Fertility preservation is receiving increasing attention as an evolving area of reproductive medicine. It aims to preserve reproductive tissue for future use. The main beneficiaries are those requiring gonadotoxic medical treatment, women undergoing destructive reproductive tract surgery, those with genetic conditions associated with premature ovarian failure, as well as women wishing to defer childbearing for ‘social’ reasons. Box 1 shows the indications for fertility preservation. Advances in the early detection of cancer and improved treatment protocols have increased patient survival significantly. The incidence of cancer in reproductive age women reported is 7% and the 5-year survival rate for them has increased during the last few years (Horner et al., 2011). Thousands of cancer survivors are women of reproductive age (Howlader et al., 2015). This, together with the trend of delayed childbearing, is resulting in a higher number of women diagnosed with cancer before their first pregnancy (Martin et al., 2006). The negative effect of cancer therapy on fertility is well known. Some chemotherapeutic agents, abdominal or pelvic radiotherapy, bone marrow transplantation and surgery for gynaecological malignancies have a high risk of gonadal damage (Noyes et al., 2011a,b; Morgan et al., 2012). It has been estimated that around 42% of female cancer patients of reproductive age may develop premature ovarian failure as a consequence of cancer therapy (Larsen et al., 2003; Eskander et al., 2011). The impact of cancer therapy on future fertility has raised concerns and fertility preservation (FP) is becoming an important component in the management of cancer patients (Lee et al., 2006; Jeruss and Woodruff, 2009). The most common malignancy in patients undergoing FP is breast cancer (Jemal et al., 2009; Horner et al., 2011). It is known that the loss of reproductive capacity negatively impacts the quality of life (Tschudin and Bitzer, 2009; Reh et al., 2011; Letourneau et al., 2012). The American Society of Clinical Oncology guidelines recommends that oncology patients of reproductive age should be counselled on the options for FP and future reproduction before the initiation of gonadotoxic therapy (Lee et al., 2006; Loren et al., 2013). Preservation of reproductive tissue is achieved through cryopreservation. Cryopreservation refers to the cooling of cells and tissues to sub-zero temperatures to achieve the complete cessation of biological activity. The temperature that is generally used for the storage of mammalian cells is – 196°C, the temperature of liquid nitrogen.1 Traditionally cryopreservation was achieved through slow freezing, although attention is now turning to vitrification. Vitrification deploys ultra-rapid cooling, in the presence of high concentrations of cryo precipitants, to solidify the cell or tissue into a glass-like state. This process avoids ice crystal formation and associated chilling injury, and is of particular importance to the cryopreservation of oocytes, whose large water content is more predisposed to ice crystal formation and damage to the fragile meiotic spindle.

Advances in cancer therapy have increased the number of women surviving a diagnosis of malignancy. Unfortunately, many such treatments are gonadotoxic and, as such, public and professional attention to fertility preservation for these women is growing. Germ cells are inherently sensitive to the toxic effects of both chemotherapy and radiotherapy. In particular, chemotherapy protocols containing alkylating agents, especially cumulative dose regimens of procarbazine and cyclophosphamide, appear to be the most gonadotoxic. null related gonadotoxicity. Age-related decline in ovarian reserve leaves older women more susceptible to the gonadotoxic effects of treatment. Therefore preservation procedures need to be tailored to the individual. Important considerations are the age and pre-existing fertility of the woman, the type of malignancy and treatment planned, the time available for preservation procedures and whether she has a male partner. Prepubescent girls are a particularly challenging group, restricted by limited options, along with ethical considerations about competence and consent issues.

Several genetic mutations are associated with POF. These often affect the X chromosome. Turner syndrome has an established association with POF and infertility. Most women with the disorder undergo ovarian failure at a very young age and many never develop any identifiable ovarian function. A Turner mosaic karyotype increases the possibility of identifying functioning ovarian tissue, which may be amenable to fertility preservation procedures. Fragile X (FMR1) premutation is also associated with POF. FMR1 is an unstable CGG triple sequence mutation located on the long arm of the X chromosome (Xq 27.3 locus). Unaffected individuals have 6–50 copies of the CGG repeat. Individuals with repeats in the range of 55–200 copies are carriers of the pre-mutation. Approximately 21% of all pre-mutation carriers develop POF, compared with only 1% of the general population. Although in principle, fertility preservation for women with genetic conditions is possible, it is not without controversy. Such women are at risk of resultant aneuploidy in the offspring. While the pre-implantation genetic diagnosis may help offset this risk, it does not negate the fact that conditions such as Turner syndrome are associated with medical comorbidities that may contraindicate pregnancy. Any decision regarding future pregnancy, therefore, needs to be carefully considered with appropriate counselling.

Ovarian surgery for benign conditions including endometriosis may diminish ovarian reserve. Several studies report a lower ovarian reserve after ovarian surgery, especially in patients with endometriomas. This may result from the incidental incision of normal ovarian tissue during cystectomy or result from damage of healthy ovarian tissue by electrocautery. Fertility preservation procedures should therefore be considered before complex or repeated ovarian surgery in women wishing to conceive in the future. Patients undergoing chemotherapy or radiotherapy for a variety of non-oncological conditions, including autoimmune connective tissue disease and haematological conditions may also benefit.

Most recently, ‘social’ fertility preservation has been assuming increasing importance. In today’s society, where increasing professional and financial opportunities are available to women, many are delaying childbearing. Given that female fertility progressively declines with age, delayed childbearing undoubtedly affects a woman’s opportunity for pregnancy. With assisted reproduction unable to fully overcome the effect of ageing on fertility loss, fertility preservation is an evolving technology that offers the potential to combat infertility secondary to ovarian ageing.

Cytotoxic chemotherapy and/or radiation therapy have been used to treat not only patients with malignant conditions but also those with various nonmalignant systemic diseases. Patients who are at risk of developing future ovarian failure may all benefit from fertility preservation technologies. Those suffering from benign ovarian diseases and undergoing radical surgery may also be added to this list. There is a growing list of diseases in which the treatment, or the disease itself, is associated with gonadal damage. Because each cancer patient’s clinical situation is unique, not one technique alone would be suitable. In our centre, we developed a comprehensive approach to fertility preservation, depending on the patient’s age, presence or absence of ovarian involvement, available time, and the indication for fertility preservation.

Cancer is the second leading cause of death in children between the ages of 1 and 14 years (Jemal et al., 2003). The cure rates from childhood cancers have improved markedly over the last three decades, thanks to cancer treatment that includes combined chemotherapy and/or radiotherapy, and HSCT. Five‐year survival is now ∼80% for all cancers combined, which is between 80 and 86% for childhood acute lymphoblastic leukaemia (ALL), and >90% for Hodgkin’s disease (Brenner et al., 2001; Jemal et al., 2003; Pui et al., 2003; Robison and Bhatia, 2003). Around 2000 patients are estimated to become long‐term survivors of ALL each year, the most common childhood malignancy (Gurney et al., 1995). In addition to leukaemias, patients who face the risk of ovarian failure due to cytotoxic treatment are those with Hodgkin’s lymphoma, neuroblastoma, non‐Hodgkin’s lymphoma, Wilm’s tumour, Ewing’s sarcoma and osteosarcoma of the pelvis and genital rhabdomyosarcoma (Nussbaum et al., 1999; Arndt et al., 2001; Franchi‐Rezgui et al., 2003; Ozaki et al., 2003; Rodl et al., 2003). Poirot et al. (2002) evaluated the feasibility of long‐term ovarian tissue cryopreservation in 31 women with a malignant and non‐malignant disease who were at risk for ovarian failure due to treatment. In their study, the age of the patients ranged between 2.7 and 34 years, and 16 of them were

Breast cancer is the most common malignant disease in reproductive age women (Weir et al., 2003). In the USA, it is estimated that ∼211 000 new breast cancer cases will have been diagnosed in 2003. The incidence of female breast cancer has increased since 1986, but the death rates decreased in the early 1990s; 2.5% per year in white women, and 1.0% per year in black women (Weir et al., 2003). One out of every 228 women will develop breast cancer before age 40 years, and ∼15% of all breast cancer cases are estimated to occur at Many of these patients will be subjected to multi‐agent, mainly cyclophosphamide‐based, cytotoxic chemotherapy (Kaufmann et al., 2003). In breast cancer, because chemotherapy is usually initiated 6 weeks after the surgery, there is adequate time for controlled ovarian stimulation to preserve fertility by oocyte or embryo cryopreservation. Because conventional ovulation induction regimens are deemed risky for breast cancer patients due to resultant surge in estradiol levels, potentially safer regimens including tamoxifen or aromatase inhibitors have been introduced (Oktay et al., 2003b). These patients may also be candidates for ovarian tissue cryopreservation, as occult ovarian metastasis is extremely rare in non‐metastatic breast cancer (Gagnon and Tetu, 1989; Hann et al., 2000).

Cancer of the cervix is a serious health problem affecting 500 000 women each year worldwide. In the year 2002, 13 000 new cervical cancer cases were diagnosed in North America and roughly half of them occurred before the age of 35 years (Waggoner, 2003). Over the past three decades, while the incidence of squamous cell carcinoma of the cervix decreased by 42%, the incidence of adenocarcinoma of the cervix increased by 29% (Smith et al., 2000). Ovarian involvement is extremely rare in squamous cell cervical carcinoma, but it is encountered in up to 12% of the cases with adenocarcinoma and adenosquamous carcinoma of the cervix (Nakanishi et al., 2001; Yamamoto et al., 2001). Ovarian transposition is commonly considered for patients with cervical cancer who will receive pelvic radiotherapy, but the success rates tend to vary with this procedure. If adjuvant radiosensitizing chemotherapy is scheduled, ovarian cryopreservation for future autotransplantation is another option. Alternatively, one ovary can be transposed, usually, the one on the opposite side of the main tumour, and the other one could be cryopreserved. Even if there is enough time for controlled ovarian stimulation, it is not considered safe to perform transvaginal oocyte aspiration in these patients, as there is a risk of profuse bleeding from the friable, cancerous cervix.

Radiotherapy is utilized to improve prognosis or to achieve local tumour control in some solid tumours presenting in the pelvis such as Ewing sarcoma, osteosarcoma, retroperitoneal sarcomas, and in some benign bone tumours (Feigenberg et al., 2001; Ferguson and Goorin, 2001; Bacciet al., 2003; Ozaki et al., 2003; Pisters et al., 2003; Rodl et al., 2003). Radiation therapy also plays an important part in the management of rectal cancer (Kapiteijn et al., 2001). These patients can resort to ovarian, oocyte or embryo cryopreservation, or oophoropexy may be considered, especially if an abdominal surgery is already necessary for the treatment of the primary disease.

Some benign ovarian diseases, either due to their extensive or progressive nature or because of bilateral occurrence and repeated surgeries, may significantly compromise ovarian reserve (Oktay et al., 2001b). Healthy pieces of the ovarian tissue can be cryopreserved in women undergoing oophorectomy for benign conditions. If disease recurrence is likely, subcutaneous transplantation of the ovarian pieces is the preferred procedure, due to ease in monitoring and the presumed simplicity of removal in the case of recurrence.

Inherited mutations, mainly BRCA‐1 and BRCA‐2, account for almost 10% of all epithelial ovarian carcinomas (Claus et al., 1996; Newman et al., 1998). In the USA, the carrier frequency of the germ‐line BRCA mutations is 0.1%, although it is as high as 1% for each mutation in the Ashkenazi Jewish population (Struewing et al., 1995). The cumulative lifetime risk of developing ovarian cancer is ∼60% in the presence of BRCA‐1 mutation, and 10–20% in women with BRCA‐2 mutation (Struewing et al., 1995, 1997; Liede et al., 2002). In addition, the lifetime risk of breast cancer in female carriers of BRCA1 mutation is 80–90%. While peritoneal cancer cannot be prevented in BRCA‐positive patients, prophylactic oophorectomy is suggested as soon as childbearing is completed, or by the age 35–40 years, to decrease the risk of ovarian and breast cancer (Kauff et al., 2002; Rebbeck et al., 2002). Ovarian tissue cryopreservation might be offered in patients who wish to delay childbearing beyond the age of 35 years. When patients desire to conceive, ovarian tissue can be transplanted, preferably subcutaneously, so that it could be easily monitored, and removed once the pregnancy occurs to avoid further cancer risk. These patients may also be candidates for future in vitro maturation using oocytes obtained from cryopreserved ovarian tissue.

Systemic lupus erythematosus (SLE) typically affects reproductive-age women, with an overall incidence between 40 and 250 per 100 000 people (Michet et al., 1985). High dose cyclophosphamide, with or without HSCT, is sometimes used in the treatment of SLE and can result in premature ovarian failure in these patients (Gladstone et al., 2002). Other autoimmune diseases reported to benefit from cytotoxic therapy with alkylating agents are Behcet’s disease, steroid‐resistant glomerulonephritis, inflammatory bowel diseases, and pemphigus Vulgaris (Russell et al., 2001; Langford et al., 2003; Nousari et al., 2003; Stallmach et al., 2003). These patients may also become candidates for ovarian, oocyte or embryo cryopreservation.

Autologous or allogeneic HSCT has been an important therapeutic tool in the management of some malignant and non‐malignant systemic diseases. Among the non‐malignant conditions reported to benefit from HSCT are some autoimmune diseases previously unresponsive to immunosuppressive therapy, diseases associated with genetically abnormal stem cells, and those associated with the deficiency of bone marrow stem cell products (Slavin et al., 2001; Burt et al., 2003). Before HSCT, preconditioning regimens are used to ablate the pre‐existing bone marrow. The most commonly used conditioning regimens for allogeneic and autologous HSCT in acute myelogenous leukaemia (AML) include cyclophosphamide/total body irradiation or busulfan/cyclophosphamide (Litzow et al., 2002). Both regimens are highly gonadotoxic; the risk of developing complete or partial ovarian failure may be >80% in children receiving a conditioning regimen for HSCT (Thibaud et al., 1998; Couto‐Silva et al., 2001). HSCT has also been used in patients with breast cancer, multiple myeloma and lymphoma (Einsele et al., 2003; Mazza et al., 2003). Among the non‐malignant conditions suggested to benefit from HSCT are systemic lupus erythematosus, aplastic anaemia, sickle cell anaemia, autoimmune thrombocytopenia, progressive systemic sclerosis, rheumatoid arthritis, juvenile idiopathic arthritis and vasculitis (Olalla et al., 1999; Tyndall and Millikan 1999; Walters, 1999; Burt et al., 2003; Cohen et al., 2003). Adult patients, if there is sufficient time, can resort to conventional IVF for oocyte or embryo cryopreservation. Ovarian cryopreservation is the only choice to preserve fertility in paediatric patients, and in patients who cannot postpone their treatment.

Currently, embryo cryopreservation is the only well‐established procedure that is commonly used in many infertility clinics worldwide. However, each cancer patient presents with a unique situation, and embryo cryopreservation may not always apply to every individual. We will outline the currently available fertility preservation options and their applicability in different clinical situations.

In cancer patients, IVF can be performed to store embryos for future use, if the patient has a partner and enough time before treatment. Survival rates per thawed embryo range between 35 and 90%, implantation rates between 8 and 30%, and cumulative pregnancy rates can be >60% (Al‐Shawaf et al., 1993; Frederick et al., 1995; Selick et al., 1995; Senn et al., 2000; Wang et al., 2001; Son et al., 2003). In breast cancer, there is typically a 6-week hiatus between surgery and chemotherapy, which would be adequate to perform ovarian stimulation and IVF. Nevertheless, since conventional ovulation hyperstimulation regimens in IVF cycles typically result in estradiol levels that may be 10‐fold higher than peak estradiol levels seen in a natural cycle, they are not recommended in breast cancer patients (Pittaway and Wentz 1983; Adashi, 1996; Pena et al., 2002; Chen et al., 2003). After its discovery in 1963, tamoxifen became an important part of the treatment of breast cancer and has been tested for the chemoprevention of this disease (Harper and Walpole, 1966, 1967; Veronesi et al., 2003). While it was originally used as a contraceptive agent in the UK, it was later found to be a useful ovulation induction agent (Klopper and Hall, 1971). Exploiting its dual effect, we recently demonstrated that tamoxifen can be safely used to perform ovarian stimulation and IVF in breast cancer patients (Oktay et al., 2003a). In that study, 12 women with breast cancer received 40–60 mg tamoxifen for a mean duration of 6.9 days beginning on days 2–3 of their menstrual cycle. Patients underwent IVF and embryo transfer with either fresh or cryopreserved embryos and were compared with a retrospective control group of breast cancer patients who had natural cycle IVF (NC-IVF). Cycle cancellation was significantly less frequent in patients receiving tamoxifen, compared to those who underwent NCIVF (1/15 versus 4/9, P < 0.05). The mean numbers of mature oocytes (1.6 versus 0.7, P = 0.03) and embryos (1.6 versus 0.6, P < 0.05) per initiated cycle were higher in the tamoxifen group compared with NC-IVF. In addition, tamoxifen stimulation resulted in the generation of an embryo in every patient (12/12) whereas only three of five patients had embryos following natural cycle IVF. The mean peak estradiol level in the tamoxifen group was significantly higher than in natural cycle IVF patients (442.4 versus 278 pg/ml). Even though tamoxifen results in an increase in peak estradiol levels, it is well known that tamoxifen can block the effects of supraphysiological levels of estrogen on breast tissue, and inhibits the growth of breast tumours by competitive antagonism of estrogen at its receptor site. Mean estradiol levels are chronically elevated in breast cancer patients who are on long‐term tamoxifen treatment and can be higher than the levels seen in patients undergoing ovarian stimulation with tamoxifen (Shushan et al., 1996; Klijn et al., 2000). Endometrial cancer is another estrogen‐sensitive malignancy, which can be seen in reproductive-age women. Because tamoxifen is stimulatory on the endometrium, it cannot be used in endometrial cancer for ovarian stimulation. For these patients, aromatase inhibitors can be used for ovarian stimulation, IVF and embryo cryopreservation (Oktay et al., 2003b). Aromatase P450 catalyses the reaction converting androgenic substances to estrogens. Letrozole is a potent and highly selective third-generation aromatase inhibitor that was developed in the early 1990s. It competitively inhibits the activity of the aromatase enzyme and has a half‐life of ∼48 h (Pfister et al., 2001). Letrozole significantly suppresses plasma estradiol, estrone and estrone sulphate levels at doses ranging from 0.1 to 5 mg/day, and it was recently shown to be superior to tamoxifen in the treatment of advanced-stage post‐menopausal breast cancer (Dowsett et al., 1995; Buzdar et al., 2001; Mouridsen et al., 2003). Aromatase inhibitors have recently been considered and tested as ovulation induction agents. In cycling bonnet monkeys, letrozole resulted in the formation of multiple follicles (Shetty et al., 1997). Clinical studies have also shown its benefit in ovulation induction alone or in combination with FSH, and letrozole has been suggested in the treatment of poor responders (Mitwally and Casper, 2002). Currently, we are testing the safety of ovarian stimulation with aromatase inhibitors in breast and endometrial cancer patients.

Embryo cryopreservation may not be an option for single women unless they choose to use sperm donation. In these patients, if they have time to complete ovarian stimulation before cancer therapy, freezing mature or immature oocytes can be considered instead. The first human live birth from cryopreserved oocytes was reported by Chen (1986). Porcu et al. (1997) reported the first human live birth after transferring embryos generated by ICSI of cryopreserved oocytes. Later, additional successful human pregnancies were declared by several investigators using the same technique (Young et al., 1998; Quintans et al., 2002; Katayama et al, 2003; Yoon et al., 2003). Unlike the cryopreservation of embryo and sperm, early results have been disappointing with low survival, fertilization, and pregnancy rates after IVF of thawed oocytes (Mandelbaum et al., 1988; Imoedemhe and Sigue, 1992; Oktay et al., 2001a). However, with recent studies suggesting increased success rates, the interest in cryopreservation of oocytes has been rekindled (Fabbri et al., 2001; Porcu, 2001; Katayama et al., 2003; Liebermann et al., 2003; Yoon et al., 2003). In earlier reports, survival and fertilization rates of frozen-thawed mature oocytes varied between 25 and 95% (Al‐Hasani et al., 1987; Gooket al., 1995; Kuleshova et al., 1999; Yoon et al., 2000; Porcu, 2001; Katayama et al., 2003) and between 13.5 and 71% (Imoedemhe and Sigue, 1992; Kazem et al., 1995; Porcu et al., 1997; Fabbri et al., 2001; Chen et al., 2002) respectively. When data from 21 studies in peer‐reviewed journals were reviewed, we found a mean survival rate of 47%, a mean fertilization rate of 52.5% and a mean pregnancy rate per thawed oocytes of 1.52%. One of the factors cited in the improvement of fertilization rates of frozen-thawed oocytes is the use of ICSI to overcome the zona hardening, which is believed to have been caused by the freezing process (Kazem et al., 1995; Tucker et al., 1996; Porcu, 2001; Katayama et al., 2003). It has also been proposed that ICSI results in embryos with greater cleavage rates (Gook et al, 1995). The safety record for oocyte cryopreservation is not extensive. The rate of abnormal fertilization ranges from 5 to 15.3% (Porcu, 2001). Of the 32 pregnancies where the perinatal outcome was reported (Oktay et al., 2001a), there has been one ventricular septal defect (VSD) and one triploid pregnancy. The latter, however, resulted after ICSI of frozen-thawed testicular sperm into cryopreserved oocytes (Chia et al., 2000). Recently, improved post‐thaw survival and fertilization rates and live births were obtained with vitrification using ethylene glycol and dimethylsulphoxide (DMSO) as cryoprotectants (Katayama et al., 2003; Yoon et al., 2003). A Review of the existing data indicates a mean survival rate of 68.4%, fertilization rate of 48.5%, and pregnancy rate of 1.7% per vitrified–thawed oocytes. It has been suggested that immature oocytes may be more resistant to cryodamage due to lower cell volume, and lack of metaphase spindle. Even though high rates of nuclear maturation have been reported with cryopreserved immature oocytes, the developmental capacity has been generally low (Toth et al., 1994; Van Blerkom and Davis, 1994; Wu et al., 2001). Spindle abnormalities, premature and partial condensation of chromosomes have also been observed after in vitro maturation of cryopreserved germinal vesicle GV stage oocytes (Park et al., 1997). Only a few pregnancies have been reported using frozen-thawed human immature oocytes thus far (Tucker et al., 1996; Wu et al., 2001).

The Ovarian cortex contains primordial follicles with oocytes arrested in the diplotene of prophase of first meiotic division. It has been suggested that relatively high surface/volume ratio, low metabolic rate and the absence of zona pellucida make primordial follicles less susceptible to cryodamage. Ovarian tissue cryopreservation and transplantation studies date back to the 1950s. Initial studies were disappointing until the discovery of effective modern cryoprotectants and the availability of automated cryopreservation machines. Glycerol was the only available cryoprotectant in the 1960s but was found ineffective for cryopreservation of human oocytes and ovarian tissue (reviewed in Oktay, 2001). With the advent of more effective cryoprotectants such as ethylene glycol, DMSO and propanediol, animal studies were repeated and successful deliveries were reported in several species (Gosden et al., 1994a; Sztein et al., 1998; Liu et al., 2001). In humans, resumption of ovarian endocrine function could be demonstrated (Oktay and Karlikaya, 2000; Radford et al., 2001), and very recently, the first embryo following subcutaneous transplantation of cryopreserved ovarian tissue was reported (Oktay et al., 2004). There has not yet been a report of pregnancy after ovarian transplantation.

Severe combined immunodeficiency disease (SCID) mice can harbour tissues from foreign species without graft‐versus‐host response, because of the T‐ and B‐cell deficiency due to a gene mutation (Bosma et al., 1983; Gosden et al., 1994b). Ovarian tissue pieces can be grafted intramuscularly, subcutaneously, or placed under the kidney capsule to improve vascularization. The SCID mouse model was first utilized to study follicle development in xenografted sheep and cat ovarian tissue (Gosden et al., 1994b). Thereafter we adapted this model to study human ovarian tissue xenografting (Oktay et al., 1998a). Approximately 1 mm3 of ovarian tissue pieces were xenografted under the kidney capsule, animals were stimulated with FSH during the last 6 weeks of the 17 week grafting period. We demonstrated estradiol production, estrogenization of the uteri and antral follicle development in FSH‐stimulated mice. In another study, we aimed to determine the long‐term survival of frozen-thawed human ovarian tissue as xenografts in ovariectomized SCID mice (Oktayet al., 2000). We did not give exogenous gonadotrophins since these animals had anatomically intact hypothalamic‐pituitary axis; however, follicles did not grow beyond the 2‐layer stage in these xenografts. The latter study indicated that endogenous gonadotrophins in ovariectomized SCID mice were not sufficient to support human follicular growth. Similarly, several groups showed follicle development, ovulation and corpus luteum formation after stimulation of xenografted human ovarian tissue using gonadotrophins in immunodeficient mice (Weissman et al., 1999; Gook et al., 2001; Kim et al., 2002; Van den Broecke et al., 2002; Abir et al., 2003). In summary, even though human ovarian xenografts provided a model to study human ovarian tissue autotransplantation, their use as a means to utilize banked ovarian tissue is in question. Concerns regarding cross‐species retroviral infections should be addressed. Moreover, this technique will require large numbers of animals to be killed since only very small fragments of ovarian tissue can be xenografted. This may not only make the technique impractical but may also raise further ethical concerns.

There have been two main approaches in autotransplant ovarian cortical pieces in humans. Orthotopic transplants involve grafting these strips near the infundibulopelvic ligaments or possibly on a post‐menopausal ovary. In the heterotopic transplant, tissues can be grafted subcutaneously at various locations including the forearm and abdominal wall. Orthotopic ovarian transplantation. While transplantation may allow a natural pregnancy to occur (Gosden et al., 1994a), it requires abdominal surgery and general anaesthesia. A laparoscopic approach makes this surgery less invasive but technically more challenging. In the first case of laparoscopic orthotopic transplantation procedure with frozen ovarian tissue in a 27-year-old woman, ovarian cortical pieces had been cryopreserved in 1.5 mol/l propanediol using a slow‐freeze protocol. After tissues had been thawed, they were sutured to two triangular frames made from an absorbable cellulose membrane. Then we laparoscopically transplanted them beneath the left pelvic peritoneum of the ovarian fossa (Figure 1). With the expectation of improving vascularization, aspirin 80 mg/day p.o. and FSH 150 IU/day i.m. were given for a week after the operation. Fifteen weeks after grafting, the patient was stimulated with daily menopausal gonadotrophins, which was gradually increased from 150 to 675 IU/day, and ovulation was confirmed by elevated progesterone levels, ultrasonographic demonstration of a corpus luteum, free fluid in the cul‐de‐sac, and change in en ometrial pattern on ultrasound. The Ovarian function could not be demonstrated beyond 9 months of follow‐up. Radford et al. (2001) reported a 36-year-old woman with stage IIIB nodular sclerosing Hodgkin’s lymphoma who had had her right ovary cryopreserved before implementation of high dose chemotherapy for the third recurrence of the disease. This patient had already been exposed to chemotherapy before ovarian cryopreservation. Nineteen months later, after she had experienced premature menopause, two ovarian cortical strips were thawed and transplanted onto the left ovary and another to the site of the removed ovary. Seven months after transplantation, the patient reported resolution of the menopausal symptoms. Five weeks later, serum estradiol rose to 352 pmol/l and pelvic ultrasonography showed a 10 mm thick endometrium and a 20 mm follicle on the right side. However, progesterone levels were never >2 nmol/l, and no ovulation was detected. Gonadotrophin levels were in the post‐menopausal range 9 months after the transplantation. The Cortex of the frozen-thawed ovary, as well as biopsy samples of the retained left ovary, did not show evidence of neoplasia. However, none of these cases can be considered as ‘ideal’ to judge the performance of this procedure, because the cryopreserved tissue was previously compromised due to ovarian surgery or chemo‐ and radiotherapy. Heterotopic ovarian transplantation. Autotransplanting tissue to a heterotopic site is a well‐known concept, and it has long been utilized for implanting fresh or frozen-thawed parathyroid tissue following total parathyroidectomy (Wells et al., 1975, 1978). Heterotopic transplantation has significant advantages; this technique does not require general anaesthesia or abdominal surgery. In addition, it is easy to monitor follicle development, and remove the transplanted tissue from a subcutaneous site when necessary. In our initial reports, we utilized the subcutaneous space above the brachioradialis facia of the forearm, and recently, the lower abdomen to transplant ovarian cortical pieces (Oktayet al., 2001b,c, 2004). We performed the first subcutaneous ovarian transplantation in a 35-year-old woman with stage III squamous cervical carcinoma before pelvic radiation (Oktay et al., 2001b). Immediately after laparoscopic oophorectomy her FSH and LH levels indicated that she was in menopause. Six weeks after the transplantation, the patient reported a painless swelling at the site of the ovarian transplantation. Estradiol levels were elevated, and ultrasound revealed one dominant follicle measuring 15 mm and four other antral follicles. Restoration of normal testosterone levels also indicated normal stromal function. After controlled ovarian stimulation with hMG, three oocytes were retrieved, two of which were immature and one was in metaphase I. The metaphase I oocyte underwent in vitro maturation but the ICSI of this oocyte did not result in fertilization. This patient had nearly 3 years of ovarian function but never ovulated spontaneously. In another 37-year-old woman, we transplanted fresh ovarian tissue subcutaneously to the forearm after oophorectomy due to recurrent benign ovarian cysts (Oktay et al., 2001b). Resumption of menstrual periods and spontaneous ovulation occurred as early as 3 months after the transplant. However, the cycle length varied from 14 to 45 days, and the graft ceased the function after 3 years (Figure 2). Very recently, after transplantation of frozen‐banked ovarian tissue underneath lower abdominal skin as a 36-year-old breast cancer survivor, we were able to reverse menopause. Percutaneous oocyte aspiration resulted in the generation of a 4‐cell embryo, the first after an ovarian transplantation procedure (Oktay et al., 2004). One of the potential limitations of ovarian tissue cryopreservation and transplantation is the loss of a large fraction of follicles during the initial ischaemia after transplantation. Previous work indicated that whereas the loss due to freezing is relatively small (Oktay et al., 1997a; Aubard et al., 1999; Baird et al., 1999), up to two‐thirds of follicles are lost after transplantation. Because of this fact, we do not recommend ovarian tissue freezing in patients aged >40 years (Oktay, 2000), and we prefer this procedure in patients

Ovarian tissue can be grafted once the patient survives malignancy or is considered cured. It is naturally of concern that the frozen-thawed tissue might harbour malignant cells, and cancer could be reseeded by ovarian transplantation. Fortunately, most of the malignant tumours of reproductive age women do not metastasize to ovaries, except some haematological malignancies such as leukaemias, Burkitt’s lymphoma, and some advanced stage solid tumours such as breast and colon cancers (Chu et al., 1981; Wyld et al., 1983; Yada‐Hashimoto et al., 2003). In children, ovarian metastasis was demonstrated in 25–50% of neuroblastoma cases in post mortem examinations (McCarville et al., 2001; Oktay et al., 2001a). Breast cancer has a low‐to‐intermediate risk of ovarian involvement in the early stages (Oktay and Sonmezer, 2004). In the absence of clinical and radiological evidence of distant metastasis, ovarian involvement is extremely rare, and most cases could be detected by a thorough clinical and radiological evaluation (Curtin et al., 1994). Most of the occult metastases belong to the less common histological type, the infiltrating lobular (15% of all breast cancer) as opposed to the invasive ductal cancer which constitutes >70% of all breast cancers (Young and Scully, 1987; Morrow, 2001; Perrotin et al, 2001; Li et al, 2003). Moreover, lobular cancer typically occurs in post‐reproductive age women (Sastre‐Garau et al., 1996; Li et al., 2003), and ovarian metastasis more commonly occurs in advanced-stage cancer (Gagnon and Tetu, 1989; Hann et al., 2000). Incidence of ovarian involvement is exceptionally low in Wilm’s tumour, Ewing’s sarcoma, lymphomas, osteosarcomas, and extragenital rhabdomyosarcomas. In squamous cell cervical cancer, ovarian involvement is < 1.0%, whereas it is reported to be between 1.7 and 12.5% in adenocarcinoma of the cervix, (Woodruff et al., 1970; Suttonet al., 1992; Nakanishi et al., 2001). The risk of ovarian involvement according to tumour type is summarized in Table II. The risk of reimplanting cancer cells via transplanted ovarian tissue was investigated in some animal and xenograft studies. In a rodent study, most animals died of the disease after a small piece of fresh or frozen-thawed ovarian tissue had been transplanted from mice with a very aggressive form of lymphoma (Shaw et al., 1996). However, this type of aggressive lymphoma is extremely rare in humans, during reproductive age. In another study, human frozen-thawed ovarian tissue from patients with Hodgkin’s disease (follicular B cell lymphoma, n = 13) and non‐Hodgkin’s lymphoma (NHL) (n = 5) was xenografted subcutaneously to NOD/LtSz_SCID mice (Kim et al., 2001). None of the animals grafted with ovarian tissue from lymphoma patients developed lymphoma, whereas, in the positive control group, all three SCID mice grafted with lymph node tissue from NHL patients developed human B‐cell lymphoma, as demonstrated by microsatellite DNA analysis. The latter indicated that ovarian transplantation is safe in NHL patients. The results were not completely reassuring for Hodgkin lymphoma patients, since, in this model, positive controls did not also transmit the disease. However, based on clinical experience, ovarian involvement is extremely rare in Hodgkin’s patients. Regardless of the magnitude of risk of ovarian involvement, a thorough histological evaluation should be performed on multiple samples taken from the ovarian tissue before and after cryopreservation. Additionally, molecular biology techniques and immunohistochemistry can be used to screen for the presence of cancer cells in the ovary (Oktay and Yih, 2001).

Ovaries can be moved out of the radiation field so that direct effects of ionizing radiation may be avoided. It is more than three decades since this procedure was first put into practice to preserve ovarian function in patients with Hodgkin’s disease receiving pelvic or para‐aortic lymph node irradiation at staging laparotomy (Ray et al., 1970; Nahhas et al., 1971; Le Floch et al., 1976; Thomas et al., 1976). If the patient is to undergo abdominal surgery, ovaries can be transposed simultaneously, or if she is to be treated non‐surgically, laparoscopic transposition can be performed before the scheduled radiotherapy (Tulandi and Al‐Took, 1998; Morice et al., 2000). The success with fertility preservation by ovarian transposition before radiotherapy varies between 16 and 90% (Hunter et al., 1980; Anderson et al., 1993; Clough et al., 1996; Morice et al., 2000; Meirow and Nugent, 2001; Bisharah and Tulandi, 2003). This variation in success rates is due to variations in the degree of scatter radiation, vascular compromise, the age of the patient, a dose of radiation, whether the ovaries were shielded, whether concomitant chemotherapy is used, and whether vaginal brachytherapy or pelvic external beam irradiation plus brachytherapy was used (Thomas et al., 1976; Hunter et al., 1980; Anderson et al., 1993; Feeney et al., 1995; Williams et al., 1999; Morice et al., 2000; Meirow and Nugent, 2001). In a recent study, laparoscopic oophoropexy was performed to preserve ovarian function before pelvic irradiation in 10 patients with Hodgkin’s disease (Williams et al., 1999). Pelvic radiation dose ranged from 1500 to 3500 cGy. All five patients who received minimal or no chemotherapy had evidence of ovarian function, four of whom achieved pregnancy. In contrast, four patients who received multiple courses of chemotherapy and one patient who received 3500 cGy to the femoral lymph nodes and pelvis with little central shielding had an ovarian failure at follow‐up. The length of follow‐up was not clearly stated. Even though ovarian transposition may decrease the risk of ovarian failure, ovaries are still subjected to a significant amount of radiation despite proper shielding. This is mainly due to scattering radiation and transmission through the shield, which may amount to as much as 8–15% of the total pelvic radiation dose (Le Floch et al., 1976). In addition, this surgical procedure is not without complications; Fallopian tube infarction, chronic ovarian pain, ovarian cyst formation, and migration of ovaries back to their original position before radiotherapy have been reported, some of which may require additional gynaecological surgeries (Gabriel et al., 1986; Williams et al., 1999; Meirow and Nugent, 2001). Anderson et al. (1993) reported subsequent oophorectomy in nine of 51 patients (17.5%) for the management of painful ovarian cysts after a mean duration of 46.8 months from the procedure.

When ovaries are transposed to an abdominal position, spontaneous pregnancy may not be possible, unless a second procedure is performed to relocate ovaries back to the pelvis (Morice et al., 1998). In addition, should these patients need IVF in the future, oocyte retrieval may become technically more challenging. Therefore candidates for ovarian transposition should be selected carefully, accounting for all the variables that may affect its success rates. It should also be borne in mind that, when gonadotoxic chemotherapy is used along with radiation, there is no strong rationale to perform this procedure. GnRH analogue co‐treatment

It has been hypothesized, largely based on the debated role of gonadal suppression in men in preserving testicular function against chemotherapy, and partially the misbelief that pre‐pubertal girls are not affected by gonadotoxic cancer treatment, that ovarian suppression can be protective. Some animal studies demonstrated a protective role of GnRH analogue treatment against chemotherapy‐induced gonadal damage (Glode et al., 1981; Ataya et al., 1989; Bokser et al., 1990). Ataya et al. (1995a) demonstrated that primordial follicle loss associated with cyclophosphamide treatment was significantly lower in Rhesus monkeys receiving GnRH analogue treatment, compared with GnRH analogue untreated animals (65 versus 29%). In another study (Ataya et al., 1995b) they failed to demonstrate any protective effect of GnRH analogue treatment against radiation‐induced ovarian injury in rhesus monkeys. Germinal epithelium is extremely sensitive to irradiation, and despite some reports showing a protective effect, it seems unlikely that radiotherapy‐induced gonadal damage can be prevented by gonadal suppression (Ortin et al., 1990; Ataya et al., 1995b; Gosden et al., 1997; Viviani et al., 1999). A few non‐randomized studies with short‐term follow‐up suggested a protective role for GnRH analogue treatment (Blumenfeld et al., 1996; 1999, 2000, 2002; Recchia et al., 2002) Blumenfeld et al. (1996) investigated the protective role of GnRH analogue co‐treatment in 18 women treated with chemotherapy for Hodgkin’s or Non‐Hodgkin’s lymphoma, compared with a historical control group of 18 women treated with similar regimens. Ten patients in the study and 11 in the control group also received mantle field irradiation after chemotherapy. Patients received GnRH analogue for a maximum of 6 months starting before chemotherapy. In the study group, 93.7% of the patients resumed spontaneous ovulation and menses within 3–8 months of termination of the combined chemotherapy/GnRH analogue co‐treatment. In contrast, only 39% in the control group resumed ovarian cyclic activity, and 61% experienced premature ovarian failure. However, mean follow‐up was 1.7 ± 1.0 years (range, 0.5–4) in the study and 7.0 ± 4.9 years (range, 1.5–8.0) in the control group. In another study, the same authors studied the protective effect of GnRH analogue in 54 cancer patients and eight women with non‐malignant diseases who received chemotherapy (Blumenfeld, 2001), and retrospectively compared these patients with 55 women who had been treated with similar chemotherapy. In almost all of the surviving patients in the GnRH analogue/chemotherapy group, spontaneous ovulation and menses occurred within 6 months. Less than 50% of the patients in the control group resumed ovarian function and regular cyclic activity. However, the methodology was not clear; whereas the control group was retrospective, the criteria for choosing the control patients were not mentioned. For the treatment group, the length of follow‐up was not mentioned. From these studies, it cannot be determined whether the lower incidence of ovarian failure is due to GnRH analogue treatment or the shorter follow‐up. In a rodent study, Meirow et al. (1999) demonstrated that, albeit the residual follicle count was related to the dose of chemotherapy, primordial follicle destruction occurred in all mice receiving different doses of cyclophosphamide irrespective of age. In that study, the dose of cyclophosphamide ranged from 20 to 100 mg/kg. It was notable that, despite significant follicle loss, the short‐term reproductive performance of cyclophosphamide‐treated mice was not affected compared with controls. The authors concluded that immediate reproductive performance was not an accurate marker for the assessment of chemotherapy‐induced ovarian damage. In a small prospective randomized study, Waxman et al. (1987) demonstrated that GnRH analogue treatment was ineffective in preserving fertility in patients receiving chemotherapy for Hodgkin’s disease. In that study, 30 men and 18 women were randomly allocated to receive GnRH analogue before, and for the duration of, cytotoxic chemotherapy. Twenty men and eight women received buserelin. After 3 years of follow‐up, all men in both study and control groups became oligo/azoospermic. Among the women, four of eight in the treatment and six of nine female controls became amenorrhoeic. In an adult ovary, ovarian reserve is made up of primordial follicles that constitute ∼90% of the total follicle pool (Lass et al., 1997; Oktay et al., 1997a). These follicles are at the resting stage with an oocyte arrested at the prophase of the first meiotic division. Primordial follicles initiate follicle growth through an unknown mechanism which is FSH independent (Gougeon, 1996; Oktay et al., 1997b, 1998a; McNatty et al., 1999; Meduriet al., 2003). FSH receptors are not expressed in primordial and primary follicles, but the expression is uniformly present in as early as 3–4 granulosa layer preantral follicles. Ovarian suppression with GnRH analogues preserves these follicles that have initiated growth. However, growing follicles not only constitute <10% of all follicles at any given time in the ovary, but once growth has been initiated, they are destined either to become atretic or to ovulate. GnRH analogue co‐treatment delays the fate of these follicles, hence giving the impression that ovarian function is protected in the short run. It has also been suggested that the effects of GnRH analogues may be explained through direct actions of GnRH in the ovary (Blumenfeld, 2003). It was shown that luteinized immortalized granulosa cell lines express GnRH receptors (Cheng et al., 2002); however, to our knowledge, the presence of these receptors on human primordial follicles or oocytes has never been demonstrated. A clinical example of why gonadal suppression may not protect ovaries is the fact that pre‐pubertal children receiving heavy chemotherapy still suffer from ovarian failure (Teinturier et al., 1998). However, since younger patients have a larger ovarian reserve, the absence of immediate ovarian failure does not mean that gonads are unaffected by the chemotherapy (Meirow et al., 1999). All of the patients who receive heavy gonadotoxic chemotherapy will eventually suffer from premature ovarian failure (Viviani et al., 1999; Grigg et al., 2000). In the absence of a prospective randomized study with sufficient power, we do not rely on ovarian suppression as an effective means of fertility preservation.

When the risk of ovarian involvement with cancer cells is high, some other experimental options may be considered in the future. It has been possible to xenograft human ovarian tissue to immunodeficient mice and grow mature follicles in these xenografts (Oktay et al., 1998a,b; Gook et al., 2001). It has also been possible to retrieve oocytes from xenografted human ovarian cortical pieces (Revel et al., 2000). However, the possibility of trans‐species viral infections has to be addressed. Primordial follicles can also be isolated from cryopreserved ovarian tissue, and it is theoretically possible to use these follicles for in vitro maturation (Oktay et al., 1997a). Even though this has been partially successful in mice, the prospect for success in humans is not clear at present. A combination of oocyte and ovarian tissue cryopreservation has also been suggested as a new strategy (Revel et al., 2003). Another possibility is in vitro growth of primordial follicles isolated from cryopreserved ovarian tissue. It has been possible to isolate primordial follicles from human ovarian tissue (Oktay et al., 1997a), but there has been no success in growing them in vitro. Progress has been made in rodents with this technique including production of oocytes competent of meiotic maturation, fertilization and preimplantation in vitro from the primordial follicle (Cortvrindt and Smitz, 2001; O’Brien et al., 2003) but whether this will translate to human studies is currently unclear. It has recently been postulated that the mechanism of age‐related physiological as well as chemo‐ or radiotherapy-induced loss in the ovarian germ cell population is mediated by programmed cell death (Perez et al., 1997; Morita and Tilly, 1999). Sphingosine‐1‐phosphate (S1P), a bioactive sphingolipid metabolite formed by sphingosine kinase, is an important lipid mediator and has many actions both inside and outside the cell. Morita et al. (2000) showed that wild‐type mice treated with S1P resisted both developmental and cancer therapy‐induced apoptosis. Radiation‐induced oocyte loss could be completely prevented by S1P therapy in wild‐type mice. The same group investigated transgenerational genomic instability and failed to demonstrate discernable propagation of genomic damage in mice pretreated with S1P before receiving ionizing radiation (Paris et al., 2002). One of the limiting factors in ovarian transplantation is the loss of a large fraction of ovarian reserve due to initial ischaemia. In theory, cryopreservation of an intact ovary with its vascular pedicle would lend itself to whole organ transplantation with vascular anastomosis, thus enabling us to avoid ischaemia. Even though the preservation of ovarian architecture and restoration of the reproductive function after transplantation of fresh or frozen-thawed intact ovary to an orthotopic or heterotopic site using microvascular anastomosis was recently demonstrated in mice and sheep (Jeremias et al, 2002; Wang et al., 2002; Bedaiwy et al., 2003; Chiu and Hu, 2003), it has not been technically possible to cryopreserve an entire human ovary with its vascular pedicle: first, the human ovary is larger than the ovaries of the animals used in the aforementioned studies; and second, it may be challenging to devise a cryopreservation protocol that will optimally preserve both the follicles and vasculature. Nevertheless, the research on cryopreserving the whole human ovary is continuing, and this technique may one day become clinically possible.

IVF with donor oocytes is another alternative in patients who suffer from premature menopause or low ovarian reserve due to cancer treatment (Polak de Fried et al., 1998). The success rates with appropriate oocyte donors are now >60% per embryo transfer. Gestational surrogacy can also be employed in patients who had undergone a hysterectomy or received pelvic radiation for cervical cancer. Patients with breast cancer who are considered high risk for recurrence, or who have to be on lifelong therapy with aromatase inhibitors, may also resort to gestational surrogacy. However, laws and regulations regarding this procedure vary significantly between countries, and between each state in the USA.