Abstract
Study question
Do women with recurrent miscarriage (RM) or implantation failure (RIF) have different levels of uterine Natural Killer (NK) cells compared to fertile controls?
Summary answer
Women with RIF but not RM are associated with significantly higher levels of CD56+ uterine NK cells compared to controls.
What is known already
Uterine NK cells (uNK) are different from peripheral NK cells (pNK) and are important in early pregnancy for development of the placenta. The association between uNK and RM/RIF is less clear, but dysfunction of uNK is believed to result in early pregnancy failure. Previous systematic reviews by Seshadri (2014) and Tang (2013) on infertile and RM patients showed no significant difference in uNK levels and highlighted need for further studies. Since, many prospective studies have been published and therefore warrant an updated systematic review. On the other hand, evidence for correlation between uNK and pNK is sparse and needs clarification.
Study design, size, duration
We have conducted a systematic review and meta-analysis to evaluate three outcomes. The primary outcome was the difference of uNK level in RM/RIF compared to controls. The secondary outcome was livebirth rate in women with RM/RIF with high compared to normal uNK level, and the tertiary outcome was correlation between uNK and pNK in RM/RIF.
Participants/materials, setting, methods
The electronic database search included MEDLINE, EMBASE, Web of Science and bibliographies from included articles from inception to December 2020 using a combination of MESH and keywords. Search, screen, and data extraction were performed by two reviewers independently. Quality assessment was conducted with ROBINS-I and meta-analysis with Revman 5.3. Out of 4636 studies screened, 43 studies (2539 women) and 3 studies each (598 and 77 women) were analysed for primary, secondary and tertiary outcomes respectively.
Main results and the role of chance
Our meta-analysis showed that CD56+ uNK were significantly higher in women with RIF but not RM compared to controls (SMD 0.60; 95% CI 0.12–1.08]. Subgroup analysis in RM patients showed no significant difference whether definition of 2 or 3 previous RM was used, in primary/secondary RM compared to controls, or in primary versus secondary RM. CD56+ uNK were significantly higher in RM/RIF when sampled during mid-luteal phase [SMD 0.56; 95% CI 0.19–0.93] but not in the early pregnancy decidua. Interestingly, there was significant difference in CD56+ uNK when analysed by immunohistochemistry [SMD 0.50; CI 0.05–0.94] but not by flow cytometry, and when CD56+ uNK were reported as percentage over total endometrial cells [SMD 0.58; 95% CI 0.10–1.07]. Further subgroup analysis showed significant difference in CD16 + [SMD 0.54; 95% CI 0.18–0.89] but not in CD56+CD16-, CD56+CD16+ or CD57.
For pregnancy outcome, there was no significant difference in livebirth rate in RM/RIF patients with high uNK compared to normal uNK [RR 1.06, 95% CI 0.86–1.30]. Mean uNK level in RM patients with subsequent miscarriage was not significantly higher than subsequent livebirth.
Finally, the pooled correlation between CD56 pNK and CD56 uNK (r = 0.42; 95% CI –0.04–0.73] was not significant in RM/RIF patients.
Limitations, reasons for caution
The meta-analysis is limited by quality of some of the studies. Some data were presented in median that was transformed to mean which may result in data skew. Other confounding factors e.g. maternal age, fetal karyotype, number of previous miscarriages and variable definition of controls may contribute to bias.
Wider implications of the findings: Clinical interpretation of uNK level needs to be treated with caution because there is significant heterogeneity in method of analysis. There may be a role for uNK measurement in RIF patients however further studies to understand pathophysiology underlying elevated uNK is warranted before recommending it as a diagnostic tool.
Trial registration number
N/A
