YOU ARE HERE:
Interrupting smartly
   Last updated: 27.10.04
The risks and benefits of interrupting successful HAART, by Edwin J Bernard
In the sixteen months since we last reported on treatment interruptions, reaction to the strategy has swung from excitement – after the February 2003 Retroviruses Conference in Boston heard promising results from the Staccato study1) – to disappointment when the July 2003 IAS Conference in Paris heard from several failed interruption studies, including the week on, week off (WOWO) arm of Staccato, and another WOWO study from the US National Institutes of Health2, which exposed flaws in the concept of fixed short- or long-cycled structured treatment interruptions, primarily due to the emergence of drug resistance.
However, this year there has been renewed interest in the concept of pulse therapy driven by CD4 count (individualised cycles on and off HAART, stopping treatment at a pre-defined CD4 level and re-starting at another, lower CD4 level), due in part to the promising preliminary results from the Italian BASTA trial3 and publicity surrounding the ongoing SMART (Strategies for Management of Anti-Retroviral Therapy) study (the largest-ever clinical endpoint study in HIV medicine), which is currently enrolling in 22 countries worldwide, including the UK.


Why interrupt?
Although it is possible to interrupt treatment at any stage of HIV disease, from primary infection to salvage therapy, this article will focus on treatment interruption in the group of people who are chronically infected with HIV, and whose HIV is currently controlled with HAART.
Proponents of treatment interruption would argue that the main benefits of interrupting successful HAART are a reduction in both the financial and long-term health costs of HAART (by reducing short- and long-term toxicities) and, possibly, a slowing of the emergence of resistance (due to better long-term adherence).
The Medical Research Council’s Dr Adrian Palfreeman, a consultant in GU Medicine at Peterborough Hospital and the lead UK investigator in the SMART study, says, “I have many patients in my clinic who ask, ‘Do I have to take these drugs forever, doctor?’ Right now, I don’t have an answer for them, which is why we’re doing this study.”
While it is true that HAART has revolutionised the impact of HIV in well-resourced nations over the past eight years, it is becoming a real concern that continuous lifelong therapy could possibly be as toxic as HIV in the long term, due to the increased risk of cardiovascular disease and the impact of HAART on the liver, especially in those co-infected with hepatitis C virus.
Relaxed approaches to adherence due to mental health and lifestyle issues, as well as the psychological impact of lipodystrophy, are leading to ‘HAART fatigue’ in a significant minority, who are developing resistance and may well run out of options sooner than new drugs can be developed.
The financial burden of therapy is also a major issue, not only in less wealthy nations, but also here in the UK, where the cost of HIV medications continues to be an issue.
Adding to the pro-interruption argument, a recently published French study4 examining the ongoing benefits of HAART over five years found that increases in CD4 counts level off after 18 months of maximally suppressive HAART, and that there is no additional action against HIV in viral reservoirs after three years of continuous treatment, leading the authors to “question the benefits of a life-long treatment for HIV infection”.


What are the risks to the CD4 count?
Treatment interruption during chronic infection almost always leads to viral load rebound. Viral load rebound raises several potential risks – CD4 cell decline and illness, failure to re-suppress HIV when treatment is re-started, and resistance.
Many observational and randomised studies have now been presented concerning the effects of treatment interruption on the CD4 count. The most convincing data come from the 64 week results of the BASTA study, which corroborated previous findings that the lower the nadir (the lowest-ever CD4 count), the faster the CD4 loss during interruption, and the more likely that therapy would need to be re-started sooner.
In this study, 69 people with a viral load below 50 copies/ml and an on-therapy CD4 cell count above 800 cells/mm3 were randomised either to stop their therapy or continue with ongoing treatment. Those in the treatment interruption arm would then restart when their CD4 cell counts fell to below 400 cells/mm3, which is the low end of the normal range of CD4 counts in healthy adults, and stop again when they reached 800 cells/mm3.
Participants who had a lowest-ever CD4 cell count below 200 cells/mm3 spent only an average of 6.9 months (and a maximum of ten months) off treatment before resuming HAART. Those whose lowest-ever CD4 cell count was between 200 – 350 cells/mm3 were off treatment for a significantly longer time – an average of 14.1 months. The longest treatment interruption, an average of 17.8 months, was possible in those people whose lowest-ever CD4 cell count was between 351 and 500 cells/mm3. Those individuals whose lowest-ever CD4 cell count was above 500 cells/mm3 did not need to resume therapy at all within the 64 week follow-up period.
The rate of CD4 cell loss also differed significantly based on lowest-ever CD4 cell count. Those with a nadir CD4 cell count below 200 cells/mm3 lost, on average, 366 cells/mm3 in the first four weeks, whereas the other three groups (200- 350, 351-500, and above 500 cells/mm3) only lost an average of 144, 179 and 137 cells/mm3, respectively, in the same period. However, the rapid CD4 loss levelled off in all groups after the first month.
Still, given the rapid drop in CD4 numbers for those with a CD4 nadir below 200 cells/mm3, there are concerns that people with a low CD4 nadir might not do as well in the SMART study. Here, people who are randomised to the treatment interruption arm (termed “the drug conservation arm”) stop HAART when their CD4 count exceeds 350 cells/mm3 – although many may have much higher levels – and restart when it falls below 250 cells/mm3, compared with 800 cells/mm3 and 400 cells/mm3 in BASTA. It requires a certain leap of faith for SMART participants to trust that falling to below 250 cells/mm3 will not lead to short- or long-term clinical disease progression. However, according to Dr Palfreeman, “We know that you’re not going to get sick until your count falls a good 100 cells below the 250 that we’re going to re-initiate on.”
Two years into the study, the DSMB (Data and Safety Monitoring Board) have reviewed the data so far “and although a quarter of the patients who have gone into this study so far had a CD4 count below 200 or prior AIDS, there were no safety concerns about rapid re-initiation of therapy in the drug conservation arm,” notes Dr Palfreeman. “I find that reassuring.”
One of the major differences between prior treatment interruption studies and SMART is that, until now, CD4 counts have been used as surrogate markers of how well the immune system is doing, whereas SMART, which will last up to nine years, will look at the actual rates of illness or disease progression. “My biggest criticism of BASTA, or indeed any study that looks at CD4 counts, is that CD4 counts are only a surrogate for clinical progression,” says Dr Palfreeman. “Clinical endpoints are really what matter; you are never going to find out from any surrogate marker anything as relevant as what happens to individuals in terms of their health. My patients are less interested in their lab results, and more interested in knowing if they will be well, and un-scarred by lipodystrophy.”


What are the risks of getting sick?
One effect of a rising viral load may be the emergence of acute retroviral syndrome – which can cause unpleasant, flu-like symptoms as HIV replication rebounds – or the re-emergence of pre-HAART illnesses. Last year, Pavie and colleagues5 reported on 30 people who interrupted therapy. One case of acute retroviral syndrome, two cases of herpes zoster, one case of herpes simplex, one case of candidiasis, two cases of low platelet counts and three cases of elevated liver function tests occurred. Low nadir CD4 count and shorter duration of suppressed virus were associated with a poorer response to treatment interruption.
Since there is also a risk of AIDS-defining illnesses if CD4 counts are below 200 cells/mm3 when treatment is interrupted, re-starting preventative treatments for opportunistic infections (e.g. Septrin to prevent PCP) would be wise in these circumstances.
However, treatment interruption also carries an element of risk of illness even among people with relatively high CD4 counts. An Italian group has reported that unsupervised treatment interruptions were associated with a greater risk of HIV disease progression or death. The ICONA cohort6 involved 2832 people on first-line treatment, of whom 553 people interrupted treatment for at least 12 weeks over approximately three years. Average CD4 count at the time of interrupting was 528 cells/mm3 and average viral load was 400 copies/ml. The study found that the likelihood of illness and/or HIV-related conditions was almost five times more likely during a break from treatment than during treatment.


What are the risks of resistance?
A further potential hazard of any treatment interruption is that stopping and starting regimens could result in drug resistance. The risk appears to be influenced by the drugs being taken at the time of interruption.
There have been particular concerns about the safety of interrupting drugs with long half-lives and/or those that require just one mutation in the virus to cause high-level resistance (this is known as having a low genetic barrier to resistance). This includes the entire NNRTI class – efavirenz, nevirapine and delavirdine – as well as the NRTIs, 3TC (lamivudine) and FTC (emtricitabine).
A recent Spanish study7 found that individuals who were interrupting an NNRTI-containing regimen were significantly more likely to experience virological rebound and develop resistance mutations than individuals who took a structured break from HAART based around a protease inhibitor (PI). In total, 21% (19 were on PI-based, and 20 on NNRTI-based, HAART) experienced a rebound in their viral load to above 10,000 copies/ml when they took a break from HAART. However, whereas all the participants who resumed PI-based HAART saw their viral load fall back to below 50 copies/ml, among the 29 participants who restarted HAART using an NNRTI-based regimen, 16 (55%) failed to achieve a viral load below 50 copies/ml. Genotypic analysis in ten of the individuals failing NNRTI therapy showed that one had wild-type virus, one had an NNRTI-associated resistance mutation, and the remaining eight had NRTI resistance.
Recently, more information has become available regarding the variability of drug levels when interrupting therapy8,9. Although guidelines have not yet been produced regarding the ideal scenario for stopping HAART, the latest information provided by the SMART study organisers provides several alternatives.

  • Those on NNRTI-based HAART with a suppressed viral load at the time of planned discontinuation: replace the NNRTI with a boosted PI three weeks prior to stopping, preferably guided by therapeutic drug level monitoring or stop the NNRTI an average of seven days prior to the NRTIs, “although the interval should probably be longer (10-21 days) for Hispanics and Blacks.”


  • Individuals on PI-based HAART or triple NRTI therapy are advised to stop all drugs at the same time. It is considered reasonable for those on NRTI-sparing HAART to stop the NNRTI first, although if there is concern about potency, intensification with another drug should take place after stopping the NNRTI.


  • Those with a detectable viral load are expected to have acquired resistance on therapy, and so stopping all drugs at the same time, regardless of drug class, is suggested.


  • Additionally, since there are concerns about the long half-lives of 3TC and FTC, replacing them with another NRTI seven days prior to interrupting therapy ought to be considered.


“I think the protocol is written in such a way that we are not going to be encouraging people to stop their antiretroviral treatment without a safe strategy,” says Dr Palfreeman. “It is better to stop when the virus is maximally suppressed, and take into account the different half-lives of the drugs. All we can do is to be guided by the current data and see what happens and try and protect our patients the best we can. And I think the risks are small, especially compared with trying to keep people on treatment for nine years and hoping they don’t get resistance.”



Will I be undetectable again when I restart?
Another crucial issue is the extent to which people can re-suppress viral load after a treatment interruption. Current evidence seems to suggest that people who have previously had an undetectable viral load on treatment will be able to re-suppress virus when they restart.
The BASTA study found that everyone who restarted therapy achieved undetectable viral loads within 24 weeks and experienced a rapid increase in their CD4 cell counts. Data from the ATHENA cohort also suggests that a break from therapy does not undermine a person’s ability to suppress virus.
However, not everyone who takes a break from treatment will re-suppress HIV to pre-interruption levels, especially if their viral load is not completely controlled prior to stopping. An observational study of treatment interruptions10 found that although the majority of individuals who resumed HAART achieved virus suppression similar to their pre-interruption levels, two out of 15 had not achieved their prior less than 50 copies/ml level by the time the study was written up. Of those with a pre-interruption viral load of 50-400 copies/ml, 13 out of 14 achieved re-suppression to less than 400 copies/ml. However, of the 16 with a pre-interruption viral load above 400 copies/ml, only eight re-suppressed to below 400 copies/ml, although five of them reached below 50 copies/ml.


What about safer sex?
A final concern regarding treatment interruptions is the possible increased risk of HIV transmission. As in blood, viral load in semen usually rebounds when treatment is stopped11, and there has been a case report in which an HIV-positive man who had been having unprotected sex with his HIV-negative partner for over a year infected his partner shortly after taking a treatment interruption12. Additionally, evidence is mounting that sexually transmitted superinfection – the re-infection of an HIV-positive person with a slightly different version or strain of HIV – appears to occur only in people not on antiretroviral therapy.


Make an informed decision
At the moment, treatment interruptions during successful HAART are not recommended outside of clinical trials. Dr Palfreeman notes, however, that in the real world, people are impatient to know the results of SMART, and some individuals will always want or need to stop their therapy. In those situations, he suggests that both individual and clinician “make an informed decision. I’ve had lots of patients who have interrupted for reasons of toxicity or because they were just completely fed up with HAART. And in that situation, what I’ve done as a clinician is try to help them stop in the most sensible way possible.”


Key conclusions
  • Structured treatment interruption is not currently recommended as a routine HIV treatment strategy

  • CD4 counts fall faster during an interruption if you have ever had a CD4 count below 200 cells/mm3.

  • The risk of acquiring drug resistance is higher if you are stopping NNRTIs and/or 3TC, although there are strategies to reduce the risk.

  • Experimenting outside research settings, and particularly without the knowledge of your doctor, is unsafe.

  • Studies of structured treatment interruption provide individuals with intensive monitoring while off treatment.

  • If you wish to interrupt your treatment outside of a study, it is advisable to consult your doctor, have viral load and CD4 tests before stopping the drugs and monitor both markers carefully during the break.



References
1. Ananworanich J. 10th CROI, Boston, abstract 64, 2003.
2. Dybul M et al. J Infect Dis. 188 (3): 388-96, 2003.
3. Maggiolo et al. AIDS 18: 439-446, 2004.
4. Viard et al. AIDS 18: 45-49, 2004.
5. Pavie J et al. 2nd IAS Conference, Paris, abstract 611, 2003.
6. d’Arminio Monforte A et al. 2nd IAS Conference, Paris, abstract 145, 2003.
7. Barreiro PM et al. 43rd ICAAC, abstract H-857, 2003.
8. Taylor S et al. 11th CROI, San Francisco, abstract 131, 2004.
9. Mackie NE et al. HIV Medicine 5, 180-184, 2004.
10. Tarwater PM et al. Clinical Infectious Diseases 37: 1541-8, 2003.
11. Liuzzi C et al. AIDS 17 (7): 1089-1092, 2003.
12. Tubiana R et al. AIDS 16 (7): 1083-1084, 2002.