ABSTRACT
Background
Pancreatic cancer (PC) disrupts metabolic and nutritional processes, including glucose homeostasis and albumin levels. Both hyperglycemia and hypoalbuminemia have been linked to poorer outcomes in various cancers, including PC. This study investigates the preoperative glucose‐to‐albumin ratio (GAR) as a potential prognostic marker of overall survival in patients undergoing pancreatic resection for pancreatic ductal adenocarcinoma (PDAC).
Methods
This single‐center retrospective analysis included patients who underwent curative‐intent pancreatectomy for PDAC and adenosquamous carcinoma of the pancreas between 2017 and 2024. GAR was calculated from preoperative laboratory tests. Kaplan–Meier and Cox regression analyses were performed for recurrence‐free and overall survival.
Results
A total of 566 patients were included (51% males and 49% females, mean age 68.3); 548 had PDAC and 18 had adenosquamous carcinoma of the pancreas. High GAR was associated with elevated CA 19–9, CEA and BMI (p < 0.0001, p < 0.0001, p = 0.02, respectively), more advanced T staging (p = 0.001), increased tumor size (p = 0.007), and AJCC staging (p = 0.04), as well as perineural invasion (p = 0.01) and lymphovascular invasion (p = 0.008). Kaplan–Meier analysis demonstrated significantly worse overall survival in high GAR patients (26.6 vs. 35.0 months, p = 0.004), while low albumin was associated with significantly worse recurrence‐free survival (15.5 vs. 20.7 months, p = 0.04). Cox regression identified high GAR (HR = 1.3, p = 0.04), lower BMI (Body Mass Index) (p = 0.02), higher CA 19–9 (p = 0.006), N staging (p = 0.002), histological tumor grading (p < 0.0001), neoadjuvant chemotherapy (p = 0.002) and perineural invasion (p = 0.03) as independent risk factors for poorer overall survival.
Conclusions
Higher preoperative GAR is associated with more advanced disease at presentation and an adverse prognosis in patients with resected PC.
Keywords: glucose‐to‐albumin ratio, pancreatectomy, pancreatic cancer, prognostic marker, survival analysis
1. Core Tip
The preoperative glucose‐to‐albumin ratio (GAR) is a simple and readily available prognostic marker for patients undergoing resection for pancreatic ductal adenocarcinoma (PDAC). In this cohort of 566 patients, elevated GAR correlated with more advanced disease at presentation, more aggressive disease features, and significantly worse overall survival post‐surgery (26.6 vs. 35.0 months). In multivariate analysis, high GAR independently predicted worse overall survival. GAR integrates two metabolic alterations of PDAC—hyperglycemia and hypoalbuminemia—into a single variable that may inform preoperative risk stratification and clinical decision‐making.
2. Introduction
Pancreatic ductal adenocarcinoma (PDAC) accounts for 90% of cases of pancreatic cancer (PC) [1]. PC has a dismal 5‐year survival rate of 13% overall, dropping to 3% for patients with advanced metastatic disease [2], and is predicted to become the second leading cause of cancer deaths by 2030 [3]. Contributing to this unfavorable prognosis is the fact that about 80% of PC cases are diagnosed with advanced‐stage disease, leaving only 20% of PC patients eligible for curative‐intent surgical resection [4]. Therefore, there is a clear need for reliable prognostic markers to guide clinical decision‐making.
PDAC is associated with endocrine and metabolic perturbations, as well as nutritional deficiencies, that contribute to morbidity and worsen prognosis. Hypoalbuminemia [5], frequently observed in PDAC, mainly arises from systemic inflammation [6] and malnutrition [7], in the context of GI tract cancers. Hypoalbuminemia predicts poor survival and treatment outcomes in GI malignancies, as well as in lung, ovarian, and breast cancers [8]. Further, in patients undergoing pancreaticoduodenectomy, preoperative albumin deficiency increases the risk of postoperative complications [9, 10] and recovery time [11]. Moreover, in PC patients, hypoalbuminemia was associated with earlier mortality [12]. In accordance, a separate study of resected PDAC patients found that postoperative serum albumin level ≥ 3.9 g/dL at 12 months was an independent predictor of improved disease‐free and overall survival [13].
Beyond hypoalbuminemia, hyperglycemia is another metabolic finding frequently encountered in patients with newly diagnosed PDAC [14, 15]. Indeed, diabetes occurs in 47% of PC patients, with most of these diagnoses made in the second‐year period prior to identifying the malignancy [16]. Persistent hyperglycemia, reflecting chronic inflammation, is associated with increased adverse events, poorer outcomes, and shorter survival in many conditions, including solid malignancies [17, 18]. Additionally, preoperative hyperglycemia is associated with negative surgical outcomes [19], including in patients without a diabetes diagnosis [20]. In the setting of PC, preoperative hyperglycemia is associated with more frequent complications and reduced overall survival [21, 22].
Glucose‐to‐albumin ratio (GAR) has emerged as a novel composite marker factoring both hyperglycemia and hypoalbuminemia. Therefore, GAR may hold additional information regarding the nutritional and metabolic reserves by capturing poor nutritional status and metabolic dysregulation, respectively. In a study of cancer patients undergoing treatment with anthracyclines, increased GAR was associated with increased risk of all‐cause mortality and cardiovascular events [23]. In addition, GAR has been associated with higher risk of kidney injury in population‐based data [24], as well as higher risk of incident diabetes, cardiovascular disease, and overall mortality [25]. Also, GAR is a predictor of intracerebral hemorrhage‐specific mortality [26], complications following hip fracture surgery [27, 28, 29], and adverse outcomes in patients undergoing percutaneous coronary intervention [30]. In this study, we aim to investigate the impact of preoperative GAR on cancer survival in a large cohort of patients undergoing curative‐intent pancreatic resection for PDAC.
3. Methods
3.1. Cohort Selection
This was a retrospective analysis of patients who underwent curative‐intent distal pancreatectomy or pancreaticoduodenectomy for resectable pancreatic cancer between 2017 and 2024 at the Jefferson Pancreas, Biliary, and Related Cancer Center. Patients with histologic diagnoses of PDAC and adenosquamous carcinoma were included in this study. Patients found to have a clinical or pathological staging of M1 at surgery and patients with perioperative mortality (< 2 months) were excluded from the analysis. Overall, 1215 patient charts were initially screened, and a total of 566 patients with resectable PDAC were included in the final analysis (Figure 1).
FIGURE 1.
Cohort Selection.
3.2. Data Collection
All data were obtained from a prospectively maintained database from the institution's electronic medical records. Preoperative laboratory values were included: Glucose (mg/dL), albumin (g/dL), CA 19–9 (U/mL), and CEA (ng/mL). Serum levels reported were the last measurement available before surgery, most often obtained as preoperative values in the week prior to surgery. Glucose‐to‐Albumin Ratio (GAR = Glucose (mg/dL)/Albumin g/dL) was computed using those last available preoperative blood glucose and serum albumin levels. Fasting status at preoperative GAR measurement was not confirmed for all patients. Other data collected include demographics, BMI (kg/m2), use of neoadjuvant or adjuvant chemotherapy, and the pathology specimen descriptors of tumor histological grade, size, T/N stage, AJCC stage, presence of lymphovascular and perineural invasion, as well as patient survival and recurrence. Survival data were calculated from the date of surgical resection.
3.3. Statistical Analysis
Statistical analyses were performed using SPSS (Version 29.0.2.0 (20), IBM Corp.). Pearson's correlation coefficient with two‐tailed significance was used for normally distributed continuous variables and binomial variables. Continuous variables were assessed for normality using histograms and Q–Q plots. CA19‐9 and CEA demonstrated right‐skewed distributions and were therefore log‐transformed prior to parametric analyses; log10(value +1) was used to accommodate zero values. All parametric tests used the log‐transformed values of these variables, while other continuous variables were analyzed on their original scale. Spearman's coefficient was used for the correlations that involved ordinal variables. Kaplan–Meier survival curves, with group comparisons, were computed using the log‐rank test. Cox regression analyses were performed for overall survival and recurrence‐free survival. For Cox regressions, a backward elimination approach was used to identify the strongest predictive model. All candidate covariates (age at surgery, neoadjuvant chemotherapy, histological grade, tumor T and N staging, tumor size, lymphovascular invasion, perineural invasion, CA 19–9, CEA, GAR above median, BMI, surgery type) were entered into the initial Cox regression model. Backward elimination was then performed by sequentially removing covariates and comparing model fit using the −2 log‐likelihood (−2LL) statistic. Elimination continued until removal of any additional variable resulted in an increase in –2LL. Potential multicollinearity among candidate covariates was ruled out using collinearity diagnostics, including condition indices and variance proportions. Patients with missing values of any covariate were temporarily excluded from optimization steps to ensure model consistency and, once the final model was selected, the model was refit using the full eligible cohort, with exclusion limited to patients with missing values for variables retained in the final model. Histological grade, nodal staging, and perineural invasion status were missing in three patients; BMI was missing in six patients. Time‐dependent Receiver‐Operator Characteristics (ROC) Curve analysis was performed in R (R version 4.2.0) using the “timeROC” and “survivalROC” packages. Statistical significance was set at a p‐value < 0.05. p‐values were derived from Chi‐square analysis for comparisons of categorical variables and from Student's t‐tests for comparisons of continuous variables between GAR‐high (above dataset median) and GAR‐low (below dataset median) groups. Minimum p‐value analysis was performed by evaluating 19 different GAR cutoffs ranging from the 5th to the 95th percentile.
4. Results
4.1. Patient Cohort Characteristics
A total of 566 patients were included in the final cohort (Figure 1) from an initial cohort of 1215 patients who received a curative‐intent pancreatectomy or pancreaticoduodenectomy at TJUH from 2017 to 2024. Approximately 51% of the cohort were male (Table 1); the mean age was 68.3 (range: 31.0–94.3) years, and the median pre‐operative BMI was 26.5 (range: 15.0–46.9) kg/m2. In the cohort, 548 (96.8%) patients were diagnosed with PDAC, and 18 (3.2%) of the patients were diagnosed with adenosquamous carcinoma. Of the 566 patients, 446 (78.8%) underwent Whipple resection, while 120 (21.2%) underwent a distal pancreatectomy. A total of 174 (30.7%) patients received neoadjuvant chemotherapy, while 414 (73.1%) patients received adjuvant chemotherapy. The average resected tumor size was 2.89 ± 1.26 cm, with perineural invasion found in 86.4% and lymphovascular invasion in 48.2% of the patients. The median preoperative values of CA 19–9 and CEA were 45.0 (IQR range: 17.0–169.0) U/mL and 2.8 (IQR range: 1.9–4.6) ng/mL, respectively. Overall survival data were available for all patients, while disease recurrence data were available for 95.1% of the cohort.
TABLE 1.
Patient and tumor characteristics (n = 566) in patients with low and high glucose‐to‐albumin ratio (GAR). Normally distributed numeric data are presented as mean ± standard deviation, while other numeric data are presented as median (interquartile range).
| Characteristic | Overall (n = 566) | GAR high (n = 283) | GAR low (n = 283) | p (GAR high vs. GAR low) |
|---|---|---|---|---|
| Age, years | 68.3 (61.9–75.3) | 69.0 (62.3–75.9) | 67.5 (60.9–75.0) | 0.07 |
| Sex | 0.23 | |||
| Male | 288 (50.9%) | 151 (53.4%) | 137 (48.4%) | |
| Female | 278 (49.1%) | 132 (46.6%) | 146 (51.6%) | |
| BMI, kg/m2 | 26.5 (23.0–29.2) | 26.9 (23.3–29.4) | 26.1 (22.7–28.9) | 0.052 |
| Race | 0.47 | |||
| White/Caucasian | 452 (79.9%) | 229 (80.9%) | 223 (78.8%) | |
| Black/African American | 62 (11.0%) | 26 (9.2%) | 36 (12.7%) | |
| Asian | 24 (4.2%) | 11 (3.9%) | 13 (4.6%) | |
| Hispanic/Latino | 24 (4.2%) | 14 (4.9%) | 10 (3.5%) | |
| Other/Unknown | 4 (0.7%) | 3 (1.1%) | 1 (0.4%) | |
| Diabetes mellitus | 167 (29.7%) | 121 (43.4%) | 46 (16.3%) | |
| Missing | 4 | 4 | 0 | |
| Surgery type | 0.04 | |||
| Whipple pancreatectomy | 446 (78.8%) | 233 (82.3%) | 213 (75.3%) | |
| Distal pancreatectomy | 120 (21.2%) | 50 (17.7%) | 70 (24.7%) | |
| Adjuvant chemotherapy | 414 (73.1%) | 208 (36.7%) | 206 (36.4%) | 0.85 |
| Neoadjuvant chemotherapy | 174 (30.7%) | 68 (12.0%) | 106 (18.7%) | < 0.001 |
| Tumor size, cm | 2.89 ± 1.26 | 3.03 ± 1.29 | 2.74 ± 1.21 | 0.007 |
| Perineural invasion | 489 (86.4%) | 257 (45.4%) | 232 (41.0%) | 0.002 |
| Lymphovascular invasion | 273 (48.2%) | 156 (27.6%) | 117 (20.7%) | 0.001 |
| CA 19–9, U/mL | 45.0 (17.0–169) | 63.5 (21.3–271.3) | 36.0 (14.0–120.0) | < 0.001 |
| CEA, ng/mL | 2.8 (1.9–4.6) | 3.0 (2.0–4.9) | 2.7 (1.7–4.4) | 0.006 |
| Histological grade (G) | 0.008 | |||
| G1 – well differentiated | 60 (10.6%) | 21 (7.4%) | 39 (13.8%) | |
| G2 – moderately differentiated | 346 (61.1%) | 181 (64.0%) | 165 (58.3%) | |
| G3 – poorly differentiated | 126 (22.3%) | 70 (24.7%) | 56 (19.8%) | |
| X/not reported | 34 (6.0%) | 11 (3.9%) | 23 (8.1%) | |
| Path. tumor T staging | 0.01 | |||
| 1 | 126 (22.3%) | 48 (17.0%) | 78 (27.6%) | |
| 2 | 289 (51.1%) | 146 (51.6%) | 143 (50.5%) | |
| 3 | 137 (24.2%) | 82 (29.0%) | 55 (19.4%) | |
| 4 | 8 (1.4%) | 4 (1.4%) | 4 (1.4%) | |
| X | 6 (1.1%) | 3 (1.1%) | 3 (1.1%) | |
| Path. tumor N staging | 0.038 | |||
| 0 | 194 (34.3%) | 97 (34.3%) | 97 (34.3%) | |
| 1 | 226 (39.9%) | 100 (35.3%) | 126 (44.5%) | |
| 2 | 143 (25.3%) | 85 (30.0%) | 58 (20.5%) | |
| X | 3 (0.5%) | 1 (0.4%) | 2 (0.7%) |
Note: Bolded values indicate statistical significance.
Abbreviations: BMI, body mass index; GAR, glucose‐to‐albumin ratio; IQR, interquartile range; PDAC, pancreatic ductal adenocarcinoma; SD, standard deviation.
4.2. Association of GAR With Tumor Characteristics and Pathological Findings
Significant differences were observed between the GAR‐high (above the dataset median of 26.44) and GAR‐low groups (≤ 26.44) (Table 1). The cutoff of 26.44 was selected based on minimum p‐value analysis (Figure S1). The mean tumor size was significantly larger in the GAR‐high group compared to the GAR‐low group (3.03 cm vs. 2.74 cm, p‐value = 0.007). Proportionally more people in the GAR‐low group had received neoadjuvant chemotherapy (18.7% vs. 12.0%, p < 0.001) and underwent a distal pancreatectomy as opposed to the Whipple procedure (24.7% vs. 17.7%, p = 0.04). Additionally, lymphovascular and perineural invasion were more frequent in the GAR‐high group (27.6% and 45.4%, respectively) compared to the GAR‐low group (20.7% and 41.0%, respectively), p = 0.001 and p = 0.002, respectively. The median levels of CA 19–9 and CEA were significantly higher in the GAR‐high group (63.5 U/mL and 3.0 ng/mL, respectively) compared to the GAR‐low group (36.0 and 2.7, respectively), p < 0.001 and p = 0.006, respectively. Histological grade was strongly associated with GAR classification (p = 0.008), with the GAR‐high group having a higher proportion of poorly differentiated tumors (24.7%) compared to the GAR‐low group (19.8%), while the GAR‐low group had a higher frequency of well‐differentiated tumors (13.8% vs. 7.4%). A similar pattern was observed for T staging (p = 0.01), where the GAR‐high group had significantly higher proportions of advanced T stage (T3–T4: 30.4%) tumors, while the GAR‐low group had higher proportions of lower‐stage tumors (T1–T2: 78.1%). Association between N stage and GAR was indicative of higher lymph node involvement in the GAR‐high group (N2: 30.0%, p = 0.04). No significant differences were observed between the GAR‐high and GAR‐low groups for age, sex, BMI, race, and adjuvant chemotherapy.
4.3. Metabolic Markers Correlate With Aggressive PDAC Features
Bivariate correlation analysis (Table 2) revealed that low serum albumin levels correlated significantly with higher CA 19–9 levels (r = −0.104, p = 0.01), higher CEA levels (r = −0.157, p = 0.0002), more advanced T (r = −0.086, p = 0.04) and nodal staging (r = −0.112, p = 0.008), higher AJCC stage (r = −0.120, p = 0.004), and age at surgery (r = −0.119, p = 0.005). Low serum albumin levels were also associated with higher rates of perineural invasion (r = −0.120, p = 0.004), lymphovascular invasion (r = −0.095, p = 0.02), and having undergone a Whipple procedure (r = −0.223, p < 0.0001). Glucose levels were positively correlated with higher CA 19–9 levels (r = 0.169, p < 0.0001), higher CEA levels (r = 0.140, p = 0.0008), T staging (r = 0.111, p = 0.08), tumor size (r = 0.125, p = 0.003), and higher rate of perineural invasion (r = 0.085, p = 0.044), suggesting that higher glucose may be associated with larger and more advanced tumors. Also, GAR was significantly associated with higher CA 19–9 levels (r = 0.189, p < 0.0001), higher CEA levels (r = 0.177, p < 0.0001), higher BMI (r = 0.097, p = 0.02), more advanced T stage (r = 0.137, p = 0.001), higher AJCC stage (r = 0.089, p = 0.04), larger tumor size (r = 0.113, p = 0.007), and perineural invasion (r = 0.16, p = 0.01) and lymphovascular (r = 0.112, p = 0.008) invasion. Other notable associations included correlation between overall survival with age at surgery (r = −0.091, p = 0.03), tumor size (r = −0.093, p = 0.03), CA 19–9 (r = −0.138, p = 0.001), BMI (r = 0.090, p = 0.03), T stage (r = −0.099, p = 0.02), N stage (r = −0.209, p < 0.0001), AJCC stage (r = −0.205, p < 0.0001), and tumor histological grade (r = −0.147, p = 0.0007). In addition, recurrence‐free survival correlated with tumor size (r = −0.098, p = 0.03), CA 19–9 (r = −0.155, p = 0.0003), BMI (r = 0.097, p = 0.03), T stage (r = −0.110, p = 0.01), N stage (r = −0.202, p < 0.0001), AJCC stage (r = −0.211, p < 0.0001), and tumor histological grade (r = −0.149, p = 0.007).
TABLE 2.
Bivariate correlations between metabolic markers and clinical/pathological characteristics.
| CA 19–9 (U/mL) | CEA (ng/mL) | T | N | AJCC Stage | Tumor grade | Age at surgery | Tumor size | BMI (kg/m2) | |
|---|---|---|---|---|---|---|---|---|---|
| Albumin | |||||||||
| Correlation a | −0.104 | −0.157 | −0.086 | −0.112 | −0.12 | −0.025 | −0.119 | −0.029 | −0.069 |
| Sig. | 0.013 | 0.0002 | 0.04 | 0.008 | 0.004 | 0.569 | 0.005 | 0.493 | 0.104 |
| Glucose | |||||||||
| Correlation a | 0.169 | 0.14 | 0.111 | 0.028 | 0.066 | 0.017 | 0.05 | 0.125 | 0.08 |
| Sig. | < 0.0001 | 0.0008 | 0.008 | 0.501 | 0.116 | 0.699 | 0.231 | 0.003 | 0.058 |
| GAR | |||||||||
| Correlation a | 0.189 | 0.177 | 0.137 | 0.054 | 0.089 | 0.051 | 0.081 | 0.113 | 0.097 |
| Sig. | < 0.0001 | < 0.0001 | 0.001 | 0.202 | 0.035 | 0.241 | 0.055 | 0.007 | 0.022 |
Spearman's rank correlation coefficient was used for ordinal variables (T, N, overall staging, tumor grade).
4.4. Body Mass Index Correlates Positively With Recurrence‐Free Survival
The median overall survival since surgery in the cohort was 23.6 (IQR: 14.2–35.7) months. Kaplan–Meier survival analysis was suggestive of an association between high BMI (upper tertile, above 27.8 kg/m2) and improved overall survival (32.3 vs. 26.6 months, median, p = 0.058) compared to patients at the bottom tertile BMI (≤ 23.9 kg/m2). Lower tertile BMI was significantly associated with worse recurrence‐free survival compared to middle and upper tertiles (14.1 vs. 17.8 vs. 22.7 months, median, p = 0.011, p = 0.015), as shown in Figure S2.
4.5. High GAR and Glucose Are Associated With Worse Overall Survival, While Lower Albumin Is Associated With Worse Recurrence‐Free Survival
Kaplan–Meier survival analysis revealed significantly worse overall survival in patients with high, compared to low, GAR (26.6 vs. 35.0 months, median, p = 0.004) (Figure 2). This finding remained consistent in a sensitivity analysis restricted to patients with PDAC, after exclusion of adenosquamous carcinoma patients (27.8 vs. 35.0 months, median, p = 0.004). Subgroup analysis by diabetes mellitus (DM), defined as preoperative HbA1c ≥ 6.5%, showed that the association between high GAR and worse overall survival remained significant in patients without DM (p = 0.014), but not in those with DM (p = 0.096), although the smaller subgroup sample sizes may have limited statistical power (Figure S3). Categorizing patients by glucose levels above or below the dataset median of 110 mg/dL also revealed worse overall survival in patients with high glucose (29.3 vs. 32.9 months, median, p = 0.045) (Figure S4). In contrast, categorizing patients by albumin levels (above or below the median of 4.2 g/dL) revealed no significant differences in overall survival (Figure S4). Neither GAR nor preoperative glucose levels were associated with recurrence‐free survival by Kaplan–Meier analysis (Figure S5). However, low albumin levels (≤ 4.2 g/dL) were significantly associated with worse recurrence‐free survival, with a median survival of 15.5 months for patients with low albumin, compared to 20.7 months for those with high albumin (p = 0.04, Figure 3). Cox regression analysis of overall survival was performed by analyzing the following factors: Age at surgery, neoadjuvant chemotherapy, histological grade, tumor T and N staging, tumor size, lymphovascular invasion, perineural invasion, CA 19–9, CEA, GAR above median, BMI, surgery type. After removal of non‐significant, non‐independent variables, the optimal model reached included: High GAR (HR = 1.295, p = 0.03), low BMI (HR = 0.972, p = 0.02), high CA 19–9 (HR = 1.236, p = 0.006), N staging (HR = 1.270, p = 0.002), neoadjuvant chemotherapy (HR = 1.513, p = 0.002), and histological grade (HR = 1.215, p < 0.0001) as independent risk factors for worse overall survival (Table 3). To evaluate whether GAR provides greater prognostic information than its individual components, we performed separate univariable and multivariable Cox regression analyses of overall survival for GAR, glucose, and albumin, modeled both as continuous and binomial (median‐based) variables (Table S1). Comparative Cox regression analyses demonstrated that median‐based GAR provided stronger prognostic information than glucose or albumin alone, with high GAR independently associated with worse overall survival and showing the best model fit in both univariable and multivariable analyses. To further assess whether the prognostic value of GAR was independent of tumor burden, stratified Cox regression analyses were performed according to AJCC staging. The association between high GAR and overall survival remained statistically significant after AJCC‐stage stratification (HR 1.328, p = 0.018), supporting an association between GAR and survival regardless of overall disease stage. A secondary analysis adjusting for preoperative total bilirubin and biliary drainage status did not alter the association between GAR and overall survival (results not shown).
FIGURE 2.
Overall Survival Functions of High vs. Low GAR. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows worse overall survival in patients with high glucose‐albumin ratio (GAR) compared with patients with low GAR (26.6 vs. 35.0 months, p = 0.004).
FIGURE 3.
Recurrence‐Free Survival Functions of High vs. Low Albumin. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows worse recurrence‐free survival in patients with below‐median albumin compared with patients with above‐median albumin (15.5 vs. 20.7 months, p = 0.04).
TABLE 3.
Cox regression analysis for overall survival.
| Factor | HR | p |
|---|---|---|
| Neoadjuvant chemotherapy | 1.513 (1.168–2.007) | 0.002 |
| N staging | 1.270 (1.091–1.477) | 0.002 |
| Perineural invasion | 1.625 (1.052–2.509) | 0.028 |
| CA 19–9 | 1.236 (1.062–1.438) | 0.006 |
| Histological grade | 1.215 (1.005–1.468) | < 0.0001 |
| BMI (kg/m2) | 0.972 (0.950–0.995) | 0.017 |
| GAR above median | 1.295 (1.029–1.631) | 0.028 |
4.6. AUC Analysis of GAR Thresholds in Survival Outcomes
Time‐dependent AUC analysis was performed for GAR using inverse probability of censoring weights computed from Kaplan–Meier estimators. For overall survival, the time points between 5 months and 60 months showed GAR to carry a favorable AUC (> 0.5), with the best time point at 50 months (AUC 59.9, 95% CI 52.6–67.3, Figure S6A) with an optimal cutoff of GAR = 29.79 as shown in the ROC curve in Figure S6B. For disease‐free survival, the time points between 5 months and 60 months showed GAR to carry a favorable AUC (> 0.5), with the best time point at 25 months (AUC 57.6, 95% CI 51.8–63.3, Figure S6C) with an optimal cutoff of GAR = 23.86 as shown in the ROC curve in Figure S6D. Minimum p‐value analysis identified the median GAR value (26.44) as the cutoff corresponding to the lowest p‐value (Figure S1).
5. Discussion
Pancreatic cancer is an aggressive disease that carries a poor prognosis, and its incidence continues to rise across the globe. The American Cancer Society estimates that pancreatic cancer will kill over 50,000 people in the United States in 2025, with almost 70,000 new cases. Most of these cases and the related mortality are attributable to PDAC, the most common histologic subtype [1]. Given the aggressive nature of PDAC, identifying reliable prognostic markers is crucial for predicting outcomes and potentially improving survival through modifiable risk factors.
The present study found that elevated preoperative glucose levels were significantly associated with larger tumor size, worse tumor T and N stage, perineural invasion, and higher CA 19–9 and CEA levels, while low serum albumin levels correlated with worse T stage and nodal status (and overall staging), presence of perineural and lymphovascular invasion, and high preoperative CA 19–9 and CEA levels. These findings suggest that both hyperglycemia and hypoalbuminemia reflect disease burden and aggressiveness in PDAC. The use of preoperative glucose‐to‐albumin ratio (GAR), a composite metric that combines both variables, was associated with tumor T and N stage and size, higher CA 19–9 and CEA, presence of perineural and lymphovascular invasion, and AJCC stage. Importantly, we found that a high preoperative GAR was associated with worse overall survival, supporting its role as a prognostic marker. Furthermore, GAR maintained its prognostic value even after correcting for BMI, neoadjuvant treatment, and other oncological and histological risk factors. Supporting these observations, high GAR was found to be a univariate predictor of survival, as well as an independent risk factor for all‐cause mortality in cancer patients receiving anthracycline‐based chemotherapy [23]. Our regression model has identified neoadjuvant chemotherapy as associated with worse overall survival, which merits some discussion. We believe that this finding should not be interpreted as a detrimental impact of the treatment, but rather as a surrogate marker of baseline disease burden, as it is preferentially offered to patients with a more advanced (borderline or locally advanced) disease.
While some molecular and serum markers are already in clinical use for risk stratification of PDAC, CA 19–9, despite its limitations, remains the only PDAC marker with FDA approval. Emerging research indicates the superior role of multiple combined markers in PDAC prognostication, such as CA 19–9 in combination with TIMP‐1, CA125, CEA, and other transcriptomic and proteomic biomarkers [31]. Additionally, broader indicators of systemic health, such as nutritional status, metabolic balance, and inflammation, are increasingly being explored [32, 33, 34, 35]. Serum albumin, a common marker of nutritional status, predicts surgical risk, burden of disease, and survival in various conditions [36, 37]. For instance, a high pretreatment C‐reactive protein‐to‐albumin ratio is predictive of poor survival in pancreatic cancer [38], while an elevated alkaline phosphatase‐to‐albumin ratio was an independent prognostic factor of worse overall survival in resectable PDAC [39]. GAR, therefore, seems to offer a new prognostic tool in our assessment of pancreatic cancer survival.
GAR merges the prognostic properties of two critical factors that influence prognosis in PDAC: Glucose and albumin, which reflect metabolic and nutritional status, respectively. Hyperglycemia, defined as elevated blood glucose levels, has been identified as a prognostic factor in PDAC. High preoperative blood glucose levels are associated with reduced overall survival and increased postoperative complications in patients undergoing resection for pancreatic cancer [21], and hyperglycemia also predicted worse overall survival in patients with advanced PDAC [40]. Preexisting chronic hyperglycemia is a risk factor for the development of pancreatic cancer [15], and it contributes to tumor progression [41] and resistance to therapy [42]. Elevated glucose promotes oncogenic pathways such as protein kinase C (PKC) and mitogen‐activated protein kinase (MAPK) by generating reactive oxygen species, driving tumor growth, resistance to apoptosis, cell proliferation, and perineural invasion [43]. In addition, oxidative stress contributes to metastasis by inducing epithelial‐to‐mesenchymal transition and vascular dysfunction [44, 45].
Beyond glucose, nutritional status plays an important role in the prognosis of cancer patients. Studies in PDAC have described an “obesity paradox”, where higher BMI correlated with improved survival, possibly reflecting better nutritional reserves [46, 47]. In that context, the present analysis revealed that higher BMI was significantly associated with longer overall and recurrence‐free survival. These findings align with the broader clinical context of PDAC‐associated cachexia and hypoalbuminemia. Hypoalbuminemia, a common finding in PDAC patients and a marker of poor prognosis in cancer [48], results from inflammation‐driven hepatic reprioritization of protein synthesis, increased vascular permeability, and catabolism [6, 49, 50, 51, 52, 53, 54, 55, 56], and decreased nutrient intake due to anorexia [57] and pancreatic exocrine insufficiency [58]. The role of albumin includes preserving oncotic pressure, substance transport, and countering inflammation and oxidative damage [5, 59]; thus, its loss is linked to peripheral edema, impaired chemotherapeutic drug transport, and inflammatory dysregulation that contributes to a tumor‐promoting microenvironment. Furthermore, hypoalbuminemia is featured in many prognostic indices that evaluate systemic inflammation and nutritional status, such as the Glasgow Prognostic Score (GPS) [60], the Prognostic Nutritional Index (PNI) [61], the CRP‐albumin‐lymphocyte (CALLY) Index [62], and the albumin‐to‐globulin ratio (AGR). In summary, both hyperglycemia and hypoalbuminemia are easily obtained and cost‐effective biomarkers that, when combined into GAR, associate with poor PDAC prognosis in a biologically plausible manner.
A key question is whether these metabolic derangements are merely prognostic markers or whether improving hyperglycemia and hypoalbuminemia, through intense perioperative glucose control or nutritional support, would impact long‐term survival. Given that the detrimental effects of hyperglycemia and hypoalbuminemia on cancer‐specific mortality likely result from prolonged metabolic and nutritional dysfunction, perioperative short‐term corrections will likely not alter long‐term postoperative survival outcomes. At the same time, while there is a biological mechanistic basis for how GAR could impact survival, GAR might simply reflect disease severity and systemic stress and therefore serve as a mere prognostic marker. Nevertheless, given the established importance of nutritional assessment and support in patients with pancreatic cancer, including during the perioperative and chemotherapy periods [63, 64], patients with concerning GAR values may warrant additional nutritional optimization. In future controlled studies, it would be valuable to assess whether preoperative optimization of GAR improves oncologic outcomes or whether patients with high GAR might benefit from modifications to their adjuvant therapy or oncologic surveillance. While our study focused on resectable PDAC, GAR's role in locally advanced and metastatic disease, as well as other types of pancreatic and biliary cancer, should be further explored. In addition, future work should also evaluate the dynamics of GAR over time and assess whether changes in this parameter are prognostic or predictive of treatment response.
This study carries several notable limitations. This was a retrospective analysis, which introduces the potential for selection and misclassification biases. The analysis was based on a large cohort of 566 patients with histologically confirmed PDAC and adenosquamous carcinoma, all operated on at the same institution. While this decreases the confounding within the cohort, it hinders the generalizability of the findings. Importantly, the GAR cutoff of 26.44 identified in this study should be considered as an initial, exploratory, and cohort‐specific metric rather than a clinically ready threshold. Our added assessment of the time‐dependent AUC of GAR as a survival marker revealed it to be quite robust for multiple cutoffs for overall survival and disease‐free survival. GAR's optimal utility, as expected, varied in timing between overall survival (50 months) and disease‐free survival (25 months). Validation of these results in multi‐institutional cohorts is needed. The data were derived from a prospectively maintained and cross‐validated database, supporting the reliability and consistency of the variables. Strict inclusion criteria ensured a homogeneous study population that only included patients with earlier‐stage, resectable pancreatic cancer; at the same time, this study excluded locally advanced and metastatic PDAC patients, thereby limiting the scope of use of GAR. We acknowledge that some patients might not have been fasting at the time of preoperative GAR measurement, and that non‐fasting or transient stress‐related elevations in blood glucose could have resulted in an overestimation of GAR. Overall, 29.7% of our cohort had pre‐existing diabetes (and HbA1C values > 6.4%), which corresponds well with the fact that 30.0% of the pre‐operative blood glucose values were above 125 mg/dL. Data regarding preoperative antidiabetic medication use were not available in our database and therefore could not be incorporated into the present analysis. Additionally, preoperative biliary obstruction and interventions such as ERCP with stenting or PTBD may influence serum albumin levels and therefore alter GAR [65]; although adjustment for preoperative total bilirubin and biliary drainage status did not significantly impact the association between GAR and overall survival, residual confounding related to biliary status cannot be entirely excluded.
6. Conclusions
Here, we report for the first time GAR as a prognostic marker of survival in pancreatic cancer patients undergoing curative‐intent surgical resection. Elevated GAR was positively associated with worse overall survival. Further research is warranted on the applicability of GAR in the prognosis of locally advanced and metastatic disease, as well as GAR's role in stratification of patient risk and guiding clinical management of pancreatic cancer.
Author Contributions
Andrea Feci: data curation, writing – original draft, investigation, formal analysis, project administration. Sophia Shah: data curation, writing – original draft, investigation. Matthew Kraft: investigation, data curation, writing – original draft. Avinoam Nevler: writing – review and editing, writing – original draft, conceptualization, methodology, supervision, formal analysis, project administration. Navin Rana: investigation, writing – original draft, data curation. Elijah Hoffman: investigation, writing – original draft, data curation. Harish Lavu: writing – review and editing, resources. Isabel Lavine: data curation, writing – original draft, investigation. Charles J. Yeo: writing – review and editing, resources. Benjamin Varughese: investigation, data curation, writing – original draft. Wilbur Bowne: writing – review and editing, resources.
Funding
The authors have nothing to report.
Ethics Statement
This study was conducted in accordance with the ethical standards of the Declaration of Helsinki and approved by the Institutional Review Board (Ethics Committee) of Thomas Jefferson University Hospital (Control #22E.126, date of approval February 17, 2022). Due to the retrospective nature of the study, the requirement for informed consent was waived by the IRB. The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1: Minimum p‐value Analysis. Association between different GAR cutoff values and the corresponding p‐values, with the minimum p‐value observed at a GAR cutoff of 26.44, corresponding to the median GAR value.
Figure S2: Overall Survival and Recurrence‐Free Survival Functions of BMI Tertiles. Kaplan–Meier analyses of patients with resected pancreatic ductal adenocarcinoma shows worse overall survival in patients with lower‐tertile (blue, N = 185) BMI (kg/m2) compared with patients with middle‐tertile (red, N = 188) BMI (26.6 vs. 30.6 months, p = 0.141) and upper‐tertile (green, N = 187) BMI (26.6 vs. 32.3 months, p = 0.058) (left). Patients with lower‐tertile (blue = 176) BMI also have shorter recurrence‐free survival compared to patients with middle‐tertile (red = 179) BMI (14.1 vs. 17.8 months, p = 0.011) and upper‐tertile (green, N = 177) BMI (14.1 vs. 22.7 months, p = 0.015) (right).
Figure S3: Overall Survival Functions of High vs. Low GAR Stratified by Diabetes Mellitus Status. Kaplan–Meier curves show that high GAR was associated with worse overall survival in patients without diabetes mellitus (32.5 vs. 25.4 months) (left), whereas the association was not statistically significant among patients with DM (p = 0.096) (right). DM was defined as preoperative HbA1c ≥ 6.5%.
Figure S4: Overall Survival Functions of High vs. Low Glucose and Albumin. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows worse overall survival in patients with above‐median glucose (32.9 vs. 29.3 months) (left), and no significant difference in overall survival in patients with preoperative above‐median albumin (right).
Figure S5: Recurrence‐Free Survival Functions of High vs. Low GAR and Glucose. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows no significant difference in recurrence‐free survival in patients with preoperative above‐median GAR (left) and above‐median glucose (right).
Figure S6: Time‐Dependent AUC and ROC Analysis of GAR for Overall and Disease‐Free Survival. Time‐dependent AUC analysis of GAR for overall survival (A) and disease‐free survival (C) from 5 to 60 months. GAR demonstrated a favorable AUC (> 0.50) across the evaluated time points, with the highest AUC observed at 50 months for overall survival (AUC 59.9%, 95% CI 52.6–67.3) and at 25 months for disease‐free survival (AUC 57.6%, 95% CI 51.8–63.3). Time‐dependent ROC curves at the optimal timepoints are shown for overall survival at 50 months (B; AUC 0.592) and disease‐free survival at 25 months (D; AUC 0.56), with corresponding optimal GAR cutoffs of 29.79 for overall survival and 23.86 for disease‐free survival.
Table S1: Univariate and multivariate Cox regression models comparing glucose, albumin, and GAR modeled as continuous and binary, median‐based variables.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Afghani E. and Klein A. P., Hematology/Oncology Clinics of North America (W.B. Saunders, 2022), 10.1016/j.hoc.2022.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. “Pancreatic Cancer — Cancer Stat Facts,” available, https://seer.cancer.gov/statfacts/html/pancreas.html.
- 3. Rahib L., Smith B. D., Aizenberg R., Rosenzweig A. B., Fleshman J. M., and Matrisian L. M., “Projecting Cancer Incidence and Deaths to 2030: The Unexpected Burden of Thyroid, Liver, and Pancreas Cancers in the United States,” Cancer Research 74, no. 11 (2014): 2913–2921, 10.1158/0008-5472.CAN-14-0155. [DOI] [PubMed] [Google Scholar]
- 4. Søreide K. and Stättner S., Principles and Practice of Surgical Oncology (Springer, 2021). [Google Scholar]
- 5. Soeters P. B., Wolfe R. R., and Shenkin A., “Hypoalbuminemia: Pathogenesis and Clinical Significance,” JPEN Journal of Parenteral and Enteral Nutrition 43, no. 2 (2018): 181–193, 10.1002/JPEN.1451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Fearon K. C. H., Barber M. D., Falconer J. S., McMillan D. C., Ross J. A., and Preston T., “Pancreatic Cancer as a Model: Inflammatory Mediators, Acute‐Phase Response, and Cancer Cachexia,” World Journal of Surgery 23, no. 6 (1999): 584–588, 10.1007/PL00012351. [DOI] [PubMed] [Google Scholar]
- 7. Almasaudi A. S., Dolan R. D., Edwards C. A., and McMillan D. C., “Hypoalbuminemia Reflects Nutritional Risk, Body Composition and Systemic Inflammation and Is Independently Associated With Survival in Patients With Colorectal Cancer,” Cancers 12, no. 7 (2020): 1986, 10.3390/CANCERS12071986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Gupta D. and Lis C. G., “Pretreatment Serum Albumin as a Predictor of Cancer Survival: A Systematic Review of the Epidemiological Literature,” Nutrition Journal 9, no. 1 (2010): 69, 10.1186/1475-2891-9-69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ahmetasevic E., Hasukic S., Agic M., and Brkic S., “Preoperative Hypoalbuminemia as a Predictor of Severe Postoperative Complications in Patients Undergoing Whipple Pancreatoduodenectomy,” Medical Archives 77, no. 5 (2023): 350, 10.5455/medarh.2023.77.350-353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Martín Santos S., Sorribas Grifell M., Busquets Barenys J., et al., “Hypoalbuminemia and Advanced Age Are Risk Factors for Delayed Gastric Emptying After Pancreaticoduodenectomy,” Nutrición Hospitalaria 40, no. 3 (2023): 517–520, 10.20960/nh.04287. [DOI] [PubMed] [Google Scholar]
- 11. Radomski M., Zenati M., Novak S., et al., “Factors Associated With Prolonged Hospitalization in Patients Undergoing Pancreatoduodenectomy,” American Journal of Surgery 215, no. 4 (2018): 636–642, 10.1016/j.amjsurg.2017.06.040. [DOI] [PubMed] [Google Scholar]
- 12. Siddiqui A., Heinzerling J., Livingston E. H., and Huerta S., “Predictors of Early Mortality in Veteran Patients With Pancreatic Cancer,” American Journal of Surgery 194, no. 3 (2007): 362–366, 10.1016/j.amjsurg.2007.02.007. [DOI] [PubMed] [Google Scholar]
- 13. Nakano Y., Kitago M., Shinoda M., et al., “Prognostic Significance of the Postoperative Level and Recovery Rate of Serum Albumin in Patients With Curatively Resected Pancreatic Ductal Adenocarcinoma,” Molecular and clinical oncology 11, no. 3 (2019): 270–278, 10.3892/mco.2019.1883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Liu Z., Hayashi H., Matsumura K., et al., “Biological and Clinical Impacts of Glucose Metabolism in Pancreatic Ductal Adenocarcinoma,” Cancers (Basel) 15, no. 2 (2023): 498, 10.3390/cancers15020498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Andersen D. K., Korc M., Petersen G. M., et al., “Diabetes, Pancreatogenic Diabetes, and Pancreatic Cancer,” Diabetes 66, no. 5 (2017): 1103–1110, 10.2337/db16-1477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Pannala R., Leirness J. B., Bamlet W. R., Basu A., Petersen G. M., and Chari S. T., “Prevalence and Clinical Profile of Pancreatic Cancer–Associated Diabetes Mellitus,” Gastroenterology 134, no. 4 (2008): 981–987, 10.1053/j.gastro.2008.01.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Hammer M., Storey S., Hershey D. S., et al., “Hyperglycemia and Cancer: A State‐Of‐The‐Science Review,” Oncology Nursing Forum 46, no. 4 (2019): 459–472, 10.1188/19.ONF.459-472. [DOI] [PubMed] [Google Scholar]
- 18. Barua R., Templeton A. J., Seruga B., Ocana A., Amir E., and Ethier J. L., “Hyperglycaemia and Survival in Solid Tumours: A Systematic Review and Meta‐Analysis,” Clinical Oncology (Royal College of Radiologists) 30, no. 4 (2018): 215–224, 10.1016/J.CLON.2018.01.003. [DOI] [PubMed] [Google Scholar]
- 19. Dougherty S. M., Schommer J., Salinas J. L., et al., “Immediate Preoperative Hyperglycemia Correlates With Complications in Non‐Cardiac Surgical Cases,” Journal of Clinical Anesthesia 74 (2021): 110375, 10.1016/j.jclinane.2021.110375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Chen J. Y., Nassereldine H., Cook S. B., Thornblade L. W., Dellinger E. P., and Flum D. R., “Paradoxical Association of Hyperglycemia and Surgical Complications Among Patients With and Without Diabetes,” JAMA Surgery 157, no. 9 (2022): 765, 10.1001/jamasurg.2021.5561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Wang S., Pan T., Wang S., Zhang X., Peng L., and Yang W., “Impact of Preoperative Blood Glucose Levels on Prognosis and Postoperative Complications in Patients With Pancreatic Cancer,” Oncology Letters 29, no. 3 (2025): 135, 10.3892/ol.2025.14880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Kohara Y., Yasuda S., Nagai M., et al., “Critical Role of Preoperative Hyperglycemia in Association With GFAT‐1 Expression in Resected Pancreatic Cancer,” Annals of Surgical Oncology 31, no. 12 (2024): 7713–7721, 10.1245/s10434-024-15985-4. [DOI] [PubMed] [Google Scholar]
- 23. Cheang I., Li Y., Zhu X., et al., “Prognosis Prediction of Cardiovascular Event With Glucose‐Albumin Ratio on Patients With Cancer and Prescribed With Anthracycline,” Cancer Medicine 13, no. 23 (2024): e70471, 10.1002/cam4.70471. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Chen D., Jiang H., and Bao J., “Association between glucose‐to‐albumin ratio and kidney injury in US adults: Results from NHANES 2005–2018, 2024”.
- 25. Hao M., Liu X., Huang Y., et al., “Ratio of glucose to albumin level, risks of incident diabetes, cardiovascular disease and mortality, 2024”.
- 26. He J., Zhang Y., Li T., et al., “Glucose‐Albumin Ratio as New Biomarker for Predicting Mortality After Intracerebral Hemorrhage,” Neurosurgical Review 46, no. 1 (2023): 94, 10.1007/s10143-023-02002-7. [DOI] [PubMed] [Google Scholar]
- 27. Wang W., Tang W., Yao W., Lv Q., and Ding W., “Glucose‐Albumin Ratio (GAR) as a Novel Biomarker of Postoperative Urinary Tract Infection in Elderly Hip Fracture Patients,” Frontiers in Medicine 11 (2024): 1366012, 10.3389/fmed.2024.1366012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Wang W., Yao W., Tang W., et al., “Glucose‐To‐Albumin Ratio as a New Predictive Indicator for Postoperative Delirium in Geriatric Hip Fracture Patients,” Journal of Arthroplasty 40 (2024): 1573–1581.e4, 10.1016/j.arth.2024.11.037. [DOI] [PubMed] [Google Scholar]
- 29. Tang W., Ni X., Yao W., et al., “Glucose‐Albumin Ratio (GAR) as a Novel Biomarker for Predicting Postoperative Pneumonia (POP) in Older Adults With Hip Fractures,” Scientific Reports 14, no. 1 (2024): 26637, 10.1038/s41598-024-60390-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Lai P., Gu X., Lin X., et al., “Association of Random Glucose to Albumin Ratio With Post‐Contrast Acute Kidney Injury and Clinical Outcomes in Patients With ST‐Elevation Myocardial Infarction,” Front Endocrinol (Lausanne) 15 (2024): 1390868, 10.3389/fendo.2024.1390868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Kane L. E., Mellotte G. S., Mylod E., et al., “Diagnostic Accuracy of Blood‐Based Biomarkers for Pancreatic Cancer: A Systematic Review and Meta‐Analysis,” Cancer Research Communications 2, no. 10 (2022): 1229–1243, 10.1158/2767-9764.CRC-22-0190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Jung H. I., “Inflammatory and Nutritional Markers in Patients With Resectable Pancreatic Cancer,” Korean Journal of Clinical Oncology 19, no. 1 (2023): 1–2, 10.14216/kjco.23001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Lu J.‐N., Zhou L.‐S., Zhang S., Li J.‐X., and Xu C.‐J., “Performance of Nutritional and Inflammatory Markers in Patients With Pancreatic Cancer,” World journal of clinical oncology 15, no. 8 (2024): 1021–1032, 10.5306/wjco.v15.i8.1021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Teja M., Garrido M. I., Ocanto A., and Couñago F., “Prognostic Impact of Inflammatory and Nutritional Biomarkers in Pancreatic Cancer,” World journal of clinical oncology 16, no. 1 (2025): 101191, 10.5306/WJCO.V16.I1.101191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Worapongpaiboon R., Siranart N., Pajareya P., and Phutinart S., “Inflammatory Markers in Predicting Survival in Pancreatic Cancer: A Systematic Review and Meta‐Analysis,” Pancreatology 25, no. 3 (2025): 385–395, 10.1016/j.pan.2025.02.014. [DOI] [PubMed] [Google Scholar]
- 36. Gremese E., Bruno D., Varriano V., Perniola S., Petricca L., and Ferraccioli G., “Serum Albumin Levels: A Biomarker To be Repurposed in Different Disease Settings in Clinical Practice,” Journal of Clinical Medicine 12, no. 18 (2023): 6017, 10.3390/jcm12186017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Nipper C. A., Lim K., Riveros C., et al., “The Association Between Serum Albumin and Post‐Operative Outcomes Among Patients Undergoing Common Surgical Procedures: An Analysis of a Multi‐Specialty Surgical Cohort From the National Surgical Quality Improvement Program (NSQIP),” Journal of Clinical Medicine 11, no. 21 (2022): 6543, 10.3390/jcm11216543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Xie Q., Wang L., and Zheng S., “Prognostic and Clinicopathological Significance of C‐Reactive Protein to Albumin Ratio in Patients With Pancreatic Cancer: A Meta‐Analysis,” Dose‐Response: A Publication of International Hormesis Society 18, no. 2 (2020): 1559325820931290, 10.1177/1559325820931290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Pu N., Gao S., Xu Y., et al., “Alkaline Phosphatase‐To‐Albumin Ratio as a Prognostic Indicator in Pancreatic Ductal Adenocarcinoma After Curative Resection,” Journal of Cancer 8, no. 16 (2017): 3362–3370, 10.7150/jca.20917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Zhu X., Xu H., Xiao Z., et al., “Hyperglycemia Predicts Adverse Prognosis in Advanced Pancreatic Cancer Patients,” Endocrine 79, no. 2 (2023): 296–303, 10.1007/s12020-022-03196-4. [DOI] [PubMed] [Google Scholar]
- 41. Sato K., Hikita H., Myojin Y., et al., “Hyperglycemia Enhances Pancreatic Cancer Progression Accompanied by Elevations in Phosphorylated STAT3 and MYC Levels,” PLoS One 15, no. 7 (2020): e0235573, 10.1371/journal.pone.0235573. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Kesh K., Mendez R., Abdelrahman L., Banerjee S., and Banerjee S., “Type 2 Diabetes Induced Microbiome Dysbiosis Is Associated With Therapy Resistance in Pancreatic Adenocarcinoma,” Microbial Cell Factories 19, no. 1 (2020): 75, 10.1186/s12934-020-01330-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Duan W., Shen X., Lei J., et al., “Hyperglycemia, a Neglected Factor During Cancer Progression,” BioMed Research International 2014 (2014): 1–10, 10.1155/2014/461917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Li W., Ma Q., Li J., et al., “Hyperglycemia Enhances the Invasive and Migratory Activity of Pancreatic Cancer Cells via Hydrogen Peroxide,” Oncology Reports 25, no. 5 (2011): 1279–1287, 10.3892/OR.2011.1150. [DOI] [PubMed] [Google Scholar]
- 45. Li W., Ma Q., Liu J., et al., “Hyperglycemia as a Mechanism of Pancreatic Cancer Metastasis,” Frontiers in Bioscience (Landmark Edition) 17, no. 5 (2012): 1761–1774, 10.2741/4017. [DOI] [PubMed] [Google Scholar]
- 46. Fu N., Jiang Y., Qin K., Chen H., Deng X., and Shen B., “Higher Body Mass Index Indicated Better Overall Survival in Pancreatic Ductal Adenocarcinoma Patients: A Real‐World Study of 2010 Patients,” BMC Cancer 21, no. 1 (2021): 1–11, 10.1186/S12885-021-09056-0/FIGURES/4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Lee B., Han H.‐S., Yoon Y.‐S., Park Y., Kang M., and Kim J., ““Obesity Paradox” as a New Insight From Long‐Term Survivors in Pancreatic Cancer Patients,” HPB 27, no. 7 (2025): 922–929, 10.1016/J.HPB.2025.03.450. [DOI] [PubMed] [Google Scholar]
- 48. Tang Q., Li X., and Sun C.‐R., “Predictive Value of Serum Albumin Levels on Cancer Survival: A Prospective Cohort Study,” Frontiers in Oncology 14 (2024): 1323192, 10.3389/fonc.2024.1323192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Dima S. O., Tanase C., Albulescu R., et al., “An Exploratory Study of Inflammatory Cytokines as Prognostic Biomarkers in Patients With Ductal Pancreatic Adenocarcinoma,” Pancreas 41, no. 7 (2012): 1001–1007, 10.1097/MPA.0b013e3182546e13. [DOI] [PubMed] [Google Scholar]
- 50. Barber M. D., Ross J. A., and Fearon K. C. H., “Changes in Nutritional, Functional, and Inflammatory Markers in Advanced Pancreatic Cancer,” Nutrition and Cancer 35, no. 2 (1999): 106–110, 10.1207/S15327914NC352_2. [DOI] [PubMed] [Google Scholar]
- 51. Odamaki M., Kato A., Kumagai H., and Hishida A., “Counter‐Regulatory Effects of Procalcitonin and Indoxyl Sulphate on Net Albumin Secretion by Cultured Rat Hepatocytes,” Nephrology, Dialysis, Transplantation 19, no. 4 (2004): 797–804, 10.1093/ndt/gfh023. [DOI] [PubMed] [Google Scholar]
- 52. Ballmer P. E., Ochsenbein A. F., and Schütz‐Hofmann S., “Transcapillary Escape Rate of Albumin Positively Correlates With Plasma Albumin Concentration in Acute but Not in Chronic Inflammatory Disease,” Metabolism 43, no. 6 (1994): 697–705, 10.1016/0026-0495(94)90117-1. [DOI] [PubMed] [Google Scholar]
- 53. O'Gorman P., McMillan D. C., and McArdle C. S., “Impact of Weight Loss, Appetite, and the Inflammatory Response on Quality of Life in Gastrointestinal Cancer Patients,” Nutrition and Cancer 32, no. 2 (1998): 76–80, 10.1080/01635589809514722. [DOI] [PubMed] [Google Scholar]
- 54. McMillan D. C., “An Inflammation‐Based Prognostic Score and Its Role in the Nutrition‐Based Management of Patients With Cancer,” Proceedings of the Nutrition Society 67, no. 3 (2008): 257–262, 10.1017/S0029665108007131. [DOI] [PubMed] [Google Scholar]
- 55. Komáromi A., Hammarkvist F., Rooyackers O., Wernerman J., and Norberg A., “Albumin Synthesis in States of Inflammation,” Intensive Care Medicine Experimental 3, no. 1 (2015): 1, 10.1186/2197-425X-3-S1-A42/METRICS. [DOI] [Google Scholar]
- 56. Preston T., Slater C., McMillan D. C., Falconer J. S., Shenkin A., and Fearon K. C. H., “Fibrinogen Synthesis Is Elevated in Fasting Cancer Patients With an Acute Phase Response,” Journal of Nutrition 128, no. 8 (1998): 1355–1360, 10.1093/JN/128.8.1355. [DOI] [PubMed] [Google Scholar]
- 57. Laviano A., Meguid M. M., and Rossi‐Fanelli F., “Cancer Anorexia: Clinical Implications, Pathogenesis, and Therapeutic Strategies,” Lancet Oncology 4, no. 11 (2003): 686–694, 10.1016/S1470-2045(03)01247-6. [DOI] [PubMed] [Google Scholar]
- 58. Vujasinovic M., Valente R., Del Chiaro M., Permert J., and Löhr J. M., “Pancreatic Exocrine Insufficiency in Pancreatic Cancer,” Nutrients 9, no. 3 (2017): 183, 10.3390/NU9030183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59. Belinskaia D. A., Voronina P. A., Shmurak V. I., Jenkins R. O., and Goncharov N. V., “Serum Albumin in Health and Disease: Esterase, Antioxidant, Transporting and Signaling Properties,” International Journal of Molecular Sciences 22, no. 19 (2021): 10318, 10.3390/ijms221910318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60. Forrest L. M., McMillan D. C., McArdle C. S., Angerson W. J., and Dunlop D. J., “Evaluation of Cumulative Prognostic Scores Based on the Systemic Inflammatory Response in Patients With Inoperable Non‐Small‐Cell Lung Cancer,” British Journal of Cancer 89, no. 6 (2003): 1028–1030, 10.1038/sj.bjc.6601242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61. Buzby G. P., Mullen J. L., Matthews D. C., Hobbs C. L., and Rosato E. F., “Prognostic Nutritional Index in Gastrointestinal Surgery,” American Journal of Surgery 139, no. 1 (1980): 160–167, 10.1016/0002-9610(80)90246-9. [DOI] [PubMed] [Google Scholar]
- 62. Iida H., Tani M., Komeda K., et al., “Superiority of CRP‐Albumin‐Lymphocyte Index (CALLY Index) as a Non‐Invasive Prognostic Biomarker After Hepatectomy for Hepatocellular Carcinoma,” HPB 24, no. 1 (2022): 101–115, 10.1016/j.hpb.2021.06.414. [DOI] [PubMed] [Google Scholar]
- 63. Trestini I., Cintoni M., Rinninella E., et al., “Neoadjuvant Treatment: A Window of Opportunity for Nutritional Prehabilitation in Patients With Pancreatic Ductal Adenocarcinoma,” World Journal of Gastrointestinal Surgery 13, no. 9 (2021): 885–903, 10.4240/WJGS.V13.I9.885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64. Liu J., Jiang S., Yang X., Li X., and Wang N., “The Significant Value of Preoperative Prognostic Nutritional Index for Survival in Pancreatic Cancers: A Meta‐Analysis,” Pancreas 47, no. 7 (2018): 793–799, 10.1097/MPA.0000000000001089. [DOI] [PubMed] [Google Scholar]
- 65. Meng Y., Li Y., Yang J., Yao W., and Zhang Y., “Clinical Efficacy of Percutaneous Transhepatic Biliary Drainage for Obstructive Jaundice in Pancreatic Cancer and Risk Factors for Postoperative Pancreatitis,” American Journal of Translational Research 18, no. 4 (2026): 3073–3084, 10.62347/RUFL1001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Minimum p‐value Analysis. Association between different GAR cutoff values and the corresponding p‐values, with the minimum p‐value observed at a GAR cutoff of 26.44, corresponding to the median GAR value.
Figure S2: Overall Survival and Recurrence‐Free Survival Functions of BMI Tertiles. Kaplan–Meier analyses of patients with resected pancreatic ductal adenocarcinoma shows worse overall survival in patients with lower‐tertile (blue, N = 185) BMI (kg/m2) compared with patients with middle‐tertile (red, N = 188) BMI (26.6 vs. 30.6 months, p = 0.141) and upper‐tertile (green, N = 187) BMI (26.6 vs. 32.3 months, p = 0.058) (left). Patients with lower‐tertile (blue = 176) BMI also have shorter recurrence‐free survival compared to patients with middle‐tertile (red = 179) BMI (14.1 vs. 17.8 months, p = 0.011) and upper‐tertile (green, N = 177) BMI (14.1 vs. 22.7 months, p = 0.015) (right).
Figure S3: Overall Survival Functions of High vs. Low GAR Stratified by Diabetes Mellitus Status. Kaplan–Meier curves show that high GAR was associated with worse overall survival in patients without diabetes mellitus (32.5 vs. 25.4 months) (left), whereas the association was not statistically significant among patients with DM (p = 0.096) (right). DM was defined as preoperative HbA1c ≥ 6.5%.
Figure S4: Overall Survival Functions of High vs. Low Glucose and Albumin. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows worse overall survival in patients with above‐median glucose (32.9 vs. 29.3 months) (left), and no significant difference in overall survival in patients with preoperative above‐median albumin (right).
Figure S5: Recurrence‐Free Survival Functions of High vs. Low GAR and Glucose. Kaplan–Meier analysis of patients with resected pancreatic ductal adenocarcinoma shows no significant difference in recurrence‐free survival in patients with preoperative above‐median GAR (left) and above‐median glucose (right).
Figure S6: Time‐Dependent AUC and ROC Analysis of GAR for Overall and Disease‐Free Survival. Time‐dependent AUC analysis of GAR for overall survival (A) and disease‐free survival (C) from 5 to 60 months. GAR demonstrated a favorable AUC (> 0.50) across the evaluated time points, with the highest AUC observed at 50 months for overall survival (AUC 59.9%, 95% CI 52.6–67.3) and at 25 months for disease‐free survival (AUC 57.6%, 95% CI 51.8–63.3). Time‐dependent ROC curves at the optimal timepoints are shown for overall survival at 50 months (B; AUC 0.592) and disease‐free survival at 25 months (D; AUC 0.56), with corresponding optimal GAR cutoffs of 29.79 for overall survival and 23.86 for disease‐free survival.
Table S1: Univariate and multivariate Cox regression models comparing glucose, albumin, and GAR modeled as continuous and binary, median‐based variables.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.