Ovarian stimulation is a key event in human IVF, and choosing properly the initial dose of gonadotropins may avoid a poor oocyte yield and, on the other side, an excessive response with high risk of ovarian hyperstimulation syndrome (OHSS). In most IVF Units, the individual dose to stimulate the ovary is tailored considering age, basal FSH level, ovarian volume and the count of small antral follicles (AFC). Prediction models that use these variables singularly or in combination reach an acceptable, even if not optimal, level of accuracy in estimating ovarian responsiveness to exogenous gonadotropins [1, 3].
In the last years, serum AMH measurement has been introduced into the clinical practice, and is presently considered the best marker of ovarian reserve , being also related to the live birth rate in IVF . AMH is a dimeric glycoprotein belonging to the transforming growth factor-beta (TGF-β) superfamily and is produced by the granulosa cells of small preantral follicles . Differently from FSH and AFC, AMH is a direct expression of the amount of small follicles that constitute the bulk of ovarian reserve [9, 10]. Furthermore, its serum levels show a very low inter- and intra-cycle variability and are independent of the menstrual cycle phase .
At present, however, automated assay methods for AMH are not yet available, and the Elisa assays currently in use have a rather high variability . These assays are expensive and not yet available in all IVF labs; their cost limits the possibility of using them in many developing countries, where relevant efforts are made to keep IVF programs as cheap as possible.
When a patient has been submitted at least once to ovarian stimulation for IVF, ovarian sensitivity index (OSI) can be easily calculated dividing the total administered gonadotropin dose by the number of oocytes retrieved at OPU. This ratio represents indeed the ovarian resistance to gonadotropins, as the lower is FSH dose, the higher is ovarian sensitivity. However, the term "ovarian sensitivity index" appears to be more immediate than "ovarian resistance" in recalling the idea of ovarian sensitivity, and therefore has been adopted herein.
The present study shows that OSI displays a strong, inverse correlation with AMH levels, and that this correlation is stronger than those between AMH and the total gonadotropin dose, or between AMH and the total number of retrieved oocytes. This is not surprising as it was shown that neither the total FSH dose administered, nor the absolute number of oocytes accurately reflect ovarian responsiveness to hormonal stimulation; their ratio is much more accurate in describing how the ovary produces oocytes in response to exogenous stimulation .
According to our observations, OSI does not accurately reflect the patient's age, even if it shows a fairly good correlation with it (r = 0.40). It rather appears to be partially independent on age: this suggests that it could be incorporated in prediction models aimed at predicting ovarian responsiveness to exogenous gonadotropins as a partially independent parameter.
Our data have been obtained using the GnRH-agonist buserelin plus rFSH in a classical "long" protocol; it must be remarked that the correlation between OSI and AMH found herein could be slightly different in case a different stimulation schedule or different drugs are used. It has been reported, in fact, that rFSH stimulation yields on the average more oocytes with a lower dose compared to what happens using human menopausal gonadotropin (hMG) .
In conclusion, the present study shows that OSI accurately reflects AMH level of women submitted to IVF with a "long" protocol. According to the present study, OSI appears to be a highly reliable index of ovarian responsiveness to rFSH and can be useful to estimate the rFSH dose in all IVF cycles following the first, provided that the same kind of protocol is used. We have studied herein a relatively small cohort of patients, although quite homogeneous: obviously larger scale (and maybe multicenter) studies are required to conclude that OSI may be used as a surrogate for AMH for predicting the ovarian response in IVF. The possibility that OSI could be a reliable surrogate of AMH measurement is particularly interesting for IVF Units where AMH assay is not yet available or is considered too expensive to be introduced, e.g. in developing countries where the economical cost of IVF program needs to be limited.