Antiviral effects of autologous CD4 T cells genetically modified with a conditionally replicating lentiviral vector expressing long antisense to HIV

Abstract

We report the safety and tolerability of 87 infusions of lentiviral vector–modified autologous CD4 T cells (VRX496-T; trade name, Lexgenleucel-T) in 17 HIV patients with well-controlled viremia. Antiviral effects were studied during analytic treatment interruption in a subset of 13 patients. VRX496-T was associated with a decrease in viral load set points in 6 of 8 subjects (P = .08). In addition, A → G transitions were enriched in HIV sequences after infusion, which is consistent with a model in which transduced CD4 T cells exert antisense-mediated genetic pressure on HIV during infection. Engraftment of vector-modified CD4 T cells was measured in gut-associated lymphoid tissue and was correlated with engraftment in blood. The engraftment half-life in the blood was approximately 5 weeks, with stable persistence in some patients for up to 5 years. Conditional replication of VRX496 was detected periodically through 1 year after infusion. No evidence of clonal selection of lentiviral vector–transduced T cells or integration enrichment near oncogenes was detected. This is the first demonstration that gene-modified cells can exert genetic pressure on HIV. We conclude that gene-modified T cells have the potential to decrease the fitness of HIV-1 and conditionally replicative lentiviral vectors have a promising safety profile in T cells.

Trial registration: ClinicalTrials.gov NCT00295477.

Figure 1

Schematic of clinical trial design. Red arrows indicate infusions of 10 billion T cells each. Blue arrows indicate time points of rectal biopsies. The second leukapheresis (pink arrows) was taken to provide baseline samples and also to serve as a backup if the first manufacture run failed. After initial data indicated that the second cycle of infusions was not increasing VRX496-T persistence, it was removed for later patients. (A) Original study design. (B) Modified study design after removal of second infusion cycle .

Figure 2

Effects of VRX496-T on viral load and CD4 counts. (A) Average CD4 counts in patients after infusion and before ATI, with 252 plotted separately as an outlier. (B) CD4 counts in patients after resuming ARVafter ATI. No difference was observed between patients never on ATI and those who resumed ARV after ATI (P = .2345). (C) Virus recrudescence after ATI. (D) Viral set point analysis. Shown are the average of historic set point readings (♢) and the after ATI viral load readings averaged at week 10 and 14 after ATI (■). Patient 252 ultimately did have a measurable viral load, but it was undetectable at weeks 10 and 14 when the post-ATI set point was assessed.

Figure 3

Conditional replication of VRX496 in vivo. Shown are all patients who had detectable conditionally replicating VRX496, which is measured by RT-PCR for vector RNA in patient plasma. Only time points at which VRX496 RNA was detected are shown. Closed symbols (black and gray) are patients who were evaluable for viral load changes.

Figure 4

Viral evolution associated with VRX496 antisense. (A) The HIV genome showing the antisense-targeted region and the positions of amplicons used for deep-sequencing analysis. Note that each of the 2 amplicons query both antisense-targeted bases and adjacent nontarget bases. (B) P values comparing the enrichment of each type of base substitution in the antisense target region in VRX496-exposed patients with control ATI patients. For each patient, the proportion of reads with at least 5% of the indicated substitution (the starting base is shown to the left, the substituted base on top) was scored. Mean values of these proportions were compared between the VRX496-exposed and control ATI groups using the Mann-Whitney test. (C) P values comparing VRX496-exposed and ATI control patients using measurements derived from the difference in rates of each base substitution between the antisense-targeted region and the adjacent antisense-nontargeted region within each patient. P values were determined using the Mann-Whitney test.

Figure 5

Persistence of VRX496-T in patients receiving 6 and 3 doses of cells in blood and gut. Individual patients are shown because of a high level of variability in persistence levels between patients. The limit of quantification (LOQ) of the PCR assay was 200 copies (red dotted line) and the limit of detection was 100 copies (blue dotted line). Incidences where vector was detected below the LOQ but above the limit of detection are plotted at 100. Graphs show persistence in patients receiving 6 (A) and 3 (B) doses. Note that patient 214 was enrolled on the 6-dose protocol but only received 3 infusions because of a quality control test delay and so is represented in panel B. (C) Long-term persistence of VRX496-modified cells in patients with engraftment beyond 1 year. Average level in blood is indicated by the closed circles and black line, with individual data points indicated. Only patients with persisting cells at 1 year and beyond are shown. (D) Levels of VRX496 in mononuclear cells in the gut versus the peripheral blood. The correlation is highly significant at P = .000017 with an R² of 0.667. Samples below the LOQ were defined as 10 for the visual purposes of this graph; changing the value of the number assigned to samples below the LOQ to 1 was evaluated and had no impact on the level of significance of the correlation.

Figure 6

Integration site distributions from pre- and postinfusion time points in or near transcription units, highly expressed genes, and cancer-associated genes. (A) Proportion of integration sites found within genes as defined by the RefSeq database normalized to the proportion of matched random control sites within genes. The line at 1 marks the expected frequency for a random distribution; levels above 1 indicate associations enriched or favored compared with random; below 1, disfavored. Ex vivo indicates integration sites from patient cells transduced ex vivo, before infusion back into the patient; Jurkat, sites from Jurkat cells infected in vitro; postinfusion, sites isolated from CD4 T cells taken postinfusion (supplemental Table 8). Ex vivo and postinfusion datasets represent pooled sites from all 5 patients and time points. (B) Proportion of sites in highly expressed genes. Genes were binned into 7 equal bins of increasing gene expression as measured by the Affymetrix HU133 plus Version 2.0 gene chip array. Genomic intervals of 1 Mb were annotated for the numbers of highly expressed genes, with low expression on the left and high expression on the right, and then the proportions of integration sites in these bins were plotted. The line at 1 shows the expectation for a random distribution. “Ex vivo” and “postinfusion” represent pooled data from all 5 patients. (C) Proportion of sites near cancer-associated genes. The plot shows the proportion of sites < 50 kb from a RefSeq gene that were also < 50 kb from a cancer-associated gene. No patient showed significant enrichment of integration sites near cancer-associated genes (Fisher exact test, 1-sided in the direction of enrichment postinfusion found no pairs significantly different).