This retrospective single-centre study documents that AMH is sometimes assayed in our district as part of pituitary-gonadal evaluation. This is probably due to the authors’ previous interest in AMH  and does not reflect the habitual endocrinological management of the pituitary-gonadal axis in our district. This is in contrast with the reported indication for ovarian testing in certain situations (e.g. polycystic ovarian syndrome, peri-menopause, infertility, prior to ovarian surgery in reproductive-age women) besides ART, before ovarian stimulation .
It is well known that age is the most important factor related to the secretion of AMH, which decreases by about 5-7 pmol/l every 3-5 years in the fertile years [2, 5]. Tal and Davies  reported the lower limits of age-appropriate serum AMH values, stratified in 5-year intervals: 3.0 ng/ml at 25 years, 2.5 ng/ml at 30 years, 1.5 ng/ml at 35 years, 1 ng/ml at 40 years and 0.5 ml at 45 years. These levels are quite similar to those observed at the 25th percentile in comparable quintiles in our population (25-30 years: 16.9 pmol/l; 31-35 years: 8.9 pmol/l; 36-40 years: 5.0 pmol/l, and 41-45 years: 1.26 pmol/l). We also observed a very strong inverse correlation between age and both AMH and FSH in our women of reproductive age, independently from thyroid function and autoimmunity. The number of subjects and the size of the age-range can influence this observation. For instance, in 314 Iranian women with a mean age of 36.7 years (±SD, 6.1 years), Kabodmehai et al.  reported that older age correlated very significantly (P < 0.0001) with low AMH, a finding that was similar to ours. Moreover, in 67 consecutive infertile Japanese women with a mean age of 35.0 years (±SD 3.5 years; range not given) Kuroda et al.  reported that age correlated negatively with AMH levels, while in 27 normal fertile women of a similar median age (34.0 years) and with an age-range of 30-39 years, it did not. Finally, in women aged 20-40 years, Kucukler et al.  reported a significant inverse correlation between AMH and age both in those with normal thyroid function and in those with newly diagnosed sub-clinical or overt thyroid dysfunction. By contrast, on evaluating AMH levels in women aged 18-35 years, Adamska et al.  did not find any age-related changes in AMH in either 46 normal or 39 euthyroid TPOAb-positive women in this age-range.
It is well known that the incidence of autoimmune thyroiditis increases with age. Therefore, it is not surprising that our TPOAb-positive women had lower AMH levels than their TPOAb-negative counterparts, as their age was greater. Samsami et al.  stratified women into two age-groups: < 35 years old and > 35 years old, and found that AMH levels were lower only in TPOAb-positive women in the latter group. That autoimmune damage to the ovaries might take longer to become detectable is suggested by a longitudinal study which showed that women with a low ovarian reserve had higher baseline levels of TPOAb, and that these levels increased over 12-year follow-up .
In our pre-menopausal women, who were arbitrarily divided into subgroups aged from 18 years to > 46 years, AMH levels did not differ between women with normal thyroid function and hypothyroid women on L-T4 therapy. Similar results emerged from the study by Polyzoz et al. . These authors evaluated AMH levels by means of a non-sensitive immunoassay (functional sensitivity 2.5 pmol/l) in women stratified by age (range not reported; mean age 32 years) and categorized ovarian reserve as low (<10th percentile), normal or high (>90th percentile) according to age-specific AMH levels; they found that TPOAb did not differ significantly among groups .
In our study, the absence of differences in age-specific AMH levels between women off/on L-T4 treatment could be due to the normalization of thyroid function in this latter group. Moreover, there was no correlation between AMH levels and TSH and f-T4 levels, nor, in L-T4 treated women, between L-T4 dosage and AMH. Then again, the impact of L-T4 treatment on AMH levels is debated. In the study by Öztürk Ünsal et al. , in which 39% of women with chronic autoimmune thyroiditis were on L-T4 treatment, AMH concentrations were similar in patients on/off L-T4. By contrast, Kuroda et al.  found that AMH levels in 35 women with Hashimoto’s disease had improved after 3 months of L-T4 treatment.
Control of body weight could be another factor in preserving ovarian function. However, data on the interrelationship between BMI and AMH levels are still debated. In our study, a weak negative correlation between AMH and BMI was observed in all women; this remained significant on multivariate analysis only in hypothyroid women on L-T4, whose BMI was higher. A significant negative correlation between BMI or waist circumference and AMH was reported in a Turkish study involving a small number of women with normal thyroid function and sub-clinical or overt hypothyroidism when they were cumulatively evaluated . Adamska et al.  also reported a negative correlation between serum AMH and the percentage of body fat mass, as estimated by bioimpedance analysis, in a group of 39 women with Hashimoto’s thyroiditis, but not in 46 control women. By contrast, no correlation between AMH and BMI was observed in two older studies [11, 18]. Recently, AMH was evaluated in women with polycystic ovarian syndrome, and was found to be significantly and negatively correlated with BMI and waist-to-height ratio . The mechanisms underlying this inverse correlation between AMH and BMI are still unclear, though the impact of insulin resistance on follicular development in PCOS has been speculated . Interestingly, it has been suggested that the inverse relationship between serum levels of AMH and BMI could be the result of hormone dilution due to higher blood volume in women with elevated BMI .
For several years, we have sought to define normal TV in our Ligurian population . Moreover, when taking our patients’ history, we routinely ask women if they know the age at which their mothers’ menopause occurred. Consequently, these data are almost always reported in our medical files. Interestingly, in the present study, AMH was significantly related to TV (negatively) and the age of spontaneous menopause in the mother (positively) on multivariate analysis. The former finding could be linked to the time-related reduction in TV on L-T4 treatment , and the latter to genetic factors . Only Adamska et al.  reported TV in their study on AMH; they found no difference in TV between women with Hashimoto’s thyroiditis and control women of same median age of 26 years, the median value being 10 ml, which is only slightly higher than that found in our women (8 ml). This difference in TV could be explained by the difference in age-range, goitre control by means of L-T4 treatment, and regional differences in iodine intake.
Our study has several limitations. The first lies in the relatively small number of women involved and the retrospective design, which made it impossible to assess any temporal relationship between AMH and thyroid function/autoimmunity. In addition, as our women were recruited in a single centre, a selection bias cannot be excluded. Endocrinological diagnoses in our women were heterogeneous (e.g. possible high AMH levels in PCOS patients) and factors other than thyroid function and autoimmunity may have influenced AMH levels. Finally, as our study did not include a group of women with untreated sub-clinical/overt hypothyroid, we cannot exclude the possibility that untreated hypothyroidism (i.e. elevated TSH levels) may be involved in the decline of AMH in women of reproductive age.
The strength of this study is that it evaluated AMH levels and thyroid parameters in a local endocrinological setting in several clinical endocrine conditions, and not only in women undergoing ART procedures.
In conclusion, in our cohort of women, age proved to be a better predictor of AMH levels than any of the other factors linked to thyroid function and autoimmunity. Our data do not support the hypothesis that sub-clinical hypothyroidism and/or autoimmunity are associated with decreased ovarian reserve. The role of BMI and thyroid volume should be better defined in a larger number of cases, in order to obtain conclusive data. Moreover, further research is needed in order to investigate thyroidal mechanisms that regulate AMH secretion and the ovarian reserve, even though two systematic reviews have recently been published [31, 32]. Finally, thorough endocrinological-metabolic evaluation should be carried out in order to facilitate the achievement of fertility when reproduction is desired .