Differential effects of recombinant human growth hormone on glomerular filtration rate and renal plasma flow in chronic renal failure

In normal subjects recombinant human growth hormone (rhGH) increases glomerular filtration rate (GFR) and effective renal plasma flow (ERPF) through the action of insulin-like growth factor-I (IGF-I). We have measured clearance of inulin and para-aminohippuric acid in 18 children with chronic renal failure (CRF) during their 1st year of rhGH treatment to look at the immediate (first 3 h), short-term (1 week) and long-term (1 year) effects of treatment. On day 1 mean (range) age was 9.1 (4.9–13.9) years, GFR 19 (9–58) and ERPF 77 (34–271) ml/min per 1.73 m². During treatment height velocity increased from 4.5 (1.7–6.5) to 9.5 (4.8–12.7) cm/year (P<0.0001). Two children required dialysis after 0.75 years and 1 child was electively transplanted after 0.5 years. There were no other serious adverse events. GFR and ERPF were unchanged in the 3 h following rhGH. GFR remained constant on day 8, 22 (6–56) and after 1 year, 20 (9–59) ml/min per 1.73 m². ERPF increased to 96 (33–276) ml/min per 1.73 m² on day 8P=0.005), and remained elevated, but not significantly so, at 99 (24–428) ml/min per 1.73 m² at 1 year. Fasting IGF-I increased from 147 (46–315) ng/ml to 291 (61–673) by day 8P<0.003), and to 341 (101–786) ng/ml at 1 year. There was no correlation between the change in IGF-I and renal function. Blood pressure, albumin excretion and dietary protein intake were unchanged by treatment. The significance of increased ERPF after 1 week of rhGH in CRF is unclear, but long-term follow-up of renal function is indicated.     

Reversible diabetes mellitus during growth hormone therapy in chronic renal failure

A 12-year-old girl with short stature due to idiopathic Fanconi syndrome and chronic renal failure was treated with recombinant human growth hormone (rhGH). There was no family history of diabetes mellitus and the glucose tolerance before treatment was normal. Intravenous glucose tolerance tests were performed before, during and after treatment. Two months after starting rhGH the early phase of insulin secretion (1-+3-min values) was diminished, and the patient developed manifest diabetes mellitus with hyperglycemia and an elevated hemoglobin A_1c. Following discontinuation of rhGH, glucose tolerance slowly returned to normal. Received May 29, 1997; received in revised form November 5, 1997; accepted November 12, 1997     

Growth preservation after brief growth hormone therapy in chronic renal insufficiency

A boy of 3 years 8 months with short stature due to chronic renal insufficiency was treated with recombinant human growth hormone (rhGH) for 20 months. Catch-up growth was achieved and the improvement of the height standard deviation score was sustained throughout an additional 4 years of follow-up without further rhGH therapy. Received March 18, 1996; received in revised form and accepted July 31, 1996     

Accelerated growth in short children with chronic renal failure treated with both strict dietary therapy and recombinant growth hormone

In a 12-month study, nine boys, aged 4.8–15.6 years, with bone ages 4.6–13 years, with moderate to severe chronic renal failure and resultant growth failure were treated with daily recombinant human growth hormone (rhGH), in conjunction with a strict low-protein/low-phosphate diet supplemented with keto and amino forms of the essential amino acids, histidine and additional energy. Improved growth had previously been observed with this dietary management over that obtained with conventional treatment for chronic renal failure. Each child had been on this diet for at least 2 years before rhGH was commenced. Mean height velocity increased from 4.6±1.3 to 9.0±1.3 cm/year (P<0.001) in the pre-pubertal group, and in the pubertal group from 5.4±1.4 to 10.4±1.8 cm/year (P<0.01). The mean height velocity standard deviation scores (SDSs) increased from –1.2±0.6 to +2.3±0.9 (P<0.001) in the pre-pubertal group and from –0.4±0.6 to +1.9±1.1 (P<0.01) in the pubertal group. Mean height SDS for chronological age increased from –2.2±0.7 to –1.5±0.5 (P<0.01) in the pre-pubertal group and from –1.9±0.7 to –1.3±0.9 in the pubertal group (P<0.02). There was no significant deterioration in renal function or renal bone disease, and bone age did not advance more than chronological age over the 12-month period.     

Growth hormone treatment started in the first year of life in infants with chronic renal failure

Infants with chronic renal failure (CRF) are at high risk of experiencing severe growth retardation. We report a study of 12 infants with CRF who have been treated with recombinant human growth hormone (rhGH) since the age of 0.5 ± 0.3 years. A control group comprised 15 infants with less severe CRF who were being treated during the same period, but who did not receive rhGH. Despite the infants in the rhGH group had more severe renal failure, they grew at least as well as those in the control group and experienced catch-up growth that started earlier and was more sustained; they also gained more weight. Between the age of 0.5 and 2.5 years, the height standard deviation score (HtSDS) improved from −2.0 ± 1.2 to −0.9 ± 0.9 in the rhGH group (p < 0.005) and from −1.6 ± 1.6 to −1.0 ± 1.9 in the control group (p=non significant, n.s.). The average gain in HtSDS was +1.1 ± 0.8 in the treated group and +0.6 ± 1.4 in the control group (p = n.s.). During the same period, the weight SDS improved from −2.2 ± 0.9 to −0.6 ± 1.2 (p < 0.005) and from −1.9 ± 1.2 to −1.3 ± 1.2 (p=n.s.) in the treatment and control groups, respectively. Nutritional intake was similar in both groups, while parathyroid hormone levels tended to increase, although not significantly, after rhGH treatment (p=n.s.). The results of this pilot study suggest that very early treatment with rhGH in patients with early-onset CRF may improve growth.     

Recombinant human growth hormone in infants and young children with chronic renal insufficiency

Children with chronic renal insufficiency (CRI) secondary to congenital structural abnormalities frequently have significant growth retardation by 2 years of age. In a multicenter placebo-controlled study of the use of recombinant human growth hormone (rhGH), 30 of 125 (24%) participants were<2.5 years of age at enrollment. Since the treatment arms of the study were balanced for age at randomization, data for these patients were examined for efficacy and safety. During the first 2 years of the study, approximately two-thirds of the patients (n=19) received rhGH 0.05 mg/kg per day subcutaneously and one-third (n=11) received placebo injections. At entry into the study, the mean (± SD) calculated creatinine clearance was 29.2±14.3 (range 12.0–63.7) ml/min per 1.73 m² in the rhGH-treated group and 23.3±15.1 (range 8.0–59.4) ml/min per 1.73 m² in the placebo-treated group. The 1st year growth rate was 14.1±2.6 cm/year for the rhGH-treated group and 9.3±1.5 cm/year in the placebo-treated group (P<0.00005). During the 2nd year of the study, the growth rate was 8.6±1.2 cm/year in the rhGH-treated group compared with 6.9±1.0 in the placebo groupP=0.025). The height standard deviation score was +2.0±0.7 for the rhGH-treated group compared with –0.2±1.1 in the placebo-treated group (P<0.00005) during the 2 years of the study. Minor adverse events occurred with similar frequency in both groups. These data suggest that rhGH is efficacious and safe in children with CRI under age 2.5 years. rhGH therapy may correct significant loss of growth at this age when used in conjunction with optimal medical management.     

Factors influencing the response to growth hormone in children with renal disease

The effects of age, height velocity over the preceding year, glomerular filtration rate (GFR) and prednisolone dose on growth response have been assessed by single and multiple linear regression analysis in 23 prepubertal children [age, mean (SD), 8.2 (2.5) years] with chronic renal failure (CRF) and 16 prepubertal children [12.1 (2.3) years] with renal transplants treated for 1 year with recombinant human growth hormone (rhGH), 30 U/m² per week. Height velocity [mean (SD), cm/year increased from 4.7 (1.3) to 9.7 (2.1) (P<0.0001) in the CRF group and 3.1 (1.6) to 7.3 (2.8) (P<0.0001) in the transplant group. In the CRF group, there was a correlation between age and height velocity, both in the pretreatment year (r=–0.755,P<0.0001) and during treatment (r=–0.421,P=0.045). There was no correlation between pretreatment height velocity or GFR and response to rhGH. In the transplanted children height velocity during the treatment year correlated with age (r=–0.647,P=0.007), prednisolone dose (r=–0.689,P=0.003), GFR (r=0.542,P=0.030) and pretreatment height velocity (r=0.655,P=0.006). Multiple regression analysis showed prednisolone dose and age to be the most important predictors of response.     

Growth hormone binding protein and free growth hormone in chronic renal failure

Chronic renal failure in children is associated with growth failure. While the pathogenesis of uremic growth failure is multifactorial, an abnormal growth hormone/insulin-like growth factor (GH-IGF) axis is an important contributory element. Patients with uremia exhibit insensivivity to the action of GH, as exemplified by high plasma GH levels, low IGF-I activity, and poor somatic growth. This insensitivity can be overcome by supraphysiological doses of exogenous GH. Plasma GH binding protein (GHBP, the circulating ectodomain of the GH receptor) levels are decreased in patients with renal failure, as are hepatic GH receptor levels in animal models. Since GHBP levels are thought to reflect GH receptor levels in tissues, it is likely that the uremic GH insensitivity in humans is mediated by a decreased number of GH receptors. Another implication of the low plasma GHBP is a disproportionate elevation of free plasma GH (the biologically active moiety) relative to total GH, lending additional support to the concept of GH insensitivity in uremia. GH kinetics are altered in renal failure because of: (1) inability to excrete GH and (2) changes in the bound fraction of GH in the circulation.     

Changes in body composition of children with chronic renal failure during growth hormone treatment

Growth hormone (GH) has different known metabolic effects, among which are lipolysis and anabolic action. We have studied the changes in body composition of children with chronic renal failure (CRF) after 1 year of daily treatment with GH. Body fat percentage and fat body mass (FBM) were derived from four site skinfold measurements; lean body mass (LBM) from total body potassium (TBK) and mid-arm muscle circumference (MAMC); bone mineral density (BMD) was measured by dual photon absorptiometry. GH treatment had a positive effect on weight, heigt and MAMC, but no effect on LBM (as reflected by TBK), FBM and BMD. Z-scores were derived in order to compare subjects with a normal population. While no significant change in z-score was noticed for weight, height, MAMC, FBM and BMD, TBK decreased during treatment. We conclude that GH therapy does not ultimately increase LBM in CRF patients compared with other GH-treated groups.     

Growth hormone therapy in chronic renal failure induces catch-up of head circumference

Growth of head circumference was studied along with height, weight, and body mass index (BMI) in 21 prepubertal patients with chronic renal failure (CRF) before and during recombinant human growth hormone (rhGH) treatment. CRF was present from birth in 15 patients, in the 6 others it was acquired and existing for at least 1 year. Five patients were on chronic dialysis, and 16 children were on conservative treatment with a median glomerular filtration rate of 17 ml/min per 1.73 m² at the start of rhGH therapy. rhGH was administered for 12 months in all patients, for 18 months in 19, and for 24 months in 12 patients. Mean height standard deviation score (SDS) increased significantly from –2.29 to –1.31 after 1 year and to –1.07 after 2 years. Mean BMI SDS was within the normal range throughout. Mean head circumference SDS improved significantly from –2.04 to –1.45 after 1 year and remained stable thereafter. Changes in head circumference differed between patients under 5 years and those over 5 years. In the former, the increase in head circumference SDS was already significant after 6 months of therapy, in the latter, significance was reached only after 1 year. It can be concluded that rhGH in CRF patients significantly improves head circumference SDS, albeit not to the same extent as height SDS. Received: 21 August 2000 / Revised: 21 February 2001 / Accepted: 26 February 2001     

Recombinant human growth hormone treatment of children with chronic renal failure: long-term (1- to 3-year) outcome

Treatment of nine boys, aged 2.8–16.3 years, with growth retardation consequent to chronic renal failure (CRF), with recombinant human growth hormone (rhGH) for 12–36 months demonstrated a significant improvement in growth velocity at each 12-month interval compared with that achieved the year prior to treatment. Despite the acceleration in growth velocity the bone age did not increase more than the increase in chronological age during the period of treatment. The mean calculated creatinine clearance did not decrease significantly during the 36 months of treatment; however, two patients required institution of dialysis at 18 and 30 months following the initiation of rhGH treatment. There was no exacerbation of the glucose intolerance of uremia following rhGH treatment. Currently, six of seven patients who have been treated for more than 24 months have achieved sufficient acceleration of growth velocity to attain a standard deviation score that was more positive than –2.00, and are above the 5th per centile for chronological age on the growth curve. These data indicate that rhGH treatment of growthretarded children with CRF results in accelerated growth velocity during the 2nd and 3rd years of treatment, and demonstrate the potential for such children to achieve normal stature for chronological age despite the continued presence of renal failure.     

Parathyroid hormone and its fragments in children with chronic renal failure

Parathyroid hormone (PTH) immunoradiometric assays (IRMA) exhibit cross-reactivity between 1-84 PTH and long carboxyl-terminal-PTH (C-PTH) molecules. C-PTH antagonizes the biological actions of 1-84 PTH and circulates in excess in chronic renal failure (CRF), partially explaining why supra-physiological PTH levels are recommended to maintain bone turnover. Furthermore, the ratio 1-84 PTH/C-PTH may be related to bone turnover. This study characterizes the 1-84 PTH/C-PTH ratio in children with varying severity of CRF and levels of PTH. Two hundred and forty-one children with CRF, managed with the aim of preventing the development of hyperparathyroidism, had PTH measured by intact IRMA and a new more specific Cyclase-Activating-PTH (CAP) IRMA. C-PTH levels were calculated by subtracting CAP-IRMA from intact IRMA. Fifty-three controls with normal renal function were also recruited. Mean intact IRMA correlated with CAP-IRMA (r=0.98), but was higher (P<0.001). The mean 1-84 PTH/C-PTH ratio was lower than controls in dialysis patients (P=0.022) and those with a glomerular filtration rate <30 ml/min per m² (P=0.033). This ratio was comparable to controls when the PTH level was normal, but was lower with PTH levels outside the normal range (P<0.01). These data suggest that CAP-IRMA gives a more accurate assessment of actual PTH levels than intact IRMA in CRF. Maintenance of normal PTH levels throughout the course of CRF permits the maintenance of a normal 1-84 PTH/C-PTH ratio, the clinical significance of which requires further investigation in children.     

Growth hormone for children with chronic renal failure and on dialysis

We studied all children with CRF who received recombinant human growth hormone (rhGH) for more than a year (mean±SD duration of therapy 3.7±2.5 years) over an 11-year period. There were 32 children. Twenty-one children were conservatively managed, with a mean glomerular filtration rate (GFR) of 24±12 mL min^–1/1.73 m² at the start of rhGH. Their height standard deviation score improved from –2.5±1.4 to –2.1±0.7 at 1 year (P=0.3), –2.0±0.7 at 2 years (P=0.01), and –1.6±0.6 at 3 years (P=0.001). After that there was no improvement. Eleven children were on dialysis, six on haemodialysis (HD) and five on peritoneal (PD). Ht SDS improved from –2.7±0.5 to –2.3±0.5 at 1 year (P=0.02). Thereafter there was no further improvement. RhGH was stopped because of transplantation in 29 patients at a mean±SD age of 12.1±4.0 years. Mean Ht SDS was –1.8±0.8 at transplant and there was no change over the following 5 years. In conclusion, treatment with rhGH resulted in improvement in Ht SDS in conservatively managed CRF for up to 3.0 years and for 1 year in children on dialysis. Discontinuation of rhGH after transplantation resulted in little change in Ht SDS.     

Early age-dependent growth impairment in chronic renal failure

We report early linear growth in 73 children (51 boys, 22 girls) with early onset of chronic renal failure (CRF). The inclusion criteria was onset of CRF before 6 months of age, two or more height measurements during the 1st year of life, follow-up for at least 3 years and continuously impaired renal function with a glomerular filtration rate below 50 ml/min per 1.73 m² at 1 year or later. Only height measurements taken during conservative treatment or dialysis were included. The data were analysed in terms of the infancy-childhood-puberty growth model. There was an age-dependent growth failure in early life leading to an attained height of –3 standard deviation score (SDS) at 3 years of age. Approximately one-third of the reduction in height occurred during fetal life and one-third during the first postnatal months. Between 0.75 and 1.5 years of age height also decreased by 1 SD as a consequence of a delayed onset of the second, the childhood, phase of growth in 36% of the patients and by an offset childhood growth pattern — i.e. a return to the infancy phase pattern after onset of the childhood phase — in 60% of the patients. Growth between 0.25–0.75 and 1.5–5 years of age was generally percentile parallel and thus less likely to be affected in CRF with early disease onset. The glomerular filtration rate was not related to the height gain in early life. We speculate that the growth failure during fetal life and the first postnatal months reflects metabolic and/or nutritional influences and the impaired growth at 0.75–1.5 years of age is related to a partial insensitivity to growth hormone.Study group members: I. Rätsch (Ancona), K. Michelis, T. Kapogiannis (Athens), F. Jung, T. Lennert (Berlin I), S. Gellert (Berlin II), T. Tulassay, P. Sallay (Budapest), T. von Lilien, D. Michalk (Cologne), M.-A. von Wendt-Göknur (Erlangen), K. E. Bonzel (Essen), R. Gusmano, E. Verrina (Genova), G. Offner (Hannover), O. Mehls, A.-M. Wingen, C. Fabian-Bach (Heidelberg, coordinators), A. Appiani, A. Bettinelli (Milan), J. Feber (Prague), G. Rizzoni, S. Picca (Rome), H. J. Stolpe, M. Wigger (Rostock), J. Kist-van Holthe, E. Wolff (Rotterdam, coordinators for the centers Amsterdam, Antwerp, Groningen, Nijmegen, Rotterdam), U. Berg (Stockholm), M. Fischbach (Strasbourg), E. Dobos (Szeged), E. Balzar (Vienna), T. Neuhaus (Zurich).     

Pubertal growth in children with chronic renal failure on conservative treatment

The pubertal growth spurt was followed for at least 3 years in 5 boys and 6 girls with chronic renal failure on conservative treatment. The peak height velocity averaged 8.6 cm/year (range 5.8–10.1 cm/year) in males and 8.2 cm/year (range 6.4–11.5 cm/year) in females. In none was the pubertal growth spurt below the 3rd percentile for chronological age. At the end of the follow-up period, all patients but 2 had stature within the normal limits of parental target. The relative variation of height averaged —0.013 standard deviation scores per year. On the whole, the pubertal growth spurt was normal in subjects with chronic renal failure on conservative treatment.     

Changes in body composition of children with chronic renal failure on growth hormone

Body composition is altered in children with chronic renal failure (CRF) and contributes to the significant growth failure seen in these children. Recombinant human growth hormone (rhGH) has been used in the past several years to improve the somatic growth of children with CRF. To determine if the growth achieved in these children occurs concomitantly with body compositional changes, seven prepubertal (n=6) and pubertal (n=1) children with chronic renal insufficiency (n=4) and end-stage renal disease (n=3) underwent measurements of total body fat (FM), fat free mass (FFM), bone mineral density (BMD), total bone mineral mass (TBBM), total body water (TBW), and total body potassium (TBK) before and 6 months after initiation of subcutaneous recombinant human growth hormone (rhGH) at 0.35 mg/kg per week. The techniques used included dual- energy X-ray absorptiometry (for measurement of FM, BMD, and TBBM), total body potassium counting (for measurement of TBK), and deuterated water for assessment of TBW. Significant increases in both height and weight were seen following 6 months of rhGH therapy. These increases were accompanied by significant re- ductions in FM (4.4±1.4 kg vs. 3.6±1.2 kg, P=0.002) and percentage fat (18.6±3.9% vs. 14.5±3.4%, P=0.04), while FFM (17.9±3.0 kg vs. 20.7±3.6 kg, P=0.04) increased significantly as did TBBM (776±171 g vs. 844±177 g, P=0.001). Increases in TBK, a measure of body cell mass, were also seen. No difference in total BMD was observed. Thus, growth in CRF is occurring with repletion of the FFM and TBBM compartments. Despite these improvements, no change was observed in the body mass index (BMI). Measurement of BMI alone does not define the compartmental catabolic losses in FFM. Received: 20 September 1999 / Revised: 31 January 2000 / Accepted: 8 February 2000     

Progression of chronic renal failure in children with dysplastic kidneys

The aim of this study is to describe progression of chronic renal failure (CRF) in children with renal malformations and to study factors influencing this progression. We reviewed retrospectively 176 children with CRF secondary to renal dysplasia, reflux nephropathy or renal obstruction with at least 5 years of follow-up. Serum creatinine was recorded at least every third month, and an estimated glomerular filtration rate (eGFR) was calculated. Number of febrile urinary tract infections (UTI), blood pressure, albuminuria (UaUc), and number of functioning kidneys was also recorded. We found that the development of renal function could be separated into three time periods: (1) During the first years of life, 82% of the children showed early improvement of their kidney function, which lasted until a median age of 3.2 years (median improvement 6.3 ml/year). (2) From the age of 3.2 years until 11.4 years, 52.5% of the studied children showed a stable kidney function, whereas in 47.5%, kidney function immediately started to deteriorate. (3) Around puberty, 42.9% started deterioration in kidney function, whereas 57.1% even after puberty showed a stable function. Patients with UaUc >200 mg/mmol deteriorated faster (−6.5 ml/min per 1.73 m² per year compared with −1.5 ml/min per 1.73 m² per year) in those with UaUc <50 mg/mmol. Children with more than two febrile UTIs, hypertension or an eGFR at onset of less than 40 ml/min per 1.73 m² deteriorated faster than the others. Most children experienced early improvement of kidney function. The further prognosis, early or late deterioration of kidney function or stable function during the whole follow-up, was related to albuminuria, number of febrile UTIs, eGFR at onset of deterioration, hypertension and puberty.     

Safety and efficacy of erythropoietin in children with chronic renal failure

 A prospective randomized study of the use of recombinant human erythropoietin (rHuEPO) in children with chronic renal disease was conducted to assess dosing requirements and side effects. Forty-four children with chronic renal failure, aged 4 months to 21 years, were studied. Twenty-five patients were pre dialysis, 10 on peritoneal dialysis, and 9 on hemodialysis. Patients received either 150 U/kg per week or 450 U/kg per week divided thrice weekly of rHuEPO for 12 weeks or until target hemoglobin (Hb) was attained. Dose was then adjusted to maintain a normal Hb. Eighty-two percent of patients reached target Hb by 7.9±5.6 weeks (mean±SD); 95% of patients in the high-dose group and 66% in the low-dose group reached target Hb within 12 weeks. The overall median rHuEPO dose at target Hb was 150 U/kg per week. Hemodialysis patients tended to require more rHuEPO to maintain a normal Hb (median 250 U/kg per week). Transfusion requirements and panel-reactive antibody levels decreased during the 12 weeks. Iron deficiency and/or hypertension occurred in 30% of children. In conclusion, rHuEPO at 150 U/kg per week is safe and effective in treating anemia in children with chronic renal disease. Received: 12 March 1998 / Revised: 24 June 1998 / Accepted: 6 July 1998     

Long-term effects of growth hormone treatment on growth and puberty in patients with chronic renal insufficiency

Several prospective trials have shown that recombinant human growth hormone (GH) accelerates growth significantly during the first years of therapy, but the effects of long-term GH therapy with regard to long-term growth response and safety have not yet been established. Forty-five Dutch prepubertal children [28 boys, 17 girls, mean (SD) age 7.8 (3.4) years] with chronic renal insufficiency (CRI) and severe growth retardation started GH therapy between 1988 and 1991 within one of the randomized Dutch trials. Long-term GH therapy, in this study a maximum of 8 years, resulted in a sustained and significant improvement of height standard deviation score (SDS) compared with baseline values (P<0.001). The mean height SDS reached the lower end (-2 SDS) of the normal growth chart after 3 years of GH therapy. During the following years the mean height SDS gradually increased, thereby approaching the mean target height SDS after 6 years of GH therapy. Three factors were significantly associated with the height SDS after 4 years of GH therapy: height SDS at the start (+) of therapy, age at the start of therapy (-), and the duration of dialysis treatment (-). Bone maturation did not accelerate during long-term GH therapy. Children on a conservative regimen at the start of GH therapy had no accelerated deterioration of renal function during 6 years of GH therapy. The average daily GH dose administered over the years had no significant influence on the glomerular filtration rate after 4 years. GH therapy had no adverse effects or significant effect on parathyroid hormone concentration, nor were there any radiological signs of renal osteodystrophy. Puberty started at a median age, within the normal range, of 12.4 years in boys and 12.0 years in girls, respectively. Long-term GH therapy leads to a sustained improvement in height SDS in children with growth retardation secondary to CRI, resulting in a normalization of height in accordance with their target height SDS, without evidence of deleterious effects on renal function or bone maturation. A GH dosage of 4 IU/m² per day appears efficient and safe. Our long-term data show that final height will be within the normal target height range when GH therapy is continued for many years. Received: 25 March 1999 / Revised: 13 January 2000 / Accepted: 20 January 2000