Fecal Calprotectin Concentrations in Cancer Patients with Clostridium difficile Infection

Abstract

Fecal calprotectin (fCPT) has been used as a surrogate marker for assessment of intestinal inflammation. We explore the utility of fCPT values as a diagnostic aid in cancer patients with suspected Clostridium difficile infection (CDI). A total of 232 stool specimens submitted for GeneXpert C. difficile PCR testing were included in the study. All specimens were tested for fCPT and toxin/GDH antigens. Clinical severity of CDI cases was determined by the IDSA/SHEA criteria. Significant differences of median fCPT values (μg/g) between CDI (n=117, Median 183.6 μg/g) and non-CDI (n=115, 145.6 μg/g, p=0.006) patients were seen. In CDI patents, significantly lower f CPT values were found in patients with mild to moderate (n=95, 182.1 μg/g) than those with severe & severe to complicated (n=22, 218.5 μg/g, p=0.014) scores, and among those that were toxin positive (n=24, 200.2 μg/g) versus toxin negative (n=86, 182.8 μg/g, p=0.044). Despite this overall trend, wide variations in fCPT values were found in all categories examined. A logistic regression analysis revealed that the fCPT values correlated independently with the severity of clinical manifestations (OR=2.021, 95%CI=1.132–3.608); however, it did not correlate with other clinical outcomes. Our study findings show that high fecal calprotectin levels correlate with toxin positive and clinically severe CDI; however wide variations in individual measurements preclude establishment of reliable cut-offs for routine diagnostic use in cancer patients.

Introduction

Clostridium difficile is an anaerobic, spore-forming, Gram-positive bacillus, which is a leading cause of infection-associated diarrhea in hospitalized patients receiving antimicrobial therapy [1]. In the human gastrointestinal tract it may exist as a silent colonizer or under certain conditions, cause active infection leading to diarrhea and related complications [2–4]. C. difficile infection (CDI) is the leading cause of hospital-acquired gastrointestinal illness and places an enormous burden on the healthcare system, with an estimated cost of 3.2 billion U.S. dollars annually [5]. The reported prevalence of CDI has increased in the past decade across the healthcare continuum [1].

Diagnosis of CDI is a major conundrum at the current time. While several different technologies exist, each has its unique set of limitations [6, 7]. It has been shown that part of the reported increase in CDI incidence is due to the wider implementation of high sensitivity molecular assays over toxin-EIA or culturing methods which are considered reference methods [8, 9]. Whereas toxin-based assays directly assess disease-mediating toxin production in vivo, PCR-based assays detect only the presence of the toxin gene. Planche et al. reviewed PCR positive stool test results from 6,522 inpatient episodes and found that patients positive for toxin were more likely to die compared with those without detectable toxin in stool [8]. Polage et al. reported that patients with positive C. difficile PCR (CDPCR) tests but negative CD toxin EIA had similar outcomes when compared to patients without detectable C. difficile, and the CD related complications occurred only among patients that were concurrently toxin positive [9].

These data suggest that molecular assays may be overly sensitive, and that a positive C. difficile PCR result in stool cannot differentiate between C. difficile colonization and infection [10]. This conclusion is especially relevant in a cancer hospital, where the likelihood of encountering toxigenic C. difficile and non-infectious diarrhea is higher [2, 11, 12].

A wide variety of biomarkers including fecal lactoferrin and fecal calprotectin (fCPT) have been used for screening for inflammatory bowel disease as well as for monitoring disease progression [13–15]. The recent guidance by UK National Institute for Health and Care Excellent (NICE) recommends using fecal biomarkers to aid in the future diagnosis of CDI [16]. Studies have reported the use of fCPT values as new diagnostic biomarkers in CDI [16–22]. In the current study, we explored the utility of fCPT values to aid in the diagnosis and monitoring of Clostridium difficile infection (CDI) in patients with underlying cancer.

Materials and Methods

Clinical specimens

A total of 232 specimens from 156 patients submitted for Cepheid Xpert C. difficile PCR assay testing from January 2012 to December 2013 were selected for the study including (i) 80 patients with single specimens (PCR-negative, 39; PCR-positive, 41), (ii) 38 patients with paired specimens from an initial PCR negative and subsequent positive sample, and (iii) 38 patients with paired specimens from an initial PCR positive to a negative second sample. Leftover stool specimens after molecular testing were aliquotted and stored at −80°C until subsequent antigen and calprotectin (fCPT) testing. Solid or formed stool specimens were rejected and the ordering clinician was notified. This study was approved by the Memorial Sloan Kettering Cancer Center (MSKCC) Institutional Review Board (WA0477–15).

C. difficile PCR assay

The GeneXpert C. difficile PCR assay (Cepheid, Sunnyvale, CA) was performed on the GeneXpert system according to the manufacturer’s instructions [23].

C. difficile antigen assay

The C. Diff Quik Chek Complete (TechLab, Blacksburg, VA) dual-antigen EIA was performed according to the manufacturer’s instructions. In brief, approximately 25μl of stool was added to a tube containing the diluents and conjugate, and the mixture was transferred to the device sample well. After incubation for 15 min at room temperature, the wash buffer and then the substrate were added to the reaction window. Results were read after 10 min. GDH antigen and/or toxins were reported positive if a visible band was seen on the antigen and/or toxin side of the device display window, respectively [24].

Fecal calprotectin assay

Fecal calprotectin (fCPT) was quantitatively detected by the Phical Test (Calpro AS, Oslo, Norway) according to the manufacturer’s instructions [25]. In brief, after three freeze-thaw cycles, stool specimens were vortexed thoroughly. The corresponding volume of Diluted Extraction Solution was combined with 80–120 mg of stool to produce a 1:10 dilution. After homogenization in an Eppendorf Thermomixer R (Fisher Scientific) at 1,000 rpm for 30 min and centrifugation at 12,200 rpm for 20 min, 20μl supernatant was transferred to a sterile Eppendorf tube containing 980μl Diluted Extraction Solution. Calprotectin levels from 100 μl of mixture were measured using an enzyme immunoassay on a 96-well microtiter plate in duplicate as previously described [25]. The fCPT value from the duplicate wells were averaged ranging between 11.9 and 278.7 μg/g based on a 6-point standard curve provided with the kit.

CDI severity scoring

A retrospective chart review was done to determine the severity of CDI. A diagnosis of CDI was established when the patient fulfilled the following criteria: (i) the presence of diarrhea, defined as passage of ≥3 unformed stools in ≤24 consecutive hours and (ii) a stool test result positive for GeneXpert C. difficile PCR [26]. CDI severity was divided into the following six categories: (i) no clinical CDI, (ii) mild CDI, (iii) mild-to-moderate CDI, (iv) moderate CDI, (v) moderate-to-severe CDI, and (vi) severe CDI based on the IDSA/SHEA guideline [26] as well as the grouping criteria described previously [27]. CDI severity was correlated with fCPT concentration.

Data statistical analysis

All samples were categorized into five groups (Table 2) based on their PCR, toxin and GDH antigen results and fCPT median, mean, and standard deviation (SD) were calculated for each group. Chi-square and Fisher exact tests were used to compare categorical variables and the Student’s t-test was used to compare continuous variables. Quantitative variables were compared by the non-parametric Mann-Whitney U-test (for 2 groups) and Kruskal-Wallis test (for >3 groups), respectively, in uni-variable analysis. Multi-variate logistic regression was performed to identify risk factors for death, as well as for compositive severe outcomes that were associated with fCPT levels. Odds ratios (OR), 95% confidence intervals (CI), and p values were calculated, and all p value less than 0.05 were considered statistically significant. On univariate analysis, variables with p<0.3 were included in a logistic regression model to further identify independent risk factors associated with fCPT levels. All data analysis was carried out in SPSS24 and p value equal or less than 0.05 was considered to be of statistical significance.

Table 2.

Median, Mean & Standard deviation of fCPT values in cancer patients with diarrhea

Group	Number	Median	p-value	Mean	SD	p-value
GDH+/TOX+/PCR+	24	200.2	0.024	390.7	398.6	0.210
GDH+/TOX−/PCR+	86	182.8		344.8	381.6
GDH−/TOX−/PCR+	7a	162.0		819.4	1664.2
GDH+/TOX−/PCR−	9	178.4		198.3	78.0
GDH−/TOX−/PCR−	106	139.8		332.7	595.7

−, negative; +, positive.

^aIncluded two specimens with extremely high fCPT values of 4585.3 and 4490.3 μg/g from one patient with gastrointestinal cancer.

Results

We first determined reproducibility of the fCPT levels of the Phical Test. Two specimens with adequate volumes were selected and tested in triplicate on three different calendar days. The average fCPT values of the two specimens were 88.7 and 248.0 μg/g. Intra- and the inter-assay coefficients of variation (CV) were 1.32–4.31% and 9.63–14.48%, respectively (Table 1).

Table 1.

Reproducibility of fCPT concentrations (μg/g) in cancer patients with diarrhea.

Sample	Day	Day 1	Intra-CV (%)	Day 2	Intra-CV (%)	Day 3	Intra-CV (%)	Inter-CV (%)
1	1	94.9	3.45	108.7	3.09	81.8	1.32	14.48
	2	89.3		103.8		80.8
	3	89.7		110.2		79.7
2	1	238.9	1.49	266.9	3.67	224.7	4.31	9.63
	2	232.2		273.0		244.4
	3	234.1		286.7		231.0

CV, coefficients of variation.

A total of 232 specimens from 156 patients were included and tested for fCPT levels by the Phical Test. These specimens were divided into five groups based on their C. difficile GDH, toxin antigen and PCR results (Table 2). The fCPT levels were the highest in the GDH+/TOX+/PCR+ group and the lowest in the GDH-/TOX-/PCR- group (200.2 ± 398.6 μg/g vs. 139.8 ± 595.7 μg/g; p=0.019). Large variations in fCPT values as indicated by high standard deviations were noticed. Significant differences in fCPT medians between the five groups were also detected (p=0.024). Notably, differences in fCPT medians between GDH+/TOX+/PCR+ and GDH+/TOX-/PCR+ groups values (200.2 μg/g vs. 182.8 μg/g) were statistically significant (p=0.044). Median values were used for subsequent further analysis.

Among the 232 specimens, 117 (50.4%) were positive by C. difficile PCR: the median fCPT values for these specimens (183.6 μg/g) was significantly higher than those that were PCR-negative (145.6 μg/g, p=0.006) (Table 3). Among the 117 PCR-positive specimens, fCPT median values were significantly higher in patients with severe and severe to complicated CDI (218.5 μg/g) compared to those with mild to moderate disease (182.1 μg/g, p=0.014).

Table 3.

Fecal calprotectin level and clinical severity in cancer patients with diarrhea

	Single group			N to P group			P to N group			Total
Clinical severity	No.	Median	p-value	No.	Median	p-value	No.	Median	p-value	No.	Median	p-value
C. difficile PCR−	39	138.7	0.014a	38	138.0	0.194a	38	164.3	0.155a	115	145.6	0.006a
Mild to moderate	39	138.7	N.D.	38	138.0	N.D.	38	164.3	N.D.	115	145.6	N.D.
Severe & severe to complicated	0			0			0			0
C. difficile PCR+	41	207.6		38	177.3		38	186.0		117	183.6
Mild to moderate	31	202.5	0.023b	30	175.5	0.590b	34	186.0	0.640b	95	182.1	0.014b
Severe & severe to complicated	10	272.5		8	181.5		4	205.4		22	218.5

N.D., not done.

^aBetween C. difficile PCR- and + groups;

^bBetween Mild to moderate and Severe & severe to complicated groups.

We further assessed characteristics, risk factors, and outcomes in relation to fCPT levels for 156 patients with the first stool samples only. On univariate analysis, statistically significant factors associated with higher fCPT levels were found for those without history of chemotherapy within 30 days (p=0.025), fever ≥38°C (p=0.048) at the time of diarrhea, lower blood white blood cell (WBC <1×10⁹/L) count (p=0.039), and those with severe-complicated CDI (p=0.036) (Table 4). A logistic regression analysis of all the characteristics with p<0.3 on univariate analysis was performed, and the only independent risk factor associated with fCPT values was severe, complicated manifestations of CDI (OR=2.021, 95% CI=1.132–3.608, p=0.017). Recurrent CDI and death within 30 days were not associated with fCPT values at the time of clinical evaluation.

Table 4.

Risk factors and outcomes associated with fecal calprotectin levels^a

	Uni-variant analysis				Multi-variant Logistic
Variable	Stratum	No.	Median	p-value	OR	95% CI	p-value
Age	≤60y	82	165.7	0.618
	>60y	74	154.7
Gender	female	69	149.7	0.259	2.009	0.905–4.461	0.087
	male	87	167.6
Cancer type	Liquid	82	156.1	0.521
	Solid, non GI	32	166.7
	Solid, GI	42	182.3
Surgery	No	87	163.2	0.629
	Yes	69	149.7
Chemotherapy	No	76	192.0	0.025	0.750	0.353–1.595	0.455
	Yes	80	139.8
Antibody therapy	No	40	150.3	0.463
	Yes	116	164.3
Laxative-use	No	112	154.7	0.286	0.700	0.290–1.687	0.426
	Yes	44	170.5
Abdominal pain	No	124	151.6	0.113	0.756	0.284–2.015	0.576
	Yes	32	169.1
Fever	<38°C	151	157.0	0.048	0.860	0.093–7.955	0.895
	≥38°C	5	209.5
Albumin	≥30mg/dl	128	156.1	0.167	1.858	0.754–4.582	0.178
	<30mg/dl	28	181.9
WBC count	<1×109/L	18	218.1	0.039	2.240	0.486–10.330	0.301
	(1–20)×109/L	128	166.7
	>20×109/L	10	128.2
Clinical recurrence	No	153	156.5	0.560
	Yes	3	178.5
Manifestation	Mild to moderate	115	145.6	0.036	2.021	1.132–3.608	0.017
	Severe	31	202.5
	Severe to compl.	10	272.5
Death in 30 days	No	138	164.3	0.316	0.986	0.303–3.212	0.982
	Yes	18	148.6

^aBased on single first specmen from each patient.

OR, odds ratio; CI, confidence level.

Discussion

Our study demonstrates that in patients with underlying cancer who experience diarrhea, fCPT values are higher in those with CDI and significantly higher among toxin positive and those with higher level of disease severity. The fCPT levels were independently associated with CDI severity in multivariate analysis. Despite these findings, due to the wide variation in individual measurements, we were unable to establish a clear cut off that would reliably distinguish between infectious and carrier states, an important clinical and epidemiological distinction [3, 12], or predict disease outcomes.

The role of fCPT is not established in the diagnosis of CDI mainly due to observed intra-individual variability in previous studies, which has mostly been attributed to sample handling and processing [28]. It has previously been reported that fCPT analysis on the first bowel movement is most appropriate [14, 15]. In our study, we processed fecal specimens that had undergone three freeze-thaw cycles followed by thorough vortexing. This allowed us to reach intra- and inter-assay CV values less than 5% and 15%, respectively. No significant fCPT value differences (4,585.3 and 4,490.3 μg/g) were revealed between the two specimens collected from the same patient with PCR result conversions, suggesting that fCPT concentration in stool was relatively stable independently of C. difficile PCR results.

Current standard-of-care molecular assays targeting C. difficile-specific toxin A or B genes do not distinguish between colonization and infection [29, 30], and thereby may result in over-diagnosis, over-treatment, and increased health care costs [8, 9]. Furthermore, only 15 to 25% of nosocomial antibiotic-associated diarrhea cases are attributed to CDI [3, 12, 31, 32]. Previous publications have demonstrated that fCPT is a useful marker in differentiating inflammatory bowel disease from functional gastrointestinal disorders [33–36]. A series of recent studies have described that the median values of fCPT is highly sensitive in differentiating CDI from non-CDI patients with hospital-acquired diarrhea [16, 22, 37]. Fecal calprotectin levels may be used as an adjunctive measure to PCR to support the diagnosis of CDI, especially in older adults [20, 21].

Our study has several limitations. We noted a great deal of variability in individual fCPT values within the same diagnostic and disease severity tier. In addition, two extremely high fCPT levels (4585.3 and 4490.3 μg/g) were detected from one patient with gastrointestinal cancer. While our study demonstrated that fCPT values independently correlated with severity at the time of evaluation, it is emphasized that fCPT levels did not correlate with other clinical outcomes including 30-day mortality and 90-day recurrence [16]. Finally, total sample numbers recruited in the study were relatively small, causing difficulty in achieving reliable conclusions. When fCPT is used to manage patients, it should be interpreted under the specific clinical context.

In summary, while the fCPT values correlate with clinically significant CDI in persons with cancer, the large variability in values precludes widespread application as a diagnostic aid. Future large studies in this population are needed to generate potential cut-off values that may be used to distinguish between carrier and disease states.