Diagnosis of Clostridium difficile Infections in Children

Abstract

The detection and diagnosis of Clostridium difficile infection in pediatric populations have some unique considerations in comparison to testing in adults. The testing methodologies, including toxigenic culture, cell cytotoxicity, antigen detection, and, more recently, molecular testing, are the same in all age groups. However, limited data exist on the specific performance characteristics in children. In this review, we focus on the challenges of testing in pediatric populations and assess the available data on test performance in these populations. Additionally, a review of the existing guidance for testing is provided.

INTRODUCTION

The roots of Clostridium difficile in pediatrics are deep, tracing back to the discovery of the organism. Hall and O'Toole first described the organism as part of the normal flora in neonates, naming it Bacillus difficilis (1). Much knowledge about C. difficile with its pathogenic potential, ranging from asymptomatic carriage to mild diarrhea, severe colitis, and toxic megacolon, has since been gained. It is well recognized that C. difficile infection (CDI) is a leading cause of nosocomial and antibiotic-associated illnesses in adults, posing a huge burden on the health care system, and thus is a major public health concern. Concomitant with the increase of CDI in adults in the past decade, the rates of hospitalized children with the infection have been increasing as well. In a study of hospitalized children in the United States, Nylund et al. showed that the number of CDI has risen from 3,565 cases in 1997 to 7,779 cases in 2006 (2). The number of cases per patient-days in the hospital has increased (3), and overall there has been a rise in hospitalizations attributable to C. difficile in children (4). While the rates of CDI have increased, the severity of disease (e.g., colectomy and in-hospital deaths) has not increased in pediatric patients as it has in adults (3).

Different approaches have been employed in the clinical laboratory to aid in the diagnosis of CDI. The changes in methods and the algorithms used to detect toxigenic C. difficile are a result of the increase in sensitivity and specificity of these assays, as well as the demand for accurate and timely results. A number of reviews have been written on the diagnosis of C. difficile (1, 5), but few have dealt specifically with the performance characteristics and the use of testing in children (5). This minireview is a summary of the clinical guidelines and available data for diagnostic testing for C. difficile focused on pediatric populations.

PATHOGENESIS AND EPIDEMIOLOGY OF CLOSTRIDIUM DIFFICILE IN CHILDREN

C. difficile is an anaerobic, Gram-positive, spore-forming bacillus that can be found in the environment and the gastrointestinal tract of animals and humans. The pathogenicity of C. difficile is attributed to the production of two protein toxins designated A (enterotoxin) and B (cytotoxin), encoded by the tcdA and tcdB genes, respectively. They are found on the pathogenicity locus (PaLoc) and both cause significant disease. Toxins A and B are glycosyltransferases that can inactivate GTP-binding proteins. The toxins enter the cytoplasm of the colonocytes by binding to receptors that are found on the luminal-facing side of these cells. Once inside the cells, they inactivate a number of proteins involved in cytoskeleton organization, triggering the apoptosis of the cells. This results in an acute inflammatory reaction, leading to diarrhea and colitis (1, 6).

Colonization with C. difficile in infants.

The colonization rate of healthy infants has been studied either by culturing techniques or by PCR. For infants <1 month of age, the organism is recovered at high rates with an average colonization rate of 37% (7) and a range of 0 to 61% being cited (8). Between 1 and 6 months of age, the colonization rate is still high at 30% and drops to about 10% by the end of the first year of life (7). The asymptomatic carriage rate continues to drop until about 3 years of age, when it stabilizes to carriage rates of 0 to 3%, similar to those found in adults (7). C. difficile toxin titers found in the stool of healthy infants are similar to those found in adults with C. difficile-associated diarrhea (9). There is a progressive increase in serum IgG antibody concentrations against toxins A and B between birth and 24 months of age (7). Together these findings suggest that toxigenic strains are colonizing the gut of infants, but they remain healthy without developing acute colitis. The exact mechanism of protection of the infants against C. difficile toxins is unclear, but a number of theories have been proposed, including the lack of toxin receptors on the surface of the intestinal cells, the protective action of breast milk, and the defense provided by other neonatal intestinal flora. In a newborn rabbit ileum model, researchers found that there were no binding sites for toxin A on the surface of the cells. Additionally, even with maximum concentrations of toxin A applied on newborn and young rabbit ileum cells, the effects were minimal compared to the severe mucosal damage caused in adult rabbits (10). Although these data derive from an animal model, they suggest a possible mechanism by which human infants may avoid disease due to lack of toxin receptors on the surface of their colonocytes, which in turn inhibits internalization of toxins that lead to colonic damage. More research is needed to better characterize the actual receptors present in human infant gut tissue. It may be that a lower number of toxin receptors are expressed on the surface of infant colonocytes or that the binding sites of these receptors are altered, leading to lower or absent binding capacity. Formula-fed infants have higher rates of colonization with C. difficile than breast-fed infants, but these differences seem to be gone by 12 months of age when weaning occurs (7). The exact mechanism of protection provided by breast milk is unclear. Various components of breast milk (galactose and colostrum) have been shown to inhibit binding of C. difficile on colonic cells (11), while human milk was found to bind and neutralize toxin A itself, rendering it inactive in the gut (12). Neutralization of toxin A would allow the asymptomatic colonization of infants with C. difficile without the side effects an active toxin A may have. Additionally, breast-fed babies have lower fecal pH, favoring the growth of Bifidobacterium, which inhibits the growth and binding of C. difficile to enterocytes (13). Gradually, other commensal colonic flora compete with C. difficile and give rise to the adult colonic flora that is carried by individuals for the rest of their lives (14).

While in most instances, detection of C. difficile in infants will be noncontributory to diarrheal illness and represents colonization, there are rare cases of C. difficile-associated deaths in this age group (15). In a previous study, Kim et al. noted that 26% of the reported C. difficile-associated disease cases in their cohort occurred in children <1 year of age. The authors acknowledged the possibility that this finding might either represent disease to a previously unrecognized degree or receipt of unnecessary treatment in a large number of children (3). Studies are needed to determine how often C. difficile in the <1 year of age population causes true disease.

Hospital-associated C. difficile infection.

The rate of symptomatic CDI in hospitalized children has been increasing, with a number of associated risk factors, including antibiotic use, immunosuppression, and bowel dysfunction (2). Unlike among adults, there are no data suggesting increased severity of disease (e.g., colectomy and in-hospital deaths attributable to CDI) among pediatric patients (2, 3).

Hospitalized children diagnosed with CDI have been shown to have poor outcomes with significant morbidity and increased risk of death (2). Additionally, CDI in pediatric patients has also been associated with increased length of stay and overall higher total hospital costs (16). It is clear that in hospitalized patients and those with other comorbidities, there is a need for a timely and accurate diagnosis of CDI so that the appropriate therapy and intervention measures are in place to prevent unfavorable outcomes. In a multicenter study across 41 children's hospitals, Vendetti et al. attempted to identify the risk factors associated with in-hospital mortality for children with diagnosed CDI. Factors that were independently associated with increased mortality included older age (>13 years), receipt of gastric acid suppression, underlying malignancy, cardiovascular disease, and hematologic/immunologic conditions (17). In this study, the laboratory data for the type of test used for the diagnosis or confirmation of CDI were not available. Therefore, the differences in the performance characteristics of the various methodologies for the detection of CDI in children in these high-risk populations were not assessed. In a separate study assessing risk factors for children with recurring CDI, malignancy, recent surgery, and antibiotic exposure were the most significant factors (18). Both enzyme immunoassays (EIAs) and nucleic acid amplification tests (NAATs) were used as laboratory tests for the detection of C. difficile; however, no data on the differences in test performance were presented. Solid organ transplantation also appears to be a risk factor for a higher incidence of CDI in children, especially those between 1 and 4 years of age (19).

As with infants, high rates of asymptomatic colonization have been reported among hospitalized children. Using PCR for C. difficile detection in hospitalized patients with and without diarrhea, Leibowitz et al. found that there was no statistically significant difference between the positivity rates of asymptomatic and symptomatic children (24% versus 19% of samples, respectively) (20). Pediatric oncology patients may also be colonized or have prolonged shedding following or in conjunction with diarrhea (21). The apparent asymptomatic colonization or shedding can complicate the diagnosis of diarrheal illness in hospitalized children.

Among patients with inflammatory bowel disease (IBD), the rates of CDI have increased between 2- and 4-fold in the past decades (22, 23). These patients can acquire C. difficile in an outpatient setting, and the presence of C. difficile can contribute to increased morbidity and mortality (1). When patients present with diarrhea during a relapse of IBD, they are often evaluated for CDI, but the best testing method for these patients has not been determined. In children with IBD, studies to characterize CDI are limited. Lamousé-Smith et al. assessed the prevalence of C. difficile in patients with IBD and non-IBD gastrointestinal disease (controls) in both diarrheal and nondiarrheal stool samples. There was no statistically significant difference in the number of samples positive for C. difficile by NAAT between the patients with IBD and the controls and between patients with active or inactive disease (24). Additionally, of the 28 PCR-positive results in this study, only 3 were positive by EIA, suggesting that the detection of C. difficile might reflect colonization and not necessarily the cause of symptoms (24).

Community-associated C. difficile infection.

The rate of detection of C. difficile in children with community-acquired (CA), or outpatient-acquired, illness is increasing. A retrospective study in a tertiary pediatric center showed that there was a statistically significant increase in the CA CDI cases: 11% in a period of 5 years (25). Historically, outpatient populations have been thought to be at lower risk of CDI. In a laboratory-based surveillance study by Wendt et al., the rate of C. difficile detection from testing of children in the community was 71% (26). However, this study and others have not correlated the positive laboratory findings with clinical data, making the link between a positive test and the causality of disease unclear (5). In a recent retrospective pediatric study comparing clinical presentation and associated risk factors in those diagnosed with hospital-acquired (HA) versus CA CDI, the authors found that many of the children diagnosed with CA CDI by PCR lacked risk factors and were more often <2 years of age, a group with a known high rate of asymptomatic colonization (27). The positive PCR results were not always confirmed with additional C. difficile testing, and evaluation for an alternate etiology was often limited, particularly for viral causes of diarrhea (27). The overall conclusion of these authors was that many of the CA CDI were misdiagnosed and treated inappropriately. Clearly, the impact of C. difficile detection in the pediatric community care setting warrants further study to determine if there is a true increase in CA CDI and if it is wise to routinely test for C. difficile, particularly with a NAAT, in the absence of any of the established risk factors.

GUIDELINES ADDRESSING C. DIFFICILE INFECTION: GENERAL TESTING RECOMMENDATIONS AND SPECIFIC RECOMMENDATIONS FOR CHILDREN

The severity of the disease, coupled with the public health implications of CDI as the most common hospital-acquired infection in the United States and other countries (28), prompted a number of professional societies to publish guidelines regarding CDI with recommendations for its diagnosis and treatment. Below, we highlight the important aspects of these guidelines as they relate either generally or specifically to testing in pediatric populations.

All existing guidance documents for the diagnosis of CDI clearly indicate that the key criterion for clinical diagnosis is the detection of toxigenic C. difficile in loosely formed stools. All guidelines follow the World Health Organization definition for diarrhea (the presence of 3 or more loose stools in a 24-hour period) and state that testing should be performed only on diarrheal unformed stool, unless ileus is suspected. One of the most pressing issues now relates to which testing methodology to use. Concerns have emerged that the NAAT is too sensitive, while the use of the toxin EIA alone lacks sensitivity. Currently there is no single test or algorithm that has been universally accepted.

Guidelines from the Infectious Disease Society of America and the Society for Healthcare and Epidemiology of America (IDSA/SHEA) do not recommend the use of the toxin EIA as a standalone test due to low sensitivity; however, other testing approaches for the detection of the organism are not clearly endorsed and issues specific to children are not addressed (29).

The American College of Gastroenterology published guidelines supplemental to those of IDSA/SHEA, recommending NAAT or a glutamate dehydrogenase (GDH) screening test as part of a two- or three-step algorithm but do not address diagnostic issues or the implications of positive results within the pediatric population (30).

The European Society of Clinical Microbiology and Infectious Diseases (ESCMID) also acknowledges that the best standard laboratory test for the diagnosis of CDI has not been established. They endorsed a two-stage algorithm where a positive first test (culture, antigen detection, or amplified molecular test) is confirmed with other tests or a reference method. This recommendation is based on the findings of a large study where various methodologies and algorithms were employed to identify the best testing strategy. According to the authors of that study, a two-stage algorithm and combination of assays improves the diagnosis of CDI, but the performance of those tests depends on the reference method (31). In these guidelines, there is no discussion of the diagnostic performances of the tests in pediatric patients (32).

The National Health Services (NHS) in Britain published guidelines for testing algorithms and reporting of C. difficile infection based on the study by Planche et al. (33). The NHS recommends testing all loose stools, irrespective of risk factors for CDI, using a two-step algorithm with a screening assay (NAAT or GDH EIA) followed by a toxin EIA if the result of the screening test is positive. Confirmed positive results are indicative of CDI, while positive results not confirmed by the toxin EIA are most likely from patients that are excreting C. difficile. However, these patients are not at risk of complications related to CDI, similar to those who test negative with the initial screening test. All confirmed positive results are reported to a national surveillance database. The guidelines suggest that children <2 years of age should not be tested, and, if the testing is performed and is positive, it is not mandatory to report the result in the national database (34).

The American Academy of Pediatrics (AAP) published guidelines in 2013 regarding CDI in children that address diagnosis and testing. They recommend testing for CDI only in children with diarrhea (3 or more loose stools in a 24-hour period). In recognition of the high colonization rates of children up to 3 years of age, testing should be limited in the following manner. It is recommended that any testing in children <1 year of age should be avoided unless there are specific gut motility disorders, such as Hirschsprung's disease, or in the setting of an outbreak. For children between 1 and 3 years of age, testing can be considered if alternative causes, such as viral infections, have been explored. For positive results in children between 2 and 3 years of age, an alternative etiology for symptoms should be pursued. Testing in children >3 years of age with a positive result indicates probable CDI. The guidelines further suggest that a test of cure is not recommended. With a recurrence of symptoms, repeat testing should not be performed before 4 weeks after the initial positive test result. A specific testing methodology is not clearly endorsed; however, NAAT is mentioned as the most sensitive method for the detection of the organism (35). The guidelines do acknowledge that there are not enough published data to assess the diagnosis of disease in children with a NAAT alone (35).

C. DIFFICILE DIAGNOSTIC TESTS

A number of laboratory tests are available to aid in the diagnosis of CDI. They differ in what is being detected with some identifying the presence of the bacteria, some detecting the presence of toxins, and some detecting nucleic acid for genes associated with toxin production. All of the tests described below can be used in both adult and pediatric patients. It is important to understand the differences in these methods and the clinical implications for a positive result in children.

TC.

Culturing for the recovery of C. difficile from stool specimens was historically the initial method used in the clinical laboratory to detect the presence of the pathogenic bacteria. In order to reduce the normal flora and increase the recovery of C. difficile, stool samples are processed by heat or alcohol pretreatment and then inoculated on C. difficile selective media (36). Both toxigenic and nontoxigenic strains may be isolated and a confirmatory test for the expression of toxin proteins is required. This approach is lengthy and technically complicated with a turnaround time of almost a week. Toxigenic culture (TC) is mainly employed in research settings and for epidemiological studies where the isolation of the organism can provide useful information, but it is not routinely used in the clinical laboratory.

CCTA.

A direct cell cytotoxicity assay (CCTA) relies on the neutralization of C. difficile toxins using an antitoxin to enhance the specificity of toxin detection. An aliquot of diluted, buffered, and filtered stool sample along with the tested stool sample mixed with C. difficile antitoxin is added into different wells with monolayers of cultured cells, and the wells are observed for the presence of a cytopathic effect (CPE). The test is considered positive if a CPE is seen in ≥50% of cells at 48 h, and the effect is inhibited in the wells containing C. difficile antitoxin (36). Nonetheless, it may be negative if the concentration of the toxin in the sample is too low or has been degraded due to improper specimen handling or even pretreatment of the patient with antibiotics prior to specimen collection (1). The assay is labor intensive, requires expertise in both cell culture maintenance and interpretation of the results, and has a long turnaround time. In studies comparing the CCTA versus TC, the CCTA has a sensitivity of 75 to 85% (37). The relatively long turnaround times of both the TC and CCTA limit their clinical utility, particularly compared to that of other methods. Both, however, are endorsed as tests against which other testing methodologies are compared (1).

EIAs.

EIAs used for the detection of C. difficile consist of those that detect either the presence of toxins A and/or B or the presence of the metabolic enzyme GDH. Both types of assays are available as microwell EIAs or in lateral flow immunochromatographic formats.

Toxin immunoassays use monoclonal or polyclonal antibodies against toxins A and/or B to detect free toxin in stool samples. Due to their low cost and ease of use, they had been adopted by a large number of clinical microbiology labs as the diagnostic test for C. difficile. It has become clear though that toxin EIAs display variable sensitivity and specificity ranging from 42.3 to 96.8% and from 84 to 100%, respectively, depending on the comparator assay used for the evaluation of the EIA (1). The IDSA/SHEA, American College of Gastroenterology, and NHS guidelines do not recommend the use of a toxin immunoassay as a standalone test for the diagnosis of C. difficile (29, 30, 34).

In a study of pediatric inpatients, the use of a toxin EIA for the diagnosis of C. difficile-related disease was evaluated. A total of 112 stool samples positive by EIA were tested for the presence of toxigenic C. difficile with TC and approximately one-third of those (40 samples) were found to be negative. Specimens that were negative by TC were also tested by PCR and GDH EIA, while some were further processed by broth enrichment or serial dilution and plated again. None of the reprocessed samples were positive, and only one of the tested specimens was positive by PCR. This high rate of false positives was attributed by the authors to the low incidence of CDI in their patient population, resulting in a positive predictive value (PPV) of 64% (38). The authors did not explore other possible explanations for the high false-positive rate with the EIA, such as low culture sensitivity or technical issues. The study results raise concerns about the use of toxin EIAs in the diagnosis of CDI in children due to the high rates of false positives; however, this seems in contrast with other publications demonstrating a lack of sensitivity with EIAs (39). The AAP guidelines also question the usefulness of EIAs in the pediatric population due to poor performance characteristics and the low positive predictive value (35).

GDH is produced in high levels by toxigenic and nontoxigenic strains of C. difficile and is known to cross-react with Clostridium sordellii (nontoxigenic strain). GDH assays have a high sensitivity as a screening test, ranging from 87.6 to 100% (1) and a negative predictive value of >97% (1). In addition, their low cost and ease of performance make them an attractive screening method for ruling out C. difficile. Due to the confirmation step needed when there is a positive GDH result, combination EIAs with the detection of GDH and toxin done in one test have been marketed. These assays can reliably predict C. difficile disease when both results are positive and can rule out disease when both are negative. A confirmatory test needs to be performed when the assay results disagree. This 2-step algorithmic approach (GDH/toxin EIA with reflex to another method, if discordant) was assessed in a pediatric cohort where the dual EIA system results were compared to those for the CCTA and NAAT. It was shown that the dual EIA results can be effectively used for accurately reporting positive and negative samples. For the samples that were discordant, the discrepant results needed to be resolved by another method to maximize the sensitivity (40). The CCTA and NAAT, as the secondary methods of result confirmation, performed similarly in this cohort, suggesting that the user-friendly NAAT is acceptable as a confirmatory test in a 2-step algorithm. In a large multicenter study in the United Kingdom that demonstrated increased mortality in association with the presence of toxins in the stool sample, the authors advocated for the use of a molecular or an enzymatic assay (GDH) as a screening test and confirmation with a toxin EIA for the most accurate prediction of disease (33). The study included patients older than 2 years, but the pediatric patients were not analyzed separately; therefore, the predictive value of a positive EIA for assessing disease in these patients is still unknown.

Molecular methods.

A number of different molecular platforms using nucleic acid amplification methods for the detection of C. difficile directly from stool samples have been FDA cleared. These include either standalone tests looking specifically for C. difficile or for C. difficile as part of larger multiplex molecular panel utilized for the detection of a variety of gastrointestinal pathogens. The available assays target the genes encoding the toxin tcdA or tcdB or a combination of both. Some also include the tcdC 117 nucleotide deletion, which can be used for the presumptive identification of the 027/NAP1/BI strain, a strain isolated both from adult and pediatric patients in North America and Europe (2, 3). This strain has been associated with increased severity of CDI in adults (1). The methodology used for the assays varies and includes qualitative real-time PCR using molecular beacons or fluorescent probe primers, loop-mediated isothermal amplification, helicase-dependent amplification, and multiplex real-time or conventional PCRs. There are currently 10 manufacturers with FDA-cleared single analyte detection assays and 2 with multiplex platforms that include the detection of C. difficile in their testing menu. Table 1 is a summary of the FDA-cleared molecular assays and the available pediatric data from their package inserts (PI). Some of the PI provide stratified data by age, but only one (illumigene; Meridian Bioscience, Cincinnati, OH) provides specific performance characteristics by age group. Table 2 summarizes published studies with performance characteristics of FDA-cleared molecular assays that were assessed either exclusively in a pediatric cohort or in a majority of pediatric samples (<21 years of age).

TABLE 1.

FDA-cleared molecular assays for the detection of C. difficile and available pediatric data from package inserts

Assay	Methodologya	Reference method	Pediatric patient data from package inserts
BD GeneOhm, BD Max Cdiff	qPCR/molecular beacons	Cytotoxicity assay	No data; no limitation of use in pediatric samples
Prodesse ProGastro	qPCR/fluorescent probe primers	Cytotoxicity assay	Data stratified by age group; patients included >2 yr
Cepheid GeneXpert	Multiplex qPCR	Toxigenic culture	Data stratified by age group; patients included >2 yr
Meridian illumigene	Loop-mediated isothermal amplification	Cytotoxicity assay	Data stratified by age group; data analyzed for patients <2 yr separately
Quidel AmpliVue	Helicase-dependent amplification	Cytotoxicity assay	No data; no limitation of use in pediatric samples
Focus Diagnostics Simplexa	qPCR/fluorescent probe primers	Toxigenic culture	No data; no limitation of use in pediatric samples
Great Basin Portrait	Helicase-dependent amplification	Direct and enriched toxigenic culture	Data stratified by age group; patients included >2 yr
Nanosphere Verigene	qPCR combined with nanoparticle array hybridization	Direct and enriched toxigenic culture	No data; no limitation of use in pediatric samples
IMDx Abbott m2000	qPCR/fluorescent probe primers	Cytotoxicity assay	Data stratified by age group; data analyzed for patients <2 yr separately
Primera ICEPlex (Modaplex)	qPCR/fluorescent probe primers	Direct toxigenic culture	No data; no limitation of use in pediatric samples
xTAG GPP Luminex	Multiplex RT-PCR/PCR	Cytotoxicity assay	Pediatric patients were included; results were not analyzed per age group
Biofire FilmArray GI panel	Multiplex nested PCR	tcdA and tcdB PCR	Pediatric patients were included; data stratified per age group

^aqPCR, quantitative PCR; RT, reverse transcription.

TABLE 2.

Published performance characteristics of available FDA-cleared molecular assays in studies including pediatric patients^a

Assay	Study (reference)	Comparative standard	No. of patient samples (age range) (% pediatric)	Sensitivity (%)	Specificity (%)	PPVb (%)	NPVc (%)
BD GeneOhm	Selvaraju et al., 2011 (41)	Toxigenic culture	200 (unavailable) (100)	89.6	96.7	89.6	96.7
	Hart et al., 2014 (42)	Toxigenic cultured	150 (11 days–17 yr) (100)	89	99	98	95
Prodesse ProGastro	Selvaraju et al., 2011 (41)	Toxigenic culture	200 (unavailable) (100)	100	93.4	82.8	100
Cepheid GeneXpert	Leibowitz et al., 2014 (20)	Toxigenic culture	262 (1–18 yr) (100)	95	91	69	99
Meridian illumigene	Ota and McGowan, 2012 (40)	Cytotoxigenic culture or compositee	141 (1–18 yr) (100)	89	98	92	97
	Hart et al., 2014 (42)	Toxigenic cultured	150 (11 days-17 yr) (100)	89	100	100	95
	Antonara et al., 2015 (43)	Quidel AmpliVue or toxigenic culture	758 (36 days–98 yr) (61)	96.1	99.8	99.2	99.2
Quidel AmpliVue	Antonara et al., 2015 (43)	Meridian illumigene or toxigenic culture	758 (36 days-98 yr) (61)	96.1	99.2	96.1	99.2
xTAG GPP Luminex	Beckmann et al., 2014 (45)	Dual EIA and DNA hybridization toxin assayf	120 (6 days–21 yr) (100)	100	100	100	100
	Patel et al., 2014 (46)	Cepheid Xpert and 16S sequencing	211 (unavailable) (100)	100	100	100	100
Biofire FilmArray GI panel	Stockmann et al., 2015 (50)	Meridian illumigene	378 (1–25 yr) (>75)	95	99	NAg	NA
	Buss et al., 2015 (49)	tcdA and tcdB PCR	1,556 (<1–>65 yr) (62)	98.8	97.1	NA	NA

^aPediatric patients were defined as those up to 21 years old; studies were either exclusively in a pediatric cohort or in a majority of pediatric samples. Additional FDA-cleared assays have not been evaluated exclusively in pediatric patients (Focus Diagnostics Simplexa, Great Basin Portrait, Verigene, IMDx Abbott m2000, and Primera ICEPlex).

^bPPV, positive predictive value.

^cNPV, negative predictive value.

^dToxigenic culture was defined by the authors as growth on cycloserine cefoxitin fructose agar (CCFA) or Cdiff CHROMagar (bioMérieux, France) confirmed with toxin gene PCR.

^eComposite standard defined by the authors as a positive result for both Meridian illumigene and positive by another enzyme immunoassay.

^fC. Diff Quik Chek Complete kit (Alere GmBH, Kolm, Germany) and DNA hybridization assay (Hain Lifescience GmBH, Nehren, Germany).

^gNA, nonapplicable.

To date there have been limited studies looking specifically at assay performance in pediatric patients. Selvaraju et al. tested BD GeneOhm (BD Diagnostics Inc., Sparks, MD) and Prodesse ProGastro (Hologic Gen-Probe Inc., San Diego, CA) in pediatric patients. Both showed better sensitivity and specificity than an EIA alone in comparison to TC. However, the authors suggest that a 2-step algorithm utilizing a dual EIA screening test (GDH and toxin) followed by confirmation with a molecular assay provides the potential of reducing laboratory test costs while at the same time reporting fast and accurate results (41). Hart et al. compared the BD GeneOhm and illumigene assays in pediatric patients to a GDH/toxin EIA, culture on cycloserine cefoxitin fructose agar (CCFA), and CCTA. They evaluated the performance characteristics of the assays as standalone tests and as part of a 2-step algorithm. The molecular methods had performance comparable to that of GDH alone and higher sensitivity than the EIAs and CCTA. The 2-step algorithms (GDH+PCR) showed reduced sensitivity compared to that of a NAAT alone (83% versus 89%). The authors postulate that this difference is probably due to the high prevalence of CDI in their patient population (mainly hematology/oncology patients), and they recommend the use of NAAT alone in high-risk patients (42). In another study comparing the illumigene and Quidel AmpliVue (Quidel Molecular, San Diego, CA) assays in a large number of pediatric patients, both tests performed similarly, suggesting that these tests can be used with this patient population. This study included a subanalysis of those <2 years of age. The detection rate of C. difficile in this group was 23%. The authors noted that these tests may be accurate in those <2 years of age but may not represent CDI, and clinical judgment is needed in interpreting the results (43).

There are currently two FDA-cleared multiplex platforms that detect an array of gastrointestinal pathogens, including C. difficile as one of the reported targets: the xTAG gastrointestinal pathogen panel (xTAG GPP; Luminex Molecular Diagnostics Inc., Toronto, ON, Canada) and the BioFire FilmArray gastrointestinal panel (FilmArray GI; BioFire Diagnostics, Salt Lake City, UT).

The xTAG GPP tests for the presence of 15 different pathogens. For the clinical trial of the xTAG GPP for FDA clearance, samples from pediatric patients were included, but the results were not analyzed per age group. According to the package insert, C. difficile was detected in 3/192 samples from healthy volunteers (ages not provided) and may have represented asymptomatic carriage (44). Data from the clinical trial (xTAG GPP package insert) showed that C. difficile was present in 52.7% of all coinfections, with the most common being norovirus codetected with C. difficile (44). In a study by Beckmann et al. using xTAG GPP testing on samples from pediatric patients presenting with symptoms of acute gastrointestinal disease, 7% were positive for C. difficile, 39% of which were from patients <1 year of age. The authors did not provide any further data regarding whether the detection of C. difficile correlated with CDI in these infants, but they did question whether the finding of C. difficile in this setting warrants treatment (45). In another study where xTAG GPP was used as the first line of detection of gastrointestinal pathogens in 211 prospectively collected specimens from a pediatric hospital, C. difficile was the number one pathogen detected (10.9%). There were 5 cases of coinfection, and C. difficile was present in 2 of these (C. difficile with norovirus and C. difficile with Salmonella spp.). The authors did not provide a breakdown of results positive for C. difficile by patient age, making it difficult to assess the C. difficile positivity rate in this cohort (46). Interestingly, in a recent study using samples from pediatric patients from a developing country where xTAG GPP was the testing platform, a high number of gastrointestinal parasites and viruses were detected. Of the samples tested, 64.7% were positive for a pathogen, but C. difficile was not detected in any of the specimens. However, the sample size was small, with only 35 samples included in the study, and the patients were older, 5 to 15 years (47).

The FilmArray GI tests for the presence of 22 pathogens, with C. difficile being one of them. According to the package insert, C. difficile was detected in 5/100 asymptomatic volunteers, 4 of which were <12 years of age (48). From the clinical trial data, the C. difficile prevalence was stratified per age group and in patients <1 year old, the organism was present in 40.5% of the samples tested, while in patients between 1 and 5 years old, it was present in 15.8% (49). C. difficile was detected in 41.6% of samples with mixed infections, making it the second most detected target in these samples (49). In a recent study by Stockmann et al., the diagnostic yield of physician ordering patterns in pediatric patients with suspected infectious gastroenteritis was examined using traditional testing methods versus the use of the FilmArray GI. The results were categorized based on whether a physician ordered testing for C. difficile alone, for other pathogens alone, or for both C. difficile and additional pathogens. Of the samples tested, 68% were from outpatients, of which 25% were positive for C. difficile, a number similar to the positivity rate of 24% seen in inpatients. For all cases where C. difficile alone was tested, 28% were positive for another pathogen that was not known to the physician based on the testing ordered. Additionally, in those children 1 to 4 years of age where a test for C. difficile alone was ordered, another pathogen was detected in 23% of the cases (50). This study had limited clinical data for review in order to identify true infections with C. difficile or the impact of C. difficile in coinfections, thus making it hard to assess whether clinical intervention would have been required. Additionally, the high prevalence of C. difficile in the outpatient population is evident, but correlation with disease was not established and additional studies are needed to interpret a positive result for C. difficile in these patients.

Similar to the study above, we analyzed data from our own hospital, comparing the ordering patterns of C. difficile testing as a standalone test versus the results of the FilmArray GI panel run in parallel as part of a clinical research study. The FilmArray GI results were not known to the physicians (Fig. 1) (unpublished data). Following the AAP recommendations to avoid testing in the 1- to 3-year-old age group, the pattern shows that C. difficile testing is not widely ordered in younger patients in our institution. However, it is considered a causative agent of diarrhea for older pediatric patients, and the number of orders increased for those >8 years of age in this cohort. Also of note, there were a considerable number of patients <3 years of age with a positive result for C. difficile by the FilmArray GI (n = 63, 20%) in which standalone testing for C. difficile had not been ordered by the physician. If a multiplex panel that includes C. difficile is used for testing on all samples, results are available even when the physician is not considering CDI in the differential diagnosis. Positive C. difficile results need to be interpreted with caution and in correlation with the clinical picture, given the high rate of asymptomatic carriage in young children and other populations such as hospitalized children as described previously. Careful consideration of reporting is needed, including instructive comments or possibly withholding the results of C. difficile testing in certain patient groups in order to mitigate the risk of inappropriate treatment.

FIG 1.

Results of C. difficile testing using a multiplex molecular gastrointestinal panel versus physicians' ordering patterns for C. difficile singleplex testing. A total of 782 stool samples were enrolled from June to September 2013 at Nationwide Children's Hospital as part of a clinical trial for FDA clearance of a multiplex gastrointestinal panel (BioFire FilmArray gastrointestinal panel). Samples were enrolled based on adequate volume of a stool sample submitted for physician-ordered routine stool culture in Cary-Blair transport media. The number of tests for C. difficile with the multiplex panel was compared to the number of physician orders for C. difficile PCR on the same sample. The FilmArray GI results were not known to the physicians. Of note, testing for C. difficile for patient care was not limited by age. CDT, Clostridium difficile testing; MP, multiplex gastrointestinal panel; pos, positive; neg, negative.

Further studies to differentiate the patients who are colonized versus those with true disease are warranted. These include more studies utilizing data from gastrointestinal multiplex panels where the exclusion of other agents of gastroenteritis and diarrhea can be explored and correlated with the patients' clinical picture. Additionally, novel algorithms investigating the role of intestinal inflammatory biomarkers to aid in the diagnosis of CDI are needed. El Feghaly et al. explored the utility of interleukin-8, lactoferrin, chemokine ligand-5 RNA, and phosphorylated p38 in children with and without diarrhea and symptomatic versus asymptomatic carriage of C. difficile. They concluded that fecal inflammatory cytokines may be useful for risk stratification in children (51).

In conclusion, the diagnosis of CDI should be made in the context of the clinical presentation of the patient, in correlation with the results of laboratory testing for detection of C. difficile. This is of particular importance in pediatric practice due to the known high levels of colonization in young children. Testing is not routinely recommended in children <1 year of age. Caution needs to be used to interpret the results in patients between 1 and 3 years of age and additional agents of infection need to be investigated (35). Nonetheless, in patients with risk factors and in special pediatric populations such as those with IBD and immunosuppressed patients (17, 18), prompt and sensitive identification of C. difficile is necessary to allow appropriate treatment, prevent transmission, and improve outcomes.

The sensitivity and specificity of testing methodologies for the detection of C. difficile have improved with a plethora of assays available. With the advent of various NAATs, including multiplex panels, there exists the possibility of increased detection of C. difficile. This higher sensitivity can be of benefit, but there is also a risk that testing may lack positive predictive value in those who are not at risk of CDI. While the NAAT is the most sensitive analytically, questions remain about its use as a first-line, standalone test. The best test or testing algorithms for the diagnosis of CDI and the ability to reliably differentiate disease from colonization are still in debate. For both pediatric and adult patients, further studies are needed to shed light on the impact of testing on clinical outcomes.

Biographies

Stella Antonara completed her Bachelor in Biology at the University of Patra, Greece, and her Ph.D. in Molecular Microbiology at Tufts University School of Medicine. She completed her CPEP Medical and Public Health Microbiology Fellowship at the National Institutes of Health in Bethesda, MD, in 2013. Upon completion of her fellowship, she became the Assistant director of Microbiology and Immunoserology at Nationwide Children's Hospital in Columbus, OH, and is currently Assistant Clinical Professor of Pathology and Laboratory Medicine at The Ohio State University College of Medicine. Her research interests include the epidemiology of Clostridium difficile in pediatrics and development of diagnostic methods in the pediatric population.

Amy L. Leber received her B.S in Microbiology and Ph.D. in Medical Microbiology and Immunology at The Ohio State University. She completed the CPEP Medical and Public Health Microbiology Fellowship at UCLA Medical Center in Los Angeles, CA. She joined Nationwide Children's Hospital in 2007 first as Associate Director then in 2013 as Director of Clinical Microbiology and Immunoserology. Dr. Leber is currently Associate Clinical Professor of Pediatric and Pathology and Laboratory Medicine at The Ohio State University College of Medicine. She is the current Editor-in-Chief for the Clinical Microbiology Handbook (ASM Press). Her research interests include development of improved diagnostics and the understanding of the pathogenesis of gastrointestinal and respiratory infections in pediatrics.