Markers of intestinal inflammation, not bacterial burden, correlate with clinical outcomes in Clostridium difficile infection

Abstract

Background: Clostridium difficile is a leading hospital-acquired infection. Many patients remain symptomatic for several days on appropriate antibiotic therapy. To assess the contribution of ongoing infection vs persistent inflammation, we examined the correlation between fecal cytokine levels, fecal C. difficile burden, and disease outcomes in C. difficile infection (CDI).

Methods: We conducted a prospective cohort study in Barnes Jewish Hospital between June 2011 and May 2012 of hospitalized adults with CDI. We determined fecal interleukin 8 (IL-8) and lactoferrin protein concentrations by enzyme immunoassay. We used real-time polymerase chain reaction (PCR) to measure relative fecal IL-8 and CXCL-5 RNA transcript abundances, and quantitative PCR to enumerate C. difficile burden.

Results: Of 120 study subjects, 101 (84%) were started on metronidazole, and 33 of those (33%) were subsequently given vancomycin. Sixty-two (52%) patients had diarrhea persistent for 5 or more days after starting CDI therapy. Initial fecal CXCL-5 messenger RNA (mRNA), IL-8 mRNA, and IL-8 protein correlated with persistent diarrhea and use of vancomycin. Time to diarrhea resolution was longer in patients with elevated fecal cytokines at diagnosis. Fecal cytokines were more sensitive than clinical severity scores in identifying patients at risk of treatment failure. Clostridium difficile burden did not correlate with any measure of illness or outcome at any point, and decreased equally with metronidazole and vancomycin.

Conclusions: Persistent diarrhea in CDI correlates with intestinal inflammation and not fecal pathogen burden. These findings suggest that modulation of host response, rather than adjustments to antimicrobial regimens, might be a more effective approach to patients with unremitting disease.

Keywords: CXCL-5; Clostridium difficile; IL-8; cytokines; inflammation.

Figure 1.

Initial cytokines and tcdB cycle threshold (CT) in patients with Clostridium difficile infection (CDI). A, Initial interleukin 8 ΔCT levels in patients with diarrhea persistent 5 days after starting CDI therapy (dashed bar) are higher than those whose diarrhea resolved (clear bar). B, Initial tcdB CT values in patients with persistent diarrhea 5 days after starting CDI therapy (dashed bar) and patients with resolved diarrhea (clear bar) do not differ. C, Initial CXCL-5 ΔCT levels in patients treated with metronidazole (clear bar) are lower than those started on vancomycin or had vancomycin added to their therapy during their hospitalization (dashed bar). D, Initial tcdB CT values in patients treated with metronidazole (clear bar) and patients started on vancomycin or had vancomycin added to their therapy during their hospitalization (dashed bar) do not differ. Boxes represent median ± 25th–75th percentiles; whiskers represent 5th–95th percentiles; dots represent outliers. P, P values of the 2-tailed Wilcoxon rank-sum test; *P < .05. Abbreviations: ΔCT, difference of cycle thresholds between internal control and cytokine tested; CT, cycle threshold; IL-8, interleukin 8.

Figure 2.

Time to diarrhea resolution in patients with elevated markers on admission compared to those with low markers. A, Time to diarrhea resolution is more prolonged in patients with high CXCL-5 (ΔCT CXCL-5 messenger RNA >0) at diagnosis (CXCL-5⁺, green line; median 12 days) compared to those with low CXCL-5 mRNA levels (CXCL-5⁻, blue line; median 4 days). B, Time to diarrhea resolution is more prolonged in patients with high interleukin 8 (IL-8) mRNA (ΔCT IL-8 mRNA >0) at diagnosis (IL-8⁺, green line; median 6 days) compared to those with negative IL-8 mRNA levels (IL-8⁻, blue line; median 4 days). C, Kaplan-Meier survival curve comparing patients with a tcdB CT in the lower quartile (CT<20.7, red line) to those with tcdB CT in the second quartile (20.71< CT <23.6, blue line), the third quartile (23.61< CT < 27.1, orange line), and the higher quartile (CT > 27.11, green line). There was no difference in the time to diarrhea resolution among the 4 groups. “+” represents censored values, and P is the P value of the log-rank (Mantel-Cox) test. *P < .05. Abbreviations: ΔCT, difference of cycle thresholds between internal control and cytokine tested; CDI, Clostridium difficile infection; CT, cycle threshold; IL-8, interleukin 8; mRNA, messenger RNA.

Figure 3.

Daily CXCL-5 and tcdB cycle threshold (CT) levels in patients treated with vancomycin compared to those treated with metronidazole. A, CXCL-5 ΔCT levels in patients treated with vancomycin (dashed bars) are persistently higher than levels of those treated with metronidazole (clear bars). Levels do not decrease significantly with time. B, tcdB CT values in patients treated with vancomycin (dashed bars) are similar to those of patients treated with metronidazole (clear bars). Bacterial burden drops at the same rate with both therapies. Boxes represent median ± 25th–75th percentiles; whiskers represent 5th–95th percentiles; dots represent outliers. Abbreviations: ΔCT, difference of cycle thresholds between internal control and cytokine tested; CT, cycle threshold.

Figure 4.

Daily CXCL-5 and tcdB cycle threshold (CT) levels in patients with persistent diarrhea compared to those whose diarrhea resolved with 5 days of therapy. A, CXCL-5 ΔCT levels in patients with persistent diarrhea (dashed bars) are persistently higher than levels of those with resolved diarrhea (dotted bars). Levels do not decrease significantly with time. B, tcdB CT values in patients with persistent diarrhea (dashed bars) are similar to those of patients with resolved diarrhea (dotted bars). Bacterial burden drops at the same rate in the 2 groups. Boxes represent median ± 25th–75th percentiles; whiskers represent 5th–95th percentiles; dots represent outliers. Abbreviations: ΔCT, difference of cycle thresholds between internal control and cytokine tested; CT, cycle threshold.