Is Anti-Müllerian Hormone a Marker of Ovarian Reserve in Young Breast Cancer Patients Receiving a GnRH Analog during Chemotherapy?

Abstract

Introduction

Anti-Müllerian hormone (AMH) is the most reliable biomarker of ovarian reserve; however, its role in predicting ovarian recovery after chemotherapy is unclear. Administration of a GnRH analog (GnRHa) during chemotherapy significantly reduces the ovarian failure rate and increases the pregnancy rate. The available data on the behavior of AMH during concurrent administration of chemotherapy and GnRHa are inconsistent. We investigated whether concurrent administration of triptorelin and adjuvant chemotherapy might reduce the expected drop of AMH.

Methods

Eligible patients were premenopausal women aged <40 years, with a diagnosis of early breast cancer, and candidates to 4–8 cycles of adjuvant chemotherapy. Triptorelin (3.75 mg i.m.) was started before chemotherapy and administered every 4 weeks thereafter. The principal endpoint was the proportion of patients with an AMH percent change ≤50% between 12 months after chemotherapy and basal levels. The secondary endpoint was the proportion of patients achieving postchemotherapy AMH levels above the threshold of 0.2 ng/mL.

Results

Fifty patients were enrolled, 31 of whom had blood samples available at baseline and 1 year after the end of chemotherapy. AMH decreased to nearly undetectable levels after chemotherapy and recovered after 12 months, but they did not exceed 1 tenth of the pretreatment levels. As for the secondary endpoint, 15 of the 31 patients recovered AMH levels above the threshold.

Conclusions

This study did not reach its principal endpoint; however, the rate of 48% of patients who recovered AMH above threshold levels favorably compared with those in studies without concurrent GnRHa, supporting a better recovery of AMH with triptorelin.

Introduction

About 7% of breast cancer patients are diagnosed before the age of 40 years, and up to 20% of newly diagnosed breast cancers occur in women of childbearing age [1]. Chemotherapy is largely used in the adjuvant systemic treatment of breast cancer patients. The principal long-term effect of cytotoxic agents in young women is impairment of ovarian function, which may result in chemotherapy-induced amenorrhea (CIA), premature menopause, and reduced fertility [2].

Anti-Müllerian hormone (AMH), a dimeric glycoprotein produced by ovarian granulosa cells, is the most reliable biomarker of ovarian reserve [3]. A large amount of data shows that AMH levels dramatically drop during chemotherapy, while data showing long-term recovery are inconsistent regarding the role of pre- or postchemotherapy AMH levels in predicting chemotherapy-induced amenorrhea [4].

The role of a GnRH analog (GnRH a) as a means to protect the ovaries during chemotherapy has been debated for a long time. After inconsistent results arising from a series of studies [5, 6, 7, 8], the POEMS study showed, in a large randomized cohort, that the administration of GnRHa concurrently with chemotherapy significantly reduced the ovarian failure rate (8 vs. 22% in the control cohort) and, most importantly, was associated with a 5-year increased pregnancy rate (23.1 vs. 12.2; p = 0.04) [9, 10].

The available data on the behavior of AMH during concurrent administration of chemotherapy and GnRHa are few and inconsistent.

In 8 early breast cancer patients receiving GnRHa and tamoxifen, AMH levels declined slowly but significantly after 12 months [11]. On the contrary in a cohort of healthy young premenopausal women (mean age 30.3 years) who were administered 1 monthly depot of GnRHa, AMH declined after 7 days but increased above pretreatment levels after 30 days [12].

In a series of young women (mean age 24 years) receiving chemotherapy and GnRHa for hematologic malignancies, AMH levels dropped significantly during treatment but recovered at the end of chemotherapy [13]. The latter finding could hardly be extrapolated to breast cancer patients given the age difference between the 2 groups.

In the ZORO study, AMH levels were measured in 17 out of 60 patients, 8 of whom were receiving goserelin plus chemotherapy and 9 of whom were receiving chemotherapy alone [5]. The median time from randomization to measurement of AMH was 4 years. Thirty-three percent of the patients in the chemotherapy-alone group versus 50% of the patients in the chemotherapy-plus-goserelin group had AMH levels >0.2 μL/L. The authors commented that these differences might have resulted from the different ages at the time of sample collection [5]. Regardless of the explanation, the hypothesis that AMH measurements could reflect preservation of the ovarian reserve exerted by GnRHa appeared plausible and deserved to be prospectively investigated in a larger sample.

Patients and Methods

We designed a multicenter prospective study to investigate whether the change in AMH during administration of triptorelin concurrently with chemotherapy might represent a surrogate marker of the gonadoprotective activity of GnRHa.

Eligible patients were premenopausal women aged <40 years who were diagnosed with breast cancer and were candidates for adjuvant chemotherapy. Premenopausal status was defined as the presence of regular menses within the past 6 months in the absence of any form of hormonal contraception or any other hormonal treatments. Patients were enrolled into 2 different cohorts according to hormone receptor (HR) status in order to evaluate separately also the possible confounding effect of different durations of GnRHa and of tamoxifen on AMH. Patients received 4–8 cycles of chemotherapy according to the internal protocols of the participating Institutions. All standard chemotherapeutic regimens commonly used for adjuvant treatment of breast cancer were allowed.

The principal endpoint was the proportion of breast cancer patients aged <40 years with a percent change (Δ) ≤50% of AMH levels at 12 months after chemotherapy as compared to basal levels.

Secondary endpoints were the proportion of patients with AMH levels >0.2 ng/mL at the 12-month time point and the comparison of 12-month AMH levels in the 2 cohorts of HR-positive (+ve) and HR-negative (−ve) patients.

All of the patients received triptorelin (3.75 mg i.m.), which was started about 1 week before chemotherapy and administered every 4 weeks thereafter; in the HR–ve cohort triptorelin was stopped after the last cycle of chemotherapy, while patients with HR+ve cancer continued triptorelin as part of the endocrine adjuvant treatment. Since triptorelin was not indicated as part of the adjuvant treatment in patients with HR–ve tumors, it was provided by Ipsen for the HR–ve cohort.

Patients underwent clinical follow-up according to the internal protocols at the different participating Institutions.

AMH was assessed before chemotherapy, at the end of chemotherapy, and 12 months after the end of chemotherapy. Since literature data report postchemotherapy AMH levels in the range of <0.1 ng/mL [14], a baseline level >0.2 ng/mL was required in order to detect a difference of at least 50%. Only patients with baseline levels of AMH >0.2 ng/mL were considered evaluable.

Blood samples were centrifuged and stored at −20°C in the local laboratory before shipment to the central laboratory (Humanitas Research Hospital, Rozzano, Italy), where the samples were stored at −80°C until the assay was performed. All of the samples from each patient were assayed at the same time in order to reduce interassay variability.

AMH levels were determined using an Access AMH assay according to manufacturer's instructions [15]; it is a simultaneous 1-step sandwich chemiluminescence immunoassay that uses 2 mouse monoclonal antibodies recognizing total AMH.

Statistical Considerations

Literature data show a decline of about 90% the baseline value of AMH after chemotherapy and a poor recovery after 1–2 years, but no cut-off value for AMH has been established as a reliable surrogate of ovarian function preservation.

Although previous data do not allow exact estimation of the effect of GnRHa, we hypothesized that a protective effect on ovaries could result in a Δ of 50% rather than 90%. We arbitrarily defined as the principal endpoint the percentage of patients achieving a reduction of of AMH levels ≤50% 12 months after the end of chemotherapy.

A proportion of patients achieving a percent change of AMH ≤50% of 10% or lower were considered to be clinically unworthy, whereas a proportion of patients of 35% or higher were assumed to be of potential interest. Further evaluations would be recommended with 0.025% rejection (1-sided α level and 95% power) if 8 or more of the 35 patients of the total evaluable patients in each cohort would obtain a percentage of AMH ≤50%. Considering a dropout rate of 10%, a sample size of 40 patients for each cohort was calculated.

Data were summarized as frequencies and proportions (categorical variables) or as medians and ranges (continuous variables). Differences between groups were evaluated using a χ² test or the Fisher exact test, when appropriate. Differences between medians were evaluated using the Mann-Whitney Wilcoxon test.

Results

Enrollment started in July 2013 and closed in June 2015 after the target sample was reached in the HR+ve tumor cohort, while the enrollment of patients with HR–ve tumors was deemed too slow to reach the target sample.

Overall 52 patients were screened and 50 were enrolled. Two patients were screening failures (1 was found to be metastatic at the staging performed after surgery and 1 was aged ≥40 years). Forty patients were enrolled into the HR+ve cohort and 10 were enrolled into the HR–ve cohort.

The principal baseline characteristics of the patients are summarized in Table 1. Final follow-up samples were missed in 15 patients. Reasons for missing samples were: patient loss to follow-up (n = 10), prophylactic oophorectomy (n = 1), and lost samples (n = 4). AMH measurements were performed only in patients with samples available at baseline and at the 12-month time point according to the principal endpoint of this study.

Table 1.

Baseline patient characteristics

Age, years	35.8 (25.5–39.7)
Parity
Yes	32
No	17
Fertility preservation procedures	9
BMI	21.5 (17.2–37.5)
Tumor size
pT1	21
pT2	24
pT3	4
pT4	1
Nodal status
pN0	20
pN1	20
pN2	7
pN3	1
pNX	2
HR status
Positive	40
Negative	10
HER2 status
Positive	15
Negative	35
Cycles of chemotherapy
4	12
6	8
8	30
Anthracycline	12
Anthracycline + taxanes	37
Anthracycline + CMF	1
Endocrine therapy
Tamoxifen	23
Exemestane	7
Unknown	10

Values are presented as medians (range) or numbers.

Thirty-four patients had blood samples available at baseline and 1 year after the end of chemotherapy. The median age was 35 years. Baseline AMH levels were higher, but not significantly, in patients aged ≤35 years as compared to those older than 35 years (2.46 [0.04–8.13] vs. 1.72 [0.13–6.0]; p = 0.09). Three patients (all in the HR+ve cohort) had baseline levels <0.2 ng/mL and were not considered evaluable.

The final analysis was performed in 31 patients, i.e., 25 with HR+ve tumors and 6 with HR–ve tumors. The median age of the 31 evaluable patients was 36 years (range 28–40).

Since the number of evaluable patients was lower than that estimated in the statistical plan, the analyses are to be considered exploratory.

The results of the AMH measurements are summarized in Table 2. AMH decreased to nearly undetectable levels after chemotherapy and recovered after 12 months but the median change was 92% (29–99.8%).

Table 2.

AMH levels overall and in the HR+ve and HR–ve tumor cohorts separately

	Overall	HR+ve	HR–ve
	(n = 31)	(n = 26)	(n = 5)
AMH at baseline, ng/mL	2.38 (0.39–8.13)	2.1 (0.13–8.13)	1.5 (0.72–6)
AMH at the end of CT, ng/mL	0.01 (0–0.14)	0.01 (0–0.14)	0.01 (0.01–0.03)
AMH at 12 months, ng/mL	0.19 (0.01–1.12)	0.11 (0.01–0.68)	0.25 (0.10–1.12)
Δ 12 months/baseline, %	92.2 (29–99.8)	93.1 (45.8–99.8)	80.2 (29–95)

Values are presented as medians (range). Δ, percent change.

When we analyzed the impact of age on AMH, a nonsignificant association with lower 12-month AMH levels (0.07 vs. 0.43 ng/mL; p = 0.091) was observed in older women, similarly to what occurred for baseline levels, but AMH change was significantly greater in older women (94.8 vs. 88.3%; p = 0.021).

We also analyzed the impact of different durations of chemotherapy on AMH posttreatment levels and change. All of the patients received anthracycline and cyclophosphamide for at least 4 cycles, and 18 patients received also CMF for 3 cycles or taxanes either on the weekly schedule or on a 3-week schedule. Since only 3 patients had received 6 cycles, we grouped patients who received 4–6 cycles (n = 13) and compared them with those receiving 8 cycles (n = 18).

Although a slight trend toward a worse recovery was observed in the 8-cycle cohort, no significant impact of different treatment durations on 12-month AMH and change was observed (AMH 0.11 vs. 0.29 ng/mL; p = 0.082).

Analyses performed separately in the HR+ve and HR–ve cohorts suggested a greater recovery in the latter, with nearly double final AMH levels and a median Δ of 80%, but the numbers were too small to allow formal comparison between the 2 groups. Fifteen out of 31 patients (48%) achieved AMH levels above the threshold of 0.2 ng/mL at the 12-month time point.

Discussion

The role of GnRH as a means to protect ovaries during chemotherapy has been debated for a long time due to conflicting results arising from studies. In 2015 the POEMS study showed a significant benefit in terms of reduced ovarian failure and the likelihood of a future pregnancy in patients receiving GnRHa during chemotherapy [9, 10].

A meta-analysis of individual patient data from 5 randomized trials including 873 patients reported ovarian failure in 14.1 versus 30.9% of cases and pregnancy rates of 10 versus 5,5% in patients in the GnRHa-treated and control groups, respectively [16].

In light of these findings the 2018 updated ASCO clinical practice guidelines on fertility preservation in cancer patients, despite not including GnRHa among proven fertility preservation methods, acknowledged that, when the former are not feasible, GnRHa might be offered to young breast cancer patients with the hope of reducing the likelihood of chemotherapy-induced ovarian insufficiency [17].

Our study did not meet its principal endpoint since AMH decreased to about 1tenth of the baseline levels in most patients, especially in the HR+ve cohort, which is consistent with previous literature data. Although the number of evaluable samples was lower than planned, especially for the HR–ve cohort, we do not think that increasing the number of samples would have changed these results.

These findings are not fully consistent with those obtained by Zhong et al. [18], who randomized breast cancer patients to receiving chemotherapy with or without concurrent goserelin and compared AMH levels before and after treatment in both groups. The authors showed a drop of AMH after chemotherapy, which was not different between the 2 groups; however, it was not significant as compared to baseline levels. On the other hand, the authors found that concomitant goserelin reduced the rate of ovarian failure, defined as amenorrhea, in the preceding 6 months [18].

As for the secondary endpoint, the proportion of patients showing AMH levels >0.2 ng/mL at the 12-month time point was 15 out of 31 (48%). This finding favorably compares with those reported in studies which assessed the AMH change 1 year after the end of chemotherapy and which did not allow concurrent administration of GnRHa during chemotherapy. Yu et al. [19] reported AMH level recovery 52 weeks after the end of chemotherapy in 4% of 26 early breast cancer patients with an age similar to that in our study.

Henry et al. [20] reported that only 3 out of 22 (13%) patients who had AMH levels assessed 1 year after chemotherapy recovered AMH levels above the level of detection. Dezellus et al. [21] conducted a large prospective study with serial measurements of AMH during and after the end of chemotherapy, retaining a high rate of samples at the 24-month follow-up (181 out of 250 baseline samples), and reported a rate of 54.6% of patients with undetectable AMH levels 24 months after the end of chemotherapy. Hamy et al. [22] showed an average monthly increase in AMH levels of 1.2% with a long follow-up (median measurement time: 20 months), although they did not specify whether the final follow-up levels were comparable to the pretreatment levels.

Conversely, in a Korean prospective study of 105 patients receiving GnRHa during chemotherapy, the proportion of patients who reached AMH levels >1 ng/mL 12 months after the end of chemotherapy, considered as a surrogate of ovarian function recovery, was 71.4%; this is substantially higher than ours and it was even higher in the 69 samples available with a longer follow-up (78%) [23]. These results are poorly comparable with ours since the median age of the patients was 32 years (nearly 3 years younger) and GnRHa was stopped after completion of chemotherapy, while tamoxifen was allowed. In the Korean study the proportion of patients achieving AMH levels >1 ng/mL at 12 months was significantly lower in tamoxifen users [23]. Similar results were reported with tamoxifen in 20 breast cancer patients and in a larger cohort of 90 breast cancer patients also treated with GnRHa, while in a population-based cohort study the AMH geometric mean was significantly greater in tamoxifen users [24, 25, 26].

According to these and our results, we cannot exclude that tamoxifen might have negatively affected AMH recovery since the final median AMH levels were double in the HR–ve cohort as compared to the HR+ve cohort, although the very low number of patients in the former group does not allow exclusion that this difference was due by chance.

In conclusion, our study, despite failing to reach primary endpoint, provides support for a better recovery of AMH in young breast cancer patients treated with concurrent GnRHa and chemotherapy although a longer follow-up may be needed to observe the full effect of GnRHa.

Statement of Ethics

This study complies with guidelines for human studies and was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. The study protocol was approved by the institutional review board at each participating site. All of the patients provided written informed consent. The protocol was approved at the coordinating center, i.e., Istituto Clinico Humanitas (reference No. CE ICH-60/13).

Conflict of Interest Statement

R.T.: participation on the advisory board of and recipient of speaker fees from MSD, Pfizer, Lilly, and the Istituto Gentili. A.M.: advisory boards of Roche, Novartis, Pfizer, Lilly, Macrogenics, and Eisai. A.S.: participation on the advisory board of and recipient of speaker fees from Sandoz, Servier, EISAI, Roche, Novartis, Gilead, Pfizer, and BMS.

Funding Sources

This study was supported by an unrestricted grant from Ipsen.

Author Contributions

R.T., V.B., M.C., A.M., C.C., and A.S. contributed to patient enrollment and management. L.G. performed the statistical analysis. M.N.M. performed the laboratory assays. All of the authors contributed to the preparation of this work and revised the final draft.