Antiretroviral therapy

Graeme Moyle MD, for NATAP

The antiretroviral therapy sessions at ICAAC revealed a mixed bag of new information, some intriguing some enlightening and much really just an update on already available data.

Themes covered included immune recovery, treatment interruptions and new drugs as well as new information on established therapies.

Immune recovery during therapy and treatment interruption

It is often assumed that the immune system recovers similarly during viral suppression regardless of the contents of the regimen used. Some preliminary data, presented previously has hinted that differences in CD4 T cell function, however, may exist between PI and NNRTI-based regimens favouring NNRTIs. Additionally, despite data supporting undiminished efficacy at least for efavirenz in high viral load, low CD4 T cell patients some physicians prefer to use a boosted PI in these circumstances.

Of note, the majority of reports of immune reconstitution disease have arisen from individuals initiated on a PI, although this may simply relate to the timing of availability of PIs, when more individuals with severe immunodeficiency were around. A case control study reported from Spain on 100 individuals initiating HAART with either one PI (IDV in 82%) or one NNRTI (NVP in 80%) plus two nucleoside analogues. [1] Patients were matched for baseline CD4+ counts <500 cells/mm3, self reported adherence, and viral load <50 copies/ml after 3 months on therapy, sustained for one year. It is unclear why the investigators did not match patients for prior dual or monotherapy use. This resulted in more fully treatment naive subjects receiving an NNRTI rather than PI (76% vs 44%, p<0.01).

A rapid mean CD4+ gain of 41.8cells/mm3 occurred within the first 12 weeks on therapy regardless of PI or NNRTI choice. Between 12 and 48 weeks, a trend towards better immune recovery was noticed among subjects receiving PI therapy with a mean CD4+ gain of 15.2 vs. 10.4 cells/mm3 for NNRTI recipients. This difference was most evident in individuals commencing therapy with a CD4 cell count <300 at baseline where the week 12-48 rise was 17.1 in the PI group vs. 6.4 in the NNRTI group (p<0.05). The rise in cells in the ‘second phase’ after the initiation of therapy is generally derived predominately from the naive cell subset.

The authors speculated that the reported PI inhibition of apoptosis in CD4 lymphocytes might explain their observations. At a practical level, however, it is unclear if these differences would be clinically relevant. In general, randomised studies of PI vs NNRTI or vs 3NRTIs have not reported significant differences in CD4 recovery and individuals responding to all these types of regimens remain generally well. Additionally, a second study in this session [2], discussed in more detail below, observed the contrary, that patients given an NNRTI based regimen had more than 100 cells/mm3 better rise in therapy than PI treated patients. Differences in these studies may relate more to design and population selection rather than accurately reflect what happens on average in clinical practice.

A more potent regimen may be better

At last year’s ICAAC data were presented suggesting that time to viral load undetectable in treatment responders may be slower in persons with high (>100,000 copies/mL) viral loads at baseline. At this year’s meeting an evaluation of the impact of baseline CD4 on the rate of viral load decline was presented. This study looked at 38 individuals naive to protease and non-nucleoside reverse transcriptase inhibitor therapy who were begun on a quadruple drug regimen of indinavir, nevirapine, zidovudine and lamivudine [3].

The study looked at viral dynamics, with plasma HIV load measured daily during the first week on therapy and the rate of decline analysed. The study included individuals with a wide range of viral loads (3.79 to 6.41log10 copies/mL, mean, 4.98, median, 5.07), CD4 cells (6 to 784 cells/mm3, mean, 332, median 288) and CD8 cells (34 to 1749 cells/mm3, mean, 750, median, 695) at baseline. The virus decay rate, ranged from 0.15 to 0.35 log10 copies/mL /day (mean and median, 0.26/day) and was, not surprisingly, significantly related to changes in viral load at 4, 8 and 12 weeks.

No correlation was found with the rate of viral decay and baseline viral load, CD4 or CD8 cell count. This study adds to the weight of evidence that the rate of viral decay with therapy is directly related to the actual antiviral potency of the regimen rather than other host or viral factors. Additionally, the data suggest that an individual should respond to any antiviral combination similarly regardless of CD4 or viral load. However in clinical studies this has not always been the case.

Criticism of treatment interruptions in chronic HIV-infection

Treatment interruption is, at best, an experimental strategy. Aside from individuals treated prior to seroconversion who have strong residual HIV specific immune responses, obvious immunologic benefits from treatment interruptions (TI) in persons optimally responding to therapy have consistently not been demonstrated. Additionally, risks of drug resistance and from sudden falls in CD4+ T cell count have been reported. However, in some of these studies there is always the occasional patient who maintains a low viral load after the interruption. The German study looked prospectively at 20 patients with CD4+ T cell counts established above 250/mm3 with viral load less than 50copies/mL for one year with, 14 individuals with low level (<5000copies/mL) rebound on therapy prior TI and 10 untreated individuals with viral loads consistently <1000copies/mL (‘long term non-progressors’ (LTNP)) [4]. Data on immune function and viral load changes were reported for days 0, 14, 28 and 56 after interrupting. After interrupting therapy the median changes of CD4 cell counts/µl at day 28 was a substantial -115 cells/mm3 below baseline values (range +19 to -333) in those previously <50 copies/mL and -159cells/mm3 (range -16 to -260) in those with previous low level viral rebound. VL increased by a median of 3.3 (0 – 4.0) and 1.2 (0.4 – 3.4) log, respectively.

Lymphoproliferative responses to HIV P24 antigen at day 56 occurred more frequently in those originally fully suppressed (9/12) than in the partial viral responsers (1/6, p = 0.04) although in both cases responses remained weaker than in the LTNP control group. Responses to other recall antigens did not shift with TI. The authors suggested that durable viral suppression prior to a TI is required to achieve a measurable immune benefit from a TI. The question remains, as this ‘benefit’ did not lead to meaningful virological control or prevent substantial CD4 decline on therapy, was it worth anything at all?

IL-2 and treatment interruption

A small Spanish treatment interruption (TI) study focused on interruptions in 12 persons treated for at least one year (median 23 months) with d4T, 3TC and indinavir during primary HIV infection (defined as 90 days from symptoms of PHI) [5]. The study planned four eight-week TIs with two to four months of retreatment accompanied by ultra-low doses of IL-2 (900,000 units /m2/day) evaluating restoration of HIV-1 specific T-cell responses (CD4 and CD8) and control of viral load.

A spontaneous drop of >0.5 log10/mL from peak in viral load was detected in six cases (50%) during the first TI cycle. Data available on the second cycle (n=7) found five cases (71%) had a similar response. Increases HIV specific CD4 lymphoproliferative responses or CD8 HIV specific cytotoxic responses were detected in two and in six cases after the first and second cycles, respectively. No genotypic resistance was detected. The authors suggested the data were supportive of this approach. Specific data detailing patients who received IL-2 were not reported.

Durability of therapy and measures of success

Success with therapy means different things to different people. Whilst regulatory authorities focus on virological and CD4 markers to approve drugs the reality for persons with HIV is about quality and length of life. Adverse events, prompting switches or interruptions of therapy are critical to well being on therapy.

A data base study from Alabama evaluated durability of initial treatment [6]. The analysis included 358 patients who began first ever therapy with a triple therapy HAART regimen after January 1996. Median viral load was 61,426 copies/ml, and CD4 175/mm3, with 77% male 50% black. Changes in medication for more than 14 days were considered a new treatment. An impressive 66% of patients remained on their first therapy at the end of one year. However, after four years only 27% remained on their original therapy and 45% had required at least four regimens.

Kaplan-Meier analysis of the proportion of patients remaining on their first regimen indicated that the mean duration of first HAART Rx was 414 days (range14-1736) and became less with each subsequent regimen, 263 days on second line and 200 days on third line therapy. Multivariate analysis showed a prior AIDS defining event was the most important predictor of therapy durability whereas age, race, sex, or baseline viral load were not predictors of duration.

The data indicate the importance of getting started with therapy before an individual had become unwell with HIV disease. As risk of opportunistic disease rises sharply below 200 CD4 cells/mm3 current guidelines appropriately recommend commencing therapy prior to reaching this CD4 level.

Does viral load continue to rise after initial rebound? Is viral load rebound always an indication to switch?

The answer to these dilemmas in part depend on what you have to switch to and the risk of accumulation of additional (cross-) resistance conferring mutations over time. The EuroSIDA database, a large database derived from 64 clinics across Europe was mined to address the stability of viral load over time in individuals who remained on a stable HAART regimen.

The evaluation does not address resistance or response to subsequent regimen issues [7]. The analysis included 471 patients who were on HAART > six months (without changes in past three months) and had viral load > 1000 but < 10,000 copies/mL (median 3.5 log10) on two consecutive occasions. CD4 count at baseline was 290 cells/mm3 with most (61%) individuals being on one PI+2 nucleoside analogues. The median change per year in viral load was: 0.20 log10/mL and in CD4 change per year was +13. Patients with a higher viral load at study entry had more rapid rises in viral load (p<0.01) whereas those with a history of viral load <500 copies/mL rebounded more slowly.

The authors discussed that the rate of rise in viral load was comparable to untreated HIV infection. The data are supportive of analyses by Deeks et al, which indicate that many patients have stable (or rising) CD4 T cell counts for prolonged periods after loss (or failure to achieve) viral suppression. However, as data from Deeks presented at the resistance workshop this year indicate, individuals remaining on therapy with detectable viral load continue to accumulate new resistance conferring mutations. For me, this approach is best reserved for persons with insufficient treatment options to make a clear next line treatment regimen.

Once daily dosing

Some physicians and patients believe that one important way in which therapy will become easier is to further reduce dosing frequency to once daily. Potential keys to establishing once daily regimens as the new standard of care include establishing that these regimens have optimal potency, as well as keeping tablet volumes low and preferably maintaining a second future once daily option. Those who argue against the once daily approach say that if a dose is missed then therapy is missed for a full 24-hour period creating greater risk of viral escape. This may not be true if we consider drugs with very long elimination half lifes.

The first study comparing once versus twice daily therapy for HIV was reported focussing on a population of Hepatitis C co-infected methadone users in Vancouver Canada. Patients were given a once daily (QD) directly observed dose or a twice-daily (BD) regimen in which one dose was given observed. The study is observational and patients do not appear to have been randomised. The basis of allocation to once or twice daily dosing was at the discretion of the managing physician. The mean baseline CD4 was 202/mm3 and viral load a substantial 215,956 copies/mL.

In the 29 patients QD group, 15 patients received ddI+3TC+nevirapine, one patients d4T+3TC+nevirapine, a further 12 a regimen based on saquinavir 1600mg/ ritonavir 100mg (8 with ddI+3TC) and in one case an efavirenz plus three nucleosides regimen. Of note, it is unclear at present if nevirapine has sufficiently robust pharmacokinetics for once a day dosing.

A small French study, Virgo, recently published, suggests that QD nevirapine regimens may be less effective than BD regimens. Amongst the 23 BD dosed individuals 11 were NNRTI based, seven PI-based and three triple class [2].

The median follow up is 18 months (all >6 months) and data were presented as an intention to treat analysis (it appears with switch or treatment modification included) of proportion of patients <50 or <400copies/ml at most recent follow-up. By this analysis, the proportions of patients <400 were 66% vs. 73% and <50 45% vs. 66% for QD and BD respectively. These differences were not statistically significant. Of note, no grade three or four (>5-fold elevations above normal) episodes of liver function abnormalities were reported. Methadone dosage adjustments were required by 31 patients. The data, whilst non-randomised and possibly not evaluating an optimal QD regimen, indicate the feasibility of directly observed therapy.

Switching to a PI-sparing regimen

Replacement of successful PI therapy to a PI-sparing regimen is now a widely accepted strategy, with supportive data for efavirenz, nevirapine and abacavir reported. Most studies have focused on patients on first line PI therapy as patients who have had prior mono- or dual nucleoside analogue therapy have been observed to do less well, at least on abacavir. It is not currently clear which is the ‘best’ agent to switch to, especially as the right agent may vary on an individualised basis.

A single (Italian) centre randomised study comparing continuation of PI therapy (n=167) with switching to abacavir (n=81) or efavirenz (n=87) reported outcomes after one year follow-up [8]. Patients had been on PI therapy for a mean 20 months with current viral load < 50 copies/ml. It was not stated if patients had previously received dual or monotherapy. All patients continued therapy with the same nucleoside analogues. Tolerability, leading to treatment interruption, discontinuation or switch was similar in the two switch arms and in both cases superior to the PI continuation group.

Viral rebound to two consecutive values above 500 copies/ml occurred in 3% of PI, 4.6% of efavirenz and 11.1% of abacavir treated patients. For this analysis, abacavir was inferior to PI continuation but not (p=0.09) significantly different to efavirenz. Efavirenz and PI continuation were also not significantly (p=0.89) different. Combining tolerability and virological endpoints, switch to efavirenz was superior to PI. No other analyses were statistically significant. Regarding lipids, here abacavir showed an advantage. Whilst triglycerides rose modestly (by 9mg/dl) with efavirenz, they fell by 9mg/dl with abacavir. Similarly, cholesterol rose 3mg/dl with efavirenz and fell 18mg/dl with abacavir. Statistical comparison of these changes was not reported.

The data indicate that efavirenz may have an edge over abacavir for virological outcomes after switching but that abacavir may have metabolic advantages. More details of the treatment history of these patients are needed to assess if prior nucleoside treatment played a role in treatment outcomes.

Is <50 copies adequate or do you have to reduce viral load to <3 copies?

One final presentation related to this issue was derived from the lopinavir/ritonavir (LPV/r) development programme. Previous data have indicated that the lower the viral load nadir on therapy the longer the duration of treatment effect. Thus there was an advantage to getting below 50 copies/mL relative to 500copies/mL. This is presumably because there is more viral replication at higher viral loads, hence viral resistance emerges more readily. Several different methodologies have suggested that viral replication persists in the majority (perhaps 2/3rds) of patients despite viral load being consistently <50copies/mL. So, 50 is 1 logarithm below 500, is there advantage to going another logarithm of virus load lower? Luc Perrin in Geneva has established a viral load assay that detects down to three copies/mL [9].

The association efficacy as assessed by this assay, relative to the standard 50 copy/mL assay was evaluated in 100 treatment-naive patients treated with LPV/r + d4T/3TC over three years. Viral load was measured at Weeks 24, 48, and 72 with the three copy assay and 12 weekly with standard assays. Two measures of response were analysed: Time to virologic failure (consecutive viral loads >400 copies/mL) at Week 144 and VL <50 copies/mL at Week 144. In this cohort, 54/96 (56%) of patients achieved at least one viral load <3 c/mL at Week 24, 48, or 72. By Kaplan-Meier analysis the proportion maintaining virologic response at Week 144 were 86% for those who achieved <3 and 84% for those who did not (p=0.70). Response was also independent of both baseline viral load and CD4 count. Whilst the authors suggested that longer follow-up is necessary to determine whether these factors will eventually have an impact on the duration of virologic response the data are suggestive that the ‘biological cut-off ‘ of current assays at 50/copies/ml may have a clinical relevance and that pushing harder for ‘deeper’ suppression has a diminishing return.


  1. Barreiro P, Soriano V, Casas E et al. Different Degree of Immune Recovery in HIV-Infected Patients Receiving Regimens with Protease Inhibitors or Non-Nucleosides.
  2. Conway B and de Vlaming S. Once-Daily vs Twice-Daily Directly Observed Therapy (DOT) for Management of HIV-Infected Individuals in a Methadone (MET) Program
  3. Polis MA, Ross R, Sidorov I et al. Plasma HIV-1 Decay Rate Constant during Initial Therapy Is Independent of Baseline CD4+ and CD8+ Lymphocyte Counts and Baseline Viral Load
  4. Hoffmann C, Mayer W Wolf E et al. Enhancement of HIV-Specific Immune Response during Structured Treatment Interruptions (STIs) in Chronic Infection Depends on Baseline Viral Load
  5. Miro JM, Plana M, Ortiz GM et al. Structured Treatment Interruptions (STI) in Patients Receiving HAART since Primary HIV-1 Infection (PHI): Spontaneous Control of Viremia in Almost Half of Cases after the First Two Cycles Off Therapy.
  6. Chen RY, Mugavero MJ, Westfal AO et al. Durability of HAART Regimens over Time
  7. Lundgren JD, Cozzi-Lepri A, Miller V et al. The Stability of the Viral Load in Subjects with Virological Failure who Remain on the Same HAART Regimen: The EuroSIDA Study.
  8. Maggiolo F, Migliorino M, Maserat R et al. Simplified Therapeutic Strategies in PI-Experienced Patients Successfully Treated with HAART
  9. Perrin L, King M, Brun S et al. Failure to Achieve HIV RNA ² 3 Copies/mL Does Not Predict Loss of Virologic Response to Kaletra (Lopinavir/Ritonavir) Therapy through 3 Years.

Source: NATAP (The US National AIDS Treatment Advocacy Project)

Links to other websites are current at date of posting but not maintained.