ABSTRACT
Background
Pancreaticoduodenectomy (PD) in dialysis patients is rare but carries a high risk of complications and mortality. This study aimed to identify preoperative factors associated with severe postoperative complications and mortality.
Methods
Using the Japanese National Clinical Database, 329 dialysis patients undergoing PD between 2016 and 2020 were retrospectively analyzed. Multivariable penalized logistic regression identified preoperative risk factors for Clavien–Dindo classification (CDC) grade ≥ 4 complications, 30‐day mortality, operative mortality, postoperative sepsis, and postoperative pancreatic fistula (POPF) (grade B or C).
Results
CDC grade ≥ 4 complications, 30‐day mortality, and operative mortality occurred in 10.3%, 5.5%, and 11.2%, respectively. Diet‐ or oral medication–treated diabetes (Odds ratio 5.19, 95% confidence interval 1.19–22.66) and insulin‐treated diabetes (7.50, 1.61–34.93) independently predicted 30‐day mortality. Serum albumin levels < 3.0 g/dL independently predicted operative mortality (2.72, 1.05–7.02), while cardiovascular disease showed a borderline association (2.15, 0.95–4.85). Elevated CRP was significantly associated with postoperative sepsis (2.47, 1.07–5.67), and pancreatic cancer was associated with a lower risk of POPF (grade B or C) (0.59, 0.37–0.94).
Conclusions
Dialysis patients undergoing PD face perioperative risks. Early mortality is driven by acute metabolic and infectious vulnerability, whereas operative mortality reflects diminished physiologic reserve related to malnutrition and cardiovascular comorbidity.
Keywords: 30‐day mortality, dialysis, operative mortality, pancreaticoduodenectomy, postoperative complication
Matsumoto and colleagues analyzed 329 dialysis patients undergoing pancreaticoduodenectomy. Diet‐controlled or oral medication‐treated diabetes and insulin‐treated diabetes were independently associated with 30‐day mortality, whereas hypoalbuminemia predicted operative mortality and elevated CRP increased postoperative sepsis risk. Early mortality reflected acute metabolic and infectious vulnerability, while later mortality reflected diminished physiologic reserve.
1. Introduction
From 1990 to 2010, the number of patients with end‐stage renal disease on hemodialysis (HD) increased from 165 to 284 per million. Hence, with a world population of 7.5 billion people, more than 2 million end‐stage renal disease patients are on HD each year [1]. These patients usually have several complications, including diabetes mellitus and cardiovascular disease, which may significantly affect survival outcomes [2].
In cardiac and vascular surgery, preoperative renal failure has been reported as a risk factor for postoperative complications and death [3]. In general surgery, two large‐scale studies conducted using The American College of Surgeons National Surgical Quality Improvement Program (ACS‐NSQIP) database reported that chronic renal failure and long‐term dialysis are independent determinants of 30‐day postoperative mortality and serious complications [4, 5]. However, there are some limitations in applying the results of these two large‐scale studies to gastrointestinal surgery procedures of varying difficulty.
Pancreaticoduodenectomy (PD), one of the most difficult surgical procedures, was first reported by Kaush in 1912 [6]. Since then, Whipple modified the technique for patients with papillary carcinoma of the Vater's papilla in 1935, [6] and it has been recognized as a common procedure for pancreatic and peripancreatic tumors. Although recent advances in surgical techniques and perioperative management have reduced the operative mortality in high‐volume centers to approximately 1%–2%, the incidence of postoperative complications remains extremely high, ranging from 30% to 65%, and it is no exaggeration to say that this surgical procedure is still in its developmental stages [7, 8, 9]. Several preoperative risk factors for postoperative complications and operative mortality after PD, including age, sex, ischemic heart disease, renal dysfunction, degree of surgical procedure, and hospital volume, have been identified in previous studies [7, 10, 11, 12, 13, 14]. However, to our knowledge, the risk of PD‐related mortality in patients on dialysis has not been adequately explored. One possible reason for such limmited investigation is the low frequency of PD among patients on dialysis (0.6%–1.0%) [7, 15, 16]. Therefore, it is difficult to analyze a coherent set of cases even at multiple centers. Given the increasing number of patients on HD, [1] it is expected that the frequency of PD will increase with an accompanying rise in the risk of PD‐related morbidity and mortality. Therefore, the morbidity and mortality rates in these patients and their risk factors need to be clarified.
Previous nationwide studies in Japan have largely relied on administrative claim–based databases such as the Diagnosis Procedure Combination (DPC), which lack detailed clinical parameters including nutritional status, inflammatory markers, functional capacity, and granular comorbidity profiles [15, 16]. In contrast, the Japanese National Clinical Database (NCD), modeled after the ACS‐NSQIP, prospectively collects comprehensive laboratory data, activities of daily living, and standardized perioperative risk variables [7, 14, 17]. This enhanced clinical resolution enables refined preoperative risk stratification and mechanistic interpretation that were not feasible in DPC‐based analyses, particularly in high‐risk populations such as dialysis patients undergoing pancreaticoduodenectomy.
This study aimed to determine the incidence of postoperative complications of grade ≥ 4 according to the Clavien‐Dindo classification (CDC), [18] 30‐day mortality rate, and operative mortality rate in patients on dialysis who underwent PD as well as their preoperative risk factors using the Japanese National Clinical Database (NCD).
2. Materials and Methods
2.1. Patient Selection and Data Source
The Japanese National Clinical Database (NCD), using nearly the same variables and definitions as the ACS‐NSQIP, [4] was founded in April 2010 as the parent body of the database system linked to the board certification systems in Japan [17].
In the current study, we analyzed PD cases among patients on dialysis registered on the NCD. We enrolled patients aged ≥ 20 years who underwent PD between January 2016 and December 2020. We excluded patients with the following: emergency surgery, missing observation items, refusal to enroll in the NCD, and acute renal failure (renal failure occurring within 24 h before surgery). The retrospective use of the data collected by the NCD for observational research was assessed and approved by the extended Ethics Committee of the Japanese Surgical Society and this study was also approved by the Ethics Committee of the Jikei University School of Medicine [33–079 (10691)]. The study procedures adhered to the ethical standards outlined in the Declaration of Helsinki and were conducted in line with the STROBE criteria for retrospective observational studies. The requirement for patient consent was waived due to the retrospective study design.
Data for individual cases were obtained from the Case Report Form of the Medical Standard Assessment for Gastrointestinal Surgeons. All variables, definitions, and inclusion criteria are accessible on the NCD website [7, 14, 19].
The following seven NCD terms were used to extract PD as a procedure: PD, PD with lymph node and plexus dissection, PD with surrounding organ resection, PD with artery or portal vein reconstruction, PD with simultaneous artery and portal vein reconstruction, PD with gallbladder malignancy surgery, and laparoscopic PD.
Patients on dialysis were defined as those with chronic or acute renal failure who received some form of dialysis—including peritoneal dialysis, hemodialysis, hemofiltration, hemodiafiltration, or ultrafiltration—within 14 days before surgery, which was the same definition of patients on dialysis in the ACS‐SQIP [5]. Patients with acute renal failure within 24 h before surgery were excluded.
In the NCD, the classification of postoperative complications (within 30 days after surgery) was assessed according to the CDC [18]. The definition of pancreatic fistula by the International Study Group of Pancreatic Fistula (ISGPF) [20] was used for the NCD since 2012.
The 30‐day mortality was defined as death within 30 days of surgery, regardless of the patient's geographical location, even if the patient had been discharged from the hospital. Operative mortality was defined as death within the index hospitalization period, regardless of the hospital stay duration (in this database, a minimum postoperative follow‐up period of 90 days is guaranteed), as well as any death up to 30 days after surgery.
The following risk factors were assessed: age, sex, activities of daily living within 30 days before surgery, body mass index, preoperative comorbidities, and preoperative blood test results. Age was stratified into three categories: 20–64 years, 65–74 years, and ≥ 75 years. Preoperative comorbidities included cardiovascular diseases, respiratory diseases, cerebrovascular diseases, diabetes mellitus, risk factors for bleeding, and long‐term steroid administration (chronic steroid use). Diabetes mellitus was originally classified into five categories in the NCD: no diabetes, diet therapy alone, oral hypoglycemic agents, insulin treatment, and untreated diabetes. For analytical purposes, diabetes status was re‐stratified into three groups: no diabetes, diet‐ or oral medication–treated diabetes, and insulin‐treated diabetes. Cardiovascular disease was defined as the presence of any of the following medical histories: congestive heart failure (≤ 30 days before surgery), myocardial infarction (≤ 6 months before surgery), angina pectoris (≤ 30 days before surgery), percutaneous coronary intervention, or cardiac surgery. Respiratory disease was defined as the presence of any of the following medical histories: dyspnea, chronic obstructive pulmonary disease, or pneumonia on chest radiography or computed tomography at administration or a positive sputum bacterial culture on admission. Cerebrovascular disease was defined as the presence of any of the following medical histories: transient ischemic attack, reversible ischemic neurological deficit, cerebrovascular accident, coma, or cerebrovascular disease of any kind, even if the details were unknown. Preoperative laboratory variables included platelet count, serum total bilirubin level, serum albumin level, and serum CRP level. Serum albumin level was subdivided into three categories (< 3.0 g/dL, 3.0–< 3.5 g/dL, and ≥ 3.5 g/dL). Primary diagnosis was categorized as pancreatic cancer or non–pancreatic cancer and incorporated into multivariable analyses as a surrogate marker of pancreatic texture.
The primary outcomes were preoperative risk factors for postoperative CDC grade ≥ 4 complications, the 30‐day mortality and operative mortality rates. Secondary outcomes were preoperative risk factors for postoperative sepsis and postoperative pancreatic fistula (POPF) (grade B or C) [20].
2.2. Statistical Analysis
Continuous data were expressed as the median (interquartile range) and compared between the two groups using the Mann–Whitney U test. Categorical data were expressed as frequencies (percentages) and compared between the two groups using Fisher’s exact test. All factors that showed significant associations in the univariable analysis were selected for the multivariable logistic regression analysis. All statistical analyses were performed using the Stata 19 software (STATA Corp, College Station, TX, USA). The “firthlogit” command in Stata was utilized to perform penalized logistic regression, addressing small sample bias by applying Firth's penalized maximum likelihood estimation method [21]. A p‐value < 0.05 was considered statistically significant.
3. Results
3.1. Patients' Characteristics
We identified 329 eligible patients receiving dialysis who underwent PD from January 2016 to December 2020. Table 1 shows the clinical variables of the 329 patients. Patients were aged 20–64 years, 65–74 years, and ≥ 75 years in 78 (23.7%), 167 (50.8%), and 84 (25.5%) patients, respectively. Two hundred and fifty patients (76.0%) were male. Sixty patients (18.1%) had cardiovascular disease and 188 (57.1%) had diabetes mellitus. No patients in the present cohort were categorized as having untreated diabetes. The incidence of CDC grade ≥ 4 complications, 30‐day mortality, and operative mortality were 10.3%, 5.5%, and 11.2%, respectively. The incidence of postoperative sepsis and POPF (grade B or C) was 9.1% and 31.0%, respectively.
TABLE 1.
Patients' characteristics and univariate analyses for postoperative complication of Grade ≥ 4 according to CDC, 30‐day mortality, and operative mortality.
| Total | Postoperative complication of Grade ≥ 4 according to CDC | 30‐day mortality | Operative mortality | ||||
|---|---|---|---|---|---|---|---|
| No | Yes | No | Yes | No | Yes | ||
| N = 329 | N = 295 | N = 34 | N = 311 | N = 18 | N = 292 | N = 37 | |
| Age | |||||||
| 65–74 years | 167 (50.8%) | 149 (50.5%) | 18 (52.9%) | 158 (50.8%) | 9 (50.0%) | 151 (51.7%) | 16 (43.2%) |
| ≥ 75 years | 84 (25.5%) | 72 (24.4%) | 12 (35.3%) | 76 (24.4%) | 8 (44.4%) | 69 (23.6%) | 15 (40.5%) |
| Sex, male | 250 (76.0%) | 221 (74.9%) | 29 (85.3%) | 234 (75.2%) | 16 (88.9%) | 219 (75.0%) | 31 (83.8%) |
| ADL with any assistance within 30 days before surgery | 18 (5.5%) | 16 (5.4%) | 2 (5.9%) | 16 (5.1%) | 2 (11.1%) | 15 (5.1%) | 3 (8.1%) |
| BMI, > 25 kg/m2 | 64 (19.5%) | 58 (19.7%) | 6 (17.6%) | 61 (19.6%) | 3 (16.7%) | 56 (19.2%) | 8 (21.6%) |
| Cardiovascular disease | 60 (18.2%) | 53 (18.0%) | 7 (20.6%) | 55 (17.7%) | 5 (27.8%) | 48 (16.4%) | 12 (32.4%) |
| Respiratory disease | 23 (7.0%) | 21 (7.1%) | 2 (5.9%) | 21 (6.8%) | 2 (11.1%) | 21 (7.2%) | 2 (5.4%) |
| Cerebrovascular disease | 20 (6.1%) | 16 (5.4%) | 4 (11.8%) | 18 (5.8%) | 2 (11.1%) | 16 (5.5%) | 4 (10.8%) |
| Diabetes mellitus | |||||||
| Diet‐controlled or oral medication | 101 (30.7%) | 88 (29.8%) | 13 (38.2%) | 93 (29.8%) | 8 (44.4%) | 85 (29.1%) | 16 (43.2%) |
| Insulin‐treated | 87 (26.4%) | 76 (25.8%) | 11 (32.4%) | 79 (25.4%) | 8 (44.4%) | 75 (25.7%) | 12 (32.4%) |
| Bleeding disorder without treatment | 9 (2.7%) | 6 (2.0%) | 3 (8.8%) | 7 (2.3%) | 2 (11.1%) | 6 (2.1%) | 3 (8.1%) |
| Chronic steroid use | 12 (3.6%) | 10 (3.4%) | 2 (5.9%) | 11 (3.5%) | 1 (5.6%) | 11 (3.8%) | 1 (2.7%) |
| Platelet count, < 120 000/uL | 35 (10.6%) | 31 (10.5%) | 4 (11.8%) | 32 (10.3%) | 3 (16.7%) | 32 (11.0%) | 3 (8.1%) |
| Serum total bilirubin level, ≥ 2.0 mg/dL | 19 (5.8%) | 16 (5.4%) | 3 (8.8%) | 18 (5.8%) | 1 (5.6%) | 15 (5.1%) | 4 (10.8%) |
| Serum albumin level | |||||||
| < 3.0 g/dL | 58 (17.6%) | 51 (17.3%) | 7 (20.6%) | 53 (17.0%) | 5 (27.8%) | 45 (15.4%) | 13 (35.1%) |
| 3.0–< 3.5 g/dL | 93 (28.3%) | 80 (27.1%) | 13 (38.2%) | 87 (28.3%) | 6 (333%) | 81 (27.7%) | 12 (32.4%) |
| Serum CRP level, > 1.0 mg/dL | 72 (21.9%) | 60 (20.3%) | 12 (35.3%) | 66 (21.2%) | 6 (33.3%) | 58 (19.9%) | 14 (37.8%) |
| Disease, pancreatic cancer | 164 (49.8%) | 152 (51.5%) | 12 (35.3%) | 157 (50.5%) | 7 (38.9%) | 149 (51.0%) | 15 (40.5%) |
| Intraoperative estimated blood loss (mL) | 649 (335–1130) | 639 (330–1130) | 853.5 (358–1180) | 647 (335–1130) | 746 (315–956) | 633 (325.5–1100) | 829 (420–1439) |
| Operative time (minutes) | 466 (397–550) | 471 (398–550) | 441.5 (379–573) | 472 (398–556) | 422.5 (373–456) | 468 (397–549.5) | 449 (394–562) |
| Length of hospital stay after surgery (days) | 34 (23–55) | 34 (23–55) | 30.5 (16–63) | 34 (23–61) | 16.5 (12–29) | 34 (23–55) | 38 (16–56) |
| Postoperative sepsis | 30 (9.1%) | 14 (4.7%) | 16 (47.1%) | 23 (7.4%) | 7 (38.9%) | 15 (5.1%) | 15 (40.5%) |
| POPF (grade B or C) | 102 (31.0%) | 83 (28.1%) | 19 (55.9%) | 91 (29.3%) | 11 (61.1%) | 85 (29.1%) | 17 (45.9%) |
Abbreviations: ADL, activities of daily living; BMI, body mass index; CDC, the Clavien‐Dindo classification; CRP, C‐reactive protein; POPF, postoperative pancreatic fistula.
3.2. Preoperative Risk Factors for CDC Grade ≥ 4 Complications
Complications between the group with CDC grade ≥ 4 complications and those with CDC grade ≤ 3 complications revealed that the former group had more patients with the following: age ≥ 75 years (35.3% vs. 24.4%), cerebrovascular disease (11.8% vs. 5.4%), bleeding disorders without treatment (8.8% vs. 2.0%), and serum CRP levels > 1.0 mg/dL (35.3% vs. 20.3%) and had fewer patients with pancreatic cancer (35.3% vs. 51.3%) (Table 1). Multivariable analysis indicated that age 65–74 years (odds ratio [OR]: 2.39, 95% confidence interval [CI]: 0.78–7.35, p = 0.13), age ≥ 75 years (OR: 3.12, 95% CI: 0.95–10.24, p = 0.06), cerebrovascular disease (OR: 2.42, 95% CI: 0.76–7.77, p = 0.14), diet‐controlled or oral medication–treated diabetes (OR: 1.84, 95% CI: 0.76–4.46, p = 0.18), insulin–treated diabetes (OR: 2.30, 95% CI: 0.90–5.88, p = 0.08), bleeding disorders without treatment (OR: 4.21, 95% CI: 0.94–18.91, p = 0.06), chronic steroid use (OR: 3.06, 95% CI: 0.67–13.94, p = 0.15), serum CRP levels > 1.0 mg/dL (OR: 1.78, 95% CI: 0.60–5.28, p = 0.08), and pancreatic cancer (OR: 0.48, 95% CI: 0.18–1.28, p = 0.05) tended to be associated with CDC grade ≥ 4 complications (Figure 1). Regarding postoperative complications, there were more patients with postoperative sepsis (47.1% vs. 4.7%, p < 0.001) and POPF (grade B or C) (55.9% vs. 28.1%, p = 0.002) in the group with CDC grade ≥ 4 complications than in the group with CDC grade ≤ 3 complications (Table 1).
FIGURE 1.
Multivariable analysis of preoperative risk factors in dialysis patients who underwent pancreaticoduodenectomy with regard to postoperative CDC grade ≥ 4 complications, 30‐day mortality, and operative mortality. ADL, activities of daily living; BMI, body mass index; CDC, the Clavien‐Dindo classification; CI, confidence interval; CRP, C‐reactive protein; OR, odds ratio.
3.3. Preoperative Risk Factors for 30‐Day Mortality
In a univariable analysis of preoperative factors, patients with diet‐controlled or oral medication–treated diabetes and insulin–treated diabetes had higher 30‐day mortality rates than those without diabetes (44.4% vs. 29.8% and 44.4% vs. 25.4%, respectively). On the other hand, there was a trend toward more patients aged ≥ 75 years (44.4% vs. 24.4%) and bleeding disorders without treatment (11.1% vs. 2.3%) among patients who died within 30 days of surgery compared to those who did not within 30 days of surgery (Table 1). Multivariable analysis indicated that diet‐controlled or oral medication–treated diabetes and insulin–treated diabetes were significantly associated with 30‐day mortality (OR: 5.19, 95% CI: 1.19–22.66, p = 0.03 and OR: 7.50, 95% CI: 1.61–34.93, p = 0.01, respectively), while age 65–74 years (OR: 3.29, 95% CI: 0.55–19.69, p = 0.19), age ≥ 75 years (OR: 5.85, 95% CI: 0.95–36.00, p = 0.06), bleeding disorders without treatment (OR: 4.52 95% CI: 0.75–27.28, p = 0.10), and chronic steroid use (OR: 4.52, 95% CI: 0.67–30.64, p = 0.12) tended to be associated with 30‐day mortality (Figure 1); however, these associations were not statistically significant. Univariable analysis of postoperative complications revealed that there were more patients with postoperative sepsis (38.9% vs. 7.4%, p < 0.001) and POPF (grade B or C) (61.1% vs. 29.3%, p = 0.008) among those who died within 30 days of surgery than among those who did not (Table 1).
3.4. Preoperative Risk Factors for Operative Mortality
Univariable analysis of preoperative factors revealed that the operative mortality rate was higher in patients with cardiovascular disease (32.4% vs. 16.4%), diet‐controlled or oral medication–treated diabetes and insulin–treated diabetes (43.2% vs. 29.1% and 32.4% vs. 25.7%, respectively), serum albumin levels < 3.0 g/dL and 3.0–< 3.5 g/dL (35.1% vs. 15.4% and 32.4% vs. 27.7%, respectively), and serum CRP levels > 1.0 mg/dL (37.8% vs. 19.9%). On the other hand, there was a trend toward more patients aged ≥ 75 years (40.5% vs. 23.6%) and bleeding disorders without treatment (8.1% vs. 2.1%) among those who died within the index hospitalization period compared to those who did not die (Table 1). Multivariable analysis indicated that diet‐controlled or oral medication–treated diabetes and serum albumin levels < 3.0 g/dL were significantly associated with operative mortality (OR: 2.46, 95% CI: 1.02–5.90, p = 0.045 and OR: 2.72, 95% CI: 1.05–7.02, p = 0.04, respectively), while age ≥ 75 years (OR: 2.82, 95% CI: 0.96–8.28), cardiovascular disease (OR: 2.15, 95% CI: 0.95–4.85), cerebrovascular disease (OR: 2.30, 95% CI: 0.70–7.59), insulin–treated diabetes (OR: 2.15, 95% CI: 0.84–5.55), serum CRP levels > 1.0 mg/dL (OR: 1.72, 95% CI: 0.78–3.78), and pancreas cancer (OR: 0.60, 95% CI: 0.29–1.23) tended to be associated with operative mortality (Figure 1). In a univariable analysis of surgical outcomes, the intraoperative estimated blood loss among patients who died within the index hospitalization period tended to exceed that in patients who did not die within the same period (633 g vs. 829 g, p = 0.099). In a univariable analysis of postoperative complications, there were more patients with postoperative sepsis (38.9% vs. 7.4%, p < 0.001), and there was a trend toward more patients with POPF (grade B or C) (61.1% vs. 29.3%, p = 0.008) among patients who died within the index hospitalization period compared to those who did not (Table 1).
3.5. Preoperative Risk Factors for Postoperative Sepsis
Univariable analysis of preoperative factors revealed that the postoperative sepsis rate was higher in patients with serum CRP levels > 1.0 mg/dL (43.3% vs. 19.7%). On the other hand, there was a trend toward more patients with serum albumin levels < 3.0 g/dL and 3.0–< 3.5 g/dL (30.0% vs. 16.4% and 30.0% vs. 28.1%, respectively) (Table 2). Multivariable analysis indicated that serum CRP levels > 1.0 mg/dL were significantly associated with postoperative sepsis (OR: 2.47, 95% CI: 1.07–5.67, p = 0.03), while being male tended to be associated with postoperative sepsis (OR: 3.20, 95% CI: 0.85–12.07) (Figure 2).
TABLE 2.
Patients' characteristics and Univariate analyses for Postoperative sepsis and POPF (grade B or C).
| Total | Postoperative sepsis | POPF (grade B or C) | |||
|---|---|---|---|---|---|
| No | Yes | No | Yes | ||
| N = 329 | N = 299 | N = 30 | N = 227 | N‐102 | |
| Age | |||||
| 65–74 years | 167 (50.8%) | 151 (50.5%) | 16 (53.3%) | 116 (51.1%) | 51 (50.0%) |
| ≥ 75 years | 84 (25.5%) | 74 (24.7%) | 10 (33.3%) | 58 (25.6%) | 26 (25.5%) |
| Sex, male | 250 (76.0%) | 222 (74.2%) | 28 (93.3%) | 169 (74.4%) | 81 (79.4%) |
| ADL with any assistance within 30 days before surgery | 18 (5.5%) | 15 (5.0%) | 3 (10.0%) | 14 (6.2%) | 4 (3.9%) |
| BMI, > 25 kg/m2 | 64 (19.5%) | 59 (19.7%) | 5 (16.7%) | 40 (17.6%) | 24 (23.5%) |
| Cardiovascular disease | 60 (18.2%) | 53 (17.7%) | 7 (23.3%) | 42 (18.5%) | 18 (17.6%) |
| Respiratory disease | 23 (7.0%) | 53 (17.7%) | 7 (23.3%) | 17 (7.5%) | 6 (5.9%) |
| Cerebrovascular disease | 20 (6.1%) | 17 (5.7%) | 3 (10.0%) | 12 (5.3%) | 8 (7.8%) |
| Diabetes mellitus | |||||
| Diet‐controlled or oral medication | 101 (30.7%) | 90 (30.1%) | 11 (36.7%) | 73 (32.2%) | 28 (27.5%) |
| Insulin‐treated | 87 (26.4%) | 77 (25.8%) | 10 (33.3%) | 60 (26.4%) | 27 (26.5%) |
| Bleeding disorder without treatment | 9 (2.7%) | 9 (3.0%) | 0 (0.0%) | 7 (3.1%) | 2 (2.0%) |
| Chronic steroid use | 12 (3.6%) | 11 (3.7%) | 1 (3.3%) | 8 (3.5%) | 4 (3.9%) |
| Platelet count, < 120 000/uL | 35 (10.6%) | 30 (10.0%) | 5 (16.7%) | 25 (11.0%) | 10 (9.8%) |
| Serum total bilirubin level, ≥ 2.0 mg/dL | 19 (5.8%) | 18 (6.0%) | 1 (3.3%) | 14 (6.2%) | 5 (4.9%) |
| Serum albumin level | |||||
| < 3.0 g/dL | 58 (17.6%) | 49 (16.4%) | 9 (30.0%) | 43 (18.9%) | 15 (14.7%) |
| 3.0–< 3.5 g/dL | 93 (28.3%) | 84 (28.1%) | 9 (30.0%) | 66 (29.1%) | 27 (26.5%) |
| Serum CRP level, > 1.0 mg/dL | 72 (21.9%) | 59 (19.7%) | 13 (43.3%) | 49 (21.6%) | 23 (22.5%) |
| Disease, pancreatic cancer | 164 (49.8%) | 150 (50.2%) | 14 (46.7%) | 123 (54.2%) | 41 (40.2%) |
| Intraoperative estimated blood loss (mL) | 649 (335–1130) | 639 (325–1106) | 867.5 (420–1180) | 670 (350–1210) | 577.5 (315–960) |
| Operative time (minutes) | 466 (397–550) | 464 (389–548) | 511 (430–611) | 477 (409–559) | 447.5 (373–534) |
| Length of hospital stay after surgery (days) | 34 (23–55) | 33 (22–52) | 63 (34–90) | 29 (19–39) | 51.5 (35–84) |
| Postoperative sepsis | 30 (9.1%) | 0 (0.0%) | 30 (100.0%) | 14 (6.2%) | 16 (15.7%) |
| POPF (grade B or C) | 102 (31.0%) | 86 (28.8%) | 16 (53.3%) | 0 (0.0%) | 102 (100.0%) |
Abbreviations: ADL, activities of daily living; BMI, body mass index; CDC, the Clavien‐Dindo classification; CRP, C‐reactive protein; POPF, postoperative pancreatic fistula.
FIGURE 2.
Multivariable analysis of preoperative risk factors in dialysis patients who underwent pancreaticoduodenectomy with regard to postoperative sepsis and POPF (grade B or C). ADL, activities of daily living; BMI, body mass index; CI, confidence interval; 6+CRP, C‐reactive protein; OR, odds ratio; POPF, postoperative pancreatic fistula.
3.6. Preoperative Risk Factors for POPF (Grade B or C)
Univariable analysis of preoperative factors revealed that the POPF (grade B or C) rate was lower in patients with pancreatic cancer (40.2% vs. 54.2%) (Table 2). Multivariable analysis indicated that patients with pancreatic cancer were significantly associated with POPF (grade B or C) (OR: 0.59, 95% CI: 0.37–0.94, p = 0.03) (Figure 2).
4. Discussion
In the present study, we evaluated the impact of preoperative factors on the development of surgical CDC grade ≥ 4 complications, 30‐day mortality, and operative mortality. Multivariable analysis indicated that diet‐controlled or oral medication–treated diabetes and insulin–treated diabetes were significantly associated with the 30‐day mortality and that diet‐controlled or oral medication–treated diabetes and serum albumin levels < 3.0 g/dL were significantly associated with the operative mortality. Moreover, age ≥ 75 years, cardiovascular disease, cerebrovascular disease, serum CRP levels > 1.0 mg/dL, and pancreatic cancer tended to be associated with operative mortality. To the best of our knowledge, this is the first study to demonstrate the impact of preoperative factors on CDC grade ≥ 4 complications, 30‐day mortality, and operative mortality in patients receiving dialysis who underwent PD.
Shinkawa et al. [15] reported that the post‐PD complication rate was higher in dialysis patients than in non‐dialysis patients, similar to that reported for general surgery [5]. The incidence of CDC grade ≥ 4 post‐PD complications in the previous nationwide cohort was 4.5% [14]. In the present study, the incidence of CDC grade ≥ 4 post‐PD complications in dialysis patients was 10.3%, suggesting a substantially higher burden of severe complications in this population. Additionally, Shinkawa et al. [15] reported 30‐day and operative mortality rates of 5.2% and 17.3%, respectively. The current cohort demonstrated a similar 30‐day mortality (5.5%) but an improved operative mortality (11.2%). This trend parallels nationwide improvements observed in the general PD population [22] and is likely attributable to advances in surgical technique, perioperative care, and quality initiatives such as board certification and centralization [23].
The present study further clarifies distinct temporal mechanisms of postoperative mortality in dialysis patients. Early mortality appears to be primarily driven by acute metabolic vulnerability related to diabetes severity and by infectious susceptibility. Elevated preoperative CRP was significantly associated with postoperative sepsis, and sepsis occurred markedly more frequently among patients who died, supporting its mediating role in early fatal outcomes. In contrast, operative mortality was more strongly influenced by reduced physiologic reserve rather than acute metabolic derangements alone. Hypoalbuminemia and cardiovascular disease, established markers of frailty and limited organ reserve [7, 14], were predominantly associated with later in‐hospital mortality, consistent with the concept of failure‐to‐rescue vulnerability reported in dialysis patients undergoing major surgery [5].
Interpretation of diabetes treatment categories in dialysis patients requires caution. Renal insulin clearance is markedly reduced in advanced kidney failure, prolonging insulin half‐life and frequently modifying therapeutic requirements [24], while dialysis‐related metabolic changes and strict dietary restrictions further complicate glycemic management [25, 26]. Consequently, pharmacologic treatment intensity may not accurately reflect underlying diabetes severity, potentially explaining why diet‐ or oral medication–treated diabetes remained significant for operative mortality whereas insulin‐treated diabetes showed only a non‐significant trend despite a similar direction of effect.
These findings highlight that perioperative risk in dialysis patients undergoing PD is not uniform but evolves over time, with distinct mechanisms driving early versus later mortality. Incorporating both acute vulnerability markers and physiologic reserve indicators into preoperative assessment may enable more precise risk stratification and targeted perioperative management.
The clinical granularity of the NCD, particularly the availability of nutritional, inflammatory, and functional parameters, enabled mechanistic interpretation of distinct mortality patterns that could not be achieved in administrative database–based studies.
This study has several limitations that should be acknowledged. The first limitation is the absence of information on the dialysis duration, modality, and timing information from the Japanese NCD. In particular, a longer duration of dialysis is known to be associated with increased vascular calcification [27]. It is expected that the duration of dialysis will significantly influence the postoperative complications, 30‐day mortality, and operative mortality. Therefore, further evaluation of the risk factors that take into account the dialysis duration, modality, and timing information is required in the future. The second limitation is the retrospective nature of our study, as this study design comes with inevitable non‐intentional biases. Thus, we encourage prospective large multicenter clinical trials to highlight the rates and risk factors for post‐PD morbidity and mortality in dialysis patients.
In conclusion, dialysis patients undergoing PD remain at exceptionally high risk of severe complications and mortality. Early postoperative death is predominantly driven by acute metabolic and infectious vulnerability, whereas later in‐hospital mortality reflects diminished physiologic reserve characterized by malnutrition and cardiovascular comorbidity. Comprehensive perioperative risk stratification integrating both acute vulnerability and reserve capacity is essential to optimize outcomes in this population.
Author Contributions
Michinori Matsumoto (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Visualization, Writing – original draft), Hiroyuki Yamamoto (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – review and editing), Toru Ikegami (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing – review and editing), Kyoko Yamamoto (Data curation, Formal analysis, Methodology, Software, Validation, Writing – review and editing), Ken Shirabe (Investigation, Methodology, Project administration, Resources, Supervision, Writing – review and editing), Yoshihiro Kakeji (Investigation, Methodology, Project administration, Resources, Supervision, Writing – review and editing), Atsushi Nanashima (Investigation, Methodology, Project administration, Resources, Supervision, Writing – review and editing), Itaru Endo (Funding acquisition, Project administration, Resources, Supervision, Writing – review and editing), Masafumi Nakamura (Funding acquisition, Project administration, Resources, Supervision, Writing – review and editing), and Masayuki Ohtsuka (Funding acquisition, Project administration, Resources, Supervision, Writing – review and editing).
Funding
This study is a research project evaluated and adopted by the Japanese Society of Hepato‐Biliary‐Pancreatic Surgery and this research was supported by the Japanese Society of Hepato‐Biliary‐Pancreatic Surgery.
Ethics Statement
This study was approved by the Ethics Committee of the Jikei University School of Medicine (approval number: 33–079 (10691)).
Consent
Individual consent was not required as the data in this study were anonymized.
Conflicts of Interest
There are no conflicts of interest for any of the authors except for Hiroyuki Yamamoto and Kyoko Yamamoto. Hiroyuki Yamamoto and Kyoko Yamamoto are affiliated with the Department of Healthcare Quality Assessment at the University of Tokyo. The department is a social collaboration department supported by grants from the National Clinical Database, Intuitive Surgical Sarl, Johnson & Johnson K.K., and Nipro Co. The corresponding authors had full access to all of the data and take full responsibility for the veracity of the data and statistical analysis. The paper is not based on any previous communication to a society or meeting.
Acknowledgments
This study was conducted by the Project Committee of the Japanese Society of Hepato‐Biliary‐Pancreatic Surgery. We are indebted to all the data managers and hospitals participating in this NCD project for their great efforts with regard to data entry. No preregistration exists for the reported study and plan of analysis. The Authors would like to thank Enago (www.enago.jp) for the English language review.
Data Availability Statement
The data that support the findings of this study are available from the National Clinical Database. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://www.ncd.or.jp/ with the permission of the National Clinical Database.
References
- 1. Wetmore J. B. and Collins A. J., “Meeting the World's Need for Maintenance Dialysis,” Journal of the American Society of Nephrology 26, no. 11 (2015): 2601–2603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Pugh J., Aggett J., Goodland A., et al., “Frailty and Comorbidity Are Independent Predictors of Outcome in Patients Referred for Pre‐Dialysis Education,” Clinical Kidney Journal 9, no. 2 (2016): 324–329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Brown J. R., Cochran R. P., MacKenzie T. A., et al., “Long‐Term Survival After Cardiac Surgery Is Predicted by Estimated Glomerular Filtration Rate,” Annals of Thoracic Surgery 86, no. 1 (2008): 4–11. [DOI] [PubMed] [Google Scholar]
- 4. Gaber A. O., Moore L. W., Aloia T. A., et al., “Cross‐Sectional and Case‐Control Analyses of the Association of Kidney Function Staging With Adverse Postoperative Outcomes in General and Vascular Surgery,” Annals of Surgery 258, no. 1 (2013): 169–177. [DOI] [PubMed] [Google Scholar]
- 5. Gajdos C., Hawn M. T., Kile D., Robinson T. N., and Henderson W. G., “Risk of Major Nonemergent Inpatient General Surgical Procedures in Patients on Long‐Term Dialysis,” JAMA Surgery 148, no. 2 (2013): 137–143. [DOI] [PubMed] [Google Scholar]
- 6. Schnelldorfer T., Adams D. B., Warshaw A. L., Lillemoe K. D., and Sarr M. G., “Forgotten Pioneers of Pancreatic Surgery: Beyond the Favorite Few,” Annals of Surgery 247, no. 1 (2008): 191–202. [DOI] [PubMed] [Google Scholar]
- 7. Kimura W., Miyata H., Gotoh M., et al., “A Pancreaticoduodenectomy Risk Model Derived From 8575 Cases From a National Single‐Race Population (Japanese) Using a Web‐Based Data Entry System: The 30‐Day and In‐Hospital Mortality Rates for Pancreaticoduodenectomy,” Annals of Surgery 259, no. 4 (2014): 773–780. [DOI] [PubMed] [Google Scholar]
- 8. Miura F., Yamamoto M., Gotoh M., et al., “Validation of the Board Certification System for Expert Surgeons (Hepato‐Biliary‐Pancreatic Field) Using the Data of the National Clinical Database of Japan: Part 2 ‐ Pancreatoduodenectomy,” Journal of Hepato‐Biliary‐Pancreatic Sciences 23, no. 6 (2016): 353–363. [DOI] [PubMed] [Google Scholar]
- 9. Mise Y., Hirakawa S., Tachimori H., et al., “Volume‐ and Quality‐Controlled Certification System Promotes Centralization of Complex Hepato‐Pancreatic‐Biliary Surgery,” Journal of Hepato‐Biliary‐Pancreatic Sciences 30, no. 7 (2023): 851–862. [DOI] [PubMed] [Google Scholar]
- 10. McPhee J. T., Hill J. S., Whalen G. F., et al., “Perioperative Mortality for Pancreatectomy: A National Perspective,” Annals of Surgery 246, no. 2 (2007): 246–253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Yoshioka R., Yasunaga H., Hasegawa K., et al., “Impact of Hospital Volume on Hospital Mortality, Length of Stay and Total Costs After Pancreaticoduodenectomy,” British Journal of Surgery 101, no. 5 (2014): 523–529. [DOI] [PubMed] [Google Scholar]
- 12. Squires M. H., Lad N. L., Fisher S. B., et al., “The Effect of Preoperative Renal Insufficiency on Postoperative Outcomes After Major Hepatectomy: A Multi‐Institutional Analysis of 1,170 Patients,” Journal of the American College of Surgeons 219, no. 5 (2014): 914–922. [DOI] [PubMed] [Google Scholar]
- 13. Norman K. S., Domingo S. R., and Wong L. L., “Pancreatic Resections in Renal Failure Patients: Is It Worth the Risk?,” HPB Surgery 2014 (2014): 938251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Aoki S., Miyata H., Konno H., et al., “Risk Factors of Serious Postoperative Complications After Pancreaticoduodenectomy and Risk Calculators for Predicting Postoperative Complications: A Nationwide Study of 17,564 Patients in Japan,” Journal of Hepato‐Biliary‐Pancreatic Sciences 24, no. 5 (2017): 243–251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Shinkawa H., Yasunaga H., Hasegawa K., et al., “Mortality and Morbidity After Pancreatoduodenectomy in Patients Undergoing Hemodialysis: Analysis Using a National Inpatient Database,” Surgery 165, no. 4 (2019): 747–750. [DOI] [PubMed] [Google Scholar]
- 16. Mori Y., Okawara M., Shibao K., et al., “Short‐Term Outcomes of Pancreatoduodenectomy in Older Individuals Over a 9‐Year Period Using Real‐World Data: A Multilevel Analysis Based on a Nationwide Administrative Database in Japan,” Journal of Hepato‐Biliary‐Pancreatic Sciences 31, no. 4 (2024): 262–272. [DOI] [PubMed] [Google Scholar]
- 17. Miyata H., Gotoh M., Hashimoto H., et al., “Challenges and Prospects of a Clinical Database Linked to the Board Certification System,” Surgery Today 44, no. 11 (2014): 1991–1999. [DOI] [PubMed] [Google Scholar]
- 18. Dindo D., Demartines N., and Clavien P. A., “Classification of Surgical Complications: A New Proposal With Evaluation in a Cohort of 6336 Patients and Results of a Survey,” Annals of Surgery 240, no. 2 (2004): 205–213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Gotoh M., Miyata H., Hashimoto H., et al., “National Clinical Database Feedback Implementation for Quality Improvement of Cancer Treatment in Japan: From Good to Great Through Transparency,” Surgery Today 46, no. 1 (2016): 38–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Bassi C., Marchegiani G., Dervenis C., et al., “The 2016 Update of the International Study Group (ISGPS) Definition and Grading of Postoperative Pancreatic Fistula: 11 Years After,” Surgery 161, no. 3 (2017): 584–591. [DOI] [PubMed] [Google Scholar]
- 21. Firth D., “Bias Reduction of Maximum‐Likelihood‐Estim Ates,” Biometrika 80, no. 1 (1993): 27–38. [Google Scholar]
- 22. Kajiwara Y., Takahashi A., Ueno H., et al., “Annual Report on National Clinical Database 2020 for Gastroenterological Surgery in Japan,” Annals of Gastroenterological Surgery 7, no. 3 (2023): 367–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Konno H., Kamiya K., Kikuchi H., et al., “Association Between the Participation of Board‐Certified Surgeons in Gastroenterological Surgery and Operative Mortality After Eight Gastroenterological Procedures,” Surgery Today 47, no. 5 (2017): 611–618. [DOI] [PubMed] [Google Scholar]
- 24. Parvathareddy V. P., Wu J., and Thomas S. S., “Insulin Resistance and Insulin Handling in Chronic Kidney Disease,” Comprehensive Physiology 13, no. 4 (2023): 5069–5076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Singhsakul A., Supasyndh O., and Satirapoj B., “Effectiveness of Dose Adjustment of Insulin in Type 2 Diabetes Among Hemodialysis Patients With End‐Stage Renal Disease: A Randomized Crossover Study,” Journal of Diabetes Research 2019 (2019): 6923543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Nasri H. and Rafieian‐Kopaei M., “Diabetes Mellitus and Renal Failure: Prevention and Management,” Journal of Research in Medical Sciences: The Official Journal of Isfahan University of Medical Sciences 20, no. 11 (2015): 1112–1120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Zhang H., Li G., Yu X., et al., “Progression of Vascular Calcification and Clinical Outcomes in Patients Receiving Maintenance Dialysis,” JAMA Network Open 6, no. 5 (2023): e2310909. [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.
Data Availability Statement
The data that support the findings of this study are available from the National Clinical Database. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://www.ncd.or.jp/ with the permission of the National Clinical Database.