A rational approach to the management of low responders in in-vitro fertilization: Opinion

Low ovarian response to stimulation occurs in ~9–24% of patients (Keay et al., 1997) and still represents one of the most intractable problems of infertility treatment. Clinicians have historically approached low responders in two ways. A first option is to refuse patients entry into in-vitro fertilization (IVF) cycles. Such an approach is, however, obviously discriminatory and serves no practical purpose except for protecting a programme's pregnancy rate (a rather self-serving and medically inappropriate choice). This leaves, as the only ethical choice, attempts to maximize outcomes in IVF cycles for low responders by attempting to improve their historically poor pregnancy and delivery rates. This paper will briefly review the various stimulation protocols (successful and unsuccessful) that have been attempted to improve IVF results in low responders. Included is our experience with a protocol that has met with some success. We would also like to propose a strategy where one does specific testing to predict low response and uses appropriate stimulation protocols to improve success rates. The original definition of low response to ovarian response by Garcia et al. was based on low peak oestradiol concentrations alone (Garcia et al., 1983). They stimulated patients with 150 IU of human menopausal gonadotrophin (HMG) i.m. daily, and defined low responders as patients with a peak oestradiol concentration of <300 pg/ml (1101.3 pmol/l). These patients produced few follicles and had fewer oocytes retrieved, fertilized and transferred, and a lower ongoing pregnancy rate compared to normal or high responders. Since then, stimulation protocols have become more aggressive and the definition of low responders has further evolved. Several programmes now define low responders as patients with peak oestradiol concentrations <500 pg/ml (1835.5 pmol/l), <4 dominant follicles on day of human chorionic gonadotrophin (HCG) administration, older patients (⩾40 years), and those with diminished ovarian reserve based on day 3 follicle stimulating hormone (FSH) concentrations or a clomiphene citrate challenge test (Table I). However, there is no universally accepted definition of low responders, although there is general agreement that this group of patients has a lower pregnancy rate than their normally responding counterparts. Other terms that have been interchangeably used to describe low responders include `poor responder', `bad responder', and `non-responder'. The poor response to stimulation with gonadotrophins is often age-related, but can also occur in young patients. One hundred and forty three low responders were analysed according to age (Hanoch et al., 1998) and it was noted that, although younger patients (⩽30 years) had similar stimulation characteristics to older women, the pregnancy rates were higher in the younger group. Tests of functional reserve of the ovaries can often be used to predict low response to standard protocols (Sharara et al., 1998). These include day 3 FSH, luteinizing hormone (LH), oestradiol, the clomiphene citrate challenge test, and inhibin B concentrations (Loumaye et al., 1990; Seifer et al., 1997; Noci et al., 1998; Sharara et al., 1998). Other tests include the day 3 FSH:LH ratio, gonadotrophin-releasing hormone agonist (GnRHa) stimulation test, and the exogenous follicle stimulating hormone ovarian reserve test (Fanchin et al., 1994; Mukherjee et al., 1996). However, despite the validity of all these tests, there still remain patients who respond poorly to stimulation despite having normal tests of ovarian reserve. Poor ovarian response may, therefore, not be predictable until a patient has failed stimulation under a standard stimulation protocol. Imaging techniques may further help in predicting poor response in patients with normal ovarian functional reserve tests. Measurement of ovarian volume (Lass et al., 1997), antral follicle count (Chang et al., 1998; Pellicer et al., 1998), and measurement of ovarian stromal blood flow with colour Doppler (Engmann et al., 1999) seem to be promising modalities. Several stimulation protocols have been proposed to try and improve outcomes in low responders (Table I). These include: (i) varying the dose or the day of cycle for initiating stimulation with gonadotrophins (Hofmann et al., 1989; Karande et al., 1990; van Hooff et al., 1993; Rombauts et al., 1998); (ii) pituitary desensitization with a GnRHa in the luteal phase of the previous cycle, followed by stimulation with a high dose of gonadotrophins (`long protocol') (Serafini et al., 1988; Belaisch-Allart et al., 1989; Land et al., 1996); (iii) initiating GnRHa and gonadotrophins together in the follicular phase `flare protocol') (Howles, et al., 1987; Karande et al., 1997a); (iv) co-treatment with oestrogens, growth hormone or birth control pills (Gonen et al., 1990; Dor et al., 1995; Russell et al., 1995); (v) using clomiphene citrate for stimulation (Awonuga and Nabi, 1997). and (vi) Natural cycle IVF (Lindheim et al., 1997). All of these strategies have met with only limited success. Three recent approaches that have shown improvement in results with low responder's merit, however, further discussion. Micro-doses of GnRHa have been used to stimulate low responders (Scott et al., 1994). The patients were initially pretreated with a cycle of oral contraceptives to prevent corpus luteum rescue. They then received 20 μg leuprolide acetate (Lupron; Tap Pharmaceuticals, North Chicago, IL, USA) twice daily starting on cycle day 2 and continuing until the administration of HCG. They also received 300 IU of gonadotrophins daily i.m., starting on cycle day 4. A majority of patients responded with higher peak oestradiol concentrations than previous cycles, an increase in number of follicles, and an increase in the number of oocytes retrieved. This protocol also successfully prevented premature LH surges. In a further evaluation of the clinical and endocrine effects of this protocol (Surrey et al., 1998), FSH concentrations were found to be significantly higher on treatment day 6 in patients on the micro-dose protocol (Lupron, 40 μg twice daily s.c.) when compared to controls. There was, however, no increase in testosterone or progesterone concentrations. The authors attribute the resultant improvement in ovarian response and clinical outcome to an increase in endogenous FSH concentrations without concomitant deleterious rises in androgen concentrations. Scott et al. have since modified their protocol by increasing the dose of GnRHa to 50 μg twice daily and starting gonadotrophin stimulation on cycle day 3. They recently reported a pregnancy rate of 40% per cycle in patients with normal ovarian reserve. In patients with diminished ovarian reserve, the pregnancy rate was similar (<5%) (Scott, 1996). An increase in ovarian response by reducing the dose of GnRHa used has been reported (Feldberg et al., 1994). In their `mini-dose' protocol, patients receive s.c. injections of 0.5 mg decapeptyl until menstruation occurred. The dose of the GnRHa was then reduced to 0.1 mg daily from day 2 until HCG was administered. Gonadotrophin stimulation was initiated in the early follicular phase in a dose of 225 IU daily i.m. Mini-dose protocol patients had a lower cancellation rate, higher peak oestradiol concentrations, utilized fewer ampoules of gonadotrophins, and had a shorter length of stimulation than controls who were stimulated without reduction in the dose of decapeptyl. They also noticed an increase in the number of oocytes harvested, increased fertilization rates, and an increased number of embryos transferred. The pregnancy rate was 28.1%, implantation rate was 9.1%, and miscarriage rate was 22.2%. They also noted no rise in progesterone concentrations, suggesting an absence of premature LH surges. The Norfolk group (Faber et al., 1998) used the work of Feldberg et al. (1994) to develop a protocol under which they terminated GnRHa (Lupron) with the onset of menses and followed up with high-dose gonadotrophin therapy (the `stop-Lupron' protocol). They argued that such a protocol maximizes ovarian response without losing the benefits of GnRHa down-regulation and presented their experience with 182 low responders undergoing 224 IVF–embryo transfer cycles. Only one patient amongst 80 cycles triggered a spontaneous, premature LH surge prior to HCG administration. They also reported a clinical pregnancy rate, ongoing pregnancy rate per transfer and implantation rate of 32%, 24% and 9% respectively. Their cycle cancellation rate was only 12.5% (28 cycles) and they noted no difference in outcome between stimulation with FSH alone versus FSH with HMG. We prospectively evaluated the `stop-Lupron' protocol (Faber et al., 1998) at our Center (Karande et al., 1997b) Improved pregnancy rate in poor responder patients with cessation of GnRHa down-regulation prior to stimulation with high dosages of gonadotrophins. The study population included 82 consecutive low responders who underwent IVF–embryo transfer between January 1996 and October 1996. Low responders were defined as patients with either a history of a cancelled cycle or low response with standard protocol [peak oestradiol <500 pg/ml (1835.5 pmol/l), or ⩽4 mature oocytes retrieved (n = 56)]. However, we also included patients with suspected abnormal ovarian reserve because we prospectively anticipated a low response in such patients. These included patients with an elevated day 3 FSH [>7 but <12 mIU/ml (7 pmol/l), n = 33] and/or elevated day 3 oestradiol concentration [>75 pg/ml (275.3 pmol/l), n = 8] in a non-treatment cycle (Smotrich et al., 1995) and patients' age ⩾40 years (n = 24). Some patients had more than one abnormality. Patients with a day 3 FSH >12 mIU/ml [equivalent to a concentration of 25 mIU/ml using the Leeco assay (Scott et al., 1991)] were excluded from the study as, historically, we have not had a single live-birth in this group of patients. Twenty-six cycles (31.6%) were cancelled due to poor ovarian response. Fifty-one low responders reached retrieval and 48 had an embryo transfer. None of the patients had a premature LH surge. Based on a presumed low oocyte quality we were liberal with the number of embryos transferred in the low responder group (4.1 ± 2) and none therefore had any excess embryos for cryopreservation. The clinical pregnancy rate per started cycle was 19.5% (16/82) and per retrieval 31.4% (16/51). Surprisingly, we had a very high incidence of multiple pregnancies (43.8%). Of the 16 pregnancies, seven were singleton, five were twin, two were triplet, and there were two quadruplet pregnancies. This is in contrast to a similar group of patients, which were stimulated with a `flare' protocol where we had a dismal success rate (Karande et al., 1997a). Increasing the number of embryos transferred in low responder patients, therefore, does not seem to be a good strategy. We are currently investigating the role of blastocyst transfer in low responder patients (Gardner et al., 1998). The mechanism by which the `stop-Lupron' protocol apparently improves ovarian responsiveness is unknown. It is possible that Lupron has a direct inhibitory effect on the ovaries and that, by reducing the dose or stopping it altogether, it removes this suppression and increases ovarian response (Parinaud et al., 1992; Kowalik et al., 1998). The mechanism of continued suppression (despite 11.1 ± 1.5 days of stimulation) of premature LH surges under this protocol is presently also still unknown and needs to be further investigated. Continuous suppression of LH after stopping leuprolide acetate has been reported (Sungurtekin and Jansen, 1995): a brief, 5 day course of GnRHa appeared to suppress endogenous GnRH activity for at least 1 week afterwards. In 19 patients, after stopping leuprolide acetate, LH was often undetectable within 48 h and remained so for at least 7 days. Moreno et al. investigated whether the routine use of intracytoplasmic sperm injection (ICSI) would improve outcomes in low responders (defined by them as patients with ⩽6 retrieved oocytes). They found no improvement in fertilization rates, fertilization failure, embryo quality, embryos transferred or reproductive outcome (Moreno et al., 1998). ICSI, therefore, does not seem to be useful in low responders. The role of routine assisted zona hatching in low responders remains controversial (Schoolcraft et al., 1997). One of the concerns in patients who fail to respond to repeated ovarian stimulation with gonadotrophins is the development of ovarian failure. Farhi et al. recently reported their experience with 12 infertile women who did not respond to repeated ovarian stimulation with gonadotrophins and subsequently developed ovarian failure within a short time span (Farhi et al., 1997). These women were initially diagnosed as having unexplained or anovulatory infertility and had a normal baseline hormonal profile. Within a mean of 9 months following the failed attempts of ovarian stimulation, the mean day 3 FSH concentrations rose from 5.4 ± 2.7 to 53.5 ± 19.7 IU/l. We are currently monitoring comparable patients to see whether we can confirm these findings. Insulin-like growth factor-I (IGF-I) concentrations were found to be reduced in the follicular fluid of low responders (Oosterhuis et al. 1998). They suggest that since the capacity of the ovary to grow follicles and mature oocytes is at least partially regulated by the IGF system, lower concentrations of IGF-I with an increased death of follicles may be the main cause of imminent ovarian failure in these patients. In conclusion, we would like to recommend that one should test ovarian reserve in all patients prior to stimulation in an effort to predict those who will respond poorly to standard stimulation. Every centre should establish criteria to screen out patients with severe `egg factor' where the chances of live-birth are practically zero (e.g. a day 3 FSH concentration repeatedly >12 mIU/ml). These patients should not undergo stimulation and should instead be recommended no treatment or offered oocyte donation. Patients with an anticipated low response should then be stimulated with one of the protocols which have been shown to be effective in low responders. Data from the literature as well as our own uncontrolled data suggest that a `stop-Lupron' protocol will improve outcome in such patients. This is, however, only one centre's uncontrolled experience, and randomized, prospective studies that test these protocols against each other would seem beneficial. Patients who fail to respond to these protocols are also candidates for oocyte donation. Strategies to stimulate low responder patients Strategies to stimulate low responder patients