The dataset for this retrospective study included 169 consecutive fresh IVF cycles performed between January 2012 and June 2014 at the Seoul National University Bundang Hospital. This study was approved by the Institutional Review Board of the Seoul National University Bundang Hospital (IRB No. B-1608-357-108). Written informed consents were obtained. The indications for IVF were unexplained infertility (n=78), tubal factor infertility (n=26), age factor infertility (n=20), endometriosis (n=18), uterine factor infertility (n=10), polycystic ovary syndrome (n=9), and male factor infertility (n=8). The body mass index, basal serum level of follicle-stimulating hormone (FSH), and a random serum level of anti-Müllerian hormone in the female were recorded if they were measured within 2 months before starting the cycle.

Controlled ovarian stimulation was performed using recombinant FSH (Gonal-F; Serono, Geneva, Switzerland) with or without highly purified urinary gonadotropin (Menopur; Ferring, Malmo, Sweden) using a luteal long protocol with a gonadotropin-releasing hormone (GnRH) agonist (0.1 mg/day of Decapeptyl; Ferring) (n=10) or a GnRH antagonist protocol (0.25 mg/day of Cetrotide, Serono) (n=157). When two or more leading follicles reached a mean diameter of ≥18 mm, 250 µg of recombinant human chorionic gonadotropin (hCG; Ovidrel, Serono) was injected. Oocytes were retrieved 36 hours after the hCG injection. Poor ovarian response (POR) was defined as the collection of three or fewer mature oocytes. Metaphase I-derived in vitro matured oocytes were counted as mature oocytes, but germinal vesicle-derived oocytes were not.

Semen was collected on the day of oocyte retrieval and standard sperm quality was assessed in the raw semen (concentration, motility, and normal form using strict criteria). Sperm quality was defined as normal (n=104) when the semen parameters were within the World Health Organization (WHO) reference values, regardless of the patient's diagnosis. The subnormal group (n=43) included semen with parameters outside the WHO criteria, but not to the point of requiring ICSI, and the abnormal group (n=22) was made up of cases requiring ICSI.

The SDF level was measured by the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) method in the raw semen, following our previously reported institutional protocol [8]. Semen samples were smeared on a silane-coated slide (DAKO Japan Co., Tokyo, Japan) and air-dried. Samples were fixed with 4% paraformaldehyde for 1 hour at 15℃−25℃, washed with phosphate-buffered saline, and then were permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate (Sigma-Aldrich, St. Louis, MO, USA). A commercial apoptosis detection kit (in situ cell death detection kit; Roche Diagnostics, Mannheim, Germany) was used to assess cell death in the samples according to the manufacturer's instructions. Counterstaining was performed using a mounting medium with 4',6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA, USA). The nuclei of sperm with fragmented DNA was stained green, whereas the nuclei of other cells were stained blue. Sperm heads with >50% of the area stained green were considered positive for DNA fragmentation, following our previously reported institutional protocol [8]. At least 500 sperm cells were counted per experimental set, and the SDF level was determined as the percentage of sperm with fragmented DNA.

The remaining semen was processed by a discontinuous gradient as described in the kit instructions (Sydney IVF density gradient media; COOK, Brisbane, Australia). After initial centrifugation of the semen (300×g for 5 minutes) to remove the seminal plasma, the obtained pellet was suspended in fresh Ham's F10 medium (1.5 mL) supplemented with 10% synthetic serum substitute (SSS; Irvine Scientific, Santa Ana, CA, USA). Pre-washed semen (1.0 mL) was layered on the top of a discontinuous gradient in a 15-mL conical tube (40%/80%). The conical tube was centrifuged at 300×g for 5 minutes, and the sperm cells collected from the bottom layer (80% layer) were washed twice by resuspension in 4 mL of Ham's F10 medium and centrifugation (300×g for 5 minutes). After two rounds of centrifugation, the supernatant was removed and the pellet was resuspended in 3 mL of Ham's F10 medium supplemented with 10% SSS. The resuspended pellet was later used in insemination.

The oocytes were inseminated by the conventional method (n=69), by ICSI (n=95), or by split insemination (n=3), depending on the quality of the sperm and oocyte. ICSI was used in 22 cycles due to a male factor and in 73 cycles due to a female factor. Fertilization was confirmed by observing a two-pronuclear zygote on the day after oocyte retrieval. A low fertilization rate was defined as ≤75% in the current study, which corresponded to the 25th percentile of the overall fertilization rate.

The embryos were transferred 3 or 5 days after oocyte retrieval. Embryo quality was evaluated by morphological criteria based on the degree of fragmentation and the regularity of blastomeres on day 3 after fertilization. The embryos were graded as follows: grade A, 0% anucleate fragments, regularity of blastomeres, and no apparent morphological abnormality; grade B, <20% anucleate fragments, regularity of blastomeres, and no apparent morphological abnormality; grade C, 20%–50% anucleate fragments, irregularity of blastomeres, and no apparent morphological abnormality; and grade D, >50% anucleate fragments, irregularity of blastomeres, and apparent morphological abnormalities. Blastocysts were evaluated on day 5 by the developmental stage and quality of the inner cell mass and trophectoderm. A good-quality blastocyst was defined as grade AA, AB, AC, BA, BB, or CA. Luteal phase support was performed using either a daily dose of 50 mg of progesterone in oil (Progest; Genefer, Seoul, Korea) or an 8% progesterone gel (Crinone, Serono), starting on the day of oocyte retrieval. Pregnancy was first assessed 14 days after oocyte retrieval by measuring the serum hCG level. In cases with positive hCG results, transvaginal ultrasonography was performed to confirm an intrauterine pregnancy and to identify the number of gestational sacs and the fetal heartbeat. Clinical pregnancy was defined as the presence of one or more gestational sacs. Miscarriage was defined as pregnancy loss before 12 weeks of gestation.

All statistical analyses were performed using SPSS PASW ver. 18.0 (SPSS Inc., Chicago, IL, USA). When analyzing the association between standard sperm quality and SDF levels, the data from 169 sperm samples were used. The data from 164 cycles were used to analyze the fertilization rate, after excluding three cases with split insemination and two cases with no mature oocyte. Associations between SDF levels and pregnancy were analyzed in the 157 cycles in which embryo transfer was performed. Associations between SDF levels and miscarriage were analyzed in the 55 cases that achieved clinical pregnancy. Correlations were tested using the non-parametric Spearman rank test. The chi-square test was used to compare proportions between two groups. If the cell numbers were <5, the Fisher exact test was applied to compare frequencies between groups. The medians of numeric data were compared using the Mann-Whitney U-test. Multiple regression analyses were performed for several numeric variables. A receiver operating characteristic (ROC) curve analysis was used to assess specific cutoff values for several numeric parameters. Results were considered significant with a two-tailed p-value <0.05.

The SDF levels ranged from 0.4% to 56.8% in the 169 sperm samples (mean, 15.1%; median, 11.8%; standard deviation, 12.2%). They did not show a normal distribution. As shown in Table 1, in the overall population, a positive correlation between SDF and the male partner's age was found, but an inverse correlation was found between SDF and sperm motility. A multivariate analysis revealed that both the male partner's age and motility were significantly associated with SDF (SDF=−0.444+[0.618×male partner's age]−[0.162×motility]). The positive correlation between SDF and the male partner's age was prominent in those with normal sperm, while an inverse correlation between SDF and motility was prominent in the group with subnormal sperm. In the group with abnormal sperm, no significant correlations were found between SDF and any of the standard sperm parameters.

In the group with normal sperm, the median SDF level (95% confidence interval [CI]) was 11.0% (8.5%–13%). The median SDF level was 13.0% (7.2%–18.7%) in the group with subnormal sperm and 14.9% (10.4%–29.8%) in the group with abnormal sperm. There was a significant difference between the median SDF level between the normal and abnormal sperm groups (p<0.05).

SDF levels were not correlated with the fertilization rate in the overall study population or in subgroups defined according to the insemination method or whether ICSI was performed due to a male or female factor (Figure 1). In the three subgroups defined according to sperm quality, no significant correlations were found between SDF and the fertilization rate in the overall population, the standard IVF group, or the ICSI group with a female or male factor (Table 2).

A ROC analysis was performed to evaluate whether SDF levels could predict a low fertilization rate (i.e., ≤75%). The cutoff value for the SDF level was ≤5.1% in the overall population, ≤4.9% in the standard IVF group, >21.3% in the ICSI group, >21.3% in the female factor ICSI group, and >12.9% in the male factor ICSI group. However, none of the values were statistically significant.

The clinical pregnancy rate was 31% (40/129) in the day 3 transfer group and 53.5% (15/28) in the day 5 transfer group. SDF levels and other standard sperm parameters did not differ according to whether pregnancy was achieved (data not shown). Two factors were significantly associated with pregnancy: female age (35 vs. 37 years, p=0.016) in the day 3 transfer group and the male partner's age (34 vs. 38 years) in the day 5 transfer group (p=0.002).

Clinical pregnancy was achieved in 55 women, but 10 pregnancies ended in spontaneous miscarriage. In the miscarriage group, the SDF levels and serum FSH levels were significantly higher than in the live birth group, and the number of mature oocytes was significantly lower (Table 3).

A ROC analysis revealed that the cutoff values for predicting miscarriage were an SDF level >13%, a basal serum FSH level >6.7 mIU/mL, and ≤3 mature oocytes; all three cutoff values were statistically significant (Table 4). Neither the male partner's age nor female age was a predictor of miscarriage. However, since both the male partner's age and female age were possible confounders, multiple regression analysis was performed, including five parameters (SDF level, basal serum FSH level, number of mature oocytes, the male partner's age, and female age). SDF was found to be the only significant predictive factor of miscarriage (odds ratio [OR], 1.051; 95% CI, 1.001–1.104).

Cycles with three or fewer mature oocytes usually indicate POR. Since high SDF levels (>13%) and POR were significant predictive factors of miscarriage, we divided the whole population into a low-SDF (≤13%) and high-SDF (>13%) group and further divided them into POR and normal ovarian response (NOR) groups (Table 5). In the low-SDF group, the miscarriage rate was similar between the POR and NOR groups (14.2% vs. 4.3%). However, in the high-SDF group, the miscarriage rate was significantly higher in the POR group than in the NOR group (60.0% vs. 13.3%, p=0.045). The pregnancy rate did not differ across the four sub-groups.