Among patients who achieve an undetectable viral load (VL) with highly active antiretroviral therapy (HAART), a subset experience intermittent episodes of virological breakthrough. These transient episodes of viraemia have been shown by several groups to be associated with increasing levels of replication-competent virus in cellular reservoirs , increased rates of evolution within viral envelope sequences [2,3], and the potential emergence of drug resistance . There are several possible mechanisms to account for these viral blips during HAART. In addition to the inherent technical variability of the VL assays , there may be incomplete viral suppression owing to periods of poor adherence or inadequate drug concentration, allowing for the emergence of drug-resistant mutants. Intermittent viraemia may also be caused by the emergence of wild-type virus as a result of the activation of pretherapy infected memory cells . However, recent data have shown that the genotype of the early rebounding plasma virus may differ from the genotype of either cell-associated HIV RNA or virus cultured from resting CD4 T cells [7,8], suggesting the possible existence of other persistent HIV reservoirs. The most appropriate management strategy of patients with intermittent viraemia remains unclear, as the clinical impact of these episodes is not known.
The main objective of this study was to examine the natural history, and in particular the prevalence, patterns and predictors of these episodes of intermittent viraemia in patients who attained an initially undetectable VL on HAART. In addition, we sought to determine the immunological and virological consequences of intermittent viraemia, and specifically whether these episodes were predictive of subsequent sustained virological rebound in patients receiving HAART.
Patients who had received a HAART regimen containing an HIV protease inhibitor or non-nucleoside reverse transcriptase inhibitor for the first time between 1 January 1996 and 31 December 1998 were identified retrospectively from three London HIV centres (King's College Hospital and Guy's and St Thomas's Hospital Trust in south London and the Chelsea and Westminster Hospital in west London). Those patients who subsequently experienced primary virological failure (i.e., VL never suppressed < 400 copies/ml) or early secondary virological failure (i.e., VL initially suppressed < 400 copies/ml but rebounding to ≥ 400 copies/ml within a year of initiating HAART) were excluded from the analysis.
Eligible patients were those who (i) attained an undetectable VL (defined as < 400 copies/ml) within 6 months of initiating HAART; (ii) had at least 12 months of follow-up following initial VL undetectability; and (iii) had at least four documented serial VL measurements following initial VL undetectability, at approximately 3 month intervals. Patients were grouped according to their subsequent virological response following initial VL undetectability: the first group were those with a sustained undetectable VL [i.e., patients in whom the VL remained undetectable (< 400 copies/ml) for at least 1 year from initiation of HAART] and the second group were those with intermittent viraemia (i.e., patients who had one or more episodes of a VL ≥ 400 copies/ml followed by a VL < 400 copies/ml).
Viral load measurements
Plasma VL assays were performed using commercially available methods according to the manufacturers’ instructions. Viral load monitoring using the AMPLICOR PCR HIV-1 MONITOR test 1.0 (Roche Diagnostics, Roche Products, Welwyn Garden City, UK) was first introduced as a clinical service in September 1996 and upgraded to the AMPLICOR test version 1.5 in October 1997. This was largely replaced by the branched chain DNA (bDNA) (HIV RNA assay version 2.0; Bayer, Newbury, UK) in October 1997 and by version 3.0 with a lower detection limit of 50 copies/ml in September 1998
Since the sensitivity and detection limits vary for the different assays (200 copies/ml for PCR version 1.0; 500 copies/ml for bDNA version 2.0; and 50 copies/ml for the Chiron version 3.0), an important first step was to standardize the different assay results to a common bDNA scale. First, any VL recorded as < 400 copies/ml was considered undetectable regardless of the assay method used. All results ≥ 400 copies/ml were then corrected to a bDNA scale, using a conversion factor of ×0.5998 for PCR VL results, based on geometric mean ratios , and ×0.5998/1.5 = 0.3999 for PCR-Ultra VL results (Roche, personal communication). Again, these corrected VL values were classified as undetectable if < 400 copies/ml. CD4 cell counts were performed using standard flow cytometry techniques.
The two groups (i.e., those with sustained VL undetectability and those with intermittent viraemia) were compared for the percentage who experienced sustained virological rebound ≥ 400 copies/ml (i.e., all subsequent VL measurements ≥ 400 copies/ml) and for the median CD4 cell count change at 12, 18 and 24 months following the initiation of HAART. The contribution of genotypic resistance to the development of intermittent viraemia was assessed from the prevalence of drug resistance mutations in the reverse transcriptase and protease genes.
The medical records of a subgroup of 96 patients with intermittent viraemia were reviewed for possible contributory factors in the month prior to the documented episode of viraemia.
Drug resistance mutations
Presence of drug resistance mutations in the reverse transcriptase and protease genes were determined in 21 of the 34 patients from the King's and St Thomas’ Hospital sites who had an initial episode of viraemia ≥ 2000 copies/ml (the limit of detection of the assay). RNA was isolated from plasma with QIAamp Mini Spin kits (QIAGEN Ltd, Crawley, West Sussex, UK), and PCR Cycle sequencing and analysis performed using the Visible genetics TRUEGENE HIV-1 Genotyping assay (TRUGENE, High Wycombe, UK). This system provides automated sequencing of all the protease and the majority of the reverse transcriptase genes of HIV-1. Mutations detected by this assay are designated as resistance mutations if they are associated with maximum reduction in drug susceptibility and as partial resistance mutations if they are associated with diminished virological responses in some, but not all, individuals and with an intermediate decrease in drug susceptibility of HIV isolates.
Baseline clinical laboratory and treatment characteristics were compared between the two groups using chi-squared and Wilcoxon Mann–Whitney U tests across the groups. The median CD4 cell count changes from baseline to that at 12, 18 and 24 months following initiation of HAART were compared using Wilcoxon Mann–Whitney U tests. Standard Kaplan–Meier estimation and log rank comparisons were used to estimate the proportion of patients with a sustained virological rebound ≥ 400 copies/ml at 18, 24 and 36 months. A Cox proportional hazards model was used to identify baseline demographic, clinical and laboratory factors among patients with intermittent viraemia that were predictive of subsequent sustained viral rebound ≥ 400 copies/ml. Data were analysed using the statistical package STATA version 6.0 (StataCorp, 1999, College Station, Texas, USA), and all the P values stated are two sided.
In total, 765 eligible patients were identified from the three London HIV clinics. The majority were male (87%) and homosexual/bisexual (75%), but about 30% of the combined cohort was non-white, mainly black African. At initiation of HAART, approximately half (48%) already had a Centers for Disease Control and Prevention (CDC) stage IV diagnosis, with a median CD4 cell count of 204 × 106 cells/l [interquartile range (IQR), 89–307] and a VL of 48 995 copies/ml (IQR, 11 966–163 005). Approximately 40% had already taken a nucleoside reverse transcriptase inhibitor drug at initiation of HAART. A protease inhibitor-based HAART regimen was initiated in 540 patients (71%) and a non-nucleoside reverse transcriptase inhibitor regimen (mainly nevirapine) in 225 (29%)
Response to therapy and prevalence of intermittent viraemia
Figure 1 summarizes the relative frequency of the four VL response patterns. Overall, 40.8% experienced primary treatment failure, with their HAART regimen failing to suppress their virus to undetectable (< 400 copies/ml) levels; 13.7% experienced early secondary treatment failure, with VL rebound ≥ 400 copies/ml) within a year of starting HAART. These two groups were excluded from all further analyses. All subsequent analyses focused on the 226 patients (29.5%) with sustained VL undetectability for at least 1 year after initiation of HAART and the 122 (15.9%) patients with intermittent viraemia after achieving an initial undetectable VL. The median follow-up after initial attainment of an undetectable VL was 27.9 months.
Pattern of intermittent viraemia
The median time from initial VL undetectability to the first episode of intermittent viraemia was 6.1 months (IQR, 3.3–10.9). The median VL at first episode was 1153 copies/ml (IQR, 636–6,053). Initial VL was ≤ 1000 copies/ml in 45%, 1001–5000 copies/ml in 26%, 5001–100 000 copies/ml in 17% and > 100 000 copies/ml in 12%. The majority of patients had a single episode of virological breakthrough, but 22 (18%) patients had two or more episodes (20 patients had two episodes and two patients had three episodes). The median VL at the second episode was 799 copies/ml (IQR, 651–4316) and this followed the initial episode after a median of 9 months (IQR, 5.8–11.8). In those patients with two or more episodes, the magnitude of the VL at the first and second episodes was 1591 (IQR, 662–5300) and 799 copies/ml (IQR, 651–4316), respectively.
Characteristics of patients with sustained undetectable viral load and those with intermittent viraemia
Table 1 compares the characteristics of patients at initiation of HAART who subsequently had a sustained undetectable VL or who experienced episodes of intermittent viraemia. In a univariate analysis, younger age (P = 0.003), female gender (P = 0.05) and more than 6 months prior nucleoside reverse transcriptase inhibitor exposure (P = 0.05) were significantly associated with the development of intermittent viraemia (Table 2). There was also an inverse relationship between baseline VL and development of intermittent viraemia. Compared with patients with a VL ≤ 20 000 copies/ml, there was a significantly reduced risk of intermittent viraemia in those with a VL 20 000–100 000 copies/ml [odds ratio (OR), 0.57; 95% confidence interval (CI), 0.33–0.99] and > 100 000 copies/ml (OR, 0.59; 95% CI, 0.35–1.01). However, in the multivariate analysis, only age and gender remained significant predictors of intermittent viraemia (P = 0.002 and 0.02, respectively).
Reasons for episodes of intermittent viraemia
There were 103 episodes of intermittent viraemia in 96 patient from the three hospital sites. In the majority of these, there was a documented event in the medical records: most commonly a period of non-adherence, an intercurrent illness or vaccination in the month prior to the episode of intermittent viraemia. The most common reasons identified were a documented period of poor adherence, drug interruption or change in 44 (42.6%), intercurrent infection or vaccination in 27 (26.2%) and drug interaction in four (6.8%). In 25 (24.3%) patients, there was no clear documented reason to account for the episode of intermittent viraemia.
Figure 2 shows the cumulative percentage with sustained virological rebound ≥ 400 copies/ml at 18, 24 and 36 months following initiation of HAART in both patients with an undetectable VL for at least 1 year after initiating HAART, and those with intermittent viraemia. The Kaplan–Meier estimates at 18, 24 and 36 months after initiation of HAART for patients with intermittent viraemia and those with an undetectable VL for > 1 year were 13.97% (95% CI, 8.9–21.5) versus 1.4% (95% CI, 0.5–4.3), 19.3% (95% CI, 8.9–21.5) versus 7.7% (95% CI, 4.5–13.0) and 31.6% (95% CI, 21.8–44.2) versus 12.9% (95% CI, 7.5–21.5), respectively (log-rank P < 0.001). Using a Cox proportional hazards models, there was a threefold higher rate of sustained virological rebound in patients with intermittent viraemia (hazards ratio, 3.15; 95% CI, 1.72–5.77, P < 0.001).
Among the 122 patients with intermittent viraemia, and a subset of 71 of these patients with no prior antiretroviral therapy experience, no factors were identified to be predictive of subsequent sustained viral rebound, using a Cox proportional hazards model (including magnitude of the VL at first episode, number of episodes of intermittent viraemia, time to first episode of intermittent viraemia, type of protease inhibitor or non-nucleoside reverse transcriptase inhibitor, prior nucleoside reverse transcriptase inhibitor experience and duration of experience, baseline CD4 cell count, VL, CDC stage IV disease, and time to initial VL undetectability).
CD4 cell count change
Patients with intermittent viraemia had a significantly lower median CD4 cell count by approximately 50 × 106 cells/l cells compared with patients with a sustained undetectable VL following initiation of HAART: at 12 months (133 versus 178 × 106 cells/l;P = 0.009), at 18 months (138 versus 224 × 106 cells/l;P = 0.0001) and at 24 months (200 versus 260 × 106 cells/l;P = 0.0029) (Fig. 3).
Genotypic resistance with intermittent viraemia
Reverse transcriptase and protease gene sequences were obtained for 16 of 21 patients with an initial episode of intermittent viraemia ≥ 2000 copies/ml. In five patients, no HIV RNA sequence was obtained. Resistance mutations in the reverse transcriptase gene were detected in 5 of 16 (31%) samples and in the protease gene of one (6%). Partial resistance mutations in the reverse transcriptase and protease genes were detected in 25% and 6% of individuals, respectively. The majority of individuals (14/16; 87%) had various combinations of accessory protease gene mutations at codon positions 10, 20, 36, 63, 71 and 77, which are considered to represent natural polymorphisms within this gene.
Intermittent viraemia, as defined by a VL ≥ 400 copies/ml, occurred in 16% of all our patients initiating HAART, and in 27% of patients who initially attained an undetectable VL. This is consistent with the reported prevalence of 20% of 241 patients initiating HAART  based on a more conservative definition of intermittent viraemia as ≥ 200 copies/ml (or 40%, when a threshold of ≥ 50 copies/ml was used). In the majority of our patients, there were documented events that may have contributed to the episodes of viraemia. A period of poor adherence or an intercurrent infection or vaccination, which are well-established causes of transient viraemia [11,12], were recorded in approximately three-quarters of all episodes of viraemia. However, levels of adherence and drug levels were not documented longitudinally in this study, and it is, therefore, difficult to quantify the level of poor adherence or low drug concentration that may have contributed to these episodes of viraemia. In addition, primary genotypic resistance mutations to reverse transcriptase and protease genes were found in almost a third of a subgroup of patients at their first episode of viraemia, although this may represent an underestimate as only those patients with viraemia ≥ 2000 copies/ml were examined. It also remains unclear as to whether this resistance was the cause or consequence of the viraemic episodes. However, several of these patients had prior experience of nucleoside reverse transcriptase inhibitor drugs, suggesting that resistance to these drugs may have been already present at baseline and contributed to subsequent viraemia. In another recent study of 15 patients with intermittent viraemia ≥ 50 copies/ml, resistance mutations in the reverse transcriptase and protease gene to one or more drugs were present at the time of relapse in 8 of 11 patients . Several of these patients had prior nucleoside reverse transcriptase inhibitor experience, and the authors concluded that half of the cases of viral breakthrough were the result of selection of resistant viruses, and the remainder were caused by the presence of wild-type virus produced from activation of memory cells infected before therapy.
In the present study, there was clear evidence for a deleterious effect of intermittent viraemia on virological and immunological responses to HAART. Intermittent viraemia ≥ 400 copies/ml was associated with a two- to threefold higher rate of sustained VL rebound and an impaired CD4 cell count rise at 12, 18 and 24 months after the initiation of HAART, relative to those patients who had maintained an undetectable VL for at least 1 year. Wood and colleagues reported a similar coupling of an optimal CD4 cell count response with sustained VL suppression . However, in four separate studies of patients with episodes of much lower-level viral breakthrough, between 50 and 400 copies/ml, no association was found with an increased rate of viral rebound [10,13,15,16].
How well can we predict which patients initiating HAART are likely to experience intermittent viraemia, and, in turn, which of these will develop sustained viral rebound? In our univariate analysis, younger age, female gender, more prolonged prior nucleoside reverse transcriptase inhibitor exposure, and a lower baseline VL were associated with an increased risk of virological breakthrough. While the impact of age and gender on VL suppression are likely to be mediated through an effect on adherence, and prior nucleoside reverse transcriptase inhibitor exposure through the presence of baseline resistance, the inverse relationship between baseline VL and the development of intermittent viraemia has not been reported previously and remains unexplained in the absence of longitudinal data on adherence and drug levels. Importantly, the development of one or even two episodes of viraemia does not imply the inevitable development of virological failure. At 2 and 3 years after the initiation of HAART, 80 and 70%, respectively, of patients who had experienced at least one episode of intermittent viraemia showed no evidence of sustained virological breakthrough. We were unable to identify any demographic, clinical or laboratory characteristics among patients with intermittent viraemia, that would predict the development of sustained VL rebound.
Overall, our findings suggest a clear clinical benefit from consistently maintaining viral suppression below a threshold of 400 copies/ml, although based on other studies not necessarily < 50 copies/ml. From these other studies, there appears to be no significant short- or medium-term adverse effects from episodes of much lower level viraemia: 50–200 copies/ml [10,13,15,16], although these observations remain in conflict with a body of work showing that achieving a VL < 50 copies/ml results in a more durable virological response compared with those whose nadir is between 50 and 400 copies/ml [17–19]. It remains possible, given the higher rates of drug resistance observed in the four studies cited [10,13,15,16] that, with longer follow-up and either repeated episodes or increasing levels of viraemia, virological failure would eventually result.
The clinical management implications of our findings include the need to optimize adherence, assess drug levels and avoid treatment interruptions, particularly where these have been identified as contributing factors to episodes of intermittent viraemia ≥ 400 copies/ml. For the same reason, structured treatment interruption studies designed to test the hypothesis that the immunological set-point can be increased and plasma viraemia be reduced for prolonged periods by stimulation of HIV-specific CD4 T cells and broadly directed cytotoxic T lymphocyte responses, should proceed with great caution. Recent data have highlighted the adverse consequences of interrupting treatment, including the emergence of drug-resistant virus, substantial declines in CD4 T cells and new or recurrent opportunistic infections . Treatment intensification (rather than a complete treatment change) is a further therapeutic approach to the management of frequent and higher-level intermittent viraemia. Recent data based on a study of only five patients suggest that the frequency of viraemic episodes can be reduced and the decay rate of the latent reservoir accelerated by increasing the potency of the antiretroviral regimen, and this provides support for treatment intensification . However, this approach needs to be guided by resistance testing and confirmed in a larger study. Future studies are also needed to address the possible contribution of latently infected cells and sanctuary sites to the episodes of intermittent viraemia.
1. Ramratnam B, Mittler JE, Zhang L. et al
. The decay of the latent reservoir of replication-competent HIV-1 is inversely correlated with the extent of residual viral replication during prolonged antiretrovral therapy. Nat Med 2000, 6: 82–85.
2. Finzi D, Hermankova M, Pierson T. et al
. Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy
. Science 1997, 278: 1291–1295.
3. Dormadula G, Zhang H, van Uitert B. et al
. Residual HIV-1 RNA in blood plasma of patients taking suppresive highly active antiretroviral therapy
. JAMA 1999, 282: 1627–1632.
4. Martinez-Picado J, DePasquale MP, Kartsonis NA. et al
. Antiretrovral resistance during successful therapy of HIV type 1 infection. Proc Natl Acad Sci USA 2000, 97: 10948–10953.
5. Brambilla D, Reichelderfer PS, Bremer JW. et al
. The contribution of assay variation and biological variation to the total variability of plasma HIV-1 RNA measurement. AIDS 1999, 13: 2269–2279.
6. Chun TW, Engel D, Berrey MM. et al
. Early establishment of a pool of latently infected, resting CD4+ T cells during primary HIV-1 infection. Proc Natl Acad Sci USA 1998, 95: 8869–8873.
7. Chun TW, Davey RT Jr, Ostrowski M. et al
. Relationship between pre-existing viral reservoirs and the re-emergence of plasma viremia after discontinuation of highly active anti-retroviral therapy. Nat Med 2000, 6: 757–761.
8. Zhang L, Chung C, Hu BS. et al
. Genetic characterisation of rebounding HIV-1 after cessation of highly active antiretroviral therapy
. J Clin Invest 2000, 106: 839–845.
9. Giles RE, Perry KR, Parry JV. et al
. Evaluation of Three Methods for Quantification of HIV 1 RNA in Plasma. Medical Devices Agency Evaluation Report MDA/98/42.
London: HSMO; 1998.
10. Havlir D, Bassett R, Levitan D. et al
. Prevalence and predictive value of intermittent viraemia
with combination antiretroviral therapy
. JAMA 2001, 286: 171–179.
11. Gunthard HF, Wong JK, Spina CA. et al
. Effect of influenza vaccination on viral replication and immune response in persons infected with human immunodeficiency virus receiving potent antiretroviral therapy
. J Infect Dis 2000, 181: 522–531.
12. Tasker SA, O'Brien WA, Treanor JJ. et al
. Effects of influenza vaccination in HIV-infected adults: a double-blind, placebo-controlled trial. Vaccine 1998, 16: 1039–1042.
13. Cohen Stuart J, Wensing AMJ, Kovacs C et al
. Mechanisms underlying transient relapses (`blips') of plasma HIV RNA in patients on HAART.Fourth International Workshop on HIV Drug Resistance and Treatment Strategies
. Spain, June 2000 [abstract 137].
14. Wood E, Yip B, Hogg RS. et al
. Full suppression of viral load is needed to achieve an optimal CD4 cell count response among patients on triple antiretroviral drug therapy. AIDS 2000, 14: 1955–1960.
15. Coakley EP, Doweiko J, Albrecht MA. Plasma HIV genotypic drug resistance profiles in subjects with stable plasma viraemia of less than 1000 copies/ml for more than 12 months.Fourth International Workshop on HIV Drug Resistance and Treatment Strategies.
Spain, June 2000 [abstract 144].
16. Greene JB, Sedorowitz M, Holzman RS et al
. Does HIV suppression below detection limits of 2nd generation PCR assays improve prognosis? An analysis of two cohorts from a large clinical practice using an observational database.XIII International Conference on AIDS.
Durban, July 2000 [abstract MoPeB2171].
17. Raboud JM, Montaner JS, Conway B. et al
. Suppression of plasma viral load below 20 copies/ml is required to achieve a long-term response to therapy. AIDS 1998, 12: 1619–1624.
18. Montaner JSG, Reiss P, Cooper D. et al
. A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV-infected patients: the INCAS trial. JAMA 1998, 279: 930–937.
19. Kempf DJ, Rode RA, Xu Y. et al
. The duration of viral suppression during protease inhibitor therapy for HIV-1 infection is predicted by plasma HIV-1 RNA at the nadir. AIDS 1998, 12: F9–F14.
20. Benson CA. Structured treatment interruption in HIV infection. AIDS Reader 2001, 11: 99–102.
21. Ramratnam B, Riveiro R, He T et al
. Antiretroviral intensification accelerates the decay of the latent reservoir of HIV-1 and increases but does not eliminate ongoing viral replication.
Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 502].