A feasibility trial of screening women with idiopathic recurrent miscarriage for high uterine natural killer cell density and randomizing to prednisolone or placebo when pregnant

STUDY QUESTION: Is it feasible to screen women with idiopathic recurrent miscarriage (RM) for high uterine natural killer (uNK) cell density and then randomize them to prednisolone or placebo when pregnant? SUMMARY ANSWER: It was feasible to recruit women with idiopathic RM into a 'screen and treat' trial despite their desire for active medication. WHAT IS KNOWN ALREADY: Clinical trials of immunotherapy in women with idiopathic RM have failed to substantiate efficacy in preventing miscarriage. Preimplantation uNK cell density is higher in women with RM and can be reduced with prednisolone. STUDY DESIGN, SIZE, DURATION: In a pilot RCT, 160 eligible women were screened with an endometrial biopsy and those with high uNK cell density were invited to return when pregnant for randomization to prednisolone (20 mg for 6 weeks, 10 mg for 1 week, 5 mg for 1 week) or identical placebo tablets. Randomization was by random number generation and patients, clinicians and outcome assessors were blinded to allocation. The study size and duration was determined by funding, which was for a feasibility trial, for 2 years, sufficient to screen 150 women and randomize 40 women. The outcome measures were recruitment rate, women's perspectives, compliance, live birth and miscarriage rates and pregnancy complications. PARTICIPANTS/MATERIALS, SETTING, METHODS: The trial was advertised nationally in the UK. Women who attended research clinics run by one consultant (SQ) with three or more consecutive idiopathic miscarriages were included. Women's perspectives of the trial were sought through a questionnaire. The endometrium was sampled 5-9 days after the LH surge, stained using immunohistochemistry for CD56 and the sub-epithelial region analysed with image analysis. Women with a high uNK cell density (>5%) were invited to contact the clinic at 4-6 weeks gestation for randomization. Compliance with medication was assessed using a daily log, and side effects recorded by the women in a diary and on a structured proforma completed in the clinic at the end of the first trimester. All women had ultrasound scans every 2 weeks until 14 weeks' gestation and growth scans at 28 and 34 weeks' gestation in addition to routine antenatal care and a follow-up in person or by telephone 6 weeks after delivery. MAIN RESULTS AND THE ROLE OF CHANCE: Despite the fact that 85% of women said they would prefer the active treatment, the trial recruitment occurred at the planned rate. Eligible women (n = 160) attended the research clinics and had the uNK test, 72 were screen positive and 40 returned when pregnant for randomization. Compliance with medication was reported to be 100%. The active treatment was associated with side effects of insomnia and flushing. The live birth rate was 12/20 (60%) with prednisolone and 8/20 (40%) with placebo (Risk Ratio 1.5, 95% confidence interval (CI) 0.79-2.86, absolute difference 20% CI-10%, +50%), and hence, this was not significant. There were no pregnancy complications or serious adverse fetal outcomes. LIMITATIONS, REASONS FOR CAUTION: This was a feasibility trial so of insufficient size to assess efficacy or safety. There was inconsistency in the start date of the trial medication and this may have affected the outcome in the active treatment group. WIDER IMPLICATIONS OF THE FINDINGS: It was feasible to recruit women with idiopathic RM into a 'screen and treat' trial despite their desire for active medication. Our data also suggest that in future trials the primary outcome measure is live birth rate after 24 weeks gestation.