Background: Preimplantation genetic diagnosis or screening (PGD, PGS) involves embryo biopsy on Day 3. Opting for one- or two-cell biopsy is a balance between the lowest risk for misdiagnosis on the one hand and the highest chance for a pregnancy on the other hand.
Methods: A prospective controlled trial was designed and 592 ICSI cycles were randomly assigned to the one-cell (group I) or the two-cell group (group II). Primary outcomes were diagnostic efficiency and embryonic development to delivery with live birth (analysed by cycle). The false-positive rate for the PCR cycles is presented as a secondary outcome (analysed by embryo).
Results: A strong significant correlation was observed between embryonic developmental stage on Day 3 and post-biopsy in vitro development on Day 5 (P < 0.0001). The influence of the intervention on Day 3 was less significant (P = 0.007): the biopsy of one cell is less invasive than the biopsy of two cells. PCR diagnostic efficiency was 88.6% in group I and 96.4% in group II (P = 0.008). For the fluorescence in situ hybridization (FISH) PGD cycles no significant difference in efficiency was obtained (98.2 and 97.5% in group I and II, respectively). Similar delivery rates with live birth per started cycle were obtained [58/287 or 20.2% in group I versus 52/303 or 17.2% in group II, P = 0.358; the absolute risk reduction = 3.05%; 95% confidence interval (CI): -3.24, 9.34]. Post-PGD PCR reanalysis showed six false positives in 97 embryos (6.2%) in group II and none in group I (91 embryos reanalysed). No false negatives were found.
Conclusions: While removal of two blastomeres decreases the likelihood of blastocyst formation, compared with removal of one blastomere, Day 3 in vitro developmental stage is a stronger predictor for Day 5 developmental potential than the removal of one or two cells. The biopsy of only one cell significantly lowers the efficiency of a PCR-based diagnosis, whereas the efficiency of the FISH PGD procedure remains similar whether one or two cells are removed. Delivery rates with live birth per started cycle were not significantly different.