Maternal–Foetal HLA‐DQB1 Incompatibility Is Associated With Pregnancy‐Induced Hypertensive Disorders in a Genetically Isolated Population

ABSTRACT

In pregnancy, semi‐allogenic foetal trophoblasts express a specific HLA profile mediating maternal leukocyte contact, crucial for placentation. Paradoxically, maternal immunomodulation requires foetal antigen recognition, especially involving certain HLA molecules. Pre‐eclampsia, a severe hypertensive complication, has been linked to antigenic similarity. Previously, we showed no selection for HLA (in)compatibility in uncomplicated naturally conceived pregnancies. However, pre‐eclamptic pregnancies were associated with increased total maternal–foetal HLA and HLA‐C matching. These associations suggest a role for HLA mismatches in immune regulation leading to an uncomplicated pregnancy. To better understand HLA homozygosity in human reproduction, we aimed to determine if there is a preferential selection for HLA compatibility in a genetically isolated population, and its relation to hypertensive complications. A nested case‐control study, comprising 125 uncomplicated pregnancies and 50 with hypertensive complications (29 with pregnancy‐induced hypertension, 21 with pre‐eclampsia) was conducted in a genetically isolated Dutch population (F_ROH 1.3–3.1). Maternal and foetal HLA‐A, ‐B, ‐C, ‐DRB1, ‐DQA1, ‐DQB1 and maternal killer‐cell immunoglobulin‐like receptor (KIR) genotyping were performed. Maternal–foetal HLA (mis)match counts were compared to expected values from randomisation of paternal HLA haplotypes over maternal haplotypes of the foetuses. Mismatched CD4⁺ T cell epitopes presented by maternal HLA class II were predicted using the PIRCHE‐II algorithm. In uncomplicated pregnancies, no difference was found between observed and expected maternal–foetal HLA (mis)matches. However, pregnancies with hypertensive complications showed significantly higher observed HLA‐DQB1 mismatches, reflected in PIRCHE‐II scores. No significant differences were found in KIR/HLA‐C frequencies. Interpretation is limited by the small sample size and the grouping of distinct hypertensive disorders. Nonetheless, maternal–foetal HLA‐DQB1 mismatch seems to play a role in the aetiology of hypertensive complications during pregnancy in this population.

1. Introduction

Pre‐eclampsia is a common but progressive and potentially severe pregnancy‐induced hypertensive disorder characterised by new‐onset hypertension and organ dysfunction. It is a leading cause of foetal and maternal morbidity and mortality worldwide [1]. While the clinical presentation of pre‐eclampsia is in the third trimester, the underlying pathophysiology is related to preceding placental dysfunction that starts early in gestation. There is growing evidence that immune recognition is involved in the aetiology of the disease [2, 3, 4].

Successful development of the semi‐allogeneic foetus and placenta within the uterus requires modulation of the maternal immune system. Foetal extravillous trophoblasts (EVT) encounter maternal immune cells at the maternal–foetal interface [5, 6, 7] and express a specific HLA profile that constitutes classical polymorphic HLA‐C as well as non‐classical oligomorphic HLA‐E, ‐F and ‐G [8]. The interaction of maternal uterine NK (uNK) cells with the invading EVT is suggested to be essential for implantation and placentation [9, 10]. The killer‐cell immunoglobulin‐like receptor (KIR) on uNK cells directly interacts with HLA‐C molecules on the surface of EVTs to facilitate spiral artery remodelling. Certain combinations of maternal KIR and (paternally inherited) foetal HLA‐C are associated with either enhancement or disruption of placentation, potentially resulting in an uncomplicated or pre‐eclamptic pregnancy, respectively [11, 12]. Furthermore, pregnancies with an HLA‐C mismatched child have been associated with an increased regulatory T cell population in the placenta [13].

The degree of complete HLA compatibility between mother and child has been associated with pregnancy outcome and risk of complications [14]. Adequate maternal immunomodulation does require a certain extent of recognition of paternal antigens and, therefore, histoincompatibility. Indeed, antigenic similarity among couples has been associated with an increased risk of pre‐eclampsia, preterm stillbirth and recurrent pregnancy losses [15, 16, 17]. Even though the trophoblast has a limited profile of HLA expression, indirect recognition of HLA peptides presented by maternal HLA class II may also play a role in T‐cell‐mediated immunomodulation [18].

We previously showed no preferential selection of maternal–foetal HLA compatibility in uncomplicated naturally conceived pregnancies [14]. In contrast, the same study showed that increased total HLA, HLA class I and especially HLA‐C compatibility is associated with pre‐eclampsia. This suggests a role for HLA mismatches in immune regulation leading to uncomplicated pregnancy. However, to increase understanding of the biological mechanisms underlying maternal–foetal HLA (in)compatibility in human reproduction, studies on parental HLA sharing between closely related individuals, rather than larger random populations, are necessary [19].

Hence, to further investigate the optimal degree of HLA compatibility for a successful pregnancy and the (clinical) consequences of increased HLA compatibility, we aimed to determine whether there is a preferential selection for HLA compatibility and specific KIR/HLA‐C combinations in a Dutch genetically isolated population with a high inbreeding coefficient and its relation to pregnancy outcome, especially pre‐eclampsia.

2. Methods

2.1. Study Design

This study was conducted as a case–control study nested within the DNA Analysis of Residents Within an Isolate in the Netherlands (DARWIN) cohort, a Dutch genetically isolated population. The village, founded in the 14th century by 7–20 families, has remained largely isolated due to strong social and religious cohesion. This isolation has resulted in a high degree of genetic homogeneity, with an inbreeding coefficient of F_ROH 1.3 (recent) and 3.1 (ancient), which is 6–14× greater than the F_ROH of the overall Dutch population [20]. The current population consists of ~22,500 inhabitants, with around 250 births annually.

The genetic homogeneity has led to an increased prevalence of rare genetic diseases caused by founder mutations, such as pontocerebellar hypoplasia type 2 (PCH2), foetal akinesia deformation sequence (FADS), rhizomelic chondrodysplasia punctata type 1 (RCDP1) and osteogenesis imperfecta type IIB/III (OI type IIB/III) [21, 22]. Since 2012, inhabitants have been able to attend an outpatient clinic of the Department of Clinical Genetics at the Amsterdam University Medical Centers (Amsterdam UMC) for (preconception) carrier screening for these diseases.

2.2. Subjects

All women who attended the outpatient clinic between 2012 and 2022 were invited to participate. Exclusion criteria included pregnancies with children affected by the aforementioned hereditary diseases, twin pregnancies, pre‐existing hypertension, and pregnancies conceived via embryo‐, sperm‐ or oocyte donation (OD).

The preconception care consultation included ancestry tracing up to four generations, resulting in a total of 16 ancestors per individual. The descent of the included children was quantified by counting the total number of the 16 ancestors that originated from the founder population. Women–child couples were selected if the combined founder descent mother and father of the child were at least 8/16 + 8/16 (i.e., half from each parent) or 4/16 + 16/16 (i.e., one parent entirely from founder descent and the other partially so).

We included the first ongoing pregnancy of all women with 259 eligible pregnancies; this resulted in 175 pregnancies. The 125 uncomplicated pregnancies were defined as a pregnancy with no pregnancy‐induced hypertension (PIH), pre‐eclampsia, HELLP syndrome (pregnancy‐associated syndrome characterised by haemolysis, elevated liver enzymes and low platelet count), preterm birth, foetal growth restriction or intrauterine infection. We included 29 pregnancies complicated by PIH and 21 pregnancies complicated by pre‐eclampsia. Pre‐eclampsia was defined as previously described by the International Society for the Study of Hypertension in Pregnancy (ISSHP), namely a new onset diastolic blood pressure ≥ 90 mmHg after 20 weeks of gestation or worsening of pre‐existing hypertension together with proteinuria (> 0.3 g/L/24 h, or EKR > 30, or 2 times ++ on the qualitative dipstick) and/or organ dysfunction [23]. PIH was defined as the new onset of hypertension (blood pressure ≥ 140 mmHg systolic or ≥ 90 mmHg diastolic) at or after 20 weeks of gestation, in the absence of proteinuria or other findings suggestive of pre‐eclampsia. We combined the cases with PIH or pre‐eclampsia, classified as hypertensive complications. The flowchart of patient inclusion is depicted in Figure 1. All medical records were reviewed, and clinical data summarised. The study protocol was approved by the ethics committee of the LUMC (23–3033), and informed consent from all women was obtained.

FIGURE 1.

Flowchart of the patient inclusion process. PE; pre‐eclampsia, PIH; pregnancy‐induced hypertension.

2.3. HLA and KIR Typing

Maternal DNA samples, isolated from peripheral blood samples collected at the outpatient clinic for carrier screening, were stored at the Department of Clinical Genetics at Amsterdam UMC. These maternal DNA samples were transferred to the Leiden University Medical Center (LUMC) for HLA typing. Saliva of the children was collected at a temporary outpatient clinic at the local midwifery practice from November 2022 to January 2023, using the Oragene DNA OG‐500 and OG‐575 collection kits for human DNA from saliva (DNA Genotek, Canada).

HLA genotyping was performed on the maternal and children's DNA samples as previously described by van't Hof et al. [14] DNA was isolated from the children's saliva samples using QIAsymphony (Qiagen, Germany). A Luminex bead‐based Reverse Sequence Specific Oligonucleotides DNA‐typing technique was used to type the samples for HLA‐A, ‐B, ‐C, ‐DR, ‐DQA1 and ‐DQB1 loci (LIFECODES typing kits, Immucor, USA). The number of maternal–foetal HLA (mis)matches was calculated at the Dutch National Reference Laboratory for Histocompatibility Testing at the LUMC. HLA class I and class II compatibility was defined at the first field level. The total number of observed antigen matches, calculated on the first field level, ranged from 5 to 10. HLA loci showing homozygosity in both mother and child were counted as 1 match and 0 mismatch.

KIR genotyping was performed for all maternal DNA samples using RT‐qPCR with sequence‐specific primers. DNA was amplified using the SYBR green protocol (Bio‐Rad, USA). The following KIR genes were typed: 2DL1, 2DL2, 2DL3, 2DL5, 2DS1, 2DS2, 2DS3, 2DS4, 2DS5, 3DL1 and 3DS1. We assigned the KIR B haplotype in the presence of one or more of the following genes: 2DL2, 2DL5, 2DS1, 2DS2, 2DS3, 2DS5 or 3DS1; group A was assigned when these genes were absent. The HLA‐C1 or HLA‐C2 group was assigned on the basis of the presence of SER77ASN80 (C1) and ASN77LYS80 (C2).

2.4. PIRCHE‐II Analysis

Using the multi‐patient solid organ transplantation module of the Predicted Indirectly Recognisable HLA Class II Epitopes (PIRCHE‐II) algorithm version 3.3.30 (PIRCHE AG, Berlin, Germany), both HLA class I and II peptides of the ‘donor’ (child) can be identified that could be presented by HLA class II of the ‘recipient’ (mother). In this case, by calculating the number of foetal HLA class I‐ and class II‐derived peptides that could be presented on maternal HLA class II, foetal HLA‐derived epitopes were identified. HLA‐A, ‐B, ‐C, ‐DRB1, ‐DRB3/4/5 and ‐DQB1 were taken into consideration as presented loci.

PIRCHE‐II peptides and their weights were calculated based on the 2011 NMDP super‐population [24]. The PIRCHE‐II score of an individual is equal to the number of predicted PIRCHE‐II peptides. PIRCHE‐II scores were determined by calculating the sum of all estimated PIRCHE‐II peptides.

2.5. Statistical Analysis

Genotypic frequencies of maternal and foetal HLA were examined for Hardy–Weinberg equilibrium (Hardy 1908; Weinberg 1963) with Pypop Software 0.7.0 [25]. All other statistical analyses were performed using SPSS Statistics 25 (IBM SPSS Software).

The expected number of HLA‐(mis)matches for each pregnancy was determined by randomisation of the paternal HLA genotype, as schematically illustrated in Figure 2. First, the paternal allele was deduced by comparing the maternal and foetal genotype for HLA‐A, ‐B, ‐C, ‐DR and ‐DQ loci. In case of identical heterozygous HLA genotypes for both mother and child, the classification between maternally or paternally inherited genes was randomised. Second, the paternally inherited haplotype of one pregnancy was randomly assigned to another pregnancy while keeping the maternal genotype and maternally inherited foetal alleles unchanged. This process was repeated for all cases and performed in triplicate for each child, thus generating three ‘artificial’ foetal genotypes per case, each combining the original maternal contribution with a randomly assigned paternal haplotype derived from another foetus in the cohort. Third, the number of HLA (mis)matches of the three complete artificial foetal HLA genotypes, in relation to the maternal genotype, was calculated by direct counting. Finally, the expected number of HLA (mis)matches was determined as the average of the HLA (mis)matches of the three artificial genotypes. The data of observed and expected HLA compatibility were analysed per locus separately.

FIGURE 2.

Visual presentation of the randomisation‐based generation of expected maternal–foetal HLA (mis)matching values. Paternally inherited alleles were determined for all but two cases. This analysis was repeated for KIR/HLA‐C combinations, using triplicate randomisation of HLA‐C genotypes combined with fixed maternal KIR types. Expected values were calculated for all combinations and for paternal HLA‐C only. Figure created in https://BioRender.com.

The analysis described was repeated for the KIR/HLA‐C combinations; maternally and paternally inherited HLA‐C genotypes were randomly divided in triplicate over the cases and combined with the fixed KIR genotypes. Expected values were calculated for all possible KIR/HLA‐C combinations and for paternally inherited HLA‐C genotypes only. The Chi‐Squared test was used to examine the differences between the expected and observed degree of HLA (in)compatibility and KIR/HLA‐C combinations. The Chi‐squared test, Mann–Whitney U test, or Fisher's exact test was used where appropriate to evaluate differences between the groups. The Bonferroni method was applied to correct p‐values for multiple testing. A p‐value of < 0.05 was considered statistically significant.

3. Results

3.1. Demographic and Baseline Characteristics

Demographic and baseline characteristics of the included pregnancies (n = 175) are depicted in Table 1. The gestational age, highest diastole and birth weight were significantly different for pregnancies with hypertensive complications compared to uncomplicated pregnancies; inherent to the clinical course of PIH and/or pre‐eclampsia. The age distribution of uncomplicated pregnancies was slightly wider and shifted to older age. A high frequency of (recurrent) pregnancy losses was observed in the included women; the proportion of women with a history of one or more pregnancy losses was 38.9% among the total 175 women. Of these, 35.3% had experienced recurrent pregnancy losses (≥ 2). This corresponded to 13.7% of the total study population (24 out of 175 women). No differences were observed in the number of total pregnancy losses, the occurrence of one or more losses, or recurrent pregnancy losses between women included with uncomplicated pregnancies and those with hypertensive complications.

TABLE 1.

Demographic and baseline characteristics.

		Uncomplicated (n = 125)	Hypertensive complications (n = 50, including pre‐eclampsia)	Pre‐eclampsia (n = 21)	p (Hypertensive complications vs. uncomplicated)
Pregnancy
Gestational age in days, median (range)		280 (259–293)	270.5 (203–291)	261 (203–289)	p < 0.001**
Mode of delivery, amount (%)
Spontaneous		102 (81.6)	36 (72)	14 (66.7)	p = 0.194*
Primary caesarean section		14 (11.2)	6 (12)	4 (19)
Secondary caesarean section		9 (7.2)	8 (16)	3 (14.3)
Highest diastole (mmHg), median (range)		76 (60–100)	100 (74–115)	100 (80–115)	p < 0.001**
Mother
Maternal age (years), median (range)		29 (22–41)	28 (22–39)	27 (22–34)	p = 0.006**
BMI 1st trimester, median (range)		24 (17.8–38.5)	24.8 (18.25–40)	25 (18–37)	p = 0.184**
Gravidity, median (range)		1 (1–5)	1 (1–4)	1 (1–3)	p = 0.297**
Parity, median (range)		0 (0–3)	0 (0–3)	0 (0–3)	p = 0.448**
Pregnancy losses
Total pregnancy losses	38.9%	45	22		p = 0.159
Women with 1 or more pregnancy losses	35.3%	45	23		p = 0.146
Recurrent pregnancy losses (≥ 2)
Primary	58.2%	16	8		p = 0.584
Secondary	41.8%
Child
Gender, %
Female		43.2	48	47.6	p = 0.564*
Male		56.8	52	52.4
Birth weight (gram), median (range)		3524 (2555–4925)	3148 (870–4426)	2920 (870–3982)	p < 0.001**
Birth weight below 5th percentile, %		0.0	14	33.3
Heritage a , %
16/16		61.6	48	47.6	p = 0.096*
14/16		8	22	28.6
12/16		19.2	28	19
8/16		4.8	2	4.8

^a Heritage being the amount of ancestors originating from the founder village until 4th generation.

^* p‐value calculated with Chi‐Square test.

^** p‐value calculated with Mann–Whitney U test. p < 0.05 is considered significant.

3.2. Genotypic Frequencies

Genotypic frequencies of the HLA alleles (including the HLA‐C1 and ‐C2 group) from all mothers and children were tested for the Hardy–Weinberg equilibrium. There was no significant deviation from the Hardy–Weinberg equilibrium, suggesting a random appearance of the genotypic frequencies of all the HLA alleles in this population (data not shown). The children from the uncomplicated pregnancies showed an increased percentage of homozygosity for HLA‐DQB1 compared to their mothers, resulting in an odds ratio of 2.70 (Bonferroni‐corrected p = 0.011, Supporting Table 1). The homozygous HLA frequencies of the mothers with uncomplicated pregnancies of the genetically isolated population were compared to mothers with uncomplicated pregnancies of a larger general Dutch population studied previously (Supporting Table 2) [14]. All expected percentages of homozygosity were higher in the genetically isolated population compared to the larger general population. The Hardy–Weinberg equilibrium and homozygosity analyses were repeated for only those pregnancies of which the child had 16/16 founder heritage, in which no significant differences were observed (data not shown).

3.3. Maternal–Foetal HLA Compatibility in (Un)complicated Pregnancies Shows No Deviation From Expected Patterns

We explored the possibility of a preferential selection for maternal–foetal HLA matching in this genetically isolated population by comparing the amount of observed HLA matches to the amount of expected HLA matches for the uncomplicated pregnancies and pregnancies with hypertensive complications (Table 2). We did not observe significant differences within or between the pregnancy groups, including sub‐analyses for pre‐eclamptic cases only (data not shown).

TABLE 2.

Observed and expected‐by‐chance amount of maternal–foetal HLA matches in uncomplicated pregnancies and pregnancies with hypertensive complications.

Uncomplicated (n = 125)						Hypertensive complications (n = 50)
	Observed (nr. of matches)		Expected (nr. of matches)		p *	Observed (nr. of matches)		Expected (nr. of matches)		p *	p * (observed vs. observed)**
	1	2	1	2		1	2	1	2
HLA‐A	82.4%	17.6%	86.1%	13.9%	1.000	94.0%	6.0%	85.3%	14.7%	0.583	0.333
HLA‐B	93.6%	6.4%	88.5%	11.5%	0.528	88.0%	12.0%	90.7%	9.3%	1.000	1.000
HLA‐C	87.2%	12.8%	83.2%	16.8%	1.000	86.0%	14.0%	84.0%	16.0%	1.000	1.000
HLA‐DRB1	87.2%	12.8%	86.1%	13.9%	1.000	88.0%	12.0%	86.0%	14.0%	1.000	1.000
HLA‐DQB1	80.8%	19.2%	75.7%	24.3%	1.000	88.0%	12.0%	76.0%	24.0%	0.329	1.000
HLA‐DQA1	81.6%	18.4%	78.4%	21.6%	1.000	78.0%	22.0%	82.0%	18.0%	1.000	1.000

^* p‐values calculated by Chi‐Square analysis. p‐values are corrected for multiple analyses (six HLA loci typed and C1/C2 group analysis) using the Bonferroni method.

^**Comparing uncomplicated pregnancies with hypertensive complications. p < 0.05 is considered significant.

3.4. Pregnancies With Hypertensive Complications Are Associated With Higher Maternal–Foetal HLA‐DQB1 Mismatching

In contrast to maternal‐foetal HLA matches, a significantly higher amount of mismatches for HLA‐DQB1 was observed than was expected by chance (Bonferroni‐corrected p = 0.017) in pregnancies with hypertensive complications (Table 3). The same trend was seen in a sub‐analysis for pregnancies complicated by pre‐eclampsia only. p‐values were not calculated due to sample size (n = 21, Supporting Table 3). No differences were found between the groups or for the uncomplicated pregnancies when comparing the observed and expected by chance numbers of HLA mismatches.

TABLE 3.

Observed and expected‐by‐chance amount of maternal–foetal HLA mismatches in uncomplicated pregnancies and pregnancies with hypertensive complications.

Uncomplicated (n = 125)						Hypertensive complications (n = 50)
	Observed (nr. of mismatches)	Expected (nr. of mismatches)			p *	Observed (nr. of mismatches)		Expected (nr. of mismatches)		p *	p * (observed vs. observed)**
	0	1	0	1		0	1	0	1
HLA‐A	36.8%	63.2%	32.8%	67.2%	1.000	28.0%	72.0%	37.3%	62.7%	1.000	1.000
HLA‐B	24.0%	76.0%	21.1%	78.9%	1.000	22.0%	78.0%	20.7%	79.3%	1.000	1.000
HLA‐C	36.0%	64.0%	40.8%	59.2%	1.000	38.0%	62.0%	39.3%	60.7%	1.000	1.000
HLA‐DRB1	28.0%	72.0%	26.1%	73.9%	1.000	24.0%	76.0%	28.0%	72.0%	1.000	1.000
HLA‐DQB1	50.4%	49.6%	48.3%	51.7%	1.000	34.0%	66.0%	55.3%	44.7%	0.017	0.344
HLA‐DQA1	52.0%	48.0%	50.4%	49.6%	1.000	50.0%	50.0%	50.7%	49.3%	1.000	1.000

^* p‐values calculated by Chi‐Square analysis. p‐values are corrected for multiple analyses (six HLA loci typed and C1/C2 group analysis) using the Bonferroni method.

^**Comparing uncomplicated pregnancies with hypertensive complications. p < 0.05 is considered significant.

3.5. KIR/HLA‐C Combination Frequencies Do no Significantly Differ Within and Between Pregnancy Groups

We analysed maternal KIR and foetal HLA‐C genotype frequencies in both pregnancy groups to determine if certain combinations occurred differently than what would be expected by chance (Table 4). Overall analysis of KIR/HLA‐C genotype combinations, including paternally inherited HLA‐C alleles only, showed no differences in observed compared to expected by chance frequencies for both pregnancy groups and in comparing observed values between the groups. No differences were seen in the comparison between uncomplicated and pre‐eclamptic pregnancies for KIR/HLA‐C genotype combination frequencies and frequencies of activating and inhibiting maternal KIR receptors (Supporting Tables 4 and 5).

TABLE 4.

Observed and expected‐by‐chance KIR/HLA‐C combinations for uncomplicated pregnancies and pregnancies with hypertensive complications.

	Uncomplicated (n = 122) a			Hypertensive complications (n = 50)
	Observed (%)	Expected (%)	p *	Observed (%)	Expected (%)	p‐value*	p‐value* (observed vs. observed)**
Genotype							0.794
KIR AA/HLA C1	23.8	22.1	0.900	18.0	21.3	0.970
KIR AA/HLA C2	18.9	18.3		20.0	19.3
KIR Bx/HLA C1	58.2	61.5		62.0	60.0
KIR Bx/HLA C2	41.8	54.3		50.0	52.0
Paternally inherited C1/C2 only							0.452
KIR AA/HLA C1	17.9	20.7	0.982	10.0	14.0	0.643
KIR AA/HLA C2	9.8	10.6		16.0	12.0
KIR Bx/HLA C1	41.5	42.0		42.0	38.0
KIR Bx/HLA C2	30.9	30.6		32.0	36.0

^a For 2 samples, no C1 or C2 group could be allocated and for 1 sample KIR A/B count not be determined; those were excluded for the KIR/HLA analysis.

^* p‐values calculated by Chi‐square analysis. p‐values are uncorrected, since all p‐values are non‐significant; correction has been omitted.

^**Comparing uncomplicated pregnancies with hypertensive complications.

3.6. The Increased Maternal–Foetal HLA‐DQB1 Mismatches in Pregnancies With Hypertensive Complications Is Reflected in the PIRCHE‐II Scores

To predict the potential immunogenic effect of the maternal–foetal HLA mismatches and the role of T cell epitope recognition in pregnancies with hypertensive complications, the PIRCHE‐II scores of the immunising paternal haplotypes were compared with uncomplicated pregnancies (Figure 3). The PIRCHE‐II scores of HLA‐DQB1 (Bonferroni‐corrected p = 0.059) and HLA class II (Bonferroni‐corrected p = 0.054) tended to be higher in the group with hypertensive complications, while this was not observed for other HLA loci, HLA class I or total HLA.

FIGURE 3.

Comparison of PIRCHE‐II scores between pregnancy groups per HLA locus. p‐values are calculated using the Mann–Whitney U test. p < 0.05 is considered significant.

4. Discussion

The complex nature of the interaction between the maternal immune system and the semi‐allogenic foetus during pregnancy and how this relates to reproductive complications continues to be a puzzling phenomenon. In the present study, we investigated whether there is a preferential selection for maternal‐foetal HLA compatibility in pregnancy and its contribution to pregnancy‐induced hypertension and/or pre‐eclampsia in a Dutch genetically isolated population. No preferential selection for maternal–foetal HLA compatibility was observed in the uncomplicated pregnancies. A significant increase in observed versus expected maternal–foetal HLA‐DQB1 mismatches was found in pregnancies with hypertensive complications. In line with this finding, higher PIRCHE‐II scores (prediction of T cell epitopes; mismatched foetal HLA‐derived peptides that can be presented by maternal HLA class II) were observed for HLA‐DQB1 and HLA class II in pregnancies with hypertensive complications compared to uncomplicated pregnancies. No differences were found in analyses on KIR/HLA‐C combination frequencies, including paternally inherited HLA‐C alleles only.

Alongside our finding of an association between maternal–foetal HLA‐DQB1 mismatching and hypertensive complications in pregnancy within this genetically isolated population, there was a significantly higher prevalence of homozygosity for HLA‐DQB1 among the children of the uncomplicated pregnancies, thus linking HLA‐DQB1 compatibility to favourable pregnancy outcomes. Several other studies have reported a link between HLA‐DQ alleles and hypertensive pregnancy complications. Honda et al. found a higher frequency of HLA‐DQB1*04 in pre‐eclamptic women of Japanese origin compared to controls; they speculated that HLA‐DQB1*04‐related amino acid residues may induce Th1 predominance—a characteristic of pre‐eclampsia [26, 27]. In the present study, only 5 women were included with the HLA‐DQB1*04 allele, and therefore, analysis of its association with pre‐eclampsia could not be confirmed. In addition, HLA‐DQB1*02 has been suggested to be in linkage disequilibrium with an HLA‐G polymorphism, which is associated with pre‐eclampsia and recurrent pregnancy loss and affects the immunomodulatory function of HLA‐G in the placenta [28].

Our study shows a significant difference in observed and expected maternal–foetal compatibility in pregnancies with hypertensive complications for the HLA‐DQB1 locus, but not for HLA‐DRB1 or HLA‐DQA1, despite the strong linkage disequilibrium of these loci. Similarly, in studies on solid organ transplantation, HLA‐DQ mismatching is the most immunogenic and potentially most pathogenic mismatch compared to other targets [29]. This may be due to both the α and the β chains being polymorphic, in contrast to HLA‐DR, which leads to increased HLA‐DQ antibody production [29]. Interestingly, PIRCHE‐II scores for HLA class II were similarly increased to those of HLA‐DQB1. This suggests that there is no direct translation from HLA match analysis to epitope presentation, as certain HLA‐derived epitopes—specifically HLA‐DQ—are more immunogenic than others [30, 31].

Other studies have found associations between other HLA class II alleles and pre‐eclampsia. In a Danish cohort, the HLA‐DPB1*04:01:01G allele was more frequent in pre‐eclamptic women, while their offspring showed a lower frequency of the DQA1*01:02:01G allele [32]. DPB1*04:01 is predicted to have low immunogenicity, meaning it has low T cell epitope self‐tolerance. Paradoxically, this makes the maternal immune system more likely to recognise mismatched T cell epitopes in the child as foreign, potentially increasing the risk of an immunogenic response [33]. While trophoblasts normally do not express HLA class II molecules, aberrant expression of HLA‐DR by the syncytiotrophoblast has been observed in placentas of women with pre‐eclampsia [34]. A consequent systemic effect might be enforced as the same research group found circulating syncytiotrophoblast‐derived extracellular vesicles with HLA‐DR in pre‐eclamptic women [35]. In response to placental inflammation, upregulated IFN‐γ may induce HLA class II expression [36].

Even when trophoblasts do not express HLA class II molecules, the mismatched HLA can still be presented by antigen‐presenting cells to maternal CD4⁺ T cells through epitopes presented by maternal HLA class II. For this reason, the potential effect of such T cell epitopes was explored, reflected by the PIRCHE‐II scores. A clear trend of increased PIRCHE‐II scores for HLA class II, and HLA‐DQB1 specifically, was observed in pregnancies with hypertensive complications. Following kidney transplantation, PIRCHE‐II scores for class II‐derived peptides have been associated with T‐cell‐mediated rejection [37]. Furthermore, it has been shown that antibodies to HLA mismatches are specific for epitopes rather than antigens [38]. We hypothesised that maternal–foetal HLA‐DQB1 mismatches can result in increased indirect CD4⁺ T cell allorecognition and negatively affect the balance of effector and regulatory T cell populations. Multiple studies have shown that HLA mismatching affects CD4⁺ T cell populations in pregnancy. For instance, maternal–foetal HLA‐C mismatching is associated with increased numbers of CD4⁺CD25^dim activated T cells and functional CD4⁺CD25^bright regulatory T cells at the maternal–foetal interface [13]. T cells with direct specificity for paternal allogeneic MHC, expressed by mouse trophoblast cells, have been observed in mice [39]. Interestingly, whether a mismatched T cell epitope leads to increased effector or regulatory T cells may depend on the specific mismatch [30]. Although we did not find a clear relationship between HLA‐C matching and pre‐eclampsia in this genetically isolated population, this may be due to factors such as the broader classification of hypertensive complications, which groups gestational hypertension and pre‐eclampsia together, as well as the relatively small sample size of pre‐eclampsia cases. These limitations could have diluted the results, making it harder to detect a robust association. Additionally, the heterogeneity of pre‐eclampsia itself should be considered [40]. However, the present study showed similar trends to previous research in (paternally‐derived) HLA‐C2 [11, 12, 14]. Larger cohort studies and replication in other inbred populations are needed to confirm these findings. It is hypothesised that the difference between the potential advantageous or disadvantageous effect of HLA mismatching strongly depends on the HLA loci and the corresponding effect on the maternal immune system. Disproportional populations of regulatory and effector T cells are a well‐recognised feature of pre‐eclampsia [41, 42]. T cell responses to maternal–foetal HLA incompatibility appear to be crucial in its pathophysiology, emphasising the importance of T cells developing into the appropriate phenotype to prevent an aberrant immune response.

It remains unknown if increased predicted CD4⁺ T cell recognition leads to B cell activation and subsequent antibody production. OD pregnancies, characterised by higher HLA mismatching and an increased risk of PIH and pre‐eclampsia, provide a model to study maternal–foetal interactions [43]. While uncomplicated OD pregnancies show increased HLA antibody production related to HLA‐DR mismatches, pre‐eclamptic cases were linked to HLA class II mismatches without corresponding antibody levels, suggesting a cellular rather than humoral immune mechanism [44, 45]. Compared to the current genetically isolated population, OD introduces the artificial scenario of high HLA incompatibility. Interestingly, in a comparable study on OD pregnancies, we observed higher maternal–foetal HLA matching than expected by chance and similar to that seen in naturally conceived pregnancies, which was associated with uncomplicated pregnancies [46]. Thus, a certain degree of HLA matching could potentially enhance the chance of a successful (OD) pregnancy.

The results of this study suggest that HLA allele frequencies appear randomly in uncomplicated pregnancies, with no deviation from Hardy–Weinberg equilibrium, even in a genetically isolated population [14]. Despite suggested increased parental HLA similarity, the HLA haplotypes within the studied population seem to be sufficiently polymorphic to not require preferential selection on HLA (in)compatibility for successful pregnancies. Nevertheless, the incidence of pregnancy‐related hypertensive complications remains unknown for this specific population. The present methodology of retrospective selection of pregnancies with live birth eliminates possible effects of maternal–foetal HLA compatibility on conception. However, we did notice a remarkably high frequency of (recurrent) pregnancy losses in this genetically isolated population, suggesting an advantage of greater disparity between maternal and paternal HLA genotypes for ongoing pregnancy [47]. Noteworthy, the four autosomal‐recessive genetic disorders that occur relatively frequently in the studied population are not related to an increased rate of pregnancy losses. Previous studies have reported similar findings, linking recurrent pregnancy loss to an increased degree of parental HLA sharing [14, 26]. However, the data remain inconsistent due to heterogeneous patient populations, varying HLA typing methods, and other contributing factors.

A limitation of this study lies in its sample size. For pre‐eclampsia, p‐values and some percentages are not provided due to the small study arm of 21 participants, where a single individual accounts for nearly 5% of the difference, making it difficult to draw strong conclusions from these numbers. Nevertheless, this sample size is relatively large for this specific population and compares favourably to other studies in the field. Furthermore, we selected the composite outcome of hypertensive complications because PIH and pre‐eclampsia share a common pathophysiology, with PIH being part of the diagnostic criteria for pre‐eclampsia or as a potential precursor [2].

Maternal–foetal HLA class II mismatching may be one of the contributing factors in the cascade leading to pre‐eclampsia, serving as an initiating factor or an additional trigger that escalates processes, depending on other factors that determine whether it progresses to clinically symptomatic disease. As discussed, there is a diversity of associations between specific HLA alleles and pre‐eclampsia that varies by population, further underscoring their role in the pathophysiology of the disease. This highlights the need for more in‐depth HLA typing; improvements have been made over the years, and it is important to note that each (mis)match is immunogenically distinct. Future studies may also benefit from incorporating non‐classical HLA class Ib molecules (e.g., HLA‐G, and ‐E), which are known to play key roles in immune tolerance and placental development through NK and T cell interactions at the foetal–maternal interface [48]. Additionally, accumulating evidence shows that pre‐eclampsia is a heterogeneous disease, suggesting the necessity for classification into several subtypes, as this heterogeneity dilutes the effects and leads to incomparable cohorts [49, 50]. Furthermore, while PIH may not always serve as a precursor to pre‐eclampsia, thus influencing its impact, we observed similar trends in gestational hypertension as we did in pre‐eclampsia alone.

In conclusion, understanding the underlying mechanisms of maternal–foetal immune interactions is crucial for elucidating the pathways that lead to both healthy and complicated pregnancies. This study's findings on high maternal–foetal HLA‐DQB1 mismatches emphasise their potential role in hypertensive complications and pre‐eclampsia, suggesting that mismatching of specific HLA alleles may influence pregnancy outcomes. These insights can significantly contribute to risk assessment for pre‐eclampsia and may help inform future research and clinical approaches.

Author Contributions

L.j.v.H., I.B.M. and L.E.E.L.O.L. wrote the study protocol. I.B.M. provided data from the outpatient clinic and the maternal DNA samples. L.j.v.H. and S.M. contacted patients, structured the database, collected the clinical samples and data, analysed the data, and performed statistical analyses. G.W.H. processed raw data, performed and supervised statistical analyses. E.T.M.P. and E.S. provided and supervised the PIRCHE‐II data. H.d.V. and W.O. facilitated patient contact in the midwifery practice. J.D.H.A. performed qPCR experiments. P.J.E.L. and D.L.R. enabled HLA genotyping. I.B.M. and L.E.E.L.O.L. are the primary editors and supervisors of the manuscript. L.E.E.L.O.L. conceptualising the study. All authors contributed to the manuscript and approved the submitted version.

Ethics Statement

The study protocol (nr. 23‐3033) was approved by the ethics committee of the LUMC and conducted according to the principles outlined in the Declaration of Helsinki. Informed consent was obtained from all participants prior to sample collection, and data were anonymised to ensure confidentiality.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Hof L. J. v., van der Hoorn M.‐L. P., Migdis S., et al., “Maternal–Foetal HLA‐DQB1 Incompatibility Is Associated With Pregnancy‐Induced Hypertensive Disorders in a Genetically Isolated Population,” HLA 106, no. 2 (2025): e70374, 10.1111/tan.70374.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.