Autoimmune diseases are manifestations of immune system dysregulation caused by a humoral or cell-mediated immune response against self-antigens and consequently loss of self-tolerance. This heterogenous group of diseases shares some common pathophysiological mechanisms and risk factors which involve genetic, environmental, immunological, and hormonal factors that contribute to the tissue injury and clinical manifestations (1). Among the most relevant and studied genetic components in autoimmune diseases are the loci coding for human leucocyte antigens (HLA) class I and II, which are located on the short arm of chromosome no. 6. In particular, HLA class II alleles - and more specifically HLA-DRB1 alleles - are considered to be more strongly involved in the susceptibility to or protection against specific autoimmune diseases than HLA class I (2, 3). However, strong associations between HLA alleles and autoantibody positivity also exist, and antibody positivity often precedes clinical manifestation of autoimmune disease by several years or decades. Therefore, one may question whether the primary role of HLA genes is predisposition to the disease through non-humoral pathways or antibody development that secondarily leads to autoimmune diseases (4).

One example of an association between HLA alleles and autoantibody positivity, is anti-citrullinated protein antibodies in patients with rheumatoid arthritis. These antibodies are mainly found in patients with a specific 5 amino acid motif located in the β-chain of the HLA-DRB1 alleles that predisposes to rheumatoid arthritis (5). Other examples are that specific HLA-DQ alleles are associated with disease-specific transglutaminase 2 antibodies in celiac disease (6); HLA-DRB1*0401 is strongly associated with autoantibodies against insulin and islet antigen 2 proteins (7); while the HLA class II haplotype HLA-DRB1*03-DQB1*02 is associated with glutamic acid decarboxylase antibodies in type 1 diabetes mellitus (8). However, the genetic background for development of autoimmune disease and autoantibody production is complex. In addition to the HLA genes, non-HLA related genetic factors and environmental factors play a role (2).

One of the non-HLA genetically determined biomarkers of autoimmunity is mannose-binding lectin (MBL) deficiency, which is mainly caused by the interaction of several gene polymorphisms on chromosome no. 10. Plasma (p-)MBL deficiency has been associated with the development and severity of several autoimmune diseases such as lupus erythematosus and rheumatoid arthritis (9–11).

In accordance with the Recurrent Pregnancy Loss (RPL) guideline of the European Society of Hunan Reproduction and Embryology (12), RPL is in this study defined as 2 or more consecutive pregnancy losses but the definition of RPL varies between different RPL guidelines and specialist societies (12, 13). In about 50% of patients, thrombophilia or endocrine, anatomic, or genetic risk factors can be found, while in the remaining subgroup of patients an aberrant immune profile is often found (14, 15). This involves autoantibody positivity or disrupted leucocyte subset distribution or function. Anti-nuclear antibodies (ANA), anti-thyroid peroxidase (TPO) antibodies and anti-phospholipid (APL) antibodies are associated with significantly increased odds ratios (OR) for RPL (16–20). Maternal carriage of HLA-DRB1*03 (21) and HLA-DRB1*07 (22) alleles and p-MBL deficiency (23, 24) have also been reported to be associated with RPL.

Due to the well-known association between HLA class II alleles and autoantibody production, as well as the association between p-MBL deficiency and some autoimmune diseases, we found it important to investigate whether autoantibody formation is an intermediate variable between the two RPL-associated genetic factors HLA-DRB1 and low p-MBL and RPL.

In total, 644 RPL patients were included of whom 88.7% had three or more pregnancy losses. The majority had pregnancy losses after natural conception while 36.2% were supported with assisted reproductive technology (ART) treatment including in-vitro fertilization (IVF) and intra-cytoplasmatic sperm injection (ICSI).

The frequencies of the HLA-DRB1*03 and -DRB1*07 phenotypes in all RPL patients were 20.8.% and 19.6%, respectively, compared with 24.9% and 18.4%, respectively, in a Danish bone-marrow donor cohort (p=0.034; OR = 0.79, 95% CI 0.64-0.98 and p=0.532; OR 1.08, 95% CI 0.87-1.35 for comparisons between patients and controls with each of the two phenotypes) (22).

When comparing baseline characteristics between patients with low or normal p-MBL levels, the HLA-DRB1*03 and/or HLA-DRB1*07 phenotype, no clinically relevant or statistically significant differences were observed ( Table 1 ).

The frequency of each autoantibody or autoantibody class did not vary between patients with low or normal p-MBL levels, and presence of the HLA-DRB1*03 or HLA-DRB1*07 phenotype, respectively. ( Table 2 ). We observed non-significantly higher frequencies of ≥2 autoantibodies in patients with low p-MBL level compared to normal p-MBL level (p=0.14) and in patients with minimum one HLA-DRB1*03 allele compared with patients with no HLA-DRB1*03 allele (p=0.11). Also, a tendency was found for a lower frequency of APL antibodies among patients with minimum one HLA-DRB1*07 allele compared to patients with no HLA-DRB1*07 allele (p=0.08) ( Table 2 ).

Among patients who who were positive for one or more autoantibodies and had low p-MBL level, the frequency of carrying an HLA-DRB1'03 allele was higher than among those with one or more autoantibodies and normal p-MBL level (OR = 2.4, 95% CI 1.2-5.0; p = 0.01) ( Table 3 ). All other pairwise comparisons of HLA-DRB1 phenotype frequencies between subsets of patients divided according to presence/absence of autoantibodies and p-MBL levels showed no significant differences ( Table 3 ). Moreover, when examining the HLA-DRB1*03/X genotypes, no differences between patients with low or normal p-MBL in women without autoantibodies were observed. However, among patients positive for ≥1 autoantibody, notable differences in HLA-DRB1*03/X genotypes were seen between patients with low or normal p-MBL levels; however, none of these differences were significant. ( Table 4 ).

ANAs were found with increased prevalence in patients with the combination of HLA-DRB1*03 and low p-MBL level compared with those without this combination (OR 2.0; 95% CI 1.0 - 3.9; p <0.05). The presence of the all other autoantibodies was also increased in patients with HLA-DRB1*03 and low p-MBL (ORs between 1.5 and 1.7) but these associations were not statistically significant ( Table 5 ).

In this study, we found that the combination of carriage of HLA-DRB1*03 and p-MBL levels ≤500 µg/l is associated with presence of autoantibodies especially ANAs in patients with RPL.

In previous large case-control studies, we have reported that HLA-DRB1*03 and HLA-DRB1*07 are conferring susceptibility to RPL (21, 22). Only the studies published by the groups of Christiansen et al. found the HLA-DRB1*03 frequency significantly increased in women with RPL compared with controls (21, 25). In a meta-analysis of 18 case-control studies investigating HLA-DR phenotype frequencies in RPL patients and controls, exclusion of patients positive for autoantibodies was done in 5 of the 7 studies finding HLA-DRB1*03 less frequent in RPL patients than in controls. In contrast, exclusion of autoantibody-positive patients was not done in any of the studies finding HLA-DRB1*03 increased in RPL patients (26). Thus, such selection bias may have distorted the observed HLA-DR distribution in the RPL cohort, and it is likely that the meta-analysis would have shown HLA-DRB1*03 significantly associated with RPL if a sensitivity analysis had been done excluding the 5 studies where autoantibody positive RPL patients were excluded.

The overall frequency of the HLA-DRB1*03 phenotype in RPL women in this study was lower than the frequency in a large previously published cohort of Danish bone marrow donors (22). We have previously reported an increasing population incidence of RPL (27) and a declining frequency of HLA-DRB1*03 among RPL patients (22). The declining frequency of the allele in RPL patients was confirmed in the present study. This pattern of temporal weakened effect of some susceptibility HLA alleles in diseases has also been described for type I diabetes mellitus (28, 29). It has been speculated whether the declining frequency of susceptibility HLA alleles in patients together with the increasing population incidence of type I diabetes are due to an increasing proportion of diabetes type I cases being caused by environmental and lifestyle changes (28, 29). In analogy with the studies in diabetes type I, we speculate whether HLA susceptibility alleles (and especially HLA-DRB1*03) with time play a declining role whereas environmental or lifestyle factors could play an increasing role in the RPL pathogenesis. However, a deeper insight into the complex and probably multifactorial pathophysiology causing RPL is needed before we can make a realistic estimate of the contribution of immunogenetic, environmental, and lifestyle factors to RPL.

More simple explanations for the failure to confirm the association previously found between HLA-DRB1*03 and RPL (21) are possible. A small statistical power may explain why the present study did not detect the association, however, now two large, independent studies of non-overlapping cohorts including 1078 (22) and 663 (this study) RPL patients, respectively, have not been able to confirm the association. Therefore, the lack of statistical power is a less likely explanation. Another possible explanation is a changed referral practice. The vast majority of the RPL patients in the prior study finding an association (21) had conceived spontaneously and had a history of predominantly clinical miscarriages confirmed by ultrasound scans whereas approximately 1/3 of RPL patients in this study had conceived by ART (IVF or ICSI) and an increased proportion of their pregnancy losses were biochemical pregnancies due to the change in the definition of RPL by European Society of Human Reproduction and Embryology’s Guideline group in the meantime between the two studies. The immunological component in the pathogenesis of the losses in our current RPL patients may have been weakened by such changes in patient characteristics.

We have in repeated studies in different populations of RPL patients and controls found that low levels of p-MBL are associated with increased risk of RPL with ORs between 1.46 and 1.79 (23, 24). Moreover, the most recent study high p-MBL levels (≥3000µg/l) seemed to protect against RPL (23). In the present study, most blood samples were taken prior to pregnancy whereas 10.2% were taken in the first trimester. A previous study found that only a minor elevation of p-MBL levels could be detected from before pregnancy to the first trimester (30). Since the huge majority of our patients’ pregnancy losses happen in the first trimester, we believe that the measurements made before or in early pregnancy reflect the p-MBL levels in the high-risk period for pregnancy loss and therefore, the MBL measurement from both timepoints are relevant. Although being considered an acute-phase protein (31), p-MBL only seems to increase in response to infections in individuals with MBL-2 genotypes predisposing to high p-MBL levels (31) so low p-MBL levels <500 µg/l are considered to be stable also during infection/inflammation.

In patients with unexplained RPL, various autoantibodies can be detected with higher prevalence than in controls, and they are thought to have a negative prognostic impact. In the European Society of Human Reproduction and Embryology’s Guideline for RPL (12), it is recommended to screen RPL patients for LA, anti-TPO antibodies, ANAs, and APL antibodies – and among APL antibodies especially aCL antibodies since plenty of studies have documented the association between aCL antibodies and RPL whereas only few studies have documented the association between β2GPI antibodies and RPL (4). Although these autoantibodies seem to affect the prognosis in RPL patients negatively, it is unclear whether they act directly to the fetus and trophoblast and cause miscarriage, whether they are markers for an overall dysregulated immune response towards pregnancy, or whether they are confounding factors associated with more basic risk factors such as genetic factors of importance for immune regulation.

As HLA-DRB1 polymorphism and MBL deficiency are both associated with autoimmunity and with RPL separately, the natural question arises whether autoantibody development is an intermediate or causal factor between the immunogenetic profile and occurrence of RPL. We tried to answer this question in a former study: In a case-control study of Danish and Czech patients with RPL, Christiansen et al. (32) found a significant difference between the prevalence of patients being HLA-DRB1*03 positive among ANA positive and ANA negative patients (55% vs 28%, p <0.05) and a significant difference in the prevalence of the allele between aCL antibody-positive patients and the background population (35% vs 21%, p < 0.05). It was concluded that HLA-DRB1*03 is a susceptibility gene for production of the two autoantibodies in Caucasian RPL women.

We performed an updated study on the association between relevant immunogenetic biomarkers and autoantibody production in a larger population of RPL patients and tested more associations than previously: between 3 genetically determined biomarkers - two HLA class II gene variants and MBL deficiency against 3 classes of autoantibodies. The study illustrates that several genetic variants interact in the pathogenesis of autoimmunity in RPL patients. Essentially, we confirmed the previously reported statistically significant association between ANA and HLA-DRB1*03 but now only in the subset of patients with MBL deficiency. This disparity may reflect the previously mentioned weakened effect of HLA class II susceptibility alleles over time seen in some autoimmune disorders.

Although we have recently reported that HLA-DRB1*07 is significantly associated with RPL (22), it is evident that this allele is not associated with production of the tested autoantibodies ( Table 2 ). The pathophysiological mechanisms causing RPL in HLA-DRB1*07 positive patients seem not to involve autoantibody production but may instead be mediated by HLA-DRB1*07 restricted T-cells directed against male-specific minor histocompatibility antigens on the fetus or trophoblast (33).

No study has so far clarified which mechanisms underly the association between RPL and low plasma MBL. There could be several mechanisms. One mechanism may be related to MBL’s ability to bind to and promote clearance of necrotic and apoptotic cells and debris probably through opsonizing these cells and particles and promoting phagocytosis by mainly macrophages (34). If apoptotic trophoblast cells or debris shed from the placenta are not rapidly cleared by this mechanism, they may instead be phagocytosed by endometrial endothelial cells, which may result in increased release of proinflammatory cytokines (35). The increased proinflammatory response may injure the endothelial cells in placenta and consequently cause vascular clotting and trophoblast/fetal ischemia. Thus, pregnant women with low p-MBL may be at a higher risk for harmful proinflammatory responses at the feto-maternal interface due to delayed clearance of apoptotic trophoblast cells/debris.

MBL may also, in theory, affect antibody/autoantibody production through more direct ways since it can bind to and influence B lymphocytes (36) and since mice studies have found increased proliferation and IgG production in MBL null mice compared to wild type in response to Group B Streptococcus vaccination (37).

Another mechanism may be related to the results from previous reports suggesting that blocking asymmetric IgG antibodies possessing mannose-rich oligosaccharide residues in the Fab residue may play a role in RPL since the levels of such antibodies increase during normal pregnancy but not in pregnancies of women with RPL (38). MBL may bind to the mannose-rich residues of the asymmetrical antibodies enhancing their blocking effect (also on autoantibody production) whereas in women with low p-MBL, the blocking effect of asymmetrical antibodies may be reduced.

More experimental studies are needed to test these hypotheses. In our Center, we are currently performing studies in order to rule out whether the clearance rate of fetal and trophoblast antigens after a delivery is depending on the p-MBL level.

The finding in this study that an interaction of HLA-DRB1*03 and MBL is associated with the presence of autoantibodies in RPL is new. Only a weak association was found between HLA-DRB1*03 and ANA, and no overall association between this allele and the other investigated autoantibodies. These findings may be surprising since previous studies have found HLA-DRB1*03 being associated with the presence of a series of autoimmune diseases. However, almost all previous studies have focused on patients with clinical autoimmune disease (39–42) where the autoimmune reactions causing their clinical symptoms are more pronounced than in otherwise healthy RPL patients who typically is less or not at all affected in terms of clinical symptoms when not pregnant potentially because of lower autoantibody titers. This may explain the absent/weak association to solitary HLA-DRB1*03 found this study ( Table 2 ). The finding that no other HLA-DRB1 allele than HLA-DRB1*03 seems to interact with low MBL is a new but negative finding, however, it does contribute to our knowledge about the pathogenesis of RPL. New independent studies are, however, needed to confirm these results.

It is recommended that a panel of relevant immunogenetic variants are investigated in patients with RPL and that models are developed for the seemingly complex interaction between the variants and the risk of RPL.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.

CN-P collected data and performed statistical analyses. CN-P and OC analyzed the results and wrote the article. RS was responsible for the HLA genotyping and p-MBL measurements. UK critically reviewed the manuscript and the statistics. All authors contributed to the article and approved the submitted version.