HiV Articles  
Plasma HIV-1 RNA Detection Below 50 Copies/mL and Risk of Virologic Rebound in Patients Receiving Highly Active Antiretroviral Therapy
  Dowload the PDF here

Clinical Infectious Diseases Advance Access published January 11, 2012

".....our study demonstrates that in treated patients monitored with the RealTime assay, a VL <50 but >40 copies/mL, and, to a lesser extent, qualitative RNA detection <40 copies/mL, indicate that treatment efficacy should be reviewed. Further studies, including possibly a controlled trial, are required to support these recommendations....."

Tomas Doyle,1,3 Colette Smith,4 Paola Vitiello,3 Valentina Cambiano,4 Margaret Johnson,2,5 Andrew Owen,6 Andrew N. Phillips,4 and Anna Maria Geretti1,3,7 Departments of 1Virology, and 2HIV Medicine, Royal Free Hampstead NHS Trust; Departments of 3Virology; 4Infection & Population Health, and 5HIV Medicine, UCL Medical School, London; 6Department of Molecular and Clinical Pharmacology, and 7Institute of Infection and Global Health, University of Liverpool, United Kingdom

(See the Editorial Commentary by Gandhi and Deeks, on pages 738-40.)


Background. Plasma human immunodeficiency virus type 1 (HIV-1) RNA suppression <50 copies/mL is regarded as the optimal outcome of highly active antiretroviral therapy (HAART). Current viral load (VL) assays show increased sensitivity, but the significance of RNA detection <50 copies/mL is unclear.

Methods. This study investigated the virologic outcomes of 1247 patients with VL <50 copies/mL at an arbitrary time point during HAART (= T0), according to whether the actual, unreported T0VL was 40-49 copies/mL, RNA detected <40 copies/mL (RNA1), or RNA not detected (RNA2), as measured by the Abbott Real Time assay. Predictors of rebound >50 and >400 copies/mL over 12 months following T0 were analyzed with Cox proportional hazards models incorporating the T0VL and demographic and clinical data.

Results. Rebound rates >50 copies/mL were 34.2% for T0VL 40-49 copies/mL, 11.3% for RNA1, and 4.0% for RNA2; rebound rates >400 copies/mL were 13.0%, 3.8%, and 1.2%, respectively. The adjusted hazard ratios for rebound .50 copies/mL were 4.67 (95% confidence interval, 2.91-7.47; P <.0001) and 1.97 (1.25-3.11; P , .0001) with T0VL 40-49 copies/mL and RNA1, respectively, relative to RNA2, and 6.91 (2.90-16.47; P <.0001) and 2.88 (1.24-6.69; P <.0001), respectively, for rebound >400 copies/mL. The association was independent of adherence levels.

Conclusions. In treated patients monitored by RealTime, a VL of 40-49 copies/mL and, to a lesser extent, RNA detection <40 copies/mL predict rebound >50 and >400 copies/mL independently of other recognized determinants. The goal of HAART may need to be revised to a lower cutoff than 50 copies/mL.

Guidelines have traditionally recommended sustained viral load (VL) suppression <50 copies/mL as the optimal outcome of highly active antiretroviral therapy (HAART) [1]. Recent guidance has recommended suppression below the detection limit of commercial assays [2, 3], but also increased the threshold for virologic failure to 200 copies/mL [3]. Although the target level of suppression was initially defined by the technical properties of the assay, clinical trials and observational studies have shown that maintaining a VL ,50 copies/mL predicts long-term virologic success and immunologic and clinical benefits [4-6]. In many treated patients, human immunodeficiency virus type 1 (HIV-1) RNA remains detectable in plasma below the cutoff of 50 copies/mL establishing residual viremia that in some populations appears to plateau at 3-10 copies/mL and is not responsive to HAART intensification [7-14].

In recent years, new commercial assays have been introduced based on real-time polymerase chain reaction that have a lower limit of quantification of 20 or 40 copies/mL and can also report qualitative RNA detection below these thresholds [15]. Although the assays have now entered routine practice, the clinical significance of RNA detection <50 copies/mL during HAART is unknown. In particular, it is unclear whether VL values that fall between the proposed plateau of residual viremia and the 50 copies/mL cutoff should be regarded as an indication that the potency and tolerability of the regimen, adherence, drug resistance, and pharmacokinetics should be reviewed. Our center introduced the RealTime HIV-1 VL assay (Abbott Molecular, Maidenhead, UK) in 2006. The assay quantifies the viral load over the range of 40-10 000 000 copies/mL (1 mL input) and reports qualitative RNA detection <40 copies/mL. The manufacturer indicates detection of 30, 20, and 10 RNA copies/mL in 96%, 88%, and 68% of cases, respectively. Following the introduction of the assay in our laboratory, VL levels of 40-49 copies/mL and qualitative RNA-detected results were reported as <50 copies/mL. The actual results were not released to the treating physician. Here, we investigated the virologic outcomes of treated patients showing a VL ,50 copies/mL at an arbitrarily selected time point during HAART (5 T0) according to whether the actual, unreported T0VL result 40-49 copies/mL, RNA detected <40 copies/mL, or RNA not detected.


In patients receiving HAART and showing VL <50 copies/mL, HIV-1 RNA detection below this cutoff predicted a risk of rebound >50 and >400 copies/mL during follow-up. The effect was independent of other recognized determinants of rebound, including adherence levels. Although further supportive evidence is desirable, the findings indicate that the goal of HAART may need to be revised to a lower cutoff than 50 copies/mL.

Using sensitive testing methods, HIV-1 RNA can be detected in plasma in a large proportion (>80%) of patients receiving HAART and showing a VL <50 copies/mL for many years [7-14]. Studies have shown that HIV-1 RNA levels decline to <50 copies/mL within approximately 12 weeks of starting therapy [8, 9]. Once levels are <50 copies/mL, they continue to decline for several months before reaching a plateau at approximately 3-10 copies/mL [7-14]. The source of this residual viremia ismuch debated [7-14, 19-22]. In recent studies, residual viremia was unresponsive to HAART intensification [11, 13]. Although not excluding ongoing replication in ''sanctuary'' sites, the lack of response provides support to the hypothesis that residual viremia does not reflect ongoing virus replication, but rather originates from virus reactivation in latently infected

cells, with rapid suppression by HAART. The alternative model proposes ongoing virus replication due to suboptimal drug levels, penetration, or activity. It seems plausible that the 2 models may coexist. In a previous study, abacavir intensification of a 2-drug regimen of efavirenz and indinavir in patients with RNA levels below 50 copies/mL and above 2.5 copies/mL lowered the levels to <2.5 copies/mL [23]. Conversely, RNA levels have been seen to increase before rebounding >50 copies/mL in patients simplifying triple HAART to ritonavir-boosted atazanavir [24]. HIV-1 genetic evolution has also been observed with RNA level above 6.5 copies/mL [12]. These findings suggest that there is a threshold level of viremia <50 copies/mL that is indicative of ongoing virus replication. Our data are consistent with this view, and although they fall short of precisely defining a cutoff, the observed relevance of detecting RNA <40 copies/mL suggests that the threshold may be even lower than 40 copies/mL.

Less than half of the patients who experienced rebound >400 copies/mL showed RAMs, which suggests poor adherence as a major driver of rebound. Resistance is only 1 of the possible negative outcomes of rebound. Previous studies have shown an association between VL >400 copies/mL during HAART and mortality [25]. Measuring adherence is complex, and all methodologies have shortcomings. We previously validated a measure of adherence based on pharmacy prescription records [16, 17]. We showed that calculating the proportion of days covered by a prescription is a simple way of estimating adherence and predicting virologic failure. In this study, adherence

was not associated with rebound in adjusted analyses, indicating that the viral load was a more powerful predictor. We also measured efavirenz plasma levels in the T0VL sample as a possible indicator of adherence. We detected no significant differences in drug levels between groups. It should be noted that, given the retrospective nature of the study, the timing of the last HAART dose relative to the collection of the T0VL sample was not available. Efavirenz levels, however, are not expected to fluctuate considerably.

We found that a single VL measurement had a strong predictive value for rebound. In clinical practice, it is recommended that a detectable VL result during HAART should be confirmed in a subsequent sample prior to making management decisions [1]. In this study, the risk of rebound was associated with confirmed RNA detection in the subsequent VL measurement. However, it should be noted that the next available VL test was performed approximately 12-16 weeks after T0, whereas prompt confirmation of viremia is generally sought in clinical practice. Further studies are required to determine the relevance of seeking immediate confirmation.

Our analysis included patients regardless of treatment duration and was adjusted for this factor. Longer duration of suppression <50 copies/mL prior to T0 was associated with a reduced risk of rebound, in line with previous reports [26-28]. To apply our results to clinical practice, one should take into account that the median times to RNA levels <50 copies/mL, <40 copies/mL, and to RNA-neg were 4.1, 4.4, and 6.2 months, respectively. This, however, does not detract from the strong independent predictive value of the T0VL for the risk of virologic rebound. Although we did not observe significant differences between efavirenz-treated and lopinavir/ritonavir-treated patients, this analysis should be repeated with a larger data set, preferably within a randomized study. The composition of the regimen appears to play a significant role in both the T0VL result and the subsequent risk of rebound. Patients on NNRTI-based regimens were more prevalent in the RNA2 group and showed a lower risk of rebound. It has been previously observed that, compared with patients on other regimens, those receiving NNRTI-based HAART were less likely to experience viremia between 50 and 400 copies/mL in the first year after achieving <50 copies/mL, and were also less likely to rebound >400 copies/mL [5]. Others have reported a lower risk of low-level viremia in patients receiving NNRTIs (especially nevirapine) compared with other regimens [29, 30]. These observations may reflect different tolerability profiles, which may influence adherence, and different pharmacokinetic properties, which may influence penetration into ''sanctuary'' sites and modify the consequences of incomplete adherence. In addition, given that NNRTI-based regimens are the recommended first-line therapy in the United Kingdom, patients on these regimens were either less experienced or may have been perceived as being at lesser risk of nonadherence than those given PIs. Controlled studies are needed to confirm the findings.

Our study has limitations. Demographic and clinical data, including adherence estimations and drug levels, were available for a subset of patients, albeit this was a large and well-defined subset. We were unable to perform further testing to identify a level of viremia below which the risk of virologic rebound was no longer apparent. We were also unable to assess the performance of other commercial VL assays, and must caution against uncritically extrapolating our findings to assays with different detection and reporting ranges. Nonetheless, our study demonstrates that in treated patients monitored with the RealTime assay, a VL <50 but >40 copies/mL, and, to a lesser extent, qualitative RNA detection <40 copies/mL, indicate that treatment efficacy should be reviewed. Further studies, including possibly a controlled trial, are required to support these recommendations.


Study Population

The population comprised 1247 patients with T0VL 40-49 copies/mL (n = 240; 19.2%), RNA+ (n 5 507; 40.7%), or RNA-neg (n = 500; 40.1%; Table 1). Overall, 891 of 1247 (71.5%) patients were from the RFH and had demographic and clinical data available, including adherence estimations. There were several differences in the characteristics of the study population at T0. Patients with RNA-neg had longer duration of HAART and suppression <50 copies/mL prior to T0, a marginally lower pre-HAART viral load, and higher CD4 lymphocyte cell counts, and were more likely to be of white ethnicity and on nonnucleoside reverse-transcriptase inhibitor (NNRTI)-based therapy. To determine the time to VL <50 copies/mL, <40 copies/mL, or RNA-neg, we performed a retrospective analysis among 78 patients who started first-line tenofovir/emtricitabine plus either efavirenz (n 5 38) or ritonavir-boosted lopinavir (n = 40). The median (95% confidence interval [CI]) time to suppression <50 copies/mL, <40 copies/mL, and to RNA-neg was 4.1 (3.3-5.1), 4.4 (3.7-5.4), and 6.2 (5.4-7.2) months, respectively. There was no significant difference between the 2 treatment regimens in the time to RNA-neg (P 5 .93).

Rebound After T0

The time to rebound according to the T0VL and each of the 4 definitions of rebound is shown in Figure 1. In the primary analyses, 211 patients experienced confirmed (or last available measurement; n = 56) rebound >50 copies/mL, and 80 patients experienced confirmed (or last available measurement; n = 21) rebound .400 copies/mL. Rebound rates differed significantly according to the T0VL. After 12 months, the Kaplan-Meier estimates of the probability (95% CI) of confirmed (or last available) rebound >50 copies/mL were 34.2% (28.1%40.3%), 11.3% (8.5%-14.0%), and 4.0% (2.3%-5.7%) for the T0VL groups 40-49 copies/mL, RNA1, and RNA2, respectively (P < .0001 for all). The estimates for the >400 copies/mL cut-off were 13% (8.6%-17.3%), 3.8% (2.1%-5.4%), and 1.2% (0.2%-2.2%) for the T0VL groups 40-49 copies/mL, RNA1, and RNA2, respectively (P < .0001 for all).

Predictors of Rebound

In the univariable analysis, predictors of confirmed (or last available) rebound >50 copies/mL in the whole population comprised the T0VL group and the length of suppression <50 copies/mL prior to T0; additional predictors identified in the RFH population were the time since starting HAART, CD4 cell count and HAART regimen at T0, age, ethnicity, risk group, and adherence levels (Table 2). In the multivariable analysis, the T0VL group and the length of suppression <50 copies/mL prior to T0 (and for the RFH population, the HAART regimen) were the only independent predictors of rebound (Table 2). Overall, rebound >50 copies/mL was least likely in patients with T0VL RNA-neg, those with longer duration of suppression <50 copies/mL prior to T0, and those receiving NNRTI-based HAART. The observations were similar in the analysis of confirmed (or last available) rebound >400 copies/mL, showing that the T0VL group and the length of VL suppression <50 copies/mL prior to T0 were strong independent predictors of rebound (Table 3).

In a subanalysis, we analyzed the VL measurement obtained at the next time point after T0, referred to as T1VL. This followed T0 by a median (range) of 12 (8-17) weeks, 15 (12-19) weeks, and 16 (12-18) weeks for the T0VL groups 40-49 copies/mL, RNA1, and RNA-neg, respectively. In the Cox proportional hazards model, relative to patients with RNA2 at T0 (n = 485), the adjusted hazard ratios (HRs; 95% CI) for confirmed (or last available) rebound >400 copies/mL were 1.52 (0.37-6.22) for RNA 40-49 copies/mL or RNA+ at T0 and RNA-neg at T1 (n = 335), and 10.42 (3.36-32.33) for RNA 40-49 copies/mL or RNA+ at T0 and any RNA detection <400 copies/mL at T1 (n = 180) (P < .0001) (Table 4). The results were consistent when analyzing the RFH population alone (data not shown).

Next, we investigated T0 efavirenz plasma levels among 186 RFH patients receiving efavirenz-based HAART, comprising 90 patients with T0VL either 40-49 copies/mL or RNA+ and 96 patients with RNA-neg. We limited the analysis to patients on efavirenz due to its stable pharmacokinetics and bedtime dosing. In the 2 groups, the median efavirenz levels at T0 were 1484 ng/mL (interquartile range [IQR], 1054-2272 ng/mL) and 1593 ng/mL (IQR, 1047-2323 ng/mL), respectively (P = .48); levels were above the recommended threshold of 1000 ng/mL in 70 of 90 (77.8%) and 74 of 96 (77.1%) patients, respectively (P 5 1.00; Figure 2). The median levels by T0VL group were 1666 ng/mL (IQR, 1278-2449 ng/mL) for 40-49 copies/mL, 1339 ng/mL (IQR, 984-2041 ng/mL) for RNA1, and 1593 ng/mL (1047-2323 ng/mL) for RNA2 (P = .11; Kruskal-Wallis test).

Drug Resistance at Rebound

Stored plasma samples were retrieved from 68 of 80 patients with rebound >400 copies/mL for resistance testing at rebound. The median interval between the first VL measurement >400 copies/mL and the resistance test was 5.5 weeks (IQR, 2-8.5 weeks). Among 63 test results, proportions showing >/=1 major RAM were 12 of 35 (34.3%), 9 of 21 (42.9%), and 2 of 7 (28.6%) with T0VL groups 40-49 copies/mL, RNA+, and RNA-neg, respectively (P 5 .75). The breakdown by drug class was nucleoside/nucleotide reverse-transcriptase inhibitors (NRTIs), 18 of 23 (78.3%); NNRTIs, 10 of 23 (43.5%); and protease inhibitors (PIs), 5 of 23 (21.7%). Overall, 17 of 23 (73.90%) patients had >/=1 RAM affecting drugs taken at the time of rebound.

  iconpaperstack view older Articles   Back to Top