Middle segment pancreatectomy can be safely incorporated into a pancreatic surgeon's clinical practice

Abstract

Middle segment pancreatectomy (MSP) is a new operation where the advantages of parenchymal preservation are counterbalanced by a high postoperative complication rate and unease among surgeons with adopting a new technique. This study reviews our experience incorporating MSP into our clinical practice focusing on the initial 34 consecutive patients operated on by one surgeon at a single institution between 1998 and 2007. Patients were divided into early (initial 17 operations) and late (subsequent 17 operations) groups for analysis. Thirty-one reconstructions were by Roux-en-y pancreaticojejunostomy and three were by pancreaticogastrostomy. Using multiple linear regression and logistic regression, we found no significant differences in performance outcomes (operative time, blood loss, tumor size, margin negative resection rate, pancreatic fistula rate, hospital length of stay, postoperative complications, and hospital readmission rate) between our early and late experience even after adjusting for potential confounding variables (patient demographics, co-morbidities, neoplasm, pancreatitis). The pancreatic fistula rate in this series was 29.4% (10/34) and they were all International Study Group on Pancreatic Fistula (ISGPF) Grade A (60%) or B (40%). In summary, MSP is an operation with a flat learning curve and acceptable morbidity rate that can be safely incorporated as a parenchymal preserving option by pancreatic surgeons in their clinical practice.

Introduction

Operations for benign lesions of the pancreatic neck and body have always posed an interesting dilemma. Standard anatomic resections such as pancreaticoduodenectomy (PD) or extended distal pancreatectomy (EDP) remove significantly more tissue than necessary to achieve negative surgical margins, while enucleation is limited to small (≤2 cm) superficial lesions not involving the pancreatic duct ¹,². Middle segment pancreatectomy (MSP), or the Dagradi-Serio-Iacono operation, is a new procedure tailored to remove the area of abnormality while preserving functioning pancreatic parenchyma ³.

Indications for this operation include benign and premalignant conditions as well as solitary metastases to the pancreas (Table I) ³. Benefits of MSP include a reduction in long-term pancreatic endocrine and exocrine insufficiency, maintenance of upper gastrointestinal tract continuity, and splenic preservation ⁴,⁵. These advantages are counterbalanced by a reportedly higher perioperative complication rate coupled with unease among surgeons in adopting a new operation ¹. The higher complication rate noted with this procedure is due to the fact that many patients who are candidates have a normal pancreas and the operation has two potential sources for pancreatic fistula: the oversewn proximal pancreatic remnant and the distal pancreatic enteric anastomosis (see Figure 1) ⁵,⁶.

Table I. Indications for middle segment pancreatectomy (3).

• Tumors localized to the neck or proximal body of the pancreas where at least 5 cm of distal pancreatic remnant remains after obtaining adequate surgical margins.

• Tumor size between 2 cm and 5 cm where enucleation entails a high risk of injury to the main pancreatic duct (Wirsung's duct).

• Smaller tumors deeply embedded in the gland and not eligible for enucleation (i.e. functioning neuroendocrine tumors).

• Benign or low-grade malignant tumors (serous and mucinous cystadenomas, neuroendocrine tumors, solid pseudopapillary tumors, non-invasive side branch intraductal papillary mucinous neoplasms) in which a conservative resection can be carried out with free margins.

• Benign cystic lesions (lymphoepithelial, dermoid, and hydatid cysts) not suitable for enucleation.

• Solitary metastases to the pancreatic neck and metastatic pancreatic neuroendocrine tumors undergoing multimodality treatment.

• Focal chronic pancreatitis with isolated and short stenosis of Wirsung's duct.

Figure 1. .

Middle segment pancreatectomy with roux-en-y pancreaticojejunostomy reconstruction.

This study describes our initial experience with MSP. It was designed to review the introduction of this procedure into our clinical practice and analyze our results as they relate to defined performance outcomes. This information will aid in determining if a surgical learning curve exists for this procedure, and whether it can be safely implemented into clinical practice.

Methods

We performed an analysis of our prospective pancreatic resection database and cross referenced this with the operating room case database to identify all patients undergoing MSP at Indiana University Medical Center between November 1, 1998 and July 1, 2007. Prior approval for this investigation was obtained through the IUPUI Institutional Review Board (0707-80) and patient records were retrospectively reviewed. To analyze performance outcomes associated with the experience of learning this operation, we identified 34 consecutive patients operated on by a single surgeon Thomas John Howard (TJH). These patients were then divided into early (initial 17 operations) and late (subsequent 17 operations) groups. In all cases, the choice of operation was made by the staff surgeon based on patient characteristics and lesion specific variables. We analyzed patient demographics, preoperative co-morbidities, indications for operation, lesion specific variables, intraoperative variables, postoperative complication rates, and hospital length of stays. Postoperative complications were classified by the grading system of DeOliveira ⁷ and pancreatic fistulas were defined by the International Study Group on Pancreatic Fistula (ISGPF) guidelines ⁸.

Continuous variables were summarized by mean and standard deviation and categorical variables were summarized by frequency and percentage. Two-sample t-test and Fisher's exact test were used to compare continuous and categorical variables, respectively. Wilcoxon rank-sum test was also conducted to compare continuous outcomes. Multiple linear regression and logistic regression were used to relate dependent variables with primary independent variables by adjusting for potential confounding factors. All analysis were conducted using SAS 9.1 (SAS, Inc., Cary, NC).

Results

Over the nine year study period, there were 1332 pancreatic resections performed at the Indiana University Medical Center, of which 38 were MSP (2.9%) done by four surgeons. A single surgeon (TJH) completed 34 of these operations. Of these thirty-four consecutive resections, thirty-one reconstructions were by pancreaticojejunostomy and three were by pancreaticogastrostomy. There were 65% females with a mean age of 49 years (±16 years) and a range from 19 to 75 years (Table II). Preoperative diabetes mellitus was present in 15% of the group while 18% had preoperative pancreatic insufficiency. Significant preoperative co-morbidities were found in 47% of patients including hypertension, diabetes, coronary artery disease, stroke, chronic obstructive pulmonary disease, and hepatitis. Indications for operation were focal chronic pancreatitis in 24 patients and pancreatic neoplasm in 10 patients. Of the pancreatic neoplasms, final surgical pathology identified: Intraductal papillary mucinous neoplasm (IPMN) five, solid pseudopapillary tumor three, adenocarcinoma one, and serous macrocystic adenoma one. The mean tumor size in the neoplasm group was 2.5 cm (±1.6 cm), ranging from 0.7 cm to 4.8 cm. Mean operative time was 244.7 minutes (±58 min) and the mean estimated blood loss (EBL) was 720.9 milliliters (±1143.7 ml). Mean length of hospital stay was 9.2 days (±4.8 days) and the hospital readmission rate was 12%. Postoperative morbidity rate was 47.1%. Using the classification system described by DeOliveira et al. ⁷, complications were graded as: No Complication 53%, Grade I 9%, Grade II 21%, Grade IIIa 18%. There were no Grade IV (life threatening complication) or V (death) complications. The pancreatic fistula rate in this series was 29.4% (10/34) and they were all ISGPF Grade A (60%) or B (40%). While there was a trend toward increased EBL early in our experience, these results can be explained by the higher percentage of patients with focal chronic pancreatitis (88.2% vs. 52.9%) in this group.

Table II. Clinical characteristics and performance outcomes for “early” and “late” patients who underwent middle segment pancreatectomy (MSP).

	Total (N=34)	Early (N=17)	Late (N=17)	p†
Clinical characteristics
% Female	22 (64.7%)	9 (52.9%)	13 (76.5%)	0.28
Age (years)	49.4 (15.5)	49.1 (14.1)	49.8 (17.3)	0.90
Pancreatitis	24 (70.6%)	15 (88.2%)	9 (52.9%)	0.06
Neoplasms (%)	10 (29.4%)	2 (%11.8)	8 (47.1%)	0.06
Co-morbidities	16 (47.1%)	7 (41.2%)	9 (52.9%)	0.73
Interventional radiology*	2 (5.9%)	0	2 (11.8%)	0.48
Performance outcomes
Operative time (min)	244.7 (58.1)	251.6 (44.4)	237.8 (70.0)	0.50
Blood loss (ml)	720.9 (1143.7)	729.4 (627.5)	712.4 (1517.8)	0.06††
Pancreatic fistula	10 (29.4%)	4 (23.5%)	6 (35.3%)	0.71
Hospital length of stay	9.2 (4.8)	9.6 (5.6)	8.7 (4.0)	0.58
Hospital readmission	4 (11.8%)	2 (11.8%)	2 (11.8%)	1
Post-operative complications	16 (47.1%)	7 (41.2%)	9 (52.4%)	0.73

*Need for interventional radiology to treat a post-operative fluid collection.

^†p-values are based on two-sample t-test for continuous variables and Fisher's exact test for binary variables.

^††p-value is based on Wilcoxon rank-sum test due to outliers. For other continuous variables, Wilcoxon rank-sum test was also conducted and the p-values are similar to the t-test.

When patients were stratified and analyzed by the type of pancreatic disease (either neoplasm or focal chronic pancreatitis) that initiated the operation, we again found no significant differences in the performance outcomes measured, namely, operative time, EBL, pancreatic fistula rate, length of hospital stay, margin negative resection rate, tumor size, hospital readmission rate, or number of postoperative complications (Table III). Because these results were stratified by disease type, we chose to divide the 10 neoplasm patients into five early and five late; similarly the 24 patients with focal chronic pancreatitis were divided into 12 early resections and 12 late. While a higher percentage of female patients had neoplasms (100% vs. 50%, p=0.01), this variable was not statistically significant. Overall, we identified a higher postoperative hospital readmission rate in the neoplasm group compared to the chronic focal pancreatitis group (30% vs. 4%). This variable was likely influenced by the higher pancreatic fistula rate in the neoplasm group (40% vs. 25%), but within each disease subset, there were no significant differences in these complications with the level of operative experience.

Table III. Patient characteristics and performance outcomes for “early” and “late” patients stratified by disease type: neoplasm or focal chronic pancreatitis.

	Neoplasm				Focal chronic pancreatitis
	Total N=10	Early N=5	Late N=5	p†	Total N=24	Early N=12	Late N=12	p††
Patient characteristics
Female	10 (100%)	5 (100%)	5 (100%)	–	12 (50%)	5 (41.7%)	7 (58.3%)	0.68
Age (yrs)	51.0 (19.9)	54.6 (19.6)	47.4 (21.9)	0.60	48.8 (13.8)	46.2 (14.4)	51.3 (13.2)	0.37
Interventional Radiology*	1 (10%)	0	1 (20%)	1	1 (4.2%)	0	1 (8.3%)	1
Performance outcomes
Operative time (min)	257.8 (62.7)	249.2 (55.0)	266.4 (75.1)	0.69	239.2 (56.6)	246.1 (48.3)	232.3 (65.2)	0.56
Blood loss (ml)	335.0 (208.2)	330.0 (243.9)	340.0 (194.9)	1††	881.7 (1329.7)	716.7 (628.2)	1046.7 (1800.7)	0.49††
Co-morbidities	5 (50%)	3 (60%)	2 (40%)	1	11 (45.8%)	4 (33.3%)	7 (58.3%)	0.41
Pancreatic fistula	4 (40%)	2 (40%)	2 (40%)	1	6 (25%)	1 (8.3%)	5 (41.7%)	0.16
Hospital length of stay	9.6 (4.8)	10.0 (5.8)	9.2 (4.2)	0.81	9.0 (4.9)	9.3 (5.4)	8.7 (4.6)	0.75
Margin negative resection	8 (80%)	4 (80%)	4 (80%)	1	NA	NA	NA	NA
Tumor size	2.5 (1.6)	2.7 (1.8)	2.2 (1.4)	0.67	NA	NA	NA	NA
Hospital readmission	3 (30%)	2 (40%)	1 (20%)	1	1 (4.2%)	1 (8.3%)	0	1
Number of post-operative complications	5 (50%)	2 (40%)	3 (60%)	1	11 (45.8%)	4 (33.3%)	7 (58.3%)	0.41

^†p-values are based on two-sample t-test for continuous variables and Fisher's exact test for binary variables.

^††p-value is based on Wilcoxon rank-sum test due to outliers. For other continuous variables, Wilcoxon rank-sum test was also conducted and the p-values are similar to the t-test.

*Patients requiring interventional radiologic drainage of postoperative fluid collections.

Using linear or logistic regression, we investigated both continuous outcome variables (operative time, EBL, hospital length of stay) and binary outcome variables (pancreatic fistula, postoperative complications, hospital readmissions) and their rate of change with the experience gained by each successive operation (xth operation = independent variable) (Table IV). We found clinically insignificant declines in operative time (−0.19), EBL (−3.9), and hospital length of stay (−0.029) which changed little when adjusting for age, co-morbidity, and disease type (Model 2). Adjusting for age and co-morbidity in the focal chronic pancreatitis group (Model 3) and age, co-morbidity, and tumor size in the neoplasm group also did not identify any significant differences.

Table IV. Four regression models investigating the effect of each successive operative experience on both continuous and binary outcome variables.

	Model 1		Model 2		Model 3		Model 4
Model †	Parameter*	p	Parameter	p	Parameter	p	Parameter	p
Continuous outcome
Operativet time (min)	–0.19	0.85	–1.06	0.41	–2.1	0.13	0.19	0.96
Estimated blood loss (ml)	–3.9	0.85	6.76	0.79	9.3	0.80	–0.85	0.95
Hospital length of stay	–0.029	0.73	–0.03	0.74	–0.11	0.35	0.15	0.61
Binary outcome
Pancreatic fistula	1.05	0.27	1.03	0.57	1.04	0.55	0.90	0.47
Post-operative complications	1.04	0.28	1.04	0.32	1.04	0.48	0.98	0.85
Hospital readmission	1.00	0.96	0.89	0.20	NA††	NA	NA	NA

*For continuous outcomes, the parameter is the regression coefficient in the linear regression model; for binary outcomes, the parameter is the estimated odds ratio in logistic regression model.

^†Model 1: only include the xth time the operation is performed as the independent variable.

Model 2: adjust for age, co-morbidity and disease type.

Model 3: fit to patients with focal chronic pancreatitis, adjusting for age and co-morbidity.

Model 4: fit to patients with neoplasm, adjusting for age, tumor size and co-morbidity.

^††Data points are too sparse for maximum likelihood estimation.

It is possible that our sample size might not have been sufficient to identify a potential association between operative experience and clinical outcomes. Nevertheless, the parameter estimates shown in Table IV suggest that such associations, even if they were to exist, are not likely to represent a clinically significant change in performance.

Discussion

Benign pancreatic lesions and focal pancreatic duct strictures are being diagnosed more frequently due to the increased use of high resolution cross-sectional imaging (computer tomography (CT) and magnetic resonance cholangiopancreatography (MRCP)), endoscopic retrograde cholangiopancreatography (ERCP), and endoscopic ultrasonography (EUS) to investigate upper gastrointestinal tract symptoms ⁹. Classically, lesions in the head of the pancreas have been treated with PD while lesions in the tail of the pancreas have been treated with distal pancreatectomy. Extended versions of these operations have been applied to resect lesions involving the neck and proximal body of the pancreas. MSP is an operation specifically designed to treat such lesions while sparing normal pancreas, duodenum, spleen, and preserving bilioenteric continuity ³. Although portions of the operation were described by Guillemin and Bessot in 1957 and Letton and Wilson in 1959, its adaptation to its current form and use in the treatment of pancreatic neoplasia was first described by Dagradi and Serio in 1984 ¹,³,⁴,⁵,⁶,¹¹,¹²,¹³,¹⁴. The operation has been performed more commonly in recent years but its narrow indications limit its application to only a small fraction of all pancreatic surgery cases. MSP accounts for less than 3% of pancreatic resections at most high volume centers, including our own ¹,¹².

This study reviews the incorporation of MSP into our clinical practice. It focuses on the initial 34 consecutive patients operated on by a single surgeon and analyzes the results in terms of early and late cohorts. This analysis was conducted to determine how the learning curve associated with adopting a complex new procedure can affect patient outcomes.

Achieving competence in any new operation involves the acquisition of both decision making ability and technical skill ¹⁵. For MSP, decision making occurs in applying the operation to appropriate patients, making intraoperative decisions based on anatomic findings, ensuring a margin negative resection, and identifying and adequately treating postoperative complications. By these metrics, we found no significant differences between our early and late experience. Thirty-three of 34 patients (97%) had benign or premalignant conditions. One patient in this series was operated on for a pancreatic duct stricture which had negative surgical margins on frozen section examination intraoperatively. On permanent histology, a T2N1M0 adenocarcinoma arising from the ductal stricture with negative surgical margins (R₀ resection) was identified. The patient declined further surgery, underwent adjuvant chemoradiotherapy, and developed metastatic disease eight months postoperatively. MSP was not designed as an oncologic operation and should not be used to treat patients with known invasive adenocarcinoma of the pancreas ³,⁴,⁵,⁶,¹²,¹³,¹⁴.

Of the three patients with positive resection margins in this series, all had side-branch IPMN with negative margins on intraoperative frozen section. IPMN was later identified involving small peripheral ducts at the margin of resection on permanent histologic examination. There were no positive margins involving the main pancreatic duct. None of these patients with a positive postoperative margin for IPMN has developed a recurrence or undergone reoperation during follow-up of 51, 41, and five months, respectively. In patients with IPMN who have main duct involvement, the possibility of intraductal extension of the lesion and the high incidence of associated malignancy make MSP a poor choice for resection ¹². For side branch IPMN, intraoperative frozen section analysis of resected lesions is mandatory to exclude malignancy and ensure negative resection margins before the pancreaticoenteric reconstruction is completed. Non-functioning neuroendocrine tumors also present an interesting dilemma because they are difficult to classify as benign or malignant on either preoperative work up or intraoperative frozen section examination ⁶. MSP in this setting carries the risk of performing what would be considered an inadequate cancer operation for a patient found to have a malignant neuroendocrine neoplasm on final pathologic evaluation. This situation would appear uncommon for small neuroendocrine neoplasms (<3 cm) where in the Crippa et al. series, there were no tumor recurrences found in 32 patients following MSP over a mean five year period of follow-up ¹².

When analyzing the technical skill component of achieving competence in an operation, direct observation and grading metrics remain poorly defined and limited to the prospective accrual of data. In a retrospective analysis, surrogates of technical skill which we term performance outcomes are variables that are directly influenced by surgical technique and can be readily measured and quantified. Technical aspects of MSP involve: dissection of the pancreas off of both the superior mesenteric portal venous confluence and proximal splenic vein; oversewing of the proximal (duodenal) pancreatic remnant; and construction of a pancreaticoenteric anastomosis (pancreaticogastrostomy or pancreaticojejunostomy) to the distal pancreatic remnant ³. Using multiple linear regression and logistic regression, we found no significant differences in performance outcomes (operative time, blood loss, tumor size, margin negative resection rate, pancreatic fistula rate, hospital length of stay, postoperative complications, and hospital readmission rate) between our early and late experience performing this operation (Table II), even after adjusting for potential confounding variables (patient demographics, co-morbidities, neoplasm, pancreatitis) (Table III). When analyzing these patient outcomes and their rate of change with the experience gained from each successive operation, we again found no significant differences (Table IV). Graphic representation of these results is a flat regression line which implies a flat surgical learning curve for this operation. These findings are predictable in that the technical skills required for MSP are used by the experienced pancreatic surgeon in both PD and distal pancreatectomy, two of the most commonly performed pancreatic resections ¹⁶,¹⁷.

An inherent limitation of MSP is the risk of pancreatic fistula that is higher than in standard pancreatic resections, presumably due to the fact that there are two cut surfaces of the pancreas to contend with rather than just one as in PD or distal pancreatectomy. Leakage from oversewing the duodenal segment would be similar to a fistula from a distal pancreatectomy, whereas leakage from the distal segment behaves like a pancreaticoenteric anastomosis ¹⁸. Unlike PD however, reconstruction in MSP is performed utilizing a defunctionalized Roux-en-y limb of jejunum. The overall pancreatic fistula rate in this series was 29.4%, with a higher fistula rate in patients with neoplasms (40%) than chronic pancreatitis (25%), presumably due to differences in pancreatic texture. All pancreatic fistulas in this series were ISGPF grade A or B. Our fistula rate is similar to the 23–36% rate reported in most surgical series ¹,⁵,¹²,¹⁴, but is much higher than the 8 and 7.5% rates reported by the groups in Lyon and Heidelberg ¹¹,¹³. These discrepancies may be related to definitions of pancreatic fistula although it is worth noting that in the two centers with low pancreatic fistula rates, reoperation rates were considerably higher at 12 and 5%, respectively, than in other published series suggesting the presence of undrained amylase-rich fluid collections (uncontrolled fistula) requiring operative drainage ¹¹,¹³. Our results with MSP also compare favorably to the largest single institiution experience with distal pancreatectomy (235 patients), where Lillemoe et al showed a postoperative complication rate of 31%, pancreatic fistula rate of 5%, reoperation rate of 6%, and perioperative mortality rate of 0.9% ¹⁹. In a collective review of the literature of 207 patients undergoing MSP compiled by Roggin et al., the overall morbidity rate was 33.3% and the pancreatic fistula rate was 22.2% ¹. Based on these data and our own clinical experience, it appears that the majority of controlled external pancreatic fistulae following MSP are ISGPF grade A or B, do not require reoperation, and can be managed successfully using conservative measures ¹,⁵,¹²,¹⁴.

Determining an individual patient's risk of postoperative diabetes mellitus following pancreatic resection relates to the functional capacity of the underlying gland, the extent of resection, and the length of postoperative observation ¹⁰. While it seems intuitive that the preservation of pancreatic mass would improve long-term pancreatic endocrine and exocrine function ²⁰, it is clear that in most patients with an otherwise normal pancreas, there is an excess of functional capacity ⁶. However, in patients with chronic pancreatitis, underlying gland dysfunction makes postoperative endocrine insufficiency more unpredictable ²¹,²². Our postoperative rate of new onset endocrine insufficiency was 5%. In a collective review by Iacono of 296 patients following MSP reported in the literature, the reported rate of postoperative exocrine insufficiency was 4% and the rate of postoperative endocrine insufficiency was 3% ³. MSP design also preserves the spleen, an organ commonly removed during distal pancreatectomy resulting in a life-long risk of overwhelming post splenectomy sepsis and reduced immune function ²³.

Surgery for chronic pancreatitis has typically focused on either pancreatic head resections, longitudinal duct decompression, or distal pancreatectomy. Some patients with chronic pancreatitis have what has been termed “focal chronic pancreatitis” characterized by a dominant pancreatic duct stricture in the neck or proximal body of the pancreas. This anatomic abnormality can also be identified in symptomatic patients after recovery from a bout of severe acute pancreatitis, presumably the result of main duct stricturing from fibrosis during the healing process ²⁴. Some have argued that in patients with chronic pancreatitis, such focal strictures are epiphenomena and surgical treatment should instead be directed at the pancreatic head, considered to be the “pacemaker” of the disease. Despite these presumptions, endoscopists have been successfully dilating and stenting these strictures with good symptom resolution and improved long-term quality of life ²⁵,²⁶. While MSP seems ideally suited to patients with focal chronic pancreatitis, most series contain very few such patients, likely due to the narrow indications (focal stricture in the pancreatic neck without complications involving adjacent organs, such as duodenal obstruction or common bile duct stenosis) for the application of this operation. In a recent prospective series, Müller et al. reported on 23 patients that underwent MSP for focal chronic pancreatitis and found that in a matched pair's analysis with patients who underwent PD or distal pancreatectomy, there were no significant differences in terms of perioperative morbidity or pancreatic fistula rates ¹³. Careful long term (>5 year) follow-up will be necessary to fully evaluate the role of MSP in chronic focal pancreatitis.

Conclusion

MSP offers a tailored approach to lesions in the neck and proximal body of the pancreas. Increasing evidence suggests that gland preservation leads to lower rates of both pancreatic endocrine and exocrine insufficiency. Although infrequently used due to its narrow indications, this operation can be safely performed without a significant learning curve by experienced pancreatic surgeons.