Modification of the Abbott Real Time PCR Assay to Detect HIV-1 Plasma RNA Viral Loads Less than One Copy per Milliliter
Steven Yukl1,2, Peilin Li1, Katsuya Fujimoto1*, Steven Deeks2,3, Teri Liegler2, Mark Pandori4, Diane V. Havlir2,3, Joseph K. Wong1,2
1. San Francisco VA Medical Center (SFVAMC), San Francisco, CA, USA
2. University of California, San Francisco (UCSF), San Francisco, CA, USA
3. San Francisco General Hospital, San Francisco, CA, USA
4. S.F. Department of Public Health, San Francisco, CA, USA
* Present address: Hokkaido University Graduate School of Medicine, Japan
Poster # 656
Abstrac t V-141
Name: S teven Y ukl
Address: 4150 Clement S treet (111W3)
S an Franc isc o, CA 94121
Phone: 415-624-9088, Fax: 415-379-5600
Email: steven.yukl@uc sf.edu
Though ART can reduce the plasma HIV-1 viral load (VL) below the detection limit of
commercial assays, with the use of more sensitive tests, it is possible to detect
residual HIV-1 RNA in the plasma of almost all patients on ART1-4. Viral loads of <50
copy/ml have been correlated with increased risk of virologic failure5, poor CD4+ T
cell reconstitution on ART6, and blood levels of soluble immune activation markers7
and T cell activation markers6,8,9. Other potential applications for detecting low-level
viremia include earlier diagnosis of acute infection, improved screening of the blood
supply, confirming infection in HIV vaccine recipients, understanding mechanisms
of viral persistence, and monitoring therapies aimed at eradication.
In the U.S., commercially available assays are approved to quantify levels of HIV >20
(Roche), 40 (Abbott), 50-75 (Chiron/Bayer), or 176 copies/ml (bioMerieux), though
actual limits of detection may be lower. For example, the Abbott RealTime (Abbott)
assay will detect 20 copies/ml in 50 of 57 (88%) cases, 10 copies/ml in 38 of 56 (68%)
cases, and 5 copies/ml in 30 of 57 (53%) cases.
The sensitivity of VL testing can be improved by using assays that can detect one or
a few copies, running samples in replicate, and concentrating virus from a larger
volume of plasma. Using centrifugation to pellet virions from up to 7ml, several
groups have modified existing commercial assays1,3 or created in-house assays2,10
with detection limit of 20 copy/ml1, <5 copy/ml2, 2.5 copy/ml3, or 1 copy/ml10.
However, centrifugation of higher plasma volumes often results in pelleting of
poorly-soluble material that seems to variably interfere with nucleic acid extraction
or with PCR reactions.
Given that density cushions of sucrose and iodixanol11 have been used to pellet
virions from culture supernatants, it was hypothesized that an iodixanol cushion
could be used to pellet virions from higher volumes of plasma while excluding
plasma constituents that affect RNA extraction or PCR efficiencies. A method is
presented here for concentrating virus from up to 30ml of plasma by centrifugation
on a density cushion of 10% iodixanol. This method can be used in conjunction with
the Abbot assay to create an easy-to-use, widely-available, reproducible test for
quantifying plasma VL <1 copy/ml.
HIV+ adults with known VL >40 copies/ml or undetectable VL were recruited from
two clinic-based cohorts in San Francisco.
Preparation of Plasma
Blood (68ml) was collected in eight 8.5 ml BD Vacutainer ACD tubes (BD),
combined, and spun at 1,000g for 10min without brake. The plasma was pipetted
off (keeping >2ml from the buffy coat), combined, spun again at 1,000g for 10min
without brake, decanted into a 50ml conical tube, and stored at 4°C overnight for
subsequent pelleting of virions.
Efficacy of Concentration Across a Range of Copy Numbers
To determine the efficacy of concentration across a range of copy numbers,
different volumes of HIV+ plasma from participants with known VL (quantified using
the Abbott assay) were spiked into 30 ml aliquots of HIV- donor plasma to achieve a
series of final HIV concentrations ranging from 0.5 to 50 copy/ml (experiment 1) or
0.9 to 1345 copies/ml (experiment 2).
Efficacy of Detection from Very Low Concentrations of Virus
To determine if the modified assay could detect very low levels of viremia, HIV+
plasma from two additional subjects with known, detectable VL was spiked into
replicate 30 ml aliquots of HIV- donor plasma to obtain 10 replicates of 0.5 copy/ml
and 10 replicates of 1.0 copy/ml (5 replicates from each subject).
Efficacy of Detection from Clinical Samples with Undetectable VL
To determine the performance characteristics in clinical samples, blood was
obtained from 12 fasting ART-treated individuals with VL that was undetectable
using the Abbott assay. For 7 of these participants, plasma was obtained at two
different time points, approximately two weeks apart, during which time there was
no significant change in clinical status.
Technique for Concentration of Virions
Virions were pelleted by centrifugation on a density cushion. Plasma was diluted
1:1 with PBS to reduce the density (note: subsequent experiments have suggested
that this step is not necessary), divided in two portions, and gently layered onto
10ml of 6% iodixanol (OptiPrep Density Gradient Medium (Sigma) diluted 1:10 in
PBS) in 50 ml polypropylene tubes (Beckton). Samples were then centrifuged at
47,810g (20,000rpm in SH3000 rotor of a Sorvall RC6) for 3hrs at 4°C without
braking. Viral pellets were resuspended in a total of 1000µl of PBS and frozen at -
HIV Quantification and Statistical Analysis
Samples were thawed and the HIV-1 RNA was measured according to the Abbott
m2000 RealTime assay protocol. Viral load values (copies/ml) determined by the
Abbott assay were adjusted for the concentration factor. Linear regressions (best-
fit method) and correlations (Pearson r for data meeting normality tests; Spearman
r for other data) were performed using GraphPad Prism 5.0.
Concentration is effective across a range of copy numbers
In 2 experiments with HIV concentrations ranging from 0.5 to 1345 copies/ml, all
samples gave detectable results and there was a strong linear correlation between the
number of copies that had been spiked into donor plasma (input) and the number of HIV
RNA copies measured from the pelleted and resuspended virions (output) (r2 = 0.998,
P=0.001 for experiment 1 and r2 = 0.97, P<0.0001 for experiment 2; see Figure 1A-B). The
slopes of the linear regression lines were 0.80 and 0.81, respectively.
Very low concentrations of virus can be reproducibly detected
For samples with a calculated input of 0.5 copy/ml, HIV RNA was detected in 6 of 10
replicates (Table 1), with a mean measured output of 0.62 copy/ml (range: 0-2.1
copy/ml). For samples with a calculated input of 1.0 copy/ml, HIV RNA was detected in
10 of 10 replicates (Table 1), with a mean value of 1.36 copy/ml (range: 0.3-2.6 copy/ml).
Virus is measurable in clinical samples with undetectable VL
Using our assay, HIV RNA was detected in 16 of 19 samples (84%) from 12 of 12 (100%)
study participants (Table 2). The calculated VL ranged from 0.4 to 11.4 copies/ml, with
most values between 0.4 and 6.2 copies/ml.
VL correlates inversely with duration of viral suppression
For 10 of 12 participants in whom clinical data were available, there was a trend towards
an inverse correlation between the plasma VL as measured by our assay (mean of
available time points) and the duration of the last period over which the VL was
consistently undetectable by commercial assays (Spearman r = -0.61; p=0.067; Figure
2A). For the 8 participants for whom data was available since the first period of
suppression, the VL measured by our assay correlated inversely (Pearson r= -0.82,
p=0.013) with the total duration of viral suppression (Figure 2B).
The Abbott RealTime assay was adapted to detect plasma VL <1 copy/ml. Because the
Abbott assay has been extensively validated for clinical samples and can
reproducibly measure plasma virus from a variety of HIV-1 subtypes, the modified
assay is a widely-available, easily-standardized method of detecting plasma VL <1
Plasma HIV levels could be detected and measured in all patients whose VL had been
undetectable using the conventional Abbott assay. In a small number of
participants with varying total duration of viral suppression (median 6.3 yrs), the
measured VL tended to correlate inversely with the duration of suppression on
ART. While studies using other assays have suggested that the residual VL
plateaus by 1-5 yrs of therapy4,8,12, the above correlation suggests that some
patients may have a more extended decline in plasma virus.
Additional modifications may serve to further increase the performance of this assay,
including use of a control viral particle to account for any variation in pelleting or
resuspension, concentration of virus from more than 30ml of plasma, or use of
detection platforms with greater sensitivity.
Assays such as this can be used to further delineate the clinical or prognostic
significance of low-level viremia (see Introduction) and should prove useful for
research aimed at eradication.
1. We report here a method of concentrating HIV-1 virions
from up to 30ml of plasma, which can be combined with the
Abbott assay to create a widely-available, reproducible
assay for quantifying plasma HIV RNA levels <1 copy/ml.
2. Viral loads measured with this assay correlated with the
duration of viral suppression, suggesting that the residual
viremia may continue to decay slowly over years on ART.
3. The assay described here could be applied to address
many other questions in clinical and basic HIV research.
Figure 1A-B: In two separate experiments (1A and 1B), different volumes of HIV+
plasma of known viral load (measured using the Abbott assay) were spiked into 30 ml
aliquots of HIV- donor plasma to get known copy numbers of HIV. The x-axis shows
the log10 of the number of copies in 30 ml of plasma (input). Virions were then pelleted
by centrifugation on iodixanol, resuspended in 1 ml of PBS, and measured again using
the Abbott assay. The y-axis shows the log10 of the number of copies in the 1 ml of
PBS (output). In 1A, error bars indicate the SEM from two independent replicates.
Table 2: Plasma HIV-1 RNA, as measured using
the modified assay, in subjects with viral load<40
copies/ml; aND = not detected
Table 2: efficacy of detection from
clinical samples with undetectable VL
Figure 1: efficacy of concentration
across a range of copy numbers
We thank the following people and institutions: 1) the study participants; 2) the study nurses at the
San Francisco VA ID Clinic (Sandra Charles and Linda Adams); 3) the PLUS study staff; and 4) the
staff at the San Francisco Department of Public Health. This work was supported in part by the U.S.
Department of Veterans Affairs (VA Merit Award [JW/SY]) and the National Institute of Health (NIH
grants 1K23AI089397 (SY), P30-AI027763 (SY), NS051145 [JW/SY], and T32 AI60530 [DH/SY]).
Table 1: Recovery of low copy numbers of HIV after dilution in
donor plasma and centrifugation; aND = not detected
Table 1: efficacy of detection from very
low concentrations of virus
Figure 2: inverse correlation between
VL and duration of viral suppression
Figure 2A-B: Correlation of plasma HIV-1 RNA (as measured by our assay) with
the duration of the last period over which the VL was consistently undetectable by
commercial assays (2A) or the total period that the VL was undetectable by
commercial assays (2B) [defined by the sum of the lengths of all time periods with
two adjacent VL that were undetectable by commercial assays].
1. Schockmel, G.A., et al. J Acquir Immune Defic Syndr Hum Retrovirol 1997
2. Dornadula, G., et al. JAMA 1999.
3. Havlir, D.V., et al. JAMA 2001.
4. Maldarelli, F., et al. PLoS Pathog 2007.
5. Geretti, A.M., et al. CROI 2010 Abstract 505.
6. Mavigner, M., et al. PLoS One 2009.
7. Ostrowski, S.R., et al. Scand J Immunol 2008.
8. Havlir, D.V., et al. J Virol 2003.
9. Yukl, S.A., et al. J Infect Dis 2010.
10. Palmer, S., et al. J Clin Microbiol 2003.
11. Dettenhofer, M. and Yu, J Virol 1999.
12. Hatano, H., et al. AIDS 2010 (24, 2535-9).