Testing for Plasma HIV RNA Levels and CD4+ T Cell Count To Guide Decisions Regarding Therapy

Decisions regarding initiation or changes in antiretroviral therapy should be guided by monitoring the laboratory parameters of plasma HIV RNA (viral load) and CD4⁺ T cell count in addition to the patient's clinical condition. Results of these laboratory tests provide clinicians with key information regarding the virologic and immunologic status of the patient and the risk for disease progression to acquired immunodeficiency syndrome (AIDS) (3, 4). HIV viral load testing has been approved by the Food and Drug Administration (FDA) for determining prognosis and for monitoring the response to therapy only for the reverse transcriptase-polymerase chain reaction (RT-PCR) assay and in vitro nucleic amplification test for HIV–RNA (NucliSens® HIV-1 QT, manufactured by Organon Teknika). Multiple analyses among >5000 patients who participated in approximately 18 trials with viral load monitoring indicated a statistically significant dose-response–type association between decreases in plasma viremia and improved clinical outcome on the basis of standard results of new AIDS-defining diagnoses and survival. This relationship was observed throughout a range of patient baseline characteristics, including pretreatment plasma RNA level, CD4⁺ T cell count, and previous drug experience. Thus, viral load testing is an essential parameter in deciding to initiate or change antiretroviral therapies. Measurement of plasma HIV RNA levels (i.e., viral load) by using quantitative methods should be performed at the time of diagnosis and every 3–4 months thereafter for the untreated patient (AIII) (Table 2). CD4⁺ T cell counts should be measured at the time of diagnosis and every 3–6 months thereafter (AIII). These intervals between tests are recommendations only, and flexibility should be exercised according to the circumstances of each patient. Plasma HIV RNA levels should also be measured immediately before and again at 2–8 weeks after initiation of antiretroviral therapy (AIII). This second measurement allows the clinician to evaluate initial therapy effectiveness because, for the majority of patients, adherence to a regimen of potent antiretroviral agents should result in a substantial decrease ("1.0 log₁₀) in viral load by 2–8 weeks. A patient's viral load should continue to decline during the following weeks and, for the majority of patients, should decrease below detectable levels (i.e., defined as <50 RNA copies/mL of plasma) by 16–24 weeks. Rates of viral load decline toward undetectable are affected by the baseline CD4⁺ T cell count, the initial viral load, potency of the regimen, adherence to the regimen, previous exposure to antiretroviral agents, and the presence of any OIs. These differences must be considered when monitoring the effect of therapy. However, the absence of a virologic response of the magnitude discussed previously should prompt the clinician to reassess patient adherence, rule out malabsorption, consider repeat RNA testing to document lack of response, or consider a change in drug regimen. After the patient is receiving therapy, HIV RNA testing should be repeated every 3–4 months to evaluate the continuing effectiveness of therapy (AII). With optimal therapy, viral levels in plasma at 24 weeks should be undetectable (5). Data from clinical trials demonstrate that lowering plasma HIV RNA to <50 copies/mL is associated with increased duration of viral suppression, compared with reducing HIV RNA to levels of 50–500 copies/mL (6). If HIV RNA remains detectable in plasma after 16–24 weeks of therapy, the plasma HIV RNA test should be repeated to confirm the result and a change in therapy should be considered (see Changing a Failing Regimen) (BIII).

When deciding on therapy initiation, the CD4⁺ T lymphocyte count and plasma HIV RNA measurement should be performed twice to ensure accuracy and consistency of measurement (BIII). However, among patients with advanced HIV disease, antiretroviral therapy should be initiated after the first viral load measurement is obtained to prevent a potentially deleterious delay in treatment. The requirement for two measurements of viral load might place a substantial financial burden on patients or payers. Nonetheless, the Panel believes that two measurements of viral load will provide the clinician with the best information for subsequent patient follow-up. Plasma HIV RNA levels should not be measured during or within 4 weeks after successful treatment of any intercurrent infection, resolution of symptomatic illness, or immunization. Because differences exist among commercially available tests, confirmatory plasma HIV RNA levels should be measured by using the same laboratory and the same technique to ensure consistent results.

A minimal change in plasma viremia is considered to be a threefold or 0.5-log₁₀ increase or decrease. A substantial decrease in CD4⁺ T lymphocyte count is a decrease of >30% from baseline for absolute cell numbers and a decrease of >3% from baseline in percentages of cells (7). Discordance between trends in CD4⁺ T cell numbers and plasma HIV RNA levels was documented among 20% of patients in one cohort studied (8). Such discordance can complicate decisions regarding antiretroviral therapy and might be caused by factors that affect plasma HIV RNA testing. Viral load and trends in viral load are believed to be more informative for decision-making regarding antiretroviral therapy than are CD4⁺ T cell counts; however, exceptions to this rule do occur (see Changing a Failing Regimen). In certain situations, consultation with a specialist should be considered.

Drug-Resistance Testing

Testing for HIV resistance to antiretroviral drugs is a useful tool for guiding antiretroviral therapy. When combined with a detailed drug history and efforts in maximizing drug adherence, these assays might maximize the benefits of antiretroviral therapy. Studies of treatment-experienced patients have reported strong associations between the presence of drug resistance, identified by genotyping or phenotyping resistance assays, and failure of the antiretroviral treatment regimen to suppress HIV replication. Genotyping assays detect drug-resistance mutations that are present in the relevant viral genes (i.e., reverse transcriptase and protease). Certain genotyping assays involve sequencing of the entire reverse transcriptase and protease genes, whereas others use probes to detect selected mutations that are known to confer drug resistance. Genotyping assays can be performed rapidly, and results can be reported within 1–2 weeks of sample collection. Interpretation of test results requires knowledge of the mutations that are selected for by different antiretroviral drugs and of the potential for cross-resistance to other drugs conferred by certain mutations. Consultation with a specialist in HIV drug resistance is encouraged and can facilitate interpretation of genotypic test results.

Phenotyping assays measure a virus's ability to grow in different concentrations of antiretroviral drugs. Automated, recombinant phenotyping assays are commercially available with results available in 2–3 weeks; however, phenotyping assays are more costly to perform, compared with genotypic assays. Recombinant phenotyping assays involve insertion of the reverse transcriptase and protease gene sequences derived from patient plasma HIV RNA into the backbone of a laboratory clone of HIV either by cloning or in vitro recombination. Replication of the recombinant virus at different drug concentrations is monitored by expression of a reporter gene and is compared with replication of a reference HIV strain. Drug concentrations that inhibit 50% and 90% of viral replication (i.e., the median inhibitory concentration [IC] IC₅₀ and IC₉₀) are calculated, and the ratio of the IC₅₀ of the test and reference viruses is reported as the fold increase in IC₅₀ (i.e., fold resistance). Interpretation of phenotyping assay results is complicated by the paucity of data regarding the specific resistance level (i.e., fold increase in IC₅₀) that is associated with drug failure; again, consultation with a specialist can be helpful for interpreting test results. Further limitations of both genotyping and phenotyping assays include the lack of uniform quality assurance for all available assays, relatively high cost, and insensitivity for minor viral species. If drug-resistant viruses are present but constitute <10%–20% of the circulating virus population, they probably will not be detected by available assays. This limitation is critical when interpreting data regarding susceptibility to drugs that the patient has taken in the past but that are not part of the current antiretroviral regimen. If drug resistance had developed to a drug that was subsequently discontinued, the drug-resistant virus can become a minor species because its growth advantage is lost (9). Consequently, resistance assays should be performed while the patient is taking his or her antiretroviral regimen, and data substantiating the absence of resistance should be interpreted cautiously in relation to the previous treatment history.

Using Resistance Assays in Clinical Practice

Resistance assays can be useful for patients experiencing virologic failure while on antiretroviral therapy and patients with acute HIV infection (Table 3). Recent prospective data supporting drug-resistance testing in clinical practice are derived from trials in which the test utility was assessed for cases of virologic failure. Two studies compared virologic responses to antiretroviral treatment regimens when genotyping resistance tests were available to guide therapy (10, 11) with the responses observed when changes in therapy were guided by clinical judgment only. The results of both studies indicated that the short-term virologic response to therapy was substantially increased when results of resistance testing were available. Similarly, a prospective, randomized, multicenter trial demonstrated that therapy selected on the basis of phenotypic resistance testing substantially improves the virologic response to antiretroviral therapy, compared with therapy selected without the aid of phenotypic testing (12). Thus, resistance testing appears to be a useful tool in selecting active drugs when changing antiretroviral regimens in cases of virologic failure (BII). Similar rationale applies to the potential use of resistance testing for patients with suboptimal viral load reduction (see Criteria for Changing Therapy) (BIII). Virologic failure regarding highly active antiretroviral therapy (HAART) is, for certain patients, associated with resistance to one component of the regimen only (13); in that situation, substituting individual drugs in a failing regimen might be possible, although this concept requires clinical validation (see Changing a Failing Regimen). No prospective data exist to support using one type of resistance assay over another (i.e., genotyping versus phenotyping) in different clinical situations. Therefore, one type of assay is recommended per sample; however, for patients with a complex treatment history, both assays might provide critical and complementary information.

Transmission of drug-resistant HIV strains has been documented and might be associated with a suboptimal virologic response to initial antiretroviral therapy (14-17). If the decision is made to initiate therapy in a person with acute HIV infection, using resistance testing to optimize the initial antiretroviral regimen is a reasonable, albeit untested, strategy (18, 19) (CIII). Because of its more rapid turnaround time, using a genotypic assay might be preferred in this situation. Using resistance testing before initiation of antiretroviral therapy among patients with chronic HIV infection is not recommended (DIII) because of uncertainty regarding the prevalence of resistance among treatment-naïve persons. In addition, available resistance assays might fail to detect drug-resistant species that were transmitted when primary infection occurred but became a minor species in the absence of selective drug pressure. Reserving resistance testing for patients with suboptimal viral load suppression after therapy initiation is preferable, although this approach might change as additional information becomes available related to the prevalence of resistant virus among antiretroviral-naïve patients.

Recommendations for resistance testing during pregnancy are the same as for nonpregnant women; acute HIV infection, virologic failure while on an antiretroviral regimen, or suboptimal viral load suppression after initiation of antiretroviral therapy are all appropriate indications for resistance testing. If an HIV-positive pregnant woman is taking an antiretroviral regimen that does not include zidovudine, or if zidovudine was discontinued because of maternal drug resistance, intrapartum and neonatal zidovudine prophylaxis should be administered to prevent mother-to-child HIV transmission (see Considerations for Antiretroviral Therapy Among HIV-Infected Pregnant Women). Not all of zidovudine's activity in preventing mother-to-child HIV transmission can be accounted for by its effect on maternal viral load (20); furthermore, preliminary data indicate that the rate of perinatal transmission after zidovudine prophylaxis might not differ between those with and without zidovudine-resistance mutations (21, 22). Studies are needed to determine the best strategy to prevent mother-to-child HIV transmission in the presence of zidovudine resistance.

Considerations for Patients with Established HIV Infection

Patients with established HIV infection are discussed in two arbitrarily defined clinical categories: asymptomatic infection or symptomatic disease (e.g., wasting, thrush, or unexplained fever for >2 weeks) including AIDS, as classified by CDC in 1993 (23). All patients in the second category should be offered antiretroviral therapy. Initiating antiretroviral therapy among patients in the first category is complex and, therefore, discussed separately. However, before initiating therapy for any patient, the following evaluation should be performed:

• complete history and physical (AII);
  
• complete blood count, chemistry profile, including serum transaminases and lipid profile (AII);
  
• CD4⁺ T lymphocyte count (AI); and
  
• plasma HIV RNA measurement (AI).
  

Additional evaluation should include routine tests relevant to preventing OIs, if not already performed (e.g., rapid plasma reagin or Venereal Disease Research Laboratory test; tuberculin skin test; toxoplasma immunoglobulin G serology; hepatitis B and C serology; and gynecologic exam, including Papanicolaou smear). Other tests are recommended, if clinically indicated (e.g., chest radiograph and ophthalmologic exam) (AII). Cytomegalovirus serology can be useful for certain patients (2) (BIII).

Considerations for Initiating Therapy for the Patient with Asymptomatic HIV Infection

Although randomized clinical trials provide strong evidence for treating patients with <200 CD4⁺ T cells/mm³ (24–26), the optimal time to initiate antiretroviral therapy among asymptomatic patients with CD4⁺ T cell counts >200 cells/mm³ is unknown. For persons with >200 CD4⁺ T cells/mm³, the strength of the recommendation for therapy must balance the readiness of the patient for treatment, consideration of the prognosis for disease-free survival as determined by baseline CD4⁺ T cell count and viral load levels, and assessment of the risks and potential benefits associated with initiating antiretroviral therapy.

Regarding a prognosis that is based on the patient's CD4⁺ T cell count and viral load, data are absent concerning clinical endpoints from randomized, controlled trials for persons with >200 CD4⁺ T cells/mm³ to guide decisions on when to initiate therapy. However, despite their limitations, observational cohorts of HIV-infected persons either treated or untreated with antiretroviral therapy provide key data to assist in risk assessment for disease progression.

Observational cohorts have provided critical data regarding the prognostic influence of viral load and CD4⁺ T cell count in the absence of treatment. These data indicate a strong relationship between plasma HIV RNA levels and CD4⁺ T cell counts in terms of risk for progression to AIDS for untreated persons and provide potent support for the conclusion that therapy should be initiated before the CD4⁺ T cell count declines to <200 cells/mm³ (Figure; Tables 4 and 5). In addition, these studies are useful for the identification of asymptomatic persons at high risk who have CD4⁺ T cell counts >200 cells/mm³ and who might be candidates for antiretroviral therapy or more frequent CD4⁺ T cell count monitoring. Regarding CD4⁺ T cell count monitoring, the Multicenter AIDS Cohort Study (MACS) demonstrated that the 3-year risk for progression to AIDS was 38.5% among patients with 201–350 CD4⁺ T cells/mm³, compared with 14.3% for patients with CD4⁺ T cell counts >350 cells/mm³. However, the short-term risk for progression also was related to the level of plasma HIV RNA, and the risk was relatively low for those persons with <20,000 copies/mL. An evaluation of 231 persons with CD4⁺ T cell counts of 201–350 cells/mm³ demonstrated that the 3-year risk for progression to AIDS was 4.1% for the 74 patients with HIV RNA <20,000; 36.4% for those 53 patients with HIV RNA 20,001–55,000 copies/mL; and 64.4% for those 104 patients with HIV RNA >55,000 copies/mL. Similar risk gradations by viral load were evident for patients with CD4⁺ T cell counts >350 cells/mm³ (Figure; Table 5) (unpublished data, Alvaro Muñoz, PhD, Johns Hopkins University, Baltimore, Maryland, 2001). These data indicate that for certain patients with CD4⁺ T cell counts >200 cells/mm³, the 3-year risk for disease progression to AIDS in the absence of treatment is substantially increased. Thus, although observational studies of untreated persons cannot assess the effects of therapy and, therefore, cannot determine the optimal time to initiate therapy, these studies do provide key guidance regarding the risks for progression in the absence of therapy on the basis of a patient's CD4⁺ T cell count and viral load.

View larger version (34K): [in this window] [in a new window]   Figure. Likelihood of developing AIDS by 3 years after becoming infected with HIV. *b-Deoxyribonucleic acid; reverse transcriptase-polymerase chain reaction. Source: Mellors JW, Muñoz A, Gigorni JV, et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med. 1997;126:946-54.

Data from observational studies of HAART-treated cohorts also provide critical information to guide using antiretroviral therapy among asymptomatic patients (27-30). A collaborative analysis of data from 13 cohort studies from Europe and North America demonstrates that among drug-naïve patients without AIDS-defining illness and a viral load of <100,000 copies/mL, the 3-year probability of progression to AIDS or death was 15.8% among those who initiated therapy with CD4⁺ T cell counts of 0–49 cells/mm³; 12.5% among those with CD4⁺ T cell counts of 50–99 cells/mm³; 9.3% among those with CD4⁺ T cell counts of 100–199 cells/mm³; 4.7% among those with CD4⁺ T cell counts of 200–349 cells/mm³; and 3.4% among those with CD4⁺ T cell counts of 350 cells/mm³ or higher (30). These data indicate that the prognosis might be better for patients who initiate therapy at >200 cells/mm³; but risk after initiation of therapy does not vary considerably at >200 cells/mm³. However, risk for progression also was related to plasma HIV RNA levels in this study. A substantial increase in risk for progression was evident among all patients with a viral load >100,000 copies/mL. In other cohort studies, an apparent benefit in terms of disease progression was reported among persons who began antiretroviral therapy when CD4⁺ T cell counts were >350 cells/mm³, compared with those who deferred therapy (31, 32). For example, in the Swiss cohort study, an approximate 7-fold decrease occurred in disease progression to AIDS among persons who initiated therapy with a CD4⁺ T cell count >350 cells/mm³, compared with those who were monitored without therapy during a 2-year period (32). However, a substantial incidence of adverse treatment effects occurred among patients who initiated therapy; 40% of patients had 1 treatment changes because of adverse effects, and 20% were no longer receiving treatment after 2 years (32). Unfortunately, observational studies of persons treated with HAART also have limitations regarding the ability to determine an optimal time to initiate therapy. The relative risks for disease progression for persons with CD4⁺ T cell counts of 200–349 and >350 cells/mm³ cannot be precisely compared because of the low level of disease progression among these patients during the follow-up period. In addition, groups might differ in key known and unknown prognostic factors that bias the comparison.

In addition to the risks for disease progression, the decision to initiate antiretroviral therapy also is influenced by an assessment of other potential risks and benefits associated with treatment. Potential benefits and risks of early or delayed therapy initiation for the asymptomatic patient should be considered by the clinician and patient [Table 5]. Potential benefits of early therapy include 1) earlier suppression of viral replication; 2) preservation of immune function; 3) prolongation of disease-free survival; and 4) decrease in the risk for viral transmission. Risks include 1) the adverse effects of the drugs on quality of life; 2) the inconvenience of the majority of the available suppressive regimens, leading to reduced adherence; 3) development of drug resistance because of suboptimal suppression of viral replication; 4) limitation of future treatment options as a result of premature cycling of the patient through the available drugs; 5) the risk for transmission of virus resistant to antiretroviral drugs; 6) serious toxicities associated with certain antiretroviral drugs [e.g., elevations in serum levels of cholesterol and triglycerides, alterations in the distribution of body fat, or insulin resistance and diabetes mellitus]; and 7) the unknown durability of effect of available therapies. Potential benefits of delayed therapy include 1) minimization of treatment-related negative effects on quality of life and drug-related toxicities; 2) preservation of treatment options; and 3) delay in the development of drug resistance. Potential risks of delayed therapy include 1) the possibility that damage to the immune system, which might otherwise be salvaged by earlier therapy, is irreversible; 2) the possibility that suppression of viral replication might be more difficult at a later stage of disease; and 3) the increased risk for HIV transmission to others during a longer untreated period. Finally, for certain persons, ascertaining the precise time at which the CD4⁺ T cell count will decrease to a level where the risk for disease is high might be difficult, and time might be required to identify an effective, tolerable regimen. This task might be better accomplished before reaching a CD4⁺ T cell count of 200 cells/mm³.

After considering available data in terms of the relative risk for progression to AIDS at certain CD4⁺ T cell counts and viral loads and the potential risks and benefits associated with initiating therapy, certain specialists in this area believe that the evidence supports initiating therapy for asymptomatic HIV-infected persons with a CD4⁺ T cell count of <350 cells/mm³ or a viral load of >55,000 copies/mL (by RT-PCR or b-deoxyribonucleic acid [bDNA] assays). For asymptomatic patients with CD4⁺ T cell counts >350 cells/mm³, rationale exists for both conservative and aggressive approaches to therapy. The conservative approach is based on the recognition that robust immune reconstitution still occurs in the majority of patients who initiate therapy with CD4⁺ T cell counts in the 200–350 cells/mm³ range, and that toxicities and adherence challenges might outweigh benefits of initiating therapy at CD4⁺ T cell counts >350 cells/mm³. In the conservative approach, increased levels of plasma HIV RNA (i.e., >55,000 by RT-PCR or bDNA assays) are an indication that more frequent monitoring of CD4⁺ T cell counts and plasma HIV RNA levels is needed, but not necessarily for initiation of therapy. In the aggressive approach, asymptomatic patients with CD4⁺ T cell counts >350 cells/mm³ and levels of plasma HIV RNA >55,000 copies/mL would be treated because of the risk for immunologic deterioration and disease progression. The aggressive approach is supported by the observation in multiple studies that suppression of plasma HIV RNA by antiretroviral therapy is easier to achieve and maintain at higher CD4⁺ T cell counts and lower levels of plasma viral load (6, 33-36). However, long-term clinical outcome data are not available to fully endorse this approach.

Data are conflicting regarding sex-specific differences in viral load and CD4⁺ T cell counts. Certain studies (37-43), although not others (44-47), have concluded that after adjustment for CD4⁺ T cell count, levels of HIV RNA are lower in women than men. In those studies that have indicated a possible sex difference in HIV RNA levels, women have had RNA levels that ranged from 0.13 to 0.28 log₁₀ lower than observed among men. In two studies of HIV seroconverters, HIV RNA copy numbers were substantially lower in women than men at seroconversion, but these differences decreased with time; and median viral load in women and men became similar within 5–6 years after seroconversion (38, 39, 43). Other data indicate that CD4⁺ T cell counts might be higher in women than men (48). However, importantly, rates of disease progression do not differ in a sex-dependent manner (41, 43, 49, 50). Taken together, these data demonstrate that sex-based differences in viral load occur predominantly during a window of time when the CD4⁺ T cell count is relatively preserved, when treatment is recommended only in the setting of increased levels of plasma HIV RNA. Clinicians might consider lower plasma HIV RNA thresholds for initiating therapy in women with CD4⁺ T cell counts >350 cells/mm³, although insufficient data exist to determine an appropriate threshold. In patients with CD4⁺ T cell counts <350 cells/mm³, limited sex-based differences in viral load have been observed; therefore, no changes in treatment guidelines for women are recommended for this group.

In summary, the decision to begin therapy for the asymptomatic patient with >200 CD4⁺ T cells/mm³ is complex and must be made in the setting of careful patient counseling and education. Factors that must be considered in this decision are 1) the willingness, ability, and readiness of the person to begin therapy; 2) the degree of existing immunodeficiency as determined by the CD4⁺ T cell count; 3) the risk for disease progression as determined by the CD4⁺ T cell count and level of plasma HIV RNA (1) [Figure; Tables 5 and 6]; 4) the potential benefits and risks of initiating therapy for asymptomatic persons, including short- and long-term adverse drug effects [Table 4]; and 5) the likelihood, after counseling and education, of adherence to the prescribed treatment regimen. Regarding adherence, no patient should automatically be excluded from consideration for antiretroviral therapy simply because he or she exhibits a behavior or other characteristic judged by the clinician to lend itself to nonadherence. Rather, the likelihood of patient adherence to a long-term, complex drug regimen should be discussed and determined by the patient and clinician before therapy is initiated. To achieve the level of adherence necessary for effective therapy, providers are encouraged to use strategies for assessing and assisting adherence: intensive patient education and support regarding the critical need for adherence should be provided; specific goals of therapy should be established and mutually agreed upon; and a long-term treatment plan should be developed with the patient. Intensive follow-up should occur to assess adherence to treatment and to continue patient counseling for the prevention of sexual and drug-injection–related transmission (see Adherence to Potent Antiretroviral Therapy).

Considerations for Discontinuing Therapy

As recommendations evolve, patients who had begun active antiretroviral therapy at CD4⁺ T cell counts of >350/mm³ might consider discontinuing treatment. Limited clinical data exist addressing whether this should be done or if it can be accomplished safely. Potential benefits include reduction of toxicity and drug interactions, decreased risk for drug-selecting resistant variants, and improvement in quality of life. Risks include rebound in viral replication and renewed immunologic deterioration. If the patient and clinician agree to discontinue therapy, the patient should be closely monitored.

Adherence to Potent Antiretroviral Therapy

The Panel recommends that certain persons living with HIV, including persons who are asymptomatic, should be treated with HAART for the rest of their lives. Adherence to the regimen is essential for successful treatment and has been reported to increase sustained virologic control, which is critical in reducing HIV-related morbidity and mortality. Conversely, suboptimal adherence has been reported to decrease virologic control and has been associated with increased morbidity and mortality (51, 52). Suboptimal adherence also leads to drug resistance, limiting the effectiveness of therapy (53). The determinants, measurements, and interventions to improve adherence to HAART are insufficiently characterized and understood, and additional research regarding this topic is needed.

Adherence to Therapy during HIV Disease

Adherence is a key determinant in the degree and duration of virologic suppression. Among studies reporting on the association between suboptimal adherence and virologic failure, nonadherence among patients on HAART was the strongest predictor for failure to achieve viral suppression below the level of detection (52, 53). Other studies have reported that 90%–95% of doses must be taken for optimal suppression, with lesser degrees of adherence being associated with virologic failure (51, 54). No conclusive evidence exists that the degree of adherence required varies with different classes of agents or different medications in the HAART regimen.

Suboptimal adherence is common. Surveys have determined that one third of patients missed doses within 3 days of the survey (55). Reasons for missed doses were predictable and included forgetting, being too busy, being out of town, being asleep, being depressed, having adverse side effects, and being too ill (56). One fifth of HIV-infected patients in one urban center never filled their prescriptions. Although homelessness can lead to suboptimal adherence, one program achieved a 70% adherence rate among homeless persons by using flexible clinic hours, accessible clinic staff, and incentives (57).

Predictors of inadequate adherence to HIV medications include 1) lack of trust between the clinician and patient; 2) active drug and alcohol use; 3) active mental illness [e.g., depression]; 4) lack of patient education and inability of patients to identify their medications [56]; and 5) lack of reliable access to primary medical care or medication (58). Other sources of instability influencing adherence include domestic violence and discrimination (58). Medication side effects can also cause inadequate adherence, as can fear of or experiencing metabolic and morphologic side effects of HAART [59]. Predictors of optimal adherence to HIV medications, and hence, optimal viral suppression, include 1) availability of emotional and practical life supports; 2) a patient's ability to fit medications into his or her daily routine; 3) understanding that suboptimal adherence leads to resistance; 4) recognizing that taking all medication doses is critical; 5) feeling comfortable taking medications in front of others [60]; and 6) keeping clinic appointments (34).

Measurement of adherence is imperfect and lacks established standards. Patient self-reporting is an unreliable predictor of adherence; however, a patient's estimate of suboptimal adherence is a strong predictor and should be strongly considered (60, 61). A clinician's estimate of the likelihood of a patient's adherence is also an unreliable predictor (62). Aids for measuring adherence (e.g., pill counts, pharmacy records, smart pill bottles with computer chips that record each opening [i.e., medication event monitoring systems or MEMS caps]) might be useful, although each aid requires comparison with patient self-reporting (61, 63). Clinician and patient estimates of the degree of adherence have been reported to exceed measures that are based on MEMS caps. Because of its complexity and cost, MEMS caps technology might be used as an adjunct to adherence research, but it is not useful in clinical settings.

Self-reporting should include a short-term assessment of each dose that was taken during the recent past (e.g., 3 days) and a general inquiry regarding adherence since the last visit, with explicit attention to the circumstances of missed doses and possible measures to prevent further missed doses. Having patients bring their medications and medication diaries to clinic visits might be helpful also.

Approaching the Patient Patient-Related Strategies

The first principle of patient-related strategies is to negotiate a treatment plan that the patient understands and to which he or she commits (Tables 7, 8, 9, and 10) (64, 65). Before writing the first prescription, clinicians should assess the patient's readiness to take medication, which might take two or three office visits and patience. Patient education should include the goals of therapy, including a review of expected outcomes that are based on baseline viral load and CD4⁺ T cell counts (4), the reason for adherence, and the plan for and mechanics of adherence. Patients must understand that the first HAART regimen has the best chance for long-term success (1). Clinicians and health teams should develop a plan for the specific regimen, including how medication timing relates to meals and daily routines. Centers have offered practice sessions and have used candy in place of pills to familiarize the patient with the rigors of HAART; however, no data exist to indicate if this exercise improves adherence. Daily or weekly pillboxes, timers with alarms, pagers, and other devices can be useful. Because medication side effects can affect treatment adherence, clinicians should inform patients in advance of possible side effects and when they are likely to occur. Treatment for side effects should be included with the first prescription, as well as instructions on appropriate response to side effects and when to contact the clinician. Low literacy is associated with suboptimal adherence, also. Clinicians should assess a patient's literacy level before relying on written information, and they should tailor the adherence intervention for each patient. Visual aids and audio or video information sources can be useful for patients with low literacy (66).

Education of family and friends and their recruitment as participants in the adherence plan can be useful. Community interventions, including adherence support groups or the addition of adherence concerns to other support group agendas, can aid adherence. Community-based case managers and peer educators can assist with adherence education and strategies for each patient.

Temporary postponement of HAART initiation has been proposed for patients with identified risks for suboptimal adherence (67, 68). For example, a patient with active substance abuse or mental illness might benefit from psychiatric treatment or treatment for chemical dependency before initiating HAART. During the 1–2 months needed for treatment of these conditions, appropriate HIV therapy might be limited to OI prophylaxis, if indicated, and therapy for drug withdrawal, detoxification, or the underlying mental illness. In addition, readiness for HAART can be assessed and adherence education can be initiated during this period. Other sources of patient instability (e.g., homelessness) can be addressed during this time. Patients should be informed and in agreement with plans for future treatment and time-limited treatment deferral.

Selected factors (e.g., sex, race, low socioeconomic status or education level, and past drug use) are not reliable predictors of suboptimal adherence. Conversely, higher socioeconomic status and education level and a lack of past drug abuse do not predict optimal adherence (69). No patient should automatically be excluded from antiretroviral therapy simply because he or she exhibits a behavior or characteristic judged by the clinician to indicate a likelihood of nonadherence.

Clinician- and Health Team-Related Strategies

Trusting relationships among the patient, clinician, and health team are essential (Table 8). Clinicians should commit to communication between clinic visits, ongoing adherence monitoring, and timely response to adverse events or interim illness. Interim management during clinician vacations or other absences must be clarified with the patient.

Optimal adherence requires full participation by the health-care team, with goal reinforcement by 2 team members. Supportive and nonjudgmental attitudes and behaviors will encourage patient honesty regarding adherence and problems. Improved adherence is associated with interventions that include pharmacist-based adherence clinics (69), street-level drop-in centers with medication storage and flexible hours for homeless persons (70), adolescent-specific training programs (71), and medication counseling and behavioral interventions (72) (Table 9). For all health-care team members, specific training regarding HAART and adherence should be offered and updated periodically.

Monitoring can identify periods of inadequate adherence. Evidence indicates that adherence wanes as time progresses, even among patients whose adherence has been optimal, a phenomenon described as pill fatigue or treatment fatigue (67, 73). Thus, monitoring adherence at every clinic encounter is essential. Reasonable responses to decreasing adherence include increasing the intensity of clinical follow-up, shortening the follow-up interval, and recruiting additional health team members, depending on the problem (68). Certain patients (e.g., chemically dependent patients, mentally retarded patients in the care of another person, children and adolescents, or patients in crisis) might require ongoing assistance from support team members from the outset.

New diagnoses or symptoms can influence adherence. For example, depression might require referral, management, and consideration of the short- and long-term impact on adherence. Cessation of all medications at the same time might be more desirable than uncertain adherence during a 2-month exacerbation of chronic depression.

Responses to adherence interventions among specific groups have not been well-studied. Evidence exists that programs designed specifically for adolescents, women and families, injection-drug users, and homeless persons increase the likelihood of medication adherence (69, 71, 74, 75). The incorporation of adherence interventions into convenient primary care settings; training and deployment of peer educators, pharmacists, nurses, and other health-care personnel in adherence interventions; and monitoring of clinician and patient performance regarding adherence are beneficial (70, 76, 77). In the absence of data, a reasonable response is to address and monitor adherence during all HIV primary care encounters and incorporate adherence goals in all patient treatment plans and interventions. This might require the full use of a support team, including bilingual providers and peer educators for non–English-speaking populations, incorporation of adherence into support group agendas and community forums, and inclusion of adherence goals and interventions in the work of chemical-dependency counselors and programs.

Regimen-Related Strategies

Regimens should be simplified as much as possible by reducing the number of pills and therapy frequency and by minimizing drug interactions and side effects. For certain patients, problems with complex regimens are of lesser importance, but evidence supports simplified regimens with reduced pill numbers and dose frequencies (78, 79). With the effective options for initial therapy noted in this report and the observed benefit of less frequent dosing, twice-daily dosing of HAART regimens is feasible for the majority of patients. Regimens should be chosen after review and discussion of specific food requirements and patient understanding and agreement to such restrictions. Regimens requiring an empty stomach multiple times daily might be difficult for patients with a wasting disorder, just as regimens requiring high fat intake might be difficult for patients with lactose intolerance or fat aversion. However, an increasing number of effective regimens do not have specific food requirements.

Directly Observed Therapy

Directly observed therapy (DOT), in which a health-care provider observes the ingestion of medication, has been successful in tuberculosis management, specifically among patients whose adherence has been suboptimal. However, DOT is labor-intensive, expensive, intrusive, and programmatically complex to initiate and complete; and unlike tuberculosis, HIV requires lifelong therapy. Pilot programs have studied DOT among HIV patients with preliminary success. These programs have studied once-daily regimens among prison inmates, methadone program participants, and other patient cohorts with a record of repeated suboptimal adherence. Modified DOT programs have also been studied in which the morning dose is observed and evening and weekend doses were self-administered. The goal of these programs is to improve patient education and medication self-administration during a limited period (e.g., 3–6 months); however, the outcome of these programs, including long-term adherence after DOT completion, has not been determined (80-83).

Therapy Goals

Eradication of HIV infection cannot be achieved with available antiretroviral regimens, chiefly because the pool of latently infected CD4⁺ T cells is established during the earliest stages of acute HIV infection (84) and persists with a long half-life, even with prolonged suppression of plasma viremia to <50 copies/mL (85-88). The primary goals of antiretroviral therapy are maximal and durable suppression of viral load, restoration and preservation of immunologic function, improvement of quality of life, and reduction of HIV-related morbidity and mortality (Table 10). In fact, adoption of treatment strategies recommended in this report has resulted in substantial reductions in HIV-related morbidity and mortality (89-91).

Plasma viremia is a strong prognostic indicator in HIV infection (3). Furthermore, reductions in plasma viremia achieved with antiretroviral therapy account for substantial clinical benefits (92). Therefore, suppression of plasma viremia as much as possible for as long as possible is a critical goal of antiretroviral therapy, but this goal must be balanced against the need to preserve effective treatment options. Switching antiretroviral regimens for any detectable level of plasma viremia can rapidly exhaust treatment options; reasonable parameters that can prompt a change in therapy are discussed in Criteria for Changing Therapy.

HAART often leads to increases in the CD4⁺ T cell count of 100–200 cells/mm³, although patient responses are variable. CD4⁺ T cell responses are usually related to the degree of viral load suppression (93). Continued viral load suppression is more likely for those patients who achieve higher CD4⁺ T cell counts during therapy (94). A favorable CD4⁺ T cell response can occur with incomplete viral load suppression and might not indicate an unfavorable prognosis (95). Durability of the immunologic responses that occur with suboptimal suppression of viremia is unknown; therefore, although viral load is a strong predictor of long-term clinical outcome, clinicians should consider also sustained rises in CD4⁺ T cell counts and partial immune restoration. The urgency of changing therapy in the presence of low-level viremia is tempered by this observation. Expecting that continuing the existing therapy will lead to rapid accumulation of drug-resistant virus might not be reasonable for every patient. A reasonable strategy is maintenance of the regimen, but with redoubled efforts at optimizing adherence and increased monitoring.

Partial reconstitution of immune function induced by HAART might allow elimination of unnecessary therapies (e.g., therapies used for prevention and maintenance against OIs). The appearance of naïve T cells (96, 97), partial normalization of perturbed T cell receptor Vb repertoires (98), and evidence of residual thymic function in patients receiving HAART (99, 100) demonstrate that partial immune reconstitution occurs in these patients. Further evidence of functional immune restoration is the return during HAART of in vitro responses to microbial antigens associated with opportunistic infections (101) and the lack of Pneumocystis carinii pneumonia among patients who discontinued primary Pneumocystis carinii pneumonia prophylaxis when their CD4⁺ T cell counts rose to >200 cells/mm³ during HAART (102-104). Guidelines include recommendations concerning discontinuation of prophylaxis and maintenance therapy for certain OIs when HAART-induced increases in CD4⁺ T cell counts occur (2).

Tools for Achieving Therapy Goals

Although approximately 70%–90% of antiretroviral drug-naïve patients achieve maximal viral load suppression 6–12 months after therapy initiation, only 50% of patients in certain city clinics achieved similar results (33, 34). Predictors of virologic success include low baseline viremia and high baseline CD4⁺ T cell count (33-35), rapid decline of viremia (6), decline of viremia to <50 HIV RNA copies/mL (6), adequate serum levels of antiretroviral drugs (6, 105), and adherence to the drug regimen (34, 51, 106). Although optimal strategies for achieving antiretroviral therapy goals have not been fully delineated, efforts to improve patient adherence to therapy are critical (see Adherence to Potent Antiretroviral Therapy).

Another tool for maximizing benefits of antiretroviral therapy is the rational sequencing of drugs and the preservation of future treatment options for as long as possible. Three alternative regimens include a protease inhibitor (PI) with two nucleoside reverse transcriptase inhibitors (NRTIs), a nonnucleoside reverse transcriptase inhibitor (NNRTI) with two NRTIs, or a 3-NRTI regimen (Table 11). The goal of a class-sparing regimen is to preserve or spare 1 classes of drugs for later use. Extending the overall long-term effectiveness of the available therapy options might be possible by sequencing drugs in this manner. Moreover, this strategy enables selectively delaying the risk for certain side effects associated with a single class of drugs. The efficacy of PI-containing HAART regimens has been reported to include durable viral load suppression, partial immunologic restoration, and decreased incidence of AIDS and death (24-26). Viral load suppression and CD4⁺ T cell responses that are similar to those observed with PI-containing regimens have been achieved with selected PI-sparing regimens (e.g., efavirenz plus two NRTIs [107] or abacavir plus two NRTIs [108]); however, whether such PI-sparing regimens will provide comparable efficacy with regard to clinical outcomes is unknown.

The presence of drug-resistant HIV among treatment-experienced patients is a strong predictor of virologic failure and disease progression (109-111). Results of prospective studies indicate that the virologic response to a new antiretroviral regimen can be substantially improved when results of previous resistance testing are available to guide drug choices (10, 11). Thus, resistance testing is a useful tool in selecting active drugs when changing antiretroviral regimens after virologic failure (see Drug-Resistance Testing).

Initiating Therapy for the Asymptomatic HIV-Infected Patient

When initiating antiretroviral therapy for the patient who is naïve to such therapy, clinicians should begin with a regimen that is expected to achieve sustained suppression of plasma HIV RNA, a sustained increase in CD4⁺ T cell count, and a favorable clinical outcome (i.e., delayed progression to AIDS and death). Clinicians should consider also the regimen's pill burden, dosing frequency, food requirements, convenience, toxicity, and drug-interaction profile compared with other regimens. Strongly recommended regimens include either indinavir, nelfinavir, ritonavir plus saquinavir; ritonavir plus indinavir; ritonavir plus lopinavir; or efavirenz in combination with one of the two NRTI combinations (Table 12). Clinical outcome data support using a PI in combination with NRTIs (24-26) (BI). Ritonavir as the single PI should be considered as an alternative agent because certain patients have difficulty tolerating standard doses of ritonavir (34) and because of the drug's multiple interactions. A similar rationale applies to saquinavir soft-gel capsule because certain patients have difficulty tolerating standard doses and because of the pill burden associated with its use; however, switching a patient off a ritonavir or saquinavir-based regimen is not necessary if they are tolerating the regimen and it is effective.

Using ritonavir to increase plasma concentrations of other PIs has evolved from an investigational concept to widespread practice. Standard doses of PIs result in trough drug levels (i.e., the lowest drug levels in the patient's system) that are only slightly higher than the effective antiviral concentration, which could allow viral replication. In contrast, protease boosting or enhancement by administering ritonavir increases the trough levels of other PIs higher than the IC₅₀ or IC₉₅, which minimizes opportunities for viral replication and potentially allows drug activity against even moderately resistant strains of virus. Additionally, these dual-PI combinations can lead to more convenient regimens regarding pill burden, scheduling, and elimination of food restrictions. They also might prevent efavirenz- or nevirapine-induced drug interactions.

Ritonavir increases plasma concentrations of other PIs by at least 2 mechanisms, including inhibition of gastrointestinal cytochrome P450 (CYP450) during absorption, and metabolic inhibition of hepatic CYP450. The 20-fold increase in saquinavir plasma concentrations with ritonavir coadministration is probably caused by CYP450 inhibition at both sites and leads to an increase in the saquinavir peak plasma concentration (C_max) (112). For lopinavir, the addition of ritonavir increases the C_max and half-life, which subsequently results in a higher trough concentration. The result is a lopinavir blood concentration curve that is 100-fold higher, compared with lopinavir alone (113). For other PIs, metabolism in the gastrointestinal tract is less critical, and the enhancement is primarily the result of CYP450 inhibition in the liver. The addition of ritonavir to amprenavir, nelfinavir, or indinavir results in substantial increases in half-life and trough levels, with a more moderate or minimal increase in C_max (114, 115).

The dose of ritonavir that is used for PI boosting is also critical for certain PIs but not others. With saquinavir and amprenavir, increases in the ritonavir dose to >100 mg two times/day do not significantly increase the PI levels (114, 116). However, increasing ritonavir doses to >100 mg two times/day provides additional enhancement for indinavir and nelfinavir (115, 117). Although pharmacokinetic data support using ritonavir-plus-PI combinations, limited data are available regarding combinations other than ritonavir plus saquinavir (118) or ritonavir plus lopinavir (119). In addition, the long-term risks and toxicities of dual-PI combinations remain unknown.

Disappointing results with antiretroviral regimens prescribed after virologic failure with a previous regimen indicate that the first regimen affords the best opportunity for long-term control of viral replication. Because the genetic barrier to resistance is greatest with PIs, certain experienced clinicians consider a PI plus two NRTIs to be the preferred initial regimen. However, efavirenz plus two NRTIs is as effective as one PI plus two NRTIs in suppressing plasma viremia and increasing CD4⁺ T cell counts (107), and certain experienced clinicians prefer this as the initial regimen because it might spare the toxicities of PIs for a substantial time (BII). Although no direct comparative trials have been reported that would allow a ranking of relative efficacy of NNRTIs, the ability of efavirenz in combination with two NRTIs to suppress viral replication and increase CD4⁺ T cell counts to a similar degree as one PI with two NRTIs supports a preference for efavirenz over other presently available NNRTIs. Abacavir plus two NRTIs, a 3-NRTI regimen, has been used successfully as well (108) (CII). However, such a regimen might have short-lived efficacy when the baseline viral load is >100,000 copies/mL. Using two NRTIs does not achieve the goal of suppressing viremia to below detectable levels as consistently as does a regimen in the strongly recommended or alternative categories and should be used only if more potent treatment is impossible (DI). Use of antiretroviral agents as monotherapy is contraindicated (DI), except when no other options exist or during pregnancy to reduce perinatal transmission. When initiating antiretroviral therapy, all drugs should be started simultaneously at full dose with the following exceptions: dose escalation regimens are recommended for ritonavir, nevirapine, and for certain patients, ritonavir plus saquinavir.

Hydroxyurea has been used investigationally in combination with antiretroviral agents for treatment of HIV infection; however, its utility in this setting has not been established. Clinicians considering use of hydroxyurea in a treatment regimen for HIV should be aware of the limited and conflicting nature of data in support of its efficacy and the importance of monitoring patients closely for potentially serious toxicity.**

Detailed information is included in this report comparing NRTIs, NNRTIs, PIs, drug interactions between PIs and other agents, toxicities, and FDA-required warning labels (Tables 13, 14, 15, 16, 17, 18, 19, and 20). Drug interactions between PIs and other agents can be extensive and often require dose modification or substitution of different drugs (Tables 17, 18, and 19) Toxicity assessment is an ongoing process; assessment 2 times during the first month of therapy and every 3 months thereafter is a reasonable management approach.