Abstract
Clostridium difficile infection (CDI) recurs in ~20% of patients. Prior studies indicated that antibody responses directed against the C. difficile toxins A and B were potentially associated with lower risk of recurrent CDI. Here we tested the hypothesis that circulating anti-toxin IgG antibody levels associate with reduced risk of recurrent CDI. A cohort study with prospective enrollment and retrospective data abstraction examined antibody levels in 275 adult patients at the University of Michigan with CDI. We developed an enzyme linked immunosorbent assay to detect IgG antibodies against toxin A and toxin B in sera obtained at the time of diagnosis. Logistic regression examined the relationship between antibody levels and recurrence, and sensitivity tests evaluated for follow-up and survivor biases, history of CDI, and PCR ribotype. Follow-up data were available for 174 subjects, of whom 36 (20.7%) had recurrence. Comparing antibody levels vs. recurrence and CDI history, anti-toxin A levels were similar, while anti-toxin B levels had a greater range of values. In unadjusted analysis, detection of anti-toxin A antibodies, but not anti-toxin B antibodies, associated with an increased risk of recurrence (OR 2.71 [1.06, 8.37], P =.053). Adjusting for confounders weakened this association. The results were the same in sensitivity analyses. We observed a borderline increased risk of recurrence in patients positive for anti-toxin A antibodies, and sensitivity analyses showed this was not simply a reflection of prior exposure status. Future studies are needed to assess how neutralizing antibody or levels after treatment associate with recurrence.
Keywords: Clostridium difficile, bacterial toxins, antibodies, healthcare-associated infections
Introduction
Clostridium difficile is a Gram positive, anaerobic, spore-forming bacillus, which can produce 2–3 toxins: TcdA, TcdB, and in some strains, binary toxin (CDT) [1,2]. The ability to produce toxin is essential to its pathogenicity [2]. Colonization with C. difficile can result in a range of outcomes from asymptomatic colonization to fulminant, life-threatening colitis [2]. Recurrent infections occur in about 20% of patients and cost $2.8 billion each year [3,4]. At day 12 after the onset of the initial CDI episode, low serum anti-toxin A IgG antibodies have been identified as a risk factor for recurrence [5]. Furthermore, immune therapy (e.g. bezlotoxumab), reduces recurrence when administered during the original episode [6]. There are other strategies to reduce recurrence, such as specific antibiotics (e.g. vancomycin tapers and fidaxomicin), and fecal microbiota transplantation [7–11]. However, use of these strategies in all patients is not practical or desirable due to expense, invasiveness, and/or undetermined safety profiles [12,13]. Thus, risk stratifying patients for recurrence upon initial infection would be helpful clinically.
Clinical risk factors for infection have been described previously, but accurate and useful predictive models have eluded researchers [14,15]. Although along with age, severity of disease, and antibiotic use, the presence of anti-toxin antibodies may be a biomarker for a reduced risk of recurrence [5,16], the role of measuring serum anti-toxin antibodies in predicting recurrence of C. difficile infection (CDI) is still debated [17,18]. This study tests the hypothesis that the presence of anti-toxin IgG antibodies in serum sampled within 48 hours of diagnosis from patients with an index episode of CDI associates with reduced risk of recurrent CDI (rCDI).
Materials and Methods
Study Design
With approval from the University of Michigan Institutional Review Board, we performed a cohort study with prospective enrollment and retrospective data abstraction, consisting of adults (18+ years) that were admitted to Michigan Medicine between January 29, 2016 and September 28, 2016. These patients were diagnosed with CDI by the clinical microbiology laboratory, and had available sera within 48 hours of that diagnosis. Individuals were excluded from the study analyses if they had a CDI diagnosis within 48 days prior to sampling. Patient data were extracted from the electronic health record as previously described [19]. Demographics, medical history, vital signs, and laboratory test results limited to 48 hours before and after C. difficile sample collection were gathered.
In order to diagnose CDI, the clinical microbiology laboratory tested stool samples for toxigenic C. difficile with a two-step algorithm. The initial step used the C. Diff Quik Check Complete test (Alere, Kansas City, MO) for C. difficile glutamate dehydrogenase (GDH) and toxin A or B by the use of an enzyme immunoassay. All initial step results that were discordant (GDH positive and toxin negative [GDH+/toxin−] or GDH negative and toxin positive [GDH−/toxin+]) were reanalyzed (reflexed) using real-time PCR for the tcdB gene with the Simplexa assay (Focus Diagnostics, Cypress, CA). Confirmation of positive tests was attempted by anaerobic culture on taurocholate-cycloserine-cefoxitin-fructose agar at 37°C, and isolates were ribotyped using a high-throughput, fluorescent PCR ribotyping protocol previously validated at multiple sites and described elsewhere [20].
ELISA
Patient sera samples were frozen at −80°C until testing. An enzyme linked immunosorbent assay (ELISA) protocol was developed using previously described methods as a guide [21–24]. Each well of 96 well plates (Corning, Ref. No. 9018, Clear, Flat Bottom, High Binding, No Lid, Polystyrene, Non-sterile) were coated with 100 μL of 0.1 μg/mL List Biological Laboratories, Inc. toxin A or toxin B. After 24 hours, plates were washed with 0.05% phosphate buffered saline (Gibco by Life Technologies) with Tween-20 (Fisher Scientific), 2% bovine serum albumin (Sigma Life Science [Aldrich]) as blocking buffer was added, plates were incubated, and the plates were re-washed. Patient serum and control serum were tested at at 1:100 and 1:25 dilutions. The positive control serum sample consisted of a 1:100 dilution of pooled human sera plus 1:1000 FcABBA human IgG [25], the negative serum control sample was 1:100 pooled human sera alone, and the true negative plate control was 0.05% phosphate buffered saline with Tween-20 without any added serum or antibody. Plates were incubated, then washed, and polyclonal rabbit anti-human IgG/HRP (Dako) was added. After incubation, 1-Step ABTS (Thermo Scientific) was added.
Absorbance was read at 410nm with reference at 650nm, to account for particulate artifacts. Samples and sample controls were averaged. ELISA units were calculated as the sample average divided by the sample control average and multiplied by 1000.
For further detail, see Appendix Section A.
Outcome & Predictor Definitions
The primary outcome of interest was recurrent CDI, defined as a second CDI diagnosis occurring within 14–100 days after the index diagnosis of CDI. A patient had to have a visit to Michigan Medicine that was not associated with a CDI diagnosis, as well as no CDI diagnosis, within the timeframe in order to be classified with no recurrent CDI. A history of CDI was defined as a prior episode >100 days before the current one. Since patients not followed beyond 14 days were not eligible for the primary outcome, in lieu of a time-to-event analysis, patients without these follow-up data were excluded from the analysis (see Data Analysis for details). The primary predictor of interest was the presence of anti-toxin A IgG antibodies or anti-toxin B antibodies, defined as an ELISA unit measurement greater than or equal to 1000, the limit of detection, at the index encounter (based on histograms and best fit with the outcomes of interest on unadjusted analyses).
Data Analysis
Descriptive statistics were examined for all clinical and laboratory-derived variables. Distributions, histograms, and associations with outcomes for continuous variables were examined to assess for threshold or non-linear effects to determine if they should be categorized or undergo transformation prior to analysis. Specifically, distance and albumin were dichotomized. Patients were given a value of 1 in ‘distance >100km’ if their listed home address zip code was greater than 100km from Michigan Medicine. Patients were classified as having ‘low albumin (Less than 3)’ if the lowest albumin lab result value was less than three in the 48 hours before and after the C. difficile collection date. For ribotypes, given the prior associations seen between ribotype 027 at our hospital and others, we created a binary variable to focus on this specific ribotype for analysis [15,26]. Antibiotics were categorized as high, intermediate, or low risk based on evidence from prior publications [27].
For the complete set of variable definitions, see Appendix B.
Time-to-event analysis was considered, but was not practical in this study. The primary reason being that a majority of people were discharged 0–14 days after the index episode, before they were even eligible to meet the outcome definition, precluding use of censoring as a means to handle this follow-up loss. Thus, patients without adequate follow-up were excluded and a logistic regression strategy was pursued. Sensitivity analysis was done to ensure that this choice did not bias the findings, the details of which are described below.
Unadjusted logistic regression relating each variable to the primary outcome was examined first. Association between all variables and both the exposure and the outcome was tested in order to check for confounding. The final multivariable model was created including predictors of interest and putative confounders.
Sensitivity Analyses
To consider if patient distance from the main facility might bias this study’s results due to differential loss to follow-up, the ‘distance > 100km’ variable was forced into the final model.
To consider if patient deaths affected the model, patients were removed from consideration if they had died within the 14–100 day window for recurrence. The final model was run using only live patients and compared to the original final model, to check for effect size alterations that could indicate a survival bias.
Because we saw an increased risk of rCDI with antibodies (see results) we wanted to ensure that antibodies were not just a marker for past C. difficile exposure, i.e. history of CDI, which itself is a known risk factor for subsequent recurrence. Thus, as a final sensitivity check, ‘history of CDI’ was introduced into the final model.
Results
The initial total cohort of patients with sera available for anti-toxin antibody measurements was 275 individuals. However, there were 101 patients who had no additional data in their electronic health record (EHR) after 14 days, so nothing could be said about their recurrence status—thus, they were excluded. Of the 174 with the available data for an assigned recurrence status, 138 of these did not experience recurrence, and 36 did experience recurrence.
Seen in Table 1, there was no significant difference in the age at C. difficile diagnosis (P =.344) or proton pump inhibitor exposure (P =.539) status in patients with and without CDI recurrence. Exposure to high risk antibiotics status was significantly different (P =.012) between the outcome groups. History of C. difficile in these outcome groups was not significant (P =.209), though ulcerative colitis was (P =<.001). The variables of cardiac arrhythmia, congestive heart failure, electrolyte fluid therapy, and inflammatory bowel disease were significant or approached significance between patients with and without recurrence (Table 1). Anti-toxin A antibodies approached significance when comparing patients with and without recurrent CDI (P =.075), and anti-toxin B antibodies did not (P =.943). Only 25 patients (10%) were infected with ribotype 027.
Table 1.
Demographic and Clinical Data for Patient Population
| All Patients | Patients without Recurrence | Patients with Recurrence | P | |
|---|---|---|---|---|
| nc | 275 | 138 | 36 | |
| Exposure of Interest | ||||
| Anti-Toxin A Antibody Positive - Count (%) | 193 (70.2) | 96 (69.6) | 31 (86.1) | .075 |
| Anti-Toxin B Antibody Positive - Count (%) | 213 (77.5) | 108 (78.3) | 29 (80.6) | .943 |
| Demographics | ||||
| Age at C. difficile Diagnosis - (mean (sd)) | 55.24 (21.14) | 54.44 (20.46) | 58.06 (19.97) | .344 |
| Male - Count (%) | 131 (47.6) | 72 (52.2) | 14 (38.9) | .218 |
| White - Count (%) | 239 (87.9) | 123 (89.8) | 30 (83.3) | .433 |
| Distance from Patient to Hospital Greater than 100km - Count (%) | 66 (24) | 32 (23.2) | 7 (19.4) | .798 |
| Severity/Comorbidity Scales | ||||
| Elixhauser Comorbidity Index - (mean (sd)) | 4.88 (3.55) | 5.61 (3.36) | 3.67 (3.62) | .003 |
| Charlson Comorbidity Index - (mean (sd)) | 3.80 (2.92) | 3.95 (2.95) | 3.64 (3.16) | .58 |
| IDSA Severity Criteria - Count (%) | 71 (26.1) | 37 (27) | 5 (14.3) | .179 |
| Disease | ||||
| History of C. difficile Diagnosis - Count (%)a | 42 (15.3) | 20 (14.5) | 9 (25) | .209 |
| Cardiac Arrhythmia - Count (%) | 107 (41.8) | 67 (50.0) | 7 (25) | .027 |
| Congestive Heart Failure - Count (%) | 73 (26.5) | 42 (30.4) | 5 (13.9) | .075 |
| Cardiovascular Accident - Count (%) | 51 (18.5) | 28 (20.3) | 6 (16.7) | .801 |
| Diabetes - Count (%) | 78 (28.4) | 39 (28.3) | 10 (27.8) | >.99 |
| Electrolyte Fluid Therapy - Count (%) | 123 (48) | 75 (56) | 9 (32.1) | .037 |
| Metastatic Cancer - Count (%) | 18 (6.5) | 9 (6.5) | 1 (2.8) | .647 |
| Myocardial Infarction - Count (%) | 50 (18.2) | 31 (22.5) | 5 (13.9) | .368 |
| Peptic Ulcer Disease - Count (%) | 31 (11.3) | 18 (13) | 5 (13.9) | >.99 |
| Pulmonary Disease - Count (%) | 95 (34.5) | 51 (37) | 12 (33.3) | .835 |
| Peripheral Vascular Disease - Count (%) | 68 (24.7) | 40 (29) | 9 (25) | .791 |
| Renal Disease - Count (%) | 58 (21.1) | 33 (23.9) | 4 (11.1) | .149 |
| Rheumatic Disease - Count (%) | 13 (4.7) | 6 (4.3) | 1 (2.8) | >.99 |
| Liver Disease - Count (%) | 76 (27.6) | 46 (33.3) | 9 (25) | .449 |
| Inflammatory Bowel Disease - Count (%) | 43 (15.6) | 16 (11.6) | 13 (36.1) | .001 |
| Ulcerative Colitis - Count (%)b | 30 (10.9) | 9 (6.5) | 11 (30.6) | <.001 |
| Lab Tests | ||||
| Albumin Value Less Than 3 - Count (%) | 81 (35.7) | 52 (44.4) | 3 (10.3) | .001 |
| Ribotype 027 Infection – Count (%) | 25 (9.1) | 9 (6.5) | 7 (19.4) | .039 |
| Medication Exposures | ||||
| High Risk Antibiotic Exposure - Count (%) | 173 (80.5) | 102 (85.7) | 14 (60.9) | .012 |
| H2 Receptor Blocker Medication Class Exposure - Count (%) | 40 (18.6) | 31 (26.1) | 1 (4.3) | .045 |
| Proton Pump Inhibitor Exposure - Count (%) | 65 (30.2) | 42 (35.3) | 6 (26.1) | .539 |
Any positive sample more than 48 days before the index case
Diagnosis of ulcerative colitis at any point in their medical history
The sum of Patients with Recurrence and Patients without Recurrence is not equal to All Patients due to a lack of medical data post-C. diff index episode that did not allow for recurrent status assignment.
Anti-toxin A antibody ELISA unit distributions were similar across outcome types (Figure 1A). Anti-toxin B antibody ELISA units had similar median values across outcome types, through there were more samples above the third quartile in patients who recurred than in patients that did not (Figure 1B). The range of values was also larger compared to anti-toxin A antibody ELISA units. Distributions of anti-toxin A ELISA units were comparable whether or not the patient had a history of CDI (Figure 1C). In Figure 1D, anti-toxin B ELISA unit levels were higher in those with a history of CDI, though the median of the distribution was similar between the groups.
Figure 1:
Anti-toxin antibody distributions. 1A: Anti-toxin A antibodies are similarly distributed across all four recurrence status categories. 1B: Anti-toxin B antibodies have similar median values across recurrence status categories, but a wider range of values. More patients fall above the third quartile in those with recurrence. 1C: Anti-toxin A antibody values are similarly distributed across history status categories. 1D: Antibody B antibody values have similar medians across history status categories, though they have a greater range.
In unadjusted logistic regression models between anti-toxin antibody levels and recurrence, detectable anti-toxin A antibodies were associated with an increased risk of recurrence (OR=2.71, P =.053) and anti-toxin B antibodies were not associated with recurrence (P =.764). These models can be seen in Appendix C. Ribotype 027 infection was associated with recurrence (OR 3.46, P =.023) and anti-toxin B antibody presence (OR 0.39, P =.033), but not anti-toxin A antibody (OR 0.73, P =.48).
The final model (Table 2) shows that adding ulcerative colitis and low albumin as confounders weakened the association between anti-toxin A antibodies and recurrence (OR=2.91, P =.112) that was seen in the unadjusted model. Having a diagnosis of ulcerative colitis increased the odds of recurrence (OR=6.06, P =.004) and having an albumin level less than three decreased the odds of recurrence (OR=0.19, P =.011). Adding ribotype to this model did not change the results for the other covariates and was not significant itself, so it was not included in the final model (Appendix D).
Table 2.
Adjusted Logistic Regression Final Model
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | P |
|---|---|---|---|---|
| Intercept | 0.12 | 0.03 | 0.37 | .001 |
| Anti-Toxin A Antibodies | 2.91 | 0.88 | 13.32 | .112 |
| Ulcerative Colitis | 6.06 | 1.84 | 21.52 | .004 |
| Low Albumin | 0.19 | 0.04 | 0.6 | .011 |
Sensitivity Analysis Results
When forcing the distance variable into the final model, the directionality of the relationships between anti-toxin A antibodies and ulcerative colitis remained the same compared to the final model alone, and the estimates did not vary by large margins (Table 3A). Low albumin switched directionality of its relationship with recurrence with the addition of the distance variable (OR=5.35, P =.011).
Table 3.
Sensitivity Analysis using Logistic Regression
| A. Model results considering Patient Distance > 100km from the Healthcare System | ||||
|---|---|---|---|---|
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | P |
| Intercept | 0.11 | 0.03 | 0.36 | <0.001 |
| Anti-Toxin A Antibodies | 2.86 | 0.86 | 13.12 | .118 |
| Ulcerative Colitis | 6.4 | 1.9 | 23.17 | .003 |
| Low Albumin | 5.35 | 0.04 | 0.6 | .011 |
| Distance > 100km | 1.26 | 0.43 | 3.49 | .659 |
| B. Model results for live patient population | ||||
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | P |
| Intercept | 0.13 | 0.029 | 0.39 | .001 |
| Anti-Toxin A Antibodies | 3.11 | 0.934 | 14.28 | .092 |
| Ulcerative Colitis | 5.29 | 1.611 | 18.75 | .007 |
| Low Albumin | 0.22 | 0.049 | 1.38 | .024 |
| C. Model results considering History of CDI | ||||
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | P |
| Intercept | 0.12 | 0.03 | 0.36 | <0.001 |
| Anti-Toxin A Antibodies | 2.62 | 0.77 | 12.2 | .159 |
| Ulcerative Colitis | 6.22 | 1.88 | 22.35 | .003 |
| Low Albumin | 0.2 | 0.04 | 0.63 | .013 |
| History of CDI | 1.64 | 0.54 | 4.78 | .369 |
Twenty patients were deceased before the recurrence period ended, therefore it is unknown whether they would have recurred if they had survived. Running the final model excluding these individuals resulting in ulcerative colitis (OR=5.29, P =.007) and low albumin (OR=0.22, P =.024) remaining significant (Table 3B). The direction of all variable relationships with the outcome remained the same compared to the final model alone. Anti-toxin A antibodies remained non-significant (OR=3.11, P =.09).
Seen in Table 3C, adding history of CDI into the model did not significantly change any relationships seen between the final model variables and recurrence. Ulcerative colitis (OR=6.22, P =.003) and low albumin (OR=0.20, P =.013) remained significant and anti-toxin A antibodies remained insignificant (OR=2.62, P =.159).
Thus, our results remained robust during sensitivity analyses.
Discussion
The increased risk of recurrence in patients positive for anti-toxin A antibodies was opposite of what was expected. In previous work, low ELISA units for antibodies against toxin A were associated with an increased risk of recurrence [5]. However, prior CDI is a known risk factor for recurrent CDI [28]. We speculated that the observed association may have been due to the positive signal acting as a mediator between history of CDI and recurrence. Yet, when adjusting the model to include prior history of CDI, the increased risk of recurrence remained. Our interpretation of this is that while anti-toxin A antibodies were more likely to be present in those patients with a history of exposure to toxigenic C. difficile, there was no evidence of protection against subsequent CDI in this study.
Kyne et. al. have previously shown that serum IgG values against toxin A on day 12 are elevated in patients with a single CDI episode, compared to those with recurrent episodes [29]. It is possible that a contrasting association between anti-toxin A IgG antibodies and recurrence was detected in this study because the sera samples were taken within 48 hours of a positive C. difficile diagnosis, and there was not sufficient time for some patients to develop elevated antibody levels and show the same relationship with recurrence seen by Kyne et al. In contrast, Kyne et al. also measured IgM antitoxin levels and these were significantly lower in patients with recurrence at only day 3. Either these patients had low IgM antitoxin because they were CDI naïve and had a blunted immune response to C. difficile toxins, or they had already been exposed to C. difficile and, thus, mounted an initial IgG antitoxin response, but this was not measured on day 3 in their study.
In regards to toxin B, a prior randomized control trial resulting in an FDA-approved product (bezlotoxumab) showed that monoclonal antibodies against toxin B can reduce recurrent CDI [30]. The clinical trial was focused on developing a monoclonal antibody, which would target a particular epitope and be neutralizing. Our results show no significant association of anti-toxin B antibodies with recurrent CDI. Our study detected all antibodies that could bind to toxin, which identified targets across many epitopes and could have resulted in our findings being discordant with this prior work. Intriguingly, in the initial research leading to the development of bezlotoxumab, antibodies to toxin A were also found to be protective against CDI and recurrence, and decreased mortality was seen in hamsters given antibodies to toxin A and B [29,31,32]. This was not borne out clinically in the human subject trial, similar to the lack of association we observed for anti-toxin A antibodies.
The findings in this study may also have been different from prior research due to the changing epidemiology of C. difficile. Studies by Kyne et.al were published nearly two decades ago, and increased rates of infection with epidemic strains, such as ribotype 027 and other community-associated strains of C. difficile, such as ribotype 106, could have altered the organism’s relationship between its environment and host [33,34].
Limitations
This study was limited as it was conducted at a single center and used retrospective data extraction. Retrospective data extraction can result in loss to follow up, survival biases, and incomplete information for the patient population. However, we attempted to mitigate these limitations through our sensitivity analyses, which did not substantially alter our findings.
The availability of sera from patients also introduced a limitation. In our study, each patient only had one sample within 48 hours of their CDI diagnosis. However, in previous work, multiple logistic regression analysis indicated that lower neutralizing anti-toxin B antibodies at day 14 and anti–toxin A neutralizing antibodies at day 28 were predictive of recurrence [30].
Infection with the epidemic BI/NAP1/027 strain is associated with a higher risk of recurrence [30]. There are low rates of C. difficile ribotype 027 infections at Michigan Medicine—in the past only about 15% of patients with CDI were infected with this strain [19], dropping to only 10% in the present study. In 2015, a study across 10 geographic regions in the United States found that ribotype 027 caused 31% of healthcare associated CDI and 19% of community associated CDI [4]. This may have played a role in our inability to find a statistically significant association between toxin antibodies and recurrence at our health center.
Conclusion
This study failed to demonstrate a relationship between circulating IgG anti-toxin antibody levels during primary CDI and the development of subsequent recurrent CDI. The presence of anti-toxin A antibodies was associated with increased risk of recurrence alone, though that was of borderline statistical significance and the relationship was lost upon adjustment for confounders. Anti-toxin B antibodies were not found to be associated with recurrence.
Future Directions
Future work should include validation of our findings in a multicenter, prospective cohort study. Looking at CDI strain type in a more granular fashion than ribotype could provide additional information to add to the model. Investigating other antibody types (IgA, IgM) as well as different areas of action (digestive tract) could uncover important biomarkers. Antibodies over time during and post-CDI could be more informative than the baseline values we assessed. Finally, it may be useful to differentiate neutralizing from non-neutralizing antibody binding in functional assays, in lieu of simple antibody levels.
Figure 2.
Venn Diagram of Potential Confounders. The left side of the diagram shows the variables that were found to have a significant relationship with the exposures of anti-toxin A antibodies or anti-toxin B antibodies. The right side of the diagram shows the variables that were found to have a significant relationship with the outcome of recurrence. The overlapping middle region of the diagram shows the variables that were significantly associated with both the exposure(s) and the outcome, identifying them as confounders.
Highlights.
Anti-toxin A antibodies were associated with an increased risk of rCDI.
Anti-toxin B antibodies were not associated with rCDI.
Sensitivity analyses showed that neither history, distance from facility, or survival affected the findings.
Acknowledgments
Funding
This work was supported by the National Institute of Allergy and Infectious Diseases [grant numbers U01AI124255 and R21AI120599].
Krishna Rao has consulted for Bio-K+ International, Inc. and Roche Molecular Systems, Inc. Dr. Rao holds a grant for an investigator initiated study supported by Merck and Co., Inc. Vincent Young has consulted for Bio-K+ International, Inc., Pantheryx, Exarca Pharmaceuticals, and Vedanta.
Appendix A:
The ELISA protocol is described in detail below.
Day One:
Make Bicarbonate Buffer by dissolving one carbonate-bicarbonate buffer capsule (Sigma; C-3041) into 100mL of de-ionized water. Make 1000ug/mL dilutions of Clostridium difficile toxin A (List Biological Laboratories, Inc.; #152C) and toxin B (List Biological Laboratories, Inc.; #155B) by adding bicarbonate buffer. To coat 96-well plates, use 10.6mL of 1ug/mL solutions of toxin A and toxin B. Two plates will be coated with toxin A, and two with toxin B. Plates should be filled in the pattern indicated in Figure 3. Cover the plates with parafilm and incubate at 4°C overnight.
Figure 3:
Plate Fill Pattern for Toxin Coating
Day Two:
Prepare 0.05% PBS-T using 250uL Tween-20 (Fisher Scientific; CN: BP337–100) and 500mL phosphate buffered saline (Gibco by Life Technologies; RN: 10010–023). Make blocking buffer using 4g bovine serum albumin (Sigma Life Science; Lyophilized Powder, Essentially Globulin Free, Low Endotoxin, ≥98%). Prepare sample and controls as shown in Table 4.
Table 4.
Sample and Control Preparations
| 1:100 Sample (500uL) | Patient Sera Sample, PBS-T |
| 1:25 Sample (500uL) | Patient Sera Sample, PBS-T |
| Positive Control (2.6mL) | 1:100 Pooled Human Sera, 1:1000 FcABBA (Feng, University of Maryland), PBS-T |
| Negative Control (2.6mL) | 1:100 Pooled Human Sera from Michigan Medicine, PBS-T |
| True Negative (1.8uL) | PBS-T |
Wash plates three times with PBS-T followed by three times with PBS. Add 200uL blocking buffer to all wells, cover with parafilm, and incubate plates at 37°C for one hour. Repeat washes with PBS-T and PBS. Add 100uL of samples and controls to wells in the pattern indicated in Figure #.
Figure 4:
Plate Fill Pattern for Sample and Control Additions
Note: 40 Samples Total can be run over the 4 plates, in duplicate and with Toxin A and B Ex. Samples 1–20 on Toxin A Plate #1, Samples 21–40 on Toxin A Plate #2, Samples 1–20 on Toxin B Plate #1, and Samples 21–40 on Toxin B Plate #2
Cover plates with parafilm and incubate at 37°C for one hour. Repeat washes with PBS-T and PBS. Add 100uL of 1:2000 polyclonal rabbit anti-human IgG/HRP (Dako; RN: P0214) to all wells. Cover plates with parafilm and incubate at 37°C for one hour. Repeat washes with PBS-T and PBS. Add 150uL ABTS 1-Step (Thermo Scientific; PN: 37615) to all wells. Incubate at room temperature for 30 minutes. Add 100uL of 1% SDS to all wells to stop the reaction. Read absorbance at 410nm and 650nm.
Notes: Temperature within the machine at the time of absorption measurement had a range of 4°C, with a standard deviation of 1°C. Plate washes were performed using a Multi-flo machine set at 250uL, with a 30 second rest between fills.
Appendix B:
Table 5.
Extended Variable Definition List
| Variable | Definition |
|---|---|
| Elixhauser Comorbidity Index | The sum of Congestive Heart Failure, Cardiac Arrythmia, Valvular Disease, Pulmonary Circulation, Peripheral Vascular Disease, Uncomplicated Hypertension, Complicated Hypertension, Paralysis, Neurologic Disorder, Pulmonary Disease, Uncomplicated Diabetes, Complicated Diabetes, Hypothyroidism, Renal Disease, Liver Disease, Peptic Ulcer Disease, AIDS, Lymphoma, Metastatic Cancer, Solid Tumor, Rheumatoid Disease, Coagulopathy, Obesity, Weight Loss, Electrolyte Fluid Therapy, Blood Loss Anemia, Deficiency Anemia, Alcohol Abuse, Drug Abuse, Psychoses, and Depression. All categories had a 1 if the patient had a record of that condition in their medical record. Missing values were ignored. |
| Charlson Comorbidity Index | An age score was applied, where if the age at C. difficile diagnosis was greater than or equal to 50 and less than 60, a 1 was assigned. If age was greater than or equal to 60 and less than 70, a 2 was assigned. If age was greater than or equal to 70 and less than 80, a 3 was assigned. If age was greater than or equal to 80, a 4 was assigned. A 0 was assigned if there was no recorded age. This score was summed with Uncomplicated Diabetes, Complicated Diabetes, Mild Liver Disease, Moderate to Severe Liver Disease, Malignancy, Metastatic Tumor, AIDS, Renal Disease, Chronic Heart Failure, Myocardial Infarction, Pulmonary Disease, Peripheral Vascular Disease, Cardiovascular Accident, Dementia, Hemiplegia, Rheumatic Disease, and Peptic Ulcer Disease. All categories had a 1 if the patient had a record of that condition in their medical record. Missing values were ignored. |
| ISDA Severity Criteria | If the highest WBC lab result value in the 48 hours before and after the C. difficile collection date was greater than 15 then this category had the value of 1. If the highest creatinine lab result vale in the 48 hours before and after the C. difficile collection date was greater than 1.5 times the value of the lowest creatinine lab result value within 30 days prior to the index admission, then this category had the value of 1. Missing values resulted in missing being assigned for this category. |
| Cardiac Arrhythmia | Record of cardiac arrhythmia in the patient’s EHR; 1 = Yes, 0 = No |
| Congestive Heart | Record of congestive heart failure in the patient’s EHR; 1 = Yes, |
| Failure | 0 = No |
| Cardiovascular Accident | Record of a cardiovascular accident in the patient’s EHR; 1 = Yes, 0 = No |
| Diabetes | Record of diabetes in the patient’s EHR; 1 = Yes, 0 = No |
| Electrolyte Fluid Therapy | Record of electrolyte fluid therapy in the patient’s EHR; 1 = Yes, 0 = No |
| Metastatic Cancer | Record of a metastatic tumor in the patient’s EHR; 1 = Yes, 0 = No |
| Myocardial Infarction | Record of myocardial infarction in the patient’s EHR; 1 = Yes, 0 = No |
| Peptic Ulcer Disease | Record of peptic ulcer disease in the patient’s EHR; 1 = Yes, 0 = No |
| Pulmonary Disease | Record of pulmonary disease in the patient’s EHR; 1 = Yes, 0 = No |
| Peripheral Vascular Disease | Record of peripheral vascular disease in the patient’s EHR; 1 = Yes, 0 = No |
| Renal Disease | Record of renal disease in the patient’s EHR; 1 = Yes, 0 = No |
| Rheumatic Disease | Record of rheumatic disease in the patient’s EHR; 1 = Yes, 0 = No |
| Liver Disease | Record of liver disease in the patient’s EHR; 1 = Yes, 0 = No |
| Inflammatory Bowel Disease | Record of a Crohn’s Disease or Ulcerative Colitis diagnosis in the patient’s EHR |
| High Risk Antibiotic Exposure | Record of a medication in Subclass - Cephalosporin Antibiotics - 3rd Generation; Cephalosporin Antibiotics - 4th Generation; Fluoroquinolone Antibiotics; Lincosamide Antibiotics; Penicillin Antibiotic, Extended-spectrum and Beta-lactamase Inhib Comb; Aminopenicillin Antibiotic - Beta-lactamase Inhibitor Combinations; Carbapenem Antibiotics (Thienamycins);Vancomycin and Derivatives administered in the patient’s EHR; 1 = Yes, 0 = No |
| H2 Receptor Blocker Medication Class Exposure | Record of a medication in H2 Receptor Blocker class administered in the patient’s EHR; 1 = Yes, 0 = No |
| Proton Pump Inhibitor Exposure | Record of a medication in Proton Pump Inhibitor class administered in the patient’s EHR; 1 = Yes, 0 = No |
Appendix C:
Table 6.
Unadjusted Logistic Regression between Predictors of Interest and Outcome of Interest
| A. Anti-Toxin A Antibodies | ||||
|---|---|---|---|---|
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | p-value |
| Intercept | 0.12 | 0.041 | 0.274 | <0.001 |
| Anti-Toxin A Antibodies | 2.71 | 1.063 | 8.373 | 0.053 |
| B. Anti-Toxin B Antibodies | ||||
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | p-value |
| Intercept | 0.23 | 0.094 | 0.501 | <0.001 |
| Anti-Toxin B Antibodies | 1.15 | 0.479 | 3.083 | 0.765 |
Appendix D:
Table 7.
Logistic Regression Considering Ribotype
| Variable | Odds Ratio | 95% Confidence Interval Lower Bound | 95% Confidence Interval Upper Bound | p-value |
|---|---|---|---|---|
| Intercept | 0.11 | 0.023 | 0.345 | <0.001 |
| Anti-Toxin A Antibodies | 2.95 | 0.860 | 14.137 | 0.118 |
| Ulcerative Colitis | 5.44 | 1.626 | 19.574 | 0.007 |
| Low Albumin | 0.19 | 0.043 | 0.621 | 0.012 |
| Ribotype 027 Infection | 2.79 | 0.626 | 12.158 | 0.165 |
Footnotes
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- [1].Hensgens MPM, Keessen EC, Squire MM, V Riley T, Koene MGJ, de Boer E, Lipman LJA, Kuijper EJ, Clostridium difficile infection in the community: a zoonotic disease?, Clin. Microbiol. Infect. 18 (2012) 635–645. doi: 10.1111/j.1469-0691.2012.03853.x. [DOI] [PubMed] [Google Scholar]
- [2].Czepiel J, Dróżdż M, Pituch H, Kuijper EJ, Perucki W, Mielimonka A, Goldman S, Wultańska D, Garlicki A, Biesiada G, Clostridium difficile infection: review, Eur. J. Clin. Microbiol. Infect. Dis. (2019). doi: 10.1007/s10096-019-03539-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Rodrigues R, Barber GE, Ananthakrishnan AN, A Comprehensive Study of Costs Associated With Recurrent Clostridium difficile Infection, Infect. Control & Hosp. Epidemiol. 38 (2017) 196–202. doi:DOI: 10.1017/ice.2016.246. [DOI] [PubMed] [Google Scholar]
- [4].Lessa FC, Mu Y, Bamberg WM, Beldavs ZG, Dumyati GK, Dunn JR, Farley MM, Holzbauer SM, Meek JI, Phipps EC, Wilson LE, Winston LG, Cohen JA, Limbago BM, Fridkin SK, Gerding DN, McDonald LC, Burden of Clostridium difficile Infection in the United States, N. Engl. J. Med. 372 (2015) 825–834. doi: 10.1056/NEJMoa1408913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Hu MY, Katchar K, Kyne L, Maroo S, Tummala S, Dreisbach V, Xu H, Leffler DA, Kelly CP, Prospective Derivation and Validation of a Clinical Prediction Rule for Recurrent Clostridium difficile Infection, Gastroenterology. 136 (2009) 1206–1214. doi: 10.1053/J.GASTRO.2008.12.038. [DOI] [PubMed] [Google Scholar]
- [6].Wilcox M, Gerding D, Poxton I, Kelly C, Nathan R, Cornely O, Rahav G, Lee C, Eves K, Pedley A, Tipping R, Guris D, Kartsonis N, Dorr MB, Bezlotoxumab Alone and With Actoxumab for Prevention of Recurrent Clostridium difficile Infection in Patients on Standard of Care Antibiotics: Integrated Results of 2 Phase 3 Studies (MODIFY I and MODIFY II), Open Forum Infect. Dis. 2 (2015). doi: 10.1093/ofid/ofv131.06. [DOI] [Google Scholar]
- [7].Ramsay I, Brown NM, Enoch DA, Recent Progress for the Effective Prevention and Treatment of Recurrent Clostridium difficile Infection, Infect. Dis. (Auckl). 11 (2018) 1178633718758023– 1178633718758023. doi: 10.1177/1178633718758023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Crook DW, Walker AS, Kean Y, Weiss K, Cornely OA, Miller MA, Esposito R, Louie TJ, Stoesser NE, Young BC, Angus BJ, Gorbach SL, Peto TEA, S. 003/004 Teams, Fidaxomicin versus vancomycin for Clostridium difficile infection: meta-analysis of pivotal randomized controlled trials, Clin. Infect. Dis. 55 Suppl 2 (2012) S93–S103. doi: 10.1093/cid/cis499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Wilcox MH, Gerding DN, Poxton IR, Kelly C, Nathan R, Birch T, Cornely OA, Rahav G, Bouza E, Lee C, Jenkin G, Jensen W, Kim Y-S, Yoshida J, Gabryelski L, Pedley A, Eves K, Tipping R, Guris D, Kartsonis N, Dorr M-B, Bezlotoxumab for Prevention of Recurrent Clostridium difficile Infection, N. Engl. J. Med. 376 (2017) 305–317. doi: 10.1056/NEJMoa1602615. [DOI] [PubMed] [Google Scholar]
- [10].Chapman B, Moore H, Overbey D, Morton A, Harnke B, Gerich M, Vogel J, Fecal microbiota transplant in patients with Clostridium difficile infection: A systematic review., J. Trauma Acute Care Surg. 81 (2016) 756–764. doi: 10.1097/TA.0000000000001195. [DOI] [PubMed] [Google Scholar]
- [11].Lofgren ET, Moehring RW, Anderson DJ, Weber DJ, Fefferman NH, A Mathematical Model to Evaluate the Routine Use of Fecal Microbiota Transplantation to Prevent Incident and Recurrent Clostridium difficile Infection, Infect. Control Hosp. Epidemiol. 35 (2014) 18–27. doi: 10.1086/674394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Bartsch SM, Umscheid CA, Fishman N, Lee BY, Is fidaxomicin worth the cost? An economic analysis, Clin. Infect. Dis. 57 (2013) 555–561. doi: 10.1093/cid/cit346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Kelly CR, Ihunnah C, Fischer M, Khoruts A, Surawicz C, Afzali A, Aroniadis O, Barto A, Borody T, Giovanelli A, Gordon S, Gluck M, Hohmann EL, Kao D, Kao JY, McQuillen DP, Mellow M, Rank KM, Rao K, Ray A, Schwartz MA, Singh N, Stollman N, Suskind DL, Vindigni SM, Youngster I, Brandt L, Fecal microbiota transplant for treatment of Clostridium difficile infection in immunocompromised patients, Am. J. Gastroenterol. 109 (2014) 1065–1071. doi: 10.1038/ajg.2014.133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Escobar GJ, Baker JM, Kipnis P, Greene JD, Mast TC, Gupta SB, Cossrow N, Mehta V, Liu V, Dubberke ER, Prediction of Recurrent Clostridium difficile Infection Using Comprehensive Electronic Medical Records in an Integrated Healthcare Delivery System, Infect. Control Hosp. Epidemiol. 38 (2017) 1196–1203. doi: 10.1017/ice.2017.176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Abou Chakra CN, Pepin J, Sirard S, Valiquette L, Risk factors for recurrence, complications and mortality in Clostridium difficile infection: A systematic review, PLoS One. 9 (2014). doi: 10.1371/journal.pone.0098400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Wullt M, Noren T, Ljungh A, Akerlund T, IgG antibody response to toxins A and B in patients with Clostridium difficile infection, Clin. Vaccine Immunol. 19 (2012) 1552–1554. doi: 10.1128/CVI.00210-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Kuehne SA, Cartman ST, Heap JT, Kelly ML, Cockayne A, Minton NP, The role of toxin A and toxin B in Clostridium difficile infection, Nature. 467 (2010) 711. 10.1038/nature09397. [DOI] [PubMed] [Google Scholar]
- [18].Lyras D, O’Connor JR, Howarth PM, Sambol SP, Carter GP, Phumoonna T, Poon R, Adams V, Vedantam G, Johnson S, Gerding DN, Rood JI, Toxin B is essential for virulence of Clostridium difficile, Nature. 458 (2009) 1176. 10.1038/nature07822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Mogle JA, Santhosh K, Rao K, Young VB, Aronoff DM, Higgins PDR, Galecki AT, Natarajan M, Winters S, Kiel MJ, LeBar W, Micic D, Walk ST, Clostridium difficile Ribotype 027: Relationship to Age, Detectability of Toxins A or B in Stool With Rapid Testing, Severe Infection, and Mortality, Clin. Infect. Dis. 61 (2015) 233–241. doi: 10.1093/cid/civ254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].V Martinson JN, Broadaway S, Lohman E, Johnson C, Alam MJ, Khaleduzzaman M, Garey KW, Schlackman J, Young VB, Santhosh K, Rao K, Lyons RH, Walk ST, Evaluation of Portability and Cost of a Fluorescent PCR Ribotyping Protocol for Clostridium difficile Epidemiology, J. Clin. Microbiol. 53 (2015) 1192 LP – 1197. doi: 10.1128/JCM.03591-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Benbrook DM, An ELISA method for detection of human antibodies to an immunotoxin, J. Pharmacol. Toxicol. Methods. 47 (2002) 169–175. doi: 10.1016/S1056-8719(02)00232-0. [DOI] [PubMed] [Google Scholar]
- [22].Kelly C, Pothoulakis C, Orellana J, LaMont JT, Human colonic aspirates containing immunoglobulin A antibody to Clostridium difficile toxin A inhibit toxin A-receptor binding, Gastroenterology. 102 (1992) 35–40. [DOI] [PubMed] [Google Scholar]
- [23].Leung DYM, Kelly CP, Boguniewicz M, Pothoulakis C, LaMont JT, Flores A, Treatment with intravenously administered gamma globulin of chronic relapsing colitis induced by Clostridium difficile toxin, J. Pediatr. 118 (1991) 633–637. doi: 10.1016/S0022-3476(05)83393-1. [DOI] [PubMed] [Google Scholar]
- [24].Warny M, Vaerman JP, Avesani V, Delmée M, Human antibody response to Clostridium difficile toxin A in relation to clinical course of infection., Infect. Immun. 62 (1994) 384 LP – 389. http://iai.asm.org/content/62/2/384.abstract. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Yang Z, Schmidt D, Liu W, Li S, Shi L, Sheng J, Chen K, Yu H, Tremblay JM, Chen X, Piepenbrink KH, Sundberg EJ, Kelly CP, Bai G, Shoemaker CB, Feng H, A novel multivalent, single-domain antibody targeting TcdA and TcdB prevents fulminant Clostridium difficile infection in mice, J. Infect. Dis. 210 (2014) 964–972. doi: 10.1093/infdis/jiu196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Rao K, Higgins PDR, Young VB, An Observational Cohort Study of Clostridium difficile Ribotype 027 and Recurrent Infection, MSphere. 3 (2018) e00033–18. doi: 10.1128/mSphere.00033-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Baggs J, Jernigan JA, Halpin AL, Epstein L, Hatfield KM, McDonald LC, Risk of Subsequent Sepsis Within 90 Days After a Hospital Stay by Type of Antibiotic Exposure, Clin. Infect. Dis. 66 (2017) 1004–1012. doi: 10.1093/cid/cix947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Rogala BG, Malat GE, Lee DH, Harhay MN, Doyle AM, Bias TE, Identification of Risk Factors Associated With Clostridium difficile Infection in Liver Transplantation Recipients: A Single-Center Analysis, Transplant. Proc. 48 (2016) 2763–2768. doi: 10.1016/J.TRANSPROCEED.2016.08.006. [DOI] [PubMed] [Google Scholar]
- [29].Kyne L, Warny M, Qamar A, Kelly CP, Association between antibody response to toxin A and protection against recurrent Clostridium difficile diarrhoea, Lancet. 357 (2001) 189–193. doi: 10.1016/S0140-6736(00)03592-3. [DOI] [PubMed] [Google Scholar]
- [30].Johnson S, Gerding DN, Bezlotoxumab, Clin. Infect. Dis. 68 (2019) 699–704. doi: 10.1093/cid/ciy577. [DOI] [PubMed] [Google Scholar]
- [31].Babcock GJ, Broering TJ, Hernandez HJ, Mandell RB, Donahue K, Boatright N, Stack AM, Lowy I, Graziano R, Molrine D, Ambrosino DM, Thomas WD Jr, Human monoclonal antibodies directed against toxins A and B prevent Clostridium difficile-induced mortality in hamsters, Infect. Immun. 74 (2006) 6339–6347. doi: 10.1128/IAI.00982-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Kyne L, Warny M, Qamar A, Kelly CP, Asymptomatic Carriage of Clostridium difficile and Serum Levels of IgG Antibody against Toxin A, N. Engl. J. Med. 342 (2000) 390–397. doi: 10.1056/NEJM200002103420604. [DOI] [PubMed] [Google Scholar]
- [33].Kim G, Zhu NA, Community-acquired Clostridium difficile infection, Can. Fam. Physician. 63 (2017) 131–132. https://www.ncbi.nlm.nih.gov/pubmed/28209679. [PMC free article] [PubMed] [Google Scholar]
- [34].Kelly CP, LaMont JT, Clostridium difficile — More Difficult Than Ever, N. Engl. J. Med. (2008) 1932–1940. doi: 10.1056/NEJMra0707500. [DOI] [PubMed] [Google Scholar]