Ultrasonography of Osteosarcoma in Dogs
Sonography Atlas of Dogs

Accurate preoperative detection, localization, and staging of the primary tumor and metastases are essential for the selection of appropriate candidates for surgery. In dogs with insulinoma, preoperative assessment usually is performed with transabdominal ultrasonography (US). There are no reports on the use of computed tomography (CT) for this purpose. The preoperative use of somatostatin receptor scintigraphy (SRS) recently has been advocated for the identification of insulinoma and gastrinoma in dogs, but its accuracy remains to be established. In this report US, CT, and single-photon emission computed tomography (SPECT) with [111In-DTPA-D-Phe1]-octreotide (a specific form of SRS) were compared for their effectiveness in detecting and localizing primary and metastatic insulinoma in dogs. Findings at surgery or postmortem examination served as control. Of 14 primary insulinomas, 5, 10, and 6 were correctly identified by US, CT, and SPECT, respectively. No lymph node metastases were detected by US or
SPECT. CT identified 2 of 5 lymph node metastases but also identified 28 false-positive lesions. Two of 4 livers were found to be positive for metastases by 1 of the imaging techniques. US can be used for the initial evaluation of dogs with hypoglycemia. Although CT identifies most primary tumors, intraoperative inspection and palpation of the pancreas is still superior. SPECT appears as effective as US and CT in detecting insulinomas. Future developments in preoperative imaging techniques might improve current methods of canine insulinoma detection.

Key words: Abdominal ultrasonography; Computed tomography; Diagnostic imaging; Dog; Pancreatic endocrine tumor; Somatostatin receptor scintigraphy

Insulin-producing pancreatic endocrine tumor, or insulinoma, is the most common of the 3 types of pancreatic
endocrine tumor (PET) described in the dog (insulinoma, gastrinoma, and glucagonoma).1–3 Over 95% of insulinomas
in dogs are malignant, and 40–50% have visibly metastasized at the time of surgery.2,4,5 Insulinomas are identified
and staged by means of preoperative transabdominal ultrasonography (US) and, to a lesser extent, by computed tomography (CT) and by inspection and palpation during exploratory laparotomy.
Data on the effectiveness of preoperative imaging techniques in dogs with insulinoma are limited to uncontrolled
studies of transabdominal US.6–8 In human medicine, several imaging techniques have been described and evaluated for preoperative identification and staging of PETs. Somatostatin receptor scintigraphy (SRS) has become an important and often 1st-line imaging method in humans with malignant PETs.9–12 This imaging modality has a high sensitivity and a high specificity.13–15 Additional advantages of this technique are whole-body scanning for extra-abdominal metastases, gamma probing for intraoperative identification of tumor foci, and selection of subjects suitable for treatment with a radiolabeled somatostatin analogue.13,16 Recently, the use of SRS with [111In-DTPA-D-Phe1]-octreotide  was  described  for  identification  of  insulinomas  and gastrinomas  in  dogs,  but  the  diagnostic  accuracy  of  this method in dogs has yet to be established.17–19 The purpose of this study was to compare transabdominal US, CT , and single-photon emission computed tomography (SPECT), a specific form of SRS, for their effectiveness in identifying  primary  tumors and metastatic disease in dogs with insulinoma.

Materials and Methods
Dogs
The dogs with insulinoma included in this study were presented at the Department of Clinical Sciences of Companion Animals of Utrecht University  from  1998  to  2001.  Not  all  eligible  dogs  were  enrolledbecause of scheduling conflicts or lack of owner consent. Transabdominal US was performed  as  part of  the  routine  diagnostic  workup. Inclusion  criteria  included  CT  and  SPECT and  verification of localization, lesion size, and stage at surgery or postmortem examination within  4  weeks  of  these  diagnostic  studies.  Histology  results  had  to  be available.
                                                                                      Food was withheld from all dogs 6 h before CT and SPECT , which
were performed consecutively. The dogs were premedicated with medetomidinea IV , and anesthesia was induced with propofolb administered IV to effect. The trachea was intubated, and anesthesia during CT was maintained  with  inhalational  isofluranec  vaporized  in  nitrous  oxide/oxygen ina circle breathing system. Anesthesia was maintained with a  continuous  infusion  of  propofol  during  SPECT .  At  the  start  of  anesthesia,  a  urinary  catheter  was  placed  in  the  bladder,  and  urine  was collected in a closed system.
Surgery was performed by 1 of the authors who is an ACVS boardcertified surgeon (JK). The surgeon was not blinded to the results of the imaging studies. After a routine exploratory celiotomy, the pancreas was visualized and palpated for evidence of primary or metastatic disease. Primary tumors were removed by a marginal suture fracture technique. Thorough exploration of the draining lymph nodes and liver by visualization and palpation was performed; abnormal lymph nodes were removed by complete excision or were biopsied. Suspected liver metastases were removed if possible or biopsied. Information on tumor localization obtained at surgery was considered accurate  if  plasma  glucose  concentrations  performed  3  months  after removal of the tumor were within the normal reference range after an overnight fast. Normal glucose concentrations suggested that no macroscopic  tumor  lesions  were  missed  at  surgery.  Postmortem  examinations were  performed  by  1  of  the  authors who  is an  ECVP board-
certified pathologist (TSGAMvdI). The pathologist was blinded to the results of the imaging studies. The maximal diameter (md) of the tumor  was  measured  in  3  perpendicular  directions  witha  ruler  during surgery or at postmortem examination (md-Control).
The Research Committee of the Department of Clinical Sciences of Companion  Animals of Utrecht University  approved  the  studies, and owner consent was obtained before enrollment.

Transabdominal Ultrasonography
Ultrasonographic examination of the abdomen was performed with a high-definition ultrasound systemd equipped with broad-band phased array transducers (5–3 and 7–4 MHz). US examinations were performed by 1 of 2 of the authors (SAEBB or GV), both European specialists in veterinary diagnostic imaging. The transabdominal US
examination was performed with the dogs in dorsal and lateral recumbency. The regions dorsal and dorsomedial to the descending duodenum, right of the portal vein, and immediately ventral to the right kidney were scanned for the right pancreatic lobe. The pancreaticoduodenal vein, which runs through the right pancreatic lobe parallel
to the descending duodenum, also aided localization. The region ventral to the portal vein was scanned for the pancreatic corpus. For examination of the left pancreatic lobe, the region caudal to the cranial duodenal flexure and greater curvature of the stomach, left of the portal vein, was scanned. The splenic vein, which passes caudodorsally
to the left pancreatic lobe, was identified to aid localization. The liver and the regions of the pancreaticoduodenal, gastric, jejunal, splenic, and hepatic lymph nodes were carefully examined for possible metastases. The maximal diameters of the primary pancreatic lesion and metastatic lesions, if present (md-US), were measured along 2 or, if
possible, 3 perpendicular axes. The written reports and video recordings of the US studies were reviewed by 1 of the authors (SAEBB) without knowledge of the surgical or postmortem findings.

Computed Tomography (CT)
Computed  tomography  was  performed  witha  3rd-generation  CT scanner.  With the  dogs in  dorsal recumbency,  scans of  the  abdomen were made from the most cranial part of the liver toa region caudalto  the  left  kidney  witha  scanning  time  of  4.5  seconds,  120  kV ,  220mA, and 10- or 5-mm-thick consecutive slices. Scans were made be-
fore  and  after  administration  ina  cephalic  vein  of  2  mL/kg  body weight of sodium and meglumine ioxotalamatef IV , witha maximum of 60 mL. Ventilation was controlled manually in all dogs, and scanswere  made  at  the  end  of  expiration.  CT  images  were  reviewed  by  1 of  the  authors  (GV)  without  knowledge  of  surgical  or  postmortem
findings. The maximum diameter of all suspected lesions on CT (md-CT)  was  determined  by  measuring  the  lesion  on  the  slice  containing the largest cross section and by estimating the craniocaudal dimensionfrom the number of consecutive scans that contained part of the lesion and the slice thickness. The density of all suspected primary and metastatic  lesions  was  measured  on  both  unenhanced  and  contrast-enhanced images and compared with the density of the liver in each dog. Dimensions  and  densities  were  measured  from  the  display  monitor with a trackball-driven cursor and CT computer software with images displayed at the same window settings (window width  400 Hounsfield units, window length  40 Hounsfield units).

Single Photon Emission Computed Tomography
Six  hours  before  SPECT ,  a  median  of  155  MBq  (range  56–214MBq) [In-DTPA-D-Phe1]-octreotideg was injected IV . The synthesis and radiolabeling of the product have been described earlier.20,21

SPECT was performed witha single-head gamma camerah equipped with a  medium-energy  parallel-hole  collimator.  The  pulse  height  analyzer windows were centered over both  111In photon peaks (172 and 246 keV) witha window width of 20%. Data from both windows were added to the acquisition frames. The camera was connected toa dedicated  open  ICON  Workstation  computer  that  used  Siemens  ICON computer  system  software.  The  acquisition  time  of  64  projections (128-   128-word  matrix)  was  determined  by  the  number  of  counts in the 1st segment (90,000–110,000 counts). SPECT images were obtained over 65 minutes. Reconstruction was performed witha Butterworth filter (cutoff 0.35, order 7) without scatter correction. Attenuation  correction  was performed  assuming  uniform attenuation with an ellipse drawn around the body. The field of view covered at least the abdomen from the cranial aspect of the diaphragm to the cranial border of the pelvis. In small dogs, a zoom factor of 1.23 was used. SPECT studies of the dogs were reviewed by 1 of the authors (YWEAP) without knowledge of the surgical or postmortem findings.

Descriptive Analysis
The diagnostic imaging techniques were assessed for their value in identifying  the  primary  pancreatic  tumor(s),  lymph  node  metastases, and liver metastases. Identification involved not only the detection of a  lesion,  but  also  the  correct  allocation  of  that  lesion  toa  certain abdominal organ. A result was considered true positive when the primary  tumor  or  lymph  node  metastases  were  identified  correctly.  Results  for  metastatic  liver  disease  were  considered  true  positive  when at  least  1  of  the  liver  metastases  was  correctly  identified.  A  false-positive  result  was  defined  asa  lesion  identified  by  imaging  but  not found at surgery or postmortem examination or that was detected but allocated  to  the  wrong organ.  A false-negative result by imaging entaileda tumorous lesion that was not detected but found at surgery or postmortem examination or that was detected but located incorrectly.
The results are presented as median and range.

Results
Dogs
Thirteen of 16 dogs with insulinoma that initially entered the study met all of the inclusion criteria: 7 were males (1
intact) and 6 were females (2 intact) of 9 different breeds. The median age was 10 years (range 6–11 years), and the
median body weight was 22 kg (range 9–36 kg). Recurrent signs of hypoglycemia were present in all dogs before ad-
mission. In all dogs, the results of routine clinicopathologic investigations were within the reference range, with the exception  of  plasma  glucose  concentrations,  which  were  repeatedly below the lower limit of the reference range (4.0mmol/L,  or  72  mg/dL).  Hyperinsulinemia  was  diagnosed on  the  basis  of  low  plasma  glucose  concentrations  in  the presence  of  inappropriately  high  concentrations of  plasma insulin (ie, plasma insulin concentrations in the upper range of,  or  above,  the  reference  range  [34–63  pmol/L,  or  5.1– 9.4 IU/mL]). Transabdominal US studies were performed a  median  of  29  days  (range  9–42  days)  and  CT/SPECT studiesa median of 14 days (range 3–28 days) before surgery or postmortem examination.Information on tumor localization and size was obtained at  surgery  (9  dogs),  postmortem  examination  (1  dog),  or
both (3 dogs). Twelve dogs hada solitary pancreatic tumor, and  1  dog  had  2  pancreatic  tumors. Four tumors were located in the right pancreatic lobe, 2 in the pancreatic corpus, and  8  in  the  left  pancreatic  lobe.  The  median  md-Control of the primary tumors was 15 mm (range 7–50 mm). One dog  had  1  pancreaticoduodenal  and  1  left  hepatic  lymph node metastasis; 3 dogs each had 1 metastasis ina splenic lymph  node.  The  5  lymph  node  metastases  hada  median md-Control  of  20  mm  (range  10–25  mm).  Three  of  the  4 dogs with lymph node metastases also had liver metastases. In  1  dog,  multiple  subcapsular  liver  lesions  (md-Control range  1–2  mm)  were  found.  Another  dog  had  30  liver metastases (md-Control range 1–15 mm). The 3rd dog had
4  liver  lesions  (md-Control:  2,  5,  14,  and  20  mm).  In  1 other dog, a solitary liver metastasis (md-Control: 12 mm)
was found without lymph node involvement. The  cumulative  descriptive  analysis  for  each  diagnostic imaging technique is summarized in Table 1, and the results for each tumor lesion found during surgery or postmortem
examination are described in Table 2.

Table 1. Cumulative descriptive analysis of the results of transabdominal ultrasonography (US),

computed tomography (CT), and single-photon emission computed tomography (SPECT) in 13

dogs with insulinoma

R, right pancreatic lobe; C, pancreatic corpus; L, left pancreatic lobe; T , whole pancreas.

a Asterisks indicatea lesion detected with the imaging technique but localized  incorrectly, 

causing  the  result  to  become  false  positive  or false negative.

b Numbers in parentheses are number of pancreatic tumor lesions (1
dog had 2 pancreatic lesions), metastatic lymph nodes, or livers with
metastases found at surgery or postmortem examination.

Transabdominal Ultrasonography
The  quality  of  the image was poor  in 9  of  13  dogs because  of gas or  other contents in the stomach, duodenum,
or  colon.  In  12  dogs,  only  parts  of  the  pancreatic  region could  be  identified,  and  in  1  dog,  no  pancreas  could  be visualized at all. Regional lymph nodes were identified only when  they  had  an  abnormal  hypoechoic  structure or were enlarged. Most of the liver could be visualized in all dogs. Primary and lymph node tumor lesions were recognized as round or lobulated, hypoechoic nodules (Fig 1). Metastatic liver  lesions  were  hypoechoic.  In  1  dog,  the  liver  lesions had hyperechoic borders.
The median md-Control of 5 pancreatic masses that were correctly identified was 20 mm (range 15–30 mm), whereas
the diameter of the 9 unidentified lesions was 15 mm (range 7–50  mm).  One  lesion  (dog  3),  considered  an  enlarged splenic  lymph  node  on  US,  could  well  have  been  an  unidentified primary insulinoma on the basis of its size (md-US 26 mm versus md-Control 27 mm) and localization (Table  2).  Another  mass  (dog  12)  was  identified  as  primary tumor  on  US  but  most  likely  wasa  splenic  lymph  node metastasis on the basis of its size. Botha false-positive and false-negative  result  were  obtained  witha  primary  lesion (dog  13)  identified  in  the  right  lobe  on  US,  whereas  the primary tumor was found in the left pancreatic lobe at surgery and postmortem examination (Tables 1, 2). In the same dog, 2 (md-Control: 10 mm) of 30 liver metastases were identified on US. In 1 dog, a false-positive finding was due to an irregular liver structure on US that was interpreted as metastatic  disease  (Table  1).  At  postmortem  examination, hepatocellular nodular hyperplasia was found.

Computed Tomography
In  all  dogs,  lesions  that  were  considered  to  be  primary pancreatic tumors were slightly hypodense compared with liver tissue. The density of these lesions increased after contrast medium injection, either uniformly or irregularly, but the lesions remained slightly hypodense compared with the contrast-enhanced liver tissue. In 11 dogs, 1 or more lymphnodes were identified as round or oval soft tissue structures that  uniformly  enhanced  after  contrast  medium  injection. These lymph nodes were suspected to contain tumor. Liver metastases were visualized as hypodense lesions. In 1 dog, these hypodense lesions were bordered by hyperdense rims.
The median md-Control of the 10 pancreatic lesions that were  correctly  identified  was  20  mm  (range  10–50  mm),
whereas  the  diameter  of  the  4  unidentified  lesions  was  7, 8, 15, and 25 mm, respectively. One of 2 lesions (dog 1), both  allocated to the pancreas, proved to bea left hepatic lymph  node  metastasis at  surgery  and  postmortem examination.  In  this  dog,  CT  also  identified  tumor  growth  into the portal vein, which was confirmed at surgery and postmortem examination. A primary pancreatic lesion (dog 13) on  CT  was considereda splenic lymph node in retrospect on  the  basis  of  its  size  (Table  2).  In  11  dogs,  30  lymphnodes  were  identified  and  considered  suspect on  CT  (median  md-CT  10  mm,  range  8–40  mm).  The  majority consisted of left hepatic (n  7) and splenic (n  12) lymphnodes. Two of the latter were tumor-positive at surgery or postmortem  examination  (md-Control: 20  and  10  mm, respectively; Table 1). One liver lesion identified on CT (md-CT:  19  mm)  was  not  associated  with  the  metastatic  liver disease  present  in  dog  1.  The  smallest  of  the  4  liver  metastases in dog 11 (md-Control: 2 mm) was not detected.

Table 2. Descriptive analysis of the results of transabdominal ultrasonography (US), computed tomography (CT), and single-photon emission computed tomography (SPECT) per tumor lesion found during surgery or postmortem examination in 13 dogs with insulinoma.

a md-Control,  maximum  diameter  of  lesion(s)  as  determined  at  surgery  or  postmortem  examination;  TP ,  true  positive;  TN,  true  negative;  FP , false positive; FN, false negative. FP* or FN*, a true tumorous lesion detected with the imaging technique but localized incorrectly, causing the result to become false positive or false negative, respectively.
b This represents 2 pancreatic lesions in 1 dog.

Fig 1. A transverse ultrasound image through the right lateral abdominal wall with the dog

in left ventral recumbency (dog 4). A round, hypoechoic mass (T) is visible medial to the

duodenum (D). This primary pancreatic tumor was not identified with CT and SPECT. The

solitary liver metastasis in this dog was identified with all 3 techniques (Table 2). Distance

between tics on the right bar is 0.5 cm

Single-Photon Emission Computed Tomography
Uptake  of  radioactivity  in  the  kidneys  and  gall  bladder was  readily  recognized.  The  radioactivity  in  the liver was weak  and  patchy.  All  dogs  hada  circumscribed  accumulation  of  radioactivity  in  the  left  epigastrium. Other accumulations   of   radioactivity   on   SPECT   images   differed among dogs but were mainly related to the intestines (Figs 2, 3).
The median md-Control of the 6 pancreatic masses that were  correctly  identified  was  24  mm  (range  15–50  mm),
whereas  the  median  diameter  of  the  8  lesions missed was 15  mm  (range  7–25  mm).  The  primary  tumors  in  dogs  5 and 10 were allocated to the pancreatic corpus on SPECT but  proved  to  be  present  in  the  right  lobe  at  surgery  (Fig 3). A primary lesion (dog 13) identified in the left lobe on SPECT  was,  in  retrospect,  more likely  to  bea  closely related splenic lymph node metastasis (Table 2).
The  accumulation  of  radioactivity  in  the  liver  (dog  11) was associated with the largest (md-Control 20 mm) of the
4 liver metastases found at postmortem examination. In the same dog, the abnormal distribution of radioactivity related to the gall bladder also could have been caused by the 2nd largest liver metastasis (md-Control: 14 mm). At postmortem examination, this metastasis was found close to the gall bladder.

Fig  2.    (A)  A  ventral  view  ofa  3-dimensional  reconstruction  of  a SPECT  study 

performed  6  hours  after  injection  of  [111In-DTPA-D-Phe1]-octreotide  (dog  9). 

Radioactivity  accumulated  in  kidneys,  gall bladder (G), gastric fundus (F), and primary

tumor in the left pancreatic  lobe  (T).  Some  radioactivity  was  detected  in  the 

intestinal  tract. (B, C) Corresponding transverse CT and SPECT images, respectively,

in the same dog. On the CT image, the right kidney (K), gastric fundus (F), spleen (S),

liver (L), and primary tumor (T) can be identified.

Fig  3.    (A)  A  ventral  view  ofa  3-dimensional  reconstruction  of  a
SPECT  study  performed  6  hours  after  injection  of  [111In-DTPA-D
-Phe1]-octreotide (dog 10). Radioactivity accumulated in kidneys, gall
bladder (G), and primary tumor in the right pancreatic lobe (T). Some
radioactivity was detected in the intestinal tract. (B, C) Corresponding
transverse CT and SPECT images, respectively, in the same dog. On
the  CT image,  the  right kidney  (K)  and  spleen  (S)  can  be  identified.
The  primary  tumor  was  identified  on  SPECT  (T),  but  only  in  retro-
spect; possibly identified on CT (T?).

Comparison of US, CT, and SPECT
Of  the  10  primary  tumors  identified  on  CT ,  2  (1  in  the right pancreatic lobe, 1 in the pancreatic corpus) also were found with SPECT and US, 2 (1 in the right lobe, 1 in the left  lobe)  with  US  but  not  with  SPECT ,  and  3  (1  in  the corpus,  2  in  the  left  lobe)  with  SPECT  but  not  with  US (Tables 1, 2). Of the 4 primary tumors not identified with CT , 1 (right lobe) was found with US (Fig 1) and 1 (right lobe) with SPECT only (Fig 3). Two primary tumors (left
lobe; md-Control: 7 and 8 mm, respectively) were not detected  by  any  of  the  imaging  techniques.  Of  the  5  lymph node metastases, only 2 were correctly identified with CT . Two  splenic  lymph  nodes  and  1  left  hepatic  lymph  node metastasis were detected with different imaging modalities but misinterpreted as primary tumors in the left lobe. The solitary  liver  metastasis  in  1  dog  was  identified  with  all techniques (Table 2). CT and SPECT identified metastases in 1 other dog, whereas US was the only technique to visualize metastases ina 3rd dog. Small metastases (md-Control  range  1–2  mm)  in  the  4th  dog  were  missed  with  all techniques.
Discussion
Surgical excision is the treatment choice for insulinomas in  dogs  despite  the  presence  of  metastatic  disease  in  40–50%  of  affected  dogs.2  Preoperative  imaging  is  necessary to select candidates for surgery, to plan surgery, and to be able to inform owners about the risk of loss of pancreatic function, the chance of therapeutic success, and the extent of surgery.22
The  accuracy  of  US  depends  on  the  experience  of  the operator, the size of the lesion, and the quality of the sonograms. Bowel gas and obesity are the most common causes of  poor-quality  transabdominal  sonograms  and  resultant failure  to  image  the  pancreas  in  humans.23,24  Additionally, image quality in the dogs in this study was negatively influenced  by  restlessness,  body  conformation  witha  deep chest,  and  intestinal  content.  Poor  image  quality  was  the main reason why few primary tumors (especially in the left lobe)  and  lymph  node  metastases were  detected  by  US. Withholding food, anesthesia, and gastric intubation to remove gas from the stomach and to administer water or saline  all  could  have  improved  the  results  of  US.  However, US in these dogs was performed as part of the routine diagnostic evaluation, even beforea definite diagnosis of insulinoma  was  made.  Because  withholding  food  and  anesthesia  pose  an  increased  risk  in  dogs  with  hypoglycemia, US was performed in these dogs without specific preparation.
Thus, we found US to have low sensitivity in detecting canine insulinoma. Ina retrospective study, the detection of
pancreatic neoplasia in dogs by transabdominal US was reported to be 75% (12/16 dogs).7 Thirteen of these 16 dogs
had insulinomas, and in 4 of these dogs, the tumor was not detected. In no more than 6 dogs, the tumor was detected
and accurately localized, whereas the other dogs were considered to have ‘‘probable pancreatic masses.’’ The detec-
tion and accurate localization of the primary insulinoma in 6 of 13 dogs is comparable to our findings in 5 of 13 dogs.
In  human  medicine,  the  sensitivity  of  transabdominal  US for  the  detection  of  insulinomas  varies  between  9  and 79%.25  This  variation  could  reflect  improved  quality  of sonographic equipment over the years, differences in study design, differences in the experience of the examiners, variation in types of neuroendocrine tumors, and small patient groups. Nevertheless, US is readily available, noninvasive, and relatively inexpensive. Ultrasound-guided percutaneous biopsy also might identify the lesion. Despite low sensitivity in the detection of insulinomas, transabdominal US stil might  be  useful  for  initial  assessment  of  dogs  with  hypoglycemia and for the exclusion of differential diagnoses of hypoglycemia such as insulinlike growth factor II–like peptide-producing extra-pancreatic tumors.26
To our knowledge, this is the 1st report on the use of CT in  dogs  with  insulinoma.  CT  identified  most  primary  in-
sulinomas (10/14). In humans, the sensitivity of CT for the localization   of   primary   insulinoma   depends   on   tumor
size.27,28  Small tumors might not distort the contour of the human  pancreas  and  could  be  missed  because  their attenuation  is  similar  to  that  of  normal  pancreatic parenchyma on  both  pre-  and  postcontrast  scans.  In  this  respect,  the relatively  thin  pancreas  in  dogs  might  be  advantageous. However, the thin pancreas also could explain why it was difficult to correctly allocatea lesion to either the pancreas or an adjacent lymph node on US and CT imaging.
Variation  in  the  size  and  number  of  unaffected  lymph nodes  and  inability  to  distinguish  tumorous  from  nontumorous  enlargement  of  lymph  nodes  on  CT  precluded  a definitive diagnosis of metastatic lymph node disease. Because not all lymph nodes could be visualized, lymph nodes that were detected with CT were considered to contain metastases,  which  resulted  in  many  (28)  false-positive  findings. This result suggests that CT might not be appropriate for identifying lymph node metastases. In humans with gastrinomas,  CT  better  detects  pancreatic  lesions  (80%)  than extrapancreatic,  extrahepatic  lesions  (35%).11  CT  was  not better than US or SPECT in detecting metastatic liver disease.  New  techniques,  such  as  spiral  CT ,  with  dual  phase contrast enhancement, have improved the sensitivity of detection of PETs in humans compared with conventional CT (85  versus 16–72%).25,29  The  introduction  of  these  techniques  to  veterinary  medicine  also  could  improve  the  accuracy of the identification of canine insulinoma.
When assessing imaging techniques for the detection of PETs  in  humans,  a  distinction  is  made  between  primary
tumor  and  metastatic  disease.12  Intraoperative manual palpation and intraoperative US have proven very accurate in identifying primary insulinomas in humans (sensitivity 90–100%).10,27,28,30,31 Also in our study, all primary insulinomas were found by careful inspection and palpation of the pancreas. Additionally, intraoperative US can detect where the insulinoma lies in relation to the pancreatic duct, common bile duct, and adjacent blood vessels, providing the surgeon with  important  information.  Given  the  low  sensitivity  of preoperative  imaging  techniques  (including  SRS14)  in  humans with insulinoma, it has been questioned whether preoperative  imaging  is  useful  at  all.10,32  In  answering  this question, it should be remembered that insulinomas are benign in about 90% of human patients, and preoperative imaging is used mainly to detect relatively small, primary pancreatic tumors. The detection of metastatic disease is of less importance.
Numerous imaging methods have been used to assess the extent  of  metastatic  disease  in  humans  with  PETs.11,22,33–35 SRS has become the primary diagnostic imaging technique for the identification and staging of malignant PETs in humans 9–12,14 and is considered more sensitive in detecting metastases than conventional US and CT .11,13,14
Lesion detection with SRS depends on the accumulation of radioactivity, which is determined primarily by the size
of  the  lesion  and  the  density  of  somatostatin  receptors.36 The  primary  tumors  detected  in  this study  were relatively small (md-Control  2 cm in 10/14) compared with malignant  PETs  such  as  pancreatic  gastrinomas  (2–3  cm).14,37 Therefore,  the  poor  to  moderate  performance  of  SRS  in dogs  with  insulinomas  could  be  related  to  the  size  of  the tumors,  but  also  toa  low  density  of  high-affinity somatostatin  receptors,  as demonstrated in about 40% of insulinomas  in  humans.11,37,38  However,  preliminary  data indicate that  most  insulinomas  in  dogs  possess  sufficient  number of high-affinity somatostatin receptors for visualization.19 A 3rd aspect that might have influenced the SPECT detection of primary tumors and metastases in dogs was the nonspecific accumulation of radioactivity in the gall bladder, kidneys, and gastrointestinal tract. In humans, a low-fiber diet and use of laxatives before SRS or SRS performed shortly (4–6  hours)  after  administration of the radiopharmaceutical  can  reduce  this  nonspecific  accumulation  of  radioactivity.39 Furthermore, tumor-related hot spots can be distinguished  from  nonspecific  bowel  radioactive  accumulation by repeating SRS aftera certain interval.13,14 In dogs, these
measures have practical limitations, such as increased risk of fasting and additional general anesthesia. Moreover, re-
cent findings suggest thata considerable amount of radioactivity is retained in the gastric and intestinal wall in dogs,
which could limit the effectiveness of these measures.40
Transabdominal US and CT provide information on anatomical relations and the location of lesions, whereas SRS
characterizes the lesion.12,41 Thus, with the combination of US or CT with SPECT , the detection, localization, and staging of the tumor could be improved. SPECT appears to be as effective as US and CT in detecting primary insulinomas and metastases. However, the combination of imaging modalities, in terms of detection rate, was only slightly better than the individual imaging modalities.
We proposea preliminary protocol for the identification and staging of insulinomas in dogs on the basis of the re-
sults of this study and studies of human patients. Transabdominal US could be used for the differential diagnosis and
initial  assessment  (especially  for  liver  metastases).  Intraoperative inspection and palpation is preferred for the identification of the primary tumor, but CT could be considered for  additional  preoperative  evaluation.  Intraoperative  US could  becomea  useful  additional  approach  for  detecting primary tumors. With improvements in study protocols and techniques, the combination of US, CT , and SRS could provide improved preoperative information about tumor status.

Footnotes
a Domitor, SmithKline Beecham Animal Health B.V ., Zoetermeer, The
Netherlands
b Diprivan, Zeneca B.V ., Ridderkerk, The Netherlands
c Forene, Abbott Laboratories B.V ., Maarsen, The Netherlands
d HDI 3000, Advanced Technology Laboratories, Woerden, The Neth-
erlands
e Tomoscan CX/S, Philips NV , Eindhoven, The Netherlands
f Telebrix  350  (350  mg  iodine/mL),  Guerbet  Nederland  B.V .,  Gorin-
chem, The Netherlands
g OctreoScan, Mallinckrodt Medical B.V ., Petten, The Netherlands
h Integrated  ORBITER  Gamma  Camera,  Siemens  Medical  Systems,
Hoffman Estates, Illinois, USA

Acknowledgments
The  studies  were  performed  at  the  Departments  of Clinical  Sciences  of  Companion  Animals,  Diagnostic  Im-
aging and Pathology of the Faculty of Veterinary Medicine, Utrecht  University,  Utrecht,  The  Netherlands.  We  thank
Mallinckrodt  Medical  B.V .,  Petten,  The  Netherlands,  for their generous donation of [111In-DTPA-D-Phe1]-octreotide. We  acknowledge  the  support  of  the  staff  of  the  Anesthesiology  Division  of  the  Faculty  of  Veterinary  Medicine, Utrecht University, Utrecht, The Netherlands.

References
1.  Gross TL, O’Brien TD, Davies AP , Long RE. Glucagon-produc-
ing pancreatic endocrine tumors in two dogs with superficial necrolytic
dermatitis. J Am Vet Med Assoc 1990;197:1619–1622.
2.  Leifer CE, Peterson ME, Matus RE. Insulin-secreting tumor: Di-
agnosis and medical and surgical management in 55 dogs. J Am Vet
Med Assoc 1986;188:60–64.
3.  Simpson KW , Dykes NL. Diagnosis and treatment of gastrinoma.
Sem Vet Med Surg (Small Anim) 1997;12:274–281.
4.  Caywood DD, Klausner JS, O’Leary TP , et al. Pancreatic insulin-
producing  secreting  neoplasms:  Clinical,  diagnostic,  and  prognostic
features in 73 dogs. J Am Anim Hosp Assoc 1988;24:577–584.
5.  Trifonidou MA, Kirpensteijn J, Robben JH. A retrospective eval-
uation of 51 dogs with insulinoma. VetQ 1998;20:S114–S115.
6.  Feldman EC, Nelson RW. Canine and Feline Endocrinology and
Reproduction, 2nd ed. Philadelphia, PA: WB Saunders; 1996:422–441.
7.  Lamb  CR,  Simpson  KW ,  Boswood  A,  Matthewman  LA. Ultra-
sonography of pancreatic neoplasia in the dog: A retrospective review
of 16 cases. Vet Rec 1995;137:65–68.
8.  Mantis  P ,  Lamb  CR.  Sensitivity  of  ultrasonography  for  canine
insulinomas. Proceedings of the 6th Annual Conference of the Euro-
pean  Association  of  Veterinary  Diagnostic Imaging, Vienna, Austria,
1999:40.
9.  Angeli  E,  Vanzulli  A,  Gastrucci  M,  et  al.  Value  of  abdominal
sonography  and  MR  imaging  at  0.5  T  in  preoperative  detection  of
pancreatic  insulinoma:  A  comparison  with  dynamic  CT  and  angiog-
raphy. Abdom Imaging 1997;22:295–303.
10.  Hashimoto LA, Walsh RM. Preoperative localization of insulin-
omas is not necessary. J Am Coll Surg 1999;189:368–373.
11.  Modlin IM, Tang LH. Approaches to the diagnosis of gut neu-
roendocrine  tumors:  The  last  word  (today).  Gastroenterology  1997;
112:583–590.
12.  Ricke J, Klose K-J. Imaging procedures in neuroendocrine tu-
mours. Digestion 2000;62:39–44.
13.  Chiti A,  Fanti S,  Savelli G,  et al.  Comparison of somatostatin
receptor imaging, computed tomography and ultrasound in the clinical
management of neuroendocrine gastro-entero-pancreatic tumours. Eur
J Nucl Med 1998;25:1396–1403.
14.  Gibril  F ,  Jensen  RT .  Comparative  analysis  of  diagnostic  tech-
niques for localization of gastrointestinal neuroendocrine tumors. Yale
J Biol Med 1997;70:509–522.
15.  Gibril  F ,  Reynolds  JC,  Chen  CC,  et  al.  Specificity  of  somato-
statin receptor scintigraphy: A prospective study and effects of false-
positive  localizations  on  management in  patients  with  gastrinomas. J
Nucl Med 1999;40:539–553.
16.  Krenning EP , Kooij PPM, Pauwels S, et al. Somatostatin recep-
tor: Scintigraphy and radionuclide therapy. Digestion 1996;57(Suppl):
57–61.
17.  Altschul M,  Simpson  KW ,  Dykes NL,  et al.  Evaluation of so-
matostatin analogues for the detection and treatment of gastrinoma in
a dog, J Small Anim Pract 1997;38:286–291.
18.  Lester NV , Newell SM, Hill RC, Lanz OI. Scintigraphic diag-
nosis  of  insulinoma  ina  dog.  Vet  Radiol  Ultrasound  1999;40:174–
178.
19.  Robben JH, Visser-Wisselaaar HA, Rutteman GR, et al. In vitro
and  in  vivo  detection  of  functional  somatostatin  receptors  in  canine
insulinomas. J Nucl Med 1997;38:1036–1042.
20.  Bakker  WH,  Albert  R,  Bruns  C,  et  al.  [111In-DTPA-D-Phe1]-
octreotide, a potential radiopharmaceutical for imaging of somatostatin
receptor-positive tumors: Synthesis, radiolabeling and in vitro valida-
tion. Life Sci 1991;49:1583–1591.
21.  Krenning  EP ,  Bakker  WH,  Kooij PPM,  et al.  Somatostatin re-
ceptor  scintigraphy  with  indium-111-DTPA-D-Phe-1-octreotide  in
man:  Metabolism,  dosimetry  and  comparison  with  iodine-123-Tyr-3-
octreotide. J Nucl Med 1992;33:652–658.
22.  Goldstone AP , Scott-Coombes DM, Lynn JA. Surgical manage-
ment of gastrointestinal endocrine tumours. Bailliere’s Clin Gastroen-
terol 1996;10:707–736.
23.  Kolmannskog  F ,  Swensen  T ,  Vatn  MH,  Larsen  S.  Computed
tomography and ultrasound of the normal pancreas. Acta Radiol Diagn
1982;23:443–451.
24.  Miller DL. Localization of islet cell tumors. In: Becker KL, ed.
Principles  and  Practice  of  Endocrinology  and  Metabolism,  3rd  ed.
Philadelphia, PA: Lippincott, Williams and Wilkins; 2001:1477–1482.
25.  Chatziioannou A, Kehagias D, Mourikis D, et al. Imaging and
localization  of  pancreatic  insulinomas.  J  Clin  Imaging  2001;25:275–
283.
26.  Boari A, Barreca A, Bestetti GE, et al. Hypoglycemia ina dog
with  leiomyoma  of  the  gastric  wall producing  an  insulin-like growth
factor II–like peptide. EurJ Endocrinol 1995;143:744–750.
27.  Boukhman MP , Karam JM, Shaver J, et al. Localization of in-
sulinoma. Arch Surg 1999;134:818–823.
28.  Kuzin NM, Egorov AV , Kondrashin SA, et al. Preoperative and
intraoperative  topographic  diagnosis  of  insulinoma.  WorldJ  Surg
1998;22:593–598.
29.  King AD, Ko GTC, Yeung VTF , et al. Dual phase spiral CT in
the detection of small insulinomas of the pancreas. BrJ Radiol 1998;
71:20–23.
30.  Lo, C-Y , Lam, K-Y , Kung AWC, et al. Pancreatic insulinomas.
A 15-year experience. Arch Surg 1997;132:926–930.
31.  Machado MCC, Monteiro da Cunha, JE, Jukemura J, et al. In-
sulinoma: Diagnostic strategies and surgical treatment. A 22-year ex-
perience. Hepatogastroenterology 2001;48:854–858.
32.  Azimuddin  K,  Chamberlain  RS.  The  surgical  management  of
pancreatic neuroendocrine tumors. Surg Clin North Am 2001;81:511–
525.
33.  Bliss  RD,  Carter  PB,  Lennard  TWJ.  Insulinoma:  A  review  of
current management. Surg Oncol 1997;6:49–59.
34.  Anderson MA, Carpenter S, Thompson NW , et al. Endoscopic
ultrasound is highly accurate and directs management in patients with
neuroendocrine tumors of the pancreas. AmJ Gastroenterol 2000;95:
2271–2277.
35.  Baba Y , Miyazono N, Nakajo M, et al. Localization of insulin-
omas.  Comparison  of  conventional  arterial  stimulation  with  venous
sampling  (ASVS)  and  superselective  ASVS.  Acta  Radiol  2000;41;
172–177.
36.  Ohrvall Ul, Westlin J-E, Nilsson S, et al. Human biodistribution
of   [111In]diethyltriaminepentaacetic   acid-(DTPA)-D-[Phe1]-octreotide
and  peroperative  detection  of  endocrine  tumors.  Cancer  Res  1995;
55(Suppl):5794s–5800s.
37.  Zimmer  T ,  Sto  ¨lzel  U,  Ba  ¨der  M,  et  al.  Endoscopic  ultrasonog-
raphy  and  somatostatin  receptor  scintigraphy  in  the  pre-operative  lo-
calisation of insulinomas and gastrinomas. Gut 1996;39:562–568.
38.  Lamberts  SWJ,  Krenning  EP ,  Reubi  J-C.  The  role  of  somato-
statin  and  its  analogs  in  the  diagnosis  and  treatment  of  tumors.  En-
docrinol Rev 1991;12:450–482.
39.  Leners  N,  Jamar  F ,  Fiasse  R,  et  al.  Indium-111-pentetreotide
uptake in endocrine tumors and lymphoma. J Nucl Med 1996;37:916–
922.
40.  Robben J, Reubi JC, Pollak Y , Voorhout G. Biodistribution of
[111In-DTPA-D-Phe1]  octreotide  in  dogs:  Uptake  in  the  stomach  and
intestines but not in the spleen points towards interspecies differences.
Nucl Med Biol 2003;30:225–232.
41.  Debray MP , Geoffroy O, Laissy JP , et al. Imaging appearances
of metastases from neuroendocrine tumours of the pancreas. BrJ Ra-
diol 2001;74:1065–1070.

Share this article / Teilen Sie diesen Artikel