Total Motile Sperm Count
In this study, couples with unexplained infertility have, after correction for confounding factors, a higher SOPR than couples in any of the WHO classes of male infertility or any of the TMSC groups below 20 × 10 spermatozoa. In between the various WHO groups with male infertility the SOPRs were not significantly different. The TMSC classification, on the other hand, shows a significant correlation with the SOPR. In addition, the TMSC is easy to calculate. Therefore, we conclude that the TMSC is more useful than the WHO classification system for expressing the severity of male factor infertility.
For daily practice, three prognostic groups can be discerned: couples with a TMSC <5 × 10, couples with a TMSC between 5 and 20 × 10 and couples with a TMSC of more than 20 × 10 spermatozoa (normospermia). It is striking to see that there were no differences in SOPR whether the TMSC was <1 × 10, 1–3 × 10 or 3–5 × 10 (data not shown). Spontaneous pregnancies occur even in the presence of extremely poor sperm quality. Most physicians will recommend those couples to start with ICSI straight away, as we did. Yet, it is remarkable that about a quarter of those couples conceived spontaneously. Once again it is shown that the semen analysis is a poor predictor of pregnancy in this low range (Van der Steeg et al., 2011).
If the best semen analysis is in the normal TMSC range, after the first semen analysis was abnormal (normalized group), the pregnancy chances of these couples seem to be at the level of the couples with consistently poorer semen quality. This shows that multiple semen analyses seem to have no additional prognostic value. One abnormal semen analysis already determines the prognosis. However this observation should be evaluated further in a larger group. This is in contrast to the Dutch national network guideline and National Institute for Health and Care Excellence (NICE) fertility guidelines (National Institute for Health and Care Excellence, 2013) that advise repeating the test if the first one is abnormal. These guidelines refer to the study of Opsahl et al. (1996), who found that about 10% of men with an abnormal semen at first analysis eventually show a normal sperm count when more samples are examined. The NICE fertility guidelines recommended performing more tests, but did not validate their statement with pregnancy rates.
The WHO reference limits are determined in a large group of fertile men, who recently fathered a child. Men with semen results below these limits, however, are not necessary infertile. On the other hand, sperm results above the reference limits do not guarantee the occurrence of a pregnancy, as other factors might influence the outcome. We think that criteria to assess semen quality should be based on pregnancy chances in infertile couples after exclusion of other causes of infertility and should be corrected for confounding factors such as female age and duration of infertility.
Ombelet et al. (1997) compared different semen parameters in fertile and infertile men and based on receiver operating characteristic curves, they concluded that sperm morphology was best able to predict which group (fertile or infertile) a person belongs to. However, no attempt was made to correlate the findings with pregnancy chance. It is interesting to see that their suggestions for reference limits come very close to the current WHO criteria. In our study morphology was not seen as discriminative, despite the fact that our hospitals participate in a national program to standardize the laboratory interpretation. This is in agreement with the recent study of Deveneau et al. (2014) in an IUI program.
The semen analysis is a good predictor if it correlates with pregnancy chance. There is hardly any study that tried to validate the semen analysis with SOPR. Polansky and Lamb (1988) did not find any significant influence of any semen characteristic on the probability of a spontaneous conception in a cohort of 1089 infertile couples. Ayala et al. (1996) showed in a cohort of 1055 infertile couples that the TOPR was significantly higher if the TMSC was more than 25 × 10 compared with a TMSC below 25 × 10 [relative risk 6.1 (95% CI: 4.7–7.9)]. In both studies total pregnancy rates were calculated regardless of female infertility factors and treatment given.
Van der Steeg et al. (2011) carried out a large multicenter cohort study and measured the pregnancy rate within 12 months. They censored couples at the start of treatment or on the last date of contact in case of expectant management. In 41% of the couples the man had a normospermia, in the remaining 59% the man was diagnosed with abnormal sperm yet, the spontaneous pregnancy rate was 24% in the first group and 23% in the second group. In their study the different WHO categories did not show a clear pattern of success rate. OAT had the lowest pregnancy rate with 12% in 12 months.
The reason for the lack of discriminating potential might be that the WHO criteria make use of cut-off points, while the pregnancy rate increases continuously with increasing values for individual parameters. By dichotomizing the sperm results the effect of the slope is ignored. For example, when a couple is diagnosed with OAT but the semen parameters are just below the cut-off values, the WHO classification will categorize this couple in the same group as a couple with extremely poor parameters. No discrimination will be made, while the TMSC takes the absolute value of three semen parameters into consideration simultaneously. This was nicely shown by van der Steeg et al. (2011). They concluded that the cut-off values as defined by the WHO are not good predictors, as the spline curves of the different sperm parameters they produced clearly show that the actual values are more informative than purely dichotomizing the parameters, as the WHO does. Van der Steeg et al. (2011) made a prediction model which incorporates several sperm parameters and the TSMC, in fact, comes close to this model. In the prognostic model for spontaneous pregnancy of Hunault et al. (2004) motility was the only sperm parameter included into the model.
A strength of our study is that the spontaneous pregnancy rate was measured and that this rate was corrected for confounding factors. This was done in an unselected longitudinal cohort of infertile couples. The study is therefore representative for couples referred with an infertility problem for the first time. We did exclude other female infertility diagnoses. The follow-up time was 3 years. Couples who had stopped treatment in our hospitals were contacted to ask whether any pregnancy had occurred after they had discontinued treatment.
A weak point of our study is that it is an observational study in which besides expectative management various treatments were given. To assess the spontaneous pregnancy chance in the different groups of male factor infertility, ideally, no treatment is given for a longer period of time. In general practice this is hardly possible. It is hard to believe that enough patients would be willing to participate in such a study, particularly when potent treatments such as IUI, IVF and ICSI are available.
In our study couples were not censored at the moment that treatment was started. Censoring couples at the time the treatment was started, as van der Steeg et al. (2011) did, might seem more appropriate at first glance. However, this gives an underestimation of the SOPR in the long run, as spontaneous pregnancies which occur during and after treatments are not included. The follow-up period in our study was 3 years. Most spontaneous pregnancies occur in the first year (Brandes et al., 2011a) yet the cumulative spontaneous pregnancy curve continues to rise, even after IUI, IVF and ICSI are discontinued.
It is difficult to say how many couples would have become pregnant if treatment was not started. We have shown that quite a number of spontaneous pregnancies occur during and after treatment, even in couples that according to the national guidelines must be offered ICSI directly. Despite the fact that we started with 90% of couples with a TMSC of <1 with ICSI, 25% conceived spontaneously over a period of 3 years.
The results of the semen analysis are not only used to determine the prognosis, but also to choose the appropriate mode of treatment. Because there is no specific WHO class with the poorest outcome, as the SOPRs among various WHO male infertility classes are comparable, it is questionable whether the WHO criteria can be used to determine the proper treatment strategy. We recommend using the TMSC classification, as is common practice in the Netherlands. Yet, we lack proper randomized controlled trials (RCTs) to choose the right therapy for each couple. In the INeS-trial (IUI, Natural cycle and Single embryo transfer) which compared (i) IUI with controlled ovarian hyperstimulation, (ii) modified natural cycle IVF with single embryo transfer (SET) and (iii) IVF with SET in couples with unexplained infertility and mild male infertility, there were no significant differences in pregnancy rate between the three groups (Bensdorp et al. 2013). As IUI is much cheaper and less invasive, this therapy should be offered as the first choice treatment (van Rumste et al., 2014). For moderate and severe male infertility, in the Netherlands the MASTER-trial (Male Subfertility Therapy Effectiveness RCT's) is currently ongoing, in which two RCTs for different TMSC classes of male infertility will be performed (Cissen and de Bruin, 2014). In the moderate range of male infertility (prewash TMSC 3 to 10 × 10) IUI is compared with expectant treatment. In the severe range of male infertility (prewash TMSC below 3 × 10 and postwash above 3 × 10) ICSI is compared with IVF. Hopefully the results of this study will shed some new light on this topic.
In conclusion, the TMSC grading appears to be a better way to classify male factor infertility than the WHO classification system. The TMSC classification should be used in prospective RCT to show how the different forms of male infertility are best treated.
Discussion
In this study, couples with unexplained infertility have, after correction for confounding factors, a higher SOPR than couples in any of the WHO classes of male infertility or any of the TMSC groups below 20 × 10 spermatozoa. In between the various WHO groups with male infertility the SOPRs were not significantly different. The TMSC classification, on the other hand, shows a significant correlation with the SOPR. In addition, the TMSC is easy to calculate. Therefore, we conclude that the TMSC is more useful than the WHO classification system for expressing the severity of male factor infertility.
For daily practice, three prognostic groups can be discerned: couples with a TMSC <5 × 10, couples with a TMSC between 5 and 20 × 10 and couples with a TMSC of more than 20 × 10 spermatozoa (normospermia). It is striking to see that there were no differences in SOPR whether the TMSC was <1 × 10, 1–3 × 10 or 3–5 × 10 (data not shown). Spontaneous pregnancies occur even in the presence of extremely poor sperm quality. Most physicians will recommend those couples to start with ICSI straight away, as we did. Yet, it is remarkable that about a quarter of those couples conceived spontaneously. Once again it is shown that the semen analysis is a poor predictor of pregnancy in this low range (Van der Steeg et al., 2011).
If the best semen analysis is in the normal TMSC range, after the first semen analysis was abnormal (normalized group), the pregnancy chances of these couples seem to be at the level of the couples with consistently poorer semen quality. This shows that multiple semen analyses seem to have no additional prognostic value. One abnormal semen analysis already determines the prognosis. However this observation should be evaluated further in a larger group. This is in contrast to the Dutch national network guideline and National Institute for Health and Care Excellence (NICE) fertility guidelines (National Institute for Health and Care Excellence, 2013) that advise repeating the test if the first one is abnormal. These guidelines refer to the study of Opsahl et al. (1996), who found that about 10% of men with an abnormal semen at first analysis eventually show a normal sperm count when more samples are examined. The NICE fertility guidelines recommended performing more tests, but did not validate their statement with pregnancy rates.
The WHO reference limits are determined in a large group of fertile men, who recently fathered a child. Men with semen results below these limits, however, are not necessary infertile. On the other hand, sperm results above the reference limits do not guarantee the occurrence of a pregnancy, as other factors might influence the outcome. We think that criteria to assess semen quality should be based on pregnancy chances in infertile couples after exclusion of other causes of infertility and should be corrected for confounding factors such as female age and duration of infertility.
Ombelet et al. (1997) compared different semen parameters in fertile and infertile men and based on receiver operating characteristic curves, they concluded that sperm morphology was best able to predict which group (fertile or infertile) a person belongs to. However, no attempt was made to correlate the findings with pregnancy chance. It is interesting to see that their suggestions for reference limits come very close to the current WHO criteria. In our study morphology was not seen as discriminative, despite the fact that our hospitals participate in a national program to standardize the laboratory interpretation. This is in agreement with the recent study of Deveneau et al. (2014) in an IUI program.
The semen analysis is a good predictor if it correlates with pregnancy chance. There is hardly any study that tried to validate the semen analysis with SOPR. Polansky and Lamb (1988) did not find any significant influence of any semen characteristic on the probability of a spontaneous conception in a cohort of 1089 infertile couples. Ayala et al. (1996) showed in a cohort of 1055 infertile couples that the TOPR was significantly higher if the TMSC was more than 25 × 10 compared with a TMSC below 25 × 10 [relative risk 6.1 (95% CI: 4.7–7.9)]. In both studies total pregnancy rates were calculated regardless of female infertility factors and treatment given.
Van der Steeg et al. (2011) carried out a large multicenter cohort study and measured the pregnancy rate within 12 months. They censored couples at the start of treatment or on the last date of contact in case of expectant management. In 41% of the couples the man had a normospermia, in the remaining 59% the man was diagnosed with abnormal sperm yet, the spontaneous pregnancy rate was 24% in the first group and 23% in the second group. In their study the different WHO categories did not show a clear pattern of success rate. OAT had the lowest pregnancy rate with 12% in 12 months.
The reason for the lack of discriminating potential might be that the WHO criteria make use of cut-off points, while the pregnancy rate increases continuously with increasing values for individual parameters. By dichotomizing the sperm results the effect of the slope is ignored. For example, when a couple is diagnosed with OAT but the semen parameters are just below the cut-off values, the WHO classification will categorize this couple in the same group as a couple with extremely poor parameters. No discrimination will be made, while the TMSC takes the absolute value of three semen parameters into consideration simultaneously. This was nicely shown by van der Steeg et al. (2011). They concluded that the cut-off values as defined by the WHO are not good predictors, as the spline curves of the different sperm parameters they produced clearly show that the actual values are more informative than purely dichotomizing the parameters, as the WHO does. Van der Steeg et al. (2011) made a prediction model which incorporates several sperm parameters and the TSMC, in fact, comes close to this model. In the prognostic model for spontaneous pregnancy of Hunault et al. (2004) motility was the only sperm parameter included into the model.
A strength of our study is that the spontaneous pregnancy rate was measured and that this rate was corrected for confounding factors. This was done in an unselected longitudinal cohort of infertile couples. The study is therefore representative for couples referred with an infertility problem for the first time. We did exclude other female infertility diagnoses. The follow-up time was 3 years. Couples who had stopped treatment in our hospitals were contacted to ask whether any pregnancy had occurred after they had discontinued treatment.
A weak point of our study is that it is an observational study in which besides expectative management various treatments were given. To assess the spontaneous pregnancy chance in the different groups of male factor infertility, ideally, no treatment is given for a longer period of time. In general practice this is hardly possible. It is hard to believe that enough patients would be willing to participate in such a study, particularly when potent treatments such as IUI, IVF and ICSI are available.
In our study couples were not censored at the moment that treatment was started. Censoring couples at the time the treatment was started, as van der Steeg et al. (2011) did, might seem more appropriate at first glance. However, this gives an underestimation of the SOPR in the long run, as spontaneous pregnancies which occur during and after treatments are not included. The follow-up period in our study was 3 years. Most spontaneous pregnancies occur in the first year (Brandes et al., 2011a) yet the cumulative spontaneous pregnancy curve continues to rise, even after IUI, IVF and ICSI are discontinued.
It is difficult to say how many couples would have become pregnant if treatment was not started. We have shown that quite a number of spontaneous pregnancies occur during and after treatment, even in couples that according to the national guidelines must be offered ICSI directly. Despite the fact that we started with 90% of couples with a TMSC of <1 with ICSI, 25% conceived spontaneously over a period of 3 years.
The results of the semen analysis are not only used to determine the prognosis, but also to choose the appropriate mode of treatment. Because there is no specific WHO class with the poorest outcome, as the SOPRs among various WHO male infertility classes are comparable, it is questionable whether the WHO criteria can be used to determine the proper treatment strategy. We recommend using the TMSC classification, as is common practice in the Netherlands. Yet, we lack proper randomized controlled trials (RCTs) to choose the right therapy for each couple. In the INeS-trial (IUI, Natural cycle and Single embryo transfer) which compared (i) IUI with controlled ovarian hyperstimulation, (ii) modified natural cycle IVF with single embryo transfer (SET) and (iii) IVF with SET in couples with unexplained infertility and mild male infertility, there were no significant differences in pregnancy rate between the three groups (Bensdorp et al. 2013). As IUI is much cheaper and less invasive, this therapy should be offered as the first choice treatment (van Rumste et al., 2014). For moderate and severe male infertility, in the Netherlands the MASTER-trial (Male Subfertility Therapy Effectiveness RCT's) is currently ongoing, in which two RCTs for different TMSC classes of male infertility will be performed (Cissen and de Bruin, 2014). In the moderate range of male infertility (prewash TMSC 3 to 10 × 10) IUI is compared with expectant treatment. In the severe range of male infertility (prewash TMSC below 3 × 10 and postwash above 3 × 10) ICSI is compared with IVF. Hopefully the results of this study will shed some new light on this topic.
In conclusion, the TMSC grading appears to be a better way to classify male factor infertility than the WHO classification system. The TMSC classification should be used in prospective RCT to show how the different forms of male infertility are best treated.