- Email: [email protected]
New automated antimüllerian hormone assays are more reliable than the manual assay in patients with reduced antral follicle count
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY
New automated antim€ ullerian hormone assays are more reliable than the manual assay in patients with reduced antral follicle count Teddy Tadros, M.D.,a,b Bruno Tarasconi, M.D.,a,b,c Jean Nassar, M.D.,a,b Jean-Luc Benhaim, Pharm.D.,d €lle Taieb, Pharm.D.,e and Renato Fanchin, M.D., Ph.D.a,b Joe ^ pital Foch, Suresnes, France; b University of Paris-Ouest, Suresnes, France; Center of Reproductive Medicine, Ho Coordenação de Aperfeiçoamento de Pessoal de Nìvel Superior (CAPES) Foundation, Ministry of Education of Brazil, dicales Benhaim, Clamart, France; and e Assistance Publique - Hôpitaux de Brasilia, Brazil; d Laboratoire d’Analyses Me ^ pital Antoine Be cle re, Clamart, France Paris (AP - HP), Department of Biochemistry, Ho a c
Objective: To compare the strength of the relationship between antral follicle count (AFC) and serum antim€ ullerian hormone (AMH) concentrations obtained with two automated and one manual AMH assays in three different AFC populations. Design: Prospective cohort study. Setting: University-affiliated IVF-ET center. Patient(s): Frozen–thawed serum samples of 211 assisted conception candidates, aged 24–43 years. Intervention(s): Serum AMH was measured using one manual (AMH Gen II) and two fully automated (Access AMH and Elecsys AMH) assays. Antral follicle count was performed under strictly standardized conditions and sorted into three groups according to tercile values: low AFC (3–12 follicles; n ¼ 73), intermediate AFC (13–20 follicles; n ¼ 65), and high AFC (21–84 follicles; n ¼ 73). Main Outcome Measure(s): Strength of correlation between AMH levels and AFC. Result(s): Overall, AMH levels were lower with Access AMH (16%) and Elecsys AMH (20%) than with AMH Gen II. Remarkably, the strength of correlations between AFC and circulating AMH levels was the same with the three assays (r ¼ 0.83). Yet in the low AFC group, serum AMH levels obtained by Access AMH and Elecsys AMH showed a stronger correlation with AFC (r ¼ 0.63 and r ¼ 0.65, respectively) than the AMH Gen II (r ¼ 0.52), a phenomenon that was not observed in the remaining AFC groups. Conclusion(s): As compared with conventional AMH Gen II assay results, [1] serum AMH concentrations were 16% and 20% lower with Access AMH and Elecsys AMH, respectively; and [2] automated assays were more strongly correlated to AFC in the subset of patients with reduced follicle count. (Fertil SterilÒ 2016;-:-–-. Ó2016 by American Society for Reproductive Medicine.) Key Words: Antim€ ullerian hormone, antral follicle count, ovarian reserve, poor responders Discuss: You can discuss this article with its authors and with other ASRM members at
A
ssociated or not to ultrasonographic counting of antral follicles, serum antim€ ullerian hormone (AMH) measurements have become the reference in the clinical appraisal of the ovarian follicular status (1), and its clinical use has been continuously gaining momentum during the last 12 years. Antim€ ullerian hormone levels reflect the activity of granulosa
cells of small antral follicles, thereby providing patients and physicians with invaluable information on ovarian aging and how to individualize controlled ovarian hyperstimulation protocols (2). Yet from a practical standpoint, in addition to cost, problems related to reliability and consistency of serum AMH measurements have raised doubts about their clinical soundness. These
Received May 25, 2016; revised August 16, 2016; accepted August 22, 2016. T.T. has nothing to disclose. B.T. has nothing to disclose. J.N. has nothing to disclose. J.-L.B. has nothing to disclose. J.T. has nothing to disclose. R.F. has nothing to disclose. ^ pital Foch, 40 rue Reprint requests: Renato Fanchin, M.D., Ph.D., Center of Reproductive Medicine, Ho Worth, 92151 Suresnes, France (E-mail: [email protected]). Fertility and Sterility® Vol. -, No. -, - 2016 0015-0282/$36.00 Copyright ©2016 American Society for Reproductive Medicine, Published by Elsevier Inc. http://dx.doi.org/10.1016/j.fertnstert.2016.08.045 VOL. - NO. - / - 2016
problems essentially are attributed, on the one hand, to an uncoordinated development of AMH assays, which displayed different callibration and standards and, on the other hand, to the compulsory operatorand technique-dependent manipulations. Recently, to overcome these limitations and to improve quality of AMH measurements, fully automated AMH assays (Access AMH [Beckman Coulter] and Elecsys AMH [Roche Diagnostics International]) have been developed and commercialized (3, 4). Yet analytical data available have been limited to the confirmation of an adequate concordance of serum AMH levels obtained by these assays (4–11).
1
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY Unfortunately, only limited and/or indirect data taking the necessary relationship between AMH and the number of AMH-producing antral follicles as a reference have been hitherto published (5–7). Because it is undoubtedly the small antral follicles that produce most of the circulating AMH concentrations (12–14), antral follicle count (AFC), when it is performed under optimal conditions, should be taken as the standard for comparing reliability of AMH assays. Indeed, serum AMH levels have been strongly correlated to AFC by numerous investigators (1, 13, 15, 16). Moreover, the relative performance of new automated and manual AMH assays in subgroups of patients displaying low, intermediate, and high AFC remains undetermined. Therefore, this insufficient knowledge spurred us to investigate the reliability and concordance of AMH levels obtained by one manual and two automated AMH assays in different AFC populations.
MATERIALS AND METHODS Subjects and Procedures We used frozen–thawed serum aliquots obtained from 211 women who were willing to enter our assisted conception program between April 2015 and July 2015. Inclusion criteria were [1] both ovaries present without morphologic abnormalities (menstruating patients with or without polycystic ovary syndrome or polycystic ovary morphology were included); [2] optimum ovarian visualization at transvaginal ultrasound scans; and [3] body mass index ranging between 18 and 25 kg/m2. All blood samples were collected between days 1 and 5 of the menstrual cycle and remained frozen from 2 to 4 months before thawing. Serum AMH determinations were performed using three different ELISAs according to manufacturer protocols by a single operator (J.-L.B.): modified AMH Gen II (Beckman Coulter), Access AMH (Beckman Coulter), and Elecsys AMH (Roche Diagnostics International). Detailed AMH assay procedures have been previously described (6). In brief, limits of detection of the three AMH assays tested were, respectively, 0.08, 0.02, and 0.01 ng/mL; limits of quantification, 0.16, 0.08, and 0.03 ng/mL; and maximum imprecision, 8.0%, 4.3%, and 3.5%. Moreover, serum E2, FSH, and LH levels were determined by an automated multianalysis system using a chemiluminescence technique (Cobas e411 Analyzer, Roche Diagnostics, Mannheim, Germany). For E2, limit of quantification was 15 pg/mL, and maximum imprecision was 8%. For FSH and LH, limit of quantification was 0.1 mIU/mL, and maximum imprecision was 3%. In parallel to blood samplings, AFC (antral follicles measuring 3–10 mm in diameter) was carefully and exhaustively determined using a 5–9-MHz transvaginal ultrasound probe (RIC 5-9-D; Voluson E8 Expert, General Electric Medical Systems, Paris, France) by a single operator (T.T.) who was unaware of AMH results. Given that the present investigation was limited exclusively to hormonal measurements in frozen–thawed serum aliquots and ultrasound scan records and that patients had previously given their informed consent for additional hormonal analysis using their stored sera, our 2
local institutional review board advised us that it did not require ethics committee submission.
Definition of AFC Groups Participants were arbitrarily sorted into three different AFC groups according to the 33rd and 66th centiles of AFC distribution (12 and 20 antral follicles): low AFC (3–12 follicles; n ¼ 73), intermediate AFC (13–20 follicles; n ¼ 65), and high AFC (21–84 follicles; n ¼ 73). These cutoffs were used as an effort to pragmatically identify patients with different ovarian follicular phenotypes.
Statistics Because data distribution was considered nonparametric, we elected to use the median as the measure of central tendency and minimum–maximum as the measure of variability. Circulating AMH levels were compared two by two among the three different assays tested by using the Wilcoxon signed-rank test. The strength of relationships between AFC and serum AMH levels provided by the three different assays was assessed by the Spearman correlation test in the total population and in the three AFC groups. Comparison of Spearman correlation coefficients was performed using the Fisher r-to-z transformation, a statistical procedure that can be applied to assess the significance of the difference between two correlation coefficients found in two independent samples. In addition, agreement between the different AMH assays was assessed graphically using the Bland-Altman plots and by Passing-Bablok regression. The Cusum test for linearity was used to test the applicability of Passing-Bablok regression. A P value of < .05 was considered statistically significant.
RESULTS Population Characteristics Characteristics of individuals included in the present analysis corresponded, as expected, to the profile of women who are usually candidates for assisted conception at our center, encompassing young and aged women and those with small and large counts of antral follicles. Median age (range) was 35 (24–43) years. In the early follicular phase, median serum E2 level was 45 (15–249) pg/mL, and serum FSH and LH levels were 6.9 (3.0–21.6) and 5.1 (1.1–21.1) mIU/mL, respectively. Overall median antral follicle count was 15 (3–84) follicles.
AMH Levels according to Manual and Automated Assays Serum AMH levels measured by the three different ELISA assays are depicted in Figure 1. As shown, each one of the tested assays resulted into statistically different AMH levels. Hence, AMH Gen II provided AMH levels at 1.97 (0.04-30.66) ng/mL [median (interquartile range)] that were higher (P< .001) than those obtained with Access AMH at 1.66 (0.04–30.46) ng/mL and with Elecsys AMH at 1.58 (0.04–26.17) ng/mL. In other words, AMH Gen II levels are, on average, 16% and 20% as high as Access AMH and Elecsys AMH, respectively. In addition, it is noteworthy that statistically different results were observed between VOL. - NO. - / - 2016
Fertility and Sterility®
FIGURE 1
Distribution of serum AMH levels, in logarithmic scale, measured by AMH Gen II, Access AMH, and Elecsys AMH assays in box-and-whisker plots. Central lines are the median, proximal parallel lines are the lower and upper quartiles, distal parallel lines are the 10th and the 90th centiles, and dots are actual AMH levels that were lower or higher than the 10th and the 90th centiles. Note that serum AMH levels measured by AMH Gen II were roughly 16% and 20% higher than those obtained with Access AMH and Elecsys AMH, respectively. In addition, serum AMH levels obtained with Access AMH were approximately 5% higher than those obtained with Elecsys AMH. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
the two automated assays (Access AMH and Elecsys AMH), with Access AMH being 5% as high as Elecsys AMH.
Assay Agreement Agreement between AMH measurement techniques assessed in Bland-Altman plots and by Passing-Bablok regression is illustrated in Figure 2. Overall, we observed adequate concordance between techniques. As we can note in the BlandAltman plots, the number of measurements situated outside the 2-SD interval was 4.7% between AMH Gen II and Access AMH, 4.3% between AMH Gen II and Elecsys AMH, and 2.8% between Access and Elecsys AMH, which may be considered acceptable. Yet it is noteworthy that agreement between assays tended to become progressively weaker as mean AMH levels obtained with the three techniques increased. Accordingly, the number of measurements situated outside the 2-SD interval tended to slightly increase from the low AFC (4.1%, 2.7%, and 5.4%, respectively) to the intermediate AFC (4.6%, 6.1%, and 3.0%, respectively) and the high AFC (6.8%, 6.8%, and 5.4%, respectively) groups. Finally, Passing-Bablok regression showed an adequate relationship between each pair of assays, and the Cusum test for linearity failed to show significant differences between AMH Gen II vs. Access AMH (P¼ .49), AMH Gen II vs. Elecsys AMH (P¼ .72), and Access vs. Elecsys AMH (P¼ .72), either in the whole data set or within each AFC group.
Reliability of the Three Different AMH Assays Correlations between AFC and circulating AMH levels obtained by AMH Gen II, Access AMH, and Elecsys AMH are shown in Figure 3. We could note that, overall, Spearman correlation coefficients were remarkably similar among the three assays (r ¼ 0.83, P< .001; r ¼ 0.83, P< .001; and r ¼ 0.83, P< .001, respectively). Because of this, comparison of Spearman correlation coefficients was not required for this analysis. VOL. - NO. - / - 2016
Spearman correlation coefficients, a representation of the strength of the relationship between AFC and serum AMH levels obtained with the three different assays in the three AFC groups, are depicted in Figure 4. As shown, in contrast to the overall population, in the low AFC group, correlation coefficients obtained with Access AMH (r ¼ 0.63) and Elecsys AMH (r ¼ 0.65) were significantly stronger (P< .03 and P< .005, respectively) than that obtained with AMH Gen II assay (r ¼ 0.52). We failed to show statistically significant differences of correlation coefficients in the intermediate and high AFC groups. This indicates an improved reliability of automated over manual AMH assays exclusively in patients presenting %12 antral follicles.
DISCUSSION The present investigation was designed to evaluate both the reliability and concordance of results of two new automated AMH assays (Access AMH and Elecsys AMH) as compared with the manual AMH assay (AMH Gen II) in a group of randomly selected assisted conception candidates presenting different ovarian phenotypes. For this, our study required a number of methodologic characteristics, such as the inclusion of a large series of patients that allowed the determination of three AFC groups, centralized hormonal measurements performed in frozen–thawed serum aliquots that have been stored for a short time interval (2–4 months), and careful and exhaustive determination of AFC as reference for hormone measurements. The choice of using AFC as a reference was based on the fact that ultrasound-detected antral follicles probably are the paramount source of circulating AMH (12–14). Indeed, studies focusing in the staining of AMH gene in granulosa cells of follicles at different development stages (12) have suggested that small antral follicles contribute to serum AMH levels the most and that preantral follicle contribution is only marginal. Further, the pharmacologic induction of growth of most of 3
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY
FIGURE 2
Agreement between AMH measurement techniques assessed in Bland-Altman plot (A–C; solid lines are the mean of differences between assays, and dotted lines are the 95% confidence interval) and by Passing-Bablok regression (D–F). As illustrated in A–C, the number of measurements situated outside the 2-SD interval was 4.7% between AMH Gen II and Access AMH, 4.3% between AMH Gen II and Elecsys AMH, and 2.8% between Access and Elecsys AMH, thereby confirming adequate concordance between techniques. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
available small antral follicles during controlled ovarian hyperstimulation resulted in a dramatic decline (not sustained by preantral follicle production) of circulating AMH levels (17). Therefore, the choice of AFC as a 4
reference is valid as long as AFC is performed under optimal, rigorous, and standardized conditions, which is not unfortunately the case in some studies that elected to standardize exclusively AMH measurements (18). VOL. - NO. - / - 2016
Fertility and Sterility®
FIGURE 3
Simple regression analysis of the relationship between AFC and serum AMH levels obtained by the AMH Gen II (A), Access AMH (B), and Elecsys AMH (C). Spearman correlation coefficients were remarkably similar among the three assays, (r ¼ 0.83, P<.001; r ¼ 0.83, P<.001; and r ¼ 0.83, P<.001, respectively). Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
VOL. - NO. - / - 2016
5
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY
FIGURE 4 1.0
Spearman correlation coefficient (r)
AMH Gen II 0.8
Elecsys AMH
Access AMH
P<0.005 P<0.03
NS NS
NS
NS
0.6
NS NS
NS
0.4
0.2
0
Low AFC (3-12 follicles)
Intermediate AFC (13-20 follicles)
High AFC (21-84 follicles)
Bar charts showing a comparison of Spearman correlation coefficients (AFC AMH), given by Fisher r-to-z transformation, in the low, intermediate, and high AFC groups. As shown, only in the low AFC group, the relationship between AFC and serum AMH levels obtained with automated assays were significantly stronger (P<.03 and P<.005) than that obtained with the manual assay. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
Our results revealed that serum AMH levels obtained by conventional or new automated assays were remarkably correlated with AFC (r ¼ 0.83). This confirms the overall adequate reliability of Access AMH and Elecsys AMH when confronted to the reference (i.e., AFC), and it expands initial data suggested by other investigators (5, 6). Furthermore, in women displaying reduced follicle counts, we noticed that the new automated AMH assays were more strictly associated to AFC than AMH Gen II assay. Because it is the population of women with reduced follicle count that constitutes, by definition, one of the main targets for accurate ovarian reserve assessment, the improved performance of automated measurements is, at least, reassuring. Yet the mechanisms underlying such a better relationship in this particular group of patients remain to be clarified. It is possible that some improved analytical features of the new automated AMH assays are at stake to explain these differences, in particular the twofold lower limit of quantification and maximum imprecision of automated AMH assays as compared with the manual AMH Gen II assay. This ameliorated performance of new automated AMH assays in low AMH concentrations may be decisive in clinical situations associated with weak AMH production for patient counseling and therapeutic orientation. Another point deserving to be emphasized is that in the present investigation we observed that, overall, circulating AMH levels obtained with Access AMH and Elecsys AMH were approximately 16% and 20% lower than with manual AMH Gen II. These data confirm and expand, in a centralized, single setting series of AMH measurements, those from previous authors showing that differences range from 9% to 22% with Access AMH (6, 7) and 12% and 28% with Elecsys AMH (3, 6, 10). Whether these lower AMH concentrations with automated AMH assays are due 6
to differences in calibration, assay phamarcokynetics, and/ or linearity (7) remains to be established. In addition, we remarked that serum AMH levels obtained with Access AMH were roughly 5% higher than those obtained with Elecsys AMH. Both these between-automated-assay differences as well as the automated–manual assay differences in AMH values should be taken into account when clinically interpreting AMH measurements. In conclusion, the present study indicated that AMH concentrations provided by the two commercially available automated AMH assays and by the manual assay tested are both concordant to each other and reliably related to the number of AMH-producing follicles. Further, clinicians should be aware that considerable differences exist in absolute AMH values provided by the three assays tested (16% and 20% as low with Access AMH and Elecsys AMH as with the manual assay) because they can influence data interpretation and therapeutic choices. Finally, it is remarkable that both automated AMH assays were significantly more strongly correlated with AFC than the manual assay exclusively in patients endowed with few antral follicles. These results should be also corroborated in additional studies focused only in IVF-ET patients using the number of oocytes retrieved (19) instead of AFC as reference. Taken together, the data reported in this article highlight a considerable advantage of the new AMH assays because patients with few follicles are one of the main targets for accurate ovarian function assessments.
REFERENCES 1.
Fanchin R, Schon€auer LM, Righini C, Guibourdenche J, Frydman R, Taieb J. Serum anti-M€ ullerian hormone is more strongly related to ovarian follicular status than serum inhibin B, estradiol, FSH and LH on day 3. Hum Reprod 2003;18:323–7. VOL. - NO. - / - 2016
Fertility and Sterility® 2.
3. 4.
5.
6.
7.
8.
9.
10.
11.
Dewailly D, Andersen CY, Balen A, Broekmans F, Dilaver N, Fanchin R, et al. The physiology and clinical utility of anti-m€ ullerian hormone in women. Hum Reprod Update 2014;20:370–85. Gassner D, Jung R. First fully automated immunoassay for anti-M€ ullerian hormone. Clin Chem Lab Med 2014;52:1143–52. Pearson K, Long M, Prasad G, Wu YY, Bonifacio M. Assessment of the Access AMH assay as an automated, high-performance replacement for the AMH generation II manual ELISA. Reprod Biol Endocrinol 2016;14:8. Anderson RA, Anckaert E, Bosch E, Dewailly D, Dunlop C, Fehr D, et al. Pro€llerian horspective study into the value of the automated Elecsys antimu mone assay for the assessment of the ovarian growing follicle pool. Fertil Steril 2015;103:1074–80. Van Helden J, Weiskirchen R. Performance of the two new fully automated €llerian hormone immunoassays compared with the clinical standard anti-Mu assay. Hum Reprod 2015;8:1918–26. Nelson SM, Pastuszek E, Kloss G, Malinowska I, Liss J, Lukaszuk A, et al. Two new automated, compared with two enzyme-linked immunosorbent, anti€llerian hormone assays. Fertil Steril 2015;104:1016–21. mu De Laveleye M, Gruson D. Anti-M€ ullerian hormone testing: evaluation of a novel method allowing more automation. Scand J Clin Lab Invest 2015;75: 681–5. Hyldgaard J, Bor P, Ingerslev HJ, Torring N. Comparison of two different methods for measuring anti-mullerian hormone in a clinical series. Reprod Biol Endocrinol 2015;13:107. € Anckaert E, Oktem M, Thies A, Cohen-Bacrie M, Daan NM, Schiettecatte J, et al. Multicenter analytical performance evaluation of a fully automated anti-M€ ullerian hormone assay and reference interval determination. Clin Biochem 2016;49:260–7. Pigny P, Gorisse E, Ghulam A, Robin G, Catteau-Jonard S, Duhamel A, et al. Comparative assessment of five serum antim€ ullerian hormone assays for the diagnosis of polycystic ovary syndrome. Fertil Steril 2016;105:1063–9.
VOL. - NO. - / - 2016
12.
13.
14.
15.
16.
17.
18.
19.
Weenen C, Laven JS, Von Bergh AR, Cranfield M, Groome NP, Visser JA, et al. Anti-M€ ullerian hormone expression pattern in the human ovary: potential implications for initial and cyclic follicle recruitment. Mol Hum Reprod 2004;10:77–83. Bentzen JG, Forman JL, Johannsen TH, Pinborg A, Larsen EC, Anderson AN. €llerian hormone during normal Ovarian antral follicle subclasses and anti-mu reproductive aging. J Clin Endocrinol Metab 2013;98:1602–11. Depmann M, van Disseldorp J, Broer SL, Eijkemans MJ, Laven JS, Visser JA, et al. Fluctuations in anti-M€ ullerian hormone levels throughout the menstrual cycle parallel fluctuations in the antral follicle count: a cohort study. Acta Obstet Gynecol Scand 2016;95:820–8. de Vet A, Laven JS, de Jong FH, Themmen AP, Fauser BC. Antim€ ullerian hormone serum levels: a putative marker for ovarian aging. Fertil Steril 2002;77: 357–62. Pigny P, Merlen E, Robert Y, Cortet-Rudelli C, Decanter C, Jonard S, et al. Elevated serum level of anti-m€ ullerian hormone in patients with polycystic ovary syndrome: relationship to the ovarian follicle excess and to the follicular arrest. J Clin Endocrinol Metab 2003;98:5957–62. Fanchin R, Louafi N, Mendez Lozano DH, Frydman N, Frydman R, Taieb J. Per-follicle measurements indicate that anti-m€ ullerian hormone secretion is modulated by the extent of follicular development and luteinization and may reflect qualitatively the ovarian follicular status. Fertil Steril 2005;84: 167–73. Nelson SM, Klein BM, Arce JC. Comparison of antim€ ullerian hormone levels and antral follicle count as predictor of ovarian response to controlled ovarian stimulation in good-prognosis patients at individual fertility clinics in two multicenter trials. Fertil Steril 2015;103:923–30. La Marca A, Giulini S, Tirelli A, Bertucci E, Marsella T, Xella S, et al. Anti€llerian hormone measurement on any day of the menstrual cycle strongly Mu predicts ovarian response in assisted reproductive technology. Hum Reprod 2007;22:766–71.
7
New automated antim€ ullerian hormone assays are more reliable than the manual assay in patients with reduced antral follicle count Teddy Tadros, M.D.,a,b Bruno Tarasconi, M.D.,a,b,c Jean Nassar, M.D.,a,b Jean-Luc Benhaim, Pharm.D.,d €lle Taieb, Pharm.D.,e and Renato Fanchin, M.D., Ph.D.a,b Joe ^ pital Foch, Suresnes, France; b University of Paris-Ouest, Suresnes, France; Center of Reproductive Medicine, Ho Coordenação de Aperfeiçoamento de Pessoal de Nìvel Superior (CAPES) Foundation, Ministry of Education of Brazil, dicales Benhaim, Clamart, France; and e Assistance Publique - Hôpitaux de Brasilia, Brazil; d Laboratoire d’Analyses Me ^ pital Antoine Be cle re, Clamart, France Paris (AP - HP), Department of Biochemistry, Ho a c
Objective: To compare the strength of the relationship between antral follicle count (AFC) and serum antim€ ullerian hormone (AMH) concentrations obtained with two automated and one manual AMH assays in three different AFC populations. Design: Prospective cohort study. Setting: University-affiliated IVF-ET center. Patient(s): Frozen–thawed serum samples of 211 assisted conception candidates, aged 24–43 years. Intervention(s): Serum AMH was measured using one manual (AMH Gen II) and two fully automated (Access AMH and Elecsys AMH) assays. Antral follicle count was performed under strictly standardized conditions and sorted into three groups according to tercile values: low AFC (3–12 follicles; n ¼ 73), intermediate AFC (13–20 follicles; n ¼ 65), and high AFC (21–84 follicles; n ¼ 73). Main Outcome Measure(s): Strength of correlation between AMH levels and AFC. Result(s): Overall, AMH levels were lower with Access AMH (16%) and Elecsys AMH (20%) than with AMH Gen II. Remarkably, the strength of correlations between AFC and circulating AMH levels was the same with the three assays (r ¼ 0.83). Yet in the low AFC group, serum AMH levels obtained by Access AMH and Elecsys AMH showed a stronger correlation with AFC (r ¼ 0.63 and r ¼ 0.65, respectively) than the AMH Gen II (r ¼ 0.52), a phenomenon that was not observed in the remaining AFC groups. Conclusion(s): As compared with conventional AMH Gen II assay results, [1] serum AMH concentrations were 16% and 20% lower with Access AMH and Elecsys AMH, respectively; and [2] automated assays were more strongly correlated to AFC in the subset of patients with reduced follicle count. (Fertil SterilÒ 2016;-:-–-. Ó2016 by American Society for Reproductive Medicine.) Key Words: Antim€ ullerian hormone, antral follicle count, ovarian reserve, poor responders Discuss: You can discuss this article with its authors and with other ASRM members at
A
ssociated or not to ultrasonographic counting of antral follicles, serum antim€ ullerian hormone (AMH) measurements have become the reference in the clinical appraisal of the ovarian follicular status (1), and its clinical use has been continuously gaining momentum during the last 12 years. Antim€ ullerian hormone levels reflect the activity of granulosa
cells of small antral follicles, thereby providing patients and physicians with invaluable information on ovarian aging and how to individualize controlled ovarian hyperstimulation protocols (2). Yet from a practical standpoint, in addition to cost, problems related to reliability and consistency of serum AMH measurements have raised doubts about their clinical soundness. These
Received May 25, 2016; revised August 16, 2016; accepted August 22, 2016. T.T. has nothing to disclose. B.T. has nothing to disclose. J.N. has nothing to disclose. J.-L.B. has nothing to disclose. J.T. has nothing to disclose. R.F. has nothing to disclose. ^ pital Foch, 40 rue Reprint requests: Renato Fanchin, M.D., Ph.D., Center of Reproductive Medicine, Ho Worth, 92151 Suresnes, France (E-mail: [email protected]). Fertility and Sterility® Vol. -, No. -, - 2016 0015-0282/$36.00 Copyright ©2016 American Society for Reproductive Medicine, Published by Elsevier Inc. http://dx.doi.org/10.1016/j.fertnstert.2016.08.045 VOL. - NO. - / - 2016
problems essentially are attributed, on the one hand, to an uncoordinated development of AMH assays, which displayed different callibration and standards and, on the other hand, to the compulsory operatorand technique-dependent manipulations. Recently, to overcome these limitations and to improve quality of AMH measurements, fully automated AMH assays (Access AMH [Beckman Coulter] and Elecsys AMH [Roche Diagnostics International]) have been developed and commercialized (3, 4). Yet analytical data available have been limited to the confirmation of an adequate concordance of serum AMH levels obtained by these assays (4–11).
1
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY Unfortunately, only limited and/or indirect data taking the necessary relationship between AMH and the number of AMH-producing antral follicles as a reference have been hitherto published (5–7). Because it is undoubtedly the small antral follicles that produce most of the circulating AMH concentrations (12–14), antral follicle count (AFC), when it is performed under optimal conditions, should be taken as the standard for comparing reliability of AMH assays. Indeed, serum AMH levels have been strongly correlated to AFC by numerous investigators (1, 13, 15, 16). Moreover, the relative performance of new automated and manual AMH assays in subgroups of patients displaying low, intermediate, and high AFC remains undetermined. Therefore, this insufficient knowledge spurred us to investigate the reliability and concordance of AMH levels obtained by one manual and two automated AMH assays in different AFC populations.
MATERIALS AND METHODS Subjects and Procedures We used frozen–thawed serum aliquots obtained from 211 women who were willing to enter our assisted conception program between April 2015 and July 2015. Inclusion criteria were [1] both ovaries present without morphologic abnormalities (menstruating patients with or without polycystic ovary syndrome or polycystic ovary morphology were included); [2] optimum ovarian visualization at transvaginal ultrasound scans; and [3] body mass index ranging between 18 and 25 kg/m2. All blood samples were collected between days 1 and 5 of the menstrual cycle and remained frozen from 2 to 4 months before thawing. Serum AMH determinations were performed using three different ELISAs according to manufacturer protocols by a single operator (J.-L.B.): modified AMH Gen II (Beckman Coulter), Access AMH (Beckman Coulter), and Elecsys AMH (Roche Diagnostics International). Detailed AMH assay procedures have been previously described (6). In brief, limits of detection of the three AMH assays tested were, respectively, 0.08, 0.02, and 0.01 ng/mL; limits of quantification, 0.16, 0.08, and 0.03 ng/mL; and maximum imprecision, 8.0%, 4.3%, and 3.5%. Moreover, serum E2, FSH, and LH levels were determined by an automated multianalysis system using a chemiluminescence technique (Cobas e411 Analyzer, Roche Diagnostics, Mannheim, Germany). For E2, limit of quantification was 15 pg/mL, and maximum imprecision was 8%. For FSH and LH, limit of quantification was 0.1 mIU/mL, and maximum imprecision was 3%. In parallel to blood samplings, AFC (antral follicles measuring 3–10 mm in diameter) was carefully and exhaustively determined using a 5–9-MHz transvaginal ultrasound probe (RIC 5-9-D; Voluson E8 Expert, General Electric Medical Systems, Paris, France) by a single operator (T.T.) who was unaware of AMH results. Given that the present investigation was limited exclusively to hormonal measurements in frozen–thawed serum aliquots and ultrasound scan records and that patients had previously given their informed consent for additional hormonal analysis using their stored sera, our 2
local institutional review board advised us that it did not require ethics committee submission.
Definition of AFC Groups Participants were arbitrarily sorted into three different AFC groups according to the 33rd and 66th centiles of AFC distribution (12 and 20 antral follicles): low AFC (3–12 follicles; n ¼ 73), intermediate AFC (13–20 follicles; n ¼ 65), and high AFC (21–84 follicles; n ¼ 73). These cutoffs were used as an effort to pragmatically identify patients with different ovarian follicular phenotypes.
Statistics Because data distribution was considered nonparametric, we elected to use the median as the measure of central tendency and minimum–maximum as the measure of variability. Circulating AMH levels were compared two by two among the three different assays tested by using the Wilcoxon signed-rank test. The strength of relationships between AFC and serum AMH levels provided by the three different assays was assessed by the Spearman correlation test in the total population and in the three AFC groups. Comparison of Spearman correlation coefficients was performed using the Fisher r-to-z transformation, a statistical procedure that can be applied to assess the significance of the difference between two correlation coefficients found in two independent samples. In addition, agreement between the different AMH assays was assessed graphically using the Bland-Altman plots and by Passing-Bablok regression. The Cusum test for linearity was used to test the applicability of Passing-Bablok regression. A P value of < .05 was considered statistically significant.
RESULTS Population Characteristics Characteristics of individuals included in the present analysis corresponded, as expected, to the profile of women who are usually candidates for assisted conception at our center, encompassing young and aged women and those with small and large counts of antral follicles. Median age (range) was 35 (24–43) years. In the early follicular phase, median serum E2 level was 45 (15–249) pg/mL, and serum FSH and LH levels were 6.9 (3.0–21.6) and 5.1 (1.1–21.1) mIU/mL, respectively. Overall median antral follicle count was 15 (3–84) follicles.
AMH Levels according to Manual and Automated Assays Serum AMH levels measured by the three different ELISA assays are depicted in Figure 1. As shown, each one of the tested assays resulted into statistically different AMH levels. Hence, AMH Gen II provided AMH levels at 1.97 (0.04-30.66) ng/mL [median (interquartile range)] that were higher (P< .001) than those obtained with Access AMH at 1.66 (0.04–30.46) ng/mL and with Elecsys AMH at 1.58 (0.04–26.17) ng/mL. In other words, AMH Gen II levels are, on average, 16% and 20% as high as Access AMH and Elecsys AMH, respectively. In addition, it is noteworthy that statistically different results were observed between VOL. - NO. - / - 2016
Fertility and Sterility®
FIGURE 1
Distribution of serum AMH levels, in logarithmic scale, measured by AMH Gen II, Access AMH, and Elecsys AMH assays in box-and-whisker plots. Central lines are the median, proximal parallel lines are the lower and upper quartiles, distal parallel lines are the 10th and the 90th centiles, and dots are actual AMH levels that were lower or higher than the 10th and the 90th centiles. Note that serum AMH levels measured by AMH Gen II were roughly 16% and 20% higher than those obtained with Access AMH and Elecsys AMH, respectively. In addition, serum AMH levels obtained with Access AMH were approximately 5% higher than those obtained with Elecsys AMH. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
the two automated assays (Access AMH and Elecsys AMH), with Access AMH being 5% as high as Elecsys AMH.
Assay Agreement Agreement between AMH measurement techniques assessed in Bland-Altman plots and by Passing-Bablok regression is illustrated in Figure 2. Overall, we observed adequate concordance between techniques. As we can note in the BlandAltman plots, the number of measurements situated outside the 2-SD interval was 4.7% between AMH Gen II and Access AMH, 4.3% between AMH Gen II and Elecsys AMH, and 2.8% between Access and Elecsys AMH, which may be considered acceptable. Yet it is noteworthy that agreement between assays tended to become progressively weaker as mean AMH levels obtained with the three techniques increased. Accordingly, the number of measurements situated outside the 2-SD interval tended to slightly increase from the low AFC (4.1%, 2.7%, and 5.4%, respectively) to the intermediate AFC (4.6%, 6.1%, and 3.0%, respectively) and the high AFC (6.8%, 6.8%, and 5.4%, respectively) groups. Finally, Passing-Bablok regression showed an adequate relationship between each pair of assays, and the Cusum test for linearity failed to show significant differences between AMH Gen II vs. Access AMH (P¼ .49), AMH Gen II vs. Elecsys AMH (P¼ .72), and Access vs. Elecsys AMH (P¼ .72), either in the whole data set or within each AFC group.
Reliability of the Three Different AMH Assays Correlations between AFC and circulating AMH levels obtained by AMH Gen II, Access AMH, and Elecsys AMH are shown in Figure 3. We could note that, overall, Spearman correlation coefficients were remarkably similar among the three assays (r ¼ 0.83, P< .001; r ¼ 0.83, P< .001; and r ¼ 0.83, P< .001, respectively). Because of this, comparison of Spearman correlation coefficients was not required for this analysis. VOL. - NO. - / - 2016
Spearman correlation coefficients, a representation of the strength of the relationship between AFC and serum AMH levels obtained with the three different assays in the three AFC groups, are depicted in Figure 4. As shown, in contrast to the overall population, in the low AFC group, correlation coefficients obtained with Access AMH (r ¼ 0.63) and Elecsys AMH (r ¼ 0.65) were significantly stronger (P< .03 and P< .005, respectively) than that obtained with AMH Gen II assay (r ¼ 0.52). We failed to show statistically significant differences of correlation coefficients in the intermediate and high AFC groups. This indicates an improved reliability of automated over manual AMH assays exclusively in patients presenting %12 antral follicles.
DISCUSSION The present investigation was designed to evaluate both the reliability and concordance of results of two new automated AMH assays (Access AMH and Elecsys AMH) as compared with the manual AMH assay (AMH Gen II) in a group of randomly selected assisted conception candidates presenting different ovarian phenotypes. For this, our study required a number of methodologic characteristics, such as the inclusion of a large series of patients that allowed the determination of three AFC groups, centralized hormonal measurements performed in frozen–thawed serum aliquots that have been stored for a short time interval (2–4 months), and careful and exhaustive determination of AFC as reference for hormone measurements. The choice of using AFC as a reference was based on the fact that ultrasound-detected antral follicles probably are the paramount source of circulating AMH (12–14). Indeed, studies focusing in the staining of AMH gene in granulosa cells of follicles at different development stages (12) have suggested that small antral follicles contribute to serum AMH levels the most and that preantral follicle contribution is only marginal. Further, the pharmacologic induction of growth of most of 3
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY
FIGURE 2
Agreement between AMH measurement techniques assessed in Bland-Altman plot (A–C; solid lines are the mean of differences between assays, and dotted lines are the 95% confidence interval) and by Passing-Bablok regression (D–F). As illustrated in A–C, the number of measurements situated outside the 2-SD interval was 4.7% between AMH Gen II and Access AMH, 4.3% between AMH Gen II and Elecsys AMH, and 2.8% between Access and Elecsys AMH, thereby confirming adequate concordance between techniques. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
available small antral follicles during controlled ovarian hyperstimulation resulted in a dramatic decline (not sustained by preantral follicle production) of circulating AMH levels (17). Therefore, the choice of AFC as a 4
reference is valid as long as AFC is performed under optimal, rigorous, and standardized conditions, which is not unfortunately the case in some studies that elected to standardize exclusively AMH measurements (18). VOL. - NO. - / - 2016
Fertility and Sterility®
FIGURE 3
Simple regression analysis of the relationship between AFC and serum AMH levels obtained by the AMH Gen II (A), Access AMH (B), and Elecsys AMH (C). Spearman correlation coefficients were remarkably similar among the three assays, (r ¼ 0.83, P<.001; r ¼ 0.83, P<.001; and r ¼ 0.83, P<.001, respectively). Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
VOL. - NO. - / - 2016
5
ORIGINAL ARTICLE: REPRODUCTIVE ENDOCRINOLOGY
FIGURE 4 1.0
Spearman correlation coefficient (r)
AMH Gen II 0.8
Elecsys AMH
Access AMH
P<0.005 P<0.03
NS NS
NS
NS
0.6
NS NS
NS
0.4
0.2
0
Low AFC (3-12 follicles)
Intermediate AFC (13-20 follicles)
High AFC (21-84 follicles)
Bar charts showing a comparison of Spearman correlation coefficients (AFC AMH), given by Fisher r-to-z transformation, in the low, intermediate, and high AFC groups. As shown, only in the low AFC group, the relationship between AFC and serum AMH levels obtained with automated assays were significantly stronger (P<.03 and P<.005) than that obtained with the manual assay. Tadros. Reliability of automated AMH assays. Fertil Steril 2016.
Our results revealed that serum AMH levels obtained by conventional or new automated assays were remarkably correlated with AFC (r ¼ 0.83). This confirms the overall adequate reliability of Access AMH and Elecsys AMH when confronted to the reference (i.e., AFC), and it expands initial data suggested by other investigators (5, 6). Furthermore, in women displaying reduced follicle counts, we noticed that the new automated AMH assays were more strictly associated to AFC than AMH Gen II assay. Because it is the population of women with reduced follicle count that constitutes, by definition, one of the main targets for accurate ovarian reserve assessment, the improved performance of automated measurements is, at least, reassuring. Yet the mechanisms underlying such a better relationship in this particular group of patients remain to be clarified. It is possible that some improved analytical features of the new automated AMH assays are at stake to explain these differences, in particular the twofold lower limit of quantification and maximum imprecision of automated AMH assays as compared with the manual AMH Gen II assay. This ameliorated performance of new automated AMH assays in low AMH concentrations may be decisive in clinical situations associated with weak AMH production for patient counseling and therapeutic orientation. Another point deserving to be emphasized is that in the present investigation we observed that, overall, circulating AMH levels obtained with Access AMH and Elecsys AMH were approximately 16% and 20% lower than with manual AMH Gen II. These data confirm and expand, in a centralized, single setting series of AMH measurements, those from previous authors showing that differences range from 9% to 22% with Access AMH (6, 7) and 12% and 28% with Elecsys AMH (3, 6, 10). Whether these lower AMH concentrations with automated AMH assays are due 6
to differences in calibration, assay phamarcokynetics, and/ or linearity (7) remains to be established. In addition, we remarked that serum AMH levels obtained with Access AMH were roughly 5% higher than those obtained with Elecsys AMH. Both these between-automated-assay differences as well as the automated–manual assay differences in AMH values should be taken into account when clinically interpreting AMH measurements. In conclusion, the present study indicated that AMH concentrations provided by the two commercially available automated AMH assays and by the manual assay tested are both concordant to each other and reliably related to the number of AMH-producing follicles. Further, clinicians should be aware that considerable differences exist in absolute AMH values provided by the three assays tested (16% and 20% as low with Access AMH and Elecsys AMH as with the manual assay) because they can influence data interpretation and therapeutic choices. Finally, it is remarkable that both automated AMH assays were significantly more strongly correlated with AFC than the manual assay exclusively in patients endowed with few antral follicles. These results should be also corroborated in additional studies focused only in IVF-ET patients using the number of oocytes retrieved (19) instead of AFC as reference. Taken together, the data reported in this article highlight a considerable advantage of the new AMH assays because patients with few follicles are one of the main targets for accurate ovarian function assessments.
REFERENCES 1.
Fanchin R, Schon€auer LM, Righini C, Guibourdenche J, Frydman R, Taieb J. Serum anti-M€ ullerian hormone is more strongly related to ovarian follicular status than serum inhibin B, estradiol, FSH and LH on day 3. Hum Reprod 2003;18:323–7. VOL. - NO. - / - 2016
Fertility and Sterility® 2.
3. 4.
5.
6.
7.
8.
9.
10.
11.
Dewailly D, Andersen CY, Balen A, Broekmans F, Dilaver N, Fanchin R, et al. The physiology and clinical utility of anti-m€ ullerian hormone in women. Hum Reprod Update 2014;20:370–85. Gassner D, Jung R. First fully automated immunoassay for anti-M€ ullerian hormone. Clin Chem Lab Med 2014;52:1143–52. Pearson K, Long M, Prasad G, Wu YY, Bonifacio M. Assessment of the Access AMH assay as an automated, high-performance replacement for the AMH generation II manual ELISA. Reprod Biol Endocrinol 2016;14:8. Anderson RA, Anckaert E, Bosch E, Dewailly D, Dunlop C, Fehr D, et al. Pro€llerian horspective study into the value of the automated Elecsys antimu mone assay for the assessment of the ovarian growing follicle pool. Fertil Steril 2015;103:1074–80. Van Helden J, Weiskirchen R. Performance of the two new fully automated €llerian hormone immunoassays compared with the clinical standard anti-Mu assay. Hum Reprod 2015;8:1918–26. Nelson SM, Pastuszek E, Kloss G, Malinowska I, Liss J, Lukaszuk A, et al. Two new automated, compared with two enzyme-linked immunosorbent, anti€llerian hormone assays. Fertil Steril 2015;104:1016–21. mu De Laveleye M, Gruson D. Anti-M€ ullerian hormone testing: evaluation of a novel method allowing more automation. Scand J Clin Lab Invest 2015;75: 681–5. Hyldgaard J, Bor P, Ingerslev HJ, Torring N. Comparison of two different methods for measuring anti-mullerian hormone in a clinical series. Reprod Biol Endocrinol 2015;13:107. € Anckaert E, Oktem M, Thies A, Cohen-Bacrie M, Daan NM, Schiettecatte J, et al. Multicenter analytical performance evaluation of a fully automated anti-M€ ullerian hormone assay and reference interval determination. Clin Biochem 2016;49:260–7. Pigny P, Gorisse E, Ghulam A, Robin G, Catteau-Jonard S, Duhamel A, et al. Comparative assessment of five serum antim€ ullerian hormone assays for the diagnosis of polycystic ovary syndrome. Fertil Steril 2016;105:1063–9.
VOL. - NO. - / - 2016
12.
13.
14.
15.
16.
17.
18.
19.
Weenen C, Laven JS, Von Bergh AR, Cranfield M, Groome NP, Visser JA, et al. Anti-M€ ullerian hormone expression pattern in the human ovary: potential implications for initial and cyclic follicle recruitment. Mol Hum Reprod 2004;10:77–83. Bentzen JG, Forman JL, Johannsen TH, Pinborg A, Larsen EC, Anderson AN. €llerian hormone during normal Ovarian antral follicle subclasses and anti-mu reproductive aging. J Clin Endocrinol Metab 2013;98:1602–11. Depmann M, van Disseldorp J, Broer SL, Eijkemans MJ, Laven JS, Visser JA, et al. Fluctuations in anti-M€ ullerian hormone levels throughout the menstrual cycle parallel fluctuations in the antral follicle count: a cohort study. Acta Obstet Gynecol Scand 2016;95:820–8. de Vet A, Laven JS, de Jong FH, Themmen AP, Fauser BC. Antim€ ullerian hormone serum levels: a putative marker for ovarian aging. Fertil Steril 2002;77: 357–62. Pigny P, Merlen E, Robert Y, Cortet-Rudelli C, Decanter C, Jonard S, et al. Elevated serum level of anti-m€ ullerian hormone in patients with polycystic ovary syndrome: relationship to the ovarian follicle excess and to the follicular arrest. J Clin Endocrinol Metab 2003;98:5957–62. Fanchin R, Louafi N, Mendez Lozano DH, Frydman N, Frydman R, Taieb J. Per-follicle measurements indicate that anti-m€ ullerian hormone secretion is modulated by the extent of follicular development and luteinization and may reflect qualitatively the ovarian follicular status. Fertil Steril 2005;84: 167–73. Nelson SM, Klein BM, Arce JC. Comparison of antim€ ullerian hormone levels and antral follicle count as predictor of ovarian response to controlled ovarian stimulation in good-prognosis patients at individual fertility clinics in two multicenter trials. Fertil Steril 2015;103:923–30. La Marca A, Giulini S, Tirelli A, Bertucci E, Marsella T, Xella S, et al. Anti€llerian hormone measurement on any day of the menstrual cycle strongly Mu predicts ovarian response in assisted reproductive technology. Hum Reprod 2007;22:766–71.
7