Peritoneal Metastases After Intraductal Papillary Mucinous Neoplasm Resection: How Common are They?

Abstract

Background:

Peritoneal metastases (PMs) following resection of pancreatic intraductal papillary mucinous neoplasms (IPMNs) are rare. Consequently, prevalence, risk factors, and prognosis are not well known. We reviewed our institution’s experience and published literature to further characterize the scope of this phenomenon.

Methods:

All pancreatectomy cases (556 patients) performed at a tertiary care center between 2010–2020 were reviewed to identify IPMN diagnoses. Patients with adenocarcinoma not arising from IPMN, or a history of other malignancies were excluded.

Results:

78 patients underwent pancreatectomy with IPMN on final pathology at our institution; 51 met inclusion criteria. Of these, there were 5 cases of PMs (4F:1M). Four had invasive carcinoma arising from IPMN and 1 had high grade dysplasia (HGD) at the index operation. Female sex and invasive histology were significantly associated with PM (p<0.05). PM rates by sex were 3% (95% CI: 0.5%−15%) in males and 22% (95% CI: 9%−45%) in females. Rates by histology were 2.9% (95% CI: 0.5–15%) for noninvasive IPMN, and 23.5% (95% CI: 9.5–47%) for invasive carcinoma arising from IPMN. Median interval from surgery to PMs was 7 months (range: 3–13 months).

Conclusions:

PMs following IPMN resection are rare but may be more common in patients with invasive histology. Although rare, PMs can arise in patients with non-invasive IPMNs. Further studies on pathophysiology and risk factors of PM following IPMN resection are needed and may reinforce adherence to guidelines recommending long-term surveillance.

Introduction

Intraductal papillary mucinous neoplasm (IPMN) of the pancreas are a spectrum of premalignant neoplasms originating from pancreatic intraductal epithelial cells. Histologically, they appear as cells with papillary projections.¹ Their estimated incidence and prevalence are about 0.14 and 0.8 per 100,000 respectively.² Given their rarity, our understanding of their natural history and recurrence pattern following resection is still evolving.^3,4

IPMNs are classified into low-grade (LGD), intermediate-grade (IGD), high-grade (HGD) dysplasia, and invasive carcinoma. Furthermore, LGD, IGD and HGD are often collectively categorized as noninvasive IPMN.¹ Few studies have evaluated the rates of local recurrence and distant metastases associated with the different cytoarchitectural IPMN types following resection.^5–7 However, no studies to date have specifically set out to describe the rates of peritoneal metastases (PMs) associated with IPMNs despite case reports of their occurrence.^5–11 Unsurprisingly, no guidelines exist for managing these patients following development of PM.¹²

At our institution, we recently cared for a 68-year-old male who underwent surgical resection for IGD IPMN with focal areas of HGD and presented on interval follow up with PMs first identified on his third surveillance CT imaging studies on a 3–6-month schedule, and subsequently confirmed on diagnostic laparoscopy (Figure 1). As such, we were motivated to estimate the rates of peritoneal metastases in non-invasive IPMN and invasive carcinoma arising from IPMN based on our institution’s experience. We hypothesized that worsening degree of dysplasia may be associated with higher rates of PM.

Figure 1:

(A) Peritoneal nodule (red arrow) noted on diagnostic laparoscopy performed to confirm imaging findings of peritoneal metastases (B) H&E of peritoneal nodule biopsy showing infiltrative atypical epithelial cells forming glands, clusters, and single cells with background mucin production consistent with metastatic adenocarcinoma with mucinous features (Magnification 200X).

Methods

Institutional chart review:

Under IRB protocols (201102712, 201202743 and 201903702) with approved waiver of consent, we analyzed all pancreatectomy cases (556 patients) performed at our tertiary care center between 2010–2020. 78 patients had a diagnosis of IPMN. After exclusion of patients with a history of other cancers (24 patients) and patients with ductal adenocarcinoma not arising from IPMN (3 patients), there were 51 patients available to be included in the analyses. All PM cases were first identified on CT scans with IV contrast. One case had a subsequent positive PET/CT imaging. One case also had an MRI and underwent diagnostic laparoscopy with peritoneal tumor biopsy showing metastatic disease. We extracted information about patient characteristics (age, sex, and history of cancer of other type), tumor characteristics (location, size, histology, duct type, margin status), report of intraoperative mucin spillage, interval time from surgery to discovery of peritoneal metastases, and time from discovery of peritoneal metastases to death. Non-invasive IPMNs included LGD (previously adenoma), IGD (previously borderline dysplasia), and HGD (previously carcinoma in-situ)⁶.

Statistics:

Statistical analyses were conducted using RStudio, Excel, and GraphPad Prism. PM rate was calculated by dividing the number of cases of PM for a histologic category by the total number of IPMNs for that category (e.g., HGD PM rate = number of PM cases with HGD/Number of all IPMNs with HGD). 95% confidence intervals were estimated by Wilson’s method. Association between categorical variables and PM was tested using the Fisher’s exact test while statistical comparison of medians of continuous variables was performed using Mood’s median test. Survival curve comparison was performed using log-rank (Mantel-Cox) test. Significance level was evaluated at 0.05 for all analyses.

Results

There was a total of 51 patients with a diagnosis of IPMN. The median age of the cohort was 67 years (38–84 years), and majority were male (33/51), Caucasian (48/51), underwent a Whipple (30/51), and had a diagnosis of HGD IPMN (18/51). Five patients (9.8%) developed PM in the background of IPMN (Table 1). Diagnosis of PM occurred at 3-, 5-, 7-, 8-, and 13-months following detection of peritoneal-omental tumor implants on CT imaging. Median age of the patients that developed PM was 70 years (56–80), 4 were female, and 3 had a positive margin following resection (2 LGD, 1 invasive carcinoma). None of the patients who developed PM had ascites at the time of initial resection, and only 2 patients underwent endoscopic ultrasound (EUS) guided biopsy prior to resection. Furthermore, intraoperative mucin spillage was not explicitly reported to have occurred during surgical resection for any of these cases.

Table 1:

Characteristics of IPMN patients with peritoneal metastases compared to those without peritoneal metastases following resection.

	Peritoneal Metastases (N = 5)	No Peritoneal Metastases (N = 46)	p-value
Median Age	70 (56–84)	66 (39–84)	0.731
Sex
Male	1(20%)	32 (70%)	0.047
Female	4(80%)	14 (30%)
Median IPMN size (cm)	5.9 (2.7–7.4)	2.5 (0.5–12.8)	0.401
Histology
Noninvasive	1 (20%)	33 (72%)	0.037
Invasive carcinoma	4 (80%)	13 (28%)
Margin status
Negative	2	31	0.267
LGD-IGD	2	7
HGD	0	4
Invasive Carcinoma	1	4
Duct type
Main	1	8	0.249
Branched	1	18
Mixed	1	16
NA	2	4
Index procedure
Whipple	2	28	0.353
Distal pancreatectomy	2	15
Total/near total	1	2
Enucleation	0	1
Intraoperative mucin spillage	0	1
Median interval from surgery to PMs	7 months (3–13 months)
Median time from PM to death	5.5 months (1–12months)
Alive: Dead (1 year from PM)	1:4

Both female sex and invasive histology were significantly associated with PM (p<0.05). Median age, median IPMN size, and margin status did not significantly differ between those who developed PM and those who did not. Rates of PM by sex were 3% (95% CI: 0.5–15%) for males and 22% (95% CI: 9–45%) for females. Rates of PM by IPMN histology were 3% (95% CI: 0.5–15%) for noninvasive IPMN, and 23.5% (95% CI: 9.5–47%) for invasive carcinoma (Table 2). Breakdown of the noninvasive IPMN group into its subcategories resulted in PM estimates of 0% (95% CI: 0–19%) for LGD-IGD, and 5.6% (95% CI: 1–26%) for HGD. Median interval from surgery to PM was 7 months (3–13 months), while the median interval from the diagnosis of PM to death was 5.5 months (1–12 months). Only 1 patient was alive beyond 1 year from surgery. Survival was significantly worse for patients who had PM compared to those without PM (Figure 2). All patients with PM received chemotherapy following diagnosis of PM but continued to have progressive disease. Table 3 summarizes the treatment each patient received, impact on disease status and their corresponding length of survival.

Table 2:

Rates of peritoneal metastases by sex and by histology

Sex	PM rates (95% CI)
Male	3.0% (0.5–15%)
Female	22.2% (9–45%)
Histology
Noninvasive	2.9% (0.5–15%)
Invasive carcinoma	23.5% (9.6–47%)

Figure 2:

Kaplan Meier curves comparing survival in patients (alive at least 30 days from index operation) with noninvasive IPMN who did not develop PM (median survival not reached), patients who had invasive IPMN (median survival not reached) and did not develop PM, and those with IPMN who developed PM (median survival = 16 months). [***p<0.0001, +p<0.05]

Table 3:

Management strategy for patients with peritoneal metastases and impact on disease status and corresponding survival length.

Patient #	Chemotherapy regimen received	Disease status on chemotherapy	Survival from PM to death (months)
1	Gemcitabine alone	Progressed	1
2	Gemcitabine/Abraxane (4 cycles), followed by single agent Irinotecan	Progressed	12
3	Gemcitabine/Nab-Paclitaxel + Pharmacologic Ascorbate (4 cycles, PACMAN Trial), followed by 5-FU/liposomal Irinotecan (12 cycles), followed by FOLFOX (4 cycles)	Progressed	16
4	Telagl enastat (CB-839)/Palbociclib (Phase 1b/2 clinical trial), followed by Gemcitabine/Cisplatin (2 cycles)	Progressed	4
5	FOLFOX (4 cycles)	Progressed	7

Discussion

Estimated rates of PM in pancreatic ductal adenocarcinomas (PDAC) are about 9.1%−11.7% and they are known to have worse prognosis than hepatic metastases.¹³ However, the rates of PM in the setting of pancreatic IPMN resection is not known and has not been explicitly reported to date. From our institutional chart review of patients that underwent IPMN resection, we found an overall PM rate of 9.8% with rates of 0% (95% CI: 0–19%), 5.6% (95% CI: 1–26%) and 23.5% (95% CI: 9.5–47%) for LGD-IGD, HGD, and invasive carcinoma respectively, and PM development was strongly associated with invasive histology at the index resection.

In the largest cohort to date, Hirono et al. (2019) performed a multi-institutional study on recurrence patterns of IPMNs in 1047 patients.⁷ Though not explicitly reported in their study, we analyzed their report and estimated rates of PM for LGD-IGD, HGD, and invasive carcinoma arising from IPMN to be 0.2% (95% CI 0.03–1.1%), 0.9% (95% CI 0.03–2.6%) and 8.5% (95% CI 5.6–13%) respectively. Similarly, Kang et al (2014) performed the largest single institution analyses of recurrence patterns of IPMN.³ We inferred rates of PM in their study to be 0.4% (95% CI 0.07–2.2%) for LGD-IGD, 2.2% (95% CI 0.4–11%) for HGD and 11.7% (95% CI 6.1–22%) for invasive carcinoma arising from IPMN. Of note, our PM rate estimates were higher than those we inferred from the studies by Hirono et al. and Kang et al. The higher rates of PM in our study may reflect differences in patient selection/referral patterns and post-resection surveillance strategy across the various centers. For example, with regards to overall PM rate, majority of the cases in our study had HGD while majority of patients in both reported studies had LGD-IGD. Moreover, if the true rates of PM are at the lower ends of our calculated confidence intervals, then it is not so different from the estimates in both studies. We also found sex-based differences in PM rates in our study. Unfortunately, there was inadequate information to infer PM rates by sex reported in the studies by Hirono et al. and Kang et al. to allow comparison. Interestingly, higher rates of PM have also been reported in female patients who underwent resection of gastric cancer though the mechanism remains unknown.^14,15 As such, future studies are needed to clarify potential drivers of this phenomenon.

A key question that remains unanswered is the mechanism of peritoneal spread particularly for noninvasive IPMNs. It is plausible that a small focus of invasive cancer is missed on initial pathology that may have already metastasized to the peritoneum microscopically.⁷ Presumably, this would occur via translymphatic rather than hematogenous spread given very limited support for the later.^16,17 Given the progressively higher PM rates associated with worsened degree of dysplasia, it is possible that disease progression in the remnant pancreas following initial resection in pancreata prone to higher-grade disease could lead to invasive carcinoma with a greater propensity for peritoneal spread.¹⁸ Nonetheless, direct “seeding” of tumor is generally favored to be the primary mechanism for peritoneal spread.¹⁷ For IPMN, there has been some speculative reports about spillage of mucin as a possible mechanism of peritoneal spread, however, no strong evidential support currently exists.^19,20 Mucin spillage was not explicitly reported in any of our PM patients nor in majority of the literature reviewed. However, unrecognized mucin spillage remains a possible explanation for later PM. Interestingly a study by Yoon et al (2014) comparing rates of PM for patients who underwent EUS guided biopsies of IPMN versus those who did not, did not find any association with PM rates.²⁰ We believe that this is likely because volume and degree of mucin spillage in these cases are negligible compared to the degree encountered during resection which is not uncommon. Certainly, there is dire need for a standardized approach to reporting mucin spillage in IPMN resection to further understand this phenomenon and improve patient outcomes. Technical precaution that can be taken to decrease risk of mucin spillage includes pre-operative planning with triple-phase CT imaging to define the extent of mucinous cystic lesions in the pancreas, intra-operative ultrasound guidance to prevent transecting mucinous cystic lesions, and meticulous dissection to prevent inadvertent entry into mucinous cystic lesions. Careful handling during specimen retrieval may also minimize spillage.

Our report has significant implications in surveillance guidelines. Briefly, the revised Fukuoka, European study group, and American College of gastroenterology (ACG) guidelines recommend that recurrence surveillance patterns for invasive carcinoma arising from IPMN be done similarly to PDAC i.e., imaging every 3–6 months for the first 2 years.^21–23 On the other hand, the American Gastroenterological Association (AGA) recommends surveillance for invasive carcinoma arising from IPMN to be done every 2 years post-resection. Furthermore, while the Fukuoka and European Study groups recommend continued surveillance for noninvasive IPMN (at 2- and 5-year for the former or every year for the latter), the AGA does not. Our report suggests PM rate is similar for invasive carcinoma arising from IPMN and PDAC and supports literature suggesting similar behavior of both diseases.^24,25 As such, it supports surveillance patterns as established by the revised Fukuoka and European guidelines but challenges the AGA surveillance guidelines. Given the high recurrence rates (>30%) that have been described for invasive carcinoma arising from IPMN, recommendations outlined by the revised Fukuoka guidelines are predominantly utilized at our institution.^3,7,24 For resected noninvasive IPMNs, our institutional practice closely resembles the European Study guideline with surveillance CT scan done every 12 months for 5 years, followed by long-term surveillance with MRI/MRCP every 2 years thereafter. If there is any high-grade dysplasia present in the resected IPMN or residual cystic lesions, surveillance would resemble those with invasive IPMN or PDAC. Of note, despite behavioral similarities between invasive carcinoma arising from IPMN and sporadic PDAC, differences in chromosomal aberrations (e.g. loss of chromosome 5q, 6q, and 11q), gene mutations (e.g. KRAS and GNAS mutation), and expression of molecular markers (e.g. p16/CDKN2A) have been reported in both diseases to support classifying them as separate entities.^26,27 Our report also suggests that surveillance recommendations should be different for LGD-IGD and HGD as established by the revised Fukuoka, European, and ACG guidelines with resected HGD patients receiving more intensive follow up than LGD-IGD patients.²¹

Limitations of our study include its small sample size, missing information regarding the IPMN duct type on 6 patients, and its basis on retrospective analyses of a single institution experience. As such, our results may not be generalizable to other centers. However, in the context of the rarity of IPMNs in the US (~0.8 per 100,000 prevalence), and that an even fewer number of patients undergo surgical resection; our sample size is reasonable for a single institution study. Furthermore, to our knowledge, no large databases or prospective studies currently exists to address our research question. In addition, we reviewed published literature to infer PM rates from large single and multi-institutional studies and are assured by the fact that our estimates of PM rates are similar.^3,7,20

Taken together, PM may occur in LGD-IGD IPMN, HGD IPMN and invasive carcinoma arising from IPMNs, and are associated with poor prognosis. Discussions with patients about the rates of this occurrence may help reinforce adherence to current consensus-based surveillance guidelines.