Can We Identify Couples That Could Benefit From IVF in Male Subfertility?

Abstract

To assess the diagnostic accuracy of baseline characteristics, semen parameters, post-wash TMC and antisperm antibodies (ASA) during the fertility work-up for in vitro fertilization (IVF) outcome in male subfertility. Retrospective cohort study. All couples with male subfertility according to the WHO criteria undergoing IVF were included from 1997-2003 at a University Hospital. All baseline male and female characteristics and all semen parameters, including after processing, during the fertility work-up were included in the analysis. Ongoing pregnancy was the main outcome measure. Logistic regression analysis was performed to construct a prediction model. The additional value of ASA and the post-wash TMC was assessed by comparing the areas under the receiver operating characteristic curve (AUC). We assessed the diagnostic accuracy of the different prediction models by comparing the predicted probabilities on an ongoing pregnancy and calculated the cycles needed to perform for one ongoing pregnancy after IVF. We included 208 couples, who underwent 355 IVF cycles. The overall ongoing pregnancy rate was 22% per cycle, 35% per couple with 11% total fertilization failure (TFF) per cycle. The number of cycles needed to establish one pregnancy was 4.5 cycles. A prediction model containing female age, the percentage progressively motile spermatozoa and the percentage sperm with normal morphology (model I) had an AUC of 0.62. Addition to model I of the presence of ASA (model II) and the post-wash TMC during the fertility work-up (model III) improved the AUC to 0.64 and 0.66, respectively, which was in both cases not statistically significant. With the use of the predicted probabilities on an ongoing pregnancy from all three models, we tested the diagnostic accuracy of the models. Model I and the selective use of model II in those cases where model I had a predicted pregnancy rate of >20% per IVF cycle, allowed for the identification of those couples that had high pregnancy rates (30% per cycle and 48% per couple) for the low TFF rates (6% per cycle). The number of cycles needed to perform to establish one pregnancy was 3.3, but this resulted in the exclusion of almost 50% of couples from the initial population from IVF. When model III was applied if model II had a predicted probability on an ongoing pregnancy <20%, the ongoing pregnancy rate was still good (29% per cycle and 45% per couple) for TFF rates of 9% per cycle. The number of cycles needed to establish one pregnancy was 3.5 cycles. In the latter approach, only 22% of the couples were excluded from IVF. In male subfertility, some couples do benefit more from IVF than others. The use of ASA prior to IVF, and the use of the post-wash TMC in case the initial pre-test probability is <20%, would increase the efficiency of IVF without affecting the total number of pregnancies in the program.