Bonanza: With Demand for Treatment Breaks at Record Level, Market for Tiny, Uninterpretable Studies Flourishes

With Demand for Treatment Breaks at Record Level, Market for Tiny, Uninterpretable Studies Flourishes

April 2000

Diverse Definitions

At least eighteen studies and speeches presented at the Seventh Retrovirus Conference included discussions related to treatment interruptions (TIs) and structured treatment interruptions (STIs) in various HIV infected populations — including primary HIV infection (three papers), chronic and virally suppressed (13 papers), and chronic and unsuppressed (two papers). On the final day, three late-breaker presentations also focused on recent developments in STI research.

In his opening overview of recent data on metabolic toxicities associated with long-term antiretroviral therapy, Bill Powderly of Washington University suggested that the heightened interest in when to start — or stop — antiretroviral therapy had re-emerged in the spotlight due to an ever-shifting HIV treatment paradigm. In the four years since Vancouver, the discovery of latently infected cell compartments, the diminution (or disappearance) of the eradication imperative, and the emergence of significant drug-related toxicities have combined to emasculate much of the macho HIE, HIH bravado. Indeed, Powderly suggested that the only people for whom therapy is absolutely indicated are those with constitutional symptoms, opportunistic infections, or a CD4 count below 200.

In HIV-infected persons with earlier or asymptomatic disease, intermittent therapy (pulsed HAART interspersed with STIs) may help prolong therapeutic efficacy. Viral load-based therapeutic goals should perhaps be supplemented or even replaced by CD4-based ones (keeping the CD4 count above a certain risk threshold; Powderly 2000). Donald Kotler commented that Powderly’s discussion of “these heretofore heretical propositions” was “a measure of our maturation in treating this disease” (Kotler 2000). Such discussions have certainly not been heretical among treatment activists in the past three years.

Several conclusions can be drawn from the STI research presented at the Retrovirus Conference: most STI research is still uncontrolled, carried out for short periods of time in small patient groups; STIs appear safe in the short-term if patients receive frequent RNA and CD4 measurements and are retreated as necessary; resumption of therapy appears to be successful if the patients were fully-suppressed before the STI; and the emergence of drug resistance has not been observed in this population. The immunologic benefit of intermittent therapy, however, remains undefined, and the clinical usefulness of HIV-specific CD4 and CD8 responses has yet to be demonstrated in this setting.

The usefulness of adjunctive approaches such as interleukin-2 (IL-2) or putative “therapeutic” vaccines such as ALVAC/gp120 (a prime-boost vaccination schedule which uses the canarypox-based “ALVAC” vaccine as the initial immunization and then the recombinant envelope based vaccine (in this case, gp120) for a series of subsequent booster immunizations; the canarypox products are best at producing cellular immune responses while the recombinant envelope products (e.g., gp120, gp160) have been shown to elicit some sort of an antibody response) remain unconvincing.

Different teams reported divergent results with IL-2 during or after HAART suppression. Reports about HAART plus ALVAC/gp120 followed by an STI are too preliminary to yield any conclusions and, in any case, were no better than results shown for HAART followed by an STI with no vaccine.

In unsuppressed patients, the incidence of “reversion to wild-type” virus in plasma is at least as high as previously reported (see TAGLine’s report from the summer 1999 Salvage Therapy Workshop), but drug-resistant virus remains archived in mononuclear cells in half or more of the cases observed. The clinical implications of reversion to wild-type in plasma — or persistence of resistance in cells — require prospective investigation. The most important — if not immediately useful — conclusion is that further well-designed, prospective, and well-controlled studies of intermittent therapy and STIs need to be conducted across the spectrum of HIV disease.

The treatment interruption (TI) and structured treatment interruption (STI) papers included preliminary (mostly uncontrolled) reports on experience with various treatment interruption designs — some structured, some more haphazard. This research can be divided into three patient populations or scenarios: acute primary infection, chronic infection with ongoing therapeutic success and chronic infection with a waning therapeutic effect.

Scenario A: Primary HIV Infection, Viral Load Successfully Contained

What the studies showed:

Treatment interruption following HAART given during acute infection boosts and broadens anti-HIV CTL responses.
Treatment interruption following 48 weeks of HAART + hydroxyurea during primary infection resulted in viral rebound regardless of inclusion of hydroxyurea. Rebound is blunted after three weeks. Rechallenge is effective.
Treatment interruption in two of four primary infection patients treated with HAART in combination with ALVAC/rgp160 resulted in viral rebound. Possible broadened anti-HIV CTL responses.

STI following HAART given during acute infection boosts and broadens anti-HIV CTL responses.
Altfield and colleagues from Mass General reported on 15 acutely infected individuals who were identified before seroconversion and treated with immediate HAART. After over one year of HAART, 7/13 individuals underwent an STI.HIV RNA, CD4, and cytotoxic T lymphocytes (CTLs) against env, gag, nef and RT, were measured weekly. Mean baseline viral load at diagnosis was 13.8 million; RNA became undetectable in all patients after a median ten weeks (range 4-19) of HAART and remained undetectable during HAART treatment. Low magnitude CTL responses against one to four epitopes were seen in 10/15 subjects (66.7%) at diagnosis and persisted at lower frequencies during HAART.

In the seven STI patients, virus rebounded within one to eight weeks [levels not given] and HAART was reintroduced. CTL responses were boosted in all STI patients, who gained a median of two new recognized epitopes and increased responses to previously recognized epitopes. Two people with no previous CTL responses gained them post-STI.

In two people who had a second STI, additional CTL increases were seen. The authors concluded that, “CTL responses are detectable during acute HIV-1 infection prior to seroconversion, but are narrowly defined at a few epitopes. These responses persist at lower levels during HAART. STIs result in a persistent augmentation of the frequency and breadth of CTL responses in these subjects, which can be further boosted by additional controlled interruptions.”

Of concern was an episode reported in the poster (but not the abstract) in which an individual simultaneously stopped HAART (for financial reasons and without telling the investigators) and received a flu shot; his viral load rebounded sky-high and he experienced the equivalent of a seroconversion syndrome (Altfeld et al., abstract and poster 357).

Comment: This paper suggests the possibility of broadening anti-HIV cellular responses by using STIs after suppressive HAART during acute infection and emphasizes the need for randomized studies of “when to start/when to stop” in the acute infection population.
STI following 48 weeks of HAART + hydroxyurea during acute infection: rebound observed regardless of hydroxyurea; rebound is blunted after three weeks; rechallenge is effective.
Zala and colleagues from Buenos Aires and Vancouver followed 18 patients treated during acute infection with d4T, ddI, nevirapine with or without hydroxyurea. After 48 weeks of treatment, patients were offered an STI with rechallenge allowed after 30 days if viral rebound was confirmed.After 48 weeks of HAART, 83% had viral load <500 and 67% had <50. Eight patients (five on hydroxyurea) interrupted therapy after a median 57 weeks (range 53-75). All eight rebounded to >5,000 copies after 21 days. The authors noted that, “Plasma viral load peaked within three weeks and decreased thereafter by a median of 0.69 log.” HU did not significantly affect the kinetics of rebound. All patients responded to rechallenge (viral load <500) within a median of four weeks. One patient on ddI/d4T/NVP discontinued treatment due to grade-three peripheral neuropathy at week 20 and his viral load remained <50 after 46 weeks off therapy (Zala et al., abstract 558).
Two of four acutely infected patients treated with HAART and immunized with ALVAC/gp160. STI blunted viral rebound and may have broadened anti-HIV CTL responses.
Jin and colleagues at New York’s Diamond Center treated four individuals with AZT/3TC/indinavir within 100 days (median 60) after they were infected. Median baseline RNA was 4.76 log. HAART suppressed viral load <50 in all subjects. After 2.5 years of therapy, subjects were vaccinated four times with the prime-boost ALVAC 1452/rgp160. After 1,162 days of HAART and one week after their final vaccination, four patients chose to stop therapy.In two individuals (50%) delayed viral rebound was observed; RNA became detectable at 68 and 85 days post vaccination; initial virus doubling times were 4.5 and 3.2 days respectively. The other two patients (50%) experienced rapid viral rebound with detectable virus within 13 and 23 days; both had viral doubling times of 1.4 days — similar to previously reported rates.

The two delayed rebounders had significant increases in CTLs to more than one viral antigen following vaccination. One rapid rebounder had a monospecific response to gag, while the other had CTL response to the vaccination. After four to eight months off therapy, the delayed rebounders had RNA levels of 3.75 and 2.52 log, while the rapid rebounders had 3.55 and >4.70 logs after four months off therapy. The authors speculated that “Therapeutic vaccination can enhance cellular immune responses that are temporally associated with suppressed HIV-1 rebound kinetics after discontinuation of HAART” (Jin et al., abstract LB12).

Comment: This small, uncontrolled study did not display results significantly different from those of Garcia et al. [see below], whose study involved three STI cycles post-viral suppression, but no “therapeutic” vaccination. The Garcia study showed 4/9 “lower rebounders” developed anti-HIV CTLs during their STIs, similar to the 2/4 “slower rebounders” reported on here. In the final question of the conference, Luís Montaner of the Wistar Institute pointed out that, as the study lacked controls, it was difficult to interpret. Of the three late-breaking STI papers, only the Deeks study was well-controlled.

Scenario B: Chronic HIV Infection, Viral Load (Moderately to Fully) Suppressed

What the studies showed:

Virus rebounding during treatment interruptions differs from that detected in latent reservoirs.
Viral setpoint not altered after two years of HAART followed by STI.
Low viral load during treatment interruption may not reflect a therapeutic effect.
ddI/hydroxyurea followed by STI may provide better immune control than HAART followed by STI.
STIs augment HIV-specific CD4 and CTL responses in some individuals with chronic HIV infection who previously controlled HIV while on HAART.
Randomized STI trial in virally suppressed chronic infection shows only transient HIV-specific immune responses in a minority of participants. Also showed STI to be safe.
HAART + IL-2 followed by an STI while continuing IL-2 elicits viral rebound followed by partial immune control, resulting in detectable viremia below pre-treatment setpoint.
Randomized HAART vs. IL-2/HAART study suggests IL-2 may not affect virus production, latent infection, or immunologic control.
Lymph node RNA rebounds in parallel to plasma RNA during STI.
Viral load does not rebound during an eight day STI after HAART suppression.
French group reports no immunologic benefit to brief treatment interruption.
Preliminary results from Swiss study indicate viral rebound may be blunted after three STIs.
Four of nine (44%) chronically infected patients, successfully suppressed on HAART, showed improved anti-HIV CD4 and CD8 responses in parallel with control of viral rebound after three STIs.

Virus rebounding during STI differs from that detected in latent reservoirs.
Tae-Wook Chun and colleagues from NIAID threw another wrench into current theory by reporting that the viral rebound observed upon treatment cessation may not come exclusively from the pool of latently infected, resting CD4 cells harboring replication-competent HIV. They used quantitative microculture and heteroduplex mobility assays to compare viral RNA (from reservoirs) before and (from plasma) after rebound in nine patients who underwent an STI in a clinical trial (the NoHRT study, Davey 1999).Rebound and reservoir RNA were identical in just two of nine patients, and diverged in seven of nine. This disparity could result from sampling error (the RNA may have rebounded from latent cells in another, unmeasured compartment), but the paper also suggested that the RNA may rebound from another, as-yet unidentified reservoir, where active replication may be ongoing despite HAART. Possibilities include tissue macrophages and microglia in the central nervous system (Chun et al., abstract 239).
Viral setpoint not altered after two years of HAART followed by treatment interruption.
Hatano and colleagues from NIAID measured plasma viral load from 12 patients who achieved good viral suppression (<50 copies/ml) on HAART and then went off antiretroviral therapy. Viral control on HAART was achieved in a median of 48 days and maintained for a mean 661 days; HAART was discontinued because of patient preference (four), toxicity (three), illness (one) or for protocol reasons. The length of the treatment interruption (TI) was a median 55 days (range 14-168, mean 66).The median difference between pre- and post-HAART viral load was 0.45 logs (mean 0.65). Eight patients had post-HAART viral load values higher than pre-HAART; four had lower values. The median duration between pre- and post-HAART values was 2,168 days (5.9 years). The authors concluded that, “After discontinuing HAART most individuals had rebounds in their viral burden approximating pre-HAART levels, even after a significant lapse of time approaching six years. Our data suggest that HAART with good viral suppression for almost two years may not alter intrinsic viral setpoints” (Hatano et al., abstract 349).
Low viral load during treatment interruption may not reflect a therapeutic effect.
Sherer and colleagues from Chicago identified 13 HIV-infected individuals who achieved a viral load below 500 copies a minimum of 90 days after stopping antiretroviral therapy (median 2.5 years, range 3 months-5.6 years). There were no striking features of the prescribed antiretroviral regimens. Only two patients received intermittent therapy, and two were treated during acute HIV infection. All remained HIV antibody positive. Pre-therapy viral loads (available for 7/13 patients) were very low (<500-9,000 copies). Three of ten patients were heterozygous for the CCR5 deletion; 2/12 had class A HIV. Devoid of any clear trends in their research results, the investigators could observe only that, “Isolated cases of individuals with low or undetectable HIV RNA upon stopping therapy do not [appear to be] evidence of an exceptional therapeutic effect” (Sherer et al., abstract 351).Comment: What was the denominator here (e.g., how many other patients went off therapy and didn’t achieve a viral load below 500 copies)?
ddI/hydroxyurea followed by STI may provide better immune control than HAART followed by STI.
Lori and colleagues conducted a case-control study comparing nine individuals receiving ddI and hydroxyurea (HU) in the PANDA cohort with eight individuals on HAART. “Unlike HAART patients, PANDAs have low but detectable viremia and vigorous HIV-specific cellular immune responses,” Lori boasted. PANDAs and matched HAART controls interrupted therapy for eight weeks. Matching criteria before the STI included the duration of previous treatment (>2 years), CD4 and CD8 counts. Failure during STI was defined as a viral load rise to over 10,000 copies or CD4 drop to below 200. Five of eight HAART patients failed by week six and had to restart therapy, whereas no PANDA had to restart during eight weeks of follow-up. [What happened after eight weeks?]Lori concluded that, “Reconstitution of HIV-specific cellular immune responses obtained during the treatment of chronically infected patients was associated with the control of viral rebound during treatment interruption” (Lori et al., abstract 352).

Comment: This case-matched control study would be more compelling if it were randomized.
STIs augment HIV-specific CD4 and CTL responses in some individuals with chronic HIV infection who previously controlled HIV with HAART.
Papsavvas and colleagues from the Wistar Institute in Philadelphia measured anti-HIV cellular (CD4 and CD8) immune responses in five chronically infected individuals who had maintained viral suppression on HAART and subsequently went on an STI and compared them with five untreated controls. During the STI, the five previously suppressed individuals were able to “significantly increase broad antiviral T-helper and interferon-gamma secreting CD8 T-cell responses as a result of complete treatment interruption.” There were “substantially fewer changes” in the control group.The STI group experienced significant increases in anti-HIV T-helper responses against p24 and gp160 “preceding significant increases in interferon-gamma secreting CD8 T-cell responses against viral envelope antigens.”

After a median 46-day STI, three subjects restarted HAART and achieved 98.86% reductions in plasma viremia by 21 days and maintained or further increased the cell-mediated anti-HIV responses. The remaining two subjects stayed off therapy, maintained high cell-mediated responses, and maintained RNA below 1,000 copies. “These observations,” the authors noted, “provide the first demonstration that CD4 and CD8-mediated cellular immune responses against autologous HIV-1 are augmented as a result of temporary treatment interruption in a subset of chronically infected individuals.” (Papasavvas et al., abstract 353).

Comment: This small, uncontrolled study suggests that CD4-mediated help may be required to stimulate anti-HIV CD8 activity.
Randomized STI trial in virally suppressed chronic infection shows only transient HIV-specific immune responses in a minority of participants, but is safe.
Ruiz and colleagues from Barcelona assessed virologic and immunologic changes during an STI in 25 chronically infected individuals who achieved long-lasting (>2 years) viral suppression and a CD4/CD8 ratio of >1 on HAART. They were randomized to stay on HAART (Group one, N = 13) or interrupt for a maximum of 30 days or until the viral load increased over 3,000 (Group two, N = 12), then resuming the same prior ART. Ninety days after rechallenge, a second STI was begun. Viral rebound kinetics were measured every two days during the two STI periods in group two and every three months in group one. T cell quantification and proliferation assays were carried out before and after the first STI; genotypic resistance was measured when treatment resumed.2/12 patients did not rebound after 30 days during the first STI. Only one maintained viral load <20 during the second STI. Among the rest, viral load became detectable (>20) for a median of 14 and 15 days during the first and second STIs, respectively. Viral load rose exponentially with a mean half-life of 1.6 and 2.2 days during the first and second STIs, respectively. CD4 and CD8 percentages did not change during STIs, but CD38+ (activation) expression rose significantly in response to viral rebound.

Four patients (33%) gained T-helper responses to recall antigens, and two of these (16.5%) developed HIV-specific p24 responses during the first STI. Only recall antigen T-helper responses were maintained by the start of the second STI. Drug resistance was not observed.

The investigators concluded that STIs were not associated with CD4 reductions or clinical complications after two years of effective viral suppression, that virus rebounded in most but not all patients, that virus was effectively controlled upon rechallenge, and that “HIV-specific helper T-cell responses may require subsequent cycles of STI to keep viral replication under control” (Ruiz et al., abstract 354).
HAART + IL-2 followed by an STI while continuing IL-2 elicits viral rebound followed by partial immune control, resulting in detectable viremia below pre-treatment setpoint.
Smith and colleagues from New York and San Jose selected nine chronically infected “aviremic” individuals with normal circulating lymphocyte counts on HAART plus daily low-dose interleukin-2 (IL-2) therapy. “Upon cessation of HAART, but continuation of IL-2, plasma HIV became detectable in all individuals in 19 ± 3 days, then increased rapidly over two weeks (doubling time 2 ± 0.2 days) to a peak of 348,123 ± 232,811 RNA molecules/ml. Remarkably, the virus concentration then decreased progressively over the next four weeks, reaching a low stable level of 26,086 ± 8,087 RNA molecules/ml, <10% of the peak concentration (p<0.001).” Moreover, this level was significantly lower than the mean viral load prior to HAART [see “viral setpoint not altered after HAART alone,” Hatano et al., above] (70,315 ± 20,459).CD4 counts dropped by 24% compared to baseline (p<0.01) just after the peak of viremia. CD8 cells rose to 200% of baseline (p<0.01) and remained high while viremia declined. The authors concluded that, “Contrary to previous assumptions, chronic HIV infection does not preclude the development of effective immune reactivity to HIV.” Additional trials of HAART cessation while continuing IL-2 “to permit endogenous HIV antigenic stimulation . . . [and] promote the expansion of antigen-activated CD8+ T cells, now appear warranted” (Smith et al., abstract 355).
Randomized HAART vs. IL-2/HAART study suggests IL-2 may not affect virus production, latent infection, or immunologic control.
Presenting data at odds with the previous IL-2 report, Stellbrink and colleagues reported on a randomized comparison of HAART alone versus HAART plus IL-2 in 56 patients with over 350 CD4 cells or recent seroconversion.Viral load and CD4 counts were measured at days 0, 14, 28 and monthly; and after treatment discontinuation at weeks one, two, four and eight. Lymph nodes were biopsied before treatment and after six months of maintaining viral load below 50 copies. 120 HIV negative controls were used for normalizing age-matched T cell subsets. Median age was 41, median CD4 count 410, median CD8 count 878, and HIV RNA 4.83 log.

IL-2 did not affect the magnitude or kinetics of plasma viral load drop, but did increase the rate of normalization of CD4 counts. Viral replication was detected in the follow-up lymph node biopsies of 11/15 non-IL-2 and 10/16 IL-2 patients (non-significant). Proviral DNA levels and decay kinetics did not differ between the two groups. After treatment discontinuation, RNA rebounded rapidly and was transiently higher than baseline. One IL-2 patient subsequently cleared viremia in the absence of therapy.

Conclusions: Despite a strong and durable immunomodulatory effect, IL-2 has no persistent effect on virus production or latent infection in vivo [or] immunological control over HIV infection” (Stellbrink et al., abstract 240).

Comment: Whereas the Smith patients were selected, the Stellbrink patients were randomized.
Lymph node RNA rebounds in parallel to plasma RNA during STI.
Orenstein and colleagues compared lymph node architecture, HIV RNA and protein expression in lymph nodes from five individuals fully suppressed on HAART and following viral rebound after treatment interruption.Lymph nodes were excised after at least one year of HAART with CD4>600 and viral load <50, and then after viral rebound. Lymph nodes at baseline on HAART were quiescent to mildly hyperplastic. They had no detectable follicular dendritic cell (FDC) associated HIV RNA or p24 protein and only rare mononuclear cells (MNC) expressing RNA (one lymph node) or p24 (two lymph nodes).

Plasma viral load after a one to two month STI ranged from 329 to 3.2 million. CD4 counts declined by 5-48% and did so most in those with the highest viral rebound. Four of five second lymph node biopsies (80%) were more hyperplastic than the initial lymph node. The fifth lymph node showed paracortical hyperplasia but no germinal centers.

Cellular HIV RNA in the second lymph node paralleled plasma viremia. Cell activation increased. “Quiescent lymph node from individuals on extended treatment with HAART rapidly became hyperplastic and activated following one to two months of treatment interruption. Virus expression in lymph node MNC parallels the rebound in plasma viremia and fall in CD4 cells.” The lymph node photographs in Orenstein’s poster were a chilling reminder of the fact that the viral rebound seen after STIs visually resembles that seen during chronic, progressive HIV infection (Orenstein et al., abstract 358).

Comment: What would happen to viral expression in lymphoid tissue if anti-HIV cellular immune responses returned in association with blunting of plasma viremia?
Viral load does not rebound during an eight-day STI after HAART suppression.
Kilby and colleagues from the University of Alabama at Birmingham (UAB) conducted an eight-day STI followed by rechallenge in five individuals with chronic infection who had maintained plasma viremia below 200 copies for at least six months. All five had detectable RNA before HAART and four were undetectable (<20) on HAART.Eight day rebound kinetics were similar in the four with RNA<20 and the one whose RNA was 100 on HAART; the latter actually did not experience a rebound. The two subjects whose virus rebounded re-suppressed after rechallenge. Three subjects had a lymph node (LN) biopsy at day eight. There was no increase in RNA+ cells above the low pre-STI level. CD4 cells dropped slightly but significantly during the STI and returned to baseline after 30 days of HAART. Activated CD4 and CD8 cells also increased, even in those whose viral RNA did not rise during the STI (Kilby et al., abstract 359).

Comment: The eight-day STI here may be too short to show much.
French group reports no immunologic benefit to brief treatment interruption.
Carcelain and colleagues from Paris studied HIV-specific T cell responses and HIV control after repeated (three or four) 7-21 day treatment interruptions in three individuals whose viral load had stayed <20 copies and whose CD4 count risen to >400 for at least two years on HAART. Five control patients whose virus rebounded during the first six months of HAART were also studied.The investigators measured RNA levels, HIV p24-specific T helper proliferation and interferon-gamma production, CD8 responses and T cell activation. Before treatment interruption, there were no significant CD4 and weak CD8 responses to HIV.

In patient one, three seven-day interruptions did not induce viral rebound or T helper stimulation. In patient two, HIV-specific CD4 responses increased only at the first rebound. In patient three, these responses occurred at each interruption but were transient. HIV-specific CD8 cell frequencies did not increase. Interferon gamma producing CD4 cells were observed after the treatment interruption, but the cells were rapidly deleted after virus replication resumed (Carcelain et al., abstract 356).

Comment: These were brief STIs. Virus may need to rebound to higher levels for longer durations to induce more durable HIV-specific responses. [See the similar low rebound observed in the Kilby study described above.] (abstract 359).
Preliminary results from Swiss study indicate viral rebound may be blunted after three STIs.
Fagard and colleagues reported early results from the Swiss Intermittent Treatment Trial (SITT). Eligible patients were antiretroviral naïve before HAART, did not experience treatment failure during HAART, maintained viral load <50 for at least six months, and were NNRTI naive. Treatment is stopped for two weeks and resumed for eight weeks, in four cycles. At week 40, treatment is suspended unless viral load rebounds to over 5,000. Endpoints include the amplitude of rebounds, and the number of patients with viral load <5,000 and CD4>400 without ART after week four.57 of a projected 120 patients enrolled between April and September 1999. Pre-HAART median CD4 count was 398, viral load 4.56 logs. Median pre-SITT HAART duration was 22 months; median CD4 count on starting SITT was 700. By the reporting deadline, 16 patients had two STIs and four had three STIs. Viral rebound in the first STI occurred in 28/43 (65%) and during the second STI in 12/16 (75%, p = NS). The two rebounds were similar in amplitude (p = 0.2).

After seven weeks’ retreatment, 3/20 evaluable and compliant patients did not achieve viral load <50 (RNA = 62, 105 and 147). One had a fever and sore throat at the end of the first STI; viral load here was 925,000. “Very early results in the few patients with three treatment stops are modestly encouraging” (Fagard et al., abstract 458).
In four of nine (44%) previously suppressed, chronically infected individuals, anti-HIV CD4 and CD8 responses increased in parallel with control of viral rebound after three STIs.
In a late-breaker presentation, Garcia and colleagues from Barcelona reported on ten chronically infected patients given three consecutive STI cycles after 52 weeks of d4T, 3TC and ritonavir or indinavir whose viral load had been <20 copies for >32 weeks and whose CD4 count was over 500.They were given three four-week STIs; after the first two, HAART was resumed for 12 weeks and stopped again if viral load went back below 20 copies; after the third cycle, they were kept off HAART until the viral load reached a new setpoint.

Baseline plasma viral load was <20 copies in all cases and <5 copies in 7/10 cases (first STI) and in 7/9 cases (second and third STI). Viral rebound occurred in all cases with a mean doubling time of 2.23, 3.38 and 3.25 days over the three STIs.

At the second STI, in 4/9 patients (44.4%) viral rebounded to levels similar to baseline and then dropped spontaneously by 0.8, 0.8, 1.3 and 2.09 logs respectively. The authors explained that, “These four patients developed strong and broad CTL responses and a strong CD4+ lymphocyte proliferative response to HIV-1 antigens.”

After the third STI, the viral load setpoint appeared significantly lower than baseline in four of seven patients. Drug resistance mutations were not detected after any of the STIs. Recovery of HIV-specific immune responses (CD4 proliferative and CD8 cytotoxic) correlated with spontaneous control of viremia.

The authors concluded that STIs “may induce effective specific cytotoxic and CD4+ lymphocyte proliferative immune responses against HIV-1 antigens associated with a spontaneous drop in plasma viral load in chronic HIV infection.” Drug resistance did not emerge during the STIs, but CD4 counts dropped, and took at least six months from rechallenge to recover (Garcia et al., abstract LB11).

Scenario C: Chronic HIV Infection, Viral Load Unsuccessfully Contained (“Salvage”)

What the studies showed:

Two of four salvage STI patients experienced increased levels of HIV RNA and white blood counts in the cerebrospinal fluid (CSF).
STIs in virologic failures after partial suppression on protease inhibitors result in reversion of protease resistance to protease sensitivity, CD4 drops and RNA rebounds, and return of a “fitter” viral phenotype.

Two of four salvage STI patients experienced increases in CSF HIV RNA and white count.
Price and colleagues from San Francisco reported on four patients, three of whom took a twelve week STI after treatment failure and had their cerebrospinal fluid (CSF) sampled via lumbar puncture (LP) at baseline, weeks three, six and 12; the fourth was observed in a study of CSF response to new or changed ART. HIV RNA was measured in peripheral blood and CSF.In two subjects plasma RNA rose by less than one log during the STI and the CSF RNA was maintained in a nearly fixed proportion. In the other two patients, however, whose plasma RNA rose by one log, their CSF RNA rose by two to three logs, as did the WBC. The two “showed a disproportionate increase in local [CSF] infection and related host cell response”; however, there were no overt neurological symptoms (Price et al., abstract 306).
STIs in virologic failures after partial suppression on PIs result in reversion of protease inhibitor-resistance to protease inhibitor-sensitivity, CD4 drops and RNA rebounds, and return of a fitter viral phenotype.
In perhaps the most impressive, detailed, and well-controlled STI study presented at the 7th Retrovirus Conference, Steven Deeks and colleagues from U.C. San Francisco addressed a number of high-priority research issues in a single study. They were seeking to:

Examine the virologic, immunologic and clinical impact of STIs compared with continuing partially-suppressive protease inhibitor-containing therapy in a heavily-pretreated population with detectable viral load and few treatment options;
Confirm or refute Veronica Miller’s provocative data regarding the reversion to wild-type virus seen in 67% (26/39) partially suppressed, highly-drug experienced patients in the Frankfurt Clinic Cohort;
Measure the kinetics of viral reversion to wild-type, if observed;
Relate viral reversion to viral rebound and CD4 declines;
Relate viral reversion in vivo to replicative capacity (hopefully a good surrogate for “fitness”) in vitro (though it is unclear whether replicative capacity = virulence)
Measure CD4 cell turnover by the deuterated glucose method

Deeks and colleagues randomized patients to undergo an STI or stay on partially suppressive protease inhibitor-containing HAART. Eligible patients had virologic failure for at least 12 months, a viral load >2,500, had been on a stable regimen for at least ten weeks, and were willing to discontinue therapy.

Eighteen patients were randomized to a twelve-week STI. Phenotypic resistance (Virologic assay), HIV RNA, and CD4 counts were measured weekly for 12 weeks, then every four weeks. Replicative fitness was measured (see abstract 233). Among the 18 STI patients, the baseline CD4 count was 254; the baseline RNA 4.6 log; the median prior protease inhibitor therapy duration 36 months (of which virologic failure had been evident for 31 months), the median decrease in protease inhibitor susceptibility 56-fold. After the three-month STI, the median CD4 decrease was 94 cells (range -28 to -128) and the median viral load increase was 0.82 log (range 0.34-0.92). (Patients were encouraged to resume therapy if the CD4 count dropped by half or the viral load rose over one log.)

Virus reverted to protease inhibitor-susceptible in 16/17 patients [higher than the 67% rate in the Frankfurt cohort]. The mean time to reversion was 8.5 weeks; however, it ranged widely from 2-15 weeks. Importantly, in most patients, phenotypic resistance to most or all drugs dropped precipitously and simultaneously. In other words, a true wild-type virus re-emerged all at once, rather than sequentially. Moreover, this wild-type reversion was highly correlated with a steep viral RNA rebound and CD4 drop which contrasted with milder changes observed in the first weeks of the STI. Greater protease inhibitor resistance and higher viral load over baseline values were associated with delayed reversion (p = 0.04, 0.003). Nucleoside resistance persisted, though at lower levels, after reversion to a protease inhibitor-susceptible virus in 7/17 patients.

“Replicative fitness (relative to wild-type virus) increased from a median 22.3% to 67.1% (p=0.004). Resistant virus identical to baseline was cultured from PBMCs 12 to 36 weeks after therapy discontinuation in four of eight patients showing phenotypic reversion in plasma virus.” Thus, drug-resistant virus remained archived in proviral DNA in half the cases measured.

It’s possible that resistant virus could occur in non-sampled cells; more importantly, virologic results of rechallenge remain to be determined. “Discontinuing therapy was associated with either decreased T cell half-life or decreased CD4 production.” These T cell turnover data were not shown in the late-breaker due to time constraints. The abstract concluded that “These data suggest that antiretroviral therapy [at least with a protease inhibitor] is associated with continued immunologic and virologic benefit, despite high level resistance. This benefit may reflect, in part, the maintenance of a less fit virus” [abstract]. “Interrupting therapy is often associated with biphasic increases in viral load and decreases in CD4 cells, reversion to wild-type virus (often abrupt but with variable time course), [and] increased replication capacity” [presentation] (Deeks et al., abstract and presentation LB10).

Comment: If more STI studies were as well-controlled, prospective, and data-intensive as this one, progress in STI research would accelerate.

Tabular Summary of STI and TI Papers Presented at the San Francisco Retrovirus Meeting
Time on ART	Pt. Char.	STI Duration	N_STI	N_cx	Observations	Author
Acute Primary Infection
>1 yr.	“Undetectable”	1-8 wks.	7	8	Rebound* w/i 1-8 wks. CTLs boosted.	Altfield
1 yr. (+HU)	RNA<500	21-30 days	8	10	Rebound w/i 21 days	Zala
2.5 yrs. (+ALVAC)	-RNA <50	13-85 days	4	0	2 rapid rebound, 2 delayed. CTLs boosted in 2.	Jin
Chronic Infection, Virus Suppressed
N/A	(see NoHRT)	N/A	9	0	Viral divergence in all but 2 pts.	Chun
22 mos.	RNA <50	14-168 days	12	0	8 pts. established a higher set point; 4, lower.	Hatano
3 mos.-5.6 yrs.	N/A	2.3 yrs. (median)	13	0	Enrolled essentially LTNPs. Descriptive.	Sherer
(ddI+HU vs. HAART)	(see PANDA)	8 wks.	9	8	5/8 HAART vs. 0/9 ddI/HU rebounded at Wk8.	Lori
N/A	“Undetectable”	46 days (median)	5	5	STI pts. showed increased CD8 and CD4 activity.	Papsavaras
>2 yrs.	CD4/CD8 =1.0	30 days	12	13	2nd STI at 90 days post rechallenge. 2/12 pts. showed no rebound during first STI. 1 of 2 showed no rebound during 2nd STI. All others rebounded after median 15 days STI.	Ruiz
(HAART+IL-2)	(IL-2 during STI)	8-10 wks.	9	0	All pts. rebounded w/i 22 days; fm Wk6 to Wk8, however, RNA levels fell over time by an avg of 1.5 log (or < 10% of peak). p<0.001	Smith
(HAART+ IL-2)	CD4>350	8 wks.	56	120	IL-2 had no effect on RNA clearance. STI pts. had rapid rebound. Rebound levels shot transiently above baseline.	Stellbrink
>1 yr.	CD4>600 RNA<50	1-2 mos.	5	0	RNA during STI ranged from 329 copies/mL to 3,200,000 copies! 4/5 lymph nodes showed increased hyperplasia. One showed paracortical hyperplasia and no germinal centers!	Orenstein
BLQ >6 mos.	RNA < 200	8 days	5	0	Two of five pts. rebounded during 8-day STI.	Kilby
> 2 yrs.	CD4>400, RNA<20	7-21 days	3	5	Multiple STIs; no rebound during first 7-day STI in 1/3.	Carcelain
22 mos.	BLQ > 6 mos.	2 wks.	57	0	Target enrollment=120. Series of four 2-week STIs, each followed by 8 wks. back on HAART. 28/43 pts. rebounded during the first STI; 12/16 rebounded during second STI.	Fagard
1 yr. BLQ > 32 wks.	CD4>500	4 wks.	10	0	Series of three 4-week STIs. At first STI viral rebound occurred in ALL patients, then dropped spontaneously in 4/9. These “lower rebounders” developed anti-HIV CTLs during STI. CD4 cell counts dropped, requiring >6 months to recover. After 3rd STI, viral setpoint was significantly lower than baseline in 4/7 pts.	Garcia
Chronic Infection, Virus Not Suppressed
N/A	“Salvage”	12 wks.	3	1	2 exp’d. modest rebound; 2, fulminant rebound.	Price
36 mos.	“Failure” >12 mos. CD4=254. RNA=4.6 log	12 wks.	18	N/A	Viral fitness increased fm 22% to 67%.	Deeks
* Depending on the study, viral “rebound” may be defined as low as 5 copies/mL and as high as 5,000 copies.

Diminishing Returns

Two-Week Viral Rebounds in Four SITT* Patients over the course of Three STIs

	1st STI	2nd STI	3rd STI
Patient #1	60,000**	30,000	2,700
Patient #2	12,600	12,100	728
Patient #3	1,327	200	973
Patient #4	<10	<10	<10
* Swiss Intermittent Treatment Trial (SITT) Fagard et al. abstract 458 ** copies/mL