Delayed gastric emptying is associated with increased risk of mortality in patients undergoing pancreaticoduodenectomy for pancreatic adenocarcinoma

Delayed gastric emptying (DGE) is common in patients undergoing pancreaticoduodenectomy (PD). The effect of DGE on mortality is less clear. We sought to identify predictors of mortality in patients undergoing PD for pancreatic adenocarcinoma hypothesizing DGE to independently increase risk of 30-day mortality. The ACS-NSQIP targeted pancreatectomy database (2014–2017) was queried for patients with pancreatic adenocarcinoma undergoing PD. A multivariable logistic regression analysis was performed. Separate sensitivity analyses were performed adjusting for postoperative pancreatic fistula (POPF) grades A–C. Out of 8011 patients undergoing PD, 1246 had DGE (15.6%). About 8.5% of patients with DGE had no oral intake by postoperative day-14. The DGE group had a longer median operative duration (373 vs. 362 min, p = 0.019), and a longer hospital length of stay (16.5 vs. 8 days, p < 0.001). After adjusting for age, gender, comorbidities, preoperative chemotherapy, preoperative radiation, open versus laparoscopic approach, vascular resection, deep surgical space infection (DSSI), postoperative percutaneous drain placement, and development of a POPF, DGE was associated with an increased risk for 30-day mortality (OR 3.25, 2.16–4.88, p < 0.001). On sub-analysis, grades A and B POPF were not associated with risk of mortality while grade C POPF was associated with increased risk of mortality (OR 5.64, 2.24–14.17, p < 0.001). The rate of DGE in patients undergoing PD in this large database was over 15%. DGE is associated with greater than three times the increased associated risk of mortality, even when controlling for POPF, DSSI, and other known predictors of mortality.


Introduction
Pancreatic ductal adenocarcinoma (PAC) is the third-leading cause of cancer death in the United States with only 11.5% surviving 5 years after diagnosis [1]. Surgical resection remains the foundation of treatment for non-metastatic PAC [2]. The perioperative mortality of patients undergoing pancreaticoduodenectomy (PD) has decreased to 1-4% [3,4]. Despite significant improvement in mortality, perioperative morbidity remains substantial. Studies report morbidity rates from 30 to 60%, with common postoperative complications consisting of delayed gastric emptying (DGE) and postoperative pancreatic fistula (POPF) [3,5].
DGE has been observed to be a complication in a high number of patients (up to 40%) undergoing PD, but whether DGE itself is associated with risk of mortality in PD patients remains unclear [6][7][8]. In contrast, POPF occurs in close to 20% of patients undergoing PD and has been shown to be associated with an increased risk of mortality [7,9,10]. Many studies have recognized the strong relationship between POPF and the occurrence of DGE [11][12][13][14][15], and have proposed that POPF grade C is an independent risk factor for DGE. [16] This study aimed to assess the risk of mortality in patients with DGE, when controlling for POPF and other significant predictors of mortality, with a hypothesis that DGE is independently associated with an increased risk of 30-day mortality.

Methods
A retrospective analysis of the American College of Surgeons-National Surgery Quality Improvement Program (ACS-NSQIP) targeted pancreatectomy database (2014-2017) was performed. All patients with PAC undergoing PD were identified. Patients with documented DGE were compared to patients without DGE. DGE was defined by NSQIP parameters as no oral intake by postoperative day fourteen or tube to external drainage, or if a nasogastric tube was present or reinserted [17].
The primary outcome was a 30-day mortality. Other outcomes measured were NSQIP-defined postoperative complications: bleeding requiring transfusions > 4 units, organ surgical site infection, deep surgical site infection, surgical site infection, sepsis, septic shock, unplanned intubation, ventilation ≥ 48 h, pneumonia, deep vein thrombosis, urinary tract infection, dehiscence, Clostridium difficile infection, acute renal failure, cardiac arrest with cardiopulmonary resuscitation, myocardial infarction, renal insufficiency, pulmonary embolism, and cerebrovascular accident. Additional outcomes measured included total hospital length of stay (LOS) and intensive care unit (ICU) LOS. The demographic variables collected included age and gender.
Frequency statistics were performed for all available variables. A Student's t test or Mann-Whitney-U test was used to compare continuous variables, and chi-square was used to compare categorical variables for bivariate analysis. Continuous data were reported as medians with interquartile range or as means with standard deviation. Categorical data were reported as percentages.
The magnitude of the association between predictor variables and mortality was first measured using a univariable logistic regression model. Covariates were chosen by author consensus and review of the literature and included sex, age greater than 65, bleeding disorder, congestive heart failure, chronic obstructive pulmonary disease (COPD), diabetes mellitus, hypertension, smoker, steroid use, weight loss, preoperative chemotherapy, preoperative radiation, vascular resection, percutaneous drain, deep surgical site infection, open operative approach, DGE, and development of POPF [18][19][20][21][22][23][24][25][26]. Covariates with statistical significance (p < 0.20) were included in a hierarchical multivariable logistic regression model and the adjusted risk for mortality was reported with an odds ratio (OR) and 95% confidence intervals (CI). Several studies have reported POPF to be associated with increased risk of DGE and associated with the development of clinical DGE [11][12][13][14][15][16]. Therefore, a separate sensitivity analysis adjusting for POPF grades A-C was performed because POPF could be a significant confounder of our analysis. NSQIP defines POPF as either a clinical diagnosis and NPO-TPN, drain continued longer than 7 days, percutaneous drainage, spontaneous wound drainage, or reoperation; or as persistent drain output of amylase rich fluid in combination with NPO-TPN, a drain continued longer than 7 days, percutaneous drainage, spontaneous wound drainage, or reoperation. NSQIP POPF grades were defined according to the International Study Group on Pancreatic Fistula (ISGPF) guidelines [27]. All p values were two-sided, with a statistical significance level of < 0.05. All analyses were performed with IBM SPSS Statistics for Windows (Version 24, IBM Corp., Armonk, NY). Patients that developed DGE after PD had higher rates of POPF grade A (16.9% vs. 6.6%, p < 0.001), POPF grade B (14.2% vs. 3.5%, p < 0.001), and POPF grade C (4.0% vs. 0.5%, p < 0.001) compared to patients without DGE. Additionally, patients with DGE had higher rates of postoperative percutaneous drain placement compared to patients without DGE (18.9% vs. 8.2%, p < 0.001). There was no difference in the operative approach and vascular resection when comparing patients that developed DGE after PD compared to those without DGE Table 2. The type of anastomosis and its effect on developing DGE could not be studied in more detail given that this was a static variable in the database and the type of anastomosis each patient underwent could not be separated.

Predictors of 30-day mortality for PAC patients undergoing PD
On multivariable analysis, DGE was independently associated with increased risk for 30 Table 3. Preoperative radiation/chemotherapy, sex, diabetes, COPD were not associated with 30-day mortality (all p > 0.05) Table 3. On sub-analysis, grades A and B POPFs were not associated with risk of mortality while grade C POPFs were associated with an increased risk of mortality (OR 5.64, 2.24-14.17, p < 0.001) Table 4.

Discussion
High rates of postoperative morbidity remain a concern for patients undergoing PD. In this retrospective analysis spanning three years, we found that DGE was independently associated with an increased risk of 30-day mortality in patients that had undergone PD for PAC. Other predictors of 30-day mortality included vascular resection, age over 65 years and hypertension. Sex, comorbidities, preoperative chemotherapy, preoperative radiation, and operative approach were not associated with 30-day mortality risk. In a separate sensitivity analysis, we found that grade C POPF was associated with a greater than five-fold increase in associated risk of mortality.
DGE is a leading postoperative complication following PD, with a high reported incidence of up to 25% [3,[6][7][8]. Many studies have demonstrated an association of DGE with increased hospital LOS, readmission, delayed adjuvant therapy, and other complications [7,11,13,15]. None of these studies found an independent association of DGE with mortality when controlling for other significant predictors of mortality following PD. Of the retrospective studies identifying postoperative PD mortality risk factors [6,[19][20][21][22], one study found DGE to be an independent risk factor for long-term pancreatic cancer survival [28]. The results of our current study demonstrate that DGE was independently associated with an increased risk of 30-day mortality following PD for PAC. This is not consistent with the ISGPS guidelines, which state that DGE is not a life-threatening complication [29]. A systematic review of clinical risk factors of DGE by Qu et al., also concluded that DGE was not life-threatening but rather a warning of other severe postoperative complications [11]. The results of the current study suggest that DGE may be addressed as an independent risk factor of mortality, rather than an association with other complications.
Multiple studies have identified risk factors of mortality following PD. Current literature shows that POPF [7,9,21] and hypertension [19,22] are associated with an increased risk of mortality, while surgical approach has no significant difference in mortality [25,30]. Factors such as age [19,22] vascular resection [21,31] comorbidities (COPD, diabetes, smoking) [19,22] neoadjuvant chemotherapy/radiation [18] and gender [19] have received mixed association with risk of mortality following PD and our study follows this trend. In the current study, we found that that POPF grade C, vascular resection, hypertension, and age greater than 65 years were associated with an increased risk of 30-day mortality. These results are consistent with several studies which demonstrated an association of POPF with mortality [19,21,22,31]. In regards to vascular resection, the results of this study add to the published literature demonstrating an increased risk of 30-day mortality [26,32,33]. When evaluating outcomes, the risk associated with hypertension, age greater than 65, and vascular resection should be incorporated into clinician and patient decision making regarding PD for PAC.
POPF remains a common and problematic complication following PD. Several retrospective studies have found Grade C POPF with a mortality rate as high as 30% after PD [7,9]. However these studies did not control for DGE. Our study adds to existing literature with the finding that grade C POPF was associated with a greater than five-fold increase in risk of 30-day mortality for patients undergoing PD for PDAC, even when controlling for DGE. This reinforces that further work addressing reduction of highgrade pancreatic fistula is needed.
Our study has several limitations including those associated with a retrospective database analysis such as reporting bias, missing data, and misclassification although attempts are made by the ACS to limit these by stringent training requirements of the institutional coders that contribute data to NSQIP. Also, ACS-NSQIP is confined to 30-day outcomes; thus, long-term outcomes are not available. Finally, a significant limitation is that NSQIP does not use the ISGPS definition of DGE, making the definition non uniform with other studies. As such, we were unable to determine if DGE occurred independent of intra-abdominal complications versus concomitant; however, we controlled for these possible confounders during our analyses. Another limitation specific to our study is the inability to separately study the type of anastomosis created for each patient (gastrojejunostomy versus duodenojejunostomy, antecolic versus retrocolic) and its relationship with the development of DGE. Despite these limitations, our study is strengthened by the extensive library of clinical outcomes and variables specific to pancreatic surgery available in this targeted version of the NSQIP database.

Conclusion
In summary, this study utilizing NSQIP found that for patients undergoing PD for PAC, DGE was independently associated with greater than three times the risk for 30-day mortality. In addition, POPF grade C, vascular resection, hypertension, and age greater than 65 years were associated with an increased risk of mortality. Specifically grade C POPF was associated with a greater than five times increased risk for 30-day mortality when controlling for DGE. We propose that DGE should be recognized as an independent risk factor for mortality. Future studies on reduction of occurrence of DGE are warranted.
Funding No funding was received for this work.

Conflict of interest
The authors declare that they have no conflicts of interest.

Research involved animals and humans
This research involved humans. However, since this retrospective study was performed using a national database with deidentified patients, risk to participants is minimal. There is no consent required.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.