hypothesis, since estimates from food frequency questionnaires may not be accurate. The multivariable analyses revealed an association between higher plasma vitamin B12 levels and a reduced risk of CL/P. A statistically significant association was also identified between high homocysteine levels and an increased risk of CL/P. No association was found between RBC folate concentrations and the risk of CL/P.

Evidence regarding the association between low plasma vitamin B12 and OFCs is mixed. A systematic review and meta-analysis of seven studies from Europe and North America found no such association but acknowledged heterogeneity among studies and potential publication bias due to small sample sizes (Blanco et al., 2016). The current study’s findings are consistent with a recent case-control study in India, which reported a positive association between low maternal plasma vitamin B12 and CL/P in offspring (Munger et al., 2021). Notably, a high prevalence of marginal plasma vitamin B12 status (defined as <221 pmol/L) was observed in Indian women (45.8% and 24.0% of cases and controls), likely due to their primarily vegetarian diets. In contrast, the prevalence of B12 deficiency (<200 pg/mL) in this study was remarkably low (2% of cases, 0% of controls), possibly reflecting different dietary habits compared to the Indian population.

The potential impact of Helicobacter pylori on our findings warrants consideration, given its high prevalence in Northeast Thailand (Tirapattanun et al., 2016). H. pylori infection has been linked to increased BMI (Suki et al., 2018) and can cause gastritis, which may result in vitamin B12 malabsorption (Cai et al., 2021; Kaptan et al., 2000). Though dietary absorption of vitamin B12 may be affected by H. pylori-related gastritis, this study utilized direct measurements of vitamin B12 levels from blood samples. With the low prevalence of vitamin B12 deficiency observed in this population, the confounding impact of H. pylori on the results is likely minimal.

Unlike vitamin B12, maternal RBC folate levels did not differ between cases and controls in this study, suggesting no association between RBC folate and CL/P risk. This aligns with a meta-analysis pooling data from 12 studies that found no link between folate (plasma and RBC) and non-syndromic OFCs (Blanco et al., 2016). Similarly, a systematic review examining the various folate measures (dietary intake, fortification, supplements, and biomarkers), reported no association with OFCs (Zhou et al., 2020). Further support comes from the previously cited Indian study (Munger et al., 2021), which showed no statistically significant difference in the proportion of low maternal plasma folate levels between cases and controls.

Folate and vitamin B12 have a close metabolic inter-relationship within the one-carbon metabolism. Vitamin B12 acts as a cofactor in two key cycles: the folate cycle and the methionine cycle, facilitating the conversion of methyltetrahydrofolate (methyl-THF) to THF. Low levels of maternal vitamin B12 or vitamin B12 deficiency could lead to ‘functional folate deficiency’ even when overall folate levels appear normal. In the absence of sufficient vitamin B12, folate can become trapped in a less active form, impeding its recycling back into the folate pool for further use (Clare et al., 2019; Rosenthal et al., 2017; Selhub, Jacques, Dallal, Choumenkovitch, & Rogers, 2008). This functional folate deficiency might explain our finding of similar folate levels in case and control mothers and the lack of association between RBC folate levels and OFCs. In essence, even with adequate folate intake, B12 deficiency might hinder its utilization, limiting the protective effect of folate against oral clefts.

Elevated homocysteine levels emerged as a risk factor for CL/P in this study, in both the main and subgroup analyses. These findings align with a meta-analysis of nine studies by Blanco et al. (2016), which reported high maternal plasma homocysteine as a risk factor for OFCs. However, the authors of the meta-analysis cautioned about potential limitations due to the relatively low number of included reports, small sample sizes, and possible publication bias. A recent study in India did not find a statistically significant association, but it was relatively small, comprising 47 cases and 50 controls (Munger et al., 2021). The present study contributes to this growing body of literature. Nonetheless, the results should be interpreted with caution, as statistically significant findings were limited to the third quartile of homocysteine, without a clear trend. This could be due to the sample size.

Homocysteine is considered a key biomarker for assessing functional folate and vitamin B12 status (Selhub et al., 2008). In one-carbon metabolism, it utilizes folate, vitamin B12, and vitamin B6 as both substrates and cofactors for the methionine and transulfuration pathways (Blanton et al., 2011). Inadequate maternal intake of these nutrients can disrupt homocysteine metabolism, leading to hyperhomocysteinemia (Garland et al., 2020; Koletzko et al., 2019; Selhub et al., 2008; Wu et al., 2012). The precise mechanisms by which hyperhomocysteinemia contributes to congenital defects remain unclear. An avian embryo model indicated that high concentrations of homocysteine alter the neural crest cells and induce abnormal organ development (Rosenquist, Ratashak, & Selhub, 1996), while human palatal cell lines and tissue culture study using animals showed palatal mesenchyme apoptosis and fusion failure due to fetal hyperhomocysteinemia (Knott, Hartridge, Brown, Mansell, & Sandy, 2003). Elevated homocysteine levels have been shown to be toxic and teratogenic during embryogenesis, which may contribute to NTDs (Brauer & Tierney, 2004). The elevated homocysteine disturbs basic cell activities and impairs cellular function, resulting in negative consequences for fetal development (D’Souza & Glazier, 2022).

A key strength of this study is the use of complementary assays to assess maternal micronutrient levels related to one-carbon metabolism. Direct measurement of plasma vitamin B12 and indirect measurement of homocysteine were utilized, providing a comprehensive assessment. Plasma vitamin B12 and holotranscobalamin (holoTC) indicate circulating vitamin B12 levels, while homocysteine and methylmalonic acid (MMA) reflect functional vitamin B12 status (Yetley et al., 2011). Evidence suggests that the best strategy for diagnosing vitamin B12 deficiency involves using vitamin B12 levels as the initial test and homocysteine or MMA levels as second-line tests (Green et al., 2017). Homocysteine testing is more widely available and less expensive compared to MMA and holoTC testing (Aparicio-Ugarriza, Palacios, Alder, & González-Gross, 2015; Carmel, 2011).

This study shares limitations common to most blood biomarker studies. A major challenge in studying biomarkers and CL/P is the variability in laboratory techniques. The use of different methods and assays across studies makes it difficult to compare results with research conducted in other countries and with varying resources.

As recognized in prior research (Colapinto et al., 2014; Rogers et al., 2018), blood folate levels can vary depending on the lab and assay used. This study employed the ECLIA to assess RBC folate. While convenient and widely used, immunoassays may overestimate RBC folate levels compared to the gold standard microbiological method (Colapinto et al., 2014; Hunt et al., 2020; World Health Organization, ‎2015). Our study design compared RBC folate levels within participant groups (quartiles) to assess the risk of CL/P. This approach mitigates the impact of potential overestimation by the ECLIA assay on our hypothesis test. Researchers comparing folate levels across studies should be mindful of potential variations due to different assay methods.

Understanding vitamin B12 status among various populations is complicated by the lack of agreement on appropriate cutoff values for different biomarkers, posing challenges in diagnosing vitamin B12 deficiency (Aparicio-Ugarriza et al., 2015). Most studies of maternal blood micronutrients used reference levels from the general population (middle-aged adults) to classify maternal vitamin B12 deficiency (Sobczyńska-Malefora, Delvin, McCaddon, Ahmadi, & Harrington, 2021). Further studies are needed to establish age-specific cutoff values for assessing vitamin B12 deficiency (Aparicio-Ugarriza et al., 2015; Green et al., 2017) in relation to CL/P in reproductive-age women. Moreover, although the higher quartiles for vitamin B12 and homocysteine in this study showed an association with CL/P, the overall trends were not statistically significant. This suggests the need for large scale studies to confirm this association.

Another limitation of the study was that oral consumption and food intake before blood sampling could affect maternal blood micronutrient levels. Since most patients lack scheduled appointments, this study identified potential cases and controls as they arrived at the clinic, recruiting eligible and interested participants on the spot. These mothers were unable to fast beforehand. Participants agreed to refrain from eating or drinking until data collection was complete. Blood samples were drawn at the end of the visit, 6–8 hours after their last meal, to minimize the impact of food intake on their micronutrient blood biomarkers. This potential source of non-differential misclassification could have obscured associations.

The most common potential threat to validity in case-control studies is differential recall bias, where mothers of children with orofacial defects may be more likely to remember past events or exposures, even if they were not actually more common. This study relies on self-reported data from mothers regarding their habits and behaviors. Undesirable habits and characteristics such as smoking, drinking alcohol, unhealthy eating habits, and weight may be under-reported, leading to random misclassification bias that could obscure the true association between these factors and the risk of CL/P. Additionally, poor recall bias is a limitation, as mothers may have difficulty accurately remembering events from the crucial period surrounding conception. To minimize potential recall limitations, several measures were implemented, including using a well-designed questionnaire administered by trained interviewers and employing a conceptional age estimation calendar to facilitate accurate recall of this critical period.

The retrospective design of this study precludes us from investigating maternal blood biomarkers at the prenatal and the precise timing of CL/P development. Based on prior retrospective studies of biomarkers and birth defects, blood biomarker measurement likely represents the mother’s periconceptional nutrition status, even years after pregnancy (Blanco et al., 2016; Little et al., 2008; Munger et al., 2011; Zhou et al., 2020). Additionally, it is shown that maternal dietary habits did not change during both the pre-pregnancy and postpartum period (Devine, Bove, & Olson, 2000). Furthermore, maternal blood biomarkers encompass the collective effects of metabolic function, environmental factors, genetic inheritance, and dietary and supplemental intake, which can influence maternal health beyond the pre-or post-pregnancy (Bezerra et al., 2015; Blanton et al., 2011; Munger et al., 2011; Zhou et al., 2020). These factors support the reasonable use of blood biomarkers to investigate the associations between CL/P and maternal micronutrients.

Conclusions

This study supported the association between higher maternal vitamin B12 levels and reduced CL/P risk. Additionally, the study found an association between elevated homocysteine levels and increased CL/P risk. These findings further underscore the critical role of micronutrients related to one-carbon metabolism in the development of CL/P.