The observer and cycle-to-cycle variability in the measurement of uterine natural killer cells by immunohistochemistry

The aim of the study was to investigate the various potential sources of variability of counting endometrial uNK cells by immunohistochemistry. Precisely timed endometrial biopsy samples were obtained from women suffering from recurrent miscarriage or recurrent implantation failure (RIF) after IVF on days LH+7–LH+9 of the cycle. uNK cells in wax embedded sections were immunostained for CD56+ and expressed as a percentage of total stromal cells. The number of uNK cells in the same sample were counted by two methods, by the same observer on two occasions (intra-observer variability), by three different observers (inter-observer variability) and finally on samples obtained from the same individual at the same time in two different cycles (cycle-to-cycle variability). There was excellent agreement (κ = 0.935) in the numbers of uNK cells obtained using both the traditional cell counting (TCC) and image analysis (IA) methods. The intra-observer variability (TCC, κ = 0.944; IA, κ = 0.935) was relatively small, although the inter-observer variability (TCC, κ = 0.832 and 0.497; IA, κ = 0.438) was modest. There was significant variation in number of cells in samples obtained from the same women in two different cycles (κ = 0.348). The measurement of uNK cells in the endometrium has a reasonable degree of precision, but the significant cycle-to-cycle variation in results is a source of concern and requires further investigation.