1 INDEX
Summary………...………2
Introduction Generalities on pancreatic cancer………...……3
Treatment strategies………...…….…….10
Role of Computed Tomography and Magnetic Resonance……….…12
Current Role of [18F]FDG PET/CT in pancreatic malignancies………..16
Purpose of the thesis………..24
Materials and methods………...25
Results………...……….27
Discussion………..30
Conclusion……….35
2
SUMMARY
Pancreatic cancer has been called the "silent killer" for its silent onset and subsequent aggressive and highly lethal behavior. Despite recent advances in radiological imaging, the strength of functional imaging lies in its ability to detect malignant disease irrespective of lesion morphology. The purpose of this thesis, which includes a review of prior literature, is to assess the ability of [18F]FDG PET/CT for distinguishing benign and malignant lesions of the pancreas; the potential of [18F]FDG PET/CT has been explored also regarding its ability to provide additional information over routine radiological imaging, in either solid and/or cystic pancreatic lesions (serous adenoma, MCNs and IPMNs) the frequency of which is increasing. To date, literature in this field is rather limited compared with other tumors, and the results reported are not always univocal.
We reviewed 72 [18F]FDG PET/CT examinations performed between January 2006 and December 2011 for diagnostic purposes in patients who had radiological suspicion of malignant pancreatic disease. Histologic confirmation was obtained in all patients.
Sensitivity and specificity of [18F]FDG PET/CT for discriminating between benign and malignant pancreatic lesions were 90% and 87%, respectively, when adopting an SUVmax cut-off of ≤ 2.4 (P<0.0001) for benign lesions; PPV was 86%, while PNV was 92%. Although [18F]FDG uptake in solid pancreatic malignancies is in general relatively lower than in other tumors, it allows anyway detection of malignancies. Moreover, SUVmax of pancreatic lesions correlates with overall survival of patients. Concomitant inflammatory disease do not affect our results.
When considering only the cystic lesions, sensitivity and specificity were 77.8% and 79.5%, respectively, based on an SUVmax cut-off ≤ 2.1 for benign lesions; the corresponding PPV was 44%, while NPV was 94% (P = 0.006). These results suggest that [18F]FDG PET/CT plays a relevant role for treatment decisions; furthermore, a patient-specific follow-up strategy could be defined when SUVmax of the lesion is lower thean 2.1, especially for those lesion that exhibit benign or uncertain radiological findings, as well as in patients for whom total pancreatectomy (with its associated higher risk of surgical complications) would be required based on radiological imaging alone.
This analysis indicates that, contrary to prior disappointing reports, [18F]FDG PET/CT is a highly useful method to characterize suspicious pancreatic masses, especially in patients with inconclusive radiological imaging. Optimized scanning protocols, including better control of blood glucose and correction for motion of organs in the upper abdomen, might further improve the diagnostic performance of [18F]FDG-PET/CT in patients with pancreatic masses.
3
INTRODUCTION
Generalities on pancreatic cancer
Pancreatic cancer has been called the "silent killer" for its silent onset and for its subsequent explosive and highly lethal behavior [1].
Approximately 90% of all pancreatic cancers originate from the pancreatic duct cells, giving rise to ductal adenocarcinoma, the most aggressive histotype. In the USA, pancreatic cancer is second only to colorectal cancer as the most common malignant tumor of the gastrointestinal tract, and represents the fourth leading cause of cancer-related death in adults. In the year 2012, 43,920 new cases of pancreatic cancer are expected (equally distributed between men and women), thus constituting 3% of all tumors in USA. The estimated number of deaths for pancraetic cancer is 37,390, representing 9% of the overall deaths for cancer. The cumulative 5-year survival rate for pancreatic cancer is 6%, but it varies according to stage of the diseaseat diagnosis: 22% in case of localized diasease, 9% in case of regional spread, and 2% in case of distant metastasis. It should be noted that only 15% of the patients have localized disease at diagnosis, while 16% have regional spread and 63% have distant metastasis [2].
Pancreatic cancer is difficult to diagnose, especially in its early stages, and surgical resection remains the only potentially curative approach; nevertheless, multimodality therapy consisting of systemic agents and external beam radiation, may improve survival. Staging of pancreatic cancers, which depends on the size and extent of the primary tumor, is continuously evolving and the latest version of the American Joint Committee on Cancer (AJCC) according to the TNM criteria is reported in Table 1 [3].
Primary Tumor (T)
TX Primary tumor cannot be assessed T0 No evidence of primary tumor Tis Carcinoma in situ
T1 Tumor limited to the pancreas, 2 cm or less in greatest dimension
T2 Tumor limited to the pancreas, more than 2 cm in greatest dimension
T3 Tumor extends beyond the pancreas but without involvement of the celiac axis or the superior mesenteric artery
T4 Tumor involves the celiac axis or the superior mesenteric artery (unresectable primary tumor)
4
Regional Lymph Nodes (N)
NX Regional lymph nodes cannot be assessed N0 No regional lymph node metastasis N1 Regional lymph node metastasis
Distant Metastasis (M)
M0 No distant metastasis M1 Distant metastasis
Anatomic Stage/Prognostic Groups
Stage 0 Tis N0 M0
Stage IA T1 N0 M0
Stage IB T2 N0 M0
Stage IIA T3 N0 M0
Stage IIB T1 N1 M0, T2 N1 M0, T3 N1 M0
Stage III T4 Any N M0
Stage IV Any T Any N M1
Table 1. AJCC Staging System of Exocrine Pancreatic Cancer
Early detection of pancreatic cancer is the only way to avoid its dismal prognosis; however, the currently available screening techniques are not so much sensitive, and prohibitively expensive [4]. Possible risk factors for pancreatic cancer include smoking, long-standing diabetes, chronic pancreatitis, familial pancreatic cancer, and advanced age. Hereditary predisposing conditions are the Peutz-Jeghers syndrome, hereditary pancreatitis, multiple endocrine neoplasia type-1 syndrome, hereditary nonpolyposis colon cancer, and the familial atypical multiple mole melanoma syndrome [5].
Early symptoms that prompt further investigations are painless jaundice (in case of bile duct obstruction) and weight loss, but they depend on the site of the tumor within the pancreas and on the degree of involvement. In advanced disease, palpable lymph nodes, hepatomegaly (due to metastasis), splenomegaly, ascites and peripheral edema (portal vein obstruction), palpable gallbladder (Courvoisier’s sign), left supraclavicular lymphadenopathy (Virchow’s node), or an abdominal mass might be appreciated.
Laboratory findings in pancreatic cancer are nonspecific. Prolonged biliary obstruction could cause alteration of liver enzymes and abnormalities of vitamin K-dependent clotting factors (due to malabsorption of fat-soluble vitamins). Pancreatic duct obstruction may result in
5 pancreatic atrophy and subsequent impaired glucose tolerance or frank diabetes in as many as 70% of patients [6].
Carbohydrate antigen 19-9 (CA 19-9) is the most commonly employed tumor-associated serum marker in pancreatic cancer. Although it has an 80% sensitivity, its serum levels increase in many benign and malignant gastrointestinal conditions, so its specificity for pancreatic cancer is relatively low (67.5%). For this reason, it is not recommended for use as a diagnostic tool nor as a sentinel of recurrent disease, although an increasing serum CA 19-9 may identify patients with progressive tumor growth. The presence of a normal CA 19-9 serum level, however, does not preclude recurrence [7].
From histologic point of view, pancreatic tumors can be distinguished according to their morphological, phenotypic and molecular features. These properties reflect the tendency to differentiation towards the three lines of cells found in normal pancreas: ductal, acinar and endocrine. The line of differentiation is the crucial factor that determines both biological characteristics and clinical behavior of pancreatic cancer [8]. However, there is considerable discrepancy between the prevalence of ductal carcinoma (90%) and acinar cell carcinoma (1%) and the percentage of acinar cells (80%) and ductal cells (5-10%) present in normal pancreas. This discrepancy is now explained with the hypothesis of a common derivation of cancers from progenitor stem cells, which locate in the "ductal" sector [9].
According to the WHO classification (Table 2), exocrine tumors of the pancreas are divided into three groups with different biological behavior: benign, uncertain potential to malignant, and malignant behavior [10].
Benign Borderline
(uncertain malignant potential) • Serous cystadenoma
• Mucinous cystadenoma
• Intraductal papillary-mucinous adenoma • Mature teratoma
• Mucinous cystic neoplasm with moderate dysplasia
• Intraductal papillary-mucinous neoplasm with moderate dysplasia
6
Malignant • Ductal Adenocarcinoma
Mucinous noncystic carcinoma Signet ring cell carcinoma Adenosquamous carcinoma
Undifferentiated (anaplastic) carcinoma
Undifferentiated carcinoma with osteoclast-like giant cells Mixed ductal-endocrine carcinoma
• Serous cystadenocarcinoma • Mucinous cystadenocarcinoma
non-invasive invasive
• Intraductal papillary-mucinous carcinoma non-invasive
invasive (papillary-mucinous carcinoma) • Acinar cell carcinoma
Acinar cell cystadenocarcinoma Mixed acinar-endocrine carcinoma • Pancreatoblastoma
• Solid-pseudopapillary carcinoma
Table 2: WHO Classification of exocrine pancreatic tumors.
Adenocarcinoma of pancreatic head (two thirds of the cases) presents on average a lower mean size (2-3 cm), compared to the rarer cancers of the body-tail (5-7 cm). The actual clinical impact of distinguishing between resectability and nonresectability is uncertain, since state-of-the-art treatment has demonstrated little impact on survival. Nevertheless, evaluating the actual disease burden is necessary to achieve a uniform definition of disease. A further, more surgically-oriented classification is based on localization of the tumor within the pancreas (head, body, tail, or uncinate process), which is important in order to plan surgical intervention.
Macroscopically, pancreatic cancer is usually characterized by a solid mass, with infiltrative margins, and hard-wood consistency. More rarely, it appear as "cystic" for the effect of necrotic-hemorrhagic changes. Cancers of the pancreatic head are usually associated with stenosis of the Wirsung’s duct and may extend to the Vater’s papilla and infiltrate the duodenum (in the more advanced stages). Body-tail carcinomas tends to invade the retroperitoneum, stomach, colon, omentum, spleen, and adrenal glands. Microscopically, pancreatic cancer is characterized by the presence of ductal-like structures dispersed in a stromal desmoplastic matrix. Tumor grading is based on cytoarchitectural criteria providing three degrees:
• G1: well-differentiated tubular structures; • G2: moderately differentiated tubular structures;
7 • G3 poorly differentiated glandular structures.
Desmoplastic reaction induces a considerable reduction of the vascular bed, which constitutes one of the signs useful for the radiological differential diagnosis between normal tissue and cancer. Histological variants are represented by all tumors that at least have a classic ductal adenocarcinoma component. They are mainly represented by:
- mucinous non-cystic carcinoma, characterized by massive mucus production and a papillary mucinous intraductal component;
- signet-ring cell carcinoma, constituted by a prevalence of mucinous cells;
- adenosquamous carcinoma, characterized by mixed aspects of adenocarcinoma, squamous cell carcinoma and frequent sarcomatous-like areas with very aggressive clinical behavior; - undifferentiated (anaplastic) carcinoma, with no recognizable distinct differentiation into
structures;
- undifferentiated carcinoma with osteoclast-like giant cells, characterized by the presence of mononuclear neoplastic epithelial and a reactive component of osteoclast-like giant cells; - mixed ductal-endocrine carcinoma, characterized by the presence of an endocrine component
higher than 30%.
The differential diagnosis of ductal adenocarcinoma arises with most forms that mught have similar clinical presentation, which are represented by chronic pancreatitis, in its various forms, and by tumors of the Vater’s papilla.
Precursors of invasive carcinoma
Three distinct non-invasive lesions can be recognized: the first is constituted by microscopic findings and includes a whole series of changes included in the category of Pancreatic Intraepithelial Neoplasia (PanIN), while the other two are macroscopical lesions, radiographically detectable, and are represented by Intraductal Papillary Mucinous Neoplasms (IPMN) and Mucinous Cystic Neoplasms (MCN)
Pancreatic intraepithelial neoplasms (PanIN) consist of small epithelial tumors of the ducts, visible only on histologic analysis, with a diameter <0.5 cm. They are characterized by columnar epithelium frequently secreting mucous, or by flat epithelium, or by micropapillary proliferations. They can present as a large spectrum of morphologic changes with varying degrees of atypia: absent atypia (PanIN-1A-B), moderate atypia (PanIN-2), or high-grade atypia and frequent mitoses.
8 Intraductal Papillary Mucinous Neoplasms (IPMN) consist of mucinous tumors frequently papillary, which can develop in the main duct or in second-order ducts, always associated with ductal ectasia macroscopically evident, with diameter greater than 1 cm. They are divided into three subtypes:
1. "central" or “main duct” with involvement of the Wirsung's duct;
2. "peripheral" or “brunch duct”, with exclusive involvement of second-order ducts; 3. "mixed", with involvement of both.
Approximately 70% of IPMNs with envolvment of the main duct show malignant degeneration [11]. The central type IPMN is often accompanied by protrusion into the lumen of the duodenum of both papilla maior and minor, with secretion of mucin. IPMNs were first recognized in 1982 and finally classified by the WHO in 1996. IPMNs have been for a long time unrecognized, classified under various denominations and included among the mucinous cystic tumors. Traditionally they were considered as rare lesions, but in the current clinical practice represent the 20%-30% of all resected pancreatic cancers. They have a slight prevalence in men and the mean age of occurrence is 65 years. Involvement of the pancreatic head is present in 70%-80% of cases. The most distinctive feature is intraductal growth, with diffuse or segmental dilatation of the main duct and/or of the second-order ducts, and absence of mechanical stenosis.
There are two main variants:
1) predominantly papillary variant, in which the duct presents an intraluminal vegetation of villous type. Detecting this type of growth constitutes a valid radiologic diagnostic feature to distinguish the tumor from pancreatitis with secondary dilatation of the ductal pancreatic system.
2) mainly ectasic variant, in which the main duct and the ectasic the second-order ducts are completely filled with mucus, but have a flat epithelium.
Microscopically, the degree of epithelial dysplasia is the element allowing subclassification into: adenomas (mild atypia), borderline (moderate grade atypia), and carcinoma (severe atypia). Invasive cancer is frequently of the "colloid" type (characterized by indolent clinical course) and only rarely of the "tubulo-glandular" type (characterized by more aggressive behavior, sometimes similar to ductal adenocarcinoma). From the molecular and immunophenotypic point of view, papillae can be distinguished as intestinal type (30%-40% of cases), biliary-pancreatic type (20% of the cases), papillae type gastric (30%-40% of cases), and oncocytic type.
9 IPMNs differ molecularly from adenocarcinoma for a reduced incidence of mutations of K-ras, p53, p16, for preserved expression of DPC4, and for inactivation in a third of the patients of the STK1/LKB1 gene. Progression from low-grade lesions to invasive carcinoma is not considered as a linear sequence similar to what is observed in the classical adenoma-carcinoma sequence of colon cancer. In fact, both morphological and immunophenotypic heterogeneity of the process of neoplastic transformation has to go through specific programs of transdifferentiation of intestinal type (indolent IPMN) or of pancreato-biliary type (aggressive IPMN). For the oncocytic variant a distinct carcinogenesis process, probably due to oxidative stress, is assumed.
The differential diagnosis of malignancy for central IPMN arises initially with respect to chronic pancreatitis and adenocarcinoma, while for peripheral IPMN tumors arises primarily with mucinous cystic neoplasms and cysts retention, as well as with inflammatory pseudocysts and cystic dystrophy of duodenal wall.
Mucinous Cystic Neoplasms (MCN, more prevalent in women) consist of encapsulated tumors unrelated to the ductal system, mainly localized in the tail of the pancreas and characterized by the presence of ovarian stroma [12].
MCNs represent approximately 30% of all cystic tumors of the pancreas. Macroscopically, these tumors are multicystic, roundish, pseudocapsulated and not connected with the main duct or the second-order ducts. Size can vary from 2 to 35 cm (mean diameter 12 cm). A smooth inner cystic surface appears in benign MCNs, while the presence of papillary growths and solid areas usually characterizes the cistoadenocarcinomas.
These tumors are classified according to degree of epithelial dysplasia: adenomas (mild atypia), borderline forms (moderate atypia), and noninvasive adenocarcinomas (severe atypia). Invasive MCNs usually exhibit features similar to those of ductal adenocarcinoma of the pancreas, even if the two are still considered as different forms of cancer.
10
Treatment strategies
Surgical resection remains the primary modality, when feasible, because it can lead to long-term survival and provides effective palliation [13]. The importance of discriminating between resectability and nonresectability is uncertain, since state-of-the-art treatment has so far demonstrated little impact on survival. Pancreatic surgery is still evolving as a multidisciplinary approach; the development of modern anesthetic techniques, appropriate cancer therapy, peri- and post-operative treatment by dedicated staff has reduced mortality in centers with a high volume of patients [14]. Only 20% of the patients have resectable tumors at diagnosis [15]. Duodeno-cephalo-pancreatectomy (DCP) is the standard surgical treatment for tumor localized in the head of the pancreas. For many years, the classic DCP, primarily described by Kausch and Whipple, was considered as the gold standard surgical approach for adenocarcinoma of the pancreatic head [16]. Subsequently, several studies showed the DCP with preservation of pylorus (PP-DCP) could be considered an equal radical surgical treatment, with long-term outcomes similar to those of classical DCP, even if adopting a less mutilating approach. Other surgical procedures include distal pancreatectomy with or without removal of the spleen (for tumors located in the body-tail region), total pancreatectomy (for multifocal tumors) and the rarer middle pancreatectomy, which is applied in sporadic cases of benign tumors located in the pancreatic body. A more aggressive surgical strategy with vascular resection or lymph node dissection [17]. has not improved overall survival. The role of surgery in metastatic disease or after neoadjuvant radiochemotherapy remains controversial.
The role of postoperative chemotherapy or chemoradiotherapy remains debated, as randomized studies provide contradictory results [18] and clinical trials are appropriate alternatives for treatment of patients and should be considered prior to selecting purely palliative approaches.
Stage I and stage II pancreatic cancer can be treated by local surgical resection, with operative mortality rates of approximately 1% to 16% [19]. Complete resection can yield 5-year survival rates of 18% - 24%, but ultimate control remains poor because of the high incidence of both local and distant tumor recurrences [20].
Standard treatment options are surgery or surgery plus postoperative fluorouracil (5-FU) chemotherapy and external beam radiation therapy. Radical pancreatic resection can be performed by the Whipple procedure (pancreaticoduodenal resection of head and uncinate process masses) or by distal pancreatectomy (tumors of the body and tail of the pancreas). Total pancreasectomy can be necessary to obtain negative margins of the surgical specimen. Other
11 therapeutic options (for example, with other chemotherapeutic agents) are under clinical evaluation [21].
Stage III pancreatic cancers are technically unresectable, since loco-regional spread is already present. Chemotherapy with gemcitabine is the standard treatment option for unresectable patients. Furthermore, these patients may benefit from palliation of biliary obstruction by endoscopic or percutaneous biliary stent placement. Gastric bypass can be performed, if necessary [22].
Stage IV patients or patients with recurrent disease show low objective response rates, so every clinical trials must be considered as a suitable treatment. Some patients have palliation of symptoms when receiving chemotherapy with 5-FU or Gemcitabine [23]. The use of capecitabine and erlotinib in patients with gemcitabine-refractory first-line treatment of advanced pancreatic cancer has been recently associated with modest improvements in overall survival [24]. Pain-relieving procedures such as celiac or intrapleural block and palliative biliary bypass or stent placement must be considered.
12
Role of Computed Tomography and Magnetic Resonance
CT, magnetic resonance imaging (MRI), magnetic resonance Cholangiopancreatography (MRCP), endoscopic ultrasounds (useful in evaluating small lesions and as a guide for biopsy) and endoscopic retrograde cholangiopancreatography are the most widely applied methods for diagnosing and staging pancreatic cancer. In fact, morphologic imaging enables to visualize any infiltration of the vessels, an important parameter for assessing tumor resectability.
Contrast enhancement CT is considered the method of choice for evaluating pancreatic lesions, due to its high spatial resolutions. The most modern equipment with detectors from 64 to 128 slices, allows to obtain volumetric images with thin collimation (0.625-1.25 mm), and also visualizes vascular structures, as well as possible infiltration of surrounding organs.
On the other hand, MRI with the advent of new hardware (1.5-3 Tesla) and Phased-array coils demonstrates high diagnostic accuracy in the study of all pancreatic pathologies. In particular, the multiparametric approach and multiplanar recostruction of MRI allow the use of several sequences of impulses and images acquisition in apnea both with or without contrast enhancement.
Ductal Adenocarcinoma
Pancreatic adenocarcinoma shows massive deformities of the pancreatic contour, poorly delineated before contrast enhancement (CE). During the "pancreatic phase" (30-50 seconds after contrast injection) adenocarcinoma appears predominantly hypodense, with smooth contours and generally no calcifications. If small, the tumor is generally surrounded by normal parenchyma and often causes dilatation of the Wirsung’s duct, with atrophy of the upstream parenchyma. When the tumor involves the pancreatic head, jaundice is often present due to obstruction of the principal bile duct; portal phase reconstructions with MIP show in detail the level of obstruction In rare cases, if the tumor is smaller than 2 cm, obstruction can only to be identified by secondary signs, such as the presence of parenchymal atrophy or swelling of the pancreatic structure. In the more voluminous lesions, the peripancreatic adipose tissue is often infiltrated with retro peritoneal desmoplasia. In the portal phase, metastatic liver lesions are recognized as hypodense areas. Peritoneal carcinomatosis is sometimes identified thanks to the presence of ascites, even of small size.
There is consensus that DCP with resection and reconstruction of the portal-mesenteric vein trunk is the standard procedure for pancreatic cancer that affects the portal-mesenteric vein trunk, if there is enohgh disease-free porta-mesenteric vein tissue available for portal-mesenteric
13 vein trunk reconstruction upstream and downstream. Moreover, the tumor must not involve the superior mesenteric artery or hepatic artery. Thus, recognizing possible infiltration of the portal vein and superior mesenteric vein is crucial for planning surgery [25-26].
Vascular invasion in pancreatic cancer has a key role in determining treatment and prognosis [27]. While in the past the most widely method used to evaluate vascular infiltration was angiography, today vascular staging is performed by tomographic methods (CT and MRI). CT signs of vascular infiltration are defined by the relations between tumor and vascular structures. CT detectable suspicion of vascular invasion are the presence of hypodense neoplastic tissue which comes in contact with the vessel, focal or circumferential obliteration of the perivascular adipose cleavage plan (encasement) between tumor and vessel, focal reduction in caliber of vessels or complete obstruction of lumen, and frank intraluminal thrombosis. In order to determine a proper CT vascular infiltration , various classifications have been proposed, based on the percentage of circumferential involvement of vessels or on indirect sign of infiltration, such as modification of the vessel caliber, the “teardrop sign” (a "drop" deformation of the vein or dilated pancreatic duodenal veins, in cases of infiltration of the mesenteric vein) [28-29].
MRI demonstrates pancreatic adenocarcinoma typically as a hypointense lesion for its necrotic and desmoplastic characteristics, compared to normal parenchyma, in both T1- a T2-weighted images with saturation of adipose tissue signal, and after injection of intravenous contrast medium, especially in the pancreatic phase. Some small tumors may show peripheral enhancement. The "double duct" sign at MRCP is considered pathognomonic of adenocarcinoma, while the so-called "duct penetrating" sign is considered typical of chronic pancreatitis. After injection of paramagnetic contrast enhancement, images can be acquired in apnea with 3D T1-weighted sequences and clipping the adipose tissue, showing vascular the structures arterial with a very accurate assessment of the extent of the tumor. Also vascular infiltration can be detected with MRI.
Resectable PCs are considered all tumors with no evidence of distant metastases, no evidence of portal mesenteric trunk involvement (contact, distortion, neoplastic thrombosis, deformation) and the presence of a detectable fatty tissue around the celiac axis, hepatic artery and superior mesenteric artery.
Borderline resectable PCs are considered all tumors with no evidence of distant metastases and limited involvement of the portal mesenteric trunk (with enough disease-free venous tract for reconstruction) and/or involvement of gastroduodenal artery up in close contact with the hepatic artery, and/or contact of tumor with superior mesenteric artery less than 180°.
14 Locally advanced PCs are considered all tumors with no distant metastases, with celiac trunk involvement and/or involvement of the superior mesenteric artery more than 180°, and/or mesenteric portal trunk without involvement with no possibility of reconstruction [30].
Another important factor for staging is infiltration of the peripancretic adipose tissue, which represents a critical factor because of the high incidence of persistence/recurrence of disease due to the difficulty of achieve a radical resection at that level [31], especially of the retroperitoneal adipose tissue in cancer of the pancreatic head. CT can identify irregularities or increased density or complete obliteration of the adipose tissue [32].
Cystic neoplasms
Cystic neoplasms include serous cystadenoma, MCTs and IPMNs. Other rare tumors of the pancreas include solid pseudopapillary tumors, endocrine tumors (that are rarely cystic), as well as metastases, teratomas and lymphangiomas [33].
Microcystic serous adenomas are typical of women 60-70 years old. At CT they appear lobulated, hypodense in the baseline (non-contraste-enhanced) study. In 20% of the cases they show calcifications in the center of the mass, corresponding to a central scar. After intravenous contrast enhancement they are hypervascular, with a tipical "beehive" appearnce. Differential diagnosis is much more difficult for serous adenomas of men with fluid density and oligocystic morphology [34]. At MRI serous microcystic cystadenomas appear hypointense in T1-weighted imaging, but hyperintense in case of intra-tumoral bleeding. The tumors are instead hyperintense on T2 weighted images, with "beehive" appearnce evident in the MRCP images. The injection of paramagnetic contrast agent enhances signal intensity of the fibrous septa, as well as that of the central scar. MRI can define the microcystic serous from borderline lesion, and recognizes the number of cysts if lower than 6 with a diameter ranging from a few mm to 2 cm. In the macrocystic and oligocystic forms, cysts are more voluminous, reaching diameters from 2 to 8 cm; in these cases the absence the central scar and fibrous septa makes differential diagnosis more problematic, especially in cases with a single cyst [35].
Mucinous cystic tumors of the pancreas prevail in the body-tail parenchyma and exceptionally communicate with the main duct. They can evolve into mucinous cystoadenocarcinomas. Cystic lesions are poorly defined in the baseline study, appearing as hypodense in the "pancreatic phase". They may have peripheral calcifications, which are raise the suspicion for malignancy. At MRI they show a fluid content, hyperintense on T2-weighted images, with variable intensity on T1-weighted images and in those with subtraction of the adipose tissue signal, reflecting mucin concentration. The injection of paramagnetic contrast
15 agent outlines the lesions, with hyperintense margins. Calcifications can be recognized in both T1- and T2-images as hypointense, while the cystic component is better shown by MRCP sequences, which allow to recognize the relationship between the lesions and the main pancreatic duct.
IPMNs prevail in the uncinate process, and in these cases the main pancreatic duct can be slightly dilated. Communication with the main duct or its branches without stenosis is a crucial sign for diagnosis; it is demonstrated by CT multiplanar reconstructions, and even better by MRCP with three-dimensional reconstructions. Usually these tumors do not present calcifications and/or ductal stenosis, which are more typical of chronic inflammation. In “main ducts IPMNs” CT usually shows diffuse dilatation of ducts with possible defects determined by both mucus and papillary lesions. At MRI, the presence of mucins and neoplastic intraductal mural nodules enhanced by paramagnetic contrast agent, bears poor prognosis. The sign of papilla protrusion in the duodenal lumen, detectable with both CT and RMI, is considered pathognomonic. Degeneration is related to advanced age, main duct involvement, diabetes, size greater than 3 cm, and the presence of multiple lesions or parietal nodules. In MRCP the use of secretin can improve diagnostic accuracy, by showing normal pancreatic exocrine function. A reduction of exocrine function is instead primarily related to chronic pancreatitis [36].
16
Role of [18F]FDG PET/CT in pancreatic malignancies
The limitation of radiologic imaging is that it only provides size-related information, without providing any metabolic information which is very useful for staging and, above all, for post-therapy restaging in many oncologic patients. Most of the studies concerning the usefulness of PET (and PET/CT) in patients with pancreatic cancer have been performed with [18F]FDG [37]. Thus, in patients with suspected/proven pancreatic lesion, [18F]FDG PET/CT is usually performed after conventional imaging. This nuclear medicine technique evaluates the functional pattern of a lesion, and has been applied since its early development also in patients with pancreatic tumors, although results were not always unequivocal. To date, the literature is rather limited compared with what has been published for other tumors.
Fluorine-18-fluoro-2deoxy-D-glucose ([18F]FDG), a glucose analog, is the most commonly used radiopharmaceutical for PET imaging. 2-Deoxy-D-glucose (DG) was first developed in 1960 as a chemotherapeutic agent, as a means to inhibit glucose consumption by cancer cells. In 1976 Wolf and his colleagues at Brookhaven National Laboratory developed the synthesis of [18F]FDG to study cerebral glucose metabolism, and the first [18F]FDG-PET brain imaging studies were performed in 1977 at University of California, Los Angeles.
[18F]FDG is trasported through the cell membrane via glucose transport proteins (GLUTs) and phosphorylated by hexokinase to [18F]FDG-6-PO4-. [18F]FDG-6-PO4- is not a substrate for the subsequent enzymatic conversion to fructose-6-phosphate by phosphohexose, and thus does not further participate in the glycolytic pathway and becomes trapped in the cell because of its negative charge. The low activity of the reverse enzyme, glucose-6-phophatase leads to accumulation in the tumor cell of [18F]FDG-6-phosphate. The extent of [18F]FDG uptake depends on glucose metabolism and the most intense activity is to be found in the normal brain cortex (9% of injeced activity within 80-100 minutes), in the sites of inflammation and infection, and in neoplastic cells. Cancer cells exhibit high rates of metabolism with increased glucose metabolism (Warburg effect). Gene-mediated regulation of GLUT transporters and hexokinase activity contributes to the increased accumulation of [18F]FDG in tumor cells relative to normal tissues; furthermore, the time-related pattern of [18F]FDG accumulation in malignancy can be different from that of benign lesions and inflammatory processes due to this different pathophysiologic condition. The mechanisms and reasons for elevated glucose metabolism in cancers are multifactorial and include tumor-related components, biochemical and molecular alterations (e.g., glucose metabolic pathway, hypoxia), and nontumor-related components (e.g., inflammation). [18F]FDG is excreted through the kidneys and about 20% of injected activity is
17 excreted in the urine within 2 hours after administration. The uptake of [18F]FDG, that competes with endogenous glucose, by tumor tissue depends ont the plasma glucose levels; in fact, in case of high blood glucose levels (as well as low insulin level), the uptake of [18F]FDG is reduced. Many other factors (e.g., chemotherapy, radiation therapy or recent administration of Granulocyte-Colony Stimulating Factor) interfere with [18F]FDG uptake, thus possibly resulting in either false-positive or false-negative imaging findings [38].
Standardized uptake values are widely and increasingly used in clinical studies in addition to visual assessments. SUV is a measurement of the uptake in a tumor normalized on the basis of a distribution volume. It is calculated as follows [39]:
The following formula is applied in the case of plasma glucose correction
In these calculations, Actvoi is the activity measured in a certain volume of interest (VOI or ROI, regions of interest), Actadministered is the administered activity corrected for the physical decay of [18F]FDG up to the start of acquisition, and BW is body weight. Patient's height, weight, and gender should be reported to allow for other SUV normalisations (LBM, BSA). It is recommended to measure plasma glucose levels using validated methodology and calculate SUV with and without plasma glucose correction in all stuides monitoring tumor response to therapy. It should be noted that the measured glucose content (Glucplasma) is normalised for an overall population average of 5.0 mmol/L, so that the SUVs with (SUVglu) and without (SUV) correction for glucose content are numerically practically identical (on average) [40].
Cancer response to treatment can be assessed by [18F]FDG PET by calculating SUV on the highest image pixel in the tumor regions (SUVmax). Alternatively, tumor volume can be estimated using a region of interest, and average SUV inside the region can be calculated and reported as such (SUVmean) or multiplied by tumor volume to calculate the total glycolytic volume, TGV [41], Nahmias and Wahl [42] reported that the use of SUVmax has worse reproducibility (3% ± 11%) than does the SUVmean value (1% ± 7%).
Calculation of SUV does not require blood sampling or dynamic imaging. The imaging must take place at a late time point, and always at the same time point if results are to be compared [43].
18 Sensitivity and specificity of [18F]FDG PET/CT in pancreatic disease could decrease if blood glucose level is higher than 130 mg/dL, optimal blood glucose level ranging from 60 to 130 mg/dL [44]. Subsequent studies showed that pancreatic adenocarcinoma may present an [18F]FDG accumulation time-pattern slower than that observed for other neoplasms. In this regard, some authors have demonstrated the usefulness of an additional late acquisition (at 2 hours) to assess progressively increasing activity of glucose transporters and esokinase [45]. In current clinical practice, acquisition of the PET images for evaluating pancreatic lesions begins at least 1 hour after radiopharmaceutical administration. Only in particular cases with equivocal findings at the standard 1-hour time point results, a later abdominal scan is performed in order to confirm/exclude the presence of any indeterminate focal uptake area.
Despite these problems that make pancreatic tumor "different" from other tumors, [18F]FDG PET/CT could be worth because of its ability to explore the whole body simultaneously, providing images from the skull to the lower limbs and therefore increasing the possibility to detect distant metastatic disease.
Thanks to recent technological advances, the use of hybrid PET/CT machines with integrated multislice CT has become an essential tool in order to increase diagnostic accuracy for all types of cancers. In clinical practice CT is performed in the low-dose mode, since it is normally used to better localize possible foci of [18F]FDG uptake. Nevertheless, it has recently been demonstrated that the use of enhanced PET/CT as a 1-stop-shop imaging protocol for assessing the resectability of pancreatic cancer is feasible and accurate. Enhanced PET/CT is significantly superior to PET/CT alone [46].
At present, the main indications to the use of [18F]FDG PET in the suspicion of pancreatic cancer is the differential diagnosis between benign and malignant lesions, staging of pancreatic adenocarcinoma, evaluation of recurrence/restaging, evaluation of response to therapy, and treatment planning for external beam radiation.
Conventional radiologic imaging, however accurate, may not be able to differentiate with certainty benign from malignant lesions, especially when the non-invasive approach is indicated. In clinical practice, [18F]FDG PET/TC is required in case of uncertain/inconclusive radiological imaging, especially for better characterizing the incidental radiological/endoscopical finding of pancreatic lesion with features of possible malignancy.
To date there is no general consensus about the role of [18F]FDG PET/TC in pancreatic cancer; reports in the literature are scarce, and the results of different studies are contradictory, especially regarding the use of the more recent hybrid [18F]FDG PET/TC equipment. Some authors showed using the semiquantitative SUVmax index that a cut-off of 3.0 would allow to
19 discriminate benign from malignant lesions, with values of sensitivity and specificity of PET higher than those of CT (92% and 85%, respectivcely, compared to 65% and 61%) [47].
Subsequent studies have emphasized that the use of [18F]FDG PET may not be sufficient for the differential diagnosis. On the other hand, SUVmax appeared to correlate well with overall survival and represented a significant prognostic information [48].
Regarding lesion characterization, the main cause of false-positive findings at [18F]FDG PET/TC is the inflammatory status due to chronic pancreatitis; the distribution of [18F]FDG-avid areas within the parenchyma can aid in the diagnosis, as diffuse high uptake in the whole pancreas is more often due to an inflammatory status, while focal [18F]FDG uptake is most frequently related to cancer [49].
Nevertheless, pancreatic cancer is sometimes accompanied by pancreatitis, so that about 24% of [18F]FDG uptake in tumor tissue may be related to inflammatory cells, meaning that it might be impossible to clearly distinguish the two different components. Furthermore, cancer cells can occasionally infiltrate diffusely the whole pancreas, so that patients with pancreatic cancer can actually show diffusely high [18F]FDG uptake in the whole organ. On the other hand, benign lesions (such as autoimmune pancreatitis) can occasionally present with a focal [18F]FDG PET uptake pattern. Moreover, there is a wide overlap in SUVs between inflammatory and malignant pancreatic disease, so that some authors suggest that only visual qualitative interpretation can provide accurate differential diagnosis [50].
False-positive findings can also occur after recent surgery or endoscopy, as well as post-radiotherapy, or in the event of an abscess, massive lymphocyte infiltration, retroperitoneal fibrosis, hemorrhage in pancreatic pseudocysts, inflammatory pseudotumors, pancreatic tuberculosis and focal high-grade dysplasia. Autoimmune pancreatitis is an additional cause of false-positive [18F]FDG PET/TC, as it can occasionally present with a focal [18F]FDG uptake pattern (although high uptake in the salivary glands can suggest the occurrence of autoimmune pancreatitis [51].
Most false-negative results can occur for small size tumors and/or in the presence of elevated serum glucose levels (it should be reminded that many of these patients suffer from pancreatic insufficiency and diabetes; high serum glucose levels compete with [18F]FDG for glucose transporters, thus reducing the sensitivity for detecting malignant lesions) [52]. Tumor cellularity has also been reported as an important cause of false negative findings: scirrhous type, cystic type or dismoplastic reaction are not rare features in pancreatic malignancy, and they are well known to be poor in cellularity even with large tumor size [53].
20 It should also to be emphasized that some histological types of pancreatic tumor besides neuroendocrine tumors (NETs) (such as mucinous cancers) do not have a high glucidic metabolism, and therefore could be the cause of false negative findings.
Regarding the accuracy of [18F]FDG PET/TC, a recent meta-analysis suggest that, although its inclusion in the diagnostic work-up may enhance the diagnosis of pancreatic malignancy, its usefulness will vary depending upon the pretest probability of the patient, the CT findings, and the provider’s testing thresholds. In fact, average sensitivity and specificity changes from 92% to 68%, respectively (after a positive CT report) to 73% and 86% (after a negative CT report), while they are reported to be 100% and 68%, respectively, after an indeterminate CT report [54].
Finally, thanks to technological advances, the preliminary experience with integrated diagnostic PET/CT show that, in lesion greater than 1 cm, [18F]FDG PET/TC still has significant diagnostic value, considering that its PPV (83%) and NPV (82%) did not differ substantially from those of ultrasound and ERCP [55].
Moreover [18F]FDG PET/TC has a role also in the management of pancreatic cancer patients. Its impact in terms of cost/benefit and changes of therapeutic strategy in 16% of cases has been shown by Heinrich et al [56].
Besides differential diagnosis between malignant and benign lesion, [18F]FDG PET/CT may have a role also in staging. About 40% of the patients with pancreatic cancer that are judged as resectable by preoperative imaging turn out to be unresectable at the time of operation. Therefore, correct staging is the most important aim of imaging in pancreatic malignancies, in order to define prognosis and to select the most appropriate management of the disease [57].
As mentioned above, the performance of [18F]FDG PET/CT regarding local staging (T parameter) is limited by its relatively poor spatial resolution; in this regard, conventional radiologic imaging certainly better demonstrates the relationship between the tumor and the adjacent organs or vascular structures.
N staging is one of the most important independent prognostic factors for the management of patients with pancreatic cancer [58]. Unfortunately, both conventional imaging and [18F]FDG PET/CT have low sensitivity for detecting lymph node LN metastasis, because of scattered radiation from tracer accumulated in the primary tumor and limited spatial resolution for detecting microscopic LN metastasis. Furthermore, false positive lymph nodes could occur in reactive locoregional lymphadenopathies after biliary interventional procedures.
For the detection of distant metastasis, the capability of scanning the entire body with a single examination is an obvious advantage of [18F]FDG PET/CT, compared with the other
21 imaging modalities. The detection of distant metastasis or of unexpected extra-pancreatic lesions by whole-body [18F]FDG PET/CT changes dramatically the management of the patient, as such patient will be spared the unnecessary surgery.
Several studies have demonstrated the superior diagnostic accuracy of whole-body [18F]FDG PET/CT in the detection of distant metastasis compared with other modalities, such as CT or US. In fact, the sensitivity of PET in this application is between 81% and 88%, versus 38%-56% for CT. Liver is the most common site of distant metastasis from pancreatic cancer, followed by the lungs and bone marrow. Direct spread into the peritoneum (peritoneal carcinomatosis) is also not uncommon, and often missed on conventional anatomical imaging. The accuracy of [18F]FDG PET/CT in detecting liver metastasis is 94%, similar as that of conventional imaging modalities (90%). False negative [18F]FDG PET/CT findings can be observed in case of small size lesions, but can also be linked to some heterogeneity in [18F]FDG uptake within the hepatic parenchyma, and/or to respiratory motion artifacts [59].
The routine use of [18F]FDG PET/CT in patients with suspected or ascertained pancreatic cancer has not been adopted by many centers, and its use is recommended only in the evaluation of diagnostically challenging cases, especially in patients with biliary narrowings without evidence of malignancy at conventional imaging. The impact of [18F]FDG PET/CT on the management of patients with pancreatic cancer has been well established in staging, as it can modify the therapeutic approach in 30% to 60% of the newly diagnosed cases [56].
[18F]FDG PET/CT could play a role also for the assessment of response to therapy. Since chemotherapy can improve survival as well as quality of life (though not changing the dismal long-term prognosis), the identification of early responders to chemotherapy is important in order to avoid unnecessary toxicity in patients who are not going to respond. Conventional morphologic imaging is not accurate to this purpose, mainly because of the presence of fibrosis induced by treatment and because of the relatively long time required for detecting reduction in size of the tumor lesions possibly induced by therapy. In this scenario, [18F]FDG PET/CT may provide a sensitive parameter for earlier assessment of response to therapy versus conventional imaging [60].
During follow-up elevated serum CA 19-9 levels have a 69% PPV for pancreato-biliary malignancy. In patients who have already been treated for pancreatic cancer but whose serum CA19-9 raises (yet whose conventional imaging is negative for recurrence), [18F]FDG PET/CT could detect unexpected sites of recurrence (such as local recurrence, liver, lungs, peritoneal dissemination and distant lymph nodes metastasis). On the other hand, [18F]FDG PET/CT could
22 also exclude the presence of recurrence in cases in which findings that are indeterminate or equivocal by other imaging modalities.
The aim of prognostic evaluation is to stratify patients into groups that would mostly benefit from adjuvant and neoadjuvant treatments. [18F]FDG PET/CT seems to be a useful tool for prognosis: a negative PET scan at one month post-chemotherapy is correlated with longer overall survival and, although there is no relationship between SUVmax and overall survival of patients, a high SUVmax seems to be associated with a shorter survival in the subgroup with inoperable tumors [61].
For radiotherapy planning, current literature shows that the accuracy of target volume delineation can be improved beyond the mere CT-based delineation, by integrating the PET data with the CT ones: about 36% of the patients, when using PET and CT information, have an enlarged field of treatment, with no evidence of increased toxicity. However, the actual impact of change management induced by [18F]FDG PET/CT on the overall survival of unresectable patients has not yet been assessed [62].
The role of PET/CT in cystic tumor
Cystic pancreatic tumors are observed with increasing frequency in asymptomatic patients as incidental findings during US or CT abdominal imaging. Within the whole group of cystic pancreas tumors, preoperative imaging, including magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasound, usually allows three main lesions to be differentiated: serous adenoma, MCNs, and IPMNs.
While in the two former tumors the indications are well recognised (radiological follow-up for serous and resection for mucinous neoplasms), IPMN still represents a more critical field, potentially bearing adenoma, in situ carcinoma or invasive carcinoma; these lesions cannot be differentiated with sufficient accuracy by the currently available imaging techniques. With no risolutive imaging, pancreatic resection, the only available therapy, may represent an over-treatment due to the low malignant potential of more than half of these lesions. [63]. The therapeutic strategy for IPMN mainly depends on the suspicion of malignancy emerging from the preoperative workup. Although some radiological features have been described that may indicate a definite risk (all “main duct” tumors and “branch duct" tumors larger than 3 cm, symptomatic or harbouring parietal nodules) no clinical, biological, biochemical and radiological factors can be considered sufficiently accurate to confirm or exclude the presence of a malignant component in cyst walls, frequently leading to an aggressive approach to lesions which finally prove to be benign [64].
23 Considering that pancreatic resection is a very invasive surgical procedure (with 5%-10% mortality and 20%-40% morbidity) and that in most IPMNs the only therapeutic procedure able to remove the disease entirely is total pancreatectomy, a diagnostic tool improving the specificity of malignancy diagnosis may represent a substantial progress in managing these lesions. The accuracy of CT scan is low in this respect, while MRCP is the gold standard diagnostic procedure, due to its ability to demonstrate the Wirsung duct anatomy and its connections to branch-sided cysts, and to exclude the presence of parietal nodules or filling defects. [18F]FDG PET/TC is a functional imaging technique that has been proposed as a valuable tool for diagnosing and staging different malignancies, including pancreatic adenocarcinoma.[65].
Regarding cystic lesion, literature is still unclear. Some published experiences have reported low [18F]FDG PET sensitivity (57%) in detecting the presence of malignant tissue in cystic lesions [66]. On the contrary, Sperti and coworkers reported a sensitivity and specificity of 94% and 97%, respectively [67].
Data from recent literature reported the possible diagnostic role of SUV in the patient with suspected IPMN and in the study of the cystic lesions. In particular an SUV >2.5 is highly suggestive that the tumor is not benign [68].
24
PURPOSE OF THE THESIS
The purpose of the thesis is to clarify the role of [18F]FDG PET/CT in the distinguishing between benign and malignant lesions of the pancreas, and to explore the role of additional information that [18F]FDG PET/CT could provide to routine radiological imaging, both in solid and cystic pancreatic lesion. In fact the strength of functional imaging lies in its ability to detect malignant disease irrespective of lesion morphology. To date, literature is rather limited in comparison with other tumors, and results are not always univocal.
25
MATERIALS AND METHODS
We reviewed 70 [18F]FDG PET/CT studies performed from 01/01/2006 to 31/12/2011 in patients who had a suspicious malignant pancreatic disease based on radiological grounds. PET/CT had been performed in all the cases during the diagnostic work-up in order to differentiate between benign and malignant lesion..
We assessed 70 patients (31 men and 39 women, age range 20-82 years, mean 65 ± 13 years) for a total of 72 pancreatic lesion to be retrospectively reviewed by [18F]FDG PET/CT. Forty-seven were cystic lesion, while 25 were solid (see Table 3). Histological analysis was obtained in all patients and disclosed 39 benign and 33 malignant pancreatic lesions. Patients with a suspicion of neuroendocrine tumor were not included in the analysis. All patients had undergone radiological imaging before the [18F]FDG PET/CT scan and were followed up until February 2012 (mean 33.3 ±19.7 months).
The [18F]FDG PET/CT scans were performed in two different Nuclear Medicine Center: the Regional Center of Nuclear Medicine at the University Hospital of Pisa and the Nuclear Medicine Unit at the National Council of Research (CNR) of Pisa. A Discovery ST/8 (GE Healthcare, Milwaukee, USA) with a 2D acquisition modality and a Discovery VCT (GE Healthcare, Milwaukee, USA) with a 3D acquisition modality were used for all the PET/CT studies. Both tomographs are produced by G.E. Healthcare.
Each patient underwent PET/CT acquisition at least 60 minutes after 18F[FDG] injection and a good hydratation. All the patients had fasted for at least 6 hours and, at the time of 18
F[FDG administration, blood glucose levels were below 180 mg/dl,. Injected activity was 5.5 MBq/ Kg for 2D mode imaging while 3.7 MBq/Kg for 3D mode imaging. Static [18F]FDG PET/CT imaging covered the upper torso from eyebrows to midthighs. Attenuation correction was performed by a low dose non contrast enhanced CT (80 mA, 120 kV; slice thickness of 3.75 mm to approximate the PET; reconstructed slice interval of 3.27 mm to match the PET slice spacing; 4 minutes per bed- PET-step in 2D acquisition, 2,5 minutes per bed- PET-step in 3D acquisition).
The reconstruction of PET data, corrected for deadtime, decay, and photon attenuation,.was performed using the OS-EM algorithm in a 128×128 matrix and a 60 cm FOV (field of view). Transaxial, coronal, and sagittal sections were obtained for imaging analysis, performed according to a colored scale. The images were analyzed visually and semi-quantitatively using the standardized uptake value (maximum SUVbw, g/ml) Any focal tracer accumulation exceeding normal regional tracer uptake was considered a pathologic finding
26 (tumor manifestation). The region of interest was placed on the area of the lesion with the highest [18F]FDG uptake.When no abnormal focal was found, the region of interest was placed on the area of suspected lesion based on referral.
Sensitivity, specificity, positive predictive value (PPV), and negative predictive value, as well as accuracy of [18F]FDG PET/CT and radiological finding were evaluated. By means of receiver operating characteristic (ROC) analysis, sensitivities and specificities for different cut-off points were calculated within the group of patients with pancreatic adenocarcinoma and with a benign tumor or a normal pancreas. P < 0.05 was considered statistically significant.
All statistical analyses were performed with MedCalc Software (Mariakerke, Belgium) on a windows based PC.
No. patients 70
Male 31
Female 39
Age years (mean ± SD) 65 ± 13
No. lesions 72
Tumor size (mm) mean 34 ± 22
range 6-110
Malignant 33
Benign 39
Solid/cystic prevalent component
Solid 24
Cistyc 48
Histology
Adenocarcinoma 23
Metastasis from other tumors 1
Intraductal Papillary Mucinous Carcinoma
(invasive/non invasive) 8
Intraductal Papillary Mucinous Neoplasia with
mild displasya (adenomas) 11
Intraductal Papillary Mucinous Neoplasia with
moderate displasya (borderline) 3
Cystoadenoma Mucinosus 5
Pseudopapillary Solid Tumor 1
Cystoadenoma Serosus 12
Cystis 4
Pancreatic Pseudocystic 2
Chronic Pancreatitis 2
Concomitant Chronic Pancreatitis 23
Benign Lesions 18
Malignant lesions 5
RESULTS
Sensitivity and specificity of and malignant pancreatic lesion considered a cut-off of SUVmax whereas PNV resulted 92%. SUV
in malignant lesions (1.9±0.9 vs 4.7 ±2.8, P < 0,001) (Fig. 2). The partial overlap of the distributions of SUVmax values obtained in benign and malignant lesions that can be observed in figure 2 is referable to the presence of false positive and false negative in both groups.
Fig. 1: Receiver Operating Characteristic
(ROC) analysis comparing
malignant lesion (n=33) with benign lesions (n=39). Cut-off of SUV
90%, specificity 87%.
In particular, 24/24 solid lesion were correctly identified by PET/CT as malignant: 23 were pancreatic adenocarcinoma (SUV
renal clear cell tumor (SUVmax
consisting in Intraductal Papillary Mucinous Carcinomas. They represent 38% of all Intraductal Papillary Mucinous Carcinomas.
Five false positive PET/CT
ducts, with mild dysplasia, 1 patient with acinar cyst and 1 patient with sierosus cystoadenoma of 110 mm).
Conventional radiology (CT and MRI) had a sensitivity and specificity of respectively 90% and 40% (P = 0.01)
ensitivity and specificity of 18F[FDG] PET/CT for the differentiation between benign and malignant pancreatic lesions resulted respectively 90% and 87% (AUC 0,91) if we max ≤ 2.4 (P < 0.0001) for benign lesions. (Fig. 1). PPV was 86%, whereas PNV resulted 92%. SUVmax in benign lesions was found to be significantly lower than in malignant lesions (1.9±0.9 vs 4.7 ±2.8, P < 0,001) (Fig. 2). The partial overlap of the values obtained in benign and malignant lesions that can be observed in the presence of false positive and false negative in both groups.
1: Receiver Operating Characteristic
(ROC) analysis comparing pancreatic
malignant lesion (n=33) with benign lesions
off of SUVmax 2.4, sensitivity
Fig. 2: Distributions of SUV
benign and malignant pancreatic
lesions (box-plot); P < 0.0001)
particular, 24/24 solid lesion were correctly identified by PET/CT as malignant: 23 were pancreatic adenocarcinoma (SUVmax5.05, range 2.6-15.2) and 1 was a
max 3.2). 18F[FDG] PET/CT resulted falsely negative in 3 cases, consisting in Intraductal Papillary Mucinous Carcinomas. They represent 38% of all Intraductal Papillary Mucinous Carcinomas.
PET/CT cases included 3 patients with IPMN of both main and brunch 1 patient with acinar cyst and 1 patient with sierosus cystoadenoma of
Conventional radiology (CT and MRI) had a sensitivity and specificity of respectively
27 differentiation between benign respectively 90% and 87% (AUC 0,91) if we (Fig. 1). PPV was 86%, ns was found to be significantly lower than in malignant lesions (1.9±0.9 vs 4.7 ±2.8, P < 0,001) (Fig. 2). The partial overlap of the values obtained in benign and malignant lesions that can be observed in the presence of false positive and false negative in both groups.
Fig. 2: Distributions of SUVmax in
benign and malignant pancreatic
; P < 0.0001).
particular, 24/24 solid lesion were correctly identified by PET/CT as malignant: 23 15.2) and 1 was a metastasis from a resulted falsely negative in 3 cases, consisting in Intraductal Papillary Mucinous Carcinomas. They represent 38% of all Intraductal
cases included 3 patients with IPMN of both main and brunch 1 patient with acinar cyst and 1 patient with sierosus cystoadenoma of
28 When considering only cystic lesions, sensitivity and specificity were 77.8% and 79.5%, respectively, by using a SUVmax ≤ 2.1 as a cut-off value for benignity (AUC 0.78) (Fig 3) . PPV was 44%, while NPV 94% (P = 0.006). SUVmax in benign lesions was found to be significantly lower than in malignant lesions (1.8±1.0 vs 4.14 ±2.6, P = 0,0003) ) (Fig. 4). The partial overlap of the distributions of SUVmax values obtained in benign and malignant lesions that can be observed in figure 4 is referable to the presence of false positive and false negative in both groups.
Fig. 3: Receiver Operating Characteristic (ROC) analysis comparing pancreatic malignant lesion (n=9) with benign lesions
(n=39). Cut-off of SUVmax 2.1, sensitivity
78%, specificity 79%.
Fig. 4: Distributions of SUVmax in benign
and malignant pancreatic cystic lesions (box-plot; P < 0.0001).
There were only 2/48 false negative results on PET/CT (both patients affected by IPMC) but 9/48 false positive cases (mean SUVmax 3, 4 patients with a SUVmax ≤ 2.4).
Conventional radiology (CT and MRI) had a sensitivity and specificity of respectively 87% and 40% (P = 0.01).
SUVmax values were analyzed as prognostic factor, and Kaplan-Meier curves were created (Fig. 5) Survival in patients with SUVmax > 2.5 was significantly lower than in patients with SUVmax ≤ 2.4 (survival at 30 months was 50%, P < 0.0001). Of all patients followed up until February 2012, 13 (18.6%) patient died during follow-up (mean survival of 11.7 months, range 1-31 months): 12 were affected by adenocarcinoma and 1 patients had IPMC. 57 (81.4%) patients were instead still alive on February 2012.
29
Fig. 5: Kaplan–Meier plots (with CI curves) depicting the overall survival for two
30 DISCUSSION
In our study [18F]FDG PET/CT was significantly more accurate in differentiating benign from malignant lesions than conventional imaging (Fig. 6).
Fig. 6: Comparison of the ragiological technique (CT and RMI) and nuclear medicine technique (PET/CT).
PET/CT is an interesting imaging modality that takes advantage of the selective uptake of labeled compounds by metabolically active malignant cells, and could be used for the characterization of tumor biochemistry. Depending on the tracer, it is possible to selectively assess several metabolic pathways or the expression of different receptors and enzymes. The glucose analogue, FDG, is a widely used radiotracer in clinical oncology. The mechanism of this tracer is based on the increased metabolism of malignant tumors associated with increased glycolysis and increased expression of glucose transporter proteins [69].
For initial diagnosis of pancreatic cancer, several studies have already shown FDG-PET to be more accurate than conventional imaging techniques, although conflicting results have also been published. In fact, most of published paper regarding the role of PET in pancreatic cancer did not use hybrid integrated PET/CT devices, especially those published less recently [70]. Differently from other tumors, both solid and cystic pancreatic tumors probably need a more accurate anatomic correlation than other tumors because the global target to background ratio of
90
40
90
87
0
20
40
60
80
100
Sensitivity
Specificity
CT/RMI
PET/CT
31 radiopharmaceutical uptake is quite low. In fact, pancreatic adenocarcinoma seems to have a consumption of carbohydrates lower than the other tumors, probably dependent on the gene expression pattern of these tumors. The grading of GLUT-1 expression plays an essential role in the process of uptake within the pancreatic cancer cells, and it is therefore closely related to FDG accumulation [71]. For this reason pancreatic adenocarcinoma can have variable behavior on FDG PET imaging , showing generally lower degrees of uptake [72]. In our study mean SUVmax in positive [18F]FDG PET/CT was 4.7 ± 2.8, a value that is generally lower than others tumors (in breast cancer mean SUVmax value has been reported 9.7, in ovarian cancer 7.6, in lung cancer 12.0 etc) [73-75]. For this reason diagnostic accuracy of [18F]FDG PET could be improved by the use of hybrid PET/CT scanners, that allow to better localize areas of focal uptake, especially in case of small lesions (Fig. 7). In the first period of PET technique development, researchers could have lost their interest in the application of this technique in pancreatic cancer due to the difficulty to disclose and to correctly identify lesions with low FDG uptake. In agreement with our findings, Wakabayashi et al. [76] reported a sensitivity of 91% and a specificity of 78% in evaluating malignancy in 30 patients with biliary stricture by FDG-PET and Kauhanen et al. [59] reported a sensitivity of 85% and a high PPV of 92% in differential diagnosis of biliary stricture malignancies.
Fig. 7: transaxial CT, transaxial PET, transaxial fused images and MIP (Maximum
32 Our study included 23 patients with histological finding of chronic pancreatitis (18 patients affected by benign disease, and in 5 patients with malignant disease). In all 18 patients with benign disease, no uptake was observed in [18F]FDG PET/CT images. Thus, chronic pancreatitis does not seem to influence [18F]FDG PET/CT results and we did not report any case of false positive result related to the presence of chronic pancreatitis. To our knowledge, only one study has been conducted concerning the diagnostic value of FDG-PET/CT in disclosing pancreatic carcinoma in patient with chronic pancreatitis [77]. In agreement with our results, this study reported that FDG-PET was negative in 67 out 77 patients with long-standing chronic pancreatitis without carcinoma.
Our false positive results did not correlate with the presence of chronic pancreatitis, while the most important factor reducing diagnostic accuracy is the presence of cystic lesions. Pancreatic cystic lesions are being detected more often as sensitive abdominal imaging tests are being used for multiple indications. Since risks of a pancreatic surgery are not negligible, resection of normal pancreas or with benign lesions should be avoided. On the other hand, the majority of these lesions are either premalignant or malignant. Thus, the decision of whether to “wait and see” or to perform surgery on cystic pancreatic lesions is difficult. Our study shows that [18F]FDG PET/CT has a NPV of 94%. Therefore, [18F]FDG PET/CT could help for the surgical decision suggesting that a follow-up strategy could be propose when SUVmax is lower than 2.1, especially in those lesions that present benign or uncertain radiological findings (Fig. 6). Therefore metabolic assessment should be proposed in all cystic lesions in order to avoid a useless pancreatectomy that often (15% of cases) may be total because of the multifocality of the disease. The dilemma whether operate or not should be supported not only by anatomical criteria but also by functional evaluation. However our results agree with some recent papers where encouraging results have been recently reported in the diagnosis of intraductal papillary mucinous neoplasms of the pancreas with a better sensitivity of PET than conventional imaging [68].
