Human immunodeficiency virus (HIV)-specific T helper responses fail to predict CD4+ T cell decline following short-course treatment at primary HIV-1 infection


  • J. Fox,

    Corresponding author
    1. Department of Genitourinary Medicine & Infectious Disease, Division of Medicine, Wright Fleming Institute, Imperial College London, London, and
    Search for more papers by this author
    • These authors contributed equally to this manuscript.

  • T. J. Scriba,

    1. The Peter Medawar Building for Pathogen Research, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, UK
    Search for more papers by this author
    • These authors contributed equally to this manuscript.

  • N. Robinson,

    1. The Peter Medawar Building for Pathogen Research, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, UK
    Search for more papers by this author
  • J. N. Weber,

    1. Department of Genitourinary Medicine & Infectious Disease, Division of Medicine, Wright Fleming Institute, Imperial College London, London, and
    Search for more papers by this author
  • R. E. Phillips,

    1. The Peter Medawar Building for Pathogen Research, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, UK
    Search for more papers by this author
  • Sarah Fidler

    1. Department of Genitourinary Medicine & Infectious Disease, Division of Medicine, Wright Fleming Institute, Imperial College London, London, and
    Search for more papers by this author

J. Fox, Department of Infectious Diseases, Faculty of Medicine, Imperial College London, St Mary's Campus, Norfolk Place, London W2 1PG, UK.


Early anti-retroviral treatment (ART) in primary human immunodeficiency virus (HIV) infection (PHI) may have unique, restorative immunological and virological benefits which could enhance clinical outcomes. However, the sustainability of these HIV-specific immune responses and their impact on clinical outcome remains unclear. We present a 3-year longitudinal clinical and immunological follow-up of a single-arm, prospective study assessing the long-term impact of a short-course of ART (SCART) during PHI. Twenty-eight subjects with defined PHI received 3 months of SCART at HIV-1 seroconversion. HIV-specific interferon-γ+ CD4+ T cell responses, CD4 cell counts and plasma viral loads were assessed prospectively. Clinical outcome was defined as the time taken from PHI to a fall in CD4 cell counts <350 cells/μl on two or more occasions. Of 28 patients, 25 (89%) had detectable HIV-specific CD4+ helper responses at baseline. Five of 11 (45%) patients had preserved HIV-specific CD4+ responses 3 years after stopping SCART. Neither the presence nor magnitude of HIV-1-specific T helper responses either at baseline or 3 years following SCART cessation predicted clinical outcome. Rebound viraemia associated with stopping SCART did not diminish HIV-1-specific CD4+ responses. Long-term (>3 years) preservation of virus-specific CD4+ cells occurred in 45% of patients receiving SCART in PHI. There was no correlation between either the presence or magnitude of these responses and clinical outcome.


Primary human immunodeficiency virus (HIV) infection (PHI) is associated with a rapid and massive rise in plasma viral load (pVL) causing preferential infection and destruction of HIV-specific CD4+ memory T cells [1]. In most people, proliferating HIV-specific CD4+ T helper cells are lost early in infection and fail to recover despite later anti-retroviral therapy (ART) and associated immune reconstitution [2,3]. Early studies suggested that the presence of proliferative HIV-specific CD4 T helper responses correlated with the steady state pVL [3,4]. However, when quantified by interferon (IFN)-γ detection assays [4–6] or human leucocyte antigen (HLA) class II tetramers [7] they did not. It is hypothesized that timely ART in PHI may preserve HIV-specific CD4+ T cell immunity [8–11]. This may enhance immune control of viral replication when therapy is subsequently discontinued, leading to an improved clinical outcome [12,13]. However, the longevity of this effect or whether this confers enhanced clinical outcome is not known. To date, two studies have reported 3-year immunological follow-up of extended ART intervention and/or structured treatment interruption (STI). Both studies failed to show reliable long-term preservation of HIV-specific CD4+ responses and found no correlation between IFN-γ-producing HIV-specific CD4+ responses and virological control [14,15]. Compared with extended ART intervention and/or STI, short-course ART (SCART) is an attractive option as it reduces both toxicity and cost [9].

Here we report on the relationship between the presence and frequency of HIV-specific CD4+ T helper responses with clinical outcome. We studied 28 individuals who received 3 months of ART at diagnosis of PHI. Initial data from this study showed that all patients were able to maintain HIV-specific CD4+ responses for 64 weeks of therapy [9]. We show here that neither the presence nor the magnitude of HIV-specific CD4+ lymphocytes as measured by IFN-γ enzyme-linked immunospot (ELISPOT) at baseline and up to 3 years after SCART cessation correlate with clinical outcome.


Twenty-eight patients with strictly defined PHI were recruited between November 1999 and July 2001 into a prospective, non-randomized SCART intervention study. Criteria for PHI were: (i) a positive HIV-antibody test and previous negative HIV-antibody test within the previous 6 months (n = 14); (ii) positive HIV DNA polymerase chain reaction (PCR) in the absence of a HIV-specific antibody (n = 4); and (3) an evolving titre-positive HIV antibody test (n = 10). SCART was commenced within 2 weeks of diagnosis and stopped at 12 weeks or once pVL <50 HIV RNA copies/ml. Genotypic viral drug resistance testing was performed in real time at baseline and 4 weeks after discontinuation of SCART.

Plasma viral load and CD4 counts were performed monthly and HIV-specific CD4+ responses measured at baseline, on the first visit following SCART cessation and for 11 individuals at 6-monthly intervals thereafter. The clinical end-point was defined as the time to reaching a CD4 count of less than 350 cells/μl on two occasions 6 months after PHI. The study was approved by the local research ethics committee and written informed consent obtained in all cases.


Plasma viral load was quantified using the Chiron 3·0 reverse transcription–PCR. The range of detection was 50–> 500 000 HIV RNA copies/ml. CD4 T cell subset counts were performed using standard fluorescence activated cell sorter (FACS) analysis. HIV genotypic analysis in pol was performed using either Visible Genetics Trugene (Visible Genetics, Evry, France) or the ViroSeq vII HIV Genotyping System (Applied Biosystems, Warrington, UK), according to the manufacturers' instructions. HIV subtyping was performed using NCBI blast algorithm software ( The impact of codon changes within pol was analysed using the χ2 and Fisher's exact tests for discrete variables.

Human immunodeficiency virus-specific CD4+ T helper responses were measured by IFN-γ ELISPOT assay at baseline and 3 years after stopping SCART. Direct ex vivo IFN-γ ELISPOT analysis was performed as described previously [9]. Briefly, fresh peripheral blood mononuclear cells (PBMC) were depleted of CD8+ T cells using anti-CD8 conjugated Dynabeads (Dynal, Bromborough, UK). The efficiency of the CD8 depletion was tested routinely by FACS staining and the post-depletion frequency of CD8 cells never exceeded 0·5%. CD8-depleted PBMC were stimulated in duplicate wells with recombinant proteins [p24, gp120, p66 at 10 μg/ml; National Institute for Biological Standards and Control, Potter's Bar, Herts, UK] and overlapping pooled peptides (20 mers overlapping by 10; Tat and Nef at 5 μg/ml). HIV-unrelated recall antigens streptokinase and streptodornase (200 U/ml; Virusys Corporation, Sykesville, MD, USA), CMV lysate (10 μg/ml; Sigma-Aldrich, St. Louis, MO, USA) and phytohaemagglutinin (5 μg/ml) were used as positive controls. Non-recombinant baculovirus and medium alone were used as negative controls. Spot quantification was automated and standardized with an ELISPOT plate reader (Autoimmune Diagnostika, Strassburg, Germany). Responses were considered positive when the mean spot-forming cells (SFC) per million CD8-depleted PBMC for duplicate wells was equal to or exceeded 50 SFC above background. A group of 10 HIV-1 seronegative individuals was used as control group to determine the background IFN-γ response. The median total HIV-specific CD4+ T helper response in this group was 19 (range 0–32) SFC per million CD8-depleted PBMC (data not shown).

CCR5Δ32 genotyping was performed by PCR using primers flanking the 32-base pairs (bp) deletion in CCR5 as described previously [16]. High-resolution HLA class I and II genotypes were determined for each patient by PCR using sequence-specific primers [17].

Correlations and differences were tested using the Spearman's non-parametric correlation test, the Mann–Whitney U-test or Student's t-test with Prism V4·0a software (GraphPad Software, San Diego, CA, USA).


The median age of the 28 individuals was 32 years (range 24–40). In all cases HIV transmission was sexual, the majority in Caucasian, homosexual men infected with B clade virus. The median time from symptoms to diagnosis of primary infection was 72 days (range 1–98). At baseline, the mean pVL was 5·00 log10[standard deviation (s.d.) ± 0·63] RNA copies/ml and the median CD4 cell count was 495 cells/μl (range 230–790). Eighteen individuals (64%) described symptoms associated with seroconversion, although none had evidence of opportunistic infections. Baseline genotyping showed that all viruses were wild-type. Table 1 shows the individual baseline characteristics of the cohort with 3-year immunological follow-up. One individual was CCR5Δ32 heterozygous.

Table 1.  Characteristics of cohort with long-term follow-up, n = 11.
Study no.Age (years)SexCountry of acquisitionViral subtypeBaseline CD4Baseline pVLTreatment at PHITime symptoms to diagnosisBaseline viral genotypeHLA class I A and B lociCCR5 statusHIV- specific CD4+ immune responses at baselineHIV- specific CD4+ immune responses at 3 yearsCD4 <350 by 3 years
  • *

    <50 = no immune responses. PHI, primary human immunodeficiency virus infection; pVL, plasma viral load; HLA, human leucocyte antigen.

125MaleUKB4804 376Combivir
28Wild-typeA*0101, B*0801,
336MaleUSAB480158 525Combivir
15Wild-typeA*0201, *0301, B*3501, *4001Wild-typeYesYesNo
42Wild-typeA*0201, *3601, B*0702, *5201Wild-typeYesNoNo
931MaleUKB590309 704Combivir
14Wild-typeA*0101, *2402, B*4101, *4402Wild-typeNoNoNo
1029MaleUKB50071 396Combivir
28Wild-typeA*0201, *0301, B*4501, *55Wild-typeNoNoYes
1724MaleUKB5302 369Combivir
90Wild-typeA*0101, *0201, B*1302, *5701HeterozygousYesYesNo
2434MalePolandB390> 500 000Combivir
2Wild-typeA*0301, *2501, B*1402, *3501Wild-typeYesYesNo
2640MaleZaireA/E23056 279Combivir
90Wild-typeA*0301, *2501, B*1402, *3501Wild-typeNoNoYes
2836MaleUKB26080 052Combivir
3Wild-typeA*0101, *0201, B*0702, *5501Wild-typeYesNoYes
3139MaleUKB380365 156Combivir
56Wild-typeA*2301, *6901, B*4901, *5501Wild-typeYesNoYes
3539MaleUKB12003 643Combivir
60Wild-typeA*0201, *3201, B*2702, *1401Wild-typeYesYesNo

Longitudinal CD4 counts, pVL and HIV-specific CD4+ T helper responses of the 11 individuals followed-up for 3 years after SCART cessation are shown in Fig. 1a–c. The baseline pVL and CD4 counts between this subset (n = 11) and the whole cohort (n = 28) were comparable (P = 0·55 and P = 0·91 respectively), as was the time to clinical end-point (P = 0·633, Fig. 1d) and numbers restarting ART because of clinical progression (27% and 39% respectively). No individual remained virologically suppressed after stopping SCART.

Figure 1.

Longitudinal follow-up of human immunodeficiency virus (HIV)-specific CD4+ T helper responses. Total HIV-specific CD4+ T helper responses were measured by interferon (IFN)-γ enzyme-linked immunospot in 11 patients and are expressed as spot-forming cells per million CD8-depleted peripheral blood mononuclear cells. (a) Longitudinal plasma viral load (pVL) of patients (n = 11). (b) CD4 count (n = 11). (c) Total HIV-specific CD4+ T helper responses. Each box plot represents the median, 75th and 25th percentiles. Error bars represent the range. At the time of short-course of anti-retroviral treatment (SCART) cessation (0 days), one patient (8%) had detectable pVL (represented as a circle). (d) Kaplan–Meier analysis of the patients with baseline (n = 28) and those with 3-year HIV-specific CD4+ T helper responses (n = 11). Vertical ticks indicate patients who have been lost to follow-up or were event-free at the latest time of analysis. E, F, G and H show the relationship between IFN-γ-producing HIV-specific CD4+ T cells and clinical outcome. (e) HIV-specific CD4+ responses at baseline (n = 28) and time to clinical end-point. Patients who had not reached end-point at the time of analysis are denoted by § (n = 10/28). (f) HIV-specific CD4+ responses 3 years after SCART cessation (n = 11) and time to clinical end-point (five had not reached clinical end-point, three restarted ART, denoted by *). P-values and regression coefficients were calculated with the Spearman non-parametric correlation test. (G) HIV-specific CD4+ responses at baseline (n = 28) and pVL 3 years after SCART cessation (excluding 11 who had recommenced ART. (h) HIV-specific CD4+ responses at 3 years (n = 11) and pVL at 3 years after SCART cessation (excluding three who had recommenced ART).

Figure 1c demonstrates the variability of HIV-specific CD4+ T helper responses in the 11-patient subset over 3 years. Regardless of treatment status 3 years after SCART cessation, five of 11 (45%) patients had detectable HIV-specific CD4+ responses. Compared with baseline (634·6 SFC per 106 CD8-depleted PBMC), the mean total HIV-specific CD4+ responses (measured by the summation of p24-, gp120-, Tat-, Nef- and p66-specific responses) at 3 years after SCART cessation was low at 149·1 SFC per 106 CD8-depleted PBMC (s.d. ± 202·3) above background.

There was no relationship between those starting ART within 3 weeks of seroconversion symptoms and those starting later (P = 0·21).

Interferon-γ-producing HIV-specific CD4+ T helper responses do not correlate with pVL or total CD4 counts [4,6,18]. Neither the magnitude nor presence of specific CD4+ responses at baseline (r = 0·2213, Fig. 1e) or at 3 years (r =  0·2032, Fig. 1f) predicted the time to clinical end-point. This was supported by the lack of correlation between pVL and HIV-specific CD4+responses either at baseline (r = 0·0068, Fig. 1g) or at 3 years post-SCART cessation (r = 0·2440, Fig. 1h). In addition, there was no association between the breadth of response at 3 years and clinical outcome (P = 0·386). Of the five individuals with detectable HIV-specific T helper responses at 3 years, one reached clinical end-point 92 weeks after stopping SCART and subsequently restarted ART. Of the six without detectable HIV T helper responses at 3 years, two reached clinical end-point (50 and 60 weeks) after stopping SCART and one subsequently restarted ART.

Some investigators have expressed concern that the resurgence of virus upon ART cessation may destroy HIV-specific CD4+T cells, perhaps irreversibly [19,20]. To address the fate of virus-specific helper cells during viral rebound, we compared the total HIV-specific CD4+ response before and after SCART cessation. At peak viraemia on stopping SCART, the median total response (415·0, s.d. ± 268·0) increased significantly (P = 0·0031) above the response at week 12 on SCART (125·0 SFC per 106 CD8-depleted PBMC; s.d. ± 131·3, data not shown). This increase at peak viraemia was more pronounced after CD4 normalization (data not shown), suggesting that rebounding virus boosts the HIV-specific T helper population, possibly though increased turnover of T cells in response to increased antigen stimulation [7].


Timely anti-retroviral therapy at PHI may preserve HIV-specific T helper responses [8–11]. Whether this intervention improves clinical outcome remains unclear. Early studies have shown that proliferation of HIV-specific CD4 T helper cells correlated with the steady state pVL [3,4]. However, when HIV-specific T helper cells were quantified by IFN-γ detection assays [4–6] or HLA class II tetramers [7], they did not. We correlated clinical end-points with CD4+ T cell IFN-γ assays following a short pulse of ART in PHI. HIV-specific CD4+ cells were present at baseline in 89% of the enrolled individuals. The presence or magnitude of HIV-specific CD4+ responses at baseline did not correlate with clinical outcome. Positive responses in 45% of individuals at 3 years after ART cessation also bore no correlation with CD4 T cell decline or steady state pVL. The lack of correlation between HIV-specific IFN-γ-producing CD4+ T cells and pVL corroborates previous studies [4–6]. There were insufficient numbers to determine if the timing between PHI diagnosis and ART intervention had any impact on the preservation of HIV-specific CD4+ responses. A recent study has suggested that earlier ART intervention may confer enhanced clinical benefit [21].

Rosenberg et al. suggested that early ART may result in control of viraemia [11]. Following this, the goal of ART in PHI was deemed to be the preservation of HIV-specific CD4+ T helper cells. However, despite initially encouraging results following intensive ART intervention and STI in PHI [11], long-term follow-up of this cohort failed to define any clear relationship between HIV-specific CD4+ helper responses and viral control [15].

Our study represents the first ART intervention study in PHI to correlate functional T helper assays with clinical outcome. These data, albeit under-powered, suggest that assays measuring IFN-γ-producing CD4+ T cells are a poor correlate of clinical outcome. This suggests a need to re-evaluate the role played by these cells or these assays in HIV pathogenesis and progression.

Measurement of other T cell functions may very well be more relevant. These include IL-2 production and CD4 proliferation which correlate inversely with pVL [22,23]. Alternatively, CD4 T cell activation, measured as CD38 expression, correlates directly with pVL and is predictive of changes in total CD4 counts [24] or PD-1 expression on CD4 T cells has shown similarly positive correlation with pVL and an inverse correlation with CD4 T cell count. Importantly, blockade of the PD-1 pathway rescues T cell function showing that the PD-1/PD-L1 pathway is immunoregulatory and defines a reversible defect in HIV-specific T-cell function [25].

We present data on a relatively small number of participants (n = 28) in an observational study. Limited longitudinal data comparing cellular immune functions in the context of clinically applicable outcomes exist. As such, these findings are highly relevant to the design and use of cellular immunology assays in the context of future clinical trials. It is inevitable that such trials will use HIV-specific immune assays; unless observations such as these are published, the design of future studies will be compromised.

Finally, this study allays concerns that rebound viraemia might destroy HIV-specific helper T cells [19,20] and concurs with a previous study using HLA class II tetramers to track HIV-specific T helper cells during rebound [7,22,23]. Other reports have also shown increased IFN-γ production by HIV-specific CD4+ cells in viraemic versus treated HIV infection [22,23,26]. Although there was an expected drop in total CD4 T cell numbers corresponding with stopping SCART, this was transient. The discontinuation of ART at PHI therefore appears safe in terms of immunology [26], drug toxicity [27] and failure to induce drug resistance [28]. These findings may be relevant to other situations of ART cessation, such as in the prevention of mother-to-child transmission, medication fatigue or inconsistent availability of ART.

The only approach to addressing clearly whether ART can improve CD4 T cell decline is with an appropriately powered, randomized controlled trial. Such a multi-national trial is currently under way (the SPARTAC Trial).


The authors thank S. Mullaney for nursing support. T. J. S. is supported by studentships from the Fogarty Foundation and the National Research Foundation (South Africa). This work was supported by the Wellcome Trust. R. E. P. and J. N. W. are Wellcome Trust Senior Fellows.