HIV treatment interruptions: a review.

INTRODUCTION

The widespread use of highly active antiretroviral therapy (HAART) has bad a significant impact on HIV by reducing mortality and morbidity. (1) Unfortunately, long-term use of antiretroviral treatment is associated with several limitations and drawbacks. Currently available antiretrovirals are unable to completely eradicate the HIV infection, and infectious virus continues to reside and to replicate latently in reservoirs. As a result, most patients will be on antiretroviral therapy for the rest of' their lives, a situation that could potentially cause severe side effects and drug toxicity. These include, but are certainly not limited to, high triglyceride and cholesterol levels, insulin resistance, diabetes, and redistribution of body fat--many of which increase the likelihood of suffering a cardiac event. Adherence is difficult because of the daily, long-term drug dosing requirements and incidence of side effects, and there is a continuous risk of an emergence of drug-resistant HIV. (2,3) Finally, because of the extremely high cost of antiretroviral treatment, 95% of HIV-infected individuals worldwide do not have access to HAART. (4)

Because of these limitations, HIV researchers are exploring alternative treatment strategies, such as "structured treatment interruptions" (STIs), by using a variety of dosing schedules, antiretroviral regimens, and patient populations. In some cases, treatment is interrupted according to a scheduled time frame, while other studies have used CD4 T cell counts or viral load as indicators of when to stop or re-initiate therapy. Alternatively, one research group has investigated the novel idea of continuous therapy, but with alternating HAART regimens. (5)

Still other researchers have simply documented the effects of treatment interruption on virologic and immunologic control from case studies whereby patients discontinued treatment temporarily for various reasons (eg, side effects, other health concerns, cost of drug, or poor compliance).

There are 3 main applications for the potential utility of STIs. First, STIs may enable a patient to receive less total drug, resulting in decreased drug-related toxicity, reduced cost of patient health management, and improved quality of life for the patient. Second, many studies have explored the concept of "autovaccination" or "autoimmunization" in conjunction with treatment interruption. Though antiretroviral therapy has enabled immunologic control of HIV and viral suppression, the use of HAART is associated with decreased HIV-specific immune responses, particularly in patients with chronic HIV infection. (6,7) Once treatment is interrupted, viral load increases shortly afterward. (8-10) Researchers hypothesized that in patients who maintained successful viral suppression with HAART for a sustained period of time, controlled and limited interruption of treatment would enable short bursts of autologous virus to boost HIV-specific immune responses, potentially resetting the viral set-point to a lower level. Repeating these on-treatment and off-treatment cycles could possibly lead to better control of the virus by the patient's own immune system, even allowing the infected individual to suppress virus in the absence of antiretroviral therapy. (2, 11) Third, STIs may be useful and beneficial in a salvage therapy regimen whereby HIV has become multi-drug resistant and patients have limited treatment options. Here, an STI may allow for the repopulation of wild-type, drug-susceptible virus, therefore improving the chances of successful salvage therapy after the interruption. (3, 12, 13)

Numerous studies have investigated treatment interruptions as an alternative approach to HIV therapy. This article will present an overview of the various strategies researchers are employing to alter the HIV therapeutic landscape with the use of treatment interruptions, and the effect these interruptions have on immunologic and virologic control of HIV. However, the benefits of STI are questionable and sometimes difficult to interpret. Comparison between studies is complicated because of varying patient populations with different treatment, immunologic, and virologic histories and characteristics. STI duration and design are discrepant: in particular, the triggers for re-initiating therapy (predetermined schedule, viral load rebound, or drop in CD4 T cell count). In addition, definitions of viral suppression and virologic failure are variable.

TREATMENT INTERRUPTION IN PATIENTS WITH PHI AND TREATMENT-NAIVE PATIENTS

Patients treated with HAART during primary HIV infection (PHI) provide a unique opportunity to examine the effects of STIs on immunologic and virologic control and HIV dynamics, particularly the viral set-point. Data suggest that early antiretroviral therapy is beneficial in patients diagnosed during PHI; however, this early and sustained treatment may lead to health problems caused by the long-term drug exposure. Controlled treatment interruptions in tiffs patient population may be a feasible alternative to minimize drug-related toxicity. Additionally, a hypothetical STI-induced autoimmunization achieved during PHI could potentially lead to an enhanced HIV-specific immune response, lowered viral setpoint, and optimal viral control even in the absence of continuous antiretroviral treatment. (2,11)

Several small studies have shown that interruption of antiretroviral treatment in this population is not detrimental, though it is uncertain if any benefits are derived. Whether treatment interruptions afford more than just a break from treatment by actually altering the dynamics of HIV and lowering the viral set-point is unclear. For patients treated during PHI, data from a variety of studies and case reports suggest that viral suppression can be maintained for at least 6 months, and in some cases, for up to 2 years after treatment interruption, (11, 14-17) though HIV RNA was still detectable, indicating that the virus had not been eradicated. (16) Studies of STIs in patients treated with HAART during early infection, rather than during the acute PHI phase, have also observed viral suppression for 1 to 2 years after treatment interruption. (18-19) In addition to sustained viral suppression, robust and broad HIV-specific immune responses that were positively associated with durable viral suppression have been detected in a small number of patients following an STI. (16, 19) Indeed, one small study observed plasma viral rebounds that became smaller and smaller with each successive interruption. (11)

HIV-infected patients who have previously been antiretroviral treatment naive also appear to suffer no harm from an interruption in treatment. An early report examined the effect of a single, 28-day treatment interruption after patients were treated for 28 days to investigate the effects of an occasional unscheduled interruption. (7) Though viral load increased once treatment was interrupted, it quickly returned to baseline levels once treatment was re-initiated, and no drug-resistance mutations were detected. Moreover, in a similar patient population, Lori and colleagues (20) observed an increase in the time to rebound after each successive treatment interruption. In fact, a lower viral set-point was induced and maintained for several months in 2 patients.

TREATMENT INTERRUPTION IN PATIENTS WITH CHRONIC HIV INFECTION

The benefits of STIs, including the potential to induce autovaccination, have been studied extensively in patients chronically infected with HIV. Initially, researchers hypothesized that asymptomatic, chronically HIV-infected individuals who had good, long-term immunologic and virologic control under HAART could experience continued suppression in the absence of antiretroviral therapy. Unfortunately, most data have not supported this concept. Some studies have reported such viral suppression, but only in a small percentage of patients. (21-23) Thus far, only Dybul and colleagues, (24) using an STI schedule of 7 days on treatment followed by 7 days offtreatment, observed sustained viral suppression (HIV RNA less than 500 copies/mL) for up to 68 weeks in all 8 patients who remained on study. In addition, no newly emergent drug-resistance mutations were detected, and patients experienced significant decreases in cholesterol and triglyceride levels. While these findings suggested that shorter treatment interruptions may be optimal because the virus may not have an opportunity to rebound, the randomized Staccato study failed to replicate these results using the same STI design. (22) Though patients were successfully treated with HAART prior to treatment interruption, over half of the patients who participated in the treatment-interruption arm experienced virologic failure. Consequently, that arm of the study was prematurely terminated.

Several larger studies have examined an STI schedule that involves interrupting current HAART therapy for 2 weeks, resuming HAART for 8 weeks, and stopping treatment at week 40 until re-initiation of treatment becomes necessary. (6,21,25) In contrast to patients with PHI, (11,16) patients chronically infected with HIV did not experience an autovaccination event in a manner that enhanced viral control or lowered the viral set-point. (6,21) However, a small percentage of patients maintained viral suppression (defined as viral load less than 5,000 copies/mL) for up to 3 months (17%) and 1 year (8%) after stopping treatment, and only 1 patient developed drug resistance that required salvage therapy, (21) though viral rebound (HIV RNA greater than 100 copies/mL) was detected within 8 days in the majority of patients. (25) Predictors of response were low pre-HAART viral load and no or few incidences of viral rebound during first 40 weeks of the study. (21)

Longer cycles of intermittent HAART show no benefit in terms of immune control or a reduction in adverse events when patients are off treatment for 4 weeks followed by 8 weeks of HAART. (26) In fact, 3 patients in the STI arm showed evidence of newly emergent drug resistance, a finding that has been reported by other research groups. (27) However, others found that drug-resistance mutations detected during STIs were intermittently present and not persistent. Moreover, detection of these mutations did not predict failure to resuppress virus once treatment was re-initiated with the same antiretroviral agents. (28)

In addition to interrupting treatment based on a prescheduled timeline, studies have also tested the idea of stopping and starting therapy based on CD4 T cell counts and viral load in patients chronically infected with HIV. As such, patients would stop treatment once their CD4 T cell counts reached a predetermined value and restart therapy once that value dropped to a specified level. The same principles are applied to STIs based on viral load; interrupting treatment once suppression is achieved and re-initiating treatment once viral rebound occurs. One of the largest trials in the history of any infections disease, known as the Strategies for the Management of Anti-Retroviral Therapy (SMART) study, was designed by investigators with the Community Programs for Clinical Research on AIDS (CPCRA) and will compare a strategy of delayed, episodic HIV therapy against one of immediate, uninterrupted therapy in patients who are treatment naive and treatment experienced. The study will follow 6000 HIV-infected subjects for as long as 8 years. The episodic treatment arm (also referred to as the "wait group") involves stopping or deferring treatment until CD4 T cell counts fall below 250 cells/m[m.sup.3], at which point subjects will initiate treatment to increase CD4 counts to at least 350 cells/m[m.sup.3] for 2 consecutive visits, after which treatment would be interrupted again. In the other study arm (referred to as the "go group"), subjects immediately begin therapy regardless of CD4 T cell count and remain on antiretroviral therapy (changing regimens as needed to suppress virus) for the duration of the trial. (29)

Other smaller studies have already reported results using these types of strategies in patients with chronic HIV infection. The Staccato study also investigated a CD4-guided STI and preliminary data show that no patient has failed this treatment arm to date. (22) In contrast, another study detected viral rebound in patients within 2 to 3 weeks after interrupting treatment based on either CD4 T cell counts or viral load, though the patients had previously maintained viral suppression for more than 2 years with HAART. (30) However, the definition of viral rebound differs dramatically from study to study. For example, the Staccato study (22) defined virologic failure as HIV RNA greater than 500 copies/mL, while virologic failure was defined as HIV RNA greater than 50 copies/mL in the study conducted by Davey and colleagues. (30) Other studies using viral load as an indicator of treatment re-initiation detected viral rebound within a few days to 1 month after interrupting treatment in the majority of patients. However, these levels became undetectable once treatment was re-initiated, (23,31,32) and drug-resistance mutations were rarely detected. (31,32) Observations from patient case studies have reported similar findings, observing rapid viral rebound that quickly drops to undetectable levels once treatment is restarted. (9) Importantly, one study reported that each consecutive STI was associated with more control of viral replication. Moreover, an increase in HIV-specific CD8 T cell responses was also detected, but only in about one-third of patients. (23)

TREATMENT INTERRUPTION IN MULTI-DRUG-EXPERIENCED PATIENTS

Treating multi-drug-experienced patients is challenging because many of these patients have exhausted all treatment options as a result of broad cross-resistance. Continuing suboptimal treatment may maintain CD4 T cell counts but also allow the emergence of additional drug-resistance mutations, further limiting treatment options. The concept of STIs in this patient population stems from the hypothesis that interrupting treatment may enable a rebound in wild-type, drug-susceptible virus and perhaps elicit an HIV-specific immune response. As a result, the patient would experience enhanced virologic suppression once a salvage regimen was started after the interruption. Temporary reversal of drug-resistance mutations can occur and, while transient, may provide an opportunity for salvage therapy to work with some increased efficacy. Further, it was anticipated that immunologic control could be maintained for brief periods without treatment because of preserved CD4 T cell increases from previous therapy. (3, 12)

One large, randomized trial (CPCRA 064) examined the effects of a 4-month STI followed by the initiation of an optimized regimen based on the presence of drug-resistance mutations in individual pa- tients. (33) Though 64% of patients in the STI group experienced a shift to wild-type virus by 4 months, there were significantly more cases of disease progression in the STI group, compared to the control group. CD4 T cell counts decreased during interruption, and though they recovered on therapy, values were consistently lower in the STI group during follow-up, compared to the control group. Similarly, the randomized Retrogene study examined the efficacy of combining several drugs, commonly referred to as "mega highly active antiretroviral therapy" or "mega-HAART," after a 3-month STI. (34) Complete reversion to wild-type HIV was detected in 35% of patients in the STI group, though reversion was not associated with viral suppression in this study.

In contrast, other smaller studies have reported positive effects of STIs prior to initiating salvage therapy. (10,13,35) A multiple-drug (approximately 8 drugs) salvage regimen, referred to as GIGHAART, initiated after an 8-week STI demonstrated a significant virologic and immunologic benefit up to 48 weeks. (35) Moreover, the inclusion of an agent in a salvage regimen to which the subjects' virus was still susceptible was beneficial and led to a durable virologic response. (10,13) These data support the idea that patients with drug-resistant HIV could benefit from an STI prior to salvage therapy, particularly if a drug to which they are not resistant is included. Another report found that viral suppression after an STI was associated with decreased viral load at the time of treatment interruption, inclusion of new drugs when re-initiating treatment, and fewer drugs to which the patient had reduced susceptibility. (36) Moreover, a shift to wild-type virus was associated with subsequent viral suppression, though patients still experienced viral rebound and a drop in CD4 T cell levels once treatment was interrupted. (36)

Why the GIGHAART study and the studies by Deeks and colleagues reported positive effects from STIs while other studies (33,34) found no real benefit to STIs (and even detrimental effects) is not quite clear: Choice of salvage regimen and the number of drugs included in the salvage regimen could be factors. In addition, duration of STI and degree of immunosuppression in the study sample may be responsible. (12)

ALTERNATING HAART REGIMENS--A DIFFERENT TYPE OF INTERRUPTION

Recently, a strategy of proactively alternating regimens while HIV is suppressed has been explored. (5) Typically, an antiretroviral regimen is changed only after it fails. Alternating treatment before it fails may provide a way to reduce viral replication capacity and accumulation of drug-resistance mutations. Older studies performed prior to the introduction of HAART studied the effects of alternating mono- or dual-therapy, but none examined this strategy with HAART. In the SWATCH study (SWitching Antiviral Therapy Combination against HIV-1), treatment-naive subjects alternated two 3-drug regimens every 3 months with continuous administration of HAART (see Figure on page 10). Researchers found that virologic failure was significantly delayed in patients who alternated treatment, compared to those subjects who received the same regimen consistently. These findings suggest that this strategy may be beneficial in HIV-infected patients, though it should be noted that this study included a somewhat small sample size and improved regimens are now available compared to the older drug regimens used in this study.

TREATMENT INTERRUPTION IN THE SETTING OF IMMUNE-BASED THERAPY

The addition of various immune-modulating agents has also been investigated in conjunction with treatment interruption. Recombinant human granulocyte macrophage-colony stimulating factor (GMCSF) was initially developed to treat chemotherapy-induced neutropenia. Research suggests that GMCSF may stimulate immune responses and improve viral control in HIV patients. When given to chronically HIV-infected patients during an STI, GM-CSF blunted viral rebound and substantially prevented decreases in CD4 T cells in the absence of antiretroviral treatment. (37)

Other studies have looked at STIs with interleukin2 (IL-2). In particular, the combination of exogenous interleukin-2 (IL-2) and HAART is associated with a reduction in the number of latently infected CD4 T cells. (38) IL-2 may assist in altering the dynamics of HIV during an STI by increasing the HIV-specific immune response. Moreover, patients with a history of IL-2 therapy may also derive benefits from an STI, compared to patients who have never received exogenous IL-2. (24) However, studies that have investigated the effect of IL-2 therapy in conjunction with an STI have shown no substantial benefit. Indeed, all such patients experienced viral rebound within 2 to 3 weeks of treatment interruption, regardless of whether they had previously received IL-2. (30) When combined with HAART, no significant differences were detected between patients who received IL-2 versus those patients who received HAART alone, in terms of eliciting an HIV-specific immune response (23) or in terms of persistent viral suppression after treatment discontinuation. (39)

The addition of hydroxyurea (HU, also known as hydroxycarbamide) has provided some benefit to patients when given in conjunction with an STI. HU possesses antiviral and cytostatic effects (40) and may blunt viral rebound during STIs. (37) The combination of IL-2 + HU + HAART prior to treatment interruption in patients with PHI was significantly associated with virologic suppression (HIV RNA less than 5000 copies/mL). (17) Case reports of patients recently infected with HIV who were treated with antiretroviral therapy combined with HU have also reported viral suppression for a year of longer following treatment interruption. (16,18) HU was associated with persistent control of HIV in a small percentage of patients with PHI in one study. (41) Importantly, 2 treatment-naive patients experienced an increase in the time to rebound after each successive treatment interruption and the induction and maintenance of a lower viral set-point for several months. (20) The addition of HU in chronically HIV-infected patients led to more cases of viral suppression, compared to patients who received HAART alone. (40) Both groups experienced increased immune responses, regardless of treatment arm, indicating that HU does not have a substantial effect on HIV-specific immune responses. However, patients in the HU group achieved a lower peak viral load rebound and lower viral load set-point.

Taken together, these findings suggest that many of these agents may be beneficial in maintaining viral suppression and perhaps lowering the viral set-point in the absence of antiretroviral therapy. However, these studies tend to have a small sample sizes and response to treatment occurred in an even smaller number of patients. In addition, many of these agents are associated with frequent, and sometimes severe, side effects. Treatment with IL-2 caused a temporary flu-like syndrome, (23) and pain, redness, and swelling were observed in the majority of patients at the sites of GM-CSF injection. (37) More serious events, such as diarrhea, hypotension, malaise, pharyngitis, periorbital edema, and back pain were also observed in patients who received GM-CSF. (37) In one study investigating the effect of HU, over half of the subjects stopped taking HU because of intolerable side effects, including peripheral neuropathy and oral ulcerations. (41)

FACTORS ASSOCIATED WITH RESPONSE TO STIs

Why some studies report benefits with STIs, while others report detrimental effects is not clear. Patient population plays a large part in this outcome and as discussed in an editorial by Aiuti and Giovannetti, (3) patients with good immunologic and virologic control, who were treatment naive before starting HAART and who are currently being treated with effective drug combinations, will probably have the best outcomes with STIs. Patients with PHI who undergo treatment interruptions fare far better than those who are multi-drug resistant or who have been chronically infected with HIV for years. Nevertheless, identifying factors that influence immunologic or virologic control during an STI would provide a means to predict whether treatment interruptions would be beneficial to specific patients.

Obvious predictive factors include pretreatment CD4 T cell nadir, plasma HIV RNA levels, and presence of drug-resistance mutations. Pretreatment CD4 T cell nadir correlates with the ability to suppress virus during an STI. (3,14) In the BASTA study, patients with high CD4 T cell nadirs remained on STIs longer than those with low CD4 T cell nadirs. The researchers concluded that STIs may be safe in patients with CD4 T cell nadirs greater than 500 cells/[mm.sup.3] and that patients with low CD4 T cell nadirs (less than 200 cells/[mm.sup.3]) are at high-risk and should not participate in any type of treatment interruption. (42) In fact, the rate of viral rebound is increased in patients with advanced HIV infection (defined on the basis of low CD4 T cell counts). (43) Plasma HIV RNA level is also a critical factor in determining how well an individual will be able to suppress virus once treatment is discontinued, (3,14) as well as presence of drug-resistance mutations prior to STI. (24)

Other, not-so-obvious factors that are associated with virologic suppression during STIs include levels of endogenous IL-15 and tumor necrosis factor-[alpha], (but not levels of endogenous IL-2, IL-7, or interferon-[alpha]). (44) An abstract at the 10th Conference on Retroviruses and Opportunistic Infections presented data showing that lower pretreatment HIV-envelope nucleotide sequence diversity and amino acid diversity were associated with improved viral control after an STI. (45) Finally, exogenous factors such as prior antiretroviral regimens (3) and the specific HAART regimen used during the re-initiation component of the STI strategy also appear to be important. For example, in the Staccato study, only 1 out of 8 patients taking an efavirenz (Sustiva)-based HAART regimen experienced virologic failure. (22)

CONCLUSION

While an abundance of studies have examined the concept of treatment interruption in HIV-infected patients with various degrees of immunosuppression and disease state, drawing any conclusions is virtually impossible except that the STI strategy is still experimental and could be dangerous in some patient populations. With the exception of studies conducted in multi-drug-experienced patients, most studies tend to be performed in low-risk patients with good immunologic and virologic control and few drug-resistance mutations. In addition, the majority of studies include small sample sizes.

In patients with PHI or who are treatment naive, STIs apparently are not detrimental and may alter the dynamics of HIV, potentially lowering the viral set-point in a select group of patients. However, this has not been shown definitively and though no obvious detriment was detected in the short term, the long-term effects of not suppressing virus fully and continually are not known and may affect clinical outcome and survival.

Overall, data from studies conducted in patients with chronic HIV infection suggest that while they experience viral rebound quickly alter treatment interruption, suppression often occurs quite easily after treatment is resumed, thus suggesting no detriment to immunologic or virologic control. However, STIs have not been shown as beneficial in this patient population because most studies, with the exception of Dybul and colleagues, (24) report sustained viral suppression only in a minority of patients, if any. Possible reasons for the positive findings observed by Dybul and colleagues, (24) include different HAART study regimens, shorter treatment interruption, absence of drug-resistance mutations prior to study enrollment, higher CD4 T cell nadir, and history of IL-2 treatment in the majority of patients (patients in other comparable studies had no history of IL-2). (25) It must be stressed that several reports show STIs to be detrimental in patients with chronic HIV infection. (22,26,27) The data are also mixed regarding multi-drug-experienced patients; however, the CPCRA 064 study observed significantly more cases of disease progression in the STI group, compared to the control group. (33) Importantly, data suggest that increased viremia experienced during an STI potentially increases the risk of HIV transmission, (46) a situation that could lead to substantial increases in HIV transmission if STIs became a routine part of clinical care.

Though the concept of STI-induced autovaccination appears to have little merit any more (see page 16), the idea of reduced total drug exposure in the absence of large drops in CD4 T cell count or unrecoverable viral rebound is still attractive as a way to reduce drug toxicity and the prohibitive costs of antiretroviral treatment. However, the assumption that less drug will automatically result in less toxicity, lower costs, and improved adherence has not been validated. (2) After more than 5 years of STI research, many questions still remain. However, one thing is certain--antiretroviral treatment, though beneficial and potentially lifesaving, is only one part of the HIV therapeutic equation. The immune system plays a part as well. Perhaps during windows such as STIs, the dynamics between immunity and virus can continue to be characterized, but with a safety net afforded by HAART.

References

(1.) Hogg RS, Heath KV, Yip B, et al. JAMA. 1998;279(6):450-454.

(2.) Hogg RS, Havlir D, Miller V, Montaner JS. AIDS. 2002;16(5):787-789.

(3.) Aiuti F, Giovannetti A. AIDS. 2003;17(15):2257-2258.

(4.) Binswanger HP. Science. 2001;292(5515):221-223.

(5.) Martinez-Picado J, Negredo E, Ruiz L, et al. Ann Intern Med. 2003;139:81-89.

(6.) Oxenius A, Price DA, Gunthard HF, et al. Proc Natl Acad Sci USA. 2002;99(21):13747-13752.

(7.) Neumann AU, Tubiana R, Calvez V, et al. AIDS. 1999;13(6):677-683.

(8.) de Jong MD, de Boer RJ, de Wolf F, et al. AIDS. 1997;11(11):F79-F84.

(9.) Harrigan PR, Whaley M, Montaner JS. AIDS. 1999;13(8):F59-F62.

(10.) Deeks SG, Wrin T, Liegler T, et al. N Engl J Med. 2001;344(7):472-480.

(11.) Rosenberg ES, Airfield M, Poon SH, et al. Nature. 2000;407(6803):523-526.

(12.) Hirschel B. N Engl J Med. 2003;349(9):827-828.

(13.) Deeks SG, Grant RM, Wrin T, et al. AIDS. 2003;17(3):361-370.

(14.) Goujard C, Deveau C, Sinet M, et al. 10th Conference on Retroviruses and Opportunistic Infections, February 10-14, 2003; Boston. Abstract 517.

(15.) Girard PM, Schneider V, Dehee A, et al. AIDS. 2001;15(2):275-277.

(16.) Lisziewicz J, Rosenberg E, Lieberman J, et al. N Engl J Med. 1999;340(21):1683-1684.

(17.) Lafeuillade A, Poggi C, Hittinger G, Counillon E, Emilie D. J Infect Dis. 2003;188(10):1426-1432.

(18.) Vila J, Nugier F, Bargues G, et al. Lancet. 1997;350(9078):635-636.

(19.) Ortiz GM, Nixon DF, Trkola A, et al. J Clin Invest. 1999;104(6):R13-R18.

(20.) Lori F, Maserati R, Foli A, Seminari E, Timpone J, Lisziewicz J. Lancet. 2000;355(9200):287-288.

(21.) Fagard C, Oxenius A, Gunthard H, et al. Arch Intern Med. 2003;163(10): 1220-1226.

(22.) Ananworanich J, Nuesch R, Le Braz M, et al. AIDS. 2003;17(15):F33-F37.

(23.) Ruiz L, Carcelain G, Martinez-Picado J, et al. AIDS. 2001;15(9):F19-27.

(24.) Dybul M, Chun TW, Yoder C, et al. Proc Natl Acad Sci USA. 2001;98(26):15161-15166.

(25.) Fischer M, Hafner R, Schneider C, et al. AIDS. 2003;17(2):195-199.

(26.) Dybul M, Nies-Kraske E, Daucher M, et al. J Infect Dis. 2003;188(3):388-396.

(27.) Martinez-Picado J, Morales-Lopetegi K, Wrin T, et al. AIDS, 2002;16(6):895-899.

(28.) Papasavvas E, Grant RM, Sun J, et al. AIDS. 2003; 17(16):2337-2343.

(29.) Simmons P. SMART: background, concepts, and design. Res Initiat Treat Action. 2001;7(1):11-14.

(30.) Davey RT Jr, Bhat N, Yoder C, et al. Proc Natl Acad Sci USA. 1999;96(26):15109-15114.

(31.) Garcia F, Plana M, Vidal C, et al. AIDS. 1999;13(11):F79-F86.

(32.) Ruiz L, Martinez-Picado, Romeu J, et al. AIDS. 2000;14:397-403.

(33.) Lawrence J, Mayers DL, Hullsiek KH, et al. N Engl J Med. 2003;349(9):837-846.

(34.) Ruiz L, Ribera E, Bonjoch A, et al. J Infect Dis. 2003;188(7):977-985.

(35.) Katlama C, Dominguez S, Duvivier C, et al. 10th Conference on Retroviruses and Opportunistic Infections, February 10-14, 2003; Boston. Abstract 68.

(36.) Miller V, Sabin C, Hertogs K, et al. AIDS. 2000;14(18):2857-2867.

(37.) Fagard C, Le Braz M, Gunthard H, et al. AIDS. 2003;17(10):1487-1492.

(38.) Chun TW, Engel D, Mizell SB, et al. Nat Med. 1999;5(6):651-655.

(39.) Stellbrink HJ, van Lunzen J, Westby M, et al. AIDS. 2002; 16(11): 1479-1487.

(40.) Garcia F, Plana M, Arnedo M, et al. AIDS. 2003;17(1):43-51.

(41.) Hoen B, Lacabaratz C, Fournier I, et al. 10th Conference on Retroviruses and Opportunistic Infections, February 10-14, 2003; Boston. Abstract 512.

(42.) Maggiolo F, Ripamonti D, Gregis G, et al. 43rd Interscience Conference on Antimicrobial Agents and Chemotherapy, September 14-17, 2003; Chicago. Abstract H-448.

(43.) Phillips AN, McLean AR, Loveday C, et al. AIDS. 1999;13(1):67-73.

(44.) Amicosante M, Poccia F, Gioia C, et al. J Infect Dis. 2003;188(5):661-665.

(45.) Joos B, Fischer M, Kuster H, et al. 10th Conference on Retroviruses and Opportunistic Injections, February 10-14, 2003; Boston. Abstract 496.

(46.) Tubiana R, Ghosn J, De-Sa M et al. AIDS. 2002;16:1083-1091.