Assessment of ovarian reserve in girls diagnosed with pediatric cancer: a presentation of baseline data from a prospective follow-up study

Introduction

Survival rates for childhood cancer continue to improve, with overall survival now exceeding 80% (1–3). However, cancer treatment can adversely affect the function of the reproductive organs. For girls and women, premature ovarian insufficiency (POI) and reduced ovarian reserve are well-established late complications of cancer treatment (4). Alkylating agent chemotherapy and radiotherapy to the pelvic organs are gonadotoxic and associated with a high risk of infertility (4–6).

With rising survival rates, an increasing number of childhood cancer survivors (CCS) are reaching reproductive age and may wish to have children. Fertility preservation could be possible for high-risk patients through careful consideration of treatment delays, patient age at diagnosis, counseling, and a multidisciplinary team approach.

Fertility preservation options are limited in prepubertal girls, with ovarian tissue cryopreservation (OTC) being the only available method (7). The International Late Effects of Childhood Cancer Guidelines Harmonization Group (IGHG) moderately supports OTC for prepubertal girls requiring treatment with high-dose alkylating agents, ovarian radiotherapy, or hematopoietic stem cell transplantation (8). For postpubertal girls, oocyte cryopreservation is an established method. Current guidelines strongly recommend offering this to patients at high risk of developing POI following cancer treatment, provided that delaying treatment does not impact prognosis (8). Although medically feasible, this can be a challenging decision to make many years before these young women might consider starting a family.

Anti-mullerian hormone (AMH) is a marker for ovarian reserve that is detectable from birth. AMH levels increase until around the age of 25 years, when levels gradually decline until menopause (9–12). Serum levels of AMH strongly correlate with the number of antral follicles in the ovary, suggesting that AMH reflects the size of the primordial follicle pool (13). Previous studies have defined normal AMH levels across the lifespan in different populations with peak values described at different ages, though data on AMH in the normal population is limited, especially in children (9, 11, 14). Kelsey et al. (11) produced a validated model of serum AMH from conception to menopause by including 3,260 data points across the range 0.3 (cord blood) to 54.3 years. A 2014 study by van Dorp et al. (15) showed that AMH levels in girls with newly diagnosed cancer were reduced compared to healthy controls, even before the start of chemo- and/or radiotherapy, suggesting that the disease itself and/or the subsequent general health status affects AMH levels. An ongoing study from our research group is currently investigating AMH as a primary surveillance modality of fertility in CCS of reproductive age (16).

A cancer diagnosis and treatment will inevitable generate considerable psychosocial stress for the child and family. The psychosocial impact of childhood cancer is widely acknowledged. Access to appropriate interventions should be facilitated to optimize the wellbeing of parents, child, and family (17). The monitoring of stress throughout the treatment trajectory is important in order to tailor the needs for each family (18).

Many initiatives have been made to create national guidelines for fertility surveillance and the detection and management of POI in CCS. Due to differences between definitions for at-risk populations, surveillance modality, and frequency of follow-up, it has been difficult to determine long-term follow-up programs in clinical practice and to reach a consensus in fertility monitoring (19). National guidelines have been published in Sweden with suggestions on how to follow-up CCS, but it is not clear how women should be monitored in terms of fertility (20).

The aim of this study is to present baseline data from the inclusion phase of a prospective follow-up study on fertility surveillance in girls aged 0–17 years old and diagnosed with cancer. For girls where the cancer treatment poses a high risk of infertility, we aim to offer fertility preservation with OTC or oocyte cryopreservation. By monitoring ovarian reserve, we intend to offer fertility preservation at follow-up visits if AMH and antral follicle count (AFC) suggest reduced fertility. We hypothesize that this could increase chances of having children as well as positively affect quality of life.

Materials and methods

Participants

Participants were enrolled in the study through the Pediatric Oncology Center at Sahlgrenska University Hospital in Gothenburg, Sweden, between May 2016 and March 2021. Inclusion criteria were a new diagnosis of pediatric cancer or other medical conditions requiring chemotherapy (for example stem cell transplantation for aplastic anemia) and scheduled treatment with chemotherapy and/or radiotherapy. Patients who were going to start treatment at a different hospital were excluded. Written informed consent was obtained from the participants' guardians. The Pediatric Oncology Clinic at Sahlgrenska University Hospital, Gothenburg, is one of six pediatric oncology clinics in Sweden. The clinic at Sahlgrenska serves an extensive geographical area of the western part of Sweden. For a proportion of patients, the maximal distance between their place of residence and the Pediatric Oncology Clinic is more than 300 km. Therefore, many patients receive a significant portion of their treatment and follow-up care at their local hospital. Due to geographic distance, it was not logistically possible to arrange blood sampling for all patients.

The participants will be part of a longitudinal follow-up program until they reach the age of approximately 25, by which time puberty will have passed and their ovarian reserve can be evaluated as an adult. AMH levels were measured at inclusion and are planned to be measured every 2 years. Psychological wellbeing will be assessed during the follow-up visits. The participating girls will fill out validated questionnaires from the age of 12 and thereafter at 15 and 18 years of age. The following tools will be used: Short Form-36 (SF-36) for quality of life, Hospital Anxiety and Depression Scale (HADS), which is used for evaluating signs of anxiety and depression, and the Sense of Coherence Questionnaire-13 (SOC-13), which measures an individual's ability to manage stress and stay healthy by assessing how they perceive life as comprehensible, manageable, and meaningful. The parents' view on the question of fertility in their daughters will also be explored through a parental questionnaire. For older girls, AFC will be assessed by vaginal sonography (age not specified, but post pubertal, when appropriate, and when she agrees to the examination); see Table 1. A risk assessment of potential risk for infertility will be done according to the Swedish treatment risk classification as described by Nyström et al. (21) and the IGHG guidelines (8) (Tables 2, 3). Girls that are considered at very high risk of developing POI will be offered fertility preservation. Ethical approval was granted by the Regional Ethics Board at the University of Gothenburg in Gothenburg, Sweden (registration number 808-15). Handling data will follow the Swedish Data Protection Act (SFS 1998:204) and the EU's General Data Protection Regulation (GDPR).

Table 1. The long-term study protocol.

AFC, antral follicle count; AMH, anti-Müllerian Hormone; eCRF, electronic case report form.

Table 2. Types of chemotherapy and radiation therapy with estimated risk of infertility due to gonadotoxic injury as described by Nyström et al. (21).

5-FU, 5-fluorouracil; BCNU, carmustine; CCNU, lomustine; Gy, Gray; HSCT, hematopoietic stem-cell transplantation; RT, radio therapy; SCT, stem-cell transplantation.

Table 3. Treatment risk of infertility in girls based on IGHG recommendations (10).

CED, Cyclophosphamide equivalent dose; HSCT, Hematopoietic stem cell transplantation.

Medical measures

All girls had an initial assessment, including the collection of a blood sample, at the pediatric oncology center in Gothenburg, Sweden. Data collected included gynecological history on pubertal development, menarche, and menstrual flow pattern (if applicable). Data were also collected on weight and height. The aim was to analyze serum AMH levels in units of μg/L (=ng/mL) before the start of cancer treatment in all participants. Bloods sampled for serum AMH were analyzed at the Clinical Chemistry laboratory at Sahlgrenska University Hospital, Gothenburg, using the Chemiluminescent Microparticle Immunoassay (CMIA) method to quantify AMH levels. The detection limit at this laboratory is 0.02 μg/L, with values < 0.02 μg/L reported as “ < 0.02 μg/L.” Although data on normal AMH levels are girls are limited, we used previously proposed cut-off values of < 1.0 μg/L to denote reduced ovarian reserve, and values < 0.025 μg/L to predict impending ovarian insufficiency (22). These values are exploratory in this population of pediatric patients and not clinically validated. In cases where cancer treatment posed a significant risk for future infertility, fertility preservation was offered, if possible, based on clinical status, following a counseling session with the patient and their guardian. An electronic case report form (eCRF) regarding demographic data, type of cancer, treatment protocol, cumulative doses of chemotherapy, radiation field and dose, and gynecological history was completed retrospectively at inclusion and will be prospectively completed for follow-up visits.

Statistical analyses

The study group characteristics at baseline were described using descriptive statistics with mean, standard deviation (SD), median, and range for continuous variables, and number and percentage (n, %) for categorical and dichotomous variables. The distribution of AMH was skewed to the right and a log-transformation was applied to obtain an approximate normal distribution. To determine the functional form of the age dependence in the two treatment groups we performed linear regression of log (AMH) on cubic polynomials stratified by treatment group. Furthermore, regression models with linear age dependence were performed, including an interaction for age and cancer treatment. Further adjustment included Tanner stage and days since start of treatment. The resulting beta-values b were converted to [exp(b)−1] x 100% to obtain the effect estimate as a percentage change per unit of the independent variable on the original AMH scale.

Results with a p-value of < 0.05 were considered statistically significant. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS), version 25 (IBM), SAS v.9.4, and MATLAB R2025b.

Results

A total of 130 girls were eligible for participation (128 with pediatric cancer, one patient with Langerhans histiocytosis, and one with aplastic anemia). Three could not participate due to the language barrier, nine girls had treatment started during weekend where biobanking was not possible and time for information about the study was limited. Five girls were too ill at diagnosis to be included. Twenty-six patients were not enrolled for unknown reasons. A total of 87 girls aged between 0 and 17 years were included.

After initial assessment of the data, one participant had no data recorded (unknown reason) and was therefore excluded. Two participants were discovered to have been planned for surgical management only, another two had recurrence of pediatric cancer initially diagnosed many years prior and were therefore also excluded. Eight of the remaining participants had no data of AMH at any stage and were also excluded (Figure 1). The ages and cancer diagnoses of the remaining 74 girls comprising our cohort are shown in Table 4. Two patients were offered fertility preservation before initiation of cancer therapy; one had oocyte cryopreservation and the other had an oophorectomy and OTC.

Table 4. Age and cancer diagnoses distribution.

^*one patient had aplastic anemia.

AMH levels

The initial AMH levels of participants in the study group were between 0.03 and 9.97 μg/L. The mean value was 1.7 μg/L and the median 1.1 μg/L. The standard deviation was 1.7 μg/L. The AMH values were further divided into five groups: below 0.025, 0.025–0.50, 0.51–1.00, 1.00–1.50, and above 1.50 μg/L. No participant had a value below 0.025 μg/L, and most participants had values above 1.50 μg/L. The baseline data shows that 34 (46%) of the girls had AMH values above 1.50 μg/L. Another 7 patients had AMH levels above 1.00 μg/L, meaning that 41 participants (55%) had an AMH above 1.00.

Of the participants, 56 (76%) had blood samples taken before initiation of cancer therapy and 18 (24%) had blood samples taken after initiation of therapy. Of the 18 patients that had AMH assessed after initiation of cancer treatment, the time after treatment ranged from 1 day to 20 months. We further analyzed two subgroups, the “pre-treatment” group with AMH levels taken prior to cancer therapy and the “treatment exposure” group with AMH after treatment started. In the pre-treatment group, 30 patients had AMH above 1.50 μg/L, 2 patients had AMH between 0.025 and 0.50 μg/L and 15 patients had AMH levels in the range 0.51–1.00 μg/L. In the treatment exposure group, 12 patients had values in the range 0.025–0.50 μg/L and 4 participants had values above 1.50 μg/L.

Characteristics of cancer patients by time since treatment start

Table 5 shows the characteristics of patients by category of time since start of treatment. Most patients had not yet started treatment (days = 0, n = 56). Mean AMH values were larger for patients not yet treated or within 1 week of treatment start compared to those with at least 2 weeks of treatment, but the age differences between the categories were large.

Table 5. Characteristics of participants by time since treatment start (0 = not yet started treatment).

^a67 girls assessed for Tanner stage.

Age dependence in the two treatment groups

Polynomial regression models for log (AMH) showed that the age dependence was linear in both the pre-treatment (n = 56) and treatment exposure (n = 18) groups. A linear regression model for log (AMH) vs. age showed no difference in age dependence between the treatment groups (p-value for interaction between age and treatment group = 0.4). Upon back-transformation to the original scale, mean AMH increased by 6.4% per year of age, independent of treatment status. Age-adjusted AMH values were on average 79% lower in patients who had already received treatment compared to the untreated group (Table 6, model 1).

Table 6. Associations between AMH and age, treatment status, Tanner stage, and time since start of treatment.

Linear regression of log AMH on age, treatment status, Tanner stage, and time since start of treatment; results are given as exp(beta-value) × 100% with 95% confidence intervals.

^aPre-treatment vs. Treatment exposure.

^*Adjusted coefficient of determination.

Influence of Tanner stage and days since start of treatment on AMH levels

Tanner stages were documented in 67 patients. Only eight girls had Tanner stage 4–5. Tanner stages 4–5 (n = 8) vs. 1–3 (n = 59) were not associated with AMH levels in a model adjusting for age and treatment status (Table 6, model 2). Patients before or within 1 week of starting treatment had higher AMH levels than those treated for more than 1 week (Table 6, model 3). There was no difference in AMH levels between girls not yet treated and those in the first week of treatment (p = 0.7). AMH levels were lowest during months 1–3 of treatment and increased after 3 months, but differences between patients treated for more than a week were not statistically significant (Figure 2). Two of the four patients in the group who were treated for more than 90 days had already completed treatment.

Age dependence of AMH levels by “pre-treatment” and “treatment exposure” status: comparison with published results for girls free of cancer

The regression line in the pre-treatment group (n = 56) was below but close to published data for healthy girls (Figure 2). In contrast, the mean values for girls in the treatment exposure group lay below the mean values for healthy girls of all ages, but the total number of observations in this group was small (n = 18). We note that the two observations with AMH >3 μg/L in the treatment group belonged to patients who had just started treatment, with days after treatment start were 1 and 5 days respectively.

Discussion

As part of a long-term follow-up study, this study presents baseline AMH data from a cohort of girls aged 0–17 years old diagnosed with cancer and assesses its potential as a prospective method for fertility surveillance in this patient group. Clinical effectiveness, including fertility preservation decisions and reproductive outcome, will be followed up over several years and reported in future studies. In the context of oncofertility practice, AMH could serve as a tool for identifying and counseling CCS at high risk of reduced fertility and referring them for fertility preservation in time.

Although 45% of the cohort had AMH values below 1.00 μg/L, it is difficult to conclude that this meant a reduced ovarian reserve due to the limited data on normal range values for AMH in children. The girls who had already started cancer therapy had significantly lower AMH mean values than the healthy girls in the model, which is likely due to the effects of the treatment.

The ages of the participants show that it was a good spread across all ages with a mean age of 7.7 years. This shows that children of all ages had been included from the Pediatric Oncology Center and decreases the selection bias, including the fact that the question of fertility could be perceived difficult to raise in cases with very young children diagnosed with cancer compared to a teenager, post-menarche.

Almost one-fourth of the participants underwent blood sampling after initiation of cancer treatment. It is evident that an ideal scenario would involve an assessment of AMH prior to the commencement of cancer treatment. The underlying cause of the delay in the collection of blood samples may be due to various logistic considerations. These include blood sampling conducted at a local hospital, the potential oversight of a request for AMH, or an inability to delay cancer treatment.

In contrast to Kelsey et al. we find a linear dependence of log(AMH) on age in the range 0–17 years (11). We acknowledge that our sample size is considerably smaller than the published data source (n = 1,101 for age = 0–17 years), and we may not have the statistical power for a regression model with higher-order terms in age. However, we note that the local maxima in the prediction line by Kelsey et al. are likely to be a consequence of the fact that the data were fitted by a polynomial of degree 20 to describe data for age 0–51 years, which means that these wiggles are not real. This hypothesis is supported in later publications that show a considerably smoother age dependence, as seen in van Dorp et al. (15), Cui et al. (14) and in our study. Two of our patients in the treatment exposure group had high AMH levels even with treatment started. One of these patients was 1 day into treatment and the other 5 days into treatment. The underlying reasons for the relatively high AMH values remain uncertain, but may be due to individual factors, the nature of the treatment, or the fact that measurements were performed early in the treatment course. In 2014, van Dorp et al. (15) showed that AMH levels in girls with newly diagnosed cancer were reduced compared with healthy controls even before cancer treatment had started. The regression line in our analysis shows that AMH values for cancer patients before treatment lie below, but close to, published data for healthy girls, a similar effect to that shown in van Dorp et al. As AMH values are typically right-skewed, with mean values larger than medians, the contrast between our group of untreated patients and girls free from cancer is probably even larger. However, the lack of data on AMH in girls without cancer made it impossible to make a statistical comparison between the groups.

In the recommendations by IGHG from 2016, the diagnostic and prognostic values of AMH to predict POI lack evidence due to insufficient studies. AMH is hence not recommended as the primary surveillance modality for evaluation of POI in survivors treated with potentially gonadotoxic chemotherapy and/or radiotherapy who desire assessment about future fertility (19). Nevertheless, later studies, such as one by Nyström et al. (23), suggested that AMH could be used as a marker for ovarian reserve in CCS. AMH is an advantageous hormone because of its stability in the menstrual cycle and it has a higher sensitivity than follicle stimulating hormone (24). Assessing ovarian reserve by measuring AFC in young girls and adolescents with transvaginal ultrasonography is both physically and ethically inappropriate, making AMH a better and more cost-efficient method. AMH is therefore promising as a potential primary modality for fertility surveillance pending longitudinal validation.

This study is important due to the increasing population of CCS reaching reproductive age and contemplating the prospect of starting a family. It has a unique prospective design, following the participants from cancer diagnosis through their adolescence and up to fertile age, assessing AMH levels, and trying to identify who would be likely to benefit from fertility preservation, and when. Prospective follow-up of health questionnaires will enable the assessment of emerging symptoms of anxiety and/or depression among the CCS group of girls and women. In addition, the increased knowledge about parental views of the question about their daughters' fertility will provide parents with better support to handle this question not only at the time of the illness but during the upbringing of their children. The study will also contribute to further knowledge on the relationship between AMH and the ovarian reserve in girls and adolescents with pediatric cancer.

The proposed study protocol is subject to several limitations, including the duration of the prospective follow-up period and its feasibility. Although our participants have a wide range of diagnoses and good span across ages 0–17, the cohort size is a limitation, especially due to risk of disease progression. It would have been desirable with a larger cohort from a perspective of generalizability; however, it was not practically feasible due to the limited numbers of new cases of pediatric cancer each year in our catchment area. The participants are children now, and the likelihood of them moving away or getting lost to follow-up cannot be disregarded. Continuity of staff during the time frame is an obvious hazard and so is the enthusiasm for the study of both participants, their guardians, and their treating clinicians. Another limitation is the lack of a healthy control group for comparison and with that the missed opportunity to control for confounders.

Conclusion

Our study proposes and investigates the feasibility of a protocol for the prospective monitoring of ovarian reserve in girls diagnosed with cancer, utilizing AMH, and starting from the time of diagnosis. In this study we have presented the baseline data of our cohort. AMH organized surveillance may facilitate the identification of individuals requiring fertility cryopreservation at a later stage in their lives, thereby serving as a model for ovarian reserve surveillance within this cohort of patients. By also investigating the psychological effects of monitoring fertility during puberty and young adulthood this could hopefully be of benefit to the individual and help health care facilities to allocate resources more efficiently.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Vesna Zadnik, Institute of Oncology Ljubljana, Slovenia

Reviewed by: Pegah Rashidian, Tehran University of Medical Sciences, Iran

Marta Marino, Università della Valle d'Aosta, Italy