1

Multicenter Evaluation of the Verigene Clostridium difficile Nucleic Acid Assay 1

Running title: Verigene C. difficile Assay 2

Karen C. Carroll1, 2 #, Blake W. Buchan3, Sokha Tan2, Paul D. Stamper1, Katherine M. Riebe3, 3

Preeti Pancholi4, Cheryl Kelly4, Arundhati Rao5, Robert Fader5, 4

Robert Cavagnolo6, Wendy Watson6, Richard V. Goering7, Ernest A. Trevino, Alice S. 5

Weissfeld, Ph.D8, Nathan A. Ledeboer3 6

1 Department of Pathology, the Johns Hopkins University School of Medicine, 2 Medical 7

Microbiology Laboratory, the Johns Hopkins Hospital, Baltimore, MD; 3Medical College of 8

Wisconsin, Dynacare Laboratories, Milwaukee WI; 4Ohio State University, Columbus OH; 9

5Scott and White Hospital, Temple TX; 6Medfusion, Lewisville, TX; 7Creighton University 10

Medical Center, Omaha, NE, 8Microbiology Specialists, Houston, TX 11

12

# Corresponding Author: 13

Karen C. Carroll, M.D. 14

Professor Pathology 15

The Johns Hopkins University School of Medicine 16

Meyer B1-193 17

600 N. Wolfe St. 18

Baltimore MD 21287 19

Kcarrol7@jhmi.edu 20

410-955-5077 21

FAX 410-614-8087 22

JCM Accepts, published online ahead of print on 2 October 2013J. Clin. Microbiol. doi:10.1128/JCM.01690-13Copyright © 2013, American Society for Microbiology. All Rights Reserved.

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

2

23

Abstract 24

25

The Verigene Clostridium difficile Nucleic Acid Test (Verigene CDF Test) (Nanosphere, 26

Northbrook, IL) is a multiplex qualitative PCR assay that utilizes a nanoparticle-based array 27

hybridization method to detect C. difficile tcdA and tcdB in fecal specimens. In addition, the 28

assay detects binary toxin gene sequences and the single base pair deletion at nucleotide 117 (ǻ 29

117) in tcdC to provide a presumptive identification of the epidemic strain ribotype 027. This 30

study compared the Verigene CDF Test with anaerobic direct and enriched toxigenic culture on 31

stool specimens from symptomatic patients among five geographically diverse laboratories 32

within the USA. The Verigene CDF test was performed according to the manufacturer’s 33

instructions, and the reference methods performed by a central laboratory included direct culture 34

onto cycloserine cefoxitin fructose agar (CCFA) and enriched culture using cycloserine cefoxitin 35

mannitol broth with taurocholate and lysozyme. Recovered isolates were identified as C. 36

difficile using gas liquid chromatography and were tested for toxin using a cell culture 37

cytotoxicity neutralization assay. Strains belonging to ribotype 027 were determined by PCR 38

ribotyping and bidirectional sequencing for ǻ 117 in tcdC. A total of 1,875 specimens were 39

evaluable. Of these, 275 specimens (14.7%) were culture positive by either direct or enriched 40

culture methods. Compared to direct culture alone, the overall sensitivity, specificity, positive 41

predictive value, and negative predictive value for the Verigene CDF test were: 98.7%, 87.5%, 42

42%, and 99.9%, respectively. Compared to combined direct and enriched culture results, the 43

sensitivity, specificity, positive predictive value, and negative predictive values of the Verigene 44

CDF test were: 90.9%, 92.5%, 67.6%, and 98.3%, respectively. Of the 250 concordantly culture 45

positive specimens, 59 (23.6%) were flagged as “hypervirulent”; 53 were confirmed as ribotype 46

027 and all 59 possessed ǻ 117 in tcdC. Time to results was approximately 2.5 h per specimen. 47

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

3

The Verigene CDF Test is a novel nucleic acid microarray that reliably detects both C. difficile 48

toxins A and B in unformed stool specimens and appears to adequately identify ribotype 027 49

isolates. 50

51

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

4

Introduction 52

The optimum testing method for the diagnosis of Clostridium difficile (C. difficile) is still 53

a matter of debate. A variety of testing methods are available and include enzyme 54

immunoassays (EIAs) for detection of toxins A and B, cell culture cytotoxicity neutralization 55

assays (CCNAs), anaerobic toxigenic culture, detection of glutamate dehydrogenase and nucleic 56

acid amplification tests (NAATs). Although EIAs provide same day results, are relatively 57

inexpensive, and easy to perform, there is general agreement in the USA and in some European 58

countries that EIAs should not be performed as stand-alone tests due to their lack of sensitivity 59

(1, 2). CCNAs and anaerobic toxigenic culture do not provide timely results and have never been 60

standardized. Laboratories are beginning to adopt nucleic acid amplification tests (NAATs) 61

alone or as a component of multi-test algorithms that include screening with glutamate 62

dehydrogenase (GDH) (3-5). Such strategies are endorsed by various professional societies in 63

their C. difficile testing and treatment guidelines (1, 6). 64

Nucleic acid amplification tests (NAATs) have evolved as improved fecal extraction 65

protocols and real-time polymerase chain reaction (PCR) methods have overcome inhibition 66

accompanied with improved assay performance (4, 7). Currently there are nine FDA-cleared 67

NAATs available in the USA with several more in development (8-14). The majority of these 68

are real-time PCR assays that detect the toxin B gene (tcdB) and include the BD-GeneOhm Cdiff 69

assay (BD-GeneOhm, San Diego CA), the GeneXpert Cdiff and Cdiff Epi tests, (Cepheid, Inc. 70

Sunnyvale, CA), the Prodesse ProGastro CD assay (Hologic Gen-Probe, Inc. San Diego, CA) 71

and the Simplexa Universal Direct assay (Focus Technologies, Inc. Cypress, CA). The 72

GeneXpert Epi assay is a multiplex PCR method that not only detects tcdB, but also detects 73

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

5

binary toxin genes and the single base pair deletion at nucleotide 117 (ǻ 117) in tcdC to provide 74

presumptive identification of the epidemic strain 027/NAP1/BI. 75

Several tests are based upon other novel chemistries including loop-mediated isothermal 76

amplification such as the illumigene assay (Meridian Bioscience, Inc., Cincinnati, OH) (11, 12), 77

helicase dependent amplification in the Portrait C. difficile assay (Great Basin, Inc., Salt Lake 78

City, UT) (13) and AmpliVue C. difficile assay (Quidel Molecular, Inc., San Diego CA), and 79

array technologies, Verigene Cdiff assay (Nanosphere, Inc. Northbrook IL). Several of these 80

assays detect conserved regions of the toxin A gene (tcdA) either alone (illumigene) [11, 12] or 81

in combination with other genetic targets (Verigene CDF test). One manufacturer, Luminex, Inc. 82

Austin, TX has included C. difficile in a multiplex assay that detects other enteric pathogens 83

(xTAG Gastrointestinal Pathogen Panel). 84

The Verigene CDF Test is a multiplex qualitative assay that utilizes PCR-amplified DNA 85

in a nanoparticle-based microarray to detect C. difficile tcdA and tcdB in fecal specimens. In 86

addition, the assay detects binary toxin gene sequences and the single base pair deletion at 87

nucleotide 117 in tcdC to provide presumptive identification of the epidemic strain 027/NAP1/BI 88

(henceforth referred to as ribotype 027). This study evaluated and established the analytical 89

performance of the Verigene CDF Test using the Verigene System compared to anaerobic 90

toxigenic culture, ribotyping and bidirectional sequencing as the reference methods. 91

(This research was presented in part at the 113th General Meeting of the American Society for 92

Microbiology, Denver, CO. and in part at the 23rd European Congress of Clinical Microbiology 93

and Infectious Diseases, Berlin, Germany). 94

Materials and Methods 95

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

6

Participating Centers and Overall Study Design: This was a prospective study from February 96

28, 2012 until June 27, 2012 to assess the analytical performance of the Verigene CDF Test 97

among five geographically diverse clinical microbiology laboratories. These included: The Johns 98

Hopkins Hospital Microbiology Laboratory, Baltimore, MD; The Medical College of Wisconsin 99

Milwaukee, WI; Ohio State University, Columbus, OH; Scott and White Hospital, Temple, TX; 100

and Medfusion Laboratories, Inc. Lewisville, TX. Each site obtained appropriate institutional 101

review board approval or waiver consistent with local human subjects research requirements. 102

All sites performed Verigene CDF testing on leftover de-identified stool samples submitted to 103

the clinical laboratory specifically for C. difficile testing according to the institution’s routine 104

practices. In addition, all sites reserved an aliquot of the stool for shipping to a central reference 105

laboratory, Microbiology Specialists, Inc. (Houston, TX) for performance of anaerobic toxigenic 106

culture methods (see reference methods section). Toxin-producing isolates were sent to 107

independent laboratories for PCR ribotyping and bidirectional sequencing (see PCR Ribotyping 108

and Bi-directional sequencing section). 109

Specimen collection and handling: A minimum of 2 mL of liquid stool or a marble-size of soft 110

stool specimen was required for testing. Each study site de-identified the fresh prospectively-111

collected specimen submitted to the laboratory for clinical testing by assigning each a unique 112

specimen identification number. Duplicate specimens from the same patient were not enrolled. 113

The collection time and date, patient age, and consistency of the specimen were recorded for 114

each specimen. Enrolled specimens were thoroughly mixed in the original container for 5 sec. 115

and then > 2 mL of specimen were pipetted into a 5 mL conical tube (tube A) for culture and 116

molecular testing within 48 h of collection. Specimens were kept at 2-8ºC until tested. A 0.5 117

mL aliquot of liquid stool or a pea-size soft stool specimen in tube A was transferred to 118

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

7

Anaerobe Tissue Transport Medium (ATTM) at the time of Verigene testing, (Anaerobe 119

Systems, Inc. Morgan Hill, CA). The ATTM was stored at room temperature and shipped each 120

weekday in an insulated shipping container to Microbiology Specialists (MSI), Inc. for anaerobic 121

toxigenic culture. 122

Reference Culture Methods: Upon arrival to MSI, the stool was inoculated onto pre-reduced 123

CCFA (cycloserine cefoxitin fructose agar) [CCFA-D] and CCMB-TAL broth (cycloserine 124

cefoxitin mannitol broth with taurocholate and lysozyme) (Anaerobic Systems). The CCFA and 125

CCMB-TAL broth were incubated anaerobically for 48 h. All TAL broths were then subcultured 126

to CCFA and CCFA-HT (CCFA with horse blood and taurocholate, CCFA-E). These plates 127

were subsequently incubated under anaerobic conditions for 48 h. On CCFA, colonies of C. 128

difficile exhibit a yellow, ground glass appearance and chartreuse fluorescence under a Wood’s 129

lamp. On CCFA-HT, colonies of C. difficile exhibit a grey, ground glass appearance. A Gram 130

stain was performed on suspicious colonies; large gram-positive rods were further tested for 131

aerotolerance and susceptibility to vancomycin. 132

Any vancomycin-susceptible, strictly anaerobic, large gram-positive rods were placed in 133

chopped meat carbohydrate (CMC) broth for 48 h. Part of the broth was then extracted for gas 134

liquid chromatography (GLC); the presence of isovaleric, isocaproic and isobutyric acids in the 135

volatile product represented a definitive identification for C. difficile. Another portion of the 136

CMC was centrifuged, and the supernatant was then removed and used to determine the presence 137

of toxin by inoculation onto human fetal foreskin cells. Toxin was neutralized with specific 138

antitoxin to C. difficile (Diagnostic Hybrids, Athens, OH) to confirm the presence of toxigenic C. 139

difficile (10). 140

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

8

For the purposes of this study, if C. difficile was isolated from the CCFA-D plate and the 141

isolate was positive by the cell cytotoxicity assay, the specimen was classified as “toxigenic C. 142

difficile positive” and the CCFA-E plate was not further analyzed. If no C. difficile was isolated 143

from the CCFA-D plate or if the isolate was negative by the cell culture cytotoxicity assay, the 144

CCFA-E plate was further analyzed. If CCFA-E was positive for C. difficile and the isolate was 145

positive for cell cytotoxicity assay, the specimen was classified as “toxigenic C. difficile 146

positive”. The specimen was reported as “negative” if CCFA-E was negative for C. difficile or 147

the isolate was tested negative by the cell cytotoxicity assay. 148

Description of the Verigene CDF Assay: 149

The Verigene System is a bench-top sample-to result platform consisting of two modules: 150

the Verigene Processor SP and the Verigene Reader. The Processor SP automates specimen 151

nucleic acid extraction, PCR amplification, and amplicon hybridization. The Processor SP 152

utilizes four single-use consumables per test: an extraction tray, amplification tray, test cartridge, 153

and pipette tip holder assembly. The user tests a specimen by loading the single-use consumables 154

into the Processor SP, pipetting the test specimen into the sample well of the extraction tray, and 155

initiating the test protocol on the Reader by scanning the test cartridge barcode and specimen 156

information. Upon test completion, the user inserts the microarray slide from the test cartridge 157

into the Reader for optical analysis and generation of test results. 158

For C. difficile testing on the Verigene System, 100 ȝl of the prepared stool sample is 159

loaded into the sample well of the extraction tray. The test specimen is prepared from a soft or 160

liquid stool sample by swabbing the stool with the provided flocked swab and transferring the 161

evenly coated swab to the provided stool prep buffer tube, which is subsequently vortexed for 15 162

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

9

seconds and centrifuged for 30 seconds. Following test initiation on the Processor SP, sample 163

bacteria are lysed and target DNA is extracted using a magnetic bead-based extraction. 164

Extraction is followed by multiplex PCR-based amplification, which generates target specific 165

amplicons. Finally, amplicon hybridization to target specific capture DNA occurs in a 166

microarray setting using mediator and gold nanoparticle probe hybridization to the captured 167

amplicons. Silver enhancement of the bound gold nanoparticle probes at the capture sites results 168

in gold-silver aggregates that are assessed optically on the Reader to produce the final test result. 169

Controls 170

A negative control is included in the array to detect whether the melt temperature of the 171

hybridization chamber is too low due to an instrument malfunction. The Verigene Reader 172

automatically checks the target oligonucleotide signal relative to the negative control 173

oligonucleotide signal. If the target signal is not adequately higher than the negative control 174

signal, a “no call” result will be generated. 175

An artificial DNA construct serves as a hybridization control and is referred to as 176

Processing Control 1 (PC1). This control material along with the primers, enzymes and 177

amplification buffers are included within the amplification tray. If the process control is not valid 178

a “no call” result will be obtained and the test should be repeated. Bacillus subtilis serves as a 179

specimen preparation and amplification control and is referred to as Processing Control 2 (PC2). 180

This control is automatically added by the Processor SP to each specimen prior to the extraction 181

step. If the process control is not valid, a “no call” result will be obtained and the test should be 182

repeated. Additional positive controls are immobilized on the test slide and are used to 183

determine that hybridization was performed correctly. The Verigene CDF Test algorithm 184

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

10

requires that these controls be valid before decisions regarding the presence or absence of any 185

other target on the panel can be determined. If these controls are not detected a “no call” result 186

will be obtained and the test is repeated. 187

Verigene CDF Test Performance: For liquid samples, 150 µL of the specimen from Tube A 188

was transferred to a small screw capped microcentrifuge tube containing stool lysis buffer. For 189

soft stool specimens, a sterile swab was fully immersed in the specimen and once evenly coated, 190

it was transferred to the stool lysis buffer tube and the swab was broken at the preformed score 191

breakpoint. In both cases the screw cap was gently tightened by hand. All specimen transfers 192

were performed in such a manner as to ensure a minimum risk of contamination. Gloves were 193

changed between pipetting and transfer of specimens and only one specimen was open and 194

processed at a time. All sample buffer tubes containing specimen were vortexed for 15 sec. and 195

microcentrifuged for 30 sec., after which 100 µL of supernatant were inoculated into the 196

specimen well of the extraction tray. Testing was performed in a blinded manner on the de-197

identified aliquots according the manufacturer’s protocol and product package insert. 198

Three external positive controls were supplied to each test site for quality control 199

purposes. These external positive controls were run on an alternating basis once every 24 h 200

during the period of time that clinical testing was conducted (e.g. CDF Control 1 was run on Day 201

1 of testing, CDF Control 2 on Day 2, and so on) with the goal being an approximately equal 202

distribution of each CDF control run during the study at each site. Several test-specific negative 203

controls are immobilized on the Test Slide and are used to guide test decisions (see Description 204

of the Verigene Test). A single external negative control (which is a non-toxigenic C. difficile) 205

was run every 24 h during the period when clinical specimen testing was conducted. 206

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

11

Results interpretation: The Verigene CDF assay provides a qualitative result for the presence 207

(detected) or absence (not detected or not applicable) of tcdA, tcdB, binary toxin genes and the 208

ǻ117 deletion in tcdC. Table 1 demonstrates the possible outcomes for valid test results. Invalid 209

results may be generated by failure to image the array slide (No Call-No Grid) or from internal 210

control failures or because the Reader could not read the target specific signals due to variability. 211

Other sources of invalid results include pre-analytical errors. All invalid results were repeated 212

within 48 h of specimen collection from a refrigerated reserved aliquot of the original stool 213

sample. 214

PCR Ribotyping: Following the culture testing at MSI, the toxigenic C. difficile positive isolates 215

were sent for strain identification by PCR-Ribotyping to an external third-party site (Creighton 216

University Medical Center, Omaha, NE) as per the study protocol. In addition to conventional 217

agarose-gel electrophoresis, for added discrimination, PCR products were analyzed using the 218

Agilent 2100 Bioanalyzer (Agilent Technologies, Clara, CA). The strains were designated into 219

two PCR-Ribotyping categories, 027 and non-027. Strain type assignments were based on a 220

comparison of isolate profiles with known C. difficile reference strains from the C. difficile 221

collection of the Cardiff-European Centre for Disease Prevention and Control [ECDC] obtained 222

from Dr. Ed Kuijper, Leiden University Medical Center, Netherlands using the BioNumerics 223

software, version 6.6 (Applied Maths, Belgium). 224

Bi-directional sequencing: During the study, DNA from the Extraction step of the Verigene 225

CDF Test was saved and stored frozen at -20 °C until needed. If the PCR amplification for tcdC 226

failed for a DNA sample, the DNA extracted during the PCR-Ribotyping was 227

tested. Sequencing templates were prepared by PCR amplification of the stored DNA using 228

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

12

sequencing primers. The same set of primers was also used for sequencing the amplified 229

material. In parallel, following central culture testing, DNA from the culture-confirmed C. 230

difficile positive isolates (that was extracted during the Verigene CDF Test) was sent to an 231

external third-party site (ACGT, Inc, Wheeling, IL) for tcdC bi-directional sequencing as per the 232

protocol. A 295 bp fragment of tcdC was sequenced. 233

Data analysis and statistical methods: Results of the Verigene CDF Test were compared to 234

direct culture on CCFA with isolates confirmed and analyzed independently by ribotyping and 235

bi-directional sequencing. In addition, the Verigene CDF Test was also compared to enriched 236

culture, with positives verified by ribotyping and bi-directional sequencing (see Reference 237

Culture Methods sections). 238

Results were entered into EXCEL (Microsoft Corporation, Redmond, Washington) and 239

the sensitivity, specificity, positive predictive and negative predictive values were calculated. 240

The exact binomial methods in SAS 9.1.3 (SAS Institute, Inc., Cary, North Carolina) were used 241

to calculate the lower and upper two-sided confidence limits. 242

RESULTS 243

Among the five study sites 1,877 specimens met the inclusion criteria. Seventy one 244

specimens (3.8%) required repeat testing: 46 specimens (2.4%) had an initial “No Call” result 245

due to assay internal control errors; 17 specimens (1.0%) had an initial “Indeterminate” call (No 246

Call-IND); and eight specimens (0.4%) had pre-analytical errors (four motor stalls, two tip 247

failures, one cracked slide and one cartridge not detected). The eight specimens that experienced 248

pre-analytical errors and the 46 “No Call” specimens all produced a valid call upon repeat testing 249

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

13

as per the protocol. However, two of the “No Call” specimens required a second repeat test. 250

Repeat testing of the 17 No Call-IND specimens called all but two specimens. Therefore, two 251

specimens had a final “Indeterminate” call and were not included in the clinical data analysis of 252

evaluable results. Thus, 1,875 specimens were analyzed in this clinical evaluation. 253

Table 2 summarizes the overall Verigene CDF Test clinical performance. Of the 1,875 254

specimens that were analyzed in this trial, there were 275 specimens positive for C. difficile and 255

1600 negative specimens by either direct or enriched culture methods after all repeat testing. The 256

majority (98.8%) of positive specimens contained genes for both toxins A and B (data not 257

shown). 258

The overall agreement between direct culture and the Verigene CDF Test for detection of 259

toxigenic C. difficile in stool was 88.4%. The overall sensitivity, specificity, positive predictive 260

values and negative predictive values of the Verigene CDF Test compared to direct culture on 261

CCFA are listed in Table 2. Overall agreement between the combined direct and enriched 262

culture results and the Verigene CDF assay was 92.3% (Table 2). 263

Table 3 lists the performance characteristics of the Verigene CDF Test compared to 264

combined culture methods stratified by investigational site. There were no major differences in 265

the performance of the Verigene assay by investigational site. 266

There were 72 non-toxigenic C. difficile isolates recovered by culture during the study. 267

Eleven of these were detected as toxigenic by the Verigene CDF test and were classified as false 268

positives. Nine of the eleven false positive results were confirmed positive for tcdC by bi-269

directional sequencing. Two of the 11 false positive results were negative for tcdC and tcdB by 270

bi-directional sequencing (data not shown). 271

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

14

Table 4 shows the distribution by Verigene CDF Test results for strain characterization. 272

Of the 250 concordantly culture and Verigene positive specimens, 59 (23. 5%) were flagged as 273

“hypervirulent”; 53 were confirmed as ribotype 027 and all 59 possessed the ǻ 117 mutation in 274

tcdC. One isolate that was ribotype 027 and contained the ǻ117 tcdC mutation was incorrectly 275

identified as non-O27 by the Verigene CDF Test. Compared to ribotyping, the Verigene CDF 276

Test correctly assigned the ribotype in 89.7% of the cases and when compared to sequencing the 277

percent agreement was 100 %. 278

DISCUSSION 279

Clostridium difficile remains a formidable pathogen that continues to cause significant 280

morbidity and mortality among hospitalized patients (15). Between 2001 and 2006, the 281

incidence of C. difficile infection in the USA more than doubled and the incidence continues 282

to rise in North America and Europe (15). In part, this increase was due to the appearance of 283

“hypervirulent’ strains such as ribotype 027. In addition to being more transmissible and a 284

hyperproducer of toxins A and B, this ribotype is notably more drug resistant compared to 285

historical strains (15). Increased severity of illness has resulted in more colectomies, 286

prolonged hospitalizations, and an increase in the observed number of recurrences (16). 287

More significantly, mortality rates have also increased and in North America and Europe are 288

reported to be as high as 22-37% (15). The economic burden is substantial, as was shown in 289

a systematic review of healthcare costs associated with patients with primary and recurrent 290

CDI (17). In 2008 US dollars, the additional costs for treating relatively mild primary CDI 291

varied from $9,822 to $13,854 (17). In Europe, the incremental increase in costs ranges in 292

Euros from €4067 to €9276 (15). 293

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

15

Over the last decade, the emergence of “hypervirulent strains” such as ribotype 027, 294

ribotype 078 and the newly described ribotypes 176 and 198 (18, 19), have spurred many 295

clinical laboratories to abandon insensitive EIAs for more sensitive methods such as NAATs. 296

The published performance characteristics among the nine FDA-cleared NAATs varies 297

depending upon the comparative method, but in general those reports that use toxigenic 298

culture as the reference method have demonstrated sensitivities ranging from 77% to 100% 299

and specificities that range from 90-100% (4,7,8,11-14, 20). The Verigene CDF test is one 300

of the newer assays to obtain FDA 510(k) clearance and is novel. The assay uses multiplex 301

PCR combined with silver enhancement of gold nanoparticle probe hybridization to capture 302

amplicons in a microarray format. The Verigene CDF test is the only FDA-cleared assay that 303

detects both toxins A and B and identifies ribotype 027 by identifying both the binary toxin 304

gene and the ǻ117 deletion of tcdC. In terms of performance characteristics, the overall 305

sensitivity and specificity of 98.7% and 87.5%, respectively compared to direct toxigenic 306

culture and 90.9% and 92.5%, respectively compared to enriched culture are comparable to 307

the published clinical performance of other NAATs (8-14). As has been noted for other 308

NAATs, the NPV for the Verigene assay ranges from 98-99.9% and essentially excludes C 309

difficile as a cause of diarrheal disease, whereas the PPV is low ranging from 42-67% 310

compared to toxigenic culture methods. Some of the false positive samples were confirmed 311

by bi-directional sequencing as containing toxin-producing strains indicating possible mixed 312

infection with toxigenic and non-toxigenic isolates. A limitation of this study is that no 313

clinical data was collected and included in the assessment of a true positive or negative 314

specimen. The percentage of combined “No-Call”, “Invalid”, and pre-analytical problems of 315

3.9% necessitating repeat testing is also similar to what has been reported for the BD 316

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

16

GeneOhm assay’s invalid rate of 0.5-7.3% (8, 12, 14), the Prodesse ProGastro CD invalid 317

rate of 2.7% (9, 14), and the illumigene rate of 0.8-4.4% (11, 12). This is slightly higher than 318

what is reported for the Quidel AmpliVue C. difficile assay (0.5%) [20]. 319

The usefulness of detecting ribotype 027 in any C. difficile assay remains controversial. 320

While some studies have demonstrated more severe disease and increased mortality associated 321

with ribotype 027 (16, 19), more recent studies have challenged this observation (21, 22). In the 322

latter studies, when other covariates were considered, the ribotype of the infecting organism was 323

not found to be a significant predictor of severe CDI (21). Currently, there are no 324

recommendations to alter infection control practices based upon ribotype. Likewise, there are 325

no formal recommendations to alter treatment based solely on the strain causing the infection. 326

Some strain types may correlate with an increased risk of recurrent disease (23). In the initial 327

fidaxomicin treatment study, the data showed that the risk of relapse may vary depending upon 328

the C. difficile strain type (24). For example, the rate of recurrence of non-027 CDI was 329

significantly lower in the fidaxomicin treated group compared to the vancomycin treated patients 330

(24). A more recent meta-analysis comparing fidaxomicin to vancomycin showed that 331

fidaxomicin reduced recurrence or death by 40% compared with vancomycin through day 40 332

(25). This observation did not hold true for ribotype 027 strains, where the rates of recurrence 333

with each agent were not different (24, 25). Considering the cost of this drug, a practical use for 334

real-time strain typing could be to include the ribotype information when triaging which patients 335

are best suited to receive fidaxomicin therapy. 336

One of the limitations to the Verigene system is the fact that it is a “one sample at a time” 337

method and the technologist has to manually remove the cartridge from the Processor into the 338

reader. While this allows for easy “on demand” testing, such a system is not amenable to large 339

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

17

through-put laboratories. The instruments can be stacked and one Reader can accommodate up 340

to 32 Processors, but this requires space and may be operationally challenging. 341

In summary, the Verigene CDF assay is a novel PCR microarray test that is unique 342

among the current FDA-cleared platforms in its simultaneous detection of the genes that encode 343

toxins A and B, binary toxin and the ǻ 117 mutation in tcdC as surrogate markers for ribotype 344

027. In comparison to published performance characteristics of the other NAATs, the Verigene 345

test performs on the high end of the range of reported sensitivities. The unresolved specificity is 346

a bit lower as might be expected for a more sensitive assay. Finally the assay correlates well 347

with PCR-ribotyping and bidirectional sequencing in identifying ribotype 027 in clinical 348

samples. 349

ACKNOWLEDGEMENTS 350

The authors thank ACGT, Inc. under the direction of Semyon Rubinchik, Ph.D. for performing 351

the genetic analysis of the recovered isolates and Jon Marion and Niku Falandino from 352

Nanosphere, Inc. who assisted with the data analysis. In addition, all authors are grateful to the 353

staff of their clinical laboratories for their support of this study. This study was sponsored by 354

Nanosphere, Inc. NAL and BWB have served as consultants to Nanosphere. 355

REFERENCES 356

1. Cohen, SH, Gerding DN, Johnson S, Kelly CP, Loo VG, McDonald LC, Pepin J, 357

Wilcox MH. 2010. Clinical practice guidelines for Clostridium difficile infection in 358

adults: 2010 update by the Society for Healthcare Epidemiology of America (SHEA) 359

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

18

and the infectious diseases society of America (IDSA). Infect.Control Hosp.Epidemiol. 360

31:431-455. 361

2. Wilcox MH. 2012. Overcoming barriers to effective recognition and diagnosis of 362

Clostridium difficile infection. Clin.Microbiol.Infect. 18 Suppl 6:13-20. 363

3. Sharp SE, Ivie WM, Buckles MR, Coover DM, Pohl JC, Hatcher PA. 2009. A 364

simple 3-step algorithm for improved laboratory detection of Clostridium difficile toxin 365

without the need for tissue culture cytotoxicity neutralization assays. Diagn. Microbiol. 366

Infect. Dis. 64:344-346. 367

4. Carroll KC. 2011. Tests for the diagnosis of Clostridium difficile infection: the next 368

generation. 369

5. Vasoo S, Stevens J, Portillo L, Barza R, Schejbal D, Wu MM, Chancey C, Singh 370

K. 2012. Cost-effectiveness of a modified two-step algorithm using a combined 371

glutamate dehydrogenase/toxin enzyme immunoassay and real-time PCR for the 372

diagnosis of Clostridium difficile infection. J. Microbiol. Immunol. Infect. Aug 23 373

[Epub ahead of print]. 374

6. Advisory Committee on Antimicrobial Resistance and Healthcare Associated 375

Infection (ARHAI). 2012. Updated guidance on the diagnosis and reporting of 376

Clostridium difficile. http://www.dh.gov.uk/publications. 377

7. Belanger SD, Boissinot M, Clairoux N, Picard FJ, Bergeron MG. 2003. Rapid 378

detection of Clostridium difficile in feces by real-time PCR. J. Clin. Microbiol. 41:730-379

734. 380

8. Stamper PD, Alcabasa R, Aird D, Babiker W, Wehrlin J, Ikpeama I, Carroll KC. 381

2009. Comparison of a commercial real-time PCR assay for tcdB detection to a cell 382

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

19

culture cytotoxicity assay and toxigenic culture for direct detection of toxin-producing 383

Clostridium difficile in clinical samples. J.Clin. Microbiol. 47:373-378. 384

9. Stamper PD, Babiker W, Alcabasa R, Aird D, Wehrlin J, Ikpeama I, Gluck L, 385

Carroll KC. 2009. Evaluation of a new commercial TaqMan PCR assay for direct 386

detection of the Clostridium difficile toxin B gene in clinical stool specimens. 387

J.Clin.Microbiol. 47:3846-3850. 388

10. Novak-Weekley SM, Marlowe EM, Miller JM, Cumpio J, Nomura JH, Vance PH, 389

Weissfeld A. 2010. Clostridium difficile testing in the clinical laboratory by use of 390

multiple testing algorithms. J. Clin. Microbiol. 48:889-893. 391

11. Noren T, Alriksson I, Andersson J, Akerlund T, Unemo M. 2011. Rapid and 392

sensitive loop-mediated isothermal amplification test for Clostridium difficile detection 393

challenges cytotoxin B cell test and culture as gold standard. J. Clin. Microbiol. 49:710-394

711. 395

12. Viala C, Le MA, Maataoui N, Rousseau C, Collignon A, Poilane I. 2012. 396

Comparison of commercial molecular assays for toxigenic Clostridium difficile 397

detection in stools: BD GeneOhm Cdiff, XPert C. difficile and illumigene C. difficile. 398

J. Microbiol. Methods 90:83-85. 399

13. Hicke B, Pasko C, Groves B, Ager E, Corpuz M, Frech G, Munns D, Smith W, 400

Warcup A, Denys G, Ledeboer NA, Lindsey W, Owen C, Rea L, Jenison R. 2012. 401

Automated detection of toxigenic Clostridium difficile in clinical samples: isothermal 402

tcdB amplification coupled to array-based detection. J. Clin. Microbiol. 50:2681-2687. 403

14. Bouza E. 2012. Consequences of Clostridium difficile infection: understanding the 404

healthcare burden. Clin. Microbiol. Infect. 18 (Suppl.6): 5-12. 405

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

20

15. Carroll KC, Bartlett JG. 2011. Biology of Clostridium difficile: implications for 406

epidemiology and diagnosis. Annu. Rev. Microbiol. 65:501-521. 407

16. Loo VG, Poirier L, Miller MA, Oughton M, Libman MD, Michaud S, Bourgault 408

AM, Nguyen T, Frenette C, Kelly M, Vibien A, Brassard P, Fenn S, Dewar K, 409

Hudson TJ, Horn R, René P, Monczak Y, Dascal A. 2005. A predominantly clonal 410

multiinstitutional outbreak of Clostridium difficile-associated diarrhea with high 411

morbidity and mortality. N. Engl. J. Med. 353:2442-2449. 412

17. Ghantoji SS, Sail K, Lairson DR, Dupont HL, Garey KW. 2010. Economic 413

healthcare costs of Clostridium difficile infection: a systematic review. J. Hosp. Infect. 414

74:309-310. 415

18. Quidel. Quidel Molecular AmpliVue C difficile product package insert. Quidel 416

Molecular, San Diego CA. December 2012. 417

19. Walker AS, Eyre DW, Wyllie DH, Dingle KE, Griffiths D, Shine B, Oakley S, 418

O’Connor L, Finney J, Vaughan A, Crook DW, Wilcox MH, Peto TEA on behalf 419

of the Infections in Oxfordshire Research Database (IORD). 2013. Relationship 420

between bacterial strain type, host biomarkers, and mortality in Clostridium difficile 421

infection. Clin. Infect. Dis. [Epub ahead of print April 10, 2013]. 422

20. Valiente E, Dawson LF, Cairns MD, Stabler RA, Wren BW. 2012. Emergence of 423

new PCR ribotypes from the hypervirulent Clostridium difficile 027 lineage. J. Med. 424

Microbiol. 61:49-56. 425

21. Walk ST, Micic D, Jain R, Lo ES, Trivedi I, Liu EW, Almassalha LM, Ewing SA, 426

Ring C, Galecki AT, Rogers MA, Washer L, Newton DW, Malani PN, Young V B, 427

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

21

Aronoff DM. 2012. Clostridium difficile ribotype does not predict severe infection. 428

Clin. Infect. Dis. 55:1661-1668. 429

22. Sirard S, Valiquette L, Fortier LC. 2011. Lack of association between clinical 430

outcome of Clostridium difficile infections, strain type, and virulence-associated 431

phenotypes. J. Clin. Microbiol. 49:4040-4046. 432

23. Figueroa I, Johnson S, Sambol SP, Goldstein EJ, Citron DM, Gerding DN. 2012. 433

Relapse versus reinfection: recurrent Clostridium difficile infection following treatment 434

with fidaxomicin or vancomycin. Clin.Infect.Dis. 55 Suppl 2:S104-S109. 435

24. Louie TJ, Miller MA, Mullane KM, Weiss K, Lentnek A, Golan Y, Gorbach S, 436

Sears P, and Shue YK. 2011. Fidaxomicin versus vancomycin for Clostridium difficile 437

infection. N. Engl. J. Med. 364:422-431. 438

25. Crook DW, Walker AS, Kean Y, Weiss K, Cornely OA, Miller MA, Esposito R, 439

Louie TJ, Stoesser NE, Young BC, Angus BJ, Gorbach SL, Peto TEA. for the 440

Study 003/004 Teams. 2012. Fidaxomicin versus vancomycin for Clostridium 441

difficile infection: Meta-analysis of pivotal randomized controlled trials. Clin. Infect. 442

Dis. 55(Suppl 2): S93-S101. 443

444

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Table 1: Interpretation of Valid Verigene CDF Test Results

tcdA tcdB Binary tcdC Result

Not detected Not Detected N/A N/A Toxigenic Clostridium difficile Not Detected

Detected Detected Detected Wild type Toxigenic Clostridium difficile Detected

Detected Detected Detected Mutant Toxigenic and PCR ribotype 027 Clostridium difficile Detected

Detected Detected Not Detected Wild type Toxigenic Clostridium difficile Detected

Detected Not Detected Detected Wild type Toxigenic Clostridium difficile Detected

Detected Not Detected Detected Mutant Toxigenic and PCR ribotype 027 Clostridium difficile Detected

Detected Not Detected Not Detected Wild type Toxigenic Clostridium difficile Detected

Not detected Detected Detected Wild type Toxigenic Clostridium difficile Detected

Not detected Detected Detected Mutant Toxigenic and PCR ribotype 027 Clostridium difficile Detected

Not detected Detected Not Detected Wild type Toxigenic Clostridium difficile Detected

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Table 2: Overall Performance Characteristics of the Verigene CDF Test Compared to Direct

Toxigenic Culture and Direct plus Enriched Toxigenic Culture

Method Number

of Specimens

Sensitivity

(% CI)

Specificity

(% CI)

Accuracy

(% CI)

PPV

(% CI)

NPV

(% CI)

Direct

Culture

1875 156/158

98.7%

(95.5-99.9)

1502/1717

87.5%

(85.8-89.0)

1658/1875

88.4%

(86.9-89.8)

156/371

42.1%

(37.0-47.3)

1502/1504

99.9%

(99.5-100)

Direct plus

Enriched

Culture

1875 250/275

90.9%

(86.9-94.0)

1480/1600

92.5%

(91.1-93.7)

1730/1875

92.3%

(91.0-93.4)

250/370

67.6%

(62.5-72.3)

1480/1505

98.3%

(97.6-98.9)

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Table 3: Direct Culture Results and Direct Culture plus Enriched Culture Results Stratified by Laboratory Site

Direct Culture Direct plus Enriched Culture Site # Sensitivity

(% CI)

Specificity

(% CI)

Accuracy

(% CI)

PPV

(%CI)

NPV

( % CI)

Sensitivity

(% CI)

Specificity

(% CI)

Accuracy

(% CI)

PPV

(% CI)

NPV

(% CI)

1 557

55/55

100%

(93.5-100)

424/502

84.5%

(81.0-

87.5)

479/557

86.3%

(83.2-

89.1)

55/133

41.4%

(32.9-

50.2)

424/424

100%

(99.1-

100)

98/106

92.5%

(85.7-96.7)

416/451

92.2%

(89.4-94.5)

514/557

92.3%

(89.7-

94.4)

98/133

73.7%

(65.4-

80.9)

416/424

98.1%

(96.3-

99.2)

2 482

36/36

100%

(90.3-100)

379/446

85.0%

(81.3-

88.2)

415/482

86.1%

(82.7-

89.1)

36/101

35.6%

(26.4-

45.8)

381/381

100%

(99.0-

100)

58/61

95.1%

(86.3-99.0)

378/421

89.8%

(86.5-92.5)

436/482

90.5%

(87.5-

92.9)

58/101

57.4%

(47.2-

67.2)

378/381

99.2%

(97.7-

99.8)

3 299

25/25

100%

(86.3-100)

251/274

91.6%

(87.7-

94.6)

276/299

92.3%

(88.7-

95.1)

25/47

53.2%

(38.1-

67.9)

252/252

100%

(98.5-

100)

36/40

90.0%

(76.3-97.2)

248/259

95.8%

(92.5-97.9)

284/299

95.0%

(91.9-

97.2)

36/47

76.6%

(62.0-

87.7)

248/252

98.4%

(96.0-

99.6)

4 371

25/27

92.6%

(75.7-

99.1)

310/344

90.1%

(86.5-

93.1)

335/371

90.3%

(86.8-

93.1)

25/59

42.4%

(29.6-

55.9)

310/312

99.4%

(97.7-

99.9)

38/44

86.4%

(72.7-94.8)

306/327

93.6%

(90.4-96.0)

344/371

92.7%

(89.6-

95.2)

38/59

64.4%

(50.9-

76.5)

306/312

98.1%

(95.9-

99.3)

5 166

15/15

100%

(78.2-100)

135/151

89.4%

(83.4-

93.8)

150/166

90.4%

(84.8-

94.4)

15/31

48.4%

(30.2-

66.9)

135/135

100%

(97.3-

100)

20/24

83.3%

(62.6-95.3)

132/142

93.0%

(87.4-96.6)

152/166

91.6%

(86.3-

95.3)

20/30

66.7%

(47.2-

82.7)

132/136

97.1%

(92.6-

99.2)

Total 1875

156/158

98.7%

(95.5-

99.9)

1502/1717

87.5%

(85.8-

89.0)

1658/187

5

88.4%

(86.9-

89.8)

156/371

42.1%

(37.0-

47.3)

1502/15

04

99.9%

(99.5-

100)

250/275

90.9%

(86.9-94.0)

1480/1600

****

92.5%

(91.1-93.7)

1730/187

5

92.3%

(91.0-

93.4)

250/37

0

67.6%

(62.5-

72.3)

1480/150

5

98.3%

(97.6-

98.9)

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

***Of the 1875 specimens evaluated, seven specimens were culture positive but were either not PCR ribotyped or sequenced or both but

are included in the performance characteristics above.

****Of the 120 specimens that were toxigenic C. difficile positive (+) by Verigene CDF testing but toxigenic C. difficile negative (-) by direct/

enriched toxigenic culture, 90 were positive by bi-directional sequencing for tcdC, 6 were positive by bi-directional sequencing for tcdB, and 24

were negative by bi-directional sequencing for both tcdC and tcdB.

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Table 4: Results for Presumptive Detection of 027 Strains by the Verigene CDF Test

Compared to PCR-Ribotyping and Sequencing

PCR-Ribotyping or Sequencing Results

Verigene CDF Test

027 Not 027

PC

R-R

iboty

pin

g

027 53 1

Not 027 5 188

Not

performed

1 3

Total 59 192

Bi-

Dir

ecti

on

al

Seq

uen

cin

g

027 59 1

Not 027 0 191

Not

performed

0 0

Total 59 192

on June 8, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from