Does Surgical Margin Impact Recurrence in Noninvasive Intraductal Papillary Mucinous Neoplasms? : A Multi-institutional Study

Abstract

Objective:

The relevance of margin positivity on recurrence after resection of intraductal papillary mucinous neoplasms (IPMNs) is poorly defined and represents one reason controversy remains regarding optimal surveillance recommendations.

Methods:

Patients undergoing surgery for noninvasive IPMN at 8 academic medical centers from the Central Pancreas Consortium were analyzed. A positive margin was defined as presence of IPMN or pancreatic intraepithelial neoplasia.

Results:

Five hundred two patients underwent surgery for IPMN; 330 (66%) did not have invasive cancer on final pathology and form the study cohort. Of these, 20% harbored high grade dysplasia. A positive margin was found in 20% of cases and was associated with multifocal disease (P = 0.02). The majority of positive margins were associated with low grade dysplasia. At a median follow-up of 36 months, 34 (10.3%) patients recurred, with 6.7% developing recurrent cystic disease and 3.6% developing invasive cancer. On multivariate analysis, margin positivity was not associated with recurrence of either IPMN or invasive cancer (P > 0.05). No association between margin status and development of recurrence at the margin was found. Only 6% of recurrences developed at the resection margin and median time to recurrence was 22 months. Of note, 18% of recurrences occurred > 5 years following surgery.

Conclusion:

Margin positivity after resection for noninvasive IPMNs is primarily due to low grade dysplasia and is not associated with developing recurrence in the remnant pancreas or at the resection margin. Long-term surveillance is required for all patients, as a significant number of recurrences developed over 5 years after the index operation.

Intraductal papillary mucinous neoplasms (IPMNs) of the pancreas are cystic, mucin-producing lesions that have been diagnosed with increasing frequency in recent years due to improvements in radiographic imaging including computed tomography (CT) and endoscopic ultrasound (EUS).^1,2 These lesions exhibit a wide range of pathology from benign disease to low grade dysplasia, high grade dysplasia, and invasive adenocarcinoma. As IPMNs may harbor premalignant or malignant disease, these lesions represent an opportunity for early detection and prevention of pancreatic cancer.

Given the need for balancing potential malignant progression with the morbidity of pancreatic resection, the management of IPMNs has received significant attention over the past 2 decades. Patient selection and indications for surgical resection have evolved with international consensus guidelines published in 2006 (Sendai criteria) and revised in 2012 and 2017 (Fukouoka criteria).^3–5 Features including cyst location associated with the pancreatic duct (main, mixed, or branch), presence of an enhanced solid component, evidence of a mural nodule, as well as epithelial subtypes have all been implicated as significant factors associated with presence of malignant disease.

Although previous studies have described patterns of recurrence after resection of IPMN, reported recurrence rates remain highly variable and significant heterogeneity exists with regard to study design and patient inclusion criteria.^6–9 Furthermore, while the association between margin positivity and recurrence after resection of invasive disease is well established, the impact of margin status on rates of recurrence following resection of noninvasive IPMNs remains controversial.^10–15 As a result, optimal postoperative surveillance strategies for these patients are subject to debate.^16–19 The objective of this study was to evaluate the impact of surgical margin on recurrence in noninvasive IPMNs using a large, multi-institutional study cohort. Utilizing patient and clinical data obtained from 8 academic medical centers, we aimed to identify predictors of margin positivity and recurrence of IPMN or invasive cancer after resection for noninvasive IPMNs.

METHODS

Study Population

A retrospective, multi-institutional, observational study was performed evaluating patients undergoing surgical resection of IPMN within the Central Pancreas Consortium (CPC) database. The CPC is a collaboration of 8 academic institutions: University of Cincinnati College of Medicine, University of Wisconsin School of Medicine and Public Health, Northwestern University Feinberg School of Medicine, University of Louisville School of Medicine, University of North Carolina School of Medicine, University of Miami Leonard M Miller School of Medicine, Vanderbilt University Medical Center, and Emory University Winship Cancer Institute. Diagnosis of IPMN was determined by pathologic review according to established guidelines by the World Health Organization.

All patients undergoing surgical resection with pathologic confirmation of IPMN were included (n = 502). Of these, patients with invasive cancer or pancreatic adenocarcinoma were excluded from analysis, resulting in a final study cohort consisting of patients with noninvasive IPMNs (n = 330). Institutional Review Board approval was obtained from each participating institution prior to initiation of the study.

Description of Patient Selection

Selection of patients for surgical resection of IPMN was typically determined in accordance with international consensus guidelines including the 2006 Sendai criteria (proposed at the 11th congress of the International Association of Pancreatology [IAP]) and the 2012 Fukouka criteria (revised at the 14th annual meeting of the IAP). In general, all main duct and mixed-type IPMNs were referred for pancreatic resection as were branch-duct IPMNs with “high-risk” and “worrisome features” including pancreatic main duct dilation, cyst size ≥3 cm, and presence of an enhancing mural nodule. Additional diagnostic studies used in the evaluation of these patients included contrast-enhanced CT, MRI, and EUS with or without cyst fluid aspiration, cytology, and biopsy. Surgical intervention was ultimately at the discretion of each participating institution and clinical judgement of the operating surgeons.

Data Variables

Retrospective chart review was performed to capture relevant clinical data from each participating institution. The following patient and disease characteristics were collected: age, gender, race, relevant medical comorbidities, disease presentation, laboratory markers, presence of high-risk or worrisome features based on radiographic and clinical findings, disease focality, surgical approach, margin status, pathologic characteristics, epithelial subtypes, ductal involvement, and degree of dysplasia. Of note, the surgical resection margin was considered positive if any degree of dysplasia or pancreatic intraepithelial neoplasia (PanIN) was identified. Additionally, the following characteristics were obtained regarding disease progression: progression or recurrence of either IPMN or invasive cancer, time to recurrence, and recurrence location. Recurrence was defined as development of disease including cyst recurrence or invasive cancer at the resection margin, new lesion(s) in the remnant pancreas, progression of a pre-existing pancreatic lesion, or distant site metastases.

Statistical Analysis

All statistical analyses were performed using SAS 9.4, SigmaPlot 11.0, and JMP Pro 11 (SAS Institute, Cary, NC) statistical software. Continuous variables were described as estimates of central tendency (median) with interquartile range (IQR) and were analyzed using Wilcoxon rank-sum test. Categorical variables were described as percentages (%) and analyzed using Pearson chi-squared test or Fisher exact test when appropriate. Variables with a P value of less than 0.05 were determined to be statistically significant. Univariate logistic regression was used to identify significant factors to be included in multivariate stepwise logistic regression analysis. A multivariate logistic regression model was performed with all included study patients to identify characteristics associated with margin positivity. Similarly, univariate Cox regression was used to identify associations between independent patient and clinical variables and disease recurrence. Multivariate Cox regression models were then generated using variables found to be significant on univariate analysis to identify predictors of recurrence. Recurrence-free survival (RFS) was estimated with Kaplan–Meier log rank analyses.

RESULTS

Patient Demographics, Disease Characteristics, and Perioperative Details Demographic information and disease characteristics are shown in Table 1. Median age at diagnosis was 67 years with an even distribution of men and women in the study population. The majority of patients were Caucasian (87.0%, n = 287). Median body mass index was 26 kg/m² (interquartile range [IQR] 23–44 kg/m²). At the time of diagnosis, most patients were either asymptomatic (44.0%, n = 155) or presented with generalized abdominal pain (32.1%, n = 106). A majority of tumors were found to be located in the pancreatic head (53.9%, n = 178). “High-risk stigmata” and “worrisome features” were found in 15.2% and 63.3% of the cohort, respectively. Operative details and perioperative outcomes are shown in Table 2. Pancreaticoduodenectomy was performed in 61.8% (n = 204) of patients and distal pancreatectomy in 30.0% (n = 99).

TABLE 1.

Patient Demographics and Disease Morphology

Characteristic	N/Median (%/IQR)
Age (yrs)	67 (59–73)
Sex
Male	162 (49.1%)
Female	168 (50.9%)
Race/ethnicity
Caucasian	287 (87.0%)
Black	25 (7.6%)
Asian	4 (1.2%)
Other	14 (4.2%)
Medical history
ASA	3 (2–4)
BMI	26 (23–44)
Hypertension	196 (59.4%)
Diabetes	86 (26.1%)
Alcohol history	76 (23.0%)
Smoking history	144 (43.6%)
Presentation
Incidental/asymptomatic	145 (44.0%)
Pain (not pancreatitis)	106 (32.1%)
Pancreatitis	49 (14.9%)
Jaundice	15 (4.5%)
Other	15 (4.5%)
Disease location
Head	178 (53.9%)
Neck	14 (4.2%)
Body	65 (19.7%)
Tail	35 (10.6%)
Other/unknown	38 (11.5%)
Disease features
Obstructive jaundice with cystic head mass	16 (4.8%)
Enhanced solid component	10 (3.0%)
Pancreatic duct > 10 mm	28 (8.5%)
Pancreatic duct 5–9 mm	53 (16.1%)
Cyst size > 3 cm	124 (37.6%)
Thickened enhanced cystic wall	5 (1.5%)
Mural nodules	13 (3.9%)
Abrupt caliber change with atrophy	33 (10.0%)
Pancreatitis	66 (20.0%)
Number of high-risk features
0	280 (84.8%)
1	47 (14.2%)
>1	3 (1.0%)
Number of worrisome features
0	121 (36.7%)
1	134 (40.6%)
>1	75 (22.7%)

TABLE 2.

Operative Details for Resected Noninvasive IPMNs

Characteristic	N/Median (%/IQR)
Surgical approach
Open	283 (86.5%)
Laparoscopic	36 (11.0%)
Laparoscopic converted to open	2 (0.6%)
Robotic	8 (2.4%)
Operation type
Total pancreatectomy	16 (4.8%)
Pancreaticoduodenectomy	204 (61.8%)
Central pancreatectomy	10 (3.0%)
Distal pancreatectomy	99 (30.0%)
Enucleation	1 (0.3%)
Length of operation (min)	282 (197–381)
Estimated blood loss (mL)	250 (100–400)
Length of stay (d)	7 (6–11)
30-d readmission	59 (17.9%)
90-d complications	147 (44.5%)
Pancreatic fistula, any grade	44 (13.3%)

Disease Pathology and Margin Status

Pathologic characteristics of resected noninvasive IPMNs are shown in Table 3; 40.1% (n = 115) were found to be main duct type, 34.1% (n = 98) were branch duct type, and 25.8% (n = 74) were mixed type. Multifocal disease was seen in 24.9% (n = 78) of patients. Pathology of resected specimens included lesions harboring low grade dysplasia (80.3%, n = 265) and high grade dysplasia (19.7%, n = 65). Of note, patients diagnosed with intermediate grade dysplasia were included in the low grade dysplasia group as recently described.²⁰ Overall, 19.7% (n = 65) were noted to have a positive surgical margin, defined as presence of IPMN or PanIN. When stratified according to degree of dysplasia at the margin, low grade dysplasia or PanIN 1/2 was present in 17.9% (n = 59) of patients and high grade dysplasia or PanIN 3 was present in 1.8% (n = 6).

TABLE 3.

Pathologic Characteristics for Resected Noninvasive IPMNs

Characteristic	N/Median (%/IQR)
Duct type
Main	115 (40.1%)
Mixed	74 (25.8%)
Branch	98 (34.1%)
Epithelial subtype
Intestinal	31 (9.4%)
Pancreatobiliary	15 (4.5%)
Gastric	65 (19.7%)
Oncocytic	4 (1.2%)
Mixed	5 (1.5%)
Other/unknown	210 (63.6%)
Disease focality
Unifocal	235 (75.1%)
Multifocal	78 (24.9%)
Degree of dysplasia
Low	265 (80.3%)
High	65 (19.7%)
Margin status
Negative	265 (80.3%)
IPMN with low grade dysplasia or PanIN 1/2	59 (17.9%)
IPMN with high grade dysplasia or PanIN 3	6 (1.8%)
Disease recurrence
Recurrence of cyst	22 (6.7%)
Recurrence of cancer	12 (3.6%)
Time to cyst recurrence (mo)	31 (5–57)
Time to cancer recurrence (mo)	21 (13–38)
Recurrence location
Resection margin	2 (5.9%)
New lesion in remnant pancreas	16 (47.1%)
Previous lesion in remnant pancreas	6 (17.7%)
Distant site	10 (29.4%)

Patterns of Recurrence

At a median follow-up of 36 months, 10.3% (n = 34) of patients undergoing resection of noninvasive IPMN developed recurrence; 22 patients recurred with de novo or progressive cystic disease while 12 patients developed invasive cancer. Of the 22 IPMN recurrences, 36% (n = 8) occurred in patients with high grade dysplasia and 64% (n = 14) occurred in patients with low grade dysplasia. Of the 12 cases with invasive recurrence, 42% (n = 5) occurred in patients who were found to have high grade dysplasia in the initial resection specimen and the remaining 58% (n = 7) occurred in patients with low grade dysplasia. Sixty-four percent (n = 14) of IPMN recurrences occurred in patients with main or mixed type IPMN while the remaining 8 occurred in patients with branch duct IPMN. Conversely, 83% (n = 10) of invasive recurrences occurred in patients with main or mixed type IPMN while the remaining 2 occurred in patients with branch duct IPMN.

Timing of Recurrence

Timing and location of recurrences are shown in Figure 1 and Table 3. Median time to disease recurrence was 22 months. Of note, median time to recurrence of noninvasive disease was 31 months compared to 21 months for invasive disease. While a majority of recurrences occurred within 2 years following surgery (55.8%, n = 19), 17.6% (n = 6) of patients developed recurrence over 5 years after resection. 47.1% (n = 16) of patients who developed recurrence were found to have a new lesion within the remnant pancreas, 17.7% (n = 6) developed disease progression within a previously known remnant lesion, and 29.4% (n = 10) developed distant site metastases. Only 2 (5.9%) recurrences were identified at the previous resection margin. Of the 6 patients presenting with “delayed” recurrence over 5 years following resection, 1 was found to have disease progression within a previously known lesion of the remnant pancreas, 3 developed new lesions within the remanant pancreas, and 2 were found to have distant site metastases.

FIGURE 1.

Timing and location of recurrence after resection of noninvasive IPMN.

Margin Status and Recurrence

The fate of patients according to margin status is shown in Figure 2. Recurrences were found in 22 (8.3%) patients with negative margins, 10 (16.9%) patients with low grade dysplasia or PanIN 1/2 at the margin, and 2 (33.3%) patients with high grade dysplasia or PanIN 3 at the margin. Of note, only 1 of the 2 patients who developed recurrence at the resection margin was noted to have a positive surgical margin at index resection, which was found to be high grade dysplasia at the margin. The patient with a negative resection margin who developed margin recurrence was found to have high grade dysplasia within the resected IPMN. Kaplan–Meier RFS curves are plotted in Figure 3. There was no statistically significant difference found with regard to RFS between patients with positive and negative surgical margins (P = 0.06). The 1-year, 5-year, and 10-year RFS for positive and negative margins were 84.9% versus 98.3%, 78.5% versus 88.9%, and 78.5% versus 77.9% respectively.

FIGURE 2.

Fate of the positive margin after resection of noninvasive IPMN.

FIGURE 3.

Comparison of RFS following resection of noninvasive IPMN according to margin positivity (P = NS).

Factors Associated With Margin Positivity

Univariate logistic regression analysis was performed examining factors associated with margin positivity including patient demographics, disease presentation, radiographic cyst characteristics, presence of “high-risk stigmata” or “worrisome features,” duct type (main, mixed, or branch duct), disease location or focality, and highest degree of dysplasia identified within the resected specimen. Statistically significant and clinically relevant characteristics were subsequently included in stepwise multivariate analysis, as shown in Table 4. The only factor that was significantly associated with increased odds of having a positive surgical margin was presence of multifocal disease (OR 2.13, 95% CI 1.12 – 4.04, P = 0.02). Of note, disease features including obstructive jaundice, enhancing solid component, or pancreatic duct dilation greater than 10 mm were not found to be significantly associated with margin positivity (all P > 0.05). Similarly, IPMN duct type (main, mixed, or branch duct) was not associated with margin positivity (P > 0.05).

TABLE 4.

Logistic Regression Model for Predictors of Margin Positivity During Resection of Noninvasive IPMN

Characteristic	OR (95% CI)	P Value
Degree of dysplasia
Low grade	Ref.
High grade	1.39 (0.64–2.99)	NS
Disease focality
Unifocal	Ref.
Multifocal	2.13 (1.12–4.04)	0.02
Duct type
Main	Ref.
Mixed	1.10 (0.67–1.80)	NS
Branch	0.89 (0.43–1.82)	NS

Factors Associated With Disease Recurrence

Analyses evaluating factors associated with disease recurrence were performed using Cox regression models. Similar to previous analyses, factors including patient demographics, disease presentation, radiographic cyst characteristics, presence of “high-risk stigmata” or “worrisome features,” duct type (main, mixed, or branch duct), disease location or focality, degree of dysplasia, and margin status were included in univariate analysis. Due to the relatively small numbers of individual types of disease at the resection margin, all patients with positive margins were included as a single group (IPMN with low grade dysplasia at margin, IPMN with high grade dysplasia at margin, PanIN 1 or 2 at margin, and PanIN 3 at margin). On stepwise multivariate analysis, as shown in Table 5, only presence of high grade dysplasia within the resected specimen was associated with disease recurrence (HR 2.88, 95% CI 1.30–6.39, P < 0.01). In addition to disease presentation, duct type, and disease focality, presence of a positive surgical margin was not found to be significantly associated with disease recurrence (HR 1.38, 95% CI 0.62–3.05, P > 0.05).

TABLE 5.

Cox Regression Model for Predictors of Recurrence Following Resection of Noninvasive IPMN

Characteristic	HR (95% CI)	P Value
Presentation
Incidental/asymptomatic	Ref.
Pain (not pancreatitis)	0.85 (0.37–1.97)	NS
Pancreatitis	0.30 (0.06–1.46)	NS
Jaundice	0.97 (0.20–4.62)	NS
Other	0.75 (0.16–3.52)	NS
Degree of dysplasia
Low grade	Ref.
High grade	2.88 (1.30–6.39)	<0.01
Disease focality
Unifocal	Ref.
Multifocal	1.64 (0.73–3.68)	NS
Duct type
Branch	Ref.
Mixed	1.40 (0.56–3.56)	NS
Main	1.15 (0.45–3.00)	NS
Margin status
Negative	Ref.
Positive	1.38 (0.62–3.05)	NS

DISCUSSION

Over the past several decades, research regarding IPMNs has focused on elucidating the risk of occult malignancy and understanding potential patterns of malignant degeneration associated with these lesions. While rates of recurrence for invasive disease are well reported, literature regarding the prognostic significance of noninvasive IPMNs remains less robust.^{7,9–11,17,21–23} Thus, the objective of the current study was to evaluate the natural history of resected noninvasive IPMNs utilizing a large multi-institutional database. We limited the current study to noninvasive disease because the incidence and timing of recurrence for noninvasive IPMNs remain illdefined. Additionally, because the relationship between margin status and disease recurrence is poorly understood, controversy remains regarding the extent of surgery required to avoid positive margin resections and the need for surveillance in these patients.

In the current report from 8 academic medical centers of the Central Pancreas Consortium, several important findings emerged. First, rates of margin positivity were found to be high at 20%, however, a majority of positive margins harbored IPMN with low grade dysplasia. Second, recurrences were infrequent following resection of noninvasive IPMN, occurring in 10% of patients with a majority being nonmalignant. Third, margin positivity was not associated with recurrence of either IPMN or invasive cancer and no association between margin status and development of recurrence at the margin was found. Finally, although a majority of recurrences occurred in the first 2 years following resection, approximately 18% of patients developed recurrence over 5 years after surgery.

When reviewing previous literature, the frequency of margin positivity depends on both the criteria used to define margin involvement and the patient population included in the study.^13–15,17 Studies evaluating a heterogeneous group of patients with both benign and invasive IPMN likely overestimate rates of margin positivity resulting in reported rates ranging from 10% to 50%.^8,22 In their meta-analysis, Leng et al¹⁰ studied a total of 1101 patients collected from 16 publications and reported a margin positivity rate of 21%. Of note, both invasive and noninvasive IPMNs were included with rates of margin positivity found to be significantly higher in patients with invasive disease. Similarly, in the largest single center series to date, investigators from Massachusetts General Hospital examined 381 cases of resected IPMN and reported a margin positivity rate of 23%.⁷ Again, a higher rate of margin positivity was noted in patients with invasive disease.

Discordance in reported rates of margin positivity is also likely a reflection of variable criteria used to define margin positivity. Series including only the presence of IPMN at the margin report rates ranging from 7% to 15% compared to series including both IPMN and PanIN at the margin ranging from 30% to 50%.^1,24–26 Although independent of each other, presence of either IPMN or PanIN at the margin constituted margin positivity in the current study as both pathologies are commonly accepted as precursor pathways for developing pancreatic ductal adenocarcinoma (PDAC). Previous studies have demonstrated that PDAC may develop in approximately 10% of patients with IPMN.²⁷ Moreover, although IPMN and PanIN have different mutational burdens, an overlap of several mutations including k-Ras, GNAS, and TP53 are seen in both precursor lesions.²⁸ In the current report, the overwhelming majority of positive margins were found to harbor low grade dysplasia. Importantly, none of the positive margins with low grade dysplasia were found to develop recurrence at the margin. We identified 2 patients who developed recurrence at the resection margin, one in a patient with high grade dysplasia at the margin and the second in a patient with a negative margin, but high grade dysplasia in the resected specimen. These data are consistent with recommendations from the Baltimore Consensus Meeting for pancreas precursor lesions in classifying low grade lesions as having no proven clinical significance.²⁰

Rates of recurrence following resection of IPMN also vary significantly depending on the patient population being followed and inclusion of patients with invasive disease.^17,29,30 Recurrence following surgery for noninvasive IPMN is not well studied, but reported rates range from 5% to 20%.^17,29 In a Johns Hopkins experience evaluating 130 patients with noninvasive IPMN, He et al²⁹ reported a recurrence rate of 17% (n = 22) at a median follow-up of 38 months. Three of these patients were found to recur with invasive disease. Similarly, investigators from Indiana University evaluated 153 patients with noninvasive IPMN and reported recurrence in 20% (n = 31) of cases.³⁰ Three of these recurrences were invasive cancers. Similar to both these studies, a majority of recurrences in our cohort occurred in the remnant pancreas while 10 patients recurred at a distant site. Interestingly, 3 of these distant site recurrences occurred in patients who had previously undergone total pancreatectomy, suggesting that pathologic evaluation of resected specimens can be difficult and occult invasive disease may be missed by even expert pathologists. Furthermore, our findings highlight that patients undergoing total pancreatectomy for IPMN may require long-term surveillance and follow-up following surgery.

Several elegant studies evaluating molecular pathways and potential mechanisms for recurrence of IPMN following surgery have been performed.^9,27 Utilizing next-generation sequencing of commonly muted genes in 13 patients who developed disease progression localized to the remnant pancreas, investigators from Johns Hopkins proposed 3 specific pathways for recurrence: 1) dissemination of residual microscopic disease at the resection margin, 2) intraparenchymal spread of neoplastic cells leading to an anatomically separate but genetically related recurrence, and 3) multifocal disease progression from a genetically distinct lesion.²⁷ Similar results were described by Tamura et al⁹ in their analysis GNAS and k-ras mutations in both primary and recurrent lesions. It may also be possible that a fourth mechanism exists involving the development of an anatomically different and clonally unrelated lesion that happens to develop similar mutational burdens.

One of the aims of the current study was to determine risk factors associated with recurrence, specifically evaluating the significance of margin positivity. We found no association between margin status and recurrence, instead only identifying presence of high grade dysplasia on final pathology as a predictor of disease recurrence. Equally importantly, we found no association between duct type, “high risk stigmata,” or “worrisome features” with disease recurrence. These findings are comparable to previous studies reporting no association between disease recurrence and margin status, duct type, or cyst features.²⁹ In contrast, there have been reports suggesting an association between margin positivity and recurrence in patients undergoing surgery for noninvasive IPMN.^7,8,24 In the study from Massachusetts General Hospital, 29 out of 299 patients with noninvasive IPMN recurred.⁷ The authors found that main duct IPMNs often recurred as invasive cancer while branch duct IPMN almost universally recurred as benign disease. Furthermore, Cox regression analysis showed that grade of dysplasia at the resection margin was an independent predictor of recurrence. It is possible that the discordance in literature may be explained by the prevalence of distinct grades of dysplasia at the resection margin. Interestingly, multiple studies demonstrating no association between margin status and recurrence reported that a majority of positive margins involved low grade dysplasia.^25,31,32 In contrast, series with a preponderance of positive margins containing high grade dysplasia demonstrate such an association.

Numerous reports have demonstrated an association between presence of high grade dysplasia within the initial surgical specimen and risk of recurrence, specifically recurrence of invasive IPMN.^29,30 In a Johns Hopkins series of 130 patients with noninvasive disease, all patients who developed recurrence with adenocarcinoma had high grade dysplasia at their initial operation.²⁹ Blackham et al⁶ evaluated 100 patients who underwent surgery for IPMN with high grade dysplasia and at a median follow-up of 35 months, found that 3 out of the 9 recurrences consisted of invasive cancer. In our current series, 65 patients were found to have high grade dysplasia in the initial resection specimen. Of these, 8% (n = 5) developed recurrence of invasive cancer. Conversely, of 265 patients with low grade dysplasia, 3% (n = 7) developed recurrence of invasive cancer. It appears that high grade dysplasia may not only increase the risk of recurrence, but may also serve as a marker for the development of future invasive disease. These findings are congruent with the characterization of high grade dysplasia as a diffuse disease process with a more aggressive tumor biology.

Determining the timing of recurrence was the final objective of this study. Fifty-six percent of recurrences occurred within the first 2 years following initial resection. Of note, median time to recurrence of noninvasive disease was 31 months compared to 21 months for invasive disease. This is consistent with previous literature demonstrating a median time to recurrence of 15 to 18 months for invasive IPMN and a significantly longer time course for benign disease.^6,7,17,29,30 Additionally, we found no relationship between degree of dysplasia and site of recurrence. Patients were equally likely to recur at extra-pancreatic sites versus the remnant gland regardless of the degree of dysplasia. More importantly, we found that 18% of recurrences occurred after 5 years following index resection. One of the current controversies regarding management of IPMN involves recommendations for appropriate duration of surveillance following resection. Recommendations, such as those published by the American Gastroenterological Association, that support discontinuation of surveillance at 5 years are called into question by our findings.¹⁹

There are several important limitations to the current study. Due to the retrospective nature, our findings are subject to selection bias. Furthermore, the study cohort consists only of patients undergoing surgical resection. As a result, the true denominator of patients presenting with IPMN was unknown and the natural history of these lesions was unavailable for review. Additionally, as this was a multi-institutional study including data from 8 different medical centers, indications for resection may have been variable, pathologic analysis may have been dependent upon heterogeneous criteria, and surveillance for recurrence was not standardized. Finally, details regarding management of recurrences were unavailable. Consequently, patients classified as developing recurrence based on radiographic or clinical findings may not have undergone re-resection to obtain pathologic confirmation.

CONCLUSION

In summary, our data suggests that margin positivity during resection for noninvasive IPMN as well as the frequency and timing of recurrence depend on the specifics of the disease being treated. IPMN cannot be thought of as a singular diagnosis as it encompasses a spectrum of differing pathologies. Likely, IPMN with an invasive component should be considered to be at increased risk for having positive surgical margins and developing disease recurrence. Presence of disease at the resection margin highlights the concept that IPMN is a multifocal process associated with a “field effect.” Our data indicate that when noninvasive IPMN are treated, positive margins are often composed of low grade dysplasia which tend not to progress at the margin and usually have an indolent disease course with a low incidence of recurrence in the remnant gland. If the positive margin demonstrates presence of high grade dysplasia, however, the risk of recurrence is increased. We have also demonstrated that disease recurrence can occur over 5 years after resection. Based on these findings, we recommend the following; 1) at the time of surgery, low grade dysplasia or PanIN 1/2 do not require re-resection, 2) high grade dysplasia or PanIN 3 should be resected to achieve a negative margin, 3) monitoring for disease recurrence should be most intense during the first 2 years after resection and should continue past 5 years.

DISCUSSANTS.

Dr Keith D. Lillemoe (Boston, MA):

I rise to congratulate our new member, Dr Ahmad, and his colleagues from the Central Pancreas Consortium for an outstanding contribution and a very well done presentation. This is one of many excellent papers coming from this group, who have both addressed a number of key important topics in HPB surgery and, most importantly, opened the door for multi-institutional collaboration in this area. Following their example, collaboration in this kind of surgery is now at an all-time high leading to a high level of productivity in tackling some of the very challenging and, unfortunately, still unresolved questions in pancreatic surgery.

Today’s paper addresses the question of the impact of margin status on the recurrence of IPMN after resection of a benign lesion. Intraductal papillary mucinous neoplasms are a relatively new diagnosis established well after most of us in the audience graduated from medical school. Thus, we still have a lot to learn about these lesions. Many have described research in the area of IPMNs comparable to blindfolded scientists examining different parts of an elephant with each investigator reporting different findings.

These authors have addressed an important and somewhat controversial area which is the risk and predictors of recurrence of disease following resection of a benign IPMN, usually by either a distal pancreatectomy or pancreaticoduodenectomy. This has been the topic of multiple other series addressed by single institutions, but this is clearly one of the largest, and being multi-institutional, is very important.

In this series, and in others, 10% of patients after resection developed a recurrence, with about one third of those patients developing invasive carcinoma. Margin status alone was not a predictor of recurrence, and only the presence of high-grade dysplasia at the margin status was a statistical predictor based on multivariate analysis. Although these findings are relatively clear, I do have a few questions.

As you know, IPMN is not just a predictor of malignancy arising in an IPMN but also for the development of pancreatic ductal carcinoma. In your 12 cases of invasive carcinoma, how many were ductal cancer versus invasive IPMN?

Next, as you noted in your manuscript, the presence of IPMN and PanIN lesions at the margin, is a bit like mixing apples and oranges. I know your numbers are small, but do you feel either of these findings make a difference in predicting recurrence and alters the management of the patients?

I assume that your reported margin status was based on final permanent sections. Is the routine use of frozen section employed by your groups? And, if so, does your data suggest that further resection at initial resection, including total pancreatectomy, be done in cases of high-grade dysplasia?

Fourth, you focus only on margin status. Do you have any outcomes related to the very common situation of progression of the multifocal cystic side branch IPMNs, which we see so often in these remnant glands?

Your data supports the need for long-term follow-up after resection which flies in the face of the AGA recommendations. How do you and your group follow these patients, and for how long?

Next, very few of your patients underwent pancreas preserving procedures like enucleation or central pancreatectomy. Do you believe these procedures have any role in the surgical management of this disease?

Finally, the Hopkins series published in 2013 reported that family history was the only risk factor for recurrent disease after resection. Did you look at family history in your patient groups?

Congratulations to Dr Ahmad and his colleagues. Thanks for sending me the manuscript in advance, and thank you to the Association for the privilege of the floor.

Response from Syed A. Ahmad:

Thank you, Dr Lillemoe. I’ll see if I can answer your questions in order.

With regards to your first question on whether the pancreatic cancers were de novo cancers or whether the cancers arose in the background of IPMN. As many of us in the audience know, IPMN carries a risk of developing pancreas cancer away from the index lesion itself. The literature would suggest that this occurs around 10% of the time. In our series, we had 12 patients who had recurrence of cancer. Ten of these were at a distant metastatic site. It is unclear whether these lesions arose from a de novo cancer in the remnant gland or whether they arose from a recurrent IPMN cyst. Two of these were in the remnant gland and were associated with an IPMN lesion.

With regards to including both IPMN and PanIN at the margin as the definition for a positive margin, this is one of the controversies that exists in the literature. If we examine the different papers commenting on this topic, about half of them include only IPMN at the margin as a definition of positive margin. The other half includes both pathologies. Those including both IPMN and PanIN report higher margin positivity rates.

The reason we included both for our definition of a positive margin was because these are the 2 precursor pathways for the development of pancreatic cancer. They often exist together and share a very similar genetic mutational burden including KRAS, GNAS, and TP53.

With regards to how our margins were analyzed, these were final pathologic sections. We do perform frozen sections in the operating room and would advocate for re-resection of PanIN 3 or high grade dysplasia at the margin.

With regards to progression of cysts left in the remnant gland, in our series, I would say that occurred about 10%. Overall, when additional cystic lesions that were not resected at the time of the index operation about 10% demonstrated progression over time. Our current follow-up algorithm is to use the * international consensus recommendations. That is, for the first 2 years patients are followed every 6 months with imaging, and then annually thereafter. In addition, based on our own study we recommend indefinite follow-up as recurrences can occur after 5 years.

We do not have any experience with pancreas preserving procedures, although this has been commented on in the literature and is an evolving topic of discussion.

With regards to risk of recurrence, we did not capture family history, although I think this is an important metric that should be recorded and followed, as it may guide us in terms of the extent of surgery to perform. With regards to when to do a total pancreatectomy, this remains a controversial topic. Perhaps it should be those patients with multifocal high grade dysplasia or patients who have an invasive cancer with IPMN and have a genetic background predisposing them for the development of pancreas cancer. For example, patients with multiple first degree relatives, family history of P16, BRCA or PRSS mutations should be considered to undergo a total pancreatectomy at the index operation.

Dr William Nealon (Manhasset, NY):

President Ellison, members and Guests

As I listened to your presentation and reviewed your data I am again struck by the inherent unpredictability of these lesions over time. As I am certain you do as well, I routinely interact with our GI colleagues who complain that they are drowning in patients who have had a diagnosis of pancreatic cysts suspicious for or confirmed to be IPMN—some too small to characterize. They will say that there are simply too many of them to follow. If we juxtapose that fact with your data we have a dilemma. Over time some percent of these potentially ignored lesions can grow in size and become invasive. We are all aware of the fact that the American Gastroenterological Association (AGA) published a position paper declaring that no surveillance was necessary for IPMNs after 5 years. They have been forced to reverse that decision. I am not surprised by your finding in this paper that high grade dysplasia at the margin predicts a poor outcome. I believe all of us would commit to further resection in that circumstance. But the meaningful message from your paper is the fact that seemingly innocent IPMN’s can progress.

So my question to you is how can we use your data and other’s data to assist in managing this potentially very large population of seemingly innocent pancreatic neoplastic cysts?

Response from Syed A. Ahmad:

Thank you Dr Nealon. We do not have data on patients we did not operate on. That is one of the weaknesses of the current study. But of the subset of patients that we did operate on that had cystic lesions left in the remnant pancreas, the answer would be that 10% of them do show progression. And most of them, at least in the follow-up period that we have, did not need surgical intervention. Thus, although they progress, they do not meet consensus recommendation for intervention.

Dr Christopher Wolfgang (Baltimore, MD):

We look to this group to really lead the charge in how to manage IPMN. A lot of good work came out of this group, and I congratulate you on this study.

You pointed out at the beginning there is a lot of confusions about margins and IPMNs, and I agree, and some of the confusion on margins, and actually mixing apples and oranges, is on terminology.

First of all, my bias is that IPMN is a field defect. If one part of the gland is affected, the entire gland is affected. And you use the terminology “recurrence.” Recurrence implies that there is nothing there and then something comes back that you took out. And I wonder if you could just comment on your thoughts of recurrence versus progression, because really when we operate on IPMN, we knowingly leave behind disease in the remnant pancreas, the majority of time.

So how do you handle a disease, known disease left in the remnant? In your terminology of recurrence, does that mean that there was no disease, and now you have clinical disease that developed? How did you handle and manage disease that was there and progressed?

And the second thing if you could comment on, IPMN-associated cancer is pancreatic ductal cancer. It just develops through a different pathway but its ductal cancer. And a lot of the studies are confusing to interpret because when they talk about recurrence, they include invasive cancer. And the pattern of recurrence follows invasive ductal adenocarcinoma. That is very different than benign disease in which there is no risk of systemic recurrence. It is only local gland progression. If you could just comment on how you handled this. Thanks.

Response from Syed A. Ahmad:

Thank you, Chris. First of all, you and your group have written a lot of the seminal papers on this topic and because of this your insight and questions are appreciated.

With regards to terminology, I agree with you that IPMN is most likely a field defect, and when we say recurrence, what we are really describing is radiographic presentation of occult disease that was always there but has now become clinically apparent.

The important question to ask is, what is the natural history of most of these occult lesions and what fraction need surgical intervention? When we first learned of this entity, surgeons were very aggressive about performing surgery based on the potential risk of malignancy. Based on the subsequent data that emerged, we realized that most of these lesions posed little risk of cancer developing and therefore surgeons became more conservative in terms of intervening. Our current consensus guidelines describe this conservative approach.

With regards to the terminology of cancer, again, I think you are absolutely correct. Cancer is cancer. But one difference we should be aware of are the different subtypes that can occur based on epithelial subtypes of IPMN, some of which may have more favorable prognosis. For example, a colloid subtype of cancer arising from a background of IPMN can behave differently than a de novo pancreatic ductal adenocarcinoma.

Dr John L. Cameron (Baltimore, MD):

Very nice presentation. Very nice study. We have known for at least a decade that IPMNs are a field defect involving the whole gland and that there are skip areas. So you would not predict that margins would make any difference on recurrence. And that is what our data showed when it was presented several years ago.

The other thing our data showed is that most of the recurrences in our series, virtually all of them, came closer to 10 years than 5 years after surgery. Now, the routine follow-up for IPMNs at our institution, and I think, at most, is every 6 months for the first 5 years, during which we have had no recurrences, and then once a year for the second 5 years, where the recurrences occurred. Maybe we should reverse that and follow once a year for the first 5 years and then more closely the second 5 years. I wonder what your thoughts are on that.

Response from Syed Ahmad:

I would say the devil is in the details. When we examine the series, you have to always look at the patient population that comprised the series. In our series, for example, we had about 25% of patients that had high grade dysplasia and about 20% of patients who had multifocal disease, all factors that can predict a high risk of recurrence, and timing of recurrence. So I think when we look at patterns of recurrence these details must be accounted for. IPMN is a very heterogeneity entity, and understanding timing and patterns of recurrence is based upon the patient population that make up the series.

I would advocate for closer follow-up the first 5 years because, at least in our own experience, that is when the majority occur. I think the fact that there are some recurrences after 5 years highlights the fact that surveillance needs to continue past 5 years. In our series of 330 patients we evaluated, only 6 recurrences occurred after 5 years. The majority occurred within 2 years.