An increasing number of patients are being diagnosed with pancreatic neuroendocrine tumors (PNETs) [1,2].
It is unknown whether this change is due to a true increase in incidence, or is the result of advances in cross-sectional imaging and widespread access to high-resolution computed tomography. However, small pancreatic neuroendocrine tumors (<2 cm) represent an increasing percentage of all diagnosed PNETs [2,3].
While malignant lesions (> 2 cm, > 2 mitosis/HPF (High Power) > 2 Ki-67 indices, or with angioinvasion) should be resected by pancreaticoduodenectomy (DP) or distal pancreatectomy (DP), the debate continues. regarding the best management for small (< 2 cm) benign-appearing neuroendocrine pancreatic tumors [4-7].
Those who support formal surgical resection argue that the risk of nodal and distant metastases and recurrence is significant and should be mitigated with formal resection [1,8-10]. In contrast, those who favor observation claim that the risk of malignancy is low and that complications of pancreatectomy are high [11].
An alternative strategy to formal resection or observation is parenchyma-sparing enucleation [12-14]. Recent studies suggest that extensive surgery for these lesions is not associated with prolonged survival, and that enucleation may be associated with reduced operative morbidity [2,13]. Small series have compared the operative outcomes of resection with enucleation, but some reports are conflicting [15,18].
It is unknown whether enucleation is an operation with lower morbidity and how the surgical management of these patients has changed over time. The objective of this analysis is to report practice patterns in surgical management and compare postoperative outcomes of resection and enucleation of small PNETs in North America.
| Methods |
> Patient population
The American College of Surgeons-National Surgical Quality Improvement Program Procedure-Targeted Pancreatectomy Participant Use File was consulted , between 2014-2017. Patients with non-functioning PNETs less than 2 cm in diameter and those classified as T1 or T2 according to the American Joint Committee on Cancer (AJCC), 7th edition, were included [21].
Patients with lymph node involvement were excluded. Patients 18 years of age or older who underwent PD were identified using primary CPT ( Current Procedural Terminology ) codes 48150, 48152, 48153, and 48154.
Those who underwent PD were identified using the primary code CPT 48120. Due to the de-identified nature of the data included in the PUF ( Participant Use File ), this study was considered exempt from review by the Institutional Review Board of the University of Pittsburgh. Medical Center .
> Variables and definitions
The perioperative variables analyzed included the surgical approach (that is, open or minimally invasive). The minimally invasive approach was defined strictly by intention to treat, and included laparoscopic and robotic approaches. The texture of the pancreatic gland was compared, in addition to the operative time and perioperative transfusions.
Postoperative outcomes examined among the three cohorts included: surgical site infection (SSI: superficial, deep, and organ-space), wound dehiscence, pneumonia, unplanned intubation, ventilator dependency > 48 hours, venous thromboembolism (VTE). , progressive renal failure or acute renal failure, urinary tract infection, stroke, cardiac arrest, myocardial infarction, sepsis or septic shock, postoperative pancreatic fistula (PPF), clinically relevant postoperative pancreatic fistula (CR-PPF), delayed gastric emptying (VGR), percutaneous drainage, unplanned reoperation, prolonged postoperative length of hospital stay (LOS), non-discharge to home, unplanned readmission, and mortality within 30 days.
The definition and classification of PPF used was based on that of the International Study Group for Pancreatic Surgery (ISGPS) [22]. As previously described, PFF was defined as persistent drainage of amylase-rich fluid (amylase > 300 U/L on postoperative day 3 or later, biochemical leakage according to the ISGPS) in combination with either a drainage remaining in in place for more than 7 days, need for percutaneous drainage placement, or reoperation; or determined by the attending surgeon in the presence of a drain in place for more than 7 days, spontaneous wound drainage, percutaneous drainage, or reoperation [21].
PFF-CR included the presence of fistula plus DEH ≥ 21 days and drain left in place for at least 14 days, organ-space SSI, sepsis, postoperative drain placement (ISGPS Grade B), or need for reoperation , shock, organ failure or death (ISGPS Graco C) [22,23].
The composite measure of severe morbidity has been described previously and includes organ-space SSI, wound dehiscence, neurological event (stroke, such as > 24 hours, peripheral neurological deficit), cardiac arrest, myocardial infarction (MI), stroke pulmonary disease (PE), ventilator dependence, progressive or acute renal failure, sepsis or septic shock [24].
> Statistical analysis
The resection and enucleation cohorts were compared for multiple demographic variables, comorbidity, operative variables, and postoperative outcomes. For postoperative outcomes, logistic regression and gamma regression models were generated to adjust for baseline differences in age, gender, body mass index (BMI), diabetes, hypertension, ASA ( American Society of Anesthesiologists ) classification, and surgical approach. (minimally invasive vs open). Additionally, a subgroup analysis of the PD, PD, and enucleation cohorts was performed. The threshold for statistical significance was set at a p value ≤ 0.05.
| Results |
> Patient population
During the study period, from 2014 to 2017, 1136 patients were diagnosed with a non-functional PNET less than 2 cm in size, or classified as T1 or T2, and underwent surgery. One hundred and twenty-seven patients (11%) underwent pancreatic enucleation; 297 (26%) to PD; and 712 (63%) to PD. Furthermore, when testing for trend analysis, no change was observed in the percentage of patients undergoing enucleation during the study time period from 2014 (10.4%) to 2017 (11.6%, p = 0.959).
The mean age of patients with enucleation was 58.6 years, and 51.2% were men. The mean age of patients undergoing PD was 60.1 years, and 55.1% were women. The mean age of patients who underwent PD was 57.3 years, and 51.9% were men. In the unadjusted analysis, the mean BMI for patients undergoing enucleation was higher compared to PD (30.5 vs 29.2 kg/m2; p = 0.033) and comparable to PD (30.7 kg/m2 ; p = 0.808).
The ASA score, incidence of diabetes mellitus, chronic obstructive pulmonary disease, hypertension, and tobacco use were comparable between the 3 cohorts. Specifically, the incidence of preoperative non-insulin-dependent diabetes mellitus was comparable between the enucleation (15%) and formal resection cohorts (12.7%; p = 0.483).
In an unadjusted analysis, the percentage of patients with tumors 2 cm or larger in patients undergoing enucleation was 13.9%, compared with those who underwent PD (47.2%) and PD (42. 2%) ( p for both < 0.001 vs enucleation).
> Operative management
The minimally invasive approach was used more frequently in patients who underwent enucleation (52.0%), compared with PD (9.8%; p < 0.001), but not with PD (72.1%; p < 0.001). Of the patients who underwent minimally invasive enucleation (n = 66), 36% underwent robotic enucleation (n = 24), and 64% laparoscopically (n = 42).
Of patients undergoing minimally invasive PD (n = 513), 24% (n = 122) was robotic, and 76% (n = 391) laparoscopic. A greater proportion of minimally invasive surgeries were performed robotically in the enucleation cohort compared to those in the PD cohort ( p = 0.034).
The duration of surgery was, on average, 170 min (SD 83) for enucleation, and was lower compared to PD (362 min; SD 135; p < 0.001) or PD (219 min; SD 98 ; p < 0.001). Similarly, perioperative transfusion requirements were lower for enucleation (1.6%), compared with PD (8.8%; p = 0.002) or PD (5.9%; p = 0.022).
> Postoperative results
Fifteen patients who underwent resection died postoperatively (1.5%), while all 127 enucleated patients survived surgery ( p = 0.058). Overall morbidity was lower after enucleation (36.2%), compared with resected patients (48.7%; p < 0.01).
When analyzed separately, that difference appeared to be driven by the PD cohort, where overall morbidity was 58.6% ( p < 0.001 vs 36.2% for enucleation).
Severe morbidity after PD was 44.5% ( p = 0.08 vs enucleation). Organ-space infections occurred less frequently after enucleation (11.8%) compared to PD (21.0%; p = 0.020). Additionally, the enucleation cohort experienced less SSI overall, compared to PD (24.7% vs 12.6%; p = 0.004).
No statistically significant difference was observed in the incidence of PPF-CR between the enucleated (13.5%) and resected cohorts (PD 20.9%; p = 0.68; PD 12.7%; p = 0.797). However, percutaneous postoperative drainage was performed less frequently after enucleation compared to PD (13.4% vs 22.4%; p = 0.028).
The mean HED was higher in the resected patients (5.7 vs 7.2; p < 0.01) and remained statistically significant when the PD and PD cohorts were compared separately. However, there were no differences in 30-day readmission rates between the enucleation and resection cohorts.
> Multivariable regression analysis
Multivariable and gamma logistic regression models were generated to adjust for baseline differences in age, gender, BMI, diabetes, hypertension, ASA classification, and surgical approach (open vs minimally invasive). A reduction in mean operative time was associated with enucleation compared to PD (mean rate 0.45; p < 0.001).
Enucleation was also associated with a reduced risk of perioperative transfusions, compared with PD (OR 0.22, p = 0.038), and PD (OR 0.21, p = 0.032). Adjusted risk models were not performed due to the rarity of the event. However, differences in the incidence of postoperative outcomes were observed between the three cohorts.
When PD or PD was compared, enucleation was associated with reduced overall morbidity (OR 0.41; p < 0.001 and OR 0.66; p = 0.040).
Furthermore, enucleation was associated with a reduction in SSI of any part (OR 0.48; p = 0.022), delayed postoperative gastric emptying (OR 0.27; p = 0.004), sepsis/shock (OR 0.32; p = 0.033), and placement of a percutaneous drain (OR 0.53; p = 0.046).
No difference was observed in the development of CR-PPF or the need for reoperation between the three cohorts. Finally, patients in the enucleation group were less likely to have prolonged HED (OR 0.44; p < 0.001) when compared to PD; the mean rate for DEH compared to DP was 0.57; p < 0.001, for the cohort of patients undergoing enucleation.
| Discussion |
This study represents the largest analysis to date of postoperative outcomes in patients undergoing surgery for small PNETs.
Compared with formal resection, enucleation of PNETs takes less time, requires fewer transfusions, and has lower postoperative morbidity compared with formal resection. This analysis also supports an early analysis of NSQIP, suggesting that mortality is lower with enucleation compared to resection [25].
Despite these findings, in North America only 11% of operations performed on patients with small PNETs are enucleations, and that percentage has not changed over time. Taken together, enucleation of small TPNEs may be an underutilized strategy.
Small series have compared the results of resection with enucleation. The reported advantages of enucleation include reduction in operative time and transfusion requirements, greater splenic preservation, and reductions in postoperative morbidity, DEH, and development of insulin-dependent diabetes mellitus. However, the findings of some analyzes do not agree with all of these proposed advantages.
In a multi-institutional review of high-volume centers in the US, Susan Pitt et al. found that patients with enucleation and resection had similar operative times, blood loss, overall morbidity, mortality, and HED [13 ].
However, for lesions in the pancreatic head, enucleation was associated with a decrease in operative time, blood loss and DEH, compared to PD ( p < 0.05), and for lesions in the tail of the pancreas, enucleation was associated with improved rates of splenic preservation.
More recently, in a multi-institutional report from the Netherlands, Jilesen et al., reported comparable morbidity between the resection and enucleation cohorts, but the incidence of postoperative insulin-dependent diabetes mellitus was lower in those treated with enucleation [17].
Given the increased risk of post-pancreatectomy insulin-dependent diabetes mellitus in patients undergoing formal resection, patients with non-insulin-dependent diabetes may benefit from enucleation, when feasible.
However, this potential benefit did not appear to play a role in the operative decision making in the present study, given that a greater proportion of patients with diabetes mellitus did not undergo enucleation, compared to formal resection.
A 2016 meta-analysis of 27 studies involving 1,316 patients with PNETs or pancreatic cystic neoplasms found no statistically significant difference in overall morbidity, reoperation rate, and mortality between the patient cohorts after enucleation and post-enucleation. resection of pancreatic tumors [26].
In our contemporary North American cohort of 1136 patients, enucleation was found to be associated with a reduction in operative time, transfusions, overall morbidity, and HED, when compared separately in risk-adjusted models with PD and the PD. Additionally, enucleation was associated with a reduction in severe morbidity, infection at any surgical site, VGR, sepsis/shock, postoperative placement of a percutaneous drain, and postoperative DEH, compared with PD.
Since postoperative outcomes after enucleation could be worse in patients with larger tumors, as a conservative approach to this analysis, patients with larger tumors (T2 tumors) who underwent enucleation were included in the comparison with patients undergoing formal resection.
After excluding patients with lymph node involvement, the percentage of patients with TPNE of 2 cm or greater was higher in the resection cohort compared to enucleation (44% vs 14%). Additionally, tumor size and stage were included as part of the multivariate analysis, so enucleation was associated with reduced overall morbidity compared with PD ( p < 0.01) and PD ( p = 0.04).
Because of the increased incidence of lymph node metastases in patients with tumors larger than 2 cm, formal resection is recommended. Why enucleation was performed in patients with larger tumors in the present study is unknown. However, given the improved outcomes after enucleation, that approach may have been preferred in frail patients, or in those with significant comorbidities.
In the present study, a higher proportion of patients who underwent enucleation had minimally invasive surgery (MIS) compared with PD, but not PD. Similarly, a report by Ore et al. found that minimally invasive enucleation was associated with shorter HFD compared with formal resection [27].
While no difference in postoperative morbidity was identified in the study by Ore et al., a meta-analysis by Guerra et al. reported reduced major morbidity after CMI enucleation (5%) compared with open enucleation ( 11%) [28].
The surgeon’s ability to employ the minimally invasive approach more frequently for patients undergoing enucleation highlights another potential benefit over formal resection. That advantage cannot be underestimated for pancreatic head injuries, where the learning curve for minimally invasive PD is likely considerably steeper than that for enucleation [29–31].
Given the potential inherent advantages of the CMI approach and the potential selection bias associated with performing a minimally invasive approach, the authors chose to include the use of CMI in their multivariable model of patient outcomes [30,32].
Even with this conservative study design, an improvement in multiple postoperative outcomes was observed in the enucleation cohort, including decreased operative time and transfusions, as well as overall morbidity and postoperative HED. However, no differences were observed in the rates of PPF or PPF-CR.
Another area of controversy is whether patients undergoing enucleation are at increased risk of developing PPF. In the meta-analysis by Zhou et al., the incidence of PPF (OR 1.96; 95% CI: 1.35-2.86) and of PPF-CR (OR 2.07; 95% CI: 1. 32-1.34) increased after enucleation [26]. Furthermore, in the already cited study by Pitt et al., it was reported that the incidence of PPF was higher after enucleation ( p < 0.01), but that the grade of the fistula was worse after resection ( p = 0.07) [13].
An analysis of patients from the University of Indiana , however , reported a similar rate of PPF in patients undergoing enucleation (33%) compared to resection (28%), but a subgroup analysis of patients undergoing PD and PD was not carried out [19]. In another study from the National Cancer Institute , Inchauste et al., reported a PPF-CR rate of 27% in patients with enucleation, compared with 20% in patients undergoing formal resection ( p = 0.4).
In the present study, the incidence of PPF or PPF-CR did not differ between the enucleation and resection cohorts in the unadjusted and risk-adjusted analyses. This finding could be related to slight variations in the definition of FPP and FPP-CR used by the NSQIP. However, the definitions used in the NSQIP are based on those of the ISGPS, and have been validated in multiple studies [22,33].
An alternative explanation is that the selection of patients to undergo enucleation has improved over time. A recent study comparing the results of deep enucleation of TPNE (those with less than 3 mm from the pancreatic duct) with standard TPNE (those with more than 3 mm from the pancreatic duct), found significant differences in the incidence of PPF (63 % vs 30%; p = 0.002) and FPP-CR (70% vs 23%; p = 0.0006) [15].
This analysis and others reinforce the importance of intraoperative ultrasound in the decision to perform enucleation. However, a limitation of the present analysis is the lack of availability of information on the use of ultrasound.
The findings of this analysis must be interpreted within the context of the limitations of the study. These limitations are mainly related to the retrospective nature of the study, which may introduce bias, as well as the data available in the NSQIP PUF. The importance of clinical judgment in determining whether to perform enucleation or formal resection, and the factors involved in those decisions, are difficult to capture retrospectively.
While tumor size is not collected as part of the NSQIP, T stage is on the final pathology study. Tumor factors beyond size, including grade, KI-67 proliferation index, presence of regional lymph node involvement, location (head, uncinate, body, or tail), and distance from the pancreatic duct, each play a role. significant in patient selection for a given approach, and are variables not collected in the NSQIP.
Other limitations include the absence of morbidity and mortality at 90 days. Therefore, the reported results may underestimate the true rate of events after enucleation and resection. Furthermore, long-term survival data are not available in the NSQIP.
Finally, the improved outcomes observed in the enucleation cohort may be due to those operations being performed at high-volume centers and/or by experienced surgeons.
However, neither surgeon experience nor hospital case volume are collected as part of the NSQIP and, therefore, these variables were not available for inclusion in the regression models and analyses.
| Conclusion |
In conclusion, enucleation of small TPNEs offers several advantages over formal resection, including reduction in operative time, transfusions, development of postoperative complications, and shortening of postoperative DEH.
Despite these benefits, enucleation is being performed nationally in a minority of patients, and that observation has not changed over time. Taken together, enucleation may be an underutilized strategy and an important way to reduce morbidity in selected patients undergoing pancreatectomy in North America.
