Journal of Virological Methods 188 (2013) 153–160
View metadata, citation and similar papers at core.ac.uk brought to you by CORE
Contents lists available at SciVerse ScienceDirect provided by Elsevier - Publisher Connector
Journal of Virological Methods
jou rnal homepage: www.elsevier.com/locate/jviromet
Simple diffusion-constrained immunoassay for p24 protein with the sensitivity
of nucleic acid amplification for detecting acute HIV infection
Lei Chang, Linan Song, David R. Fournier, Cheuk W. Kan, Purvish P. Patel, Evan P. Ferrell, Brian A. Pink,
∗
Kaitlin A. Minnehan, David W. Hanlon, David C. Duffy, David H. Wilson
Quanterix Corp, 113 Hartwell Ave, Lexington, MA 02421, USA
a b s t r a c t
Article history: Nucleic acid amplification techniques have become the mainstay for ultimate sensitivity for detecting
Received 8 May 2012
low levels of virus, including human immunodeficiency virus (HIV). As a sophisticated technology with
Received in revised form 22 August 2012
relative expensive reagents and instrumentation, adoption of nucleic acid testing (NAT) can be cost
Accepted 29 August 2012
inhibited in settings in which access to extreme sensitivity could be clinically advantageous for detection
Available online 2 October 2012
of acute infection. A simple low cost digital immunoassay was developed for the p24 capsid protein of HIV
based on trapping enzyme-labeled immunocomplexes in high-density arrays of femtoliter microwells
Keywords:
and constraining the diffusion of the enzyme–substrate reaction. The digital immunoassay was evalu-
Digital ELISA
Immunoassay ated for analytical sensitivity for HIV capsid protein p24, and compared with commercially available NAT
methods and immunoassays for p24, including 4th-generation antibody/antigen combo assays, for early
Single molecule array
p24 detection of HIV in infected individuals. The digital immunoassay was found to exhibit 2000–3000-fold
HIV greater analytical sensitivity than conventional immunoassays reactive for p24, and comparable sensitiv-
ity to NAT methods. Assaying serial samples from 10 HIV-infected individuals, the digital immunoassay
detected acute HIV infection as early as NAT methods, and 7–10 days earlier than conventional immunoas-
says. Comparison of assay results between the digital immunoassay and a quantitative NAT method
2
from HIV infected serum exhibited a linear correlation R > 0.99. The data indicate that by constraining
diffusion of the signal generation step of a simple sandwich immunoassay and enabling the digital count-
ing of immunocomplexes, dramatic improvements in sensitivity to virus can be obtained to match the
sensitivity of NAT at a fraction of the cost.
© 2012 Elsevier B.V. Open access under CC BY-NC-ND license.
1. Introduction sensitivity detection of acute HIV infection. NAT has significantly
shortened the window between initial infection and its detection
Since the development of the polymerase chain reaction in through amplification of viral nucleic acid rather than detecting
the 1980s (Mullis et al., 1986), amplification of specific nucleic the presence of antibody following seroconversion (Busch, 2007).
acid sequences for genetic identification has become an indis- Although NAT has become the mainstay for detection of viruses,
pensible tool in medical research and diagnostics, including the the technology is relatively complex and high cost (Westreich et al.,
detection and diagnosis of infectious disease. With its imple- 2008), and can be cost inhibitory or prohibitive in settings where
mentation for blood screening in the US in 1999 (Busch et al., access to high sensitivity could be clinically advantageous. These
2000), nucleic acid amplification testing (NAT) for detection of scenarios include blood donor screening in lower-resource settings
viral pathogens has helped safeguard blood for transfusion by and clinical screening in higher incidence areas, where a significant
providing the most sensitive, economically feasible detection of number of cases of acute infection can be missed by less sensi-
infected blood donations possible. For HIV detection, NAT meth- tive immunoassay tests. Rapid detection and reporting of acute HIV
ods in use for blood screening are capable of analytical sensitivities infection represents a key opportunity to control the march of the
of approximately 60 HIV RNA copies/mL (30 viruses/mL) with indi- disease because viral transmission is 10 times more likely during
vidual donors (Assal et al., 2009; Proceix ULTRIO Product Insert, the acute phase than in the chronic phase (Wawer et al., 2005).
2011). In the clinical setting, NAT is the gold standard for high Immunoassays, on the other hand, are simpler and lower
cost than NAT, and can be deployed in more diverse environ-
ments. Conveniently, nature provides its own amplification of
∗ protein molecules from each virus in the form of antibodies to the
Corresponding author.
E-mail address: [email protected] (D.H. Wilson). virus and the viral proteins. Immunoassays to HIV antibodies are
0166-0934 © 2012 Elsevier B.V. Open access under CC BY-NC-ND license. http://dx.doi.org/10.1016/j.jviromet.2012.08.017
154 L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160
straightforward, but they require the immune response and are genetically engineered Saccharomyces cerevisiae. One amino acid
unable to detect acute infection when antibodies are not yet (met) was inserted at the N-terminus to enable expression.
present. 3rd generation antibody assays (Constantine et al., 1994)
detect the earliest stage of the immune response (immunoglobu- 2.2. Cultured virus
lin M), reducing the serological window to approximately 3 weeks
(Fiebig et al., 2003). Viral proteins are present at the start of infec- The cultured HIV-1 Group M, Subtype B virus was purchased
tion, and each HIV particle produces approximately 2000 copies from Seracare Life Sciences (Part number PN242B, lot 9899P-10,
9 ®
of p24 capsid protein (NIAID, 2011), which works in favor of 1.15 × 10 HIV-1 RNA/mL by Roche AMPLICOR HIV-1 MonitorTM
immunoassay detection of acute infection. 30 viruses/mL equates Assay, v. 1.5) and diluted in delipidated, defibrinated human
to approximately 60,000 p24 molecules per mL, or a concentration serum (Seracare PN: HS-210). This virus is harvested from the
of 3 fg/mL. Unfortunately, current, state-of-the-art conventional culture supernatant without additional purification; therefore the
immunoassays are not sufficiently sensitive to detect these low standard could contain free p24 in addition to virion-incorporated
7
numbers. Immunoassays for p24 (Kontio, 1991) and 4th genera- p24. The diluted virus, at 1 × 10 copies HIV-1 RNA/mL, was inactiv-
tion combo assays that combine reactivity for p24 and anti-HIV ated by the addition of Triton X-100 to 0.4% and incubation of the
◦
antibodies (Weber et al., 1998) can only detect the presence sample for 1 h at 37 C (Ukkonen et al., 1988).
of circulating p24 antigen after it has elevated to greater than
11,000–70,000 fg/mL (Marcel et al., 2011). While this can reduce 2.3. Seroconversion panels
the detection window another 4 or 5 days relative to serology
tests, immunoassays for p24 remain thousands-fold less sensi- Sets of seroconversion serum samples from HIV infected indi-
tive for virus than HIV RNA detection, which reduces the window viduals were purchased from SeraCare (PRB956, PRB958, PRB967,
an additional week or more relative to antigen detection (Fiebig RB968, PRB969, PRB972) and from Zeptometrix (ZM-HIV9016, ZM-
et al., 2003). Thus conventional immunoassays remain blind to HIV9018, ZM-HIV9024, ZM-HIV9031).
most of the acute phase window that is detectable by NAT. A simple
immunoassay capable of the sensitivity of NAT for HIV could repre- 2.4. Digital immunoassay
sent another step forward in broader, more cost effective detection
of acute HIV infection, which would help further control the spread Digitization of immunoassay analyte detection using single
®
of the disease. molecule arrays (Simoa ) has been described (Rissin et al., 2010,
Thermodynamically, antigen–antibody interactions should 2011). In brief, it involves performing a standard paramagnetic
enable sufficient sensitivity for detection of fg/mL concentrations microbead-based enzyme-linked immunosorbent assay (ELISA),
of antigen with little amplification. The problem for unamplified singulation of individual microbeads in high-density arrays of 50 fL
methods has been acquiring sufficient signal of this interaction. microwells, and confinement within the microwells of fluorescent
Conventional immunoassays employing optical signal detection product generated by enzymes labeling the immunocomplexes.
(such as chemiluminescence) require millions to thousands of Preventing diffusion of the fluorescent product out of the wells per-
millions of signal molecule events before signal can be reliably mits rapid buildup of fluorescence that is readily visualized with a
discerned by a luminometer or fluorometer. This limitation arises CCD camera. Immunocomplexes are labeled with -galactosidase
because the ensemble signal is diluted in a typical reaction vol- (G) during the ELISA, and 30 s of conversion of enzyme substrate
ume of hundreds of microliters required for measurement. When (resorufin -d-galactopyranoside, RGP) into fluorescent product
the signal molecules are constrained from diffusing beyond a small (resorufin) by a single molecule of G provides sufficient signal
local volume, then lower numbers of signal molecules would be suf- to differentiate wells containing beads with an enzyme label from
ficient for optical detection due to their local high concentration. wells containing beads without an enzyme. At very low p24 con-
With an isolated enzyme-labeled antigen, the smaller the volume of centrations, Poisson statistics predict that bead-containing wells
confinement, the higher the local concentration of signal molecules in the array will contain either a single labeled p24 molecule or no
for detection. By confining enzyme labels within femtoliter vol- p24 molecules, resulting in a digital signal of either “active” (posi-
umes, single molecules are readily detectable by a low cost CCD tive) or “inactive” (negative) wells (Rissin and Walt, 2006). Positive
camera (Rissin et al., 2010). This simple concept is used to power a wells are counted and averaged for each array to give ‘average
standard sandwich immunoassay for p24 antigen to NAT-level sen- enzymes per bead’ (AEB), which represents the fundamental mea-
sitivity for detection of acute HIV infection. The sensitivity of the surement unit for Simoa. At higher p24 concentrations, when all
immunoassay is demonstrated by comparison to NAT and 4th gen- wells become occupied by at least one labeled p24 molecule, dig-
eration immunoassay with serial seroconversion sample sets from ital measurements transition to conventional non-digital (analog)
HIV infected individuals. The assay is standardized to quantitate measurements of total fluorescence intensity. These analog mea-
p24 protein, and correlation between p24 concentration and viral surements are converted into AEB based on the known average
RNA copies/mL by quantitative NAT is shown with HIV infected fluorescence intensity generated by single enzymes, as determined
serum. The data suggest that low cost digital immunoassay technol- by digital measurements made at low p24 concentrations. With
ogy could have important implications for HIV screening efficacy single-molecule sensitivity, concentrations of labeling reagents can
and economics. be lowered, resulting in reduced nonspecific background. This
enables high signal-to-background ratios at extremely low ana-
lyte concentrations. Because large numbers of positive wells can
2. Materials and methods be simultaneously counted on a single array, lengthy sequen-
tial bead counting is not needed for statistically powered data.
2.1. Recombinant p24 This approach makes Simoa digital immunoassays compatible
with the fast processing cadence of a high throughput automated
The recombinant p24 was obtained from Novartis Vaccines & immunoassay analyzer. The immunoassay described in this report
Diagnostics (Emoryville, CA). The HIV-1 p24 antigen is a polypep- utilizes standard low-cost reagents to demonstrate that a short
tide encoded by the p24 gag region (Steimer et al., 1986) of the diffusion-constrained signal amplification step enables thousands-
HIV-1 SF2 genome (Sanchez-Pescador et al., 1985) (231 amino fold greater sensitivity for p24 than conventional immunoassays.
acids; HIV-1 amino acid residues 139–369) and is produced in Together with nature’s amplification of 2000 p24 molecules per
L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160 155
HIV, this simple immunoassay can detect less than 100 viruses/mL,
comparable to the sensitivity of NAT for detection of acute HIV
infection.
2.5. Digital immunoassay reagents
Three reagents were developed: paramagnetic p24 cap-
ture microbeads, biotinylated detector, and a streptavidin:-
galactosidase (SG) conjugate. The capture beads were comprised
of a monoclonal anti-p24 antibody (Capricorn HIV-018-08204),
covalently attached by standard coupling chemistry to 2.7 m
carboxy paramagnetic microbeads (Varian). The antibody-coated
6
beads were diluted to a concentration of 5 × 10 beads/mL in PBS
buffer with a surfactant and BSA. Biotinylated detector reagent
was comprised of a second monoclonal anti p24 antibody (Zep-
tometrix 801136), which was biotinylated using standard methods
and diluted to a concentration of 0.2 g/mL in a PBS diluent con-

taining a surfactant and goat serum, GS (PBS/GS). S G was prepared
by covalent conjugation of purified streptavidin (Thermo Scien-
tific) and G (Sigma) using standard coupling chemistry. For assay,
Fig. 1. Dose–response of digital immunoassay over three logs of spiked p24 in
aliquots of a concentrated SG stock were diluted to 15 pM in
human plasma. Y-Axis refers to the average number of label enzymes per individ-
PBS/NCS with 1 mM MgCl2.
ual microbead (AEB) captured in each well of the array. Each data point depicts the
mean of three replicates. Standard deviations for each set of replicates are within the
circles. (a) Log-log representation of the data with 0 pg/mL calibrator omitted. (b)
2.6. Calibration Non-log depiction of the 0–0.1 pg/mL range (0 pg/mL calibrator included) showing
the low background obtained with digital quantification. Analyses of 11 calibration
curves gave a mean signal:background ratio at 0.01 pg/mL of 2.00. Statistics depict
The assay was calibrated using recombinant p24 protein diluted 2
least squares fit of all calibrators (R > 0.999). LoD was estimated as 2.8 fg/mL as
in pooled human serum (Seracare HS210). Calibrator levels were 0,
described in the text.
0.001, 0.010, 0.022, 0.100, 1.000 and 10.00 pg/mL.
2.9. Conversion between p24 units and RNA copies/mL
2.7. Immunoassay procedure
Estimates on the number of p24 units composing the HIV capsid
∼
have ranged from 1000 to 3000 (Briggs et al., 2004; Summers et al.,
Bead-sample incubations and labeling of immunocomplexes
1992). We used an assumption of 2000 p24 units/virion, per Piatak
were conducted as described (Rissin et al., 2010), except that sam-
et al. (1993) and accepted by National Institute of Allergy and Infec-
ples were assayed without pre-dilution. In brief, the prototype
tious Disease (NIAID, 2011). HIV contains 2 copies of RNA/virion.
digital p24 assay requires 100 L of sample and was performed
in three steps: analyte capture with 500,000 anti-p24 coated
microbeads for 120 min, biotinylated detector antibody binding 3. Results

for 60 min, and labeling with S G for 30 min. Assays were run in
conical 96-well ELISA plates (Axygen) using a Tecan EVO 150 for 3.1. Dose–response and limit of detection
fluidics handling and bead washing between assay steps. Serum
samples were tested with no sample pretreatment. Following assay A digital ELISA for p24 utilizing commercially available antibod-
and bead collection with a magnet, beads were loaded onto the ies and recombinant p24 for calibration was developed as described
arrays for imaging in a loading buffer comprised of PBS and 0.01% previously for other proteins (Rissin et al., 2010; Wilson et al., 2011;
Tween-20, MgCl2, and sucrose. Zetterberg et al., 2011). A representative dose–response over three
logs of spiked recombinant p24 in human serum is depicted in
Fig. 1. The assay demonstrated a highly linear response, with least
2
2.8. Array imaging squares R > 0.999. In a study of 11 calibration curves on differ-
ent days, the mean signal:background ratio (as AEB) at 0.01 pg/mL
Beads from the ELISA were loaded onto the arrays as described was 2.00 (SD 0.35). The analytical limit of detection (LoD) was esti-
(Rissin et al., 2010). Wells containing beads with labeled p24 were mated by reading the p24 concentration from the calibration curve
visualized by the hydrolysis of enzyme substrate (RGP, Invitro- at the AEB signal corresponding to 3SD above the mean signal of
gen) by G into fluorescent product. Enzyme-containing wells were zero calibrator (n = 3). LoD was calculated for each of 11 calibration
imaged by fluorescence microscope fitted with a CCD camera. The runs from triplicate measurements of the zero calibrator and the
images were analyzed to determine AEB as described (Rissin et al., lowest p24-containing calibrator (0.01 pg/mL). The mean LoD was
2011). At <70% active beads relative to total beads (low p24), the 4.87 fg/mL (SD 2.89 fg/mL) (Fig. 2). Day-to-day reproducibility of a
signal output is a count of active beads corrected for a low sta- Simoa digital ELISA has recently been described in detail (Wilson
tistical probability of multiple enzymes/bead (Rissin and Walt, et al., 2011).
2006). At >70% active beads (higher p24), the probability of multiple
enzymes/bead increases, and average fluorescence of the wells is
3.2. Cultured standardized virus vs. recombinant p24
converted to AEB based on the average intensities of wells contain-
ing single enzymes determined at lower concentrations. The AEB
Cultured HIV-1 Group M, Subtype B virus was standardized for
unit thus works continuously across the digital and analog realms
RNA copies/mL and diluted in human serum as described in Section
(Rissin et al., 2011).
2. Fig. 3 depicts a log-log representation of assay dose–response to
156 L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160
0.20 b
0.10 0.08 0.1 0.06 0.04
0.02
10 10010001x10 1x10
AEB 0.02
0.01 8 a
0.01 6
0.01 4 0.01 2
0.01
0.01 0.00 8
0.00 6
0.00 4
020 040060080010001200
1000 1x104
RNA copies/mL
Fig. 3. Log-log depiction of assay response to cultured HIV-virus in human serum,
standardized for RNA copies/mL. (a) depicts low-end data down to virus-free serum
Fig. 2. Analytical LoD estimate for the digital immunoassay. A mean LoD of in non-log space. Estimated LoD was 90 copies/mL (3.6 fg/mL p24 equivalents). (b)
4.87 fg/mL (SD 2.89) was estimated from each of 11 calibration runs performed depicts data from a separate study comparing cultured virus (open circles) and p24
on different days. For comparison, LoD ranges for p24-reactive conventional antigen (closed circles).
immunoassays are depicted (11,000–70,000 fg/mL Biomerieux VIDAS p24 II, VIDAS
DUO Ultra, Abbott AxSYM HIV Combo, ARCHITECT HIV Combo) along with the LoD
of the PROCLEIX system (60 RNA copies/mL). comparison between digital immunoassay results with a quan-
titative bDNA NAT method, an immunoassay for p24, and two
4th generation HIV combo immunoassays with four sets of late-
cultured virus, and Fig. 3a highlights the low-end data down to
emerging seroconversion samples from HIV infected individuals.
virus-free serum in non-log space. Calculated analytical sensitivity
The data shown for the comparator methods were obtained from
was equivalent to 90 RNA copies/mL (or 45 viruses/mL), consistent
the sample vendor. Red represents the point in time at which the
with estimates using spiked p24 antigen. Antigen from both sources
assays first detected the HIV infection. The digital immunoassay
exhibited good agreement using an equivalency conversion based
detected the presence of p24 protein on the same blood draws that
on 2000 p24 molecules per virus (Fig. 3b), in agreement with the
the NAT method detected the presence of amplified HIV RNA. On
stoichiometry accepted by the National Institute of Allergies and
the other hand, the conventional immunoassays reactive for p24
Infectious Disease (NIAID, 2011).
did not detect the presence of antigen until one to four blood draws
later (average of 7–9 days). With seroconversion panel 9031, the
3.3. Detection of acute HIV infection Abbott tests exhibited reactivity at draw #17 (3 draws, 15 days
later, not depicted) and the Coulter p24 assay exhibited first reac-
The sensitivity of the digital immunoassay for detection of tivity at draw #18 (4 draws, 22 days later). Noteworthy is that
acute HIV infection was examined with 10 sets of seroconversion seroconversion panel 9031 comprised more frequent blood draws,
serum samples obtained from commercial sources. Table 1 depicts a and the initial reactive draw was early in the NAT-detectable phase
Table 1
Earliest detection of HIV infection.
HIV infected donor Seroconversion Days since Digital p24 Zeptometrix data
sample designation first draw immunoassay
fg/mL Chiron bDNA Coulter p24 Abbott Architect Abbott Prism
copies RNA/mL immunoassay HIV combo HIV combo
S/CO S/CO S/CO
9016-07 23 1 9016-08 27 1127.4 7473 0.32 0.21 0.20 9016-09 30 14089.9 69,101 3.36 1.41 1.11 9018-06 19 9018-07 22 45.8 304 0.09 0.14 0.07 2 9018-08 26 5067.3 15,280 0.21 0.78 0.44 9018-09 29 52692.4 193,100 1.55 5.46 4.18 9024-10 46 3 9024-11 50 1743.7 12,840 0.60 0.38 0.25 9024-12 54 Not tested >500,000 12.41 30.84 25.94 9031-13 126 9031-14 130 13.7 197 0.51 0.06 0.07 4 9031-16 137 1516.5 10,507 0.49 0.28 0.23 9031-18 152 Not tested 173,075 5.671 6.10 4.41 Mean days 57.3 57.3 66.3 64.5 64.5 L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160 157 105 Y = M0 + M1*X M0 -1.501 1 104 M1 1.1817 R 0.99464 1000 100 10 Digital immunoassay (fg/mL p24) 1 4 5 6 10 100 1000 1x10 1x10 1x10 bDNA (RNA copies/mL) Fig. 4. Correlation between digital immunoassay and quantitative bDNA NAT Fig. 5. Histogram of digital immunoassay results from 45 HIV-negative samples. method. Samples tested were from four panels of seroconversion samples from HIV Samples were tested in replicates of 3 with two reagent and calibrator batches on infected individuals. Digital immunoassay data depicted are the mean of three repli- different days. The dashed lines a, b, and c depict the mean AEB of the zero calibrator cates. bDNA data are as reported by the commercial sample vendor (Zeptometrix). (0.00559), 4SD from the HIV-negative sample population mean (0.00710), and 3SD from the mean zero calibrator signal (0.00781) respectively. The LoD estimated at 0.00781 AEB corresponds to 5.75 fg/mL p24. of the infection (197 RNA copies/mL). The digital p24 assay also detected the infection in this blood draw, measuring 13.7 fg/mL p24. First detection of this infection by conventional immunoassay in Fig. 4 were as reported by the sample vendor, and details around came 2–3 weeks later. the testing were not available. The results between the two meth- Table 2 summarizes earliest detection data across six additional ods demonstrated linear proportionality across a 5-log range, and sets of seroconversion samples for the digital immunoassay, three the relationship between p24 and RNA copies/mL in this compar- NAT methods, and two conventional immunoassay methods reac- ison may reflect differences in standardization between the two tive for p24. The digital immunoassay and NAT methods detected NAT methods. the presence of infection approximately 9–11 days earlier than the conventional immunoassay methods. The time-to-detection for the digital immunoassay (20.5 days) was within the range reported 3.5. HIV-negative samples for the NAT methods (18.8–21.3 days). Taken together, these data indicate that the digital immunoassay is able to detect acute HIV A preliminary evaluation of HIV-negative samples was con- infection with the sensitivity of NAT. ducted on EDTA plasma from 45 presumed normal human donors. All samples were NAT-negative for HIV RNA (PROCLEIX ULTRIO, 3.4. Correlation between p24 and quantitative NAT Novartis Diagnostics). Approximately half the samples (24) were tested with the p24 digital immunoassay using one batch of A comparison between the digital p24 immunoassay and a reagents and calibrators, and the remaining samples were tested quantitative NAT method across four sets of seroconversion pan- on a separate date using a second batch of reagents and calibra- els (Table 1) is depicted in Fig. 4. The two methods exhibited tors. The distribution of immunoassay results from all 45 samples 2 excellent correlation (R 0.99). The relationship between measured is depicted in Fig. 5. All but one sample gave an AEB signal below p24 and RNA copies/mL reported for the bDNA method differed the mean signal for the zero calibrator (Fig. 5a). A comparison of the by approximately 3.5-fold from that obtained by testing a stan- statistics of the HIV-negative samples with the AEB corresponding dardized preparation of cultured virus (Fig. 3). In that study, the to the mean LoD estimated from the zero calibrators (as 3SD above cultured virus was standardized for HIV-1 RNA/mL by the Roche the zero background) is depicted in Fig. 5b–c. The negative pop- ® AMPLICOR HIV-1 MonitorTM Assay, v. 1.5, and the standardiza- ulation mean was over four standard deviations lower than the tion was confirmed in the digital immunoassay by agreement with AEB corresponding to the mean LoD (5.75 fg/mL). Using either cri- the accepted stoichiometry of p24 molecules to virus. The NAT data terion (4SD from negative sample mean or estimated LoD) as a Table 2 Days of initial detection of HIV infection since first draw. HIV infected Seroconversion Digital p24 Seracare data donor panel designation immunoassay Roche CAP/ Siemens bDNA Abbott HIV Perkin Elmer p24 Abbott Architect CTM v1.0 NAAT NAAT RNA NAAT immunoassay HIV combo immunoassay 5 PRB968 15 15 15 15 26 26 6 PRB956 40 40 40 40 47 47 7 PBR958 2 0 0 0 9 7 8 PBR967 0 3 7 7 17 17 9 PBR969 55 55 55 53 63 63 10 PBR972 11 0 11 3 18 18 Mean days 20.5 18.8 21.3 19.7 30.0 29.7 158 L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160 preliminary cut-off for HIV reactivity, the specificity of this limited pathogens (e.g. HIV/HBV/HCV triplex and West Nile virus), and study was 100%. public expectation of maximum blood safety has led to the estab- lishment of NAT for blood bank screening in higher income countries despite its cost. This is not the case, however, for many 4. Discussion European and developing countries (Laperche, 2005; Ekiaby et al., 2010), where feasibility studies weigh cost effectiveness, infra- The data presented in this report demonstrate that by isolat- structure requirements, affordability, and disease prevalence vs. ing single labeled immunocomplexes and confining fluorescent less sensitive alternatives. The practicality of extending NAT pool- product from a brief amplification step to a small volume, a ing to clinical screening has also been explored. Stekler et al. and simple immunoassay with a commercially available antibody Hutchinson et al. concluded that with regular 3rd generation anti- pair exhibited a 2000–3000-fold reduction in limit of detec- HIV antibody testing, NAT for acute HIV infection detection makes tion vs. conventional immunoassays reactive for p24 antigen. sense only in higher risk settings such as sexually transmitted dis- This sensitivity improvement, together with the natural 2000- ease clinics (Stekler et al., 2009; Hutchinson et al., 2010). fold amplification provided by the p24:virus stoichiometry, enable Economics aside, timely detection and reporting represent immunoassay sensitivity to virus comparable to NAT for detection another challenge limiting the potential epidemiological impact of of acute HIV infection. Preliminary analytical LoD was shown to NAT detection of acute HIV infection. The acute phase represents be 4.87 fg/mL, approximating 122 RNA copies/mL (∼60 virus/mL). a period of high viremia and hyperinfectiousness, contributing Within the sample volume of 100 L utilized for the assay, this disproportionately to the spread of the disease (Wawer et al., equates to detection of only six viruses. For comparison, commer- 2005; Gray et al., 2001). Turnaround time for executing a pool- cially available 4th generation HIV combo assays exhibit LoDs of ing strategy, testing, and deconvolution of a positive pool can delay approximately 11–70 pg/mL (11,000–70,000 fg/mL) (Fiebig et al., reporting an acute infection for days, making timely intercession 2003). Across 10 sets of seroconversion samples from HIV infected of a patient’s potential to transmit virus difficult. The time spent individuals, the digital immunoassay matched NAT for early detec- determining a positive result and locating the infected individ- tion of HIV. ual may largely negate the sensitivity advantage gained by NAT. If A low-cost, easily deployable alternative to NAT could have the method by which early stage infection is caught were capable important implications for HIV screening economics and efficacy. of a short, patient-specific turnaround of results, the reduction of NAT economics in different screening settings is a key issue, the detection window could have greater practical significance, as prompting many cost-effectiveness examinations (Stramer et al., there would be greater likelihood of conveying acute HIV infection 2004; Marshall et al., 2004; Jackson et al., 2003; Custer et al., status to the patient and preventing further transmission dur- 2005; Stekler et al., 2009; Hutchinson et al., 2010; Laperche, 2005; ing the hyperinfectious stage. Today’s automated high-sensitivity Ekiaby et al., 2010). In blood bank screening, despite donor pool- immunoassays typically have an assay time of an hour or less, ing to reduce cost, NAT is cost ineffective for serology-negative and results can be turned around while the patient waits. By detection of acute HIV infection. In the US, estimates of cost infusing sample with antibody-coated microbeads, even extremely per Quality Adjusted Life Year (QALY) range from $1.5–4.3 mil- low abundance antigen can be quickly and efficiently captured lion unadjusted for inflation (Marshall et al., 2004; Jackson et al., and labeled with good ELISA-grade antibodies. This kinetic prin- 2003) which greatly exceeds normal medical standards (gen- ciple was recently shown to apply to a very high sensitivity erally $100,000–200,000 USD (Hutchinson et al., 2010)). Overall Simoa digital immunoassay (Chang et al., 2012). Additionally, with cost effectiveness ratios improve with the inclusion of additional Fig. 6. Revised HIV screening algorithm proposed by the Center for Disease Control at the APHL HIV Diagnostics Conference, 2010. Adapted from Rollins (2010). L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160 159 immunoassay, virus detection does not require sample pretreat- specimens, adding $7.16/specimen to the initial immunoassay cost ment during early stage infection when competing antibodies are (Gous et al., 2010). These studies highlight both the epidemiological weak or absent, further enabling a short turnaround time. benefits of NAT-sensitive screening for acute HIV infection, as well While studies on the practicality of NAT in the clinical set- as the challenges and economic pressures around NAT adoption. ting have focused on immunoassay-negative samples, recent HIV A low cost immunoassay with the sensitivity of NAT could repre- screening recommendations utilize NAT in a confirmatory capacity sent an opportunity to further impact the spread of transmitted and (Owen et al., 2008; Rollins, 2010). Expanded screening is aimed transfusion-borne HIV in lower resource, high-risk areas. at the 20% of HIV infected individuals unaware of the HIV sta- The prototype digital p24 immunoassay described in this report tus (Marks et al., 2006), and a recent algorithm proposed in the utilizes standard low cost bead-based ELISA reagents and com- U.S. by the Centers for Disease Control (modeled on Western Euro- mercially available antibodies. Array technology readily lends to pean practice for the past decade) starts with a 4th generation HIV multiplexing, and from a reagent standpoint, the cost/test for a combo immunoassay (Fig. 6). A premise of the algorithm is to uti- ‘combo’ digital p24/HIV-1/2 antibody assay would be expected to lize 4th generation immunoassay screening due to its reactivity be comparable to conventional HIV combo immunoassays – cur- to sufficiently elevated p24 prior to seroconversion (Hecht et al., rently one-third to one-tenth the cost of NAT (Karris et al., 2012). 2002) as well as anti-p24 immunoglobulins following seroconver- For low cost automation and deployment of digital immunoas- sion. Fig. 6 represents inclusion of acute HIV infection detection says, Kan et al. recently reported proof of principle of low cost capability to assay point-of-entry testing for clinical HIV screening. plastic disposable disks composed of 24 flow cells that integrate Because HIV combo immunoassays do not discriminate between microfluidics with high-density Simoa microarrays of 216,000-fL possible sources of a reactive result, additional discrimination sized wells (Kan et al., 2011). This array disk consumable is aimed reflex testing is recommended. Ideally, a screening test should offer at enabling cost-effective integration of Simoa technology with three things: high sensitivity for both acute and chronic infection, conventional automated immunoassay instrumentation perform- low cost, and short turnaround time to facilitate patient inter- ing paramagnetic bead-based assays. More extensive validation of vention. NAT provides the highest sensitivity for acute, but can the digital p24 immunoassay described and further development miss chronic infection when viral load is undetectable. Laboratory of a high-throughput automated analyzer instrument is ongoing. immunoassays and point-of-care devices offer cost and turnaround time advantages, but can miss acute infection. 4th generation Disclosure combo immunoassays have presented the best trade-off among these imperatives. Although 4th generation combo immunoassays The co-authors are employees of Quanterix Corporation. shorten the detection window relative to serological tests, a sen- sitivity gap remains: patients given a clean bill by 4th generation Acknowledgments immunoassay may actually be highly infectious. Failure to diag- nose acute HIV infection represents a key public health problem The authors thank Todd Campbell, Ray Meyer, Tomasz Piech, (Pao et al., 2005; Brenner et al., 2007). Assured by apparent HIV and Alfred Schroff for their experimental help and manuscript sug- negative status, acutely infected individuals represent a high trans- gestions. mission risk and a lost opportunity to control the spread of the virus. <10% of the 300–400 annual HIV infection diagnoses in Seattle, for References example, were of acute HIV infection (Stekler et al., 2009). A simple immunoassay with rapid turnaround time that is capable NAT sen- Assal, A., Barlet, V., Deschaseaux, M., Dupont, I., Gallian, P., Guitton, C., Morel, P., sitivity for acute HIV infection could represent another step toward David, B., De Micco, P., 2009. Comparison of the analytical and operational per- leak-proof HIV screening in a more cost and clinically effective way. formance of two viral nucleic acid test blood screening systems: Procleix Tigris and cobas s 201. Transfusion 49, 289–300. Because of their relative simplicity, immunoassays are more AVERT. AIDS and HIV Information from AVERT. AVERT, Horsham, United Kingdom. practically deployable than NAT in diverse environments. HIV http://www.avert.org/safricastats.htm (accessed November 2011). screening in developing countries and low-resource settings with Brenner, B.G., Roger, M., Routy, J.P., Moisi, D., Ntemgwa, M., Matte, C., Baril, J.G., high HIV incidence faces the issues described above, but on a larger, Thomas, R., Rouleau, D., Bruneau, J., Leblanc, R., Legault, M., Tremblay, C., Charest, H., Wainberg, M.A., 2007. High rates of forward transmission events after more urgent scale. NAT is a sophisticated technology with adop- acute/early HIV-1 infection. The Journal of Infectious Diseases 195, 951–959. tion barriers on both supplier and user sides. Although screening Briggs, J.A., Simon, M.N., Gross, I., Kräusslich, H.G., Fuller, S.D., Vogt, V.M., John- donated blood with NAT has been in practice since the late 90s, son, M.C., 2004. The stoichiometry of Gag protein in HIV-1. Nature Structural & Molecular Biology 11, 672–675. 40% of worldwide blood donations remain unscreened or screened Busch, M.P., Kleinman, S.H., Jackson, B., Stramer, S.L., Hewlett, I., Preston, S., 2000. by less sensitive immunoassay (Global blood safety and availability Nucleic acid amplification testing of blood donors for transfusion-transmitted WHO fact sheet, 2009; Editorial, 2007), and 5–10% of new HIV infec- infectious diseases. Report of the Interorganizational Task Force on Nucleic Acid Amplification Testing of Blood Donors. Transfusion 40, 143–159. tions are transfusion-borne (Busch et al., 2005). In South Africa, the Busch, M.P., 2007. Evolving approaches to estimate risks of transfusion-transmitted prevalence of HIV infection in the general population is almost 11% viral infections: incidence-window period model after ten years. In: Dax, E.M., (AVERT, 2011), but diagnosis of acute HIV infection is rare (Stevens Farrugia, A., Vyas, G.N. (Eds.), Advances in Transfusion Safety – vol. IV, Develop- ments in Biologicals (Basel), vol. 127. S. Karger Publishers, Inc., Farmington, CT, et al., 2005). Since NAT is the only methodology currently capa- pp. 87–112. ble of definitively diagnosing acute HIV infection, feasibility of NAT Busch, M.P., Glynn, S.A., Stramer, S.L., Strong, D.M., Caglioti, S., Wright, D.J., Pap- implementation in these very high incidence settings has been palardo, B., Kleinman, S.H., 2005. A new safety strategy for estimating risks of transfusion-transmitted viral infections based on rates of detection of recently examined (Stevens et al., 2005; Karim et al., 2007). Karim et al. infected donors. Transfusion 45, 254–264. demonstrated the clinical benefit of a pooling strategy for identify- Chang, L., Rissin, D.M., Fournier, D.R., Piech, T., Patel, P.P., Wilson, D.H., Duffy, D.C., ing acute HIV infection in a high-risk cohort in South Africa (Karim 2012. Single molecule enzyme-linked immunosorbent assays: theoretical con- et al., 2007). A similar study on patients attending primary health- siderations. Journal of Immunological Methods 378, 102–115. Constantine, N.T., van der Groen, G., Belsey, E.M., Tamashiro, H., 1994. Sensitivity of care clinics in Johannesburg showed nearly 1% of individuals were HIV-antibody assays determined by seroconversion panels. AIDS 8, 1715–1720. acutely infected, translating to an annual incidence rate of 12.9% Custer, B., Busch, M.P., Marfin, A., Petersen, L., 2005. The cost-effectiveness of (Gous et al., 2010). Gous and colleagues reported 4 cases of acute screening the U.S. blood supply for West Nile virus. Annals of Internal Medicine 143, 486–492. HIV infection missed by 4th generation immunoassay testing of Editorial, 2007. Improving blood safety worldwide. Lancet 370, 1879–1974. 3005 samples in a Johannesburg general hospital. The pooling strat- Ekiaby, M.E., Lelie, N., Allain, J.P., 2010. Nucleic acid testing (NAT) in high egy spread a $21,568 USD cost for 211 NAT results across the 3005 prevalence–low resource settings. Biologicals 38, 59–64. 160 L. Chang et al. / Journal of Virological Methods 188 (2013) 153–160 Fiebig, E.W., Wright, D.J., Rawal, B.D., Garrett, P.E., Schumacher, R.T., Peddada, L., Procleix ULTRIO Assay Product Insert. 501807-REG Rev. 3. http://www.fda.gov/ Heldebrant, C., Smith, R., Conrad, A., Kleinman, S.H., Busch, M.P., 2003. Dynamics downloads/biologicsbloodvaccines/bloodbloodproducts/approvedproducts/ of HIV viremia and antibody seroconversion in plasma donors: implications for licensedproductsblas/blooddonorscreening/infectiousdisease/ucm092120.pdf diagnosis and staging of primary HIV infection. AIDS 17, 1871–1879. (accessed November 2011). Global blood safety and availability facts and figures from the 2007 blood safety Rissin, D.M., Kan, C.W., Campbell, T.G., Howes, S.C., Fournier, D.R., Song, L., Piech, T., survey. WHO Fact Sheet #279, November 2009. Patel, P.P., Chang, L., Rivnak, A.J., Ferrell, E.P., Randall, J.D., Provuncher, G.K., Walt, Gous, N., Scott, L., Perovic, O., Venter, F., Stevens, W., 2010. Should South Africa be D.R., Duffy, D.C., 2010. Single-molecule enzyme-linked immunosorbent assay performing nucleic acid testing on HIV enzyme-linked immunosorbent assay- detects serum proteins at subfemtomolar concentrations. Nature Biotechnology negative samples? Journal of Clinical Microbiology 48, 3407–3409. 28, 595–599. Gray, R.H., Wawer, M.J., Brookmeyer, R., Sewankambo, N.K., Serwadda, D., Wabwire- Rissin, D.M., Fournier, D.R., Piech, T., Kan, C.W., Campbell, T.G., Song, L., Chang, L., Mangen, F., Lutalo, T., Li, X., vanCott, T., Quinn, T.C., 2001. Probability of HIV- Rivnak, A.J., Patel, P.P., Provuncher, G.K., Ferrell, E.P., Howes, S., Pink, B.A., Min- 1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant nehan, K.A., Wilson, D.H., Duffy, D.C., 2011. Simultaneous detection of single couples in Rakai, Uganda. Lancet 357, 1149–1153. molecules and singulated ensembles of molecules enables immunoassays with Hecht, F.M., Busch, M.P., Rawal, B., Webb, M., Rosenberg, E., Swanson, M., Chesney, broad dynamic range. Analytical Chemistry 83, 2279–2285. M., Anderson, J., Levy, J., Kahn, J.O., 2002. Use of laboratory tests and clinical Rissin, D., Walt, D., 2006. Digital concentration readout of single enzyme molecules symptoms for identification of primary HIV infection. AIDS 16, 1119–1129. using femtoliter arrays and Poisson statistics. Nano Letters 6, 520–523. Hutchinson, A.B., Patel, P., Sansom, S.L., Farnham, P.G., Sullivan, T.J., Bennett, B., Rollins, G., 2010. What’s the best testing strategy for HIV infection? Clinical Labora- Kerndt, P.R., Bolan, R.K., Heffelfinger, J.D., Prabhu, V.S., Branson, B.M., 2010. tory News 36, 1–11. Cost-effectiveness of pooled nucleic acid amplification testing for acute HIV Sanchez-Pescador, R., Power, M.D., Barr, P.J., Steimer, K.S., Stempien, M.M., Brown- infection after third-generation HIV antibody screening and rapid testing in the Shimer, S.L., Gee, W.W., Renard, A., Randolph, A., Levy, J.A., 1985. Nucleotide United States: a comparison of three public health settings. PLoS Medicine 7, sequence and expression of an AIDS-associated retrovirus (ARV-2). Science 227, e1000342. 484–492. Jackson, B.R., Busch, M.P., Stramer, S.L., AuBuchon, J.P., 2003. The cost-effectiveness Steimer, K.S., Puma, J.P., Power, M.D., Powers, M.A., George-Nascimento, C., Stephans, of NAT for HIV, HCV, and HBV in whole-blood donations. Transfusion 43, J.C., Levy, J.A., Sanchez-Pescador, R., Luciw, P.A., Barr, P.J., 1986. Differential 721–729. antibody responses of individuals infected with AIDS-associated retroviruses Kan, C.W., Rivnak, A.J., Campbell, T.G., Piech, T., Rissin, D.M., Mösl, M., Peterc¸ a, A., surveyed using the viral core antigen p25gag expressed in bacteria. Virology Niederberger, H-P., Minnehan, K.A., Purvish, P.P., Ferrell, E.P., Meyer, R.E., Chang, 150, 283–290. L., Wilson, D.H., Fournier, D.R., Duffy, D.C., 2011. Isolation and detection of single Stekler, J.D., Swenson, P.D., Coombs, R.W., Dragavon, J., Thomas, K.K., Brennan, C.A., molecules on paramagnetic beads using sequential fluid flows in micro fabri- Devare, S.G., Wood, R.W., Golden, M.R., 2009. HIV testing in a high-incidence cated polymer array assemblies. Lab on a Chip 12, 977–985. population: is antibody testing alone good enough? Clinical Infectious Diseases Karim, S.S., Mlisana, K., Kharsany, A.B., Williamson, C., Baxter, C., Karim, Q.A., 2007. 49, 444–453. Utilizing nucleic acid amplification to identify acute HIV infection. AIDS 21, Stevens, W., Akkers, E., Myers, M., Motloung, T., Pilcher, C., Venter, F., 2005. High 653–655. prevalence of undetected, acute HIV infection in a South African primary health Karris, M.Y., Anderson, C.M., Morris, S.R., Smith, D.M., Little, S.J., 2012. Cost savings care clinic. In: IAS Conf. HIV Pathol. Treatmt., Rio de Janeiro, Brazil (Abstr. associated with testing of antibodies, antigens, and nucleic acids fordiagnosis of MoOa0108). acute HIV infection. Journal of Clinical Microbiology 50, 1874–1878. Stramer, S.L., Glynn, S.A., Kleinman, S.H., Strong, D.M., Caglioti, S., Wright, D.J., Dodd, Kontio, S., 1991. Sensitive one-step enzyme immunoassay for HIV-1 p24 antigen in R.Y., Busch, M.P., 2004. Detection of HIV-1 and HCV infections among antibody- human blood specimens and cell culture supernatants. Journal of Immunological negative blood donors by nucleic acid-amplification testing. The New England Methods 139, 257–263. Journal of Medicine 351, 760–768. Laperche, S., 2005. Editorial. Blood safety and nucleic acid testing in Europe. Euro Summers, M.F., Henderson, L.E., Chance, M.R., Bess, J.W., South, T.L., Blake, P.R., Sagi, Surveillance 10 (2). I., Perez-Alvarado, G., Sowder, R.C., Hare, D.R., 1992. Nucleocapsid zinc fingers Marcel, M., Grèzeb, M., Bailly, A., Izopet, J., 2011. Analytical sensitivity of four HIV detected in retroviruses: EXAFS studies of intact viruses and the solution-state combined antigen/antibody assays using the p24 WHO standard. Journal of Clin- structure of the nucleocapsid protein from HIV-1. Protein Science 1, 563–574. ical Virology 50, 57–60. Ukkonen, P., Korpela, J., Suni, J., Hedman, K., 1988. Inactivation of human immu- Marks, G., Crepaz, N., Janssen, R.S., 2006. Estimating sexual transmission of HIV from nodeficiency virus in serum specimens as a safety measure for diagnostic persons aware and unaware that they are infected with the virus in the USA. AIDS immunoassays. European Journal of Clinical Microbiology & Infectious Diseases 20, 1447–1450. 4, 518–523. Marshall, D.A., Kleinman, S.H., Wong, J.B., AuBuchon, J.P., Grima, D.T., Kulin, N.A., Wawer, M.J., Gray, R.H., Sewankambo, N.K., Serwadda, D., Li, X., Laeyendecker, Weinstein, M.C., 2004. Cost-effectiveness of nucleic acid test screening of vol- O., Kiwanuka, N., Kigozi, G., Kiddugavu, M., Lutalo, T., Nalugoda, F., Wabwire- unteer blood donations for hepatitis B, hepatitis C and human immunodeficiency Mangen, F., Meehan, M.P., Quinn, T.C., 2005. Rates of HIV-1 transmission per virus in the United States. Vox Sanguinis 86, 28–40. coital act, by stage of HIV-1 infection, in Rakai, Uganda. The Journal of Infectious Mullis, K.B., Faloona, F.A., Scharf, S., Saiki, R.K., Horn, G., Erlich, H.A., 1986. Specific Diseases 191, 1403–1409. enzymatic amplification of DNA in vitro: the polymerase chain reaction. Cold Weber, B., Fall, E.H., Berger, A., Doerr, H.W., 1998. Reduction of diagnostic window Spring Harbor Symposia on Quantitative Biology 51, 263–273. by new fourth-generation human immunodeficiency virus screening assays. National Institute of Allergies and Infectious Disease. Structure of HIV. Journal of Clinical Microbiology 36, 2235–2239. http://www.niaid.nih.gov/topics/HIVAIDS/Understanding/Biology/pages/ Westreich, D.J., Hudgens, M.G., Fiscus, S.A., Pilcher, C.D., 2008. Optimizing screening structure.aspx (accessed November 2011). for acute human immunodeficiency virus infection with pooled nucleic acid Owen, S.M., Yang, C., Spira, T., Ou, C.Y., Pau, C.P., Parekh, B.S., Candal, D., Kuehl, D., amplification tests. Journal of Clinical Microbiology 46, 1785–1792. Kennedy, M.S., Rudolph, D., Luo, W., Delatorre, N., Masciotra, S., Kalish, M.L., Wilson, D.H., Hanlon, D.W., Provuncher, G.K., Chang, L., Song, L., Patel, P.P., Fer- Cowart, F., Barnett, T., Lal, R., McDougal, J.S., 2008. Alternative algorithms for rell, E.P., Lepor, H., Partin, A.W., Chan, D.W., Sokoll, L.J., Cheli, C.D., Thiel, human immunodeficiency virus infection diagnosis using tests that are licensed R.P., Fournier, D.R., Duffy, D.C., 2011. Fifth-generation digital immunoassay for in the United States. Journal of Clinical Microbiology 46, 1588–1595. prostate-specific antigen by single molecule array technology. Clinical Chem- Pao, D., Fisher, M., Hué, S., Dean, G., Murphy, G., Cane, P.A., Sabin, C.A., Pillay, D., 2005. istry 57, 1712–1721. Transmission of HIV-1 during primary infection: relationship to sexual risk and Zetterberg, H., Mörtberg, E., Song, L., Chang, L., Provuncher, G.K., Patel, P.P., Fer- sexually transmitted infections. AIDS 19, 85–90. rell, E., Fournier, D.R., Kan, C.W., Campbell, T.G., Meyer, R., Rivnak, A.J., Pink, Piatak, M., Saag, M.S., Yang, L.C., Clark, S.J., Kappes, J.C., Luk, K.C., Hahn, B.H., Shaw, B.A., Minnehan, K.A., Piech, T., Rissin, D.M., Duffy, D.C., Rubertsson, S., Wilson, G.M., Lifson, J.D., 1993. High levels of HIV-1 in plasma during all stages of infec- D.H., Blennow, K., 2011. Hypoxia due to cardiac arrest induces a time-dependent tion determined by competitive PCR. Science 259, 1749–1754. increase in serum amyloid  levels in humans. PLoS ONE 6, e28263.