Abstract
Introduction:
The impact of obesity on postoperative outcomes after pancreatoduodenectomy remains understudied.
Methods:
All pancreatoduodenectomy patients were abstracted from the 2014–2018 ACS NSQIP datasets and were stratified into three BMI categories: non-obese (BMI 18.5–29.9); class 1/2 obesity (BMI 30–39.9); and class 3 severe obesity (BMI≥40). Analyses tested associations between patient factors and four 30-day postoperative outcomes: mortality, composite morbidity, delayed gastric emptying (DGE), and postoperative pancreatic fistula (POPF). Multivariable logistic regression models tested independent associations between patient factors and these four outcomes.
Results:
16,823 patients were included: 12,234 (72.7%) non-obese; 4,030 (24%) obese; and 559 (3.3%) with severe obesity. Bivariable analyses demonstrated significant associations between obesity, severe obesity and greater proportions of numerous preoperative comorbidities as well as greater likelihood of post-operative complications, including POPF, DGE, composite morbidity, and mortality (all p≤0.001). After adjusting for significant covariates, obesity was independently associated with POPF (OR 1.49, 95% CI: 1.33–1.67, p < 0.001), DGE (OR 1.16, 95% CI: 1.05–1.28, p = 0.004), composite morbidity (OR 1.28, 95% CI: 1.18–1.38, p < 0.001), and mortality (OR 1.79, 95% CI: 1.36–2.36, p < 0.001).
Conclusions:
Obesity and severe obesity are significantly associated with worse short-term outcomes after pancreatoduodenectomy. Preoperative considerations such as weight management strategies during individualized treatment planning could improve outcomes in this population.
INTRODUCTION
Obesity is an epidemic affecting 42.4% of people in the United States (US).1 From 1999 to 2018, the rate of obesity drastically increased with over 1 in 10 adults now considered to have severe obesity with a body mass index (BMI) ≥ 40 kg/m2.1 Obesity and its accompanying comorbidities, such as hypertension, diabetes, coronary artery disease, and multiple types of cancer,1, 2 are inevitably associated with overall worse health and surgical outcomes.1 Pancreatic disease is no exception, as obese patients have been described to be at increased risk for both benign and malignant pancreatic conditions.2, 3
Patient outcomes after pancreatoduodenectomy continue to improve with current 30-day and 90-day morbidity and mortality approximating 30–50% and 1.5–5%, respectively.4 The effect of obesity on morbidity, including higher rates of infections, risk of reintubation, and longer length of hospital stay, has been described previously for patients selected for gastrointestinal operations.5, 6 Given the observed increase in postoperative complications among gastrointestinal surgery patients with obesity, an increased risk of complications would be expected in obese patients selected for pancreatoduodenectomy. However, to date, outcomes data are mixed with a number of published studies describing no difference in postoperative outcomes between obese and non-obese patients,2, 7, 8 while other studies highlight worse outcome measures among the obese.9–14
Given this equipoise further investigation aimed to understand the impact of obesity on postoperative outcomes among patients selected for pancreatoduodenectomy is warranted. In this study we aim to examine associations between obesity and postoperative outcomes including morbidity and mortality, as well as procedure specific complications - delayed gastric emptying (DGE), and postoperative pancreatic fistula (POPF), in patients undergoing pancreatoduodenectomy. We hypothesized that obesity and severe obesity are associated with significantly increased postoperative morbidity and mortality after pancreatoduodenectomy. Identification and quantification of risk associated with obesity in this patient population can improve significantly preoperative counseling including risk assessment of complications, as well as, allow for treatment adjustment and sequencing in individually selected patients in whom consideration for preoperative weight management strategies prior to operation could be pursued.
METHODS
Study Population
All patients age 18 and older who underwent pancreatoduodenectomy and were included in the 2014–2018 American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) Procedure Targeted Pancreatectomy Participant Use Data Files(PUF) were abstracted and merged with ACS NSQIP PUF for each corresponding year and included in this retrospective cohort study. The current National Institutes of Health (NIH) definition categorizes obesity as class 1 (BMI 30–34.9 kg/m2), class 2 (BMI 35–39.9 kg/m2), and class 3 (BMI ≥ 40.0 kg/m2).15 Accordingly, patients were stratified into three BMI categories: non-obese (BMI 18.5–29.9 kg/m2), class 1 or 2 obesity (BMI 30.0–39.9 kg/m2), and class 3 severe obesity (BMI > 40.0 kg/m2). Patients who preoperatively had a BMI < 18.5 kg/m2 were excluded as well as patients with certain preoperative conditions: open/infected wounds, systemic inflammatory response syndrome, sepsis, shock, disseminated cancer, American Society of Anesthesiologists (ASA) Class 5, ventilator dependence, emergency operative case, or transfusion of ≥ 1 unit of packed red blood cells in the 72 hours prior to surgery. Clinical variables in the PUF and Targeted Pancreatectomy datasets were merged for data analysis.
Study Variables
The primary objective of this study was to compare postoperative morbidity and mortality among patients within the three BMI categories. Secondary outcomes included comparisons of POPF and DGE among patients in each of these BMI categories. Patient demographics included age, sex, race (White, Black, other/unknown), and ethnicity (Hispanic, non-Hispanic/unknown). Patient comorbid conditions included diabetes (with or without insulin dependence), congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), hypertension (HTN), smoking, dialysis dependence, steroid use, dyspnea, ascites, functional status (independent, dependent), > 10% weight loss in the last six months, and ASA classification. Overall ACS NSQIP defined probability of morbidity and mortality were additionally abstracted to define calculated preoperative risk of morbidity and mortality in patient groups. Pancreatectomy-specific variables abstracted from the Targeted Pancreatectomy dataset included pathologic diagnosis (categorized as pancreatic adenocarcinoma, chronic pancreatitis, IPMN-invasive, IPMN non-invasive, mucinous cystic neoplasm, neuroendocrine neoplasm, ampullary carcinoma, cystadenocarcinoma, distal cholangiocarcinoma, duodenal carcinoma, serous cystadenoma, solid pseudopapillary neoplasm, and unknown histology), pancreatic gland texture (characterized as soft, intermediate, or hard), main pancreatic duct size (categorized as < 3 mm, 3–6 mm, or > 6 mm). “Other diagnoses” category included IPMN without invasive component, neuroendocrine neoplasms, mucinous cystic neoplasms, ampullary carcinoma, cystadenocarcinoma, distal cholangiocarcinoma, duodenal carcinoma, serous cystadenoma, solid pseudopapillary neoplasm, and unknown histology.
Outcome measures including morbidity, mortality, DGE, and POPF were limited to the 30-day ACS NSQIP follow-up period. Morbidity was defined as a composite variable including surgical site infection (superficial, deep, organ space), systemic sepsis, dehiscence, pneumonia, unplanned intubation, failure to wean from ventilator after > 48 hours, deep vein thrombosis/thrombophlebitis (DVT), pulmonary embolism (PE), progressive renal insufficiency, acute renal failure, urinary tract infection (UTI), cerebrovascular accident with deficits (CVA), myocardial infarction, and cardiac arrest. Additionally, composite organ-specific complications were developed including: 1) cardiovascular complications, defined as a cardiac arrest, myocardial infarction, or cerebrovascular accident; 2) venous thromboembolism (VTE) including both DVT and PE; 3) infectious complications including superficial, deep, organ space infections, UTI, and pneumonia.
The primary definition of POPF used in this study was any grade BL (biochemical leak), B, or C POPF according to the International Study Group for Pancreatic Surgery (ISGPS) definition.16 Additionally, as a sensitivity analysis, a more restrictive definition of POPF was used as a clinically-significant grade B or C POPF. Grade B or C POPF was categorized using ACS NSQIP Targeted Pancreatectomy definitions including any clinically significant or persistent drainage that required percutaneous drainage, reoperation, drain placement for longer than 7 days, or initiation of total parental nutrition according to the ACS NSQIP Targeted Pancreatectomy dataset definition aligns with International Study Group for Pancreatic Surgery (ISGPS) definition.16 While ISGPS definition uses a 21 day drain cut-off, ACS NSQIP abstracts POPF definition based on 7 day cut-off period. DGE was defined as nasogastric tube (NGT) usage and/or reinsertion and/or inability to tolerate solid foods after postoperative day fourteen according to the ACS NSQIP Targeted Pancreatectomy dataset definition. Other outcome measures included return to the operating room, hospitalization exceeding 30 days, and readmission within 30 days.
Statistical Analysis
Data were summarized as median (interquartile range) or frequency (percentage). Continuous variables were compared with the Wilcoxon rank-sum test. Categorical variables were compared with chi-squared test. A series of four multivariable logistic regression models were developed to test associations between patient specific factors and the outcome measures of composite morbidity, mortality, DGE, and POPF. Patient demographics and comorbidities that were significant (p < .05) during bivariable analysis were included in all risk-adjusted models. Additionally, surgical drain use, duct size, and gland texture were included in the risk-adjusted POPF model. Age was modeled as a categorical variable: < 65 years old, 65–74, 75–84, or ≥85. All covariates included in risk-adjusted models for morbidity, mortality, and DGE had <1% missing data. For the POPF model, missing data was present for pancreatic duct size (20.2%) and pancreatic gland texture (23.6%). A complete case analysis was performed in computation of each model; patients with missing duct size and gland texture variables in POPF model were excluded from model analysis. A sensitivity analysis for POPF model was performed in which missing pancreatic duct size and gland texture variables were included into the model as “unknown” category. Categorization of these missing data as “unknown” did not alter model performance and did not significantly change effects of obesity and severe obesity in the POPF model.
Four multivariable logistic regression models were developed in an exploratory analysis of only pancreatic adenocarcinoma patients to test associations between patient specific factors including neoadjuvant chemotherapy and radiation and outcome measures of composite morbidity, mortality, DGE, and POPF. The same criteria for establishing significance (p < .05) in the primary risk-adjusted models was used.
Statistical significance was determined by an alpha level of < 0.05 with two-sided hypothesis testing. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC). The University of Virginia Institutional Review Board classified this study as exempt from review (#22540).
RESULTS
Demographic and preoperative factors
A total of 16,823 patients were included in the study and categorized into three BMI groups: 12,234 as non-obese, 4,030 as class 1/2 obese, and 559 as class 3 severe obesity. Demographic characteristics and preoperative comorbid factors in the three patient groups are summarized in Table 1. Patients with class 3 severe obesity, compared to those without obesity and class 1 or 2 obesity, tended to be younger, of female sex, of Black race, and of non-Hispanic ethnicity (all p < 0.01). Patients with severe obesity had significantly greater proportions of comorbid conditions including insulin-dependent diabetes, CHF, smoking, dyspnea, hypertension, weight loss, and ASA class ≥3 (all p < 0.01). The ACS NSQIP probability of morbidity was greater in patients with severe obesity compared to non-obese patients (34.5% vs. 27.2%, p < 0.001).
Table 1.
Demographic characteristics and preoperative comorbid conditions
| BMI 18.5–29 | BMI 30–39 | BMI ≥ 40 | p-value | |
|---|---|---|---|---|
| (n = 12,234) | (n = 4,030) | (n = 559) | ||
| Age, years, median (IQR) | 67.0 (59.0–74.0) | 65.0 (57.0–71.0) | 61.0 (53.0–68.0) | <.001 |
| Categorical age, years | <.001 | |||
| <65 | 4997 (40.9) | 1949 (48.4) | 353 (63.2) | |
| 65–74 | 4361 (35.7) | 1509 (37.4) | 154 (27.6) | |
| 75–84 | 2586 (21.1) | 542 (13.5) | 50 (8.9) | |
| ≥ 85 | 290 (2.4) | 30 (0.74) | 2 (0.36) | |
| Male | 6746 (55.1) | 2142 (53.2) | 217 (38.8) | <.001 |
| Race | <.001 | |||
| White | 9262 (75.7) | 3159 (78.4) | 437 (78.2) | |
| Black | 847 (6.9) | 382 (9.5) | 73 (13.1) | |
| Other/unknown | 2125 (17.3) | 489 (12.2) | 49 (8.82) | |
| Hispanic ethnicity | 577 (4.7) | 211 (5.2) | 17 (3.0) | 0.005 |
| BMI, kg/m2, median (IQR) | 25.1 (22.7–27.3) | 32.8 (31.3–35.2) | 43.3 (41.2–46.3) | <.001 |
| Diabetes | <.001 | |||
| Yes, insulin | 2903 (23.7) | 1366 (33.9) | 211 (37.8) | |
| Yes, non-insulin | 9331 (76.3) | 2664 (66.1) | 348 (62.3) | |
| CHF | 33 (0.27) | 20 (0.5) | 5 (0.89) | 0.008 |
| Smoking | 2248 (18.4) | 579 (14.4) | 61 (10.9) | <.001 |
| Dyspnea | 498 (4.1) | 273 (6.8) | 69 (12.3) | <.001 |
| Hypertension | 6028 (49.3) | 2606 (64.7) | 386 (69.1) | <.001 |
| >10% weight loss in last 6 mo | 2040 (16.7) | 413 (10.3) | 33 (5.9) | <.001 |
| ASA class | <.001 | |||
| 1 | 50 (0.47) | 7 (0.17) | 2 (0.36) | |
| 2 | 2917 (23.8) | 748 (18.6) | 55 (9.8) | |
| 3 | 8499 (69.5) | 3010 (74.7) | 455 (81.4) | |
| 4 | 752 (6.2) | 262 (6.5) | 46 (8.2) | |
| Dialysis dependent | 29 (0.24) | 9 (0.22) | 2 (0.36) | 0.83 |
| Steroid use | 305 (2.5) | 101 (2.5) | 7 (1.3) | 0.17 |
| COPD | 476 (3.9) | 170 (4.2) | 27 (4.8) | 0.39 |
| Ascites | 29 (0.24) | 12 (0.3) | 0 (0) | 0.39 |
| Independent functional status | 12122 (99.1) | 3992 (99.1) | 555 (99.3) | 0.52 |
| Probability of morbidity, % | 27.2 (22.7–32.6) | 29.6 (25.1–35.3) | 34.5 (29.6–40.7) | <.001 |
| Probability of mortality, % | 1.2 (0.64–2.38) | 1.0 (0.54–1.9) | 1.0 (0.55–2.0) | <.001 |
Data presented as No. (%) unless otherwise specified.
BMI = Body Mass Index, CHF = Congestive Heart Failure, ASA = American Society of Anesthesiologists, COPD = Chronic Obstructive Pulmonary Disease, IQR= interquartile range
Pathologic diagnoses and operative factors are summarized in Table 2. Patients with severe obesity had longer operating times compared to obese and non-obese patients (p < 0.001). Patients with severe obesity also had the highest proportions of soft pancreatic gland texture (45.8%, p < 0.001), <3 mm pancreatic duct size (32.4%, p < 0.001), and operative drain use (93.0%, p < 0.001).
Table 2.
Clinical characteristics and operative factors
| BMI 18.5–29 | BMI 30–39 | BMI ≥ 40 | p-value | |
|---|---|---|---|---|
| (n = 12,234) | (n = 4,030) | (n = 559) | ||
| Diagnosis | <.001 | |||
| Pancreatic adenocarcinoma* | 7496 (61.3) | 2146 (53.3) | 270 (48.3) | |
| Chronic pancreatitis | 552 (4.5) | 136 (3.4) | 16 (2.9) | |
| Other diagnoses† | 4186 (34.2) | 1748 (43.4) | 273 (48.8) | |
| Pancreatic gland texture | <.001 | |||
| Hard | 4154 (34.0) | 1136 (28.2) | 144 (25.8) | |
| Intermediate | 1079 (8.8) | 379 (9.4) | 41 (7.3) | |
| Soft | 4076 (33.3) | 1589 (39.4) | 256 (45.8) | |
| Unknown | 2925 (23.9) | 926 (23.0) | 118 (21.1) | |
| Pancreatic duct size | <.001 | |||
| <3 mm | 2775 (22.7) | 1122 (27.8) | 181 (32.4) | |
| 3–6 mm | 5223 (42.7) | 1691 (42.0) | 230 (41.1) | |
| >6 mm | 1733 (14.2) | 415 (10.3) | 60 (10.7) | |
| Unknown | 2503 (20.5) | 802 (19.9) | 88 (15.7) | |
| Operating time, min, median (IQR) | 351 (276–436) | 377 (294–464) | 396 (301–490) | <.001 |
| Drain(s) | <.001 | |||
| Yes | 10717 (87.6) | 3633 (90.2) | 520 (93.0) | |
| No | 1503 (12.3) | 388 (9.6) | 39 (7.0) | |
| Approach | 0.23 | |||
| Open | 11215 (91.7) | 3685 (91.5) | 501 (89.6) | |
| Minimally invasive | 1018 (8.3) | 344 (8.5) | 58 (10.4) |
Data presented as No. (%) unless otherwise specified.
BMI- Body Mass Index
IQR- interquartile range
Pancreatic adenocarcinoma: includes invasive intraductal papillary mucinous neoplasm (IPMN)
Other diagnoses: IPMN without invasive component, neuroendocrine neoplasms, mucinous cystic neoplasm, ampullary carcinoma, cystadenocarcinoma, distal cholangiocarcinoma, duodenal carcinoma, serous cystadenoma, solid pseudopapillary neoplasm, unknown histology
Postoperative outcomes
Unadjusted postoperative outcome measures are summarized in Table 3. Composite morbidity was significantly greater in patients with severe obesity compared to obese and non-obese patients (41.0% vs. 38.2% vs. 32.5% p < 0.001). Mortality was greatest in obese patients, followed by patients with severe obesity and lowest among non-obese patients (2.2% vs. 1.8% vs. 1.3%, p < 0.001). Both DGE (20.2%) and POPF (29.3%) were highest among patients with severe obesity, followed by patients with obesity, and lowest among patients without obesity (p < 0.001). Similarly, development of grade B or C POPF were highest among patients with severe obesity (27.9%), followed by patients with obesity (21.8%), and lowest among patients without obesity (14.5%), p < 0.001. Composite cardiovascular complications did not differ by BMI group (2.2% vs 2.6% vs 3.2%, p=0.18). Patients with obesity and severe obesity had greater proportion of VTE (3.4% vs 4.4% vs 5.6%, by BMI group, p<0.001). Similarly, patients with obesity and severe obesity had greater proportion of infectious complications (24.8% vs 30.4% vs 35.6%, p<0.001). Thirty-day readmission was greatest among patients with severe obesity (21.5%) compared to obese patients (19.0%), and non-obese patients (15.9%), p < 0.001.
Table 3.
Patient outcomes after pancreatoduodenectomy
| BMI 18.5–29 | BMI 30–39 | BMI ≥ 40 | p-value | |
|---|---|---|---|---|
| (n = 12,234) | (n = 4,030) | (n = 559) | ||
| Superficial SSI | 937 (7.7) | 352 (8.7) | 57 (10.2) | 0.014 |
| Deep SSI | 155 (1.3) | 64 (1.6) | 21 (3.8) | <.001 |
| Organ space SSI | 1,699 (13.9) | 787 (19.5) | 121 (21.7) | <.001 |
| Sepsis | 1,088 (8.9) | 422 (10.5) | 58 (10.4) | 0.008 |
| POPF | 1,894 (15.5) | 934 (23.2) | 164 (29.3) | <.001 |
| Grade B or C POPF | 1,775 (14.5) | 880 (21.8) | 156 (27.9) | <.001 |
| DGE | 1,883 (15.4) | 704 (17.5) | 113 (20.2) | 0.001 |
| Dehiscence | 130 (1.1) | 62 (1.5) | 11 (2.0) | 0.01 |
| Pneumonia | 425 (3.5) | 156 (3.9) | 17 (3.0) | 0.40 |
| Reintubation | 416 (3.4) | 162 (4.0) | 27 (4.8) | 0.05 |
| Failure to wean from ventilator > 48 hrs | 301 (2.5) | 136 (3.4) | 26 (4.7) | <.001 |
| DVT | 315 (2.6) | 128 (3.2) | 21 (3.8) | 0.04 |
| Pulmonary embolism | 126 (1.0) | 65 (1.6) | 12 (2.2) | 0.002 |
| Any VTE* | 410 (3.4) | 179 (4.4) | 31 (5.6) | <.001 |
| Progressive renal insufficiency | 47 (0.38) | 39 (0.97) | 14 (2.5) | <.001 |
| Acute renal failure | 97 (0.79) | 56 (1.4) | 8 (1.4) | 0.002 |
| UTI | 329 (2.7) | 103 (2.6) | 23 (4.1) | 0.10 |
| Any infection† | 3,037 (24.8) | 1,225 (30.4) | 199 (35.6) | <.001 |
| CVA | 33 (0.27) | 14 (0.35) | 2 (0.36) | 0.70 |
| Cardiac arrest requiring CPR | 117 (0.96) | 56 (1.4) | 13 (2.3) | 0.001 |
| Myocardial infarction | 142 (1.2) | 39 (0.97) | 4 (0.72) | 0.40 |
| Any cardiovascular complication‡ | 272 (2.2) | 103 (2.6) | 18 (3.2) | 0.18 |
| Return to OR | 624 (5.1) | 230 (5.7) | 38 (6.8) | 0.09 |
| Any Complication | 3972 (32.5) | 1538 (38.2) | 229 (41.0) | <.001 |
| Length of Stay, days, median (IQR) | 8 (6–12) | 8 (6–13) | 8 (7–14) | <.001 |
| Remained hospitalized after 30 days | 306 (2.5) | 126 (3.1) | 16 (2.9) | 0.10 |
| Readmission within 30 days | 1,940 (15.9) | 764 (19.0) | 120 (21.5) | <.001 |
| 30-day Mortality | 163 (1.3) | 88 (2.2) | 10 (1.8) | <.001 |
Data presented as No. (%) unless otherwise specified.
SSI = surgical site infection, POPF = postoperative pancreatic fistula, DGE = delayed gastric emptying, DVT = deep vein thrombosis, VTE = venous thromboembolism, UTI = urinary tract infection, CVA = cerebrovascular accident, CPR = cardiopulmonary resuscitation, OR = operating room; BMI= Body Mass Index, IQR= interquartile range
Any VTE: composite includes DVT and PE
Any infection: composite includes superficial, deep, and organ space surgical site infection, pneumonia, and UTI
Any cardiovascular complication: cardiovascular composite includes cardiac arrest, myocardial infarction, or cerebrovascular accident
Multivariable models estimating postoperative outcome measures
The results of multivariable morbidity and mortality models are presented in Figure 1. After adjusting for demographics (age, sex, race, ethnicity), preoperative comorbid conditions, and ASA class, both class 1/2 obesity (OR 1.28, 95% CI: 1.18–1.38, p < 0.001) and class 3 severe obesity (OR 1.45, 95% CI: 1.22–1.74, p < 0.001) were independently associated with greater composite morbidity. Similarly, after adjusting for demographics, preoperative comorbid conditions, and ASA class, patients with class 1/2 obesity were at significantly greater risk of death compared to non-obese patients (OR 1.79, 95% CI: 1.36–2.36, p < 0.001). There was a trend towards greater odds of mortality among patients with class 3 severe obesity compared to non-obese patients, though this did not reach statistical significance (OR 1.67, 95% CI: 0.86–3.25, p = 0.13).
Figure 1.
Risk-adjusted logistic regression models for post-operative morbidity and mortality. Adjusted associations of patient demographics and comorbidities, including obesity category, with postoperative morbidity (A) and mortality (B).
Diamonds represent adjusted odds ratio point estimates and error bars represent 95% CIs. CHF = congestive heart failure, ASA = American Society of Anesthesiologists, BMI = body mass index
The results of multivariable POPF and DGE models are presented in Figure 2. After adjusting for demographics, comorbid conditions, pancreatic gland texture, pancreatic duct size, and operative drain use, both obesity class 1/2 (OR 1.49, 95% CI: 1.33–1.67, p < 0.001) and class 3 severe obesity (OR 2.05, 95% CI: 1.61–2.61, p < 0.001) were associated with a higher risk of POPF. Interestingly, greater than 10% weight loss in the six months prior to surgery was associated with a significantly decreased risk of experiencing POPF (OR 0.79, 95% CI: 0.68–0.93, p = 0.004) but was not associated with composite morbidity, mortality, or DGE. A sensitivity analysis restricting POPF to only clinically relevant grade B or C POPF is reported in Supplementary Table 1 and demonstrates similar adjusted associations between obesity and POPF: with obesity class 1/2 (OR 1.48, 95% CI: 1.32–1.67, p < 0.001) and class 3 severe obesity (OR 2.07, 95% CI: 1.62–2.65, p < 0.001) associated with greater risk. After adjusting for demographics, comorbid conditions, and ASA class, patients with class 1/2 obesity (OR 1.16, 95% CI: 1.05–1.28, p = 0.004) and class 3 severe obesity (OR 1.45, 95% CI: 1.16–1.80, p = 0.001) were at higher risk for DGE compared to non-obese patients.
Figure 2.
Risk-adjusted logistic regression models for post-operative pancreatic fistula and delayed gastric emptying. Adjusted associations of patient demographics and comorbidities, including obesity category, with POPF (A) and DGE (B).
Diamonds represent adjusted odds ratio point estimates and error bars represent 95% CIs. CHF = congestive heart failure, ASA = American Society of Anesthesiologists, BMI = body mass index
Pancreatic adenocarcinoma patient subgroup
Among 9,560 patients diagnosed with pancreatic adenocarcinoma: 7,235 were non-obese, 2,063 had class 1/2 obesity, and 262 had class 3 severe obesity. Receipt of neoadjuvant chemotherapy within 90 days of resection was less common among severely obese and obese patients in comparison with non-obese patients (27.1% vs 27.0% vs 31.3%, P = .002); however, receipt of neoadjuvant radiation therapy did not significantly differ among groups (9.9% vs 11.3% vs 13.1%, P = .15).
Morbidity, mortality, POPF, and DGE multivariable models including neoadjuvant chemotherapy and neoadjuvant radiation variables in patients with pancreatic adenocarcinoma are summarized in Supplementary Tables 2–5. Composite morbidity was significantly associated with class 1/2 obesity (OR 1.28, 95% CI: 1.15–1.42, p < 0.001) and class 3 severe obesity (OR 1.40, 95% CI: 1.08–1.82, p = 0.01) (Supplementary Table 2). Similarly, mortality was associated with class 1/2 obesity (OR 1.28, 95% CI: 1.15–1.42, p < 0.001) (Supplementary Table 3) in comparison to non-obese patients. Class 1/2 obesity (OR 1.37, 95% CI: 1.14–1.64, p < 0.001) and class 3 severe obesity (OR 1.57, 95% CI: 1.02–2.41, p = 0.04) were associated with POPF development compared to non-obese patients after adjusting for significant covariates (Supplementary Table 4). Obesity and severe obesity was not associated with greater risk of DGE in this population (Supplementary Table 5).
DISCUSSION
Preoperative obesity and severe obesity are associated with significantly greater 30-day postoperative morbidity including composite morbidity, pancreatic fistula, and delayed gastric emptying. In addition, obesity is associated with significantly greater 30-day postoperative mortality. Not surprisingly, patients with obesity and severe obesity had significantly greater burden of preoperative comorbid conditions. Congestive heart failure, dyspnea, and ASA class ≥ 3 were all greater in patients with obesity and severe obesity. A recently published study, using 2010–2015 ACS NSQIP data, propensity-matched obese patients (BMI ≥ 30) with controls (BMI < 30) undergoing pancreatoduodenectomy. This study demonstrated an association between obesity and higher proportions of postoperative complications: organ space wound infection, failure to wean from the vent longer than 48 hours, renal insufficiency, and septic shock.9 Data in our study support these findings but also significantly expand descriptions of postoperative morbidity to include pancreas-specific postoperative complications (POPF and DGE) and mortality. In addition, our study stratifies obesity into clinically relevant groups to provide granular data to describe outcomes in both patients with class 1/2 obesity (BMI 30–39.9) and patients with class 3 severe obesity (BMI≥40). Greater morbidity in patients with obesity and severe obesity is driven by both pancreas related complications (POPF and DGE) and systemic complications (VTE, infections, and others).
Inconsistencies in definitions of obesity affect interpretation of a number of prior published studies. Some groups have dichotomized obesity with a BMI cutoff of 25 kg/m2 8, 11, 14 – a value too low for patient populations in the United States and Europe. Other studies dichotomized patients at a BMI of 30 kg/m2, 7, 9, 10, 17 without further sub-stratification of patients into obesity and severe obesity. We categorized patients into three clinically meaningful NIH-defined BMI groups to examine impact of obesity and severe obesity in patients selected for pancreatoduodenectomy. Importantly, severe obesity and obesity were independently associated with clinically important postoperative outcomes including composite morbidity, mortality, POPF and DGE. Interestingly associations between severe obesity and worse postoperative outcomes were observed despite overall younger age of patients with severe obesity.
Both obesity and severe obesity were significantly associated with greater risk for development of POPF in this study. Associations between visceral obesity and POPF after pancreatoduodenectomy have been described previously.10–14, 17 Unlike this analysis, most of these studies published previously have focused on single-center outcomes and do not have the benefit of a larger sample size from a national, multi-center database. Pancreatic parenchymal factors including pancreatic duct size and gland texture are established important factors associated with pancreatic fistula development.18 In addition, a recently published study demonstrated that pancreatic depth, frequently greater in patients with obesity and severe obesity, is an important factor associated with greater risk of pancreatic fistula.19 Importantly,in our POPF modeling, we adjusted for both pancreatic gland texture and duct size. After adjusting for expected statistically significant effects of both pancreatic duct size and gland texture, both obesity and severe obesity were independently associated with POPF. Lower risk of pancreatic fistula in patients with pancreatic adenocarcinoma and chronic pancreatitis selected for pancreatoduodenectomy is an established fact in pancreatic surgery. Indeed, in our multivariable models, a composite of other diagnoses requiring pancreatoduodenectomy had a greater adjusted risk of POPF compared to patients with pancreatic adenocarcinoma and chronic pancreatitis.
DGE was also associated with preoperative obesity and severe obesity in our study. Previous investigations evaluating role of obesity in development of DGE have been contradictory. While one single institution retrospective study demonstrated significant positive associations between obesity and DGE,10 another single institution retrospective study did not identify independent associations between obesity and DGE.20 In our retrospective multi-institutional study, DGE was strongly associated with both preoperative obesity and severe obesity, even after adjusting for other significant contributory variables.
Similar to other ACS NSQIP analyses, this study has a number of limitations inherent to registry structure and data collection. Postoperative variable collection is limited to the 30-day period post-resection. While the majority of short-term outcomes such as infectious complications and post-pancreatectomy POPF and DGE occur within first 30 days, 90-day mortality is known to exceed 30-day statistics.4, 21 With an overall low number of patients with BMI ≥ 40 kg/m2 who died (10 of 559 patients), a type II error could have affected measurement of associations between severe obesity and mortality. Another limitation includes missing data for pancreatic duct size and gland texture for approximately 20% of patient cohort. We performed a sensitivity POPF model analysis by including missing duct size and gland texture variables as an “unknown” category. Overall results of this analysis and associations between POPF and obesity / severe obesity did not change significantly. Given multitude of variables which could influence duct size and gland texture, imputation was not pursued; however, given the known predilection for pancreatic fat infiltration among patients with obesity and the significant effects of bothlglan dtextur e/ duct siz evariables an dobesit yvariables,th eeffects in our model are unlikely to change significantly. We also used an overall definition of POPF (including BL, Grade B, and Grade C) as the main outcome measure. A sensitivity analysis examining Grade B and C POPF only was performed without significant differences in data or outcome measures. While ACS NSQIP and ISGPS definitions of POPF align, recording of drain duration variable cut-off at 7 days (ACS NSQIP) rather than 21 days (ISGPS) requires ACS NSQIP data coding to be clinically meaningful. We chose to include patients with a drain present for greater than 7 days for management of fistula, as Grade B POPF, given the likelihood that a drain present after day 7 will remain for the following two weeks to allow fistula to heal. A number of other potentially relevant variables such as patient’s socioeconomic status, surgeon technique, and pancreatic gland depth are not collected by the registry and thus were not available for analysis.
Despite these limitations, this study demonstrates significant associations between preoperative patient obesity and severe obesity and greater post-pancreatoduodenectomy morbidity, mortality, POPF, and DGE. So, is there anything that could be done to mitigate this risk? Obese patients are likely to continue to experience worse outcomes unless conscious efforts are made to investigate strategies to mitigate morbidity and mortality in this population. Prehabilitation has gained significant traction to decrease postoperative complications among patients considered for major abdominal surgery.22 A number of recent systematic reviews examining prehabilitation strategies prior to major abdominal operations have demonstrated significant reductions in overall morbidity by optimizing patient preoperative physical, functional, nutritional, and psychological wellbeing.22, 23 Modern dietician-led intervention techniques can also help with preoperative weight loss among selected group of patients.24 Clinical judgment will be necessary to select patients in whom prehabilitation and/or preoperative weight loss would be medically permissible. While these strategies could be considered in patients with pancreatic adenocarcinoma who are successfully tolerating neoadjuvant multi-modality therapy or patients with pre-malignant neoplasms considered for prophylactic resection, weight loss reduction could be medically prohibitive in a patient with obstructing periampullary malignancy who already had 10% body weight loss and remains severely obese. Notably in patients with pancreatic adenocarcinoma majority of the patients during the study period were treated with upfront resection. As the field of pancreatic cancer treatment continues to expand to neoadjuvant therapy and total neoadjuvant therapy for patients with anatomically and biologically resectable PDAC, opportunities for monitored weight loss reduction strategies should increase. As such, decisions regarding weight loss strategies prior to pancreatoduodenectomy are nuanced, should be performed on a case-by-case basis, and require further investigation.
CONCLUSION
With the increasing rate of obesity in the US, obesity will remain a prominent factor in treatment of surgical patients with pancreatic disease. This study demonstrates that obesity and severe obesity are associated significantly with worse short-term outcomes after pancreatoduodenectomy, and provides impetus for considering active steps to mitigate the burden of postoperative morbidity and mortality in this growing at-risk population. Further investigation and utilization of preoperative counseling, weight management strategies, and vigilant care coordination could improve outcomes in patients with obesity and severe obesity.
Supplementary Material
FUNDING
This study was supported in part by funding from the National Cancer Institute T32 CA163177 award to Courtney M. Lattimore, MD and William J. Kane, MD.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
This study describes worse short-term outcomes, including both overall and pancreas-specific complications, in patients with obesity and severe obesity who had pancreatoduodenectomy compared to non-obese patients. Further investigation and utilization of preoperative counseling, weight management strategies, and vigilant care coordination could improve outcomes in patients with obesity and severe obesity considered for pancreatoduodenectomy.
COI / DISCLOSURES
American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. None of the authors have personal, financial, or other conflicts of interest to disclose.
REFERENCES
- 1.Hales CM, Carroll MD, Fryar CD, Ogden CL. Prevalence of Obesity and Severe Obesity Among Adults: United States, 2017–2018. NCHS Data Brief. 2020(360):1–8. [PubMed] [Google Scholar]
- 2.Berrington de Gonzalez A, Sweetland S, Spencer E. A meta-analysis of obesity and the risk of pancreatic cancer. Br J Cancer. 2003;89(3):519–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Malli A, Li F, Conwell DL, Cruz-Monserrate Z, Hussan H, Krishna SG. The Burden of Systemic Adiposity on Pancreatic Disease: Acute Pancreatitis, Non-Alcoholic Fatty Pancreas Disease, and Pancreatic Cancer. JOP. 2017;18(5):365–8. [PMC free article] [PubMed] [Google Scholar]
- 4.Narayanan S, Martin AN, Turrentine FE, Bauer TW, Adams RB, Zaydfudim VM. Mortality after pancreaticoduodenectomy: assessing early and late causes of patient death. J Surg Res. 2018;231:304–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Mullen JT, Davenport DL, Hutter MM, Hosokawa PW, Henderson WG, Khuri SF, et al. Impact of Body Mass Index on Perioperative Outcomes in Patients Undergoing Major Intraabdominal Cancer Surgery. Annals of Surgical Oncology. 2008;15(8):2164. [DOI] [PubMed] [Google Scholar]
- 6.Ri M, Miyata H, Aikou S, Seto Y, Akazawa K, Takeuchi M, et al. Effects of body mass index (BMI) on surgical outcomes: a nationwide survey using a Japanese web-based database. Surg Today. 2015;45(10):1271–9. [DOI] [PubMed] [Google Scholar]
- 7.Benns M, Woodall C, Scoggins C, McMasters K, Martin R. The impact of obesity on outcomes following pancreatectomy for malignancy. Ann Surg Oncol. 2009;16(9):2565–9. [DOI] [PubMed] [Google Scholar]
- 8.Ito Y, Kenmochi T, Irino T, Egawa T, Hayashi S, Nagashima A, et al. The impact of obesity on perioperative outcomes of pancreaticoduodenectomy. Hepatogastroenterology. 2012;59(120):2618–22. [DOI] [PubMed] [Google Scholar]
- 9.Chang EH, Sugiyama G, Smith MC, Nealon WH, Gross DJ, Apterbach G, et al. Obesity and surgical complications of pancreaticoduodenectomy: An observation study utilizing ACS NSQIP. Am J Surg 2020;220(1):135–9. [DOI] [PubMed] [Google Scholar]
- 10.Ekström E, Ansari D, Williamsson C, Andersson R, Tingstedt B, Aronsson L, et al. Impact of body constitution on complications following pancreaticoduodenectomy: A retrospective cohort study. Int J Surg. 2017;48:116–21. [DOI] [PubMed] [Google Scholar]
- 11.El Nakeeb A, Hamed H, Shehta A, Askr W, El Dosoky M, Said R, et al. Impact of obesity on surgical outcomes post-pancreaticoduodenectomy: a case-control study. Int J Surg. 2014;12(5):488–93. [DOI] [PubMed] [Google Scholar]
- 12.House MG, Fong Y, Arnaoutakis DJ, Sharma R, Winston CB, Protic M, et al. Preoperative predictors for complications after pancreaticoduodenectomy: impact of BMI and body fat distribution. J Gastrointest Surg. 2008;12(2):270–8. [DOI] [PubMed] [Google Scholar]
- 13.Ramsey AM, Martin RC. Body mass index and outcomes from pancreatic resection: a review and meta-analysis. J Gastrointest Surg. 2011;15(9):1633–42. [DOI] [PubMed] [Google Scholar]
- 14.Zou SY, Wang WS, Zhan Q, Deng XX, Shen BY. Higher body mass index deteriorates postoperative outcomes of pancreaticoduodenectomy. Hepatobiliary Pancreat Dis Int. 2020;19(2):163–8. [DOI] [PubMed] [Google Scholar]
- 15.Defining Adult Overweight and Obesity: Centers for Disease Control and Prevention; [Available from: https://www.cdc.gov/obesity/adult/defining.html.
- 16.Bassi C, Marchegiani G, Dervenis C, Sarr M, Abu Hilal M, Adham M, et al. The 2016 update of the International Study Group (ISGPS) definition and grading of postoperative pancreatic fistula: 11 Years After. Surgery. 2017;161(3):584–91. [DOI] [PubMed] [Google Scholar]
- 17.Tsai S, Choti MA, Assumpcao L, Cameron JL, Gleisner AL, Herman JM, et al. Impact of obesity on perioperative outcomes and survival following pancreaticoduodenectomy for pancreatic cancer: a large single-institution study. J Gastrointest Surg. 2010;14(7):1143–50. [DOI] [PubMed] [Google Scholar]
- 18.Martin AN, Narayanan S, Turrentine FE, Bauer TW, Adams RB, Zaydfudim VM. Pancreatic duct size and gland texture are associated with pancreatic fistula after pancreaticoduodenectomy but not after distal pancreatectomy. PLoS One. 2018;13(9):e0203841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Iseki M, Noda H, Watanabe F, Kato T, Endo Y, Aizawa H, et al. A deep pancreas is a novel predictor of pancreatic fistula after pancreaticoduodenectomy in patients with a nondilated main pancreatic duct. Surgery. 2020. [DOI] [PubMed] [Google Scholar]
- 20.Robinson JR, Marincola P, Shelton J, Merchant NB, Idrees K, Parikh AA. Peri-operative risk factors for delayed gastric emptying after a pancreaticoduodenectomy. HPB (Oxford). 2015;17(6):495–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Swanson RS, Pezzi CM, Mallin K, Loomis AM, Winchester DP. The 90-day mortality after pancreatectomy for cancer is double the 30-day mortality: more than 20,000 resections from the national cancer data base. Ann Surg Oncol. 2014;21(13):4059–67. [DOI] [PubMed] [Google Scholar]
- 22.Hughes MJ, Hackney RJ, Lamb PJ, Wigmore SJ, Christopher Deans DA, Skipworth RJE. Prehabilitation Before Major Abdominal Surgery: A Systematic Review and Meta-analysis. World Journal of Surgery. 2019;43(7):1661–8. [DOI] [PubMed] [Google Scholar]
- 23.Lyons NB, Bernardi K, Olavarria OA, Dhanani N, Shah P, Holihan JL, et al. Prehabilitation among Patients Undergoing Non-Bariatric Abdominal Surgery: A Systematic Review. J Am Coll Surg. 2020;231(4):480–9. [DOI] [PubMed] [Google Scholar]
- 24.Griffin SB, Ross LJ, Burstow MJ, Desbrow B, Palmer MA. Efficacy of a dietitian-led very low calorie diet (VLCD) based model of care to facilitate weight loss for obese patients prior to elective, non-bariatric surgery. J Hum Nutr Diet. 2020. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.