Endometrial cell counts in recurrent miscarriage: a comparison of counting methods.

AIMS: Studies of uterine natural killer (uNK) cells require reliable measurements of uNK cell density among diverse endometrial tissue. The aim of this study was to compare cell counting manually with two computer-aided methods based on a public domain software package, ImageJ. METHODS AND RESULTS: Immunohistochemistry (IHC) of CD56(+) uNK cells was performed on endometrium from recurrent miscarriage patients. Numbers of stromal cells per high-power field (HPF) were counted by two observers using: (i) manual tally counter and graticule; (ii) ImageJ 'point picker' tool; and (iii) ImageJ 'particle analysis' tool. Coefficients of variation (CV) and Bland-Altman plots were used to evaluate interobserver differences. Evaluation of %uNK using ImageJ particle analysis for stromal cell counts and point picker tool for uNK counts was undertaken. Point picker and particle analysis were significantly better than manual counting [interobserver CVs mean (standard deviation) 6.1% (3.3%); 4.7% (3.9%), 8.2% (6.5%), respectively]. Mean inter- and intra-observer CVs for %uNK were 10.3% (6.6%), 8.5% (4.9%) and 6.8% (4.3%), respectively. Bland-Altman analysis revealed no systematic differences in cell counts with the number of cells in the image for each method. CONCLUSIONS: Compared to manual cell counting, computer-aided image analysis yields more reproducible results for the assessment of uNK cells density using IHC.

Uterine natural killer cells and angiogenesis in recurrent reproductive failure.

BACKGROUND: Increased numbers of phenotypically unusual CD56(bright) CD16- uterine natural killer (uNK) cells have been associated with recurrent reproductive failure. uNK cells produce angiogenic growth factors and are potential regulators of decidual angiogenesis in early pregnancy. The final common mechanism for early pregnancy loss is thought to be early onset of the maternal circulation and excessive placental oxidative stress. We tested the hypothesis that increased uNK cells in preimplantation endometrium are associated with altered angiogenesis. METHODS: Women with recurrent reproductive failure (n = 122) were investigated with uterine artery Doppler and endometrial biopsy. Immunohistochemistry was used to identify uNK, endothelial and vascular smooth muscle cells and image analysis was used to assess location, density and differentiation. RESULTS: uNK cell density was positively correlated with the formation of blood (P = 0.005, r = 0.5) and lymphatic vessels (P = 0.0001, r = 0.6), spiral arteriole smooth muscle differentiation (P = 0.01, r = 0.5) and endometrial oedema (P = 0.004). The functional effect of this was a reduced uterine artery resistance to blood flow. CONCLUSIONS: These data suggest that uNK cells may regulate angiogenesis in non-pregnant endometrium. The mechanisms of reproductive failure associated with increased uNK cell density appear to be increased angiogenesis and peri-implantation blood flow, which may lead to early maternal circulation and hence pregnancy failure due to excessive oxidative stress.