Abstract
Background/Aim: Whether the presence of bacteria in the bile or postoperative infections sustained by microorganisms with different levels of drug-resistance are associated with changes in the oncologic prognosis of patients undergoing surgery for pancreatic cancer has not been thoroughly investigated. The aim was to study the association of bile contamination, postoperative infections, and multi-level resistance with long-term outcome. Patients and Methods: Prospectively maintained databases were queried for patients who underwent pancreatoduodenectomy (PD). Patients who underwent preoperative biliary stenting prior to PD and an intraoperative bile culture were included. The levels of bacterial resistance of intraoperative bile cultures and of specimens of postoperative infections were stratified into multidrug sensitive (MDS), multidrug-resistant (MDR), and extensive drug-resistant (XDR). Results: A total of 267 patients met the inclusion criteria. The Kaplan–Meier survival curves for overall survival (OS) of patients having no bacteriobilia or positive cultures with MDS versus MDR/XDR bacteria were not statistically different (log-rank=0.9). OS of patients stratified for no postoperative infection or infections by MDS was significantly better than those having MRD/XDR isolates (log-rank=0.04). A Cox multivariate model showed that having MRD/XDR postoperative infections was and independent variable for worse OS (HR=1.227; 95%CI=1.189-1.1918; p=0.036). Conclusion: Postoperative drug resistant infections are a significant risk factor for poor OS after pancreatoduodenectomy for ductal adenocarcinoma.
Ductal adenocarcinoma of the pancreatic head is first diagnosed based on jaundice appearance in more than 70% of the cases (1). Current recommendations suggest avoiding routine preoperative biliary stenting (PBS) and limiting the procedure to symptomatic jaundice (bilirubin level ≥250 mmol/l) or planned neoadjuvant treatments (2). However, persistent cholestasis may evoke hepatic inflammation, liver cell damage, impaired protein synthesis, imbalanced T-cell homeostasis, altered neutrophil phagocytosis, and coagulopathy. This homeostasis derangement has been associated with increased risk of postoperative infection, anastomotic leakage, poor wound healing, and excessive perioperative bleeding (3-6). Moreover, preoperative jaundice has been shown to be an independent adverse prognostic factor for long-term survival in pancreatic cancer patients (6, 7). Consequently, PBS to lower bilirubin levels and to restore liver functions prior to pancreatoduodenectomy (PD) has been used in the attempt of limiting the risk of postoperative morbidity and improving oncologic outcomes. However, stent placement creates a communication between the biliary system and the intestinal tract, facilitating upstream bacterial migration. The resulting bacteriobilia with microorganisms harboring different levels of antimicrobial resistance have been independently associated with higher rates of postoperative infectious complications and other life-threatening surgery-related short-term morbidity (8-11). Moreover, some studies have reported that the use of PBS has an unfavorable impact on the long-term survival of patients with resected pancreatic cancer (12, 13). The occurrence of severe postoperative complications, and in particular the infectious ones sustained by resistant bacteria, represents another risk factor for cancer recurrence, and for any mortality reason in the long-term (14-16). Whether finding a contamination of the bile with bacteria or microorganisms sustaining postoperative infections with different levels of antimicrobial resistance may be associated to changes in oncologic prognosis has not been thoroughly investigated. Thus, we aimed to study the potential association of bile contamination and postoperative infections with multi-level resistant microorganisms with long-term survival, in patients undergoing PD for pancreatic ductal adenocarcinoma (PDAC).
Patients and Methods
Study overview and patient selection. This study used anonymized patient data. The study protocol followed the ethical guidelines of the 1975 Declaration of Helsinki (as revised in Brazil 2013). Local Ethical Committees reviewed the protocol and deemed that formal approval was not required owing to the retrospective, observational, and anonymous nature of this study. Results are reported according to Strengthening the Reporting of Observational Studies in Epidemiology (17).
Prospectively maintained databases from three European academic medical centers (San Gerardo Hospital, Milano-Bicocca University, Monza; Humanitas Research Hospital, Rozzano-Milano; University Medical Center Schleswig-Holstein, Luebeck) were retrospectively queried for patients who underwent PD from January 2014 to December 2018.
All patients who underwent PBS prior to PD were included. Patients undergoing PDs for histology other than PDAC, lacking intraoperative bile culture, or deceased within 90 days after surgery were excluded. Biliary drainage was performed according to international recommendations (2) when bilirubin level was equal or greater than 15 mg/dl (250 mmol/l) or when neoadjuvant treatment was planned. Information on patient demographic, date of stent placement and related post-procedure complications, surgery-related morbidity, final pathology reports, and perioperative oncologic treatments were retrieved from the datasets or medical records.
Study endpoints. The primary endpoint was to study the potential association among bile contamination, development of infections in the postoperative course, levels of antimicrobial resistance, and overall survival (OS) in patients undergoing PD for PDAC. The secondary endpoint was to study the association of the above variables with relapse-free survival (RFS).
Surgical technique, bacterial cultures, and antimicrobial resistance levels. Pylorus-preserving pancreatoduodenectomies or classic Whipple procedures were performed or supervised by experienced surgeons. Antibiotic prophylaxis was carried out with I.V. first- or second-generation cephalosporin and repeated after 4 h during operation.
Bile was harvested intraoperatively at the time of biliary duct transection. The levels of antibiotic resistance of the bile and specimens related to postoperative infections were defined according to the classification of Magiorakos et al. (18) and were used to stratify microbes into multidrug sensitive (MDS), multidrug-resistant (MDR), and extensive drug-resistant (XDR).
Postoperative morbidity. Postoperative infectious morbidity was defined as the occurrence of any of the following infections: wound, organ-space, urinary tract, pneumonia, sepsis, infected pancreatic fistula, blood, or any vascular device (19). Complications were considered as any 30-day deviation from the patient recovery, prolonging the length of hospitalization, requiring supplementary care or readmission. Major morbidity was defined as any complication with a Clavien–Dindo (20) classification grade ≥ III. Long-term variables and outcomes
All patients were followed using measurement of serum carbohydrate antigen 19-9, abdominal ultrasound, contrast computed tomography or magnetic resonance imaging, and office visits. Briefly, each patient was followed up every 3 months for the first two years and then every six months, or on clinical demand. OS was defined as the time interval in months from surgery to death; if alive, patient data were censored at the last available visit. RFS was defined as the time interval in months from surgery to recurrence (local or distant) or death. In case of no recurrence or death, patient data were censored at the last available follow-up. Patient surveillance was closed at the end of May 2020. We used the 8th edition of the American Joint Committee on Cancer staging system for PDAC.
Statistical analysis. Overall sample description was performed using number and percentages for categorical variables or median and interquartile range (IQR) for continuous variables. Estimates of OS over time for each type of bile contamination were obtained using the Kaplan–Meier (KM) method and compared using the log-rank test. OS curves were also compared according to the presence of postoperative infection and to the pattern of resistance. The latter comparison was also evaluated in terms of hazard ratio (HR) using a univariate cox model. Moreover, the association between postoperative infections and OS was estimated in a multivariate Cox model by adjusting for several other potential prognostic factors selected based on the available literature and clinical plausibility (15, 21-24) (each factor was also tested in a univariate Cox model). The same analyses for OS were repeated for the RFS endpoint after excluding patients with positive resection margins (<1 mm).
Results
During the study period, 643 PDs were performed at the three participating centers, and 267 patients met the inclusion criteria and were analyzed (Figure 1).
Patient selection chart.
The baseline characteristics, pathology features, and intra- and postoperative parameters are summarized in Table I. The results of intraoperative bile cultures indicated that 75 (28.1%) patients had no contamination, MDS bacteria were retrieved from bile in 90 patients (33.7%), and 102 (38.2%) had MDR/XDR isolates. The proportion of postoperative infection was 53.2% (142/267). Of these, 51 (35.9%) were sustained by MDS microorganisms and 91 (64.1%) by MDR/XDR bacteria. The most frequent bacteria retrieved from bile were: E. coli (19.5%), K. pneumoniae (15.7%), and E. faecalis (13.5%). Similar frequencies were observed for isolates from postoperative infection sites (Table I).
Descriptive statistics of the studied cohort.
Table II describes the relationship between isolates from bile and from postoperative infection sites. There was no strong relationship between finding bacteria in the bile and the occurrence of postoperative infections and vice versa, namely sterile bile and no postoperative infectious complications. Similarly, the same level of resistance was not frequently confirmed by comparing bile and postoperative isolates.
Contingency table for microorganisms retrieved from bile and from postoperative infection sites
The median follow-up time was 25 months (IQR=13-40). Figure 2A depicts the KM survival curves for OS of patients having no bile contamination or positive bile cultures with MDS (combined) versus MDR/XDR bacteria (log-rank test=0.9). The KM survival curves for OS were calculated for patients stratified for having no postoperative infection or experiencing infectious morbidity sustained by MDS (combined) versus MRD/XDR germs (Figure 2B; log-rank test=0.04). The RFS KM curves for no contamination/MDS versus MDR/XDR bile contamination are shown in Figure 3A (log-rank=0.9) and the results of postoperative infections stratified for resistance levels are shown in Figure 3B (log-rank=0.05).
Overall survival curves. A) Kaplan–Meier curves for overall survival (OS) of patients having no bile contamination or positive cultures with multidrug sensitive (MDS) versus multidrug resistant/extensive drug-resistant (MDR/XDR) bacteria (log-rank=0.9). B) Kaplan–Meier curves for OS of patients stratified for having no postoperative infection or experiencing infectious morbidity sustained by MDS versus MRD/XDR germs (log-rank=0.04).
Recurrence-free survival curves. A) Kaplan–Meier curves for recurrence-free survival (RFS) of patients having no bile contamination or positive cultures with multidrug sensitive (MDS) versus multidrug resistant/extensive drug-resistant (MDR/XDR) bacteria (log-rank=0.9). B) Kaplan– Meier curves for RFS of patients stratified for having no postoperative infection or experiencing infectious morbidity sustained by MDS versus MRD/XDR germs (log-rank=0.05).
Table III reports the results of the Cox univariate and multivariate models for OS. Bacteriobilia was not an independent risk factor for OS, while the development of a postoperative infection sustained by MDR/XDR bacteria was associated with a significant increase in the risk of death (HR=1.227; p=0.036) and in particular when an infection was sustained by K. pneumoniae (HR=1.767; p=0.023). As expected, tumor stage and grading, lymph node metastases, positive resection margins, and no or incomplete postoperative chemotherapy were all associated with a significant increased risk of long-term mortality. The results of the RFS analysis showed a similar trend when compared to OS. However, due to the smaller sample size (n=125), the prognostic effect of some variables at multivariate Cox did not reach statistical significance (Table IV).
Results of the Cox uni- and multi-variate models for overall survival.
Results of the Cox univariate and multivariate models for recurrence-free survival.
Discussion
The present findings confirm a significant association of postoperative infections and dismal oncologic prognosis, that has been already reported for other tumor histotypes and locations (25-28). Postoperative infections may evoke a systemic immunosuppression by the continuous and excessive release of cytokines with pro-inflammatory ability (29-31). Persisting high levels of IL-6 may decrease the number, maturation, and activity of cytotoxic T-lymphocytes, natural killer, and other immunocompetent circulating cells such as dendritic antigen-presenting cells (32-35). More recently, it was found that tumor infiltrating T cells in human PDAC, particularly CD4-helper T cells, produce proinflammatory mediators such as tumor necrosis factor alpha and IL-17A (36). Moreover, the pro-inflammatory T-helper17 cells produce high levels of IL-21, a pleiotropic cytokine with immunosuppressive activity and high concentration of T-helper17 cells and IL-21 positive T cells, in specimens of PDAC was associated with poor patient survival (37).
An additional attractive interpretation of the present results is the direct role of intestinal microbiome on cancer growth. Indeed, it has been reported that gut dismicrobism may blunt anti-neoplasm immune responses (38), and specific bacteria metabolites are associated with reduced chemotherapy tolerance (39). Furthermore, specific gammaproteobacteria, found in PDAC specimens can modulate the tumor sensitivity to chemotherapy (40).
Despite the association of persistent inflammation/infection and tumor progression seems sufficiently documented, the role and degree of antimicrobial resistance on long-term survival in cancer patients has not been thoughtfully explored. This potential association may be relevant to be investigated because MDR bacteria have been recognized as a source of major postoperative complications (8, 10, 11, 41-44) and when they occur, are associated with dismal oncologic outcome in patients with PDAC (15, 16).
The uniqueness of the present study is having documented that only MDR bacteria, and in particular K. pneumoniae, seems to play a role in affecting long-term survival after pancreatic resection for ductal adenocarcinoma. Our data partially support the finding of Weniger et al. (45) who analyzed 211 patients with intraoperative bile cultures and reported an association between bile infection sustained by different pathogens and increased disease recurrence after pancreatoduodenectomy for PDAC. Moreover, they reported that K. pneumoniae infections promoted chemoresistance to adjuvant gemcitabine and the eradication of K. pneumoniae with quinolone treatment improved the oncologic outcome.
A strong association of bacteriobilia and a higher risk of having postoperative infectious complications and other life-threatening surgery-related complications in the postoperative course has been repeatedly reported (8-11). Bacteriobilia is inevitably related with biliary stenting (9, 10). This repeated observation suggests an unfavorable effect on the long-term survival in patients bearing PBS and undergoing resection of a pancreatic cancer, and indirectly advocates a causal effect of bile microbes on oncologic outcome (12, 13). However, the present results suggest a non-significant effect of bacteriobilia and antimicrobial resistance patterns of the bile on the oncologic prognosis of patients bearing a preoperative biliary stent and undergoing PD for PDAC. These findings appear in contrast with previous publications. Darnell et al. (46) reported that cholangitis after biliary stenting was associated with a 2.5-fold increased hazard ratio for all-cause long-term mortality in patients who received a Whipple operation for PDAC. The relevant role of an hyperinflammatory response after PBS on oncologic outcomes of periampullary carcinoma has been also suggested by Fujiwara et al. (47). They developed a prognostic score that combine PBS and C-reactive protein (CRP) as a marker of inflammation. The results advised that high CRP levels after biliary stenting was an independent predictor of tumor recurrence and overall long-term survival. The diversity in results may have different suitable explanations. PBS may convert into an overt cholangitis with systemic inflammatory response, or in an uneventful and asymptomatic bacteriobilia with colonizing microorganisms having different patterns of drug resistance as suggested by our results. Colonizing microorganisms can trigger a cooperative bond with the host by stimulating both adaptive and innate immune response (48). Molecular patterns expressed on the bacterial cell membrane can trigger recognition receptors and this may induce a symbiotic relationship to prevent pathogens from becoming infective. Gram-positive bacteria through peptidoglycans and Gram-negative bacteria through lipopolysaccharides expressed on their membranes, can induce the cytoplasmic release of NF-kB. After entering the nucleus, NF-kB stimulates the synthesis and release of chemokines, which in turn induce the migration of neutrophils to the site of bacteria colonization. This mechanism restrains microorganisms to access to the blood stream through a self-limited inflammatory response (48). The observation that roughly 60% of our patients with positive bacterial cultures of the bile did not develop a postoperative infection, and 35% of the cases of MDR/XDR bile contamination had an infection with non-resistant bacteria is consistent with the above hypothesis of symbiotic immune tolerance and that in specific circumstances some microorganisms act as simple colonizers without transforming into pathogens, and may justify the lack of association between bacteriobilia and oncologic outcome. Likely, this symbiotic interaction leads to immune homeostasis or even to complete bacterial clearance, and accordingly to no expression of disease. Conversely, malfunction of these immune mechanisms may eventually translate into an overt infection and excessive inflammation. NF-κB synthesis from activated B cells has been frequently reported to be upregulated in PDAC. There is of evidence that NF-κB pathways have a relevant role in the progression of pancreatic cancer because inhibition of NF-κB, in experimental studies (49), increases tumor sensitivity to chemotherapy drugs by inducing apoptosis and inhibiting angiogenesis with a resulting reduction in tumor growth (50).
In a broader interpretation, the results of Darnell et al. (46) and Fujiwara et al. (47) advise that only patients developing an overt PBS-related cholangitis or hyperinflammation irrespective of the pattern of resistance, may be regarded as having failed an adaptive response to bile colonizing microbes.
The study has some limitations. As all retrospective studies, some variables related to OS and RFS might have been missed. Moreover, we evaluated associations between events, and not cause-effect relationships. Therefore, it is possible that patients having postoperative infection with MDR bacteria simply represent a subgroup with hidden baseline frailty characteristics or with more aggressive cancer biology. Second, we intentionally excluded patients who received surgery without PBS. However, the objective of the study was to evaluate patients with high risk of developing bile contamination and its possible consequences on outcomes. At present, nearly half of the patient candidates to PD necessitate a preoperative jaundice treatment (51). Given the promising results of neoadjuvant treatments (52) the proportion of patients with PBS is likely to increase in the future. Thus, we found worthwhile, within the stent bearing population, trying to identify subjects at higher risk of dismal survival.
In summary, the present results imply a role of postoperative infections sustained by MDR microorganisms, and in particular K. pneumoniae, in the OS of patients receiving PD for PDAC. In contrast, bacteriobilia does not seem to play a role in affecting the oncologic prognosis.
Footnotes
Authors’ Contributions
Luca Gianotti, Kim C. Honselmann, Marco Angrisani, Davide P. Bernasconi: Conceptualization, Methodology, Software. Francesca Gavazzi, Marta Sandini, Tobias Keck, Ulrich Wellner, Louisa Bolm, Natalie Petruch, Giovanni Capretti, Gennaro Nappo Davide P. Bernasconi: Data curation, Writing- Original draft preparation. Kim C. Honselmann, Marco Angrisani, Francesca Gavazzi, Marta Sandini: Visualization, Investigation. Tobias Keck, Alessandro Zerbi: Supervision. Davide P. Bernasconi, Marta Sandini: Software, Validation. Luca Gianotti, Kim C. Honselmann, Marta Sandini, Tobias Keck, Ulrich Wellner, Louisa Bolm, Davide P. Bernasconi, Alessandro Zerbi: Writing-Reviewing and Editing.
Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.
- Received January 17, 2022.
- Revision received March 21, 2022.
- Accepted March 23, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.








