Schneider, Michael F. MS*; Margolick, Joseph B. MD, PhD†; Jacobson, Lisa P. ScD*; Reddy, Susheel MPH‡; Martinez-Maza, Otoniel PhD§,‖,¶; Muñoz, Alvaro PhD*
Departments of *Epidemiology
†Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD
‡Division of Infectious Diseases, The Feinberg School of Medicine at Northwestern University, Chicago, IL
Departments of §Obstetrics & Gynecology
‖Microbiology, Immunology & Molecular Genetics, David Geffen School of Medicine at UCLA, Los Angeles, CA
¶Department of Epidemiology, UCLA School of Public Health, Los Angeles, CA.
Correspondence to: Alvaro Muñoz, PhD, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe Street, Room E7648, Baltimore, MD 21205 (e-mail: email@example.com).
Supported by the National Institute of Allergy and Infectious Diseases, with additional supplemental funding from the National Cancer Institute: UO1-AI-35042, UL1-RR025005, UO1-AI-35043, UO1-AI-35039, UO1-AI-35040, and UO1-AI-35041.
J.B.M. has received funds from the International AIDS Society (IAS); L.P.J. has received funds from Bristol-Myers Squibb, and A.M. has received funds from Pfizer.
The authors have no other funding or conflicts of interest to disclose.
Received October 17, 2011
Accepted December 14, 2011
Plasma HIV-1 RNA levels among individuals receiving highly active antiretroviral therapy (HAART) have been shown to follow a bimodal distribution.1 The rightmost peak of the distribution reflects individuals with suboptimal virologic response (eg, due to viral drug resistance or poor adherence to therapy), whereas the leftmost peak, which includes the great majority of individuals on HAART, reflects the maximal suppression that can be achieved. The latter peak has not been well characterized because it is composed primarily of values that are below the limits of detection (LD) of assays that have been commercially available (eg, 50 copies per milliliter). Recently, however, more sensitive assays have been developed, either research assays that can measure down to 1 copy per milliliter or commercial assays that can measure down to 20 copies per milliliter.2–6 These assays have the potential to delineate more clearly the leftmost peak of the HIV-1 RNA distribution in treated populations. The characterization of this leftmost peak is important to understand not only the effectiveness of HAART but also the biology of optimally treated HIV-1 infection.
Two assays from Roche Molecular Diagnostics used to measure HIV-1 RNA are the AmpliPrep/Amplicor MONITOR version 1.5,7 with LD of 50 copies per milliliter, and the recently US Food and Drug Administration–approved COBAS AmpliPrep/COBAS TaqMan version 2.0,8 whose LD is 20 copies per milliliter. The lower LD of the TaqMan assay, allowing for the quantification of plasma HIV-1 RNA values between 20 and 49 copies per milliliter, reduces the number of left-censored observations and thus improves estimation of the distribution of HIV-1 RNA levels below 20 copies per milliliter. In 2010, the Multicenter AIDS Cohort Study (MACS), which had been measuring HIV-1 RNA using the Amplicor assay, adopted the TaqMan assay for routine use. The goal of the current study was to characterize the distribution of suppressed HIV-1 RNA values among HAART-experienced individuals using an assay with an LD of 20 copies per milliliter and to show the misrepresentation provided by a less sensitive assay with an LD of 50 copies per milliliter.
The MACS was initiated in 1984 to study the natural history of HIV-1 infection among homosexual and bisexual men at 4 sites in the United States (Chicago; Baltimore/Washington, DC; Los Angeles; and Pittsburgh). The study design and characteristics of the study participants have been described previously.9,10 Briefly, a total of 3554 participants were either HIV-1 infected at enrollment (81%) or became HIV-1 infected during follow-up (19%). Of the 1854 HIV-1–infected men who were observed after January 1995, 1488 (80%) initiated HAART11 before October 2010. Appropriate institutional review boards approved study protocols and consent forms, and each study participant gave written informed consent.
Fresh specimens from 200 consecutive HIV-1–infected participants (100 from the Baltimore/Washington DC site and 100 from the Los Angeles site) seen between June 2010 and September 2010 were tested with both TaqMan and Amplicor HIV-1 RNA assays. One hundred twenty-seven samples of plasma stored at −70°C in local freezers and taken from participants seen between May 2009 and May 2010 were also tested with both HIV-1 RNA assays. Our analyses included the 306 (94% of 327) samples taken from 273 individuals who had initiated HAART. HAART was reported at 263 (86%) of the 306 study visits included in analyses; combination therapy at 15 (5%) study visits, monotherapy at 2 (<1%) study visits, and no therapy at 14 (5%) of study visits. At 12 (4%) of the 306 study visits, data on antiretroviral therapy used concomitantly with the HIV-1 RNA measurement were not available. At the 263 study visits in which HAART was reported, at least 1 protease inhibitor was reported in 133 (51%) instances and 1 nonnucleoside transcriptase inhibitor was reported 143 (54%) times. The median time since HAART had been initiated was 11.8 years [interquartile range (IQR) = 7.6–13.4 years].
HIV-1 RNA Assays
Both the COBAS TaqMan HIV-1 Test, Version 2.0, and the Roche Amplicor HIV-1 MONITOR Test, Version 1.5, are tests for the quantification of HIV-1 RNA based on in vitro amplification of the highly conserved HIV-1 gag gene.7,8 The Amplicor is a semiautomated, conventional polymerase chain reaction assay that has a dynamic range of 50–750,000 copies per milliliter. The TaqMan is a completely automated, real-time polymerase chain reaction assay that has a broader dynamic range (20–10,000,000 copies per milliliter) and also targets the HIV-1 long terminal repeat region.
Let Y denote the left-censored random variable log10(HIV-1 RNA) using the TaqMan assay with LD = log10(20) copies per milliliter, which hereafter we will refer to as L20. We allow Y to follow a bimodal mixture of 2 Gaussian distributions, one of which is the distribution of log10(HIV-1 RNA) measurements among individuals with lower levels (leftmost peak) and the second of which corresponds to the distribution of log10(HIV-1 RNA) measurements among individuals with higher levels (rightmost peak). Maximum likelihood methods were used as described by Chu et al.12 Similar methods were used for the measurements from the Amplicor assay with LD as L50 = log10(50) copies per milliliter.
To determine the correlation between the TaqMan and Amplicor measurements in both the leftmost and rightmost peaks (ie, 2 correlations), we allowed the random vector Y = (Y1, Y2) = (log10[TaqMan-measured HIV-1 RNA], log10[Amplicor-measured HIV-1 RNA]) to follow a bivariate mixture of 2 Gaussian distributions. Maximum likelihood methods were used as described by Chu et al.12
To determine the goodness of fit of each model, we compared the predicted percentage of HIV-1 RNA measurements falling below different levels with the observed percentages. More importantly, appropriately modeling the distribution of HIV-1 RNA below the LD allows for the estimation of the percentage of individuals with lower than, for example, 1 copy per milliliter, and the percentage expected to have less than 1 copy in the whole plasma volume (∼3 L). Acknowledging the immunological finding that HIV cannot be completely eradicated from HIV-infected individuals with current therapies,6,13–15 the percentage of individuals with less than 1 copy per 3 L of plasma is expected to be very low.
The median age of the 273 men was 52.4 years (IQR = 46.4–58.6 years); 141 (52%) were White and 71 (26%) were African American. The median CD4 cell count was 568 cells per cubic millimeter (IQR = 414–741 cells per cubic millimeter). Forty-seven (17%) had a history of an AIDS-defining condition before the study visit at which their HIV-1 RNA was measured.
The HIV-1 RNA values and CD4 cell counts of the MACS men included in this convenience sample are internally consistent with the distributions of HIV-1 RNA and CD4 cell counts of the entire MACS HIV-infected population. Specifically, at their last study visit between October 2009 through September 2010, 79% of the entire cohort had undetectable HIV-1 RNA using the Amplicor assay (similar to the percentage of Amplicor measurements that were ≤50 copies per milliliter in our study population described below), and their median CD4 cell count was 557 cells per cubic millimeter (IQR = 406–728 cells per cubic millimeter), similar to the median of 568 cells per cubic millimeter (IQR = 414–741 cells per cubic millimeter) seen among individuals included in this study.
Using the Amplicor assay, 254 (83%) of the HIV-1 RNA measurements were less than or equal to 50 copies per milliliter. Two hundred sixteen (71%) of the HIV-1 RNA measurements using the TaqMan assay were less than or equal to 20 copies per milliliter, and 30 (10%) were between 21 and 50 copies per milliliter. Of the 306 samples tested, 214 (70%) were ≤L20 for TaqMan and ≤L50 for Amplicor; 50 (16%) were >L20 and >L50; 40 (13%) were >L20 and ≤L50; and 2 (1%) were ≤L20 and >L50. The median duration of undetectable HIV-1 RNA in those who were ≤L20 for TaqMan and ≤L50 for Amplicor was approximately 4.5 years.
The bimodal mixture of 2 Gaussian distributions used to describe the distribution of log10(HIV-1 RNA) values for the Amplicor and TaqMan assays are shown in the top and bottom panels of Figure 1, respectively. The goodness of fit of the bimodal distributions to the observed data is readily apparent from the nearly perfect agreement between the observed and predicted cumulative percentages shown for both assays.
Figure 2 shows the fitted bimodal distributions for the Amplicor (blue line) and TaqMan (black line) assays. The 9% (=80% − 71%) of the HIV-1 RNA measurements that were between 21 and 50 copies per milliliter using the TaqMan assay represent the measurements that would have been undetectable if the LD had been 50 copies per milliliter instead of 20 copies per milliliter. The 2 distributions were very similar for values >1000 copies per milliliter (ρR = 0.99), but diverged substantially for HIV-1 RNA levels <50 copies per milliliter (ρL = 0.63). The leftmost peak of the TaqMan distribution was centered at a higher value (median = 3.78 copies per milliliter) than that of the Amplicor distribution (median = 0.59 copies per milliliter). Furthermore, the more sensitive TaqMan assay predicted that 23%, 0.2%, 0.003%, and 0.0003% of values would be <1 copy per milliliter, <1 copy per deciliter, <1 copy per liter, and <1 copy per 3 L, respectively, consistent with clinical observations indicating that HIV-1 RNA cannot be completely eradicated by current HAART regimens. This is in sharp contrast to the higher estimates provided by the less sensitive Amplicor assay, whereby the corresponding predicted percentages were 51%, 9%, 2%, and 1%, respectively.
The present study confirms our previous report that the distribution of post-HAART plasma log10(HIV-1 RNA) measurements is bimodal.1 It also extends the previous report by incorporating data obtained using the TaqMan assay, which has an LD of 20 copies per milliliter rather than 50 copies per milliliter for the Amplicor assay used in the previous report, and thus allowed for a much better estimation of the distribution of HIV-1 RNA at very low values.
Use of the TaqMan assay provided data for 9% of the HIV-1 RNA distribution that were not measured using the Amplicor assay, that is, the values between 21 and 50 copies per milliliter. With these added data, the leftmost peak of the distribution of TaqMan measurements was steeper and farther to the right (median = 3.78 copies per milliliter) than the distribution of HIV-1 RNA using the Amplicor assay (median = 0.59 copies per milliliter) and predicted many fewer individuals with levels of viremia <1 copy per milliliter than the Amplicor assay.
The improved estimates are consistent with the findings of studies indicating that HIV-1 RNA cannot be eradicated completely from circulating CD4 T cells by present HAART regimens.13–15 In addition, using the TaqMan assay, our estimates of the leftmost peak and the percentage of observations predicted to be <1 copy per milliliter are strongly supported by the measurements obtained in 2 studies that used a specialized assay that quantified plasma HIV-1 RNA values as low as 1 copy per milliliter. Maldarelli et al16 found a mean HIV-1 RNA of 3.1 copies per milliliter in people who had maintained HIV-1 RNA values <50 copies per milliliter for 1–2 years while receiving HAART, with >80% of samples having detectable viremia. Similarly, Palmer et al4 studied 293 plasma samples from 40 people who maintained HIV-1 RNA values <50 copies per milliliter for several years while receiving HAART and found a median value of 3.34 copies per milliliter among the 77% of samples yielding values ≥1 copy per milliliter. These estimates are in excellent agreement with the estimate of 3.78 copies per milliliter and 23% of values below 1 copy per milliliter in the present study. Taken together, these data suggest that the TaqMan assay with LD = 20 copies per milliliter is adequate for characterizing the distribution of suppressed HIV-1 RNA values. On the other hand, the Amplicor assay with LD = 50 copies per milliliter produces a stark misrepresentation of the distribution of suppressed HIV-1 RNA at very low levels.
The improved estimate of the distribution of HIV-1 RNA among individuals achieving viral suppression on HAART permits more precise quantitation of the effectiveness of HAART. Although plasma HIV-1 concentrations can now be directly measured down to 1 copy per milliliter using specialized assays, such assays require large amounts of plasma (eg, 7 mL) and may not be feasible in large studies or at most clinical sites. Therefore, the methods and improved estimates provided in the present study may be useful in studies of the effect of residual HIV-1 viremia in people receiving HAART and in the assessment of medications or therapeutic approaches that may suppress HIV replication even more effectively than current HAART regimens or that reduce the latent reservoirs of HIV-1.
Data in this article were collected by the MACS with centers (Principal Investigators) located at The Johns Hopkins Bloomberg School of Public Health (Joseph B. Margolick); Howard Brown Health Center and Northwestern University Medical School (John P. Phair); University of California, Los Angeles (Roger Detels); University of Pittsburgh (Charles R. Rinaldo); and Data Analysis Center (Lisa Jacobson).
1. Li X, Chu H, Gallant JE, et al.. Bimodal virological response to antiretroviral therapy for HIV infection: an application using a mixture model with left censoring. J Epidemiol Community Health. 2006;60:811–818.
2. Sizmann D, Glaubitz J, Simon CO, et al.. Improved HIV-1 RNA quantitation by COBAS AmpliPrep/COBAS TaqMan HIV-1 Test, v2.0 using a novel dual-target approach. J Clin Virol. 2010;49:41–46.
3. Glaubitz J, Sizmann D, Simon CO, et al.. Accuracy to 2nd International HIV-1 RNA WHO Standard: assessment of three generations of quantitative HIV-1 RNA nucleic acid amplification tests. J Clin Virol. 2011;50:119–124.
4. Palmer S, Maldarelli F, Wiegand A, et al.. Low-level viremia persists for at least 7 years in patients on suppressive antiretroviral therapy. Proc Natl Acad Sci U S A. 2008;105:3879–3884.
5. Dinoso JB, Kim SY, Siliciano RF, et al.. A comparison of viral loads between HIV-1-infected elite suppressors and individuals who receive suppressive highly active antiretroviral therapy. Clin Infect Dis. 2008;47:102–104.
6. Dinoso JB, Kim SY, Wiegand AM, et al.. Treatment intensification does not reduce residual HIV-1 viremia in patients on highly active antiretroviral therapy. Proc Natl Acad Sci U S A. 2009;106:9403–9408.
7. Amplicor HIV-1 MONITOR®Test, version 1.5 [package insert]. Branchburg, NJ: Roche Molecular Systems, Inc.; 2002.
8. COBAS® AmpliPrep/COBAS® TaqMan® HIV-1 Test, version 2.0 [package insert]. Branchburg, NJ: Roche Molecular Systems, Inc.; 2010.
9. Kaslow RA, Ostrow DG, Detels R, et al.. The Multicenter AIDS Cohort Study: rationale, organization, and selected characteristics of the participants. Am J Epidemiol. 1987;126:310–318.
10. Dudley J, Jin S, Hoover D, et al.. The Multicenter AIDS Cohort Study: retention after 9½ years. Am J Epidemiol. 1995;142:323–330.
12. Chu H, Moulton LH, Mack WJ, et al.. Correlating two continuous variables subject to detection limits in the context of mixture distributions. Appl Statist. 2005;54:831–845.
13. Finzi D, Hermankova M, Pierson T, et al.. Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy. Science. 1997;278:1295–1300.
14. Chun TW, Stuyver L, Mizell SB, et al.. Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy. Proc Natl Acad Sci U S A. 1997;94:13193–13197.
15. Bailey JR, Sedaghat AR, Kieffer T, et al.. Residual human immunodeficiency virus type 1 viremia in some patients on antiretroviral therapy is dominated by a small number of invariant clones rarely found in circulating CD4+ T cells. J Virol. 2006;80:6441–6457.
16. Maldarelli F, Palmer S, King MS, et al.. ART suppresses plasma HIV-1 RNA to a stable set point predicted by pretherapy viremia. PLoS Pathog. 2007;3:e46.
© 2012 Lippincott Williams & Wilkins, Inc.