Serous Cystadenomas

Serous cystadenomas (SCA) were first characterized by Compagno and Oertel in 1978 as microcystic and glycogen rich, and they distinguished these lesions from mucinous cysts (MCNs and IPMNs) (Compagno & Oertel, 1978a). SCAs occur with variable size, and when large (>10 cm) may cause symptoms from local compression. Grossly, these cysts are characterized by septations and thick fibrous walls and have what appear to be innumerable small cysts containing clear, thin fluid. A classic honeycomb appearance is often present, as is a calcified central scar with or without hemorrhage (Fig. 57.1A and B). On imaging, these lesions may appear as a single, large cyst (oligocystic) (Fig. 57.1C and D); if so, they are very difficult to distinguish from mucinous lesions. Because of the fibrous nature of this lesion, a solid component is often visualized; if noted in the presence of the other characteristic findings of SCA, this should not be concerning for malignancy. Histologically these lesions are characterized by a bland, cuboidal epithelial lining without nuclear polymorphism or mitoses. Typically, no mucin is present in the epithelial lining, but glycogen may be detected in the cytoplasm or in the amorphous cellular material.

FIGURE 57.1 Imaging characteristics of serous cystadenoma. A, Classic honeycomb appearance. B, Central calcification. C, Oligocystic serous cystadenoma. D, Large cystic neoplasm with peripheral calcification, a thickened wall, and pancreatic ductal dilation. This patient underwent distal pancreatectomy and splenectomy because of the inability to exclude malignancy. Final diagnosis was serous cystadenoma.

SCAs are generally considered benign lesions, and only 25 cases of serous “cystadenocarcinoma” have been reported in the world literature as of 2010 (King et al, 2009; Matsumoto et al, 2005). Many of these case reports describe the “malignant” serous cystadenoma as locally invasive, and only 9 (36%) of 25 cases had evidence of metastasis. Despite the presence of metastasis, long-term prognosis remains excellent in these patients; it appears that the true incidence of malignancy within SCA is less than 1%. The institutional database at Memorial Sloan-Kettering Cancer Center (MSKCC) includes approximately 200 patients who have had resection for SCA, and none has developed or presented with metastatic disease. The more common problem caused by these lesions is local invasion that results in symptoms such as early satiety, obstructive jaundice, and pain. It stands to reason that large lesions are more likely to produce symptoms; however, many patients with lesions greater than 10 cm in diameter will have their lesion identified incidentally; for example, with a mean diameter of 11 cm, 71% of patients had symptoms compared with a mean diameter of 5 cm, with which 53% had symptoms (Compagno & Ortel, 1978a; Tseng et al, 2005).

Mucinous Cystic Neoplasms

Mucinous cystic neoplasms (MCNs) are mucin-producing cystic tumors that lack communication with the pancreatic duct and contain mucin-producing columnar epithelium. Ovarian-like stroma surrounding the columnar epithelium is considered a pathognomonic finding and is the presumed reason that MCNs are almost exclusively found in females. The stroma is a characteristic that pancreatic MCNs share with some mucinous cysts of the ovary and the liver. Pancreatic MCNs are most commonly found in the body and tail and can range in size from small (2 cm) to large (25 cm). Despite their distal location, patients are often symptomatic (76% in a review of multiple series) at the time of diagnosis (Goh et al, 2006).

Any macrocystic lesion in the distal pancreas of a female patient should be highly suspicious for MCN. These lesions are often unilocular but may contain septa within the primary lesion (Fig. 57.2A to C). When these lesions are multilocular, they are typically macrocystic, in contrast to serous cystadenomas. These lesions may also have peripheral “eggshell” calcifications. Any evidence of mural nodularity in these lesions should be concerning for invasion (Fig. 57.2D). Upon gross inspection, these tumors are round, with a smooth surface and fibrous pseudocapsule.

FIGURE 57.2 Imaging characteristics of mucinous cystic neoplasm. A, Unilocular lesion in the tail of the pancreas in a young female. B, Peripheral “eggshell” calcifications. C, Oligocystic lesion in tail of pancreas. D, Large macrocystic lesion with central calcification and mural nodularity.

Unlike serous cystadenomas, MCNs have a risk for harboring malignancy. Rates of invasive disease range from 10% to 50%, which may be an underestimate because of the fact that both benign and malignant epithelium may coexist within the same cyst. Only an extensive pathologic evaluation will detect both entities, which is critical for accurate diagnosis. Clinical factors that increase the risk for malignancy in MCNs are not well characterized, but one study found that older age was more often associated with cancer (Crippa et al, 2008). Factors on imaging associated with malignancy within MCNs include the presence of septations, larger size, and mural nodularity (Allen et al, 2006; Zamboni et al, 1999). Because these lesions are so rare, their natural history is not well defined, although the extent of invasion was demonstrated to be associated with recurrence and death from disease in one study (Allen et al, 2006; Zamboni et al, 1999). In one of the largest studies of MCNs to date, 5-year disease-specific survival for patients with malignancy was 57% (Goh et al, 2006).

Intraductal Papillary Mucinous Neoplasms

Before the first classification system published by the World Health Organization (WHO) (Klöppel et al, 1996), IPMN was described under a variety of names, including mucinous ductal ectasia, papillary carcinoma, and villous adenoma. In contrast to other mucinous and cystic tumors, IPMNs occur equally in men and women and are more often found in older individuals, with the peak incidence between 60 and 70 years of age (Brugge et al, 2004a). Among cystic neoplasms diagnosed in large series, IPMNs represent approximately 15% to 30% of all lesions (see Table 57.2; Allen et al, 2006; Ferrone et al, 2009; Kosmahl et al, 2004; Spinelli et al, 2004). Because of the recent evolution in the understanding of IPMNs, many early reports may include a mixture of patients with IPMNs and MCNs and should be interpreted with caution.

Current nomenclature as defined by the WHO divides IPMN into the categories of adenoma (low-grade dysplasia), borderline (moderate dysplasia), carcinoma in situ (high-grade dysplasia), and carcinoma (Furukawa et al, 2005). As clinical experience with IPMN increases, it is becoming more evident that the underlying process presents as a spectrum of neoplasia with significant variation (Schnelldorfer et al, 2008). The clinical and radiologic presentation, malignant potential, and disease-specific outcomes are highly variable. There also appears to be a progression from adenoma to intraductal carcinoma to invasive carcinoma that can be predicted based on molecular studies.

IPMN is indeed a process that involves the entire pancreas; however, radiographically detectable disease may involve the main duct alone, branch ducts alone, or both (mixed variant). The main duct variant of IPMN is typically seen with diffuse ductal dilation (Fig. 57.3). The presence of a discrete mass or solid component should be concerning for malignancy. Branch duct IPMNs, particularly when smaller than 3 cm, do not have any defining features. They may be unilocular, multilocular, and may be present without dilation of the main pancreatic duct. In the absence of septations, mural nodules, or a solid component, branch duct IPMNs may be indistinguishable from pancreatic retention cysts, MCNs, small cystic endocrine tumors, or even pancreatic pseudocysts.

FIGURE 57.3 Imaging characteristics of intraductal papillary mucinous neoplasm (IPMN). A and B, Main duct IPMN with involvement of the entire main pancreatic duct. C, Branch duct IPMN. D, Chronic dilation of the main pancreatic duct in a patient with conventional ductal adenocarcinoma. The patient did have a component of IPMN in the resected specimen after pancreatoduodenectomy.

IPMNs are now a well-characterized mucinous cystic tumor of the pancreas evidenced by ductal involvement and mucin production. Recently, a histopathologic consensus conference defined IPMN as a grossly visible, noninvasive, mucin-producing neoplasm arising from either the main or side branches of the pancreatic duct with variable degrees of ductal dilation (Fig. 57.4A; Furukawa et al, 2005). On histologic analysis, IMPN lesions are characterized by papillary projections of columnar lined epithelium with varying degrees of dysplasia. Mucin is typically abundant both within the cytoplasm of the lining epithelial cells as well as within the acellular fluid matrix (Fig. 57.4B).

FIGURE 57.4 Gross (A) and microscopic (B) features of intraductal papillary mucinous neoplasm.

IPMN-Associated Malignancy

Malignancy arising from IPMN is well documented, and several large series have been reported in the literature (Table 57.4; D’Angelica et al, 2004; Nara et al, 2009; Salvia et al, 2004; Schnelldorfer et al, 2008; Shin et al, 2010; Sohn et al, 2004; Woo et al, 2009). Invasive cancer is found in 20% to 50% of resected IPMN specimens, which is the reason that these lesions are typically resected. Factors shown to be associated with the presence of invasive cancer in large series include symptoms at presentation, large size, mural nodularity, main duct dilation, and elevated levels of carbohydrate antigen 19-9 (CA19-9).

The reported outcome associated with invasive cancer in patients with IPMN of 5-year survival of 43% to 60% is better than that of ductal adenocarcinoma (15%) (see Chapter 58A, Chapter 58B ) following resection. Poultsides and colleagues (2010) reported the Johns Hopkins experience of IPMN-associated invasive cancers and found that compared with patients with ductal adenocarcinoma, IMPN patients had a lower rate of advanced T stage and lymph node metastasis. Patients with ductal adenocarcinoma more often had high-grade lesions and evidence of perineural invasion, vascular invasion, and positive margins. Despite patients with IPMN-associated cancers having an overall better outcome, if any of the adverse tumor-related factors were present, outcome was similar to patients with ductal adenocarcinoma. It can be inferred from this study that the improved outcome of IPMN-associated cancers is a factor of earlier stage of disease at resection.

Variants of IPMN-associated cancer also have different prognoses. A matched controlled study from our institution demonstrated that patients with colloid-subtype invasive IPMN had significantly better survival than both matched patients with conventional pancreatic carcinoma and patients with invasive tubular IPMN. Fifty-nine patients with IPMN-associated cancer were matched one-to-one with patients with ductal adenocarcinoma with factors from the MSKCC pancreatic cancer nomogram (unpublished data). Among IPMN-associated cancers, tubular histology (compared with the colloid variant) and lymph node involvement were negative prognostic factors associated with outcome. Patients with conventional adenocarcinoma more often presented with obstructive jaundice and had more specimens with evidence of perineural/vascular invasion and regional lymph node metastasis. Survival among patients with colloid variant carcinoma (87% estimated 5-year survival) was significantly better compared with patients with tubular cancers (55%) and ductal adenocarcinoma (23%) (Fig. 57.5).

FIGURE 57.5 Outcome of intraductal papillary mucinous neoplasm–associated malignancy

(Memorial Sloan-Kettering Cancer Center, unpublished data). PDA, pancreatic ductal adenocarcinoma.

Serous Cystadenoma

When radiographic and/or endoscopic findings are characteristic of SCA, resection should be reserved for young patients with significant growth or for those whose lesion is clearly symptomatic. As previously discussed, asymptomatic patients have a risk of malignancy of less than 1%. For these lesions the risk associated with pancreatectomy is much greater than the risk of cancer, and it is well documented that the risk of malignancy in SCA is exceedingly rare. The scenario that presents a challenge is the young asymptomatic patient because the natural growth history of SCA is not well defined.

Tseng and colleagues (2005) radiographically followed 24 patients with SCA for a median of 23 months and reported a 0.6 cm per year median growth rate for these lesions and noted that larger lesions (≥4 cm) had a higher growth rate (1.98 cm/year) compared with lesions smaller than 4 cm (0.12 cm/year). The authors recommended that all patients with SCAs larger than 4 cm be resected, regardless of symptoms. At MSKCC, we have noted a growth rate for SCA of approximately 0.5 cm per year but have not found an association between larger size and faster growth (Allen et al, 2006). It is our current practice to observe asymptomatic patients, regardless of lesion size, with the possible exception of those patients who have large lesions that are marginally resectable at presentation, where tumor growth would preclude the ability to perform resection. However, even in the latter circumstance, resection of asymptomatic SCA is difficult to justify given its benign nature.

Mucinous Cystic Neoplasms

Because of the malignant potential of MCNs, it is recommended that these lesions be resected in most cases. MCNs are more often located in the distal pancreas and therefore may be amenable to splenic preservation and laparoscopic resection, which has been demonstrated to be safe and feasible. The remnant pancreas is not known to be at risk following resection of MCN, and therefore patients do not require long-term follow-up. Patients with features concerning for malignancy should have a formal pancreatectomy. Patients with mucinous cystadenocarcinomas should undergo postoperative surveillance the same as patients with pancreatic cancer.

Intraductal Papillary Mucinous Neoplasms

Resection is reserved for selected patients with IPMNs. When radiographic or endoscopic studies identify dilation of the main pancreatic duct and/or any other concerning features, the standard approach is resection. Concerning radiographic features in the setting of branch duct IPMN include a solid component or mural nodule, septations, or size greater than 3 cm (Allen et al, 2006). Lesions in the head of the pancreas are resected with pancreatoduodenectomy, and in the tail, distal pancreatectomy with or without splenectomy is used. In centers where pancreatectomy is commonplace, central pancreatectomy may be considered for lesions in the body of the pancreas. Total pancreatectomy is reserved for fit patients with diffuse IPMN; however, this procedure should be reserved for highly selected patients, in whom surveillance of the affected remnant duct would not be feasible.

The goal of resection of IPMN is to achieve a complete resection. One of the controversies that exist is the extent of pancreatic resection (Lai & Lau, 2005). IPMN is considered a diffuse process; therefore some concern has been raised that removal of only part of the pancreas would be inadequate. The clinical significance of a glandular recurrence in a patient with invasive IPMN is small because invasive IPMN is a disease characterized more by distant recurrence, similar to that of ductal adenocarcinoma. However, patients who undergo resection of noninvasive IPMN are at risk for recurrence in the remnant pancreas. In a recent report from MSKCC, White and colleagues (2007) reviewed 78 patients who had a resection for noninvasive IPMN; at a median follow-up of 40 months, a recurrence rate in the remnant pancreas of 8% was reported. Only 2% of patients (1 in 50) with a negative margin had recurrence, but 17% of those (4 of 23) with a margin positive for IPMN had recurrence (P = .02). Because 92% of this cohort of patients did not have a recurrence, it would seem that the risk associated with total pancreatectomy outweighs the risk for recurrence. However, routine surveillance of these patients with cross-sectional imaging is recommended. In this study, recurrences occurred at a median of 22 months (range, 8 to 62 months) from the time of initial resection and were noninvasive. Completion pancreatectomy is indicated for treatment of any detected recurrence.

One of the challenges of managing patients with IPMN is that many patients are young and are seen with small cysts in the head of the pancreas. Our institution and others have demonstrated the exceedingly rare incidence of invasive malignancy in cysts smaller than 3 cm (Allen et al, 2006; Lee et al, 2008; Sahani et al, 2006). We have not seen invasive IPMN less than 3 cm in the setting of isolated branch duct disease without concerning radiographic features. Because of the low risk of invasive cancer in small mucinous cysts, our current approach to these lesions is selective. Resection is indicated for patients with symptoms, which are infrequently present in patients with cystic neoplasms. A second indication for resection is for cysts with concerning features, such as a solid component or increasing size (Fig. 57.6; Allen et al, 2006). If the decision is made for nonoperative management, our current surveillance protocol includes high-quality cross-sectional imaging with multidetector CT scan or magnetic resonance imaging (MRI) every 6 months for 2 years (Landa et al, 2009). After 2 years of demonstrable stability, patients can graduate to annual imaging. Indications for resection in patients undergoing surveillance include significant tumor growth or development of other features concerning for malignancy.

FIGURE 57.6 A, Intraductal papillary mucinous neoplasm (IPMN) in the head and uncinate process of the pancreas in a 68-year-old woman. Because the lesion was small without associated concerning features, the patient was initially only monitored. B, Over time, the lesion increased in size, and the patient underwent pancreatoduodenectomy. Final pathology revealed a main duct IPMN with moderate dysplasia.

Solid Pseudopapillary Tumor and Other Cystic Pancreatic Neoplasms

Solid pseudopapillary tumors (SPTs) are also known as Frantz tumors or Hamoudi tumors after the pathologists who first described this unique entity (Frantz, 1959; Hamoudi et al, 1970). SPTs are rare tumors that account for up to 2.5% of resected pancreatic neoplasms. They predominate in young women and are often discovered when patients have symptoms. In a review of SPTs, Papavramidis and Papavramidis (2005) described the clinicopathologic factors of 718 patients collected from the English literature. The female/male ratio was 9.8 : 1, and the mean age was 22 years. Only 15% of the patients in the review were asymptomatic at presentation; 47% had pain, and 35% had a mass. Tumors were often large, and 34% were greater than 10 cm in diameter. The most common location for SPT was the tail (36%) and head (34%) of the pancreas, and metastases were present in approximately 20% of patients.

Cross-sectional imaging characteristically reveals encapsulated lesions with irregular areas of hypodensity secondary to necrosis or hemorrhage (Fig. 57.7). The wall often reveals calcifications. Because of the cystic degeneration exhibited by these lesions, they are often difficult to differentiate from pseudocysts or other pancreatic cystic neoplasms. What differentiates these lesions from other cystic tumors is a combination of solid and cystic components with gradual degenerative changes resulting in pseudopapillae formation (Tang et al, 2005). SPTs are distinct from the normal pancreatic parenchyma, and like other cystic neoplasms, they rarely cause ductal dilation because of the lack of invasion.

FIGURE 57.7 Imaging characteristics of solid pseudopapillary tumor (SPT) in the pancreas of a 35-year-old woman. A large lesion in the head of the pancreas was causing chronic abdominal pain; note the thickened wall and area of hemorrhage classic for SPT (A to D). This patient underwent endoscopic ultrasound to confirm SPT prior to pancreatoduodenectomy.

Grossly, tumors appear tan or yellow and are well circumscribed and smooth on palpation. When opened, the tumors show irregular cystic cavities with some areas of necrosis or hemorrhage (Fig. 57.8). Most SPT can be diagnosed with routine histologic evaluation by the presence of polygonal epithelial cells arranged in a discohesive pattern; the boundaries resemble a mass effect as opposed to being invasive. Immunohistochemical staining for vimentin, keratin, neuron-specific enolase, CD10, and progesterone is common (Martin et al, 2002). These lesions rarely express chromogranin or estrogen. In contrast to ductal adenocarcinoma, abnormal β-catenin expression has been shown to be present in SPTs, but KRAS, TP53, and SMAD4 appear to be unaltered (Abraham et al, 2002). Aggressive phenotypes of SPT have been demonstrated and histologically exhibit a high mitotic rate, nuclear atypia, spindling of tumor cells, and anaplastic giant tumor cells consistent with sarcomatoid carcinoma (Tang et al, 2005).

FIGURE 57.8 Gross appearance of solid pseudopapillary tumor (resected specimen from patient in Fig. 57.7).

Clinically, SPTs are indolent and are typically cured with complete surgical extirpation. Many series have demonstrated long-term survival after resection (Lam et al, 1999; Martin et al, 2002; Reddy et al, 2009). Significant vascular invasion is the most common reason for the inability to resect these lesions, and metastatic disease has been noted to be present in up to 15% of patients. Criteria for defining SPTs as “malignant” may be locally advanced disease with vascular invasion precluding resection or lymph node or hepatic metastasis. Long-term survival is often possible despite the presence of malignant factors, such as metastatic disease, and should not preclude resection in selected cases (Lam et al, 1999; Martin et al, 2002; Reddy et al, 2009).

Indeterminate Cystic Neoplasm

The spectrum of cystic lesions discussed above represents many of the incidental lesions seen on cross-sectional imaging. However, a significant number of patients present with lesions that lack diagnostic certainty. Because of the inability to determine the exact histologic diagnosis without pancreatectomy, and because of the risk for malignancy of some cyst subtypes, some authors advocate a routine resection of pancreatic cysts (Horvath & Chabot, 1999; Ooi et al, 1998; Siech et al, 1998; Spinelli et al, 2004). Advocates of this approach argue that the clinical distinction between benign and malignant lesions is unreliable, and that there is a risk for the development of cancer; therefore all medically fit patients should undergo pancreatic resection. Despite the fact that this approach allows definitive treatment of all malignant lesions, it exposes patients with benign lesions to the risk associated with pancreatectomy without known benefit.

A selective approach to resection of pancreatic cystic neoplasms has been advocated by several reports, including those from our institution (Allen et al, 2003, 2006; Fernandez-del Castillo et al, 2003). Advocates of this approach argue that with improved knowledge of specific histologic subtypes and with the advances in diagnostics, a group of patients can be identified who are at very low risk (<1%) of harboring malignancy. Most authors who have reported their experience with this approach have recommended radiologic surveillance over resection for small, incidentally discovered cysts that lack a solid component or that have other concerning features of malignancy. Although this approach avoids the risks associated with pancreatectomy in patients with benign lesions, it does not guarantee that a cancer is being mistakenly observed. Operative morbidity following pancreatectomy is consistently as high as 45%, and mortality rates range from 2% to 10% (Büchler et al, 2000; Mezhir et al, 2009; Muscari et al, 2006; Winter et al, 2006). Published data suggest that with current imaging and endoscopic techniques, a group of patients with pancreatic cysts can be identified with a malignancy rate of less than 3% (Allen et al, 2006). It is in this select group of patients that the risk of death from pancreatectomy approaches or exceeds the risk of the lesion harboring a malignancy.

Despite the well-characterized lesions described above, many patients are seen with a cystic neoplasm that does not fit a specific diagnosis. In these patients, a nonoperative histologic diagnosis may not be made; therefore treatment recommendations for these patients are primarily based on imaging findings. The majority of patients who fit into this category have small, asymptomatic lesions, and their management is controversial. Our approach to such patients has been published in recent reports (Allen et al, 2003, 2006). Allen and colleagues (2006) reviewed 539 patients, the majority of which (369 [68%]) were followed up without operative intervention. Compared with patients treated with resection, observed patients were more likely to have small (median diameter, 2.4 cm) asymptomatic cysts without a solid component. At a median follow-up of 24 months, the average change in diameter of the observed lesions was 0.2 cm. Changes leading to resection developed in 29 patients (8% of 369 initially observed), and no patients with malignant mucinous tumors were reported in this group. Eight patients (2%) who did not undergo immediate resection developed malignancy, which is lower than the mortality rate associated with pancreatic resection, even in high-volume centers. Of note, in five of the eight patients who developed malignancy in the observed group, the cysts appeared to have represented retention cysts that developed adjacent to a radiographically undetectable malignancy. The decision-tree analysis illustrating the approach used is shown in Fig. 57.9.

FIGURE 57.9 Selective management of cystic neoplasms of the pancreas. Rad., radiologic.

(From Allen PJ, Brennan, MF, 2007: The management of cystic lesions of the pancreas. Adv Surg 41:211-228.)

As a result of our findings from this retrospective review, our current approach to patients with incidental cysts less than 2.5 cm in diameter is to initially obtain high-quality cross-sectional imaging with triphasic multidetector CT with 2-mm cuts through the pancreas or an MRCP. The imaging is reviewed at a multidisciplinary hepatopancreatobiliary conference by surgeons, gastroenterologists, and radiologists with expertise in pancreatic diseases. A detailed analysis of the cyst characteristics and the adjacent parenchyma is undertaken. Any findings concerning for malignancy—such as a solid component, mural nodules, or any pancreatic parenchymal abnormality—should result in EUS for further diagnosis or resection.

The importance of evaluating the adjacent pancreatic parenchyma cannot be overstated in that a number of patients in our series who developed adenocarcinoma in the observed group had small retention cysts adjacent to invasive cancer. Incidentally discovered cystic neoplasms smaller than 3 cm without solid component of main duct dilation are typically followed radiographically every 6 months for 2 years and then annually thereafter.