Abstract
Background: Measurement of antimüllerian hormone (AMH) is used to assess ovarian reserve. Circulating levels of AMH correlate with antral follicle count, with relatively high levels indicating an ample reserve of primary and preantral follicles in the ovary. AMH levels are stable with dilution and freezer storage, and are not altered by hemolysis or menstrual cycle day in young women of reproductive age. We sought to examine whether glucose challenge or food intake modifies AMH levels compared with fasting.
Methods: Residual plasma samples were available from 54 pregnant women under fasting conditions and then 1, 2, and 3 h after ingestion of a 100-g glucose challenge. These samples were collected as part of routine clinical care to identify gestational diabetes (GDM) at 24–28 weeks of gestation. Twelve of these women met criteria for GDM based on an increased glucose level at a minimum of 2 time points. A second set consisted of serum samples collected from 8 nonpregnant women at fasting and 1 h after a meal. Levels of AMH were measured using an ultrasensitive assay (Ansh Labs, Webster, TX). A 2-way ANOVA (sample timing and GDM status) or matched t-test was performed. AMH measurements were subject to a logarithmic transformation before analysis.
Results: Median AMH levels in pregnant women at 1, 2, or 3 h after glucose challenge did not differ compared with AMH levels at fasting or by diagnosis of GDM. Similarly, there was no difference in median AMH levels in nonpregnant women of reproductive age at fasting and after a meal.
Conclusion: AMH levels are not altered by glucose or food intake.
IMPACT STATEMENT
Antimüllerian hormone (AMH) levels are measured to assess ovarian reserve. Within-woman levels of AMH are stable regardless of fasting or glucose or food intake, so random sample collection is suitable. These data contribute to our knowledge regarding the potential effects of preanalytical variables on AMH measurement.
Measurement of antimüllerian hormone (AMH)4 is used to assess ovarian reserve in women of reproductive age (1). Circulating levels of AMH are correlated with antral follicle count, with relatively high levels indicating an ample reserve of primary and preantral follicles in the ovary. Very high levels of AMH are observed in women with polycystic ovary syndrome (PCOS), a condition of menstrual irregularity, hyperandrogenemia, and/or polycystic ovaries, often accompanied by marked insulin resistance (2). As women approach menopause, AMH levels drop precipitously and become unmeasurable in postmenopausal women.
Several preanalytical and demographic variables have been studied previously for their effects on the measurement of AMH. For example, levels of AMH are stable with dilution and freezer storage, and are not affected by hemolysis (3) in current assay methods. AMH results are comparable in plasma (sodium citrate) and serum samples (4). Levels of AMH are not affected by menstrual cycle day (5) in young women of reproductive age. On the other hand, smoking, use of oral contraceptive pills (6), and increased weight or body mass index (7) are commonly associated with reduced AMH levels. There may also be racial and ethnic variability in normal AMH levels; African American women tend to have lower levels than age-matched white women (8).
The question of whether samples for AMH must be collected with consideration of food intake has not been studied. The goal of the present study was to determine whether glucose or food intake modifies AMH levels. This information can be used to guide clinical sample collection guidelines for AMH measurement, and may be particularly useful knowledge for testing AMH under conditions of metabolic dysfunction, such as PCOS, insulin resistance, or diabetes.
Materials and Methods
A convenience sample of residual plasma (sodium fluoride/oxalate) was available from 54 pregnant women under fasting conditions and then 1, 2, and 3 h after ingestion of a 100-g glucose challenge. This protocol was part of routine clinical care at the Women and Infants Hospital to identify gestational diabetes (GDM) between 24 and 28 weeks of gestation. Twelve of the 54 pregnant women met the criteria for GDM based on an increased glucose level (fasting, >95 mg/dL; 1 h, >180 mg/dL; 2 h, >155 mg/dL; and 3 h, >140 mg/dL) at a minimum of 2 time points. Glucose was measured at the time of clinical care using the hexokinase method on the ARCHITECT automated Aeroset immunoassay (Abbott Laboratory, Chicago, IL). This residual plasma study was performed with approval of the Institutional Review Board for Human Studies at Women and Infants Hospital. Samples were retrieved, aliquoted, and stored frozen at −20 °C for up to 3 years. AMH testing was performed after a second thaw.
Another sample set was collected from 8 nonpregnant women of reproductive age. An early morning sample was taken after an overnight fast, and then another sample was taken later that morning approximately 1 h after breakfast. These samples were collected at ARUP Laboratories after participant consent using a protocol approved by the Institutional Review Board at the University of Utah. Serum was separated and stored frozen at −20 °C for up to 1 month and then thawed once for AMH testing.
Samples were tested in duplicate for levels of AMH using the Ansh Laboratories (Webster, TX) ultrasensitive ELISA, according to manufacturer instructions. All assays were performed at Women and Infants Hospital. AMH levels are stable with freezer storage and up to 4 freeze–thaw cycles using this assay (3). Samples from 2 pregnant women were excluded from analysis because of nondetectable AMH levels (<0.08 ng/mL) at all time points. It was not possible to quantify whether AMH levels changed in response to glucose in these cases.
Group comparisons between pregnant women with and without GDM, and between pregnant and nonpregnant women were performed using t-test or a test of proportions (z-test). Correction for body weight was performed by converting data to multiples of the median for each patient group and then adjusting by a linear reciprocal weight equation. Levels of AMH were compared at fasting and 1, 2, and 3 h after glucose ingestion using a 2-way ANOVA (sample timing and GDM status) in the group of pregnant women, or matched t-test in the group of nonpregnant women. AMH data were subject to logarithmic transformation before analysis. Two-tailed P < 0.05 was considered statistically significant.
Results
The group of pregnant women had an average age of 31 years and weighed 75.4 kg (Table 1). Most were white (96%) and did not smoke cigarettes (92%). Women diagnosed with GDM weighed more than those without GDM (P = 0.04). All other variables (age, proportion of smokers, African American race, and AMH levels) were similar between women with and without GDM.
Selected demographic (mean ± SD) and AMH (median ± log SD) results in pregnant women at fasting and after glucose ingestion, stratified by GDM diagnosis.
As shown in Fig. 1A and Table 1, levels of AMH did not differ between fasting conditions and at 1, 2, or 3 h after glucose ingestion. AMH levels were similar at all time points regardless of normal or increased glucose levels; the results were not different in women with or without GDM. AMH levels have been reported to be lower in women with heavier body weights (7). Yet, after adjustment for body weight, AMH levels remained comparable in pregnant women with and without GDM (P = 0.86).
AMH levels at fasting and 1, 2, and 3 hours after glucose challenge (A). Closed circles indicate women eventually diagnosed with GDM, and open circles indicate women without GDM. AMH levels are not significantly affected by timing of sample collection (P = 0.99) or diagnosis of GDM (P = 0.78). AMH levels in relation to plasma glucose levels (2 h after glucose challenge shown as a representative time point) (B). Correlations of glucose and AMH levels are not significant (r = 0.09, P = 0.56 without GDM or r = −0.44, P = 0.15 with GDM).
Correlation analysis revealed that there was no significant relationship between glucose and AMH levels at any time point. For example, at the 2-h time point (Fig. 1B), the correlation coefficients for glucose and AMH levels were r = 0.09 (P = 0.56) in women without GDM and r = −0.44 (P = 0.15) in women with GDM.
The group of nonpregnant women had a mean age of 30 years (SD ± 4 years) and a mean body weight of 69.0 kg (±13.6 kg). AMH levels at fasting were significantly higher in nonpregnant (median = 3.86 ng/mL; log SD = 0.38) than in pregnant women (median =1.38 ng/mL; log SD = 0.43; P = 0.003). Adjustment for body weight did not change this result; AMH levels remained significantly higher in the nonpregnant than pregnant group (P = 0.0019). On the other hand, AMH levels in nonpregnant women were not different at fasting (median = 3.86 ng/mL; log SD = 0.38) and after a meal (median = 2.79 ng/mL; log SD = 0.36; P = 0.22).
Discussion
Preanalytical variables must be routinely considered in clinical chemistry testing. Variation in analyte levels can occur as a result of normal physiological variation, as well as in sample collection, transportation, processing, and storage procedures (9). Previous studies have established that AMH levels are stable with hemolysis, dilution, freezer storage, and menstrual cycle day in women of reproductive age. In the present study, AMH levels were similar under fasting conditions and after ingestion of glucose or a meal. Therefore, testing to evaluate ovarian reserve by AMH measurement in women does not need to be controlled for food intake.
We found no relationship of AMH and glucose levels in pregnant women. Furthermore, there was no difference in AMH results between women with and without GDM. This finding is corroborated by a large observational study of AMH levels in girls and women, with ages ranging from newborn to >51 years (10), which showed that fasting AMH and glucose levels were not correlated. AMH levels were also independent of insulin, cholesterol, triglycerides, and low- and high-density lipoproteins.
Although AMH and glucose are not correlated in healthy women or those with GDM, the relationship of these hormones may be different in other groups. For example, Kim et al. reported that young women with type 1 diabetes had lower AMH levels than nondiabetic controls (11). After the age of 35 years, AMH levels were similar between groups. The authors speculate that autoimmune mechanisms may have a role in the reduced circulating ovarian AMH in women with type 1 diabetes. They also suggest that early therapeutic interventions to obtain glycemic control could explain why the effects of diabetes on AMH are observed at early, but not later, ages. However, another report refuted this, finding that the reduced level of AMH among women with type 1 diabetes persisted after the age of 33 years, regardless of glycemic control (12).
A dysregulated relationship between AMH and glucose may also be present among women with PCOS. Many studies now document increased serum AMH levels in association with PCOS (2). PCOS patients are often obese and have insulin resistance. In this pathophysiological state, a positive relationship of serum AMH and insulin resistance (based on the homeostasis model of assessment of insulin resistance) has been observed in some (13), but not all (14), studies. Further evaluation of the relationship between AMH and glucose in women with PCOS is needed.
Although our data provide evidence that fasting is not required for determination of AMH in healthy women, there are other physiological variables known or suspected to influence its measurement. As observed presently, pregnancy is related to a suppression of AMH levels as compared with age-matched, nonpregnant women (15). The cause of this reduction in AMH during pregnancy is unknown, as is the timing and trajectory of recovery in the postpartum period. It is also possible, but not likely, that sample type contributed to a difference in AMH levels between pregnant (plasma) and nonpregnant (serum) women. We have previously reported comparable levels of AMH in plasma and serum (4), but a direct comparison was not performed in this study.
The strengths of the present study include examination of AMH levels in women having a fixed glucose load, with multiple samples collected at carefully timed intervals. Glucose levels were available for all pregnancy samples, and women with a range of glucose levels (with and without GDM) were included in the analyses. A previous large report of Chinese girls and women analyzed correlation between fasting AMH and glucose (10) but did not test a glucose or food challenge. A limitation of the present study is the small sample of nonpregnant women and the absence of glucose measurements in this group. Although the present results suggest that food intake did not alter the serum AMH level, this finding should be examined in a larger group of nonpregnant women, including those with PCOS, in men, and in children.
Finally, an important consideration in AMH measurement currently is the method of immunoassay. There are several manual ELISA methods available, and some of these have undergone protocol modifications over time. One early AMH assay did not demonstrate linearity of sample values and has since been modified to correct this (AMH Gen II, technical update, Beckman Coulter, August 2013). Newer methods provide linear assay results and include automated platform test options. However, AMH assays have not been calibrated uniformly, leading to significant variation in levels between methods (16, 17). It is critical for laboratorians and medical professionals to evaluate AMH levels against method-specific reference data at this time.
Footnotes
↵4 Nonstandard abbreviations:
- AMH
- antimüllerian hormone
- PCOS
- polycystic ovary syndrome
- GDM
- gestational diabetes.
- Presented in part at the Annual Meeting of the American Society for Reproductive Medicine, Salt Lake City, UT, October 2016.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form.
Employment or Leadership: None declared.
Consultant or Advisory Role: G. Palomaki, Ansh Labs.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Other Remuneration: G.M. Lambert-Messerlian, Ansh Labs, travel and reagent support.
Role of Sponsor: No sponsor was declared.
- Received March 8, 2017.
- Accepted June 8, 2017.
- © 2017 American Association for Clinical Chemistry
References
In this issue
