BACKGROUND Uterine natural killer cells (uNK) are the most abundant lymphocytes found in the decidua during implantation and in first trimester pregnancy. They are important for early placental development, especially trophoblast invasion and transformation of the spiral arteries. However, inappropriate uNK function has been implicated in reproductive failure, such as recurrent miscarriage (RM) or recurrent implantation failure (RIF). Previous studies have mainly focussed on peripheral NK cells (pNK), despite the well-documented differences in pNK and uNK phenotype and function. In recent years, there has been an explosion of studies conducted on uNK, providing a more suitable representation of the immune environment at the maternal–foetal interface. Here, we summarize the evidence from studies published on uNK in women with RM/RIF compared with controls. OBJECTIVE AND RATIONALE The objectives of this systematic review and meta-analysis are to evaluate: differences in uNK level in women with RM/RIF compared with controls; pregnancy outcome in women with RM/RIF stratified by high and normal uNK levels; correlation between uNK and pNK in women with RM/RIF; and differences in uNK activity in women with RM/RIF compared with controls. SEARCH METHODS MEDLINE, EMBASE, Web of Science and Cochrane Trials Registry were searched from inception up to December 2020 and studies were selected in accordance with PRISMA guidelines. Meta-analyses were performed for uNK level, pregnancy outcome and uNK/pNK correlation. Narrative synthesis was conducted for uNK activity. Risk of bias was assessed by ROBINS-I and publication bias by Egger’s test. OUTCOMES Our initial search yielded 4636 articles, of which 60 articles were included in our systematic review. Meta-analysis of CD56+ uNK level in women with RM compared with controls showed significantly higher levels in women with RM in subgroup analysis of endometrial samples (standardized mean difference (SMD) 0.49, CI 0.08, 0.90; P = 0.02; I2 88%; 1100 women). Meta-analysis of CD56+ uNK level in endometrium of women with RIF compared with controls showed significantly higher levels in women with RIF (SMD 0.49, CI 0.01, 0.98; P = 0.046; I2 84%; 604 women). There was no difference in pregnancy outcome in women with RM/RIF stratified by uNK level, and no significant correlation between pNK and uNK levels in women with RM/RIF. There was wide variation in studies conducted on uNK activity, which can be broadly divided into regulation and receptors, uNK cytotoxicity, cytokine secretion and effect of uNK on angiogenesis. These studies were largely equivocal in their results on cytokine secretion, but most studies found lower expression of inhibitory receptors and increased expression of angiogenic factors in women with RM. WIDER IMPLICATIONS The observation of significantly increased uNK level in endometrium of women with RM and RIF may point to an underlying disturbance of the immune milieu culminating in implantation and/or placentation failure. Further research is warranted to elucidate the underlying pathophysiology. The evidence for measuring pNK as an indicator of uNK behaviour is sparse, and of limited clinical use. Measurement of uNK level/activity may be more useful as a diagnostic tool, however, a standardized reference range must be established before this can be of clinical use.

COVID-19 vaccination protects against the potentially serious consequences of SARS-CoV-2 infection, but some people have been hesitant to receive the vaccine because of reports that it could affect menstrual bleeding. To determine whether this occurs we prospectively recruited a cohort of 79 individuals, each of whom recorded details of at least three consecutive menstrual cycles, during which time they each received at least one dose of COVID-19 vaccine. In spontaneously cycling participants, COVID-19 vaccination was associated with a delay to the next period, but this change reversed in subsequent unvaccinated cycles. No delay was detected in those taking hormonal contraception. To explore hypotheses about the mechanism by which these menstrual changes occur, we retrospectively recruited a larger cohort, of 1,273 people who had kept a record of their menstrual cycle and vaccination dates. In this cohort, we found a trend toward use of combined hormonal contraception being protective against reporting a delayed period, suggesting that menstrual changes following vaccination may be mediated by perturbations to ovarian hormones. However, we were unable to detect a clear association between the timing of vaccination within the menstrual cycle and reports of menstrual changes. Our findings suggest that COVID-19 vaccination can lengthen the menstrual cycle and that this effect may be mediated by ovarian hormones. Importantly, we find that the menstrual cycle returns to its pre-vaccination length in unvaccinated cycles.

Uterine natural killer cells (uNK) play an important role in promoting successful pregnancy by regulating trophoblast invasion and spiral artery remodelling in the first trimester. Recently, single-cell RNA sequencing (scRNAseq) on first-trimester decidua showed that uNK can be divided into three subsets, which may have different roles in pregnancy. Here we present an integration of previously published scRNAseq datasets, together with novel flow cytometry data to interrogate the frequency, phenotype, and function of uNK1–3 in seven stages of the reproductive cycle (menstrual, proliferative, secretory phases of the menstrual cycle; first, second, and third trimester; and postpartum). We found that uNK1 and uNK2 peak in the first trimester, but by the third trimester, the majority of uNK are uNK3. All three subsets are most able to degranulate and produce cytokines during the secretory phase of the menstrual cycle and express KIR2D molecules, which allow them to interact with HLA-C expressed by placental extravillous trophoblast cells, at the highest frequency during the first trimester. Taken together, our findings suggest that uNK are particularly active and able to interact with placental cells at the time of implantation and that uNK1 and uNK2 may be particularly involved in these processes. Our findings are the first to establish how uNK frequency and function change dynamically across the healthy reproductive cycle. This serves as a platform from which the relationship between uNK function and impaired implantation and placentation can be investigated. This will have important implications for the study of subfertility, recurrent miscarriage, pre-eclampsia, and pre-term labour.