STUDY QUESTION
What effects did treatment using hyaluronic acid (HA) binding/selection prior to ICSI have on clinical outcomes in the Hyaluronic Acid Binding sperm Selection (HABSelect) clinical trial?
SUMMARY ANSWER
Older women randomized to the trial’s experimental arm (selection of sperm bound to immobilized (solid-state) HA) had the same live birth rates as younger women, most likely a result of better avoidance of sperm with damaged DNA.
WHAT IS KNOWN ALREADY
Recent randomized controlled trials (RCTs) investigating the efficacy of HA-based sperm selection prior to ICSI, including HABSelect, have consistently reported reductions in the numbers of miscarriages among couples randomized to the intervention, suggesting a pathological sperm-mediated factor mitigated by prior HA-binding/selection. The mechanism of that protection is unknown.
STUDY DESIGN, SIZE, DURATION
The original HABSelect Phase 3 RCT ran from 2014 to 2017 and included 2752 couples from whom sperm samples used in control (ICSI) and intervention (Physiological IntraCytoplasmic Sperm Injection; PICSI) arms of the trial were stored frozen for later assessment of DNA quality (DNAq). The trial overlapped with its mechanistic arm, running from 2016 to 2018.
PARTICIPANTS/MATERIALS, SETTING, METHODS
As miscarriage reduction was a significant secondary outcome of the trial, samples (n = 1247) selected for the mechanistic analysis were deliberately enriched for miscarriage outcomes (n = 92 or 7.4%) from a total of 154 miscarriages (5.6%) among all (n = 2752) couples randomized by stratified random sampling. Values from fresh semen samples for sperm concentration (mml), percentage forward progressive motility and percentage HA-binding score (HBS) were obtained before being processed by differential density gradient centrifugation or (rarely) by swim-up on the day of treatment. Surplus sperm pellets were recovered, aliquoted and cryopreserved for later analysis of DNAq using slide-based Comet, TUNEL, acridine orange (AO) and the sperm chromatin dispersion (SCD) assays. Following their classification into normal and abnormal sample subcategories based on reference values for sperm concentration and motility, relationships with HBS and DNAq were examined by Spearman correlation, Student’s t-tests, Mann Whitney U tests, and logistic regression (univariable and multivariable). Parsimonious selection enabled the development of models for exploring and explaining data trends. Potential differences in future cumulative pregnancy rates relating to embryo quality were also explored.
MAIN RESULTS AND THE ROLE OF CHANCE
Results from the 1247 sperm samples assayed for HBS and/or DNAq, generated data that were considered in relation to standard physiological measures of (sperm) vitality and to treatment outcomes. All measures of HBS and DNAq discriminated normal from abnormal sperm samples (P < 0.001). SCD correlated negatively with the Comet (r = −0.165; P < 0.001) and TUNEL assays (r = −0.200; P < 0.001). HBS correlated negatively with AO (r = −0.211; P < 0.001), Comet (r = −0.127; P < 0.001) and TUNEL (r = −0.214; P < 0.001) and positively with SCD (r = 0.255; P < 0.001). A model for predicting live birth (and miscarriage) rates included treatment allocation (odds ratio: OR 2.167, 95% CI 1.084–4.464, P = 0.031), female age (OR 0.301, 95% CI 0.133–0.761, P = 0.013, per decade) and the AO assay (OR 0.79, 95% CI 0.60–1. 02.761, P = 0.073, per 10 points rise). A model predicting the expected rate of biochemical pregnancy included male age (OR 0.464, 95% CI 0.314–0.674, P < 0.001, per decade) and the SCD assay (OR 1.04, 95% CI 1.007–1.075, P = 0.018, per 10 point rise). A model for conversion from biochemical to clinical pregnancy did not retain any significant patient or assay variables. A model for post-injection fertilization rates included treatment allocation (OR 0.83, 95% CI 0.75–0.91, P < 0.001) and the Comet assay (OR 0.950, 95% CI 0.91–1.00, P = 0.041).
LIMITATIONS, REASONS FOR CAUTION
HABSelect was a prospective RCT and the mechanistic study group was drawn from its recruitment cohort for retrospective analysis, without the full benefit of randomization. The clinical and mechanistic aspects of the study were mutually exclusive in that measures of DNAq were obtained from residual samples and not from HA-selected versus unselected sperm. Models for fitting mechanistic with baseline and other clinical data were developed to compensate for variable DNAq data quality. HABSelect used a solid-state version of PICSI and we did not assess the efficacy of any liquid-state alternatives. PICSI reduced fertilization rates and did not improve the outlook for cumulative pregnancy rates.
WIDER IMPLICATIONS OF THE FINDINGS
Notwithstanding the interventional effect on fertilization rates and possibly blastocyst formation (neither of which influenced pregnancy rates), poor sperm DNAq, reflected by lower HBS, probably contributed to the depression of all gestational outcomes including live births, in the HABSelect trial. The interventional avoidance of defective sperm is the best explanation for the equalization in live birth rates among older couples randomized to the trial’s PICSI arm. As patients going forward for assisted conception cycles globally in future are likely to be dominated by an older demographic, HA-based selection of sperm for ICSI could be considered as part of their treatment plan.
STUDY FUNDING/COMPETING INTEREST(S)
The study was supported by the National Institute for Health Research (NIHR) EME (Efficacy and Mechanism Evaluation)-11-14-34. National Research Ethics Service approval 11/06/2013: 13/YH/0162. S.L. is CEO of ExamenLab Ltd (company number NI605309).
TRIAL REGISTRATION NUMBER
ISRCTN99214271.
