University Clinic of Schleswig Holstein, Campus Luebeck, Ratzeburger Allee 160, 23858 Luebeck, Germany
1 To whom correspondence should be addressed: Email: georg.griesinger{at}frauenklinik.uni-luebeck.de
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: cetrorelix/GnRH-antagonist/ovarian stimulation/recombinant FSH/recombinant LH
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
For the GnRH-agonist long ovarian hyperstimulation protocol, the effect of a profound follicular phase endogenous LH level suppression under recombinant (rec) FSH stimulation has been associated with an adverse reproductive outcome in some studies (Westergaard et al., 2000; Esposito et al., 2001
; Copola et al., 2003
), however, this association has been scrutinized in others (Loumaye et al., 1997
; Balasch et al., 2001
; Penarrubia et al., 2003
). The residual LH level below which follicular development and treatment outcome might be impaired is denominated as the LH threshold level.
Additional LH activity might optimize COH when GnRH-analogues and gonadotropin preparations depleted in LH content are combined. More specifically, it was indicated that the duration necessary to achieve ovulation induction criteria might be shorter and the gonadotropin consumption lower (Filicori et al., 1998, 2003
; Balasch et al., 2003
), when HMG rather than highly purified HMG or recFSH are used in pituitary suppressed patients.
The effect of LH suppression in GnRH-antagonist ovarian hyperstimulation protocols is also an issue of concern. Midcycle antagonist administration leads to a marked decrease in LH levels (Albano et al., 1997), down to the lower detection limit of commercial immunoassays for LH in some patients (Finas et al., 2003
). A dose finding study on the GnRH-antagonist ganirelix revealed very low pregnancy rates under high antagonist doses which in turn are associated with profound LH suppression (Ganirelix dose-finding study, 1998
). Furthermore, in comparative phase III trials the odds of clinical pregnancy was comparatively (but not significantly) lower in GnRH-antagonist study arms in which recFSH void of any LH activity was utilized for stimulation as compared to trials that utilized human menopausal gonadotropin (HMG) (Al-Inany and Aboulghar, 2003
).
In response to these observations, a group comparative trial was undertaken based on the hypothesis that using a combination preparation of recFSH and recLH might optimize ovarian hyperstimulation parameters and infertility treatment outcome in GnRH-antagonist multiple dose cycles, possibly in part by alleviating the pituitary suppressive effect of antagonist administration at a critical point of follicular development.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Patient population
Main inclusion criteria were: indication for treatment with IVF or ICSI; age between 20 and 39 years; body mass index between 18 and 35 kg/m2; regular menstrual cycle, ranging from 24 to 35 days, intra-individual cycle variability of 3 days; use of fresh as well as frozen-thawed sperm retrieved by testicular biopsy. Main exclusion criteria were: >3 previous unsuccessful assisted reproduction technique attempts; previous poor response to gonadotropin stimulation defined as <3 preovulatory follicles; history of ovarian hyperstimulation syndrome (OHSS) grade IIIII; polycystic ovarian syndrome; any other endocrine disorder; no natural luteal phase prior to treatment cycle; abnormal uterine cavity as evaluated by ultrasonography; presence of a clinically significant systemic disease.
Protocol
Ovarian stimulation started on day 2 of the natural cycle with 150 IU recFSH (Gonal-F; Serono, Geneva Switzerland) in the control group (recFSH group), and 150 IU recFSH (Gonal-F) plus 75 IU recLH (Luveris; Serono, Geneva Switzerland) in the study group (recFSH/recLH group). All injections were given once daily s.c. in the morning by the patient. After 5 days of gonadotropin treatment, GnRH-antagonist cetrorelix 0.25 mg (Cetrotide; Serono, Geneva Switzerland) administration was started by once daily s.c. injection in the morning. Gonadotropin and antagonist treatment was continued up to and including the day of HCG administration. From day 6 onwards, the recFSH dosage could be increased at the discretion of the physician to 225 IU or 300 IU according to the ovarian response as assessed by ultrasonography and estradiol (E2) values. In case of a dose increment to 300 IU recFSH in the recFSH/recLH group, the recLH dose was concomitantly adjusted to 150 IU.
RecHCG 250 µg (Ovitrelle; Serono, Geneva, Switzerland) was administered s.c. as soon as three follicles were 18 mm in diameter, and 3436 h thereafter oocyte retrieval was performed. All follicles
13 mm were punctured. After IVF or ICSI according to standard procedures, no more than three embryos were to be replaced on day 2 after oocyte retrieval. Spare two-pronuclei oocytes were frozen. The outcome of cycles utilizing embryos generated from those oocytes is not presented further herein.
Luteal phase support started the morning after oocyte retrieval and was provided with daily 90 mg micronized progesterone (Crinone 8%; Serono, Geneva Switzerland). Additionally, 5000 IU urinary HCG (Choragon; Ferring Arzneimittel GmbH, Kiel, Germany) were administered once on the day of embryo transfer in case E2 levels on the day of HCG were 2500 pg/ml. When pregnancy was achieved, luteal phase support in the form of daily vaginal or twice weekly i.m. progesterone administration was continued until fetal heartbeat was visualized by vaginal ultrasonography.
Compulsory criteria for cycle cancellation on any monitoring visit were: premature luteinization, defined as progesterone rise 1.7 ng/ml; premature LH rise, defined as
12.1 mIU/ml; E2 drop, defined as a reduction of E2 levels of at least 50% between two monitoring visits; imminent OHSS, defined as
15 intermediate follicles (1216 mm) on stimulation day 8 or
20 large follicles (1620 mm) on day 10 or later. An optional criterion for cycle cancellation was poor response, defined as mono- or bifollicular development, in which case patients were free to cancel the cycle or to continue HCG injection.
Assessments
Hormonal and ultrasound assessment was performed on stimulation day 6 just before first antagonist administration and thereafter as necessary, as well as on the day of HCG administration. For determining the optimal time point for HCG administration, a follicular growth of 2 mm per day was assumed. Five medical doctors involved in this trial performed the ultrasound examinations. Ultrasound and hormonal assessment, as well as oocyte retrieval and embryo transfer was performed 7 days a week. Serum E2, LH and progesterone levels were assayed with the electrochemiluminescence immunoassay ECLIA (Roche Diagnostics Inc., Germany) on the Roche Elecsys 2010 automated immunoassay analyser by the local laboratory. Intra-assay and inter-assay coefficients of variation were <2.0% and <7% for E2, <2.5% and <5% for LH, and <2.5% and <5% for progesterone, respectively.
Retrieved oocytes were classified as metaphase II oocytes, metaphase I oocytes, germinal vesicle stage oocytes or degenerate oocytes for ICSI patients. Embryos were classified as top quality, medium quality and low quality and the cumulative embryo score in a modified version is presented, as described previously (Ludwig et al., 2000).
Outcome measures
Primary endpoint: number of days of gonadotropin treatment. Secondary endpoints: gonadotropin consumption in international units; hormone (E2, LH, progesterone) levels on day 6 and the day of HCG; proportion of metaphase II to total number of retrieved oocytes (in ICSI cases); proportion of two pronuclei oocytes to number of injected or inseminated oocytes (fertilization rate); cumulative modified embryo score; biochemical pregnancy, defined as HCG 10 mIU/ml 14 days after embryo transfer; clinical pregnancy, defined as an ongoing pregnancy at 12 weeks of gestation; implantation rate, defined as number of gestational sacs with fetal heartbeat divided by the number of embryos transferred.
Power analysis
The sample size was based on the primary outcome measure duration of gonadotropin treatment (days). The hypothesis was that the addition of recLH would enhance ovarian response and lead to a shorter time necessary to reach ovulation induction criteria. The study was powered to detect a difference of 1 day between the two treatment modalities. In a previous GnRH-antagonist multiple-dose trial (Ludwig et al., 2002) with a similar ovarian stimulation study protocol and similar patient population, the number of days of gonadotropin stimulation was 12.04 ± 1.7 (mean ± SD) in the GnRH-antagonist multiple-dose protocol study arm with recFSH for ovarian stimulation and a fixed start of the antagonist on day 6. Group sample sizes of 47 and 47 patients who reach HCG administration achieve 81% power to detect a difference of 1 day in the number of gonadotropin treatment days between the null hypothesis that both group means are 12.0 days and the alternative hypothesis that the mean of group 2 is 11.0 days with assumed group SD of 1.7 and 1.7 and with a significance level (alpha) of 0.05 using a two-sided two-sample t-test.
Statistical tests
Chi-square test was used to analyse nominal variables in the form of frequency tables. Normally distributed metric variables (KolmogorovSmirnov test) were tested by a two-sided t-test for independent samples, whereas non-normally distributed metric variables were analysed by MannWhitney U test. Alpha <0.05 was considered significant. Descriptive and efficacy analyses were performed on an intention-to-treat (ITT) basis as appropriate, taking into account all randomized patients. Computations were performed using SPSS statistical software version 12.0.1 for Windows (SPSS lac., 19892002).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Causes of infertility in the patient population are depicted in Table I. Proportions of couples with solely male, solely female or combined infertility were similar between the groups, as was the proportion of women with primary and secondary infertility.
|
|
|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Furthermore, this is indicative that GnRH-antagonist administration leads to suppression in endogenous LH to such an extent that low dose recLH supplementation will significantly affect E2 biosynthesis of the maturing follicles. Although the amount of LH activity necessary for optimal follicular development is unknown, it has been suggested that <1% of follicular LH receptors need to be occupied in order to allow normal steroidogenesis (Chappel and Howles, 1991). However, the concept of normal ovarian steroidogenesis is not well defined and it is unlikely that the lower E2 values in the recFSH only group represent a correlative of impaired follicular development or impaired oocyte developmental competence. Regarding follicular development, we found no shortening of the time until HCG criteria were met under exogenous LH administration. Although SDs were higher than the assumed SDs in the power calculation, it can be assumed that additional low dose LH will not increase the follicular growth rate in this antagonist protocol setting, since no positive trend with the treatment group was observed [12.0 (±2.4) days in the study group vs 11.4 (±2.1) days in the control group]. Similarly, regarding oocyte developmental competence, we found no evidence of a positive effect of higher peak serum E2 values as a consequence of LH supplementation, as evidenced from a similar cumulus oocyte number, similar proportions of mature oocytes, fertilization rates and mean embryo scores in the two groups.
The fact that early follicular phase LH supplementation and relatively higher LH levels measured on stimulation day 6 did not translate into higher E2 levels on stimulation day 6 in the recLH supplemented group, contradicts the observation that serum E2 levels and endogenous LH levels before the onset of antagonist administration are positively correlated (Kolibianakis et al., 2003). This finding might be attributed to the fact that granulosa cells (the main source of E2) will not acquire their own LH receptors before the mid- to late follicular phase under the influence of FSH (Erickson et al., 1979
), whereas endogenous background LH levels in the earlier follicular phase might already be sufficient to occupy most of the constitutionally expressed LH receptors of the theca cells. Concomitantly, multifollicular recruitment appeared slower in the recLH group, resulting in a lower number of follicles in the recLH supplemented group on stimulation day 6, which is likely to have contributed to this observation. The mean number of follicles
13 mm was 1.5 (±2.6) and 2.3 (±3.6) (not significantly different) in the recLH supplemented group and recFSH only group, respectively. This obviously prompted the physicians to increase the gonadotropin dose in the recLH supplemented group more often than in the non-supplemented group. It may therefore be speculated that early follicular phase LH supplementation might alter multifollicular recruitment and support dominant follicle selection, while suppressing smaller follicles. However, two points must be considered: allowing gonadotropin dose adjustments dependent on individual response and according to the physician's discretion unquestionably renders assessment of follicular growth dynamics and ovarian response more difficult, and introduced variability in the present study. Second, hormone values and ovarian sonomorphology in early follicular phase were not assessed in this work; therefore the above stated has to be considered hypothetical. The finding of higher progesterone levels in the recFSH only group on stimulation day 6 is of unclear origin, although it might be attributed to the relatively higher number of growing follicles in the recFSH only group. Values in both groups were in the lower third of the reference range for the follicular phase and the clinical significance of this finding is likely to be low.
It has been suggested previously that there is no need in further exploring LH supplementation in ovarian hyperstimulation (Hull et al., 1994). The introduction of highly purified and recFSH and the widespread use of these compounds combined with GnRH-agonists without additional LH supplementation has not led to a drop in overall success rates, as evidenced from annual statistics from large national programmes, such as the German IVF Registry (Data collection 19992002). However, a number of retrospective studies have indicated that low levels of circulating follicular phase LH in GnRH-agonist long protocol cycles might be associated with impaired E2 synthesis and/or a low oocyte yield, low fertilization rate, low pregnancy rate or high abortion rate, respectively (Fleming et al., 1998
, 2000
; Westergaard et al., 2000
; Esposito et al., 2001
; Humaidan et al., 2002
). In contrast with these findings, two recent studies (Balasch et al., 2001
; Penarrubia et al., 2003
) failed to find a lower LH threshold value in agonist-suppressed patients discriminative between conception and non-conception cycles, or early pregnancy loss and ongoing pregnancy cycles. Furthermore, no association between LH levels and parameters of ovarian response was found. Part of this disagreement between studies has been attributed to different relative potencies of different agonists used. Two interventional trials (Humaidan et al., 2004
; Marrs et al., 2004
) failed to find a significant positive impact as regards treatment outcome of recLH supplemented ovarian stimulation in long agonist protocols. An interventional trial in the antagonist protocol in which increasing the dose of HMG (containing both FSH and LH) at the time of GnRH antagonist introduction was tested, also failed to demonstrate a positive effect of this measure regarding number of oocytes retrieved, embryos obtained, implantation rate or clinical pregnancy rate (Aboulghar et al., 2004
).
Notably, little is known about the association of follicular phase LH levels with ovarian response and treatment outcome in antagonist cycles. With no intrinsic activity at the GnRH-receptor, GnRH-antagonists lead to a direct and sudden suppression of endogenous gonadotropins within 46 h of administration. In contrast with the long agonist protocol this suppression does not occur before stimulation day 6 in the fixed protocol and is undulatory rather than steady, with a nadir every 46 h after antagonist administration. This, together with the pulsatile fashion in which LH is secreted, makes correlation of a single LH measurement with treatment outcome difficult. In the present trial, endogenous LH levels on day of HCG in the non-LH supplemented group were on average above suggested LH threshold values (0.7 IU/l, Fleming et al., 2000; <0.5 IU/l, Westergaard et al., 2000
). However, this finding most likely reflects the study protocol rather than the true maximum nadir values for LH after antagonist administration, because LH assessment was pursued
24 h after the last antagonist administration. Recently, it has been reported that profound suppression of LH on day 8 of stimulation in antagonist treated cycles is associated with a significantly higher chance of achieving an ongoing pregnancy (Kolibianakis et al., 2004
). Similarily, another recent study reported a higher number of oocytes retrieved, embryos obtained, and embryos cryopreserved in patients with LH <0.5 mIU/ml on the day of HCG in antagonist cycles (Merviel et al., 2004
).
Another recent publication suggested LH supplementation as an alternative way of indirectly evaluating the consequences of LH depletion caused by the antagonist (Cedrin-Durnerin et al., 2004). In short, findings by this group were for the most part concordant with our results, although a flexible single dose cetrorelix protocol was used and LH was only administered concomitantly with the antagonist.
Because there is no reliable or cost-effective way to detect which women might possibly need additional LH administration, it was suggested to systematically add LH to ovarian stimulation protocols (Levy et al., 2000). In line with this proposal and also addressing the recent withdrawal of urinary derived HMG preparations from some markets, the present trial aimed at evaluating the effect of a combination preparation of recombinant gonadotropins on ovarian response and IVF treatment outcome. The trial did not aim to specifically assess the effect of LH supplementation concomitantly to LH suppression by a GnRH-antagonist, because the study design (LH administration in early and late follicular phase, respectively) inevitably would imply multiplicity of hypothesis testing within the trial. In conclusion, our findings do not support systematic exogenous low dose LH throughout the follicular phase in GnRH-antagonist multiple-dose cycles in an unselected patient population.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Albano C, Smitz J, Camus M, Riethmuller-Winzen H, Van Steirteghem A and Devroey P (1997) Comparison of different doses of gonadotropin-releasing hormone antagonist Cetrorelix during controlled ovarian hyperstimulation. Fertil Steril 67, 917922.[CrossRef][ISI][Medline]
Al-Inany H and Aboulghar M (2003) Gonadotrophin-releasing hormone antagonists for assisted conception (Cochrane Review). In The Cochrane Library, Issue 1. Update Software, Oxford.
Balasch J, Vidal E, Penarrubia J, Casamitjana R, Carmona F, Creus M, Fabregues F and Vanrell JA (2001) Suppression of LH during ovarian stimulation: analysing threshold values and effects on ovarian response and the outcome of assisted reproduction in down-regulated women stimulated with recombinant FSH. Hum Reprod 16, 16361643.
Balasch J, Penarrubia J, Fabregues F, Vidal E, Casamitjana R, Manau D, Carmona F, Creus M and Vanrell JA (2003) Ovarian responses to recombinant FSH or HMG in normogonadotrophic women following pituitary desensitization by a depot GnRH agonist for assisted reproduction. Reprod Biomed Online 7, 3542.[Medline]
Cedrin-Durnerin I, Grange-Dujardin D, Laffy A et al. (2004) Recombinant human LH supplementation during GnRH antagonist administration in IVF/ICSI cycles: a prospective randomized study. Hum Reprod 19, 19791984.
Chappel SC and Howles C (1991) Reevaluation of the roles of luteinizing hormone and follicle-stimulating hormone in the ovulatory process. Hum Reprod 6, 12061212.[Abstract]
Coppola F, Poti ER, Barusi L, Ferrari B, Salvarani MC and Vadora E (2003) Profound luteinizing hormone suppression induces a deleterious follicular environment during assisted reproduction technology. Fertil Steril 79, 459460.[CrossRef][ISI][Medline]
Erickson GF, Wang C and Hsueh AJ (1979) FSH induction of functional LH receptors in granulosa cells cultured in a chemically defined medium. Nature 279, 336338.[ISI][Medline]
Esposito MA, Barnhart KT, Coutifaris C and Patrizio P (2001) Role of periovulatory luteinizing hormone concentrations during assisted reproductive technology cycles stimulated exclusively with recombinant follicle-stimulating hormone. Fertil Steril 75, 519524.[CrossRef][ISI][Medline]
European Recombinant Human, L.H. Study Group (1998) Recombinant human luteinizing hormone (LH) to support recombinant human follicle-stimulating hormone (FSH)-induced follicular development in LH- and FSH-deficient anovulatory women: a dose-finding study. J Clin Endocrinol Metab 83, 15071514.
Filicori M, Cognigni GE, Taraborrelli S, Spettoli D, Ciampaglia W, de Fatis CT and Pocognoli P (1998) Luteinizing hormone activity supplementation enhances follicle-stimulating hormone efficacy and improves ovulation induction outcome. J Clin Endocrinol Metab 84, 26592663.[CrossRef][ISI]
Filicori M, Cognigni GE, Pocognoli P, Tabarelli C, Ferlini F, Perri T and Parmegiani L (2003) Comparison of controlled ovarian stimulation with human menopausal gonadotropin or recombinant follicle-stimulating hormone. Fertil Steril 80, 390397.[CrossRef][ISI][Medline]
Finas D, Felberbaum R, Roiha K, Wiese J, Schultze-Mosgau A, Griesinger G, Kuepker W and Diedrich K (2003) LH-pulsatility profiles within 24 h after first administration of the GnRH-antagonist Cetrorelix (0.25 mg) on day 6 of stimulation. In Abstracts of the 19th Annual Meeting of the ESHRE, Madrid, Spain, p. xviii 172.
Fleming R, Lloyd F, Herbert M, Fenwick, J Griffiths, T and Murdoch A (1998) Effects of profound suppression of luteinizing hormone during ovarian stimulation on follicular activity, oocyte and embryo function in cycles stimulated with purified follicle stimulating hormone. Hum Reprod 137, 17881792.[CrossRef]
Fleming R, Rehka P, Deshpande N, Jamieson ME, Yates RW and Lyall H (2000) Suppression of LH during ovarian stimulation: effects differ in cycles stimulated with purified urinary FSH and recombinant FSH. Hum Reprod 15, 14401445.
Ganirelix dose-finding study group (1998) A double-blind, randomized, dose-finding study to assess the efficacy of the gonadotrophin-releasing hormone antagonist ganirelix (Org 37462) to prevent premature luteinizing hormone surges in women undergoing ovarian stimulation with recombinant follicle stimulating hormone (Puregon). Hum Reprod 13, 30233031.[Abstract]
Hull MG, Armatage RJ and McDermott A (1994) Use of follicle-stimulating hormone alone (urofollitropin) to stimulate the ovaries for assisted conception after pituitary desensitization. Fertil Steril 62, 9971003.[ISI][Medline]
Humaidan P, Bungum L, Bungum M and Andersen CY (2002) Ovarian response and pregnancy outcome related to mid-follicular LH levels in women undergoing assisted reproduction with GnRH agonist down-regulation and recombinant FSH stimulation. Hum Reprod 17, 20162021.
Humaidan P, Bungum M, Bungum L and Yding Andersen C (2004) Effects of recombinant LH supplementation in women undergoing assisted reproduction with GnRH agonist down-regulation and stimulation with recombinant FSH: an opening study. Reprod Biomed Online 8, 635643.[ISI][Medline]
Kolibianakis EM, Albano C, Camus M, Tournaye H, Van Steirteghem AC and Devroey P (2003) Relationship between LH and oestradiol in IVF cycles before GnRH antagonist initiation. Reprod Biomed Online 7, 190193.[Medline]
Kolibianakis EM, Zikopoulos K, Schiettecatte J, Smitz J, Tournaye H, Camus M, Van Steirteghem AC and Devroey P (2004) Profound LH suppression after GnRH antagonist administration is associated with a significantly higher ongoing pregnancy rate in IVF. Hum Reprod 19 [Epub ahead of print].
Levy DP, Navarro JM, Schattman GL, Davis OK and Rosenwaks Z (2000) The role of LH in ovarian stimulation: exogenous LH: let's design the future. Hum Reprod 15, 22582265.
Loumaye E, Engrand P, Howles CM and O'Dea L (1997) Assessment of the role of serum luteinizing hormone and estradiol response to follicle-stimulating hormone on in vitro fertilization treatment outcome. Fertil Steril 67, 889899.[CrossRef][ISI][Medline]
Ludwig M, Schoepper B, Katalinic A, Sturm R, Al-Hasani S and Diedrich K (2000) Experience with the elective transfer of two embryos under the conditions of the German Embryo Protection Law: results of a retrospective data analysis of 2573 transfer cycles. Hum Reprod 15, 319324.
Ludwig M, Katalinic A, Banz C, Schroder AK, Loning M, Weiss JM and Diedrich K (2002) Tailoring the GnRH antagonist cetrorelix acetate to individual patients' needs in ovarian stimulation for IVF: results of a prospective, randomized study. Hum Reprod 17, 28422845.
Marrs R, Meldrum D, Muasher S, Schoolcraft W, Werlin L and Kelly E (2004) Randomized trial to compare the effect of recombinant human FSH (follitropin alfa) with or without recombinant human LH in women undergoing assisted reproduction treatment. Reprod Biomed Online 8, 175182.[ISI][Medline]
Merviel P, Antoine JM, Mathieu E, Millot F, Mandelbaum J and Uzan S (2004) Luteinizing hormone concentrations after gonadotropin-releasing hormone antagonist administration do not influence pregnancy rates in in vitro fertilization-embryo transfer. Fertil Steril 82, 119125.[CrossRef][ISI][Medline]
Penarrubia J, Fabregues F, Creus M, Manau D, Casamitjana R, Guimera M, Carmona F, Vanrell JA and Balasch J (2003) LH serum levels during ovarian stimulation as predictors of ovarian response and assisted reproduction outcome in down-regulated women stimulated with recombinant FSH. Hum Reprod 18, 26892697.
Westergaard LG, Laursen SB and Andersen CY (2000) Increased risk of early pregnancy loss by profound suppression of luteinizing hormone during ovarian stimulation in normogonadotrophic women undergoing assisted reproduction. Hum Reprod 15, 10031008.
Submitted on July 16, 2004; resubmitted on November 23, 2004; accepted on December 14, 2004.