Serum AMH level as a marker of acute and long-term effects of chemotherapy on the ovarian follicular content: a systematic review

The toxicity of alkylating agents has been well studied. Experimental studies in rats and rhesus monkeys, treated with cyclophosphamide, have shown that the depletion of primordial follicles is rapid and drastic [14‐16]. Human ovarian histology following chemotherapy treatments evidenced atrophy, reduced follicle store [5, 10, 17] and especially, primordial follicle pool loss [18, 19]. Furthermore, one of these studies reported a deep AMH decrease in mice treated with cyclophosphamide concomitantly with the histological follicular depletion [16]. Some authors suggest that this primordial follicle loss can arise in two different ways. The first would be a direct eliminatory effect of chemotherapeutic agents on primordial follicles. The second would be indirect, through an excessive recruitment of the primordial follicles into the growing pool due to a decrease in Anti-Müllerian Hormone (AMH) after growing follicle damage [7]. Indeed, it has been shown that AMH-knockout mice exhibit premature activation and rapid depletion of ovarian follicle reserve [20]. In addition, a recent study showed that cyclophosphamide triggers upregulation of the PI3K pathway, initiating a wave of follicle recruitment and growth, and ultimately, burnout of the ovarian follicle reserve [21].

Oocyte and somatic cells in the ovary can be potential targets of chemotherapeutic agents, with differential sensitivity to the various types of drugs. In animal models, cyclophosphamide is well known to induce apoptosis of pregranulosa and granulosa cells in all classes of follicles [22]. Oocytes are damaged only in primordial and small antral follicles [14, 22]. A recent study in mice investigated the effect of doxorubicin on the ovarian follicles and suggested that damage to the oocytes may be mainly due to somatic cell failure [23]. Using a human ovarian xenografting model, Oktem and Oktay [8] showed that primordial follicle density is rapidly affected by apoptosis after cyclophosphamide injection (12%, 53% and 93% of follicle loss at 12, 24 and 48 h after the first injection, respectively). Human oocytes are drastically damaged (100% at 12 h post injection), followed by granulosa cells (63% at 12 h). The fact that alkylating agents are not cell-cycle specific and thus, do not require cell proliferation for their cytotoxic action can explain why resting follicles are also damaged. While the follicular toxicity of alkylating agents such as cyclophosphamide is well defined [2, 3, 13], that of multi-agent regimens remains poorly documented.

In addition to the apoptotic phenomenon of granulosa cells, ovarian stromal tissue could also be damaged by chemotherapy due to cortical fibrosis and blood vessel injury [7, 10, 11]. These phenomena could participate secondarily to the follicular alteration through impairment of follicular vascularization.

Most of the studies that have addressed the question of reproductive capacity after cancer have focused on markers such as cycle length, pregnancy occurrence and/or basal FSH values. Nevertheless, recent studies have shown that follicular depletion may occur despite recovery of regular menstrual cycles [24‐26]. This suggests that more accurate indicators are needed in order to properly inform the women about their fertility after treatment. Recent studies have suggested that AMH could be a valuable indicator of follicular depletion in breast cancer [27‐29] and lymphoma patients [24].

AMH is a dimeric glycoprotein and belongs to the TGF β family which acts on tissue growth and differentiation. Its name comes from its first known function: the regression of Müllerian ducts during male fetal differentiation. In women, serum AMH appears to be exclusively of ovarian origin since AMH is undetectable in serum 3 to 5 days following bilateral ovariectomy [30]. AMH is produced by granulosa cells of preantral and small antral follicles and its main physiological role seems to be the inhibition of the initial follicular recruitment from the primordial to the antral pool [31]. It has been extensively studied in Assisted Reproductive Therapy processes. It is now well established that AMH is the more accurate marker of the ovarian reserve [32]. In women, serum AMH levels are almost undetectable at birth and progressively increase to the adult level. It appears to be stable up to 25–30 years and then to decrease throughout the remaining reproductive life until being undetectable after spontaneous menopause [30, 32‐36]. AMH level on day 3 of the menstrual cycle has been shown to be strongly correlated with the antral follicle count (AFC) [33] and with the primordial follicule pool [37‐41]. Furthermore, AMH measurement is reproducible and easy to obtain. Highly sensitive ELISA assays are currently available to measure AMH levels in women: the Gen II assay and the Immunotech MIS-AMH assay. Both are produced by the same company (Beckman Coulter). They use different monoclonal antibodies targeting different regions of AMH and different calibrators thus explaining that the two assays generate quite different results and are not interchangeable. The Diagnostic Systems Laboratories (DSL) assay is not used anymore, but has been used in several studies cited in this review.

Finally, as AMH concentration does not change significantly during the menstrual cycle [30, 42, 43] and is weakly influenced by short term (< 6 months) gonadotropin suppressing treatments [44, 45] unlike other follicular markers such as FSH, estradiol, AFC, Inhibin B [30, 42, 43], AMH is thus considered as the more specific and reproducible marker of the ovarian reserve.

This review aims to highlight whether serum AMH could be a valuable indicator of the acute and long-term effects of chemotherapy regimen on the ovarian follicular content and how it could help in better understanding the chemo-induced toxicity and in fertility preservation strategies.

It has been shown that post-chemotherapy AMH levels are significantly lower than those of age-matched controls confirming the idea that AMH is a more accurate, direct and specific indicator of the follicle loss. Nevertheless, only the comparison of pre/post-chemotherapy AMH levels would allow the evaluation of the individual degree of follicular depletion. Studies which addressed this issue are summarized in Table 1.

In 22 young, regularly menstruating women with breast cancer, Lutchman-Singh et al. [46] demonstrated that, for each woman, the post-chemotherapy serum AMH level was significantly lower than the pre-treatment level. In this study, pre-treatment AMH levels were not different between the patients and the control population. Aslam et al. [47] in a smaller case–control study had the same results. In contrast, two other studies [48, 49] concluded that ovarian reserve seemed to be already compromised prior to chemotherapy for haematological malignancy. Indeed, in both these studies AMH was significantly lower in the treated group before chemotherapy compared to controls, even though the latter group was older. In addition, Lawrenz et al. [48] reported that the number of retrieved oocytes for fertility preservation was significantly lower in patients with lymphoma compared to those with breast cancer. They hypothesized that haematological malignancies may affect the ovarian reserve in women before chemotherapy, as it was already suggested in men [50] for semen quality, through unknown mechanisms.

To date, few studies have been designed to prospectively evaluate the longitudinal variations of AMH levels in patients treated by chemotherapy. This approach allows for studying the timing of follicle loss and of putative recovery. Five of these studies have been conducted in pre-menopausal, breast cancer patients [26‐29, 51], one in lymphoma patients [24], two in several types of cancer [52, 53] and two in childhood cancer [54, 55]. These studies are summarized in Table 2 for adult cancers and Table 3 for childhood cancers.

A marked and prompt fall in serum AMH levels as early as the first month of chemotherapy has been described in four longitudinal studies [24, 26, 53, 54]. Decanter et al. [24] investigated 30 young lymphoma patients (mean age: 24 years) divided into two groups according to the presence or not of alkylating agents in their chemotherapy protocol. AMH levels decreased significantly 15 days after the start of treatment, even in the non-alkylating group. Rosendahl et al. [53] found the same results as soon as week 1 after initiation of chemotherapy in 17 patients treated for lymphoma, breast cancer or Ewing sarcoma, of which 13 had an ovary removed for cryopreservation. Brougham et al. [54] recently described the same drastic AMH fall in 22 prepubertal and pubertal girls (median age: 4,4 years) treated by various chemotherapy or radiotherapy protocols for different types of cancer as did Anderson in 59 early breast cancer patient (mean age: 42.6) as soon as one cycle of chemotherapy [26]. Mörse et al. [55] and Dillon et al. [52] described this fall of AMH but only three months after the first cycle of chemotherapy in, respectively, 34 young patients (mean age: 9,5 years) and 46 patients (mean age: 26,1 years) treated for different kinds of cancer. At the end of chemotherapy, these studies showed significantly decreased or undetectable AMH levels, independently of protocols. The study of Dillon et al. [52] is the only study to prove an association between alkylating agent exposure and post-therapy impairment of AMH but in a very heterogeneous group.

The same significantly decreased or undetectable AMH levels at the end of chemotherapy was obtained in studies conducted in breast cancer patients by Anderson et al., Anders et al., Yu et al. and Henry et al. [27‐29, 51]. Anderson et al. in 2006 [28] investigated 50 breast cancer patients (mean age: 41 years) who were treated by different regimens of adjuvant chemotherapy, all including cyclophosphamide. In most of the patients, AMH levels remained undetectable after treatment. Anders et al., Yu et al. and Henry et al. confirmed these results in 44, 26 and 27 premenopausal breast cancer patients, respectively [27, 29]. The drastic AMH fall just after the beginning of chemotherapy suggests that most of the growing follicles are damaged. These results are in keeping with those previously described in the animal models with alkylating agents [14‐16].

Conversely, the pattern of the ovarian recovery phase greatly differs depending on the protocols. Decanter et al. [24] showed different ovarian recovery patterns between the groups with or without alkylating agents. Patients who were treated with a non-alkylating regimen recovered their pre-treatment AMH levels, as early as the 6th month of follow-up. In contrast, in the alkylating regimen group, AMH values during all the follow-up remained significantly different from the pre-treatment value, with very low or undetectable levels. Rosendahl et al. [53] also reported significantly higher AMH values in patients who did not receive alkylating agents. All women treated for breast cancer received alkylating agents and, in the three studies, serum AMH levels after chemotherapy remained very low or undetectable in all of them. Likewise, in a study by Brougham et al. [54], different recovery phases were also described between the high gonadotoxic risk group and the medium/low risk group. These groups were defined depending on chemotherapy drugs, cumulative dose and exposure (or not) to radiotherapy involving the ovaries. Indeed, AMH remained undetectable even 3 years after the end of chemotherapy in the high risk group. In contrast, in the low/medium risk group, the recovery occurred as soon as 6 months after treatment. The work of Mörse et al. [55] confirmed these results with undetectable AMH levels 18 months after the beginning of chemotherapy in patients who received irradiation below the diaphragm and/or stem cell transplantation. But after 15 months of follow-up in the 7 patients treated for acute lymphatic leukaemia (5 with alkylating agents), AMH levels returned to pre-treatment or higher values. In this last study, distinction between high and low gonadotoxic risk was not possible because of the heterogeneity of the recruitment and the small number of patients in each group. Finally, a recent work of Dillon et al. [52] concluded that the rate of recovery of AMH after the end of chemotherapy was impacted by the alkylator use. Almost all women treated for breast cancer received alkylating agents and, in the five studies, serum AMH levels remained very low or undetectable during the entire follow-up after chemotherapy [26‐29, 51].

Only the dynamics of AMH level during the recovery phase can discriminate between protocols and suggest different degrees and/or mechanisms of ovariotoxicity. Presumably, the primordial follicle pool is less damaged in the case of non-alkylating protocols, therefore allowing the rapid emergence of newly growing and AMH-secreting follicles. Intriguingly, AMH levels decreased drastically just after the beginning of chemotherapy even in the non-alkylating protocol group. Whether this fall of AMH could reflect either the follicle depletion or a functional granulosa cell impairment remains to be established.

Infertility and early menopause are associated with chemotherapy even in women who resumed menses after treatment [56]. The probability of early menopause and infertility depends on the type of cancer, the treatment, and the age at diagnosis [56]. Relationships between post-chemotherapy AMH levels and ovarian function, as indicated by menstrual activity, are difficult to establish. Indeed, among patients in whom AMH levels remained undetectable at least one year after the end of chemotherapy, a significant number had already recovered spontaneous menses [24, 25, 27‐29, 53, 57]. Conversely, all the patients with persisting amenorrhea at one year of follow-up had undetectable AMH levels [24, 25, 28, 57]. It may be that ELISA AMH assays currently available are not sensitive enough to detect very low values of AMH in women who recovered menses despite “undetectable” AMH level in the present dosage. There are very limited data about AMH and spontaneous conception. One study, by Hagen et al., showed no reduced fecundity in young women with a low AMH level compared to normal AMH [58] but another study showed that rather older women did show a relationship between AMH and spontaneous fertility [59]. In addition, it is now well established that a low AMH level (<0.7 μg/l) is of poor prognosis in assisted reproductive therapy [32]. Nevertheless, whether a low post-chemotherapy AMH level predicts the same low chances of conception remains to be elucidated.

Anderson et al. recently showed that pre-treatment AMH level is the strongest predictor (better than age) of menstrual activity at 1, 2 and 4–5 years of follow-up in two studies including 59 and 42 breast cancer patients [26, 28]. Similarly, Rosendahl et al. [53] showed that high pre-treatment AMH levels were predictive of higher AMH levels during recovery of ovarian function after chemotherapy, independently of patient’s age, removal or not of one ovary and type of chemotherapy. In the same way, Dillon et al. [52], very recently, showed a better rate of AMH recovery after the end of chemotherapy when pre-treatment AMH level was over 2 ng/ml. Likewise, Anders et al. [27] reported that pre-treatment AMH concentration was lower in women treated for breast cancer who were amenorrheic one year after chemotherapy, although Yu et al. have reported no such difference [29]. Noteworthy, pre-treatment AMH level can be lower in haematological malignancies before chemotherapy [48]. The possible causes of the reduced AMH level in haematological malignancies prior to chemotherapy are still unclear. Maybe a functional impairment of granulosa cells by compromised general health or systemic inflammation can explain or contribute to the reduced AMH level. Very recently, Dorp et al. [60] described specifically these reduced AMH levels in very young girls with newly diagnosed cancer. They hypothesized that a decrease in AMH production, an increased metabolism of AMH or a more rapid decline of the primordial follicle pool might be involved in this reduced AMH level. This must be taken into account for fertility preservation counselling.

Thus, the pre-treatment AMH level may predict the long-term ovarian function after chemotherapy but the prognosis of conception when the AMH level is low after chemotherapy needs to be ascertained. Indeed, even in young healthy women, the low concentration of serum AMH is not predictive of reduced fecundability [58]. Only a long-term and systematic follow-up of ovarian reserve by AMH and of spontaneous fertility in a large population will allow these issues to be addressed.