Pros and Cons of Alternative Diagnostic Testing Strategies for C. difficile Infection Christopher R. Polage, MD, MAS Associate Professor of Pathology and Infectious Diseases UC Davis
Disclosures Test materials - Cepheid, TechLab, Alere Research Grants - BioFire Diagnostics Consulting - Honoraria - BioFire Diagnostics, genepoc, Seres, SlipChip, Meridian Alere
C. difficile is a leading cause of healthcareassociated infection 453,000 U.S. cases/yr. 200% 1999-2011 1% hospital admissions 2-5 excess hospital days $1.2-5.9 B excess cost/yr. Dubberke CID 2012, Lessa NEJM 2015, Magill NEJM 2014
C. difficile infection (CDI) is mediated by toxins & inflammation
Main types of tests Toxin tests Detects free protein toxin in feces Examples: toxin immunoassay (EIA), cell cytotoxin assay Bacterial or organism tests Detects bacteria in feces (usu. strain with toxin genes) Positive result does not imply active toxin production Examples: toxigenic culture, NAAT (PCR, LAMP), GDH EIA
The CDI diagnostic test debate The major controversy in the diagnosis of symptomatic gastrointestinal infection due to C. difficile is whether laboratory evidence of the C. difficile organism in culture is sufficient or if evidence of one of the C. difficile toxins in stool should be required. Gerding & Brazier CID 1993 50% 50%
A two-phase CDI epidemic 1999-2008: True epidemic Novel hypervirulent strain 95% labs used toxin test only Toxin+ = CDI Toxin-/C. difficile+ not detected Occasional reports missed CDI cases 2009-2011: Test-related epidemic First FDA-cleared PCR test Labs switch to PCR/NAAT tests Toxin+ and Toxin-/C. difficile+ detected CDI Rates by 1.5-2X
Why is there so much confusion about performance of CDI tests? There is no true reference test for clinical disease Most signs and symptoms of CDI are non-specific Non-CDI diarrhea & C. difficile colonization are more common than CDI Negative experience with toxin assays Examples: toxin- toxin+ conversion; toxin- complicated CDI Fueled desire for absolute sensitivity regardless of specificity Lab studies without clinical correlation Promoted belief that all toxin-/naat+ or TC+ = CDI Exaggerated insensitivity of toxin tests
Shifting priorities, ordering practices, & test methods over time Historically (1979-2005) Priority: specificity High bar for testing & diagnosis Toxin tests favored Culture = slow, non-specific Belief: most toxin- pts don t have CDI Recently (2005-2013) Priority: sensitivity Low bar for testing & diagnosis Shift to organism tests NAAT = available, rapid Belief: most or all toxin- /PCR+ pts have CDI
TC overdiagnosis >>> Toxin- CDI Gerding et al. Arch Intern Med 1986 149 case patients 108 diarrhea-free controls Stringent inclusion criteria 6 stools/36 hrs no other obvious cause 64% got endoscopy
TC overdiagnosis >>> Toxin- CDI Gerding et al. Arch Intern Med 1986 1. Toxin 149 case sensitivity patients depends on reference for disease: 1. 108 Diarrhea diarrhea-free + C. diff controls ( CDAD ): CTA = 70% 2. Stringent Pseudomebranous inclusion colitis (PMC): CTA = 90% 2. Cases criteria diagnosed only by stool culture showed essentially 6 stools/36 no differences hrs from controls, 21% of whom no other obvious cause had asymptomatic colonization. We estimate that only 20% 64% of got these endoscopy cases had diarrhea due to C difficile.
TC overdiagnosis >>> Toxin- CDI Planche et al. Lancet Infect Dis 2013 Group 1 (Tox+/C. diff+) Group 2 (Tox-/C. diff+) Group 3 (Tox-/C. diff-) 1 vs. 2 1 vs. 3 2 vs. 3
PCR overdiagnosis >>> Toxin- CDI Polage et al. JAMA IM 2015
PCR overdiagnosis >>> Toxin- CDI Polage et al. JAMA IM 2015 Outcome Tox+/PCR+ (n=131) C. difficile positive Tox-/PCR+ (n=162) C. difficile negative Tox-/PCR- (n=1123) P-value Complication a 10 (7.6%) 0 3 (0.3%) <0.001 Death b, c 11 (8.4%) 1 (0.6%) 0 <0.001 Complication or death 18 (13.7%) 1 (0.6%) 3 (0.3%) <0.001 a ICU care, megacolon, colectomy related to CDI b Death within 30 days related to CDI c All-cause 30 day mortality: 14 (10.7%); 23 (14.2%); 98 (8.7%); P=0.08
2015 observational study with outcomes Polage et al. JAMA Intern Med 2015 Toxin-/PCR+ patients had Shorter duration of symptoms Few or no CDI-related complications Similar outcomes to Toxin-/PCR- patients with little or no treatment in most cases
More data supporting the utility of toxin tests 14 of 18 (78%) studies show less symptoms / better outcomes in toxin- pts CDI drugs have better efficacy in toxin+ pts. in clinical trials Lower PCR Ct predicts toxins, severe CDI, & outcomes Polage 2012; Planche 2013; Longtin 2013; Beaulieu 2014; Bunnell 2016; Davies 2015; Reigadas 2016; Fang, Polage, Wilcox 2017 Toxin- complicated CDI is uncommon overall UC Davis 0% Spain 0% Quebec 0% Pittsburgh 0.3% Switzerland 0.6% Cleveland 1.5% Spain 3.9% Polage 2015; Polage 2012; Reigadas 2016; Beaulieu 2014; Dallal 2002; Erb 2015; Guerrero 2011; Origuen 2017
Potentially conflicting studies Case reports/ case series of toxin- patients with complicated CDI Chicago - Lashner et al. Am J Gastroenterol 1986; PMID: 3766495 Canada - Pepin et al. CMAJ 2004; PMID: 15337727 Pittsburgh - Dallal et al. Ann Surg 2002 Studies with clinical data showing no statistical difference UCLA - Humphries et al. J Clin Microbiol 2013; PMID: 23269736 Memorial Sloan-Kettering - Kaltsas et al. J Clin Microbiol 2012 Michigan - Rao et al. Clin Infect Dis 2015 Cleveland VA - Guerrero et al. Clin Infect Dis 2011 UK - Berry et al. J Hosp Infect 2014
Why some studies don t see differences b/w toxin+ & toxin-/naat+ pts Type II statistical error (inadequate power) Few complicated CDI cases Bias & misattribution Failure to attribute non-specific clinical outcomes Failure to consider or exclude alternative causes Lack of control group Shared risk factors between CDI, C. difficile colonization, & non-cdi diarrhea
Causes of toxin-/pcr+ results Colonization in patient with No or borderline symptoms (inappropriate testing) OR Alternative cause of diarrhea (laxative, meds, etc.) Prior antibiotic treatment with activity against C. difficile Usually metronidazole for non-cdi indication Prior empiric therapy for CDI Pre-analytic toxin degradation (rare)
What s going on?
Overdiagnosis causes harm Matta, Greenberg, & Singh JAMA Intern Med 2015 70 yo F w/ metastatic RCC and multiple episodes of C. difficile PCR+ diarrhea x 2 yrs. Presents w/ 12-15 nonbloody, watery BMs/day, abd. cramps, afebrile, WBC 13.7K Started on PO vancomycin & metronidazole for rcdi Day 6 diarrhea unchanged, diagnosis re-evaluated
Overdiagnosis causes harm Matta, Greenberg, & Singh JAMA Intern Med 2015 Chart review showed prior treatment courses had minimal effect on diarrhea. Prior tests for other enteric pathogens were negative; Abd CT showed mild colitis; colonoscopy was non-diagnostic and negative for pseudomembranes. Review of the medication history revealed that the diarrhea first started after axitinib was prescribed for RCC and diarrhea is a common adverse effect.
Overdiagnosis causes harm Matta, Greenberg, & Singh JAMA Intern Med 2015 After discussion with her oncologist, the axitinib and antibiotics were discontinued. The patient showed dramatic improvement over the next 3 days. Diagnosis: Medication-induced diarrhea w/ persistent C. difficile colonization
More harms of overdiagnosis Antibiotics are not benign Disrupt gut-microbiota homeostasis -> MDROs, sepsis? Misdiagnosis delays recognition of true cause of illness & outbreaks Hospitals get penalized for excess CDI cases
Testing option 1: Use toxin immunoassay alone Pros Reasonably good correlation with clinical disease Detects most severe/complicated CDI Less overdiagnosis versus TC or NAAT Cons Misses some CDI cases Can miss occasional severe or complicated CDI cases (uncommon/rare)
Testing option 2: Use PCR/NAAT alone Pros High analytical sensitivity for toxigenic C. difficile High NPV Faster than toxigenic culture DNA less susceptible to pre-analytic degradation vs toxins Cons Low clinical specificity Doesn t distinguish between colonization & CDI Probably leads to overdiagnosis & overtreatment
Testing option #3: ESCMID algorithm High sensitivity/ high NPV test High specificity toxin test Evaluate -/+ pts before treating Crobach et al. CMI 2016
Pros & cons of ESCMID algorithm Pros Best of both worlds? High NPV first test Toxin EIA second test helps interpret positive results clinically Reduces PCR/NAATrelated overdiagnosis Cons Requires two tests for GDH/NAAT+ patients
Testing option 4: Use modified PCR cutoff to predict toxins, need for treatment, & outcomes
Pros & cons of modified PCR threshold Pros Better clinical predictive value vs current FDAapproved PCR/NAAT Reduces PCR overdiagnosis Cons Early, limited data Not cleared by FDA LDT May still overdiagnose some patients Significance of toxin- /modified PCR+ pts. is unknown
Potential for diagnostic stewardship to reduce inappropriate testing
Most doctors do not test selectively Banaei et al. NEJM 2015
Most doctors do not treat selectively Buckel et al. ICHE 2015
Frequently asked questions How does the UK/European diagnostic approach differ from 2-step testing in the U.S.? Should toxin-/pcr+ patients with diarrhea be put in contact isolation? How do I decide whether to treat a toxin-/pcr+ pt?
Summary The insensitivity of toxin tests is exaggerated Organism tests overcall CDI substantially Overdiagnosis does cause harm ESCMID 2-step approach & novel strategies can help reduce PCR/NAAT-related overdiagnosis
Coming soon IDSA Clinical Practice Guidelines for Clostridium difficile Infection in Adults and Children - IDSA/SHEA (2017 Update) New or confirmation of old recommendations on: Epidemiology and surveillance Diagnosis Infection control and prevention Treatment
Questions?