Acute infectious diarrhea is one of the commonest diseases in the world [1]. Most cases are self-limited and caused by viral pathogens; therefore, routine antibiotics are not recommended [1]. Nevertheless, empirical antibiotic therapy for acute infectious diarrhea is still widely used in clinical practice [2]. Indiscriminate antibiotic use can lead to antibiotic resistance, allergic reactions, drug-related toxicities, Clostridium difficile infection, and increased medical costs [3,4]. Therefore, it is necessary to quickly identify whether a bacterial infection is present in the early stages of acute bacterial colitis. To date, treatment strategies for patients with suspected infectious diarrhea are determined mainly by correctly classifying the severity of the clinical features of the patients; however, there is no objective biochemical indicator for distinguishing bacterial colitis from nonbacterial colitis [1].

Stool-based tests such as bacterial culture, microscopy, and stool antigen tests are conventional diagnostic approaches for identifying enteric pathogens in patients with bacterial colitis [1]. However, these tests are time-consuming, have inadequate sensitivity, and require specialized equipment for analysis [1,5]. C-reactive protein (CRP) and procalcitonin (PCT) are representative serological markers for various inflammatory conditions and sepsis [6]. CRP is an acute-phase reactant used to diagnose and follow up on diverse bacterial infections [6]. PCT is a calcitonin precursor made of 116 amino acids [7]. PCT is rarely expressed under normal conditions, but it is activated in response to bacterial infection and mediated by endotoxins or interleukins (ILs) and tumor necrosis factor-α [7]. Regarding speed, simplicity, and expense, if bacterial colitis can be differentiated by serologic biomarkers such as PCT and CRP, it will be beneficial in making clinical treatment decisions.

To date, there is insufficient data on the clinical feasibility of PCT and CRP in diagnosing acute bacterial colitis. Herein, we evaluated the value of serum PCT and CRP in differentiating between bacterial and nonbacterial colitis.

In total, 638 patients were included in this study: 186 in the bacterial colitis group and 452 in the nonbacterial colitis group. Table 1 presents the patients’ baseline characteristics. The proportion of patients aged ≥65 years was significantly higher in the bacterial colitis group than in the nonbacterial colitis group (60.8% vs. 50.2%, p=0.015). Patients with chemotherapy were significantly higher in the bacterial colitis group than in the nonbacterial colitis group (16.1% vs. 8.4%, p=0.007). Serum CRP level was significantly higher in the bacterial colitis group than in the nonbacterial colitis group (mean±standard deviation, 11.98±9.75 vs. 10.11±9.34 mg/dL, p=0.023). The mean serum PCT level was 8.46 ng/mL in the bacterial colitis group and 7.98 ng/mL in the nonbacterial colitis group; there was no significant difference between the groups. Other laboratory values and clinical parameters, including WBC count, platelet count, creatinine level, fever, ICU admission, use of immunosuppressant, and presence of bacteremia, were not significantly different between the two groups. The causative pathogens detected by PCR testing are presented in Table 2. In the bacterial colitis group, C. perfringens was the commonest pathogen (n=70, 37.6%), followed by C. difficile toxin B (n=65, 34.9%), Campylobacter spp. (n=42, 22.6%), and Salmonella spp. (n=11, 5.9%).

Findings of the ROC curve analysis of PCT and CRP in differentiating between bacterial and nonbacterial colitis are shown in Table 3 and Fig. 1. The area under the curve (AUC) of PCT for diagnosing bacterial colitis was 0.557 (95% confidence interval [CI], 0.509–0.605). At a cutoff of 0.52 ng/mL, the sensitivity and specificity of PCT were 54.8% and 54.6%, respectively. The AUC of CRP was 0.561 (95% CI, 0.512–0.610). At a cutoff level of 8.80 ng/mL, the sensitivity and specificity of CRP were 52.2% and 54.2%, respectively. Combining PCT and CRP did not improve the diagnostic performance beyond that of PCT or CRP alone (AUC, 0.522; 95% CI, 0.474–0.571).

Our study demonstrated that serum PCT and CRP levels were not potential markers for the early distinction between bacterial and nonbacterial colitis. The combination of PCT and CRP did not show a better diagnostic performance in discriminating bacterial colitis from nonbacterial colitis than PCR or CRP alone.

In Korea, among nosocomial infections, the rate of methicillin-resistant Staphylococcus aureus reached 80%, and the vancomycin-resistant enterococci rate of Enterococcus faecium reached 24% [9]. If the bacterial infection can be predicted early, patient treatment can be determined in advance to prevent unnecessary antibiotics. This will ultimately help address issues such as the overuse of antibiotics, rising healthcare costs, and antibiotic resistance. For this reason, several biomarkers are being actively studied as tools to distinguish bacterial infections (e.g., leucocyte count, erythrocyte sedimentation rate, CRP, soluble triggering receptor expressed on myeloid cells 1, serum PCT, IL-6, IL-8, IL-27, etc.) [10].

An ideal biomarker should not be only highly sensitive and specific but also easy to use, fast, and inexpensive [11,12]. CRP, the most commonly used serologic biomarker for infection, rises 12 to 24 hours after infection and reaches its maximum level after 24 hours. On the other hand, PCT is immediately detectable within 3 to 4 hours after the infection and reaches its maximum level 6 to 12 hours later [13]. When infection is controlled, PCT levels decrease rapidly. In addition, PCT seems to be useful for diagnosing bacterial infection because its level is reduced by IL-γ, a viral infection mediator [14]. Due to the immediate response of PCT to bacterial infection, PCT is thought to be a promising biomarker for early diagnosis of bacterial infection, monitoring of response to antibiotics, or determination of the need to change antibiotics [13]. CRP can also increase during exacerbation of viral infection or autoimmune disease [15], but PCT is less affected by conditions, such as neutropenia and reduced immunity, than CRP [14].

The diagnostic value of PCT is a focus of research as PCT is reportedly useful for early diagnosis and monitoring of various diseases, especially bacterial infections. In a systematic review and meta-analysis, PCT showed better diagnostic accuracy than CRP in hospitalized patients with suspected bacterial infections [6]. PCT is reportedly a potential marker for septic shock in acute cholangitis [16]. Additionally, PCT plays an important role in differentiating between edematous pancreatitis and necrotizing pancreatitis [17], diagnosing respiratory distress syndrome [18], and monitoring infections in transplant patients [19].

Although the feasibility of PCT has been studied in various diseases, only a few studies have addressed its role as a diagnostic biomarker for bacterial colitis. One previous study analyzed the discriminative value of PCT in differentiating between inflammatory and noninflammatory diarrhea [20]. In that study, serum PCT had a significant predictive value (odds ratio [OR], 1.321; AUC, 0.797) and a better predictive value than CRP (OR, 1.145; AUC, 0.697) [20]. Contrarily in our study, neither PCT nor CRP helped discriminate between bacterial and nonbacterial colitis. Unlike in our study, in a previous study, inflammatory colitis was diagnosed using colonoscopy and imaging, and no tests for microbial pathogens were used. The results may differ because the studies used different definitions of bacterial colitis; however, our study had a larger sample size and used multiple PCR testing for diagnosis, which is more sensitive than culture [1,21]. In a study investigating whether PCT could differentiate between infectious gastroenteritis and inflammatory bowel disease (IBD), PCT and CRP appeared to be good diagnostic markers for gastroenteritis. Still, there was no significant difference in IBD monitoring using PCT. However, similar to our study, studies have shown that PCT is not an appropriate diagnostic tool for bacterial infection. In another study, the feasibility of PCT levels in discriminating Salmonella infections was assessed: PCT had a low diagnostic value [22].

In this study, C. perfringens and C. difficile accounted for most colitis-associated bacterial infections. Contrarily, E. coli is known to be the commonest cause of infectious diarrhea in Korea, followed by S. aureus, Salmonella, Vibrio parahaemolyticus, C. perfringens, Bacillus cereus, Campylobacter jejuni, and Shigella [8]. Since this study targeted patients who visited a general tertiary hospital, the distribution of the causative bacteria may differ slightly from that of the general population.

This study has some limitations. First, there may have been a selection bias, given the study’s retrospective design. Second, PCT reaches peak levels 6 to 24 hours after infection; however, in this retrospective study, blood sampling may have been performed when PCT levels were not sufficiently elevated. Third, PCT levels could have been higher in patients with cancer or immune diseases, but the patients’ comorbidities and medications were not included in the analyses in this study. Fourth, because the fecal PCR test is not a test tool that can accurately diagnose all pathogens, even in patients with bacterial colitis, bacterial pathogens may not be detected by fecal PCR testing. In particular, although C. difficile-associated diarrhea (CDAD) has pathophysiological differences from other bacterial colitis, mild CDAD patients may have been included in the nonbacterial colitis group in our study. Nevertheless, the advantage of this study is that it identified the feasibility of PCT, CRP, and stool PCR testing in a relatively large number of patients. This is the first study to investigate the feasibility of PCT and CRP in differentiating between bacterial and nonbacterial colitis based on the stool PCR test.

In conclusion, our study showed that neither serum PCT nor CRP levels helped differentiate between bacterial and nonbacterial colitis. Although serum PCT or CRP may be helpful in the clinical judgment of bacterial colitis when considered in conjunction with history, physical examination, and other laboratory values, caution is warranted in differentiating between bacterial and nonbacterial colitis using PCT or CRP alone. Larger prospective studies will help validate these results.