Two different immunoassays produced highly discordant results when used to measure anti–Müllerian hormone in feline blood serum and urine

Abstract

Anti–Müllerian hormone (AMH) is a useful biomarker for a variety of veterinary conditions relating to the gonads. For female mammals, these include spayed or intact status, ovarian remnant syndrome, granulosa cell tumor, and ovarian response to gonadotropin stimulation for assisted reproductive technologies. We compared 2 different AMH immunoassays that produced markedly discordant results, although the original aim of our research was to refine an earlier study that intended to determine whether AMH concentrations in feline blood serum and urine are correlated. The previous study reported measurable AMH concentrations in all 27 urine samples tested, which were not correlated with the corresponding serum concentrations. Our studies differ in that we used the AMH ELISA (assay A; AL-116, Ansh Labs) currently in use in our diagnostic laboratory, which differs from the immunoassay (assay B; E0078Ca, BT Lab) used in the original study. In contrast to assay B, assay A detected no AMH in urine collected from 19 cats immediately before ovariohysterectomy. We re-tested these same urine samples using assay B, and all had measurable AMH. However, a negative serum sample that is routinely run in assay A for quality control purposes also had measurable AMH in assay B. A second run of assay B found measurable AMH concentrations in 20 serum samples that had previously tested below the detection limit of assay A. Assay B also failed the parallelism validation test. Our results indicate that assay B is not valid for feline or canine AMH testing.

The utility of anti–Müllerian hormone (AMH) testing in small mammal veterinary medicine was first reported in 2011, ²² and its clinical applications have steadily grown over the last 14 y.^14,20 Because the gonads of both sexes are the sole source of AMH, the detection or absence of AMH in blood serum effectively distinguishes between intact and gonadectomized animals.^{3,22,23,25,27} In animals that were meant to have undergone complete gonadectomy (spay or castration), AMH testing has become the gold standard for diagnosing ovarian remnant syndrome (ORS), incomplete orchiectomy, and cryptorchidism.^{1,10,19,21,22,27} In some patients, an unusually high serum AMH concentration raises the suspicion that the gonadal tissue harbors a granulosa or Sertoli cell tumor (GCT or SCT).^2,4,10,15,26 When a GCT or SCT has been removed surgically, AMH testing is an excellent means of monitoring for residual tumor, dissemination, and recurrence. AMH testing has also been evaluated in felids in the field of assisted reproductive technologies (ARTs), specifically those ARTs for which ovarian stimulation via exogenous gonadotropin administration is used.^5,6 For all of the aforementioned applications of AMH testing, the analysis of this hormone has been performed on blood serum or plasma, which requires a needle stick. Non- or less-invasive methods to perform AMH testing remain to be identified.

Urine is the only liquid sample type, other than serum or plasma, for which AMH testing has been performed in cats and dogs.^17,18 The ability to accurately detect AMH in urine would have utility in veterinary medicine because it would enable noninvasive sample collection from fractious animals and possibly from captive exotic felids and canids that regularly spray or void in well-defined areas within their enclosures.^8,9 For domestic animals that are being evaluated for the presence or absence of gonads or remnant or cryptic gonadal tissue, qualitative results (positive or negative) would suffice. For these sorts of cases, absolute concentrations of AMH in urine wouldn’t necessarily need to be correlated with the concentrations in serum or plasma, other than to be detectable or non-detectable in the presence or absence of gonads or gonadal tissue, respectively. Conversely, for the diagnosis of GCTs, SCTs, and in advance of exogenous gonadotropin administration as part of ART regimens, urinary AMH results would need to be quantitative and to be correlated with AMH concentrations in serum or plasma. Although AMH was detected in urine samples from 27 intact cats and 42 intact dogs,^17,18 in neither species were urine and serum AMH concentrations correlated. However, these publications did not indicate the concentration of each urine sample at the time of sample collection.

Our original objective was to refine a prior study design ¹⁸ to determine whether AMH concentrations in feline serum and urine are correlated when urine AMH concentrations are expressed relative to creatinine (Cr) or specific gravity (SG). We elected to use the canine/feline AMH ELISA (assay A; AL-116, Ansh Labs) that is used for clinical testing in the Cornell University diagnostic endocrinology laboratory (Ithaca, NY, USA). Upon discovering that urine samples from 18 of 19 intact female cats had AMH concentrations below the limit of detection (LOD) of assay A, we re-ran the same 19 serum and urine samples in the canine AMH ELISA (assay B; E0078Ca, BT Lab) that had been used in the prior studies.^17,18 Whereas assay A has been validated for AMH testing of feline and canine serum to determine the presence or absence of gonads or gonadal tissue,^7,16 we found no validation information published to date for assay B.^17,18 The only validation information listed in the manufacturer’s instructions for assay B is that for the intra- and inter-assay CVs (<8% and <10%, respectively). Therefore, we attempted to validate assay B. Here we report our comparison of AMH concentrations in feline serum and urine following a side-by-side evaluation of these 2 AMH ELISAs.

Materials and methods

Animals

We enrolled 19 adult female domestic cats that were presented to the spay and neuter clinic at the Cornell University Small Animal Community Practice (Ithaca, NY, USA). All activities, save for the collection of a blood sample from each cat, were performed as standard-of-care for a patient receiving ovariohysterectomy, and therefore, were exempt from oversight by the Cornell University Institutional Animal Care and Use Committee (IACUC). Blood sample collection was covered by IACUC protocol 2007-0146. Per routine, following anesthetic induction, urine was expressed from the bladder during patient surgical preparation for ovariohysterectomy and collected into plain red-top collection tubes. During this same period, a 3-mL blood sample was collected into a plain red-top collection tube, and the blood and urine specimens were brought to the nearby diagnostic endocrinology laboratory. The blood samples were allowed to clot at room temperature for 30 min, centrifuged at 2,100 × g for 5 min, aliquoted into 2 equal-volume serum samples, and stored frozen at −20°C until analyzed. The urine samples were centrifuged at 2,100 × g for 5 min to clear them of sediment, aliquoted into 2 equal-volume samples, and stored frozen at −20°C until analyzed, which is comparable to the sample processing performed in the prior study. ¹⁸

AMH analysis by assay A of serum and urine from intact cats

Assay A is a quantitative 3-step sandwich ELISA with a reported analytical sensitivity for AMH of 0.11 pmol/L (0.015 ng/mL) and a LOD that varies by lot of 0.58–0.71 pmol/L (0.04–0.10 ng/mL). Within a single ELISA plate, 19 paired serum and urine samples were analyzed for AMH using assay A according to the manufacturer’s instructions and the laboratory’s standard QC practices that are applied to clinical AMH testing. The QC practices comprise the inclusion of assay A positive and negative controls and the laboratory pooled serum sample positive and negative controls, which, at the time of this assay run, were pooled canine serum samples that had previously produced unequivocally positive or negative AMH results (≥1.44 or <0.71 pmol/L [≥0.20 ng/mL or <0.10 ng/mL], respectively). The laboratory’s internal QC pools had been aliquoted and stored frozen at −20°C, and an aliquot was thawed individually for every run of assay A. The QC criteria for the approval of each assay include negative controls with AMH concentrations that are less than the established cutoff for negative results (<0.71 pmol/L [<0.10 ng/mL]), the manufacturer’s positive control having an AMH concentration within the lot-specific range, and the laboratory pooled positive control having an AMH concentration within 2 SDs of the mean established by ≥20 runs of assay A.

Urine SG was determined by refractometry (JorVet) using the residual volume after loading the ELISA plate. The SG for each of the 19 samples was greater than the refractometer’s range (1.050), and the urine samples were diluted 2-fold with milli-Q water and re-analyzed to yield SGs that were within the range of the instrument. The second urine aliquot was submitted to the clinical pathology laboratory for Cr analysis, which was performed by an enzymatic method on a clinical analyzer (Cobas 501c; Roche).

AMH analysis by assay B of serum and urine from intact cats

Assay B is a quantitative 3-step sandwich ELISA with a reported sensitivity for AMH of 2.27 pmol/L (0.32 ng/mL); the LOD was not reported by the manufacturer. After discovering that 18 of the 19 urine samples produced negative AMH results in assay A, we analyzed the samples using assay B according to the manufacturer’s instructions. For the first of 2 runs of assay B, we included negative QC samples that we routinely analyze in our clinical AMH testing, which is by assay A. One negative control is provided by the manufacturer, and the other is an internal control that is composed of pooled canine serum samples, each of which previously had unequivocally negative AMH results when run in assay A. Because both the external and internal negative controls yielded unequivocally positive results in assay B [i.e., well above the reported sensitivity of the assay and well within the range of the standard curve (3.55–2,130 pmol/L [0.5–300 ng/mL])], we ran a second assay B ELISA plate from the same lot using a mix of AMH-positive and -negative serum samples from cats and dogs to compare the results between assays A and B.

AMH analysis by assay B of clinical serum samples from cats and dogs

After discovering that assay B produced unequivocally positive AMH results for assay A negative control and our laboratory internal negative control, we selected a mix of recently tested, archived clinical samples from cats and dogs to run in assay B. These serum samples had produced unequivocally positive or negative AMH results in assay A (5 positive and 17 negative cats; 8 positive and 3 negative dogs). Three AMH-negative samples (2 feline, 1 canine) were selected specifically because the sample submission forms indicated that the animals were showing no signs of heat and were undergoing AMH testing simply to confirm their spayed status before being adopted out from a shelter. Within this same assay B, we performed a serial dilution of a feline serum sample that had an AMH concentration of 85.0 pmol/L (11.9 ng/mL) in assay A. This sample underwent four 2-fold serial dilutions using a standard assay buffer that is routinely used in our diagnostic endocrinology laboratory. This commercial assay buffer is 0.05 M phosphosaline, pH 7.4, 0.025 M EDTA, 0.08% sodium azide, and 1% RIA-grade bovine serum albumin (MilliporeSigma). However, we note that assay B instructions for use state, “Sample can’t be diluted with this kit. Owing to the material we use to prepare the kit, the sample matrix interference may falsely depress the specificity and accuracy of the assay.”

Data analysis

Owing to the markedly discordant AMH results between assays A and B, we limited the statistical analyses to Pearson-product correlations for serum samples that had detectable AMH concentrations in both assays and for the urine Cr vs SG data. Additionally, we performed a Passing–Bablok regression of the serum AMH concentrations from assays A and B for the 19 study subjects (JMP Pro, v.17.0.0; SAS Institute). All other data presentations are descriptive.

Results

AMH results of serum and urine from intact cats for assay A

Assay A met all of the laboratory’s QC criteria and would have been approved if it had been run for clinical testing. The serum AMH concentrations for the 19 intact cats when analyzed in assay A had a range of 11.8–72.9 pmol/L (1.65–10.2 ng/mL), and thus, all of them would have been reported as positive (≥1.43 pmol/L [≥0.20 ng/mL]) if they had been clinical samples. Conversely, the urine AMH concentrations for 18 of 19 cats were <0.71 pmol/L (<0.10 ng/mL), and thus, 94.7% would have been reported as negative. One urine sample had an AMH concentration of 1.07 pmol/L (0.15 ng/mL), which would have been reported as inconclusive. Note: the quantitative ranges for assay A qualitative results have been established for serum, but not for urine. However, to qualify as a positive result, presumably the urine AMH concentration would at the very least need to be greater than the assay LOD (0.58 pmol/L [0.04 ng/mL]).

We found that urine Cr concentration and SG were significantly correlated (r = 0.58; p = 0.01), but the values were not useful given that urine AMH concentrations were essentially undetectable in assay A.

AMH results of serum and urine from intact cats for assay B

Serum AMH concentrations for the 19 intact cats when analyzed in assay B had a range of 70.5–246 pmol/L (9.86–34.5 ng/mL), and assay B AMH concentrations were not significantly correlated with those from assay A (r = 0.24; p = 0.33), nor was the Passing–Bablok fit of the regression statistically significant (τ = 0.11, p = 0.51; [AMH]•B = 20.99 − 0.74 × [AMH]•A). Moreover, the assay A negative control and our laboratory internal negative control (pooled canine serum) also had measurable AMH concentrations in assay B of 95.1 and 288 pmol/L (13.3 and 40.4 ng/mL), respectively. Urine AMH concentrations for the 19 intact cats analyzed in assay B had a range of 87.4–224 pmol/L (12.2–31.4 ng/mL), which is in contrast to assay A.

AMH results of clinical serum samples from cats and dogs for assay B

The most striking discordance between assays A and B is that 20 clinical serum samples (17 cats, 3 dogs) that were unequivocally negative in assay A had measurable AMH concentrations according to assay B, with a range of 72.0–306 pmol/L (10.1–42.9 ng/mL; Table 1). According to the second run of assay B, the AMH concentrations for the canine and feline negative controls were 139 and 125 pmol/L (19.4 and 17.4 ng/mL), respectively (Table 1). The 5 feline and 8 canine clinical serum samples that had been reported as positive by assay A (3.06–121 pmol/L [0.43–16.9 ng/mL]) had an AMH range of 78.5–284 pmol/L (11.0–39.8 ng/mL) in assay B (Table 1). One feline sample with an inconclusive AMH result of 1.27 pmol/L (0.18 ng/mL) in assay A had an AMH concentration of 108 pmol/L (15.1 ng/mL) in assay B (Table 1).

Table 1.

Results of anti–Müllerian hormone (AMH) testing by assay B (E0078Ca; BT Lab) using internal controls and clinical serum samples that had been analyzed in assay A (AL-116; Ansh Labs) and reported to clients.

Species	Sample ID	Assay A result	Assay A, pmol/L	Assay B, pmol/L
	Neg control	Neg	<0.71	138
Canine	Pos control	Pos	12.0	167
Canine	Neg control	Neg	<0.71	124
Feline	Pos control	Pos	14.9	99.5
Feline	Pos sample, serial dilution
	Neat	Pos	85.0	142
	1:2	ND	ND	194
	1:4	ND	ND	131
	1:8	ND	ND	98.5
	1:16	ND	ND	80.4
Feline	1	Neg	<0.71	92.8
Feline*	2	Neg	<0.71	72.0
Feline*	3	Neg	<0.71	155
Canine*	4	Neg	<0.71	306
Canine	5	Neg	<0.71	125
Canine	6	Neg	<0.71	125
Feline	7	Neg	<0.71	144
Feline	8	Neg	<0.71	160
Feline	9	Neg	<0.71	98.2
Feline	10	Neg	<0.71	120
Feline	11	Neg	<0.71	152
Feline	12	Neg	<0.71	98.8
Feline	13	Neg	<0.71	165
Feline	14	Neg	<0.71	138
Canine	15	Neg	<0.71	144
Feline	16	Neg	<0.71	219
Feline	17	Neg	<0.71	139
Feline	18	Neg	<0.71	204
Feline	19	Neg	<0.71	157
Feline	20	Neg	<0.71	81.8
Feline	21	Inc	1.27	107
Feline	22	Pos	2.62	136
Canine	23	Pos	3.06	157
Feline	24	Pos	3.71	107
Canine	25	Pos	5.90	126
Canine	26	Pos	6.32	78.5
Canine	27	Pos	13.9	141
Feline	28	Pos	14.8	124
Feline	29	Pos	37.7	140
Canine	30	Pos	53.9	100
Canine	31	Pos	89.4	114
Feline	32	Pos	120	284
Canine	33	Pos	120	113

Inc = inconclusive; ND = not done; Neg = negative; Pos = positive. Both qualitative and quantitative (pmol/L) results from assay A are listed. One AMH-positive feline sample ( in italics) with an AMH concentration of 85.0 pmol/L in assay A was serially diluted in a buffer solution and run in assay B.

^*Denotes samples from animals for which AMH analysis was performed in assay A to confirm spayed status prior to adoption, and the submission forms stated that no signs of heat had been observed.

The serial dilution of a feline serum sample with an AMH concentration of 85.0 pmol/L (11.9 ng/mL) by assay A was not parallel to the assay B standard curve. The AMH concentration of this same undiluted sample by assay B was 142 pmol/L (19.9 ng/mL); however, the result of the first 2-fold dilution was 194 pmol/L (27.2 ng/mL), and, for the 4-fold dilution, the result was 131 pmol/L (18.3 ng/mL; Table 1). It was only after this run of assay B that we became aware that the buffer used to perform the serial dilution contained sodium azide (0.08%), which can have an inhibitory effect on horseradish peroxidase (HRP) and can lead to a reduction in signal for ELISAs that use HRP. ²⁴ However, the presence of sodium azide in the dilution buffer cannot explain the serial dilution results for assay B. This is because the AMH concentration following the first 2-fold dilution increased by 37%, and if the sodium azide had an impact on the outcome, the concentration would have decreased by more than 2-fold.

Discussion

We originally undertook our project to refine the experimental design of a published study ¹⁸ that reported serum and urine AMH concentrations from female domestic cats are measurable in both sample types, but the serum and urine values were not correlated. Our intended refinement was to account for how dilute or concentrated each feline urine was at the time of sample collection by measuring urine Cr concentration and SG. However, this refinement lost relevance when we found that measurable concentrations of AMH were not detected by assay A in 18 of 19 urine samples from intact female cats. These results are in stark contrast to the outcomes with assay B, for which the prior study ¹⁸ reported that all 27 feline urine samples had measurable AMH concentrations. One possible explanation for the discrepancy between assays A and B is that AMH is excreted in the urine as a metabolite(s) that cross-react(s) with the antibodies in assay B, but not the antibodies in assay A. This possibility led to the acquisition of 2 assay B ELISA kits that matched the catalog number (E0078Ca) used in the prior study, ¹⁸ although the lot numbers are different (Kaya S, pers. comm., 2025 Jan 16).

Whereas we found that measurable concentrations of AMH were detected by assay B in both serum and urine samples from 19 of 19 cats, we also noted that the internal and external (assay A) negative controls had measurable AMH concentrations. Recognizing that our internal negative control was a pooled canine serum sample and that the matrix of the assay A negative control might be incompatible with assay B, we performed a second run of assay B that included 17 feline and 3 canine serum samples that had previously tested unequivocally negative in assay A, which had been validated.^7,16 Each of these 20 serum samples was falsely positive for AMH in assay B, and if these results had been reported to the submitting veterinarians, the incorrect interpretations would have been that the patients are gonadally intact or have retained gonadal tissue. This could potentially lead to unnecessary exploratory surgeries. In cases of ORS or cryptorchidism, the gonadal tissue can be difficult to find and surgically remove, ¹¹ which adds to the complexity and duration of surgeries. For these reasons, it is imperative for AMH assays to have a rate of false positivity that is as close to zero as possible. Whereas assay A has performed very well in this regard, assay B failed to pass this critical test.

We recognize that we have run only 2 assay B kits of the same lot, but as of this writing, we have not received evidence from the manufacturer that indicates this particular AMH ELISA can distinguish between the presence and absence of gonads or gonadal tissue in either cats or dogs. It is not possible to determine whether the discordance between our results and those of the prior studies^17,18 can be attributed to different lots of assay B, because in the prior studies^17,18 serum samples from gonadectomized animals were not included. Rather, those studies^17,18 only analyzed samples collected from gonadally intact animals, which would have rightly been expected to produce positive AMH results. Our first run of assay B also included serum and urine samples from gonadally intact cats, and if not for the inclusion of the external and internal negative controls, we might have assumed that assay B had worked as intended, and that assay A’s inability to detect AMH in urine was due to a lack of cross-reactivity with the AMH metabolite(s) excreted in feline urine.

Our study reiterates the importance of ensuring the validity of assays that are used in veterinary medicine, regardless of whether their use is for clinical or purely research purposes. Until the manufacturer of assay B or some other third-party user demonstrates that this AMH ELISA can reliably and accurately distinguish between gonadally intact animals and those for which gonads or gonadal tissues are completely absent, we recommend against its use for veterinary testing. We are not aware of any currently available AMH assays that can reliably detect AMH in feline or canine urine. However, AMH is a member of the transforming growth factor–beta superfamily, and other members of this superfamily (e.g., tumor necrosis factor, inhibin A), have been detected and quantified in human urine.^12,13 Therefore, the ability to noninvasively detect/measure AMH and/or its metabolite(s) in felids and canids through the analysis of their urine remains a possibility.