Semen Levels of Spermatid-Specific Thioredoxin-3 Correlate with Pregnancy Rates in ART Couples

Spermatid specific thioredoxin-3 (SPTRX3 or TXNDC8) is a testis/male germ line specific member of thioredoxin family that accumulates in the superfluous cytoplasm of defective human spermatozoa. We hypothesized that semen levels of SPTRX3 are reflective of treatment outcome in assisted reproductive therapy (ART) couples treated by in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI). Relationship between SPTRX3 and treatment outcome was investigated in 239 couples undergoing ART at an infertility clinic. Sperm content of SPTRX3 was evaluated by flow cytometry and epifluorescence microscopy, and correlated with clinical semen analysis parameters, and data on embryo development and pregnancy establishment. High SPTRX3 levels (>15% SPTRX3-positive spermatozoa) were found in 51% of male infertility patients (n = 72), in 20% of men from couples with unexplained, idiopathic infertility (n = 61) and in 14% of men from couples previously diagnosed with female-only infertility (n = 85). Couples with high SPTRX3 produced fewer two-pronuclear zygotes and had a reduced pregnancy rate (19.2% pregnant with >15% SPTRX3-positive spermatozoa vs. 41.2% pregnant with <5% SPTRX3-positive sperm; one-sided p<0.05). The average pregnancy rate of all 239 couples was 25.1%. Live birth rate was 19.2% and lowest average SPTRX3 levels were found in couples that delivered twins. Men with >15% of SPTRX3-positive spermatozoa, a cutoff value established by ROC analysis, had their chance of fathering children by IVF or ICSI reduced by nearly two-thirds. The percentage of SPTRX3-positive spermatozoa had predictive value for pregnancy after ART. Gradient purification and sperm swim-up failed to remove all SPTRX3-positive spermatozoa from semen prepared for ART. In summary, the elevated semen content of SPTRX3 in men from ART couples coincided with reduced incidence of pregnancy by IVF or ICSI, identifying SPTRX3 as a candidate biomarker reflective of ART outcome.

: Smoking has a modest effect on sperm SPTRX3 levels. Heavy smokers consuming 15-30 cigarettes per day had higher average levels of sperm SPTRX3 than other groups shown.
Of note, two thirds (10/15) of these heavy smokers had indication/clinical diagnosis of male infertility. The numeric difference between smokers consuming 2-5 cigarettes per day and smokers consuming 15-30 cigarettes per day, was not statistically significant (p>0.1).

Figure S3
: Scatter diagram illustrating the relationship between subjective, light microscopic assessment of sperm SPTRX3 content (% spermatozoa with SPTRX3-positive heads; x-axis) and flow cytometry (%M3 SPTRX-value; y-axis). This simple light microscopic analysis was conducted as a potential precursor of a test suitable for clinical andrology laboratories. Samples from 150 randomly chosen donors, processed for flow cytometry, were re-evaluated by epifluorescence microscopy for the percentage of SPTRX3 positive sperm heads, sperm tails, and total % of SPTRX3 positive spermatozoa. Correlations were found between all three categories of light microscopic evaluation and % of SPTRX3-positive spermatozoa as measured by flow cytometry (%M3). The highest correlation coefficient (r=0.46) was between % SPTRX3positive sperm heads by light microscopy and %M3 SPTRX3 by flow cytometry (see Supplemental Table 4 B).

Materials & Methods for Western Blotting and Computerized Densitometry
Western blotting was performed to validate the SPTRX3 antibody and to explore the relationship between the total amount of SPTRX3 in spermatozoa of men with good and poor clinical semen parameters. As described previously [36], semen samples were washed, extracted in 50 mM Tris HCl, 500 mM NaCl, 0.5% Triton X-100, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and a 1:1,000 dilution of protease inhibitor cocktail (chymostatin, leupeptin, antipain, and      Pregnancy outcomes were available for 57/60 pregnant couples. One of those 57 pregnancies was an ectopic pregnancy in a case of female-only infertility; it is not included in the above table. Live birth was achieved in 81% (46/57) of those analyzed pregnant couples.
*One of those three male factor patients had a vasectomy   (B). SPTRX3 values, but not the clinical indication, were predictive of good zygotic development after IVF or ICSI. Couples with lowest SPTRX3 levels (A, top row, <5% M3) produced the highest percentage of normal, two-pronuclear (2 PN) zygotes out of all fertilized oocytes (A, column 8), and also when calculated based on all oocytes harvested (A, column 9). These couples also produced the most embryos suitable for transfer or cryopreservation (column 13). Note that couples with low or medium levels of SPTRX3 produced more 2PN zygotes per couple on average than couples with >15% SPTRX3 (A, column 10). Idiopathic couples in which men recorded more than 15% SPTRX3-positive spermatozoa had the lowest yields of twopronuclear zygotes. Numbers shown in red are highest & lowest values for each column. No significant correlations were found between parameters of early embryo development and flow cytometric SPTRX3 levels.        *Age information was not provided for one female patient **Three couples were first treated by IVF before ICSI ***Three couples were first treated by IVF before ICSI ****One couple was first treated by IVF before ICSI    SAS proc discrim procedure was applied to conduct discriminant analysis to understand the relationship between the pregnancy rate and all the sperm quality parameters. When all data is considered and all sperm quality parameters are considered, there is no significant discriminant function. The stepwise discriminant analysis identifies one significant predictor, PR, which leads to a discriminant function with canonical correlation of 0.13 (p-value=0.05). The cross-validated classification accuracy rate based on this discriminant function is 75%. The PR variable separates the two groups well with mean value of 21.6 for the non-pregnant group and 24.8 for the pregnant group.
When considering the IVF group only, no significant discriminant function is detected when all sperm quality parameters are considered. The stepwise discriminant analysis identifies Volume and M2% as significant predictors and the discriminant function based on these two variables is significant with p-value=0.008 and has canonical correlation of 0.3. The discriminant function is highly positively correlated with M2% (r=0.71) and negatively correlated with Volume (r=-0.68).
The cross-validated classification accuracy rate of this discriminant function is 73%. The means of M2% are 81.4 versus 84.2 for non-pregnancy group and pregnant group, respectively. The volume has mean 3.4 for non-pregnancy group and 2.8 for pregnancy group.
When considering the ICSI treatment group only, the stepdisc procedure identified four predictors: V, PR, M and C. The discriminant function based on these four predictors has p-value=0.014 and canonical correlation of 0.31. The cross-validated classification accurate rate is 78%. The structure matrix shows that this discriminant function is positively correlated with C (r=0.52) and negatively correlated with V (r=-0.5). The C variable separates the two groups well with 44.9 (non-pregnancy group) versus 26.7 (pregnancy group). The V variable has mean of 3.1 (no pregnancy) versus 3.6 (pregnancy).
If only the biomarker parameters are considered, the discriminant analysis identified %M2 and CVM3 as significant predictors for the IVF group ( Table S11). The discriminant function has pvalue 0.025 with canonical correlation of 0.26. It has positive correlation with %M2 (r=0.82) and CVM3 (r=0.77). The cross validated classification accuracy rate is 72% and the two variables %M2 and CVM3 separates the pregnancy group from the no-pregnancy group well.
If only the biomarker parameters are considered, the discriminant analysis identified Median M3 as a significant predictor for the ISCI group. The discriminant function has p-value 0.05 with canonical correlation of 0.17. The cross validated classification accuracy rate is 80% and the variable Median M3 separates the pregnancy group from the no-pregnancy group well with means of 232 versus 212 for the two groups.

ANALYSIS II: Discriminant analysis was conducted to study the relationship of the treatment assignment, SPTRX3 and the sperm quality parameters. SAS proc discrim procedure was used to conduct discriminant analysis to study the relationship between the treatment assignment and the sperm quality parameters.
The discriminant function is highly significant (   S13 gives the means of the ten variables by the treatment groups and shows that the ten variables separate the two groups well. For instance, %M2 has a mean of 82.25 for the IVF group and a mean of 76.41 for the ICSI group. Table S13:

ANALYSIS III: Logistic regression analysis using SAS proclogistic procedure was applied to study the relationship of pregnancy rate with the sperm quality parameters with consideration of female age and treatment assignments.
Considering only the patients assigned to the IVF treatment, both %M2 SPTRX3 and %M3 SPTRX3 were significant predictors for pregnancy rate, with or without adjustment for the female age factor. None of the convention parameters or other flow cytometric parameters was significant even with adjustment for the female age. The odds of getting pregnant in the IVF group were reduced by 10.2% for every unit of increase in %M3 (p-value=0.03). After adjusting for the factor of female partner's age (35 years or younger vs. older than 35), the odds of getting pregnant in the IVF group were reduced by 11.1% for every unit of increase in %M3 (p-value=0.03). The odds of getting pregnant when %M3 was below 5% were 6.7 times better than for the %M3 levels above 15% (p-value=0.05). After adjusting for female age, the adjusted odds ratio was 7.5 with p-value=0.04. We found that %M2 was also a significant predictor for pregnancy rate for IVF group with p-value=0.03 and the odds ratio is 1.1 for each unit increase.
After adjusting for the female age, the p-value=0.02 and the odds ratio was 1.1 (i.e. the odds of getting pregnant in IVF group increased by 13.6% for every unit increase of %M2).
For the ICSI group, the pregnancy rate was not affected by any of the new or conventional sperm quality parameters or the female age factors. When considering all patients together, %M3 and %M2 were not significant predictors for pregnancy rate (p-value=0.38 and 0.50 respectively). However, when we considered categorizing the %M3 levels into four groups, below 5%, between 5-10%, between 10-15%, and above 15% SPTRX3-positive spermatozoa, the estimated odds ratio between the below 5% group and the above 15% group was 2.94 with p-value=0.06. In other words, the odds of getting pregnant when the SPTX3 levels was below 5% were 2.94 times better than that for the SPTX3 levels above 15%. After adjusting for female age, the adjusted odds ratio for the below 5% SPTX3 group over the above15% SPTX3 group was 3.3 with p-value=0.04. Using 85% as a cutoff for %M2, the below 85% group and above 85% group had significantly different pregnancy rate when considering all patients together. The odds of getting pregnant for the above 85% of %M2 group were 113.2% better than that for the below 85% group (p-value=0.02) and the pregnancy rate was 41.5% versus 22.72%. After the adjustment for the female age, the odds ratio was 2.1 with p-value=0.02.

SPTRX3 analysis by flow cytometry.
SAS proc discrim procedure was used to study the relationship between the male infertility indicator variable and all the sperm quality parameters. The discriminant function detected is highly significant with p-value<0.0001 and has the canonical correlation of 0.83, which is the maximal multiple correlation of the linear combination of the sperm quality parameters with the male infertility indicator variable. TABLE S14 (below) presents the canonical structure matrix of the discriminant analysis, which shows that the discriminant function is positively correlated with progressive motility (PR; r=0.84), total motility (M; r=0.60) and % normal spermatozoa (NX; r=0.46), C (r=0.46), % Total (r=0.35), %M2 (r=0.33) and negatively correlated with %M3 (r=-0.29) Mean M3(r=-0.35) and Median M3(r=-0.30). The cross-validated classification accuracy rate based on the discriminant function is 90%. TABLE S15 shows means of the above nine variables separately for infertile and fertile male. Based on Table 13, all nine variables (PR, M, NX, C, % Total, % M2, % M%, Mean M3 and Median M3) clearly separate the two groups. For example, PR separates the two groups with a mean of 11.5 for infertile male and 29 for fertile male. %M3 has mean 18.6 for infertile group versus mean 11.6 for the fertile group, which is in accordance with the proposed property of SPTRX3 protein as a biomarker specific to defective spermatozoa associated with male infertility.