The use of near-infrared (NIR) spectroscopy as an adjunct to morphology for metabolomic profiling of culture media for selecting embryos does not improve the live birth rate significantly in comparison to embryo selection by morphology alone, states a multi-center study.
Proof of principle studies have suggested the use of the NIR and the derivation of a viability score as a tool for assessing embryo viability (Seli et al., 2007, 2010, 2011; Nagy et al., 2008, 2009; Scott et al., 2008; Vergouw et al., 2008, 2011; Ahlstrom et al., 2011). In 2012, Vergouw et al. conducted a double-blind RCT that showed Day 3 embryo selection by metabolomic profiling of culture medium with NIR spectroscopy as an adjunct to morphology did not improve ongoing pregnancy and live birth rates compared with morphology alone.
However, Sfontouris et al. (2013) found significantly higher ongoing implantation rates for Day 5 transfers, but similar ongoing pregnancy and live birth rates, in their RCT in which embryo selection by metabolomic profiling with NIR spectroscopy plus morphology was compared with morphology alone; however, that study was terminated early and therefore lacked adequate power.
The purpose of the current study by C.G. Vergouw and colleagues was to utilize individual patient data meta-analysis (IPD MA) to assess the efficacy of metabolomic profiling of culture medium with NIR spectroscopy technology as an adjunct to embryo morphology in IVF. “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,” wrote Vergouw et al. (“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,” HR, 2014;29(3):455-461).
Included studies were those that reported results of an RCT comparing embryo selection by morphology to embryo selection by morphology and use of NIR spectroscopy of spent embryo culture medium. A literature search was performed in PubMed, the Cochrane Library, and the WHO International Clinical Trials Registry using relevant terms. Selected studies utilized a randomized design with a population of women who underwent IVF or ICSI treatment.
Authors of eligible studies were asked to send in their data set, from which data relevant to the current study were obtained. Study quality of selected studies was independently assessed by two authors using the Cochrane checklist for evaluation of RCTs (Dutch Cochrane Collaboration, 2012).
Statistical analysis utilized SAS Proc Glimmix, SPSs 20.0, and STATA 12 software and included intention-to-treat (ITT) analysis, testing by the I ² statistic in a random effects model, and logistic regression analysis. Four RCTS were identified as eligible for inclusion in the present IPD MA; four databases were obtained (Hardarson et al., 2012; Vergouw et al., 2012; Sfontouris et al., 2013; Economou, no date given), three of which had been published previously (Hardarson et al., 2012; Vergouw et al., 2012; Sfontouris et al., 2013).
Study size of original clinical trials varied between 55 and 417 patients (ITT population); the four databases combined included 924 IVF/ICSI cycles of 924 patients. The control group was made up of 484 patients for whom embryos for transfer were selected by morphology alone, while the treatment group was made up of 440 patients for whom embryos were selected by morphology
plus NIR spectroscopy. Data regarding live birth was available for all 924 patients.
For the control group, the live birth rate was 34.7% (168 of 484), while that of the NIR group was 33.2% (146 of 440). The results of the IPD MA logistic regression analysis gave a pooled odds ratio (OR) of 0.98 [95% confidence interval (CI) 0.74 - 1.29], which indicated the live birth rates did not differ significantly between the two groups. Data of the four studies showed no significant
heterogeneity; consequently, a fixed effect model was applied in the IPD MA.
The following variables were used for model building and evaluation of potential confounders (all were provided by all four studies): maternal age, medical cause of infertility, number of previous IVF attempts, treatment type, number of oocytes at OPU, fertilization method, number of fertilized oocytes, number of good quality embryos, number of embryos transferred, and day of
embryo transfer. The test for interaction effects of the treatment group with all potential confounding variables was not significant. The strongest confounding factors were: maternal age, medical cause of infertility, number of previous IVF attempts, number of oocytes at OPU, and number of fertilized oocytes. Multivariable logistic regression analysis including these confounding variables showed the study group was not related to live birth (OR 0.97, 95% CI 0.73 - 1.29).
“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 birth rates,” concluded Vergouw et al.
Address correspondence to C.G. Vergouw, Department of Obstetrics and Gynecology, VU University Medical Center, Amsterdam, The Netherlands; e-mail: firstname.lastname@example.org.