Abstract

(See the editorial commentary by Taiwo and Bosch, on pages 1189–91.)

Background. The importance of human immunodeficiency virus (HIV) blip magnitude on virologic rebound has been raised in clinical guidelines relating to viral load assays.

Methods. Antiretroviral-naive individuals initiating combination antiretroviral therapy (cART) after 1 January 2000 and achieving virologic suppression were studied. Negative binomial models were used to identify blip correlates. Recurrent event models were used to determine the association between blips and rebound by incorporating multiple periods of virologic suppression per individual.

Results. 3550 participants (82% male; median age, 40 years) were included. In a multivariable negative binomial regression model, the Amplicor assay was associated with a lower blip rate than branched DNA (rate ratio, 0.69; P < .01), controlling for age, sex, region, baseline HIV-1 RNA and CD4 count, AIDS-defining illnesses, year of cART initiation, cART type, and HIV-1 RNA testing frequency. In a multivariable recurrent event model controlling for age, sex, intravenous drug use, cART start year, cART type, assay type, and HIV-1 RNA testing frequency, blips of 500–999 copies/mL were associated with virologic rebound (hazard ratio, 2.70; P = .002), whereas blips of 50–499 were not.

Conclusions. HIV-1 RNA assay was an important determinant of blip rates and should be considered in clinical guidelines. Blips ≥500 copies/mL were associated with increased rebound risk.

Significant reductions in morbidity and mortality have been achieved in human immunodeficiency virus type 1 (HIV-1)–infected individuals since the advent of combination antiretroviral therapy (cART) [1, 2]. The goal of treatment of HIV-1–infected individuals has been suppression of viral load (measured as HIV-1 RNA) to levels below assay limits [3, 4], although this has been reassessed in recent treatment guidelines by the Department of Health and Human Services (DHHS) [3]. This is in part owing to the fact that in some cases, patients achieving virologic suppression experience intermittent increases in their plasma HIV-1 RNA, referred to as “blips” [5–8]. Though the exact nature and etiology of blips is uncertain, hypotheses have been put forward to explain them, namely, the occurrence of HIV replication bursts from either stable reservoirs [9] and/or ongoing cycles of replication [10]. Additionally, some evidence points to contributions from random biological fluctuation, random statistical variation, or methodological laboratory issues [11–14]. Finally, it has been previously noted that HIV-1 RNA assay variability is higher at lower HIV-1 RNA levels [15], which may also be a contributing factor to blips.

The published literature is conflicting with regard to whether virologic blips predict adverse clinical outcomes such as virologic failure or rebound (Table 1). Although most studies have found no relationship between isolated blips and virologic failure [5, 7, 16–21], others have reported an association between blips and an increased risk of subsequent virologic failure (Table 1) [6, 22–24]. Part of the discrepancy in findings is due to inconsistencies in definitions of blips and virologic failure and differences in testing methods [25]. As was recently reported in persistent low-level HIV viremia, links between blips and the development of resistant genotypes [22, 26–28] have been described. Links have also been described between blips and a dampened CD4+ cell count rise [18, 22]. The effect of the magnitude of a virologic blip on the occurrence of virologic rebound has been examined previously [17, 21, 23], with 2 of these studies noting that blips of higher magnitude increased the likelihood of virologic rebound [21, 23]. In addition, whereas recurrent event methodology has been used to study predictors of virologic failure [29, 30], to our knowledge, no studies have used this technique to examine the influence of blips on subsequent virologic failure. Recurrent event methodology allows the inclusion of multiple periods of suppression from each individual in the analysis, yielding increased statistical power. Last, since most of the literature on the consequences of virologic blips is from the early cART era, it is important to examine the consequences of blips among patients on modern cART therapy.

Table 1.

Review of the Literature on Virologic Blips and Risk of Subsequent Virologic Failure

Author No. Treatment Experience cART Regimen Study Period Follow-up Definition of Blip (copies/mL) Patients With Blip, % Definition of Virologic Failure (copies/mL) Association of Blip With Virologic Failure Comments 
Raboud [17165 Naive Double therapy or NNRTI- or PI-based triple therapy 1994–1997 12 mo >20 or >50 52% >500 at 52 weeks Blip at single visit did not reduce virologic success at week 52 compared with those with no viral rebound (P = .38). Rebounds at 2 consecutive visits decreased the probability of virologic success.Generalized estimating equation model demonstrated that magnitude of blip was not predictive of whether VL returned to undetectable levels. 
Easterbrook [22765 Naive to cART; possible prior monotherapy PI- (71%) or NNRTI-based (29%) cART 1996–1998 29.5 mo >400 15.9% Sustained VL >400 HR, 3.15 (95% CI, 1.72–5.77), P < .001 Dampened CD4 rise in blip group. 
Havlir [16254 Naive to PI, 3TC; possible prior use of ZDV and other medications Received 1, 2, or 3 drugs from: IDV, ZDV, d4T, 3TC 1997–1998 19 mo >50 40% 2 consecutive VL >200 HR, 1.28 (95% CI, .59–2.79), P = .53 Used maintenance simplification strategy for cART. 
Mira [5330 NA 2 NRTI + PI or NNRTI 1997–2000 33 mo 51–1000 11% 2 consecutive VL >200 HR, 0.48 (95% CI, .12–2.04), P = .3 Time to VF not significantly shorter in suppressed group (P = .12). 
Moore [23553 NA PI- or NNRTI-based cART NA 13 mo >50 35% 2 consecutive VL >400 Blip > 400: 51% experienced VF, blip ≤400: 8% VF No significance parameters included. 
Sklar [7448 Naive and treatment-experienced NA 1997–2000 16 mo >50 27.2% Low-level: 50–400, high-level >400 RR, 0.82 (95% CI, .49–1.38)  
Martinez [1843 Prior PI-based regimen NNRTI-based cART NA 18 mo >50 18.6% 2 consecutive VL >200 No VF seen Patients with blips had significantly lower CD4 count at follow-up. 
Masquelier [24219 Naive PI-based cART 1998–1999 24 mo >500 9.1% (1) 2 consecutive VL >500 or (2) persistence >500 after first year of follow-up 25% with blip in first year and 5% without blip in year 1 had VF in year 2 (P = .03)  
Sungkanuparph [19380 Naive and experienced NNRTI- and PI-based cART 1998–2003 23.5 mo 50–1000 33.7% 2 consecutive VL >1000 HR, 1.75 (95% CI, .77–3.97), P = .183 If multiple blips seen, only first episode considered in analysis. Examined time to VF (P = .271) for all subgroups. 
Podsadecki [20223 Naive LPV/r + 2 NRTIs NA 22 mo 50–1000 27% (1) 2 consecutive VL ≥50, or final measured VL ≥50; (2) >1000 at study end or last visit; (3) 2 consecutive VL >200 or at study end. No association between blips and VF for all 3 definitions (P ≥ .33) If multiple blips seen, only first episode considered in analysis. 
Garcia-Gasco [21655 NA Triple-drug cART 1999–2006 NA 51–500 28.6% >500 VF occurred in 9.1% of patients with blip Percentage of patients with VF increased with increasing magnitude of blip: OR, 1.281 (95% CI, 1.087–1.509), P = .003 
Author No. Treatment Experience cART Regimen Study Period Follow-up Definition of Blip (copies/mL) Patients With Blip, % Definition of Virologic Failure (copies/mL) Association of Blip With Virologic Failure Comments 
Raboud [17165 Naive Double therapy or NNRTI- or PI-based triple therapy 1994–1997 12 mo >20 or >50 52% >500 at 52 weeks Blip at single visit did not reduce virologic success at week 52 compared with those with no viral rebound (P = .38). Rebounds at 2 consecutive visits decreased the probability of virologic success.Generalized estimating equation model demonstrated that magnitude of blip was not predictive of whether VL returned to undetectable levels. 
Easterbrook [22765 Naive to cART; possible prior monotherapy PI- (71%) or NNRTI-based (29%) cART 1996–1998 29.5 mo >400 15.9% Sustained VL >400 HR, 3.15 (95% CI, 1.72–5.77), P < .001 Dampened CD4 rise in blip group. 
Havlir [16254 Naive to PI, 3TC; possible prior use of ZDV and other medications Received 1, 2, or 3 drugs from: IDV, ZDV, d4T, 3TC 1997–1998 19 mo >50 40% 2 consecutive VL >200 HR, 1.28 (95% CI, .59–2.79), P = .53 Used maintenance simplification strategy for cART. 
Mira [5330 NA 2 NRTI + PI or NNRTI 1997–2000 33 mo 51–1000 11% 2 consecutive VL >200 HR, 0.48 (95% CI, .12–2.04), P = .3 Time to VF not significantly shorter in suppressed group (P = .12). 
Moore [23553 NA PI- or NNRTI-based cART NA 13 mo >50 35% 2 consecutive VL >400 Blip > 400: 51% experienced VF, blip ≤400: 8% VF No significance parameters included. 
Sklar [7448 Naive and treatment-experienced NA 1997–2000 16 mo >50 27.2% Low-level: 50–400, high-level >400 RR, 0.82 (95% CI, .49–1.38)  
Martinez [1843 Prior PI-based regimen NNRTI-based cART NA 18 mo >50 18.6% 2 consecutive VL >200 No VF seen Patients with blips had significantly lower CD4 count at follow-up. 
Masquelier [24219 Naive PI-based cART 1998–1999 24 mo >500 9.1% (1) 2 consecutive VL >500 or (2) persistence >500 after first year of follow-up 25% with blip in first year and 5% without blip in year 1 had VF in year 2 (P = .03)  
Sungkanuparph [19380 Naive and experienced NNRTI- and PI-based cART 1998–2003 23.5 mo 50–1000 33.7% 2 consecutive VL >1000 HR, 1.75 (95% CI, .77–3.97), P = .183 If multiple blips seen, only first episode considered in analysis. Examined time to VF (P = .271) for all subgroups. 
Podsadecki [20223 Naive LPV/r + 2 NRTIs NA 22 mo 50–1000 27% (1) 2 consecutive VL ≥50, or final measured VL ≥50; (2) >1000 at study end or last visit; (3) 2 consecutive VL >200 or at study end. No association between blips and VF for all 3 definitions (P ≥ .33) If multiple blips seen, only first episode considered in analysis. 
Garcia-Gasco [21655 NA Triple-drug cART 1999–2006 NA 51–500 28.6% >500 VF occurred in 9.1% of patients with blip Percentage of patients with VF increased with increasing magnitude of blip: OR, 1.281 (95% CI, 1.087–1.509), P = .003 

Abbreviations: 3TC, lamivudine; d4T, stavudine; cART, combination antiretroviral therapy; CI, confidence interval; HR, hazard ratio; IDV, indinavir; LPV/r, lopinavir/ritonavir; NA, not available; NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; OR, odds ratio; PI, protease inhibitor; RR, relative risk; VF, virologic failure; VL, viral load; ZDV, zidovudine.

The purposes of this study were to determine correlates of rates of virologic blips with particular focus on associations with the HIV-1 RNA assay used and to examine the association of an HIV viral blip with virologic rebound. Viral blips were categorized by magnitude: 50–199 copies/mL, 200–499 copies/mL, and 500–999 copies/mL.

METHODS

Study Design and Cohort Description

This was a retrospective, observational cohort study using information collected through the Canadian Observational Cohort (CANOC) collaboration [31]. CANOC is a collaboration of 8 HIV cohort studies from Ontario, Quebec, and British Columbia. Eligibility criteria for inclusion into CANOC include documented HIV infection, residence in Canada, minimum age of 18 years, initiation of a first antiretroviral regimen comprised of a minimum of 3 individual agents after 1 January 2000, and at least 1 measurement of HIV-1 RNA and CD4+ cell count within 6 months prior to cART initiation. Patient selection and data extraction are performed at the data centers of the participating cohort studies. Nonnominal demographic, laboratory, and clinical data from each cohort are then pooled at the Project Data Centre in Vancouver, British Columbia. All participating cohorts have received approval from their institutional ethics boards to contribute nonnominal, patient-specific data. Ownership of individual cohort data remains with the contributing cohort, and cohort data can only be used for studies approved by the CANOC Steering Committee.

Statistical Methods

For participants to be eligible for inclusion in the present analysis, they must have achieved virologic suppression as defined by an HIV-1 RNA value of <50 copies/mL on 2 consecutive occasions at least 30 days apart.

An HIV-1 viral blip was defined as an HIV-1 RNA value of 50–999 copies/mL preceded and followed by another value below assay limits (<50 copies/mL). The limit of 999 copies/mL was introduced to reduce the effect of large periods of nonadherence. Two or more HIV-1 RNA values meeting the definition of a blip occurring within the same 30-day period were considered to be part of the same blip episode [8, 32, 33].

Rates of blips per year were calculated for each person by dividing the number of blips during the study period by the number of years of follow-up for that person. Negative binomial regression models were used to measure the effects of HIV-1 RNA assay type on rates of blips per person-year of follow-up after adjusting for other covariates.

Virologic rebound was defined as either an HIV-1 RNA value ≥50 copies/mL at 2 consecutive visits at least 30 days apart or an HIV-1 RNA value ≥1000 copies/mL. Within each period of virologic suppression, time to rebound was calculated as the time from the date of virologic suppression to the date of virologic rebound. When there was no viral rebound after achieving virologic suppression, the participant was censored at the last HIV-1 RNA measurement. Multiple periods of virologic suppression were included in the analysis for each participant. Gap-time recurrent event models [34], which allowed multiple periods of virologic suppression per person to be included in the analysis, were used to estimate the effect of a blip on the hazard of an HIV-1 RNA rebound after adjusting for other covariates. Time-dependent covariates were used to model the effects of virologic blips on the hazard of virologic rebound. The time-dependent blip was defined and categorized to the corresponding levels based on the HIV-1 RNA in the test. If a blip occurred at consecutive tests (within 30 days), the higher HIV-1 RNA level was used. The association of the magnitude of a blip with subsequent virologic rebound was investigated by including a categorical variable for blip magnitude in the recurrent event model of time to virologic rebound. In the model, the covariate “number of HIV-1 RNA tests per year” was defined as the yearly rate of HIV-1 RNA tests for a patient from his or her first virologic suppression to the last follow-up.

Univariate negative binomial regression models were used to assess the relative risks (with 95% confidence intervals) for predictors of the primary outcome—number of blips per year—adjusted by HIV-1 RNA testing frequency. The variables that were a priori believed to be related to the number of blips, and those variables that were significant at P < .20 in the univariate analyses, were candidates for inclusion in the final multivariable negative binomial regression models. The same algorithm was applied to the analysis of recurrent events.

RESULTS

Study Population

A total of 4560 individuals were enrolled in CANOC as of 5 January 2009, and 3550 participants achieved virologic suppression and were included in the analysis. The median duration of follow-up was 2.7 years (interquartile range [IQR], 1.2–4.4). The median number of HIV-1 RNA tests per individual was 11 (IQR, 6–18). Fifty-six percent of participants had HIV-1 RNA measured with a polymerase chain reaction ultrasensitive Amplicor HIV-1 Monitor version 1.5 assay (Roche Diagnostics) (henceforth referred to as the ultrasensitive Amplicor assay), and 44% had their HIV-1 RNA measured with a branched DNA (bDNA) test method (Chiron Corp). Other demographic and clinical characteristics of these participants at the time of initiation of their first cART regimen are described in Table 2.

Table 2.

Demographic and Clinical Characteristics of Participants at Time of Initiation of Combination Antiretroviral Therapy

Characteristics Sample (N = 3550) 
Age, y 40 (34–47) 
Male 2904 (82%) 
Race  
    Caucasian 918 (54%) 
    Black 187 (11%) 
    First Nations 90 (5%) 
    Asian 52 (3%) 
    Mixed/Other 449 (24%) 
Risk factor  
    Men who have sex with men 1081 (35%) 
    Injection drug use 505 (16%) 
Region  
    Ontario 1184 (33%) 
    Quebec 881 (25%) 
    British Columbia 1485 (42%) 
Year cART started  
    Prior to 2001 430 (12%) 
    2001–2004 1870 (53%) 
    After 2004 1250 (35%) 
Type of initial cART  
    NNRTI-based 1549 (44%) 
    Boosted PI–based 1432 (40%) 
    Single PI-based 345 (10%) 
    Other 224 (6%) 
Hepatitis C coinfection 607 (21%) 
Hepatitis B coinfection 581 (21%) 
AIDS-defining illness 473 (13%) 
CD4 count  
    CD4 count (cells/μL) 190 (100–280) 
    CD4 ≥200 (cells/μL) 1735 (49%) 
Viral load  
    Log10 copies/mL 4.9 (4.3–5.1) 
No. of HIV RNA tests per year  
    ≤3 464 (13%) 
    3–≤4 830 (23%) 
    4–≤6 1354 (38%) 
    >6 902 (25%) 
Characteristics Sample (N = 3550) 
Age, y 40 (34–47) 
Male 2904 (82%) 
Race  
    Caucasian 918 (54%) 
    Black 187 (11%) 
    First Nations 90 (5%) 
    Asian 52 (3%) 
    Mixed/Other 449 (24%) 
Risk factor  
    Men who have sex with men 1081 (35%) 
    Injection drug use 505 (16%) 
Region  
    Ontario 1184 (33%) 
    Quebec 881 (25%) 
    British Columbia 1485 (42%) 
Year cART started  
    Prior to 2001 430 (12%) 
    2001–2004 1870 (53%) 
    After 2004 1250 (35%) 
Type of initial cART  
    NNRTI-based 1549 (44%) 
    Boosted PI–based 1432 (40%) 
    Single PI-based 345 (10%) 
    Other 224 (6%) 
Hepatitis C coinfection 607 (21%) 
Hepatitis B coinfection 581 (21%) 
AIDS-defining illness 473 (13%) 
CD4 count  
    CD4 count (cells/μL) 190 (100–280) 
    CD4 ≥200 (cells/μL) 1735 (49%) 
Viral load  
    Log10 copies/mL 4.9 (4.3–5.1) 
No. of HIV RNA tests per year  
    ≤3 464 (13%) 
    3–≤4 830 (23%) 
    4–≤6 1354 (38%) 
    >6 902 (25%) 

Data are no. (%) or median (interquartile range).

Abbreviations: cART, combination antiretroviral therapy; HIV, human immunodeficiency virus; NNRTI, nonnucleoside reverse transcriptase inhibitor, PI, protease inhibitor.

Correlates of Viral Blips

Seven hundred fifty-six of the 3550 participants (21%) had ≥1 blips during the study period. The overall rate of blips was 0.11 per year of follow-up. Eight hundred seventy-four (85%) of the virologic blips were between 50 and 199 copies/mL, 71 (7%) were between 200 and 349 copies/mL, 38 (4%) were between 350 and 499 copies/mL, and 40 (4%) were between 500 and 999 copies/mL. The largest blip for 717 participants was between 50 and 499 copies/mL, whereas 39 participants experienced a blip between 500 and 999 copies/mL. Results of univariate negative binomial models are shown in Table 3. Patients whose HIV-1 RNA was measured with the ultrasensitive Amplicor assay were approximately half as likely to experience blips. Lower rates of blips were also associated with a higher CD4 count at cART initiation and lower rates of HIV-1 RNA testing. Higher rates of blips were associated with Ontario residence (which coincides with use of the bDNA assay), boosted protease inhibitor–based regimens, an HIV risk factor of men who have sex with men, higher HIV-1 RNA at cART initiation, and an AIDS-defining illness at cART initiation.

Table 3.

Negative Binomial Model of Number of Blips per Year

 Unadjusted Analysisa
 
Adjusted Analysis
 
Covariateb RR (95% CI) P Value RR (95% CI) P Value 
Age (per 10 years) 1.05 (.97–1.12) .21 1.04 (.97–1.12) .27 
Male 0.83 (.68–1.02) .07 0.91 (.74–1.11) .33 
Caucasian 1.11 (.89–1.38) .34   
Risk factor     
    MSM 1.30 (1.11–1.53) <.01   
    IDU 1.04 (.83–1.29) .75   
Region     
Ontario 1.88 (1.63–2.17) .0001 1.48 (1.17–1.87) <.01 
British Columbia/Quebec   
Type of cART     
    Boosted PI-based 1.43 (1.22–1.67) <.0001 1.35 (1.15–1.58) <.001 
    Single PI-based 1.22 (.96–1.55) .11 1.15 (.90–1.46) .27 
    Other 1.37 (1.05–1.78) .02 1.36 (1.05–1.75) .02 
    NNRTI-based   
Year of cART initiation     
    Prior to 2001 0.90 (.71–1.14) .37 1.08 (.85–1.38) .53 
    2001–2004 1.06 (.88–1.27) .53 1.15 (.96–1.38) .14 
    After 2004   
First VL suppression after year 2005 1.09 (.90–1.33) .37   
Hepatitis C coinfection 0.87 (.71–1.05) .15   
Hepatitis B coinfection 0.86 (.71–1.06) .16   
AIDS-defining illness 1.21 (1.00–1.45) .05 1.11 (.92–1.34) .27 
CD4 count at cART initiation     
    ≥200 cells/μL 0.88 (.77–1.01) .08   
CD4 (per 100 cells/μL) 0.94 (.90–.98) <.01 0.97 (.92–1.01) .16 
VL at cART initiation     
    ≥105 copies/mL 1.46 (1.27–1.68) <.0001 1.41 (1.22–1.62) <.0001 
    Log10 copies/mL 1.23 (1.13–1.34) <.0001   
No. of HIV RNA tests per year     
    ≤3 0.36 (.28–.47) <.0001 0.31 (.24–.40) <.0001 
    3–≤4 0.52 (.42–.63) <.0001 0.44 (.35–.54) <.0001 
    4–≤6 0.59 (.49–.70) <.0001 0.53 (.44–.64) <.0001 
    >6   
Assay     
    Ultrasensitive Amplicor assay 0.56 (.49–.65) <.0001 0.69 (.55–.88) <.01 
    bDNA   
 Unadjusted Analysisa
 
Adjusted Analysis
 
Covariateb RR (95% CI) P Value RR (95% CI) P Value 
Age (per 10 years) 1.05 (.97–1.12) .21 1.04 (.97–1.12) .27 
Male 0.83 (.68–1.02) .07 0.91 (.74–1.11) .33 
Caucasian 1.11 (.89–1.38) .34   
Risk factor     
    MSM 1.30 (1.11–1.53) <.01   
    IDU 1.04 (.83–1.29) .75   
Region     
Ontario 1.88 (1.63–2.17) .0001 1.48 (1.17–1.87) <.01 
British Columbia/Quebec   
Type of cART     
    Boosted PI-based 1.43 (1.22–1.67) <.0001 1.35 (1.15–1.58) <.001 
    Single PI-based 1.22 (.96–1.55) .11 1.15 (.90–1.46) .27 
    Other 1.37 (1.05–1.78) .02 1.36 (1.05–1.75) .02 
    NNRTI-based   
Year of cART initiation     
    Prior to 2001 0.90 (.71–1.14) .37 1.08 (.85–1.38) .53 
    2001–2004 1.06 (.88–1.27) .53 1.15 (.96–1.38) .14 
    After 2004   
First VL suppression after year 2005 1.09 (.90–1.33) .37   
Hepatitis C coinfection 0.87 (.71–1.05) .15   
Hepatitis B coinfection 0.86 (.71–1.06) .16   
AIDS-defining illness 1.21 (1.00–1.45) .05 1.11 (.92–1.34) .27 
CD4 count at cART initiation     
    ≥200 cells/μL 0.88 (.77–1.01) .08   
CD4 (per 100 cells/μL) 0.94 (.90–.98) <.01 0.97 (.92–1.01) .16 
VL at cART initiation     
    ≥105 copies/mL 1.46 (1.27–1.68) <.0001 1.41 (1.22–1.62) <.0001 
    Log10 copies/mL 1.23 (1.13–1.34) <.0001   
No. of HIV RNA tests per year     
    ≤3 0.36 (.28–.47) <.0001 0.31 (.24–.40) <.0001 
    3–≤4 0.52 (.42–.63) <.0001 0.44 (.35–.54) <.0001 
    4–≤6 0.59 (.49–.70) <.0001 0.53 (.44–.64) <.0001 
    >6   
Assay     
    Ultrasensitive Amplicor assay 0.56 (.49–.65) <.0001 0.69 (.55–.88) <.01 
    bDNA   

Abbreviations: bDNA, branched DNA; cART, combination antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; MSM, men who have sex with men; IDU, injection drug user; NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor; RR, rate ratio.

a

All models were adjusted by number of HIV RNA tests per year.

b

All covariate values were at the time of cART initiation.

In the multivariable negative binomial model (Table 3), the ultrasensitive Amplicor assay remained associated with lower rates of blips, after adjusting for region, age, sex, type of cART regimen, year of cART initiation, rate of HIV-1 RNA measurement and CD4 count, HIV-1 RNA, and AIDS-defining illness at cART initiation.

Correlates of Virologic Rebound

A total of 919 (26%) participants had a virologic rebound during the study period: 771 had 1 rebound, 128 had 2 rebounds, 17 had 3 rebounds, and 3 had 4 rebounds. Of the total 1090 periods of virologic rebound, 923 included an HIV-1 RNA level of ≥1000 copies/mL, and 167 periods included at least 2 consecutive HIV-1 RNA measurements of ≥50 copies/mL but no HIV-1 RNA measurements of ≥1000 copies/mL. Only 14% of the virologic rebounds were preceded by a virologic blip.

In the univariate recurrent event model of time to virologic rebound (Table 4), a virologic blip ≥500 copies/mL was associated with an increased hazard ratio of rebound, whereas blips of 50–199 and 200–499 copies/mL were not. Covariates that were associated with increased rates of virologic rebound included measurement with the ultrasensitive Amplicor assay, an HIV risk factor of injection drug use, and earlier year of cART initiation. Male sex, older age, and less frequent HIV-1 RNA testing were associated with decreased rates of virologic rebound.

Table 4.

Stratified Gap-Time Proportional Hazards Model of Time to Virologic Failure

 Unadjusted Analysisa
 
Adjusted Analysis
 
Covariates HR (95% CI) P Value HR (95% CI) P Value 
Age (per 10 years)b 0.75 (.70–.81) <.0001 .76 (.70–.82) <.0001 
Male 0.51 (.44–.58) <.0001 .60 (.51–.70) <.0001 
Injection drug use as HIV risk factor 1.91 (1.65–2.20) <.0001 1.76 (1.51–2.04) <.0001 
Blipb     
    Blip (VL 500–999 copies/mL) 2.54 (1.40–4.62) .002 2.70 (1.44–5.06) .002 
    Blip (VL 200–499 copies/mL) 1.13 (.70–1.84) .61 1.00 (.53–1.88) >.99 
    Blip (VL 50–199 copies/mL) 0.93 (.76–1.14) .47 1.08 (.85–1.36) .53 
    No blip   
Assay     
    Roche Ultrasensitive Amplicor 1.42 (1.25–1.63) <.0001 1.39 (1.16–1.66) .0003 
    Chiron 3 bDNA   
Year of cART initiation     
    Prior to 2001 2.25 (1.80–2.82) <.0001 2.41 (1.84–3.16) <.0001 
    2001–2004 1.69 (1.39–2.05) <.0001 1.86 (1.46–2.38) <.0001 
    After 2004   
cART typeb     
    PI-based 1.10 (.97–1.25) .14 1.11 (.96–1.28) .15 
    Other 1.04 (.79–1.38) .76 0.80 (.59–1.08) .14 
    NNRTI-based   
No. of HIV RNA tests per year     
    ≤3 0.62 (.51–.76) <.0001 0.57 (.45–.72) <.0001 
    3–≤4 0.54 (.46–.65) <.0001 0.56 (.46–.69) <.0001 
    4–≤6 0.64 (.55–.76) <.0001 0.66 (.55–.79) <.0001 
    >6   
 Unadjusted Analysisa
 
Adjusted Analysis
 
Covariates HR (95% CI) P Value HR (95% CI) P Value 
Age (per 10 years)b 0.75 (.70–.81) <.0001 .76 (.70–.82) <.0001 
Male 0.51 (.44–.58) <.0001 .60 (.51–.70) <.0001 
Injection drug use as HIV risk factor 1.91 (1.65–2.20) <.0001 1.76 (1.51–2.04) <.0001 
Blipb     
    Blip (VL 500–999 copies/mL) 2.54 (1.40–4.62) .002 2.70 (1.44–5.06) .002 
    Blip (VL 200–499 copies/mL) 1.13 (.70–1.84) .61 1.00 (.53–1.88) >.99 
    Blip (VL 50–199 copies/mL) 0.93 (.76–1.14) .47 1.08 (.85–1.36) .53 
    No blip   
Assay     
    Roche Ultrasensitive Amplicor 1.42 (1.25–1.63) <.0001 1.39 (1.16–1.66) .0003 
    Chiron 3 bDNA   
Year of cART initiation     
    Prior to 2001 2.25 (1.80–2.82) <.0001 2.41 (1.84–3.16) <.0001 
    2001–2004 1.69 (1.39–2.05) <.0001 1.86 (1.46–2.38) <.0001 
    After 2004   
cART typeb     
    PI-based 1.10 (.97–1.25) .14 1.11 (.96–1.28) .15 
    Other 1.04 (.79–1.38) .76 0.80 (.59–1.08) .14 
    NNRTI-based   
No. of HIV RNA tests per year     
    ≤3 0.62 (.51–.76) <.0001 0.57 (.45–.72) <.0001 
    3–≤4 0.54 (.46–.65) <.0001 0.56 (.46–.69) <.0001 
    4–≤6 0.64 (.55–.76) <.0001 0.66 (.55–.79) <.0001 
    >6   

Abbreviations: bDNA, branched DNA assay; cART, combination antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; VL, viral load.

a

All models were adjusted by number of HIV RNA tests per year.

b

Time-dependent covariates.

In the multivariable recurrent event model of time to virologic rebound (Table 4), a virologic blip ≥500 copies/mL remained associated with an increased hazard ratio of rebound after adjusting for assay type, age, sex, an HIV risk factor of injection drug use, year of cART initiation, type of cART regimen, and rate of HIV-1 RNA testing.

DISCUSSION

In this study of HIV-infected individuals who had achieved virologic suppression on their first cART regimen, virologic blips were relatively common but were less likely when HIV-1 RNA was measured with the ultrasensitive Amplicor assay than with the bDNA assay. Virologic blips of 500–999 copies/mL were associated with subsequent virologic failure, which was defined as either confirmed virologic rebound ≥50 copies/mL or a single HIV-1 RNA value ≥1000 copies/mL. Virologic blips <500 copies/mL were not associated with an increased risk of virologic failure.

Our finding that lower rates of blips were recorded with the ultrasensitive Amplicor assay than with the bDNA assay suggests that the specific assay used needs to be considered when interpreting blips. As newer real-time polymerase chain reaction HIV-1 RNA assays become available, clinicians, patients, and researchers will have to be mindful that blips with a new assay may not have the same interpretation as those with the previous assay. The relationship of HIV-1 RNA assay to the frequency of virologic blips has been examined in prior studies with the finding that the adoption of a newer assay with a lower limit of detection typically leads to increased blip rates [13, 14]. Reflecting this, the most recent DHHS guidelines have recommended a new cutoff for virologic failure of ≥200 copies/mL HIV-1 RNA [3]. Though our study found a significant association between blips of 500–999 copies/mL HIV-1 RNA using the more conservative definition of virologic failure of ≥50 copies/mL HIV-1 RNA, we also ran the same analyses (not shown) using a definition of virologic failure of ≥200 copies/mL and found the same association.

Additional results presented in this study provide important information on other correlates of blips. The strong influence of baseline parameters associated with reduced HIV disease progression (eg, high CD4 count, low HIV-1 RNA) on rates of blips suggests that blips represent biological phenomena rather than random fluctuation. It is possible that blips reflect the degree that a given individual has an HIV-1 RNA consistently close to the 50-copy limit of detection of the assay used.

In the current literature examining the consequences of virologic blips, there is little uniformity with regard to definitions of blips and virologic failure, study populations, duration of follow-up, cART regimens, and the HIV-1 RNA assay used. The lack of consistency in definitions makes comparisons of results problematic. Our results showing that blips <500 copies/mL were not associated with subsequent virologic rebound agree with a number of studies in the literature that have not found a relationship between blips and subsequent virologic failure [5, 7, 16–20]. It should be noted, however, that most previous studies used a higher HIV-1 RNA threshold to define virologic failure and only 1 other study considered 2 consecutive HIV-1 RNA values ≥50 copies/mL to be virologic failure [20]. Our findings that blips of higher magnitude are associated with later virologic rebound agree with 3 studies that used a blip definition with a lower HIV-1 RNA limit of ≥400 copies/mL and found an association between the occurrence of blips and virologic failure [22–24]. Our results that the risk of virologic failure was only increased for blips ≥500 copies/mL, with no gradation of risk for blips <500 copies/mL were in contrast to those of Garcia-Gasco et al [21], who demonstrated an increasing risk of virologic failure in magnitudes of blips <500 copies/mL and in contrast to those of Raboud et al [17], who did not document an increase in rate of virologic failure with increasing blip magnitude. Our results are in agreement with those of Moore et al [23], who documented higher rates of virologic failure in individuals with blips > 400 copies/mL than in those with blips ≤ 400 copies/ml.

Boosted protease inhibitor use was associated with increased rates of virologic blips, likely because regimens in this class tend to be prescribed to individuals who have more advanced disease or who have trouble with adherence.

It is possible that the association of male sex with lower risk of virologic failure is due in part to women taking antiretroviral therapy (ART) for the prevention of mother-to-child transmission of HIV and who experience virologic rebound when they stop ART upon childbirth. Although explicit pregnancy data were not available for the CANOC cohort, we conducted a sensitivity analysis, assuming that women whose ART regimen contained zidovudine and lasted <1 year and who had a CD4 count ≥350 cells/μL at therapy initiation had been taking ART solely for the prevention of mother-to-child transmission of HIV. Twenty-nine women met this criterion. When these women were excluded from the analysis, the results were unchanged.

Strengths of our study include its large sample size and clinical follow-up spanning 10 years, which allowed us to investigate temporal trends in rates of blips and virologic rebound among individuals on modern cART. The use of recurrent event methodology, including multiple periods of virologic suppression for a single individual, increased the statistical power of the study. Examination of the effect of the magnitude of a virologic blip on the likelihood of subsequent virologic failure was also a strength of the study.

Our study has several limitations. Data were not available on adherence to cART regimens. Although prior studies have yielded conflicting results as to whether or not incomplete adherence leads to blips [10, 20, 32, 35], being able to control for adherence would have strengthened our study because blips due to a lapse in adherence may be less concerning than blips experienced while fully adherent. Geographic region was confounded with assay because all HIV-1 RNA tests were measured with the ultrasensitive Amplicor assay in British Columbia and with bDNA in Ontario and Quebec. Newer HIV-1 RNA assays such as the Roche TaqMan assay were not included in our analysis. Although such analyses would have been of interest to countries using this assay, many jurisdictions, including Ontario and Quebec, continue to use bDNA test methods. Finally, it should be emphasized that our study focused only on single blip episodes (but multiple periods of viral suppression). Thus, it is likely that the occurrence of repeated blip episodes within a single individual may portend an even higher hazard ratio of virologic failure.

It is likely that the clinical significance of virologic blips on virologic and clinical outcomes will continue to be a topic of interest to HIV-infected individuals and their treating physicians. The importance of virologic blips will need to be reviewed periodically as both the efficacy and ease-of-use of available antiretroviral regimens improve, as different methodologies for measuring HIV-1 RNA become available, and as HIV-1 RNA measurement becomes more common in resource-limited settings.

Notes

Acknowledgments.

We would like to thank all the participants for allowing their information to be a part of the CANOC collaboration.

The CANOC Collaboration includes: Community Advisory Committee: Sean Hosein (Chair), Bruno Lemay, Shari Margolese, Evelyne Ssengendo; Investigators: Gloria Aykroyd (Ontario HIV Treatment Network [OHTN]), Louise Balfour (University of Ottawa, OHTN Cohort Study (OCS), OCS Coinvestigator), Ahmed Bayoumi (University of Toronto, OCS Coinvestigator), John Cairney (University of Toronto, OCS Coinvestigator), Liviana Calzavara (University of Toronto, OCS Coinvestigator), Curtis Cooper (University of Ottawa, OCS Coinvestigator), Kevin Gough (University of Toronto, OCS Coinvestigator), Silvia Guillemi (British Columbia Centre for Excellence in HIV/AIDS, University of British Columbia), P. Richard Harrigan (British Columbia Centre for Excellence in HIV/AIDS, University of British Columbia), Marianne Harris (British Columbia Centre for Excellence in HIV/AIDS), George Hatzakis (McGill University), Robert Hogg (British Columbia Centre for Excellence in HIV/AIDS, Simon Fraser University), Don Kilby (University of Ottawa, Ontario HIV Treatment Network), Marina Klein (Montreal Chest Institute Immunodeficiency Service Cohort, McGill University), Richard Lalonde (Montreal Chest Institute Immunodeficiency Service Cohort and McGill University), Viviane Lima (British Columbia Centre for Excellence in HIV/AIDS, University of British Columbia), Mona Loutfy (University of Toronto, Maple Leaf Medical Clinic, OCS Coinvestigator), Nima Machouf (Clinique Medicale l’Actuel, Université de Montréal), Ed Mills (British Columbia Centre for Excellence in HIV/AIDS, University of Ottawa), Peggy Millson (University of Toronto, OCS Coinvestigator), Julio Montaner (British Columbia Centre for Excellence in HIV/AIDS, University of British Columbia), David Moore (British Columbia Centre for Excellence in HIV/AIDS, University of British Columbia), Alexis Palmer (British Columbia Centre for Excellence in HIV/AIDS), Janet Raboud (University of Toronto, University Health Network, OCS Coinvestigator), Anita Rachlis (University of Toronto, OCS Coinvestigator), Stanley Read (University of Toronto, OCS Coinvestigator), Sean Rourke (Ontario HIV Treatment Network, University of Toronto), Marek Smieja (McMaster University, OCS Coinvestigator), Irving Salit (University of Toronto, OCS Coinvestigator), Darien Taylor (Canadian AIDS Treatment Information Exchange, OCS Coinvestigator), Benoit Trottier (Clinique Medicale l’Actuel, Université de Montréal), Chris Tsoukas (McGill University), Sharon Walmsley (University of Toronto, OCS Coinvestigator), and Wendy Wobeser (Queens University, OCS Coinvestigator).

Financial support.

This work was supported by the Canadian Institutes of Health Research (CIHR), Canadian HIV Trials Network, as well as an Emerging Team Grant from the CIHR. J. M. R., C. C., and S. L. W. have Career Scientist Awards from the Ontario HIV Treatment Network. M. R. L. receives salary support from the CIHR. P. R. H. is a GSK/CIHR Chair in Clinical Virology at the University of British Columbia. M. B. K. is supported by a Chercheur-Boursier clinician senior career award from the Fonds de recherche en santé du Québec. J. S. G. M. is supported by an Avant-Garde Award from the National Institute of Drug Abuse, National Institutes of Health (1DP1DA026182).

Potential conflicts of interest.

All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

1.
Palella
FJ
Delaney
KM
Moorman
AC
, et al.  . 
Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection
New Engl J Med
 , 
1998
, vol. 
338
 (pg. 
853
-
60
)
2.
Hammer
SM
Saag
MS
Schechter
M
, et al.  . 
Treatment for adult HIV infection: 2006 recommendations of the International AIDS Society—USA Panel
J Am Med Assoc
 , 
2006
, vol. 
296
 (pg. 
827
-
43
)
3.
Panel on Antiretroviral Guidelines for Adults and Adolescents
Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents
 , 
2011
Department of Health and Human Services
(pg. 
1
-
166
4.
Thompson
MA
Aberg
JA
Cahn
P
, et al.  . 
Antiretroviral treatment of adult HIV infection: 2010 recommendations of the International AIDS Society—USA panel
J Am Med Assoc
 , 
2010
, vol. 
304
 (pg. 
321
-
33
)
5.
Mira
JA
Macias
J
Nogales
C
, et al.  . 
Transient rebounds of low-level viraemia among HIV-infected patients under HAART are not associated with virological or immunological failure
Antivir Ther
 , 
2002
, vol. 
7
 (pg. 
251
-
6
)
6.
Greub
G
Cozzi-Lepri
A
Lederberger
B
, et al.  . 
Intermittent and sustained low-level HIV viral rebound in patients receiving potent antiretroviral therapy
AIDS
 , 
2002
, vol. 
16
 (pg. 
1967
-
9
)
7.
Sklar
PA
Ward
DJ
Baker
RK
, et al.  . 
Prevalence and clinical correlates of HIV viremia (“blips”) in patients with previous suppression below the limits of quantification
AIDS
 , 
2002
, vol. 
16
 (pg. 
2035
-
41
)
8.
DiMascio
M
Markowitz
M
Louie
M
, et al.  . 
Viral blip dynamics during highly active antiretroviral therapy
J Virol
 , 
2003
, vol. 
77
 (pg. 
12165
-
72
)
9.
Ho
DD
Toward HIV eradication or remission: the tasks ahead
Science
 , 
1998
, vol. 
280
 (pg. 
1866
-
7
)
10.
Nettles
RE
Kieffer
TL
Kwon
P
, et al.  . 
Intermittent HIV-1 viremia (blips) and drug resistance in patients receiving HAART
JAMA
 , 
2005
, vol. 
293
 (pg. 
817
-
28
)
11.
Nettles
RE
Kieffer
TL
Update on HIV-1 viral load blips
Curr Opin HIV AIDS
 , 
2006
, vol. 
1
 (pg. 
157
-
61
)
12.
Stosor
V
Palella
FJ
Berzins
B
, et al.  . 
Transient viremia in HIV-infected patients and use of plasma preparation tubes
Clin Infec Dis
 , 
2005
, vol. 
41
 (pg. 
1671
-
4
)
13.
Smit
E
Bhattacharya
S
Osman
H
Taylor
S
Increased frequency of HIV-1 viral load blip rate observed after switching from Roche Amplicor to Cobas Taqman Assay
J Acquir Immune Defic Syndr
 , 
2009
, vol. 
51
 (pg. 
364
-
5
)
14.
Lima
V
Harrigan
R
Montaner
JSG
Increased reporting of detectable plasma HIV-1 RNA levels at the critical threshold of 50 copies per milliliter with the Taqman assay in comparison to the Amplicor assay
J Acquir Immune Defic Syndr
 , 
2009
, vol. 
51
 (pg. 
3
-
6
)
15.
Brambilla
D
Granger
S
Bremer
J
Variations in HIV RNA assays at low RNA concentration [abstract 774]
Program and abstracts of the 7th Conference on Retroviruses and Opportunistic Infections
 , 
2000
(San Francisco, CA). Alexandria, VA: CROI
16.
Havlir
DV
Bassett
R
Levitan
D
, et al.  . 
Prevalence and predictive value of intermittent viremia with combination HIV therapy
JAMA
 , 
2001
, vol. 
286
 (pg. 
171
-
9
)
17.
Raboud
JM
Rae
S
Woods
R
Harris
M
Montaner
JSG
Consecutive rebounds in plasma viral load are associated with virological failure at 52 weeks among HIV-infected patients
AIDS
 , 
2002
, vol. 
16
 (pg. 
1627
-
32
)
18.
Martinez
V
Marcelin
AG
Morini
JP
, et al.  . 
HIV-1 intermittent viraemia in patients treated by non-nucleoside reverse transcriptase inhibitor-based regimen
AIDS
 , 
2005
, vol. 
19
 (pg. 
1065
-
9
)
19.
Sungkanuparph
S
Overton
ET
Seyfried
W
Groger
RK
Fraser
VJ
Powderly
WG
Intermittent episodes of detectable HIV viremia in patients receiving nonnucleoside reverse-transcriptase inhibitor-based or protease inhibitor-based highly active antiretroviral therapy regimens are equivalent in incidence and prognosis
Clin Infect Dis
 , 
2005
, vol. 
41
 (pg. 
1326
-
32
)
20.
Podsadecki
TJ
Vrijens
BC
Tousset
EP
Rode
RA
Hanna
GJ
Decreased adherence to antiretroviral therapy observed prior to transient human immunodeficiency virus type 1 viremia
J Infect Dis
 , 
2007
, vol. 
196
 (pg. 
1773
-
8
)
21.
Garcia-Gasco
P
Maida
I
Blanco
F
, et al.  . 
Episodes of low-level viral rebound in HIV-infected patients on antiretroviral therapy: frequency, predictors and outcome
J Antimicrob Chemother
 , 
2008
, vol. 
61
 (pg. 
699
-
704
)
22.
Easterbrook
PJ
Ives
N
Waters
A
, et al.  . 
The natural history and clinical significance of intermittent viraemia in patients with initial viral suppression to <400 copies/mL
AIDS
 , 
2002
, vol. 
16
 (pg. 
1521
-
7
)
23.
Moore
AL
Youle
M
Lipman
M
, et al.  . 
Raised viral load in patients with viral suppression on highly active antiretroviral therapy: transient increase or treatment failure?
AIDS
 , 
2002
, vol. 
16
 (pg. 
615
-
8
)
24.
Masquelier
B
Pereira
E
Peytavin
G
, et al.  . 
Intermittent viremia during first-line, protease inhibitors-containing therapy: significance and relationship with drug resistance
J Clin Virol
 , 
2005
, vol. 
33
 (pg. 
75
-
8
)
25.
Harrigan
R
Sherlock
C
Hogg
R
, et al.  . 
What should be the primary virological endpoint in clinical trials of ARV efficacy? [Paper 661]
Program and abstracts of the 18th Conference on Retroviruses and Opportunistic Infections
 , 
2011
(Boston, MA). Alexandria, VA: CROI
26.
Cohen Stuart
JW
Wensing
AM
Kovacs
C
, et al.  . 
Transient relapses (“blips”) of plasma HIV RNA levels during HAART are associated with drug resistance
J Acquir Immune Defic Syndr
 , 
2001
, vol. 
28
 (pg. 
105
-
13
)
27.
Macias
J
Palomares
JC
Mira
JA
, et al.  . 
Transient rebounds of HIV plasma viremia are associated with the emergence of drug resistance mutations in patients on highly active antiretroviral therapy
J Infect
 , 
2005
, vol. 
51
 (pg. 
195
-
200
)
28.
Taiwo
B
Gallien
S
Aga
E
, et al.  . 
Antiretroviral drug resistance in HIV-1–infected patients experiencing persistent low-level viremia during first line therapy
J Infect Dis
 , 
2011
, vol. 
204
 (pg. 
515
-
20
)
29.
Flandre
P
Pincon
C
Aboulker
JP
, et al.  . 
Analyzing plasma HIV-1 RNA measurements as multiple recurrent events in clinical trials
HIV Clin Trials
 , 
2006
, vol. 
7
 (pg. 
116
-
24
)
30.
Srasuebkul
P
Ungsedhapand
C
Ruxrungtham
K
, et al.  . 
Predictive factors for immunological and virological endpoints in Thai patients receiving combination antiretroviral treatment
HIV Med
 , 
2007
, vol. 
8
 (pg. 
46
-
54
)
31.
Palmer
AK
Klein
MB
Raboud
J
, et al.  . 
the CANOC Collaboration
Cohort profile: the Canadian Observational Cohort collaboration
Int J Epidemiol
 , 
2011
, vol. 
40
 (pg. 
25
-
32
)
32.
Percus
JK
Percus
OE
Markowitz
M
, et al.  . 
The distribution of viral blips observed in HIV-1–infected patients treated with combination antiretroviral therapy
Bull Math Biol
 , 
2003
, vol. 
65
 (pg. 
263
-
77
)
33.
DiMascio
M
Ribeiro
RM
Markowitz
M
, et al.  . 
Modeling the long-term control of viremia in HIV-1 infected patients treated with antiretroviral therapy
Math Biosci
 , 
2004
, vol. 
188
 (pg. 
47
-
62
)
34.
Prentice
RL
Williams
BJ
Peterson
AV
On the regression analysis of multivariate failure time data
Biometrika
 , 
1981
, vol. 
68
 (pg. 
373
-
9
)
35.
Miller
LG
Golin
CE
Liu
H
, et al.  . 
No evidence of an association between transient HIV viremia (“blips”) and lower adherence to the antiretroviral medication regimen
J Infect Dis
 , 
2004
, vol. 
189
 (pg. 
1487
-
96
)

Author notes

Presented in part: 19th Annual Canadian Association of HIV Research Conference, Saskatoon, Saskatchewan, Canada, May 2010. Abstract O008.
a
All additional research team members are listed in the “Acknowledgments” section.