P-285 An analysis of automated morphometric measurements finds that a combination of a large blastocyst size and a short tB-tSB time interval doubles the implantation rate

Abstract

Abstract Study question Are automated blastocyst morphometric measurements combined with morphokinetic pattern associated with implantation rate? Summary answer Automated blastocyst morphometric measurements with morphokinetic pattern demonstrated that a larger blastocyst size and a shorter time interval tB-tSB are associated with higher implantation potential. What is known already Optimization of embryo selection is important for increasing implantation potential. Transfer of a high-quality blastocyst based on conventional morphological parameters has been shown to improve IVF clinical outcome. Novel parameters of blastocyst quality, including morphokinetics from time-lapse monitoring and manual analysis of morphometric parameters, have demonstrated promising results regarding implantation potential. However, manual measurement of morphometric parameters is a time-consuming task and is subjected to intra- and inter-observer variations. The introduction of automated morphometric measurements would remove subjective blastocyst analysis and further improve implantation rates. Study design, size, duration A nested retrospective case control analysis of 608 day-5 transferred blastocysts was conducted and included women who underwent IVF treatment in three public IVF units between 2014 and 2017. Participants/materials, setting, methods Automated morphometric blastocyst analysis was measured at the mean time of tEB-tPNf (85.82 h) by training a pixel-wise segmentation model (MaskRCNN) on time-lapse videos. Morphometric blastocyst parameters included the following: blastocyst size (μm), inner cell mass (ICM) size (μm), ICM to blastocyst size ratio, and ICM shape. Annotation variables included all the time intervals (hours) from time of pronuclei fading (tPNf) to the expanded blastocyst (tEB). Main results and the role of chance The mean blastocyst size for implanted embryos was significantly larger compared to non-implanted embryos (152.10 ±19.22μm vs 144.25±18.52μm, respectively, p < 0.001), while the mean interval times of tSB-tPNf, tB-tPNf, tEB-tPNf, tB-tSB, and tEB-tSB were significantly shorter (tSB-tPNf: 72.10±5.60h vs 73.30±5.80h, respectively, p = 0.016; tB-tPNf: 80.08±5.96h vs 82.54±5.92h, respectively, p < 0.001; tEB-tPNf: 84.95±5.43h vs 86.58±4.93h, respectively, p = 0.001; tB-tSB: 8.21±2.90h vs 9.49±3.62h, respectively, p < 0.001; and tEB-tSB: 13.50±3.00h vs 14.58±3.75h, respectively, p = 0.001). In a multivariable logistic model that measured the independent effect of ICM size, blastocyst size, tB-tSB, and woman age on implantation potential, blastocyst size was found to be positively associated with implantation potential (OR = 1.017, 95% CI 1.006-1.027, p = 0.002) while tB-tSB and woman age were found to be negatively associated (OR = 0.918, 95% CI 0.861-0.980, p = 0.010 and OR = 0.903, 95% CI 0.874-0.932, p < 0.001, respectively). Blastocyst size larger than the mean and a tB-tSB interval shorter than the mean had a 2.028 greater chance of implantation compared to blastocysts that did not meet these criteria (OR = 2.028, 95% CI 1.420-2.894, p < 0.001). In a multivariable logistic model adjusted for woman age, the chance for implantation among blastocysts meeting the aforementioned criteria was 1.7 greater (adjusted OR 1.714, 95% CI 1.182-2.485, p = 0.005). The AUC value for implantation prediction was 0.69 (p < 0.01). Limitations, reasons for caution The study's limitations include its retrospective nature and the absence of some patient characteristics. Wider implications of the findings A blastocyst selection based on the combination of automated blastocyst size measurements and manual tB-tSB time interval may double the implantation rate. The inclusion of automated morphometric measurements to the blastocyst selection algorithm may reduce intra- and inter-observer variations and should be incorporated into models for implantation prediction. Trial registration number 0006-20-HMO