Initiating Antiretroviral Therapy During HIV Infection: Title and subTitle BreakConfusion and Clarity

Highly active antiretroviral therapy (HAART) has changed the landscape of human immunodeficiency virus (HIV) care in the developed world. Many patients with access to antiretroviral therapy (ART) have benefited from the dramatic reductions in mortality and morbidity, and HIV disease has become one of relative chronicity for most but not all infected patients.^{1 - 3}

The success of HAART has now led to research into approaches to rid virally suppressed patients of residual HIV reservoirs.⁴ Nonetheless, as with several chronic diseases, the treatment often has significant adverse effects.⁵ This is the case with virtually all drugs in the various classes of antiretroviral compounds approved by the Food and Drug Administration.⁵ As such, physicians have dealt with a pendulum effect in decisions regarding when to initiate therapy during HIV infection.^{5 - 7}

Following the development of HAART, many physicians were quite aggressive in treating patients at virtually any stage of this human retroviral disease, almost regardless of the CD4 T-lymphocyte count and plasma HIV RNA level. Because of the increasingly reported serious adverse effects of the diverse drug constituents of HAART, studies were conducted to attempt to determine the time at which initiation of ART was most efficacious, based on clinical end points and surrogate markers. Clinical guidelines^{5 - 7} are now suggesting potential benefit in initially withholding ART for certain therapy-naive patients based on baseline CD4 T-lymphocyte counts and plasma viral RNA levels. Nonetheless, this approach has remained extremely controversial in clinical retrovirology. In this issue of THE JOURNAL, Phillips et al⁸ and Hogg et al⁹ report 2 large studies that add some clarity to this contentious issue.

The article by Phillips et al⁸ examines the correlation between the plasma HIV RNA response to ART and baseline CD4 T-lymphocyte counts and plasma HIV RNA levels. This analysis of 3 large cohort studies in Europe evaluated 3430 therapy-naive patients. The authors found no difference in achieving undetectable plasma viral RNA levels, defined as less than 500 copies/mL at 32 weeks, regardless of baseline CD4 T-lymphocyte count or plasma HIV RNA level. Of note, a relatively large percentage (85%) of the total patients in this study achieved this level of viral suppression at 32 weeks. There was also no difference in viral rebound, regardless of baseline CD4 T-lymphocyte count or plasma viral load. However, a baseline plasma viral RNA level greater than 100 000 copies/mL did yield a slower rate of viral suppression after treatment.

Plasma viral RNA levels are usually stable, to within 0.1 log copies/mL per year in patients with HIV infection who are not yet treated with antiretroviral agents. In untreated groups of patients, the risk of progression to acquired immunodeficiency syndrome (AIDS) before the CD4 T-lymphocyte count is less than 200 × 10⁶/L and before the plasma viral load is greater than 10 000 copies/mL is low.^{5 - 7} The study by Phillips et al is important because it evaluates relatively large numbers of therapy-naive patients followed for more than 2 years, and also confirms and extends a previous study by Cozzi Lepri et al.¹⁰ As such, if only these large cohort analyses are considered, it could be suggested that therapy should be withheld for indefinite periods before initiation in patients, at least when solely based on a potential response in plasma viral load. Nonetheless, there is still a concern that immune recovery may not be as complete once CD4 T-lymphocytes have been depleted beyond a certain point. Furthermore, younger HIV-infected individuals may have an increased level of CD4 T-lymphocyte restoration after initiation of HAART compared with older patients,¹¹ possibly due to better preserved thymic tissue and function. Thus, subgroups of patients may be quite heterogeneous with regard to the meaning of prognostic indicators.

Moreover, it is important to note that Phillips et al used a cut-point level of 500 copies/mL rather than the more sensitive level of 50 copies/mL to define undetectability in plasma viral load. The authors did have viral load data using the 50 copies/mL level in a subgroup of 609 patients who were seen in 2 clinics after January 1998. Although this group is substantially smaller than the overall study cohort, the analysis confirms no association between baseline HIV RNA and CD4 T-lymphocyte count and viral suppression to this lower level of 50 copies/mL. In other smaller studies, viral response and its duration were directly correlated with baseline plasma viral load and time to undetectability of plasma viral load.¹² Duration of viral suppression has been observed to be greater in patients with a viral load below 50 copies/mL than in those with levels of 50 to 400 copies/mL after initiating HAART.¹³ In addition, consideration of the difference in effects between men and women, with women possibly progressing to AIDS at somewhat lower plasma viral loads than men,¹⁴ may be useful when assessing the implications of the study by Phillips et al for clinical decision making for HIV-infected individuals. However, differences between men and women in the correlation between plasma viral load and disease progression remain controversial. A large prospective study to confirm the data in the present article is needed, but will be difficult to perform.

In the second article, Hogg et al⁹ conducted a population-based study of 1219 therapy-naive HIV-infected individuals in Canada to determine rates of disease progression stratified by baseline CD4 T-lymphocyte count and plasma HIV RNA level prior to initiating HAART. Patients with CD4 T-lymphocyte counts of less than 50 × 10⁶/ L and those with counts of 50 × 10⁶/ L to 199 × 10⁶/ L were 6.67 and 3.41 times, respectively, more likely to die than patients with CD4 T-lymphocyte counts of 200 × 10⁶/ L or greater. The CD4 T-lymphocyte count was the only independent prognostic indicator for progression to AIDS or death. Patients with baseline CD4 T-lymphocyte counts of 200 × 10⁶/ L or greater had low rates of progression to AIDS or death, and these outcomes were clustered among those with a CD4 T-lymphocyte cell count of less than 200 × 10⁶/ L. This finding is consistent with several previous studies.⁵ A survival benefit was observed at a cutoff of 100 000 copies/mL of plasma viral RNA, but only in univariate and not multivariate analyses. This lack of prognostic importance of plasma HIV RNA levels differs from most natural history studies,^{15 - 17} but this difference is likely due to a relatively robust effect of HAART in suppressing plasma viral RNA in most patients.

The complexity of HIV pathogenesis has been best defined in natural history studies. Factors, such as viral subtype, patient HLA profiles, possible differences based on sex, and levels of T-cell activation, interact with plasma HIV RNA and CD4 T-lymphocyte levels to affect prognosis of therapy-naive patients.^{14 ,18 - 20} Prognostic factors in studies of untreated HIV-infected individuals may differ from those in studies of patients treated with HAART, since even with viral rebound to baseline levels there is decreased clinical progression if most regimens of HAART are maintained after rebound.²¹ This phenomenon may be due to alterations in viral fitness, viral virulence, and immune modulations secondary to HAART.²²

Of importance, both of the studies reported in this issue of THE JOURNAL are confounded by possible lead-time biases, as the exact time of seroconversion was impossible to precisely analyze. Thus, even though inferences regarding overall survival benefit cannot be made, the study by Hogg et al confirms the prognostic value of baseline CD4 T-lymphocyte counts demonstrated in 3 recently reported cohorts.^{23 - 25} The conclusion of the study by Hogg et al is that initiation of HAART should begin prior to CD4 T-lymphocyte counts decreasing to less than 200 × 10⁶/ L. Of note, rates of progression to AIDS and death at a median of 28 months were independent of protease inhibitor use, as well as patient age and sex.

So how can clinicians use these 2 studies to yield more clarity when determining when to initiate ART in HIV-infected individuals? First, the studies sought to evaluate different hypotheses. In the article by Phillips et al, the end point was plasma virological control and in this case there was little evidence that lower CD4 T-lymphocyte counts or higher plasma viral loads were associated with a poorer antiviral response. In the study by Hogg et al in which the outcomes were development of AIDS or death, there was a clear association between occurrence of these clinical end points and a baseline CD4 T-lymphocyte count of less than 200 × 10⁶/ L. This critical finding suggests that virological response to ART is not a certain indication for good clinical outcome during relatively long-term follow-up, if the CD4 T-lymphocyte count is profoundly attenuated prior to initiating therapy.²⁶ Thus, the study by Hogg et al also demonstrates that plasma viral load is not the only predictor of the effects of triple-drug therapy for HIV-infected patients.

Based on these studies, several somewhat divergent clinical conclusions could be drawn. Nevertheless, it seems that a reasonable conclusion would be to focus primarily on the CD4 T-lymphocyte count for determining initiation time for ART in many infected patients. A CD4 T-lymphocyte count of 200 × 10⁶/L in peripheral blood appears to be the critical level in the study by Hogg et al⁹ and in previous analyses,⁵ and it would seem prudent to ensure that ART is initiated before CD4 T-lymphocyte counts decrease below this level. Acknowledging that this parameter does not apply to all patient subgroups, most patients could be monitored closely, rather than immediately beginning therapy with drugs that have potential significant adverse effects over several years of therapy (eg, lipodystrophy, mitochondrial toxicity, lipid abnormalities, osteopenia, and lactic acidosis).²⁷

Consistent with published guidelines, which use somewhat different treatment-initiation benchmarks, the present studies appear to support postponing the initiation of ART for some patients until the CD4 T-lymphocyte count approaches 200 × 10⁶/ L at least. This recommendation would not be appropriate for patients evaluated within 6 months of primary HIV seroconversion. For those patients, immediate initiation of HAART may preserve HIV-specific T-helper cell function. If an individual patient's CD4 T-lymphocyte count is declining relatively rapidly, the physician should monitor this progression carefully to intervene with HAART before the CD4 T-lymphocyte level decreases below 200 × 10⁶/ L and the chances of immune restoration are reduced. Further studies will be necessary to determine whether certain subgroups of patients, based on age or dynamics of CD4 T-lymphocyte decline, should be followed closely and therapy initiated at different stages of clinically documented immune suppression.