Initiation of Highly Active Antiretroviral Therapy at CD4⁺ T Lymphocyte Counts of >350 cells/mm³: Disease Progression, Treatment Durability, and Drug Toxicity

The optimal CD4⁺ T lymphocyte count at which to initiate HAART in asymptomatic HIV-infected persons is unknown. Among patients receiving HAART, the risk of death is increased for persons who do not begin to receive it until their CD4⁺ T lymphocyte counts are <200 cells/mm³ [1]. However, even in persons for whom HAART is initiated at low CD4⁺ T lymphocyte levels, virologic suppression can be achieved [2]. In one study, HAART improved disease progression when initiated at a CD4⁺ T lymphocyte level of <200 cells/mm³, but its impact was less clear when initiated at a level of >200 cells/mm³ [3]. In 2 other studies, however, HAART slowed clinical disease progression for persons with a CD4⁺ cell count of >350 cells/mm³, compared with persons who either did not receive HAART or for whom initiation was delayed until the CD4⁺ cell count was less than that value [4, 5]. Despite these conflicting data, HIV treatment guidelines have recently been revised to recommend that antiretroviral therapy should be initiated at CD4⁺ T lymphocyte counts of either ⩽200 cells/mm³ [6] or <350 cells/mm³ [7]. These recommendations are driven in large part by the toxicities associated with HAART, which include lipodystrophy syndrome, glucose intolerance, hyperlipidemia, and lactic acidosis [8, 9]. In addition, drug resistance can develop in patients with incomplete virologic suppression. Even prolonged suppression of viral replication is unlikely to eradicate HIV-1 infection [10].

We sought to assess the validity of not initiating HAART at a CD4⁺ T lymphocyte count of >350 cells/mm³ by comparing clinical disease progression between HIV-infected persons who did and those who did not begin to receive HAART at a level of >350 cells/mm³. In those for whom HAART was initiated, we assessed treatment durability and drug toxicity.

Patients and methods. We identified 2 groups of HIV-infected patients from the Johns Hopkins HIV Clinic Cohort (Baltimore, MD). Group 1 included all patients enrolled in the clinic on or after 1 July 1996 who began to receive HAART when they had a CD4⁺ T lymphocyte count of 350–499 cells/mm³ and who received >90 days of treatment. A HAART regimen was defined as 2 nucleoside reverse-transcriptase inhibitors (NRTIs) in combination with >1 protease inhibitor (PI) or a nonnucleoside reverse-transcriptase inhibitor (NNRTI). Regimens containing 3 NRTIs were also considered to be HAART. Patients who received hard-capsule saquinavir therapy but no other PI or NNRTI were excluded because of the relatively poor bioavailability of this drug when used alone. Group 2 included all patients enrolled in the clinic on or after 1 July 1996 who had a CD4⁺ T lymphocyte count of 350–499 cells/mm³ at baseline but did not receive HAART while in this CD4⁺ cell count stratum. This group included persons who began to receive HAART when they had a CD4⁺ T lymphocyte count of <350 cells/mm³, persons who received non-HAART regimens, and persons who never received therapy. All patients were assessed for symptoms of HIV disease at baseline. Persons did not need to be antiretroviral naive to be included in the study. Follow-up started with initiation of HAART (group 1) or at the first clinic visit (group 2) and was continued through 30 June 2001.

Clinical events were identified through diagnoses made by the patient's clinician, but all diagnoses were subsequently reviewed according to standardized criteria, and only diagnoses that met those criteria were retained in the analysis [11]. Clinical data were obtained from all patient visits.

Disease progression was defined as death or a new AIDS-defining illness, on the basis of the 1993 Centers for Disease Control and Prevention clinical case definition [12], excluding the criterion of a CD4⁺ T lymphocyte count of <200 cells/mm³. CD4⁺ T lymphocyte levels were quantified by flow cytometry; HIV-1 RNA levels were quantified by RT-PCR (Roche Molecular Systems). Categorical variables were compared with the Yates-corrected χ² test. Continuous variables were compared with use of the Wilcoxon rank sum test. We performed a Kaplan-Meier analysis of the time to a new AIDS-defining illness or death (combined end point) from either the date of initiation of the first HAART treatment regimen (for group 1) or the date of the initial clinic visit (for group 2); curves were compared using the log-rank test.

Results. The demographic and clinical characteristics of the study population are summarized in table 1. There was no difference in median age between the 2 groups; however, those for whom HAART was initiated while they had CD4⁺ T lymphocyte counts of 350–499 cells/mm³ were more likely to be male and of nonblack race and to have homosexual sexual contact as their HIV risk category than were persons who did not receive HAART while in this CD4⁺ cell count stratum. The median baseline CD4⁺ T lymphocyte count and HIV-1 RNA level did not differ between the 2 groups. The median date of enrollment was 3 September 1997 for persons in group 1, and it was 29 January 1998 for persons in group 2. The median duration of follow-up was longer among patients for whom HAART was initiated early (i.e., at CD4⁺ cell counts of 350–499 cells/mm³) than among those who did not, but there was no difference in loss to follow-up between the 2 groups. The median number of clinic visits per year was 6.5 for group 1 and 5.2 for group 2 (P =.09).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Kaplan-Meier analysis of time to a new AIDS-defining illness or death for the 2 groups of study patients. Group 1 comprised patients for whom HAART was initiated when their CD4⁺ T lymphocyte counts were 350–499 cells/mm³, and group 2 comprised patients for whom HAART was initiated at counts of <350 cells/mm³ or was never initiated. The curves represent the probability that patients would not experience a new AIDS-defining illness or death. The numbers of persons at risk during each 250-day interval are given below the graph. There was no significant difference between the 2 groups in the risk of clinical disease progression (P =.21, log-rank test).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Demographic and clinical characteristics of the study patients.

Of the 174 persons for whom HAART was not initiated at a CD4⁺ T lymphocyte count of 350–499 cells/mm³, 40 (23%) initiated therapy at counts of <350 cells/mm³. There were also 28 patients (16%) in group 2 who received non-HAART antiretroviral regimens. There was no significant difference in disease progression between patients who received non-HAART regimens and those who received no therapy, so group 2 was not stratified when compared with group 1.

The risk of clinical disease progression did not differ between group 1 and group 2 (figure 1; P =.21, log-rank test); the number of AIDS-defining events was low in both groups (20 and 23, respectively; table 2), and there was no difference between the risk for group 1 and for those patients for whom initiation of therapy was delayed or who did not receive HAART (P =.24). When death alone was used as the clinical end point, there was no difference in the risk of disease progression between groups 1 and 2 (P =.10, log-rank test). There was also no difference in the risk for disease progression between groups 1 and 2 when follow-up was considered to begin with the initial clinic visit (P =.17) or when patients who received non-HAART regimens were excluded from group 2 (P =.10).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2

Incidence of AIDS-defining opportunistic illness and death among the study patients.

Of the 159 patients in group 1, eighty-five (53%) had an HIV-1 load of >400 copies/mL at their most recent clinic visit (71 [45%] had had virus load measurements of >400 copies/mL on at least 2 occasions), and 65 (41%) experienced an adverse drug reaction that necessitated a change in therapy. The median increase in CD4⁺ T lymphocyte count was 36 cells/mm³. At the end of follow-up or at the most recent visit, 81 patients (51%) were receiving their second HAART regimen, and 8 (5%) were receiving their third regimen.

Discussion. Despite similar baseline CD4⁺ T lymphocyte counts and HIV-1 RNA levels among patients who did or did not begin to receive HAART at a CD4⁺ T lymphocyte count of >350 cells/mm³, highly active therapy did not slow clinical disease progression. However, there were several important adverse consequences of HAART. Most patients did not have durable virologic suppression. Almost one-half of the patients developed an adverse drug reaction requiring a change in regimen, and most patients were switched to a second HAART regimen during follow-up.

The median duration of follow-up was longer for patients who received HAART than for those who did not. This is probably because treatment was more likely to be withheld from the study population during the latter portion of the study period because of recent changes in treatment guidelines. The difference was not due to differential loss to follow-up. Because, early in the study period, clinicians were not aware of (and therefore did not assess for) toxicities that are now believed to be attributable to HAART (e.g., lactic acidosis, glucose intolerance, and hypertriglyceridemia), it is possible that the incidence of adverse drug reactions was underestimated. However, the firm end point of change in antiretroviral therapy due to an adverse event was uniform throughout the study period and likely represents clinically significant adverse events.

The sample size of the present cohort study was relatively small, such that small differences (<15%) in the incidence of clinical disease progression could not be detected statistically. However, even if the sample size were large enough to demonstrate a statistically significant difference in the incidence of disease progression, such a difference would likely be small compared with the high rates of nondurable virologic suppression and drug toxicity among persons initiating HAART (i.e., the magnitude of the risk-to-benefit ratio would not change).

It is important to keep in mind that this was a retrospective observational cohort study rather than a prospective, randomized, controlled clinical trial. There were differences in the demographic characteristics of groups 1 and 2 that could potentially influence the results of the study. However, to date, no clear differences in disease progression have been reported with respect to sex, race, or HIV risk category, so such differences may not be critical. Factors that have been associated with differences in HIV disease progression (CD4⁺ T lymphocyte count, HIV-1 RNA level, and age) did not differ between the 2 groups. In addition, the median duration of follow-up of 21-31 months in our study was relatively short; differences in disease progression between groups 1 and 2 could become apparent after longer follow-up time. Although observational studies can contain biases that affect comparisons between groups, a randomized, controlled trial to assess when to initiate HAART has not been (and, because of ethical concerns, likely never will be) performed. Thus, we must attempt to make inferences from observational studies.

The findings of the present study strongly support the current treatment guideline recommendation that HAART not be initiated at a CD4⁺ T lymphocyte count of >350 cells/mm³. Given the decreased likelihood of achieving virologic suppression with a second HAART regimen, compared with the initial regimen [13], it would appear most prudent to initiate HAART at CD4⁺ T lymphocyte levels at which such therapy will have a significant impact on clinical disease progression and to work to ensure that patients adhere to therapy, which will maximize the likelihood of durable virologic suppression.

• Received September 6, 2002.
• Accepted December 2, 2002.

• © 2003 by the Infectious Diseases Society of America