When to start and stop?

New American and European guidelines for initiating antiretroviral therapy in chronic asymptomatic HIV infection are converging. Previous thinking that therapy should be initiated early in the course of infection has shifted to an apparent consensus for recommending initiation of therapy once the CD4 lymphocyte count decreases to consistently below 350±300 (and some would say 250±200) cells/mL. Also, the plasma viral load threshold for when to initiate has increased from 5000 to 30 000±100 000 HIV-1 RNA copies/mL. A wise man once told me that the amount of data required to change guidelines should be signi®cantly more substantive than the data used to form the guidelines in the ®rst place. This paradigm has certainly been ful®lled for HIV therapy guidelines. Back in 1996, when potent antiretroviral therapy was introduced, we knew very little about how best to use it strategically. There are certainly many holes in our understanding still, but since that time new data and new interpretations have emerged. Studies in untreated patient populations followed for several years told us that the plasma viral load and the CD4 lymphocyte count at a given point in time predicted the risk of developing an HIV-related disease 3 years later [1]. No clear threshold was identi®ed and the observations were extrapolated to say that therapy should be initiated early in the course of chronic asymptomatic HIV infection. Although studies like this made important contributions to our understanding of the dynamic interaction between the plasma viral load and the CD4 lymphocyte count, they were also potentially misleading in addressing the question of when to start. In a clinical situation, you would never envision a situation in which blood tests were taken 3 years apart in order to determine when to start therapy. Rather, you would do it regularly, every 3 months. By doing so, you would identify and initiate therapy for patients with the most rapid decline in CD4 lymphocyte count and, conversely, leave those with a stable and relatively high CD4 cell count without therapy. The plasma viral load remains relatively stable in individual asymptomatic patients over long periods of time (median increase of around 0.1 log10 HIV-1 RNA copies/mL per year in untreated patients). Hence, for most patients (except those who display a consistent and marked increase), it is not logical to use the plasma viral load alone as a marker for initiating therapy. Conversely, this measure can be used to predict the subsequent risk of CD4 cell decline and therefore identify patients for whom stringent follow-up is advisable. Applying this way of thinking allows you to decrease the CD4 count level for initiating therapy without placing patients at an excess risk of contracting HIV-related diseases. Emerging concerns about late-onset metabolic adverse events for several of the drug classes has also led to a more conservative approach to initiating therapy [2]. In fact, for many clinicians this concern may be the decisive factor in delaying therapy. Finally, concerns about whether ef®cacy diminishes with HIV progression have been addressed recently. The data available suggest that this concern is not warranted as long as the CD4 count remains above 200 cells/mL and the plasma viral load below 100 000 copies/ mL [3]. Our increased scienti®c understanding of when to start therapy raises a number of questions. What therapeutic strategy should be used in a patient who began therapy at a higher CD4 cell count than is currently recommended? Should such a patient continue or stop therapy? and is a CD4 count of 200 cells/mL while on therapy really enough? Care is needed when extrapolating knowledge and recommendations from drug-naive patients to patients currently on therapy. Structured treatment interruption (STI) is a possible strategy, but we do not yet know the full scale of the downside of STI. This downside could include the risk of acute HIV±like syndromes, clinical disease especially if the CD4 cell count plummets, lack of immune reconstitution and inability to regain virological control and antiretroviral resistance once therapy is reinitiated. An important clinical lesson so far is that, in STI, the most rapid decline in CD4 cell count occurs in those who have experienced the largest increase in the CD4 count while on therapy. So, the better the therapy has worked the more you lose by stopping it. Clearly, what we need now is to complete the ongoing randomized studies of large patient series addressing STI before this strategy is widely recommended. Contrary to the guidelines for starting and stopping antiretroviral therapy, those dealing with starting and stopping disease-speci®c prophylaxis have been less controversial. Revised guidelines have been issued recently [4], expanding on the number of situations where disease speci®c prophylaxis can be safely stopped. A key concern has been the effectiveness of the immune reconstitution Correspondence: Dr Jens D Lundgren, Department of Infectious Diseases, Hvidovre Hospital, 2650 Hvidovre, Copenhagen, Denmark. E-mail: j.d.lundgren@inet.uni2.dk