RBMOnline - Vol 6. No 2. 185 190 Reproductive BioMedicine Online; www.rbmonline.com/article/725 on web 23 December 2002 Article Improvement in consistency of response to ovarian stimulation with recombinant human follicle stimulating hormone resulting from a new method for calibrating the therapeutic preparation Professor Jean-Noel Hugues trained and works in France where he qualified first as a medical doctor and later obtained his PhD in Biochemistry and the Biology of Reproduction. In 1995 he became Head of the Reproduction Medicine Unit at the Jean Verdier Hospital in Paris. Research interests include clinical pharmacology and pharmacokinetics, human embryology, physiology of development, internal medicine and endocrinology. He is a member of the French Societies of Endocrinology, Andrology, and Reproduction Medicine. Dr Jean-Noel Hugues Jean-Noel Hugues 1,5, David H Barlow 2, Zev Rosenwaks 3, Isabelle Cédrin-Durnerin 1, Susan Robson 4, Lionel Pidoux 4, Ernest Loumaye 4 1 Assistance publique Hopitaux de Paris, Hopital Jean Verdier, Avenue du 14 Juillet, F-93143 Bondy, France 2 University of Oxford, Nuffield Department, Obstretrics and Gynecology, Oxford Radcliffe Hospital, The Women s Centre, Headington, Oxford, OX3 9DU, UK 3 Cornell University, 505 East 70 and York Ave. New York, NY 10021, USA 4 Clinical Development Department, Serono International, 15 bis Chemin des Mines, CH-1211, Geneva, Switzerland 5 Correspondence: e-mail: jean-noel.hugues@jvr.ap-hop-paris.fr Abstract Traditionally, therapeutic preparations of gonadotrophins are quantified with a rat in-vivo bioassay in biological international units (IU). This method was developed to cope with variability of production batch quality. The bioassay, however, presents some limitations, and differences in clinical responses when using different batches of urine-derived gonadotrophins have been reported. The production of human FSH by recombinant technology now allows the use of advanced physicochemical methods for quantifying FSH, which can be measured in µg of FSH proteins (in mass). The study reported here was designed and conducted to assess the clinical relevance of this new method for quantifying therapeutic preparation of FSH. Four bulk lots of recombinant human FSH (r-hfsh) were used to prepare batches filled by IU (FbIU) and four batches filled by mass (FbM). These eight batches were compared in a double-blind, randomized study in patients undergoing assisted reproductive technology. One hundred and thirty-one patients were enrolled in this study and met protocol criteria (66 in the FbM group and 65 in the FbIU group). The starting dose of recombinant human FSH (r-hfsh) was either 150 IU or 11 µg/day. Both preparations induced multiple follicular development and all patients underwent oocyte retrieval. The number of follicles 11 mm was 14.85 and 14.91, serum oestradiol concentration on day of human chorionic gonadotrophin (HCG) administration was 6524 and 6350 pmol/l, number of oocytes retrieved was 10.76 and 11.28, number of two-pronuclear (2 PN) oocytes was 5.2 and 5.00, number of viable embryos (replaced or cryopreserved) was 4.15 and 3.72, and clinical pregnancy rate was 30.3 and 26.2% respectively in the FbM and FbIU groups. Overall, the patients response consistency was found to be superior with FbM (P = 0.039), and in particular for clinical pregnancy rates (P < 0.001). This new method for quantifying r-hfsh delivers an improved consistency in clinical outcome. Keywords: assisted reproduction treatment, bioassay, FSH, gonadotrophins, recombinant technology 185
186 Introduction FSH is a complex heterodimeric glycoprotein secreted by the pituitary gland. Therapeutic preparations of human FSH are used broadly to stimulate ovarian follicle growth in infertile women, and spermatogenesis in hypogonadotrophic hypogonadal men. Until 1995, all human FSH pharmaceutical preparations were extracted from post-menopausal urine i.e. Menotropin or human menopausal gonadotrophin (HMG) and urofollitropin or u-hfsh. Marked heterogeneity with respect to bioactivity, immunoreactivity, LH content and isohormone profiles has been reported between different commercial urinary gonadotrophin preparations (Harlin et al., 1986; Rodgers et al., 1995) and, more importantly, between batches of the same preparation (Cook et al., 1988; Stone et al., 1989; Braileanu et al., 1998). Variability in the quality of the gonadotrophin preparations has been shown to contribute to differences in patient response (Stone et al., 1989; Rector et al., 1993). In order to overcome the variability of urine-derived products, FSH activity in batches of gonadotrophin preparations is assessed by the Steelman and Pohley rat in-vivo bioassay (Steelman and Pohley, 1953). Briefly, each production batch is tested in rats in order to quantify its FSH activity content. The end-point of the test is the rat ovarian weight gain. Measuring rat ovarian weight has inherent limitations, hence the European Pharmacopoeia defines an activity range within which an FSH batch is acceptable for clinical use. The practical consequence is that a labelled 75 IU FSH ampoule may theoretically contain anything between 48 and 117 IU FSH (finished product FSH activity, fiducial limits) (European Pharmacopoeia). Over the last 30 years, attempts to replace the cumbersome invivo bioassay by an immunoassay or an in-vitro bioassay have been undertaken (Rose et al., 2000). However, because the biological activity of FSH is heavily determined by its glycosylation, which defines its pharmacokinetic characteristics as well as its activity at the receptor level (Galway et al., 1990; Chappel, 1995; Vitt et al., 1998; Creus et al., 2001), these attempts have failed to provide a reliable predictor of FSH therapeutic activity (Rose et al., 2000). Recombinant DNA technology allows the in-vitro production of human FSH (r-hfhs) with very high purity and consistent quality (Recombinant Human FSH Development Group, 1998). Recombinant FSH has been demonstrated to be the most effective agent yet developed for stimulating human follicular development (Daya et al., 2001). However, r-hfsh release specifications still include the Steelman and Pohley bioassay. Because of its high purity, each production batch of r-hfsh (Gonal-F ; Serono, Aubonne, Switzerland) has been routinely characterized and controlled using physicochemical techniques, including size exclusion high performance liquid chromatography (SE-HPLC), isoelectric focusing (IEF) (Siebold, 1996) and glycan mapping (Gervais et al., 2003). The first method allows assessment of the integrity of FSH molecules as well as quantifying the amount of FSH, while the last two allow the characterization of FSH glycoforms present in the preparation. This approach is consistent with the current trend for therapeutic proteins in finding alternative approaches to biological assay to assess their activity (Jeffcoate, 1994; Mulders et al., 1997). The result of the characterization of more than 100 batches of this r-hfsh preparation has demonstrated that for this product there is a consistent glycosylation profile in the different production batches, and that there is a constant relationship between the FSH mass (µg) and its biological activity (Driebergen et al., 2002). Because measuring FSH µg by SE- HPLC is more precise, it has been postulated that substituting this assay for the bioassay to calibrate each batch of r-hfsh would deliver more consistent FSH activity. This clinical trial has been designed and conducted to assess the clinical relevance of the improved manufacturing process. Materials and methods This was a multicentre, double-blind, randomized, parallel group study to compare the safety and efficacy of r-hfsh preparations (Gonal-F ) filled and released by µg (FbM) and filled and released by IU (FbIU) in stimulating multiple follicular development prior to in-vitro fertilization (IVF). The study was conducted in three clinical centres according to Good Clinical Practice (GCP) standards, including patient s written informed prior to enrolment in the study. The protocol was approved by each of the three centre s Institutional Review Board or Ethics Committee. Study population Patients admitted to the study were infertile women desiring pregnancy and justifying assisted reproductive techniques (IVF, ICSI), provided that they conformed to the following eligibility criteria: (i) infertility justifying assisted reproduction treatment, (ii) a male partner with semen analysis obtained within 6 months prior to beginning gonadotrophin (GnRH)-agonist therapy with >1.0 10 6 motile spermatozoa (motility grade A or B) per ml in the whole ejaculate and an oocyte fertilization rate 20% during any previous assisted reproduction treatment attempt, (iii) aged 18 38 years (inclusive), a spontaneous ovulatory menstrual cycle of 25 35 days, (iv) early follicular phase (day 2 4) serum FSH <12 IU/l, serum LH <13.5 IU/l, serum prolactin <800 miu/l, serum oestradiol <75 pg/ml, (v) presence of both ovaries, (vi) a body mass index (BMI) <30, (vii) not more than three previous consecutive assisted reproduction treatment cycles without a clinical pregnancy, (viii) no previous assisted reproduction treatment cycle indicating a poor response to gonadotrophin stimulation, (ix) in the early follicular phase, a baseline endovaginal ultrasound scan showing no more than 13 follicles 4 mm and 8 mm on the largest section through one ovary. The primary objective of this study was to confirm the equivalence of rfsh filled and released by mass or by bioactivity in stimulating multiple follicular development in women undergoing IVF and embryo transfer. The set of patients to be used for all statistical analyses was the per protocol dataset. The number of patients in this study was determined on practical and statistical grounds. A maximum difference of 2 in the number of follicles 11 mm was considered as a relevant efficacy concern. Using a one-sided level of significance of 0.05 and assuming that the common standard deviation is 5.4, the power to accept the equivalence between the two treatments was 70% for 65 evaluable patients per group.
Randomization When a patient had signed the informed consent form, was found to be eligible for the study and pituitary down-regulation with GnRH-agonist was confirmed, she was allocated a unique patient identification number in sequential, chronological order. This identification number assigned the patient in a blinded fashion to one of the eight batches of rfsh filled and released by either mass (FbM) or bioactivity (FbIU) according to a computer generated randomization list. The randomization was stratified by treatment and by centre, so that all centres received r-hfsh from the eight batches. Study medication and administration regimen Eight batches of r-hfsh were used in the study. These batches were obtained from four different lots of r-hfsh, with each lot used to produce one batch filled and released by µg and one batch filled and released by bioactivity (Figure 1). All r-hfsh lots were produced at the same production site (Laboratoires Serono, Aubonne, Switzerland). All batches were produced at the same production site (Serono Pharma Bari, SPB, Italy). The finished product was supplied in ampoules delivering either 75 IU of r-hfsh or its mass equivalent, i.e. 5.5 µg FSH, 30 mg sucrose and phosphate buffer in a lyophilized form. Treatment with rfsh began after at least 10 days of GnRHagonist therapy when pituitary desensitization was achieved, and was administered subcutaneously once daily in the abdominal wall. The starting fixed dose was 150 IU FSH/day (i.e. two ampoules/day) for 5 days (inclusive). The dose could be adapted as of day 6 of stimulation, according to the ovarian response monitored by ultrasound and serum oestradiol concentrations. The maximum dose allowed was 450 IUFSH/day (i.e. six ampoules/day). In order to achieve final follicular maturation before oocyte retrieval, urinary human chorionic gonadotrophin (u-hcg) was administered subcutaneously or intramuscularly at a dose of up to 10,000 IU. Oocytes were retrieved 34 38 h after u-hcg administration and fertilized in vitro. Not more than three of the resulting embryos were replaced. Natural progesterone was administered vaginally, every day starting after oocyte retrieval and continuing either up to menstruation, or if the patient became pregnant, for at least the first 3 weeks of pregnancy. The patient was followed up and the treatment outcome (pregnancy or menstruation) was recorded. If, at any time, a clinician suspected that a patient might be at risk of ovarian hyperstimulation syndrome, the dosage of r- hfsh was reduced or administration discontinued. Results Patient disposition and demographics A total of 131 patients were enrolled, randomized to r-hfsh treatment and included in the per protocol population. Sixty-six patients were allocated to r-hfsh FbM and 65 to r-hfsh FbIU. Treatment outcomes were as follows: all patients qualified for and received HCG and underwent oocyte retrieval, and 63 and 61 had at least one embryo transferred in the FbM and FbIU group respectively. In terms of demographic characteristics, the two groups were well balanced. Patients ages were 30.8 ± 4.0 and 31.4 ± 3.5; BMI was 23.1 ± 3.2 and 22.5 ± 3.1; the proportion of non-smoker was 72.7 and 72.3%; the proportion of primary infertility was 60.6 and 56.9%; duration of infertility was 3.7 ± 2.9 and 3.7 ± 2.5 years; mean cycle duration was 28.7 ± 1.5 and 29.0 ± 1.7 days, the proportion of patients without previous assisted reproduction treatment was 74.2 and 73.8%; baseline serum FSH was 7.4 ± 1.4 and 7.4 ± 1.7 IU/l; baseline serum LH was 3.8 ± 1.3 and 3.8 ± 1.2 IU/l; and number of follicle between 4 and 8 mm in diameter in both ovaries was 7.3 ± 4.0 and 8.5 ± 5.2 in the FbM and the IU groups respectively. In addition, no difference was recorded in baseline sperm characteristics between the treatment group partners. The duration of GnRH agonist administration (from start to HCG) was comparable between the two groups, with 28.3 ± 6.5 days and 28.9 ± 6.5 days for the FbM and the FbIU groups respectively. Figure 1. Manufacture scheme for four batches of r-hfsh filled by mass (µg) and four batches filled by biological units (IU). 187
Table 1. Ovarian stimulation results, embryo transfer and pregnancy rates (mean ± SD). Mass Bioassay P-value n 66 65 Duration of treatment (days) 10.44 ± 1.39 10.52 ± 1.39 0.689 No. follicles 11 mm on day 6 1.44 ± 2.51 1.48 ± 2.19 0.922 No. follicles 11 mm on day HCG* 14.85 ± 6.42 14.91 ± 6.41 0.964 No. follicles 14 mm on day hcg 10.03 ± 4.51 10.22 ± 4.18 0.817 Oestradiol on S6 (pmol/l) 686 ± 584 684 ± 676 0.972 Oestradiol on day of HCG (pmol/l) 6524 ± 3413 6350 ± 3164 0.779 HCG received (%) 100 100 1.000 Oocytes retrieved 10.76 ± 4.67 11.28 ± 5.52 0.491 2 PN oocytes fertilized 5.20 ± 3.60 5.00 ± 3.20 0.802 Viable embryos 4.15 ± 2.99 3.72 ± 2.73 0.339 Overall pregnancy rate (%) 36.4 33.9 0.855 Clinical pregnancy rate (%) 30.3 26.2 0.699 OHSS (%) 4 (6.1) 6 (9.2) 0.531 The overall responses to r-hfsh treatment and outcome are presented in Table 1. Both preparations achieved significant ovarian stimulation, resulting in a large number of embryos and a clinical pregnancy rate close to 30% per treated cycle. Variability of the ovarian response to different batches of r- hfsh was compared for all efficacy parameters. Figure 2 shows the mean number of oocytes retrieved in patients treated with the four batches of FbM and the corresponding four batches of FbIU. Figure 3 shows similar presentation for pregnancy rates. For both parameters, the r-hfsh FbM batches appear to deliver a more consistent therapeutic effect. This consistency was quantified and tested comparing the difference between the highest and the lowest mean value in each treatment group for efficacy parameters (Table 2). For the majority of the efficacy parameters, FbM performed more consistently than FbIU, as illustrated by a smaller difference between the mean results obtained in each four patient subgroups. Overall, as well as for clinical pregnancy rates, FbM consistency was found to be statistically superior to FbIU (P = 0.039 and 0.001 respectively). The safety assessment showed the expected adverse events associated with ovarian stimulation and assisted reproduction technology, with the most frequently reported being headaches and abdominal pain and discomfort. Six cases of ovarian hyperstimulation syndrome (OHSS) (four mild and two moderate) were reported in the FbIU group and four (three mild and one severe) were reported in the FbM group (P = 0.531) (Table 2). Discussion The purity of human FSH produced by recombinant DNA technology has made possible the use of state-of-the-art methods for characterizing the integrity of FSH, the FSH glycosylation pattern and for quantifying the FSH content of individual production batches. Extensive experience with rfsh production has revealed a very consistent pattern of integrity and glycosylation overtime (Driebergen et al., 2002; Gervais et al., 2003). This has led to the conclusion that replacing the standard bioassay, which was developed to cope with the variability of urine-derived FSH preparations, by an SE-HPLC 188 a Figure 2. Mean number of oocytes retrieved in patients treated with each batch of r-hfsh. a: four batches of r-hfsh fill by mass (FbM); b: four batches of r-hfsh fill by bioassay (FbIU). b
Pregnancy rate % Pregnancy rate % a b Figure 3. Pregnancy rates in patients treated with each batch of r-hfsh. a: four batches of r-hfsh fill by mass (FbM); b: four batches of r-hfsh fill by bioassay (FbIU). Table 2. Difference between the highest and the lowest mean values observed using four different r-hfsh batches in each treatment group. Mass Bioassay P-value n 66 65 Serum FSH on day 6 1.33 2.09 0.362 Serum FSH on day of HCG 1.81 4.55 0.292 No. follicles 11 mm on day HCG 1.67 3.47 0.134 No. follicles 14 mm on day HCG 0.68 1.94 0.271 Oestradiol on S6 (pmol/l) 300 346 0.457 Oestradiol on day of HCG (pmol/l) 1599 2815 0.308 Oocytes retrieved 1.49 5.41 0.191 2PN oocytes fertilized 1.35 2.07 0.228 Viable embryos 1.84 1.54 0.767 Total no. of FSH ampoules administered 4.84 3.33 0.798 Duration of stimulation (days) 0.52 0.65 0.596 Overall pregnancy rate 13.40 35.30 0.275 Clinical pregnancy rate 5.55 29.42 0.001 OHSS 12.50 5.51 0.440 is not only possible but could further increase the batch-tobatch consistency of rfsh. This clinical study assessed the possible clinical benefit of applying this new calibration method by comparing the clinical response to four batches calibrated in µg (FbM) with four batches calibrated in IU (FbIU). The study results support the selected conversion factor, as illustrated by the clinical outcome in the two treatment groups. More important for clinicians are the results regarding consistency of the clinical response between batches. It is well established that patient s responses to ovarian stimulation treatment are variable. Parameters contributing to this variability are numerous and include patient s age, ovarian reserve, pre-treatment with GnRH analogues and oral contraceptive pill, and polycystic ovaries. The levels of ovarian response directly and indirectly impact on the clinical pregnancy rates, since the latter is directly related to the number of embryos obtained in vitro and to the number of embryos replaced, both of which are strong prognostic factors for conception (Templeton and Morris, 1998). Adding additional variability in this context is unlikely to be beneficial. This study shows that in a homogeneous population of patients receiving assisted reproduction treatment, the improved manufacturing process was associated with an improvement in the consistency of the ovarian response, including significantly improved consistency in the clinical pregnancy rate between batches of gonadotrophin. The variability between clinical outcome resulting from treatment with different FbM batches was reduced, indicating that in a well-defined patient population the quality of the gonadotrophin preparation may have a predominant role in the consistency of clinical response. This study did not include a urinary gonadotrophin arm. However, published data indicate that a larger variation may be expected in clinical response when using different batches of urinederived gonadotrophins. In a retrospective study, comparing nine batches of HMG, Stone et al. reported a mean number of large follicles at OPU ranging between 6.4 ± 0.8 and 10.4 ± 1.4 and a pregnancy rate ranging between 0 and 21% (Stone et al., 1989). 189
190 In conclusion, the consistent quality of the recombinant DNA technology manufacturing process for human FSH has allowed for the first time, the delivery of FSH based on µg of protein. This new method results in further improvement in r-hfsh preparations by delivering increased consistency in clinical outcome. Acknowledgements The authors wish to thank Drs Gillian Lockwood, Lee Lim, Isaac Kligman and Dehan Chen for their highly valued contribution to the conduct of this clinical study, Patrick Engrand PhD for statistical input in study design and analysis, Michele Sauvage, Melissa Allen and Andrea Goodyer for their excellent study monitoring, Lional Pidoux and Hélène Favier for developing an electronic data capture system, and Serono Bari manufacturing site for producing these eight batches of Gonal-F. References Braileanu GT, Albanese C, Card C, Chedrese PJ 1998 FSH bioactivity in commercial preparations of gonadotropins Theriogenology 49, 1031. Chappel SC 1995 Heterogeneity of follicle stimulating hormone: control and physiological function. Human Reproduction Update 1, 479 587. Cook AS, Webster BW, Teranova PF et al. 1988 Variation in the biologic and biochemical characteristics of human menopausal gonadotropin. Fertility and Sterility 49, 704 712. Creus S, Chaia Z, Pellizzari EH et al. 2001 Human FSH isoforms: carbohydrate complexity as determinant of in-vitro bioactivity. Molecular and Cellular Endocrinology 28, 41 49. Daya S, Ledger W, Auray JP et al. 2001 Cost-effectiveness modelling of recombinant FSH versus urinary FSH in assisted reproduction techniques in the UK. Human Reproduction 16, 2479 2480. Driebergen R, Basset R, Baer G et al. 2002 Improvements in quantification of r-hfsh activity: SE-HPLC vs the in vitro rat bioassay. Human Reproduction 17, P-480. Galway AB, Hsueh AJ, Keene JL et al. 1990 In vitro and in vivo bioactivity of recombinant human follicle-stimulating hormone and partially deglycosylated variants secreted by transfected eukaryotic cell lines. Endocrinology 127, 93 100. Gervais A, Hammel Y-A, Pelloux S et al. 2003 Glycosylation of recombinant gonadotrophins: characterisation and batch-to-batch consistency. Glycobiology, in press. Harlin J, Khan SA, Diczfalusy E 1986 Molecular composition of luteinizing hormone and follicle-stimulating hormone in commercial gonadotropin preparations Fertility and Sterility 46, 1055 1061. Jeffcoate SL 1994 The role of bioassays in the assessment of recombinant proteins. Developments in Biological Standardization 83, 159 164. Mulders JW, Derksen M, Swolfs A et al. 1997 Prediction of the in vivo biological activity of human recombinant follicle stimulating hormone using quantitative isoelectric focusing Biologicals 25, 269 281. Recombinant FSH Product Development Group 1998 Recombinant follicle stimulating hormone: development of the first biotechnology product for the treatment of infertility Human Reproduction Update 4, 862 881. Rector NA, Markusen TE, Stone BA et al. 1993 Numbers and quality of oocytes after induction of multiple folliculogenesis in women and in mice with different lots of human gonadotropins. Fertility and Sterility 60, 1082 1087. Rodgers M, McLoughlin JD, Lambert A et al. 1995 Variability in the immunoreactive and bioactive follicle stimulating hormone content of urinary menopausal gonadotrophin preparations. Human Reproduction 10, 1982 1986. Rose MP, Gaines RE, Balen AH 2000 Definition and measurement of follicle stimulating hormone. Endocrine Reviews 21, 5 22. Siebold B 1996 Physicochemical characterisation of recombinant human follicle stimulating hormone. Human Reproduction 11, 109 115. Steelman SM, Pohley FM 1953 Assay of the follicle stimulating hormone based on the augmentation with human chorionic gonadotropin. Endocrinology 53, 604 616. Stone BA, Quinn K, Quinn P et al. 1989 Responses of patients to different lots of human menopausal gonadotropins during controlled ovarian hyperstimulation. Fertility and Sterility 52, 745 752. Templeton A, Morris JK 1998 Reducing the risk of multiple births by transfer of two embryos after in vitro fertilisation. New England Journal of Medicine 339, 573 577. Vitt UA, Kloosterboer HJ, Rose UM et al. 1998 Isoforms of recombinant follicle-stimulating hormone: comparison of effects on murine follicular development in vitro. Biology of Reproduction 59, 854 861. Received 11 July 2002; refereed 31 October 2002; accepted 5 December 2002.