To the best of our knowledge, this is the first study on the correlation between pbNK and autoimmune antibodies in unexplained recurrent miscarriage. Previous studies demonstrated that pbNK is recruited to the decidua during pregnancy and that enhanced cytotoxicity or an increased number of pbNK cells could directly attack and have lethal effects on embryonic trophoblast cells, resulting in insufficient recasting of uterine spiral arterioles (23), which is related to miscarriage. NK cells play a continuous role in pregnancy loss (24), and an increase in the number of pbNK cells is considered to be related to a higher miscarriage rate. However, some investigators demonstrated that numbers of pbNK cells did not increase in patients with RM when compared to controls (15, 25). Our early study confirmed found that the number of pbNK cells did not increase with increasing number of miscarriages in patients with RM (22).
Previous studies have found that the increased number of pbNK cells in patients with RM is related to APS (19) and autoimmune thyroiditis (20, 21). Considering that previous studies including our published one did not strictly exclude patients with autoimmune diseases such as APS and autoimmune thyroiditis, which led to an inaccurate conclusion that the number of pbNK cells in patients with RM is higher than that in the control group, we excluded patients with autoimmune thyroiditis as well as other autoimmune diseases, including APS, SLE, immune vasculitis, Sjogren's syndrome and CTD, due to their possible impact on the pbNK cell count. Our results showed that after excluding patients with autoimmune diseases, the number of pbNK cells and their subsets did not increase with increasing number of miscarriages. This study completely excluded patients with autoimmune diseases, so the conclusion is more reliable.
There are many previous studies on the correlation between positive autoantibodies and RM, most of which focus on APS and immune thyroiditis; however, there are few studies on the correlation between simple positive autoantibodies and RM with the exclusion of subjects with APS and all diagnosed autoimmune diseases, and the results are inconsistent (26). The sample size was relatively large among the published studies, which excluded all the known effects on RM that may be related to autoantibody positivity. This study was based on a RM population and excluded patients with autoimmune diseases but included patients who simply tested positive for antibodies. These patients did not meet the diagnostic criteria for autoimmune diseases consisting of autoimmune thyroiditis, APS, SLE, immune vasculitis, Sjogren's syndrome or CTD after a medical consultation with the Department of Rheumatology but merely had one or more test-positive antibodies.
Previous studies found that the increased pbNK cell count in patients with RM was related to positive antiphospholipid antibody and positive TPO-Ab (20, 27). The possible mechanisms were twofold: 1. The production of antiphospholipid antibodies and antithyroid antibodies induces an immune response, resulting in the activation and promotion of NK cells, and these positive antibodies, together with pbNK cells, jointly provoke pregnancy loss; 2. The production of these antibodies strengthens the role of elevated pbNK cells with high cytotoxicity, which is associated with pregnancy loss. The sample size of our study was large, and patients with autoimmune diseases, including subacute APS and thyroiditis, which are known to cause RM, were excluded. Known effects directly on changes in pbNK cell counts were excluded. Additionally, positivity for three autoantibodies, namely, ANA, TPO-Ab and Tg-Ab, which were not simultaneously assessed in previous studies, was evaluated, and the results were more reliable. Additionally, this study completely excluded patients with other known causes of RM and reduced the interference of factors that may lead to changes in the number of pbNK cells. However, the level of pbNK cells has been confirmed to be affected by factors such as menstrual cycle or steroid use (28). Although patients taking steroids within 3 months were excluded from this study, whether all patients underwent pbNK cell examination at the same day of the menstrual cycle was not guaranteed, which was a possible defect of this study. But we set a time limit for the peripheral blood collection within early follicular phase of menstrual cycle (Day 2 to Day 5), when the levels and fluctuations of reproduction-related hormones exert little effect on the NK cells.
Multivariable logistic regression suggested that positive TPO-Ab was a promoting factor for pbNK cell count, while ANA was an inhibiting factor for pbNK cell count. Logistic regression that using candidate variables including age, BMI, basal levels of FSH, basal levels of LH and estradiol, TSH, fT3, fT4, tT3, tT4 and quantitative results of ANA, TPO-Ab and Tg-Ab were also performed and no significant different was found (Supplementary table 1). The results suggested that the production of autoantibodies rather than their titers has an effect on the fluctuation of pbNK cells. Changes in the pbNK cell count are a result of the miscarriage-associated autoimmune response rather than a causal factor for RM.
We presume that patients with positive antithyroid antibodies are potential patients with autoimmune diseases, and clinical manifestations often proceed by the presence of characteristic organ-specific antibodies that might occur in serum even a few years before symptom onset and diagnosis (29). There is literature showing that a high number of patients with juvenile rheumatoid arthritis have almost the complete absence of NKbright cells in circulation (30); moreover, it was found that the number of pbNK cells in patients with autoimmune arthritis does not increase but decreases (31). The increased number of pbNK cells in uRM patients who tested positive for TPO-Ab and ANA are likely to be the responses to autoimmune abnormalities, and the production of TPO-Ab and ANA indicate the onset of symptoms to clinically diagnosed thyroid or immune diseases over time, when the normal pbNK cell count of the patients with RM may be a compensatory response rather than a real pathological change.
Hence, the increased number of pbNK cells is not a factor of uRM and cannot be applied as a therapeutic index in patients with uRM. Endocrine therapy targeted at thyroid function under the condition of dynamic monitoring during pregnancy in uRM patients with both simple positive TPO-Ab and an increased number of pbNK cells may be of more clinical value. Further full-scale follow-up of pregnancy outcomes is needed to reach clearer conclusions to guide our daily clinical work. We hoped to know the pregnancy outcomes of uRM patients with simple positive TPO-Ab and elevated pbNK cell counts after treatment compared with untreated patients; however, we found that, driven by fear of reoccurrence of miscarriage and excessive anxiety, most of the patients who have been followed up to now have undergone diverse and inconsistent therapies such as immunoglobulin, fat emulsion, hormones or immunosuppressants in different medical institutions, with few patients remaining untreated, which increases the complexity and bias of the results and makes it difficult to conduct further analysis as planned. Subsequent studies should pay attention to dynamic monitoring of thyroid function before and during pregnancy rather than to the decline in pbNK cell count or autoantibody titers, to determine the diversion from immunotherapy to endocrine therapy directed at thyroid function in uRM patients with both simple positive TPO-Ab and an increased number of pbNK cells, as this would be of more clinical value. Changes in pbNK cell count are a result of the miscarriage-associated autoimmune response rather than a causal factor for RM. Hence, the increased number of pbNK cells in patients with uRM cannot be applied as a therapeutic index for immunological abnormalities in patients with uRM.
In conclusion, changes in pbNK cell count are considered to be a result of the miscarriage-associated autoimmune response rather than a causal factor for RM. Hence, the increased number of pbNK cells in patients with uRM cannot be applied as a therapeutic index for immunological abnormalities in patients with uRM.