Fax: (212) 746-8848
Quantitative assessment of the impact of chemotherapy on ovarian follicle reserve and stromal function
Version of Record online: 11 OCT 2007
Copyright © 2007 American Cancer Society
Volume 110, Issue 10, pages 2222–2229, 15 November 2007
How to Cite
Oktem, O. and Oktay, K. (2007), Quantitative assessment of the impact of chemotherapy on ovarian follicle reserve and stromal function. Cancer, 110: 2222–2229. doi: 10.1002/cncr.23071
- Issue online: 2 NOV 2007
- Version of Record online: 11 OCT 2007
- Manuscript Accepted: 13 JUN 2007
- Manuscript Revised: 11 JUN 2007
- Manuscript Received: 15 MAR 2007
- alkylating agents;
- primordial follicle;
- stromal function;
- histologic assessment
Various chemotherapy agents, especially of the alkylating category, have been associated with premature ovarian failure but there is no quantitative evidence of chemotherapy-induced ovarian damage in humans. The aim was to quantify the impact of chemotherapy on primordial follicle reserve and stromal function in human ovary with a prospective controlled quantitative histologic and in vitro study.
Samples from 26 patients who were undergoing ovarian tissue cryopreservation for fertility preservation were assessed histologically and for in vitro estradiol production. Of the 26 patients, 10 had previously received chemotherapy whereas 16 had not (control). There were 17 age-matched patients. Primordial follicle counts and in vitro estrogen production were compared between age-matched control and chemotherapy patients as well as between those who did and did not receive alkylating agents.
Patients who received chemotherapy had significantly lower primordial follicle counts than control patients (5.4 ± 1.32 vs 9.6 ± 2.2, P = .05). Furthermore patients treated with alkylating regimens had significantly lower primordial follicle counts compared with those who received nonalkylating agents (2.9 ± 1.1 vs 7.9 ± 1.6, P < .05) and with those who did not receive any chemotherapy (2.9 ± 1.1 vs 9.6 ± 2.2, P < .05). In vitro, ovarian cortical pieces from individuals who were previously exposed to chemotherapy (chemotherapy group) produced significantly less estradiol compared with those who were not (control group). In the chemotherapy group, there was no difference in in vitro estradiol production between those who received alkylating agents and those who did not.
This is the first quantitative evidence of the impact of chemotherapy on ovarian infrastructure, and shows that, whereas alkylating agents can significantly reduce ovarian reserve, both alkylating and nonalkylating regimens may affect ovarian stromal function. Cancer 2007. © 2007 American Cancer Society.
One of the major quality of life issues for young cancer survivors is preserving gonadal function and fertility. On the basis of clinical studies, it is known that the alkylating agents are associated with the highest risk of infertility.1 Nevertheless, nearly the entire body of information on the impact of cancer treatments on human fertility is based on the crude assessment of menstrual function, which is not a sensitive marker of fertility.2 Although animal studies have demonstrated the impact of these agents on primordial follicle reserve,3, 4 there could be significant discrepancies between the ovarian biology of rodents and humans. Furthermore, based on studies in the rodent, the main mode of infertility and ovarian failure by chemotherapy is believed to be through follicular destruction, but the impact of cancer treatments on ovarian stromal cells has not been studied. Because ovarian stromal cells play an important role in ovarian endocrine function, and speculatively in post-chemotherapy damage repair,5 assessment of chemotherapy-associated damage is important.
The main aim of this study was to quantify and compare the impact of chemotherapy on primordial follicle density in human ovarian cortical pieces obtained from cancer patients who were undergoing ovarian tissue cryopreservation before or after receiving chemotherapy. Furthermore, we ascertained whether the stromal function is altered in patients who were exposed to chemotherapy by ovarian estradiol production in vitro.
MATERIALS AND METHODS
Ovarian samples were obtained from 26 patients who were undergoing ovarian cryopreservation for fertility preservation. The research protocol was approved by the Institutional Review Board of Cornell University. None of the subjects had an underlying metabolic or endocrinologic illness, used any hormonal treatment, smoked, or had ovarian metastasis. Twenty-three subjects were diagnosed with various types of solid and hematologic cancers, of whom 13 had not yet received any chemotherapy (control), whereas 10 had prior chemotherapy. Of those 10 patients an equal number received alkylating and nonalkylating regimens. The remaining 3 had various hematologic illnesses and had undergone tissue sampling before myeloablative conditioning for hematopoietic stem cell transplantation.
Alpha-MEM culture medium and fetal bovine serum were purchased from Gibco (Invitrogen, Carlsbad, Calif). Recombinant follicle-stimulating hormone (FSH) was from Serono (Rockland, Mass). The estradiol kit was from DPC (Diagnostic Products, Los Angeles, Calif).
From each subject, 1 5 × 5 × 3 mm ovarian cortical piece was fixed overnight, paraffin-embedded, and serially sectioned at 7 μm thickness to determine primordial follicle density. primordial follicles were counted in every fifth section (35 μm apart) using an Olympus BX-41 microscope under × 200 magnification (11–28 sections per sample). To avoid duplicate counting in each section, only the primordial follicles with visible oocyte nuclei were recorded. Follicle density was determined per mm2 of tissue surface and a mean value was obtained from all sections.
Ovarian Tissue Culture
One ovarian cortical piece from each subject was cultured in alpha-MEM culture medium supplemented with 10% fetal bovine serum and 300-mIU/mL recombinant FSH for 7 days. Due to tissue limitation, ovarian cortical cultures could be performed in only 18 subjects. Half of the culture medium was removed and replaced daily and stored at −80°C until assayed for estradiol (E2) by radioimmunoassay (RIA). primordial follicle density was evaluated in serial sections before and after the 7-day culture.
Analysis of variance (ANOVA) or Mann-Whitney U-tests were used in comparing means of various groups, where appropriate. Spearman test was used for correlation analyses (SPSS v. 11, Chicago, Ill). Graphic bars and curves were created using GraphPad Prism software (San Diego, CA). A P-value ≤.05 was considered significant.
Validation of Follicle Counts
Patient ages, chemotherapy protocols, and their primordial follicle counts are shown in Table 1. The mean age ± standard error (SE) of all patients was 27 ± 2 with a range of 4 to 44. The baseline follicle density inversely correlated with age in both controls and chemotherapy patients (r = −0.98, P < .0001, and r = −0.82, P < .001, respectively). When all control and chemotherapy patients were analyzed together the statistical significance still persisted (r = −0.96, P < .0001, Fig. 1). The correlation of age with follicle density has been previously shown6 and validates the accuracy of our assessment of follicle counts.
|Patients||Age, y||Diagnosis||Chemotherapy||Follicle density|
|1||4||Diamond-Blackfan syndrome||—||25 ± 3.2|
|2||6||Thalassemia major||—||24 ± 1.87|
|3||10||Acute lymphoblastic leukemia||—||22.4 ± 5.7|
|4||17||Myelodysplasia||—||18.25 ± 5.17|
|5||18||Hodgkin disease||—||15.3 ± 1.8|
|6||18||Acute myelocytic leukemia||2 × ADE-GMTZ||14.4 ± 1.6|
|7||22||Hodgkin disease||—||16.6 ± 3.5|
|8||22||Non-Hodgkin lymphoma||1 × CHOP||6.17 ± 0.7|
|9||23||Hemophagocytic lymphohistiocytosis||4 × Etoposide+intrathecal MTX||6.1 ± 0.8|
|10||24||Hodgkin disease||6 × ABVD||7.6 ± 1.7|
|11||24||Synovial sarcoma||VACA+RT||4.52 ± 0.89|
|12||28||Acute myelocytic leukemia||Idarubicin cytarabine||6.77 ± 1.3|
|13||28||Hodgkin disease||—||6.83 ± 2.5|
|14||30||Breast cancer||—||7 ± 1.6|
|15||30||Ependymoma||—||6.75 ± 0.9|
|16||32||Hodgkin disease||6 × ABVD||4.8 ± 1.5|
|17||32||Breast cancer||4 × AC||2.37 ± 1.1|
|18||33||Breast cancer||—||5.66 ± 0.9|
|19||33||Breast cancer||—||5.23 ± 1.1|
|20||33||Non-Hodgkin lymphoma||7 × CHOP+GnRH agonist||1.5 ± 0.65|
|21||36||Breast cancer||—||0.79 ± 0.3|
|22||36||Breast cancer||—||0.55 ± 0.18|
|23||38||Hodgkin disease||1 × BEACOPP||0.22 ± 0.15|
|24||39||Breast cancer||—||0.15 ± 0.2|
|25||42||Ovarian mass||—||0.12 ± 0.1|
|26||44||Endometrial cancer||—||0.125 ± 0.1|
Impact of chemotherapy on primordial follicle density
Patients who received any chemotherapy (n = 16) versus those who did not (n = 10) had similar mean ages (26.7 ± 3.1 vs 27.4 ± 1.9, P = .15). When primordial follicle densities of these 2 groups were compared regardless of the type of regimen, those who received chemotherapy had significantly lower follicle counts (5.4 ± 1.32 vs 9.6 ± 2.2, P = .05).
Comparison of baseline primordial follicle density between age-matched control and chemotherapy patients
To further delineate the impact of individual regimens on ovarian follicle density we compared the follicle counts of those who were exposed to chemotherapy to age-matched controls, when possible. The effect of these chemotherapy regimens on ovarian reserve in comparison to age-matched controls is illustrated in Figure 2 and representative sections of these ovarian samples are shown in Figure 3. The youngest patient in the chemotherapy group was an 18-year-old with acute myelogenous leukemia (AML). This patient had already received 2 courses of a nonalkylating regimen (ADE-GMTZ; arabinoside-C 2000 mg/m2 every 12 hours for 5 days; daunorubicin 45 mg/m2/d for 2 days; etoposide 350 mg/m2 for 5 days; and gemtuzumab ozogamicin). Her baseline primordial follicle density was comparable to an age-matched patient with Hodgkin disease who had not yet received chemotherapy (14.4 ± 1.6 vs 15.3 ± 1.8, P > .05). A 22-year-old patient who had received 1 course of standard CHOP (cyclophosphamide 750mg/m2; doxorubicin 50 mg/m2; vincristine 1.4 mg/m2; and prednisone 40 mg/m2) for non-Hodgkin lymphoma had a baseline primordial follicle density significantly lower than the density of an age-matched patient who did not yet receive chemotherapy for Hodgkin disease (6.17 ± 0.7 vs 16.6 ± 3.5, P < .001). After the ovarian tissue sampling this 22-year-old patient received 3 additional courses of CHOP chemotherapy every 3 weeks, followed by 4 courses of ICE (ifosfamide-carboplatin-etoposide) treatment before hematopoietic stem cell transplantation (HSCT). At 6-month follow-up after the HSCT the patient had an FSH value of 67.3 IU/mL, confirming premature ovarian failure. In another patient, age 33, who had received 7 courses of standard CHOP for non-Hodgkin lymphoma, baseline primordial follicle density was significantly lower than the 2 age-matched control patients with breast cancer (1.5 ± 0.65 vs 5.23 ± 1.1, P < .01 and 1.5 ± 0.65 vs 5.66 ± 0.9, P < .01). Although this patient had received a GnRH agonist treatment (leuprolide acetate) during the courses of chemotherapy, her follicle density was significantly lower than the untreated controls (Fig. 3). Furthermore, despite the GnRH agonist use, her FSH was 20.8 mIU/mL on cycle Day 3 (upper limit of normal 11.8), consistent with compromised ovarian function.2
Of the 2 patients aged 24, the baseline primordial follicle density of the patient with Hodgkin disease who received a nonalkylating regimen (6 courses of ABVD: adriamycin 25 mg/m2; bleomycin 10 mg/m2; vinblastin 6 mg/m2; dacarbazine 375 mg/m2 on Days 1 and 15 repeated every 4 weeks) before ovarian sampling was significantly higher than a patient of the same age with synovial carcinoma who had received an alkylating regimen (VACA, cyclophosphamide 1200 mg/m2 weeks 1, 4, 7; doxorubicin 60 mg/m2 weeks 1, 7; actinomycin D 1.5 mg/m2 week 4; vincristine 1.5 mg/m2 weeks 1–4) plus inguinal radiotherapy (600 cGy) 10 years prior (7.6 ± 1.7 vs 4.52 ± 0.89, P < .05). This 24-year-old patient with synovial sarcoma of the left hip had received her chemotherapy when she was 14 years old and her current ovarian reserve was similar to that of a 33-year-old control (4.52 ± 0.89 vs 5.66 ± 0.9, P > .05). After the ovarian cryopreservation procedure the patient received 2 courses of ifosfamide-etoposide (ifosfamide 1800 mg/m2 and etoposide 100 mg/m2). Fifteen months after receiving this chemotherapy her FSH was elevated to 19.6 from the pretreatment level of 6.83 IU/mL.
A 32-year old patient who had received 6 courses of ABVD for Hodgkin disease had a significantly higher follicle count than an age-matched patient treated with 4 courses of AC (adriamycin 60 mg/m2 IV on Day 1, cyclophosphamide 600 mg/m2 IV on Day 1, a total of 4 cycles every 21 days) for breast cancer (4.8 ± 1.5 vs 2.37 ± 1.1, P < .05).
The baseline follicle density of a 28-year-old patient who had received cytarabine (100 mg/m2 per day by continuous infusion for 7 days) and idarubicin (9 mg/m2 3 times) as an induction therapy for AML before ovarian sampling was comparable to another age-matched control patient with Hodgkin disease (6.77 ± 1.3 vs 6.83 ± 2.5, P > .05).
Overall impact of alkylating agents on ovarian reserve
To further validate the matched comparisons, we compared the mean follicle densities between those who received alkylating agents and who those did not. The mean age difference between those treated with alkylating regimens (CHOP, AC, VACA, and BEACOPP) and those who had received nonalkylating regimens (ABVD, idarubucin-cytarabin, etopside-methotrexate, and ADE-GMTZ) did not reach statistical significance (29.8 ± 2.9 vs 25 ± 2.3, P > .05). However, the mean primordial follicle count of those who had received an alkylating agent was significantly lower (2.9 ± 1.1 vs 7.9 ± 1.6, P < .05). The primordial follicle density in those who received nonalkylating agents was similar to that of controls (7.9 ± 1.6 vs 9.6 ± 2.2, P > .05).
Assessment of the Impact of Chemotherapy on Ovarian Stromal Function
Comparability of ovarian culture samples in chemotherapy and control groups
Among the patients whose tissue samples were used for in vitro culture the mean age of patients, baseline and postculture follicle density, as well as the mean percentage of follicle loss were similar in the treatment and control groups (Table 2).
|Control Mean ± SEn = 9||CT Mean ± SE n = 9||P|
|Age, y||32.3 ± 2.8||27 ± 1.9||.9|
|Baseline follicle density||5.52 ± 2.1||5.56 ± 1.3||.9|
|Postculture follicle density||4.07 ± 1.8||3.54 ± 1.1||.8|
|Menstrual day at the time of ovarian sampling||13.25 ± 3.1||12.75 ± 3.1||.6|
|Mean estradiol level, pg/mL||1886 ± 545||1232 ± 485||.01|
|Mean volume of ovarian pieces cultured, cm3||0.066 ± 0.01||0.080 ± 0.01||.9|
Comparison of in vitro estradiol production between chemotherapy and control groups
In culture, ovarian cortical pieces that were previously exposed to chemotherapy secreted significantly less estradiol compared with controls. Daily levels of estradiol in both groups are illustrated in Figure 4A. The mean estradiol levels on each day of culture (Fig. 4A, P < .05) and for the entire duration of the culture period (Table 2, P = .01) were significantly lower in the chemotherapy group compared with controls. On correlation analysis, mean estradiol levels inversely correlated with age in the control group (r = −0.7, P < .05) whereas such a correlation did not exist with the chemotherapy group (r = −0.36, P = .3; Fig. 4B,C, respectively). As expected, neither daily estradiol values nor mean estradiol levels correlated with primordial follicle counts in both groups. In postculture serial sections, 100% of follicles were in primordial stage, ruling out the possibility that the difference in estradiol production originated from developing follicles. Because primordial follicles do not have aromatase activity, estradiol production could have only originated from the stromal cells. As a negative control, estradiol measurements were also performed in the blank culture media and were undetectable (< 10 pg/mL).
Comparison of in vitro estradiol production between patients treated with alkylating and nonalkylating regimens
To determine whether alkylating agents showed a selective toxicity toward stromal cells as they did for primordial follicles, we compared in vitro estradiol production between those samples that were exposed to alkylating agents and those that were not. The mean estradiol levels from culture Days 1–7 of chemotherapy patients who had received alkylating regimens was not significantly different from those treated with nonalkylating combinations (1429 ± 299 vs 1075 ± 94, P > .05).
Our report contains novel observations with practical implications for young women undergoing chemotherapy. Specifically, to our knowledge we present the first quantitative evidence in humans that chemotherapy regimens that contain alkylating agents result in significant loss of ovarian reserve compared with controls. Furthermore, and more interestingly, we show that chemotherapy regimens may alter ovarian stromal function regardless of whether they contain an alkylating agent, and this effect appears to be irrespective of the magnitude of germ cell damage. Previous studies to assess the gonadotoxic potential of different chemotherapy regimens on reproductive function were mainly focused on menstrual activity. As well-established from earlier studies of reproduction, menstruation is a very weak indicator of fertility.7 Whereas ovarian reserve markers such as FSH, estradiol, and anti-Mullerian hormone (AMH) measurements7 as well as antral follicle counts8 can give a better estimate of ovarian reserve before and after chemotherapy, there are no comprehensive studies evaluating the impact of chemotherapy regimens with these markers nor are these markers direct measures of ovarian reserve. Ovarian reserve consists of quiescent primordial follicles that are established before birth, even though this dogma has been recently challenged by 2 studies.9, 10
The numbers dwindle with age, as a result of follicle growth initiation and apoptotic death.6 Thus, the most direct assessment of ovarian reserve is by histological evaluation of primordial follicle counts. We previously developed and validated methods of primordial follicle counts in mammalian ovaries, including humans.11, 12 By using this histologic reserve assessment method on samples obtained during ovarian tissue cryopreservation procedures, we were able to quantitatively evaluate the impact of various chemotherapy regimens on ovarian reserve.
It has been well recognized that alkylating agents have the most significant negative impact on fertility1 and our work provided the histologic confirmation of this previously observed ovarian impact of alkylating agents. In rodent models, cyclophosphamide caused a dose-dependent loss in primordial follicles even at doses as low as 20 mg/kg.13 In a mouse model, a single dose of 200 mg/kg of cyclophosphamide resulted in an 87% reduction in primordial follicle count 72 hours after its intraperitoneal administration.4 Although we did not have a healthy control group, there is no evidence that cancer diagnosis itself affects primordial follicle reserve. In our series the impact of CHOP was strikingly similar to what was observed in rodent studies4 and in a xenograft model recently described by us.12
Although the incidence of amenorrhea is reported to be < 40% in young women treated with AC,14, 15 a 32-year-old subject treated with 4 cycles of AC for breast cancer in this study had an ovarian reserve that was approximately half that of a similar age patient who had received 6 cycles of ABVD for Hodgkin disease. As illustrated by this case, even the very young patients who continued regular menstruation after receiving the AC regimen may have diminished ovarian reserve, which will inevitably translate to premature ovarian failure.7
Data on the gonadotoxic effects of the ABVD combination suggest that the likelihood of treatment-related infertility is small in women under the age of 40 provided that no pelvic/abdominal radiation is given.16, 17 In our study there were 2 patients aged 24 and 32 who were diagnosed with Hodgkin disease and treated with 6 cycles of ABVD. Although there were no age-matched untreated controls with Hodgkin disease, both had significantly higher follicle counts when compared with 2 patients at ages 24 and 32 who were treated with regimens containing cyclophosphamide. This finding is consistent with the previous clinical observations that the ABVD protocol is associated with lower rates of amenorrhea. However, the 24-year-old patient had received inguinal radiation in addition to alkylating agents, which could have also contributed to lower primordial follicle counts.
The 18-year-old patients treated with 2 cycles of ADE-GMTZ for AML had an ovarian reserve comparable to an age-matched control, suggesting a less gonadotoxic profile for this combination regimen. Consistent with our histologic assessment, in 1 study 2 patients aged 29 and 31 who received 1 to 3 courses of Ara-C and daunorubicin did not develop premature ovarian failure in 1 and 7 years of follow-up.18 Likewise, in the case of idarubicin-cytarabin induction chemotherapy for AML, the follicle densities were comparable to an age-matched control patient, suggesting that this regimen does not have a significant impact on ovarian reserve.
Whereas the impact of chemotherapy on primordial follicles has been carefully assessed in animal studies, there have been no studies on the impact of these treatments on ovarian stromal function. primordial follicles appear to be hormonally inactive,19, 20 and in the absence of developing follicles the source of ovarian estradiol is ascribed to stromal cells. Using the in vitro estradiol production as a surrogate, we show for the first time that the previous exposure to chemotherapy resulted in altered ovarian stromal function. There were no developing follicles in the pre- and postculture specimens, indicating that altered estradiol production truly represented stromal cell population. Interestingly, aging had a similar negative effect on in vitro estradiol production as chemotherapy, suggesting that chemotherapy-induced damage may share similarities with age-related changes. Strikingly, estradiol production was diminished regardless of the regimen, indicating that ovarian stromal damage may not be specific to a certain class of agents. All these findings show that the compromised stromal function in patients treated with nonalkylating agents might not necessarily be associated with a striking loss in primordial follicle counts, in contrast to what is observed in patients treated with alkylating agents. However, considering static follicle counts as the sole measure of gonadotoxicity may lead to an underestimation of ovarian damage, as these stromal alterations may culminate in premature ovarian failure in the long run. All of these are novel observations that are being validated with a human ovarian xenograft model.12
In conclusion, we provide the first quantitative histologic evidence for chemotherapy-induced ovarian damage and show that cyclophosphamide-based alkylating regimens resulted in greater reduction ofovarian reserve in comparison to nonalkylating regimens.
Furthermore, we show that regardless of whether they include an alkylating agent, most chemotherapy regimens may have detrimental effects on ovarian stromal function. The clinical implications of the latter may become more obvious as the role of the niche in gonadal stem cell physiology is better understood.21, 22 In the meantime, the impact of chemotherapy on stromal cell function is being delineated in other organs, such as bone marrow.23
- 3Oocyte apoptosis is suppressed by disruption of the acid sphingomyelinase gene or by sphingosine-1-phosphate therapy. Nat Med. 2000; 10: 1109–1114., , et al.
- 14Amenorrhea following adjuvant chemotherapy for breast cancer. Proc Am Soc Clin Oncol. 1995; 14: A158., , , et al.