Structured Treatment Interruptions for the Management of HIV Infection

The introduction of highly active antiretroviral therapy (HAART) was a milestone in the treatment of human immunodeficiency virus (HIV) infection, and has caused a significant decline in mortality rates among patients. However, the long-term use of HAART is expensive for many patients and it has generated drug toxicity and drug-adherence problems. Patients discontinuing HAART usually experience rapid viral rebound and loss of CD4 T lymphocytes^{1 - 2} because HIV suppression with HAART does not result in reconstitution of HIV-specific immune responses. Structured treatment interruptions (STI) may offer an alternative to continuous HAART. The STI approach involves repetitive on-and-off cycles of HAART to enhance the utility of therapy. While there is ground for cautious optimism in some cases, no comprehensive set of indicators are available to support the benefit of STI for an individual patient.

The STI approach has been studied in 3 distinct clinical scenarios: acute infection, chronic drug-suppressed infection, and virological drug failure (Table 1). The first group of patients are those treated shortly after HIV infection, who have a relatively intact immune system characterized by vigorous HIV-specific T-cell–mediated immune responses. The objective of STI in this patient population is to induce long-term immune control of HIV (ie, to preserve or reconstitute immune responses so HAART can be discontinued safely for a long period). In the second group of patients with chronic infection (representing the majority of patients), the present objective of STI is to ensure long-term drug-mediated control of HIV. If patients can limit their yearly exposure to antiretroviral drugs and, consequently, decrease HAART-associated toxic effects without jeopardizing the efficacy of the treatment, they might be able to take drugs over a longer period. For the third group, the rationale of STI for patients in whom HAART is no longer effective is that it is intended to restore drug sensitivity so that HAART can be reinitiated for subsequent effective inhibition of virus replication.

The use of STI is not part of the routine current clinical practice, and it should be distinguished from other forms of drug discontinuation, such as nonstructured sporadic interruptions due to acute toxic effects, lack of compliance, drug holidays, etc. The STI approach requires careful planning, is typically characterized by several interruptions, and needs close physician support and monitoring, as well as frequent laboratory tests. A large number of ongoing clinical trials worldwide (available at: http://www.clinicaltrials.gov) are currently testing the validity of the concept and exploring different STI approaches.

We used MEDLINE to find all peer-reviewed publications beginning in January 1999 to August 2001 that contained studies of patients treated with HAART in which treatment interruption was investigated. Within the same time frame of publication, we also explored an online database formerly known as AIDSLINE to extend our search to abstracts from authoritative international conferences. The online bibliographic database provided access to the published literature on acquired immunodeficiency syndrome (AIDS) and HIV and is now available at: http://gateway.nlm.nih.gov. We gave priority to peer-reviewed articles. The search terms used were combinations of structured treatment interruptions, supervised treatment interruptions, structured intermittent therapy, STI, SIT, HIV, SIV, therapeutic immunization, DNA, PHI, acute infection, chronic infection, virological failure, and salvage therapy. We based the analysis of the data on the type of drug treatment, baseline patient status, number of treatment interruptions, duration of treatment and interruption, changes in viral load, and immune system parameters (when available). We classified the patients according to their stage of infection and examined their response to STI in 3 distinct clinical scenarios: acute infection, chronic drug-suppressed infection, and virological drug failure.

Initial excitement for STI occurred with our anecdotal report of a patient in Berlin, Germany, who was able to control viral replication for 176 days (presently >4 years) after cycling on and off therapy twice.^{3 ,11} This patient had started treatment with HAART based on hydroxyurea, didanosine, and indinavir early after infection.³ At the time, it was noticed that the patient had vigorous T-cell–mediated immunity in the absence of neutralizing antibodies against HIV. The significance of the absence of neutralizing antibodies was unclear; however, later studies confirmed that control of HIV induced by STI during primary infection was associated in most cases with a cellular and not a humoral immune response.⁶ It was hypothesized that intermittent, controlled exposure to HIV might have boosted the HIV-specific immune response (autoimmunization hypothesis illustrated in Figure 1).

The case of the Berlin patient was followed by a report of a patient who had begun treatment within 2 weeks of infection, and had controlled viral replication but experienced toxic effects in the following 6 months of treatment.¹³ The patient stopped treatment and developed an acute HIV syndrome within 35 days, which resolved between 10 and 14 days. Resolution of the acute HIV syndrome coincided with emergence of a CD8 T-cell–mediated immune response and control of viremia. The authors speculated that the acute illness might have been a sign of the CD8 T-cell–mediated immune response.

A year later, another research group longitudinally followed 6 patients treated with different drug combinations and with variable adherence to their drug regimens.¹⁴ After stopping therapy, 3 of the 6 patients temporarily suppressed plasma viremia for 4 to 24 months. Based on the comparison of virological response, time of intervention, adherence, and HIV-specific immune responses, the authors concluded that potent HIV-specific cytotoxic T-lymphocyte response might be associated with viral control. After the example of the Berlin patient, our group described 3 other patients undergoing STI.¹⁵ Two of those patients were able to progressively elongate the time free from viral rebound during therapy interruption (one after the second STI, similar to the Berlin patient, the other after 5 STI), whereas the third one could not. It became clear that achieving control of HIV in the absence of therapy might require several STI in some patients and that STI might not be helpful for some patients.

Confidence among the scientific community in studying STI increased with the report of STI involving 8 subjects treated during acute infection in the Boston cohort.⁶ The protocol was to restart therapy if the viral load exceeded 5000 copies/mL of HIV RNA for 3 consecutive weeks, or 50 000 copies/mL at 1 time. All of these patients achieved long-term control of HIV replication in the absence of drug treatment after 1 or more treatment interruptions. Between the interruptions, there was an increase in the quantity and the breadth of the T-cell–mediated immune responses. Only 1 patient developed HIV-neutralizing antibodies. This study suggested the feasibility of autoimmunization.

At the same time, a randomized controlled trial comparing HAART with STI was performed using macaques acutely infected with simian immunodeficiency virus.⁹ The trial was designed to have a fixed-schedule STI (3 weeks on and then 3 weeks off HAART) because this kind of schedule can easily be translated to routine clinical use. The randomized controlled nature of the study comparing STI and HAART allowed the first demonstration that the intermittent interruption of therapy is the factor responsible for boosting immunity. All animals in the STI arm exhibited enhanced virus-specific T-cell–mediated immune responses that were associated with increasing control of viral rebound during subsequent interruptions. Animals treated continuously with HAART did not show increased immune responses and were unable to control viral rebound when treatment was withdrawn after 24 weeks. This study also used a new diagnostic assay based on the measurement of the interferon γ produced by CD8 T cells after virus-specific stimulation, known as the Virus-Specific Immune Response assay.⁹ The assay proved to be quantitative for the absolute number and percentage of subtypes of functional HIV-specific T cells. The quantity of HIV-specific CD8 cells expressing interferon γ correlated with the degree of immune control after treatment interruption in macaques.

Two potential STI strategies might be viable. The first would be to resume drug treatment after HIV reappears in the plasma. This approach is simple to interpret, but complicated to administer. A second approach would be to cycle HAART according to a fixed schedule. Whether 4 consecutive 6-week cycles (3 weeks on and 3 weeks off HAART) are enough to induce viral control in an HIV-infected patient treated early after infection must be determined in clinical trials. However, such an approach would be relatively easy to administer to a large group of patients. The STI approach now offers the hope that acutely infected patients might be able to stop therapy and still contain viral rebound. As it turns out, extension of these results to patients in later stages of the disease will be more complicated because STI in this setting cannot take advantage of the strong, relatively undamaged immune system, which is still present in the acute phase.

The acute phase of HIV infection does not last long. In a matter of weeks anti-HIV antibodies appear in the plasma. Complete seroconversion represents a critical point for the immune system. By the time the patient is chronically infected, the immune system is damaged, and the damage progresses during the chronic phase.¹⁶ Several reports indicated that the cellular arm of the immune response might not be as "boostable" by the autologous virus after this point and this might affect the rate of immunologic success of STI.^{7 ,17 - 18} Patients' immune responses to STI during the chronic phase are therefore expected to be variable, and if STI is to be widely used for such patients, more reliable predictors of immune system performance must be developed.

The Swiss-Spanish Study⁴ has been the largest clinical trial to investigate STI in chronically infected patients. Despite 2 weeks off and then 8 weeks on HAART for 4 cycles, participants did not show a great deal of improvement in viral load rebound or CD4 cell count. After 4 cycles, therapy was interrupted until viral rebound (>5000 copies/mL) occurred. Viral load remained less than 5000 copies/mL after 12 weeks without therapy in only 9 of 54 patients. All patients who had high baseline viremia (>100 000 copies/mL) experienced viral rebound. One promising perspective on these results, however, is that most patients were able to sustain viral control and CD4 cell counts without experiencing too much resistance during subsequent reinitiation of antiretroviral drugs. Those who failed to control their viral load to less than 50 copies/mL after resumption of treatment (24/128 patients) had high viral loads and low CD4 cell counts at baseline.

An observational study from the Philadelphia cohort⁷ followed 5 chronically infected HAART-suppressed patients. Both HIV-specific CD4 and CD8 responses increased following STI, and patients who had the stronger cellular immune responses at the beginning of a given interruption period had longer delays of viral rebound. These results suggest that patients treated with HAART might also be able to recover HIV-specific T-cell responses during chronic infection, and that variations in the timing and length of interruptions could have an effect on immune response.

French investigators¹⁹ performed 11 STI (median [range], 7 [4-24] days) in 3 HIV-infected patients after receiving 18 to 21 months of HAART. Treatment was resumed after 1 week or when virus became detectable. Seven viral rebounds were observed (median HIV RNA level, 4712 copies/mL) with a median of 7 days during which CD4 and CD8 cell counts did not significantly change. After treatment resumed, the viral load returned to below 200 copies/mL within 3 weeks. The T-cell–mediated immune responses appeared simultaneously with viral rebound, but this immunity was only transient and sometimes it disappeared before therapeutic control of the virus had occurred. No delay in viral rebound or increase in virus rebound rate was observed after repeated STI.

A Spanish group²⁰ randomly assigned 12 patients to interrupt HAART 3 times and 14 patients to continue their previous HAART. Plasma virus doubling time was significantly elongated during the second and third STI, and this was associated with a significant increase in the frequency of HIV-specific CD8 T cells. However, an increase in viremia of greater than 50 copies/mL was faster during the second and third STI.

In conclusion, STI during chronic infection has the potential of reconstituting the immune system, although this outcome remains questionable for any given patient. In the majority of patients, reexposure to autologous virus during STI transiently mobilizes HIV-specific T cells, but this T-cell population is rapidly depleted when control of viral replication is lost. These kinetics might explain the failure to delay subsequent viral rebound and raise concerns about strategies based on STI to restore durable HIV-specific T-cell responses in chronic HIV infection. Furthermore, the quality of immune reconstitution during chronic infection appears to be different from that obtainable during acute infection. This could be the result of several factors, including a skewed maturation of effector CD8 T lymphocytes,²¹ a damaged T-cell receptor V beta repertoire,^{22 - 23} disruption of the lymphoid organs,^{24 - 25} or establishment of latent viral reservoirs.^{26 - 28} Whether the damage inflicted by HIV on the immune system can be reversed at any given time during the course of infection (and if so, to what expected degree of recovery), remains to be seen. The issue is not simply academic because an incomplete (quantitatively and qualitatively) recovery of the immune system might be inadequate to maintain viral suppression in the absence of therapy.

An intervention involving treatment interruption contains inherent risks (Table 2). In the case of HIV infection, withdrawal of drug treatment unaccompanied by plans to restart treatment or monitor for viral rebound allows the virus to rebound out of control, possibly repopulating the cellular reservoirs, and an acute retroviral syndrome may be manifested. These developments may occur in patients who have had long-term, successful treatment with HAART.¹³ Similar problems might be encountered during STI if treatment interruption and treatment resumption cycles are not adequately controlled. Long interruption periods have been associated with a decline in CD4 cell counts and AIDS-related clinical events in patients with an advanced stage of the disease, especially in the salvage therapy setting.^{5 ,8} Although such results have not yet been reported, STI bears the potential for an induction and reinduction of the cytochrome P-450 system by different medications, as well as the recurrence of acute adverse effects of therapy (eg, nausea, diarrhea, rash, central nervous system effects, and hypersensitivity reaction). Finally, clinicians may find that they must consider social and psychological factors affecting patients. When patients have unrealistic expectations, the relief from a heavy daily regimen of medication may reduce their willingness to return to therapy, or adhere to the therapy schedule. Resolution of all these questions will have to wait for randomized controlled trials.

Management of STI during the chronic stage of HIV infection is more complicated than in the acute stage. At this time it appears that STI may not provide immune control sufficient to allow long treatment breaks for most chronically infected patients. Along this line, a group at the National Institutes of Health is attempting a 1 week on and then 1 week off fixed-schedule STI, with neither the hope nor intention of stimulating the immune system by autoimmunization, since the short interruptions would not provide enough time for the virus to rebound. The trial design is based on their previous observations indicating that HIV cannot be detected in the plasma before 7 to 10 days following treatment discontinuation.²⁹ Trial investigators¹⁰ reported promising preliminary results after 24 weeks regarding the fixed-schedule STI with 1 week on and then 1 week off. Patients in the study continued to control viral replication and had no significant changes in CD4 or CD8 T-cell counts. These results suggest that once the plasma viral load is reduced to less than 50 copies/mL, 50% less HAART administered in short, intermittent cycles may maintain CD4 T-cell counts and suppression of plasma HIV RNA. Although immunological benefits would not occur, there would be a substantial reduction in cost, relief from a continuous medication schedule, and a decrease in drug-related toxicity due to the reduced intake of drugs. The authors recommend caution in interpreting the preliminary data since they have yet to confirm that their strategy does not produce adverse effects.³⁰ One concern is that poor adherence to HAART has been significantly associated with the onset of resistance and virological failure.^{31 - 32} It has been reported that approximately 80% of patients who were adherent to less than 70% of drug doses experienced virological failure.³³ In contrast, 82% of patients with excellent adherence (those who took >95% of medications as prescribed) experienced virological success.

In patients with uncontrolled HIV infection whose regimens fail to suppress viral load, STI may induce a change in the most prevalent viral quasi species from multidrug-resistant to wild type, potentially improving sensitivity to subsequent salvage therapy regimens. Whether these changes are long-lived enough to have clinical significance, or are outweighed by the resulting damage to the immune system, is a critical question.

In the Frankfurt cohort,⁵ heavily pretreated patients with failing therapies received an STI for at least 2 months followed by a therapy restart. A shift to wild-type virus was observed in two thirds of the participants. This shift was associated with improved short-term virological response. On the negative side, these patients experienced a significant decrease in CD4 cell counts during their STI. One fourth of the patients did not recover their pre-STI CD4 cell values, and three fourths of the patients had viral load rebounds after therapy was restarted. In addition, AIDS-related events occurred.

In the San Francisco cohort,⁸ 16 patients experiencing virological failure with HAART were randomized to undergo a 12-week STI or remain taking their failing therapies. These patients experienced a median increase in viral load of about 0.84 log copies/mL and a median decrease in CD4 cell count of about 128 cells/mm³. It was found that a substantial phenotypic shift in the patients' HIV from drug-resistant to drug-sensitive occurred around weeks 6 through 8, but the resistant virus was still detectable in the circulating lymphocytes of most patients.

Thus far, the use of STI in patients with failing therapies has not been met with great success. The periods of the interruption phase have been relatively long, premised on the theory that a relatively long interruption is needed for the virus to revert from having mutations conferring drug resistance back to the wild type. Whether it is absolutely necessary to wait 8 weeks and sustain immune system damage to effect a change in the viral species during a STI is unknown. A shorter STI may be long enough to shift the viral population genotypic characteristics and yet short enough to reduce damage to the immune system. Currently, the data do not provide encouragement for the use of STI for patients with virological failure.

A fundamental question in evaluating the use of STI is when to restart treatment after viral rebound; furthermore, what is the definition of viral rebound in this context? Do we have to maintain the viral load at an undetectable level (and what is the definition of undetectable: less than 500 copies/mL, less than 50 copies/mL, or less than 10 copies/mL)? Is the "panic" value of 5000 copies/mL used in several trials^{4 ,6 ,15} a reasonable one? Or should an STI be declared a success if the viral load set point of less than 30 000 copies/mL, the threshold above which initiation of HAART is presently recommended,³⁴ can be maintained? Indeed, one could argue that the criteria for restarting therapy in patients receiving an STI intervention should be the same as the criteria for starting therapy in patients who are treatment-naive.

At present, STI may be most beneficial for patients treated with HAART during the acute phase of infection, as they are most likely to be able to develop enhanced immune responses. The immunologic advantage for patients with chronic suppression is less pronounced. However, preliminary results suggest that STI can increase immune responses in some chronically infected patients. One goal is to improve the ability to determine immune competence in these patients, and to develop tests that may then predict which patients will be the best candidates for STI. With this sort of information, physicians could recommend a choice of therapy and individualize it based on the immune status of the patient. A high level of HIV-specific T cells would suggest that the patient is a good candidate for STI, with the hope that treatment could then be safely interrupted for a relatively long period. A moderate level of HIV-specific T cells would predict that several cycles of a fixed-schedule STI would be necessary to mount a better immune response. A low or undetectable HIV-specific T-cell count before and/or after the STI would reveal a compromised immune system, and the use of continuous HAART could be recommended. The predictive value of the Virus-Specific Immune Response assay⁹ and other HIV-specific immune diagnostic assays³⁵ warrant evaluation in randomized controlled clinical trials.

Some investigators are engaged in trying to expand the time free from viral rebound during STI in chronic infection. There are a number of strategies that might be used to enhance the efficacy of STI. For example, the length of both the treatment and the interruption periods might be adjusted for different patient populations. The number of cycles might be increased and the requirements for restarting therapy might be modified. Adjunct therapies, such as the use of immune modulators, might enhance the immune responses that are obtained. A Spanish group³⁶ analyzed the use of hydroxyurea, an immune-system modulator, with HAART. Twenty chronically infected patients were randomized into 2 groups receiving stavudine, didanosine, indinavir, and hydroxyurea (hydroxyurea group) or stavudine, didanosine, and indinavir (HAART group). Patients underwent 5 consecutive cycles of STI, which involved 2-month treatment periods. Hydroxyurea was discontinued during the first 3 interruption periods along with all other drugs, but continued during the last 2 cycles despite interruption of other drugs. Immune responses were increased in both groups after the first 3 interruption periods, and interestingly, viral control began to emerge in the hydroxyurea group when hydroxyurea was continued during the interruption periods. At the end of the study, which all patients completed, T-lymphocyte proliferative responses were analyzed in 11 of 20 patients. One of 6 patients in the HAART group and 5 of 5 patients in the hydroxyurea group had CD4 T-lymphocyte proliferative responses specific to the HIV protein p24.

Another potential immune modulant could be micophenolic acid, which, like hydroxyurea, reduces T-cell activation.³⁷ Different cytokines that enhance T-cell–mediated immune responses might also be used as immunomodulators. For example, interleukin 2 has the potential of co-stimulating HIV-specific T cells responding to HIV antigen. Interleukin 2 might be used to amplify HIV-specific immune responses, provided a minimal amount of HIV antigen is present (eg, in combination with the STI).³⁸ Clinical trials to evaluate the effect of immunomodulatory drugs in combination with STI are warranted.

It is possible that HIV vaccines and STI could be used in combination as a therapeutic technique to extend drug-free periods. Many existing vaccine candidates, not only those designed for therapeutic vaccination, such as HIV-1 Immunogen,³⁹ poxvirus-based vaccines,⁴⁰ or the dendritic cell/Langerhans cell–mediated DNA immunization technology (DermaVir),^{41 - 42} but also those intended for use in prevention,^{43 - 44} can be tested. Autologous (by STI) and heterologous (by vaccine) antigens might act synergistically to enhance HIV-specific immune responses. It is conceivable that autologous and heterologous antigens will act in different cellular compartments and/or stimulate qualitatively different responses.

These are exciting prospects for HIV-infected patients. However, it is too early to recommend the routine use of STI in any clinical setting because the end result remains unpredictable, and the potential hazards to patients have not yet been properly evaluated. At the very least, STI must still be optimized in prospective clinical trials that assess the overall risk-benefit ratio of this novel strategy for various patient populations. Moreover, new assays that more reliably predict immune system performance are needed to evaluate patients and identify the STI therapies most likely to benefit them. In the future, the manipulation of the immune system could become a fourth "drug," which could be used as part of effective and comprehensive treatment strategies. Until then, patients and clinicians must wait for additional information before embracing STI outside the closely monitored clinical trial setting.