The initial response was favorable, with hematologic remission after 2 weeks, complete cytogenetic remission after 6 months, and major molecular response (≥3 log reduction in Bcr-Abl transcripts measured by quantitative polymerase chain reaction) after 9 months. Imatinib toxicity was limited to early transient neutropenia requiring a 4-week period of reduced dose, gastritis requiring omeprazole therapy, and subjective fatigue. After 20 months of therapy, minimal progression was noted (loss of major molecular response), and the imatinib dose was increased to 400 mg twice daily. Worsening fatigue and the return of abdominal pain were noted but a rapid response was observed, with Bcr-Abl transcripts falling to undetectable levels (complete molecular remission) after 3 months of imatinib at 400 mg twice daily.

At the time of minimal progression, stem cell transplantation was planned. Prior to the use of conditioning chemotherapy (busulfan, fludarabine, and rabbit antithymocyte globulin), a regimen known to be ovotoxic, the patient's mother requested that the patient be seen to discuss fertility preservation options.

Ultrasound evaluation at the patient's initial visit showed an age-appropriate ovarian reserve with an antral follicle count of 26 and ovarian volumes of 7.0 cc for the right ovary and 8.8 cc for the left ovary with an emerging dominant follicle of 13 mm noted on the left. She received extensive counseling regarding the likely impact of the planned chemotherapy regimen on her future ovarian function and likely need for hormone replacement and possible fertility preservation options, including oocyte cryopreservation, embryo cryopreservation, and ovary tissue cryopreservation. After counseling with the patient and her mother, the patient chose ovarian stimulation with oocyte cryopreservation.

Her ovarian stimulation involved pituitary suppression with a gonadotropin-releasing hormone (GnRH) antagonist and daily ovarian stimulation with 150 IU follicle-stimulating hormone (FSH) and 75 IU Menopur (a mixture of FSH and luteinizing hormone [LH]) (Table 1). During ovarian stimulation, her serum estradiol was noted to be lower than anticipated given the number of developing follicles, and on the day of human chorionic gonadotropin (hCG) administration her serum estradiol was 906 pg/mL, with 22 follicles measuring ≥13 mm for an average of 41 pg/mL per mature-sized follicle. A typical value for this ratio is 150–200 pg/mL per mature follicle. Additionally, on the day of the planned hCG administration, her follicles were noted to have a somewhat hazy appearance consistent with blood or debris within the follicles. An ovum retrieval was attempted on that day in case the patient had spontaneously begun to ovulate despite pituitary suppression. Despite aspiration of multiple follicles, no eggs were retrieved. hCG was given that evening as planned. The next day, albeit low, her estradiol level continued to rise, which was inconsistent with spontaneous premature ovulation. At 36 hours after hCG administration, only eight eggs were obtained, of which six were mature meiosis II (MII) oocytes.

A discussion was held with the patient, her mother, and the patient's hematologist regarding the unusual ovarian stimulation findings in the context of CML and imatinib use. Based on her impending stem cell transplant, the improved quality of her remission after imatinib dose escalation, and the question of an imatinib effect on her ovarian response, her hematologist felt it reasonable to interrupt imatinib for repeat stimulation.

The intrafollicular imatinib concentration in the first FF sample was 2,149 ng/mL, and the concentration of its active N-desmethyl metabolite was 506 ng/mL, consistent with equilibrium between the plasma and follicular compartments. There was no imatinib in the second FF samples, showing that the drug had been completely cleared from the patient's system at the time of the second oocyte retrieval.

In order to determine whether imatinib was cytotoxic to human granulosa cells, we treated primary human granulosa cell cultures with varying concentrations of imatinib (approximating treatment level serum concentrations ranging up to 4 μM) and found no significant difference in cell survival as assessed by granulosa cell protein content from a similarly plated number of granulosa cells treated with the gonadotropins FSH and hCG (Fig. 1). We used this in vitro culture system to further examine the classical granulosa cell steroidogenic function of aromatization of androgens using testosterone as a substrate. No significant differences were seen in the ability of granulosa cells to produce estradiol from testosterone when stimulated with gonadotropins with imatinib levels up to 4 μM (Fig. 2).

Previous studies have suggested that imatinib can have potential reproductive effects in both males and females. Animal studies have shown that imatinib can potentially inhibit postnatal testicular development and sperm capacitation [3, 4]. There are two reported cases of impaired spermatogenesis potentially related to imatinib use, one in a young adult [5] and one wherein imatinib was used long term prior to puberty [6]. However, there have been multiple reports of men on imatinib conceiving with no significantly higher risk for anomalies in the offspring [7]. The use of imatinib by women has potentially greater implications for pregnancy and offspring. Clearly, women taking imatinib retain at least some level of fertility, as evidenced by the numerous case reports of women conceiving while on imatinib therapy. The largest case series assessed 180 pregnancies, and of the pregnancies with known outcomes, there were serious fetal malformations in 9.6%, with approximately half of the pregnancies resulting in a healthy infant [8]. There has been one prior controversial report of an association between imatinib and premature ovarian failure [9, 10], but gonadal toxicity has not been specifically studied in either standard or high-dose imatinib trials, and this remains an area ripe for exploration.

The most commonly employed methods of female fertility preservation involve ovarian stimulation to increase the number of mature oocytes beyond what is achieved in a natural cycle. At the onset of a regular menstrual cycle, an initial rise in FSH secreted from the pituitary selects a follicle for development. Once a follicle is selected, estradiol is produced, which creates a negative feedback loop on the pituitary and subsequently suppresses FSH such that only a single mature follicle develops. Toward the end of follicular development, a positive feedback loop is established whereby the maturing follicle produces sufficient estrogen to sensitize the pituitary and result in a burst of gonadotropin secretion (termed the LH surge) to induce the final stages of oocyte maturation and initiate the ovulatory process. With ovarian stimulation, one main goal is to override this natural feedback mechanism by administering exogenous gonadotropins and a GnRH receptor blocker to cause development of multiple mature follicles while preventing the endogenous LH surge [11]. Physiologically, GnRH secreted by the hypothalamus acts on the pituitary to cause both FSH and LH secretion. Prevention of the endogenous LH surge is accomplished either by long-term (multiple weeks) GnRH agonist administration to downregulate pituitary GnRH receptors or through the short-term use of GnRH antagonists during the latter part of the ovarian stimulation treatment.

For patients using imatinib, the use of gonadotropin stimulation to recover multiple oocytes is required either for classical fertility preservation, involving oocyte or embryo cryopreservation (when gonadotoxic chemotherapy is anticipated), or for use with gestational surrogacy because teratogenicity in animal studies and limited human data make maternal use of imatinib during pregnancy contraindicated [12]. We were fortunate to have a patient who consented to analysis of her intraovarian follicular response and to have pre-hCG FF from each of her ovarian stimulation cycles. Analysis of the FF showed that imatinib and its active metabolite were present in concentrations suggesting equilibrium between the plasma and ovarian follicle compartments. Given the molecular weight of imatinib (589.7) and the permeability of the follicular basement membrane, this is not necessarily unexpected [13]. In the presence of imatinib, ovarian hormone production and oocyte recovery were compromised despite a comparable number of developing ovarian follicles.

The apparent difference in intrafollicular estradiol concentrations, despite identical gonadotropin stimulation, suggests that imatinib adversely affects steroidogenesis, possibly through an inhibitory effect on either the theca or granulosa cells. The lower number of oocytes retrieved while on imatinib is more difficult to explain and potentially more problematic clinically. We hypothesized that the granulosa cells underwent apoptosis leading to oocyte atresia and lower egg recovery [14, 15]. We therefore assessed the effect of imatinib on human luteinized mural granulosa cells in primary culture for any global effects on cell growth and survival by measuring cellular protein levels. As shown in Figure 1, we did not see any significant effect of imatinib on luteinized granulosa cells under these conditions. With treatment using a pharmacologically relevant range of imatinib concentrations, we also did not see a negative effect of increasing concentrations of imatinib on steroidogenesis in our luteinized granulosa cell culture system (Fig. 2). Based on these results, imatinib may be having an effect on either theca cells or granulosa cell proliferation during the follicular phase. Additional studies involving theca cells and nonluteinized granulosa cells could be performed to measure steroidogenic activity and levels of apoptosis with imatinib treatment to elucidate potential mechanisms for the effects we observed.

Imatinib mesylate (Gleevec®; Novartis Pharmaceuticals Corporation, East Hanover, NJ) is the first in a family of orally available, rationally designed specific kinase inhibitors, highly effective in CML and other select diseases. Nilotinib (Tasigna®; Novartis) and dasatinib (Sprycel®; Bristol-Myers Squibb, Princeton, NJ) have been developed and approved for use in imatinib-resistant and imatinib-intolerant patients; both have published data from the frontline setting and nilotinib is now approved by the U.S. Food and Drug Administration for use in newly diagnosed patients [16, 17]. Imatinib, the prototype compound, is a phenylaminopyridine derivative that inhibits constitutively activated Bcr-Abl tyrosine kinase by binding to the active site ATP-binding cleft [18].

CML, at presentation and in the setting of imatinib intolerance or resistance (over time, ~30% of de novo imatinib-treated patients), is felt to be reliant on Bcr-Abl activation, and Bcr-Abl–specific inhibitors are the mainstay of treatment of an increasing population of patients with Ph⁺ leukemia. In addition to Bcr-Abl⁺ leukemias, imatinib has been approved for the treatment of metastatic gastrointestinal stromal tumors (GISTs) [19]. Although not yet in clinical use, several trials have evaluated the use of imatinib in combination with other chemotherapeutic agents for more prevalent cancers, such as prostate, breast, and ovarian cancer [20–23].

More recent studies have shown that imatinib is not completely specific for its intended target and that it inhibits several other kinases to varying degrees. Imatinib is known to bind with high affinity to Kit and platelet-derived growth factor receptor (PDGFR)-α and PDGFR-β, with the former being the rationale for using imatinib to treat c-Kit⁺ GISTs. Kinase inhibitors are typically selective but not absolutely specific for their intended targets. A recent analysis of the interactions of 38 kinase inhibitors (including imatinib) for a representative subset of the human kinome using 317 kinases showed that, in addition to Abl, Kit, PDGFR-α, and PDGFR-β, imatinib also bound to 25 other kinases with varying affinities [24]. Other Bcr-Abl inhibitors used with increasing frequency may have narrower (nilotinib) or broader (dasatinib) target profiles, and “off-target” (e.g., non-Bcr-Abl) effects continue to be explored.

The targets of imatinib—Abl, c-Kit, PDGFR-α, and PDGFR-β—were found to be expressed at relatively uniform levels in 10 normal human premenopausal ovaries and at variable levels in granulosa tumor–derived cell lines [25]. That publication further characterized the effect of imatinib on granulosa tumor cell line proliferation, showing dose-dependent decreases in cell proliferation and viability and increased apoptosis, with an EC50 (half maximal effective concentration) that implicated off-target effects of imatinib on granulosa cells. This is consistent with the effects seen in our patient. Although there is considerable interpatient variability in imatinib pharmacokinetics, typical adult steady-state peak blood concentrations are in the range of 2,000–4,000 ng/mL [26, 27]. Interestingly, the distribution coefficients for many of the putative interactions of imatinib with other kinases were within these steady-state plasma concentrations, raising the possibility of clinically relevant off-target effects. It would be very interesting to perform microarray analyses of the gene-expression profiles of granulosa cells (mural and cumulus) exposed to imatinib, compared with those from unexposed patients, to understand the global effects imatinib could have on the somatic compartment of the developing follicle. Analyses of this type may also improve out understanding of which signaling pathways, if any, are altered in the ovarian follicle by imatinib treatment.

Recent studies in rodents have shown that imatinib targets are present in the murine ovary [28]. Interestingly, the consequence of imatinib use in these animal studies was protection of early postnatal murine ovaries from cisplatin-induced loss of follicular reserve of primary and primordial follicles. The mechanism for this phenomenon appears to be that blocking c-Abl tyrosine kinase activity prevents phosphorylation of p63, a key component for activation of proapoptotic genes within this system. Therefore, imatinib has been proposed as a unique method for preserving ovarian reserve during the administration of ovotoxic chemotherapy. However, this proposal should proceed with caution. Clearly, there are other effects on the ovary, and it is possible that by inhibiting ovarian responsiveness and function, imatinib may provide protection by creating a quiescent ovarian state, favorable under the threat of cytotoxic agents but paradoxically detrimental for fertility under similar conditions. Given an elimination half-life of ~18 hours for the active drug, and up to 40 hours for active N-desmethylated piperazine metabolites (http://www.micromedex.com), it would appear reasonable to stop imatinib for at least 2 weeks prior to attempting ovarian stimulation. Given that the time to recruit a primary follicle into the gonadotropin-responsive pool is on the order of a few months [14], our data show no evidence that very early follicle recruitment or development (prior to the antral stage) is compromised, because the oocytes obtained from our patient's second ovarian stimulation had already begun their early initial development while our patient was still on imatinib.

Provision of study material or patients: Mitchell P. Rosen, Alberuni M. Zamah, Marcelle I. Cedars

Collection and/or assembly of data: Alberuni M. Zamah, Merrill J. Egorin

Data analysis and interpretation: Mitchell P. Rosen, Alberuni M. Zamah, Michael Mauro, Brian J. Druker, Kutluk Oktay, Merrill J. Egorin

Manuscript writing: Mitchell P. Rosen, Alberuni M. Zamah, Michael Mauro, Brian J. Druker, Kutluk Oktay