Preoperative chronic kidney disease and complications after pancreatoduodenectomy: a retrospective cohort study

Chronic kidney disease is a prevalent condition in surgical patients. Possible associations with increased postoperative morbidity and mortality have not been clearly demonstrated in patients undergoing pancreatoduodenectomy. The aim of this study was to assess the risk of postoperative complications in patients with reduced kidney function undergoing pancreatoduodenectomy.


Introduction
5][6] Importantly, CKD confers large costs to healthcare systems and contributes to comorbidity and mortality in the general population. 6,7The widely used definition of CKD proposed by the workgroup Kidney Disease: Improving Global Outcomes (KDIGO) utilizes estimated glomerular filtration rates (eGFR), based on creatinine measurements, as the marker to identify and classify CKD. 8 In surgical patients a low preoperative eGFR, reflecting a decreased kidney function, is associated with increased mortality and a higher frequency of postoperative complications.Such association has been shown primarily after cardiac and general surgery. 9,10In a large, multi-institutional survey from the United States, the prevalence of CKD among surgical patients was higher compared to general population. 11Further, patients with decreased eGFR experienced worse postoperative outcomes, including short-term mortality and major complications. 11Given the role of the kidneys' function in maintaining fluid homeostasis, patients with CKD may also be at increased risk of postoperative fluid retention.Postoperative weight increase, a common measure of fluid retention, has also been shown to correlate with several postoperative complications, including a longer hospital stay, intensive care unit (ICU) admission and in-hospital mortality. 12ancreatoduodenectomy is a demanding surgical procedure, with high rates of postoperative complications. 13Risk stratification according to predetermined predictors of adverse outcomes remains the key to promptly applying possible mitigation strategies.Evidence suggests that patients with a reduced preoperative kidney function are at higher risk of adverse complications following pancreatoduodenectomy. 14,15he aim of this study was to explore the relationship between preoperative CKD and major postoperative complications in patients undergoing pancreatoduodenectomy at a high-volume, tertiary referral centre.

Study population
Patients 18 years or older who underwent a pancreatoduodenectomy at the Karolinska University Hospital, Stockholm, Sweden, from January 2008 to June 2019, were included in this study.Patients were excluded if they had a history of previous left-sided pancreatic surgery.Electronic medical records were reviewed and data was retrospectively collected.This study was approved by the Regional Ethics Committee, Stockholm, Sweden (reference number 2020/05238) and informed consent was not deemed necessary.The study was performed in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. 16

Exposures and covariates of interest
The primary exposure of interest was preoperative kidney function.Kidney function was defined using eGFR based on serum creatinine measurements the day before surgery, part of routine preoperative procedures, and was calculated using the CKD Epidemiology Collaboration (CKD-EPI) equation, 17 and eGFR was then used as a continuous measurement and categorized using the KDIGO classification for CKD. 8 According to current KDIGO guidelines, eGFR <60 mL/min/1.73m 2 is recognized as CKD without other markers of kidney damage.Due to a relatively small number of patients with advanced CKD, patients with CKD stage G3b or worse were combined to a single group (equivalent to eGFR <45 mL/min/1.73m 2 ).Information on patient characteristics at the time of surgery were collected and included age, sex, preoperative weight, body mass index (BMI), American Society of Anaesthesiologists (ASA) Physical Status Classification System and smoking status (never, former, current).Further, information on history of diabetes, hypertension, previous myocardial infarction atrial fibrillation and heart failure was collected (binary).

Outcomes of interest
The primary outcomes of interest were postoperative complications and postoperative weight change.Postoperative complications included Clavien-Dindo grade ≥ IIIa, 18 admission to the ICU, a hospital-stay ≥14 days, delayed gastric emptying, postoperative pancreatic fistula, post-pancreatectomy haemorrhage and bile leakage.0][21][22] Postoperative weight change was used as a marker for overhydration, defined as the difference between preoperative weight and weight measured the first postoperative day.Information on weight on the second postoperative day was used if the weight at the first day after surgery was missing.A weight increase of ≥2 kg was considered clinically significant as this cut-off has previously been associated to an increased risk of postoperative complications. 23

Statistical analysis
Descriptive statistics for continuous variables were presented by mean and standard deviation (SD) or median and interquartile range (IQR).Categorical variables were summarized using frequency and percentage.Continuous variables were compared using analysis of variance or Student's t-test, while categorical variables were analysed using the chi-squared test or Fisher's exact test.Cross-sectional differences by preoperative eGFR categories were assessed using univariable and multivariable logistic regression, presenting odds ratios (OR) with 95% confidence intervals (CI).Multivariable logistic regression was used to adjust for potential confounders and included age at the time of surgery, BMI, sex, presence of hypertension, diabetes, atrial fibrillation, previous myocardial infarction, and heart failure (all binary), and smoking status (current, previous, never).Univariable and multivariable logistic regression models were also used to investigate the difference in odds of postoperative outcomes with eGFR on a linear scale, using increments of 10-units of preoperative eGFR, since a single unit of eGFR is a minor difference in the clinical context.Covariates included in adjusted models were decided a priori.For all analyses, P < 0.05 was considered significant.Analyses were conducted using Stata version 16 software (StataCorp).
All included patients had a creatinine measurement taken before the time of surgery and 909 patients (94%) had their weight measured the first or second day after surgery.No patient had CKD stage G5 or end-stage kidney disease preoperatively.

CKD categories and postoperative complications
When categorizing by CKD categories, patients with CKD G3b or worse had a higher adjusted odds of ICU admission (P = 0.038) and hospital stay ≥14 days (P = 0.002), compared to patients with a better preoperative kidney function (Table S1).Further, there was a trend towards higher adjusted odds of postoperative pancreatic fistula (P = 0.007) and post-pancreatectomy haemorrhage (P < 0.001) among patients with CKD G3b or worse, compared to patients with higher preoperative kidney function (Table S2).

Discussion
In the present study, the results suggest that patients with decreased preoperative kidney function undergoing pancreatoduodenectomy may be more likely to experience postoperative complications.In particular, a higher preoperative eGFR was associated to a lower risk of ICU admission, hospital stay >14 days, delayed gastric emptying, postoperative pancreatic fistula, and post-pancreatectomy haemorrhage.There was a trend of a higher risk for ICU admission, postoperative pancreatic fistula and post-pancreatectomy haemorrhage with more advanced CKD stage.In addition, patients with CKD were more likely to experience a weight change ≥2 kg, the latter an outcome associated with adverse postoperative events, compared to patients with normal kidney function.
5][26] Given its prevalence, CKD is a clinically relevant entity in surgical patients as well. 11Indeed, preoperative CKD has been reported to be associated with several  adverse postoperative outcomes after different types of surgery, including major cardiac complications, 27 anastomotic leakage, 28 postoperative infection 29 and mortality. 30There are several possible mechanisms that may lead CKD patients to adverse outcomes.Studies report high rates of comorbidity and polypharmacy in patients with CKD, which are also present in moderate stages of the disease 31 and may put these patients at increased risk of postoperative complications.Moreover, evidence suggests that patients with advanced CKD may have a compromised immune response, thus being more vulnerable to postoperative infections, abscesses and sepsis.For instance, earlier studies report macrophage and natural killer cell dysfunction in end-stage kidney disease 32,33 and patients with CKD may therefore be at increased risk of infectious diseases. 34here are some studies that have explored preoperative kidney function in pancreatic resections.One earlier study including 1061 patients who underwent pancreatic resection found that among 709 patients undergoing pancreatoduodenectomy, lower preoperative kidney function was independently associated with major postoperative complications, defined as Clavien-Dindo grade ≥ III and respiratory failure. 35In another study based on US national database, preoperative CKD categories in patients undergoing pancreatic resections was investigated, using a composite measurement of postoperative complications including unplanned intubation, pulmonary embolisms, sepsis, myocardial infarction among others. 36he study found that patients with CKD, including those with moderately reduced kidney function, had a higher risk of major complications. 36In accordance with our results, few patients with more advanced CKD were operated, and only 23 out of 16 173 included patients had an eGFR <15 mL/min/1.73m 2 .This probably reflects a surgical selection bias towards healthier patients that are candidates for the procedure, 35,36 but patient with more advanced CKD undergoing pancreatoduodenectomy appears to have a higher risk of postoperative complications. 37Interestingly, given that the number of patients on dialysis is expected to grow, Barbas et al. investigated postoperative complications and mortality after hepatic or pancreatic resection among patients on maintenance dialysis.The authors found that patients with end-stage kidney disease and on dialysis had a higher rate of postoperative complications, but not increased mortality. 38ur results are similar to previous studies investigating preoperative CKD in patients undergoing pancreatoduodenectomy, but expand on these by also assessing other postoperative complications, such as delayed gastric emptying, postpancreatectomy haemorrhage, bile leakage as well as postoperative weight change.For example, we found that patients with CKD were more likely to gain ≥2 kg postoperatively, likely reflecting fluid retention or overhydration.Early fluid retention in patients undergoing pancreatoduodenectomies has previously been associated to increased postoperative morbidity, and may represent an important factor in the early perioperative setting. 23,39In this present study, we were unable to include information on perioperative fluid administration, which limits the assessment of postoperative weight change.While some evidence suggests that a liberal intraoperative fluid balance may be associated with an increased rate of postoperative morbidity after pancreatoduodenectomy, there are also convincing data showing how, in case of a high-risk pancreas, a near-zero fluid balance could lead to pancreatic stump ischemia, with increased postoperative pancreatic fistula and hyperamylasaemia rates. 40,41atients with CKD and consequent high risk of early fluid retention may require a personalized intraoperative fluid management according to renal function and patient's pancreas-specific risk factors.Moreover, early treatment of fluid retention in CKD patients may help increase postoperative outcomes, preventing not only generic but also pancreas-specific comorbidities.Postoperative pancreatic fistula is a common and sometimes severe complication following pancreatoduodenectomy. 42Higher preoperative eGFR was associated with lower odds of developing postoperative pancreatic fistula.In addition, worse CKD category were associated with increasing odds of developing a fistula.Despite the possible causeeffect relationship remains unclear, previous evidence suggests that patients with kidney disease have impaired wound healing, 43,44 which could increase the risk of anastomotic leak.
Our study has some limitations.Preoperative serum creatine measurements were taken routinely the day before surgery and were for the purpose of this study used to determine preoperative eGFR and CKD stage.There is a possibility that these values change shortly before surgery, therefore not reflecting the habitual CKD stage.However, the results still indicate that there is an association between preoperative eGFR and postoperative adverse outcome.Other limitations of this study are the retrospective design and the low number of patients accepted for surgery with advanced CKD, which made impossible to include patients with pre-dialysis CKD or on kidney replacement therapy.We also lack information on intraoperative blood loss, pancreatic gland texture and duct size, factors that may influence the development of delayed gastric emptying, postoperative pancreatic fistula and postpancreatectomy haemorrhage.
In conclusion, this study found that patients undergoing pancreatoduodenectomy with decreased preoperative kidney function were more likely to experience major postoperative complications and postoperative weight increase.Preoperative kidney function assessment is important in risk stratification before pancreatoduodenectomies.

Fig. 1 .
Fig. 1.Unadjusted and adjusted odds ratios for postoperative complications comparing preoperative CKD and normal kidney function.

Table 1
Characteristics of included patients by chronic kidney disease categoriesPreoperative eGFR (ml/min/1.73m 2 ) eGFR, estimated glomerular filtration rate; IQR, interquartile range; BMI, body mass index; SD, standard deviation; ASA, American Society of Anaesthesiologists' physical status classification system.

Table 2
Characteristics of included patients by preoperative kidney function Values for continuous data are mean AE SD or median (IQR) and count (%) for categorical data.eGFR, estimated glomerular filtration rate; IQR, interquartile range; BMI, body mass index; SD, standard deviation; ASA, American Society of Anaesthesiologists' physical status classification system.

Table 3
Preoperative kidney function and odds ratios for postoperative complications after pancreatoduodenectomy Preoperative eGFR <60 versus ≥60 mL/min/1.73m 2 Unadjusted and adjusted logistic regression of postoperative complications after pancreatoduodenectomy comparing normal to decreased preoperative eGFR.†Adjusted for age, body mass index (both continuous), smoking status (current, previous, never), sex, presence of hypertension, diabetes, atrial fibrillation, previous myocardial infarction, and heart failure (all binary).

Table 4
Odds of postoperative complications per increase in 10 mL/min/1.73m 2 of preoperative eGFR Unadjusted and adjusted logistic regression of postoperative complications after pancreatoduodenectomy per change in 10 mL/min/1.73m 2 of eGFR.†Adjusted for age, body mass index (both continuous), smoking status (current, previous, never), sex, presence of hypertension, diabetes, atrial fibrillation, previous myocardial infarction, and heart failure (all binary).