The Journal of Molecular Diagnostics, Vol. 17, No. 6, November 2015

jmd.amjpathol.org

Comparative Evaluation of Broad-Panel PCR Assays for the Detection of Gastrointestinal Pathogens in Pediatric Oncology Patients Zhengming Gu,* Haiqing Zhu,* Alicia Rodriguez,* Mohammad Mhaissen,y Stacey Schultz-Cherry,y Elisabeth Adderson,y and Randall T. Hayden* From the Departments of Pathology* and Infectious Diseases,y St. Jude Children’s Research Hospital, Memphis, Tennessee Accepted for publication June 2, 2015. Address correspondence to Randall T. Hayden, M.D., Department of Pathology, St. Jude Children’s Research Hospital, 262 Danny Thomas Place, Rm C5025, Mailstop #250, Memphis, TN 381052794. E-mail: randall.hayden@ stjude.org.

Broadly multiplexed molecular amplification assays offer an unprecedented ability to diagnose gastrointestinal infection in immunocompromised patients. However, little data are available to compare the performance of such systems in this population. A total of 436 stool samples were collected from 199 predominantly immunocompromised pediatric oncology patients. Remnant samples were tested in parallel with the use of the premarket (investigational use only) versions of two broadly multiplexed PCR assays (BioFire and Luminex), and the results of samples corresponding to the first episode per patient were compared with those from laboratory-developed molecular assays, culture, and antigen detection. Overall performance of the multiplexed systems was comparable, with BioFire and Luminex detecting 94 and 99 positives (P Z 0.34), respectively. Stratifying by analyte, BioFire assay detected 51 samples positive for Clostridium difficile, whereas Luminex assay detected 60 (P Z 0.01). Biofire and Luminex detected 28 and 38 norovirus-positive samples (P Z 0.002), respectively. Astrovirus- and adenoviruspositive samples were detected in higher numbers by in-house PCR than by BioFire; the same was observed for adenovirus with Luminex. Differences observed with other analytes were minimal, did not reach statistical significance, or lacked the numbers needed to detect a difference between systems. Broadly multiplexed PCR offers an effective means of detecting a variety of gastrointestinal pathogens in pediatric oncology patients, with assay performance comparable among the tests examined. (J Mol Diagn 2015, 17: 715e721; http://dx.doi.org/10.1016/j.jmoldx.2015.06.003)

Diarrhea is a common cause of morbidity and increased length of hospital stay among immunocompromised patients.1,2 The rapid and accurate diagnosis of infectious causes can have implications for antimicrobial stewardship, utilization of laboratory and other diagnostic testing services, length of stay, and clinical outcome.3,4 Conventionally, such testing is accomplished through a combination of several individual tests that use bacterial culture, antigen detection, and in some cases PCR.4,5 Together these tests can create a substantial burden in terms of workload and logistics for the clinical laboratory. Turnaround time varies markedly, depending on the analyte and methods used, and sensitivity may be poor. Broad panel PCR assays have recently become available. First marketed for testing of respiratory specimens,6e10 their Copyright ª 2015 American Society for Investigative Pathology and the Association for Molecular Pathology. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmoldx.2015.06.003

availability promises a paradigm shift in how we diagnose infectious diseases. These tests typically include 15 to 25 targeted pathogens, all detected rapidly and in some cases with minimal hands-on time by technologists.10,11 Tests available for respiratory pathogens have worked well, albeit with some variation in performance characteristics among different assays.7,11,12 Tests aimed at gastrointestinal pathogens have only recently become available,13e15 although at least one publication has recently compared such systems, the included patients (although not specified) appeared to represent a general medical Supported by ALSAC and BioFire. Disclosures: BioFire provided reagents and instrumentation. Luminex provided reagents. R.T.H. previously served on the Scientific Advisory Board for BioFire.

Gu et al population.14 The pediatric oncology population represents a unique demographic group with its own epidemiology, offering potentially different challenges to assay design and performance. At the same time, this group is perhaps uniquely suited to benefit from the availability of the rapid turnaround time of a highly sensitive test that simultaneously targets a wide range of infectious agents.

Materials and Methods Experimental Design A total of 436 stool samples were collected from 199 predominantly immunocompromised pediatric oncology patients who received routine diagnostic testing for gastrointestinal pathogens over an 18-month period (January 2010 through June 2011). After routine clinical testing, remnant samples were stored at 80 C for 30 to 44 months before use in this study. The study was approved by the Institutional Review Board of the St. Jude Children’s Research Hospital. Remnant samples were tested in parallel with the use of two broadly multiplexed PCR assays, including premarket (investigational use only or IUO) versions of the FilmArray Gastrointestinal Panel (BioFire Diagnostics, Salt Lake City, UT) and the xTAG Gastrointestinal Pathogen Panel (GPP) (Luminex Corporation, Austin, TX), and by real-time PCR assays for detection of human astrovirus, norovirus, and sapovirus (ANS). The IUO versions of the multiplexed assays differ from subsequently cleared versions by the U.S. Food and Drug Administration (in vitro diagnostic or IVD) primarily in that fewer organisms are reported with each of the IVD versions; those unique to each IUO test are indicated. The qualitative results generated from the two broad panel assays were compared with each other and with available results of inhouse standard-of-care tests and ANS PCR assays (standardof-care and ANS testing collectively referred to as IHT) (Table 1).

FilmArray Gastrointestinal Panel

Table 1

Targets Detected by BioFire, Luminex, and IHT

Target

BioFire

Luminex

IHT

Adenovirus F 40/41 Aeromonas Astrovirus Campylobacter Clostridium difficile Cryptosporidium Cyclospora cayetanensis Entamoeba histolytica EAEC EPEC ETEC lt/st Giardia lamblia Norovirus GI/GII Plesiomonas shigelloides Rotavirus Salmonella Sapovirus STEC stx1/stx2** Shigella Shigella/EIEC Vibrio Vibrio cholerae Yersinia enterocolitica

X Xy X X X X X X X X X X X X X X X X

X

X* Xz Xz Xz Xx X{

X X X X

X

X X X X

X X X X X X

X Xyy

Xzk Xzk Xzk X{ Xx Xz X{ Xz Xx X Xz Xk Xz Xz Xz

*Clinical testing detected all subtypes of adenovirus, not limited to types 40 and 41. y Not included in the IVD product BioFire. z Culture. x PCR. { EIA. k Clinical testing did not differentiate various subgroups for E. coli from a positive culture. **BioFire detects Shiga-like toxin-producing E. coli; Luminex detects STEC and E. coli O157; IHT detects target in routine culture methods and EIA. yy Not included in the IVD product for Luminex. EAEC, enteroaggregative E. coli; EIA, enzyme immunoassay; EIEC, enteroinvasive E.coli; EPEC, enteropathogenic E. coli; ETEC, enterotoxigenic E. coli; IHT, in-house testing; IVD, in vitro diagnostic; STEC, Shiga-like toxin-producing E.coli.

FilmArray system. Total run time was approximately 1 hour with approximately 5 minutes of hands-on time.

The FilmArray Gastrointestinal Panel (IUO) is a prefabricated pouch that contains lyophilized reagents capable of detecting 22 pathogens, including 13 bacteria, five viruses, and four protozoa (Table 1). Note that the research use only and IVD versions of the FilmArray assay differ only by which organisms were included in the cleared version. All other aspects of method design and analysis remain the same. Fecal specimen (about 1 g of solid or 1 mL of liquid stool specimens) was suspended in 15 mL of Cary Blair Transport Medium (remel, Lenexa, KS; per package insert) and mixed well. After incubation at room temperature for 30 minutes, an aliquot of 200 mL of sample suspension was used to prepare a FilmArray pouch, according to manufacturer’s instructions, and was loaded on a pouch for specimen processing, target amplification, and identification on the

xTAG GPP (Luminex) detects gastrointestinal pathogens, including eight bacteria, three viruses, and three protozoa in a single reaction (Table 1). The assay consists of RT-PCR, followed by probe hybridization and detection with the use of the xTAG universal tag sorting system on the MAGPIX instrument (Luminex). Fecal specimens (about 0.1 g solid or 0.1 mL liquid) were transferred into SK38 bead tubes (Luminex) that contained 1 mL of easyMag lysis buffer (bioMérieux Inc., Durham, NC) and mixed well with the addition of 10 mL of Luminex xTAG MS2 internal control. After incubation for 10 minutes at room temperature, the mixture was centrifuged at 14,000  g for 2 minutes. An aliquot of 200 mL of supernatant fluid was nucleic acid

716

jmd.amjpathol.org

xTAG GPP

-

The Journal of Molecular Diagnostics

Comparison of Broad-Panel PCR Assays Table 2

Primer and Probe Sequences of the In-House Tests

Virus, primer/probe

Sequence

Astrovirus 1e8 Forward primer Reverse primer Probe (California fluoresce orange 560) Astrovirus MLB1 Forward primer Reverse primer Probe (California fluoresce orange 560) Astrovirus VA2 Forward primer Reverse primer Probe (California fluoresce orange 560) Sapovirus Forward primer 1 Forward primer 2 Reverse primer Probe (6-carboxyfluorescein, 6-FAM)

Molecular targets and nucleotide position

50 -CCAGRCTCACAGAAGAGCAAC-30 50 -AAGAAGTGYGTAGGCTAGCAAG-30 50 -CATCGCATTTGGAGGGGAGGACC-30 50 -TGCGGTGTCGCAGACAAT-30 50 -TGTTGAAACTGAATGTGCAGTCTTT-30 50 -CAGCAACACTCGGCACCGTTGG-30 50 -TGTTACCTTCTGGTGARGTCACYAAA-30 50 -TGTCTTCTTCCYGCGGTTTGAA-30 50 -AACCCTTCAGGYCAGGTGTCAACAACT-30 50 -GACCAGGCTCTCGCYACCTAC-30 50 -TTGGCCCTCGCCACCTAC-30 50 -CCCTCCATYTCAAACACTAHTTTG-30 50 -TGGTTYATAGGYGGTAC-30

RNA-dependent RNA polymerase 4305e4325* 4351e4373* 4328e4350* Capsid protein 4034e4051y 4094e4118y 4053e4175y RNA-dependent RNA polymerase 3595e3620z 3680e3701z 3633e3659z RNA-dependent RNA polymerase 5074e5094x 786e803{ 5177e5154x 5101e5117x

*Positions in complete astrovirus genome (AF292073). y Positions in complete astrovirus MLB1 genome (FJ222451). z Positions in complete astrovirus VA2 genome (HQ674645). x Positions in complete sapovirus (NC_006269). { Positions in complete sapovirus (U95644).

extracted and eluted into 60-mL volume by using protocol A 1.0.2 on the NucliSENS easyMAG system (bioMérieux Inc.). An aliquot of 10 mL of extract was amplified, and the product was detected with fluorescently labeled beads specific to each target as described by Khare et al.14 Results were read on the MagPix instrument with the use of the supplied xPONENT software version 4.2 (Luminex). Total run time was approximately 6.5 hours with 2.5 hours of hands-on time, including the sample processing and data analysis.

IHTs Enteric bacterial culture with the use of standard methods was routinely performed in the clinical microbiology laboratory, together with Shiga-like toxin (ImmunoCard STAT! EHEC; Meridian Bioscience, Inc., Cincinnati, OH), rotavirus (Premier Rotaclone; Meridian Bioscience, Inc.), Cryptosporidium, and Giardia detection by enzyme immunoassay (ImmunoCard STAT!; Meridian Bioscience, Inc.), and Clostridium difficile by PCR (GeneXpert system; Cepheid, Sunnyvale, CA). Real-time PCR was used for detection of human ANS and adenovirus. Adenovirus was detected with a laboratory-developed test, multiplexed PCR assay that detected all serotypes (not limited to enteric adenovirus), as previously described.16 Detection of ANS was also performed with laboratory-developed test methods, described in the following section. Norovirus analytespecific reagent (ASR) primers and probes (Cepheid) were multiplexed with laboratory-developed primers and probes for detection of astrovirus types 1 to 8 and for detection of noncanonical strains MLB1 and VA2. The detection of noroviruses was verified with the norovirus ASR primer and probe in a multiplexed PCR assay to detect known norovirus GI (NVGI)

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

and GII (NVGII) strains. Sequence-confirmed astrovirus strains, types 1 to 8 and noncanonical strains MLB1 and VA2, were used to validate the assay for detection of these viruses in a multiplexed format. Analytical evaluation of these assays found 100% sensitivity and specificity (based on evaluation of reference strains). Sapovirus was detected with primers and probes on the basis of previously published sequences17 in a reaction separate from norovirus and astrovirus; the sensitivity and specificity of this assay were determined to be 95.7% and 100%, respectively. Both reactions were run in multiplex with internal control template and detection reagents (Cepheid). Primers and probes for detection of human astrovirus were selected from conserved regions of the viral genomes with the use of Primer Express software version 3.0 (Applied Biosystems, Foster City, CA). The nucleotide sequence and position of each primer and probe are listed in Table 2. Primers and probes for detection of human sapovirus were selected from the region of the RNA-dependent RNA polymerase gene17 with minor modifications. ASRs, NVGI, Table 3 Testing

Overall Results for BioFire, Luminex, and In-House

First episode per patient

BioFire

Luminex

In house*

Negative Positive Targets detected, n (%) 1 2 3 4

105 94

100 99

152 47

80 (85.1) 12 (12.8) 2 (2.1) 0

80 (80.8) 11 (11.1) 7 (7.1) 1 (1)

46 (97.9) 1 (2.1) 0 0

*Astrovirus, norovirus I and II, and sapovirus tested by PCR.

717

Gu et al statistic, with significance reaching P < 0.05. Exact McNemar’s test was used in cases when discordant cells were