Article information

Korean J Pancreas Biliary Tract. 2025;30(3):119-127

Publication date (electronic) : 2025 July 31

doi : https://doi.org/10.15279/kpba.2025.30.3.119

Wootaek Seo^,*, Hyeon-Gi Kim^,*, Hee-Eon Lim, Kwangrok Jung, Jong-Chan Lee, Jin-Hyeok Hwang, Jaihwan Kim

Department of Internal Medicine, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Seongnam, Korea

서우택^,*, 김현기^,*, 임희언, 정광록, 이종찬, 황진혁, 김재환

서울대학교 의과대학 분당서울대학교병원 내과

Corresponding author : Jaihwan Kim Department of Internal Medicine, Seoul National University Bundang Hospital, Seoul National University College of Medicine, 82 Gumi-ro 173 beongil, Bundang-gu, Seongnam 13620, Korea Tel. +82-31-787-7075 Fax. +82-31-787-4051 E-mail: drjaihwan@snu.ac.kr

*Wootaek Seo and Hyeon-Gi Kim contributed equally to this work as the first authors.

Received 2025 April 23; Revised 2025 May 22; Accepted 2025 May 26.

Abstract

Trans Abstract

INTRODUCTION

Cancer can lead to various complications in patients [1]. Common issues include cachexia, infection, and pain [2,3]. Among these, biliary tract infection and pyogenic abscess are frequent causes of fever in patients with pancreatic cancer [4]. However, fever can also arise from non-infectious causes, such as neoplastic or drug-induced fever, including reactions to antibiotics. Non-infectious fever is relatively common in clinical practice [5,6]. Given the risk of sepsis or potential disruption of anti-cancer treatment, prompt and accurate identification of bacterial infections is essential [7,8].

Diagnosing non-septic fever can be challenging due to the complexity of medical conditions and the high incidence of pyogenic infections in patients with pancreatic cancer. Blood cultures are important for distinguishing between infectious and non-infectious causes of fever, but they have limitations in sensitivity, specificity, and turnaround time [9]. Therefore, there is a growing need for reliable biomarkers that can quickly and accurately predict bacterial infections.

Procalcitonin is a biomarker used to detect bacterial infections [10]. It is a protein produced by the thyroid gland and released into the bloodstream in response to bacterial infections [11]. Levels of procalcitonin can be measured through a simple blood test, and elevated levels may indicate the presence of a bacterial infection. While its effectiveness has been extensively studied across various patient populations [12], its reliability as a marker for bacterial infections in febrile patients with pancreatic cancer remains unclear.

This study aimed to evaluate the utility of procalcitonin as a biomarker in febrile patients with pancreatic cancer.

METHODS

RESULTS

1. Comparison of biomarkers for predicting bacterial infection

A total of 149 pancreatic cancer patients experienced fever, leading to 199 individual febrile episodes. Twelve patients had recurrent febrile episodes, resulting in duplicate counting in the total event number. Five patients experienced three recurrences, three had four recurrences, and two patients had five and six recurrences, respectively.

The optimal threshold for WBC count was determined to be 11,450/µL, with a sensitivity of 50.0% and a specificity of 63.8% for detecting positive blood cultures. Similarly, the identified cutoff for CRP level was 12.36 mg/dL, with sensitivity and specificity values of 52.0% and 71.1%, respectively. The optimal threshold for procalcitonin level was determined to be 1.84 ng/mL, with a sensitivity of 60.0% and a specificity of 81.9%.

The AUC values for WBC count, CRP, and procalcitonin in predicting bacterial infection based on blood culture results were 0.550, 0.580, and 0.763, respectively (Fig. 1). Comparative analyses were conducted to assess differences in AUC values among these biomarkers. The AUC difference was 0.031 for WBC versus CRP (p=0.6128), 0.212 for WBC versus procalcitonin (p=0.0004), and 0.182 for CRP versus procalcitonin (p=0.001). Combining the three biomarkers did not improve infection prediction compared to using each biomarker individually (AUC values of combined biomarkers; CRP and WBC, 0.555; procalcitonin and CRP, 0.543; procalcitonin and WBC, 0.559).

Fig. 1.

The receiver operating characteristics curve analysis shows the diagnostic performance of WBC count and CRP and procalcitonin levels in predicting bacterial infection, with area under the curve values of 0.550, 0.580, and 0.763, respectively. WBC, white blood cell; CRP, C-reactive protein.

2. Correlations between biomarkers

Scatter plots illustrating the correlations among various biomarkers are shown in Fig. 2. Notably, a significant correlation coefficient (r) of 0.31 (p<0.001) was observed between CRP and procalcitonin levels (Fig. 2A). Similarly, correlation coefficients of 0.26 (p<0.001) and 0.16 (p=0.021) were observed between WBC count and procalcitonin levels (Fig. 2B) and between WBC count and CRP levels (Fig. 2C), respectively.

Fig. 2.

Scatter plots depict the correlation between biomarker levels. (A) The correlation coefficient between CRP and procalcitonin was 0.31 (p<0.001). (B) The correlation coefficient between WBC count and procalcitonin was 0.26 (p<0.001). (C) The correlation coefficient between WBC count and CRP was 0.16 (p=0.021). CRP, C-reactive protein; WBC, white blood cell.

3. Comparison of baseline characteristics based on procalcitonin cutoff levels

Patients were divided into two groups based on the procalcitonin cutoff value: ≥1.84 ng/mL (higher group) vs. <1.84 ng/mL (lower group) (Table 1). The mean age differed significantly between the two groups (67.5 years in higher group vs. 63.7 years in lower group; p=0.007). However, there were no significant differences in gender, past medical history, surgical resectability, current cancer therapy, and chemotherapy regimen.

Table 1.

Baseline characteristics of patients

A significant difference was also observed in bacterial culture results between the higher and lower procalcitonin groups. Gram-negative bacterial infections were significantly more frequent in the higher procalcitonin group and less common in the lower procalcitonin group, whereas negative blood culture results were more common in the lower group. In contrast, gram-positive infections showed no significant difference between the two groups.

The length of hospital stay after fever onset was significantly longer in the higher procalcitonin group (9.36 vs. 7.37 days; p=0.0266) (Fig. 3).

Fig. 3.

Box plot comparing hospital stay duration after fever onset between the higher and lower procalcitonin groups. Patients with higher procalcitonin levels (≥1.84 ng/mL) had significantly longer hospital stay after fever onset compared to those with lower procalcitonin levels (<1.84 ng/mL) (mean, 9.36 vs. 7.37 days; p=0.0266).

4. Causes of fever

Table 2 outlines the various clinical causes of fever. Cholangitis was the most common cause (45 cases; 22.6%), followed by pneumonia (20 cases; 10.1%). In 10 cases (5.0%), the cause of the fever remained undetermined. When stratified by procalcitonin levels, the distribution of fever causes showed no statistically significant differences between the two groups.

Table 2.

Causes of fever

5. Identified bacterial pathogens

Bacterial pathogens were detected in 64 isolates, as presented in Table 3. The most commonly identified pathogen was the gram-negative Escherichia coli (19 isolates; 29.7%), followed by the gram-positive Enterococcus spp. (14 isolates; 21.9%). When stratified by procalcitonin levels, Escherichia coli was significantly more prevalent in the higher procalcitonin group (42.1% vs. 11.5%; p=0.009). Conversely, Pseudomonas aeruginosa was significantly more prevalent in the lower procalcitonin group (0.0% vs. 15.4; p=0.024). Other bacteria showed no significant differences between the groups.

Table 3.

Pathogens isolated from blood cultures

The distribution of bacterial culture results according to procalcitonin levels is summarized in Table 4. Higher procalcitonin levels were significantly associated with gram-negative bacterial infections, whereas cultures with no bacterial growth were more prevalent in the lower procalcitonin group.

Table 4.

Distribution of blood culture results

DISCUSSION

This study aimed to evaluate the utility of procalcitonin as a biomarker for predicting bacterial infections in febrile patients with pancreatic cancer. While previous studies have explored the clinical significance of procalcitonin in febrile patients with cancer [6,17], studies especially focused on pancreatic cancer remained limited. Our study addresses this gap by investigating procalcitonin not only as a diagnostic tool but also as a potential prognostic marker for clinical outcomes in this patient population.

Our findings indicate that procalcitonin demonstrated higher AUC values compared to CRP or WBC count. Additionally, it demonstrated superior discriminative ability in the early stages of febrile episodes. These results suggest that procalcitonin could serve as a more reliable biomarker for predicting bacterial infections in febrile patients with pancreatic cancer, facilitating early triage and potentially improving outcomes for this vulnerable group. Notably, we identified a subset of patients who exhibited elevated procalcitonin levels despite having normal or borderline WBC and CRP values, and were subsequently diagnosed with bacteremia. This suggests that procalcitonin may provide complementary diagnostic information independent of conventional inflammatory markers, particularly in diagnostically ambiguous situations.

Based on the upper reference level of procalcitonin at 0.5 ng/mL, previous research in patients with hematological cancers suggested an optimal cutoff level of 0.5 ng/mL [17]. However, a study focusing on patients with advanced solid tumors identified a higher optimal cutoff value of 1.52 ng/mL [6]. Our study further supports this trend, identifying 1.84 ng/mL as the optimal cutoff for procalcitonin, yielding the highest positive predictive value (sensitivity, 60.0%; specificity, 81.9%). This relatively elevated cutoff may be attributed to the clinical characteristics of our cohort, which included a substantial proportion of patients with metastatic pancreatic cancer, a high prevalence of biliary infections, and a considerable number receiving chemotherapy. These factors are likely to contribute to an enhanced systemic inflammatory response. Accordingly, procalcitonin thresholds may need to be interpreted with consideration of specific patient populations and clinical settings, as supported by recent evidence suggesting infection-specific variability in optimal cutoff values [18].

Building on these observations, the association between elevated procalcitonin levels and bacterial infections suggests its potential utility in clinical decision-making. Patients with procalcitonin levels of ≥1.84 ng/mL were not only more likely to have a bacterial infection but also experienced longer hospital stays, suggesting its possible role as a prognostic marker. In addition, a higher proportion of these patients received broad-spectrum antibiotics, indicating that procalcitonin levels may reflect not only infection severity but also the intensity of therapeutic interventions. Consistently, previous studies have shown that higher procalcitonin levels are associated with worse clinical outcomes, including extended hospitalization [19].

Moreover, the significant differences in bacterial culture results across procalcitonin groups reinforce its relevance in distinguishing infection types. The predominance of gram-negative infections in the high procalcitonin group supports the notion that procalcitonin may serve as an indicator of gram-negative bacterial infections rather than infections caused by other species. This is consistent with previous research showing significantly higher procalcitonin levels in gram-negative sepsis compared to gram-positive sepsis [20]. Conversely, the higher frequency of negative blood cultures in the low procalcitonin group suggests that alternative non-infectious causes of fever should be considered in these patients. These findings highlight the potential of procalcitonin to enhance current diagnostic approaches in febrile pancreatic cancer patients, enabling more targeted antibiotic strategies and reducing unnecessary antimicrobial exposure.

Scatter plots were used to visually assess the correlations among the three biomarkers. While the correlation coefficients between biomarker levels were relatively modest, the associated p-values indicated statistical significance. These findings suggest that each biomarker may be used interchangeably despite differences in their discriminative abilities in febrile patients with pancreatic cancer.

This study has several limitations. First, procalcitonin levels can be elevated in the absence of infection due to factors such as severe stress, trauma, surgery, or cardiogenic shock [21,22]. Additionally, severe liver dysfunction has been shown to influence procalcitonin levels [23]. While procalcitonin is typically not elevated in viral infections, it can be elevated in fungal infections [18,20]. In this study, a few non-bacterial infections were identified through microbiological tests, including polymerase chain reaction for viral pathogens and blood cultures for fungal organisms. Specifically, two cases of coronavirus disease 2019 and one fungal infection were included. Although potential confounding factors, their small number likely had minimal impact on the overall findings. Second, individuals with CKD exhibit elevated baseline levels of inflammatory biomarkers due to increased production of inflammatory cytokines [24,25]. However, the study included only five patients with CKD, potentially limiting the generalizability of the findings. Third, although our study demonstrated meaningful results for procalcitonin, the use of empirical antibiotics remains practically unavoidable in current clinical practice [26].

In conclusion, procalcitonin exhibited superior diagnostic accuracy in distinguishing positive bacterial blood cultures in febrile patients with pancreatic cancer, particularly in early fever stages. Its elevation reinforced its role as a bacterial infection marker and a prognostic indicator of infection severity. These findings support procalcitonin’s utility as both a diagnostic and prognostic biomarker for guiding clinical decision-making

Article information Continued