| |
A placebo-controlled, dose-ranging study of a growth hormone releasing factor in HIV-infected patients with abdominal fat accumulation
|
| |
| |
AIDS: Volume 19(12) 12 August 2005 p 1279-1287
Falutz, Juliana; Allas, Sorayae; Kotler, Donaldc; Thompson, Melanied; Koutkia,
Polyxenij; Albu, Jeanineb; Trottier, Benoith; Routy, Jean-Pierref; Cote,
Pierreg; Abribat, Thierrye; Grinspoon, Stevenj
From the aMontreal General Hospital Immuno-Deficiency Treatment Centre, McGill University Health Center, Montreal, Quebec, Canada
bDivision of Endocrinology
cDivision of Gastroenterology, St Luke's Roosevelt Hospital Center and Columbia University College of Physicians and Surgeons, New York, New York
dARCA (AIDS Research Consortium of Atlanta), Atlanta, Georgia, USA
eTheratechnologies, St. Laurent
fDivision of Hematology and Immunodeficiency Service, Royal Victoria Hospital, Montreal
gClinique medicale du Quartier Latin, Montreal
hClinique Medicale l'Actuel, Montreal, Quebec, Canada
jMass General Hospital Program in Nutritional Metabolism and Harvard Medical
School, LON207, Mass General Hospital and Harvard Medical School, Boston,
Massachusetts, USA.
Abstract
Objective: To investigate the effects of TH9507, a novel growth hormone releasing factor, on abdominal fat accumulation, metabolic and safety parameters in HIV-infected patients with central fat accumulation.
Design and methods: Randomized, double-blind, placebo-controlled trial enrolling 61 HIV-infected patients with increased waist circumference and waist-to-hip ratio. Participants were randomized to placebo or 1 or 2 mg TH9507 subcutaneously, once daily for 12 weeks. The primary outcome was change in abdominal fat, assessed by dual energy X-ray absorptiometry and cross-sectional computerized tomography scan. Secondary endpoints included change in insulin-like growth factor-I (IGF-I), metabolic, quality of life, and safety parameters.
Results:
TH9507 resulted in dose-related physiological increases in IGF-I (P < 0.01 for 1 mg (+48%) and 2 mg (+65%) versus placebo).
Trunk fat decreased in the 2 mg group versus placebo (0.8, -4.6 and -9.2%; placebo, 1 and 2 mg, respectively, P = 0.014 for 2 mg versus placebo), without significant change in limb fat.
Visceral fat (VAT) decreased most in the 2 mg group (-5.4, -3.6 and -15.7%; placebo, 1 and 2 mg, respectively) but this change was not significant versus placebo.
Subcutaneous fat (SAT) was preserved and did not change between or within groups. Lean body mass and the ratio of VAT to SAT improved significantly in both treatment groups versus placebo.
Triglyceride and the cholesterol to high-density lipoprotein ratio decreased significantly in the 2 mg group versus placebo. Treatment was generally well tolerated without changes in glucose.
Conclusions: TH9507 reduced truncal fat, improved the lipid profile and did not increase glucose levels in HIV-infected patients with central fat accumulation. TH9507 may be a beneficial treatment strategy in this population, but longer-term studies with more patients are needed to determine effects on VAT, treatment durability, and safety.
Introduction
HIV-infected patients treated with highly active antiretroviral therapy (HAART) commonly experience changes in fat distribution that include increased visceral and central fat accumulation [1], as well as loss of subcutaneous fat in association with insulin resistance and dyslipidemia [2,3]. Recent data suggest potentially increased cardiovascular disease and myocardial infarction rates in patients treated with prolonged antiretroviral therapy [4]. The mechanisms of fat redistribution in this population are not known and changes in visceral and subcutaneous fat may be due to independent processes [5]. Nonetheless, relative or absolute central adiposity may confer excess cardiovascular risk. In non-HIV-infected patients [6] and among HIV-infected patients with changes in fat distribution [7], increased waist-to-hip ratio (WHR) and central fat accumulation is related to increased metabolic risk indices. Recent studies also suggest that growth hormone levels are reduced in HIV-infected patients, and correlate inversely with excess visceral fat accumulation [8,9]. Studies using higher dose, pharmacologic growth hormone (GH) administration have resulted in reduced visceral adiposity in this population, but are associated with increased insulin resistance and side effects [10-13]. In contrast, strategies using GH secretagogues, may result in physiological increases in GH secretion and reduced truncal fat with fewer side effects [14]. In this study, we assessed the effect of TH9507, a novel growth hormone releasing factor (GRF) analog, over 12 weeks in HIV-infected men and women with evidence of fat redistribution and increased truncal adiposity.
Results
Participant characteristics
Between May 2002 and November 2003, 88 patients were screened. Sixty-one subjects were randomized, 21 to placebo, 19 to TH9507 1 mg and 21 to TH9507 2 mg. Five subjects discontinued in the placebo group, two in the 1 mg group and six in the 2 mg group, for a 79% completion rate.
At baseline no significant differences were seen between the groups in demographic characteristics (Table 1), including use of antiretroviral therapy. The percentage of patients with diabetes (2 h-glucose ≥ 11.1 mmol/l) and impaired glucose tolerance (7.8 mmol/l ≦ 2 h-glucose < 11.1 mmol/l) was not different between the groups. Among the entire study group, 20% of subjects demonstrated impaired glucose tolerance (IGT) and 5% demonstrated diabetes mellitus at baseline. Use of lipid-lowering medications did not differ among the groups.
Body composition
Body composition variables were not different between the groups at baseline. Trunk fat decreased in the 2 mg group versus placebo (0.8, -4.6 and -9.2%; for placebo, 1 and 2 mg groups, respectively, P = 0.014 for 2 mg group versus placebo) (Fig. 1, Table 2), whereas extremity fat did not change in comparison to placebo. Visceral fat (VAT) decreased most in the 2 mg group [-5.4, -3.6 and -15.7%; for placebo, 1 and 2 mg, respectively, (P = 0.03 for change within group) but this change was not significant versus placebo]. Subcutaneous fat (SAT) did not change significantly between the groups and the ratio of VAT: SAT decreased more in the treatment groups versus placebo [0.01 (0.10), -0.23 (0.47), -0.14 (0.18); for placebo, 1 and 2 mg groups, respectively] (Table 2) (P = 0.008 for 2 mg versus placebo and P = 0.043 for 1 mg versus placebo]. The ratio of trunk to leg fat did not decrease compared to placebo, but decreased within the treatment groups (Table 2).
Lean body mass increased in both treatment groups versus placebo [-0.5 (1.6) kg, 0.7 (2.0) kg, and 1.7 (2.3) kg, mean (SD) for placebo, 1 and 2 mg, respectively, P = 0.002 for change in 2 mg group versus placebo, P = 0.047 for the change in 1 mg group versus placebo, Table 2]. Total fat decreased in the 2 mg group compared to placebo [0.3 (1.7) kg, -0.4 (1.8) kg and -1.4 (2.0) kg; for placebo, 1 and 2 mg, respectively, P = 0.013 for 2 mg group versus placebo]. Weight did not change significantly between the groups.
Biochemical indices
Biochemical indices did not differ between the groups at baseline except for the ratio of total cholesterol: HDL, which was higher in the 1 mg group (Table 2). IGF-I increased significantly with the 1 and 2 mg dose versus placebo [21.8 (33.2) ng/ml, 79.0 (68.4) ng/ml, 102.5 (79.3) ng/ml; last observation values for placebo, 1 and 2 mg, respectively, P < 0.01 for each active group versus placebo]. Similar changes in IGF-I were seen using 12-week data.
Using last observation values, triglyceride levels decreased in the 2 mg group compared with placebo [-0.2 (1.3) mmol/l, -0.9 (4.2) mmol/l, -0.9 (1.2) mmol/l; for placebo, 1 and 2 mg, respectively, P = 0.013 for 2 mg versus placebo] and the ratio of cholesterol: HDL improved in both treatment groups versus placebo [0.3 (1.1), -0.3 (0.7), -0.3 (0.6); for placebo, 1 and 2 mg, respectively, P = 0.013 for 2 mg group versus placebo and P = 0.051 for 1 mg group versus placebo]. HDL increased within the 2 mg group, but did not change significantly compared to placebo in either treatment group. Changes in triglycerides, the ratio of total: HDL cholesterol, and HDL were not significantly different between the groups using 12-week data.
Changes in fasting and 120 min glucose were not significant between or within the treatment groups. Among subjects with normal glucose tolerance at baseline (n = 45), three subjects in the 2 mg group versus none in the 1 mg group versus three in the placebo group developed IGT and no patient developed DM. Among patients with IGT at baseline (n = 12), no subject developed DM. Changes in fasting insulin and HOMA-R were not different between the groups but fasting insulin and HOMA-R increased from baseline within the 2 mg group. HbA1C decreased within the placebo group [-0.3(0.5)% P = 0.03)] and increased within the 1 mg group [0.2 (0.4)%, P = 0.02 for within group change and P = 0.002 versus placebo]. No change was seen within the 2 mg group [-0.0 (0.5)%, P = 0.60] or in comparison to placebo.
Bone markers
Osteocalcin increased significantly within the 2 mg group, whereas no changes with NTX were seen (Table 2).
Quality of life
Baseline quality of life indices were not different among the groups except for social well-being (Table 3). Significant differences over time between groups were not seen for the six subscales of the general portion of the quality of life questionnaire [16]. Changes in reported abdominal bloating were significant between groups and bloating decreased significantly within the 2 mg group. Although not significant between groups, abdominal pain decreased significantly within the 1 mg group, and enlarged abdominal girth decreased significantly within the 2 mg group (Table 3).
Immunologic parameters
CD4 cell counts [28(104), 24(147) and -40(77): for placebo, 1 and 2 mg, respectively] did not change significantly between (P = 0.126) or among groups. Viral load remained undetectable in the majority of patients and did not change significantly within treatment groups.
Adverse events and discontinuation
Discontinuation rates were not different between the groups (24, 11 and 29%; for placebo, 1 and 2 mg, respectively). One subject in the placebo group (arthritis), none in the 1 mg group and three in the 2 mg group (rash, arthralgia, paresthesia) experienced adverse events leading to treatment discontinuation. There were 18 severe adverse events that occurred in two patients (9%) in the placebo, 5 (26%) in the 1 mg group and three (14%) in the 2 mg group. Musculoskeletal adverse events such as pain and arthralgias were noted in 24, 26 and 29% of subjects in the placebo, 1 and 2 mg groups, respectively (Table 4). Carpal tunnel symptoms were not noted in any patient. Edema and/or peripheral swelling were noted in one patient in the 2 mg group only. Headache was noted in 14, 21 and 29% in the placebo, 1 and 2 mg groups, respectively. Paresthesias were noted in 0 (0%), 1 (5%) and 3 (14%) of subjects in the placebo, 1 and 2 mg groups, respectively. Blood pressure and heart rate did not change between or within the groups. One patient in the placebo group compared to three in the 2 mg group withdrew from the study related to adverse events.
Safety laboratory values, including hemoglobin, white blood cell, liver function tests, and creatinine did not differ between the groups (data not shown). Anti-TH9507 antibodies were not detected after 12 weeks in any patient.
Discussion
HIV-infected patients on chronic antiretroviral therapy often demonstrate changes in fat distribution, including increased abdominal adiposity in association with insulin resistance and dyslipidemia [1,3], which may increase risk for CVD in this population [4,22,23]. Recent studies suggest that GH secretion is decreased in these patients in association with increased visceral adiposity [8,9]. Strategies using high dose, pharmacologic GH have been shown to decrease visceral fat but aggravate hyperglycemia in this population [11,12]. We hypothesized that a strategy to increase GH within the physiologic range would improve fat distribution and decrease truncal and visceral fat with fewer adverse effects. We therefore assessed the effect of TH9507, a growth hormone releasing factor (GRF) analog on abdominal fat accumulation and relevant metabolic indices in patients with HIV lipodystrophy. GH stimulating analogs increase GH pulsatility but may lead to less toxicity if feedback inhibition of IGF-I on the pituitary is intact. The GRF analog used in this study is not commercially available but was assessed under an Food and Drug Administration investigational new drug (FDA IND).
TH9507, a GRF analog, resulted in significant but physiologic increases in IGF-I, an integrated measure of GH secretion. Trunk fat demonstrated a graded dose-response effect, with a significant reduction in response to 2 mg. Visceral fat decreased most in the 2 mg group, but this change was not significant compared to placebo. In contrast, the ratio of VAT: SAT improved significantly compared to placebo. Larger studies are necessary to determine the statistical and clinical significance of reduced VAT in response to GRF, but the percent decrease in VAT in this study was similar in magnitude, 15% over 3 months, to that shown by GH in the recent STARS study of pharmacologic GH [10]. Furthermore, the lack of a significant effect on extremity and subcutaneous fat may be an advantage of GRF compared to other strategies using pharmacologic GH, in which both visceral and subcutaneous fat decrease [10]. Preservation of extremity and subcutaneous fat is of critical importance to patients with significant lipoatrophy.
The results using a GRF analog in this study are comparable with those seen in response to GHRH 1-29 in a recently published study in men with HIV lipodystrophy, in which truncal fat, but not extremity or subcutaneous fat decreased in response to physiologic increases in GH [14]. The current study extends the findings of Koutkia et al. in a larger group of patients, including men and women, using graded doses of a novel 1-44 amino acid GRF analog that is dosed once rather than twice a day and distinct from the GHRH compound used by Koutkia et al. Although the lower of the two doses increased IGF-I significantly, this dose did not result in a significant decrease in trunk fat, suggesting that there may be a threshold increase in IGF-I necessary to reduce truncal fat.
The 2 mg GRF dose significantly improved triglyceride levels and the cholesterol: HDL ratio. This is a significant advantage of a GRF analog, not seen with other treatment strategies for HIV lipodystrophy [24,25]. Similar beneficial effects on triglyceride were seen with lower, alternating day, but not higher doses of GH in a study reported by Kotler et al. In contrast, higher doses of GH were shown to reduce total and LDL cholesterol in patients with HIV lipodystrophy [10]. Growth hormone has been shown to decrease cholesterol and triglyceride levels in GH-deficient patients and among otherwise healthy men chosen for abdominal obesity [26,27]. Taken together, our data suggest that treatment with TH9507 resulted in an improved lipid profile in dyslipidemic, abdominally obese patients with HIV lipodystrophy.
An important issue regarding the use of GH or related strategies in HIV lipodystrophy is glucose control. Patients with HIV lipodystrophy are often insulin resistant, and a significant percentage, more than one-third, may have impaired glucose tolerance [3]. In this study, even with the higher dose of 2 mg, there were no significant differences in fasting glucose or 2-h glucose in response to a standard glucose tolerance test. Furthermore, there was no significant increase in the number of patients who went from normal glucose tolerance to impaired glucose tolerance. Of note, and in contrast to recently published studies with GH [10], patients with IGT were permitted to enter this study (e.g. 25% of subjects had IGT or diabetes at baseline), and among this subgroup, there was not a significant shift toward the development of diabetes mellitus. Further studies of longer duration will be necessary to determine the effects of TH9507 on glucose metabolism and insulin resistance. The lack of any significant increase in glucose or glucose tolerance, even among patients with baseline IGT, suggests that physiologic increases in GH resulting from this strategy are not likely to aggravate glucose homeostasis in a clinically relevant way, but changes in insulin will need to be assessed in larger studies before an effect on insulin resistance can be ruled out.
TH9507 was also associated with other benefits in this study. Osteocalcin, a marker of bone formation increased within the 2 mg group, whereas NTX, a marker of bone resorption did not, suggesting a net positive effect on bone turnover. Reduced bone density has been described among patients with HIV disease in inverse association with visceral and truncal adiposity [28,29]. Relative reductions in GH secretion may contribute to reduced bone density in some patients with lipodystrophy [30], and a positive effect on bone formation is an additional benefit of TH9507.
Quality of life assessment did not demonstrate significant differences between groups in global domains of physical and psychological functioning, but did suggest that patients receiving GRF, particularly at the 2 mg dose, felt improvement in abdominal fullness, bloating and girth, consistent with the decreases in truncal fat. This is an important consideration for HIV-infected patients for whom such complaints may affect quality of life.
Specific symptoms of GH excess, such as carpal tunnel syndrome, edema, swelling, pain and arthralgias were not noted in a larger proportion of the patients receiving active medication, but headaches and to a lesser extent paresthesias were seen more often in the subjects receiving the 2 mg dose, although not to a degree necessitating discontinuation from the study or therapeutic intervention. Blood pressure did not increase. Discontinuation rates from the study were not different between the groups. Our data suggest that TH9507 was generally well tolerated except for a dose-related increase in headaches, but larger studies of longer duration are needed to definitively determine the safety of GRF in the treatment of HIV-infected patients.
Although this study demonstrated significant effects on trunk fat and other metabolic variables, the number of subjects studied was relatively small, the duration of treatment relatively short, and mode of HIV transmission was not determined. Furthermore, the majority of subjects were men, but prior studies with this compound in non HIV-infected patients do not suggest gender-related differences in IGF-I response at the 2 mg dose [31]. The durability of GRF effect after discontinuation has not been determined, and it is possible that maintenance therapy will be required to sustain the effects on central fat and other metabolic indices.
In summary, treatment of HIV-infected patients with lipodystrophy and central obesity with TH9507, a GRF analog, results in physiologic increases in GH secretion. At a 2 mg dose, this strategy results in significant reductions in truncal fat and increases in lean body mass, without reductions in extremity and subcutaneous fat. VAT decreased 15% over 3 months in the 2 mg group, but changes in VAT were not significant compared to placebo. Glucose did not increase, even among patients with impaired glucose tolerance. In addition, triglyceride and cholesterol: HDL ratio improved in response to the 2 mg dose compared to placebo. Further studies are needed to assess the utility of this novel treatment strategy for patients with HIV and central fat accumulation.
Methods
Patients
HIV-infected males age 18-65 years and non-menopausal females age > 18 years with truncal fat accumulation considered to be part of the HIV lipodystrophy syndrome and waist circumference ≥ 95 cm for men and ≥ 94 cm for women, and a WHR ≥ 0.94 for men and ≥ 0.88 for women were enrolled. Inclusion criteria were based on Lemieux et al. demonstrating excess visceral adiposity in association with increased waist girth [15]. We excluded subjects with a body mass index (BMI) ≦ 20 kg/m2, CD4 cell count ≦ 100 X 106 cells/l, viral load ≥ 10 000 copies/ml, history of opportunistic infection or HIV-related disease within 3 months of the study, history of prostrate cancer or prostate-specific antigen (PSA) > 5 ng/ml in male subjects, history of breast cancer or abnormal mammography within 6 months of the study in female subjects, known hypopituitarism or history of pituitary surgery, radiation or significant head trauma, untreated hypothyroidism, prior history of Type I diabetes mellitus, any prior use of GH or GH-related products within 6 months of the study, systemic steroid administration or megestrol acetate within 60 days of the study, fasting glucose > (8.3 mmol/l)150 mg/dl, aspartate aminotransferase (AST) or alanine aminotransferase (ALT) > 3 X normal, creatinine > 1.5 X normal, or hemoglobin < 9 g/dl. Subjects receiving testosterone or estrogen within the prior 6 months or who had participated in another clinical trial with an investigational agent within 30 days were also excluded from the study. Subjects were excluded for ongoing use of major illicit drugs, including cocaine and heroine, and excess alcohol intake. All subjects were required to be on a stable antiretroviral regimen for 8 weeks prior to enrollment. Subjects receiving lipid-lowering medications were required to be on a stable regimen for 3 months prior to enrollment. Subjects were not permitted to begin anti-diabetic medication, use systemic corticosteroids for > 10 days, or begin estrogen or testosterone preparations during the study.
Clinical research protocol
Subjects underwent a screening visit to determine eligibility. All subjects gave written consent to participate in the study, and the study was approved by the Institutional Review Board at each participating site. BMI, anthropometic measurements, thyroid stimulating hormone (TSH) prolactin, viral load, CD4 cell count, PSA, mammography and pregnancy tests were performed as per protocol.
At the baseline visit, anthropometric measurements, dual energy X-ray absorptiometry (DEXA) and cross-sectional computerized tomography (CT) at L4-L5 were performed. Insulin-like growth factor-I (IGF-I), glucose, insulin, lipid profile [cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglyceride, non-HDL and the ratio of cholesterol: HDL], haemoglobin A1C (HbA1C), insulin-like growth factor binding protein -3 (IGFBP-3), and bone markers [serum osteocalcin and n-telopetide (NTX-1)], CD4, viral load, and a pregnancy test were determined in the fasting state. In addition, insulin and glucose response to a 75 g standard glucose challenge were assessed and a quality of life questionnaire was administered [16].
After baseline evaluations were complete, subjects were randomized equally to one of three groups, placebo, 1 mg TH9507 or 2 mg TH9507, using a permuted block algorithm stratified for gender. Treatment assignment was blinded to investigators and patients. Active drug and placebo were provided as lyophilized powder in vials of identical appearance. Each vial was reconstituted with sterile water immediately prior to injection. Patients were instructed to inject each morning at approximately 0900 h.
Subjects returned for a safety visit at week 1 to assess adverse events and compliance and at 2 weeks to assess weight, anthropometric measures, lipid profile, IGF-I, IGFBP-3, glucose, adverse events and compliance. Subjects returned for a visit at week 6 identical to baseline and a visit at week 9 to assess compliance and adverse events. Subjects returned for a last study visit at week 12 - or, whenever possible, at the time of early termination - identical to the baseline visit.
Study drug
TH9507 (Theratechnologies, Inc, St-Laurent, Canada) is a synthetic analog that comprises the 44-amino acid sequence of human growth hormone releasing factor (hGRF) on which a hexenoyl moiety, a C6 side chain has been anchored on Tyr 1 at the n-terminal. The in vitro half-life of TH9507 is 3-8 h compared to 0.56 h for hGRF. Daily 1 and 2 mg doses have been shown to increase IGF-I to the physiological range seen in younger adults [17]. Th9507 has been studied in 53 patients with Type II diabetes mellitus (DM), and was not shown to aggravate overall glycemic control when administered at a daily dose up to 2 mg [18]. Administration of the 1 and 2 mg doses of TH9507 for 3 months in 109 patients with chronic obstructive pulmonary disease resulted in similar overall incidence of adverse events in all groups, including placebo [19].
Biochemical indices
Serum IGF-I and IGFBP-3 was measured using a radio immunoassay kit (Esoterix Inc., Calabasas Hill, California, USA). HDL, LDL, total cholesterol, triglyceride, glucose, HbA1C, PSA and TSH were determined by standard techniques. Serum insulin concentration was measured using the microparticle enzyme immunoassay (MEIA) (Abbot Laboratories, Abbott Park, Illinois, USA). NTX was measured by enzyme-linked immunosorbent assay kit (Osteomark; Ostex International Inc., Seattle, Washington, USA). Osteocalcin was measured using an enzymatic immunoassay kit (Metra osteocalcin; Quidel Corporation, San Diego, California, USA). CD4 and viral load were performed by routine methodology. Homeostatic model assessment (HOMA) for insulin resistance was calculated as follows [Insulin (μU/ml) X glucose (mmol)]/22.5] [20].
Body composition
Whole body and regional DEXA and cross-sectional abdominal CT scans were performed based on previously established protocols and were standardized across sites. Digitized images were sent to a central reading center, St. Lukes Roosevelt Hospital, for analysis by independent experts without knowledge of treatment assignment. Each site was certified in the proper technique. Total body DEXA scans were performed with analysis of lean body mass, total fat mass, and regional fat mass in the trunk and extremities [21].
Statistical methods
Sample size was planned to assess a statistically significant change of 10% or more in visceral fat. The sample size was estimated to be between 17 and 22 patients/group, for alpha values of 0.05 to 0.1 and 80% power. Baseline data were compared between the groups by analysis of variance (ANOVA) for continuous variables and χ2 test for categorical variables. Treatment effect over time was compared between groups using analysis of covariance (ANCOVA), with treatment and baseline terms or the χ2 test. For cholesterol parameters, a term for use of lipid-lowering agents was also included in the ANCOVA analysis. Where appropriate, data were rank transformed prior to analysis. Pairwise P-values comparing change over time between individual treatment groups were adjusted by the Bonferonni method if the overall treatment term was not significant. Changes within each group were determined by t-test. For quality of life indices, overall treatment effect was determined by ANOVA and change within each group determined by t-test. The intention-to-treat (ITT) population was defined as all subjects who received at least one dose of the study treatment. Descriptive statistics and analyses for all efficacy and safety endpoints were performed on the ITT population. End of study, 12-week data were used to calculate change from baseline in body composition endpoints. For biochemical indices with interval data collected at mid-study timepoints, change from baseline was compared using last observation available after baseline. Imputation for missing data and interim analyses were not performed. Results are mean (SD) unless otherwise noted. All statistical tests were performed with a two-sided type I error level of 0.05 using SAS Version 8.0 (SAS Institute, Cary, North Carolina, USA) unless otherwise specified.
|
|
| |
| |
|
|
|
