Eur J Clin Pharmacol (2000) 56: 293±297
Ó Springer-Verlag 2000
PHARMACOKINETICS AND DISPOSITION
J. Y. V. Yee á S. B. Duull
The effect of body weight on dalteparin pharmacokinetics A preliminary study
Received: 30 July 1999 / Accepted in revised form: 16 March 2000
Abstract Aim: To investigate whether there were signi®cant dierences in the volume of distribution (V) and clearance (CL) of dalteparin in obese versus normalweight patients, and thereby determine whether dosing of dalteparin should be based on total body weight, lean body weight or an adjusted body weight in obese patients. Methods: Patients (ten obese and ten normal weight) treated with dalteparin were matched for age, gender, lean body weight and creatinine CL. Two steady-state plasma dalteparin concentrations were taken from each patient and assayed in duplicate. The pharmacokinetic values of V and CL were estimated, for each patient, using the Bayesian maximum a posteriori method with the program ABBOTTBASE. Results: The mean V in obese patients was approximately 60% larger than in normal-weight patients, but this was not statistically signi®cant (P 0.11; twotailed). The mean value of V (8.4 l) in the normal-weight patients was similar to that reported in the literature. The mean dierence in values of CL (18% larger in obese patients) was not clinically or statistically signi®cant. A poor correlation was seen between V and lean body weight (r2 0.05). There was a moderate correlation between V and total body weight (r2 0.52) and between V and adjusted body weight (r2 0.55); adjusted body weight [lean body weight + 0.4(total body weight ± lean body weight)]. Total body weight and adjusted body weight provided a better correlation with CL (r2 0.39, 0.32, respectively) than did lean body weight (r2 0.01). J. Y. V. Yee (&) Senior Clinical Pharmacist, Pharmacy Department, Christchurch Hospital, Private Bag 4710, Christchurch, New Zealand e-mail:
[email protected] Tel.: +64-3-3640840; Fax: +64-3-3640838 S. B. Duull Drug Information Pharmacist, Clinical Pharmacology Department, Christchurch Hospital, New Zealand
Conclusion: These results suggest that doses of dalteparin in obese patients should be based on total body weight or an adjusted body weight, but not lean body weight. This study highlights some potential dierences in the pharmacokinetics of dalteparin in individuals who are obese, and further work is necessary to quantify these dierences in more detail. Key words Dalteparin á Low-molecular-weight heparins á Pharmacokinetics
Introduction The term ``low-molecular-weight heparins'' (LMWHs) is applied to all fractions or fragments of heparin that have a mean molecular weight of less than 6 kDa. However, within the group, there are signi®cant dierences in molecular size, methods of preparation, and the pharmacokinetic and pharmacodynamic properties [1, 2]. Dalteparin exhibits a ®rst-order, one-compartment, pharmacokinetic pro®le [2]. A linear relationship was observed between peak plasma LMWH concentration and incidence of bleeding [3, 4, 5, 6, 7, 8], although the rates of bleeding reported in these studies were highly variable. Low peak plasma concentrations between 0.1 IU ml)1 and 0.2 IU ml)1 were associated with a low incidence of haemorrhage [7, 8]. These studies suggest the need to characterise volume of distribution (V) accurately, since dose and V are the determinants of the peak plasma concentration. The apparent V of LMWHs is expected to be close to plasma volume, due to the high molecular weight, low lipid solubility, high plasma protein binding and low tissue binding. Estes [9] showed that unfractionated heparin distributes primarily into plasma volume, which is approximately 2±5% of normal body weight. Similarly, Andrassy and Eschenfelder [10] reported that LMWHs are con®ned to the intravascular space and their V correlates with plasma volume. This and other information to date would argue therefore that obese
294
patients might be appropriately dosed based on lean body weight. Janvier et al. [11] reported that there was no relationship between the peak plasma concentration of enoxaparin, another LMWH, and deep vein thrombosis resolution, as indicated by changes in venograms. Levine et al. [3] reported that trough plasma concentration of LMWHs were linked to clinical eectiveness (assessed by inhibition of thrombosis). This relationship, however, does not appear to have been the subject of further investigations. The importance of the average steady-state concentration or trough concentration, as a predictor of pharmacodynamic response, remains to be established. To date, there are no pharmacokinetic studies of LMWHs in obese patients. If the values of the pharmacokinetic parameters in obese patients are dependent on lean body weight, it is likely that obese patients dosed on total body weight will be overdosed and be at increased risk of bleeding. Our hypothesis is that the V and clearance (CL), and hence the dose of dalteparin, in obese patients should be based on lean body weight. The aim of the study was to investigate whether there were signi®cant dierences in V and CL of dalteparin in obese versus normal-weight patients, and thereby determine whether obese patients should be dosed based on total body weight, lean body weight or an adjusted body weight.
Methods Subjects The study was approved by the Southern Regional Health Authority Ethics Committee and each subject gave written informed consent. Ten obese and ten normal-weight patients of Caucasian descent were selected. The patients were matched for age, gender and lean body weight, and had normal renal function (creatinine CL >4.3 l h)1). Any patient with intrinsic coagulation disorders, (de®ned as abnormal international normalised ratios), recent childbirth or renal disease was excluded from the study. The inclusion criteria for obesity and normal-weight patients were de®ned as: Obese patients. Body mass index ³30 Normal weight patients. Body mass index 20±29.9 where body mass index is expressed in units of kg m)2 (i.e. weight height)2) [12]. Dalteparin administration/dosage Dalteparin (Fragmin, Pharmacia and Upjohn; Kabi-2165, FR-860) was administered via the subcutaneous route in the abdomen of the patient. The doses used were based on current hospital guidelines (deep vein thrombosis 200 IU kg)1 day)1; pulmonary embolism, unstable angina 120 IU kg)1 twice daily). Prescribers varied in their decision to use lean, adjusted, or total body weight to dose obese patients. This decision was not in¯uenced by the investigators. Body weights were de®ned as follows: lean body weight (LBW) (height)150 cm) ´ 0.9 + 45 kg (female) or +50 kg (male) [13]; adjusted body weight (ABW) LBW + CF ´ (TBW±LBW) where CF 0.4 and TBW total body weight. A correction factor (CF) of 0.4 was chosen as an estimate of the
extracellular ¯uid distribution in adipose tissue, as it has been described for aminoglycosides [14]. Investigations Two venous blood samples for measurement of the plasma dalteparin concentration were taken from each patient on the second or a subsequent dose of dalteparin. In most subjects, steady-state concentrations would have been achieved (assuming a half-life of 4±6 h) [15]. The ®rst sample was taken at approximately (but not earlier than) 4 h after the dose was administered, and the second sample just prior to administration of the next dose. The two blood samples were taken at least 4 h apart. The exact times of administration, dosing history and time of sample collection were recorded. Blood was collected into citrate tubes and cooled immediately on ice to minimise release of heparin antagonists from blood cells. Plasma was separated from cells by centrifugation for 20 min at 2000 g at 2 °C as soon as possible after blood collection. Plasma samples were stored at )80 °C, at which temperature dalteparin is stable for at least 6 months [15]. Assay IL test heparin (Xa) (Instrumentation Laboratory, Milan) The IL test heparin (Xa) is an assay which directly measures the concentration of dalteparin using the Automated Coagulation Laboratory (ACL) ± Futura II analyser (Instrumentation Laboratory, Milan), a colorimetric spectrophotometer. Each assay run included two quality control (QC) concentrations (0.4 IU ml)1 and 0.7 IU ml)1). Dalteparin concentrations higher than 1.0 IU ml)1 were diluted 1:2 in buer and the assay repeated. All patient samples were analysed in duplicate. The r2 of the standard curve for validation was 0.988 and, for all subsequent standard curves, the r2 was at least 0.98. The coecient of variation of the assay was chosen to be that of the QC 0.4 IU ml)1 (6.5%). This value was rounded to 7%, arbitrarily, for the pharmacokinetic analysis. Values of assay variability at some concentrations were higher and at others lower than this value. However, since the patient samples were analysed in duplicate, variability from this source will be characterised better, hence the choice of 7% is reasonable. A priori model for dalteparin in ABBOTTBASE A linear one-compartment model with ®rst-order absorption was used to describe the time course of dalteparin following subcutaneous administration. The standardised (per kg) values of V and non-renal CL for the a priori model (Table 1) are based on an adult weight of 60 kg. Other prior information for the parameters are given in Table 1. Creatinine CL was estimated by the CockroftGault equation [16], but using the lower of lean body weight and total body weight [17]. Statistical analyses and interpretation All statistical analyses and linear regression were performed using the program GraphPad Prism v2 (1995). Descriptive statistics The patient's demographics were described using mean and standard deviation (SD), as were the pooled values of the pharmacokinetic parameters. The estimated values of V and CL for each patient were standardised to total body weight, and the mean coecient of variation (%), CV% and 95% con®dence intervals were derived for each group.
295 Table 1 A priori model for dalteparin in ABBOTTBASE. ka absorption rate constant; F bioavailability factor (SC); CLNR non-renal clearance; renal slope renal clearance/ creatinine clearance
Parameter
Mean* value (%)
CV** (%)
Lower limit
Upper limit
References
V (l kg)1) CLNR (l h)1 kg)1) Renal slope ka (h)1) F CV of assay
0.09 0.0023 0.2 0.63 0.87 ±
30 40 40 Fixed Fixed 7
0.03 0.0014 0.01 ± ± ±
0.5 0.143 1.0 ± ± ±
18, 18, 18, 18, 15 ±
19, 23, 24, 25, 26 19, 23, 24, 25, 26 25 23
* Geometric mean ** Assuming a log-normal distribution Comparative statistics The characteristics age, height, total body weight, lean body weight and creatinine CL were compared using a paired t-test. The pharmacokinetic parameters were compared using both parametric (ttest) and non-parametric (Wilcoxon signed rank) tests. If the 95% con®dence interval of the mean dierence in V included zero, then no dierence between the groups could be concluded. If the test did not support a signi®cant dierence between the mean values of CL and V between groups, then a power analysis was undertaken in order to determine the b error. A b error of less than 0.2 is considered suciently low, such that determination of a real dierence between groups, could be shown, should one exist.
The value of V was larger in seven obese patients, and smaller in three obese patients. V was approximately three times bigger in the patient (obese 7) than the matched normal-weight patient. This patient had the largest body mass index in the group. When V was expressed in litres, the ratio of V in the obese group relative to the normal-weight group was 1.63. The mean value for CL was larger in the obese group than in the normal group, but the dierence was less than 20% and not clinically signi®cant (1.3 l h)1 vs 1.1 l h)1).
Regression analyses
Comparative analyses
The pharmacokinetic parameter values of the 20 patients were pooled. A linear regression analysis was undertaken to determine whether there was a signi®cant correlation between: (1) V and lean body weight, total body weight or an adjusted body weight; (2) CL and lean body weight, total body weight or an adjusted body weight.
The dierence in V between the obese and normal group, using a paired t-test, was not statistically significant (P 0.11; two-tailed value). Similarly, the results were not statistically signi®cant using the Wilcoxon signed rank test. The former analysis was calculated to have a b error of 0.64 (power of 0.36), and 58 subjects would have been required to demonstrate a dierence of this size with a 0.05. When CLobese was compared with CLnormal, the difference was small (0.19 l h)1). Using a paired t-test, the results were not statistically signi®cant, (P 0.27; twotailed value). Similarly, the results were not statistically signi®cant using the Wilcoxon signed rank test. This
Results Patient selection The characteristics of the patients are shown in Table 2. Obese patients were approximately 1.5 times heavier than normal-weight patients. Nine patients were being treated for unstable angina, eight for pulmonary embolism and three for deep vein thrombosis. Only one pair of patients were matched for disease states. All patients had normal values of the international normalised ratio.
Table 3 Summary of volume of distribution (V) and clearance (CL) values for obese versus normal-weight patients, derived using the ABBOTTBASE program. CV% coecient of variation (%)
Statistical analyses of pharmacokinetic data Descriptive analyses (Table 3) The mean value of V was approximately 50% larger in the obese group (12.4 l) than in the normal group (8.4 l). Table 2 Analysis of pairing between two study groups. CLCR creatinine clearance; SCr serum creatinine; LBW lean body weight; TBW total body weight
Mean CV% Lower 95% CI Upper 95% CI
V (l)
CL (l h)1)
Obese Normal Ratio
Obese Normal Ratio
12.39 54.13 8.90 15.88
1.30 45.38 0.88 1.72
8.36 51.08 6.14 10.58
1.63 55.06 0.99 2.27
1.11 25.22 0.91 1.32
1.22 53.28 0.76 1.69
Descriptor
Obese mean (SD)
Normal mean (SD)
Mean paired dierence (95% CI)
Age (years) Height (cm) SCr (mmol l)1) CLCR (l h)1) Gender LBW (kg) TBW (kg)
52.0 (13.2) 169.2 (9.4) 0.073 (0.023) 4.79 (0.12) 5 Males, 5 females 64.1 (12.3) 106.4 (22.1)
52.1 (14.0) 170.5 (7.4) 0.081 (0.014) 4.93 (0.19) 5 Males, 5 females 66.1 (8.7) 69.7 (9.3)
)2.3 ()7.7, 3.1) )1.30 ()3.75, 1.15) )0.008 ()0.02, 0.004) )0.14 ()0.47, 0.18) )2.01 ()5.4, 1.4) 36.6 (18.6, 54.7)
296
analysis was calculated to have a b error of 0.87 (power of 13%), and 220 subjects would have been required to demonstrate a dierence of this size with an a 0.05. The power of the study is not of great interest as the mean dierence is too small to be of clinical signi®cance. Regression analysis The pharmacokinetic data of the 20 patients were pooled and regression analyses were undertaken to determine whether the parameters V and CL were signi®cantly correlated with various descriptors of patient size. V was moderately well correlated with total body weight (r2 0.52), and the line of best ®t (with 95% CI) for V versus lean body weight is described by y intercept 2.55 ()14.8, 19.9), x intercept )21.1 and slope 0.1 ()0.14, 0.39), r2 0.05. The line of best ®t (with 95% CI) for V versus total body weight is described by y intercept )4.4 ()11.8, 2.9), x intercept 26.4 and slope 0.17 (0.09, 0.25), r2 0.52. The line of best ®t (with 95% CI) for V versus adjusted body weight is described by y intercept )14.1 ()25.3, 3), x intercept 43.2 and slope 0.33 (0.18, 0.47), r2 0.55. The correlation coecients of CL versus lean body weight, total body weight and adjusted body weights were r2 0.01, r2 0.39 and r2 0.32, respectively.
Discussion The mean value of V, estimated using ABBOTTBASE, for the normal group was 8.4 l, which is similar to the values of 7.3 l and 7.7 l reported by Simonneau et al. [18] and Collignon et al. [19], respectively. This mean value is larger than plasma volume, indicating some extravascular distribution of LMWHs. Lane and Dawes and Pavuk reported that dalteparin has a lower anity for plasma and matrix proteins than does unfractionated heparin, which is reported to have a V of 5 l (showing very limited extravascular distribution) [20, 21]. This observation may explain some of the dierences in pharmacokinetics between unfractionated heparin and LMWHs. The apparently bigger V in obese patients (approximately 50%) appeared to be proportional to the dierence in total body weight between the two groups. Regression analyses indicated a signi®cant correlation between both V and CL, and total body weight, as well as with an adjusted body weight, that is supported by the reduced between-subject variability described as the coecient of variation when V and CL were standardised by total body weight. This suggests that obese patients should be dosed on total body weight or an adjusted body weight, rather than lean body weight. The use of an adjusted body weight, as opposed to total body weight, may be preferable in morbidly obese patients, until the relationship of V to either total body weight or adjusted body weight can be clari®ed in a larger study.
The use of an adjusted body weight, using a correction factor of 0.4, appears to be acceptable for other hydrophilic drugs, such as gentamicin [14], as extracellular ¯uid is expected to expand directly with body weight in morbidly obese patients. Morse and Soeldner [22] reported that the amount of extracellular water in adipose tissue in obese patients was 15% of total body weight, with no signi®cant dierence between obese (122±153% of lean body weight) and non-obese (