To the Editor:
We read with great interest the paper “Antimüllerian hormone in gonadotropin releasing-hormone antagonist cycles: prediction of ovarian response and cumulative treatment outcome in good-prognosis patients” (1) by Arce et al. In that paper the authors make a secondary analysis of the MEGASET study, concluding that antimüllerian hormone (AMH) can predict oocyte yield, categories of ovarian response, and live birth from IVF, and that AMH was associated with these outcomes, irrespective of the type of gonadotropin used. Moreover, as Nelson pointed out in the same issue of the journal (2), in Arce and colleagues’ analysis it was observed that the antral follicle count (AFC) was not associated with any of these outcomes.
There are two aspects that we would like to comment on. First, as already mentioned by Nelson, previous studies have observed that AMH and AFC were essentially equivalent in ovarian response prediction. We recently analyzed our population of oocyte donors (3) that presented similar inclusion criteria to that of the study by Arce (1): normal ovarian reserve (OR) by AFC and basal follicle-stimulating hormone (FSH). In this population treated with antagonists, AMH levels (analyzed by the Gen 2 ELISA) correlated significantly with the number of metaphase II oocytes (MII) retrieved. The AMH cutoff to predict a retrieval <6 MII was 2.31 ng/ml. AMH showed a mild capacity to discriminate poor response (AUC 0.675). We carried out a multiple regression analysis including age, AFC, and FSH in order to obtain a poor ovarian response prediction model and the AUC was 0.668 (95 % CI 0.540-0.796); if AMH was added to the model, the AUC was 0.713 (95% CI 0.596-0.830), slightly improving the prediction capacity. According to our results, measuring AMH is not an advantage for those reproductive medicine centers with easy access to AFC and FSH. In Arce’s paper, blood samples were analyzed at a central laboratory, whereas the AFC was performed at each investigational site by different observers, and a sonographer-dependent variability has been suggested.
It is not infrequent to see patients coming from other reproductive centers with hormone analysis results that were performed in another laboratory. Repeating the same analysis would lead to unnecessary cost, especially if one has easy access to high-quality ultrasound. This availability is a prerequisite in any reproductive center, although not all clinics may have an approved laboratory to determine AMH.
On the other hand, we find remarkable the fact that the cutoff of AMH to predict hyperresponse is lower for hpHMG (28 pmol/L) than for FSH (31 pmol/L); if we take into account that with FSH the number of retrieved oocytes was significantly higher than with hpHMG, we would expect the opposite finding.
Finally, we would like to state that Arce’s analysis is very interesting and provides a useful tool to individualize treatments; nevertheless, we think that those reproductive centers that do not have easy access to AMH testing but do have skill in AFC measurement can still rely on AFC as a marker of ovarian response.
Francisca Martínez, Ph.D., Marta Devesa, M.D., Pedro Nolasc Barri, Ph.D.
Institut Universitari Dexeus, Barcelona, Spain
1. Arce JA, La Marca A, Klein BK, Andersen AN, Fleming R. Antimüllerian hormone in gonadotropin releasing-hormone antagonist cycles: prediction of ovarian response and cumulative treatment outcome in good-prognosis patients. Fertil Steril 2013; 99:1644–53.
2. Nelson SM. Antimüllerian hormone: is the writing on the wall for antral follicle count? Fertil Steril 2013;99:1563-4.
3. Martínez F, Clua E, Carreras O, Tur R, Rodríguez I, Barri PN. Is AMH useful to reduce low ovarian response to GnRH antagonist protocol in oocyte donors? Gynecol Endocrinol 2013, accepted.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2013.05.049
The authors respond:
We appreciate the comments by Martínez et al., allowing us to provide further information. Three recent large, international multicenter trials (1, 2, 3) have consistently found that antimüllerian hormone (AMH) is the best predictor of ovarian response in GnRH antagonist and long GnRH agonist cycles with respect to number of oocytes retrieved and categorization of low and high responders. These investigations showed that antral follicle count (AFC) by itself is a poorer predictor than AMH and further that AFC provided no added predictive value beyond AMH. Even though a number of earlier smaller, single-center studies point to a similar predictive value of AMH and AFC in terms of poor and excessive response prediction, introduction of AMH in the recent large, multicenter studies (1, 2, 3) have given substantial evidence to support the views of Nelson (4) that AMH could be considered the single biomarker of choice for prediction of ovarian response to gonadotropin treatment.
In their letter, Dr. Martínez et al. question whether AMH and AFC were given the same opportunities in our analysis (1), since AMH was analyzed centrally and AFC locally. The thinking would be that the lack of predictive value of AFC in a multicenter study could be attributed to operator variability across centers. We have analyzed data separately for each clinic participating in our study and overwhelmingly find that AMH indeed is a better predictor of oocyte yield than AFC, not only across the study but also within individual clinics. When analyzing clinics enrolling ≥10 patients (exclusion of clinics with few observations), 16 of 18 clinics had numerically higher correlation coefficients with AMH than AFC. When further restricting the analysis to high-recruiting clinics with ≥25 patients, the data showed that at seven of eight centers AMH had a numerically higher correlation coefficient than AFC. Thus, when looking at each clinic’s data, the general conclusion is that AMH is a better predictor of ovarian response to gonadotropin therapy than AFC. We must also comment that from a statistical methodological perspective, AFC was given the same opportunity as AMH in the analyses (1), such as simultaneous inclusion of both parameters in a multiple regression model. This model showed that AMH was strongly significant (p<0.0001) while AFC was not significant (p=0.0938), indicating that AFC provided no added predictive value beyond AMH.
In our opinion, based upon these and similar examinations, AFC remains prone to inter- and intra-operator bias within clinics and there appears to be systematic differences among clinics in the way AFC is done (mixed technology and absence of agreed criteria). Furthermore, the assay variation of AMH (Gen 2 ELISA) is well documented and AMH exhibits high intra- and inter-cycle reproducibility compared with AFC. Thus, AMH seems to be a more robust marker of ovarian response to gonadotropin treatment than AFC. It is also important to note that earlier concerns related to AMH assay standardization have been addressed (5) and that establishment of an international standard will facilitate its use in clinical practice.
Regarding the comment on AMH cutoff for prediction of high ovarian response, we consider the values for the two gonadotropin preparations (28 and 31 pmol/L for hphMG and rFSH, respectively) to be in a similar range, ~30 pmol/L, and furthermore consistent with what has previously been reported in GnRH agonist cycles (30 pmol/L for both hphMG and rFSH) (2).
Finally, Nelson remarked that medicine is trying to move toward predictive, preventive, and personalized treatment (4). We believe that in order for the area of reproductive medicine to follow this path, efforts must be made to provide consistent and standardized methods, independent of operators and clinics, upon which treatment strategies can be established for each individual patient seeking infertility treatment.
Joan-Carles Arce, M.D., Ph.D.
Reproductive Health, Ferring Pharmaceuticals A/S, Copenhagen, Denmark
Antonio La Marca, M.D., Ph.D.
Mother-Infant Department, University of Modena and Reggio Emilia, Modena, Italy
Bjarke Mirner Klein, Ph.D.
Global Biometrics, Ferring Pharmaceuticals A/S, Copenhagen, Denmark
Anders Nyboe Andersen, M.D.
Fertility Clinic, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark
Richard Fleming, Ph.D.
Department of Reproductive and Maternal Medicine, University of Glasgow, Glasgow, United Kingdom
1. Arce J-C, La Marca A, Klein BM, Nyboe Andersen A, Fleming R. Antimüllerian hormone in gonadotropin releasing-hormone antagonist cycles: prediction of ovarian response and cumulative treatment outcome in good-prognosis patients. Fertil Steril 2013;99:1644–53.
2. Anckaert E, Smitz J, Schiettecatte J, Klein BM, Arce J-C. The value of anti-müllerian hormone measurement in the long GnRH agonist protocol: association with ovarian response and gonadotropin-dose adjustments. Hum Reprod 2012;27:1829–39.
3. Nyboe Andersen A, Witjes H, Gordon K, Mannaerts B. Predictive factors of ovarian response and clinical outcome after IVF/ICSI following a rFSH/GnRH antagonist protocol with or without oral contraceptive pre-treatment. Hum Reprod 2011;12:3413–23.
4. Nelson SM. Antimüllerian hormone: is the writing on the wall for antral follicle count? Fertil Steril 2013;99:1563–4.
5. Nelson SM, Anderson RA, Broekmans FJ, Raine-Fenning N, Fleming R, La Marca A. Anti-Müllerian hormone: clairvoyance or crystal clear? Hum Reprod 2012;27:631–6.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2013.05.050