Objectives: To determine the levels of residual HIV DNA and RNA in blood and gut reservoirs in aviremic patients, assess correlations among compartmental measurements of HIV burden, and evaluate association with clinical parameters.
Design: Cross-sectional analysis of baseline data only, on 40 patients enrolled in phase II study evaluating efficacy of autologous gene-modified CD4+ and CD8+ T cells. All patients were on stable antiretroviral regimen with undetectable plasma HIV RNA (< 50 copies/ml).
Methods: Measurements repeatedly performed over 8–12 weeks pre-intervention: blood HIV DNA, analysis of rectal mucosa-associated lymphoid tissue for both HIV RNA and HIV DNA, and quantitative co-culture of HIV from CD8-depleted peripheral blood mononuclear cells (PBMC).
Results: Quantifiable levels of HIV detected in compartments despite undetectable levels of plasma HIV RNA: HIV co-culture of PBMC (88%), blood HIV DNA (95%), rectal biopsy HIV DNA (95%), rectal biopsy HIV RNA (65%). A significant correlation existed among various measures of HIV burden (HIV co-culture, blood HIV DNA, rectal biopsy HIV RNA and DNA) but not between assays and clinical parameters [duration of highly active antiretroviral therapy (HAART), type of HAART]. All assays had comparable or less variability than in plasma viral load assays; HIV co-culture had the highest coefficient of variability whereas the blood HIV DNA assay had the lowest and was considered the most reliable assay.
Conclusions: The data support safety, feasibility and high compliance of quantifying reservoirs of residual HIV in treated subjects with undetectable plasma HIV RNA. Lack of correlation between levels of HIV in residual reservoirs and duration of HAART suggests treatment-mediated viral suppression alone does not lead to reproducible decay in HIV reservoirs.
From the aUniversity of California, Los Angeles, the bSan Francisco General Hospital, San Francisco, California, the cMassachusetts General Hospital, Boston, Massachusetts; dViRx Inc., San Francisco, eAdvanced Research Associates, Mountain View, California, fLos Alamos National Laboratory, Los Alamos, New Mexico, the gUniversity of California, San Diego, hRoche Molecular Systems, Alameda and jCell Genesys, Inc., Foster City, California, USA.
Correspondence to Peter A. Anton, M.D., Center for HIV and Digestive Diseases, UCLA AIDS Institute, MRL 2734, 675 Charles E. Young Dr. South, Los Angeles, CA 90095, USA. Tel: +1 310 206 5797; fax: +1 310 206 8824; e-mail: firstname.lastname@example.org
Received: 17 December 2001; revised: 25 July 2002; accepted: 7 August 2002.
Using highly active antiretroviral therapy (HAART), it is now possible to reduce plasma HIV-1 RNA to undetectable levels . In addition to the pronounced effect on HIV-1 viremia, clinical trials have now shown that combination antiretroviral therapy (ART) delays disease progression and prolongs life [2,3]. Although healthy, asymptomatic patients with normal levels of CD4+ T cells often have low levels of plasma viral RNA, active viral replication and T-cell turnover is still present in lymph nodes [4,5]. Recent studies have demonstrated that viable, replication-competent virus persists for at least 2 years in the face of HAART, despite complete suppression of HIV-1 RNA in blood and in some cases, lymphoid tissue [6–8]. Assays to detect HIV-1 RNA and DNA levels in lymphoid tissue have recently been described and these assays can now be used to follow the status of HIV infection in these tissues [9–12]. What has not been clarified is the degree to which quantified levels of virus in compartments correlate with and possibly predict clinical status and whether these measures may function as surrogate markers in clinical trials and/or care.
As the gut is the largest lymphoid organ in the body and a known reservoir for HIV, it is reasonable to investigate this compartment as a readily accessible source of lymphoid tissue in subjects with undetectable plasma viral load. The interest in quantifying viral burden of gastrointestinal-associated lymphoid tissue (GALT) HIV RNA and HIV DNA is compounded by the enhanced vulnerability to HIV infection of lymphocytes at this site compared to circulating peripheral blood mononuclear cells (PBMC) [13,14] based on activation state, memory phenotypes, expression of HIV coreceptors, and increased levels of soluble inflammatory mediators (e.g. tissue chemokines) .
We have recently completed a phase II study to evaluate the efficacy of autologous CD4-zeta gene-modified CD4+ and CD8+ T cells in HIV-infected subjects with undetectable plasma HIV RNA (< 50 copies/ml) and CD4+ cell counts > 200 × 106 cells/l on HAART for > 6 months . Such subjects have been estimated to have a total body load of latently infected CD4+ T cells with replication-competent HIV of < 107 cells . The virologic endpoints of the trial focused on quantitation of reservoirs of HIV-infected cells in blood and lymphoid (mucosal) tissues before and after cell infusions and is reported elsewhere . Assays at baseline and throughout the trial included measurement of HIV DNA in blood, analysis of rectal mucosa-associated lymphoid tissue for changes in viral burden (both HIV RNA and HIV DNA), and quantitative co-culture of HIV from CD8-depleted PBMCs using an enhanced limiting dilution co-culture assay .
We now report the results from baseline, steady-state, quantitative measurements of HIV burden using five assays in blood and gut compartments. Our primary objective in this analysis was to assess the relationship between viral burden in a variety of long-lived cellular reservoirs, and to describe the relationship between virus burden and clinical status (years of HIV infection, duration of antiviral therapy and CD4+ cell count). Our secondary objective was to determine the feasibility of using novel assays for measuring outcome in patients without detectable plasma viremia.
Materials and methods
Subjects and study design
The subjects reported here are those recruited for an investigator-blinded, randomized trial of gene therapy using gene-modified versus unmodified T-cell infusions in HIV-infected subjects at five clinical sites (reported elsewhere) . Only baseline data are reported here. Subjects were required to be undetectable by ultrasensitive plasma viral load assay (< 50 copies/ml) (Roche Amplicor HIV-1 Monitor® Test; Roche Molecular Systems, Alameda, California, USA) in two of three biweekly screening measurements, to have been on at least 24 weeks of stable treatment with a minimum of three ART medications, including at least one protease or non-nucleosides reverse transcriptase inhibitor (i.e. HAART), and have CD4+ T-cell count > 200 × 106 cells/L. A historical record of ART exposure was recorded as well as duration on HAART. For later analysis, the subject's ART was recorded as optimal or sub-optimal based on whether the initial regimen when ART-naive followed recommendations from the Department of Health and Human Services. Subjects with AIDS-defining complications, significant co-morbid illness, or recent history of treatment with immunomodulatory agents within 2 years were excluded.
During the pre-treatment period (8–12 weeks) of the clinical trial, subjects underwent serial evaluation of residual viral reservoirs using the assays described below as well as monitoring of plasma viral load and CD4+ cell counts. For purposes of this report, the number of baseline samples evaluated included all available samples from the screening period and the pre-infusion period and were as follows: plasma viral load: six values; CD4+ T-cell counts: four values; HIV co-culture: two values; blood proviral DNA: five values; rectal mucosa HIV RNA: two values and rectal mucosa HIV DNA: two values. No therapeutic intervention other than continuation of the stable HAART regimen occurred during the pre-treatment period.
This protocol, all amendments and informed consent forms were reviewed by the Institutional Review Boards at University of California, San Francisco (UCSF), University of California, Los Angeles (UCLA), Massachusetts General Hospital, Independent Review Consulting, Inc. Written informed consent was obtained from each subject at the time of initial screening and prior to initiation of any research activities.
Blood HIV DNA assay
Analysis was performed at Roche Molecular Systems (Alameda, California, USA) using cryopreserved PBMCs that were thawed, washed twice with Specimen Wash Solution and resuspended in extraction buffer containing proteinase K and the quantitation standard. Cells were then incubated at 60°C for 30 min and 100°C for 30 min. Amplifications were performed using the Amplicor HIV-1 Monitor® version 1.5 master mix that contains primer SK145 and SKCC1B. A DNA standard was co-amplified with each sample. Quantification was performed on microwell plates with colorimetric detection as currently used in the Amplicor HIV-1 Monitor® test. The level of HIV-1 proviral DNA in PBMCs was normalized to the amount of genomic DNA as determined by Hoechst dye. Assay sensitivity was 10 copies/106 cells .
Quantitative HIV-1 co-culture assay
The co-culture assay was a modification of a previously described assay . Briefly, the subject's PBMCs were isolated from 60 ml whole blood, CD8-depleted using goat anti-mouse IgG magnetic beads (Dynal, Lake Success, New York, USA) and anti-CD8 antibody (Coulter Immunotech, Hileah, Florida, USA), stimulated for 24 hours with plate-bound anti-CD3 (Ortho Diagnostics, Raritan, New Jersey, USA) and anti-CD28 (Coulter Immunotech), and then plated in two-fold limiting dilution in duplicate beginning at 5–20 × 106 cells/well in six-well plates. Cells were cultured for 4 weeks with the addition of 5–10 × 106 phytohemagglutinin-stimulated (Sigma, St. Louis, Missouri, USA), CD8-depleted pooled donor PBMC at days 1, 7, and 14. Supernatant from each well was then harvested and analyzed for HIV p24 by ELISA (Coulter Immunotech) and scored as positive or negative. Raw data were analyzed and expressed as the number of infectious units per million input cells (IUPM) using a statistical analysis Fortran program developed specifically to analyze the results of limiting dilution assay data . The assay ranged from 0.01 to 20 IUPM with a mean standard deviation equal to 0.27 log10.
Rectal mucosal HIV-1 quantitative assays
Flexible sigmoidoscopy was performed twice in the rectosigmoid area, 2 weeks apart, to obtain stable baseline values. At each procedure, six biopsies were obtained circumferentially at a standard level of 30 cm from the anal margin to avoid potential regional variation and to exclude confounding conditions such as inflammation secondary to infectious or traumatic proctitis. Biopsies were collected using large cup endoscopic biopsy forceps (Microvasive Radial Jaw no. 1589, outside diameter 3.3 mm; Boston, Massachusetts, USA) and immediately placed in liquid nitrogen. Each biopsy was approximately 2–3 mm3 (average 10–12 mg each) and was collected into 2 ml pre-labeled cryovials (Nalgene, Rochester, New York, USA). Biopsies were immediately frozen in liquid nitrogen, and stored at −80°C. In order to enhance the precision and reproducibility of these assays, mRNA isolation was performed independently on two separate biopsies at each time-point; DNA isolation was performed independently on two separate biopsies; two biopsies were kept in reserve as needed for repeats. For both HIV-RNA and DNA analyses, isolations and analysis were performed in duplicate at each time-point.
mRNA isolation and quantitation of HIV-1 RNA from rectal biopsy tissue
mRNA was isolated in duplicate from the frozen rectal biopsy tissue (each isolation from two biopsies) using the Quickprep Micro mRNA Purification kit (Amersham Pharmacia Biotech, Piscataway, New Jersey, USA). Briefly, the frozen tissue was placed under liquid nitrogen for several minutes, removed and immediately pulverized with a steel mortar and pestle (Fisher Scientific, Pittsburgh, Pennsylvania, USA) that had been pre-cooled on dry ice for at least 1 h. Using a sterile, disposable scalpel (Fisher Scientific), the resultant powdered tissue was transferred rapidly to a 50 ml centrifuge tube (Becton Dickinson, Franklin Lakes, New Jersey, USA) and placed in dry ice. The powdered tissue was dissolved by repeated pipetting into 400 μl of the extraction buffer provided, prior to the addition of 800 μl of the supplied elution buffer and transferred to a sterile 2 ml Eppendorf tube. The tissue extract was then cleared by centrifugation for 1 min at maximum speed and transferred to a fresh sterile tube prior to the addition of 1 ml of Oligo(dt)-Cellulose. After gentle mixing for 3 min, the Oligo(dt)-Cellulose complex was washed as per the kit instructions using the MicroPlex 24 Vacuum and eluted twice into a final volume of 400 μl. Finally, the mRNA was precipitated overnight following the addition of 10 μl glycogen, 40 μl K acetate and 95% ethanol. The resulting pellet was dissolved in diethyl pyrocarbonate (DEPC) water and the yield determined spectrophotometrically at 260 nm. An average biopsy was found to yield approximately 1 μg of mRNA.
HIV-1 RNA was quantified by reverse transcriptase (RT)-polymerase chain reaction (PCR) using the Thermo-stable RTth reverse transcriptase RNA PCR kit (PE Biosystems, Foster City, California, USA). Gene-specific reverse transcription of 100 ng of mRNA was accomplished using the HIV-1 LTR primer AA55 (100 ng/ml) (5′-CTGCTAGAGATTTT CCACACTGAC-3′) and PCR was performed with P-32 end-labeled HIV-1 LTR primer 667 (5′-GGCTAACTAGGGAACCCACTG-3′). RNA was then subjected to 30 cycles of amplification in a 4800 series Perkin Elmer thermal cycler with each cycle consisting of a 1 min denaturation step at 94°C followed by a 2 min annealing step at 65°. RNA standards (10–3000 copies), constructed by the in vitro transcription of cloned HIV-1 DNA, were included with each series of reactions. Standards were diluted in 100 ng of HIV-1 seronegative tissue RNA prepared by our extraction protocol. Radiolabeled PCR products were resolved on a 6% polyacrylamide gel followed by autoradiography on phosphorscreens (Amersham Biosciences, Piscataway, New Jersey, USA) and quantified with the use of IMAGEQUANT software (Amersham Biosciences). Pixel volumes were extrapolated to copy number based on the comparative values obtained from the linear portion of the generated standard curve. Copies are expressed per mg of tissue total RNA; sensitivity was 10 copies per PCR reaction.
DNA isolation and quantitation of HIV-1 DNA from rectal biopsy tissue
Endoscopic biopsies were acquired as described. DNA was isolated in duplicate (each isolation from two biopsies) pulverized using the mortar and pestle system outlined above using urea lysis buffer to extract nucleic acid from the powdered tissue. Briefly, the pulverized tissue was dissolved into 140 μl urea lysis buffer containing 70 μl DEPC water by repeated pipetting. Following a series of phenol-chloroform (× 3) and chloroform (× 2) extractions, DNA was precipitated overnight at −20°C in absolute ethanol (2.5 × volumes) with 5 mol/l NaCl (0.1 × volume). The resulting pellet was washed once with 70% ethanol, air-dried and dissolved in 50 μl of Rnase/Dnase free water. The yield was determined spectrophotometrically and an average biopsy was found to yield between 20 and 40 μg DNA.
HIV-1 DNA was quantified by PCR using the HIV-1 LTR primers AA55 and 667 described above. In general, 100 ng of tissue DNA, in the presence of 1.5 units of DNA Taq Polymerase and 0.5 mmol/l MgCl2, was subjected to 30 cycles of PCR under the conditions previously described. DNA standards (10–3000 copies) constructed of cloned HIV-1 DNA (Pykjrcsf) diluted in 100 ng of seronegative tissue DNA were run simultaneously. Samples (10–20 ng) of the tissue DNA sample were amplified using the β-globin primers LA1 (5′-ACACAACTGTGTTCACTAGC-3′) and LA2 (5′-CAACTTCATCCACGTTCACC-3′) and the results extrapolated against a standard curve constructed from genomic DNA quantified spectrophotometrically. HIV-1 DNA was normalized to 2 × 106 copies β-globin to permit comparison between biopsies and expressed as copies per 1 × 106 cells; the sensitivity was three copies per PCR reaction.
All baseline measurements for the blood and rectal HIV RNA and DNA assays were averaged to generate a mean baseline value for each assay for each patient. For the HIV co-culture data the raw limiting dilution data for both baseline measurements was pooled to calculate a single estimate for the baseline value using the limiting dilution Fortran analysis program developed by Dr. C. Macken . Due to the wide spread in the data, mean baseline values were transformed to log10 scale prior to performing the correlation analyses and the correlation coefficients presented are based on using the log-transformed data rather than the raw data. The rectal biopsy HIV RNA and DNA data required more steps to calculate each baseline value. For rectal biopsy HIV RNA, each patient had two biopsies, each run in duplicate at each time-point. Duplicates of each biopsy were averaged with subsequent averaging of the values from the two separate biopsies to obtain a single value for each visit. For rectal biopsy HIV DNA, two biopsies were analyzed in duplicate at each time-point and results were averaged as above to generate a single value for each time-point. In situations where data was reported as undetectable, a value of 50% of detection limit was assigned for summary analyses [for blood proviral DNA: 5 copies/mg DNA; for rectal HIV RNA: 5 copies/mg mRNA; for rectal HIV DNA: 5 copies/1 × 106 cells; for HIV co-culture: because of the analytical methodology used to calculate the point estimate for HIV co-culture, undetectable values were set at the limit of detection for each individual point estimate (i.e. a value reported as < 0.018 set to 0.018; a value of < 0.035 set to 0.035), as determined by the statistical limiting dilution Fortran analysis program developed by Dr. C. Macken . The SAS system was used to perform all analyses. Differences were analyzed parametrically using the analysis of variance and non-parametrically using the Wilcoxon test and the analysis-of-variance of ranks. Pearson correlation coefficients for numeric random variables and the probabilities associated with these statistics were calculated using the SAS procedure PROC CORR. No additional statistical adjustments were made for the multiple statistical assessments. Differences in viral burden as measured in each of the four reservoir assays between subjects with durable plasma HIV suppression and those with transient blips were analyzed by Wilcoxon rank sum.
The demographics and baseline characteristics for the 40 subjects evaluated for correlation of clinical features with assay results are presented in Table 1. The mean age of the 40 treated patients was 40.9 years old (range, 28–59 years). The mean time from original diagnosis of HIV to study entry was 7.0 years (range 1.2–17.9 years). The mean duration of taking any HIV medication (ART) was 3.6 years (range 0.6–10.9 years), and the mean time taking HAART medication was 1.6 years (range 0.4–3.0 years).
All patients had undetectable plasma viral load on at least two of three screening measurements taken 2 weeks apart (< 50 copies HIV-1 RNA/ml). Thirty-two of 40 subjects (80%) were undetectable on all baseline screening measurements performed over a 15 week period. Transient viral load blips were detected in eight subjects (53–148 copies/ml) over this same period (six measurements). In all cases plasma HIV RNA was detected at only one isolated time-point. Since patients were not followed with longitudinal viral load measurements prior to study entry the actual duration of undetectable viral loads is unknown. However, all subjects were required to be on a stable HAART regimen for at least 6 months prior to study entry suggesting sustained virologic control on these regimens. Mean CD4+ T-cell counts for all subjects was 422 × 106 cells/l (95% confidence interval, 375–469).
Subject compliance with study procedures was high. Specifically with the invasive flexible sigmoidoscopy and biopsies, 40 of 40 (100%) completed both baseline procedures (80 flexible sigmoidoscopies) and 37 of 40 (92.5%) subjects completed all five sigmoidoscopic procedures for the entire study [195 of 200 (97.5%) procedures completed]. No significant side effects or adverse events related to the procedure were reported.
HIV quantitative assays
Rectal tissue HIV RNA
Twenty-six out of 40 (65%) patients had detectable HIV RNA in at least one rectal biopsy at baseline. Approximately one-third of subjects were undetectable in this assay at both baseline measurements. In 39 subjects (95%), both biopsies were either positive or negative for HIV RNA. One subject (5%) showed mismatched results between the two biopsy procedures (< 10 and 136 copies). Absolute levels of HIV RNA were low in most subjects where HIV RNA was detected (range from 0.7 to 2.42 log10 copies mRNA/mg total mRNA) (Table 2).
Rectal tissue HIV DNA
Thirty-eight of 40 (95%) subjects had detectable levels of HIV DNA in at least one of their two rectal biopsy samples at baseline. One patient had detectable HIV DNA (55 copies) in only one of two biopsy procedures. The remaining 37 subjects had HIV DNA detected in both biopsies. The absolute levels of detection ranged from 0.7 to 3.47 log10 HIV DNA copies/1 × 106 cells.
Blood HIV DNA
The blood HIV DNA assay showed 38 of 40 (95%) patients had detectable HIV DNA in PBMC at baseline in at least one measurement during five visits. Two subjects were consistently undetectable in five independent measurements performed over a 12 week period. Five additional subjects (13%) demonstrated intermittent low level positive results (all < 25 copies). The range of values was 0.7–2.54 copies/mg DNA (Table 2).
All 40 subjects had at least one quantitative HIV co-culture assay performed. Thirty-five of 40 (88%) had two baseline measurements performed. Assays were aborted prematurely due to technical difficulties (e.g. contamination) in the other five cases. The results showed that 35 of 40 (88%) of patients had infectious HIV cultured from PBMC at baseline in at least one assay. Five subjects were undetectable in both baseline measurements and four additional subjects were undetectable in one of two assays (Table 2).
All 40 subjects had HIV detected in at least one of these four reservoir assays at baseline (Fig. 1). Two subjects (5%) were undetectable in all assays (viral load, HIV co-culture, blood HIV DNA, rectal tissue HIV RNA) except the rectal tissue HIV DNA assay. Two additional subjects had undetectable levels of rectal tissue DNA and RNA at both baseline measurements. However, both subjects had detectable HIV DNA in the blood as well as successful culture of infectious HIV from PBMC. The remaining 36 subjects (90%) had HIV detected in both blood and gut reservoirs.
The number of visits used to evaluate the variability of the assays were as follows: CD4+ T-cell counts, four time-points; HIV co-culture, two time-points; blood HIV DNA, five time-points; rectal biopsy HIV DNA, two time-points; rectal biopsy HIV RNA, two time-points. The results of assay variability analysis for the 40 study patients showed that blood HIV DNA data had the lowest coefficient of variation (CV) value (8.932) and was considered as the most reliable assay. Rectal biopsy HIV DNA, rectal biopsy HIV RNA and CD4+ T-cell count also had low CV values (11.423, 15.490 and 19.385, respectively). Among all the assays, the HIV co-culture assay had the highest CV value (−41.042); however, this assay had a mean standard deviation of 0.27 log which is comparable to plasma viral load assays currently in standard use.
Correlations among viral reservoir measurements using HIV quantitative assays
The baseline assays of viral burden (HIV co-culture, blood HIV DNA, rectal biopsy HIV DNA and RNA) were correlated to evaluate representativeness of measurement in one compartment compared with another (Table 3). The results showed that all measures of virologic reservoirs correlated with each other. HIV co-culture and blood HIV DNA had the highest correlation (r = 0.54, P = 0.0003). HIV co-culture was also highly correlated with rectal biopsy HIV DNA (r = 0.48, P = 0.0017) and rectal biopsy HIV RNA (r = 0.45, P = 0.0036). Blood HIV DNA correlated with rectal biopsy HIV DNA (r = 0.42, P = 0.007) and with rectal biopsy HIV RNA (r = 0.31, P = 0.05). Finally, rectal biopsy HIV DNA correlated with rectal biopsy HIV RNA (r = 0.31, P = 0.05) (Table 3, Fig. 2).
Correlation of viral reservoir measurements with patient's clinical characteristics
Table 3 also presents the correlation matrix of patient disease characteristics (years of HIV infection, months of taking HAART medications, and months of taking any HIV antiretroviral medications) with parameters measuring viral burden (HIV co-culture, blood HIV DNA, rectal biopsy HIV DNA and RNA) and immune function (CD4+ T-cell counts) for the 40 patients. The results show that the CD4+ T-cell count was not correlated with any measure of HIV reservoirs. A modest inverse correlation was seen between rectal biopsy HIV RNA and both years of HIV infection (r = −0.35, P = 0.03) and months of ART (r = −0.33, P = 0.04) (Table 3, Fig. 2). Otherwise, there were no significant correlations between clinical parameters of patient disease activity and measures of HIV reservoirs. Given the small sample size in this study, no attempt was made to make statistical adjustments for the variation in the number of data-points available for each assay or the multiple univariate correlations performed.
Patients were retrospectively classed into clinical subsets to determine whether significant correlations between descriptive characteristics and viral reservoirs would emerge. Assuming viral blips may define a cohort of subjects with less rigorous viral control, the 32 subjects with no history of viral blips in plasma viral load during all six baseline visits and the eight with blips were independently assessed for correlations between clinical parameters and laboratory assays. No significant correlations were found. Similarly, not all subjects had been treated with optimal ART prior to their 6-month pre-enrollment HAART regimens. Prior antiviral treatment regimens for all patients were reviewed and designated as optimal (n = 20) or sub-optimal (n = 20) by one of us (S.D.) and laboratory correlations between the two groups were performed. Again, no significant correlations between measurements of blood and tissue HIV burden and clinical parameters emerged.
In order to determine whether subjects with viral blips (n = 8) had greater viral burden in the various compartments or ‘reservoirs’ than subjects with durable and complete viral suppression (n = 32), an analysis was performed comparing levels of virus in each of the four reservoir assays in these two subgroups. There were no significant differences in levels of viral burden in any of the blood or gut reservoirs between those with a history of viral blips or those without.
Antiretroviral therapy prevents de novo infection and inhibits viral replication but does not eradicate long-lived cellular reservoirs of virus. Immune based therapies that are cytolytic and have the potential to eradicate virus from latent cellular reservoirs may be necessary for viral eradication. Reports have suggested that replication competent HIV can be isolated from the vast majority of patients with prolonged plasma virus suppression, confirming that residual HIV reservoirs remain [11,19,20]. Assessments of these reservoirs with reliable and reproducible methods with minimal variability are critical in providing measurable endpoints for adjunctive therapies aimed at further reducing total body viral burden.
In this assessment of steady-state measurements of viral reservoirs, the vast majority of subjects had quantifiable levels of HIV measured using the HIV co-culture (88%), blood HIV DNA (95%), and rectal biopsy HIV DNA (95%) assays. Furthermore, most patients (65%) had detectable levels of HIV RNA in rectal biopsies despite having undetectable levels of HIV RNA in plasma. The procedure for obtaining rectal biopsies is relatively painless, fast, has few complications and allows for frequent sampling . An important observation in this trial is the high degree of adherence to the invasive procedures performed (100% of 80 baseline and 97.5% of the entire study's 200 scheduled flexible sigmoidoscopies).
Eradication efforts need to address these mucosal tissue sites directly as such tissues may continue to foster low-level replication due to, among other factors, the inherent difference in susceptibility to infection of GALT T cells compared with PBMC [13,14], the increased activation status of the resident lymphocytes and the presumed, although not documented, lower concentrations of antiviral drugs in tissue than in plasma. This latter point is suggested by reports of isolated cases of slightly different HIV resistance panels in gut- and blood-derived virus, despite the majority of isolates having concordant profiles . Regardless of the pathogenic variables, the demonstration of persistent viral detection in the mucosal compartment of aviremic subjects coupled with the documentation by co-culture that the presence of PBMC-associated viral DNA sequences was significantly correlated with replicating virus, suggests that at least a fraction of the mucosal viral RNA and DNA reflects replication-competent HIV. To confirm the presence of replication-competent HIV in GALT in aviremic patients, mucosal HIV co-culture assays would be necessary. However, such assays are not available and would be technically challenging due to the large numbers of lymphocytes required.
However, one can use the acquired information to make rough estimates of the percentage of HIV DNA in PBMCs that reflects replication-competent versus non-replicative DNA by comparing the number of copies of HIV DNA detected in PBMC with the number of infectious units (cells) determined by the limiting dilution co-culture assay. With the assumption that 1 μg DNA reflects 150 000 cells, comparison of the number of PBMC with HIV DNA with the number of PBMCs with infectious virus demonstrates a two to three orders of magnitude greater number of HIV DNA-containing cells than infectious virus-containing cells. This suggests that in PBMC, ∼ 99.9% of detectable HIV DNA is of non-replicative potential (under the stimulating co-culture assay). Nevertheless, it also demonstrates that infectious potential remains. In viewing the gut environment, it is plausible to imagine low-level replication and maintenance of infection given the proximity to highly vulnerable cellular targets. Even without having tissue co-culture data to support this, the blood analysis suggests that there may be sufficient infectious potential of residual HIV in the gut to implicate GALT as an important reservoir of latent infection.
There was a significant correlation between the various measures of HIV burden (HIV co-culture, blood HIV DNA, rectal biopsy HIV RNA and DNA) suggesting that these are interrelated measures of total body HIV burden. However, there was no significant correlation between HIV reservoir assays and clinical parameters including duration of HIV infection, ART or HAART usage or number of CD4+ T lymphocytes. A curious exception to this was the isolated inverse correlation between HIV tissue RNA level and years of HIV infection although the relevance of this is unclear given the multiple statistical analyses performed. Given the small sample size in this study no attempt was made to make statistical adjustments for the variation in the number of data-points available for each assay or the multiple univariate correlations performed. However, given the multiple associations across all reservoir assays and the relative lack of association between reservoir assays and clinical parameters, it is unlikely that these observations are due to chance alone.
There was no significant correlation between tissue HIV RNA and years of any ART exposure. The lack of correlation between levels of HIV in residual blood and tissue reservoirs and patient clinical characteristics supports previous published data focused on quantitative HIV co-culture data only in a smaller cohort of patients . Our analysis extends this previous observation to include measures of both blood and gut mucosal reservoirs and suggests that HAART therapy alone does not lead to a reproducible decay in HIV reservoirs; other factors such as host immune response, viral fitness, size of reservoirs, tissue penetration etc. are likely to be important co-factors that need to be addressed in future efforts to reduce viral reservoirs [19,22–24].
Under optimal treatment conditions, approximately 90% of the HIV-infected population undergoing therapy with HAART can be rendered aviremic, at least for a period of time . Given that such therapy is now generally accepted to be incapable of eradicating HIV infection, new surrogate endpoints are needed to evaluate the next generation of HIV therapeutics designed to eliminate these residual reservoirs of virus. The assays employed here were evaluated with this in mind. A reliability analysis was carried out for each of the assays (HIV co-culture, blood HIV DNA, rectal biopsy HIV DNA, rectal biopsy HIV RNA) based on two or more baseline values to determine whether such assays are reliable with low variability and, therefore, potentially suitable for use as surrogate efficacy endpoints in trials in patients with undetectable plasma viremia. This analysis was based on the assumption that the multiple values taken over an 8–12 week period in this well-characterized population prior to initiation of a clinical trial represented steady-state levels of viral burden. All assays studied had comparable or less variability than that seen in currently used plasma viral load assays. HIV co-culture had the highest coefficient of variability whereas the blood HIV DNA assay had the lowest coefficient of variation and was considered to be the most reliable assay. The variability in the HIV co-culture assay was similar to that of current plasma viral load assays in clinical use.
The inherent virologic failure rate in subjects who initially achieve an undetectable plasma viral load on HAART and are maintained on optimal regimens is estimated at 4% over 16 weeks. This failure rate may be higher in those with prior history of suboptimal regimens, those with higher pre-treatment viral loads, or longer time to initial viral suppression . Patients successfully treated with HAART may have transient ‘blips’ in viral load (defined as transient increase in plasma HIV RNA to > 50 copies/ml at isolated time-points). During the pre-treatment period for this clinical trial, eight subjects experienced viral load ‘blips'. Such subjects might reflect a different immunologic and virologic cohort. However, when a subset analysis was performed on these two virologic cohorts (persistently undetectable and transient viral blips), no new correlations between clinical parameters and viral reservoirs emerged. Similarly, when we retrospectively assessed the ART history of the subjects prior to the study's 6 month pre-screening period as to whether it was ‘optimal’ or ‘sub-optimal’ and then evaluated if either sub-group's clinical parameters correlated with their viral reservoirs, no new patterns emerged. Interpretations of these results are stronger in that there were 20 subjects in each group and suggest that the extent of antiviral coverage does not reflect differently in viral reservoirs. In addition, there was no evidence of increased viral burden in blood and gut compartments in those with periodic plasma viral blips. Taken together, whether the antiviral regimen was fully suppressive or of optimal design did not yield any clinically relevant correlates with viral activity and reservoir reserve.
Plasma viral load may not provide information regarding tissue HIV reservoirs nor residual viral activity in patients responding well to HAART. Thus, there is a need for additional efficacy endpoints for trials assessing the impact of antiviral regimens in subjects with undetectable viremia on HAART. As plasma viral load proved to be an important surrogate of clinical activity of antiviral regimens in those with detectable plasma viral burden, once levels fall below detection in plasma, sampling from those tissue sites shown to have persistent viral presence will need to be assessed. Monitoring of rectosigmoid tissue viral burden may prove to be a useful biologic surrogate endpoint with low assay and intra-subject variability and excellent patient compliance  for use in future intervention trials in aviremic subjects. The procedure is minimally invasive and may identify changing viral and immune parameters that may be unappreciated when sampling blood alone, as exemplified at the conclusion of this clinical trial in which reductions in gut HIV DNA, but not blood HIV-DNA were demonstrated following infusion of gene-modified HIV targeted T cells (results reports elsewhere) . Blood HIV DNA measurements may also be a useful parameter to follow and demonstrated the greatest sensitivity and reliability in this study but again, may not accurately reflect the tissue dynamics during treatment . Although scientifically interesting and the only assay to directly measure the replication capacity of residual HIV, quantitative HIV co-culture assays are laborious and expensive and, therefore, not feasible for large clinical studies.
The data presented here support the feasibility of the described assays in quantifying HIV in various compartments and demonstrate their inter-relatedness in subjects with undetectable plasma HIV. The rectal biopsy HIV RNA assay had the lowest sensitivity, HIV co-culture assay the highest variability and blood HIV DNA the lowest variability. While the assays did correlate with each other, the degree of viral burden in these different compartments did not correlate with clinical parameters. Therapeutic strategies aimed at eradicating HIV in both early-stage disease and in those with undetectable plasma viral load will need to be active in mucosal lymphoid tissue where the majority of HIV replication is taking place. These assays can be used to follow the status of HIV infection in different tissues but shed no illumination on distinguishing clinical factors contributing to the size of the residual viral reservoir in fully suppressed patients. The size of these reservoirs is not clearly correlated with historical clinical descriptors. Whether they emerge as predictors of future clinical course remains to be seen.
We would like to acknowledge the detailed laboratory contributions of Julie Elliott (UCLA Center for HIV and Digestive Diseases).
Sponsorship: This work was supported by Cell Genesys, Inc. in collaboration with Hoechst Marion Roussel. P.A. was supported in part by Mucosal Immunology Core, UCLA CFAR NIH AI28697, Macy's Foundation and NIH AI01610. This work was carried out in part in the General Clinical Research Centers at the following institutions: UCLA (RR-00865), SFGH (5-MO1-RR00083-37) and UCHSC (RR-00051).
1.Gulick RM, Mellors JW, Havlir D, Eron JJ, Gonzales C, McMahon D, et al
. Treatment with indinavir, zidovudine, and lamivudine in adults with human immunodeficiency virus infection and prior antiretroviral therapy. N Engl J Med
2.Collier AC, Coombs RW, Schoenfeld DA, Bassett RL, Timpone J, Baruch A, et al
. Treatment of human immunodeficiency virus infection with saquinavir, zidovudine, and zalcitabine. N Engl J Med
3.Hammer SM, Squires KE, Hughes MD, Grimes JM, Demeter LM, Currier JS, et al
. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. N Engl J Med
4.Embretson J, Zupancic M, Ribas JL, Burke A, Racz P, Tenner-Racz K, et al
. Massive covert infection of helper T lymphocytes and macrophages by HIV during the incubation period of AIDS. Nature
5.Pantaleo G, Graziosi C, Demarest JF, Butini L, Montroni M, Fox CH, et al
. HIV infection is active and progressive in lymphoid tissue during the clinically latent stage of disease. Nature
6.Wong JK, Hezareh M, Günthard HF, Havlir DV, Ignacio CC, Spina CA, et al
. Recovery of replication-competent HIV despite prolonged suppression of plasma viremia. Science
7.Finzi D, Hermankova M, Pierson T, Carruth LM, Buck C, Chaisson RE, et al
. Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy. Science
8.Chun T-W, Stuyver L, Mizell SB, Ehler LA, Mican JAM, Baseler M, et al
. Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy. Proc Natl Acad Sci USA
9.Faust RA, Henry K, Dailey P, Melroe H, Sullivan C, Erice A, et al
. Outpatient biopsies of the palatine tonsil: access to lymphoid tissue for assessment of human immunodeficiency virus RNA titers
. Otolaringol-Head Neck Surg
10.Haase AT, Henry K, Zupancic M, Sedgewick G, Faust RA, Melroe H, et al
. Quantitative image analysis of HIV-1 infection in lymphoid tissue. Science
11.Chun T-W, Carruth LM, Finzi D, Shen X, DiGiuseppe JA, Taylor H, et al
. Quantification of latent tissue reservoirs and total body viral load in HIV-1 infection. Nature
12.Anton PA, Poles MA, Elliot J, Mao SH, McGowan I, Lenz H-J, et al
. Sensitive and reproducible quantitation of mucosal HIV-1 RNA and DNA viral burden in patients with detectable and undetectable plasma viral HIV-1 RNA using endoscopic biopsies. J Virol Methods
13.Anton PA, Elliot J, Poles MA, McGowan IM, Matud J, Hultin LE, et al
. Enhanced levels of functional HIV-1 co-receptors on human mucosal T cells demonstrated using intestinal biopsy tissue. AIDS
14.Lapenta C, Boirivant M, Marini M, Santini SM, Logozzi M, Viora M, et al
. Human intestinal lamina propria lymphocytes are naturally permissive to HIV-1 infection. Eur J Immunol
15.Poles MA, Elliot J, Vingerhoets J, Michiels L, Scholliers A, Bloor S, et al
. Despite high concordance, distinct mutational and phenotypic drug resistance profiles in human immunodeficiency virus type 1 RNA are observed in gastrointestinal mucosal biopsy specimens and peripheral blood mononuclear cells compared with plasma. J Infect Dis
16.Deeks SG, Wagner B, Anton PA, Mitsuyasu RT, Scadden DT, Huang C, et al
. A phase II randomized study of adoptive immunotherapy with HIV-specific gene-modified vs. unmodified T-cells in HIV infected adults with undetectable plasma viremia on anti-retroviral therapy. Mol Ther
17.Christopherson C, Kidane Y, Conway B, Krowka J, Sheppard H, Kwok S. PCR-based assay to quantify human immunodeficiency virus type 1 DNA in peripheral blood mononuclear cells. J Clin Microbiol
18.Macken C. Design and analysis of serial limiting dilution assays with small sample sizes. J Immunol Methods
19.Ramratnam B, Mittler JE, Zhang L, Boden D, Hurley A, Fang F, et al
. The decay of the latent reservoir of replication-competent HIV-1 is inversely correlated with the extent of residual viral replication during prolonged anti-retroviral therapy. Nat Med
20.Chun T, Finzi D, Margolick J, Chadwick K, Schwartz D, Siliciano RF. In vivo fate of HIV-infected T cells: quantitative analysis of the transition to stable latency. Nat Med
21.Cranston RD, Anton PA, McGowan IM. Gastrointestinal mucosal biopsy in HIV disease and AIDS. Gastrointest Endosc Clin N Am
22.Chun T-W, Fauci AS. Latent reservoirs of HIV: obstacles to the eradication of virus. Proc Natl Acad Sci USA
23.Finzi D, Blankson J, Siliciano JD, Margolick J, Chadwick K, Pierson T, et al
. Latent infection of CD4+ T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy. Nat Med
24.Wong JK, Günthard HF, Havlir DV, Zhang ZQ, Haase AT, Ignacio CC, et al
. Reduction of HIV-1 in blood and lymph nodes following potent antiretroviral therapy and the virologic correlates of treatment failure. Proc Natl Acad Sci USA
25.Havlir DV, Marschner IC, Hirsch MS, Collier AC, Tebas P, Bassett RL, et al
. Maintenance antiretroviral therapies in HIV-infected subjects with undetectable plasma HIV RNA after triple-drug therapy. N Engl J Med
Keywords:© 2003 Lippincott Williams & Wilkins, Inc.
HIV-1; quantitative; reservoirs; mucosa; tissue; gastrointestinal-associated lymphoid tissue; viral load; clinical trials; assay correlation; assay variability