|Nefrologia 2011;31(5):573-578 | Doi. 10.3265/Nefrologia.pre2011.Jul.10893|
|Diagnostic efficiency of a hyperbolic model in predicting digoxin concentrations based on glomerular filtration rates|
|Eficiencia diagnóstica de la predicción de concentraciones de digoxina en función de la tasa de filtración glomerular mediante un modelo hiperbólico|
|Enviado a Revisar: 17 Mar. 2011 | Aceptado el: 25 Jul. 2011 | En Publicación: 20 Sep. 2011|
|Jaime Gonzalez-López1, José Carlos Tutor2|
|1Servicio de Farmacia. Hospital Clínico Universitario, Instituto de Investigación Sanitaria (IDIS). Santiago de Compostela, A Coruña (Spain)|
2Unidad de Monitorización de Fármacos, Laboratorio Central. Hospital Clínico Universitario, Instituto de Investigación Sanitaria (IDIS). Santiago de Compostela, A Coruña (Spain)
|Correspondencia para José Carlos Tutor, Unidad de Monitorización de Fármacos, Laboratorio Central, Hospital Clínico Universitario, Instituto de Investigación Sanitaria (IDIS), 17706, Santiago de Compostela, A Coruña, Spain|
|Figure 1 - Correlation between glomerular filtration rate and obtained digoxin concentration (A), digoxin clearance (B) and the ratio (digoxin clearance) / (glomerular filtration rate).|
|Figure 2 - Correlation and regression between predicted Css [OM] and predicted Css [MM] (A), predicted Css [OM] and obtained Css (B), and predicted Css [MM] and obtained Css (C).|
|Figure 3 - Correlation of the ratio (predicted Css [OM])/(predicted Css[MM]) with GFR|
Digoxin has been used for over two centuries, and concern over its toxicity can be traced back to its introduction into therapy. However, considering its beneficial effect in reducing mortality and hospitalisation of patients with progressive heart failure, its cost-effectiveness and easy availability worldwide, digoxin should not be considered a drug of the past but rather a drug of the present and even one of the future.1
Digoxin is usually administered orally, reaching maximum serum concentrations two to three hours after administration. However, trough concentration (immediately before the next dose) is used for therapeutic monitoring, or at least once the distribution phase is complete.2 The determination of serum digoxin concentration is essential for safe and proper use of the drug, which has a narrow therapeutic index, and this is the primary cause of morbidity and mortality associated with its use. It is currently recommended that doses be set to reach serum concentrations between 0.5ng/ml and 1.1ng/ml.1,3-5 A significant fraction of the absorbed dose of digoxin is eliminated by the kidneys, with its systemic clearance being an important determinant of the maintenance dose.1,6 Renal function impairment therefore frequently causes the onset of toxic effects.1
Glomerular filtration rate (GFR) is widely accepted as an appropriate measure of renal function, and adjusting the dosage of digoxin based on GFR is widely recommended in the literature.6-11 However, recent studies suggest that inappropriate dosing of digoxin is common in patients with various degrees of renal failure.7,12 Comparative studies have been performed using conventional pharmacokinetic procedures for calculating clearance (CL) and the prediction of digoxin concentrations, estimating GFR based on cystatin C13 and creatinine using Cockcroft-Gault and MDRD (modification of diet in renal disease) formulas.13,14 This article, which employs data from 400 adult patients, indicates the Diagnostic efficiencyof a hyperbolic model for calculating digoxin concentration based on GFR.15
PATIENTS AND METHOD
We studied a group of 400 patients of both sexes (158 males and 242 females) with heart failure, with a mean age (± standard deviation of the mean [SDM]) of 78.6±0.64 years (range, 24-97 years), who were treated as outpatients in our hospital’s emergency, cardiology and internal medicine departments. All patients received digoxin orally in the form of tablets, with a dosage of 0.125mg to 0.25mg every 24 to 48 hours, which was not changed for at least 20 days before taking blood samples. The samples were taken after the distribution phase was complete, 24 to 48 hours after the last doses, which meant that digoxin levels corresponded to the steady-state minimum concentration (Css). The study was performed according to the good practice standards for human research of the Health Department of the Government of Galicia, and was approved by the Clinical Research Ethics Committee of the University Hospital Complex of Santiago de Compostela.
Serum digoxin concentration was determined by fluorescence polarisation immunoassay, using reagents from Abbott Laboratories (Abbott Park, IL, USA). Determination of serum creatinine was performed using an Advia 2400 Chemistry System (Siemens Healthcare Diagnostics Inc., Newark, NJ, USA). Cystatin C was determined by PENIA (particle enhanced nephelometric immunoassay) in a BN ProSpec nephelometer (Siemens Healthcare Diagnostics, Inc.). GFR was calculated from the serum concentrations of creatinine and cystatin C using the Cockcroft-Gault16 and de Hoeck et al17 formulas, respectively. The apparent clearance of digoxin was calculated using a conventional pharmacokinetic procedure18: CL=(dose/dosage interval)/Css, with the results expressed in ml/min. The prediction of digoxin concentration was made according to the Konishi et al hyperbolic model15: Css/D=1/(2.22CGGFR+25.7), where Css= level of digoxin (ng/ml), D= dose (µg/day) and CGGFR=GFR calculated using the Cockcroft-Gault formula (ml/min).
Statistical analysis was performed using the StatGraphics Plus (v5.0) program, using the Shapiro-Wilks test for data distribution assessment. We used the Pearson’s correlation coefficient for Gaussian distributions, and the Spearman correlation for all other distributions. Regression analysis was performed using the non-parametric Passing-Bablock method. The hyperbolic fit of the relationship between the ratio Css/dose and CGGFR was performed using the Microcalc Origin® (v8.0) program.
In a group of 300 study patients (120 males and 180 females) with a mean age of 79.2±0.5 years (range, 24-97 years), there were significant correlations between digoxin Css, digoxin CL, and the ratio (digoxin CL)/(GFR), with GFR as indicated in Figure 1. According to earlier studies,9,11,12 digoxin Css was inversely correlated to GFR (Figure 1A). However, although digoxin CL is significantly dependent on GFR (Figure 1B), the modest coefficient of determination (r2=0.257) between both variables indicated that only 26% of the interindividual variability of digoxin CL may be explained by changes in GFR.19 This determination coefficient was not significantly improved by a dichotomy of the data by gender, the normalisation of digoxin CL by kg of weight (ml/min/kg) or the use of MDRD equations with four or six variables for calculating GFR. Values of GFR under 70ml/min would produce an exponential increase in the relative proportion of non-renal CL, compared to the total digoxin CL (Figure 1C).
Serum concentrations of cystatin C were measured in 60 of these patients. The correlations obtained between digoxin CL and GFR calculated from creatinine (r=0.403; P<.005) and cystatin C (r=0.500; P<.005) suggest a modest improvement of the coefficient of determination between digoxin CL and GFR calculated by cystatin C.
Just as Konishi et al15 developed their predictive model on a group of Japanese people, we calculated the constants for the hyperbolic equation for our group of 300 patients with an eye towards optimisation, obtaining the following expression: Css/D=1/(2.42±0.18CGGFR+29.44±4.06). Figure 2 shows the results for the group of 100 remaining patients (38 males and 62 females) with a mean age of 77.2±1.0 years (range, 24 to 92 years) for the relationship between digoxin concentrations obtained experimentally and the predicted values using the original Konishi et al model15 and the model modified for our population group of 300 patients. Digoxin concentrations predicted by the original model were on average 10% higher than those provided by the model modified for our population group (Figure 2A). The proportional difference between the concentrations predicted by each model (expressed as a ratio) showed a highly significant negative correlation with GFR (Figure 3).
We obtained modest correlation coefficients and wide dispersion between obtained digoxin concentrations and those predicted by the original (Figure 2B) and modified (Figure 2C) estimation models. The obtained concentrations and those predicted by both models showed 65% concordance in their classification as subtherapeutic, therapeutic and supratherapeutic.
Currently, the US Food and Drug Administration, the National Kidney Foundation22 and various authors20-24 recommend the use of the Cockcroft-Gault formula for calculating GFR in pharmacokinetic studies and for dosage adjustments.
Consequently, in this study, GFR calculation was performed using the Cockcroft-Gault formula,16 which is also the method used by Konishi et al15 in the development of their predictive model.
Recent studies have demonstrated that patients with varying degrees of renal failure often receive inappropriate doses of digoxin.7,12 The GFR calculation would provide better dosage adjustment for these cases.6-11 As with earlier studies,9,11,12 our patients had an inverse correlation between digoxin concentration and GFR (Figure 1A). There was a 54% prevalence of cases with improper dosages of digoxin (of which 94% had GFR<60ml/min) and with serum concentrations higher than 1.1ng/ml, which is considered the safety limit in chronic treatment.1,3-5 For this reason, it should be noted that elderly patients with renal dysfunction are particularly susceptible to high concentrations of digoxin, with increased susceptibility to cardiac toxicity, which consequently increases mortality risk.25
However, the results shown in Figure 1B indicate the poor coefficient of determination (r2=0.257) between digoxin CL and GFR, which suggests that differences in this variable would only explain 26% of the interindividual variability of digoxin CL. Although the serum concentration of cystatin C has been reported as having certain advantages over that of creatinine in calculating GFR for predicting digoxin levels,10,13 the use of cystatin C in calculating GFR did not improve in a practical way the coefficient of determination between this variable and digoxin CL. Moreover, the results shown in Figure 1C show that, for a GFR<60-70ml/min, the interindividual variability of the ratio (digoxin CL)/GFR increased significantly, with a significant increase in the relative proportion of non-renal CL compared to the total CL of the drug. These results would indicate a priori that patients with GFR<60-70ml/min have clearly limiting factors for correctly predicting digoxin concentrations (or doses) based on GFR.
Digoxin concentrations predicted by the original Konishi et al model15 were, on average, 10% higher than those predicted by the model modified for our population group (Figure 2A). However, none of the cases reached a difference of 15%, which is the accepted limit of deviation according to the standards of validation of methods for the determination of drugs and their metabolites in biological media.26 Consideration should be given to the possibility of greater bioavailability of the tablets administered to the Japanese patients, as well as our different analytical methodology in determining digoxin. However, since the proportional difference between concentrations predicted by both models increases as GRF decreases (Figure 3), as does the relative proportion of non-renal CL compared to total digoxin CL (Figure 1C), the highest concentrations predicted by the original model may be due to a lower non-renal digoxin CL in the Japanese population.
The Diagnostic efficiencyof a laboratory test is the percentage of total results that are true (excluding false positives and false negatives), and it is accepted as a general rule that a test is not clinically valid if its Diagnostic efficiencyis below 80%.27 The agreement in the classification as subtherapeutic, therapeutic and supratherapeutic between the obtained and predicted digoxin concentrations was only 65%, which indicates that both the original predictive model and the model modified for our population have an unacceptable diagnostic accuracy.
In conclusion, although the importance of adjusting dosage based on GFR in order to avoid digoxin intoxication has been noted,7-9 this variable would only explain 25% of the interindividual variability of digoxin CL. Moreover, in cases with GFR<60-70ml/min, the non-renal CL significance increases very variably for digoxin elimination, as it depends on several factors28-30 that are difficult to account for in the predictive models. These facts lead to the conclusion that the Diagnostic efficiencyof predictive models for digoxin levels (or dosage) based on GFR are clinically unacceptable. According to Schentang et al,31 it seems more practical to assume that digoxin bioavailability and CL (if there are no significant changes in GFR or concomitant medication with possible pharmacokinetic interaction) remain stable for the same patient in the medium term, and in a steady-state there is a linear relationship between dosage and Css. We could therefore, based on the Css obtained experimentally for a dose D that must be adjusted, make a prediction of the Css* for a new dose D* using the expression: Css*/D*=Css/D.
1. Pervaiz MH, Dickinson MG, Yamani M. Is digoxin a drug of the past ?. Clev Clin J Med 2006;73:821-33.
2. Bernard DW, Bowman RL, Grimm FA, Wolf BA, Simson MB, Shaw LM. Nighttime dosing assures postdistribution sampling for therapeutic drug monitoring of digoxin. Clin Chem 1996;42:45-9.[Pubmed]
3. Rathore SS, Curtis JP, Wang Y, Bistow MR, Krumholz HM. Association of mortality risk with high serum digoxin concentration and outcomes in patients with heart failure. JAMA 2003;289:871-8.[Pubmed]
4. Morris SA, Hatcher HF, Reddy DK. Digoxin therapy for heart failure: an update. Am Fam Physician 2006;74:613-8.[Pubmed]
5. Ahmed A, Pitt B, Rahimtoola SH, Waagstein F, White M, Love TE, et al. Effects of digoxin at low serum concentrations on mortality and hospitalization in heart failure: a propensity matched study of the DIG trial. Int J Cardiol 2008;123:138-46.[Pubmed]
6. Mangoni AA, Jackson SHD. Age-related changes in pharmacokinetics and pharmacodynamics: basic principles and practical applications. Br J Clin Pharmacol 2003;57:6-14.[Pubmed]
7. Markota NP, Markota I, Tomic M, Zeleneka A. Inappropriate drug dosage adjustments in patients with renal impairment. J Nephrol 2009;22: 97-501.[Pubmed]
8. Bauman JL, DiDomenico RJ, Viana M, Fitch M. A method of determining the dose of digoxin for heart failure in the modern era. Arch Intern Med 2006;166: 2539-45.[Pubmed]
9. Cepeda Piorno J, Pobes Martínez de Salinas A, González García ME, Fernández Rodríguez E. Utilidad de la ecuación MDRD para detectar insuficiencia renal oculta y disminuir el riesgo de sobredosificación digitálica. Nefrologia 2009;29:150-5.[Pubmed]
10. Nakamura T, Ioroi T, Sakaeda T, Horinouchi M, Hayashi N, Saito K, et al. Serum cystatin C levels to predict serum concentration of digoxina in Japanese patients. Int J Med 2006;3:92-6.
11. Hallberg P, Melhus H, Hansson LO, Larsson A. Cystatin C vs creatinine as markers of renal function in patients on digoxin treatment. Upsala J Med Sci 2004;109:247-54.
12. Pita-Fernández S, Lombardía-Cortiña M, Orozco-Veltran D. Clinical manifestations of elderly patients with digitalis intoxication in the emergency department. Arch Gerontol Geriatr 2011 (in press).
13. García A, Hermida J, Tutor JC. Estimation of the glomerular filtration rate from serum creatinine and cystatin C with regard to therapeutic digoxin monitoring. J Clin Pharmacol 2007;47:1450-56.[Pubmed]
14. Vázquez-Hernández M, Bouzas L, Tutor JC. Glomerular filtration rate estimation using the Cockcroft-Gault and Modification of Diet in Renal Disease formulas for digoxin dose adjustment in patients with hearth failure. Upsala J Med Sci 2009;114:154-9.
15. Konishi H, Shimizu S, Chiba M, Minouchi T, Koida M, Yamaji A. Predictive performance of serum digoxin concentration in patients with congestive heart failure by a hyperbolic model based on creatinine clearance. J Clin Pharm Ther 2002;27:257-65.[Pubmed]
16. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-4.[Pubmed]
17. Hoeck FJ, Kemperman FAW, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transpl 2003;18:2024-31.
18. Winter ME. Basic Clinical Pharmacokinetics, 4th edition. Philadelphia: Lippincot Williams & Wilkins; 2004.
19. Mehvar R. Estimation of pharmacokinetic parameters based on the patient-adjusted population data. Am J Pharm Educ 2006;70:article 96.
20. Spruill WJ, Wade WE, Cobb HH. Comparison of estimated glomerular filtration rate with estimated creatinine clearance in the dosing of drugs requiring adjustments in elderly patients with declining renal function. Am J Geriatr Pharmacother 2008;6:153-60.[Pubmed]
21. Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney function: measured and estimated glomerular filtration rate. N Engl J Med 2006; 354: 2473-83.[Pubmed]
22. Roblin I, De Sobarnitsky S, Basselin C, Vial F, Bard E, Dufrene, et al. Estimated glomerular filtration rate fro drug dose adjustment: Cockcroft-Gault or abbreviated MDRD equation? Clin Biochem 2009;42:111-3.[Pubmed]
23. Wargo KA, Eiland EH, Hamm W, English TM, Philippe HM. Comparison of modification of diet in renal disease and Cockcroft-Gault equations for antimicrobial dosage adjustments. Ann Pharmacother 2006;40:1248-53.[Pubmed]
24. Golik MV, Lawrence KR. Comparison of dosing recommendations for antimicrobial drugs based on two methods for assessing kidney function: Cockcroft-Gault and modification of diet in renal disease. Pharmacotherapy 2008;42:1758-65.
25. Vaz Pérez A, Otawa K, Zimmermann AV, Stockburger M, Müller-Werdan K, Werdan K, et al. The impact of impaired renal function on mortality in patients with acutely decompensated chronic heart failure. Eur J Heart Fail 2010;12:122-8.[Pubmed]
26. Shah V, Midha KK, Findlay JWA, Hill HM, Hulse JD, McGilveray IJ, et al. Bioanalytical method validation. A revisit with a decade of progress. Pharm Res 2000;17:1551-7.[Pubmed]
27. Gornall AG. Basic concepts in laboratory investigation. En: Gornal AG, ed. Applied Biochemistry of Clinical Disorders, 2nd edition. Philadelphia: JB Lippincott Company; 1986:3-13.
28. Nakamura T, Kakumoto M, Yamashita K, Takara K, Tanigawara Y, Sakaeda T, et al. Factors influencing the prediction of steady state concentrations of digoxin. Biol Pharm Bull 2001;24:403-8.[Pubmed]
29. Tsujimoto M, Kinoshita Y, Hirata S, Otagiri M, Othani H, et al. Effects of uremic serum and uremic toxins on hepatic uptake of digoxin. Ther Drug Monit 2008;30:576-82.[Pubmed]
30. Tsujimoto M, Dan Y, Hirata S, Ohtani H, Sawada Y. Influence of SLCO1B3 gene polymorphism on the pharmacokinetics of digoxin in terminal renal failure. Drug Metab Pharmacokinet 2008;23:406-11.[Pubmed]
31. Schentag JJ, Bang AJ, Kozinski-Tober JL. Digoxin. En: Burton ME, Shaw LM, Schentag JJ, Evans WE, eds. Applied Pharmacokinetics and Pharmacodynamics, 4th edition. Philadelphia: Lippincott Williams & Wilkins; 2006:410-39.