Eur Radiol (2017) 27:507–517 DOI 10.1007/s00330-016-4394-3
GASTROINTESTINAL
Outcome and CT differentiation of gallbladder neuroendocrine tumours from adenocarcinomas Tae-Hyung Kim 1 & Se Hyung Kim 1,2,3 & Kyoung Boon Lee 4 & Joon Koo Han 1,3,5
Received: 2 November 2015 / Revised: 15 February 2016 / Accepted: 2 May 2016 / Published online: 25 May 2016 # European Society of Radiology 2016
Abstract Objectives To retrospectively investigate clinical outcome and differential CT features of gallbladder (GB) neuroendocrine tumours (NETs) from adenocarcinomas (ADCs). Materials and methods Nineteen patients with poorlydifferentiated (PD) NETs and 19 patients with PD ADCs were enrolled. Clinical outcome was compared by the KaplanMeier method. We assessed qualitative and quantitative CT features to identify significant differential CT features of PD NETs from ADCs using univariate and multivariate analyses. Receiver operating characteristic (ROC) analysis was used for quantitative CT features. Results PD NETs showed poorer prognosis with significantly shorter median survival days than ADCs (363 vs. 590 days, P = 0.03). On univariate analysis, NETs more frequently manifested as GB-replacing type and showed well-defined margins accompanied with intact overlying mucosa. On multivariate analysis, well-defined margin was the sole significant CT differentiator (odds ratio = 27.817, P = 0.045). Maximum size
* Se Hyung Kim
[email protected]
1
Department of Radiology, Seoul National University Hospital, Seoul, Korea
2
Department of Radiology, Seoul National University Hospital and Seoul National University College of Medicine, 101 Daehangno, Jongno-gu, Seoul 03080, Korea
3
Department of Radiology, Seoul National University College of Medicine, Seoul, Korea
4
Department of Pathology, Seoul National University Hospital, Seoul, Korea
5
Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea
of hepatic and lymph node (LN) metastases was significantly larger in NETs (11.0 cm and 4.62 cm) than ADCs (2.40 cm and 2.41 cm). Areas under ROC curves for tumour-to-mucosa ratio, maximum size of hepatic and LN metastasis were 0.772, 0.932 and 0.919, respectively (P < 0.05). Conclusion GB PD NETs show poorer prognosis than ADCs. Well-defined margin, larger hepatic and LN metastases are useful CT differentiators of GB NETs from ADCs. Key Points • A well-defined margin is a useful CT differentiator of GB NETs from ADCs. • Hepatic and LN metastases are significantly larger in NETs than in ADCs. • Primary tumour and hepatic metastasis of NETs are more hyperattenuated than ADCs. Keywords Gallbladder . Neuroendocrine tumours . Adenocarcinomas . CT . Differentiation
Abbreviations ADCs Adenocarcinomas CT Computed tomography GB Gallbladder LNs Lymph nodes NETs Neuroendocrine tumours PD Poorly-differentiated WD Well-differentiated
Introduction Neuroendocrine tumours (NETs) of the gallbladder (GB) represent only 0.2 % of all gastrointestinal NETs and 2 % of GB cancers [1–4]. Little is known about the biological behaviour of
508
GB NETs; however, their prognosis is dismal with a 5-year survival rate of 0–21.3 % [5]. The incidence and awareness of this uncommon tumour of GB NETs has increased due to advances in imaging technology and increased frequency of imaging work-up; consequently, this disease is sometimes discovered in its early stage. No consensus has been reached in terms of standard treatment; however, complete en bloc surgical resection (if resectable) is widely accepted as the only curative therapeutic method [6]. However, curative surgery might not be indicated in most cases because most GB NETs present along with metastasis at the time of diagnosis [7]. If unresectable, chemotherapy may be of predictable value in poorly-differentiated (PD) NETs, especially small-cell carcinoma, like their counterparts in the lung [8]. Biotherapy such as somatostatin analogues and interferons along with peptide receptor radionucleotide therapy has been proposed for new medical treatments of GB NETs [9]. Therefore it is clinically important to differentiate GB NETs from adenocarcinomas (ADCs) usually treated with conventional chemotherapeutic agents. Computed tomography (CT) has been the most widely used evaluation tool of primary GB tumour extent and distant metastasis due to well-standardized protocol and easy accessibility [10]. However, CT features of GB NETs as well as differential imaging features of GB NETs from ADCs have not yet been adequately investigated despite several case reports regarding the CT features of GB NETs [7, 11, 12]. Therefore, our study retrospectively investigates the differential CT features of GB NETs from ADCs.
Materials and methods The retrospective study was approved by the ethics committee of our institute and did not require informed consent.
Eur Radiol (2017) 27:507–517
were proven to have large-cell NETs while six were proven to have small-cell NETs. The remaining eight patients were not specifically classified. To serve as the control group, we obtained a list of 89 patients who were diagnosed as GB ADCs from our pathology database between March 2000 and June 2014. Among them, we only included patients who satisfied the following inclusion criteria: (a) patients who had preoperative CT images and (b) patients whose primary lesions were located in the GB. After excluding 50 patients whose CT images were not available (n = 35) and in whom the origin of the primary lesions were uncertain (n = 15), we obtained a list of 39 patients with GB ADCs. Our statistician matched the patients having GB ADCs with patients proven to have NETs by 1:1 in terms of age, sex, tumour differentiation and tumour stage. Finally, 19 patients with GB PD ADCs (M:F = 9:10, mean age, 63.3 years; age range, 43–89 years) pathologically proven through surgery (n = 14) or biopsy (n = 5) were included.
Clinical and histological features Clinical features of all patients were analysed by one author (KTH, with 2 years of experience in abdominal imaging) using electronic medical records (EMR) of our hospital. He reviewed clinical features, pathological TNM staging, type of surgery and type of chemotherapy. In addition, clinical outcomes including patients’ current status (dead, alive or followup loss) and survival days after the initial diagnosis of GB malignancy were collected. Overall follow-up survival data were completed by reviewing EMR of our hospital as well as by contacting the Resident Service Division of the Ministry of Public Administration and Security. The endpoints of this study were either the patient’s death or 3 September 2015. Table 1 indicates the clinical and histological features of 38 patients with NETs or ADCs.
Patients Between September 2002 and August 2014, we found 23 patients with GB NETs proven either through surgery or biopsy in our pathology database using the search terms BGB and neuroendocrine,^ Bgallbladder and neuroendocrine,^ BGB and carcinoid,^ Bgallbladder and carcinoid^. We included patients who satisfied the following inclusion criteria: (a) patients who had preoperative CT images and (b) patients whose primary lesions were located in the GB. We excluded two patients who had no CT images and two patients in whom the origin of the primary lesions were uncertain. Finally, 19 patients with pathologically proven GB NETs (M:F = 6:13; mean age, 62.2 years; age range, 33–83 years) were enrolled in our study. All patients were proven to have grade 3 PD NETs. Five of the 19 patients
CT techniques A variety of CT systems were utilized in our retrospective study. All CT examinations except one were performed with MDCT scanners. CT scanning parameters used in this study were: detector collimations of 0.6–0.625, 0.75, 1.25 and 2.5 mm for the 64- (n = 18), 16- (n = 13), 8- (n = 3) and 4channel (n = 4) MDCT scanners, respectively; gantry rotation times, 0.5–0.75 s; tube voltage, 120 kVp; tube current, 150– 200 mAs; section thickness 2.5–5.0 mm; reconstruction interval, 2.0–5.0 mm; and field of views, 300–390 mm. For the contrast-enhanced study, 1.5 ml/kg of a 370 mg∙I/ ml iodinated contrast agent (Ultravist 370, Bayer Schering Pharma, Berlin, Germany) was injected at a rate of 3–5 ml/s.
Eur Radiol (2017) 27:507–517
509
Table 1 Clinical and histological features of 38 patients with gallbladder (GB) poorly differentiated (PD) neuroendocrine tumours (NETs) or adenocarcinomas (ADCs) NETs (n = 19) M:F
6:13
ADCs (n = 19) 9:10
Mean age
62.2 years
Biopsy vs. operation Biopsy
13 (68.4 %)
5 (26.4 %)
6 (31.6 %)
14 (73.6 %)
Operation Histology Large-cell NETs
63.3 years
Small-cell NETs PD tubular ADCs
6 (31.5 %) 0 (0 %)
0 (0 %) 19 (100 %)
Unspecified
8 (42.2 %)
0 (0 %)
Clinical T stage (cT)
0.929
T1 T2 T3
0 (0 %) 4 (21.0 %) 14 (73.6 %)
0 (0 %) 5 (26.3 %) 13 (68.3 %)
T4
1 (5.4 %)
1 (5.4 %)
Pathological T stage (pT) T1 T2 T3 T4 Pathological N stage (pN) N0 N1
0.796
<0.0001 0 (0 %)
N2 Clinical M stage (cM) M0 M1
0.508 0.009
5 (26.3 %)
Clinical N stage (cN) N0 N1
P value
0.831 4 (21.0 %) 1 (5.4 %)
4 (21.0 %) 2 (10.7 %)
14 (73.6 %)
13 (68.3 %)
10 (52.6 %) 9 (47.4 %)
10 (52.6 %) 9 (47.4 %)
0 (0 %) 2 (33.3 %)
0 (0 %) 4 (28.6 %)
4 (66.7 %) 0 (0 %)
10 (71.4 %) 0 (0 %)
2 (33.3 %) 1 (16.7 %)
4 (28.5 %) 2 (14.4 %)
3 (50 %)
8 (57.1 %)
6 (100 %) 0 (0 %)
10 (71.4 %) 4 (28.6 %)
6 (100 %) 0 (0 %) 0 (0 %)
8 (57.1 %) 2 (10.7 %) 4 (28.6 %)
1.000
1.000
0.958
N2 Pathological M stage (pM) M0 M1 Type of surgery Extended cholecystectomy PPPD Palliative operation Chemotherapy No Yes
3 (15.7 %) 16 (84.3 %)
7 (36.8 %) 12 (63.2 %)
Type of chemotherapy Conventional chemotherapy Biotherapy*
11 (68.7 %) 5 (31.3 %)
12 (100 %) 0 (0 %)
0.267
0.159
0.141
0.053
*Biotherapy includes treatment with somatostatin analogue, interferon or peptide receptor radionucleotide therapy P values in italics indicate statistical significance c clinical staging, p pathological staging, PPPD pylorus-preserving pancreaticoduodenectomy
510
CT images were obtained at a single phase of the portal venous phase in five examinations, dual phases that included the late arterial and portal phase in 21, and triple phases that included the precontrast, late arterial and portal venous phase in 12. For the late arterial phase, a delay time of 22–29 s was used after the attenuation of the descending aorta reached 100 Hounsfield units (Hus) using the bolus tracking technique (29 s for 64-, 27 for 16-, 25 for 8- and 22 for 4-channel). Portal venous phase scanning was performed 70–75 s after the administration of contrast agent. CT Image Analysis All CT features were assessed by two radiologists (KTH and KSH, with 2 and 18 years of experience, respectively) on a picture archiving and communication system (PACS) workstation monitor in a consensus manner. Both radiologists were aware that the study population had GB NETs or GB ADCs, but were blinded to the histological subtype (NETs vs. ADCs or large-cell vs. small-cell type). The following CT findings of the primary GB lesions were evaluated: location of the tumour, maximum size, tumour morphology, homogeneity, degree of enhancement on each phase, the presence of intact overlying mucosa, the preservation of GB lumen, and the presence of direct liver invasion. Location was divided into cystic duct, neck, body, fundus or diffuse. Tumours located at more than two sites were designed as diffuse involvement. Tumour morphology was divided into three types: polypoid, wall thickening or GB replacing. The degree of enhancement was evaluated as high-, iso- or lowattenuation on each phase, referencing the enhancement of the normal GB mucosa. Overlying mucosa was considered to be intact when innermost mucosal layer was entirely and partially free of disease. Complete disruption of the enhancing thin mucosal layer and irregularly thickened mucosal layer that covered the entire tumour were considered negative for an intact overlying mucosa [13]. Preservation of GB lumen was considered positive when the GB lumen was filled with bile fluid entirely or partially. Direct liver invasion was considered when the border between the tumour and adjacent hepatic parenchyma was obliterated. For lymph node (LN) metastasis, maximum size, degree of enhancement, the presence of necrosis, the presence of infrarenal or extraabdominal LN involvement, and the presence of sandwich sign were assessed. The degree of enhancement was evaluated as high-, iso- or low-attenuation on each phase, referencing the enhancement of the normal paravertebral muscle. The presence of necrosis was considered present when the total volume of the necrotic portion was > 50 % of the entire volume of the metastatic LNs. The presence of sandwich sign was considered positive when confluent lymphadenopathy was present on both sides of the mesenteric vessels without obliteration of vessels.
Eur Radiol (2017) 27:507–517
For hepatic metastases, maximum size, degree of enhancement and homogeneity were assessed. Hepatic metastasis were confirmed by histology (n = 15) or based on a combination of CT findings and follow-up studies (n = 4). The presence of extrahepatic metastasis such as peritoneal seeding (n = 3), bone metastasis (n = 2) or psoas muscle metastasis (n = 1) was also recorded. All extrahepatic metastases were confirmed based on a combination of CT findings and follow-up studies determined by a consensus of the two radiologists. For all primary and metastatic lesions, quantitative analysis was performed by measuring the mean HUs at the most strongly enhanced portion of the lesions as well as at the adjacent normal structures through drawing regions of interest (ROIs) on each phase using a PACS workstation. The lesion-to-normal ratio was then calculated using the following equation: (HU at the most strongly enhancing portion of the lesions) / (HU at the normal structures). In addition, the size ratio between GB lesion and hepatic or LN metastasis was also calculated using the following equation: (maximum size of hepatic or LN metastasis) / (maximum size of GB lesion).
Statistical analysis Survival curves were calculated by the Kaplan-Meier method and the difference was analysed with the log-rank test. The prevalence of each CT finding between GB NETs and GB ADCs were compared using Fisher’s exact test or Student’s
Fig. 1 Kaplan-Meier survival curves comparing survival rates between gallbladder (GB) poorly-differentiated (PD) neuroendocrine tumours (NETs) (Bold line) versus GB adenocarcinomas (ADCs) (Dotted line) (P value = 0.03)
Eur Radiol (2017) 27:507–517
511
t-test. Thereafter, binary logistic regression analyses using the enter method were performed to identify the most significant differential CT features. Receiver operating characteristic (ROC) curve analysis was performed to ascertain the optimal cut-off value of the lesion-to-normal ratio as well as the size of hepatic or LN metastasis to differentiate GB NETs from ADCs. All statistical analyses were performed using SPSS 21.0 for Windows (SPSS Inc., Chicago, IL, USA). A P value < 0.05 was considered to indicate statistical significance. Table 2 Results of univariate analysis for CT features between gallbladder (GB) poorly differentiated (PD) neuroendocrine tumours (NETs) and adenocarcinomas (ADCs)
Results Clinical and histological findings Table 1 summarizes all clinical and histological findings. For tissue confirmation in PD NETs, 13 patients underwent biopsy (68.4 %, 13/19) and six patients underwent operation (31.6 %, 6/19) while in patients with ADCs, five patients (26.5 %, 5/19) underwent biopsy and 14 patients (73.6 %, 14/19) underwent
NETs (n = 19)
ADCs (n = 19)
Tumour size (cm)
3.57 ± 1.30
3.73 ± 1.57
Location Neck
3 (15.7 %)
2 (10.7 %)
Body Fundus
4 (21.3 %) 6 (31.5 %)
4 (21.3 %) 8 (41.6 %)
Diffuse
6 (31.5 %)
5 (26.4 %)
0.729 0.902
Morphology of tumour Polypoid Wall thickening GB replacing Tumour margin Well-defined
P value
0.02 0 (0 %) 4 (21.1 %) 15 (78.9 %)
0 (0 %) 12 (63.1 %) 7 (26.9 %)
18 (94.7 %)
2 (10.6 %)
1 (5.3 %)
17 (89.4 %)
Present Absent Preservation of GB lumen
14 (73.6 %) 5 (26.4 %)
2 (10.6 %) 17 (89.4 %)
Preserved Not preserved Direct liver invasion Present Absent Liver metastasis
18 (94.7 %) 1 (5.3 %)
18 (94.7 %) 1 (5.3 %)
15 (78.9 %) 4 (21.1 %)
14 (73.6 %) 5 (26.4 %)
11 (57.9 %)
8 (42.1 %)
0.330
8 (42.1 %) 11.00 ± 5.46
11 (57.9 %) 2.40 ± 1.00
<0.0001
14 (73.7 %) 5 (26.3 %) 4.62 ± 1.53 7 (50 %) 12 (85.7 %) 9 (64.3 %)
15 (78.9 %) 4 (21.1 %) 2.41 ± 1.39 7 (46.7 %) 7 (46.7 %) 9 (60.0 %)
2 (66.7 %) 1 (33.3 %) 0 (0 %)
1 (33.3 %) 1 (33.3 %) 1 (33.3 %)
Ill-defined Intact overlying mucosa
Present Absent Mean diameter of the largest metastasis (cm) LN metastasis Present Absent Mean diameter of the largest LN (cm) Necrosis Infrarenal involvement Sandwich sign Other metastasis Peritoneal seeding Bone metastasis Psoas muscle metastasis P values in italics indicate statistical significance LN lymph node
<0.0001
<0.0001
1.000
1.000
1.000 <0.0001 0.858 0.05 0.812 1.000
512 Table 3 Results of quantitative analysis between gallbladder (GB) poorly differentiated (PD) neuroendocrine tumours (NETs) and adenocarcinomas (ADCs)
Eur Radiol (2017) 27:507–517
NETs (n = 19) Primary GB tumour Late arterial phase
n = 19
ADCs (n = 19)
P value
n = 19
Mean HU of enhancing portion
94.31 ± 23.05
79.29 ± 25.26
0.085
Mean HU of normal mucosa Mean tumour-to-mucosa HU ratio
53.63 ± 14.12 1.81 ± 0.42
58.35 ± 11.91 1.38 ± 0.44
0.305 0.007
91.00 ± 16.73 87.47 ± 26.31
101.32 ± 21.34 77.89 ± 11.66
0.106 0.084
1.09 ± 0.23 n = 11
1.32 ± 0.29 n=8
0.308
105.20 ± 11.23
91.33 ± 29.93
0.319
66.80 ± 15.28 1.65 ± 0.41
73.50 ± 18.84 1.14 ± 0. 46
0.072 0.036
Portal phase Mean HU of enhancing portion Mean HU of normal mucosa Mean tumour-to-mucosa HU ratio Liver metastasis Late arterial phase Mean HU of metastasis Mean HU of normal liver parenchyma Mean metastasis-to-liver HU ratio Portal phase Mean HU of metastasis
104.00 ± 55.49
78.75 ± 29.29
0.259
114.18 ± 21.63 0.90 ± 0.40 n = 14
110.25 ± 20.01 0.76 ± 0.43 n = 15
0.692 0.474
Mean HU of metastasis Mean HU of normal paravertebral muscle Mean metastasis-to-muscle HU ratio
58.00 ± 15.22 53.25 ± 7.29 1.09 ± 0.27
58.00 ± 15.65 58.43 ± 7.75 0.99 ± 0.20
1.000 0.094 0.263
Portal phase Mean HU of metastasis Mean HU of normal paravertebral muscle
61.36 ± 12.40 59.50 ± 9.32
68.60 ± 13.86 63.87 ± 6.75
0.151 0.158
1.03 ± 0.16
1.08 ± 0.20
0.523
Mean HU of normal liver parenchyma Mean metastasis-to-liver HU ratio LN metastasis Late arterial phase
Mean metastasis-to-muscle HU ratio P values in italics indicate statistical significance LN lymph node, HU Hounsfield unit
operation (P = 0.009). Age and distributions of clinical and pathological TNM staging were not significantly different between the two groups (P > 0.05). Conventional chemotherapy was performed in 16 patients (84.3 %, 16/19) with PD NETs and 12 patients (63.2 %, 12/19) with ADCs. Biotherapy including transarterial chemoembolization (n = 1), somatostatin analogs (n = 2) and interferons (n = 2) was performed in five PD NETs. The median survival day (363 days; range, 297–428 days) of GB NETs was significantly shorter than that (590 days; range, 160–1019 days) of ADCs (P = 0.030). Figure 1 presents the Kaplan-Meier survival curve. CT findings of primary tumour in the gallbladder Table 2 presents the results of univariate analyses of CT findings between PD NETs and ADCs. NETs more frequently manifested as GB-replacing type (78.9 %, 15/19) compared with ADCs, which more often manifested as a wall thickening
type (63.1 %, 12/19) (P = 0.02). NETs (94.7 %, 18/19) more frequently showed well-defined margins than ADCs (10.6 %, 2/19) (P < 0.0001). In addition, NETs (73.6 %, 14/19) were more often accompanied with intact overlying mucosa than ADCs (10.6 %, 2/19) (P < 0.0001). Other CT findings were not significantly different between the two groups. Table 3 lists the results of the quantitative analysis. The mean tumour-to-mucosa ratio in NETs (1.81 ± 0.42) was Fig. 2 Receiver operating characteristic (ROC) curves of quantitative values to differentiate gallbladder neuroendocrine tumours from adenocarcinomas. (a) For the maximum size of liver metastasis, the area under the ROC curve (Az) was 0.932 (P = 0.002). When a cut-off value was set at 6.15 cm, 81.8 % sensitivity and 100 % specificity were achieved. (b) For the maximum size of lymph node metastasis, Az value was 0.919 (P < 0.0001). When a cut-off value was set at 2.95 cm, 92.9 % sensitivity and 86.7 % specificity were achieved. (c) For the tumour-tomucosa HU ratio on the late arterial phase, the Az value was 0.772 (P = 0.008). When a cut-off value was set at 1.42, 87.5 % sensitivity and 64.7 % specificity were achieved. HU Hounsfield Unit
Eur Radiol (2017) 27:507–517
513
514
Eur Radiol (2017) 27:507–517
significantly greater than (1.38 ± 0.44) in ADCs on the late arterial phase (P = 0.007). For the tumour-to-mucosa ratio, ROC analysis revealed that area under the curve (Az) was 0.772 with 87.5 % sensitivity and 64.7 % specificity at a cut-off value of 1.42 for the differentiation of GB NETs from ADCs (P = 0.008) (Fig. 2). Hepatic and lymph node metastases from PD NETs versus ADCs Both hepatic and LN metastases from NETs appeared significantly larger than those from ADCs (liver, 11.00 ± 5.46 vs. 2.40 ± 1.00 cm; LN, 4.62 ± 1.53 cm vs. 2.41 ± 1.39 cm, P < 0.0001, respectively). For hepatic metastasis, ROC analysis revealed that Az was 0.932 with 81.8 % sensitivity and 100 % specificity at a cut-off value of 6.15 cm for the differentiation of GB NETs from ADCs (P = 0.002) (Fig. 2). For LN metastasis, Az value was 0.919 with 92.9 % sensitivity and 86.7 % specificity at a cut-off value of 2.95 cm (P < 0.0001) (Fig. 2). Similar to the results of the primary tumour, the metastasis-to-liver ratio was significant greater in PD NETs (1.65 ± 0.41) than that (1.14 ± 0. 46) in ADCs (P = 0.036). However, ROC analysis revealed no statistical significance for the metastasis-to-liver ratio (P = 0.051). Binary logistic regression analysis revealed a well-defined margin as the only significantly independent differential CT feature of PD NETs from ADCs with an odds ratio of 27.817 (95 % confidence interval, 1.075–719.864; P = 0.045). Representative examples are presented in Figs. 3, 4, 5 and 6.
Discussion NETs of the GB account for only 0.2 % of all gastrointestinal NETs and 2 % of GB cancers [1–3]. Little is known about the biological behaviour of GB NETs; however, their management and prognosis have been reported to be different from ADCs. Therefore, the exact differentiation between NETs and ADCs can be important in clinical practice due to the different treatment strategies of these tumours [14]. The diagnosis of GB NETs is usually made during pathological analysis; however, the exact pathological differential diagnosis of PD NETs from PD ADCs through biopsy alone is sometimes insufficient even for expert pathologists due to the PD nature of the tumours and the insufficient biopsy specimens. Therefore, PD NETs are usually confirmed after extensive futile operation. CT could have a huge clinical impact if it provided differential imaging features between the two disease entities; however, it is currently impossible to differentiate preoperatively between GB ADCs and NETs with imaging techniques alone [5, 7, 11, 12]. To the best of our knowledge, this is the first original article describing the CT findings of GB NETs and suggesting some promising differential CT findings of GB NETs from GB ADCs.
Fig. 3 A 65-year-old woman with poorly-differentiated neuroendocrine tumour. (a, b) Late arterial (a) and portal (b) phase CT images show that a well-defined low attenuating mass (arrows) at the fundus of the gallbladder. Intact overlying mucosa is seen as a thin enhancing line at the inner surface of the lesion. Note multiple enlarged lymph nodes (arrowheads) at the portocaval space and peripancreatic area
In our study, a well-defined margin was the most important CT discriminator of PD NETs from the ADCs. We found that 18 of 19 patients with GB NETs (94.7 %) showed welldefined margins compared to only two out of 19 GB ADCs (10.6 %). Normal GB epithelium does not contain neuroendocrine cells; therefore, the gross or microscopic appearance is often a cauliflower-shaped solid yellow lesion that arises at subepithelial lamina propria, with subsequent infiltration through the muscle layer and extension into the perimuscular connective tissue [15]. We assume that the microscopic location of tumour cells starting from subepithelial lamina propria may be the histological background of the well-defined margin on CT. Similar results were reported in gastric NETs which share the same histological origin of the tumour. Kim et al. [13] reported that a well-defined margin was more often observed in gastric NETs (55.6 %, 10/18) than in gastric ADCs (0 %, 0/19) (P < 0.0001). In addition to the tumour margin, the morphology of GB mass was another CT differential feature. GB NETs are more often manifested as GB-replacing type while ADCs as wall thickening type. We believe that such different morphology is also responsible for different tumour margins because the mass replacing type tends to show a well-defined margin while the wall thickening type tended to show poorly-defined margin due to its infiltrative nature. In our study, the presence of intact
Eur Radiol (2017) 27:507–517
515
Fig. 5 A 62-year-old man with poorly-differentiated adenocarcinoma. (a, b) On late arterial (a) and portal (b) phase CT images, ill-defined low-attenuating wall thickening (arrows) is visualized at the body of the gallbladder. Overlying mucosa is completely disrupted. Small conglomerated lymph nodes (arrowheads) are seen at both para-aortic areas
Fig. 4 An 83-year-old man with poorly-differentiated neuroendocrine tumour. (a, b) Late arterial (a) and portal (b) phase CT images show a well-defined, hypervascular mass (arrows) at the body of the gallbladder. Intact overlying mucosa is seen as a thin enhancing line at the inner surface of the lesion. Large hypervascular metastatic masses (arrowheads) are seen at the right lobe of the liver. Note a large necrotic metastatic lymph node (*) at portocaval space. (c) Coronal CT image obtained at portal phase demonstrates a GB-replacing mass (arrow) with extensive liver metastasis (arrowheads)
overlying mucosa was a significant CT differential finding of GB NETs. Intact overlying mucosa may be more frequently observed because NETs originate from subepithelial lamina propria, unlike ADCs, which arise from the innermost epithelial layer. Therefore, we cautiously insist that GB NETs can be first considered if a GB-replacing mass shows a well-defined margin and has intact overlying mucosa. GB PD NETs resemble their more common counterparts arising in the bronchopulmonary system. They exhibit similar histological features and aggressive behaviour with a high propensity for extensive growth, early LN and distant metastases, and have an exceedingly poor prognosis [7, 8]. Our results are
well in line with previous literatures [7, 8]. The mean diameter of the largest hepatic metastases and LN metastases were significantly larger in NETs than in ADCs, suggesting bulky hepatic metastases and LN metastases. This finding coincides with previous observations that GB NETs show a more bulky burden of hepatic and LN metastases than ADCs in a number of case reports [5, 7, 12]. It is possible that GB ADCs may present with huge hepatic metastases and extensive LN metastases at its late stage; however, NETs can be considered rather than ADCs when huge hepatic metastases and extensive lymphadenopathy with GB tumour are shown at initial presentation. In addition to the larger size, hepatic metastases from GB NETs were significantly more hyper-attenuated in the late arterial phase than those from ADCs. GB NETs also showed higher late arterial attenuation than ADCs. The mean tumourto-mucosa HU ratio and the mean hepatic metastasis-toparenchyma HU ratio in the late arterial phase were significantly higher in GB NETs than those of GB ADCs. Our results are in agreement with previous studies that described hypervascularity of NETs and hepatic metastases of NETs [13, 16, 17]. Our results on CT differentiation of hepatic metastases between GB NETs and GB ADCs may be especially clinically significant due to the comparable positive survival gain possible from various treatment options such as somatostatin analogues, interferons and transarterial chemoembolization to
516
Fig. 6 A 77-year-old woman with poorly-differentiated adenocarcinoma. (a, b) On oblique coronal images obtained at late arterial (a) and portal (b) phases, ill-defined wall thickening (arrow) is seen at the fundus of the gallbladder. Overlying mucosa is completely disrupted. Small hypovascular metastatic nodules (arrowheads) are also demonstrated in the liver
the metastatic NETs [18, 19]. As our study results demonstrate, contrary to ADCs, NETs is one of the hypervascular tumours in the GB. Therefore, late arterial phase images should be obtained for the differentiation of NETs from ADCs. For the evaluation of lymph node and distant metastasis including liver metastasis, portal venous phase is also mandatory. Therefore, we would like to insist that at least threephase CT including precontrast, late arterial (pancreatic) and portal phases should be used for the evaluation of GB diseases. Regarding the prognosis, two articles which directly compare the survival rates between the two tumours have been recently published and their results are conflicting [20, 21]. Chen et al. [20] compared ten GB NETs with 316 ADCs and reported that the median survival time of NETs was significantly shorter than ADCs (3 vs. 6 months, P = 0.038). Yun et al. [21] compared four GB small-cell NETs with 42 ADCs and reported a better overall 5-year survival rate of NETs without statistical significance (100 % vs. 74.8 %, P = 0.896). In our study, the median survival time of NETs was significantly shorter than ADCs. We believe that a larger number of enrolled NET patients and strict matching between
Eur Radiol (2017) 27:507–517
NETs and ADCs in terms of age, sex and tumour stage produced superior evidence concerning poorer prognosis of NETs than ADCs. Our study had several limitations. First, this study included a small number of patients. However, the two groups from our study were statistically well balanced in terms of patient age, sex ratio and tumour stage. Second, we included patients over long periods, which created differences in CT scanners, CT parameters, dose of contrast agent and acquired phases in each patient. However, as all CT examinations were performed with MDCT scanners and ≤ 5-mm reconstruction interval, we believe that the quality of the CT images in our study were sufficient to evaluate CT findings such as the margin of the tumour. Third, although most (33/38, 86.8 %) CT examinations was performed with dynamic protocol CT including late arterial and portal phases, not every CT was multiphasic. Therefore, a further prospective study with an optimized CT protocol is strongly required to prove our study results. Fourth, 42.2 % (8/19) of NETs were pathologically unspecified in our study. Future studies with specified pathology of NETs are warranted to provide additional information regarding the specified pathology. In conclusion, a well-defined margin of the GB mass can be a useful CT finding that can differentiate GB NETs from ADCs. In addition, GB NETs have significantly larger hepatic and LN metastases than ADCs with a stronger enhancement of GB mass as well as liver metastases on the late arterial phase.
Acknowledgments The scientific guarantor of this publication is Joon Koo Han. The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. This study received funding from the Seoul National Hospital Research Funding (Fund No. 04-2015-620). No complex statistical methods were necessary for this paper. Institutional Review Board approval was obtained. Written informed consent was waived by the Institutional Review Board. Methodology: Retrospective, diagnostic or prognostic study, performed at one institution.
References 1. 2.
3.
4.
Modlin IM, Lye KD, Kidd M (2003) A 5-decade analysis of 13,715 carcinoid tumors. Cancer 97:934–959 Modlin IM, Shapiro MD, Kidd M (2005) An analysis of rare carcinoid tumors: clarifying these clinical conundrums. World J Surg 29:92–101 Yao JC, Hassan M, Phan A et al (2008) One hundred years after "carcinoid": epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol 26: 3063–3072 Chen H, Shen YY, Ni XZ (2014) Two cases of neuroendocrine carcinoma of the gallbladder. World J Gastroenterol 20:11916– 11920
Eur Radiol (2017) 27:507–517 5.
6.
7.
8. 9.
10.
11.
12.
13.
Eltawil KM, Gustafsson BI, Kidd M, Modlin IM (2010) Neuroendocrine tumors of the gallbladder: an evaluation and reassessment of management strategy. J Clin Gastroenterol 44:687–695 Buscemi S, Orlando E, Damiano G et al (2015) "Pure" large cell neuroendocrine carcinoma of the gallbladder. Report of a case and review of the literature. Int J Surg. doi:10.1016/j.ijsu.2015.12.045 Jun SR, Lee JM, Han JK, Choi BI (2006) High-grade neuroendocrine carcinomas of the gallbladder and bile duct: Report of four cases with pathological correlation. J Comput Assist Tomogr 30: 604–609 Gustafsson BI, Kidd M, Chan A, Malfertheiner MV, Modlin IM (2008) Bronchopulmonary neuroendocrine tumors. Cancer 113:5–21 Nikou GC, Lygidakis NJ, Toubanakis C et al (2005) Current diagnosis and treatment of gastrointestinal carcinoids in a series of 101 patients: the significance of serum chromogranin-A, somatostatin receptor scintigraphy and somatostatin analogues. Hepatogastroenterology 52:731–741 Joo I, Lee JY, Baek JH et al (2014) Preoperative differentiation between T1a and >/=T1b gallbladder cancer: combined interpretation of high-resolution ultrasound and multidetector-row computed tomography. Eur Radiol 24:1828–1834 Obuz F, Altay C, Sagol O, Astarcioglu H, Oztop I, Igci E (2007) MDCT findings in neuroendocrine carcinoma of the gallbladder: case report. Abdom Imaging 32:105–107 El Fattach H, Guerrache Y, Eveno C et al (2015) Primary neuroendocrine tumors of the gallbladder: Ultrasonographic and MDCT features with pathologic correlation. Diagn Interv Imaging 96: 499–502 Kim SH, Kim SH, Kim MA, Shin CI, Han JK, Choi BI (2015) CT differentiation of poorly-differentiated gastric neuroendocrine
517
14.
15.
16.
17.
18.
19.
20.
21.
tumours from well-differentiated neuroendocrine tumours and gastric adenocarcinomas. Eur Radiol 25:1946–1957 Kim JH, Eun HW, Kim YJ, Lee JM, Han JK, Choi BI (2015) Pancreatic neuroendocrine tumour (PNET): Staging accuracy of MDCT and its diagnostic performance for the differentiation of PNET with uncommon CT findings from pancreatic adenocarcinoma. Eur Radiol. doi:10.1007/s00330-015-3941-7 Deehan DJ, Heys SD, Kernohan N, Eremin O (1993) Carcinoid tumour of the gall bladder: two case reports and a review of published works. Gut 34:1274–1276 Paulson EK, McDermott VG, Keogan MT, DeLong DM, Frederick MG, Nelson RC (1998) Carcinoid metastases to the liver: role of triple-phase helical CT. Radiology 206:143–150 Dromain C, de Baere T, Baudin E et al (2003) MR imaging of hepatic metastases caused by neuroendocrine tumors: comparing four techniques. AJR Am J Roentgenol 180:121–128 Vogl TJ, Naguib NN, Zangos S, Eichler K, Hedayati A, Nour-Eldin NE (2009) Liver metastases of neuroendocrine carcinomas: interventional treatment via transarterial embolization, chemoembolization and thermal ablation. Eur J Radiol 72:517–528 Harring TR, Nguyen NT, Goss JA, O'Mahony CA (2011) Treatment of liver metastases in patients with neuroendocrine tumors: a comprehensive review. Int J Hepatol 2011:154541 Chen C, Wang L, Liu X, Zhang G, Zhao Y, Geng Z (2015) Gallbladder neuroendocrine carcinoma: report of 10 cases and comparision of clinicopathologic features with gallbladder adenocarcinoma. Int J Clin Exp Pathol 8:8218–8226 Yun SP, Shin N, Seo HI (2015) Clinical outcomes of small cell neuroendocrine carcinoma and adenocarcinoma of the gallbladder. World J Gastroenterol 21:269–275