Assisted reproductive technology use and outcomes in childhood cancer survivors.

The CCSS is a multi-institutional retrospective cohort study with longitudinal follow-up of survivors of childhood cancers (acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), other leukemias, astrocytoma, medulloblastoma, other central nervous system (CNS) malignancy, Hodgkin lymphoma (HL), Non-Hodgkin lymphoma (NHL), kidney tumors, neuroblastoma, soft tissue sarcoma, Ewing sarcoma, osteosarcoma, or other bone malignancy) diagnosed prior to age 21 at one of 31 institutions in the U.S. and Canada. Eligible survivors were diagnosed with cancer between January 1, 1970 and December 31, 1999 and survived at least five years after diagnosis. Siblings of a random sample of survivors were recruited to serve as a comparison group. The design and methods of the study have been previously described. 14 , 15 The institutional review boards of the participating institutions approved the CCSS protocol, and participants provided informed consent. SART CORS includes comprehensive data collected through voluntary submission, verified by SART, and reported to the Centers for Disease Control and Prevention (CDC) in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493). In 2004, following a contract change with the CDC, SART gained access to the SART CORS data system for the purposes of conducting research. In 2019, 81% of clinics were SART members reporting 90% of all IVF cycles in the U.S. 16 Further details regarding SART CORS can be found in the Supplement . For this study, a cohort of 9,885 female survivors and 2,419 siblings from the CCSS were identified as eligible for linkage and included only women who were of reproductive adult age (18–55 years) during the SART CORS data collection time frame of 2004–2015 and excluded Canadians. Of female survivors in the CCSS cohort, 95% were of reproductive age for at least a year of follow-up. Of a potential 182,215 person-years of follow-up during childbearing age, 90,831 (50%) were during 2004–2015. We used previously published algorithms for the linkage process. 17 , 18 The algorithm used a woman’s date of birth, last name, first name and social security number to match repeat cycles within a single clinic and between clinics and to probabilistically link these SART CORS records to the CCSS records. Validity checks were run on potential matches and reviewed for accuracy. Detailed information on cancer treatments received within five years of initial diagnosis was abstracted from medical records at each CCSS institution, including chemotherapy cumulative doses and body region-specific radiation dosimetry. 15 Alkylating agents were summarized as cyclophosphamide-equivalent dose and anthracyclines were standardized as doxorubicin-equivalent dose. 19 , 20 Detailed procedures for ART are provided in Supplement 1 . Information on patient characteristics (demographics, cancer characteristics and treatment exposures for survivors) and ART treatment was summarized for survivors and siblings, including age at time of retrieval, race, maximum FSH, and BMI. Characteristics were compared between survivors who received ART and those who did not, as well as comparisons between survivors and siblings who received ART using chi-square tests. The primary outcome for the analysis was any live birth, analyzed as a percent of all ovarian stimulation cycles and referred to as live birth rate. Live birth rate was evaluated separately for women who used autologous oocytes and those who used donor oocytes. To examine success per intervention, cycle information was summarized for each ovarian stimulation per woman, and overall success per patient was summarized as number of live births. Among survivors, information was evaluated by cancer diagnosis group, whether the patient had undergone hematopoietic cell transplant (HCT), and by cumulative dose of alkylating agents. Multiple imputation was used to populate missing treatment data, summarizing 10 different sets of treatment information. Generalized linear models with Poisson distribution and a log link were used to model the probability of a live birth per stimulation cycle. We assessed the relative risk of live birth in all survivors, as well as mutually exclusive categories of survivors defined by treatment exposures in comparison to siblings. In addition, a comparison to survivors not exposed to radiation or alkylating chemotherapy was performed. Each model was a priori adjusted for patient age at the time of oocyte retrieval. We examined BMI in a sensitivity analysis within the subset of participants for whom it was available due to only 60% with available values at the time of the stimulation cycle and limited utility of BMI data available at up to 5 years prior to ART in an additional 25% of participants. BMI was not identified to be a significant predictor nor was it an important confounding variable and was not included in final models. Relative risks (RR) and 95% confidence intervals (CI) were reported. Models were evaluated among all stimulations for women with only autologous cycles, and a robust sandwich error was included to account for multiple stimulations per participant.

A total of 9,885 female childhood cancer survivors and 2,419 siblings met reproductive age criteria during the window of availability of SART CORS data. After linkage, 137 survivors (1.4%) and 33 siblings (1.4%) were identified as participating in ART. Personal, sociodemographic and cancer treatment characteristics of the female childhood cancer survivors and siblings are summarized in Table 1 . With respect to non-treatment factors, among survivors, those receiving ART were more likely to have tried to get pregnant for more than one year without success, have normal BMI, have higher household income, have higher attained education levels, live in a state which had implemented an IVF insurance mandate and never smoked. The most common infertility diagnosis among survivors undergoing ART was diminished ovarian reserve (54, 39.4%), followed by male factor (29, 21.2%) and “Unexplained” (14, 10.2%). ( Supplemental table 2 ). Among the 137 survivors who underwent ART, self-report CCSS questionnaire data revealed 100 (73%) with normal ovarian function, 10 (7.3%) with acute ovarian failure, 9 (6.6%) with premature menopause, 8 (5.8%) with surgical premature menopause, 6 (4.4%) with acute ovarian failure prior to menarche, and 4 (2.9%) with unknown ovarian function. For siblings, the most common infertility diagnoses included diminished ovarian reserve and male factor (9, 27.3%) followed by polycystic ovarian syndrome (6, 18.2%). Other ART indications included endometriosis, tubal factor, uterine factor, unexplained infertility, and other. In the survivor group, 137 women underwent 224 ovarian stimulations involving donor or autologous eggs, which included 155 fresh autologous ovarian stimulation cycles and 88 frozen or donor embryo transfer cycles ( Figure 1 ). Thirty-seven siblings underwent 47 fresh autologous ovarian stimulation cycles, 4 donor ovarian stimulation cycles, and 13 frozen embryo transfer cycles. The median age at time of egg retrieval in the fresh autologous ovarian stimulation cycle was 30 years (range 19–44 years; Table 2 ) in survivors compared to 34 years (range 24–43 years) in the sibling group. Ovarian stimulation and transfer cycle characteristics are further outlined in Table 2 . Ninety-nine women underwent ovarian stimulation with autologous oocytes and two used a gestational carrier ( Table 2 ). Neuroblastoma survivors had the highest percent of live births [89% (8/9)], followed by bone cancer [64% (7/11)], soft tissue sarcoma [50% (4/8)], CNS tumor [45% (5/11)], NHL [43% (3/7)], leukemia [41% (9/22)], HL [44% (7/16)], and kidney tumors [38% (5/13)] ( Supplemental Table 1 ). Overall, the likelihood of live birth after autologous egg, non-gestational carrier ART cycles was not statistically different in survivors as compared to siblings (RR=0.72; 95% CI −0.46 – 1.11; Table 4 ). Survivors who received pelvic RT had a lower likelihood of live birth compared to siblings, even after adjusting for age of the woman at the time of the egg retrieval (RR=0.33; 95% CI - 0.15 – 0.73). Cranial RT exposure was also associated with a lower likelihood of live birth, both before and after adjusting for oocyte age (RR= 0.44; 95% CI- 0.20– 0.97). Exposure to alkylating agents did not significantly reduce live birth rate among survivors compared to siblings ( Table 4 ) or when compared to survivors who received no alkylating agents ( Table 4 ). A total of 14 survivors had a history of TBI or HCT. None of the six survivors who were previously treated with TBI conceived, regardless of whether autologous or donor oocytes were used. Six survivors with a history of HCT without TBI underwent 10 embryo transfers, all using donor oocytes and none using gestational carrier. Three of these survivors with prior autologous HCT had live births. A subset of survivors and siblings pursued treatment with donor oocytes. Sixty-seven cycles in 38 survivors and four cycles in one sibling involved donor oocytes. Among the eight patients who had received pelvic RT and pursued pregnancy with donor oocytes, only one had a live birth (pelvic RT dose 20 Gy). None of the five women who had received pelvic RT with a dose >20 Gy had a live birth using donor oocytes. Six survivors and no siblings used gestational carriers, with the four survivors described above using donor oocytes. The other two survivors did not have prior HCT or TBI but used autologous oocytes to create embryos for transfer into gestational carriers. One woman with a history of Wilms had one embryo transferred resulting in a live birth. One woman with HL transferred an embryo, but no pregnancy resulted.

This study showed that childhood cancer survivors are as likely to utilize ART as siblings. Survivors with prior pelvic RT experienced markedly decreased chance of live birth after ART and none of the patients with prior TBI had a live birth. Thus, gestational surrogacy might be considered after pelvic/abdominal RT or TBI. Because exposure to alkylating agents was not associated with decreased odds of live birth, women with prior exposure to alkylating agents alone should be considered candidates for ART treatments using autologous oocytes.

To our knowledge, this is the first study to objectively examine both ART utilization and ART outcomes in female childhood cancer survivors. In our study, 1.4% of childhood cancer survivors or siblings utilized ART at some point in their studied reproductive period. Although the overall likelihood of live birth was not statistically significantly different between survivors and siblings, survivors treated with pelvic RT and cranial RT had a decreased chance of live birth after ART compared to siblings. Exposure to alkylating agents in childhood cancer survivors undergoing ART was not associated with a lower likelihood of live birth. This is unexpected because alkylating agents decrease ovarian reserve and ovarian reserve determines the number of eggs available for recruitment in each ovarian stimulation cycle. Thus, when ovarian reserve is decreased, there are fewer eggs to create embryos per ovarian stimulation cycle and thus, fewer chances at pregnancy per cycle. Despite the overall low utilization noted in our study, others have reported that childhood cancer survivors desire children as much as the general population. A recent consensus publication from the International Late Effects of Childhood Cancer Guideline Harmonization Group emphasized that even prior to cancer treatment, parents and patients with childhood, adolescent, and young adult cancers strongly desire information about the effects of cancer treatment on fertility and options for fertility preservation and thus, this should be part of the pretreatment discussion, especially in the setting of alkylating agent doses of at or greater than 6000–8000mg/m2, HSCT, and/or pelvic RT. In a 2018 study, van Dijk et al 23 examined data from the DCOG LATER-VEVO study, a nationwide retrospective cohort study of female fertility among Dutch childhood cancer survivors and found that female survivors were as likely as the control group (survivors sisters and females from the general population) to desire children, but more likely to consult a fertility specialist (12% CCSs vs 10% controls (OR = 1.7, 95% CI: 1.3–2.4). The finding of similar utilization of ART treatments among survivors and siblings in this database linkage study contrasts with the retrospective, questionnaire data presented in a prior study of fertility treatments in CCSS patients. Using self-reported survey data, Barton et al in 2013 13 compared infertility, infertility treatment and pregnancy outcomes of 3,531 CCSS women and 1,366 sibling female controls. Survivors reported an increased risk of infertility at earlier ages compared to sibling controls. Both survivors and siblings were equally likely to seek infertility treatment (315/455, 69.2% vs. 100/137, 73.0%), but survivors were less likely to be prescribed infertility medication (87/208, 41.8% vs. 56/75, 74.7%) which suggests a lower utilization of ART. This study was limited by its utilization of self-reported data, lack of details regarding IVF treatment, and included only the original CCSS cohort of women diagnosed between 1970 and 1986. The first birth from IVF in the US was in 1981, so IVF was less available to the participants in the original CCSS cohort. A second study published by Melin et al in 2020 21 compared utilization of fertility treatment by Finnish adult and childhood cancer survivors to siblings. This study found higher fertility treatment utilization by survivors compared to siblings. The authors employed population-based national registries for population, birth, cancer, and prescription medication to determine the utilization of prescription drugs related to ovulation induction or ART. However, the majority of data came from adult cancers diagnosed between age 25 to 40 years. Of 3,650 survivors age 0 to 24 years at cancer diagnosis, 93 underwent ovulation induction treatment and 94 underwent ART treatment. Additionally, the study examined fertility treatment utilization but not outcome. The results of our study may differ from those of previous studies due to more stringent reporting requirements by ART programs, which would not include patients evaluated but determined to be poor ART candidates. It is also possible that ovulation induction may have resulted in pregnancies for some survivors, ART unnecessary. However, the small number of childhood cancer survivors utilizing ovulation induction medication in the Melin et al study argues against ovulation induction (OI) success as a significant factor in lowering ART utilization by survivors. Perceived impaired fertility by both survivors and fertility providers may account for under-utilization by childhood cancer survivors. This is supported by studies that have noted patient perceptions of fertility in childhood cancer survivors is worse than actual ovarian reserve testing. 22 , 23 Additionally, childhood cancer survivors, particularly those with a history of CNS tumors are less likely to partner or marry when compared to siblings or the general population. 24 CNS tumor survivors are disproportionately missing from the ART group compared to the general CCSS cohort. Although this may be due to chance in the overall small sample size, it also may be because CNS tumor survivors are even less likely to partner or marry than those with other cancer diagnoses, thus would be less likely to pursue family building, with or without ART. 24 , 25 Our study did not show a significant effect of alkylating agent chemotherapy on ART outcomes. In a prior CCSS report, 8 the authors found no association between cyclophosphamide equivalent dose and fertility among female childhood cancer survivors, either by tertile of equivalent dose (upper tertile vs no exposure) or by dose linear model. In another study from the CCSS, 2 pregnancy and live birth rates were lower among women experiencing nonsurgical premature menopause been the ages of 31 and 40 years, but not between ages 21 and 30 years. Alkylating chemotherapeutic agent exposure has been implicated in lower age of menopause for both female childhood and young adult cancer survivors. 26 , 27 Of note, only twelve survivors received no RT with CED ≥8 grams/m2 in our study, thus the lack of association between high dose alkylator exposure and live births could be due to our very small sample size. Additionally, the survivors with alkylator exposure who experienced the sharpest decrease in ovarian reserve may have been counseled not to pursue ART, thus our cohort may not be representative of the overall population of survivors with alkylator exposure. Our study adds to the body of literature that illustrates lower pregnancy rates in women who had pelvic radiation, though this is the first to specifically demonstrate this in ART. Notably, none of the six survivors with history of TBI had a live birth using autologous eggs. A metanalysis 6 demonstrated a significant association between radiation exposure and infertility (RR=1.28, 95% CI 1.6–1.42), acute ovarian failure (RR=9.51, 95% CI 5.03–17.96), and low anti- Mullerian hormone (AMH) (<1ng/mL) (RR=14.79, 95% CI 3.36–66.64). A more recent study, 28 illustrated in a Finnish cohort that childhood cancer survivors with prior pelvic RT and childhood cancer survivors without prior pelvic RT had smaller uterine volumes than the general population. Survivors with prior pelvic RT had an increased risk of lower birth weight and preterm delivery along with the small uterine volume. Cranial RT decreased the likelihood of live birth in our study, despite the use of ART. The mechanism by which cranial RT is proposed to affect fertility is by damaging GnRH neurons and/or anterior pituitary cells resulting in irregular ovulation due to hypogonadotropic hypogonadism. Ovarian stimulation with exogenous gonadotropins treats this mechanism of infertility and thus should improve chance of live birth in those with cranial RT to make the chances comparable to those without cranial RT history. However, prior studies by Green et al have shown increased miscarriages in childhood cancer survivors with a history of prior cranial RT. 29 , 30 The proposed mechanism for this observation is decreased endogenous gonadotropins needed to stimulate progesterone production from the corpus luteum, the primary source of progesterone in the first 7 to 9 weeks of pregnancy. After 9 weeks, the placenta takes over progesterone production. This is unlikely to be a contributing mechanism in our population because all ART treatment involves exogenous progesterone supplementation through 10 weeks of pregnancy. Thus, there are likely other factors contributing to the decreased likelihood of live birth with cranial RT that have yet to be elucidated. There are a few limitations of our study. First, the SART CORS dataset began in 2004, covering only 50% of the Childhood Cancer Survivor Study cohort’s female survivor reproductive years (aged 18–55 years), so survivors who used ART prior to 2004 are not captured in this analysis. Also, we did not address the utilization of non-ART infertility treatments because there is no central reporting mechanism to collect these data in the US. Although, to our knowledge, this is the largest objective study of ART outcomes in childhood cancer survivors, the small sample size limited the statistical power to detect a difference in live birth rates between survivors and siblings, especially in the high-dose alkylator exposure group, a factor known to affect nonART pregnancy outcomes. 8 There may be additional survivors who utilized ART outside the 2004–2015 period in which SART CORS data was available for matching, so the focus of this analysis was therefore on live births per stimulation cycle and not on likelihood of utilizing ART. As US ART utilization continues to increase over time, future studies will be able to better assess CCS ART utilization. Additionally, body mass index (BMI), an important predictor of fertility, 31 and congenital anomalies in offspring, a previously reported risk associated with ICSI only in ART, 32 were not reliably reported in our SART CORS dataset.

Over 85% of children diagnosed with cancer will become five-year survivors. 1 Understanding the impact of cancer treatments on their future fertility is an urgent priority. Prior studies have demonstrated that both chemotherapy and radiation therapy (RT) can lead to premature ovarian insufficiency or diminished ovarian reserve, even when regular menses return post-treatment. 2 – 4 Decline in ovarian reserve occurs in a dose-dependent manner with both RT and alkylating chemotherapeutic agents. 5 , 6 Additionally, pregnancy rates for both female and partners of male childhood cancer survivors are affected by the type and dose of chemotherapy administered or total dose of radiation exposure. 7 , 8 Previously, limited data have suggested that female cancer survivors seek treatment from an infertility specialist sooner and at a younger age than siblings and age-matched controls. 9 Subsequently, they may face a higher risk of infertility and poorer in vitro fertilization (IVF) outcomes when they seek infertility treatment. 10 – 13 Assisted reproductive technologies (ART) are infertility treatments that remove eggs and sperm from the body, create embryos in the laboratory, and transfer the embryo into the patient or a gestational carrier’s uterus to achieve pregnancy. Although two prior studies examined ART utilization among female childhood cancer survivors using self-reported questionnaires, 9 , 13 none assessed post-treatment ART outcomes. We sought to determine the utilization and success rates of ART, as well as pregnancy outcomes, in childhood cancer survivors by linking the Childhood Cancer Survivor Study Cohort (CCSS) and the Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS), a nationwide, clinic-based registry that coordinates mandatory reporting to the Center for Disease Control (CDC). The objective was to investigate ART cycle characteristics, including live birth outcomes of childhood cancer survivors who underwent ART compared to siblings who also underwent ART.