Preoperative C-reactive protein and textbook outcome following pancreaticoduodenectomy for pancreatic head adenocarcinoma

Abstract

BACKGROUND

Textbook outcome (TO) captures an ideal postoperative course after pancreaticoduodenectomy (PD) and is increasingly used for surgical quality assessment. However, standardized TO-based data after PD from Türkiye remain limited, restricting comparison with international benchmarks. We hypothesized that preoperative systemic inflammation, reflected by C-reactive protein (CRP), would be associated with TO achievement after PD for pancreatic head adenocarcinoma.

AIM

To determine the TO achievement and preoperative factors associated with TO after PD for pancreatic head adenocarcinoma.

METHODS

This study included 151 patients who underwent PD between March 2015 and January 2022. After excluding those with benign pathology and pathology other than pancreatic head adenocarcinoma, 100 patients were analyzed. TO was defined using the van Roessel framework. A preoperative-only multivariable logistic regression model was constructed, and CRP was modeled as log₂(CRP + 1). Receiver operating characteristic analyses were performed separately for CRP alone and multivariable models.

RESULTS

TO was achieved in 53 patients (53.0%). The most frequent unmet criteria were severe complications (31.0%), 30-day readmission (21.0%), and in-hospital mortality (10.0%). Ninety-day mortality was 12.0%. In the primary preoperative model, each doubling of CRP was independently associated with lower odds of TO achievement odds ratio, 0.46; 95% confidence interval: 0.33-0.64; P < 0.001). This association remained stable in the bilirubin-adjusted sensitivity model (odds ratio = 0.45; 95% confidence interval: 0.32-0.63; P < 0.001). CRP alone predicted non-TO with an area under the curve of 0.815.

CONCLUSION

Higher preoperative CRP was independently associated with lower odds of achieving TO after PD and may support risk stratification, but prospective validation is required.

Key Words: C-reactive protein; Pancreatic head adenocarcinoma; Pancreaticoduodenectomy; Postoperative complications; Risk stratification; Surgical quality; Textbook outcome

Core Tip: Textbook outcome (TO) provides a multidimensional benchmark for pancreaticoduodenectomy, but standardized TO data from Türkiye are scarce. In this single-center cohort of pancreaticoduodenectomy for pancreatic head adenocarcinoma, TO was achieved in 53.0% of patients. The main causes of TO failure were severe complications, readmission, and in-hospital mortality. Preoperative C-reactive protein remained the most consistent preoperative signal associated with TO achievement, and this association was not explained by biliary drainage or bilirubin. These findings position C-reactive protein as a practical marker for preoperative risk stratification rather than as a validated decision-making tool.

INTRODUCTION

Pancreaticoduodenectomy (PD) is the cornerstone of curative-intent treatment for resectable pancreatic head adenocarcinoma[1]. The extent of resection, vascular anatomical variations, and the need for multistep gastrointestinal reconstruction make PD a complex procedure associated with a high risk of morbidity[2,3]. Indeed, in a prospective cohort including 4223 patients from 67 countries, 2501 patients underwent PD. In the PD subgroup, the overall complication rate was 73.6%, the major complication rate (Clavien-Dindo ≥ IIIa) was 28.2%, and the 90-day mortality rate was 6.2%[2].

Conventional surgical quality assessment relies on individual metrics such as mortality, major complication rate, length of hospital stay, and readmission[4,5]. However, for outcomes with low event rates, such as mortality and failure to rescue, detecting statistically significant inter-institutional differences at current center volumes remains challenging[4]. Moreover, in procedures with heterogeneous complication profiles such as PD, individual metrics do not fully capture the overall postoperative course[4,6]. These limitations have led to growing interest in composite quality measures such as textbook outcome (TO) in pancreatic surgery[6].

TO is an “all-or-nothing” composite quality measure, originally described in colorectal surgery and subsequently adapted to pancreatic surgery[5,6]. It requires the simultaneous achievement of all predefined components of an ideal postoperative course in the same patient[5,6]. TO has emerged as an increasingly used tool for quality assessment in pancreatic surgery and for evaluating the effectiveness of perioperative interventions such as enhanced recovery after surgery protocols[7-9]. Nevertheless, composite outcome measures for PD have not been fully standardized; the European TO framework and the North American concept of optimal pancreatic surgery include different component sets, and subsequent efforts have sought to harmonize these definitions[10,11]. In the present study, we adopted the van Roessel definition, which is based on national audit data and international expert consensus. Accordingly, TO was defined as the absence of clinically relevant postoperative pancreatic fistula, bile leak, or postpancreatectomy hemorrhage (all grade B/C); the absence of severe complications (Clavien-Dindo ≥ III); no readmission within 30 days; and no in-hospital mortality[6]. TO achievement has been associated with short-term surgical quality and long-term survival[12].

A recent systematic review and meta-analysis demonstrated that several patient- and process-related variables, including preoperative biliary drainage and increased intraoperative blood loss, may be associated with failure to achieve TO after PD[13]. However, most TO evidence has been generated from international audit-based or multicenter cohorts, whereas data from Türkiye remain limited. Published PD series from Türkiye have predominantly reported conventional single endpoints, such as short-term morbidity, mortality, hemorrhagic complications, nutritional status, and mortality risk factors, rather than standardized TO-based composite quality assessment[14-18]. To our knowledge, no standardized TO-based assessment of PD outcomes from Türkiye has been reported in the indexed English-language literature. This limits the comparability of national PD outcomes with contemporary international benchmarks based on a harmonized multidimensional quality framework[6,8]. Therefore, evaluating TO achievement and its associated factors in a PD cohort from Türkiye may provide a more comparable estimate of short-term surgical quality and help identify clinically relevant risk signals for TO failure. The aim of this study was to determine the TO rate in a single-center PD cohort from a tertiary center in Türkiye, to describe the distribution of unmet TO components, and to evaluate preoperative clinical factors associated with TO achievement.

MATERIALS AND METHODS

Study design and patient population

This study evaluated short-term surgical outcomes after PD using TO as the primary composite endpoint. Of 151 patients who underwent PD between March 2015 and January 2022, 2 were excluded because of benign pathology, and 49 because of pathology other than pancreatic head adenocarcinoma. The final analytic cohort included 100 patients with pancreatic head adenocarcinoma. The patient selection process is shown in Figure 1.

Figure 1 Patient selection flow diagram. Of 151 patients who underwent pancreaticoduodenectomy between March 2015 and January 2022, 2 were excluded because of benign pathology, and 49 because of pathology other than pancreatic head adenocarcinoma. The final analytic cohort comprised 100 patients with pancreatic head adenocarcinoma.

This retrospective research project was reviewed and approved by the Clinical Research Ethics Committee of Çukurova University Faculty of Medicine for compliance with research ethics (decision No. 51, March 13, 2026). The approval covered the retrospective review of institutional data from patients who underwent PD between March 2015 and January 2022. No study-specific intervention or patient contact was performed. Because anonymized retrospective institutional data were used, study-specific informed consent was not required for this retrospective study. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Surgical technique

Patients underwent either classical or pylorus-preserving PD. Pancreatic reconstruction was performed by pancreatojejunostomy in all cases. No patient underwent portal vein, superior mesenteric vein, or arterial resection/reconstruction.

Postoperative management

At our institution, intra-abdominal drains were routinely placed after PD and managed according to clinical findings, drain output, and drainage appearance. Drain fluid amylase was not measured routinely in all patients. Postoperative complications, including clinically relevant postoperative pancreatic fistula, were retrospectively classified according to the available medical records and standard definitions. Discharge was considered when patients were clinically stable, afebrile, able to tolerate oral intake, had adequate pain control, and had no unresolved complications requiring inpatient care. Readmission was defined as any unplanned hospital admission within 30 days after discharge due to a postoperative or surgery-related complication. Emergency department visits without hospitalization, planned outpatient visits, adjuvant treatment-related admissions, and elective admissions were not considered readmissions.

Data collection and variables

Patient data were retrospectively collected under the following categories: Demographic and clinical characteristics, preoperative hematologic and biochemical parameters, biliary drainage status, intraoperative findings, and short-term postoperative outcomes. Variables included in the core comparative analyses were age, sex, body mass index, American Society of Anesthesiologists (ASA) score, preoperative laboratory parameters [C-reactive protein (CRP), albumin, total bilirubin, and carbohydrate antigen 19-9], biliary drainage status, surgical technique (classical vs pylorus-preserving PD), and intraoperative transfusion. Intraoperative transfusion was defined as any red blood cell transfusion administered during surgery. None of the patients in the study cohort received neoadjuvant therapy.

Definition and classification of complications

Postoperative pancreatic fistula was classified according to the 2016 updated International Study Group of Pancreatic Surgery definition, postpancreatectomy hemorrhage according to the International Study Group of Pancreatic Surgery definition, and bile leak according to the International Study Group of Liver Surgery definition[19-21]. Postoperative complications were graded according to the Clavien-Dindo classification[22]. In-hospital mortality was defined as any death occurring during the index hospitalization after PD, regardless of postoperative day. Ninety-day mortality was recorded descriptively for international comparability and included both in-hospital and post-discharge deaths occurring within 90 days after surgery. Only in-hospital mortality was included as a TO component, whereas post-discharge deaths within 90 days were reported separately and did not change TO classification. The timing of death and recorded mortality-associated clinical events were reviewed from the available medical records.

Primary endpoint: TO definition

The primary endpoint of the study was TO. TO was defined according to the framework proposed by van Roessel et al[6], based on national audit data and international expert consensus. Accordingly, TO was defined as the simultaneous fulfillment of the following six criteria in the same patient: Absence of clinically relevant postoperative pancreatic fistula (grade B/C), absence of bile leak (grade B/C), absence of postpancreatectomy hemorrhage (grade B/C), absence of severe complications (Clavien-Dindo ≥ III), no readmission within 30 days after discharge, and no in-hospital mortality. Failure to meet any one of these criteria resulted in classification into the non-TO group. Reasons for TO failure were assessed cumulatively, acknowledging that a single patient could fail to meet more than one criterion.

Statistical analyses

Continuous variables were presented as median and interquartile range (IQR), and categorical variables as frequency and percentage [n (%)]. For comparisons between the TO and non-TO groups, the Mann-Whitney U test was used for continuous variables and Fisher’s exact test for categorical variables. Ninety-five percent confidence intervals (CIs) were calculated for proportions.

Multivariable logistic regression analysis was performed with TO achievement as the dependent variable. The primary model was designed as a preoperative-only risk-stratification model and included age, ASA class ≥ III, preoperative biliary drainage, preoperative CRP, and preoperative albumin. Variables were selected a priori according to clinical relevance, preoperative availability, previous literature, and the need to avoid overfitting. Sex and carbohydrate antigen 19-9 were not included in the multivariable models to preserve model parsimony in this limited cohort, as they were not prespecified core variables for the preoperative risk-stratification model and showed no meaningful imbalance between TO groups. CRP was modeled as log₂(CRP + 1) because of its skewed distribution, and the odds ratio (OR) was interpreted per doubling of CRP.

Intraoperative transfusion was not included in the primary model because it may reflect surgical complexity, intraoperative blood loss, or early perioperative deterioration rather than baseline risk. Because CRP may be influenced by biliary obstruction or drainage-related inflammation, a bilirubin-adjusted sensitivity model was constructed by replacing albumin with preoperative total bilirubin while retaining age, ASA class ≥ III, preoperative biliary drainage, and log₂-transformed CRP. A CRP × biliary drainage interaction term was also tested separately to assess whether the association between CRP and TO differed according to biliary drainage status.

The individual components of TO were not included as independent variables to avoid circular associations. Receiver operating characteristic (ROC) analysis was performed separately for CRP alone and for the multivariable models; therefore, CRP-alone and model area under the curve (AUC) values were reported as distinct analyses. All variables included in the descriptive, comparative, ROC, and multivariable analyses were available for all 100 patients; therefore, a complete-case analysis was performed, and no imputation was required. Statistical analyses were performed in Python 3.13.5 (Python Software Foundation, Wilmington, DE, United States) using pandas 2.2.3 and statsmodels 0.14.6. All tests were two-sided, and P < 0.05 was considered statistically significant.

RESULTS

General characteristics of the cohort

Baseline demographic, preoperative, and operative characteristics stratified by TO status are summarized in Table 1. Preoperative CRP levels were markedly higher in the non-TO group than in the TO group [median: 22.2 (IQR: 6.9-38.5) vs 3.1 (IQR: 1.2-8.8) mg/L; P < 0.001]. Intraoperative transfusion was also more frequent in the non-TO group than in the TO group (53.2% vs 32.1%; P = 0.043), whereas preoperative albumin showed a nonsignificant trend toward lower levels in the non-TO group than in the TO group (P = 0.057). Other demographic, clinical, biochemical, drainage-related, and operative variables were comparable between the two groups.

Table 1 Baseline characteristics of the overall cohort and comparison according to textbook outcome status.

Variable	Total (n = 100)	TO (n = 53)	Non-TO (n = 47)	P value
Age, years	65.0 (56.8-70.2)	65.0 (54.0-71.0)	65.0 (60.0-69.5)	0.525
Sex				1.000
Male	61 (61.0)	32 (60.4)	29 (61.7)
Female	39 (39.0)	21 (39.6)	18 (38.3)
BMI, kg/m2	25.9 (23.3-28.1)	26.4 (23.6-28.1)	25.1 (23.2-27.2)	0.269
ASA score				0.700
I	5 (5.0)	3 (5.7)	2 (4.3)
II	83 (83.0)	45 (84.9)	38 (80.9)
III	12 (12.0)	5 (9.4)	7 (14.9)
CA19-9, U/mL	127.1 (29.6-371.3)	103.7 (17.6-294.0)	150.6 (42.6-413.7)	0.192
Preoperative CRP, mg/L	7.3 (2.1-22.4)	3.1 (1.2-8.8)	22.2 (6.9-38.5)	< 0.001
Preoperative albumin, g/L	33.8 (28.8-39.1)	34.8 (30.4-39.9)	32.0 (25.2-38.0)	0.057
Preoperative total bilirubin, mg/dL	1.7 (0.7-3.7)	1.3 (0.6-3.4)	2.0 (0.8-4.3)	0.215
Preoperative biliary drainage				0.681
Yes	64 (64.0)	35 (66.0)	29 (61.7)
No	36 (36.0)	18 (34.0)	18 (38.3)
Pylorus-preserving PD				1.000
Yes	41 (41.0)	22 (41.5)	19 (40.4)
No	59 (59.0)	31 (58.5)	28 (59.6)
Intraoperative transfusion				0.043
Yes	42 (42.0)	17 (32.1)	25 (53.2)
No	58 (58.0)	36 (67.9)	22 (46.8)

TO: Textbook outcome; BMI: Body mass index; ASA: American Society of Anesthesiologists; CA19-9: Carbohydrate antigen 19-9; CRP: C-reactive protein; PD: Pancreaticoduodenectomy.

TO rate and frequencies of unmet TO criteria

TO was achieved in 53 (53.0%) of 100 patients. The frequencies of unmet individual TO criteria are presented in Table 2. Severe complications were the most frequently unmet criterion, followed by 30-day readmission and in-hospital mortality. Bile leak, postpancreatectomy hemorrhage, and clinically relevant postoperative pancreatic fistula were less frequent component-level failures.

Table 2 Textbook outcome achievement and unmet criteria.

Variable	Value
Textbook outcome achieved, n (%)	53 (53.0)
Clinically relevant POPF (grade B/C), n (%)	4 (4.0)
Bile leak (grade B/C), n (%)	8 (8.0)
Postpancreatectomy hemorrhage (grade B/C), n (%)	5 (5.0)
Severe complications (Clavien-Dindo ≥ III), n (%)	31 (31.0)
30-day readmission, n (%)	21 (21.0)
In-hospital mortality, n (%)	10 (10.0)
Length of hospital stay, days, median (IQR)
TO group	12.0 (10.0-14.0)
Non-TO group	17.0 (13.0-23.0)
P value	< 0.001

POPF: Postoperative Pancreatic Fistula; IQR: Interquartile range; TO: Textbook outcome.

In-hospital mortality occurred in 10 patients (10.0%), and all in-hospital deaths occurred within 30 days after surgery. Two additional patients died after discharge within 90 days, resulting in a 90-day mortality rate of 12.0%. Among patients with in-hospital mortality, the median time to death was 9 days (range: 1-30 days). No in-hospital death was associated with postoperative pancreatic fistula or coded as sepsis in the available records. Postpancreatectomy hemorrhage was recorded in 4 patients, and 3 patients required reoperation. Recorded systemic complications among patients with in-hospital mortality included pulmonary embolism in 4 patients, myocardial infarction-related events in 3 patients, and bleeding-related complications in 3 patients. The two post-discharge deaths occurred on postoperative days 32 and days 36; one occurred outside the index hospital, and the other followed aspiration-related respiratory arrest. Length of hospital stay was significantly shorter in the TO group than in the non-TO group (median: 12 days vs 17 days, P < 0.001).

Cumulative distribution of reasons for TO failure

The cumulative distribution of unmet TO criteria among the 47 patients in the non-TO group is shown in Figure 2. Severe complications were the dominant reason for TO failure, both overall and among patients with single-criterion failure. Overall, 24 patients (51.1%) failed to achieve TO because of a single unmet criterion, whereas 23 (48.9%) had two or more concurrent unmet criteria.

Figure 2 Cumulative distribution of unmet textbook outcome criteria among the 47 patients in the non-textbook outcome group. Percentages were calculated using the non-textbook outcome group as the denominator (n = 47). Criteria are cumulative and not mutually exclusive; therefore, one patient could fail more than one criterion. Single-criterion failure occurred in 24 patients (51.1%), and multiple concurrent failures occurred in 23 patients (48.9%). The most frequent isolated failure was severe complications (Clavien-Dindo ≥ III; 11/24, 45.8%). TO: Textbook outcome.

Preoperative multivariable models and ROC analyses

In the primary preoperative multivariable logistic regression model, TO achievement was used as the dependent variable. The model included age, ASA class ≥ III, preoperative biliary drainage, preoperative CRP, and preoperative albumin. Preoperative CRP was modeled as log₂(CRP + 1), and each doubling of CRP was independently associated with lower odds of TO achievement (OR: 0.46; 95%CI: 0.33-0.64; P < 0.001). The other prespecified covariates were not independently associated with TO achievement (Table 3). The AUC of the primary preoperative model was 0.819.

Table 3 Preoperative multivariable logistic regression models for textbook outcome achievement.

Variable	Primary preoperative model OR (95%CI)	P value	Bilirubin-adjusted sensitivity model OR (95%CI)	P value
Age, years	0.99 (0.94-1.05)	0.745	0.98 (0.93-1.04)	0.524
ASA class ≥ III	1.50 (0.31-7.22)	0.612	1.46 (0.30-7.09)	0.640
Preoperative biliary drainage	1.61 (0.55-4.73)	0.390	1.61 (0.52-4.98)	0.406
Preoperative CRP, log2(CRP + 1)	0.46 (0.33-0.64)	< 0.001	0.45 (0.32-0.63)	< 0.001
Preoperative albumin, g/L	1.04 (0.96-1.12)	0.383	-	-
Preoperative total bilirubin, mg/dL	-	-	0.94 (0.79-1.13)	0.518

ASA: American Society of Anesthesiologists; CRP: C-reactive protein; OR: Odds ratio; CI: Confidence interval.

To assess whether the CRP signal reflected biliary obstruction or drainage-related inflammation, CRP levels were compared according to preoperative biliary drainage status and correlated with total bilirubin. Preoperative CRP did not differ significantly according to biliary drainage status and showed no significant correlation with total bilirubin. In contrast, total bilirubin was significantly higher in patients with preoperative biliary drainage (P < 0.001), supporting the clinical relevance of drainage status as an obstruction-related variable.

In the bilirubin-adjusted sensitivity model, which included age, ASA class ≥ III, preoperative biliary drainage, preoperative total bilirubin, and log₂(CRP + 1)-transformed CRP, CRP remained independently associated with lower odds of TO achievement (OR = 0.45; 95%CI: 0.32-0.63; P < 0.001). The remaining covariates were not independently associated with TO achievement (Table 3). The AUC of the bilirubin-adjusted sensitivity model was 0.820.

The CRP × biliary drainage interaction term was not statistically significant (OR = 1.16; 95%CI: 0.58-2.31; P = 0.669), indicating no evidence that the association between CRP and TO differed according to biliary drainage status. In the ROC analysis of preoperative CRP alone for predicting non-TO, the AUC was 0.815 (95%CI: 0.722-0.896). The optimal cutoff value was 13.0 mg/L, yielding 66.0% sensitivity and 90.6% specificity. These CRP-alone ROC results were distinct from the AUC values of the multivariable models (Figure 3).

Figure 3 Receiver operating characteristic curve. A: Receiver operating characteristic curve of preoperative C-reactive protein alone for predicting non-textbook outcome. The area under the curve was 0.815 (95% confidence interval: 0.722-0.896). The Youden-derived cut-off was C-reactive protein ≥ 13.0 mg/L, with 66.0% sensitivity and 90.6% specificity. The diagonal dashed line represents the reference line; B: Receiver operating characteristic curves of the multivariable models for textbook outcome achievement. The area under the curve was 0.819 for the primary preoperative model and 0.820 for the bilirubin-adjusted sensitivity model. The diagonal reference line is shown for comparison. AUC: Area under the curve; CI: Confidence interval; ROC: Receiver operating characteristic; TO: Textbook outcome; CRP: C-reactive protein.

DISCUSSION

This study evaluated the TO rate and clinical factors associated with TO failure in a single-center series of patients who underwent PD for pancreatic head adenocarcinoma. Preoperative CRP level was independently associated with TO achievement. This finding suggests that a readily accessible inflammatory biomarker may serve as a risk-stratification marker in short-term quality assessment following PD.

The TO rate of 53% observed in our series is comparable to the 58.3% reported by van Roessel et al[6] in Dutch national audit data for patients undergoing PD and the 58.3% reported by Nicholas et al[8] from New Zealand. Using a harmonized ideal outcome definition incorporating a different component set, Augustinus et al[10] reported a rate of 54% following PD. Higher TO rates have been reported in multicenter laparoscopic PD series[7]. This discrepancy may be attributable to differences in surgical experience, center volume, and case selection criteria. The alignment of our series with these international benchmarks suggests that standardized TO assessment may provide a useful framework for evaluating short-term surgical quality in single-center settings. However, differences in center volume, case mix, and perioperative resources should be considered. The significantly shorter length of hospital stay in the TO group further supports the concept that TO reflects not merely the absence of complications but also a more rapid postoperative recovery.

Component-level analysis of the non-TO group revealed that severe complications were the predominant reason for failure. van Roessel et al[6] reported a severe complication rate of 30.4% in their cohort. The severe complication rate observed in our series is comparable to that of the van Roessel cohort, indicating that the burden of major complications constitutes a substantial component of the complication profile of PD. Similarly, Wu et al[7] reported that Clavien-Dindo ≥ III was one of the leading causes of TO failure in their multicenter laparoscopic PD series. The readmission rate represented the second most frequent reason for TO failure. Although it was slightly higher than the 16.9% reported in the van Roessel PD cohort, it remained broadly comparable[6]. This observation suggests that readmission emerges as a recurring barrier to TO across different series. Notably, half of the non-TO patients failed to meet only a single criterion, and in approximately half of these cases, the isolated cause was severe complications. This finding suggests that early recognition and management of major complications may be a relevant target for future quality-improvement strategies. However, whether such interventions improve TO rates requires prospective evaluation.

The relatively high in-hospital mortality observed in this cohort should be interpreted as a major contributor to TO failure and as a marker of early postoperative vulnerability. The available records did not suggest a predominance of pancreatic fistula-related mortality; rather, hemorrhagic events and systemic cardiopulmonary complications were prominent among patients who died. This pattern suggests that TO failure in our cohort may reflect broader perioperative deterioration rather than a single procedure-specific complication pathway. Both baseline patient risk and major early postoperative complications may have contributed to this process. However, because granular data on operative complexity, blood loss, fistula-risk profile, and postoperative rescue processes were limited, the mechanisms underlying this mortality burden could not be fully characterized.

The most consistent association in this cohort was observed between higher preoperative CRP and lower odds of achieving TO. Each doubling of preoperative CRP was associated with an approximately 54% decrease in the odds of TO achievement in the preoperative-only model, and this estimate remained virtually unchanged in the bilirubin-adjusted sensitivity model. In parallel, preoperative CRP alone showed clinically relevant discrimination for non-TO in ROC analysis. Taken together, these findings suggest that CRP represents a consistent preoperative risk signal associated with failure to achieve TO. Importantly, the CRP estimate remained stable after adjustment for biliary drainage and total bilirubin. In addition, the CRP × biliary drainage interaction was not significant. These findings suggest that the association was not explained solely by biliary obstruction or drainage status. This finding is consistent with previous studies reporting associations between elevated preoperative CRP or CRP-based inflammatory markers and adverse outcomes after PD[23,24].

The biological interpretation of this association should remain cautious. This association may be biologically plausible through pathways related to endothelial dysfunction, impaired tissue repair, and altered anastomotic healing[25,26]. However, these mechanisms were not directly evaluated in the present cohort and should be considered hypothesis-generating rather than explanatory. Because detailed data on drainage duration, objective cholangitis criteria, microbiological findings, and serial inflammatory markers were unavailable, the predominant source of CRP elevation could not be determined. In routine practice, elevated preoperative CRP may help identify patients who require closer perioperative assessment for occult infection, biliary drainage-related inflammation, nutritional compromise, or increased postoperative monitoring needs. However, the present data do not establish whether delaying surgery or applying specific CRP-guided interventions improves TO rates. Therefore, the role of CRP in guiding perioperative risk assessment or decision-making requires prospective validation.

Intraoperative red blood cell transfusion was more frequent in the non-TO group in the unadjusted comparison. Because this variable represents an intraoperative process rather than a baseline patient characteristic, it was not included in the preoperative multivariable risk-stratification model. Increased intraoperative blood loss has been identified as one of the principal risk factors for non-TO in previous literature[13]. Given that direct intraoperative blood loss volume was not recorded, intraoperative transfusion should be interpreted as an indirect marker of operative complexity or bleeding burden rather than as an independent predictor of TO failure in the present analysis. Although preoperative albumin tended to be lower in the non-TO group, it was not independently associated with TO achievement in the preoperative multivariable model. Larger studies may clarify whether albumin adds prognostic information beyond CRP in preoperative risk stratification after PD.

Preoperative biliary drainage was not associated with TO achievement in our cohort, either in the unadjusted comparison or in the multivariable models. This finding differs from the meta-analysis by Yuan et al[13], in which preoperative biliary drainage was identified as a risk factor for non-TO. This discrepancy may be related to unmeasured differences in drainage duration, cholangitis-related inflammation, and institutional drainage practices, which could not be evaluated in the present dataset. The persistence of the CRP association in the bilirubin-adjusted model supports the view that this signal was not explained solely by biliary drainage or obstruction-related variables.

This study has several limitations. The retrospective and single-center design limits causal inference and restricts the generalizability of the results. The multivariable models were intentionally restricted to a small number of prespecified clinically relevant variables to reduce overfitting. Nevertheless, the sample size remained limited and may have been insufficient to detect independent associations for variables with moderate effect sizes. Important perioperative and gland-related variables, including blood loss, pancreatic texture, duct diameter, operative time, surgeon or center volume, and formal fistula risk scores, were unavailable in this retrospective dataset. Intraoperative transfusion was included as an indirect marker of bleeding burden and operative complexity, but cannot replace direct blood loss measurement. Therefore, residual confounding by unmeasured operative complexity and fistula risk cannot be excluded. No patient underwent vascular resection or reconstruction.

CONCLUSION

This study presents a standardized TO-based quality assessment in a PD cohort from a tertiary center in Türkiye. The TO rate of 53% was broadly consistent with international benchmarking cohorts. Higher preoperative CRP was independently associated with lower odds of achieving TO and showed clinically relevant discrimination for non-TO. These findings suggest that CRP may represent a readily available preoperative risk-stratification marker. However, its role in guiding perioperative decision-making or targeted optimization strategies requires prospective multicenter validation.

ACKNOWLEDGEMENTS

The authors would like to thank all members of the Department of General Surgery and the Department of Surgical Oncology, Faculty of Medicine, Çukurova University, for their valuable contributions to the perioperative management and postoperative follow-up of the patients included in this study. The authors also acknowledge the staff of the medical records and archive units of Çukurova University Faculty of Medicine for their assistance during data retrieval.