Pancreatic Ductal and Acinar Cell Neoplasms in Carney Complex: A Possible New Association

Abstract

Context:

Carney complex (CNC) is a rare disease inherited as an autosomal dominant trait, associated with various tumors, and caused most frequently by inactivation of the PRKAR1A gene.

Objectives:

In our recent investigation of a large cohort of CNC patients, we identified several cases of pancreatic neoplasms. This possible association and PRKAR1A's possible involvement in pancreatic tumor have not been reported previously.

Patients and Methods:

Nine patients (2.5%) with CNC and pancreatic neoplasms in an international cohort of 354 CNC patients were identified; we studied six of them. Immunohistochemistry and PRKAR1A sequencing were obtained.

Results:

Three men and three women with a mean age of 49 yr (range 34–75 yr) had acinar cell carcinoma (n = 2), adenocarcinoma (n = 1), and intraductal pancreatic mucinous neoplasm (n = 3). Five patients had a germline PRKAR1A mutation, including two patients with acinar cell carcinoma, for whom mutations were found in a hemizygous state in the tumor, suggesting loss of heterozygosity. PRKAR1A expression was not detected in five of the six pancreatic neoplasms from CNC patients, whereas the protein was amply expressed on other sporadic pancreatic tumors and normal tissue.

Conclusion:

An unexpectedly high prevalence of rare pancreatic tumors was found among CNC patients. Immunohistochemistry and loss-of-heterozygosity studies suggest that PRKAR1A could function as a tumor suppressor gene in pancreatic tissue, at least in the context of CNC. Clinicians taking care of CNC patients should be aware of the possible association of CNC with a potentially aggressive pancreatic neoplasm.

Carney complex (CNC) (MIM 160980), initially described in 1985 by Dr. J. Aidan Carney (1), is a rare multiple neoplasia syndrome inherited as an autosomal dominant trait. The main causing gene is PRKAR1A, which codes for protein kinase A (PKA), cAMP-dependent, regulatory, type I, α-subunit (R1α). PRKAR1A is mutated in the heterozygote state in about 75% of the patients with the syndrome (2–4) and is located on chromosome 17q22-24. R1α is the main regulator of the cAMP-signaling pathway by controlling PKA responses to cAMP. Inactivating PRKAR1A mutations in CNC lead to PKA activation and increased cellular proliferation and tumor development (5). PRKAR1A functions as a tumor suppressor gene, although not a classic one because it is associated with hyperplasia in the haploinsufficient state; somatic mutations or loss of heterozygosity (LOH) have been described in CNC tumors as second hits and are usually associated with more aggressive tumors (6, 7).

CNC is characterized by the development of various endocrine tumors, including primary pigmented nodular adrenocortical disease (PPNAD) and pituitary, thyroid, testicular, and ovarian tumors. A number of nonendocrine tumors have also been described, from psammomatous melanotic schwannoma to liver tumors, and the spectrum of neoplasms associated with this disease appears to be expanding (2, 8). Pancreatic neoplasms have now been reported in individual cases (7–9) and associated with significant mortality (8). One of the reported cases was a pancreatic acinar cell carcinoma (7), an extremely rare form of pancreatic neoplasm that accounts for only 1% (or less) of all pancreatic tumors. This tumor had, in addition to a germline PRKAR1A-inactivating mutation, somatic LOH of the 17q22-24 locus, suggesting that PRKAR1A may function as a tumor suppressor gene in pancreatic tissue.

In this investigation, we systematically searched a cohort of 354 CNC patients, the largest part of whom we reported recently (2), for cases of pancreatic neoplasms. Of a total of nine cases, six had accurate clinical data and biological material available; we studied those latter cases. The results confirm a possible association of pancreatic neoplasms with CNC and point to the potential role of PRKAR1A as a tumor suppressor gene in pancreatic tissue.

Subjects and Methods

Patient selection and clinical evaluation

We searched in a series of patients with CNC genotyped for PRKAR1A that has been, in part, previously reported (2) for cases of pancreatic neoplasm. Informed consent for genetic diagnosis, tumor analysis and access to the data collected was obtained, from all patients, as part of studies approved by Institutional Review Boards of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health; Mayo Clinic; and the Cochin Hospital. Clinical, hormonal, and radiological investigations were done as previously recommended to screen for the main CNC manifestations (8).

Pathological examination and immunohistochemical labeling

Pathological examination was performed on all pancreatic tumors available (n = 6), with special attention to tumor differentiation. For three tumors, nontumoral pancreas was not available as an internal control. In the remaining three samples, both tumor material and the extratumoral normal ductal and acinar pancreatic tissue were available. Normal pancreas, intraductal papillary mucinous neoplasm (IPMN), pancreatic ductal adenocarcinoma, and pancreatic acinar cell carcinoma from patients without CNC were used as additional controls. Immunohistochemistry for PRKAR1A, using primary mouse monoclonal antibody raised against the RIα subunit (1:250; Becton Dickinson Transduction Laboratories, Le Pont de Claix, France), was performed as previously described (10). Briefly, sections (4 μm thick) from formalin-fixed tissue embedded in paraffin were mounted on Superfrost/Plus glass slides. The paraffin was removed by incubating the sections in xylene and then rehydrating them. For antigen retrieval, sections were heated in a bain-marie (double boiler) for 40 min at 98 C in 10 mmol sodium citrate buffer (pH 6.0). The slides were incubated with the antibody for 60 min at room temperature and subsequently with the streptavidin-biotin-peroxidase complex. The marker was detected by the enzymatic precipitation of 3,3′-diaminobenzidine tetrahydrochloride in 0.5 mmol Tris. Finally, the slides were counterstained with Mayer's hematoxylin.

PRKAR1A molecular genetic analysis

Pancreatic tumor DNA was purified from frozen tissue as previously reported (7) and from formalin-fixed paraffin-embedded tissues using the High Pure PCR Template Preparation kit (Roche Diagnostics, Indianapolis, IN) according to the manufacturer's protocol. DNA and RNA were extracted from whole blood using standard procedures. For PRKAR1A, sequencing of the 12 exons and proximal intronic regions of the gene was completed as previously reported (7). DNA samples from CNC patients that were negative for sequencing defects underwent Southern hybridization analysis for the identification of large intra- or perigenic deletions and/or other rearrangements (11). After separation of B lymphocytes and their transformation by Epstein-Barr virus, for mutations predicted to lead to nonsense-mediated mRNA decay (NMD), cycloheximide treatment of transformed lymphocytes was performed as previously reported to demonstrate this mechanism of PRKAR1A inactivation (3, 7, 12, 13). For mutations that escape NMD, the presence of the mutant mRNA or protein was demonstrated on transformed leukocytes or tumor tissues, as previously reported (7, 14, 15). Nucleotides were numbered in accordance with the reference sequence for PRKAR1A (GenBank accession no. NM_002734), as reported by Kirschner et al. (3, 12). The Genescan and Genotyper software programs were used for data collection and export.

Results

Clinical characteristic of patients with CNC and their pancreatic tumors

Among the 354 patients, nine (2.5%) were reported to have a pancreatic lesion, including three patients (patients 1, 4, and 5) reported previously (7, 9). Pancreatic lesions diagnosed were unspecified pancreatic cancer (n = 3), IPMN (n = 3), acinar cell carcinoma (n = 2), and pancreatic adenocarcinoma (n = 1). Accurate clinical information and biological material were not available for three patients who had an otherwise unspecified pancreatic cancer, and they were excluded from the study. The remaining six patients were three men and three women with a mean age of 49 yr (from 34–75 yr); three died of their pancreatic tumor. Clinical characteristics of the six patients and histopathological description of their pancreatic tumors are detailed in Table 1.

Table 1.

Clinical and genetic characteristics of patients with CNC and pancreatic tumors

Patient no.	Sex/age (yr) at diagnosis of pancreatic tumor	Clinical manifestation at diagnosis	Procedure	PRKAR1A mutation			Associated tumors	Status (age)
				Germline	Expected effect on the proteina	Somatic
1	F/34	Pancreatitis	Whipple	c.708+1G→T heterozygous	Expressed (shorter PRKAR1A protein 328 amino acids)	c.708+1G→T homozygous	Cardiac, skin, and breast myxoma, lentiginosis, PPNAD, melanocytic schwannoma, thyroid and ovarian lesion, mesenteric myxoma, leiomyoma of the bladder, splenic lipoma, mammary adenofibroma, liver adenoma	Deceased at 35
2	M/65	Abdominal pain	Total pancreatectomy	c.891+3 a→g heterozygous	Skip exon 10	c.891+3a→g homozygous	Lentiginosis, large-cell calcifying Sertoli cell tumor	Deceased at 67
3	F/75	Abdominal pain	None (biopsy–autopsy)	No mutation (polymorphism variant: c.349−5dup)	NA	NA	Lentiginosis, PPNAD, pilonidal cyst	Deceased at 76
4	M/50	Pancreatitis, abdominal pain	Distal pancreatectomy	c.190 C→T/p.Q64X heterozygous	NMD, no protein	NA	Cardiac, skin myxoma, large-cell calcifying Sertoli cell tumor, thyroid tumor	Alive at 57
5	M/54	Pancreatitis, abdominal pain	Total pancreatectomy	c.190 C→T/p.Q64X heterozygous	NMD, no protein	NA	Cardiac myxoma	Alive at 57
6	F/48	Incidentaloma	Distal pancreatectomy	c.709(−7−2)del6 heterozygous	NMD, no protein	NA	PPNAD, ovarian lesion	Alive at 52

F, Female; M, male; NA, not available.

^aAccording to Ref. 15.

Genetic analysis of patients with CNC and pancreatic tumors

Molecular genetic data for the six patients are detailed in Table 1. Germline PRKAR1A-inactivating mutations were observed in five of the six patients; a polymorphism variant of the PRKAR1A gene was present in the remaining patient, i.e. c.349-5dup (this is the most frequent intronic variant of the PRKAR1A gene that is detected in approximately 15% of the studied alleles).

Among the five mutations, at the protein level, NMD was found for three of the sequences; this leads to absent protein expression of the mutant allele (3). In the remaining mutation, leading to skipping of exon 6 of the gene, a shorter PRKAR1A protein is expressed that leads to an increase of the cAMP/PKA-dependent phosphorylation of target molecules in an in vitro study (7). The last mutation leads to a splice retaining part of intron 9 on the RNA level and skipping exon 10 on the protein level (3, 15).

The germline mutations were found in the hemizygote state in two pancreatic acinar cell carcinomas available for study, suggesting LOH. We were able to confirm LOH in one of them, studying seven microsatellite markers located on17q22-24 (7).

Histopathological studies of pancreatic tumors in CNC patients

Results are detailed in Table 2, and representative images of the staining are shown in Fig. 1. There was no staining for the PRKAR1A protein in five of the six pancreatic tumors, whereas membranous and cytoplasmic staining for the molecule was present in surrounding normal tissues. The tumor, with only a weak staining for the PRKAR1A protein, was an acinar cell carcinoma having a splice mutation affecting intron 9 to exon 10 and classified as pathogenic (15).

Table 2.

Histological type and immunohistochemical characteristics of pancreatic tumors of patients with CNC, sporadic pancreatic carcinomas, and normal pancreas

Patient	Pancreatic tumor type/AJCC	PRKAR1A staining
		Pancreatic lesion	Normal pancreas
1	Acinar cell carcinoma/T3N1M0	No expression	NA
2	Acinar cell carcinoma/T3N0M0	Weak cytoplasmic	NA
3	Adenocarcinoma/T2NxM1	No expression	NA
4	IPMN main and branch duct, low-grade dysplasia	No expression	Membranous and cytoplasmic (moderate in ducts, intense in acinar cells)
5	IPMN main and branch duct, high-grade dysplasia	No expression	Membranous and cytoplasmic (moderate in ducts, intense in acinar cells)
6	IPMN branch duct, gastric metaplasia, low-grade dysplasia	No expression	Membranous and cytoplasmic (moderate in duct, intense in acinar cell)
	Normal pancreas		Moderate membranous and weak cytoplasmic in ducts; membranous intense and cytoplasmic weak on acinar cells
	Sporadic adenocarcinoma	Moderate membranous and weak cytoplasmic
	Sporadic IPMN	Moderate membranous and weak cytoplasmic
	Sporadic acinar cell carcinoma	Diffuse moderate cytoplasmic

AJCC, American Joint Committee on Cancer; NA, not available.

Fig. 1.

Histological features, and PRKAR1A immunohistochemistry (IHC) in pancreatic tumors of patients with CNC. Left panels, Hematoxylin-eosin-saffron (HES) staining (×200); right panels, immunohistochemical labeling for PRKAR1A in tumoral pancreas (×200): 1, pancreatic acinar cell carcinoma showing no PRKAR1A labeling; 2, pancreatic acinar cell carcinoma showing a weak cytoplasmic PRKAR1A labeling; 3, pancreatic adenocarcinoma showing no PRKAR1A labeling; 4, IPMN, main and branch duct with low-grade dysplasia showing PRKAR1A labeling only in the nontumoral pancreas; 5, IPMN, main and branch duct with high-grade dysplasia, showing PRKAR1A labeling only in the nontumoral pancreas; 6, IPMN, branch duct with gastric metaplasia and low-grade dysplasia, showing PRKAR1A labeling only in the nontumoral pancreas.

In normal pancreas, staining for PRKAR1A was moderate in ductal and intense in acinar cells. Additionally, membranous and cytoplasmic staining for PRKAR1A was seen in sporadic IPMN, adenocarcinoma, and acinar cell carcinoma (Table 2).

Discussion

CNC is a multiple neoplasia syndrome characterized by the development of various tumors. Herein, we describe a possible new association of CNC with pancreatic tumors. Pancreatic tumors with various histology, i.e. adenocarcinoma, acinar cell carcinoma, and IPMN, were found in 2.5% (nine of 354) of an international registry of CNC patients.

In the United States, according to the National Cancer Institute, pancreatic cancer prevalence can be estimated at around 10 per 100,000 (16) and IPMN prevalence at around 25 per 100,000 (17). It is difficult to accurately estimate prevalence and incidence in our population, because we cannot precisely assess how long our patients have been followed. But we strongly believe that the number of observed cases is surprisingly high, especially in a cohort of patients with a mean age at last follow-up around 35 yr, when the median age at diagnosis for cancer of the pancreas is 72 yr of age (http://seer.cancer.gov/statfacts/html/pancreas.html). Additionally, two of 354 patients (0.5%) were diagnosed with pancreatic acinar cell carcinoma in contrast with fewer than one case per million people living in the United States (18). Indeed, pancreatic acinar cell carcinomas are about 100 times less frequent than pancreatic adenocarcinoma, because they represent only 1% of all pancreatic carcinoma, according to a Surveillance, Epidemiology, and End Results (SEER) database search from 1988–2003 (18).

PRKAR1A is the main causative gene of CNC, mutated in about 75% of patients (2). There was no PRKAR1A staining by immunohistochemistry in five of the six pancreatic tumors of CNC patients, whatever their histology, whereas PRKAR1A was present widely in normal pancreatic tissues as well as in sporadic tumors (from non-CNC patients). The only pancreatic tumors of CNC patients having a weak cytoplasmic staining had a pathogenic splice mutation likely leading to an abnormal and nonfunctional protein (15). Germline heterozygous PRKAR1A mutations were observed in five of the six patients (83%). Within the tumor, in the two pancreatic acinar cell carcinomas, only the mutated allele was detected. LOH has been previously demonstrated in one of these (7) and is strongly suggested in the other one because the mutation was in a hemizygous state in the tumor, whereas it appears heterozygous in the leukocytes. In the remaining four tumors, amplification of DNA from paraffin-fixed tissue was not technically possible. In these patients, the absence of protein in immunohistochemistry could be explained by LOH, a second inactivation of PRKAR1A (second hit), or a dominant-negative effect of the mutated PRKAR1A allele, but this remains to be proven.

These observations favor the hypothesis that pancreatic tumors are associated with CNC and that they are due to inactivation of PRKAR1A. They are consistent with previous reports of mice heterozygous for a conventional null allele of PRKAR1A (19), which develop a spectrum of tumors that overlaps with those observed in CNC patients, including pancreatic malignant tumors in two of the 44 mice (4.5%) (Stratakis, C. A., and E. Saloustros, unpublished data). In this study, allelic loss occurred in a subset of tumor cells, but this was not assessed in the pancreatic tumors studied here.

Beyond LOH and complete inactivation, PRKAR1A haploinsufficiency could also act as a weak tumorigenic signal (20) that would act synergistically with other oncogenic signals in a tissue-specific manner. This would in part explain the various pathological types of pancreatic tumors. Wnt/β-catenin signaling activation follows PRKAR1A haploinsufficiency in at least some tissues (20). Wnt signaling is known to be activated in pancreatic tumors (21–23), including acinar cell carcinoma (24). It has also been recently shown that PRKAR1A inactivation in adrenocortical cells decreased SMAD3 mRNA and protein levels, altering also the cellular response to TGF-β (25), in particular, causing resistance to TGF-β-induced apoptosis. The TGF-β pathway (26) and SMAD4 mutations (27) play a major role in pancreatic tumors. In pancreatic carcinoma, a tumor suppressor gene, SMAD4, is known to be inactivated in approximately 50% of human pancreatic adenocarcinomas (27–29) and in some nonfunctioning endocrine pancreatic carcinomas (30), through mutations and LOH, altering the cellular response to TGF-β. SMAD3 with SMAD2 are required for SMAD4 to form a functional heterocomplex for translocation to the nucleus. As hypothesized in the adrenal, pancreatic tumorigenesis in CNC could be due to interactions and cross-regulation between the PKA and TGF-β signaling pathways. It is possible that PRKAR1A inactivation could also decrease SMAD3 in pancreas, leading to resistance to TGF-β tumor-suppressing actions.

Additional evidence for PRKAR1A's involvement in pancreatic tumor formation, and especially that of pancreatic acinar cell carcinoma, is the frequent LOH in 17q22-24 found in an allelotyping study (31) of sporadic acinar cell carcinomas. Two of the nine pancreatic tumors described in patients with CNC were acinar cell carcinomas, a rare form of pancreatic cancer (with fewer than one case per million people living in the United States), accounting for less than 1% of all pancreatic cancers (32–34). Diagnosis of the tumor was based on morphological pattern and strong immunohistochemical positivity for chymotrypsin (35). Little is known about the pathophysiology of these tumors, but Wnt/β-catenin pathway has been described as being activated in about 20% of cases (24), through APC or CTNNB1 mutations. As already mentioned, Wnt/β-catenin pathway activation has been described in various CNC tumors, including PPNAD (36, 37) through CTNNB1 somatic mutations. In addition to acinar cell carcinoma (24), involvement of the Wnt/β-catenin pathway has been suggested to occur in a variety of pancreatic tumors, including pancreatoblastoma (38), pancreatic ductal adenocarcinoma (39–41), and solid pseudopapillary neoplasm (21, 22). It is noteworthy that pancreatic acinar cell carcinoma can harbor endocrine differentiation in about 40% of the cases; this component of the tumor can vary from a few scattered cells to more than 25% of the cells constituting the neoplasm (35, 42). In some cases, the tumor may have both acinar and ductal differentiation (43). However, these specific patterns were not observed in the available slides of our patients.

From a clinical perspective, pancreatic neoplasms caused 20% of overall mortality (8) and 35% of all cancer-related deaths among patients with CNC. The mean age at last follow-up of this particular cohort (2) was 34 yr, which was the age of the younger patient with pancreatic tumors diagnosed in this study. Thus, it is possible that the occurrence of pancreatic lesions in patients with CNC may increase with increasing follow-up or could have been underestimated because IPMN is usually nonsymptomatic, and no specific pancreatic screening has been done in this cohort, and consequently, only symptomatic disease might have been diagnosed. Due to the severity of pancreatic cancer, which has an overall 5-yr survival below 5% (44), and late stage at presentation, special attention should be paid to the pancreas of patients with CNC who are over 30 yr of age. Whether a systematic screening should be performed in those patients remains to be determined. First, no genotype-phenotype correlation or identification of a high-risk subgroup of patients with CNC was identified in our study, and consequently, this screening should be proposed to all CNC patients. Second, the efficiency of such a policy remains unknown, even in patients at high risk of pancreatic cancer (45), and the treatment of early lesion will need to be balanced with the risks of surgery and overtreatment. If systematic screening is planned, this should be done using magnetic resonance cholangiopancreatogram, as it is for relatives of familial pancreatic cancer patients.

We are aware of some limitations of our study. This includes the small number of patients with pancreatic tumor that make possible a fortuitous association or the lack of accurate information for three of the nine patients despite our multiple attempts to get additional information. Nevertheless, these limitations are inherent to a very rare disease, for which patients have been recruited worldwide and treated in various and numerous centers. Consequently, it is very unlikely that more accurate information would be available in the future or that more accurate information would be disclosed by another group or consortium. A possible bias of our study might be due to the systematic screening of these patients. Nevertheless, none of the tumors were discovered because of family history of pancreatic cancer, or systematic screening for pancreatic malignancy, because investigators at this time were not aware of the possible association between CNC and pancreatic neoplasm, and overall, only one of the six tumors was incidentally discovered. Second, we observed pancreatic tumors from varied histology, and evidence of a possible causal relation between PRKAR1A mutation and pancreatic tumorigenesis is mainly strong for patients with acinar cell carcinoma. A fortuitous association with pancreatic adenocarcinoma or IPMN, which are increasingly diagnosed (47), remains possible. Finally, our hypothesis remains to be proven by basic science evidence including dedicated animal models and functional studies.

In conclusion, we report the possible association of pancreatic nonendocrine tumors of various types, in particular acinar cell carcinoma, with CNC. The tumors demonstrate loss of PRKAR1A expression, in addition to the germline mutations, which suggests that PRKAR1A could function as a tumor suppressor gene in pancreatic tissue. Clinicians taking care of patients with CNC should bear in mind this additional possible tumor predisposition for their patients.