A. De Maria, Istituto Nazionale per la Ricerca sul Cancro, Largo Rosanna Benzi 10, Genova 16132, Italy. E-mail: firstname.lastname@example.org
Anti-retroviral treatment (ART) usually results in efficient control of virus replication and in immune reconstitution. Among potential adverse effects, impairment of immune responses in terms of CD4+ T cell counts has been attributed to some ART regimens, as with didanosine–tenofovir. We studied the functional integrity of adaptive and innate immunity during didanosine–tenofovir-containing ART. Two groups of extensively pretreated patients completing at least 48 weeks of ART containing either lamivudine–didanosine (n = 21) or tenofovir–didanosine (n = 25) were identified. In addition to standard clinical immune and virological parameters, we performed a flow cytometric analysis of natural killer (NK) cells, of memory and naive CD4+ T cells and of T cell receptor αβ+ T cells co-expressing inhibitory NK receptors. Functional analysis consisted in specific and total interferon-γ production by NK cells and of recall antigen proliferation of peripheral blood mononuclear cells. Comparable clinical immunological reconstitution and virological control were confirmed in the two groups of patients in the absence of clinically relevant adverse effects. The proportion of CD4+CD45RA+ T cells and of functionally inhibited killer immunoglobulin-like receptor T cell receptor αβ+ cells, the proliferation to recall antigens as well as NK cell phenotype and function as determined by interferon-γ production in patients treated with tenofovir–didanosine were comparable to those treated with a different regimen. Thus, no differences in functional innate or adaptive immune reconstitution are detected in drug-experienced human immunodeficiency virus-infected patients on tenofovir–didanosine nucleoside reverse transcription inhibitor regimens.
Potent anti-retroviral treatment (ART) induces rapid control of viral replication to undetectable levels with reductions of plasma viral loads within weeks/months from starting treatment [1,2]. Biphasic increases in CD4+ T cell numbers are seen with a slower kinetics. An initial redistribution of CD4+ memory cells within the first weeks of treatment is followed by a slower increase in CD4+ naive cells that requires months to years of treatment . Full reconstitution of the immune system during successful ART is not obtained in the majority of patients. The late and often incomplete dynamics of CD4+ peripheral increase [3,4] is particularly evident in some patients, referred to as ‘immunological discordant’[5,6]. Other forms of incomplete recovery may be viewed as possible shortcomings of immune reconstitution, and include the persistence of T cell activation [7,8] and incomplete recovery of peripheral dendritic cell (DC) phenotype [9,10] in the presence of successful suppression of viral replication.
When prescribing any line of ART to naive or drug-experienced patients, multiple drug combinations are recommended according to known clinical and virological parameters including, among others, drug efficacy, combination regimen potency, suitability for improved patient adherence, likelihood of durability of action due to high genetic barriers to the accumulation of critical viral mutations, forgiveness in case of occasional dosing mistakes and likelihood or quality of adverse effects [11–14].
Some anti-retroviral drugs may modulate immune responses. Interactions of tenofovir with didanosine metabolism through depletion of purine nucleoside phosphorylase results in increases in plasma didanosine and in up to twofold increases in intracellular didanosine concentrations [15,16]. This mechanism has been associated with a paradoxical reduction of CD4+ T cell counts and lymphocytopenia, despite the maintenance of viral suppression in patients receiving standard-dose didanosine combined with tenofovir and nevirapine [17,18]. More reassuringly, recent data report stability of CD4+ T cells in large numbers of patients treated in controlled studies with tenofovir and lamivudine-containing regimens  when reduced didanosine dosing is administered [20,21].
Although CD4+ cell numbers are a validated surrogate marker of immune competence , the question regarding quality of overall immune response may be raised for patients who would be assigned to the tenofovir–lamivudine combination even after genotypic testing, due to previous reports cautioning against its use [17,18,23]. To some extent, new drug classes introduced in clinical practice, including integrase inhibitors, entry inhibitors or CCR5 inhibitors, may also affect immune function. Entry inhibitors or CCR5 inhibitors may affect CD4+ T lymphocyte function through interactions with homeostatic mechanisms that regulate lymphocyte trafficking and distribution. Also, different drugs have variable effects on CD4+ T lymphocytes and monocytes, with notable differences in terms of anti-viral activity  and of possible effects of ART on the immune system .
In addition to known perturbations of adaptive responses also innate immune responses are affected during human immunodeficiency virus (HIV)-1 infection. Perturbations include reduced numbers and function of mDC [9,10,26,27], natural killer (NK) cell activation , decreased activating receptor expression and function , decreased interferon (IFN)-γ and tumour necrosis factor-α production  and increased NKG2C expression . NK cells have editing functions on DCs  with potential effects on the shape of adaptive T and B cell responses, as they depend on DC for antigen presentation [27,33]. Defective DC editing has also been proved for HIV-1-infected viraemic patients, while suppressed patients have partial recovery of these defects .
In view of the possible interaction between ART and the immune response outlined above, and of persisting doubts on CD4 cell function in spite of conserved CD4+ cell numbers after didanosine+tenofovir treatment, a functional evaluation was considered to be relevant. We report here on a study of adaptive and innate immune parameters in these patients.
Participants and methods
Patients on treatment with ddI were identified as described previously . Patient exclusion criteria were: concomitant treatment with immunosuppressors such as hydroxyurea, cyclosporin or mycophenolate or with immune modulators including interleukin-2 and IFN-α. Patients with improper adherence or who discontinued or interrupted treatment for adverse events were also excluded from the evaluation. All the patients administered tenofovir were treated with ddI 250 mg once daily, while ddI dosage was 400 mg once daily in the patients administered other nucleoside reverse transcription inhibitor (NRTI) drugs. All other drug dosages followed published guidelines.
Of 730 HIV-1-infected ART-treated patients followed-up to April 2006, 46 had completed at least 48 weeks of treatment and were included in the present evaluation. Of these, 21 patients were treated with lamivudine–didanosine and 25 with tenofovir–didanosine. A third drug (protease inhibitors of NNRTI) was given in all cases. Compared to a previous 24-week clinical report , an increased number (seven patients 3TC+ddI, six patients TDF+ddI) qualified for the 48-week end-point (Table 1).
Table 1. Demographic and clinical characteristics of the patients.
P = not significant. HAART: highly active anti-retroviral therapy; PI: protease inhibitors; s.d.: standard deviation.
(Mean + s.d.)
Years of infection
(Mean + s.d.)
No. of previous drugs
(Mean + s.d.)
No. of previous regimens
(Mean + s.d.)
Nadir CD4+ cells(#/µl)
(Mean + s.d.)
Nadir CD4+ cells(%)
(Mean + s.d.)
Zenith viral load
Baseline CD4+ cells(#/µl)
(Mean + s.d.)
Baseline CD4+ cells(%)
(Mean + s.d.)
Viral load baseline
During the 48-week observation period, changes in HIV viral load and in peripheral CD4+ lymphocyte counts, serum transaminase concentrations, serum amylase, lipase and creatinine were monitored. The appearance of proteinuria and serum creatinine concentrations were also monitored. Glomerular filtration rate (GFR) was derived using the Cockroft–Gault formula after measurement of age, weight, serum creatinine and serum albumin concentrations.
The original aim of the study was the comparison between a regimen containing didanosine plus lamivudine with a regimen containing didanosine plus tenofovir based on clinical and immunovirological parameters. In consideration of persisting doubts on CD4 cell function, in spite of conserved CD4+ cell numbers after didanosine+tenofovir treatment, a functional evaluation was considered to be relevant. For this reason, an immunological study was offered to the patients and 22 agreed to the evaluation (10 treated with lamivudine–diadanosine and 12 with didanosine–tenofovir) after giving their informed consent.
Peripheral blood mononuclear cells (PBMC) were isolated by density gradient centrifugation on Lympholyte (Cedarlane, Ontario, Canada) and were used immediately in proliferation assay and in the immunophenotyping on CD45 subpopulations. The remaining PBMC fraction was cryopreserved for further phenotypical and functional studies to characterize NK and T cell receptor (TCR)αβ+ cells. Culture medium was RPMI-1640 supplemented with 5% autologous plasma for proliferation assay or with 10% fetal calf serum, l-glutamine (2 mM) and 1% antibiotic mixture (Pen-Strep 5 mg/ml, stock solution) for intracellular IFN-γ production.
Protein purified derivative (PPD) from Mycobacterium tuberculosis (Statens Seruminstitut, Copenhagen, Denmark) was used at 5 µg/ml final concentration. Candida albicans (Ca) was grown in RPMI-1640 medium for 2 days. Pneumocystis carinii (Pc) was prepared from homogenized lungs of immunosuppressed rats by gradient fractionation. Both pathogens were washed twice in phosphate-buffered saline, autoclaved and used at 2 × 106 bodies/ml final concentration. This concentration induced maximal PBMC stimulation, as determined in preliminary titrations experiments. Phytohaemagglutinin (PHA; Sigma-Aldrich, Milan, Italy) was used at 10 µg/ml final concentration. Aliquots of density gradient separated PBMC were resuspended in RPMI-1640 supplemented with 5% autologous plasma. Resuspended PBMC (200 µl at 2 × 106/ml) were cultured in duplicates in flat-bottomed microtitre plate wells (Nunc; Sardstedt, Nümbrech, Germany) with or without antigens for 4 days in two independent experiments. Wells were pulsed with 0·5 µCi [3H]-thymidine (Amersham, Amersham, UK) on day 4 and harvested after 6 h. Wells with PHA were pulsed similarly on day 4 in parallel with recall antigens. The dry filters were counted in a Canberra Packard Matrix 9600 beta counter without scintillation fluid. Antigen-specific tests by proliferation assay are considered positive: (i) if the stimulation index (SI) calculated by proliferation (Kcpm) with antigen/proliferation (Kcpm) without antigen is above 2; (ii) if the proliferation in presence of antigen is over 1 Kcpm; and (iii) if the spontaneus proliferation results are below 0·5–1 Kcpm. Here, data are grouped according to treatment and represented quantitatively as median SI in box-plots.
The following panel of anti-human monoclonal antibodies (mAb) was used for surface staining: anti-TCRαβ: WT31 [immunoglobulin G1 (IgG1)]; anti-CD16: c127(IgG1); anti-NKp30: antizyme 29 (Az20) (IgG1); anti-NKp46: Bab 281 (IgG1); anti-killer immunoglobulin-like receptor (KIR)2dL2/S2: GL183 (IgG1); anti-KIR2dL1/S1: 11pb6 (IgG1); anti-KIR3dL1: Z27 (IgG1) and anti-NKG2A: Z270 (IgG1) and Z199 (IgG2b); anti-LIR1: F278(IgG1). In addition mAb anti-NKG2C MAB1381(IgG2b) from R&D Systems Inc. (Minneapolis, MN, USA), anti-TCR Panαβ clone BMA 031 (IgG2b), anti-CD56PC5 and anti-CD56PC7 from Beckman Coulter (Immunotech-Coulter, Marseille, France); anti-CD3 fluorescein isothiocyanate (FITC), anti-CD3 peridinin chlorophyll (PerCP), anti-CD45RAFITC, anti-CD4PE from BD PharMingen (San Jose, CA, USA) were purchased. FITC-conjugated (Southern Biotechnology, Birmingham, AL, USA) and phycoerythrin-conjugated antibodies (BD PharMingen) goat anti-mouse anti-isotype were purchased. For intracellular staining anti-IFN-γ antigen-presenting cells was used (BD PharMingen).
Samples were analysed by three- and two-colour flow cytometry analysis (FACSCalibur; Becton Dickinson, Mountain View, CA, USA) using the Cell QuestPro program as described previously . For each analysis 10 000 events were acquired.
Intracellular IFN-γ production
A total of 500 000 PBMC were harvested and resuspended in complete medium plus 10 U/ml rIL-2 (Proleukin; Chiron Corporation, Emeryville, CA, USA) in the presence of 50 000 P815 target cells and either anti-CD19 (BU19, IgG1) or anti-NKp30 at 37°C overnight. Phorbol myristate acetate (PMA)/ionomycin (0·5 ng/ml) for 4 h were used as maximal stimulation. After 16 h, GolgiPlug (BD Pharmingen, San Jose, CA, USA) was added (4 h, 37°C). Anti-CD3FITC and anti-CD56PC7 mAbs were used for surface staining followed by permeabilization/fixation according to BD Citofix/Citoperm™ protocol and subsequently intracellular staining for IFN-γ in the presence of permeabilizing solution (0·1% saponin phosphate-buffered saline). Cells were analysed on a flow cytometer (FACSCalibur; Becton Dickinson) and for each analysis 20 000 events were counted in CD3- gated lymphocytes.
All statistical analyses were performed using Statview 4·2 software. Continuous variables were compared with the non-parametric Mann–Whitney U-test, nominal variables were compared with Fisher's exact test. Significant results were determined with a two-tailed formulation; logistic regression analysis was applied as described previously .
The overall immune status of the patients at the time of the switch to either tenofovir–didanosine or lamivudine–didanosine reflected the high number of previous failed regimens. The two groups of patients were comparable according to demographic parameters and to baseline HIV-related conditions (Table 1). At the time of the switch there was a comparably low CD4+ cell nadir (13% and 9%, or 137/µl and 145/µl, respectively) and marginally satisfactory absolute CD4+ cell numbers and percentages (<300/µl and <20%, respectively, in both arms) (Table 1). Accordingly, after 48 weeks of either regimen there was an increase of CD4+ cell numbers to clinically satisfactory median values (415/µl and 345/µl, lamivudine–didanosine and tenofovir–didanosine, respectively (Fig. 1a). This represented a comparable median increase of 18% and 15% (lamivudine–didanosine and tenofovir–didanosine, respectively) when expressed as the change in absolute CD4+ cell numbers between study entry and week 48 (Fig. 1). Among patients achieving virological control, nine were discordant responders with poor immune recovery at 48 weeks (<10% CD4+ cell recovery). Of these, four were on lamivudine–didanosine with a median 3% CD4+ cell increase and five were on tenofovir–didanosine with a median 7% CD4+ cell increase. When virological control was considered, the proportion of patients controlling HIV replication in the tenofovir–didanosine arm was 56%, as opposed to 61% in the lamivudine–didanosine group [χ2 = not significant (n.s.)].
No acquired immune deficiency syndrome (AIDS) events or deaths were registered during the observation on treatment in both groups. Serum creatinine, transaminase, lipase and amylase concentrations remained within normal limits throughout week 48. To estimate potential decreases in renal function, GFR was calculated on data available at the beginning of the study and at week 48 of entry on the current treatment regimen. There were no significant differences at baseline (109·2 versus 104 ml/min mean GFR for lamivudine–didanosine versus tenofovir–didanosine arms, respectively), with irrelevant changes after 48 weeks of treatment (102·3 versus 104·1 ml/min, respectively). Logistic regression analysis to evaluate factors contributing to immunological control failed to show an association between the type of backbone regimen and the probability of increased. Higher CD4+ cell numbers, low number of previous lines of treatment, younger age and previous length of protease treatment contributed significantly to the whole model fit (P < 0·0014).
Phenotypic analysis of peripheral blood NK and T cells
Flow cytometric analysis showed similar proportions of CD4+CD45RA+ naive cells in the two groups of patients (Fig. 2a), indicating comparable thymic recovery.
Functional inhibition of antigen-specific CD3+CD8+TCRαβ+ PBMC occurs during HIV-1 infection . Successful highly active anti-retroviral therapy induces a staged reduction of KIR expression on T lymphocytes with recovery of antigen-specific CD8+cytotoxic lymphocyte function . To address these points, flow cytometric analysis of KIR or NKG2A expression was performed on CD3+TCRαβ+ PBMC. As shown in Fig. 2b, KIR expression on TCRαβ+ was comparable, and was superimposable in both groups to those found in healthy individuals and in successfully treated patients [35,36].
The proportion of CD56bright NK cells (2·5% versus 3·4%, lamivudine–didanosine versus tenofovir–lamivudine, respectively, n.s.) and of CD56-CD16+ cells (15·9% versus 7·4%, respectively, n.s.) (Fig. 2c) were also comparable. Similarly, the expression of the major activating NK cell receptors NKp46, NKp30 and of human leucocyte antigen-E-specific activating (NKG2C/CD94) or inhibitory (NKG2A/CD94) receptors were comparable in the two patient groups (Fig. 2d). Although the differences were clearly not significant, the trend towards lower proportions of ‘dysfunctional’ CD56-CD16+ NK cells  and of inhibitory receptor expression (CD94/NKG2A) in the tenofovir–didanosine-treated group could reflect a positive influence of this regimen on innate immune parameters that may deserve further attention in larger groups of patients.
Thus, heavily drug-experienced patients treated for 48 weeks with a tenofovir+didanosine containing N(t)RTI backbone do not show evident phenotypic perturbations of T or NK cells.
Functional analysis of adaptive and innate cell functions
To evaluate the predominantly CD4+ cell-dependent proliferation to recall antigens, fresh PBMC were plated and assayed in the presence of antigen (PPD, Ca, Pc) or polyclonal (PHA) stimuli in vitro in [3H]-thymidine incorporation assays. As shown by analysis of the stimulation index ratios, recall antigen proliferation of PBMC in the tenofovir–didanosine was conserved. (Fig. 3a). Challenge with PHA represented a positive control but does not reflect peak PHA proliferation, which is usually observed after 2 days of in vitro stimulation. Therefore, PHA SI values are lower than those usually observed in 48-h assays. Further, Ca-specific proliferation was correlated directly with CD4+ memory cells in patients treated with tenofovir–lamivudine (Spearman's rank correlation P < 0·05). Similar results were obtained for PPD and pneumocystis. In addition, in a multiple regression model that included the proportion of CD4+CD45RA+ PBMC, of NKp30+ NK cells, of CD56bright NK cells, of CD56-CD16+ NK cells and the production of IFN-γ upon PMA-ionomycin stimulation as independent parameters, a correlation was found using in vitro proliferation (SI) as the dependent parameter to Ca (P = 0·0334) and to PPD (P = 0·0295) for patients treated with tenofovir–didanosine, but not in the comparator group.
NK cell function was studied by the production of IFN-γ by peripheral blood NK cells (CD3-gated PBMC) using intracytoplasmic flow cytometric staining upon polyclonal (PMA-ionomycin) or single-receptor (NKp30) stimulation. As shown in Fig. 3b, IFN-γ production was detected in a relevant fraction of NK cells after polyclonal stimulation in both groups of patients. Comparable proportions of cells also produced IFN-γ upon specific triggering, as assayed by mAb-mediated receptor cross-linking.
These functional assays are in line with phenotypic analyses and provide evidence for the first time that neither antigen-specific nor NK cell functional defects are detected in patients treated with tenofovir–didanosine-containing regimens over prolonged periods of times.
Contraindications or cautionary use of ART drug combinations to achieve suppression of viral replication and subsequent immune recovery are related usually to several factors, including immediate or delayed toxicities directly connected to the use of the molecule, to its pharmacokinetic profile/disposition, to drug–drug interactions or to a combination thereof.
Recent data on the effect of protease inhibitors on PBMC represent one first step at understanding some of these aspects . The interaction of tenofovir and didanosine initially attracted relevant attention due to initial unexpected declines of peripheral CD4+ T cells in patients with successful virological control [17,18]. Although the interaction has received subsequent clearance for clinical use, particularly for drug-experienced patients who benefit from adjusted didanosine dosing [20,21], clinical implications of a possible residual functional CD4+ or innate immunity impairment remain unaddressed.
The present work addresses this point and analyses patients who remained on a didanosine–tenofovir backbone combination for at least 48 weeks. The clinical findings confirm that earlier observations after 24 weeks of treatment  are in line with a recent report showing virological and immunological responses when this drug combination is chosen according to a genotypic resistance test , and also agree with those derived from larger groups of patients treated similarly over at least 48 weeks . To verify for possible interference with CD4+ cell recovery after successful control of viraemia, we also compared the number of discordant immunological responses (i.e. patients with successful suppression of viraemia without relevant increases of CD4+ cell numbers ), and this analysis also showed that the two treatment arms were comparable.
Functional T cell analysis was performed for the first time in this work, as previous reports concentrated only on CD4+ cell numbers [17,18,20,21,23,38,39]. Patients on didanosine–tenofovir combinations had comparable recall antigen-specific proliferative activity compared to control didanosine–lamivudine-treated patients. Memory CD4+ T cell proportions correlated positively with in vitro recall antigen proliferation to Ca and PPD. This suggests that there is a reliable functional correlate of quantitative CD4+ measurements in patients on tenofovir–didanosine NRTI backbone and represents a reassuring clinical finding in light of slightly lower mean CD4 numbers found in patients treated with this combination [20,39]. The additional finding of comparable expression of inhibitory NK cell receptors on CD3+TCRαβ+ lymphocytes in the two groups also contributes to support the view that there is no residual activation or functional inhibition of T cells mediated by human leucocyte antigen class I-specific inhibitory NK receptors in patients on long-term didanosine–tenofovir treatment.
The present study aimed at verifying feasibility and comparability of functional data in patients treated with different ART regimens. Limitations are posed by the availability of only a subgroup of patients for immunological study and by the absence of stored cryopreserved samples before the switch to the ART regimen. The clinical characteristics of the two cohorts at study entry were superimposable, however, and are therefore likely to reflect a homogeneous patient population. For this reason, direct functional comparisons at the time of the present observation may represent an acceptable functional comparison, even if more reliable information would have been obtained comparing samples gathered at the time of the switch to the current ART regimen.
Mechanism(s) underlying CD4+ T cell depletion in patients on standard-dose didanosine+tenofovir have been suggested to involve purine metabolism interactions . Here, we tested an alternative hypothesis of a defective NK cell shaping of adaptive immune responses leading to a possible skewing of CD4+ T cell function . In this regard, IFN-γ production upon NKp30 cross-linking is one of the main mechanisms that leads to NK cell activation and subsequent DC editing , promoting efficient antigen-specific expansion of CD4+ T cells [41,42]. Neither activating receptor (NCR) perturbations nor decreased IFN-γ production upon receptor cross-linking could be detected in patients treated with tenofovir–didanosine compared to the control arm treated with lamivudine–didanosine. Based on this evidence, no difference would be therefore anticipated in terms of NK–DC cross-talk and ultimately on antigen-specific CD4+ T cell function between patients treated according to the two NRTI regimens. This is supported further by the present finding of comparable PBMC proliferation to recall antigens. Together, these functional and phenotypic observations represent a reassuring note for patients still on this combination treatment or for those with only this option left.
In future, routine combined strategies assaying antigen-specific proliferation, CD8+ cytotoxic lymphocyte functional inhibition and NK cell phenotype function could be useful to monitor other ART regimens that may impact particularly on immune function .
This work has been supported by grants awarded by Istituto Superiore di Sanità (Programma Nazionale AIDS n. 40G.41/40F.55 and 45G.11), Italian Concerted Action for AIDS vaccine, Accordi di Collaborazione Scientifica n. 40D61 and 45D/1·13, Ministero della Salute; Ministero dell'Istruzione, dell'Università e della Ricerca.