Current guidelines for the treatment of HIV infection recommend the use of two nucleos(t)ide analogues (NA) along with a non-nucleoside analogue (NNA) or a protease inhibitor (PI) [1,2]. Tenofovir (TDF) and didanosine (ddI) are among the most frequently prescribed NA because both are administered in a convenient fashion (just one pill each per day), show relatively high genetic barrier for resistance, have a quite acceptable safety profile, and remarkable antiviral potency . However, their co-administration has given rise to concerns given that TDF significantly increases ddI plasma levels and, accordingly, ddI doses are recommended to be reduced from 400 to 250 mg per day in patients weighting more than 60 kg who take both NA concomitantly . At the same time, several reports have highlighted an increased risk of pancreatitis and of hyperglycemia in patients treated with TDF plus ddI [5–7]. More intriguing is the recent evidence of unexpected CD4+ T-cell declines in patients treated with this dual NA combination despite having undetectable viral load [8,9].
In a large multicenter study we assessed the CD4 cell outcome in patients receiving different NA combinations, including TDF plus ddI, ddI alone, TDF alone, and others. Moreover data were analysed according to the third anti-HIV agent administered and the dose of ddI. Finally, the relationship to plasma levels of ddl was further examined. We propose a new mechanism by which TDF plus ddI may cause CD4+ T-cell depletion in HIV-infected individuals despite providing complete virus suppression.
Patients and methods
Two groups of HIV-infected individuals were retrospectively analysed. The first included all patients who initiated a PI-sparing antiretroviral regimen between September 2002 and June 2003 at one HIV referral clinic, and had at least one subsequent visit within the following 12 months, demonstrating complete virus suppression at each visit. Only drug-naive individuals who started antiretroviral treatment or subjects who simplified a prior successful regimen were included in the study. The second group of patients was represented by individuals on successful antiretroviral therapy at five hospitals who had been recruited during the same period in a Spanish trial in which a simplification strategy with ddI, TDF and efavirenz was being evaluated . Lastly, only patients who continued with the same drugs during the whole study period were assessed.
Patients who commenced a PI-based regimen were excluded from the analysis as some reports have claimed that PI may exert an immunological benefit beyond its antiviral effect [11–13]. Moreover, patients on virological failure under another regimen who initiated a new antiretroviral combination as salvage therapy were excluded from this analysis, as other confounding factors may have influenced the likelihood of achieving virological control and CD4 recovery in antiretroviral-experienced patients [14,15].
All individuals who received ddI took 400 mg if their weight was above 60 kg and 300 mg if lower than 60 kg. After the release of the recommendation in early 2003  to reduce the dose of ddI in subjects taking TDF concomitantly, subjects weighting more than 60 kg switched to 250 mg daily, and those weighting less than 60 kg switched to 200 mg daily.
Plasma levels of ddI were measured at week 12 after initiation of therapy and at month 12 if a reduction in the ddI dose was performed. A validated high-performance liquid chromatography method was used . Given that ddI was taken at night-time, approximately 12 h before the blood was drawn next morning, trough levels were calculated following standard recommendations , and expressed as nanograms/ml.
Continuous and categorical data were expressed as median and interquartile ranges (IQR) and percentages, respectively. Univariate analysis was performed to compare baseline characteristics in drug-naive and patients who simplified their treatment, according to the use of ddI and/orTDF. Categorical data were compared with the χ2 test (with Yates and Fisher corrections when appropriate). The Kruskal–Wallis test was used to compare continuous variables. Changes in CD4+ T-cell counts (absolute number and percentage) were assessed for each patient at different time-points with respect to baseline. CD4+ T-cell changes over time in different treatment arms were compared using parametric (Student's t) or non-parametric (Wilcoxon) tests, as required.
In order to assess which variables were independently associated with changes in the CD4+ T-cell counts, a multivariate analysis (multiple linear regression) was performed at different time points (β value: change in the CD4+ T-cell count for each independent variable). Following the results of the first exploratory analyses, the independent parameters which were finally included in the multivariate analysis were: weight, ddI dose, baseline CD4+ T-cell count, time on high ddI doses and lactate levels.
In the subset of patients for whom plasma ddI levels were available, another multivariate analysis was performed, including this variable. All statistical analyses were conducted using the SPSS package (v9.0, Chicago, Illinois, USA) and only P values below 0.05 were considered as significant.
A total of 570 individuals were assessed. They were split according to treatment modality (98 drug-naive versus 472 simplified). The baseline characteristics were well balanced between groups and are summarized in Table 1. Overall, the third agent used together with two NA was a NNA (n = 378 patients) or a NA (n = 192 patients).
At 12 months, the median CD4+ T-cell counts was significantly lower in patients included in the simplification on TDF + ddI when compared to those taking ddI or TDF alone (Fig. 1). Moreover, in drug-naive individuals, those under ddI + TDF experienced a significantly lower gain in the median CD4+ T-cell count when compared with patients in other groups, taking either ddI or TDF (Fig. 1). This occurred in drug-naive and simplified patients irrespective of the third agent included in the triple combination. Interestingly, the greater CD4+ T-cell decrease was seen in the group of patients receiving three NA. Changes in CD4+ percentages followed the same trend (data not shown).
Considering the changes in CD4+ T cells after 12 months of treatment with the different regimens, a significant loss of CD4+ T lymphocytes was seen only in patients receiving combinations including TDF and ddl (Table 2). Moreover, eight subjects showed a decline of CD4+ T cells to less than 200 × 106 cells/l (having had more than 350 × 106 cells/l at baseline) and all had received TDF + ddI combinations. Four were on simplification regimens and four were drug-naive, and as required for entry in this study, all had undetectable viremia on therapy.
In the subset of patients on ddI + TDF, a total of 269 (91.2%) initiated ddI at higher doses than currently recommended (400 mg once daily if > 60 kg and 250 mg once daily if < 60 kg) and 26 subjects (8.8%) commenced with the reduced ddI dose (250 mg once daily if > 60 kg and 200 mg once daily if < 60 kg). Many patients who initiated ddI therapy with the higher dose reduced it during the following 12-month study period, as new recommendations became available. After adjusting by weight, patients who had been exposed to higher doses of ddI showed more pronounced CD4 declines than those treated with lower doses (Fig. 2).
The extent in which CD4+ T-cell declines occurred after 12 months of treatment with the different ddI + TDF regimens, after adjusting for the ddI dose, is recorded in Table 2. Six out of the eight subjects who showed a decline of CD4+ T cells to less than 200 × 106 cells/l (having had more than 350 × 106 cells/l at baseline) had received high ddI doses. On the other hand, patients included in the simplification group taking three NA developed more pronounced CD4 drops, despite all being with the lower ddI dose, although this difference did not reach statistical significance.
When we examined which variables were significantly associated with CD4+ T-cell declines, in the multivariate analyses only low weight (P = 0.03), higher baseline CD4+ T-cell counts (P < 0.001), and time on high ddI doses (P < 0.001) were associated with a significant loss of CD4+ T cells.
Finally, plasma ddI levels at steady state were measured in 57 patients under TDF–ddI combinations. The median estimated plasma trough level was 163 ng/ml (IQR, 48–312). When subjects receiving the high dose (n = 42) were compared with those with the low dose (n = 15), significantly higher median ddI plasma levels were seen in the former versus the latter group: 183 ng/ml (IQR, 73–368) versus 49 ng/ml (IQR, 19–196) (P = 0.036). There was a trend towards an association between plasma ddI levels and the extent of CD4+ T-cell decline: for each increase of 10 ng/ml in ddI plasma levels, a mean decrease of 1.67 CD4+ T cells per microlitre was expected (β, −1.67; 95% confidence interval, −3.41 to +0.06; P = 0.058). Moreover, when ddI plasma levels were included in the multivariate analysis as an additional variable, this trend persisted (P = 0.058), whereas time on a high ddI dose was no longer associated with CD4+ T-cell loss (P = 0.74).
This study demonstrates that HIV-infected individuals receiving ddI + TDF-based combinations show CD4+ T-cell declines (in both absolute number and percentages) despite complete virus suppression. This effect generally occurs after 6 months of therapy and worsens with time. It occurs earlier and is more pronounced when TDF + ddI are taken together with another NA as third agent as well as when using high ddI doses. Accordingly, high ddI plasma levels correlated with loss of CD4+ T cells. Finally, CD4+ T-cell declines were not seen in patients receiving any other antiretroviral regimen, including those in which either TDF or ddI were included.
Given that ddI and TDF are both adenosine analogues, we hypothesize that a synergistic effect of their metabolites might cause an imbalance in the purine pool within CD4+ T lymphocytes. As these cells experience a rapid turnover in HIV infection, any impairment in cell replication might translate into loss of CD4+ T cells by a mechanism which is independent of virus replication. This cytostatic effect of TDF + ddI combinations on CD4+ T lymphocytes essentially resembles the T-cell immunodeficiency seen in the purine nucleoside phosphorylase (PNP) deficiency, a rare autosomal recessive genetic disorder [18–20]. The homozygotes present as severe combined immunodeficiency, with recurrent infections and death in infancy. Failure to clear purine metabolites and accumulation of deoxy-guanosine-triphosphate (dGTP) in these children result in an inhibition of the ribonucleotide reductase enzyme, which in turn, inhibits DNA synthesis and impedes cell division. Although the PNP enzyme is found in most body tissues, it has the highest levels in lymphoid cells, which may explain why CD4+ T lymphocytes are selectively targeted in HIV-infected patients receiving ddI and TDF together. It is our hypothesis that because TDF metabolites inhibit PNP , the use of the alternative purine pathway by ddI metabolites is compromised in HIV-infected individuals on ddI + TDF. There is a shift to production of high levels of dGTP, which results in an inhibition of the ribonucleotide reductase, and consequently DNA synthesis is blocked, causing CD4+ T-cell declines (see Fig. 3).
The fact that CD4+ T-cell declines were generally seen after 6 months on ddI + TDF therapy suggest that some compensatory mechanisms are involved, at least in the short term, which ultimately fail as extended periods of therapy are given.
The greater CD4+ T-cell decline seen in patients who took ddI + TDF with a third NA is a remarkably finding. The fact that other triple NA combinations did not cause CD4+ T-cell drops, including those in which TDF or ddI were provided separately, suggests that administration of additional NAs might further exacerbate the metabolic interaction between ddI and TDF within the cells. As the plasma lactate levels did not correlate with CD4+ T-cell declines and no cases of symptomatic hyperlactatemia were seen in our series, deleterious cell damage following the mechanism of some anti-metabolite compounds used for cancer therapy rather than through mitochondrial toxicity could be the more reasonable explanation for our findings. However, this should be further investigated, since apoptosis of T lymphocytes may ultimately be favoured when mitochondrial DNA synthesis is limited [22,23].
Higher ddI doses were associated with more profound CD4+ T-cell declines in patients undergoing ddI + TDF-based combinations. Even more, time on high ddI doses was independently associated to loss of CD4+ T cells in the multivariate analysis. Interestingly, this association was lost when ddI plasma levels were included in the analysis, suggesting that high ddI concentrations in the blood were the best surrogate marker of toxic exposure to ddI. The relatively small size of the population for which we could obtain plasma ddI levels at the steady state most likely precluded the achievement of a significant correlation between plasma ddI levels and net CD4+ T-cell loss. This observation is somewhat unexpected, given that it is generally believed that plasma levels of NA may not correlate with its effects, including antiviral and side effects, as they need to be phosphorylated within the cells to become active . However, in patients on ddI and TDF, the interaction between these two drugs might hypothetically favour a better correlation between ddI concentrations within and outside cells. Two different mechanisms have been proposed to explain why TDF increases plasma ddI levels. The first postulated that TDF might favour the absorption of ddI at gastrointestinal level  whereas the second suggests that TDF metabolites, throughout inhibition of PNP within the cells, delay ddI catabolism . Hypothetically, the use and the saturation of a single metabolic pathway for ddI might tend to favour a better correlation between drug metabolite levels within and outside the cells.
Our findings are relevant for selection of antiretroviral combinations and discourage the use of TDF and ddI in combination. The recognition of CD4+ T-cell declines in patients taking these medications together adds to the other recently discussed concerns about other side effects, including a higher risk of pancreatitis [5,6], hyperglycemia , and lactic acidosis , as well as of higher risk of virological failure with selection of the K65R mutation [27–29]. Therefore, when possible, the combination of TDF and ddI should be avoided.
This work was supported in part by grants from Fundación IES (Investigación y Educación en SIDA) and Red de Investigación en SIDA (RIS). We would like to thank Marta Boffito for helpful comments.
1. Yeni P, Hammer S, Hirsch M, Saag M, Schechter M, Carpenter C, et al
. Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel. JAMA 2004; 292:251–265.
2. Panel on Clinical Practices for Treatment of HIV Infection convened by the US Department of Health and Human Services. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents
, accessed November 2nd, 2004).
3. Jimenez-Nacher I, de Mendoza C, Garcia-Diaz B, Soriano V. Different impact of efficacy, toxicity and convenience on the prescription of antiretroviral drugs
. Seventh Congress on Drug Therapy in HIV Infection
. Glasgow; November 2004 [abstract P66].
4. Kearney B, Isaacson E, Sayre J. Didanosine and tenofovir DF drug-drug interaction: assessment of didanosine dose reduction
. In: Program and Abstracts of 10th Conference on Retroviruses and Opportunistic Infections
., Boston, MA February 2003 [abstract 533].
5. Martinez E, Milinkovic A, de Lazzari E, Ravasi G, Blanco JL, Larrousse M, et al
. Pancreatic toxic effects associated with co-administration of didanosine and tenofovir in HIV-infected adults. Lancet 2004; 364:8–10.
6. Kirian M, Higginson R, Fulco P. Acute onset of pancreatitis with concomitant use of tenofovir and didanosine. Ann Pharmacother 2004; 38:1660–1663.
7. García-Benayas T, Barrios A, Sánchez-Conde M, Rendón A, R-Nóvoa S, Sánchez-Somolinos M, et al
. Higher risk of hyperglycaemia in patients under didanosine and tenofovir-containing regimens
. In: Program and Abstracts of the 12th Conference on Retroviruses and Opportunistic Infections
, Boston, MA. February 2005 [abstract 197] (in press).
8. Negredo E, Molto J, Burger D, Viciana P, Ribera E, Paredes R, et al
. Unexpected CD4 cell count decline in patients receiving didanosine and tenofovir based regimens despite undetectable viral load. AIDS 2004; 18:459–463.
9. Negredo E, Bonjoch A, Paredes R, Clotet B. Concurrent administration of tenofovir and didanosine compromises immunologic recovery in treatment-experienced patients. Results from the TORO studies
. Seventh International Congress on Drug Therapy in HIV Infection
, Glasgow; November 2004 [abstract PL3].
10. Barrios A, Negredo E, Vilaró, Domingo P, Estrada V, Labarga P, et al
. Safety and efficacy of a qd regimen based on TDF, ddI and EFV
. In: Program and Abstracts of 11th Conference on Retroviruses and Opportunistic Infections
, San Francisco, CA, February 2004 [abstract 566].
11. Barreiro P, Soriano V, Casas E, Gonzalez-Lahoz J. Different degree of immune recovery using antiretroviral regimens with protease inhibitors or non-nucleosides. AIDS 2002; 16:245–249.
12. Barreiro P, González-Lahoz J, Soriano V. CD4+ T-cell gain with non-nucleoside or protease inhibitors: convenience may not always be the most convenient. J Acquir Immune Defic Syndr 2004; 36:758–759.
13. Manfredi R, Calza L, Chiodo F. First-line efavirenz versus lopinavir/ritonavir-based HAART for naive patients. AIDS 2004; 18:2331–2333.
14. Ciancio B, Trotta M, Lorenzini P, et al
. The effect of number of mutations and of drug-class sparing on virological response to salvage genotype-guided antiretroviral therapy. Antiviral Ther 2003; 8:611–616.
15. Mocroft A, Phillips A, Friis-Moller N, Colebunders R, Johnson A, Hirschel B, et al
. Response to antiretroviral therapy among patients exposed to three classes of antiretrovirals: results from the EuroSIDA study. Antiviral Ther 2002; 7:21–30.
16. Knupp C, Stancato F, Papp E, Barbhaiya R. Quantification of didanosine in human plasma and urine by high performance liquid chromatography. J Chromat Biomed Applic 1990; 533:282–290.
17. Kappelhoff B, Crommentuyn K, de Maat M, Mulder J, Huitema A, Beijnen J. Practical guidelines to interpret plasma concentrations of antiretroviral drugs. Clin Pharmacokinet 2004; 43:845–853.
18. Cohen A, Doyle D, Martin D, Ammann A. Abnormal purine metabolism and purine over production in a patient deficient in purine nucleoside phosphorylase. N Engl J Med 1976; 295:1449–1454.
19. Myers L, Hershfield M, Neale W, Escolar M, Kurtzberg J. Purine nucleoside phosphorylase deficiency (PNP-def) presenting with lymphopenia and developmental delay: successful correction with umbilical cord blood transplantation. J Pediatr 2004; 145:710–712.
20. Markert ML. Purine nucleoside phosphorylase deficiency. Immunodefic Rev 1991; 3:45–81.
21. Ray A, Olson L, Fridland A. Role of purine nucleoside phosphorylase interactions between 2′,3′-dideoxyinosine and allopurinol, ganciclovir or tenofovir. Antimicrob Agents Chemother 2004; 48:1089–1095.
22. Arpaia E, Benvebniste P, Di Cristofano A, et al
. Mitochondrial basis for immune deficiency: evidence from purine nucleoside phosphorylase- deficient mice. J Exp Med 2000; 191:2197–2207.
23. Barreiro P, Garcia-Benayas T, Rendon A, Rodriguez-Novoa S, Soriano V. Combinations of nucleoside/nucleotide analogues for HIV therapy. AIDS Rev 2004; 6:234–243.
24. Becher F, Landman R, Mboup S, et al
. Monitoring of didanosine and stavudine intracellular triphosphorylated anabolite concentrations in HIV-infected patients. AIDS 2004; 18:181–187.
25. Robbins B, Wilcox C, Frodland A, et al
. Metabolism of tenofovir and didanosine in quiescent or stimulated human peripheral blood mononuclear cells. Pharmacotherapy 2004; 23:695–701.
26. Murphy M, O'Hearn M, Chou S. Fatal lactic acidosis and acute renal failure after addition of tenofovir to an antiretroviral regimen containing didanosine. Clin Infect Dis 2003; 36:1085–1802.
27. Jemsek J, Hutcherson P, Harper E. Poor virologic responses and early emergence of resistance in treatment-naive, HIV-infected patients receiving a once daily triple nucleoside regimen of didanosine, lamivudine and tenofovir DF
. In: Program and Abstracts of 11th Conference on Retroviruses and Opportunistic Infections
, San Francisco, CA, February 2004 [abstract 51].
28. Valer L, Martín-Carbonero L, de Mendoza C, Corral A, Soriano V. Predictors of selection of K65R: tenofovir use and lack of thymidine analogue mutations. AIDS 2004; 18:8–10.
29. Leon A, Martinez E, Mallolas J, Laguno M, Blanco JL, Pumarola T, et al
. Early virological failure in treatment-naïve HIV-infected adults receiving didanosine and tenofovir plus efavirenz or nevirapine. AIDS 2005; 19:213–215.