Predicting Recurrence of C. difficile Colitis Using Bacterial
Virulence Factors: Binary Toxin Is the Key David B. Stewart & Arthur Berg & John Hegarty Received: 20 April 2012 / Accepted: 11 October 2012 / Published online: 20 October 2012 #2012 The Society for Surgery of the Alimentary Tract Abstract Background Recurrent Clostridium difficile colitis is common, yet the ability to predict recurrence is poorly developed. Methods Patients 18 years of age treated at our institution for C. difficile of any severity were consecutively enrolled. C. difficile colitis was defined as symptoms of colitis with a positive PCR stool test. Each bacterial isolate was studied for virulence factors: tcdC mutations via PCR; the presence of genes for toxins A, B, and binary toxin using restriction fragment length polymorphism; and identification of ribotype 027 by PCR. Chi-squared tests, t tests, and logistic and linear regression were used to determine which virulence factors predicted recurrence. Results Sixty-nine patients (male, 57 %) were studied, with a mean age of 6413 years. Twenty-one (30 %) patients were initially diagnosed as outpatients. There was no difference (p>0.05) between virulence factors among inpatients and outpatients. The presence of a binary toxin gene was the single virulence factor independently associated with recurrence (p00.02). The combination of a tcdC mutation with binary toxin gene resulted in the highest odds of recurrence (OR, 5.3; 95 % CI, 3.526.09). Conclusion Binary toxin gene is a predictor of recurrent infection. Its presence may require longer antibiotic regimens in an effort to lower already elevated recurrence rates. Keywords Clostridiumdifficile . Recurrence . Binary toxin . Virulence Introduction The recent state of Clostridium difficile infection (CDI) is characterized by both conceptual upheaval and frequent shifts in the collective understanding of its pathophysiology. 1 This flux is attributable to rapid changes in disease patterns toward a higher incidence and prevalence of C. difficile colitis (CDC), 2,3 coupled with more frequent occurrences of fulmi- nant CDC which, 4,5 contrary to past disease behavior, now more regularly requires surgical intervention as a stopgap to failing medical therapy. While death from CDC is an uncom- mon sequela, a commonplace adverse outcome involves re- current CDC (RCDC). It is estimated that a minimum of 20 30 % of CDC patients will develop at least one recurrent episode, with an incidence that approaches 5065 % subse- quent to the first recurrence. 6,7 RCDC often requires extended periods of treatment, which exposes patients to lengthy and expensive courses of antibiotics which may ultimately prove inadequate toward eradicating the infection. 8 Several patient-related and environmental risk factors for both primary and recurrent C. difficile infection have been suggested, such as the absence of proper hand-washing among health care providers 9 as well as the injudicious use of broad spectrum antibiotics. 10 More recently, the Food and Drug Administration released a warning 11 that the use of proton-pump inhibitors may promote CDI, which may prove to have implications for RCDC as well. Despite these and other associations, the ability to actually predict CDC disease course, including recurrence of infection, from clin- ical factors alone remains merely provisional. This manuscript was presented at the SSAT Plenary Session III at the May 2012 Digestive Disease Week Meeting in San Diego, CA, USA. D. B. Stewart (*) : A. Berg : J. Hegarty Department of Surgery, Penn State College of Medicine, 500 University Drive, H137, Hershey, PA 17033, USA e-mail: dstewart@hmc.psu.edu J Gastrointest Surg (2013) 17:118125 DOI 10.1007/s11605-012-2056-6 By comparison to purely clinical research, much less information is available to explain how bacterial genetics and toxin profiles might influence the clinical progression of CDC, including the ontogeny of recurrent infections. In particular, whether there is a difference in recurrence rates as influenced by the presence of the relatively recently discovered binary toxin is completely unknown. While there is currently a heightened interest in the molecular biology of C. difficile, the determinative influences of alleged bacterial virulence factors compared to more commonly appealed to clinical factors, and the interplay between bacterial and patient characteristics, is at present an underdeveloped sche- ma. In terms of correlating bacterial factors with RCDC, there is almost no previous research to guide clinical decisions. The aim of the present study was to characterize how C. difficile virulence factors studied from non-duplicated patient-derived stool samples correlated with the incidence of RCDC. Materials and Methods This study was performed solely at the authors institution and with the Institutional Review Board (IRB) approval. The study is linked to an IRB-approved C. difficile tissue bank of one of the authors (DS), which collects C. difficile- positive clinical stool samples from inpatients and outpa- tients 18 years of age or older who consent to the use of their specimen for scientific research. Bacterial Isolates All C. difficile isolates were cultured and cryopreserved from patient-derived stool samples which were sent to our institutions clinical microbiology labora- tory for testing. The presence of the C. difficile gene for toxin A (tcdA) was confirmed through the use of a loop- mediated isothermal DNA amplification technique, as part of each patients clinical evaluation for CDC. Reference C. difficile strains representing ribotypes 021 and 078 (both NAP7 and NAP8 variants) were kindly pro- vided by Dr. Brandi Limbago (Centers for Disease Control and Prevention, Atlanta, GA, USA) in order to verify iden- tification of C. difficile for all PCRs subsequently performed. Isolation of C. difficile C. difficile-positive stool samples were shocked in 95 % alcohol for 30 min followed by culture using CDC anaerobe and phenylethyl alcohol anaer- obe agars (Remel, Lenexa, KS) containing 5 % sheep blood, under anaerobic conditions (85 % N 2 , 10 % CO 2 , 5 % H 2 ) using an Anoxomat system. Presumptive C. difficile isolates were identified by their characteristic malodor, characteristic colony morphology, and by chartreuse fluorescence under exposure to a Woods lamp. These isolates were then sub- cultured on CDC anaerobe 5 % sheep blood agar. Isolated colonies were then biochemically identified using RapID ANA II panels (Remel, Lenexa, KS). Positive C. difficile isolates were tested for susceptibility to metronidazole and vancomycin by the Etest method (bioMrieux, Durham, NC). Etest strips were positioned directly on the agar surface and were incubated at 37 C for 48 h. The minimum inhib- itory concentration (MIC) (in micrograms per milliliter) for metronidazole and vancomycin was measured at the ellipse of inhibited growth. Isolates were then stored and suspended in trypticase soy broth (BBL Becton Dickinson, Franklin Lakes, NJ) containing 15 % glycerol at 80 C. DNA Extraction Bacterial genomic DNAwas purified from washed bacterial cell pellets obtained from plate cultures following a 48-h period of anaerobic growth at 37 C. DNA was extracted using the UltraClean Microbial DNA isolation kit (MO-BIO, Carlsbad, CA) and was checked for concen- tration and purity with a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE). Detection of Binary Toxin Gene Primers Tim6, Struppi6, cdtBpos, and cdtBrev (Table 1) were used for multiplex amplification of the binary toxin gene (CDT) as well as the C. difficile-specific gene cdd3. Cycle conditions con- sisted of 35 cycles of denaturation for 30 s at 94 C, annealing for 45 s at 50 C, with extension for 1 min at 72 C, followed by a final extension cycle for 5 min at 72 C. PCR products were separated by electrophoresis on 1.5 % TrisacetateEDTA (TAE) agarose gels and were assessed following ethidium bromide staining. Toxinotyping Isolates were toxinotyped for the major toxins using primers A3C, A4N, B1C, and B2N (Table 1) as de- scribed by Rupnik, 12 with modified cycling conditions. The first 3 kb of the toxin B gene (B1) and 3 kb of the C-terminal region of the toxin A gene (A3) were amplified by PCR using cycle conditions consisting of 40 cycles of denatur- ation for 5 s at 95 C, annealing for 5 s at 48 C, and with extension for 3 min at 72 C, followed by a final extension cycle for 7 min at 72 C. Toxin fragments B3 and A1 were confirmed on 1 % TAE agarose gels at 6 V/cm for 1 h and were then subsequently digested by restriction enzymes HincII, AccI, and EcoRI at 37 C for 3 h. Restriction patterns defining the toxinotypes were determined by final electrophoresis on 1.5 % TAE agarose gels at 8 V/cm for 2 h followed by ethidium bromide staining. PCR Ribotyping Genetic variations between isolates were further characterized by 16S23S rRNA intergenic spacer PCR performed using the method outlined by Stubbs. 13 PCR products were separated on 2.5 % MetaPhor agarose J Gastrointest Surg (2013) 17:118125 119 gels (Lonza, Allendale, NJ) at 150 V in 0.5 chilled Tris borateEDTA (TBE) buffer for 4 h followed by ethidium bromide staining. Ribotypes were determined by compari- son of gel electrophoresis banding patterns to reference strains. tcdC Deletions Large deletions in the negative regulator of toxin production (tcdC) protein were identified as described by Spigaglia and Mastrantoni, 14 with slight modifications. Primers tcdC1 and tcdC2 (Table 1) were used to amplify a 250-bp region of the tcdC gene. PCR was carried out in 20- L reactions containing HotStar Plus master mix (Qiagen, Valencia, CA) containing 1.5 mM MgCl 2 , 10 pmol of pri- mers, and 1 ng genomic DNA. The DNA template was denatured at 95 C for 5 min, and DNA was amplified for 35 cycles consisting of denaturation for 30 s at 94 C, annealing for 30 s at 52 C, and extension for 30 s at 72 C. Reactions were terminated following a final exten- sion step for 4 min at 72 C. tcdC deletions were identified by electrophoresis on 2.5 % TBE agarose gels at 7.5 V/cm for 3 h and visualized by ethidium bromide staining. Clinical Data Each patient enrolled in the tissue bank also provided consent for review of their medical record. For all patients, age, gender, ethnicity, and the presence of comor- bidities were recorded. A Charlson Comorbidity Scale (CCS) 15 score was calculated for both inpatients and out- patients, as was a measurement of the severity of each patients CDC using the guidelines from the Infectious Dis- eases Society of America (IDSA) criteria 16 as listed in Table 2. Clinical data collected included whether or not hospital admission was required at any time during the treatment of CDC, whether hospital admission was for CDC, whether hospital admission was to an unmonitored setting versus an intensive care unit (ICU), and whether inpatients required transfer to the ICU after initial admission for the treatment of CDC. Whether the patient was using proton-pump inhibitors (PPI) prior to their CDC diagnosis was also recorded. Recurrent CDC was defined as two consecutive positive C. difficile stool studies no closer than 21 days apart. Out- comes of interest were the development of at least one episode of RCDC, the number of episodes of RCDC for patients with recurrent infections, and the need for hospital admission during the treatment of any episode of RCDC. Chi-squared tests, t tests, analysis of variance (ANOVA), and logistic and linear regression were used to evaluate the relationship between bacterial and patient factors and out- comes of interest. Results Summary information is provided for study subjects in Table 3, stratified based on those patients who did and did not experience RCDC. A total of 69 non-duplicate isolates Table 1 Primers used for PCR testing of C. difficile isolates Primer 53 sequence Target Size(s) (bp) Reference Tim6 TCCAATATAATAAATTAGCATTCCA cdd3 622 Stubbs et al. 26 Struppi6 GGCTATTACACGTAATCCAGATA cdtBpos CTTAATGCAAGTAAATACTGAG cdtB 510 cdtBrev AACGGATCTCTTGCTTCAGTC tcdC1 GCACCTCATCACCATCTTC tcdC 196250 Spigaglia and Mastrantonio 14 tcdC2 TGGTTCAAAATGAAAGACGAC 16S CTGGGGTGAAGTCGTAACAAGG 16S (1,4451,466; 3 end) 200700 Stubbs et al. 13 23S GCGCCCTTTGTAGCTTGACC 23S (20 to 1; 5 end) A3C TATTGATAGCACCTGATTTATATACAAG A3 fragment of tcdA 3,100 Rupnik et al. 12 A4N TTATCAAACATATATTTTAGCCATATATC B1C AGAAAATTTTATGAGTTTAGTTAATAGAAA B1 fragment of tcdB 3,100 B2N CAGATAATGTAGGAAGTAAGTCTATAG Table 2 Infectious Disease So- ciety of America criteria to cat- egorize severity of C. difficile infection Episode Severity Clinical definition Initial Mild or moderate WBC <15,000 and serum creatinine <1.5baseline Severe WBC 15,000 or serum creatinine 1.5baseline Severe (complicated) Ileus, hypotension/shock, megacolon First recurrence Same as above Second recurrence Same as above 120 J Gastrointest Surg (2013) 17:118125 were studied, with 28 (41 %) patients experiencing at least one episode of RCDC. There was no statistically significant dif- ference (p>0.05) between patients with and without RCDC based on age, gender, ethnicity, or CCS. Based on IDSA classification for severity of CDI, those patients experiencing at least one recurrence of CDC had a higher proportion of patients with severe and severe (complicated) CDC compared to patients without RCDC, though these differences were not statistically significant. There was no difference (p00.12) in the incidence of PPI therapy between patients with and with- out recurrences. Patients who did not experience RCDC had a higher incidence of systemic hypertension (p00.03) and dia- betes (p00.04) while having a lower incidence of coronary artery disease (p00.04) compared to patients with RCDC. Of the 69 study patients, 48 (70 %) received inpatient care during their first episode of CDC. Of these 48 patients, 20 (42 %) patients had a primary admission diagnosis of CDC. Ten of the 20 patients required admission to the ICU upon their admission to the hospital (RCDC, six versus no RCDC, four; p00.14), and 5 of these 20 patients required transfer to the ICU after their admission to the hospital (RCDC, four versus no RCDC, one; p00.07). There was no differ- ence between those with and without RCDC regarding the proportion of patients initially managed as outpatients for their index episode of CDC (p00.06). None of the patients in the study population required colectomy or died from CDC. Approximately 22 % of the entire study population had received either extended spectrum penicillins such as amoxicillin/clavulanic acid, or a quinolone, within 60 days of their index episode of CDC, with no difference (p00.34) noted in the incidence of previous antibiotic use between the two groups. Table 3 Characteristics of 69 patients with C. difficile colitis Characteristic No RCDC (n041) RCDC (n028) Significance (p value) Mean age (years) 6211 669 0.85 Gender (male) 35 (85 %) 16 (57 %) 0.08 Caucasian ethnicity 39 (95 %) 24 (86 %) 0.10 Charlson comorbidity score 2.80.8 3.20.7 0.07 IDSA severity Mild/moderate 25 (61 %) 10 (36 %) 0.06 Severe 13 (32 %) 15 (54 %) 0.07 Severe (complicated) 3 (7 %) 3 (10 %) 0.07 Use of proton-pump inhibitors 30 (73 %) 21 (75 %) 0.12 Comorbidities Systemic hypertension 35 (85 %) 15 (53 %) 0.03 Diabetes 25 (61 %) 13 (46 %) 0.04 Coronary artery disease 15 (37 %) 17 (60 %) 0.04 COPD 7 (17 %) 5 (18 %) 0.23 Obesity (BMI >30) 30 (73 %) 20 (71 %) 0.98 Chronic renal disease 8 (20 %) 10 (36 %) 0.07 Outpatient management during index CDC episode 10 (24 %) 11 (39 %) 0.06 Admitting service Medicine 37 (90 %) 20 (71 %) 0.76 Surgery 4 (10 %) 8 (29 %) Table 4 Bacterial virulence factors and correlation with re- current C. difficile colitis Virulence factor Incidence within study population (n069) Association with recurrence (p value) Association with need for admission for first CDC episode (p value) Association with admission for RCDC (p value) Toxin A 61 (88 %) 0.56 0.92 0.78 Toxin B 66 (96 %) 0.73 0.96 0.60 Binary toxin gene (CDT) 42 (61 %) 0.02 0.04 0.02 tcdC mutation 39 (56 %) 0.18 0.07 0.04 Ribotype 027 26 (38 %) 0.32 0.13 0.02 J Gastrointest Surg (2013) 17:118125 121 Table 4 presents the results of a univariate analysis of bacterial virulence factors correlated with the occurrence of at least one episode of RCDC. The majority of C. difficile isolates harbored at least one virulence factor, with 63 % harboring at least two virulence factors and with 48 % having three or more virulence traits. Out of all bacterial virulence factors, only the presence of the binary toxin gene (CDT) predicted a recurrent episode of CDC (p00.02). Neither ribotype 027 (p00.32), mutations of the tcdC gene (p00.18) nor the presence of genes for toxins A (p00.56) and B (p00.73) was associated with RCDC. The presence of the CDT gene (p00.04) was the only bacterial virulence factor associated with hospital admission for CDC treatment during the index episode of CDC. The presence of the CDT gene was found in 100 % of the PCR ribotype 027 strains (n026) as well as in 100 % of the PCR ribotype 078 strains (n07), being absent in ribotypes 001 (n03), 014 (n01), 017 (n01), and in several previously unreported strains (HMC 1117). A total of 28/69 (40 %) patients developed at least one episode of RCDC. Among the RCDC patient group, the mean number of recurrences was 2.01.8 (range, 16). Of these RCDC patients, the presence of the CDT gene (p0 0.02) and the presence of a tcdC mutation (p00.04) or PCR ribotype 027 (p00.02) were each associated with need for admission to the hospital specifically for the treatment of RCDC. Based on multivariable logistic regression, the com- bination of tcdA and tcdB with the presence of the binary toxin gene resulted in a higher odds of RCDC (odds ratio (OR), 3.1; 95 % confidence interval (CI), 2.973.33), with a mean number of RCDC episodes of 1.41.2. The combina- tion of a tcdC mutation with the presence of the binary toxin gene resulted in the highest odds of RCDC (OR, 5.3; 95 % CI, 3.526.09) and was associated with the highest mean number of recurrent episodes of infection (mean, 2.70.6). Among the entire study population, the mean MIC for vancomycin was 0.650.75 gmL 1 while the mean MIC for metronidazole was 0.20.2 gmL 1 . There was no difference in MICs between those with RCDC and those without recurrent infection (vancomycin: p0 0.87; metronidazole: p00.94) nor was there a difference in MICs for patients treated as inpatients (p00.99) and outpatients (p00.53) for their index episode of CDC. Based on ANOVA, there was no statistically significant difference (p00.23) in the use metronidazole, vancomy- cin, or the use of dual-agent therapy with both anti- biotics between the RCDC and no RCDC groups (Table 5). Discussion Binary toxin was only recently discovered, being first de- scribed by Popoff in a 1988 description of a female patient with CDC. 17 Further interest in the toxin has only developed in the past several years, and due to this latent attention, there is a smaller body of literature describing the structure, biology, and function of this toxin. While toxins A and B are members of a toxin family known as the large clostridial toxins, 18 which collectively have a molecular mass of 250 to 308 kDa and which are encoded by genes within the path- ogenicity locus (PaLoc), 19 CDT is encoded by a region remote to the PaLoc and is a member of the iota-like subclass of the clostridial binary toxin family. 20 Unlike toxins A and B, whose mechanism of action involves trans- ferring a glucose moiety from UDP-glucose to a family of small GTPases such as Rho, Rac, and CDC42, 21 binary toxin acts as an actin-ADP-ribosylating toxin, transferring an ADP moiety from NADP/NADPH to G-actin, thus pre- venting its normal polymerization into F-actin which leads to the disruption of the cytoskeleton, cellular dehydration, and cell death. 22,23 CDT consists of two components, CDTa and CDTb, the former being an enzymatically active unit and the latter being an inert transporter that aids the active component in crossing the cellular membrane. 24 There has been no scientific inquiry as to how binary toxin might promote recurrent disease, as the association between CDT and recurrence is a novel concept. There are several previously described characteristics attributed to CDT which, due to a paucity of clinically oriented research with this toxin, are not firmly established Table 5 Analysis of variance comparing the choice of antibi- otics during the index episode of CDC (p00.23) Antibiotic regimen during index CDC episode CDC without recurrence (n041) Patients who developed RCDC (n028) Oral metronidazole 22 (54 %) 16 (57 %) Oral metronidazole with non-CDC antibiotics 5 (12 %) 4 (14 %) Intravenous metronidazole 9 (22 %) 3 (11 %) Oral vancomycin 3 (7 %) 4 (14 %) Oral or intravenous metronidazole with oral vancomycin 1 (2.5 %) 1 (3.5 %) Oral vancomycin with non-CDC antibiotics 1 (2.5 %) 0 122 J Gastrointest Surg (2013) 17:118125 as typical for CDT and which are challenged by the present study. While it is often cited that the prevalence of the CDT gene ranges between 6 and 12.5 % 25,26 in the present study, the prevalence was much higher, being identified in 61 % of isolates. It is possible that there are regional variations in the prevalence of both ribotype and toxinotype among C. diffi- cile isolates, and if this was the case, then accurate measure- ment of the prevalence of virulence factors would require a multi-institutional effort with geographically diverse sour- ces of bacteria to provide an accurate assessment. The same phenomenon would also apply regarding the preponderance of ribotype 027, frequently referenced as the hyperviru- lent form of C. difficile. While previous reports 3,4 have described this type of C. difficile as an epidemical strain, suggesting that its emergence in an institution precedes outbreaks of particularly fulminant forms of infection, in the present study, 38 % of the isolates were of this ribotype. It is noteworthy that none of the 027 patients in this study required colectomy or died from CDC and that this ribotype was not a univariate predictor of recurrent infection. This suggests that, in some institutions, ostensibly notorious ribotypes such as 027 are actually quite common causes of CDC as opposed to representing a periodically encountered variant and that disease behavior may not be accurately, mechanistically viewed from the sole vantage point of al- legedly epidemical ribotypes. Other studies have reached similar conclusions, where predicting virulence has not been successful with such factors as tcdC mutations or ribotype. 27 The fact that, in the present study, the so-called virulence factors were not associated with fulminant colitis would suggest that their mere presence is perhaps necessary but not sufficient to produce severe forms of CDC. Though previously published data on CDC have often considered recurrence to be characteristic of virulent infections, sur- geons may wish to distinguish between virulence, defined as severe forms of colitis, and recurrence as reflecting an infection that is difficult to eradicate. This difference in parlance is important to recognize in order to avoid ambi- guities between the surgical, infectious disease and molec- ular biological literature on CDC. Much concern has recently been expressed regarding the possible etiologic role of PPI therapy in the development of CDC. 28 Though the provocating effect of PPI therapy to- ward development of CDC has not been a universal observation, 29 the issue of whether PPIs promote recurrent CDC is even less clear. A recent case-control study by Kim and colleagues identified PPI therapy as the only clinical factor associated with RCDC, without providing any infor- mation regarding bacterial toxin typing or genetic factors. 30 In contradistinction, the present study demonstrated no sig- nificant difference in rates of RCDC based on the presence of PPI therapy. Though it is commonly assumed that acid suppression is the principal manner through which this class of medications promotes CDC, 31 there is almost no infor- mation available regarding the direct effect that PPIs may have on either the colon or C. difficile. This is in keeping with the burgeoning yet seminal area of current interest in C. difficile research focused on identifying those environmental factors affecting bacterial behavior within the large intestine. As an example of the importance of environmental cues on the behavior of C. difficile, preliminary evidence 32 has sug- gested that antibiotics may have direct effects promoting growth and the production of toxins in C. difficile apart from disturbances to the microbiome caused by these anti- biotics. Further research on the effects of PPIs, antibiotics, and other medications on the colon, commensal bacteria and C. difficile is necessary to understand the risks these medi- cations may pose. Conclusion The presence of the binary toxin gene in C. difficile is an independent predictor of recurrent CDI, which is a finding that has not been previously reported. The combination of the binary toxin gene and mutations in the tcdCgene is associated with 430 % higher odds of recurrent CDI. C. difficile isolates which produce binary toxin may require longer antibiotic regimens in an effort to lower already elevated recurrence rates within the general CDC patient population. References 1. Bartlett JG, Perl TM. The new Clostridium difficilewhat does it mean? N Engl J Med 2008;353(23):2503-05. 2. Dallal RM, Harbrecht BG, Boujoukas AJ, et al. Fulminant Clostridium difficile: an underappreciated and increasing cause of death and complications. Ann Surg 2002;235(3):363-372. 3. Pepin J, Valiquette L, Alary ME, et al. Clostridium difficile-asso- ciated diarrhea in a region of Quebec from 1991 to 2003: a changing pattern of disease severity. CMAJ 2004;171(5):466-472. 4. Warny M, Pepin J, Fang A, et al. Toxin production by an emerging strain of Clostridium difficile associated with outbreaks of severe disease in North America and Europe. Lancet 2005;366:1079-84. 5. McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin gene-variant strain of Clostridium difficile. N Engl J Med 2005;353:2433-41. 6. McFarland LV, Surawicz CM, Rubin M, et al. Recurrent Clostridium difficile disease: epidemiology and clinical character- istics. Infect Control Hosp Epidemiol 1999;20:43-50. 7. McFarland LV, Elmer GW, Surawicz CM. Breaking the cycle: treatment strategies for 163 cases of recurrent Clostridium difficile disease. Am J Gastroenterol 2002;97(7):1699-75. 8. McFarland LV, Surawicz CM, Rubin M, Fekety R, et al. Recurrent Clostridium difficile disease: epidemiology and clinical character- istics. Infect Control Hosp Epidemiol 1999;43-50. 9. Jabbar U, Leischner J, Kasper D, Gerber R, et al. Effectiveness of alcohol-based hand rubs for removal of Clostridium difficile spores from hands. Infect Control Hosp Epidemol 2010;31(6):565-70. J Gastrointest Surg (2013) 17:118125 123 10. Stevens V, Dumyati G, Fine LS, Fisher SG, van Wijngaarden E. Cumulative antibiotic exposures over time and the risk of Clostridium difficile infection. Clin Infect Dis 2011;53(1):42-8. 11. FDA U.S. Food and Drug Administration. Accessed on 4/2/2012; http://www. fda.gov/Safety/MedWatch/SafetyInformation/ SafetyAlertsforHumanMedicalProducts/ucm290838.htm 12. Rupnik M, Avesani V, Janc M, von Eichel-Streiber C, Delme M. A novel toxinotyping scheme and correlation of toxinotypes with serogroups of Clostridium difficile isolates. J Clin Microbiol 1998;36(8):2240-7. 13. Stubbs SL, Brazier JS, ONeill GL, Duerden BI. PCR targeted to the 16S-23S rRNA gene intergenic spacer region of Clostridium difficile and construction of a library consisting of 116 different PCR ribotypes. J Clin Microbiol 1999;37:461-63. 14. Spigaglia P, Mastrantonio P. Molecular analysis of the pathogenic- ity locus and polymorphism in the putative negative regulator of toxin production (TcdC) among Clostridium difficile clinical iso- lates. J Clin Microbiol 2002;40(9):3470-5. 15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: de- velopment and validation. J Chronic Dis 1987;40:373-83. 16. Cohen SH, Gerding DN, Johnson S, et al. Clinical practice guide- lines for Clostridium difficile infection in adults: 2010 update by the Society for Healthcare Epidemiology of America (SHEA) and the Infectious Diseases Society of America (IDSA). Infect Control Hosp Epidemiol 2010;31:431-55. 17. Popoff MR, Rubin E, Gill DM, Boquet P. Actin-specific ADP- ribosyltransferase produced by a Clostridium difficile strain. Infect Immun 1988;56:2299-306. 18. Thelestram M, Chaves-Olarte E. Cytotoxic effects of the Clostridium difficile toxins. Cur Top Microbiol Immunol 2000;250:85-96. 19. Rupnik M, Grabnar M, Geric B. Binary toxin producing Clostridium difficile strains. Anaerobe 2003;9:289-294. 20. Barth H, Aktories K, Popoff MR, Stiles BG, et al. Binary bacterial toxins: biochemistry, biology, and applications of common Clostridium and Bacillus mechanism of ADP ribosylating toxins. FEBS J 2006;273:4579-4593. 21. Sundriyal A, Roberts AK, Ling R, McGlashan J, et al. Expression, purification and cell cytotoxicity of actin-modifying binary toxin from Clostridium difficile. Protein Expr Purif 2010;74:42-48. 22. Vandekerckhove J, Schering B, Barmann M, et al. Clostridium perfringens iota toxin ADP ribosylates skeletal muscle actin in Arg-177. FEBS Lett 1987(225):48-52. 23. Reuner KH, Presek P, Boschek CB, et al. Botulinum C2 toxin ADP-ribosylates actin and disorganizes the microfilament network in intact cells. Eur J Cell Biol 1987;43:134-140. 24. Davies AH, Roberts AK, Shone CC, Acharya KR. Super toxins from a super bug: structure and function of Clostridium difficile toxins. Biochem KJ 2011;436:517-526. 25. Geric B, Johnson S, Gerding DN, Grabnar M et al. Frequency of binary toxin genes among Clostridium difficile strains that do not produce large clostridial toxins. J Clin Microbiol 2003;41:5227-32. 26. Stubbs S, Rupnik M, Gilbert M, Brazier J, et al. Production of actin specific ADP-ribosyltransferase (binary toxin) by strains of Clostridium difficile. FEMS Microbiol Lett 2000;186:307-312. 27. Goldenberg AD, French GL. Lack of association of tcdC type and binary toxin status with disease severity and outcome in toxigenic Clostridium difficile. J Infect 2011;62(5):355-362. 28. McCarthy DM. Adverse effects of proton pump inhibitor drugs: clues and conclusions. Curr Opin Gastroenterol 2010;26:624-631. 29. Naggie S, Miller BA, Zuzak KB, et al. A case-control study of community-associated Clostridium difficile infection: no role for proton pump inhibitors. Am J Med 2011;124(3):276.e1-276.e7. 30. Kim YG, Graham DY, Jang BI. Proton pump inhibitor use and recurrent Clostridium difficile-associated disease. J Clin Gastroenterol 2012;46(5):397-400. 31. Dial S, Delaney JA, Barkun AN, et al. Use of gastric acid- suppressive agents and the risk of community-acquired Clostridium difficile-associated disease. JAMA 2005;294:2989- 95. 32. Dupuy B, Govind R, Antunes A, Matamouros S. Clostridium difficile toxin synthesis is negatively regulated by TcdC. J Med Microbiol 2008;57:685-689. Discussant Dr. Sekar Dharmarajan (St. Louis, MO): Thank you Mr. Chairman. While the surgical literature is replete with studies that correlate clin- ical patient factors with morbidity, need for surgery, and mortality from C. difficile infection, Dr. Stewart and colleagues have taken the novel and thought-provoking approach of characterizing bacterial virulence factors and correlating these with morbidity from C. difficile in the form of recurrent infection. I have three broad areas of comments/ questions. The first surrounds the definition of recurrent C. difficile colitis. The authors define recurrent C. difficile infection as two consecutive pos- itive C. difficile stool samples no closer than 21 days apart. Without a negative intervening sample, how do we know this is recurrent C. difficile infection as opposed to persistent C. difficile infection, as we know that C. difficile is notoriously difficult to eradicate? Specifically, do you have any data on how far apart temporally the positive stool samples were in the patients with recurrent C. difficile? More interest- ingly, do you have any data on the bacteriology of the recurrent C. difficile isolates to see how they are compared to the original isolate? The second area is with regard to the broader generalizability or applicability of the study. As the authors state in their manuscript, the prevalence of the binary toxin gene in this study was five- to tenfold higher than that previously reported in the literature. Similarly, ribotype 027, which has been found in previous studies to be hypervirulent, was not associated with any morbidity or mortality in the present study. Is it possible that, while binary toxin is the key to predicting recurrent infection at Hershey, these factors may differ institution to institution? Finally, the last area is with regard to patient or host factors that certainly must contribute to morbidity and mortality from C. difficile infection. While the present study found no clinical factors that pre- dicted recurrent infection, I wonder if the authors have plans or could comment on studying the interaction of patient genotypes at genetic loci that may render susceptibility to infection with bacterial virulence factors on morbidity from C. difficile. Similarly, as the authors com- ment in their manuscript, the host microbiome has become an increas- ingly important area of research in determining outcomes from a variety of disease processes, and I wonder if the authors had any comment on their plans to study this. Thank you for the opportunity to review your extremely well- written manuscript. Selection Bias: There is no patient with fulminant C. difficile colitis as defined by need for surgery or death. Closing Discussant Dr. David B. Stewart: I would like to thank the SSAT for the opportunity to present our research, and I would also like to express my gratitude to Dr. Dharmarajan for being willing to serve as a discussant for our presentation. 124 J Gastrointest Surg (2013) 17:118125 Dr. Dharmarajans first point is perhaps the most critical issue which he raised as a discussant. There are no consensus-based defi- nitions, for all research endeavors on the topic, which define recurrent C. difficile colitis (RCDC) in any scientifically meaningful manner. This lack of standardization would potentially allow for an inflated estimate of RCDC, if the time interval between two consecutive CDC episodes was diminished to a great enough degree. In fact, in reviewing the literature which deals with recurrent C. difficile, the reader will encounter both heterogeneity in the definition of recurrence as well as a degree of arbitrariness which includes extremely short time intervals (<10 days) as well as criteria that range from clinical symptoms of diarrhea to tests confirming the presence of toxigenic C. difficile. In our case, it seemed appropriate to set the definition of RCDC at a point at least far enough after the diagnosis of the index episode to allow for the standard course of 1014 days of antibiotics to be completed. We then allowed for an additional week of time to elapse in order to better ensure that we avoided collapsing recurrence into persistence. While our approach is not universal in acceptance, we believe it is fair and that it has a conceptual appeal given what we have described. Further, our recurrence rates fall into the same general range as those previously published in larger studies. There are no data available to currently guide the definition of when persistence ends and recurrence begins, and so there are no data available to answer the issue of temporal relationships for persistence or recurrence. In part, this very relevant question will not be answered until we have large C. difficile registries that can provide, at minimum, regional epidemiologic information regarding endemic ribotypes and virulence characteristics. Such resources do not exist in this country, though they do exist in seminal forms in Canada. The issue related to whether a recurrent episode of CDC is due to a different ribotype is something our group will be focusing on in the future. There are no relevant data on this topic currently available, though the issue is critical. Our group has unpublished data which demonstrate that C. difficile responds to drugs such as proton-pump inhibitors in a ribotype-dependent manner, such that some ribotypes will produce more toxins in the presence of PPIs and some will express toxin genes to a lesser degree. Some are actually unaffected by PPIs. If this is confirmed in larger studies, it will introduce an additional complexity in understanding C. difficile and in how environmental cues, including the drugs we use to eradicate the organism, may lead to different clinical outcomes. There is a biologic basis for such phenomena, and in the same manner that we push for personalized medicine with respect to diseases like cancer, we need personalized microbiology to a much greater degree than is presently available. The second comment deals with whether those bacterial genetic signatures that were associated with higher rates of recurrence in our study might not be relevant in a different population of C. difficile. Until our findings are prospectively evaluated in a multiregional fash- ion, then this concern will not be definitively addressed. However, the idea that binary toxin would negatively affect clinical outcomes in CDC has scientific plausibility, given that previous reports have sug- gested that this is a more potent form of toxin. Dr. Dharmarajans last comment deals with how host factors, including other microorganisms within the gut, may influence the behavior of C. difficile. The issue of the gut microbiome is cer- tainly the hot topic of our day in microbiology, and for good reason. However, our understanding of the microbiome concept is currently so piecemeal that the more we learn, the less we know. There is at least one publication from several years ago which demonstrated a higher rate of RCDC when the human subject was found to harbor certain single nucleotide polymorphisms in the gene for IL-8, though this publication used a definition for recurrence, which may have lent toward artificially higher recur- rence rates. Our own group is now using metabolomics to study how different environmental elements within the gut, such as vary- ing levels of electrolytes and micronutrients, might affect transcrip- tional control of virulence genes in C. difficile. We have also begun obtaining rectal swabs on CDC patients in order to study the mRNA present in the organism at the time the sample is obtained, in an effort to study the transcriptome and how differential gene expression is associated with clinical outcomes such as recurrence, fulminant colitis, and death from CDC. One final commentthese were consecutively enrolled patients which were presented in our study. Though none of our study patients required colectomy or died, there were patients with severe CDC based on the IDSA classification schema. Surgical intervention for CDC, while much more common than in previous eras, is still relatively infrequent, which makes studying this particular patient group more difficult when attempting to obtain informed consent for a stool sample is required and when the patient is critically ill. J Gastrointest Surg (2013) 17:118125 125