INSERM UMR-S 943, Paris, France.
Received 8 December, 2009
Accepted 9 December, 2009
Correspondence to Philippe Flandre, INSERM UMR-S 943, 56 boulevard, Vincent Auriol, BP 335, 75625 Paris, France. E-mail: email@example.com
Gallant and Moore  recently reported changes of the estimated glomerular filtration rate (eGFR) in 201 and 231 patients initiating a tenofovir (TDF)-containing antiretroviral regimen or any alternative nucleoside analogues nucleotide reverse transcriptase inhibitors (NRTIs), respectively. I read this article with great interest but it is difficult to have a clear opinion on this topic based on the two successive studies from the Johns Hopkins HIV database.
As mentioned in their recent report, they demonstrated in a previous study that the use of TDF was associated with a greater decline in renal function compared with the use of other NRTIs. This study was based on 344 and 314 patients starting therapy with TDF or other NRTIs as part of their regimen . The creatinine clearance was calculated using the Cockcroft–Gault equation. Patients were included between 1 January 2001 and 31 December 2003 and the main result indicated an absolute change in creatinine clearance of −13.3 and −7.5 ml/min (P = 0.005) for patients in the TDF and NRTIs groups, respectively. In their more recent report, eGFR was computed using the Modification of Diet in Renal Disease (MDRD) equation and there was no significant difference in eGFR change between the TDF and NRTIs groups at month 6 (P = 0.26), month 12 (P = 0.76) and month 24 (P = 0.76).
It is difficult to have a full picture of the data used in their more recent study. Patients were recruited after 1 January 2002 but we do not know the date of the last inclusion. Patients initiating an antiretroviral regimen between the 1 January 2002 and 31 December 2003 are then eligible for the two studies. In their Table 1, we do not know whether means or medians were used for baseline eGFR, CD4 cell count and HIV RNA. The ranges for these values are not provided . We do not know the number of measures of creatinine recorded by patient. This is very important because in observational studies this may introduce a major bias if the number and the timing of these measures are not well balanced between groups. Many studies used a confirmed eGFR of 60 ml/min/1.73 m2 or less as endpoint [3–6]. The use of events defined as a 25% and 50% decline in eGFR from the baseline level make the comparison with the other studies difficult. We do not know how many patients reached a 25% and 50% decline in eGFR. The use of a simple Cox regression model is questionable because general therapeutic settings, including the use of different protease inhibitors, have changed from 2002 to 2008. It is well known that a Cox model with delayed entry (the date to first creatinine measurement) would be more appropriate. A linear mixed-effects model should be used to investigate the association between treatment received and decreases in eGFR rather than the highly correlated tests provided in their article (figures 2 and 3) . Their results indicate that patients taking TDF and a protease inhibitor boosted by ritonavir (PI/r) had a greater median decline in eGFR than those taking TDF and a non-NRTI at M6 but we do not know which protease inhibitors were used .
It is known that the Cockcroft–Gault formula generally gives higher eGFR values in comparison with the MDRD formula . It is reasonable, however, to postulate that the different findings between the two studies are not explained by the use of a different formula. In fact, in a subsequent analysis of their first study, median changes in eGFR, computed according to the MDRD formula, in the TDF arm and in the NRTI arm at month 6 were −14 and −2 ml/min/1.73 m2, respectively (P = 0.01); at month 12, they were −18 and −10 ml/min/1.73 m2, respectively (P = 0.005); at month 18, they were −19 and −11 ml/min/1.73 m2, respectively (P = 0.005) . Thus, I understood that, at month 6, the median change in eGFR in patients receiving other NRTIs shifts from −2 ml/min/1.73 m2 in the 314 patients included in the first study to −12 ml/min/1.73 m2 in the 231 patients enrolled in the second study. It will be most interesting to understand why these two studies involving patients from both the same area and database provided such different results and conclusions.
1. Gallant JE, Moore RD. Renal function with use of a tenofovir-containing initial antiretroviral regimen. AIDS 2009; 23:1971–1975.
2. Gallant JE, Parish MA, Keruly JC, Moore RD. Changes in renal function associated with tenofovir disoproxil fumarate treatment, compared with nucleoside reverse-transcriptase inhibitor treatment. Clin Infect Dis 2005; 40:1194–1198.
3. Mocroft A, Kirk O, Gatell J, Reiss P, Gargalianos P, Zilmer K, et al
. Chronic renal failure among HIV-1: infected patients. AIDS 2007; 21:1119–1127.
4. Kalayjian RC, Franceschini N, Gupta SK, Szczech LA, Mupere E, Bosch RJ, et al
. Suppression of HIV-1 replication by antiretroviral therapy improves renal function in persons with low CD4 cell counts and chronic kidney disease. AIDS 2008; 22:481–487.
5. Roe J, Campbell LJ, Ibrahim F, Hendry BM, Post FA. HIV care and the incidence of acute renal failure. Clin Infect Dis 2008; 47:242–249.
6. Sorli ML, Guelar A, Montero M, Gonzalez A, Rodriguez E, Knobel H. Chronic kidney disease prevalence and risk factors among human immunodeficiency virus-infected patients. J Acquir Immune Defic Syndr 2008; 48:493–494.
7. Gallant JE, Parish MA, Keruly JC, Moore RD. Tenofovir and changes in renal function: reply to Gupta [letter]. Clin Infect Dis 2005; 41:571.