Abstract
Introduction
One primary goal of antiretroviral treatment (ART) is suppression of viral load (VL) to below the level of detection. 1 An accurate quantification of HIV-1 VL is an indispensible tool in the management of HIV-1 infected patients to determine the appropriate time to initiate therapy and the response to treatment. While a number of methods of VL measurement have been approved, clinicians currently use one of two techniques: signal amplification with branched DNA (bDNA) or nucleic acid amplification with polymerase chain reaction (PCR). Early studies comparing these methods demonstrated higher VLs using PCR methods. 2 Later versions of the bDNA methods had improved sensitivity but still reported lower VLs compared with PCR testing.3,4
On November 27, 2006, the method for VL detection in our institution changed from Bayer Versant HIV RNA version 3.0 (bDNA 3.0) to Roche COBAS Amplicor HIV-1 Monitor version 1.5 (a PCR method). The quantification of the bDNA method has been shown to be 0.3 log-folds lower than that of Amplicor.3,4 As this increase may result in increased numbers of patients with detectable VL measurements using the PCR method, the magnitude and observed duration of this effect is unknown. In addition, the effect on physician management has not been studied. We evaluated the frequency and magnitude of viral load increase as well as physician response and long-term virologic outcomes in patients with newly detected viral load measurements after the change in method.
Methods
We performed a retrospective case-control study of adult HIV+ patients on ART with matched VL performed from 2 time periods: June 1, 2006 to November 26, 2006, when VL was assessed using the Versant bDNA assay and November 27, 2006, to November 27, 2007, using the Amplicor PCR assay. Cases were defined as patients with undetectable VL (<50 copies/mL) using bDNA that was detectable by RT-PCR (≥50 copies/mL) after the switch. Control patients were undetectable in both time frames. We collected the following baseline information on cases and controls: demographics, years since HIV diagnosis, line of ART (eg first, second, or third regimen), and type of ART (protease-inhibitor containing, nonnucleoside reverse inhibitor containing, or triple nucleoside reverse transcriptase inhibitor). We recorded patient CD4 counts, VL, physician response to increased VL (no change, resistance testing, increased frequency of VL monitoring, or change in ART), and results of resistance testing if performed. Patients without CD4 and VL in both time frames and patients with detectable VL during the first time period were excluded.
The frequency of viral breakthrough during the second period was recorded. Cases and controls were compared to determine risk factors for viral breakthrough using χ2 for categorical variables and Student t-test and ANOVA for continuous variables. Long-term (1 year) virologic outcome was compared between cases and controls to determine whether the breakthrough was persistent. Data were analyzed using SPSS v12.0.
Results
Among 170 HIV+ outpatients, 105 patients were excluded as 51 had a detectable VL using bDNA and 54 of 105 did not have paired testing during both time frames of the study. The remaining 65 patients were either identified as cases (n = 38, 58.5%) or controls (n = 27, 41.5%). The comparison of demographics and clinical features are shown in the Table 1 . The case patients were more likely to be male and had a significantly longer period of HIV diagnosis. The baseline CD4 (bDNA period) was similar in the 2 groups. The CD4 among cases increased from 524 ± 257.0 to 558 ± 297.0 in the PCR period (P = .19). Among controls, CD4 increased from 441.7 ± 202.8 to 527.9 ± 296.1 (P < .03). The mean change in CD4 (ΔCD4) in control and case groups was 86.2 ± 194.8 and 33.3 ± 153.9, respectively (P = .27). The mean viral load in the case group was 684 ± 1230 copies/mL (2.5 ± 0.6 log).
Comparison of Cases (n = 38) and Controls (n = 27)
a Regimen containing protease inhibitors.
b Regimen containing non-nucleotide reverse transcriptase inhibitors.
c Regimen containing triple nucleotide reverse transcriptase inhibitors.
d Regimen containing enfurvitide.
* Regimens exceed 100% as patients may have been on combinations of these agents.
Among the 38 cases, 12 (31.6%) did not come for follow-up office visits in the immediate PCR period and the physician response to viral breakthrough could not be evaluated. After the switch, repeat VL testing and resistance testing were ordered in 4 of 26 (15.4%) and 6 of 26 (23.1%), respectively, of the cases. Among the 6 with resistance testing, 5 had VL > 1000 copies/mL. Of the 6 patients with resistance testing ordered, 3 patients did not have sufficient viral copies to run the test at the time it was ordered, 1 had resistance to the existing treatment regimen (VL 7190 copies/mL at time of breakthrough), 1 had a susceptible pattern, and 1 patient had the test ordered but not performed. At 1-year follow-up, laboratory data were available for 35 (92%) of cases and 25 (92.6%) of controls. By 1 year, further VL testing was undetectable in 29 of 35 (82.8%) cases and 23 of 25 (92%) controls (P = 0.30, χ2). All 6 cases with detectable viral load had resistance testing ordered. Among these 6, 2 patients had resistant virus to the existing treatment, 2 had insufficient VL to run the test, 1 patient had a susceptible virus, and 1 patient failed to have test performed. In the cases group, the mean VL for patients with resistant virus was 8160 copies/mL, and mean for the other 4 patients was 610 copies/mL. Both control patients with detectable viral load had resistance testing ordered. One patient had a VL of 20 300 copies/mL and had resistant virus to the existing treatment and the other insufficient VL to run the test.
Patients were maintained on the same HAART regimen unless their resistance testing revealed resistance to that regimen.
Discussion
The use of VL monitoring along with highly active ART has resulted in significant decreases in AIDS-related morbidity and mortality and slowed the rate of progression to AIDS among HIV-infected patients. The evolution of VL assays has allowed for lower levels of viral detection, but this increased sensitivity may be associated with increased detection of clinically insignificant increases in VL. As a result of the differing results of VL assays, it is not recommended to use different assays interchangeably to monitor VLs of patients.3,4 In addition to lower values seen with Versant 3.0 bDNA assay versus the Amplicor PCR method, Versant also produces lower values compared to the Cobas Ampliprep/Cobas TaqMan HIV-1 PCR assay.5,6 Among 136 patients with undetectable VL using Versant 3.0, 14 (10.3%) were detectable on paired testing using the TaqMan assay. In addition to the variations seen between bDNA and PCR assays, there is also variation among PCR assays. The TaqMan assay has been reported by several centers to have increased detection of VL among patients undetectable using Amplicor.7–12
The effect of the variations caused by changes in VL assay on clinicians practice and the long-term persistence of the increased rate of detectable VL have not been studied. In our institution, the switch to RT-PCR using Amplicor 1.5 from bDNA resulted in a finding of detectable VL in more than half of the patients with testing done in both time periods. The detectable VLs resulted in increased frequency of VL testing in 15.4% and resistance testing in 23.1% of case patients who returned for follow-up visits. When compared to cases, the ΔCD4 in the control group did not reach statistical significance (P = .27). By 1-year of follow-up, there was no difference in the rates of undetectable VL in the case and control groups indicating that most increases in VL were clinically insignificant.
There are several limitations to the study. First, we do not know whether the transient increase of VL to detectable levels in cases represent a “blip” rather than true differences in the sensitivities of both assays. Di Mascio et al 13 found the frequency of blips on each VL measurement to be 9% in a study of 123 patients with well-controlled HIV. This is substantially lower than our rate of 58.5% (38/65) with detectable VL after the switch. While the higher sensitivity seen with PCR assays provides better detection of VL at lower levels than bDNA assays, it has resulted in lower reproducibility of VL results and overestimation of HIV-1 RNA values around the lower quantification limit.3–5,9,11-12,14
A second limitation is that paired testing was not performed on each sample, as this was a retrospective study, to confirm the discrepancy between the assays. Our findings of higher rate of detectable VL after the switch to PCR concur with findings of a study showing a decrease of detected viremia from 39% to 5% when switching from PCR to bDNA method. 14 Another limitation is the lack of resistance testing for all samples with detectable VL. However, most of these detectable VL episodes were <1000 copies/mL and resolved by 1-year follow-up. Only episodes with VL >1000 copies/mL resulted in subsequent detection of resistant virus. Resistance testing is not recommended in patients with VL <1000 copies/mL or without sustained viremia >1000 copies/mL. 1
In conclusion, the increases in VL are a reflection of the variability in the result of 2 different assays that are not predictive of long-term virologic failure. Patients with true virologic failure had more significant increases in VL (range: 6860 to 20 300 copies/mL) compared to the mean VL among cases that was related to the new assay. Clinicians should be aware of this variability and perform resistance testing only on previously undetectable individuals with sustained increases in VL or VL >1 000 copies/mL after changing assays. In addition, these patients do not need more frequent VL testing if the VL is <1000 copies/mL when switching from bDNA to PCR methods.
Footnotes
Data presented in part at the 45th Annual Meeting of the Infectious Diseases Society of America, San Diego, California, October 4-7, 2007. The study was approved by the IRB committee at St. John Hospital and Medical Center.
The author(s) declared no conflicts of interest with respect to the authorship and/or publication of this article.
The author(s) received no financial support for the research and/or authorship of this article.