Human Reproduction vol.15 no.6 pp.1231–1234, 2000
Recombinant human FSH versus highly purified urinary FSH: a randomized study in intrauterine insemination with husbands’ spermatozoa
R.Matorras1, V.Recio, B.Corco´stegui and F.J.Rodrı´guez-Escudero Human Reproduction Unit, Department of Obstetrics and Gynaecology, Hospital De Cruces, Paı´s Vasco University, Baracaldo, Vizcaya, Spain 1To
whom correspondence should be addressed at: Marı´a Diaz de Haro 7, 6 iz, 48013 Bilbao, Spain. E-mail:
[email protected]
A randomized trial was carried out comparing recombinant FSH (rFSH) and highly purified urinary FSH (uFSH) in intrauterine insemination (IUI) with husbands’ spermatozoa. A total of 45 women received rFSH (139 cycles), while 46 women received uFSH (155 cycles). The starting dose was 150 IU/day s.c., beginning on the second day, and on days 6–7 the dose was adjusted according to ovarian response, assessed by vaginal ultrasound and plasma oestradiol concentration. The pregnancy rate according to the intention to treat was 57.8% in rFSH versus 52.2% in uFSH, the corrected pregnancy rates 56.8% and 52.2%, and the cumulative pregnancy rates 69.6% and 61.0%, but the differences were not statistically significant. The per cycle pregnancy rate was 18.12% in rFSH and 15.48% in u-FSH, also not statistically significant. In the rFSH group, the consumption of FSH ampoules per cycle was significantly lower (19.20 ⍨ 7.02 versus 23.80 ⍨ 10.78; P < 0.0001). The ratio of oestradiol/FSH ampoules was significantly higher in rFSH (56.45 ⍨ 31.26 versus 46.41 ⍨ 29.25; P < 0.001). These data indicate that, in IUI cycles, rFSH has a higher potency than uFSH. Key words: intrauterine insemination/oestradiol/ovarian stimulation/pregnancy rate/recombinant FSH
Introduction Ovarian stimulation plays an important role in intrauterine insemination (IUI) with husbands’ spermatozoa. It is well established that gonadotrophin stimulation results in higher pregnancy rates than does no stimulation or clomiphene stimulation (Tredway et al., 1990; Matorras et al., 1997; Guzik et al., 1999). Although until recently all available human FSH pharmaceutical preparations were extracted from postmenopausal urine, recombinant human FSH (rFSH) became available a few years ago (Chappel et al., 1992; Olijve et al., 1996). The manufacture of rFSH has been reported to have the following advantages: that production is rendered independent of urine collection; that batch-to-batch consistency can be guaranteed; that the injection of potentially allergenic proteins © European Society of Human Reproduction and Embryology
can be avoided; and that the risk of potential transmission of infectious diseases may also be avoided (Goa and Wagstaff, 1998). Recently, the first trials in IVF were performed, and rFSH demonstrated a superior ovarian response (Out et al., 1995; Bergh et al., 1997). In a recent meta-analysis (Daya and Gunby, 1999) of 12 randomized trials (the majority of which have not yet been published in peer reviews) in IVF–intracytoplasmic sperm injection (ICSI), it was concluded that the pregnancy rate (PR) was statistically higher with rFSH than with urinary FSH. However, these aspects have not been analysed in IUI. The aim of our study was to evaluate the efficacy and the potency of rFSH compared with highly purified urinary FSH (uFSH) when used for ovarian stimulation in infertile patients undergoing IUI. Materials and methods Patients were recruited between September 1997 and September 1998 and included in the study if they satisfied the following criteria: (i) a history of infertility of ⬎2 years; (ii) woman’s age between 18 and 40 years; (iii) at least one normal patent tube; (iv) willingness to participate and comply with the protocol for the duration of the study; and (v) completion of the informed consent form. The main indications were: (i) male factor defined as a subnormal sperm analysis following World Health Organization criteria (WHO, 1992) (except for motility, which was studied according to our reference values) (Matorras et al., 1998), in which after preparation with density gradient centrifugation through silicone-coated silica particles (PureSperm; Nidacon International AB, Gothenburg, Sweden), it was possible to obtain at least 5⫻106 million motile spermatozoa/ml; (ii) failure to obtain pregnancy in six cycles of programmed intercourse, under ovarian stimulation with gonadotrophins; or (iii) idiopathic infertility. The study was approved by the Institutional Board. The pre-study work-up for the women included pelvic and systemic examination, blood chemistry, endometrial biopsy, plasma progesterone and prolactin concentration, postcoital test, hysterosalpingography, pelvic ultrasound and laparoscopy. The following pathologies were detected in the women, some of whom had more than one condition: 22.0% mild tubal factor, 31.9% endometriosis, 12.1% ovulatory disorders, 5.5% hyperprolactinaemia, 9.9% cervical factor, 17.0% uterine myomata (⬍4 cm), and 3.3% immunological factor. Semen samples were obtained after 4 days of sexual abstinence and studied according to the WHO standards (WHO, 1992), except for progressive motility which was analysed following our laboratory reference values. Asthenozoospermia was defined if there were ⬍30% of progressive motile spermatozoa (Matorras et al., 1998). The main male diagnoses were: oligozoospermia (4.4%), asthenozoospermia (58.2%), teratozoospermia (29.7%), and immunological factor (5.5%). In all cases of male factor infertility there was asthenozoospermia, alone or in combination with the other male conditions described above. The IUI method has been reported previously (Matorras et al.,
1231
R.Matorras et al.
1995, 1997). Ovarian cycle stimulation was started on the second day of the menstrual cycle, with 150 IU of rFSH (Gonal F; Laboratorios Serono, Madrid, Spain) or with 150 IU of uFSH (Metrodin HP; Laboratorios Serono). After a short period of training, patients selfadministered either drug s.c. Ovarian monitoring was carried out by means of vaginal ultrasound and oestradiol monitoring, starting on day 6–7 of the menstrual cycle, and adjusting the dose of gonadotrophins according to response. Human chorionic gonadotrophin (HCG, 5000 IU) (Profasi; Laboratorios Serono) was given on the day on which there were two or more follicles of 艌17 mm diameter, the oestradiol concentration being ⬎400 pg/ml (conversion factor to SI units, 3.671). If the two aforementioned criteria were not reached, the cycle was cancelled because of low response. Patients with declining oestradiol concentrations were also cancelled. If there were more than six follicles 艌17 mm diameter and/or oestradiol concentrations ⬎2000 pg/ml, then the cycle was cancelled because of hyper-response (risk of hyperstimulation/multiple pregnancy). In patients in whom two cycles were cancelled, in the following cycles the ovarian stimulation was performed under gonadotrophin-releasing hormone (GnRH) agonist treatment with 0.75 mg/day leuprolide acetate (Procrin; Abbott, Madrid, Spain), beginning on the 22nd day of the previous cycle. A single insemination per cycle was performed with 0.3–0.5 ml of prepared spermatozoa, 36 h after HCG administration. Sperm preparation was performed with PureSperm (Centola et al., 1998; Claassens et al., 1998). In all cases the luteal phase was supplemented as reported previously (Matorras et al., 1996, 1997): with HCG in cases with oestradiol concentrations ⬍1500 pg/ml (2500 units on day ⫹1, ⫹3, ⫹5 and ⫹7 following IUI); or with vaginal micronized progesterone in cases with oestradiol concentrations ⬎1500 pg/ml (Utrogestan; Laboratoires BesinsIscovesco, Paris, France) (100 mg/12 h during the 15 days following IUI). A total of six cycles of IUI was performed if pregnancy was not obtained previously. In all cases, pregnancy was defined by the visualization of a gestational sac at the sixth to seventh week of amenorrhoea.
Table I. Number of patients receiving ovarian stimulation by recombinant follicle stimulating hormone (rFSH) or urinary FSH (uFSH) according to the number of stimulation and insemination cycles undergone Total no. of stimulation/ insemination cycles undergone
1 2 3 4 5 6
No. of patients receiving stimulation by: rFSH
uFSH
15 7 2 10 2 9
12 8 6 3 5 12
Table II. Main demographic characteristics of patients receiving ovarian stimulation by recombinant human follicle stimulating hormone (rFSH) or highly purified urinary FSH (uFSH)a Characteristic
rFSH
Women’s age (years) Men’s age (years) Duration of infertility (years) Previous children (%) Previous abortions (%) Ovulatory disorder (%) Mild–moderate tubal factor (%) Endometriosis (%) Male factor (%) Sperm concentration (⫻106/ml) Progressive motility (%) Normal forms (%) Sperm concentration after PureSperm (⫻106/ml) Progressive motility (%) after PureSperm
33.27 36.14 4.64 9.5 21.7 11.1 22.2 28.9 57.77 84.58 32.92 29.46 59.66
uFSH ⫾ 3.41 ⫾ 3.67 ⫾ 1.96
⫾ ⫾ ⫾ ⫾
29.25 15.84 10.78 28.47
75.52 ⫾ 9.72
33.93 35.95 5.33 12 20 13.6 20.0 34.7 58.69 85.51 30.83 30.98 62.7
⫾ 3.07 ⫾ 3.65 ⫾ 2.49
⫾ ⫾ ⫾ ⫾
38.03 11.87 12.29 29.75
73.83 ⫾ 15.96
are reported as percentages or as mean ⫾ SD. There were no significant differences between the two patient groups.
aValues
Design of the study The 91 couples studied were divided randomly into two groups, by means of sealed envelopes: 45 in the rFSH group (out of 163 started cycles, 24 were dropped, corresponding to 139 IUI cycles) and 46 in the uFSH group (out of 182 started cycles, 27 were dropped, corresponding to 155 cycles). The randomization was performed by a person not involved in the study. Computer-generated random numbers were placed into sealed envelopes. Patients were blinded with regard to the type of treatment, as were the ultrasound staff, oestradiol analysis and sperm laboratory. The prescribing gynaecologist was not blinded. Patients were randomized at the beginning of the study and remained in the same treatment during all the IUI cycles. The frequency of the number of cycles in each group was as follows: one cycle (45 rFSH, 46 uFSH); two cycles (30 and 34); three cycles (23 and 26); four cycles (21 and 20); five cycles (11 and 17); and six cycles (nine and 12). The numbers of patients in each group undergoing the varying numbers of stimulation cycles are shown in Table I. Characteristics of both groups Both groups were comparable with regard to the duration of infertility, the proportion of secondary infertility, the proportion of normal women and other infertility parameters (Table II). Nor were there any other differences in the women’s characteristics. The following male conditions were similar in both groups: asthenozoospermia (57.8% and 58.7%), oligozoospermia (2.2% and 6.5%), teratozoosper-
1232
mia (37.8% and 21.7%) and immunological factor (2.2% and 8.7%). rFSH and uFSH groups were comparable regarding sperm characteristics, both in the unprepared specimen and the prepared specimens (Table I). Statistical analysis Statistical analysis was performed by means of χ2, Mann–Whitney, two-tailed Fisher and Student’s t tests following the standard criteria of applicability. Pregnancy rates were tested by means of the odds ratio (OR) and its 95% confidence interval (CI). Cumulative pregnancy rates were calculated following life-table analysis (Cramer et al., 1979). Statistical significance limit was defined as α ⫽ 0.05.
Results No significant differences were found between the demographic characteristics of the two groups (Table I), thus suggesting that they were adequately randomized. With regard to pregnancy rates, one spontaneous pregnancy occurred in one rFSH patient, after failure to obtain pregnancy in one IUI cycle, another cycle being cancelled. Thus, the PR has been calculated both including this case (per intention to
Recombinant versus purified FSH in intrauterine insemination
Table III. Pregnancy rates (PR) during treatment with recombinant follicle stimulating hormone (rFSH) and highly purified FSH (uFSH)a rFSH Per woman (%) n Per intention to treat PR Corrected PR Cumulative PR Per cycle (%) n First cycle PR Crude PR Corrected PR
uFSH
45 57.8 (26/45) 56.8 (25/44) 69.6
46 52.2 (24/46) 52.2 (24/46) 61.0
139 28.88 (13/45) 17.99 (25/139) 18.12 (25/135)
155 23.91 (11/46) 15.48 (24/155) 15.48 (24/155)
aValues
are reported as percentages. There were no significant differences between the two patient groups. Values in parentheses correspond to numbers of cases.
Table IV. Ovarian cycle parameters of intrauterine insemination cycles during treatment with recombinant follicle stimulating hormone (rFSH) and highly purified FSH (uFSH)a rFSH No. of days of stimulation Insemination day No. of FSH ampoules No. of ultrasounds No. of oestradiol analyses Oestradiol on the day of HCG (pg/ml) No. of follicles 艌16 mm diameter Cancellation rate (%) Oestradiol/FSH ampoules ratio No. of gestational sacs at 6 weeks Mean no. of conceptional cycles
10.49 14.38 19.20 2.22 2.60 976.4
uFSH ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.67 2.64 7.02 0.84 0.85 451.8
3.75 ⫾ 2.28
10.54 14.48 23.80 2.19 2.52 924.0
P ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.54 2.59 10.78 0.84 0.96 389.4
4.5 ⫾ 2.19
NS NS ⬍0.0001 NS NS NS 0.004
14.7 14.8 NS 56.45 ⫾ 31.26 46.41 ⫾ 29.25 ⬍0.001 1.41 ⫾ 0.96 1.45 ⫾ 0.8 NS 2.12 ⫾ 1.51
2.08 ⫾ 1.38
NS
are mean ⫾ SD. NS ⫽ not significant.
aValues
treat PR) and excluding it (corrected PR) (Table III). The PR calculated on a per woman basis were somewhat higher in rFSH, both on a per intention to treat basis (57.8% versus 52.2%) and on a per woman corrected PR (56.8% versus 52.2%; OR ⫽ 1.20, 95% CI ⫽ 0.48–3.0) and cumulative PR (69.6% versus 61.0%), but without statistical significance. The per cycle crude PR was 17.99% in rFSH versus 15.48% in uFSH, which again lacked statistical significance (Table II). Nor were there differences concerning the per cycle corrected PR (18.12% versus 15.48%; OR ⫽ 1.21, 95% CI ⫽ 0.62– 2.33). The first cycle PR was 28.88% (13/45) in rFSH versus 23.91% (11/46) in uFSH (not significant; OR ⫽ 1.29, 95% CI ⫽ 0.46–3.64). With regard to ovarian cycle monitoring parameters (Table IV), there were no differences in terms of duration of stimulation, insemination day, oestradiol determinations and ultrasounds performed. Oestradiol concentrations were similar in both groups. The number of FSH ampoules consumed per cycle was lower in the rFSH group (19.20 ⫾ 7.02 versus 23.80 ⫾ 10.78; P ⬍ 0.0001); the number of follicles 艌16 mm diameter was also lower in the rFSH group (3.75 ⫾ 2.28
versus 4.5 ⫾ 2.19; P ⫽ 0.004). The oestradiol/FSH ampoules ratio was significantly higher in rFSH (56.45 ⫾ 31.26 versus 46.41 ⫾ 29.25; P ⬍ 0.001), and the oestradiol/follicle 艌16 mm diameter ratio was also significantly higher (264 ⫾ 120 versus 211 ⫾ 99; P ⬍ 0.001). The cancellation rate was similar in both groups (24/163 or 14.7%, and 27/182 or 14.8% in rFSH and uFSH respectively), the cancellation rate due to an inadequate ovarian response also being similar (14.1% and 12.1%). The causes of cancellation were similar in both groups: hyper-response (1.2% and 1.1%), low response (6.7% and 5.5%), and fall in oestradiol concentrations (4.3% and 3.3%). There was an additional 0.6% cancellation due to non-medical reasons in rFSH (one case because of personal problems) and 2.7% in uFSH (three cases because of personal problems, one case because of failure to obtain the sperm specimen, and one case because of a car accident). The number of FSH ampoules consumed in cancelled cycles was also lower in the rFSH group (17.04 ⫾ 5.46 versus 25.5 ⫾ 13.68; P ⫽ 0.006), the number of oestradiol determinations and ultrasounds being similar. No cases of ovarian hyperstimulation syndrome (OHSS) or other adverse effects were detected in the rFSH or uFSH groups. The proportion of multiple pregnancies was similar in both groups, as follows: one twin, one triplet and two quadruplets (one of which evolved spontaneously to a twin) in rFSH; and five twin, one triplet and one quadruplet in uFSH. The abortion rate was 12.5% (3/24) in uFSH compared with 28% (7/25) in rFSH; these were not significantly different. There was one ectopic pregnancy in uFSH and two ectopic pregnancies in rFSH. It was calculated that with a similar study design (one case/ one control), to demonstrate that the per woman PR found here were statistically different, with a power of 90% and an α-value of 0.05, a total of 1227 study couples and 1227 control couples would be required. Similarly, under the same methodological conditions, totals of 2131 study cycles and 2131 control cycles would be required to achieve statistical significance with regard to per cycle PR. Discussion IUI with husbands’ spermatozoa is a widely employed, albeit controversial technique. Although early studies reported low PR, present rates are much higher. This improvement in PR is mainly due to the use of advances in IVF methodology (semen preparation, ovarian stimulation, cycle monitoring) in IUI. At the present time, IUI is also being used in artificial insemination by donor (Matorras et al., 1996). IUI is usually performed under gonadotrophin ovarian stimulation. A number of gonadotrophin protocols have been described (Guzik et al., 1999), but as far as we know rFSH has not been studied in a randomized trial. A number of theoretical clinical advantages of recombinant gonadotrophins have been reported, including lack of infective risk, homogeneity among batches, and avoidance of the injection of allergenic proteins. It has been suggested that rFSH might have a superior in-vivo activity, as better ovarian responses have been obtained in IVF cycles in 1233
R.Matorras et al.
comparison with uFSH (Out et al., 1995; Bergh et al., 1997; Jacob et al., 1998). Recently, in a meta-analysis of 12 randomized trials (the majority of which have not yet been published in peer reviews) (Daya and Gunby, 1999), rFSH and urinary FSH (either urinary FSH or highly purified urinary FSH) were compared in IVF-ICSI protocols, the conclusion being made that the PR was statistically higher in rFSH than in uFSH. In our study, no statistical difference was obtained in PR, although the rates appeared somewhat higher in the rFSH group (18.12% versus 15.48% per cycle). However, the possibility of a β error cannot be discarded, since to demonstrate that such differences were of statistical significance would require studying about 2500 women and 4200 cycles. It should be noted that the OR obtained in our study for rFSH, both per cycle and per woman (1.21 and 1.20), were similar to the overall total OR (1.20) reported in the aforementioned metaanalysis (Daya and Gunby, 1999), although in the metaanalysis statistical significance was reached. With regard to the potency of the two FSH preparations, we did not find significant differences concerning the duration of stimulation, the insemination day or oestradiol concentrations. As the treatment was adjusted according to response, these findings indicate that ovarian stimulation was similar in both groups. However, the number of ampoules used was about 20% lower in rFSH; thus, when the oestradiol/ampoules ratio was calculated, a higher value was found among rFSH patients, with an increase of 33% compared with uFSH. This finding was consistent with the higher ovarian response reported in IVF programmes with rFSH (Out et al., 1995; Bergh et al., 1997; Jacob et al., 1998), as well as in WHO group II anovulatory patients (Balasch et al., 1998). In our study, in spite of similar oestradiol concentrations, the number of follicles 艌16 mm diameter was somewhat lower in rFSH, with statistical significance (P ⫽ 0.004), the oestradiol/follicle ratio being higher (P ⬍ 0.001). One could speculate that a better follicle development occurred in terms of oestradiol production when stimulated by rFSH. Our results are in contrast to the lower ratio of oestradiol per follicle in rFSH reported in IVF (Jacob et al., 1998). Although in IUI the follicle count is less accurate than in IVF when follicle aspiration takes place, if this finding is confirmed we consider that the difference might also be due to a different follicle response occurring secondary to the different number of follicles obtained, namely 11 in the study by Jacob et al. and 3.75 in the present study. With regard to other parameters of clinical interest, the cancellation rate, the multiple pregnancy rate and oestradiol analyses and ultrasounds performed were similar in both groups. There was no case of OHSS in either of the two groups. In a larger population, it is possible to calculate the estimated frequency of OHSS for rFSH to be ⬍0.7% (⬍1/139). It is concluded that in IUI, rFSH has a somewhat higher potency than uFSH, as shown by the higher oestradiol concentrations obtained per ampoule. On the other hand, rFSH was found to be a safe therapy, from which PR at least equal to those obtained with uFSH were obtained. The apparently higher per cycle PR (not significant), which was consistent 1234
with that reported in a meta-analysis of IVF–ICSI data (Daya and Gunby, 1999) needs to be confirmed in much larger studies. References Balasch, J., Fabregues, F., Pen˜ arrubia, J. et al. (1998) Follicular development and hormonal levels following highly purified or recombinant folliclestimulating hormone administration in ovulatory women and WHO group II anovulatory infertile patients. J. Assist. Reprod. Genet., 15, 552–559. Bergh, C., Howles, C.M., Borg, K. et al. (1997) Recombinant human follicle stimulating hormone (r-hFSH; Gonal F) versus highly purified urinary FSH (Metrodin HP): results of a randomized comparative study in women undergoing assisted reproductive techniques. Hum. Reprod., 12, 2133–2139. Centola, G.M., Herko, R., Andolina, E. and Weisensel, S. (1998) Comparison of sperm preparation methods: effects on recovery, motility, motion parameters, and hyperactivation. Fertil. Steril., 70, 1173–1175. Chappel, S., Kelton, C. and Nugent, N. (1992) Expression of human gonadotropins by recombinant DNA methods. In Genazzani, A.R. and Petraglia, F. (eds), Proceedings of the 3rd World Congress on Gynecological Endocrinology. Parthenon Publishing Group, Carnforth, UK, pp. 179–184. Claassens, O.E., Menkveld, R. and Harrison, K.L. (1998) Evaluation of three substitutes for Percoll in sperm isolation by density gradient centrifugation. Hum. Reprod., 13, 3139–3143. Cramer, D.W., Walker, A.M. and Schiff, I. (1979) Statistical methods in evaluating the outcome of infertility therapy. Fertil. Steril., 32, 80–86. Daya, S. and Gunby, J. (1999) Recombinant versus urinary follicle stimulating hormone for ovarian stimulation in assisted reproduction. Hum. Reprod., 14, 2207–2215. Goa, K.L. and Wagstaff, A.J. (1998) Follitropin alpha in infertility. A review. Bio. Drugs, 9, 235–260. Guzik, D.S., Carson, S.A., Coutifaris, C. et al. (1999) Efficacy of superovulation and intrauterine insemination in the treatment of infertility. N. Engl. J. Med., 340, 177–183. Jacob, S., Drudy, L., Conroy, R. and Harrison, R.F. (1998) Outcome from consecutive in-vitro fertilization/intracytoplasmic sperm injection attempts in the final group treated with urinary gonadotrophins and the first group treated with recombinant follicle stimulating hormone. Hum. Reprod., 13, 1783–1787. Matorras, R., Corco´ stegui, B., Pe´ rez, C. et al. (1995) Sperm morphology analysis (strict criteria) in male infertility is not a prognostic factor in intrauterine insemination with husbands sperm. Fertil. Steril., 63, 604–647. Matorras, R., Gorostiaga, A., Diez, J. et al. (1996) Intrauterine insemination with frozen sperm increases pregnancy rates in donor insemination cycles under gonadotropin stimulation. Fertil. Steril., 65, 620–625. Matorras, R., Pe´ rez, C., Corco´ stegui, B. et al. (1997) Treatment of the male with follicle-stimulating hormone in intrauterine insemination with husband’s spermatozoa: a randomized study. Hum. Reprod., 12, 24–28. Matorras, R., Genolla´ , J., Mendoza, R. et al. (1998) Total immunoreactive alpha inhibin in human seminal plasma, sperm quality and in vitro fertilization rates. Int. J. Fertil., 43, 171–176. Olijve, W., De Boer, W., Mulders, J.W.M. et al. (1996) Molecular biology and biochemistry of human recombinant follicle stimulating hormone (Puregon). Mol. Hum. Reprod., 2, 371–382. Out, H.J., Mannaerts, B.M.J.L., Driessen, S.G.A.J. et al. (1995) A prospective, randomized, assessor-blind, multicenter study comparing recombinant and urinary follicle stimulating hormone (Puregon vs Metrodin) in in-vitro fertilization. Hum. Reprod., 10, 2534–2540. Tredway, D.R., Chan., P., Henig, I. et al. (1990) Effectiveness of stimulated menstrual cycles and Percoll sperm preparation in intrauterine insemination. J. Reprod. Med., 35, 103–108. World Health Organization (1992) Laboratory Manual for the Examination of Human Semen–Cervical Mucus Interaction. 3rd edn. The Press Syndicate of the University of Cambridge, Cambridge, pp. 43–45. Received on October 18, 1999; accepted on February 23, 2000
