Prevalence and clinical correlates of HIV viremia (`blips') in patients with previous suppression below the limits of quantification

Objective: To examine the prevalence and clinical correlates of subsequently measurable viremia in HIV-infected patients who have achieved viral suppression below the limits of quantification (< 50 copies/ml).

Design: Non-randomized dynamic cohort study of ambulatory HIV patients in nine HIV clinics in eight cities.

Patients: Patients had two consecutive HIV-1 RNA levels < 50 copies/ml (minimum, 2 months apart) that were followed by at least two more viral level determinations while remaining on the same antiretroviral therapy (ART) between January 1997 and June 2000 (median 485 days). Transiently viremic patients were defined having a subsequently measurable viremia but again achieved suppression < 50 copies/ml.

Results: Of the 448 patients, 122 (27.2%) had transient viremia, 19 (4.2%) had lasting low-level viremia and 33 (7.4%) had lasting high-level viremia (defined as 50-400 and > 400 copies/ml, respectively). Only 16 (13.1%) of those who had transient viremia later had persistent viremia > 50 copies/ml. The occurrence of transient viremia did not vary with whether the patient was ART-naive or experienced (P = 0.31), or currently taking protease inhibitors or not (P = 0.08). On consistent ART, the median percentage increase in CD4 cell count was statistically different between subgroups of our cohort (Kruskal-Wallis, P = 0.002).

Conclusions: Transiently detectable viremia, usually 50-400 copies/ml, was frequent among patients who had two consecutive HIV-1 RNA levels below the limits of quantification. In this analysis, such viremia did not appear to affect the risk of developing lasting viremia. Caution is warranted before considering a regimen as 'failing' and changing medications.

Introduction

Several studies have suggested that achieving plasma HIV-1 RNA levels below the limits of detection (as low as < 20 copies/ml) within several weeks of initiating a new regimen is predictive of long-term success of highly active antiretroviral therapy (HAART) [^1,2]. However, many HIV-infected patients who achieve viral suppression below the limits of quantification while on HAART have measurable, but transient, viremia subsequently. It is still unclear, however, whether such transiently detectable viremia is detrimental to longer-term virologic control. On one hand, measurable viremia may be a harbinger of drug resistance and treatment failure. It has been suggested that low-levels of viremia may alter viral dynamics and change the slope of the decay curve of latently infected cells [³]. On the other hand, transiently measurable viremia may simply reflect variability in the assay, such as that resulting from specimen processing, or could be caused by extraneous factors, such as immunization or intercurrent illness. This study further characterizes this phenomenon in the setting of clinical practice including patients both naive and experienced to antiretroviral drugs and on both protease inhibitor (PI)-based and non-PI-based regimens in terms of long-term virologic and immunologic outcomes.

Methods

The HIV Outpatient Study (HOPS)

The Centers for Disease Control and Prevention's (CDC) ongoing HIV Outpatient Study, into which patients are continuously recruited, has to date collected data on the course of disease for more than 5500 HIV-infected, non-hospitalized patients, who have been seen in about 106 000 outpatient visits since 1992 [⁴]. The present analysis includes data on patients seen from 1 January 1997 to 30 June 2000. The study sites are nine clinics (seven private and two public) in eight United States cities (Chicago, Illinois; Denver, Colorado; Oakland and San Leandro, California; Philadelphia, Pennsylvania; Stony Brook, New York; Tampa, Florida; Washington, DC) that provide care for at least 150 HIV-infected patients each per year. HOPS participating physicians routinely care for hundreds of HIV-infected patients and have extensive experience treating HIV. This study has been reviewed and approved by CDC and local institutional review boards since its inception.

Information in five general categories is abstracted from the chart for each outpatient visit and entered electronically by trained research coordinators. The data are compiled centrally, reviewed, and routinely processed for quality control. Because the participating physicians are the source of primary care for these patients, all symptoms, diagnoses, and treatments since the previous visit, including interim changes, are noted at each clinic visit. The categories of information abstracted are as follows: demographic characteristics and risk factors for HIV infection; symptoms; diagnosed diseases (both definitive and presumptive diagnoses); medications prescribed, including the dose and duration; and laboratory values, including CD4 cell counts and measurements of plasma HIV-1 RNA [⁴].

Patients

During the period of present analysis, there were 3772 active patients in the HOPS database, 448 of whom met the inclusion criteria for this analysis. To be included, patients must have had two consecutive HIV-1 RNA levels below the limits of detection (< 50 copies/ml), 2 or more months apart, and at least two subsequent viral load measurements with no change in antiretroviral therapy (ART). Dose modification of an antiretroviral agent was not considered a change in therapy, nor was a treatment interruption of less than 30 days. In this select cohort from the HOPS, no patient had a recorded complete interruption in ART. A patient was not included if the only detectable viral load was the last one he or she had during the study period. Measurements of HIV-1 RNA were performed using the reverse transcriptase polymerase chain reaction RT-PCR technique (Amplicor HIV-1 Monitor, Roche Diagnostics Corporation, Indianapolis, Indiana, USA) in 91.1% of the assays and by the b-DNA technique (Chiron Corporation, Emeryville, California, USA) in the remainder.

For the purpose of analysis, patients were classified into one of four groups. The first group were 'suppressed' patients - that is, patients who maintained HIV-1 RNA levels below the limits of quantification on all determinations throughout the study period.

The second group, 'transiently viremic' patients, were those who had measurable viremia (`blips') after two determinations < 50 copies/ml and who again achieved suppression below the limits of quantification. This group were examined to see whether, during the remainder of the period of observation, there was a subsequent episode of transient or lasting detectable viremia after the first blip and return below the limits of quantification.

The third and fourth groups of analysis were patients with persistent viremia, i.e., 'lasting rebound' viremia after two determinations < 50 copies/ml; these patients never again achieved viral suppression below the limits of quantification during the study period. Because persistent viremia may be clinically important, these patients were further grouped as 'high level', if the level of measurable viremia was ever > 400 copies/ml, and 'low-level' if measurable viremia was persistently at levels of 50-400 copies/ml. These numerical cut-off points were chosen based on the limits of detection of the widely available viral load assays.

Statistical analysis

Data were analyzed with SAS software (version 6.12; SAS Institute, Cary, North Carolina, USA). Patient characteristics, prevalence, and clinical correlates of measurable viremia were compared across groups, classified by level of virologic control, using chi-square analyses for categorical variables and the non-parametric Kruskal-Wallis test of multiple medians for continuous variables. All medians are reported with interquartile ranges (IQR) Q1-Q3 in the tables. Comparisons of risk between subgroups of the cohort for subsequent viremia after the initial HIV-1 RNA measurements below the limits of detection were made with risk ratios.

Results

Patient characteristics

The analyses include data collected during the period 1 January 1997 to 30 June 2000 on 448 HIV-infected persons who met study inclusion criteria. Patients were classified according to their level of virologic control. Except for payer status, no statistically significant differences were found between groups compared with regard to a number of demographic and clinical variables examined, including age, sex, race, primary risk behavior, highest level of education, baseline CD4 cell count and HIV-1 RNA level (Table 1).

Table 1

Given the dynamic nature of this cohort, the study period differed for each patient. The median length of observation was 485 days (69 weeks), which did not differ statistically among the groups. However, the number of follow-up viral load measurements after achieving viral suppression below the limits of quantification did differ among compared groups (Kruskal-Wallis test, P = 0.0002) (Table 1).

Prevalence and clinical correlates of transient and lasting viremia

Of all 448 patients, 274 (61.2%) maintained viral loads below the limits of quantification throughout the study. The only association with maintenance of undetectable viral loads was health payer status: 232 (64.6%) of 359 privately insured patients, but only 30 (47.6%) of 63 patients with public subsidy (Medicare, Medicaid), maintained viral levels < 50 copies/ml throughout the study period [relative risk (RR), 1.36; 95% confidence interval (CI), 1.04-1.78; Table 1).

Transient viremia occurred in 122 (27.2%) of the study participants and nine (7.4%) of them had more than one blip (i.e., 22.5 blips/100 person-years). Seventy-eight percent of the time, these episodes were of low-level viremia, 50-400 copies/ml. After the first detectable measurement, 106 (87%) of patients returned below the limits of quantification on the next measurement; all transiently viremic patients returned below the limits of quantification by the third measurement after the blip. On continued follow-up, however, 16 (13.1%) of the original 122 patients experiencing transient viremia subsequently had a 'rebound' with viremia > 50 copies/ml that never returned below the limits of quantification during the study period. Of these 16 patients, the previous episode of transient viremia was low-level in 75%; none was a patient who had experienced more than one blip episode.

Nineteen (4.2%) of the 448 patients in the study population had a rise from 'undetectable' viral loads to lasting levels of 50-400 copies/ml (low-level rebound), and 33 (7.4%) had lasting viremia > 400 copies/ml (high-level rebound). During the period of study, 16 (13.1%) of the 122 patients who had transient viremia developed persistent low or high level (`rebound') viremia at some later point, compared with 52 (16.0%) of the remaining 326 patients in the cohort, but the difference was not statistically significant (RR, 0.82; 95% CI, 0.49-1.38).

The median first detectable RNA level was 96 copies/ml among transiently viremic patients, 94 copies/ml among the lasting low-level viremic patients and 426 copies/ml among the lasting high-level viremic patients (Kruskal-Wallis test, P = 0.0001). Median CD4 cell counts at the time of first detectable RNA were 486.5 × 10⁶ cells/l among transiently viremic patients, 628 × 10⁶ cells/l among the patients with lasting low-level viremia, and 550 × 10⁶ cells/l among the patients with lasting high-level viremia. There was no significant difference between the groups for this parameter (Kruskal-Wallis test, P = 0.25).

Two important clinical parameters, antiretroviral experience and composition of the antiretroviral regimen (PI-containing or non-PI-containing) were compared for their association with transient viremia. There was no evidence to suggest that blips were more likely to occur among patients who were previously taking or not taking ART (ART-naive versus ART-experienced patients; P = 0.31), or in patients on PI regimens versus those on non-PI regimens (P = 0.08). However, patients with persistent rebound viremia had been exposed to a greater number of antiretroviral agents than those who remained suppressed or who experienced a transient episode of viremia (Kruskal-Wallis test, P = 0.01). (Table 2).

Table 2

CD4 cell count response to consistent antiretroviral therapy

As an index of immune reconstitution, CD4 cell counts were compared from before the start of the antiretroviral regimen to the last recorded measurement at the end of the study period. Recall (Table 1), the median length of follow-up did not significantly differ among the groups. Interestingly, on consistent ART, the median percentage increase in CD4 cell count was statistically different between subgroups (Kruskal-Wallis test, P = 0.002; Table 3 and Fig. 1).

Fig. 1

Table 3

Discussion

Transiently detectable viremia, usually 50-400 copies/ml, was found to be a common phenomenon, occurring in about one-quarter of patients who had achieved viral suppression below the limits of quantification while remaining on the same antiretroviral regimen. These blips of viremia appear to represent no increased risk for subsequent virologic rebound. Rebound to persistent levels of viremia of 50-400 copies/ml occurred in fewer than 5% of patients, and lasting rebound viremia > 400 copies/ml, 'virologic failure,' occurred in fewer than 10%. There was no statistically significant evidence that patients who had had a previous episode of transient viremia were at a greater risk of developing persistent viremia than those who did not experience transient viremia.

The phenomenon of transient viremia has been previously examined in a small case series [⁵], in the setting of clinical trials [⁶], and in two retrospective cohort studies [^7,8]. While comparison between studies is problematic as study definitions are not consistent, investigators in one clinical trial (ACTG 343) found that intermittent viremia of 50-200 copies/ml occurred in 40% of patients, while intermittent viremia to levels > 200 copies/ml occurred in 20% of patients. They also concluded that intermittent viremia was not associated with a greater risk of viral rebound [⁶]. Data on low-level viral rebound and blips of 50-500 copies/ml has also been reported from the Swiss HIV Cohort Study and the Frankfurt HIV Clinical Cohort (seen in 32.5% of patients, which is 37.4 episodes/100 person-years) [⁸]. Thus, overall prevalence of this phenomenon appears similar across all studied populations.

In the present study, our data suggest that CD4 cell count response to ART, but not antiretroviral experience (as opposed to no prior ART treatment) or PI use, was correlated with transient viremia. The immune response (CD4 cell rise) among the transiently viremic patients seen in this study contrasts somewhat with other reports [^7,9]. For example, investigators from three London HIV centers have suggested that CD4 cell count response to therapy was not as large in those experiencing virological blips as in those maintaining undetectable viral loads [⁷]. Similar findings were demonstrated in a study from the Johns Hopkins cohort, with those having a sustained virologic response to HAART experiencing the greatest rise in CD4 cell count at 24 weeks of study [⁹].

A rise from an 'undetectable' RNA level on any given measurement is not necessarily an indication of failure of an antiretroviral regimen. The great majority of patients will again have 'undetectable' viral loads without any change in ART. Although patients with lasting rebound may have higher viral loads at first detection, there is no way to predict at this point which patients will subsequently return to undetectable levels. In the present study, all those whose viremia was transient returned below the limits of detection by the third follow-up measurement. Therefore, only a small proportion of patients (7.4%) had lasting viremia > 400 copies/ml, requiring adjustments in their antiretroviral regimen.

The patients and physicians participating in the HOPS were diverse and reasonably representative of the HIV-infected population receiving medical care in the United States. Since the HOPS database is a dynamic cohort, patients are followed for various times; additional data points could result in patients being reclassified to another group. Interestingly, however, the longest median length of follow-up was among those patients who experienced transient viremia. As would be expected from a clinical perspective, viral load measurements were obtained more often in those patients who experienced transient viremia and lasting viremia > 400 copies/ml. It is possible that blips were discovered because of more frequent sampling in those groups. However, when the length and frequency of follow-up were compared among the transiently viremic patients and the patients with low- and high-level lasting viremia, from the time of first detectable viremia, there were no differences between the groups. It is important to note that baseline CD4 cell counts and HIV-1 RNA levels were comparable among the groups. The power of the study to detect differences between groups was limited by the small number of patients with lasting, low-level and high-level viremia.

With the newly revised US Department of Health and Human Services guidelines issued in February 2001, the 'when to start' question has been revisited [¹⁰]. There are fewer data on the 'when to switch' question, i.e. what defines a failing regimen. Clinically, there appears to be a difference between having ever achieved viral suppression below the limits of quantification and remaining below this threshold after initial suppression. This is particularly relevant given the current limited therapeutic options available to many patients and the mounting evidence of toxicities of some of these agents. Caution is warranted before considering a regimen 'failing' and changing therapies. This clinically oriented observational study does not attempt to explain the etiology of this phenomenon, which is likely related to factors such as adherence or pharmacologic variables, such as achievable drug levels. Alternatively, these blips may represent truly intermittent viremia of unknown cause. Clearly, longer follow-up of patients is necessary to address such issues. The likely reality is that low levels of ongoing viral replication are constantly occurring below the limits of detection of commercially available assays; the defined thresholds are arbitrary. However, the occurrence of detectable blips should trigger the clinician to readdress the issue of adherence with the patient [¹¹].

Acknowledgments

The authors thank Alan Greenberg and Ida Onorato (Centers for Disease Control and Prevention, Atlanta, Georgia) and Gary Simon and David Parenti (George Washington University Medical Center, Washington, DC) for their support, helpful advice, and editorial suggestions.

Sponsorship: This study was supported by cooperative agreement (UC64/CCU5096889-03) between the Centers for Disease Control and Prevention and the Health Research Network of APACHE Medical Systems.

Note: Approved by the Office for Protection from Research Risks, National Institutes of Health, Single Project Assurance S-017605-01, Project Number U64/CCU513976-01, Epidemiological Research Studies of AIDS & HIV Infection, June 1, 1999, in compliance with Department of Health & Human Services regulations (45 CFR 46).