A single-centre evaluation of two new anti-müllerian hormone assays and comparison with the current clinical standard assay

Transcription

1 Human Reproduction, Vol.29, No.5 pp , 2014 Advanced Access publication on February 26, 2014 doi: /humrep/deu036 ORIGINAL ARTICLE Reproductive endocrinology A single-centre evaluation of two new anti-müllerian hormone assays and comparison with the current clinical standard assay Paul Welsh 1, *, Karen Smith 2, and Scott M. Nelson 3 1 Institute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UK 2 Clinical Biochemistry, MacEwan Building, Glasgow Royal Infirmary, Glasgow, UK 3 Maternal and Reproductive Medicine, School of Medicine, University of Glasgow, Glasgow, UK *Correspondence address. BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Place, Glasgow G12 8TA. Tel: ; paul.welsh@glasgow.ac.uk Submitted on October 11, 2013; resubmitted on January 24, 2014; accepted on February 3, 2014 study question: Are the new Ansh Labs Ultra-Sensitive anti-müllerian hormone (AMH) and picoamh ELISA assays suitable for clinical use and is the Ultra-Sensitive assay comparable to the Beckman Coulter AMH Gen II assay? summaryanswer: The Ultra-Sensitive assay appears to have different calibration to the Gen II assay, but has performance characteristics generally suitable for clinical use. what is known already: The Gen II assay is the most commonly used AMH assay in routine biochemistry at present, but persistent calibration/interference problems have been reported. study design, size, duration: Serum from patients referred for AMH measurement was assayed (in duplicate) using the Gen II assay in Glasgow Royal Infirmary between January and February We randomly selected 193 stored serum samples to re-run (in duplicate) using the Ultra-Sensitive AMH Ansh Labs assay, blinded to the original result. Samples that returned low results were run on the picoamh Ansh Labs assay. Performance characteristics and linearity of the new assays were also assessed. participants/materials, setting, methods: All serum samples from patients referred for AMH at Glasgow Royal Infirmary between January and February 2013 were eligible for inclusion. Investigators were blinded to any identifiable information regarding the patients, including sex, age and reason for AMH measurement. main results and the role of chance: Intra-assay and inter-assay coefficients of variation of the Ultra-Sensitive and picoamh assays were 6.0 and 10.7%, respectively, over a range of concentrations. The assays had mean linearity of 98 and 97% over the dilution range of 1:2 1:16 and 1:2 1:8, respectively. The limit of detection of the ultrasensitive assay was calculated to be 0.34 pmol/l. For 166 samples which provided a quantitative result on the Gen II and Ultra-Sensitive Ansh Labs assays, the median (interquartile range) was 12.2 ( ) pmol/l and 20.0 ( ) pmol/l, respectively (P,0.0001). The Passing Bablok regression equation (in pmol/l) was y (Ultra-Sensitive) ¼ Gen II. More samples were below the clinical cut-off of 5.4 pmol/l using the Gen II assay (a difference between paired proportions of 15.0%, P, 0.001). Fifteen of the 22 undetectable samples yielded a measurable concentration result on the picoamh assay. limitations, reasons for caution: The present study is a pragmatic assessment of the new assay under ideal conditions. Lot-tolot variation could not be assessed. Demographics and outcomes of patients referred for AMH measurement were not known. wider implications of the findings: The Ansh Labs Ultra-Sensitive assay performance characteristics are similar to the Gen II assay and may be suitable for clinical and epidemiological use. Enhanced sensitivity of the Ansh Labs picoamh assay enables measurement of low AMH concentrations. These results re-emphasize the need for an AMH international standard. study funding/competing interest(s): Ansh Labs provided kits for this study free of charge. The manufacturer played no part in conducting assays or data analysis. S.M.N. has received speaker s fees and participated in advisory boards for Beckman Coulter, Merck Serono, MSD and Ferring regarding AMH. P.W. is supported by British Heart Foundation fellowship FS/12/62/ We declare no other financial relationship or competing interests. & The Author Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please journals.permissions@oup.com

2 1036 Welsh et al. Key words: AMH / anti-mullerian hormone / validation studies / immunoassay / ovarian reserve Introduction Anti-Müllerian hormone (AMH) declines throughout adult female reproductive life mirroring the decay in the ovarian reserve (Nelson et al., 2011). A developing clinical indication for AMH assay is the prediction of ovarian response to gonadotrophin stimulation in IVF: the recent National Institute for Health and Clinical Excellence (NICE) guideline setting thresholds of 5.4 and 25.0 pmol/l for a low and high response, respectively (NICE, 2013). As such AMH levels can guide the decision as to whether to proceed with IVF therapy and to guide treatment during hormone dosing to reduce the incidence of ovarian hyperstimulation, and recent data suggest that this approach is costeffective (Yates et al., 2011). AMH is thus a potentially important assay in managing treatment, and assay performance in terms of reproducibility and calibration is critical in enabling adherence to guidelines. Unfortunately, no international standard for AMH is available as yet, which is problematic for assessing assay performance over time. Recently published work has validated a Beckman Coulter Gen II assay for AMH (Kumar et al., 2010; Wallace et al., 2011). This assay was designed to harmonize assays by diagnostic systems laboratory (DSL) and immunotech laboratory (IOT) ELISAs, both of which were acquired by Beckman Coulter. When the Gen II assay was released it was standardized to the IOT assay, but it was also strongly correlated with the DSL assay, although there was a 40% increase in AMH measurements compared with DSL (Kumar et al., 2010; Wallace et al., 2011). Since then however there has been a release of field safety notices from Beckman Coulter, stating that all kit lots ( ) are affected by interference from complement, causing a time-dependent 70% reduction in AMH measurement (MHRA, 2013). This has caused considerable clinical confusion, and has shaken confidence in the assay (Clark et al., 2014 ). Ansh Labs have developed two new AMH ELISA assays for clinical and research use: an ultra-sensitive AMH ELISA for standard use (approximate range 1 62 pmol/l) and a picoamh assay for use in samples with lower AMH (approximate range pmol/l). A potential advantage of the Ansh Labs platform is the use of characterized recombinant human AMH protein in assay calibrators (Kumar et al., 2013) and linear epitope antibodies which may allow for stable assay performance over time. The aim of this study was to validate the performance of the Ultra-Sensitive Ansh Labs assays and the picoamh to assess manufacturer claims that the picoamh assay has a lower detection limit and to compare the results of the Ultra-Sensitive Ansh Labs assay with the Beckman Coulter Gen II assay. To achieve this, we used a pragmatic design, utilizing samples from patients referred for clinical AMH measurement. As a post hoc analysis we compared the results from the Gen II lot (suspected to be susceptible to complement interference by the manufacturer) with Gen II lot Materials and Methods Patient serum and measurement of AMH The BeckmanCoulter (High Wycombe, UK) Gen II (lot ) assay was used to measure AMH in 350 serum samples from clinical patients at the Department of Biochemistry at Glasgow Royal Infirmary between January and February In general,.90% of AMH assays conducted at Glasgow Royal Infirmary are for females referred for potential IVF treatment. Serum samples were initially separated and routinely stored at 2808C until assay for clinical reporting. These Beckman Coulter assays were conducted in duplicate using a robotic platform (Grifols Triturus, Hampshire, UK) using undiluted serum as per the manufacturer s instructions (first thaw). Samples were then re-stored at 2808C. In April 2013, 193 of these patient samples were randomly selected for measurement, in duplicate, in the present study using the Ansh Labs (Webster, Tx, USA) Ultra-Sensitive AMH ELISA (lot ) (second thaw). Samples that yielded low results were also subsequently re-assayed using the Ansh Labs picoamh assay (lot A) (third thaw). Selected samples were also re-run (third thaw) on another Beckman Gen II assay (lot ) following release of the field safety notice to test the impact of complement interference on our original data. These samples were run with a pre-dilution protocol, according to manufacturer instructions. The non-clinical study assays were performed by a single experienced technician who was blinded to the Gen II result. The local ethics committee advised the study did not require formal ethical approval on the basis of it constituting a potential service development, and it utilized samples where we were blinded to any donor information. Assay performance and quality control For both the Ultra-Sensitive AMH and picoamh assays, two levels of manufacturer quality control (C1 and C2) were run in duplicate on each plate. In addition, two patient samples were run as a quality control in duplicate on every plate and were also run six or more times on a single plate to yield an estimate of intra-assay variation. The limit of blank (LoB) and limit of detection (LoD) was tested broadly according to EP17 guidelines (Armbruster and Pry, 2008) on the Ultra-Sensitive assay, using duplicates from the 21 lowest detectable patient samples. LoB and LoD were not tested in the picoamh assay due to only a small number of kits being available to re-assay low AMH samples. Regression techniques were used to estimate a limit of quantification (LoQ) in both Ultra-Sensitive and picoamh assays, defined as the AMH where the expected sample coefficients of variation (CV) is 20%. Linearity was checked using different patient samples, diluting samples with pooled serum from patients with an undetectable AMH. For linearity checks on the Ultra-Sensitive assays, three samples were tested in the range of pmol/l at dilutions of 1:2 1:16. On the picoamh assay, three samples were tested in the range of pmol/l at dilutions of 1:2 1:8 for the picoamh assay). Statistical analyses Where appropriate, a conversion of 1 ng/ml ¼ 7.14 pmol/l was used. For quantitative analysis of continuous data, data were restricted to samples giving a result more than the LoD. For comparisons using categorical cut-offs ( 5.4 and 25.0 pmol/l) difference between paired proportions was assessed using the McNemar test, and Cohen s Kappa statistic of interrater agreement calculated (Carletta, 1992). Comparisons between two methods were assessed using Lin s concordance correlation coefficient (using log transformed data to satisfy assumptions of normal distribution) and Bland Altman plots using Pitman s test to assess likelihood of bias. The 95% limits of agreement in the Bland Altman plot assume no bias between the two methods and are included here only for visualization purposes. Passing Bablok regression was used to describe the relationship between two methods using data (Passing and Bablok, 1983). Unlike least squares linear regression, Passing Bablok regression does not assume freedom of variables from error, nor does it make any assumptions about

3 Anti-Müllerian hormone assay evaluation 1037 the distribution of the data and hence it may be less influenced by outliers. All statistics were performed using Stata v11.2. Results Ansh Lab Ultra-Sensitive AMH and picoamh assays performance and quality control Reproducibility of both the assays was acceptable; CVs of manufacturer and in-house quality control material were 6% for intra-assay variation and 10.7% for inter-assay variation (Table I). For the Ultra-Sensitive AMH assay, the mean recovery was 96% at 1:2 dilution, 112% at 1:4, 94% at 1:8 and 89% at 1:16 (overall mean 98%; Fig. 1a). The LoB was calculated to be 0.13 pmol/l, while the LoD was 0.34 pmol/l (manufacturer reported LoD of 0.16 pmol/l). Our data did not reveal an LoQ at 20% CV for the Ultra-Sensitive assay; linear regression of sample concentration against duplicate CV (for samples in the range of,7.0 pmol/l) suggested no evidence of an inverse relationship of sample concentration with CVs. The mean CV of duplicates of the 21 lowest non-zero samples (range pmol/l) was 6.3%. For the picoamh assay recovery was 88% at 1:2 dilution, 103% at 1:4 and 101% at 1:8 (overall mean 97%; Fig. 1b). Again, for this assay no LoQ was evident; linear regression of sample concentration against duplicate CVs suggested no evidence of an inverse relationship. The mean CV of duplicates of 18 samples in the range of,2.5 pmol/l was 3.3%. Ultra-Sensitive AMH versus Beckman Gen II method comparison Samples from 22 patients returned AMH results which were below the LoD (19 of which were below the LoD on both assays). A further five samples had results over the detectable range. After excluding these samples for analysis of continuous data, median [interquartile range (IQR)] AMH was 12.2 pmol/l in the Gen II assay ( pmol/l) and 20.0 pmol/l ( pmol/l) in the Ansh Labs assay (P, ). The concordance between log-transformed values was rho ¼ 0.78 (95% CI ). The Passing Bablok regression equation was (Fig. 2): y (Ultra- Sensitive assay in pmol/l) = Gen II (95% CI for slope ). Bland Altman analysis showed evidence of bias in absolute measurements (Fig. 3), such that samples with higher AMH concentration had higher estimates on the Ultra-Sensitive Ansh Labs assay. The correlation between the difference and the mean was r ¼ 0.61, P, Of the 193 patients results, 51 had results in both assays which were below the 5.4 pmol/l clinical cut-off, 30 patients had results which fell below the cut-off using the Gen II but not in the Ultra-Sensitive assay and one sample was below the cut-off only on the Ultra-Sensitive assay (a difference between paired proportions of 15.0%, P, 0.001; kappa 0.65, P, 0.001). Using the cut-off of 25 pmol/l, 54 had a result above this level on both assays, three patients were above this level on the Gen II assay only and 21 on the Ultra-Sensitive assay only (a difference in paired proportions of 9.3%, P, 0.001; kappa ¼ 0.73, P, 0.001). Ultra-Sensitive AMH versus picoamh method comparison All samples which yielded a value,5.4 pmol/l on the Ultra-Sensitive assay were re-run using the picoamh kit. Of the 22 samples which yielded no detectable result previously, seven yielded no detectable result on the picoamh assay, but 15 did yield a value (median 0.22 pmol/l, IQR pmol/l). Comparing samples which yielded a value on both assays, the median value was 3.1 pmol/l (IQR ) on the Ultra-Sensitive assay and 2.2 pmol/l (IQR ) on the picoamh assay. The concordance between log-transformed measurements was moderately strong (rho ¼ 0.85; 95% CI ). Table I Performance of the Ansh Labs assay for AMH: CV. Sample Manufacturer value (pmol/l) Intra-assay variation... Inter-assay variation... n Mean (pmol/l) CV (%) n Mean (pmol/l) CV (%)... Ultra-Sensitive AMH Ansh Labs C points plates Ansh Labs C points plates Internal control 1 n/a 22 points (1 plate) plates Internal control 2 n/a 22 points (1 plate) plates PicoAMH Ansh Labs C points plates * Ansh Labs C points plates * Internal control 1 n/a 6 points (1 plate) plates * Internal control 2 n/a 8 points (1 plate) plates * *Due to the low number of replicates, these values arereported for illustrativepurposes only, although theyare broadlyin agreement with thatgiven by the manufacturer. CV, coefficient of variation; C1/2, manufacturer s quality control 1/2.

4 1038 Welsh et al. Figure 1 Linearity of expected versus observed AMH recovery from sequential dilutions in (a) the Ultra-Sensitive assay and (b) the picoamh assay. original results from Beckman Gen II lot , 37 samples selected at random were re-run on a new Beckman Gen II kit (lot ) following manufacturers resolution of this issue. Two samples returned values less than the LoD on both plates. Of the other 35 samples, the median (IQR) AMH results were 7.6 pmol/l ( pmol/l) on lot and 7.7 ( ) on lot (Fig. 4). The concordance was strong (rho ¼ 0.95; 95% CI ). The Passing Bablok regression yielded a line: Beckman Lot = Beckman lot (95% CI of slope ). Using Gen II lot as the referent method, data generated from comparison with the Ultra-Sensitive Ansh Labs samples were not substantially different from data generated using Gen II lot as the referent method (data not shown). Discussion Figure 2 Passing Bablok regression of Ansh Labs Ultra-Sensitive assay on Beckman Gen II assay. Comparison of Beckman Gen II before and after field safety notice Following our experiments, Beckman Coulter released a field safety notice regarding AMH Gen II kit lots including lot due to apparent complement interference in serum samples causing falsely low results. This may have partly explained the lower results obtained on the Gen II assay than on the Ultra-Sensitive Ansh Labs assay. To validate our This study demonstrates that the new Ansh Labs assay Ultra-Sensitive assay has performance characteristics similar to the Beckman Gen II assay, which renders it broadly suitable for clinical and research use. The Ultra-Sensitive assay is broadly comparable to the Gen II assay in terms of relative rank of results, but there were absolute differences, such that the Ansh Labs assay gave results which were 40% higher. This is similar to the change that was observed when the Gen II assay was compared with the original DSL ELISA (Wallace et al., 2011). This difference in measured AMH concentration results in a considerable change in the proportions of samples that would be categorized as having low levels of AMH according to clinical guidelines for fertility (26.9% on the Ansh Labs assay and 42.0% on the Beckman Gen II assay) (NICE, 2013). Furthermore, we demonstrate that the picoamh

5 Anti-Müllerian hormone assay evaluation 1039 Figure 3 Bland Altman plot relating differences in the Ultra-Sensitive and Gen II assays. The bold blue line denotes linear data trend. Average bias is denoted by the mean difference line. The red lines denote 95% limits of agreement assuming no bias. Figure 4 Passing Bablok regression of Beckman Gen II lot (post-reported complement interference issue) on lot (during reported complement interference issue). assay does have a lower limit of sensitivity than either the Gen II or Ultra-Sensitive assay, and may therefore be a useful assay in epidemiological studies or clinical situations where expected values of AMH are low. The NICE guidelines suggest clear cut-offs for the measurement of AMH to give an indication of ovarian reserve, and these guidelines are likely to impact treatment decisions. Therefore, accurate measurement of AMH across time and between labs is important for developing a firm evidence base to establish treatment cut-offs. Recent trial evidence has suggested AMH is a cost-effective method of directing IVF treatment (Yates et al., 2011). It is worthwhile noting that other groups have found that the Beckman Gen II assay has not performed over time as predicted by the kit manufacturer: one group reported that results were 20% lower rather than 40% higher originally predicted in comparison with the DSL assay (Rustamov et al., 2012). Our results, the first we are aware of to be published on this topic, show that the effect of apparent complement interference recently reported in field safety notices (MHRA, 2013) is not substantial in our hands. This may be at least partly due to batching and storage of samples for AMH measurement in Glasgow Royal Infirmary biochemistry department, rather than immediate measurement. There are ongoing cross-lab efforts to pool data and more widely investigate this issue in the Gen II kits. In the context of ongoing issues with the Gen II kits it is difficult to definitely assess the clinical consequences of the apparent differences in calibration between the Beckman and Ansh Labs kits. It is possible that changes in the performance of the Beckman Gen II assay over time at least partially explain the bias we observe in our Bland Altman plot; the Gen II assay may now be reporting results which are more closely calibrated with the original DSL ELISA (Wallace et al., 2011). Thus, the apparent overestimation of AMH by the Ansh Labs Ultra-Sensitive assay may actually be underestimation from the Gen II assay, as observed in other data (Nelson et al., 2014). The Ansh Labs assays are standardized to a wellcharacterized (by HPLC and mass spectrometry) recombinant human AMH protein. This is a theoretical advantage of the assay, which should enable long-term robust calibration. To definitely assess assay

6 1040 Welsh et al. performance over time and lot-to-lot reliability for all manufacturers, an AMH international standard is urgently required. Of the 22 samples which had an undetectable AMH, 19 patients had an undetectable AMH result on both Ansh Labs Ultra-Sensitive and Beckman Gen II assays, hence we find no strong evidence that one assay has a better analytical sensitivity than the other. However, our study also validates the Ansh Labs picoamh assay for clinical and research use. There are no current clinical indications for measurement of AMH in the range of,5 pmol/l, although availability of the assay for research use will allow more detailed clinical research investigations into the implications of lower ovarian reserve (Nelson, 2013). Evolving areas of clinical and research interest for AMH as a biomarker include assessment of the ovarian reserve following chemotherapy or radiotherapy treatments (Henry et al., 2014), fertility in polycystic ovarian syndrome (Homburg et al., 2013), the perimenopausal transition and transgenerational fertility trends (Bentzen et al., 2013), and the process of sexual maturation from childhood (Hagen et al.,2012). Strengths and limitations of the study require consideration. This study represents a pragmatic comparison of AMH measurement methods using an unselected group of patients referred for clinical AMH measurement for any reason. The fact we could not identify demographic or other data regarding our patients reduces the risk of selection bias in designing the study, but means we cannot conduct subgroup analysis and we cannot be certain as to whom our results are generalizable to (although.90% of AMH assays conducted at Glasgow Royal Infirmary are for females referred for potential IVF treatment). We also did not have any data on outcomes, which means the association of AMH with outcomes for both assays cannot be compared. In the absence of a gold standard assay or international standard it is impossible to definitively state that one assay is better than another, or indeed which is calibrated more accurately. To add to the present study and confirm clinical utility, more data are required investigating the use of the Ansh Labs assays in wellcharacterized prospective cohort studies with well-defined outcomes. As we suggest, recent clinical problems with the Beckman Gen II assay suggest the need for tighter regulation of this increasingly important biochemistry assay using international standards and immunoassays that perform consistently in the long term. Plans for a National External Quality Assessment Service scheme in the UK are being developed to achieve this. Lot-to-lot variation could not be assessed, and we could not run Beckman standards on the AMH kit, as the bovine AMH they utilize would not be detected by the human AMH-specific Ansh Labs antibodies. A further limitation of our study is that we were unable to specify an LoQ based on a 20% CV. This is likely due a lack of data measuring low AMH sample on repeat plates (i.e. a lack of sufficient inter-assay variation data). However, the tentative calculated LoD calculated here is broadly in agreement with that given by the manufacturer. In conclusion, the present results suggest that the Ansh Labs Ultra- Sensitive AMH assay may be broadly comparable to the Beckman Gen II assay in terms of technical performance for clinical or epidemiological use. Enhanced sensitivity of the Ansh Labs picoamh assay also enables assessment of low AMH concentrations in epidemiological and, potentially, emerging clinical applications. However, these data clearly re-emphasize the need to develop an international standard for AMH given the important and evolving role of AMH in clinical management of IVF. Acknowledgements We thank Elaine Butler for technical assistance. Authors roles P.W. involved in study conception, design, data analysis, interpretation, drafting the manuscript. K.S. and S.M.N. involved in study conception, design data interpretation and critical revision of the manuscript. Funding Ansh Labs provided kits for this study free of charge. The manufacturer played no part in conducting assays or data analysis. P.W. is supported by British Heart Foundation fellowship FS/12/62/ Conflict of interest S.M.N. has received speaker s fees and participated in advisory boards for Beckman Coulter, Merck Serono, MSD and Ferring regarding AMH. We declare no other financial relationship or competing interests. References Armbruster DA, Pry T. Limit of blank, limit of detection and limit of quantitation. Clin Biochem Rev 2008;29(Suppl. 1):S49 s52. Bentzen JG, Forman JL, Larsen EC, Pinborg A, Johannsen TH, Schmidt L, Friis-Hansen L, Nyboe Andersen A. Maternal menopause as a predictor of anti-mullerian hormone level and antral follicle count in daughters during reproductive age. Hum Reprod 2013;28: Carletta J. Assessing agreement on classification tasks: the kappa statistic. Comput Linguist 1992;22: Clark CA, Laskin CA, Cadesky K. Anti-Mullerian hormone: reality check. Hum Reprod 2014;29: Hagen CP, Aksglaede L, Sorensen K, Mouritsen A, Andersson AM, Petersen JH, Main KM, Juul A. Individual serum levels of anti-mullerian hormone in healthy girls persist through childhood and adolescence: a longitudinal cohort study. Hum Reprod 2012;27: Henry NL, Xia R, Schott AF, McConnell D, Banerjee M, Hayes DF. Prediction of post-chemotherapy ovarian function using markers of ovarian reserve. Oncologist 2014;19: Homburg R, Ray A, Bhide P, Gudi A, Shah A, Timms P, Grayson K. The relationship of serum anti-mullerian hormone with polycystic ovarian morphology and polycystic ovary syndrome: a prospective cohort study. Hum Reprod 2013;28: Kumar A, Kalra B, Patel A, McDavid L, Roudebush WE. Development of a second generation anti-mullerian hormone (AMH) ELISA. J Immunol Methods 2010;362: Kumar A, Kalra B, Patel AS, Shah S. Development of a well characterized ultra-sensitive human anti-müllerian hormone (AMH) ELISA AACC 2013 A Documents/AACC_13_AM_A279-A363.pdf. MHRA (Medicines and Healthcare products Regulatory Agency). Urgent Field Safety Notice FSN AMH Gen II ELISA (REF A79765), con pdf. Nelson SM. Biomarkers of ovarian response: current and future applications. Ovarian Reserve Oocytes 2013;99:

7 Anti-Müllerian hormone assay evaluation 1041 Nelson SM, Messow MC, Wallace AM, Fleming R, McConnachie A. Nomogram for the decline in serum antimullerian hormone: a population study of 9,601 infertility patients. Fertil Steril 2011;95: Nelson SM, Iliodromiti S, Fleming R, Anderson R, McConnachie A, Messow CM. Reference range for the antimüllerian hormone Generation II assay: a population study of 10,984 women, with comparison to the established Diagnostics Systems Laboratory nomogram. Fertil Steril 2014; 101: NICE (National Institute for Health and Clinical Excellence). Fertility: Assessment and treatment for people with fertility problems. NICE Clinical Guideline 156, /62769/62769.pdf. Passing H, Bablok W. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I. J Clin Chem Clin Biochem 1983; 21: Rustamov O, Smith A, Roberts SA, Yates AP, Fitzgerald C, Krishnan M, Nardo LG, Pemberton PW. Anti-Mullerian hormone: poor assay reproducibility in a large cohort of subjects suggests sample instability. Hum Reprod 2012;27: Wallace AM, Faye SA, Fleming R, Nelson SM. A multicentre evaluation of the new Beckman Coulter anti-mullerian hormone immunoassay (AMH Gen II). Ann Clin Biochem 2011;48: Yates AP, Rustamov O, Roberts SA, Lim HY, Pemberton PW, Smith A, Nardo LG. Anti-Mullerian hormone-tailored stimulation protocols improve outcomes whilst mreducing adverse effects and costs of IVF. Hum Reprod. 2011;26: