No evidence that embryo selection by near-infrared spectroscopy in addition to morphology is able to improve live birth rates: results from an individual patient data meta-analysis

Study question: What is the value of embryo selection by metabolomic profiling of culture medium with near-infrared (NIR) spectroscopy as an adjunct to morphology, compared with embryo selection by morphology alone, based on an individual patient data meta-analysis (IPD MA)?

Summary answer: The IPD MA indicates that the live birth rate after embryo selection by NIR spectroscopy and morphology is not significantly different compared with the live birth rate after embryo selection by morphology alone.

What is known already: Retrospective proof of principle studies has consistently shown that high NIR viability scores are correlated with a high implantation potential of embryos. However, randomized controlled trials (RCTs) have generally shown no benefit of the NIR technology over embryo morphology, although there have been some conflicting results between pregnancy outcomes on different days of embryo transfer.

Study design, size, duration: This IPD MA included all existing RCTs (n = 4) in which embryo selection by morphology was compared with embryo selection by morphology and the use of NIR spectroscopy of spent embryo culture medium by the Viametrics-E(™).

Participants/materials, setting, methods: Searches of PubMed, the Cochrane Library and the WHO International Clinical Trials Registry were conducted and the sole manufacturer of the Viametrics-E(™) was consulted to identify clinics where an RCT comparing embryo selection by morphology to embryo selection by morphology and the use of the Viametrics-E(™) (NIR viability score) was performed. A total of 20 citations were potentially eligible for inclusion, two of which met the inclusion criteria. The manufacturer of the Viametrics-E(™) provided two additional clinical sites of use. In total, four RCTs were identified as eligible for inclusion. The IPD MA was based on a fixed effect model due to the lack of heterogeneity between included studies. Differences between study groups were tested and reported using logistic regression models adjusted for significant confounders. The pooled analysis of the primary outcome led to a total sample size of 924 patients: 484 patients in the control group (embryo selection by morphology alone) and 440 patients in the treatment group (embryo selection by morphology plus NIR spectroscopy).

Main results and the role of chance: The live birth rates in the control group and the NIR group were 34.7% (168 of 484) and 33.2% (146 of 440), respectively. The pooled odds ratio (OR) was 0.98 [95% confidence interval (CI) 0.74-1.29], indicating no difference in live birth rates between the two study groups. The data of the four studies showed no significant heterogeneity (I(2) = 26.2% P = 0.26). The multivariate regression analysis including all confounders show that maternal age (OR 0.90, 95% CI 0.87-0.94) and the number of previous IVF cycles (OR 0.83, 95% CI 0.71-0.96) were significantly related to live birth. The study group (i.e. embryo selection by morphology or embryo selection by morphology plus NIR) was not related to live birth (OR 0.97, 95% CI 0.73-1.29).

Limitations and reasons for caution: The availability of at least two similar best quality embryos as an inclusion criterion prior to transfer in the two largest RCTs might have caused a selection bias towards a better prognosis patient group.

Wider implications of the findings: There is at present no evidence that NIR spectroscopy of spent embryo culture media in its current form can be used in daily practice to improve live birth rates.