Assessment of Tumor Growth in PancreaticNeuroendocrine Tumors in von Hippel LindauSyndrome

Allison B Weisbrod, MD, Mio Kitano, MD, Francine Thomas, PhD, David Williams, PhD,Neelam Gulati, BS, Krisana Gesuwan, MS, Yixun Liu, PhD, David Venzon, MS, Ismail Turkbey, PhD,Peter Choyke, MD, Jack Yao, PhD, Steven K Libutti, MD, FACS, Naris Nilubol, MD, FACS,William M Linehan, MD, Electron Kebebew, MD, FACS

BACKGROUND: The incidence of pancreatic neuroendocrine tumors (PNETs) is increasing, but only a subsetof these heterogeneous tumors will progress to malignant disease, which is associated with apoor prognosis. Currently, there are limited data on the natural history of these tumors and itis difficult to determine which patients require surgical intervention because the risk ofmetastatic disease cannot be accurately determined.

STUDY DESIGN: We conducted a prospective study of 87 patients with von Hippel Lindau syndrome-associated solid pancreatic lesions to determine the natural history of these tumors withbiochemical testing, follow-up anatomic and functional imaging, and advanced imaginganalysis, with a median follow-up of 4 years.

RESULTS: Approximately 20% of consecutive tumor measurements during follow-up were decreased insize and 20% showed no change. This included 2 of 4 surgically proven malignant tumors,which had a net decrease in tumor size over time. Tumor volume, as derived from greatestdiameter and volumetric measurements, showed good correlation to pathology tumor mea-surement of surgically resected tumors (Spearman rank correlation r ¼ 0.72, p ¼ 0.0011,and r ¼ 0.83, p < 0.0001, respectively). Tumor density measurement had an inverserelationship with tumor size (Spearman rank correlation �0.22, p ¼ 0.0047). A tumordensity cutoff of 200 was 75% specific for malignant tumors.

CONCLUSIONS: Pancreatic neuroendocrine tumors demonstrate a nonlinear growth pattern, which includesperiods of no growth and apparent decrease in size by imaging. These growth patterns arevariable and are not associated with tumor grade and malignancy. Tumor density, asmeasured in this cohort, may offer a specific diagnostic tool for malignant disease. (J AmColl Surg 2014;218:163e169. � 2014 by the American College of Surgeons)

Disclosure Information: Nothing to disclose.

Funding: This research was supported by the intramural research programof the Center for Cancer Research, National Cancer Institute, NationalInstitutes of Health.

Received September 20, 2013; Revised October 30, 2013; AcceptedOctober 31, 2013.From the Endocrine Oncology Branch (Weisbrod, Kitano, Gulati, Gesu-wan, Nilubol, Kebebew) and Urologic Oncology Branch (Linehan), Centerfor Cancer Research; the Diagnostic Radiology Department, Clinical Cen-ter (Thomas, Williams, Liu, Turkbey, Choyke, Yao), and the Biostatisticsand Data Management Section, Office of the Clinical Director (Venzon),National Cancer Institute, National Institutes of Health, Bethesda, MD;and the Department of Surgery, Montefiore Medical Center and the AlbertEinstein College of Medicine, Bronx, NY (Libutti).Correspondence address: Electron Kebebew, MD, FACS, EndocrineOncology Branch, National Cancer Institute, Bethesda, MD 20892-1201.email: kebebewe@mail.nih.gov

163ª 2014 by the American College of Surgeons

Published by Elsevier Inc.

Pancreatic neuroendocrine tumors (PNETs) are a hetero-geneous group of rare tumors with a low but increasingincidence, accounting for 1.3% of all pancreatic tu-mors.1-5 They may be functional, manifesting in one ofseveral characteristic clinical syndromes, or nonfunc-tional.2-5 They may be sporadic or can occur within thecontext of multiple endocrine neoplasia type 1, von Hip-pel Lindau (VHL) syndrome, neurofibromatosis type I,or carcinoid syndrome.2-5 Pancreatic neuroendocrinetumors display a variety of histologic characteristics, pre-senting as a spectrum of well-differentiated to poorlydifferentiated neuroendocrine cells, with only a subsetprogressing to malignant disease.2-5

There are limited prospective data on the optimal man-agement of localized PNETs. Benign tumors, excludinginsulinomas, do not threaten patient survival and may

ISSN 1072-7515/13/$36.00

http://dx.doi.org/10.1016/j.jamcollsurg.2013.10.025

Abbreviations and Acronyms

CgA ¼ chromogranin APNET ¼ pancreatic neuroendocrine tumorPPP ¼ pancreatic polypeptideVHL ¼ von Hippel Lindau syndrome

164 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome J Am Coll Surg

not require surgical resection if symptoms associated withhormonal hypersecretion are medically controlled.6,7

Malignant tumors, on the other hand, exhibit a poorprognosis.6-8 Survival rates average 1 to 3 years once me-tastases are identified.1 Early surgical intervention offersthe best potential for curative therapy or the preventionof metastatic disease but is not without morbidity.6,7,9

Unfortunately, it is difficult to distinguish benigntumors from those that are potentially malignantwithout histopathologic analysis, with a significant num-ber of tumors still of undetermined malignant potentialbased on the World Health Organization (WHO) clas-sification.10 Chromogranin A (CgA) and pancreaticpolypeptide (PPP) are often used as serum tumormarkers for PNETs once disease has been surgicallyconfirmed; however, they provide little preoperativeprognostic information.5,8 Radiographic evidence ofadvanced disease, such as local tumor invasion or evi-dence of metastasis, are currently the only reliableclinical predictors of malignancy. However, these find-ings provide little benefit to the patient because curativeresection may not be possible at that point. Previousstudies have suggested tumor size and/or growth rateas useful indicators of malignancy.11 However, recentstudies have shown that even small tumors (�2 cm)may have aggressive behavior.5,12,13

Patients with VHL syndrome have an approximate20% risk of developing 1 or more PNETs over their life-time. As such, they undergo routine lifetime screeningand surveillance to allow for early intervention forPNETs, which have malignant potential.14 Computedtomography is the most commonly used imagingmodality for screening and surveillance, due to its highsensitivity (about 94%) for identifying PNETs.15-18 Fortu-nately, PNETs are easily identifiable, with characteristicradiographic features, appearing as an early enhancingwell-circumscribed solid mass with rounded or lobulatedborders and a rich vascular supply, and characteristichyperintensity on the arterial phase of CT scan.17,19 Theaim of this study was to review a prospectively maintaineddatabase and identify PNETs that develop in patientswithin the context of VHL and follow their naturalcourse of disease progression in hopes of identifying ameans to distinguish patients who are more appropriatefor active surveillance compared with those who may

benefit from early surgical intervention for tumors withhigh malignant potential.

METHODSAs part of a prospective clinical protocol approved by theNational Cancer Institute’s Institutional Review Board,134 patients with pancreatic manifestations of VHLwere enrolled and underwent comprehensive biochemicaltesting and anatomic and functional imaging studiesannually at the National Institutes of Health (NIH) Clin-ical Center.

Radiographic imaging and volume and densityassessment

As part of our clinical research protocol, we performed aroutine pancreatic protocol (2-mm slices) abdominal CTand magnetic resonance imaging, both with contrast(both protocols inject contrast at 3mL/second,with arterialphase typically 10 to 20 seconds after injection). These areperformed annually in patients with solid pancreatic lesionsand every 2 years in those with complex or cystic pancreaticlesions. These defined intervals comprised the majority ofall collected data points, and the measurements from CTscan were used to determine growth rate. If patientspresented with concerning symptoms warranting furtherCT imaging between specified surveillance intervals, thesestudies were also included at 4- to 6-month intervals. Fourof the 134 patients enrolled in the study had their CT scansrepeated within 6 months.Each imaging study was assessed by at least 2 indepen-

dent reviewers (3 independent reviewers for the first 107patients enrolled), who catalogued all solid radiographiclesions of the pancreas that displayed hyperintense signalduring early arterial phase (Fig. 1). Size and location wererecorded for each lesion. Key images were stored for eachtumor to provide a reference baseline image for consistentanalysis. For each lesion, the largest diameter of each tumorwas recorded by each reviewer and then averaged to providea consensus measurement. Volume measurements werederived using the equation: volume ¼ diameter3.In addition, a software-based measurement for volume

and density was calculated for each lesion identified onCT scan (Fig. 2). Using the stored key images providedby our independent reviewers, each tumor was centrallylabeled within a computer workstation. Syngo.via foroncology (Siemens Health Care Corporation) was usedto collect software-calculated measurements of volumeand density for each identified pancreatic lesion.Tumor measurements were used for comparison

against different laboratory and radiographic data ob-tained within a 6-month time frame. Tumor greatest

Figure 1. Solid pancreatic lesion on CT scan. A solid radiographiclesion is identified by independent reviewers, whose averagedmeasurements are the basis for 2-dimensional volume estimations.

Vol. 218, No. 2, February 2014 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome 165

diameter was compared with serum chromogranin A(CgA) and pancreatic polypeptide (PP) values.

Patient population

Of 134 patients in our study, 30 were excluded from thesample population after all 3 reviewers independentlyfound no evidence of a solid pancreatic tumorda lesionthat in this population is most consistent with a neuro-endocrine tumor. An additional 17 patients wereexcluded for having only a single imaging record, there-fore not providing enough data to measure tumorgrowth. This resulted in a cohort of 87 patients.Twenty-one patients in our cohort had a total of 26lesions surgically resected, a subset of the populationfor which histopathology data was available and whichwere classified according to the WHO tumor grade clas-sification system. Malignancy was defined as the pres-ence of metastasis to lymph nodes or distant sites(liver) proven on histologic examination. The WHOgrading system was used in all the primary tumors

Figure 2. Software-derived tumor volumes. Eacdelineates tumor margins from surrounding densi

removed. Patients followed within our study cohortreceived only surgical intervention over the study timeperiod for management of their disease. None weretreated with systemic therapy including octreotide.

Statistical analysis

All statistical analyses were completed through collabora-tion with biostatisticians in the Biostatistics and DataManagement Section of the National Cancer Institute.Tumor measurements were used for comparison againstdifferent laboratory and radiographic data obtainedwithin a 6-month interval. For descriptive information,data were calculated as means with standard deviations.Because standard deviation is expected to increase as themean increases, a coefficient of variation was calculatedas the standard deviation divided by the mean, andthen expressed as a percentage. Tumor sizes (linear andvolume measurements) were compared using theSpearman rank correlation test. The Spearman rank cor-relation test was also used for growth rate comparisonsand for associations with pathology tumor measurements.Repeated measures analysis of variance was used to adjustfor correlation between the multiple tumors of some pa-tients in tests of serum CgA, PPP, and 18F-FDOPA PETSUVs. For analyses of tumor growth measurements,tumor sizes were log-transformed before analysis. Growthdistributions at 3-month intervals were estimated bylinear interpolation of measurements over observed inter-vals. The central tendency and the variation in growthrates across patients were tested using Kruskal-Wallisand Ansari-Bradley tests. The Wilcoxon signed rank testwas applied to changes in tumor density.

RESULTSA total of 163 tumors were followed within our cohort of87 patients for a total of 377 growth intervals. Forty-one(47%) of the patients manifested a single tumor, with amedian tumor burden of 2 for the entire cohort.

h tumor is labeled, and then the softwareties to derive volume measurements.

Table 1. Demographic and Clinical Characteristics ofStudy Cohort

Characteristic Data

Age, y, mean � SD 47.6 � 13.7

Male:female 35:52

Race/ethnicity, n

Caucasian 79

African American 3

Asian 1

Unknown 4

Total radiographic solid enhancing lesions, n 163

Median tumor burden per patient, n 2

Surgically resected lesions, n 26

Histopathologic diagnosis, n

MCA 3

WHO Grade I 5

WHO Grade II 7

WHO Grade III 8

Unspecified PNET 3

MCA, microcystic serous cystadenoma; PNET, pancreatic neuroendocrinetumor; WHO, World Health Organization.

166 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome J Am Coll Surg

Demographic, clinical, and histopathology diagnoses forthe study cohort are summarized in Table 1. During thistime period, 29 new lesions were identified over an aggre-gate follow-up time period of 3,249 patient months, pro-ducing a new tumor incidence of 1 per 112 patientmonths.

Tumor growth and accuracy of volume calculations

Inter-reader correlation in linear tumor measurements washigh (correlation coefficients >0.86). The range of readerdeviations from the mean for each lesion was �0.7 to 0.7,the median coefficient of variation was 9.4%, and no

Figure 3. Observed changes in tumor size for 163 solLindau syndrome. The graph on the left displays growon the right displays the growth patterns for tumomicrocystic serous cystadenoma.

individual reader had deviations statistically differentfrom a zero mean. The mean over readers was thereforean appropriate summary for each measurement.Tumor growth was followed for the entire cohort

(Fig. 3). Tumor growth patterns were nonlinear. Approx-imately 20% of the differences between consecutive mea-surements were radiographic decreases in size and 20%were no change. Figure 4 shows growth for the mean, me-dian, fifth, and 95th percentiles of tumors at 3-month in-tervals. Growth rate was independent from tumor size(correlation r ¼ �0.08, p ¼ 0.11) (ie, larger lesionsdid not tend to show a more rapid growth pattern).The growth rates in patients with multiple tumorsshowed no significant clustering or differences betweenmeans (Kruskal-Wallis test, p ¼ 0.45); however, some pa-tients showed highly variable growth rates while othershad more consistent tumor growth over time (p <0.0001).The accuracy of tumor volume measurements was

assessed for both 2-dimensional imaging and volumetricsoftware calculations. The modalities showed strong cor-relation with each other in the measurements at the firsttumor scan (correlation coefficient 0.80, p < 0.0001).Additionally, both measurements showed a strong corre-lation with pathology measurements of resected tumors(volumetric software r ¼ 0.83, p < 0.0001 and2-dimensional imaging r ¼ 0.72, p ¼ 0.0011).

Tumor density

Overall, tumor density displayed a weak tendency to decreasewith time (median changes<0 in 3of 4 intervals, p¼ 0.049corrected for multiple comparisons), and was inverselyrelated to baseline tumor volume (correlation �0.22,

id radiographic lesions in patients with von Hippelth patterns for all tumors in our cohort. The graphrs with known histopathologic diagnoses. MCA,

Figure 4. The mean, median, fifth, and 95th percentiles of theobserved growth of solid radiographic pancreatic lesions in patientswith von Hippel Lindau syndrome are depicted over time.

Vol. 218, No. 2, February 2014 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome 167

p ¼ 0.0047) (Fig. 5). There was a trend toward decreasingmean density with increased tumor aggressiveness. Six of 7malignant tumors decreased in density over time. Three of7 malignant tumors had the greatest mean density for all tu-mors with a histopathologic diagnosis. The remaining 4 ma-lignant tumors demonstrated densities below the median ofthemean densities for all tumors evaluated by histology. Thissuggests that a high tumor density may be specific for malig-nant classification on histology. A tumor density of 200 was75% specific for malignant classification (n ¼ 16).

Clinical correlations

Growth rates were not significantly different by 18F-FDOPA uptake status or WHO tumor grade in resectedtumors. Furthermore, no threshold in growth rate wasobserved to distinguish malignant tumors (patients whopresented or developed metastasis during follow-up)

Figure 5. Relationship of tumor density and tumor volume. MCA,microcystic serous cystadenoma.

from benign and intermediate grades (localized tumors).Two lesions in the surgically proven malignant subsethad a net decrease in tumor size over time.Tumor greatest diameter was compared to serum

CgA and PP values. Serum CgA values demonstrated a sta-tistical trend toward an inverse relationship with tumor size(p ¼ 0.053, n ¼ 137); no association was seen betweentumor size and serum PP levels (p ¼ 0.81, n ¼ 165).

DISCUSSIONOur study showed that PNETs can demonstrate variablegrowth. Across all time periods in our cohort, 20% ofconsecutive measurements of tumors demonstrated nogrowth and an additional 20% were smaller by imaging.This observation was consistent across different patientsand different tumors, with evidence that individualtumors may have periods of negative growth, no growth,indolent growth, and aggressive growth over time.To our knowledge, this is the first large cohort study with

a relatively long follow-up time to examine the variablenatural history of PNETs. There have been reports of par-tial and complete spontaneous regression of malignantneuroendocrine tumors in the literature, including patientswith metastatic disease.20-26 This has also been observed inother types of tumors. Lindell and associates27 observed thegrowth characteristics by CT scan of 48 histologicallyproven lung cancers, finding periods of negative growthin 4 tumors. Korst and colleagues28 identified a negativegrowth rate in 28% of 69 patients with pulmonary nodules<3 cm, including 1 histologically proven lung cancer thatshowed negative growth over 3 consecutive intervals.Finally, Honda and colleagues29 identified negativedoubling time in 5 of 40 pulmonary adenocarcinomas.Negative growth has been published in longitudinal

studies on PNETs. Kann and coworkers30 describedthe natural progression of 84 asymptomatic PNETs<15 mm in size in 20 patients with multiple endocrineneoplasia type 1 (MEN-1). Endoscopic ultrasound wasused to measure greatest tumor diameter every 6 monthsfor a maximum of 44 months. The findings were similarto ours, identifying periods of variable growth. Thisincluded tumors with alternating periods of positiveand negative growth, as well as individual tumors thatconsistently grew smaller over time. None of the patientsin the study by Kann and colleagues30 had surgical inter-vention to provide histopathologic confirmation of indi-vidual lesions; however, subsets of the PNETs weredocumented by pathology and had excellent tumor sizeand volume correlation to the surgically resected tumor.It is uncertain what caused PNETs to decrease in size

in our study cohort. This may have been a fluctuation of

168 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome J Am Coll Surg

a constant tumor mass, with change in density causing areciprocal inverse change in tumor size (mass ¼volume � density). A weak but significant statisticalrelationship to this effect was seen in our population;however, it does not account for multiple individualtumors. It may be suggested that measurement errorplayed an additional factor, with fluctuations in sizecaused by inter-reader variations. However, there was astrong correlation in measurement between independentmeasurements in this study. Additionally, it is has beendemonstrated that tumor size changes that meet athreshold of �1.73 mm are reliably detected by CTscan images; a threshold that was surpassed by theshrinking tumors over multiple intervals.31 Therefore,even though a small factor of measurement error is likelypresent in any study, our data are suggestive of a realdecrease in tumor size.A separate radiographic variable, tumor density, had

high specificity for malignant PNETs. Within our cohort,a tumor density of 200 on arterial phase images demon-strated a specificity of 75% for malignant classification, asevidenced by the presence of metastasis on histology.Although tumor density is a standard variable of measure-ment within the CT software package, we are not awareof previous publications that described its use as a markerof malignant disease. Additional testing in a larger patientpopulation is needed to see if CT tumor density providesuseful information for surgical decision-making withadequate patient follow-up time. We currently recom-mend surgical intervention in patients with VHL whohave a localized solid enhancing pancreatic lesion >3 cm(>2 cm in the head) on CT scan or when there is suspicionof metastatic disease. So, the tumor density measurementmay reduce the need for surgical intervention in low risklesions, which accounted for 5 of the resected tumors(Table 1).An interesting result of this study was that serum CgA

showed a statistical trend toward an inverse relationshipto tumor size. Neuroendocrine cells characteristically syn-thesize and secrete CgA, making it a helpful biomarkerfor neuroendocrine tumors for follow-up.32,33 SerumCgA levels have been shown to correlate with tumorburden and are often used for patient surveillance.32,33

Our findings suggest that as the lesions in our cohortgrew, an increasing percentage of the tumor volumemay not have produced CgA. A change may occur withtumor growth that either stops production or secretionof CgA. An alternative reason may be that as the lesiongrows, there is a greater percentage of nonmetabolicallyactive tissue (ie, necrosis or fibrosis).There are several limitations to this study. First, tech-

nical error may provide inaccurate representation of

tumor size. Even though a standardized CT protocolwas used for all patients within our cohort, certain factorscannot be controlled, including exact timing of contrastadministration and identical level of CT images at eachfollow-up visit. These variations are generally understoodand accepted under standard practice but may confoundour results. Nonetheless, these variations may account fordifferences in tumor volume of 2% to 20%.34-37 An addi-tional limitation is the equation used to derive tumor vol-ume from tumor greatest diameter. Volume ¼ diameter3

may not be the most appropriate mathematical descrip-tion for each tumor, which likely takes an irregular ratherthan perfectly geometric shape. With these recognizedpitfalls, it is reassuring that both the greatest diameter-derived volumes and the volumetric software-derivedvolumes showed high correlation to 3-dimensional mea-surements documented on actual tumor sample volu-metric software (Spearman rank correlation r ¼ 0.72,p ¼ 0.0011, and r ¼ 0.83, p < 0.0001, respectively).

CONCLUSIONSIn summary, our study showed that PNETs can demon-strate variable growth on serial imaging. Tumor densitymay provide a specific indicator for malignancy in ourcohort and should be validated in additional populations.Additional study is warranted to establish the optimalmeans for measuring changes in tumor behavior thatmay necessitate surgical intervention.

Author Contributions

Study conception and design: Weisbrod, Kitano,Kebebew

Acquisition of data:Weisbrod, Kitano, Thomas,Williams,Gulati, Gesuwan, Liu, Venzon, Turkbey, Choyke, Yao,Libutti, Nilubol, Linehan, Kebebew

Analysis and interpretation of data: Weisbrod, Kitano,Thomas, Williams, Gulati, Gesuwan, Liu, Venzon,Turkbey, Choyke, Yao, Libutti, Nilubol, Linehan,Kebebew

Drafting of manuscript: Weisbrod, Kitano, Thomas,Williams, Gulati, Gesuwan, Liu, Venzon, Turkbey,Choyke, Yao, Libutti, Nilubol, Linehan, Kebebew

Critical revision: Weisbrod, Kebebew

REFERENCES

1. Yao JC, Eisner MP, Leary C, et al. Population-based study ofislet cell carcinoma. Ann Surg Oncol 2007;14:3492e3500.

2. Oberg K. Pancreatic endocrine tumors. Semin Oncol 2010;37:594e618.

3. Eriksson B, Oberg K. Neuroendocrine tumours of thepancreas. Br J Surg 2000;87:129e131.

4. Oberg K, Eriksson B. Endocrine tumours of the pancreas. BestPract Res Clin Gastroenterol 2005;19:753e781.

Vol. 218, No. 2, February 2014 Weisbrod et al Tumor Growth in von Hippel Lindau Syndrome 169

5. Kulke MH, Benson AB 3rd, Bergsland E, et al. Neuroendo-crine tumors. J Natl Compr Canc Netw 2012;10:724e764.

6. Jensen RT, Cadiot G, Brandi ML, et al. ENETS ConsensusGuidelines for the management of patients with digestiveneuroendocrine neoplasms: functional pancreatic endocrinetumor syndromes. Neuroendocrinology 2012;95:98e119.

7. Falconi M, Bartsch DK, Eriksson B, et al. ENETS ConsensusGuidelines for the management of patients with digestiveneuroendocrine neoplasms of the digestive system: well-differentiated pancreatic non-functioning tumors. Neuroendo-crinology 2012;95:120e134.

8. Halperin DM, Kulke MH. Management of pancreatic neuro-endocrine tumors. Gastroenterol Clin North Am 2012;41:119e131.

9. Grant CS. Surgical management of malignant islet cell tumors.World J Surg 1993;17:498e503.

10. Graziani R, Brandalise A, Bellotti M, et al. Imaging of neuro-endocrine gastroenteropancreatic tumours. Radiol Med 2010;115:1047e1064.

11. Blansfield JA, Choyke L, Morita SY, et al. Clinical, genetic andradiographic analysis of 108 patients with von Hippel-Lindaudisease (VHL) manifested by pancreatic neuroendocrine neo-plasms (PNETs). Surgery 2007;142:814e818; discussion818 e811e812.

12. Haynes AB, Deshpande V, Ingkakul T, et al. Implications ofincidentally discovered, nonfunctioning pancreatic endocrinetumors: short-term and long-term patient outcomes. ArchSurg 2011;146:534e538.

13. Ferrone CR, Tang LH, Tomlinson J, et al. Determining prog-nosis in patients with pancreatic endocrine neoplasms: can theWHO classification system be simplified? J Clin Oncol 2007;25:5609e5615.

14. Kim JJ, Rini BI, Hansel DE. Von Hippel Lindau syndrome.Adv Exp Med Biol 2010;685:228e249.

15. Noone TC, Hosey J, Firat Z, Semelka RC. Imaging and local-ization of islet-cell tumours of the pancreas on CT and MRI.Best Pract Res Clin Endocrinol Metab 2005;19:195e211.

16. O’GradyHL, Conlon KC. Pancreatic neuroendocrine tumours.Eur J Surg Oncol 2008;34:324e332.

17. Rockall AG, Reznek RH. Imaging of neuroendocrine tumours(CT/MR/US). Best Pract Res Clin Endocrinol Metab 2007;21:43e68.

18. Kitano M, Millo C, Rahbari R, et al. Comparison of 6-18F-fluoro-L-DOPA, 18F-2-deoxy-D-glucose, CT, and MRI inpatients with pancreatic neuroendocrine neoplasms with vonHippel-Lindau disease. Surgery 2011;150:1122e1128.

19. Rha SE, Jung SE, Lee KH, et al. CT and MR imaging findingsof endocrine tumor of the pancreas according to WHO classi-fication. Eur J Radiol 2007;62:371e377.

20. Ip YT, Pong WM, Kao SS, Chan JK. Spontaneous completeregression of gastric large-cell neuroendocrine carcinoma:mediated by cytomegalovirus-induced cross-autoimmunity?Int J Surg Pathol 2011;19:355e358.

21. Wooff JC, Trites JR, Walsh NM, Bullock MJ. Completespontaneous regression of metastatic merkel cell carcinoma: a

case report and review of the literature. Am J Dermatopathol2010;32:614e617.

22. Hassan SJ, Knox M, Griffin M, Kennedy MJ. Spontaneousregression of metastatic Merkel cell carcinoma. Ir Med J2010;103:21e22.

23. Kubo H, Matsushita S, Fukushige T, et al. Spontaneousregression of recurrent and metastatic Merkel cell carcinoma.J Dermatol 2007;34:773e777.

24. Agard C, Guerzider P, Bouillard J, et al. [A metastatic neuro-endocrine tumor with exceptional outcome!]. Rev MedInterne 1998;19:438e441.

25. Christ F, Siewert B, Koch O. [Spontaneous regression of livermetastases in neuroendocrine tumors of the abdomen]. Ultra-schall Med 1991;12:41e44.

26. Fujimaki T, Mishima K, Asai A, et al. Spontaneous regressionof a residual pineal tumor after resection of a cerebellar ver-mian germinoma. J Neurooncol 1999;41:65e70.

27. Lindell RM, Hartman TE, Swensen SJ, et al. Five-year lungcancer screening experience: CT appearance, growth rate, loca-tion, and histologic features of 61 lung cancers. Radiology2007;242:555e562.

28. Korst RJ, Lee BE, Krinsky GA, Rutledge JR. The utility ofautomated volumetric growth analysis in a dedicated pulmo-nary nodule clinic. J Thorac Cardiovasc Surg 2011;142:372e377.

29. Honda O, Johkoh T, Sekiguchi J, et al. Doubling time of lungcancer determined using three-dimensional volumetric soft-ware: comparison of squamous cell carcinoma and adenocarci-noma. Lung Cancer 2009;66:211e217.

30. Kann PH, Balakina E, Ivan D, et al. Natural course of small,asymptomatic neuroendocrine pancreatic tumours in multipleendocrine neoplasia type 1: an endoscopic ultrasound imagingstudy. Endocr Relat Cancer 2006;13:1195e1202.

31. Revel MP, Bissery A, Bienvenu M, et al. Are two-dimensionalCT measurements of small noncalcified pulmonary nodulesreliable? Radiology 2004;231:453e458.

32. Portela-Gomes GM, Grimelius L, Wilander E, Stridsberg M.Granins and granin-related peptides in neuroendocrine tu-mours. Regul Pept 2010;165:12e20.

33. Modlin IM, Gustafsson BI, Moss SF, et al. ChromograninAebiological function and clinical utility in neuro endocrinetumor disease. Ann Surg Oncol 2010;17:2427e2443.

34. Wormanns D, Kohl G, Klotz E, et al. Volumetric measure-ments of pulmonary nodules at multi-row detector CT:in vivo reproducibility. Eur Radiol 2004;14:86e92.

35. Revel MP, Lefort C, Bissery A, et al. Pulmonary nodules: pre-liminary experience with three-dimensional evaluation. Radi-ology 2004;231:459e466.

36. Ko JP, Rusinek H, Jacobs EL, et al. Small pulmonary nodules:volume measurement at chest CTephantom study. Radiology2003;228:864e870.

37. Winer-Muram HT, Jennings SG, Meyer CA, et al. Effect ofvarying CT section width on volumetric measurement oflung tumors and application of compensatory equations. Radi-ology 2003;229:184e194.