Islet Cell Tumors of the Pancreas:
`
`Pathologic—lmaging Correlation Among
`Size, Necrosis and Cysts, Calcification,
`
`Malignant Behavior, and Functional Status
`
`Peter C. Buetow‘ '2
`Tremont V. Parrino2'3
`James L. Buck"?
`Linda Pantongrag-Brown ‘ '2
`Pablo R. Ros”
`Abraham H. Dachman"5
`David F. Cruesse
`
`Received March 22. 1995; accepted after revi-
`sion June 15. 1995.
`
`The opinions and assertions contained herein
`are the private views of the authors and are not to
`be construed as official or as reflecting the views of
`the Departments of Defense. Army. or Navy.
`Presented at the annual meeting of the Ameri-
`can Roentgen Ray Society, Washington. DC, April-
`May 1995.
`
`‘Department of Radiologic Pathology, Armed Forc-
`es Institute of Pathology, Bldg. 54. Rm. M-121,Alaska
`Ave. and Fern St., N.W., Washington, DC 20306-
`6000. Address correspondence to P. C. Buetow.
`
`2Departments of Radiology and Nuclear Medi-
`cine, Uniformed Services University of the Health
`Sciences, Bethesda. MD 20814-4799.
`
`“Department of Radiology, National Naval Med-
`ical Center, Bethesda, MD 20814-4799.
`
`OBJECTIVE. The purpose of our study was to correlate the imaging and pathologic
`features of islet cell tumors with regard to tumor size, necrosis and cysts, calcifica-
`tion, malignant behavior, and functional status.
`MATERIALS AND METHODS. We retrospectively reviewed the clinical, pathologic, and
`imaging features of all 133 cases of pathologically proved islet cell tumors of the pancreas
`seen at the Armed Forces institute of Pathology. Clinical data, including the patients‘ symp-
`toms and serologic characteristics, were used to distinguish hyperfunctioning tumors
`(those causing symptoms related to elevated serum polypeptide levels) from nonhyper-
`functioning tumors; hyperfunctioning tumors were divided further into insulin-producing
`and non—insulin-producing types. All patients had at least one cross-sectional imaging
`study, including CT (n = 118), sonography (n = 42), or MR imaging (n = 22). Clinical, patho-
`logic, and imaging features were evaluated and correlated with tumor size, necrosis and
`cysts, calcification, local invasion, vascular invasion, metastases, and functional status.
`RESULTS. Islet cell tumors with areas of necrosis or cystic change found patholog-
`ically and on imaging studies (56/133) were larger (8.4 cm in mean transverse diame-
`ter) than homogeneous solid lesions (2.9 cm in mean transverse diameter) and were
`predominantly non—insulin producing (48/56) and nonhyperfunctioning (36/56). Of the
`43 insulinomas, 35 were small (2.2 cm in mean transverse diameter), solid, and homo-
`geneous. Larger size also was associated with calcification and malignant behavior,
`including local invasion, vascular invasion, and distant metastases.
`CONCLUSION. Our findings show that cystic and necrotic islet cell tumors are usu-
`ally non—insulin-producing and nonhyperfunctioning neoplasms and larger than the
`typically solid and small insullnomas. Calcification, local invasion, vascular invasion,
`and metastatic disease are more commonly seen with larger neoplasms.
`
`AJR 1995;165:1175-1179
`
`Islet cell tumors constitute a broad group of endocrine neoplasms. They are
`best divided on clinical grounds into those that are hyperfunctioning (the so-called
`functioning islet cell tumors) and those that are nonhyperfunctioning (the so-called
`nonfunctioning islet cell tumors). One would expect hyperfunctioning tumors to be
`manifested earlier and to be small
`in comparison with nonhyperfunctioning
`tumors, which one would expect to be manifested later by virtue of mass effect,
`local invasion, or metastases. Furthermore, one would expect larger tumors to
`have areas of necrosis or cystic degeneration. We studied the imaging and patho-
`logic features of 133 cases and correlated the findings of size, necrosis and cysts,
`calcification, malignant behavior, and functional status of the islet cell tumors.
`
`‘Department of Radiology, University of Florida.
`Gainesville. FL 32610.
`
`Materials and Methods
`
`5Depanmemo1 Radiology, University of Chicago.
`Chicago, IL 60637.
`“Department of Preventive Medicine. Uni-
`formed Services University of the Health Sciences.
`Bethesda, MD 20814-4799.
`0361—803XI95/1655-1175
`©American Roentgen Ray Society
`
`We retrospectively reviewed the clinical records, radiologic appearance, and pathologic
`findings of all 133 islet cell tumors studied by cross-sectional imaging at the Armed Forces
`Institute of Pathology between January 1980 and January 1994. These tumors occurred in
`124 patients; multiple tumors were found in seven patients. All cases were pathologically
`proved by surgical resection or surgical inspection and by biopsy results. Results of immuno-
`histochemical staining performed to determine specific hormone production within each tumor
`varied from case to case. A single case of nesidioblastosis was excluded from the study.
`
`- st-Ward Pharm.
`Exhibit 1007
`Page 001
`
`West-Ward Pharm.
`Exhibit 1007
`Page 001
`
`

`

`1176
`
`BUETOW ET AL.
`
`AJR:165, November 1995
`
`Clinical data regarding the patients’ presenting signs and symptoms
`were recorded. Criteria used to diagnose functioning islet cell tumors
`were as follows: insulinoma-—symptomatic hypoglycemia with inappropri-
`ately elevated plasma insulin levels; gastn'noma—symptoms and signs of
`peptic ulcer disease and elevated semm gastrin levels; glucagonoma—
`diabetes mellitus, dermatitis, and painful glossitis; sornatostatinoma——ele-
`vated serum somatostatin levels, diabetes mellitus, gallbladder disease.
`and steatorrhea; vipoma—profuse secretory diarrhea and elevated vase
`active intestinal polypeptide levels; and adrenocorticotropin hormone
`(ACTH)-producing
`fumors—Cushing’s
`syndrome. Tumors
`causing
`slightly elevated serum peptide levels or positive immunohistochemical
`staining results but not causing symptoms were classified as nonhyper-
`functioning tumors. Tumors causing elevated pancreatic polypeptide lev-
`els also were classified as nonhyperfunctioning tumors. Three patients
`with type 1 multiple endocrine neoplasia syndrome had gastrinomas.
`One hundred eighteen patients had CT, 42 had sonography, and 22
`had MR imaging. Imaging features were analyzed independently by two
`radiologists and were correlated with pathologic and surgical findings.
`These features included the size and location of the mass; the presence
`of cystic changes; findings of local invasion into surrounding organs or
`adenopathy; vascular invasion into the splenic vein, superior mesenteric
`vein, or splenic artery; and the presence of metastatic disease involving
`the liver or other distant organs. MR imaging and CT findings also were
`used to identify solid nonenhancing areas and solid homogeneously
`enhancing masses. CT alone was used to detect the presence of calcifi-
`cation. Any discrepancies were resolved by consensus. With regard to
`location,
`in two cases in which the lesions were extremely large and
`
`replaced the entire pancreas, the center was recorded as being within
`the body of the pancreas. Imaging studies were performed on a wide
`variety of equipment over many years, and protocols regarding imaging
`parameters, contrast injections, and other details were not standardized.
`Correlation was made among the description and photographs of the
`gross specimen, the histopathologic findings, and the intemal morphology
`on imaging studies. Tumors were classified as solid homogeneous
`masses when no necrosis or cystic change was seen pathologically and
`when a uniform solid tumor was identified by CT, sonography, or MR imag-
`ing. All the remaining tumors were classified as heterogeneous and as
`having areas of necrosis; some of these tumors were classified further as
`cystic. Tumors were classified as cystic when fluid-filled or empty cavities
`were seen pathologically and were correlated with areas of water density
`on CT scans, anechoic areas with acoustic enhancement on sonograrns.
`or areas of water signal intensity on T1 - and T2-weighted MR images. Cor-
`relation also was made between the imaging and pathologic findings for
`the clinically hyperfunctioning tumors (which were grouped as insulin pro-
`ducing and non—insulin producing) and the nonhyperfunctioning tumors.
`
`Results
`
`The sizes, imaging features, and distribution of the various
`hyperfunctioning and nonhyperfunctioning islet cell
`tumors
`are given in Table 1. Of the hyperfunctioning islet cell tumors,
`insulinomas were the most common and the smallest, with a
`mean diameter of 2.2 cm. Most insulinomas (35 of 43) were
`solid homogeneous masses (Fig. 1). All eight lesions that had
`
`TABLE 1: Imaging Features of Hyperfunctioning and Nonhyperfunctioning Islet Cell Tumors
`
`DE
`
`_aAANOOOO)-54>
`
`|\)\l
`
`C-A;\l|\)O(a>f\)CaJ®
`
`I00
`
`No. of Tumors with the Following Imaging Features“:
`HE
`
`12
`36
`
`56
`
`SH
`
`35
`15
`
`5 24 2
`
`14
`
`77
`
`Tumor
`
`No. of
`Patients
`
`No. of
`Lesions
`
`Size
`
`lnsulinoma
`Gastrinoma
`
`Glucagonoma
`Stomatostatinoma
`Vipoma
`ACTH” producing
`Non—insulin producing“
`Nonfunctioning
`Total
`
`39
`18
`
`1 1
`5
`4
`2
`40
`45
`1 24
`
`43
`18
`
`1 1
`5
`4
`2
`40
`50
`
`Z133‘
`
`G)U'IU'l4>\lO‘.i-bI'\)
`
`onI
`
`‘Size = mean diameter in centimeters, SH = solid homogeneous appearance, HE = heterogeneous appearance, C = cystic appearance, CA = calcification,
`LI = local invasion, VI = vascular invasion, DM = distant metastases.
`°ACTH = adrenocorticotropin hormone.
`‘Subtotal for functioning tumors, excluding insulinomas.
`
`Fig. 1.—22-year-old woman who had lnsulino-
`ma and whose symptoms included confusion,
`agitation, blurred vision, and hypoglycemia. in-
`sulinomas are usually small solid lesions, as in
`this case.
`A, lntraoperatlve sonogram shows solid Iso-
`echoic mass wlthln tail of pancreas (TAIL PANC);
`mass measured approximately 2 cm In diameter.
`5, contrast-enhanced CT scan shows homo-
`geneously enhanclng solid mass wlthln tail of
`pancreas.
`
`West Ward Pharm.
`Exhibit 1007
`Page 002
`
`West-Ward Pharm.
`Exhibit 1007
`Page 002
`
`

`

`AJFi:165, November 1995
`
`ISLET CELL TUMORS OF PANCREAS
`
`1177
`
`cystic components seen on gross examination and on imag-
`ing studies were larger than the mean. Calcification,
`local
`invasion, vascular invasion, and distant metastases were less
`common than with nonhyperfunctioning tumors.
`The 18 gastrinomas were larger than the insulinomas
`(mean diameter, 4.2 cm). The majority of these lesions (15 of
`18) also were solid homogeneous masses. The three lesions
`that had cystic characteristics measured 1, 6, and 3 cm. Vipo-
`
`mas were similar in size to gastrinomas (4.1 cm) and were all
`solid homogeneous masses. The other hyperfunctioning islet
`cell tumors, including glucagonomas. somatostatinomas, and
`ACTH-producing tumors, were larger (Figs. 2 and 3).
`The nonhyperfunctioning islet cell
`tumors had a mean
`diameter of 7.7 cm. Thirty-six of these tumors had heteroge-
`neous areas, and over half (n = 27) had cystic degeneration
`(Fig. 4). Only 14 of the 50 nonhyperfunctioning tumors were
`
`Fig. 2.—69-year-old woman who had glu-
`cagonoma and whose symptoms included rash
`and painful glossitls. Six-centimeter mass noted
`within tail of pancreas had crescentic area 01 cys-
`tic degeneration (arrow). Note central area at cal-
`cltlcation (arrowhead). Note also evidence ot
`vascular invasion into splenic vein, with multiple
`perisplenlc varices. Metastasis was noted within
`liver (not shown). These findings were document-
`ed at attempted surgical resection. calcification
`and vascular Invasion are common tindings with
`such non-insulin-producing tumors as well as
`with nonhypertunctioning islet cell tumors.
`
`Fig. 3.-42-year-old man who had necrotizing
`migratory erytherna caused by glucagonoma. CT
`scan shows10-cm heterogeneousmasswlth inter-
`nal tocus of cystic degeneration. Hypodense liver
`metastases also were noted. cystic degeneration
`and metastatic disease are common tlndlngs with
`nonhyperfunctioning islet cell tumors as well.
`
`Fig. 4.—45-year-old man who had nonfunc-
`tional islet cell tumor and whose symptoms in-
`cluded 8—month history of vague abdominal pain.
`Large lesions such as this one are predisposed
`to cystic degeneration and aggressive behavior
`(splenic vein Invasion was noted on this image
`as well as on other images [not shown]).
`A, sonogram shows well-circumscribed 10-
`cm mass with large anechoic central area and
`acoustic enhancement compatible with cystic
`degeneration.
`B. Contrast-enhanced CT scan shows central
`area at fluid attenuation similar to attenuation of
`gallbladder. Thick line of enhancing viable tumor
`was noted around periphery.
`Cand D, Contrast-enhanced T1 -weighted (6851/
`15 [TRll’E]) (C) and T2-weighted (6751190) (D) axial
`MR images continn central cystic degeneration.
`Decreased signal intensity on T1-weighted Image
`and markedly Increased signal intensity on T2-
`weighted Image parallel signal Intensity of CSF.
`Enhancing viable tumor was noted peripherally on
`contrast-enhanced T1 -weighted image (C).
`
`Wes -Ward Pharm.
`Exhibit 1007
`Page 003
`
`West-Ward Pharm.
`Exhibit 1007
`Page 003
`
`

`

`BUETOW ET AL.
`
`AJFi:165, November 1995
`
`Fig. 5.—60-year-old man with nonhyperfunc-
`tioning islet cell tumor. Tumor shows cystic de-
`generation, calcification, and liver metastasis.
`A, Lobulated 13-cm mass was noted within
`head of pancreas. Note central areas of cystic
`degeneration with attenuation similar to that of
`gallbladder. Small punctate area of calcification
`also was noted (arrow).
`B, Unenhanced CT scan shows large metasta-
`sis within right lobe of liver (arrows), typical fea-
`ture of nonhyperfunctioning islet cell tumors.
`
`solid homogeneous masses. Local invasion, vascular inva-
`sion, calcification, and distant metastases were more com-
`mon than with insulinomas (Fig. 5).
`Twenty-seven of the 40 tumors with cystic components
`were nonhyperfunctioning islet cell tumors. Heterogeneous
`areas corresponding to necrosis or cystic degeneration were
`noted in 56 tumors with a mean diameter of 8.4 cm. The
`
`mean diameter of the cystic tumors was 7.9 cm. Thirty-six of
`the tumors with heterogeneous areas were nonhyperfunc-
`tioning islet cell tumors. Solid homogeneous masses were
`seen in 76 cases. Only 14 of these lesions were nonhyper-
`functioning islet cell tumors. The mean diameter of the solid
`homogeneous tumors was 2.9 cm.
`Using an analysis of variance and Duncan’s multiple-
`range test, we found the difference in mean diameter
`between the two largest groups, insulinomas and nonhyper-
`functioning tumors, to be statistically significant (p = .0001).
`Using a t test, we found a statistically significant correlation
`between the size of the tumors and whether they were cystic
`(p < .0005), solid homogeneous (p = .0001), or heteroge-
`neous (p = .0001 ). Similarly, we found statistically significant
`differences in size between tumors that were cystic (mean
`diameter, 7.9 cm) or heterogeneous (mean diameter, 8.4
`cm) and solid homogeneous tumors (mean diameter, 2.9
`cm) (p = .0001 for both comparisons).
`A statistically significant correlation between the size of
`the tumors and the presence of all the characteristics that
`we analyzed also was documented: calcification (mean
`diameter, 8.2 cm) (p < .0O005), local invasion (mean diame-
`ter, 6.5 cm) (p = .00006), and vascular invasion (mean diam-
`eter, 7.2 cm) (p = .0019). The larger the tumor, the more
`likely it was to contain all of these features.
`We also found statistically significant differences in the
`internal morphology of the tumors, depending on whether
`they were hyperfunctioning or nonhyperfunctioning. Hetero-
`geneity was significantly correlated with nonhyperfunctioning
`tumors (p < .0005, x2 test).
`Liver metastases were least common with insulinomas.
`
`Liver metastases were seen with both ACTH-producing
`tumors, with six of 10 glucagonomas (Fig. 3), and with 11 of
`50 nonhyperfunctioning islet cell tumors (Fig. 5).
`
`Discussion
`
`All islet cell tumors are capable of producing polypeptides.
`Although individual tumors may elaborate more than one
`hormone to various degrees in different parts of the tumor,
`the vast majority of patients have symptoms related to the
`overproduction of a single hormone or have no hormone-
`related symptoms at all [1-5]. However, because of this vari-
`ability, islet cell tumors can be divided, on clinical grounds,
`into hyperfunctioning and nonhyperfunctioning tumors.
`Hyperfunctioning tumors may be subdivided into insulin-pro-
`ducing and non—insulin-producing types.
`Our data demonstrate that size, function, consistency, and
`malignant behavior are integrally related. The most common
`hyperfunctioning islet cell tumors, the insulinomas, were typi-
`cally small homogeneous masses measuring about 2 cm in
`diameter. Cystic changes and necrosis were uncommon fea-
`tures. On the other hand, nonhyperfunctioning islet cell
`tumors were typically larger, heterogeneous tumors with a
`propensity for cystic degeneration and central areas of necro-
`sis; these tumors measured almost 8 cm in diameter. These
`differences were statistically significant, although there was
`some overlap. The other hyperfunctioning islet cell tumors,
`including gastrinomas, glucagonomas, somatostatinomas,
`vipomas, and ACTH-producing tumors, were, as a group,
`larger than the insulinomas and shared a number of the char-
`acteristics noted for nonhyperfunctioning islet cell tumors.
`in our study, larger tumors, regardless of their hyperfunc-
`tioning or nonhyperfunctioning status, were more likely to
`be associated with cystic and necrotic areas, calcification,
`local
`invasion, vascular
`invasion, and liver or distant
`metastases. Other associations,
`in fewer patients, have
`been found between nonhyperfunctioning tumors and the
`presence of calcification [6, 7], increased size and necrosis
`[8], and metastases [9]. Seven of 10 islet cell tumors with
`calcification were shown to have malignant characteristics
`[10]. Venous invasion was shown in 10 of 76 patients with
`islet cell tumors; seven of these 10 patients had hepatic
`metastases [11]. However, not all of these findings have
`been specifically addressed in a large series and correlated
`with the size. consistency, and functional status of
`the
`
`Page 004
`
`West-Ward Pharm.
`Exhibit 1007
`Page 004
`
`

`

`AJFl:165, November 1995
`
`ISLET CELL TUMORS OF PANCREAS
`
`tumor, as we have done in this study. Larger surgical and
`pathologic series have not addressed these associations
`directly [6, 7, 12, 13].
`Direct correlation between the size of the tumor and cys-
`tic or necrotic areas has not yet been formally addressed,
`probably because the research done so far has included
`predominantly surgical series in which the external appear-
`ance was emphasized [12]. Cross-sectional imaging allows
`assessment of the internal structure as well. Before this
`
`study, cystic changes within islet cell tumors were viewed as
`rare, with only 20 cases previously being reported [14]-
`partly because the surgical community is not familiar with
`these tumors [15] and partly because the tissue examined
`by pathologists and prepared for slides is, appropriately,
`from the solid portion of
`the tumor. Also,
`these earlier
`reports concerned tumors that were predominantly cystic,
`not partially cystic.
`The cases considered in this study were collected from a
`large referral population. Thus, our findings are unavoidably
`biased in that regard. However, the proportion of hyperiunc-
`tioning islet cell tumors in our series parallels that noted in
`larger epidemiologic studies [5, 16, 17], supporting the
`veracity of the sampling even if
`it was not from a single
`patient population. An additional deficiency is that the imag-
`ing studies and protocols were acquired without the benefit
`of standardized equipment. However, our analysis of the
`gross specimens and surgical reports helped ensure that the
`radiologic interpretation of these cases was correct.
`In summary, we have demonstrated an association among
`size, consistency, behavior, and function of islet cell tumors.
`Smaller tumors tend to be homogeneous masses without
`local invasion or distant metastases and typically are insuli-
`nomas. Larger tumors more commonly demonstrate cystic
`changes, necrosis, calcification,
`local
`invasion, vascular
`invasion, and distant metastases and are nonhyperlunction-
`ing lesions or are associated with a less clinically evident
`functional syndrome than that seen with insulinomas. Knowl-
`edge of these features is important to radiologists in identify-
`ing and characterizing masses within the pancreas.
`
`ACKNOWLEDGMENTS
`
`We are indebted to all the radiologists and pathologists whose
`contributions of outstanding cases to the Armed Forces Institute of
`Pathology have made this article possible.
`
`REFERENCES
`
`1. Kloppel G, Heitz PU. Pancreatic endocrine tumors. Pathol Fles Pract
`1988;1:155-172
`2. Oertel JE, Oertel YC, Heftess CS. The pancreas. In: Sternberg SS, ed.
`Diagnostic surgical pathology. New York: Haven, 1994: 1 41 9-1 458
`. Larsson Ll, Grimelius L, Hakaanson R, et al. Mixed endocrine pancreatic
`tumors producing several peptide hormones. Am J PathoI1975;79:271—284
`. Wynick D, Williams SJ, Bloom SR. Symptomatic secondary hormone syn-
`dromes in patients with malignant pancreatic endocrine tumors. N Engl J
`Med 1988:31 9:605—607
`. Solcia E, Sessa F, Flindl G, Bonato M, Capella C. Pancreatic endocrine
`tumors: non-functioning tumors and tumors with uncommon lunction. In:
`Dayal Y, ed. Endocrine pathology of the gut and pancreas. Boca Flaton,
`FL: CBC, 1991:105—131
`. Kent RB. van Heerden JA, Weiland LH. Non-functioning islet cell tumors.
`Ann Surg1981;193:185—190
`. Eckhauser FE, Cheung PS, Vinik A, et al. Non-functioning malignant neu-
`roendocrine tumors of the pancreas. Surgery 1986:100:978—987
`. Fugazzola C, Procacci C, Andreis lAB, et al. The contribution of u|trasonog-
`raphy and computed tomography in the diagnosis of non-functioning islet
`cell tumors of the pancreas. GastrointestRadiol1990;14:139—144
`. Eelkema EA, Stephens DH, Ward EM, Sheedy PF ll. CT features of non-
`functioning islet cell tumors. AJFl1984:943—948
`lmhol M, Frank P. Pancreatic calciiications in malignant islet cell tumors.
`Radiology 1 977; 1 22:333—337
`. Bok EJ, Cho KJ, Williams DM, et al. Venous involvement in islet cell
`tumors of the pancreas. AJFt1984;142:319—322
`. Thompson GB, van Heerden JA, Grant CS, et al. islet cell tumors of the
`pancreas: a twenty year experience. Surgery1988;104:1011—1117
`. Broughan TA, Leslie JD, Soto JM, et al. Pancreatic islet cell tumors. Sur-
`gery 1986;99:671—677
`. Takeshita K, Furui S. Makita K. et al. Cystic islet cell tumors: radiologic
`findings in the three cases. Abdom lmaging1994;19:225—228
`. Davtyan H, Nieberg R, Reber HA. Pancreatic cystic endocrine neo-
`plasms. Pancreas 1990;5:230—233
`. Watson FlGP, Johnston CF, O'Hara NMT, at al. The frequency ol gas-
`trointestinal endocrine tumors in a well-defined population——Northern Ire-
`land 1970-1985. 0 J Med 1989;72:647—657
`. Service FJ, McMahon MM. O'Brien PC, Ballard DJ. Functioning insuli-
`noma—incidence, recurrence, and long term survival of patients: a 60
`year study. Mayo Clin Proc 1991 ;66:711—719
`
`.
`
`-Ward Pharm.
`Exhibit 1007
`Page 005
`
`West-Ward Pharm.
`Exhibit 1007
`Page 005
`
`