Reply to ‘Value of basal FSH concentrations: prognostic implications for pregnancy outcome’
1 Research Institute for Endocrinology, Reproduction and Metabolism, Department of Obstetrics and Gynaecology, Vrije Universiteit Medical Centre, Amsterdam and 2 Department of Obstetrics and Gynaecology, St Franciscus Gasthuis, Rotterdam, The Netherlands
* Email: j.vanmontfrans{at}zonnet.nl
Sir,
Dr Styne-Gross and colleagues correctly note an inconsistency in one of our previous papers (van Montfrans et al., 2004
). In this study, we accidentally reported having used the Amerlite ILMA for FSH, whereas in fact the immunometric assay from Delfia, Wallac Turku, Finland was used. Test characteristics were described correctly, however. The cut-off value for elevated FSH in the Amerlite assay (10.0 IU/l) that was used previously in our institution (van Montfrans et al., 2000) corresponded to 12.5 IU/l in the Delfia assay, and described the threshold for decreased success in our IVF programme.
Several publications in the recent literature describe the concept of diminished ovarian reserve and its relationship to aneuploidy (e.g. Nasseri et al., 1999; van Montfrans et al., 1999; Freeman et al., 2000). Two factors complicate studying this relationship. First, no direct markers for ovarian reserve are available. Secondly, there is no uniform definition of diminished ovarian reserve. It may lead to early menopause and subfertility while pursuing spontaneous pregnancy or during assisted reproduction technology (ART), which are all separate clinical entities. Cut-off values for normal and abnormal cannot be interchanged automatically between these entities.
Our study described basal FSH values in relation to an indirect marker of ovarian function, i.e. the rate of (early) pregnancy loss (van Montfrans et al., 2004). As no data on basal FSH and (very) early pregnancy loss (i.e. measurable HCG during a presumed menstrual period without a noted increase in cycle length) have been published previously, no cut-off value for abnormal FSH was available either. This implies studying the putative relationship by measuring the effect of basal FSH on pregnancy outcome, preferably in a prospective study design. Analysing the data in such a study implies regression analysis, or a comparable test, to evaluate the existence of such an effect. A 
2 test can be performed as well, with the 5% of participants with the highest FSH levels classified as being abnormal. Classifying the highest 5% of values as abnormal is in common use in medical statistics when no prior data on cut-off levels are available.
We used the cut-off level of 12.5 IU/l for FSH, correlating with decreased fertility in IVF in our clinic, corresponding to the highest 6% of values in the current study. Strictly speaking, it would have been better to use only the highest 5% of values and ignore the cut-off value in the IVF programme, since we are studying a separate clinical entity. It appeared, however, that both values (cut-off level in IVF and highest 5%) were practically identical, and both were mentioned in the paper. In fact, the FSH cut-off value used by Levi et al. (2001) has also been calculated using data of patients with an abnormally low response in an IVF programme instead of pregnancy outcome as suggested in their paper. The results of our study (both the regression analysis and the 2 test) showed no relationship between basal FSH and pregnancy outcome at all, strongly pleading against any relationship.
Is this study design comparable with testing thyroid-stimulating hormone values to predict tachycardia in a normal population? No, since in a normal population, the rate of the outcome marker (tachycardia in the example) can be expected to be low and therefore a relationship is less likely to be discovered, whereas in our study, the incidence of the outcome variable of interest [(early) pregnancy loss] was 47%. Given this percentage and the fact that no relationship to basal FSH was discovered, a relationship between basal FSH and early pregnancy loss becomes highly unlikely.
Would the outcome be different if women with higher basal FSH values than in our paper were studied? In these women with ‘extremely’ elevated FSH, the incidence of (early) pregnancy losses could be higher than in our participants. However, the incidence of women with FSH levels significantly above the levels we measured would in any population (with or without subfertility) be so low that it would not alter our understanding of early pregnancy loss. Does variation in basal FSH values automatically have no significance in relation to ovarian function when the majority of values evaluated are normal? No, as we have for example shown previously that differences in basal FSH concentrations were related to the incidence of Down's syndrome in a case–control study (van Montfrans et al., 1999).
The authors express their concern that women with elevated basal FSH should be counselled regarding decreased fertility and increased risk for (early) pregnancy loss. Excluding IVF patients, proper counselling of these women remains a problem in our opinion. The majority of studies reporting fertility in this group were retrospective, and decreased fertility was not reported unanimously (van Montfrans et al., 2000; van Rooij et al., 2004). As for the relationship between ovarian reserve and pregnancy loss or aneuploidies, this relationship was only described in retrospective studies. As far as the study by Levi et al. (2001) is concerned, we believe that its retrospective design with the possibility of verification bias, the fact that only clinical pregnancies noted in a computer database containing FSH data were included (and thus pregnancies diagnosed elsewhere may have been missed) and the inclusion of oligomenorrhoeic patients preclude using its results in patient counselling in the general population. The study of Levi et al. is a retrospective study in a different population evaluating a different primary outcome parameter from that in our prospective study.
Our study, being the first prospective study into ovarian reserve and pregnancy loss, found no relationship. We conclude that robust data from prospective studies evaluating the relationship between ovarian reserve and pregnancy loss or aneuploidies remain to be published, before counselling as suggested by Dr Styne-Gross and colleagues becomes good clinical practice.
References
Freeman SB, Yang Q, Allran K, Taft LF and Sherman SL (2000) Women with a reduced ovarian complement may have an increased risk for a child with Down syndrome. Am J Hum Genet. 66, 1680–1683.[CrossRef][Web of Science][Medline]
Levi AJ, Raynault MF, Bergh PA, Drews MR, Miller BT and Scott RT Jr (2001) Reproductive outcome in patients with diminished ovarian reserve. Fertil Steril 76, 666–669.[CrossRef][Web of Science][Medline]
Nasseri A, Mukherjee T, Grifo JA, Noyes N, Krey L and Copperman AB (1999) Elevated day 3 serum follicle stimulating hormone and/or estradiol may predict fetal aneuploidy. Fertil Steril 71, 715–718.[CrossRef][Web of Science][Medline]
van Montfrans JM, Dorland M, Oosterhuis GJ, van Vugt JM, Rekers-Mombarg LT and Lambalk CB (1999) Increased concentrations of follicle-stimulating hormone in mothers of children with Down's syndrome. Lancet 353, 1853–1854.[CrossRef][Web of Science][Medline]
van Montfrans JM, Hoek A, van Hooff MH, de Koning CH, Tonch N and Lambalk CB (2000) Predictive value of basal follicle-stimulating hormone concentrations in a general subfertility population. Fertil Steril 74, 97–103.[CrossRef][Web of Science][Medline]
van Montfrans JM, van Hooff MH, Huirne JA, Tanahatoe SJ, Sadrezadeh S, Martens F, van Vugt JM and Lambalk CB (2004) Basal FSH concentrations as a marker of ovarian ageing are not related to pregnancy outcome in a general population of women over 30 years. Hum Reprod 19, 430–434.
van Rooij IA, De Jong E, Broekmans FJ, Looman CW, Habbema JD and Te Velde ER (2004) High follicle-stimulating hormone levels should not necessarily lead to the exclusion of subfertile patients from treatment. Fertil Steril 81, 1478–1485.[CrossRef][Web of Science][Medline]
Submitted on June 21, 2004; accepted on July 2, 2004.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||