Comparison of Droplet Digital PCR and Seminested Real-
Time PCR for Quantification of Cell-Associated HIV-1 RNA
Maja Kiselinova1., Alexander O. Pasternak2., Ward De Spiegelaere1, Dirk Vogelaers1, Ben Berkhout2,
1HIV Translational Research Unit (HTRU), Department of Internal Medicine, Ghent University and Ghent University Hospital, Ghent, Belgium, 2Department of Medical
Microbiology, Laboratory of Experimental Virology, Center for Infection and Immunity Amsterdam (CINIMA), Academic Medical Center, University of Amsterdam,
Amsterdam, The Netherlands
Cell-associated (CA) HIV-1 RNA is considered a potential marker for assessment of viral reservoir dynamics and antiretroviral
therapy (ART) response in HIV-infected patients. Recent studies employed sensitive seminested real-time quantitative
(q)PCR to quantify CA HIV-1 RNA. Digital PCR has been recently described as an alternative PCR-based technique for
absolute quantification with higher accuracy compared to qPCR. Here, a comparison was made between the droplet digital
PCR (ddPCR) and the seminested qPCR for quantification of unspliced (us) and multiply spliced (ms) CA HIV-1 RNA. Synthetic
RNA standards and CA HIV-1 RNA from infected patients on and off ART (N=34) were quantified with both methods.
Correlations were observed between the methods both for serially diluted synthetic standards (usRNA: R2=0.97, msRNA:
R2=0.92) and patient-derived samples (usRNA: R2=0.51, msRNA: R2=0.87). Seminested qPCR showed better quantitative
linearity, accuracy and sensitivity in the quantification of synthetic standards than ddPCR, especially in the lower
quantification ranges. Both methods demonstrated equally high detection rate of usRNA in patient samples on and off ART
(91%), whereas ddPCR detected msRNA in larger proportion of samples from ART-treated patients (p=0.13). We observed
an average agreement between the methods for usRNA quantification in patient samples, albeit with a large standard
deviation (bias=0.0560.75 log10). However, a bias of 0.9460.36 log10was observed for msRNA. No-template controls were
consistently negative in the seminested qPCR, but yielded a positive ddPCR signal for some wells. Therefore, the false
positive signals may have affected the detection power of ddPCR in this study. Digital PCR is promising for HIV nucleic acid
quantification, but the false positive signals need further attention. Quantitative assays for CA HIV RNA have the potential to
improve monitoring of patients on ART and to be used in clinical studies aimed at HIV eradication, but should be cross-
validated by multiple laboratories prior to wider use.
Citation: Kiselinova M, Pasternak AO, De Spiegelaere W, Vogelaers D, Berkhout B, et al. (2014) Comparison of Droplet Digital PCR and Seminested Real-Time PCR
for Quantification of Cell-Associated HIV-1 RNA. PLoS ONE 9(1): e85999. doi:10.1371/journal.pone.0085999
Editor: Paul Richard Harrigan, University of British Columbia, Canada
Received July 12, 2013; Accepted December 3, 2013; Published January 21, 2014
Copyright: ? 2014 Kiselinova et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: MK is supported from the ‘Special Research Grant – BOF grant’ of Ghent University (Grant n: 01N02712). LV is supported by the Research Foundation –
Flanders (FWO) (Grant n: 1.8.020.09.N.00). AOP is supported by the research grants 2011020 and 2012025 from the Dutch AIDS Fonds. The funders had no role in
study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: Linos.Vandekerckhove@ugent.be
. These authors contributed equally to this work.
Current antiretroviral therapy (ART) effectively suppresses
HIV-1 plasma viremia by inhibiting viral replication. In most
patients, plasma viremia is suppressed below the detection limit
(20–40 viral copies/ml of plasma) of currently available diagnostic
assays . However, even in the settings of optimal therapy,
residual low-level viremia persists in a large subset of patients [1,2].
Because monitoring of ultra-low plasma viremia is technically
challenging, and it is unclear whether the low-level viremia has an
impact on long-term therapy response, new diagnostic markers
and tools will be needed to support HIV care and clinical guidance
with the next generation of antiretroviral therapies [1,3].
Recently, cell-associated (CA) HIV-1 RNA was demonstrated to
be a predictive marker of ART outcome in 26 patients .
Additionally, CA HIV-1 RNA was found to denote productive
HIV-1 infection in patients after therapy cessation and in patients
with modest nonadherence to ART [5,6]. Importantly, as
expression of CA HIV-1 RNA is believed to directly reflect the
reactivation of latent HIV reservoir in vivo, it was recently used to
monitor clinical trials aiming to purge the latent reservoir [7,8].
The role of CA HIV-1 RNA and its potential use as a virological
biomarker for monitoring the response to ART and to novel
therapeutic strategies has recently been reviewed in depth
elsewhere . With the current effort to find a strategy for HIV
eradication, an easy and straightforward assay to assess therapy
effectiveness is needed. In this framework, CA HIV-1 RNA is a
promising candidate biomarker for future diagnostic purposes.
Despite promising data indicating the importance of monitoring
CA HIV-1 RNA load in patients on ART, only a limited number
of studies have been conducted on these markers. This is mainly
due to the technical difficulties to monitor the low amounts of
HIV-1 RNA. In recent years, quantification of CA HIV-1 RNA
has been performed using assays based on quantitative reverse
transcription real-time PCR (RT-qPCR) [5,7,8,10]. However, this
PLOS ONE | www.plosone.org1 January 2014 | Volume 9 | Issue 1 | e85999
technique suffers from increased technical variation at the lower
ranges of detection . Moreover, small differences in efficiency
in the lower ranges of the standard curve may further bias
quantitative results [12,13]. To overcome these shortcomings,
Pasternak et al. developed a seminested real-time qPCR proce-
dure that enables CA HIV-1 RNA measurement in patient
samples with a lower limit of quantification and with increased
accuracy at the lower quantitative range compared to one-step
qPCR based assays . By performing two successive PCR
reactions, the specificity is maintained and the limit of quantifi-
cation is considerably reduced. The introduction of this method
revealed its value in multiple in vivo studies [4,6,14,15]. However,
an accurate standard curve is still necessary for seminested qPCR
quantification. This requires careful calibration and assumes
consistent amplification efficiencies between the biological samples
and the standards. A quantitative technique that does not rely on a
standard curve is therefore desirable.
Digital PCR (dPCR) has been described as an alternative PCR-
based technique for absolute quantification with higher accuracy
compared to qPCR [12,13,16]. The dPCR technique is based on
limiting dilution of samples across a large number of separate PCR
reactions. If the input sample is sufficiently diluted, not all
reactions will harbor template DNA. This will allow absolute
quantification using Poisson statistics without requiring a standard
curve [12,17]. In addition, decreased PCR efficiency is better
tolerated in dPCR as the end-point fluorescence suffices to
perform absolute quantification. Because of technical obstacles
and costs of making multiple reactions, dPCR has not been widely
implemented so far. However, thanks to recent technological
developments including microfluidics to form droplet in oil
suspension, dPCR is now possible in high throughput at lower
costs. To date, several studies on cancer and viral infections report
a higher degree of sensitivity and precision of dPCR than qPCR
[13,16,18,19]. In addition, Henrich et al. reported equal sensitivity
of ddPCR and qPCR for detection of HIV-1 DNA in patient
samples . On the other hand, one study evaluated CMV viral
load quantification and reported reduced sensitivity of dPCR
compared to qPCR . Taken together, the published data point
to the potential clinical use of dPCR for sensitive and accurate
absolute quantification of nucleic acids.
In this study, we compared seminested qPCR and digital
droplet PCR (ddPCR) for quantification of CA HIV RNA. We
first quantified the synthetic RNA standards, corresponding to
unspliced (us) and multiply spliced (ms) CA HIV-1 RNA, by
ddPCR and seminested qPCR. Based on the quantification of
these standards, raw data-to-RNA conversion factors were
generated for both methods. These conversion factors were
subsequently used, in the patient samples, to convert the raw
outputs of seminested qPCR (quantification cycles, Cq) and
ddPCR (cDNA copy numbers) to the HIV RNA copy numbers.
This allowed making a comparison between ddPCR and the
seminested qPCR for quantification of CA HIV-1 RNA in the
samples from HIV-infected patients on and off ART.
Materials and Methods
Thirty-four peripheral blood mononuclear cell (PBMC) samples
used in this study were from patients who were participating in the
Primo-SHM cohort at the Academic Medical Center (AMC) in
Amsterdam, the Netherlands (n=23) and patients who are in
follow-up at the Aids Reference Center of Ghent University
Hospital (n=11). Samples were collected from patients receiving
ART with undetectable viral loads (,50 copies/ml) (n=14 in
Amsterdam cohort and n=7 in Ghent cohort), and from therapy-
naı ¨ve patients (n=9 in Amsterdam cohort and n=4 in Ghent
cohort). The majority of patient samples (28 out of 34) were
derived from patients infected with HIV-1 subtype B, two samples
were subtype CRF01_AE, one was subtype CRFO2_AG, and for
three samples the subtype was unknown (Table S1). Ethical
approval was obtained from Ethics Committees of the University
Hospital Ghent and of the AMC. All participants had provided
written informed consent.
Nucleic Acid Isolation, DNase Treatment and cDNA
Cell-associated HIV-1 RNA from patient samples of Ghent
University Cohort was extracted from 56106PBMCs using
TRIzolH Reagent (AmbionH) and eluted in 20 ml nuclease-free
water (AmbionH) as previously described . RNA purity and
integrity was assessed using automated electrophoresis system
(ExperionTM, Bio-Rad) . Total cell-associated nucleic acids
from patient samples of the Amsterdam cohort were extracted
from 2.5–56106PBMCs according to the isolation method of
Boom et al. and eluted in 50 ml nuclease-free water (AmbionH)
. 12 ml of the eluted RNA samples were first subjected to
DNase treatment (DNA-freeTMkit, AmbionH), to remove HIV-1
DNA which could interfere with the quantification, and subse-
quently added to the reverse transcription (RT) mix. RT was
performed in the total volume of 20 ml reaction and contained 200
units of SuperScriptTMIII reverse transcriptase, 20 units of
RNaseOUTTMribonuclease inhibitor, 150 ng of random primers,
and 20 nmoles of deoxynucleoside triphosphates (all- Invitrogen) at
42uC for 60 min, followed by heat inactivation of the reverse
transcriptase for 10 min at 70uC. Patient-derived cDNA prepara-
tions were used for the usRNA and the msRNA assays by ddPCR
and seminested qPCR. For all samples, same amounts (4 ml) of the
same cDNA preparations (hereafter referred to as ‘‘input unit’’)
were always used for both ddPCR and qPCR, except for 11
patient samples with limited amounts of material, where 1 ml of
cDNA template was used for the seminested qPCR and the results
were normalized to 4 ml for the purpose of subsequent compar-
isons. Samples were tested in single replicate, because of the
limited availability of patient samples (Graph S1).
For both usRNA and msRNA assays and for both ddPCR and
qPCR methods, no-template controls (NTCs) with water were
included in every run. To assess possible false positive droplets for
the ddPCR run, a total of 42 NTCs were assessed. From these, 21
NTCs were assessed for the usRNA assay and 21 for the msRNA
assay. To discern possible PCR contamination from system
artefacts, eight NTCs per assay were prepared with an amplifi-
cation-deficient ddPCR mix, which contained only one primer
(only forward or only reverse) and a probe. These eight wells were
surrounded with other NTCs. Additionally, to assess the possibility
of cross-contamination between wells in the droplet read-out, 4
positive controls were inserted between the NTC wells. The read-
out of the dPCR proceeds in a sequential manner, hence the 4
positive controls were placed in the wells just before the NTCs.
Primers and Probes
For both PCR methods, previously described primer and probe
sets for usRNA and msRNA were used [5,14]. A table that lists all
primers and probes is provided as supplementary data (Table S2).
HIV-1 usRNA was quantified using the GAG1, GAG2, and
SK431 primers that amplify a region within the HIV-1 gag, and
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org2 January 2014 | Volume 9 | Issue 1 | e85999
the GAG3 hydrolysis probe was used. Spliced HIV-1 msRNA was
quantified using the ks1, mf84, and mf83 primers that amplify a
region containing the tat/rev exon-exon junction, and the ks2-tq
Droplet Digital PCR
HIV-1 RNA was quantified using the QX100TMDroplet
DigitalTMPCR system (Bio-Rad, Pleasanton, CA). The ddPCR
mix for the usRNA assay consisted of: 10 ml 2x ddPCRTMsuper
mix for probes (Bio-Rad); 200 nM of GAG1 and GAG2 primers;
400 nM GAG3 probe mix and 4 ml of the cDNA into a final
volume of 20 ml. The total mix was placed into the 8 channel
cartridge, 50 ml of droplet generating oil was added and droplets
were formed in the QX100TMdroplet generator (Bio-Rad).
Droplet in oil suspensions were transferred to an EppendorfH 96
well plate (Eppendorf, Germany) and placed into the T100TM
Thermal Cycler (Bio-Rad). Cycling conditions were as follows:
95uC for 5 min, followed by 40 cycles of 95uC for 15 sec and 58uC
for 60 sec. The ddPCR mix for the msRNA assay consisted of
10 ml 2x ddPCRTMsuper mix for probes (Bio-Rad); 250 nM of
ks1 and mf83 primers, 500 nM of ks2-tq, and 4 ml of the cDNA
into a final volume of 20 ml. PCR cycling conditions were the same
as for the usRNA assay, except the annealing temperature which
was 60uC. The droplets were subsequently read automatically by
the QX100TMdroplet reader (Bio-Rad) and the data was analyzed
with the QuantaSoftTManalysis software 18.104.22.168 (Bio-Rad).
Seminested Real Time PCR
For the seminested qPCR, two rounds of PCR amplification
were performed for usRNA and msRNA assays. For the usRNA
assay, the first round of the PCR was performed on a conventional
PCR machine (GeneAmpH PCR System 9700; Applied Biosys-
temsH) in 25 ml of PCR mix, which contained 4 ml of cDNA
template, 20 mM Tris (pH 8.3), 50 mM KCl, 2 mM MgCl2,
0.4 mM of deoxynucleoside triphosphates, 1 U of AmpliTaqH
(Applied BiosystemsH), and 50 ng each of GAG1 and SK431
primers. The PCR cycling settings were: 94uC for 3 min, followed
by 15 cycles of 94uC for 30 s, 55uC for 30 s, and 72uC for 1 min.
The product of the first PCR was used as a template in the second,
seminested qPCR amplification, performed on the ABI PrismH
7000 qPCR machine (Applied BiosystemsH) using TaqManH
detection chemistry. Two microliters of the first PCR product
were diluted to 50 ml with PCR mix containing 25 ml of
2*PlatinumH Tag qPCR mix (Invitrogen), 1 ml of ROX reference
dye (Invitrogen), 1 mM MgCl2, 0.2 mM of each of primers (GAG1
and GAG2), and 0.2 mM dual hybridization probe GAG3. Real-
time PCR cycling settings were: 50uC for 2 min, 95uC for 10 min,
45 cycles of 95uC for 15 s and 60uC for 1 min. For the msRNA
assay, the same protocol was used. The first PCR was performed
with the primer pair ks1 and mf83, which amplifies the msRNA
species encoding the Tat and Rev proteins. Subsequently, the
seminested qPCR of the msRNA assay was performed with the
primers mf84 and mf83 and the fluorescent hydrolysis probe ks2-
tq. Cycling conditions were the same, except that 50 amplification
cycles were done instead of 45 in the second PCR.
Preparation of Standard Curves
As external standards, synthetic runoff RNA transcripts,
corresponding to the HIV gag and tat/rev regions, were used
. The concentrations of RNA standards were determined
spectrophotometrically and recalculated to RNA copies/ml.
Master stocks of the standards were frozen in aliquots at 280uC
until use. Duplicate standard curves for each assay were made
from separate master stocks from which serial dilutions were made
to obtain a 7-point standard curve. Serial dilutions of standards for
usRNA and msRNA assays were quantified using the ddPCR and
the seminested qPCR technique.
Conversion of the Raw Data
The raw quantitative output of ddPCR was the cDNA copy
number in the input sample, whereas qPCR provided the Cq
value which is based on the fluorescent amplification curve. For
patient samples, the raw outputs of both methods were converted
to the RNA copy numbers using the standard curves as conversion
factors (Table 1, Fig. 1) . The quantified HIV RNA copy
numbers were log-transformed. The final output measures, for
patient samples, were the log10RNA copy numbers per input unit
(4 ml of the input cDNA) for both ddPCR and qPCR (Graph S1).
Statistical analysis was performed using GraphPad Prism
software 5.01 (http://www.graphpad.com). Linear regression
was used to analyze the standard curves. Pearson correlation
analysis and Bland-Altman tests were used to assess the
quantitative agreement between ddPCR and seminested qPCR
measurements in patient samples. For these analyses, undetectable
values were censored to one copy. Fisher’s exact tests were used to
compare the detectability of HIV RNA in patient samples between
Detection of HIV-1 RNA in Standards
Serially diluted usRNA and msRNA standards were measured
in duplicate for both ddPCR and seminested qPCR methods.
Table 1 shows the raw quantification output of serially diluted
synthetic RNA standards by ddPCR (cDNA copy numbers) and
seminested qPCR (Cq values). The quantitative accuracy was
higher for seminested qPCR as compared with ddPCR, as the
coefficients of variation (CV) were smaller for seminested PCR for
almost all standard dilutions, especially in the lower ranges of
quantification. The detectability of higher copy numbers of
standards was the same between the two methods, but for the
lower concentrations, seminested qPCR showed better sensitivity
than ddPCR for both usRNA and msRNA standards (Table 1).
Quantitative linearity of both methods was assessed by
quantification of standards for usRNA and msRNA. UsRNA
standard curve showed good efficiency (E) and linearity i.e.
E=90.7%, R2=0.96 by ddPCR and E=89.2%, R2=0.99 by
seminested qPCR, respectively (Fig. 1A and 1B). The correlation
between ddPCR and seminested qPCR on standards had a
coefficient of determination R2of 0.97 (Fig. 1C). Linearity of
standard curve for msRNA was better in seminested qPCR:
R2=0.93 in ddPCR and R2=0.998 in seminested qPCR.
However, PCR efficiency was better in ddPCR (90.0%) than in
seminested qPCR (61.9%) (Fig. 1D and 1E). The correlation
between the two methods had a coefficient of determination R2of
0.92 (Fig. 1F).
Detection of HIV-1 RNA in Patient Samples
Thirty-four clinical samples were evaluated, with 21 samples
from patients on ART with undetectable viral load (,50 copies/
ml) and 13 samples from therapy-naı ¨ve patients. UsRNA and
msRNA quantification was performed with both methods.
UsRNA was quantified in all 34 patient samples and msRNA
was quantified in 23 samples (15 from patients on ART and 8 from
therapy naı ¨ve patients).
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org3January 2014 | Volume 9 | Issue 1 | e85999
The detectability of usRNA in patient samples was equally high
for both methods: ddPCR and seminested qPCR detected usRNA
in 31 out of 34 samples (91%) (Table S1). From therapy-naı ¨ve
patients, both methods detected usRNA in 12 out of 13 samples
(92%). From patients on ART, usRNA was detected in 19 out of
21 samples (90%) by both methods. MsRNA was detected more
frequently with the ddPCR (14 positive samples out of 23, 61%)
than seminested qPCR (9 positive samples out of 23, 39%)
(p=0.24). This difference is attributable to samples from patients
on ART: whereas the detectability of msRNA in therapy-naı ¨ve
patients was equal between methods (6 out of 8 samples (75%)
were positive by both methods), msRNA was detected in 8 out of
Figure 1. Quantification of serially diluted synthetic RNA standards by ddPCR and seminested qPCR. Panels (A) – (C) show usRNA data,
and panels (D) – (F) show msRNA data. (A, D) Quantification of standards by ddPCR. The log10-transfomed RNA copy numbers of serially diluted
synthetic RNA standards were plotted against the corresponding log10-transformed cDNA copy numbers determined by ddPCR and fitted with a
linear regression model. (B, E) Quantification of standards by seminested qPCR. The log10-transfomed RNA copy numbers of serially diluted synthetic
RNA standards were plotted against the corresponding quantification cycle (Cq) values of seminested qPCR on a semi-log scale and fitted with a
linear regression model. (C, F) Pearson correlation between ddPCR and seminested qPCR output values for the serially diluted standards. The log10-
transformed cDNA copy numbers determined by ddPCR were plotted against the corresponding quantification cycle (Cq) values of seminested qPCR
on a semi-log scale. For every dilution, an average value of two independent measurements is shown.
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org4 January 2014 | Volume 9 | Issue 1 | e85999
15 samples from ART-treated patients (53%) by ddPCR and in 3
out of 15 such samples (20%) by seminested qPCR (p=0.13).
Detection of usRNA and msRNA with both methods in patient
samples on and off ART is listed in the supplementary table S1.
Analysis of patient samples by ddPCR and seminested qPCR
revealed a correlation between the two methods (R2=0.51 for
usRNA and R2=0.87 for msRNA; Fig. 2A and 2B). For usRNA,
the mean difference (bias) between the RNA copy numbers
generated with seminested qPCR and ddPCR, assessed with
Bland-Altman test, was 0.0560.75 log10(95% Limits of Agree-
ment, 21.43 to 1.53 log10) and a corresponding bias for msRNA
was 20.9460.36 log10 (95% Limits of Agreement, 21.64 to
20.24 log10; Fig. 2C and 2D).
No-template controls (NTCs) were used in all assays for both
methods. In the seminested qPCR protocol, all NTC remained
negative in the usRNA and msRNA assays. However, in ddPCR,
positive events of 0.16 copies/reaction (2 positive droplets) and
0.22 copies/reaction (3 positive droplets) were detected in 1 well
out of 3 for the usRNA and msRNA assay, respectively. The
positive NTC in the usRNA assay had 3 droplets with similar
fluorescence range as for the patient samples (Fig. S1). The positive
NTC in msRNA had 2 droplets with higher fluorescence than the
patient samples (Fig. S2).
To better understand the nature of false-positive events that
were observed, we assessed another 42 NTC replicates. From
these NTCs, 1 or 2 positive droplets were registered in 9 wells out
of the total 42. From these, only 1 well with 1 positive droplet
originated from the usRNA assay, and the remaining 8 wells were
from the msRNA assay. Two wells from the msRNA assay had 2
positive droplets detected and in the remaining 6 wells, only 1
positive droplet was registered. Interestingly, the reactions in the
wells with 2 positive droplets and in 1 well with 1 positive droplet
were prepared with the amplification-deficient ddPCR mix. The 4
out of 42 NTCs that were placed after a positive control with high
input of amplicons were all negative.
In the present study, synthetic HIV RNA standards and CA
HIV RNA in patient-derived samples were quantified with
seminested qPCR and ddPCR. To the best of our knowledge,
this is the first report of HIV RNA measurement by ddPCR in
patient-derived samples. In the first part of the study, synthetic
HIV RNA standards were measured by both seminested qPCR
and ddPCR. Subsequently, in the second part, patient-derived
samples were quantified with both methods.
The correlation of measurements between seminested qPCR
and ddPCR was good for both standard curves (Figs. 1C, 1F).
However, in absolute numbers, the cDNA copy numbers
quantified by ddPCR were lower than the corresponding RNA
copy numbers assessed by UV spectrophotometry. One explana-
tion for this is the suboptimal efficiency of RT, in which not all
RNA molecules are reverse transcribed into cDNA. The efficiency
of RT (cDNA yield) has been shown to vary widely, depending on
the enzyme used and the priming strategy [24,25]. Another
possible explanation for the discrepancy between the RNA and
cDNA copy numbers is molecular dropout , a recently
described dPCR phenomenon, in which the target molecule is
present in the partition but fails to amplify. Because we directly
used aliquots of RT reactions as input material for ddPCR, the
PCR step could have been inhibited by RT components . To
correct for the RT efficiency and PCR inhibition, dPCR
quantification of RNA should be performed in combination with
a calibrator that would provide a conversion factor for the raw
dPCR data to RNA copy numbers . In our study, the raw
data-to-RNA conversion factors were calculated based on the
standard curves. These conversion factors were used for the
patient samples to convert the Cq to RNA copy numbers for the
seminested qPCR, and the cDNA copy numbers to RNA copy
numbers for the ddPCR.
The second part of the study consisted of quantification of
patient-derived material with both seminested qPCR and ddPCR.
Thus far, there are no clinically validated methods for measure-
ment of CA HIV RNA. However, the seminested qPCR method
has been validated and compared with Cobas Amplicor HIV-1,
the clinical assay for plasma HIV RNA quantification .
Moreover, the seminested qPCR method has been extensively
used for CA HIV RNA quantification in patient-derived material
[4,6,14,15]. Hence, we compared seminested qPCR and ddPCR
for CA HIV RNA measurements in patient samples.
For usRNA quantification in patient samples, the difference
between seminested qPCR and ddPCR was 0.0560.75 log10
Table 1. Quantification of standards for usRNA and msRNA with ddPCR and seminested qPCR.
ddPCRSeminested qPCR ddPCR Seminested qPCR
1ean 6 SD[i]
Mean 6 SD
Mean 6 SD
Mean 6 SD
Copy nr. of
5.0360.030.7 14.7160.49 3.3 4.9160.000.1 17.1260.01 0.0
3.5560.29 8.3 17.5360.211.2 3.5960.12 3.322.0160.190.9
2.5160.041.5 21.3760.602.82.4960.21 8.526.7460.150.6
1.3960.07 5.225.3460.21 0.81.3660.3425.1 31.9560.37 1.2
0.9860.032.6 27.9760.010.00.3460.2057.934.4460.73 2.1
0.5560.5090.530.4560.17 0.60.21n/a38.5160.15 0.4
0.09n/a32.5062.05 6.30.65 n/a41.30 n/a
[i] SD, standard deviation.
[ii] CV, coefficient of variation.
[iii] Cq, quantification cycle.
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org5January 2014 | Volume 9 | Issue 1 | e85999
(Fig. 2C). On average, it is less than the accepted threshold of
clinically significant variability (0.5 log10). However, the standard
deviation was relatively large and the linearity of the correlation
was only R2=0.51 (Fig. 2A). The suboptimal correlation could be
due to the fact that the majority of samples tested were derived
from well suppressed patients on ART, in whom usRNA levels are
very low. It is well known that in samples with very low copy
numbers, random variation due to sampling error (Poisson’s error)
becomes significant [28,29]. This indicates that the comparison of
methods on samples from patients on suppressive ART is difficult.
Consequently, a second assessment was made, comparing only
those patient samples with more than 100 copies/input unit
detected with both methods. This resulted in an improved
correlation, R2=0.87 (data not shown) supporting the hypothesis
that the mediocre correlation in the usRNA assay was primarily
due to sampling variation in samples with low HIV-1 loads.
For msRNA quantification in patient samples, we observed a
higher difference between the measurements by seminested qPCR
and ddPCR (–0.9460.36 log10), meaning that msRNA values
measured by seminested qPCR were lower than the corresponding
ddPCR measurements by an average factor of 8.7. The
underestimation of measurements with seminested qPCR could
be due to primer-template or probe-template mismatches. Such
mismatches have a direct effect on qPCR-based quantification
, but dPCR is less susceptible to these effects . This is
supported by a recent comparison of the effects of mismatches on
quantifying HIV DNA with qPCR and ddPCR .
While the detectability of usRNA in patients on ART and in
therapy-naı ¨ve patients was equally high (90–92%) with both
methods, msRNA was detected with ddPCR in a higher
proportion of patients on ART compared to the seminested
qPCR. However, this difference did not achieve statistical
significance, possibly due to small patient numbers. Moreover,
we cannot conclude that this effect is due to higher sensitivity of
the ddPCR, because samples containing single positive droplets in
ddPCR may have been false positives due to the observed false
positive reactions in the ddPCR NTCs. The detectability of
msRNA in therapy-naı ¨ve patients with higher msRNA loads was
equal between the methods.
The major limitation of this study is the positive signals obtained
in the NTCs in the ddPCR experiments for both usRNA and
Figure 2. Quantification of usRNA and msRNA in patient samples. (A, B) Correlations between ddPCR and seminested qPCR measurements
of usRNA (A) and msRNA (B) in patient samples are shown. The units of measurement are log10copies RNA per input unit (4 ml of input cDNA) for
both ddPCR and qPCR. Samples that were undetectable with both methods (n=1 for usRNA and n=8 for msRNA) are not shown. (C, D) Bland-Altman
plots comparing the ddPCR and seminested qPCR measurements of usRNA (C) and msRNA (D) in patient samples. Mean differences and 95% Limits
of Agreement are shown on the graphs.
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org6 January 2014 | Volume 9 | Issue 1 | e85999
msRNA assays. Although the majority of NTCs were negative,
several NTCs were recorded with #3 positive events. The false
positive signals may have contributed to the higher detectability of
msRNA in patients on ART by ddPCR. The current ddPCR
technique does not allow sequencing of the samples in order to
establish whether the false positives represent artefacts. The
problem of false positives could be alleviated by setting up a limit
of detection for ddPCR, which would correspond to the maximal
number of positive droplets per NTC well (3 in this study). In this
case, some of the samples of patients on ART that were positive for
usRNA, would not be scored as positives, as they yielded #3
positive droplets. Likewise, all samples from patients on cART that
were positive for msRNA would not be scored as positive because
they yielded #2 positive droplets (Table S3).
Our study is not the first occasion on which false positive events
in NTCs were reported in ddPCR experiments. Previously, two
independent groups reported positive signals in NTCs, and Strain
et al. reported on average of 0.1 to 0.4 false positive events per
NTC well after analyzing more than 500 NTCs [16,31]. These
false positive events are detected randomly, they are not assay-
dependent, and they have different fluorescence height (Fig. S1
and S2). Sometimes we observed false positive events with
extremely high fluorescence compared to the real positives,
suggesting that these are artefacts (Fig. S2). This is supported by
the experiments on the NTCs which showed that false positives
also occurred in reactions where lab contamination can be
excluded. In addition, these experiments revealed that possible
carry-over during sample processing and read-out is also unlikely.
At the moment, the false negative events, appearing in experi-
ments with ddPCR, preclude its wider use for quantification of
extremely low viral loads, and this problem needs to be further
Recent interest in ddPCR as a method of nucleic acid
quantification largely stems from the fact that ddPCR is a direct
method that does not rely on an external standard curve, as qPCR
does. However, although ddPCR does provide absolute quantifi-
cation of target DNA (or cDNA), it is important to realize that, at
this point, its application to absolute quantification of RNA is still
under development. When using the two-step RT-dPCR method,
where RNA is reverse transcribed to cDNA before sample
partitioning, the quantified absolute cDNA copy number has to
be back-converted to the RNA copy number. In this study, this
cDNA-to-RNA conversion was performed based on the standard
curve, which enabled the direct comparison of RNA copy
numbers in patient samples between the two methods, but made
the ddPCR quantification of RNA as relative on the standard
curve as the seminested qPCR was. The use of pre-validated
calibrators will facilitate higher accuracy of RNA quantification
with ddPCR. An alternative is to use one step RT-dPCR methods,
in which an RNA sample is partitioned prior to RT. However,
accurate calibrators will likely be needed even in this case, because
the efficiency of RNA quantification by dPCR was recently shown
to be assay- and transcript-dependent .
Further exploration of the use of ddPCR for accurate CA HIV
RNA measurement is necessary. Quantitative assays for CA HIV
RNA have the potential to improve the monitoring of patients on
ART and to be used in clinical studies aimed at HIV eradication,
but should be cross-validated by multiple laboratories prior to
msRNA assays in ddPCR. DdPCR droplet read-out for the
usRNA assay. Last three columns (H03, H04 and H05) show the
NTC’s. In column H04 three positive droplets are registered. The
other two columns are the other 2 NTC, which are negative. The
readout from A01 until G03 is for patient samples.
Negative template controls for us- and
assay. The last column (H05) shows the NTC with two positive
droplets registered. H03 and H04 columns are the other two
NTC, which are negative. The readout from A01 until G03 is for
DdPCR droplet read-out for the msRNA
usRNA and msRNA with ddPCR and seminested qPCR
Patient samples information; detection of
Primers and probes used in this study.
usRNA and msRNA assays in patient samples.
Raw data from the ddPCR experiments for
msRNA in clinical samples on ddPCR and seminested
Workflow used to measure usRNA and
We thank the study participants for their involvement.
Conceived and designed the experiments: MK AOP WDS LV. Performed
the experiments: MK AOP. Analyzed the data: MK AOP WDS.
Contributed reagents/materials/analysis tools: MK AOP WDS. Wrote
the paper: MK AOP WDS. Critically reviewed the final version: MK AOP
WDS DV BB LV. Read and approved the manuscript: MK AOP WDS
DV BB LV.
1. Doyle T, Geretti AM (2012) Low-level viraemia on HAART: significance and
management. Curr Opin Infect Dis 25: 17–25.
2. Palmer S, Maldarelli F, Wiegand A, Bernstein B, Hanna GJ, et al. (2008) Low-
level viremia persists for at least 7 years in patients on suppressive antiretroviral
therapy. Proc Natl Acad Sci U S A 105: 3879–3884.
3. Lewin SR, Rouzioux C (2011) HIV cure and eradication: how will we get from
the laboratory to effective clinical trials? AIDS 25: 885–897.
4. Pasternak AO, Jurriaans S, Bakker M, Prins JM, Berkhout B, et al. (2009)
Cellular Levels of HIV Unspliced RNA from Patients on Combination
Antiretroviral Therapy with Undetectable Plasma Viremia Predict the Therapy
Outcome. PLoS One 4: e8490.
5. Fischer M, Joos B, Hirschel B, Bleiber G, Weber R, et al. (2004) Cellular viral
rebound after cessation of potent antiretroviral therapy predicted by levels of
multiply spliced HIV-1 RNA encoding nef. J Infect Dis 190: 1979–1988.
6. Pasternak AO, de Bruin M, Jurriaans S, Bakker M, Berkhout B, et al. (2012)
Modest nonadherence to antiretroviral therapy promotes residual HIV-1
replication in the absence of virological rebound in plasma. J Infect Dis 206:
7. Elliott J, Solomon A, Wightman F, Smith A, Palmer S, et al. (2013) The Safety
and Effect of Multiple Doses of Vorinostat on HIV Transcription in HIV+
Patients Receiving cART. 20th Conference on Retroviruses and Opportunistic
Infections Atlanta, Georgia.
8. Archin NM, Liberty AL, Kashuba AD, Choudhary SK, Kuruc JD, et al. (2012)
Administration of vorinostat disrupts HIV-1 latency in patients on antiretroviral
therapy. Nature 487: 482–485.
9. Pasternak AO, Lukashov VV, Berkhout B (2013) Cell-associated HIV RNA: a
dynamic biomarker of viral persistence. Retrovirology 10: 41.
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org7 January 2014 | Volume 9 | Issue 1 | e85999
10. Lewin SR, Vesanen M, Kostrikis L, Hurley A, Duran M, et al. (1999) Use of Download full-text
real-time PCR and molecular beacons to detect virus replication in human
immunodeficiency virus type 1 infected individuals on prolonged effective
antiretroviral therapy. J Virol 73: 6099–6103.
11. Bernardo V, Ribeiro Pinto LF, Albano RM (2013) Gene expression analysis by
real-time PCR: experimental demonstration of PCR detection limits. Anal
Biochem 432: 131–133.
12. White RA, 3rd, Blainey PC, Fan HC, Quake SR (2009) Digital PCR provides
sensitive and absolute calibration for high throughput sequencing. BMC
genomics 10: 116.
13. White RA, 3rd, Quake SR, Curr K (2012) Digital PCR provides absolute
quantitation of viral load for an occult RNA virus. J Virol Methods 179: 45–50.
14. Pasternak AO, Adema W, Bakker M, Jurriaans S, Berkhout B, et al. (2008)
Highly sensitive methods based on seminested real-time reverse transcription-
PCR for quantitation of human immunodeficiency virus type 1 unspliced and
multiply spliced RNA and proviral DNA. J Clin Microbiol 46: 2206–2211.
15. Pasternak AO, Jurriaans S, Bakker M, Berkhout B, Lukashov VV (2010) Steady
increase in cellular HIV-1 load during the asymptomatic phase of untreated
infection despite stable plasma viremia. AIDS 24: 1641–1649.
16. Strain MC, Lada SM, Luong T, Rought SE, Gianella S, et al. (2013) Highly
Precise Measurements of HIV DNA by Droplet Digital PCR. PLoS One 8:
17. Hindson BJ, Ness KD, Masquelier DA, Belgrader P, Heredia NJ, et al. (2011)
High-Throughput Droplet Digital PCR System for Absolute Quantitation of
DNA Copy Number. Anal Chem 83: 8604–8610.
18. Shen F, Sun B, Kreutz JE, Davydova EK, Du WB, et al. (2011) Multiplexed
Quantification of Nucleic Acids with Large Dynamic Range Using Multivolume
Digital RT-PCR on a Rotational SlipChip Tested with HIV and Hepatitis C
Viral Load. J Am Chem Soc 133: 17705–17712.
19. Whale AS, Huggett JF, Cowen S, Speirs V, Shaw J, et al. (2012) Comparison of
microfluidic digital PCR and conventional quantitative PCR for measuring copy
number variation. Nucleic Acids Res 40: e82.
20. Henrich TJ, Gallien S, Li JZ, Pereyra F, Kuritzkes DR (2012) Low-level
detection and quantitation of cellular HIV-1 DNA and 2-LTR circles using
droplet digital PCR. J Virol Methods 186: 68–72.
21. Hayden RT, Gu Z, Ingersoll J, Abdul-Ali D, Shi L, et al. (2013) Comparison of
droplet digital PCR to real-time PCR for quantitative detection of cytomega-
lovirus. J Clin Microbiol 51: 540–546.
22. Messiaen P, De Spiegelaere W, Alcami J, Vervisch K, Van Acker P, et al. (2012)
Characterization of LEDGF/p75 genetic variants and association with HIV-1
disease progression. PLoS One 7: e50204.
23. Boom R, Sol CJ, Salimans MM, Jansen CL, Wertheim-van Dillen PM, et al.
(1990) Rapid and simple method for purification of nucleic acids. J Clin
Microbiol 28: 495–503.
24. Stahlberg A, Kubista M, Pfaffl M (2004) Comparison of reverse transcriptases in
gene expression analysis. Clin Chem 50: 1678–1680.
25. Stangegaard M, Dufva IH, Dufva M (2006) Reverse transcription using random
pentadecamer primers increases yield and quality of resulting cDNA.
BioTechniques 40: 649–657.
26. Whale AS, Cowen S, Foy CA, Huggett JF (2013) Methods for applying accurate
digital PCR analysis on low copy DNA samples. PLoS One 8: e58177.
27. Sanders R, Mason DJ, Foy CA, Huggett JF (2013) Evaluation of digital PCR for
absolute RNA quantification. PLoS One 8: e75296.
28. Bustin SA, Nolan T (2004) Pitfalls of quantitative real-time reverse-transcription
polymerase chain reaction. Journal of biomolecular techniques 15: 155–166.
29. Picard C, Silvy M, Gabert J (2006) Overview of real-time RT-PCR strategies for
quantification of gene rearrangements in the myeloid malignancies. Methods in
molecular medicine 125: 27–68.
30. Lefever S, Pattyn F, Hellemans J, Vandesompele J (2013) Single-nucleotide
polymorphisms and other mismatches reduce performance of quantitative PCR
assays. Clin Chem 59: 1470–1480.
31. Pinheiro LB, Coleman VA, Hindson CM, Herrmann J, Hindson BJ, et al. (2012)
Evaluation of a Droplet Digital Polymerase Chain Reaction Format for DNA
Copy Number Quantification. Anal Chem 84: 1003–1011.
ddPCR & Seminested qPCR for HIV RNA Quantification
PLOS ONE | www.plosone.org8 January 2014 | Volume 9 | Issue 1 | e85999