Comparison of Three Adenovirus Quantitative PCR Assays with ATCC Reference Strains and Clinical Samples

Adenoviruses (AdV) have been associated with a variety of human diseases and are recognized as causing significant morbidity and mortality in immunocompromised or transplant patients. Quantification of AdV DNA in plasma is notoriously difficult due to the genetic diversity of the 71 different serotypes identified to date. There is no World Health Organization standard available to harmonize quantitative data, so results between labs vary widely.

ABSTRACT

Adenoviruses (AdV) have been associated with a variety of human diseases and are recognized as causing significant morbidity and mortality in immunocompromised or transplant patients. Quantification of AdV DNA in plasma is notoriously difficult due to the genetic diversity of the 71 different serotypes identified to date. There is no World Health Organization standard available to harmonize quantitative data, so results between labs vary widely. In this study, we compared a laboratory-developed multiplex PCR assay with primers and probes specific for each group (A to G) and subgroup E4 (Octaplex) to one with a single primer and probe set (modified from N. Jothikumar et al., Appl Environ Microbiol 71:3131–3136, 2005) and one utilizing bisulfite pretreatment of DNA to reduce variation prior to amplification (Genetic Signatures). Our Octaplex assay detected all low-copy-number clinical samples, while the other two assays had subsets of samples that did not amplify. The modified Jothikumar assay failed to efficiently amplify three of the high-copy-number cultured strains, while the Genetic Signatures 3base assay had a positive bias, resulting in higher copies/ml (>0.5 log₁₀) for all culture fluids tested. All three assays resulted in endpoint detection of the available 51 AdV types. Using two different materials to generate a standard curve revealed that the Octaplex TaqMan assay and the modified Jothikumar assay both consistently gave adenovirus levels lower than the commercial platform for AdV culture fluids but not patient samples. This study highlights the differences in detection of AdV between laboratories that can be attributed to both the PCR method, as well as the reference material used for quantitation.

INTRODUCTION

Human adenoviruses (AdV) are a group of nonenveloped viruses containing a double-stranded genome (34 to 36 kb in size) (1). A majority of infections occur in children, but immunocompromised adult populations are also at risk of infection or reactivation in the conjunctiva, gastrointestinal tract, and respiratory tract. Adenovirus infections are usually self-limiting, except within immunosuppressed populations, where disease progression can be severe (2). The reactivation and dissemination of AdV, cytomegaloviruses, Epstein-Barr viruses, and other viruses can have devastating effects, resulting in increased morbidity and mortality (3, 4).

There are currently more than 70 AdV types (5–7) in seven groups (human adenovirus A to G), with AdV-D containing the most types (8, 9). Specific AdV types have been associated with different clinical presentations (8). For example, types 40 and 41 (group F) have been implicated as a leading cause of acute gastroenteritis (10), whereas groups B and C have been linked to outbreaks of pharyngoconjunctival fever (11).

Quantification of AdV can provide greater diagnostic value than qualitative detection of AdV in blood samples (12–14), and early detection of AdV DNA in plasma and sterile fluids has been shown useful for identifying patients at risk for invasive AdV disease. Quantification of AdV DNA in plasma has been beneficial for monitoring patients’ risk for invasive AdV disease, progression, and response to treatment (15–18).

The high divergence between the individual strains of human adenovirus has been a major obstacle for the development of quantitative AdV PCR assays that cover all identified AdV types. An early assay described by Heim et al. (19) described a quantitative assay for 51 types of AdV from blood. A single primer/probe set, which amplified hexon sequence, was designed to achieve amplification of all AdV types known at the time. Quantification was determined using AdV-2 and AdV-41 DNA stock preparations. A variety of other adenovirus PCR single-primer set assays have since been designed and many have been published, but because of the high diversity within adenovirus sequences, most assays are very similar to each other and directed to small, relatively conserved areas in the penton or hexon regions of the virus. Because there is still viral diversity in these locations, the primers often contain degenerate bases or utilize other strategies to overcome occasional mismatched bases (20). A new strategy to overcome mismatches has been pioneered by Genetic Signatures, which utilizes bisulfite reduction of the DNA prior to extraction/PCR. The bisulfite reduces cytosine to uracil, which is then seen in subsequent PCR amplification steps as a thymine. This reduces the number of degenerate bases by limiting the needed matches to three bases instead of four. An alternative strategy for adenovirus PCR design has been described that involves creating a multiplex mix of primer sets, each specific for one or more of the AdV groups (21, 22).

In addition, the absence of a universal standard for standardization adds another level of complexity to the ability to accurately quantify all AdV consistently between labs. There is a paucity of quantitative reference material and no World Health Organization (WHO) standard available to harmonize quantitative data, so results between labs can vary widely. Currently, most material available for assay design/clinical standardization is made from the AdV-5 strain (group C), although nonquantified culture fluids for 51 of the strains are available from ATCC. We evaluated two different standard materials (AdV-3 and AdV-5) as part of this study.

Our primary goal was to update our existing laboratory-developed five-primer set multiplex assay (21) to improve our detection of more recently described strains. After analysis of all available sequences (>400 in total), we modified some of our existing primer sets and added three additional primer sets (eight total) to cover all known sequences. This new “Octaplex” assay was then compared to a single-plex assay from the literature and the Genetics Signatures assay.

MATERIALS AND METHODS

Reference materials.

The AcroMetrix adenovirus plasma panel was purchased from Thermo Fisher Scientific (Waltham, MA). This material of unknown AdV type came in a panel of five vials of culture fluid contained 10³, 10⁴, 10⁵, 10⁶, and 10⁷ copies/ml. Each tube in the panel was extracted on both the MagNA Pure 96 (Roche Molecular, Pleasanton, CA), as well as the GS-mini (Genetic Signatures, Newtown, Australia), for comparison of methods and generation of standard curves. As part of this study, the AcroMetrix panel material was sequenced and found to be an AdV-3 strain.

Human adenovirus 5 type C was purchased from American Type Culture Collection (ATCC; Manassas, VA; catalog no. VR-1516) and characterized by the Adenovirus Reference Material Working Group (ARMWG) to be 5.8 × 10¹¹ particles/ml (23–25). This material was serially diluted, and six 10-fold dilutions were extracted on both the MagNA Pure 96, as well as the GS-mini, for comparison of methods and generation of standard curves.

Samples.

A total of 51 culture fluids (prototype strains of AdV 1 to 51) purchased from the ATCC and 69 plasma specimens containing adenovirus recovered from leftover samples submitted for qualitative or quantitative adenovirus testing at the University of Washington Clinical Virology Lab (collected from 2014 to 2017) were tested. The plasma specimens were isolated mostly from posttransplant cases with clinical illness. Most ADV strains were represented in the ATCC culture fluids, but the more newly identified E4, G, and D strains (14 new D strains with strain numbers of >52) were not available. In cases where sample volume was limiting, extraction was only performed on the MagNA Pure 96. To further address the E group, five proficiency samples from previous surveys from Quality Control Molecular Diagnostics (QCMD; Scotland, UK) and the College of American Pathology (CAP; Northfield, IL) which were consensus typed as E were analyzed. For control purposes, one type B and two type C samples from some of the same shipment dates were also tested. No clinical material was available for the group G strain.

Sample extractions.

(i) Octaplex and Jothikumar assays. A total of 200 μl of plasma or culture fluid was loaded into a Roche MagNA Pure96, and DNA was extracted according to the manufacturer’s instructions. A measured amount of plasmid containing a sequence of jellyfish DNA used as an internal positive extraction control (cloned by Blue Heron, Bothel WA) was added to the lysis buffer. The final elution volume was 100 μl, and eluate was split and used for the Octaplex assay, the modified Jothikumar assay, and an assay for the internal control. All extractions performed gave adequate detection of the internal control. The 95% assay cutoff limit was 300 copies/ml for the Octaplex assay (data not shown).

(ii) Genetic Signatures 3base assay. Extractions used the GS-mini extractor according to the manufacturer’s instructions. The bisulfite reduction was done as a pre-step with an aliquot of 100 μl of plasma or culture fluid, which was mixed with 400 μl of extraction buffer (EasyScreen sample processing kit SP001) containing a lambda phage internal process control. The samples were heated at 95°C for 20 min, before being loaded into the extractor. The final elution volume was 50 μl, and all extractions performed gave adequate detection of the internal control.

Real-time PCR assays.

(i) Modified Jothikumar PCR assay. This assay was performed using a QuantiTect multiplex PCR mastermix (Qiagen, Germantown, MD) and an ABI 7500 (Thermo Fisher, Waltham, MA) real-time PCR system, as described by Jothikumar et al. (20), with several modifications. Amplification reaction mixtures contained 10 μl of template DNA, 200 nM primers, and 120 nM fluorogenic probe in a total volume of 50 μl. The forward and reverse primers and probe are found in Table 1. To slightly increase the assay performance, we made two modifications to the assay, one was a change to a nucleotide in the reverse primer to reflect sequence homology of all aligned sequences. In addition, the FAM probe at the 3′ end was changed from a black whole quencher to an MGBNFQ (Minor Groove Binder Non-Fluorescent quencher) quencher to improve binding strength and specificity. The PCR protocol had the following conditions: hot-start denaturation step of 95°C for 10 min, followed by 40 cycles with a 95°C denaturation for 60 s, 55°C annealing for 30 s, and 72°C elongation for 30 s.

TABLE 1.

Primers and probes used in this study

Primer or probe	Sequencea
Primers
A forward	CCG GKC TGG TGC AAT TCG
A reverse	CGA TCC ACG GGC ACA AA
B forward	TGG ACA TGA CYT TYG AGG TGG AT
B reverse	CGT CGA ADA CTT CRA ARA GAA GA
C forward	TYG ACA CCA CCC GTG TGT AC
C reverse	TGC TGT GGT CGT TCT GGT AGT T
D forward	ATG ATG CCG CAG TGG GCG TAC AT
D reverse	TCA GGT ACT CCG AGG CGT CCT
E forward	TCA ACC ACC ACC GYA AYG C
E reverse	TGG ATG TGG AAT GGC ACG TA
E forward type4	GCT TAA CTT GCT TGT CTG TGT ATA TGT G
E reverse type4	GCC TTT CTC TTC ACT CCT CCT TCT
F forward	TGT TYG AAG TTT TCG ACG TYG T
F reverse	SAG GTA GAC GGC CTC GAT GA
G forward	TGT TYG AAG TYT TCG ACG TBG T
Jothikumar forward	GGA CGC CTC GGA GTA CCT GAG
Jothikumar reverse	ACN GTG GGG TTT CTA AAC TTG TT
Probes
A probe	CCA CGG ACA CCT ACT TCA CCC TGG G
B probe	CCA TGG ATG AGC CCA CCC TGC TTT
C probe	TGG ACA ACA AGT CAA CGG ATG TGG CA
D probe	TAC ATG CAC ATC GCC GGG CA
E probe	TAC CGC TCC ATG CTC CTG GGCA
E4 probe	TCC GCC GCC GCT GCT GTC
F probe	CGC ATC CAC CAG CCS CAC C
Jothikumar probe	CTG GTG CAG TTC GCC CGT GCC A

^aUnderlined bases indicate substitutions or new primers added to capture new AdV published genomes. Primer sets adapted from Huang et. al. (21) and Jothikumar et. al. (20). The boldface underlined base in the Jothikumar reverse sequence was modified to match all adenovirus sequences examined during the alignment studies.

(ii) In-house Octaplex PCR assay. Real-time TaqMan PCR was performed using QuantiTect multiplex PCR master mix (Qiagen) using 10 μl of DNA template and a total reaction volume of 35 μl. All assays were multiplexed with the adenovirus probes labeled with FAM dye and a primer/probe set specific for the jellyfish internal control material labeled with VIC (26). All primer and probe sequences used for the assay area contained in Table 1. The primer and probe concentrations were 250 and 62 nM, respectively. The PCR conditions were a 10-min activation step at 95°C, followed by 40 cycles of 95°C for 15 s and 60°C for 45 s. Amplifications were carried out using an ABI 7500 PCR system (Thermo Fisher).

(iii) Genetic Signatures PCR. An ASR PCR master mix containing a proprietary mixture of primers was provided by Genetic Signatures, and the probe was thawed and added in compliance with their technical document. We added 10 μl of extracted DNA to 12.7 μl of master mix for a total volume of 23 μl. PCR was performed on a Mic qPCR cycler (BioMolecular Systems, El Cajon, CA) using the following template: a 95°C activation step for 15 min, followed by 45 cycles of denaturation at 95C for 2 s, annealing at 55°C for 30 s, extension at 60°C for 30 s, and extension at 65°C for 30 s.

Analysis of all PCR data were done using either the ABI 7500 or the BioMolecular Systems software. Linear regression, Bland-Altman analysis, and graph generation was done with GraphPad Prism v7.0.3.

Next-generation sequencing.

Deep sequencing was performed on discrepant samples, which gave lower than expected PCR amplification with either the modified Jothikumar or the Genetic Signatures assays. DNA extracted as described above from culture fluids was quantified using Qubit 2.0 (Thermo Fisher). Sequencing libraries were prepared using 50 ng of DNA and quarter-volume reactions of a KAPA HyperPlus kit (Roche, Clifton, NJ). Samples were fragmentized at 37˚C for 7 min and ligated with 15 μM Y-stub adapters at 20°C for 30 min. Ligated samples were cleaned up using 0.8× Ampure XP beads and eluted into 20 μl of ddH₂O. Then, 10 μl of the eluate was amplified with 10 cycles of PCR using KAPA HiFi Ready Mix with TruSeq dual-indexed primers. Postamplification libraries reactions were purified with 0.8× Ampure XP beads and eluted in 20 μl of ddH₂O. Libraries were run on a 1.2% agarose FlashGel (Lonza, Basel, Switzerland) as a quality check to verify amplification and visualize size distribution and then quantified with Qubit 2.0 (Thermo Fisher). Libraries were sequenced on 1 × 192bp run on Illumina MiSeq (Illumina, San Diego, CA). Samples were quality and adapter trimmed using cutadapt and aligned to a concatenation of adenovirus reference genomes using Geneious R10 (Newark, NJ).

RESULTS

Optimizing and initial evaluation of our new Octaplex assay.

In order to evaluate the predicted performance of our existing assay against the full diversity of AdV as currently understood, we compared the current primer/probe sets with published genomes. All available adenovirus full-length genomes and hexon sequences (>400 in total), including the newest additions (AdV 52 to 71) were compared to the existing primer/probe sets utilizing Geneious software. We identified several mismatches with the potential to reduce detection of certain subtypes with our existing primer sets and both redesigned existing primer sets and added additional primers to cover all known sequences. Primers/probes to detect a newly described E lineage (E₄) (27, 28), all D types, and group G were added to the existing mix to ensure detection of all lineages.

We evaluated the new primer/probe sets against culture fluids as either a single-plex, 2-3plex (as in the existing assay [21]), and as an octaplex with all eight primer/probe sets in a single mix. The sequence of the Octaplex primers/probes can be found in Table 1, and a summary of strain cross-reactions seen with the primer sets when used individually with one of each of the 51 ADV strains can be seen in Fig. 1. Compared to the single-plex reactions, we did not observe any inhibition or loss of amplification when multiple primers and probes were multiplexed. Our existing assay utilized a degenerate group B primer/probe set to detect both group B and D groups. This current evaluation revealed that all group D samples were amplified on average 3.7 cycle thresholds (C_T) later with the group B primer/probe set (about 1.3 logs underquantified; see Fig. 2); these data confirmed the necessity of including the group D-specific primer/probe set in our final improved assay for accurate quantification of group D samples.

FIG 1.

Cross-reactivity of the 51 ATCC strains with each group-specific primer set. The columns represent the eight primer/probe sets tested, and the six rows represent the strain groups. The ratio number reflects the number of strains detected/total number of strains tested. The color intensity represents the number of samples amplified with each primer mix. The darker blue represents the most amplification, and the lighter blue represents the least amplification.

FIG 2.

Cross-reactivity of 31 group D ATCC strains with the group B primer/probe set. Each dot represents an individual group D strain and the reactivity with B and D primer/probe sets. The gray dotted line represents a perfect correlation between primer sets. Regression analysis of these data gave a slope of 0.8174 and an R² of 0.8708.

Comparison of C_T data for reference strains and patient samples.

A comparison of amplification results for each of the assays with ATCC strains as well as patient samples is shown in Table 2. The Octaplex assay and the modified Jothikumar assay amplified all 51 AdV culture fluids (see Table S1 in the supplemental material). However, for three culture fluids (AdV B3, B34, and F41), the C_Ts with the modified Jothikumar assay were later (>6 C_T) than the Octaplex assay. Due to limited extraction volumes for some of the samples, the Genetic Signatures assay could only be tested with 29 of the 51 culture fluids, but the assay detected all 29 samples. C_T data for all these reactions can be found in Table S1.

TABLE 2.

Initial amplification results with all three assays

Parameter	No. of results
	Octaplex assay	Modified Jothikumar assay	Genetic Signatures assay
ATCC strains (n = 51)
Total tested	51	51	29
Total positive	51	51a	29
Negative	0	0	0
Patient samples (n = 69)
Total tested	69	67	63
Positives	69	62	55
Negatives (initial result, >1,000 copies/ml)	0	4	6
Negatives (initial result, <1,000 copies/ml	0	1	2
Quantity not sufficient	0	2	6

^aThree samples had significantly lower C_T values than expected.

Of the 69 samples selected because of their positivity with our initial qPCR assay, the new Octaplex assay detected all 69 samples, while the modified Jothikumar assay detected 62 of 67 plasma samples tested (two samples had insufficient sample volume for testing) (Table 2). The five samples that were not detected by the modified Jothikumar assay had viral loads 10² to 10³ copies/ml (Table S2). These five failures were retested using the modified Jothikumar assay with leftover DNA, and all repeated as negative (data not shown). The Genetic Signatures assay detected 55 of 63 plasma samples tested (six samples had insufficient sample volume for testing). The eight negative samples had viral loads of around 10³ copies/ml and were negative upon retesting. The C_T data for these samples can be found in Table S2.

Reference materials affect assay performance.

Both the Genetic Signatures and Octaplex PCRs amplified the AcroMetrix reference material efficiently with similar C_Ts (average difference of 0.8 C_T) to create a standard curve (Table 3) with the expected 3.3 cycles of separation for 10-fold dilution series. The serial dilutions run on the modified Jothikumar assay gave significantly later C_Ts for three of the dilutions (average of 9.0 ± 0.3 C_T) and failed to amplify the other two. Thus, we could not accurately generate a standard curve for the AcroMetrix reference material using the modified Jothikumar assay. Subsequent sequencing showed this material to be an AdV B serotype 3 strain which was also shown to be inadequately detected by the modified Jothikumar assay. In contrast, all three assays amplified the VR-1516 reference material efficiently to generate a standard curve, with all three assays giving similar C_Ts (average difference ≤ C_T) and the expected C_T separations for the 10-fold dilution set.

TABLE 3.

Amplification C_T values for reference materials

Testing parameter and dose (copies/ml)	CT obtained
	Octaplex assay	Modified Jothikumar assay	Genetic Signatures assay
AcroMetrix panel
1 × 103	36.39	not detected	36.32
1 × 104	33.11	not detected	34.20
1 × 105	29.78	38.51	31.05
1 × 106	26.36	35.74	28.40
1 × 107	22.92	32.09	23.25
ATCC VR-1516
5.8 × 104	33.08	33.84	34.19
5.8 × 105	29.23	30.29	30.06
5.8 × 106	25.70	26.60	26.21
5.8 × 107	21.99	23.14	22.11
5.8 × 109	14.30	16.46	15.19
5.8 × 1011	8.90	9.98	8.99

To investigate the effect of reference material selection on quantification, we used each reference material to construct a standard curve for quantification. When we looked at the quantification of AdV in the prototype strain culture fluids (serotypes 1 to 51) using the AcroMetrix reference material to generate a standard curve, we observed higher copies/ml using the Genetic Signatures assay compared to the Octaplex assay (mean difference of 1.0 ± 0.2 log, P < 0.0001), especially at higher concentrations (Fig. 3A). For quantifying patient samples (Fig. 3B), the average difference between assays was lower (mean difference of 0.11 ± 0.69 log). Eight low-quantity samples by the Octaplex were negative on the Genetic Signatures assay. These differences seen between the two assays may partly be a result of different extraction efficiencies between the MP-96 for Octaplex and the GS-mini. The modified Jothikumar PCR assay quantities could not be compared to this data set due to failure to amplify the reference material.

FIG 3.

Comparison of quantitative results calculated from an AcroMetrix panel standard curve. Solid black lines represent the calculated regression line and dotted gray lines a perfect fit correlation line. (A) Thirty-nine ATCC strains. Regression analysis gave a slope of 1.097 and an R² of 0.9779. (B) Sixty-three patient samples. Regression analysis gave a slope of 1.201 and an R² of 0.8227.

Since all three assays gave similar results with the ATCC VR-1516 reference material, we could compare both culture strains and patient samples calculated from a single standard curve material (Fig. 4). For the culture fluids which predominantly had high copies, overall correlations are seen in Fig. 4A, C, and E. The Octaplex assay had excellent agreement (slope = 1.05; R² = 0.87) with the modified Jothikumar assay quantitative results, with the exception of the three strains discussed previously. For the Octaplex assay versus Genetic Signatures assay, the slope of the line was not parallel (slope 0.79) to a perfect correlation; results had a bias of about 0.2 logs at high quantities and nearly 1.0 log at low quantities. The modified Jothikumar versus Genetic Signatures assay comparison shows good correlation with the exception of the modified Jothikumar failure which skewed the regression analysis significantly (slope = 0.66; R² = 0.8), partially due to the inability to test several of the low-quantity samples with the modified Jothikumar assay. As for the patient data comparisons (Fig. 4B, D, and E), excellent correlation (R > 0.8) was seen with all comparisons for all samples tested. The small number of negatives in the modified Jothikumar and Genetics Signatures assays skewed the regression lines slightly away from perfect correlated lines.

FIG 4.

Comparison of quantitative results calculated from the ATCC VR-1516 reference standard curve. Solid black lines represent the calculated regression line, and dotted gray lines represent a perfect-fit correlation line. Panels A, C, and E contain ATCC adenovirus strains (n = 51); panels B, D, and F contain patient samples. (A and B) The Octaplex assay versus the Jothikumar assay yielded a regression slope of 1.027 and an R² of 0.8679 in panel A and a slope of 1.053 and an R² of 0.9118 in panel B. (C and D) The Octaplex assay versus the Genetic Signatures assay yielded a regression slope of 0.79944B and an R² of 0.8794 in panel C and slope of 1.188 and an R²  of 0.8093 in panel D. (E and F) The Jothikumar assay versus the Genetic Signatures assay yielded a regression slope of 0.6659 and an R² of 0.8014 in panel E and a slope of 1.076 and an R² of 0.8227 in panel F.

All three assays detect proficiency materials.

CAP and QCMD proficiency samples were pulled from previous years’ shipments and tested using all three assays. All three assays readily detected group B and C samples and, for the E samples, gave improved amplification over previous years. When using the human adenovirus C serotype 5 reference material (ATCC VR-1516) as the reference material for the standard curve (see the data subsequently presented below), all three assays gave higher values (average of 0.8 ± 0.2 log, P < 0.00001) than those reported for the PT consensus (Fig. 5). This may indicate that VR-1516, one of the only available standard materials, is not being used for standard curves by the majority of labs performing proficiency testing with QCMD and CAP materials.

FIG 5.

Quantitative comparison between the CAP/QCMD proficiency testing consensus final results and quantities from the Octaplex, Jothikumar, and Genetic Signatures assays. A regression analysis including all data generated a slope of 0.9685 and an R² of 0.888. A Bland-Altman analysis gave an average bias of 1.221 logs.

Discrepant samples contained mismatches in Jothikumar primers and probes.

AdV serotype B3 amplified 10 cycles later and AdV B34 and F41 amplified seven cycles later in the Jothikumar assay compared to the Octaplex assay. To investigate why the modified Jothikumar assay substantially underquantitated three culture fluid strains (B3, B34, and F41) and the AcroMetrix reference materials, we deep sequenced these samples and aligned the Jothikumar primer/probe sets to the obtained consensus sequence (Fig. 6). Mismatches present in the 3′ region of both the forward primer and the probe presumably account for the late amplification for AdV B3, whereas two additional mismatches in the reverse primer presumably account for late amplification in AdV B34. Three mismatches in the probe were likely responsible for partial failure of AdV F41.

FIG 6.

DNA sequence comparison for ATCC AdV strains that gave partial failures with the Jothikumar assay, adenovirus strains B3, B34, and F41. The line labeled Jothikumar is the primer/probe sequences from their publication. Lines that begin with “ATCC” are the sequences obtained at the University of Washington Medical Center by next-generation sequencing. Consensus sequences were obtained by alignment of all available GenBank sequences for those strains. Nucleotides different from the Jothikumar primer/probe sequences are highlighted in red. (Italic/bold denotes one nucleotide changed from the published assay to better represent published sequences.)

DISCUSSION

We compared not only the effect of assay properties but also the effect of different commercial reference quantification materials on quantification of AdV. We aligned >400 human AdV published genomes, including the new additions, and discovered primer/probe mismatches that have the potential to reduce detection of certain subtypes. To ensure detection of all types, we modified an existing multiplex AdV TaqMan assay by altering and multiplexing existing primers and probes to include all published adenovirus hexon gene sequences. We compared our Octaplex assay to one published assay (modified from Jothikumar) and one commercial platform (Genetic Signatures). The Jothikumar assay used one set of degenerate primers and one probe in combination with low stringency annealing conditions to detect all AdV types, while the commercial platform used a novel 3base technology to detect all AdV types. We found all three assays detected the available 51 AdV types, with differences seen in quantitation. Using two different standard curve materials, we found that the Octaplex TaqMan assay and the modified Jothikumar assay consistently gave lower quantified adenovirus levels (<0.5 ± 0.2 log) compared to the Genetic Signatures platform for AdV culture fluids but not patient samples.

For the samples run on the Jothikumar assay that failed to amplify, we repeated the PCR at least once, but we did not try alternative master mixes to optimize conditions. Furthermore, we modified the original published Jothikumar assay for our instrumentation, and the original assay’s performance was not tested. Similarly, for the samples run on the Genetic Signatures assay that failed to amplify, we repeated the PCR at least once, but followed the manufacturer’s instructions for use and did not alter PCR conditions.

Overall, all three assays showed similar performance quantifying patient AdV samples. Our Octaplex assay detected all patient samples while the modified Jothikumar and Genetic Signatures assays missed a few samples in the range of 10² to 10³ copies/ml. All three assays performed equally well in quantifying previously missed CAP/QCMD samples, though all three protocols yielded results roughly 1 log higher than the consensus reported when using commercial AdV quantification material ATCC VR-1516 (type C).

There have been a number of qPCR assays described that capture a wide breadth of AdV types, using degenerate primers and probes based on known consensus sequence (18, 19, 29). “Generic” detection of AdV has been difficult to due to high genetic diversity and continuing addition of new subtypes (30). Another approach that has been described relies on multiplexing many primer sets and probes into one or more reactions for broad-range AdV detection (31–36). One caveat of using low-stringency amplification conditions for quantification is that degenerate primers may amplify different viral types with different efficiencies based on their sequences, resulting in some types amplifying preferentially (18, 19, 29). Drawbacks of multiplexing too many sets of primers can include a loss of efficiency or inhibition due to primer-primer interaction. Despite these inherent intricacies, many detection methods for AdV have been published (15–18, 37, 38).

To evaluate whether one large multiplex reaction could replace the two separate reactions currently in use, DNA was amplified from all ATCC strains with each individual mix, triplex, pentaplex, and octaplex assay without changing the concentration of the primers/probes or the PCR conditions. One possibility is that the cross-reactivity between primers/probe sets might be cumulative and multiple amplifications occurring from different primers could result in earlier detection. This phenomenon does not seem to have a deleterious effect on qualitative detection but should be considered for quantification purposes. All AdV culture fluids were detected using the single-plex and octaplex versions of this assay. Overall each specific primer/probe set detected its own AdV type most efficiently, with the lowest C_T, suggesting that primer/probe sets show reasonable sensitivity for their given type. The exception was the primer/probe set designed against group D, which showed amplification for most AdV types. However, the C_Ts for group D were much lower than the C_Ts for other groups, suggesting they amplify group D most efficiently. The greatest cross-reactivity occurred between group B and D, which has been previously described (21).

Because we selected patient samples to study based on their positivity with our older qPCR multiplex assay, it is possible that the Jothikumar and Genetic Signatures assays might identify other adenovirus positives missed by the older qPCR assay. Identifying more rare strains would require the analysis of hundreds of additional patient samples, which was beyond the scope of this study. Unfortunately, some of the strains that have been identified by sequencing of patient samples (Adv 52 to 71, mostly strain D) are not readily available as either direct sample or cultured virus and so could not be tested in vitro but only analyzed by alignments of their reported DNA sequences (in silico).

In these studies, we identified both group D and group E samples whose quantities appeared to be better determined by the newer assays due to the higher quantitative values obtained. Accurate quantitation of all adenovirus strains is currently difficult, due to the lack of standardized materials. None of the ATCC strain materials have been accurately quantified, so quantitative standard/reference materials are currently limited to the ARMWG AD5 and commercially available panels such as the AcroMetrix panel which had an unknown strain type. Unfortunately, our Octaplex assay standardized with the ARMWG material gave significantly higher values for the samples from the CAP/QCMD surveys, leading to a lack of confidence in the quantitative values obtained. In the absence of available WHO standards for each of the known adenovirus groups, droplet digital PCR quantification of adenovirus would probably be the most accurate method to determine the true quantity of samples (39, 40). Given the complex nature of the Octaplex assay with the number of cross-reactions seen for some primers, this would probably best be done with individual assays with single-group primer/probe sets.

These results have critical implications for the development of international standards for adenovirus quantitation. There is currently a scarcity of reference material to harmonize the quantification of AdV between laboratories. Quantitative results need to be comparable across time and methods, either by using the same method, producing the same value across methods or by using reliable and stable conversions. Universally available standards and reference materials specific to quantitative molecular technologies are critical but few in number. The absence of adequate universal standards for test development and the use of laboratory-derived tests used in the absence of standardized industry-produced assays contribute to the discrepancies in reporting (41).

It is difficult to standardize among the numerous AdV groups when only one type is used for quantification. Commercial products on the market only contain one AdV type, and yet amplification efficiency for one type may not be carried over/inferred for other types. Current WHO International Standards address some of these needs, but there are limited organisms represented by the materials available (National Institute for Biological Standards and Control [http://www.nibsc.ac.uk]). We expect that these issues will be similar for the accurate quantitation of even more divergent viruses commonly encountered in the clinical virology laboratory such as enteroviruses and rhinoviruses (42).

In summary, we showed both the qPCR assay and the standardization material used for establishing a standard curve can affect AdV quantitative reporting. These results highlight the need for standardization of AdV quantitative standard material among laboratories performing AdV quantification regardless of the PCR method being used.