Veronica Miller and her German research colleagues previously reported at last year's Salvage Therapy Workshop on the concept of treatment interruptions. The idea of treatment interruptions, or structured therapy interruptions (STI), has received much attention in the last 1.5 years. Every AIDS conference has had a number of researchers report their studies on the results they've observed from treatment interruptions. The way I see it, there are various types of interruptions or STIs (structured therapy interruptions): (1) for the person who started therapy early, reduced their viral load to undetectable and remains undetectable-- for this group researchers are testing the idea that multiple structured interruptions may stimulate an immune response which could control HIV without HAART; (2) STIs for individuals with chronic HIV, who've had HIV for a while, who have undetectable viral load, and may have started treatment later in the course of disease or in the medium range when CD4s were about 300; (3) individuals who have extensive treatment experience and drug resistance, have been through several regimens, their viral load is dectectable and their CD4s may be low (in the 100-200 range) or higher 200-350 range).

The ACTG, and other researchers have focused attention on developing studies to explore interruptions. Researchers developing and testing immune-based therapies such as Remune, IL-2, and vaccines in combination with HAART have designed studies exploring IBTs+HAART with "strategic therapy interruptions, in individuals who have undetectable viral load and either started treatment early or in acute infection, or in individuals who started therapy later but have undetectable viral load. The reason STIs are being tested in this population first is because STIs may be less likely to be effective in individuals with more advanced HIV. And researchers want to test it in people who are more likely to respond to see if it can work. At this year's 7^th Retrovirus conference there were numerous small studies exploring treatment interruptions. This idea received initiative or momentum because many patients were taking therapy interruptions or as many called them "drug holidays". The Berlin Patient and other cases of clinical and more basic research showed a handful of patients who stopped therapy and were able to control their viral load at a low level without HIV therapy. More basic human and animal research showed that in some cases stopping therapy after initiating therapy during acute infection could stimulate an immune response --HIV specific CD4s and/or CTLs-- that appeared associated with the control of HIV viral load. Although the collective results from these studies are preliminary, these study results have so far been mixed. There have been a few reports of some patients who had low viral loads lasting for a period of time following one or more interruptions, and the one report of the Berlin Patient who appeared to have a longer lasting low viral load response. But some researchers have pointed out that this can occur in a proportion of untreated patients. So, questions remain-- is the immune response seen in animals and humans after a therapy interruption really associated with durable clinical effectiveness? What is the longer-term effect from possibly repopulating reservoirs with HIV when viral load goes up during an STI? For an STI to be durably effective maybe an immune-based therapy is required to stimulate the desired immune response? Further study is needed to better understand STIs. Experimenting on your own outside a clinical study may be risky.

Authors: V Miller, C Sabin, K Hertogs, J Martinez-Picado, R D'Aquilla, B Larder, S Bloor, Sturmer, C Rottman, AN Phillips, S Staszewski

Veronica Miller's study is a retrospective analysis of patients who took a treatment interruption for whatever reason. They were not treated with an interruption as part of a strategy. The abstract reports that of 195 treatment interruptions, 169 were first treatment interruptions, 20 second, 4 third, and 2 fourth. 49 were extensively PI pre-treated, 14 were less extensively PI-experienced, and 106 patients were minimally PI experienced. Miller and colleagues previously have reported observing that for treatment experienced patients for whom resistance to HIV drugs was detectable and after a lengthy interruption, resistance could become undetectable in blood for a good proportion of patients in their analysis. Their data suggested that patients would respond better after restarting therapy following the interruption, than if they switched their regimen without a therapy interruption. Here she reports that although there may be an initial virologic response, there is a high rate of virologic rebound in this cohort of patients (73% in the original cohort of 33 patients, and 86% rebounded in the expanded patient group numbering 163). Miler's average reported time to recover CD4 declines experienced during interruption raises concerns.

Granted, most of the patients in these 2 cohorts (149) were first treatment interruptions, and one could postulate that the response rate could improve following several structured interruptions, but questions and doubts were raised at this meeting as well as previous to this meeting about the clinical effectiveness of interruptions. One point raised at this meeting was that the overall response rate in Miller's study, when considering the rebound, is no better and possibly worse than not using an interruption. Immediately switching therapy utilizing resistance testing and/or MEGA-HAART may provide the same response or better. So therefore, why risk the CD4 decline and viral load increase that appears associated with interruptions. At Retrovirus, Steve Deeks reported on his study of 18 patients randomized to discontinue therapy that the median CD4 drop was 94 cells (range -28 to -126), and the median increase in VL was +0.82 log (range -0.34 to -.92 log). Their baseline CD4s averaged 245 and viral load 4.6 log (about 40,000 copies/ml). He also reported that NRTI resistance persisted often at much reduced levels, after reversion to a fully PI susceptible virus in 7 patients. Deeks also reported at Retrovirus that although resistance was undetected in plasma following interruption, resistant virus identical to baseline was cultured from PBMCs 12 to 36 weeks after therapy discontinuation in 4 of 8 patients showing phenotypic reversion. The longer-term implications of this particular finding may not be fully appreciated yet.

On the other hand, patients are interrupting their therapy, or as often put taking "drug holidays", because they want a break from the side effects and toxicities. They also feel they need a break from the difficult demands of maintaining adherence to the daily routine of taking complicated drug regimens. If a person's CD4s are relatively high when they start an STI and their viral load is 1000 copies/ml or undetectable, the risks in interrupting therapy may be less. But I think Miller's and Deek's data on CD4 decline and Miller's data on CD4 recovery reported below suggest that there is a clear risk for the person with relatively low CD4s and higher viral load.

At this meeting, Miller started her presentation by quickly summarizing previous findings on patients with treatment failure:

• Treatment interruptions, at least in treatment failures, were associated with a drop in CD4s and a rise in viral load
• Some patients with drug resistance showed a shift to wild-type as the dominant population. These first 2 observations were also reportedly observed by Steve Deeks in his late-breaker presentation at the 7^th Retrovirus Conference.
• Miller reported observing improved virologic response for patients with a shift in wild-type in their initial response after restarting therapy following the interruption
• Miller said the durability of the response was unclear
• Also, there was concern whether the CD4 count could recover since interruption was associated with a decline in CD4s. The CD4 reduction could be substantial

Miller pointed out that this is a work in progress and today's discussion will focus on further characterization of the shift to wild-type, the durability of the virologic response, and the recovery of CD4s. Today's presentation will include a follow-up from the original group of patients and from an expanded cohort (group) including all patients from the Frankfurt HIV Cohort with viral load >5000 who stopped therapy for a minimum of 2 months while they were on three drugs or more. Miller said the patients could be divided into 3 groups: heavily PI experienced (>2 PIs, 1 year), medium PI experienced (>1 PI, >1 year) , light or no PI experience (<1 year).

Miller showed a chart of one representative patient in whom by 60 days after stopping therapy they observed a complete shift to wild-type (no resistance detected) in all classes of drugs. Sensitivity to three drug classes returned.

The expanded cohort dataset is based on 177 patients with a first STI. They have some resistance data for 72 patients. They have post-STI data for 55: for 43 patients they have pre- and post STI data; for 12 they only have data for post STI; and 5 patients had sensitive virus at pre- and post STI (Miller suggested possibly these 5 were not taking their drugs as prescribed or were having absorbtion problems).

There was evidence of a shift in all three treatment groups in 35/55 (64%) of patients, and this was distributed in all 3 treatment groups regardless of how much PI experience they had. In 4 patients (7%), there was a shift to wild-type for PI and NRTI but not for NNRTI. For 5 patients (9%), there was no shift in any drug class. Interestingly, all of these 5 patients were in the heavily PI-treated group.

• In the original set of 33 responders, the average follow-up was 385 days (range 42-799). And, 73% (24) experienced a viral load rebound in a median of 78 days. Although it wasn't statistically significant there was a suggestion that patients with a shift had a more durable response (rebound rate of 75% for those with a shift and 100% for those without a shift).
• In the expanded set of 165 patients, 103 of 163 (62%) experienced a virologic response. The response was associated with the height of the viral load at the end of the STI and the change of viral load during the STI, or the viral load at the start and end of the STI. However, 86% experienced viral rebound. The rebound was associated with their CD4 count at the start of their STI, the previous CD4 nadir, the number of drugs exposed to, and the number of classes exposed to. The association with resistance is not yet included.

In the original set of 33 patients 74% recovered their CD4s with an estimated time of 251 days, and their was no difference according to shift or no shift. In the expanded set of 163 patients, 75% experienced CD4 recovery to within 90% of baseline in a median of 3 months.

Miller's Conclusions

• Treatment interruptions will happen in patients with treatment failure for a variety of reasons, but should it be a strategy? Miller said we can't answer that yet
• Changes in resistance patterns may be complex and dependent on treatment history
• The effect of resistance and shift to wild-type on durability of response remains to be determined
• The clinical effect of STI remains to be determined
• Several randomized clinical trials are underway to specifically address these questions
• A joint observational database project is in the making comprised of about 8 cohorts with 600-700 patients, looking at the effects of STI in patients with treatment failure