Delaying a Treatment Switch in Antiretroviral-Treated HIV Type 1-Infected Patients with Detectable Drug-Resistant Viremia Does Not Have a Profound Effect on Immune Parameters: AIDS Clinical Trials Group Study A5115

Abstract

Some patients are unable to achieve and maintain an undetectable plasma HIV-1 RNA level with combination antiretroviral therapy (ART) and are therefore maintained on a partially suppressive regimen. To determine the immune consequences of continuing ART despite persistent viremia, we randomized 47 ART-treated individuals with low to moderate plasma HIV-1 RNA levels (200–9999 copies/ml) to either an immediate switch in therapy or a delayed switch (when plasma HIV-1 RNA became ≥10,000 copies/ml). After 48 weeks of follow-up, naive and memory CD4⁺ T cell percents were comparable in the two groups. The proportion of subjects with a lymphocyte proliferative response to Candida, Mycobacterium avium- intracellulare complex, or HIV-gag was also not significantly different at week 48. Delaying a treatment switch in patients with partial virologic suppression and stable CD4⁺ T cells does not have profound effects on immune parameters.

The immediate goal of antiretroviral therapy (ART) is to reduce plasma HIV-1 RNA levels to below detectable limits. Failure to achieve this goal is associated with the gradual accumulation of drug-resistance mutations, which can compromise future drug options. Some patients, however, are unable to achieve this goal, and are maintained on a partially suppressive regimen. Moreover, the “switch” criteria in many areas of the world is a plasma HIV-1 RNA >10,000 copies/ml.¹ Although this approach is associated with the accumulation of drug-resistance mutations,² it is associated with much slower rates of immunologic and clinical progression, at least as compared to situations in which all treatment is interrupted.^3,4

The immune consequences of a delayed switch strategy are uncertain. Most of the data generated to date is retrospective, and therefore subject to survival bias (i.e., those who are better able to maintain peripheral CD4⁺ T cell counts despite viral replication are more likely to be maintained on a stable regimen). Also, the measurement of peripheral CD4⁺ T cell counts is an imperfect surrogate for immune function. More comprehensive assessments are now available, but these measurements generally need to be performed in the context of prospective studies because of the need for real-time processing.

AIDS Clinical Trials Group (ACTG) Protocol 5115 was a randomized pilot study that evaluated the optimal time to switch treatment in ART-treated patients with plasma HIV-1 RNA of greater than 200 copies/ml but less than 10,000 copies/ml for at least 52 weeks.⁵ Subjects were randomized to a switch threshold of either ≥200 copies/ml (immediate switch) or ≥10,000 copies/ml (delayed switch). Subjects in the delayed switch arm were also encouraged to modify therapy if their CD4⁺ T cells declined by greater than 20% from baseline. We previously reported that 50% of subjects in the immediate switch arm had plasma HIV-1 RNA <50 copies/ml at 48 weeks.⁵ Ten of 23 subjects in the delayed switch arm met the switch criteria, and 7 had plasma HIV-1 RNA <50 copies/ml at 48 weeks. Subjects in both arms had similar outcomes at 48 weeks postrandomization in terms of drug resistance accumulation, peripheral CD4⁺ T cell numbers, and frequency of activated (CD38⁺) CD8⁺ T cells.⁵

While CD4⁺ T cells and CD38⁺CD8⁺ T cell frequency are validated surrogates for clinical outcomes in HIV infection,^6–8 these measures do not provide a complete picture of immune function. CD4⁺ T cells are made up of subsets that are differentially affected by HIV and have variable impact on disease progression. Naive and memory CD4⁺ T cell frequency can predict clinical progression and CD4⁺ T cell change in treatment-naive individuals^9,10 and after the initiation of ART.^11,12 Impairments in lymphocyte proliferative response to antigens are also associated with disease progression^13,14 and are not readily restored with ART.^15,16 ACTG 5115 provided an opportunity to determine whether these immune parameters are affected by a delay in treatment switch in patients with partial virologic suppression. We also explored the association between these parameters and study outcomes.

The study population and design of ACTG 5115 have been reported elsewhere.⁵ Peripheral blood mononuclear cells (PBMCs) were obtained from subjects and isolated by density centrifugation at baseline and every 16 weeks through week 48. Four-color flow cytometry was done using antibodies for CD3, CD4, CD45RA, CD45RO, and CD62L that were conjugated to fluorescein isothiocyanate, phycoerythrin, peridinin chlorophyll-protein, or allophycocyanin (Becton Dickinson-Pharmingen, San Jose, CA). The naive CD4⁺ T cell subset was defined as the percent of CD3⁺CD4⁺ T cells that was CD45RA⁺CD62L⁺, while the memory T cell subset was defined as the percent that was CD45RA^–CD45RO⁺. Lymphocyte proliferation in response to pokeweed mitogen (0.1μg/ml; Sigma Chemical, St. Louis, MO), Candida CASTA antigen (10 μg/ml:, Greer Laboratory, Lenoir, NC), Mycobacterium avium-intracellulare complex (MAC) antigen (1:10 dilution; BioWhitaker Laboratory, Walkersville, MD), and HIV p24 antigen/control (5 μg/ml; Protein Sciences, Meriden, CT) was done on fresh PBMCs as described previously.¹⁷ Results were expressed as a stimulation index (SI), defined as the ratio of the median counts per minute of the wells with antigen to the median counts per minute of the wells without antigen. An SI ≥ 3 was considered a positive response.

Distributions of immunology measures at weeks 0 and 48 between binary groups were compared using the Wilcoxon rank sum test. The p-values reported did not adjust for multiple comparisons. Univariate and multivariate linear or Cox regression models were used to explore baseline immunologic risk factors for various study outcomes.

Of the 45 subjects (22 from the immediate switch arm and 23 from the delayed switch arm) that had baseline immunology data, only 36 subjects (17 from the immediate switch arm and 19 from the delayed switch arm) had available week 48 advanced flow data and only 35 subjects (18 from the immediate switch arm and 17 from the delayed switch arm) had available week 48 pathogen-specific response data. The 36 subjects with available naive and memory CD4⁺ T cell frequency were statistically comparable to the 9 subjects with incomplete samples in terms of baseline plasma HIV-1 RNA, CD4⁺ T cells, %CD38⁺CD8⁺ T cells, naive and memory CD4⁺ T cells frequencies, intravenous drug use, sex, race, number of prior ART failure, and pathogen-specific response. However, excluded subjects had marginally higher %CD38⁺HLA-DR⁺CD8⁺ T cells (median 47% for excluded subjects vs. 33.5% for included subjects, p = 0.04). The 35 subjects with available pathogen-specific response measures were statistically comparable to the 10 subjects with incomplete samples in terms of the factors mentioned above, except that the excluded subjects had higher %CD38⁺HLA-DR⁺CD8⁺ T cells (median 48% for excluded subjects vs. 33% for included subjects, p = 0.02) and a higher proportion of female subjects (5/10 for excluded subjects vs. 5/35 for included subjects, p = 0.03).

Naive and memory CD4⁺ T cell percentages were not significantly different between subjects in the two treatment arms at either baseline or week 48. Median (Q1–Q3) naive T cell percents in the immediate switch arm at baseline and week 48 were 34.5 (24–34)% and 33.5 (28–44)%, respectively, while median (Q1–Q3) naive T cell percents in the delayed switch arm at baseline and week 48 were 31.0 (27–43)% and 31.0 (21–35)%, respectively. Median (Q1–Q3) memory CD4⁺ T cell percents in the immediate switch arm at baseline and week 48 were 55.5 (51–64)% and 62.5 (55–71)%, respectively, while median (Q1–Q3) memory CD4⁺ T cell percents in the delayed switch arm at baseline and week 48 were 66.0 (55–74)% and 66.0 (61–74)%, respectively. The median (Q1–Q3) changes in naive CD4⁺ T cell percents over 48 weeks were 1.0 (−7.0–4.0)% in the immediate switch arm and −1.5 (−6.0–0.5)% in the delayed switch arm. The median (Q1–Q3) changes in memory CD4⁺ T cell percents over 48 weeks were 0.0 (−5.0–4.0)% in the immediate switch arm and −1.5 (−4.5–0.5)% in the delayed switch arm. No statistically significant difference was detected between the arms.

At baseline, while the proportion of subjects with a positive or negative lymphocyte proliferative response to Candida or HIV-gag was similar between the two arms, there was a trend for more subjects in the delayed switch arm to have a positive response to MAC (p = 0.07) (Table 1).By week 48, the proportion of subjects that had a switch from a negative to positive MAC response was higher in the immediate switch arm (8 of 18, 44%) than in the delayed switch arm (2 of 17, 12%) (p = 0.03). While treatment assignment may account for the difference in the proportion of subjects that had a change in MAC-specific response, the higher proportion of MAC responders in the delayed switch arm at baseline may have made it less likely to demonstrate an increase in MAC response in this arm. There were no consistent differences between the groups in terms of the change in response to either Candida or HIV-gag from baseline to 48 weeks (Table 1). At week 48, the proportion of subjects with a positive or negative response to Candida, MAC or HIV-gag was similar between the two arms.

Table 1.

Lymphocyte Proliferative Stimulation Indices (SI) to Candida, MAC, and HIV-gag^a

Antigen	Week		Immediate switch	Delayed switch	p
Candida	0	No. (%) with SI > 3	13 (59.1%)	16 (69.6%)	0.54
	48	No. (%) with SI > 3	11 (57.9%)	13 (76.5%)	0.30
	Change from week 0 to 48b	No. with increase (%)	3 (16.7%)	3 (17.7%)	0.67
		No. with decrease (%)	3 (16.7%)	1 (5.9%)
		No. unchanged (%)	12 (66.7%)	13 (76.5%)
MAC	0	No. (%) with SI > 3	5 (22.7%)	12 (52.2%)	0.07
	48	No. (%) with SI > 3	9 (47.4%)	11 (64.7%)	0.34
	Change from week 0 to 48b	No. with increase (%)	8 (44.4%)	2 (11.8%)	0.03
		No. with decrease (%)	3 (16.7%)	1 (5.9%)
		No. unchanged (%)	7 (38.9%)	14 (82.4%)
HIV-gag	0	No. (%) with SI > 3	2 (9.1%)	1 (4.4%)	0.61
	48	No. (%) with SI > 3	4 (21.1%)	1 (5.9%)	0.34
	Change from week 0 to 48b	No. with increase (%)	3 (16.7%)	1 (5.9%)	0.79
		No. with decrease (%)	1 (5.6%)	1 (5.9%)
		No. unchanged (%)	14 (77.8%)	15 (88.2%)

^aA positive response was defined as SI > 3. An increased response was defined as a change from a negative to a positive response, while a decreased response was defined as a change from a positive to a negative response. Distributions of SI between the immediate and delayed switch arms were compared using the Wilcoxon rank sum test. The exact Chi-square test was used to compare the proportions with an increased, decreased, or unchanged lymphocyte proliferative response between the two arms. Reported p values are two-sided.

^bThe comparisons were not adjusted for multiple comparisons.

We next explored the relationship between baseline immune parameters (%CD38⁺HLA-DR⁺CD8⁺ T cells, %naive CD4⁺ T cells, %memory CD4⁺ T cells, and pathogen-specific immunity) and the following study endpoints: the occurrence of CDC class B or C events, the change in plasma HIV-1 RNA, and the change in CD4⁺ T cells, and found no consistent associations in univariate regression models. There was a trend, however, suggesting that a higher naive CD4⁺ T cell percent was associated with longer time to virologic failure after a treatment switch (p = 0.06).

A surprising finding in ACTG 5115 was that those who switched immediately had lower CD4⁺ T cells 16 weeks after switching compared to those who delayed their switches.⁵ At 16 weeks postswitch, CD4⁺ T cells in subjects in the immediate switch arm decreased by 31 cells/mm³ while they increased by 75 cells/mm³ in subjects in the delayed arm (p = 0.03). To investigate the determinants of CD4⁺ T cell change postswitch, we included the following factors in univariate and multiple linear regression models: treatment arm assignment, plasma HIV-1 RNA, CD4⁺ T cells, proportion of naive/memory CD4⁺ T cells, %CD38⁺HLA-DR⁺CD8⁺ T cells, future drug options index,⁵ HIV-specific lymphocyte proliferative response, and the change in HIV-1 RNA at 16 weeks postswitch. The diversity of drug regimens used both at baseline and at the time of switch⁵ did not allow for a meaningful analysis of a drug regimen's effect on outcomes. In the univariate regression models, we found significant associations with treatment arm assignment (p = 0.03), %CD38⁺HLA-DR⁺CD8⁺ T cells (p = 0.05), and HIV gag-specific lymphocyte proliferative response (p = 0.01). If all these three variables were simultaneously included in a multiple regression model, all remained significant [treatment arm assignment (p = 0.004), %CD38⁺HLA-DR⁺CD8⁺ T cells (p = 0.01), and HIV gag-specific lymphocyte proliferative response (p = 0.05)]. Linear regression analysis showed that higher percent naive CD4⁺ T cells and lower CD8⁺ T cell activation at the time of the switch resulted in higher CD4⁺ T cell counts at 16 weeks postswitch.

ACTG 5115 was a pilot randomized prospective study that explored the consequences of delaying a treatment switch in ART-treated patients with persistently detectable viremia under 10,000 copies of HIV-1 RNA/ml. This is an important issue for treatment-experienced patients who may not have access to a regimen that is able to fully suppress HIV replication. The impact of persistent low-to moderate-level viremia on immune function is of particular interest to individuals being treated in resource-poor regions, where newer antiretroviral agents are often not available and the switch criteria may be a plasma HIV-1 RNA >10,000 copies/ml.¹ The current analysis reinforces findings from the primary study in that delaying a treatment switch in patients with plasma HIV-1 RNA levels below 10,000 copies/ml was not associated with significant immunologic harm, at least as defined by changes in naive and memory CD4⁺ T cell frequency, or Candida- and HIV-specific lymphocyte proliferative responses. However, this observation occurred in a setting in which only 50% of subjects in the immediate switch arm achieved virologic suppression. Contemporary regimens that include drugs that were not available at the time of the study can achieve higher rates of virologic suppression in treatment-experienced patients with drug-resistant viremia,¹⁸ and may affect the extent of immune reconstitution in these patients. Nevertheless, these observations are consistent with other studies that show that clinical disease progression among treated individuals with low to moderate levels of drug-resistant viremia is relatively slow.^19–22

In summary, while delaying a treatment switch in the context of partial virologic suppression is associated with the accumulation of drug resistance mutations, it does not have a profound impact on immune parameters in treatment-experienced patients, at least during 48 weeks of observation. Although our sample size is limited, the results of this prospective randomized study are consistent with clinical data from larger cohorts, and argue that in the absence of a fully suppressive regimen, strategies aimed at maintaining viral loads below 10,000 copies RNA/ml should be considered.

Acknowledgments

This work was funded by the following grants from the National Institutes of Health: AI68636 (ACTG); U01-AI069471 (A.R.T.); U01-AI068634 and U01-AI38855 (H.J. and Y.Z.); U01-AI069472 and K24-RR16482 (D.R.K.); U01 AI069484, U01-AI062563, P30-AI64518, and K24-AI0744 (J.A.B.); AI068636 (A.L.L.); and U01-AI069494 (S.A.R.). Other A5115 team members included Hairong Huang (Harvard School of Public Health, Boston, MA) and Edward P. Acosta (University of Alabama, Birmingham). The protocol team is grateful to the individuals who volunteered to participate in this study. The team is also grateful to the following for their contribution to the study's conduct: Rush University Immunology Support Laboratory: Betty Donoval and A5115 Sites/Site Personnel: Lee McClurkin, RN, and Janet Mueller, BS MT (Duke University); Karen Tashima, MD, and Helen Sousa, LPN (Miriam Hospital); Margaret Travis, RN (Rush University Medical Center); Sylvia Stoudt, RN, and Pat Cain, RN (Stanford University); Beverly Putnam, RN ANP, and M. Graham Ray, RN MSN (University of Colorado Health Sciences Center); Lorna Nagamine, RN, and Debra Ogata-Arakaki, RN (University of Hawaii); Jose G Castro, MD, and Margaret A Fischl, MD (University of Miami); Joseph J. Eron, MD, and David Currin RN (University of North Carolina); Nancy Mantz, MSN CRNP (University of Pittsburgh); William A. O'Brien, MD, MS, and Cheryl Mogridge, ACRN (University of Texas, Galveston); Mamta Jain, MD, and Todd Morgan, RS (University of Texas, Southwestern Medical Center); David Haas, MD, Jie Wang, RN, and Michael Morgan, FNP (Vanderbilt University).

Disclosure Statement

No competing financial interests exist.