Pancreatic neuroendocrine tumors in patients with tuberous sclerosis complex

Authors


Elizabeth A. Thiele, MD, PhD, Herscot Center for Tuberous Sclerosis Complex, Massachusetts General Hospital, 175 Cambridge Street, Suite 340, Boston, MA 02114, USA.
Tel.: +1 617 726 0241;
fax: +1 617 726 0230;
e-mail: ethiele@partners.org

Abstract

Larson AM, Hedgire SS, Deshpande V, Stemmer-Rachamimov AO, Harisinghani MG, Ferrone CR, Shah U, Thiele EA. Pancreatic neuroendocrine tumors in patients with tuberous sclerosis complex.

We explored pancreatic neuroendocrine tumors (PanNETs) associated with tuberous sclerosis complex (TSC) to determine their incidence in the TSC population; define their clinical, radiological, and pathological characteristics; and investigate their association with underlying genotypes. Retrospectively reviewed abdominal imaging of 219 patients with TSC, evaluating the incidence, size, and architecture of pancreatic lesions. Pathology records at Massachusetts General Hospital (MGH) were reviewed for all PanNET diagnoses in patients with TSC. Literature was reviewed for TSC-related PanNET cases. Nine patients with TSC were found to have a pancreatic lesion(s) on abdominal imaging and six patients have been diagnosed with a PanNET by pathology at MGH. Twelve cases of TSC-associated PanNETs have been reported in the literature. Of these 18 PanNET cases, one third were cystic, and the average age at resection was 26 years. Germline TSC2 mutations were found in all patients for whom genetic data were available (n = 3). We did not identify pancreatic angiomyolipomas in this series. Our results suggest that PanNETs are the most common pancreatic lesion in patients with TSC. Focal pancreatic mass lesions, solid or cystic, in patients with TSC should be considered possible PanNETs, and resection of the lesion may be clinically indicated.

Conflict of interest

Nothing to declare.

Introduction

Pancreatic neuroendocrine tumors (PanNETs) are rare, representing 1–2% of all pancreatic cancer, with an annual incidence of 2.2 in 1,000,000 individuals (1). Although they carry a much better prognosis than pancreatic adenocarcinoma, survival rates remain 64% at 5 years and 44% at 10 years (2). The current standard of care for PanNETs is surgical resection. PanNETs can occur sporadically or in the setting of an autosomal dominant genetic syndrome, principally, multiple endocrine neoplasia type 1 (MEN1) but also von Hippel Lindau, neurofibromatosis type 1, and tuberous sclerosis complex (TSC) (3–6). To date, the connection between PanNETs and TSC has not been formally investigated beyond several single case reports (7–18). In this study, we examine a large series of TSC patients to define the clinical characteristics of TSC-associated PanNETs.

TSC is an autosomal dominant genetic disorder, characterized by the development of hamartomas, benign tumors, and rarely, malignant tumors in multiple organs including the brain, heart, eyes, kidney, skin, and lungs. The underlying genetic abnormality is a germline mutation in TSC1 and TSC2 tumor suppressor genes, resulting in loss of the respective encoded proteins hamartin or tuberin and activation of the mammalian target of rapamycin complex 1 (mTORC1) pathway, which modulates cell growth and survival (19). Most lesions associated with TSC are hamartomatous or benign.

The pancreas is not a main focus for TSC-related pathology, and therefore, it is not typically monitored clinically. However, TSC patients are followed by abdominal imaging for angiomyolipomas (AMLs) and cysts, which are common in TSC patients, predominantly in the kidney, liver, and lungs. When a pancreatic lesion is incidentally identified on abdominal imaging in TSC, it is often assumed to represent an angiomyolipoma or benign cyst. In this study, we reviewed a series of TSC-associated PanNETs to determine the incidence of these lesions in the TSC population and defined the clinical, radiological, and pathological characteristics of these pancreatic lesions and their association with underlying genotypes.

Materials and methods

At the Herscot Center for TSC, 710 cross-sectional imaging studies, including contrast enhanced computerized tomography (CT) and magnetic resonance imaging (MRI) scans of 219 patients (92 males and 127 females) with TSC were reviewed to calculate the incidence of focal pancreatic lesions. The scans were performed over a period of 17.5 years with the primary intention of monitoring for renal pathology and as such, this radiological series followed late arterial and/or portal venous phase contrast protocols. All 710 studies were read by two radiologists and the following findings were recorded as consensus: presence or absence of a focal pancreatic lesion, size on axial sections, architecture (cystic/solid), post-contrast enhancement, location, and pancreatic duct status. Lesion size was reported based on the greatest diameter. When serial scans were available, the lesions were assessed for change, with change defined as a difference of 0.5 cm or greater between serial scans. The radiological findings were correlated with clinical parameters as well as each patient's TSC genotype.

A retrospective review of pathology records at Massachusetts General Hospital (MGH) for all surgical and autopsy specimens from 1977–2011 identified 390 PanNETs, of which six occurred in patients who also had TSC. Retrieval of clinical, radiological, surgical, and genetic data from medical charts was performed in accordance with institutional review board (IRB) approval, and the pathology was reviewed and assessed using the 2010 classification and the diagnostic standard procedure suggested by the World Health Organization (20). Additionally, a literature review was conducted, and 12 case reports of TSC-associated PanNETs were identified, published between 1959 and 2009. Genotypic and phenotypic data were compiled along with the clinical history for each case.

Results

For the 219 TSC patients whose CT or MRI abdominal images were reviewed, the average age at first scan was 22.2 years (range 0.1–73.3 years). Seventy-four percent of the cohort had multiple studies available. Pancreatic lesions were seen in nine patients, six males and three females. Of the 91 patients with a confirmed TSC2 mutation, five were found to have a pancreatic lesion (5.5%), and of the 41 individuals with a TSC1 mutation, two were found to have pancreatic lesion (4.9%). A solitary lesion was observed in eight of the nine cases, while one individual had two cystic lesions identified. The average lesion size was 1.7 cm (range 0.4–5.6 cm), lesion size remained stable over time in six cases, increased in one, and decreased in two. In seven of the cases, the pancreatic lesions had a cystic component (77.8%), four of which were purely cystic. All identified lesions showed heterogeneous enhancement (9/9) and none (0/9) exhibited dilation of the pancreatic duct. In six of the cases, pancreatic lesions were identified on the first imaging study available; therefore, age of onset cannot be stipulated. However, in three cases, the lesions were first seen on a follow-up image after a previously negative scan. These initial lesions were first observed in patients at ages 7, 35, and 56 years.

Five of these nine patients with TSC-associated pancreatic focal pathology have not yet had a surgical biopsy of their lesion (Table 1). The remaining four patients with pancreatic lesions identified in this radiologic retrospective review have undergone surgical resection. Upon retrospective review of the MGH pathology records, two additional cases of PanNETs in patients with TSC were identified (Cases 15 and 18) for whom radiological studies were not available for review, one of which was an autopsy case. All six lesions have been diagnosed as well-differentiated pancreatic neurendocrine neoplasms (grade 1 or 2) on histopathological examination (Table 2). All five resection cases are currently doing well without disease recurrence at 2.1, 20.4, 22.3, 27.1 months and 12.8 years, respectively, post surgical resection.

Table 1.  Pancreatic lesions incidentally identified by abdominal CT or MRI in patients with tuberous sclerosis complex
Age (years)/ sex/siteRadiologySize (cm)Size change/ timeTSC mutation
  1. B, body; CT, computerized tomography; F, female; H, head; M, male; MRI, magnetic resonance imaging; T, tail; TSC, tuberous sclerosis complex.

  2. aTwo cystic lesions identified which on a follow-up scan after 11 months had grown together and transitioned from a cystic to a solid architecture. This single solid mass then decreased in size over 3.3 years (5.4 to 4.9 cm, -9%).

  3. bHistory of renal cell carcinoma.

7 F, TSolid1.6Stable/2.1 years TSC1
32 M, BaCystic/solid5.4-9%/3.3 years TSC2
35 F, BbCystic0.5Stable/6 months TSC2
43 M, HCystic0.4Stable/2.6 years TSC1
59 F, HCystic0.6Stable/3.6 years 
Table 2.  Pancreatic neuroendocrine tumors in patients with tuberous sclerosis complex Thumbnail image of

Since 1959, there have been 12 individual cases of TSC-associated PanNETs reported in the literature (7–18). Three of the reports include TSC germline mutation data and two include loss of heterozygosity (LOH) analysis of the resected tumor tissue (Table 2). The 18 cases of TSC-associated PanNETs diagnosed on pathology, 6 presented here and 12 from the literature are compiled in Table 2. The average patient age at surgical resection was 26.4 years. Two thirds of the cohort was men. A third of the pancreatic lesions had a cystic component either on radiology or on pathology. Two of the reported cases were found to have local or distant metastases (11,14). Table 3 compares the TSC-related PanNETs with a previously reported surgical series of 168 PanNETs that were resected at MGH between 1977 and 2005 (please note, 12 of the 168 tumor resections were in patients with MEN1 and one patient occurs in both series) (21). Table 3(21–22) also compares the six TSC-related cystic PanNETs with a series of 29 cases of cystic PanNETs that were resected at MGH between 1977 and 2006, again, one patient occurs in both series.

Table 3.  Pancreatic neuroendocrine tumors and cystic pancreatic neuroendocrine tumors
 TSC-associated case series, n = 18General population case series (21), n = 168a
Pancreatic neuroendocrine tumors
  1. TSC, tuberous sclerosis complex.

  2. aPanNETs that were surgically resected (21).

  3. bAt resection.

  4. cCystic PanNETs that were surgically resected (22).

  5. dAll cystic PanNETs that were surgically resected.

Average age (years)b2656
Male to femaleM 67%M 51%
 F 33%F 49%
  n = 18 
Cystic33%17%c
  n = 18 
Mean size5.1 cm3.6 cm
Range0.8–21 cm n = 120.5–17 cm
  n = 12 
Lesion location  
 (Uncinate/Head/Neck) (Body/Tail)23%36%
 (Body/Tail)77%64%
  n = 13 
Functional44%42%
Non-functional56%58%
  n = 18 
Incidentally detected50%51%
Metastatic13%6.5%
  n = 15 
  TSC-associated case series, n = 6 General population case series (22), n = 29d
Cystic pancreatic neuroendocrine tumors
Average age3353
Male to female4M (67%)M 50%
 2F (33%)F 50%
Average female age (years)15 
Average male age (years)40 
Mean size (cm)7.04.9
  n = 5 
Lesion location  
 (Uncinate/Head/Neck)0%24%
 (Body/Tail)100%76%
  n = 5 
Functional17%20%
Non-functional83%80%
Incidentally detected83%27%

Discussion

The incidence of PanNETs in this TSC series was 1.8% based on a pathologic (n = 4) diagnosis or 18,000 per 1,000,000, as compared to the rate of 2.2 per 1,000,000 in the general population (1). Further, in the MGH patient cohort with a confirmed PanNET diagnosis on pathology, the incidence of TSC was 1.5%, or 92 in 6000, as compared the rate of 1 in 6000 seen in the general population. The results of this study suggest that the pathophysiology of TSC may include pancreatic endocrine tumors. The current clinical assumption that pancreatic lesions in patients with TSC are AMLs implies that lesions are benign and therefore require no immediate treatment. This series, however, suggests that the majority of these pancreatic lesions in patients with TSC are in fact not AMLs, but rather, PanNETs. These lesions hold malignant potential and therefore require a more aggressive treatment protocol. The literature does contain four case reports of pancreatic AML variants [perivascular epithelioid cell tumors (PEComa), clear cell ‘sugar’ pancreatic tumors] (23–26); however, these lesions have never been documented in an individual with TSC. Therefore, based on the current evidence and the results of this study, we submit that PanNETs are the dominant pancreatic pathology in the setting of TSC.

A comparison of the clinical, radiological, and pathological features of TSC-related PanNETs in this report with those of PanNETs occurring in the general population shows two interesting findings: the lesions in the TSC cohort occur at a younger age and are more frequently cystic. Risk of aggressive behavior is difficult to assess. Although most of our cases showed no pathological features predictive of aggressive behavior, our follow-up window was too short and the series was too small to make any conclusions. In addition, there are two TSC-related PanNET case reports that described local or distant metastatic behavior, so although it may not be common, aggressive behavior in a TSC-related PanNET is possible (11,14). In the general population, cystic lesions are 3.5 times more likely to occur in patients with multiple endocrine neoplasia type 1 (an inherited tumor syndrome) when compared with solid lesions (22). But when comparing the TSC-related cystic PanNETs to the cystic lesions of the general population, the TSC cohort is still younger (22). The parallel behavior and similar cystic architecture of TSC and MEN1-related PanNETs suggest a common or related pathophysiologic mechanism. The most significant difference between TSC-related PanNETs and all other syndromic PanNETs is that in TSC, the majority of these lesions were found to be solitary and not multifocal.

In our cohort of six TSC-related PanNETs, genetic data were available for three individuals, and all three were found to have a TSC2 germline mutation in or just upstream of the GTPase-activating protein (GAP) domain (exons 33–36). In the three cases from the literature that reported genotypic data, TSC2 germline mutations were also described (14, 16, 17). Additionally, for two of these cases, the tumor tissue was found to have LOH at the TSC2 locus (14, 16). Our study confirms that the loss of TSC2 may play an integral role in PanNET pathology. More specifically, based on these data, the GAP domain of TSC2 has surfaced as a potential biomarker for PanNET development. The nonsense mutation reported in case 1, the youngest documented TSC-related PanNET, accounts for 0.9% of all TSC2 point mutations (16, 27). Also, case 6 has a TSC2 in-frame deletion mutation, the loss of a single amino acid in the GAP domain; her TSC phenotype, however, is quite severe. This case could highlight a possible PanNET locus of susceptibility and driving mutation for neoplastic development in the setting of TSC.

The literature regarding the role of the mTOR pathway in the development of sporadic PanNETs is rapidly evolving. Decreased TSC2 protein expression has been found in 53.4% of well-differentiated endocrine tumors (28). This translated to the exploration of mTOR inhibitors as adjuvant therapy for PanNETs as well as neuroendocrine tumors of other organs. Human PanNET cell lines, when exposed to everolimus, a sirolimus analogue, underwent increased apoptosis and decreased proliferation in a dose-dependent way (28–30). A phase 3 clinical trial of everolimus as monotherapy for PanNETs (RAD001 In Advanced Neuroendocrine Tumors; RADIANT-3) showed a 6.4-month improvement in progression free survival for patients with advanced PanNETs of low or intermediate grade (31). Additionally, the genetic profile of sporadic PanNETs continues to unfold. In a recent study, mutations in DAXX/ATRX, MEN1, and mTOR pathway genes, surfaced as the dominant driving mutations in PanNETs (32). Eighteen thousand protein-coding genes were analyzed in 10 sporadic PanNETs tumors, two of which were found to have a genetic mutation in TSC2. In the second phase, by Sanger sequencing, 8.8% of 56 sporadic tumors also showed a TSC2 mutation, 7.3% had mutations in PTEN, and 1.4% in PIK3CA(32). These results provide further evidence that aberrant mTOR activity likely acts as a driver in the development a subset of sporadic PanNETs and suggest that patients with PanNETs could be stratified for chemotherapeutic treatments (32).

Cystic lesions of the pancreas associated with TSC were first described in 1933 (15, 33). Pancreatic lesions in patients with TSC, solid or cystic, should be considered possible PanNETs, and resection of the lesion may be clinically indicated. In light of these data, we have changed our clinical practice for routine imaging in TSC from the more limited renal imaging series to a more complete abdominal protocol.

Acknowledgements

This study was approved and supported by the Herscot Center for Tuberous Sclerosis Complex, and NIH/NINDS P01 NS024279. Special thanks to Agnies van Eeghen, MD, and Susana Camposano, MD, for their help with language translations.

Ancillary