Utility of Various Functional and Anatomic Imaging Modalities for Detection of Ectopic Adrenocorticotropin-Secreting Tumors

Abstract

Context: Because ectopic ACTH-secreting (EAS) tumors are often occult, improved imaging is needed.

Objective: Our objective was to evaluate the utility of [¹¹¹In-DTPA-d-Phe]pentetreotide scintigraphy [octreotide (OCT)] imaging at 6 mCi [low OCT (LOCT)] and 18 mCi [high OCT (HOCT)], [¹⁸F]fluorodeoxyglucose (FDG)-positron emission tomography (PET) and [¹⁸F]l-3,4-dihydroxyphenylalanine (F-DOPA)-PET scans, computed tomography (CT), and magnetic resonance imaging (MRI).

Design and Setting: The study was a prospective evaluation at a clinical research center.

Patients: Forty-one subjects participated, 30 (17 female) with resected EAS tumors and 11 (three female) with occult EAS, based on inferior petrosal sinus sampling results and imaging studies.

Intervention: Intervention included CT and MRI of neck, chest, abdomen, LOCT (with or without HOCT) and FDG- or F-DOPA-PET without CT every 6–12 months.

Main Outcome Measure: Tumor identification was the main outcome measure.

Results: Most recent results were analyzed. Eighteen patients had tumor resected on the first visit; otherwise, surgery occurred 33 ± 25 (9–99) months later. Tumor size was 1.9 ± 1.7 (0.8–8.0) cm; 83% were intrathoracic. CT, MRI, LOCT, HOCT, FDG-PET, and F-DOPA-PET had sensitivities per patient of 93% [95% confidence interval (CI) = 79–98%], 90% (95% CI = 74–96%), 57% (95% CI = 39–73%), 50% (95% CI = 25–75%), 64% (95% CI = 35–85%), and 55% (95% CI = 28–79%) and positive predictive values (PPV) per lesion of 66, 74, 79, 89, 53, and 100%, respectively. LOCT and PET detected only lesions seen by CT/MRI; abnormal LOCT or F-DOPA-PET improved PPV of CT/MRI. By modality, the fraction of patients with one or more false-positive findings was 50% by CT, 31% by MRI, 18% by L/HOCT, and 18% by FDG-PET. Eight occult EAS patients had 64 ± 58 (9–198) months follow-up; others had none.

Conclusions: High sensitivity and PPV suggest thoracic CT/MRI plus LOCT scans for initial imaging, with lesion confirmation by two modalities.

The initial imaging of patients with presumed ectopic ACTH-secreting tumors should include thoracic CT and MRI followed by LOCT.

Ten percent of Cushing’s syndrome is caused by ectopic ACTH-secreting (EAS) tumors (1,2,3,4). Although first described in patients with overt and metastatic tumors, more recent series include slow growing, usually intrathoracic, occult tumors (5,6). Identification and surgical removal of the tumor remains the optimal treatment for the management of EAS (7). Thus, localizing these tumors is crucial.

Despite the use of anatomical imaging techniques like computed tomography (CT) and magnetic resonance imaging (MRI), up to 50% of EAS tumors cannot be found (8). [¹¹¹In-DTPA-d-Phe]Pentetreotide scintigraphy [octreotide (OCT)], a functional imaging modality, has shown promise in localizing these tumors (9,10,11,12,13,14,15,16). The ability of OCT to identify the tumors depends on multiple factors, including the dose of the radiopharmaceutical, the type and degree of somatostatin receptor expression, and tumor size (9,10,11). [¹⁸F]Fluorodeoxyglucose (FDG)-positron emission tomography (PET) has not successfully localized occult tumors, probably due to their low metabolic activity (12,13). [¹⁸F]l-3,4-dihydroxyphenylalanine (F-DOPA)-PET scans may help localize neuroendocrine tumors (14,15,16). The goal of this study was to examine the ability of OCT at a standard 6-mCi dose [low OCT (LOCT)] and at a high 18 mCi dose [high OCT (HOCT)], FDG- and DOPA-PET, and CT and MRI to identify EAS tumors.

Patients and Methods

Between 1999 and 2009, after providing informed consent, 63 patients referred for presumed ectopic ACTH-secreting tumors were studied prospectively at the National Institutes of Health (NIH) Clinical Center under a protocol (Clinical Trials NCT00001849) approved by the Institutional Review Board and NIH Radiation Safety Committee. Six eucortisolemic patients were excluded. The remaining 57 patients underwent evaluation for the cause of ACTH-dependent Cushing’s syndrome, which included CRH and dexamethasone suppression tests, pituitary MRI, and inferior petrosal sinus sampling (IPSS) (1).

Sixteen patients whose IPSS results suggested Cushing’s disease did not undergo further imaging studies. Each had an ACTH-positive tumor resected at transsphenoidal surgery. The remaining 41 patients underwent clinical and research imaging. CT, MRI, and LOCT imaging studies were reviewed by staff radiologists for routine clinical care; decisions regarding surgery were based on these findings and those of research radiologists in cases 5, 17, and 29, augmented by review of the investigational HOCT and PET scans. Surgery was recommended when at least one nonresearch study was clearly abnormal.

After an interim analysis showed that FDG-PET was not helpful (13), DOPA-PET was substituted. Three-Tesla (T) chest MRI imaging, gated cardiac studies, and single-photon emission CT (SPECT)/CT for octreoscans were added when they became available.

Imaging

Patients underwent CT and MRI of neck, chest, abdomen, pelvis and LOCT every 6–12 months until a tumor was identified. At intervals of no less than 12 months, HOCT was obtained if LOCT was inconclusive, and up to two PET scans were obtained.

Over the years, CT was performed with 8-, 16-, and 64-detector systems. Contrast agents were administered orally and iv. From 1999–2009, CT section thickness from the neck to the adrenal glands decreased from 5 to 1 mm; otherwise, section thickness was 10 mm. When it became available, 64-slice gated cardiac CT was performed with and without iv contrast.

T1 spin echo and T2 MRI images were obtained with and without gadolinium using a 1.5-T scanner. The 3-T chest MRI (using scan thickness of 5–8 mm depending on the patient’s size) and cardiac MRI (using a six-element cardiac phased-array receiver coil, and vector electrocardiographic gating) were performed when available if the patients could tolerate the examination.

Generally, LOCT images were obtained 4 and 24 h and HOCT images 4, 24, and 48 h after injection; rarely, further delayed images were obtained. Planar whole-body images were obtained at 4 h, and SPECT was obtained at all time points. SPECT/CT was used when it became available in 2008.

PET images were obtained after 6 h fasting with a GE Advance PET scanner and iterative reconstruction. Forty minutes after FDG (20 mCi) injection, images were acquired from upper thighs to forehead. Initially, 10-min two-dimensional emission and 8-min transmission scans were acquired. Later, the transmission time was changed to 3 min. PET was performed without CT coregistration. For DOPA-PET, five patients were pretreated with 200 mg carbidopa. Images were acquired from upper thighs to mid-forehead 30 min after injection (12 mCi F-DOPA) and from mid-abdomen to the thorax 105 min after injection. The emission time was 5 min with 3-min transmission scans.

Data collection and statistical analysis

Clinical nuclear medicine and radiology staff were not blinded to other available imaging results; they provided prospective readings. MRI and CT were prospectively and retrospectively reviewed by research staff, who considered all imaging results. LOCT, HOCT, and PET scans were read prospectively in a blinded fashion. All prospective data were considered for patient care purposes.

All retrospective reviews were done without knowledge of the pathology findings, except for patient 26. In Table 1, we note whether the retrospectively reviewed results were positive and the original clinical report negative [cases 5 (CT), 13 (F-DOPA), 14 (MRI and LOCT), 26 (CT), and 27 (LOCT)].

Table 1.

Imaging studies and surgical pathology results of the patients with Cushing’s syndrome due to ectopic ACTH secretion

Case no.	Age (yr)a	Sex	CT	MRI	PET	LOCT	HOCT	Diagnosis	Maximum tumor size (cm)	Tumor ACTH stain	Disease stage
1b	61	Female	Widely metastatic lesions: chest, abdomen	NDg	FDG: L supraclavicular; LU abdomen; DOPA: ND	L supraclavicular	ND	Diffuse SCLC	NA, multiple LyN	NEG (biopsy was stained)	Meta
2	32	Male	RLL nodule, R subcarinal LyNc,d (R pleural nodule, R peribronchial LyN)f	RLL nodulec,h	FDG: ND; DOPA: RLL	RLL (LLL)	ND	RLL NET	1.6	POS	LyN
3b	42	Female	RML nodule; R paratracheal LyN	RML nodule	FDG: Focus RML; DOPA: ND	RML	ND	RUL NET	1.3	NA	LyN
4b	53	Female	Liver lesions, mesenteric LyNc (R pericardial nodule)	Liver lesions, mesenteric LyNsc	FDG: FN; DOPA: ND	Mid abdomen: 2 foci, liver: 2 foci	ND	Primary ileal NET	4	POSj	Meta
5b	57	Male	L neck lesion seen on chest CTk	L neck lesion	FDG: L neck focus; DOPA: ND	L neck (bilateral apices lung uptake due to pulmonary fibrosis)	ND	Rec neck NET	4	POS	LyN
6	27	Male	Thymic lesion	Thymic lesionh (paratracheal LyN)	FDG: ND; DOPA: Med	Med	ND	Thymic carcinoma	8	POS	Localized
7b	36	Male	Mass in Ant cranial fossa (multiple lung nodulesl)	Large mass anterior cranial fossa (multiple lung nodulesl)	FDG: Ant cranial fossa (multiple chest focic,l); DOPA: ND	Ant cranial fossa	ND	Olfact	5.8m	POS	Localized
8b	42	Male	Nodule beneath carina	FN	FDG: FN; DOPA: ND	Upper Med	ND	Mediastinal LyN (Rec PulmC)	2.2	POS	Localized
9	23	Male	RML noduled	RML noduleh (liver nodule)	FDG: ND; DOPA: R lung	RLL (LL)	ND	R PulmC	1	POS	Localized
10	56	Female	RUL noduled, 3 hilar LyNn (R thyroid nodule)	RUL nodule, hilar LyNi,n	FDG: R lung DOPA: ND	FN	R lung, R hilar	RUL NET	0.8	POS	LyN
11b	22	Female	Adrenal mass (RML, RLL)	Adrenal mass seen on chest MRI	FDG: R adrenal (bilateral hilar, liver focus) DOPA: ND	FN	FN	R Pheoo	2.0	NEG	Localized
12b	39	Male	Nodules: LUL, Medc (LyN: cervical, mediastinal)f	L apex nodule (cervical LyN)	FDG: Mid L Medc; DOPA: ND	2 foci mid L Med	ND	L PulmC	1.6	NA	LyN
13	41	Female	Nodule behind L ventriclec,d	Nodule behind L ventriclec,h	FDG: ND DOPA: Posterior to L ventriclek	FN	FN	LLL NET	1.6	POS	Localized
			(LyN: peribronchial, mediastinal)f	(LyN, mediastinal)
(Continued)

Table 1A.

Continued

Case no.	Age (yr)a	Sex	CT	MRI	PET	LOCT	HOCT	Diagnosis	Maximum tumor size (cm)	Tumor ACTH stain	Disease stage
14b	37	Male	RLL nodulec,d,e (L pleural nodule)	RLL nodulec,k (L pleural nodule)	FDG: FN; DOPA: ND	RLLk	RLL	R PulmC	0.8	POS	LyN
15	22	Female	LLL nodulec,d,p (R pulm nodule)	LLL nodulec,h,p	FDG: ND; DOPA: FN	FN	FN	LLL NET	0.8	POS	LyN, tumor thrombus in blood vessel
16	50	Male	Med nodulec,d,q (paratracheal LyN)f	Med nodulec	FDG: ND; DOPA: FNr	FN	Ant Med	Thymic NET	2	POS	Localized
17	55	Female	Retrocardiac noduled,p,q (paraesophageal LyN)	Retrocardiac noduleq	FDG: ND; DOPA: FNr	FNs	FNs	LLL NET	1	POS	Localized
18	55	Female	RUL noduled,q	RUL noduleh,q	FDG: ND; DOPA: RUL noduler	FNs	RUL nodules,t	RUL NET, multifocal	0.9	POS	Localized
19b	41	Female	Med LyNsd,p	Med LyNsp	FDG: ND; DOPA: ND	R Ant; Med, R hilar	ND	Med LyN (Rec PulmC)	NA, multiple LyN	POS	LyN, Rec disease
20	36	Female	RML nodule	FN (R Pleural nodule)	FDG: ND; DOPA: ND	FN (Bil post medial lung uptake)	ND	RML NET	0.8	POS	Localized
21	48	Female	RML nodulec,d,q	RML nodulec,h,q	FDG: ND; DOPA: ND	FNs	FNs (R posterior lung focus)	RML NET	0.65	POS	Localized
22	68	Female	LLL noduled (LUL, R lung nodules)	LLL noduleh (subcarinal LyN)	FDG: ND; DOPA: ND	LLL	ND	PulmC	0.8	POS	Localized
23	30	Female	2 Ant Med lesionsc,d (thyroid nodule calcified)	3 Ant Med lesionsc,h (thyroid nodule; adnexal cyst)	FDG: ND; DOPA:FN	Med focus	ND	Atypical thymic carcinoma, multifocal	2.5	POS	Lymphatic invasion (pathological diagnosis)
24b	44	Male	Med nodule	Med nodule	FDG: ND DOPA:ND	ND	ND	Thymic NET	0.8	POS	Localized
25b	62	Female	FNc	Lesion in tail of pancreas	FDG: ND; DOPA: ND	FN	ND	Pancreatic NET	1.5	POS	Localized
26b	55	Male	L lung nodulec,k (mediastinal LyN)f	FN	FDG: ND; DOPA:ND	ND	ND	Multifocal tumorlets parenchymal L lung NET	0.8	POS	Localized
27	26	Female	FNd	Med LyN, 1.5-T MRI FN	FDG: FNc; DOPA: ND	Med focusk	Med focus	Med LyN Rec PulmC	NA, multiple LyN	POS	LyN
(Continued)

Table 1B.

Continued

Case no.	Age (yr)a	Sex	CT	MRI	PET	LOCT	HOCT	Diagnosis	Maximum tumor size (cm)	Tumor ACTH stain	Disease stage
28	60	Female	Retrocardiac noduled,q	Retrocardiac noduleh,q	FDG: ND; DOPA:FN	FN	ND	LLL NET	3	NEG	Localized
29	58	Male	3 R lung nodulesd, Med LyNd,p	3 RLL nodulesh,q, Med LyNh,p,q	FDG: ND; DOPA: 3 foci in RLLm	2 foci in RLLn	2 foci in RLLn	RLL NET mulitfocal	1	POS	LyN hilar, subcarinal
30	62	Male	Retrocardiac noduled (3 R lung nodules)	Retrocardiac nodule	FDG: ND; DOPA: ND	FNn	FNn	RML NET	1	POS	Localized

False-positive results are presented in parentheses and italics. For disease stage, localized indicates no local or distant metastasis. Ant, Anterior; FN, false negative; L, left; LL, left lower; LU, left upper; LUL, left upper lobe; LyN, infiltrated lymph nodes; Med, mediastinum; Meta, distant metastases; NA, not available/not applicable; ND, not done 6 months before surgery; NEG, negative; NET, neuroendocrine tumor; POS: positive; Olfact, olfactory eshtesioneuroblastoma; Pheo, pheochromocytoma; Post, posterior; PulmC, pulmonary carcinoid; R: right; RLL, right lower lobe; RML, right middle lobe; Rec, recurrence; RUL, right upper lobe; SCLC, small-cell lung cancer; ThyC, thymic carcinoid. 

^aAges of the patients were calculated based on the date of the most recent imaging. 

^bReported previously by Pacak et al. (13). 

^cHyperplastic adrenals. 

^d,e For CT imaging: 

^dCT was done by virtual bronchoscopy or high-resolution technique; 

^econventional 5-mm CT is negative. 

^fLymph nodes are negative on pathological report. 

^g–i For chest MRI imaging, unless otherwise noted, 1.5-T MRI was obtained for all studies; 

^gMRI exam was not performed because the patient had an intracranial clip after operation for a cerebral aneurysm; 

^hseen on both 1.5- and 3-T MRI; 

ⁱonly 3-T MRI obtained. 

^jNo ACTH staining done on excised tumor, but stain from biopsy of a metastatic hepatic lesion was positive. 

^kNot reported by clinical radiologist in the initial read but found positive on retrospective review by research radiologist. 

^lPulmonary infection (nocardiosis). 

^mSize was obtained from MRI. 

ⁿTwo lymph nodes positive for histoplasma. 

^oThe patient with pheochromocytoma had positive [¹³¹I]MIBG scintigraphy. 

^pNot reported by clinical radiologist but found positive on prospective review by research radiologist. 

^qCardiac imaging (CT or MR) was performed. 

^rF-DOPA performed with oral preadministration of 200 mg carbidopa. 

^sSPECT/CT obtained. 

^tSeen at 48 h only. 

Patients were divided into tumor-found and occult disease groups. In tumor-found patients, imaging findings for individual lesions were classified as true or false positive according to the actual tumor location. In occult patients, all imaging findings were classified as questionable lesions because they were not confirmed pathologically. Findings with nontumor explanations (e.g. fractures, calcified pulmonary nodules, liver cysts, hemangiomas, adrenal hyperplasia, and vascular ectasia or anomalies) were excluded from analysis.

The largest tumor diameter measured at pathology was recorded, except for case 7, where the MRI measurement was used. Tumor size was not available for patients with metastatic disease and/or multiple enlarged lymph nodes.

Data are presented as mean ± sd. Logarithmic transformation was applied for nonnormally distributed data before analysis. An unpaired t test compared mean tumor sizes. P values < 0.05 were considered significant. The sensitivity and positive predictive value (PPV) with 95% confidential intervals (CI) for each imaging modality or combination was calculated for individual patients (not lesions) (17). In the tumor-found group, we analyzed only the studies performed during the last, presurgical, evaluation. In the occult disease group, studies performed at the most recent follow-up visit are shown. The duration of follow-up was calculated from the dates between the initial and the most recent visits, except in one patient who was evaluated initially with CT and MRI in 1990 under another protocol.

Results

Thirty patients (17 female) aged 45 ± 13.6 (23–69) yr had EAS based on biopsy or removal of a neuroendocrine tumor (Table 1). Twenty-five tumors stained positively for ACTH. The others stained positively for neuroendocrine tumor markers; ACTH immunohistochemistry was negative in three patients or not done in two patients. Four of these five patients had normal or low ACTH and cortisol values after tumor removal; one patient with negative ACTH immunochemistry had inoperable disease.

Eleven patients (three female) aged 54.1 ± 14.4 (33–82) yr had presumed occult EAS, based on negative IPSS and imaging results (18) (Table 2). Sixteen patients, including three patients with occult EAS, were reported previously (13).

Table 2.

Imaging studies of the patients with OCC-EAS

Case no.	Agea (yr)	Sex	CT	MRI	FDG-PET	F-DOPA-PET	LOCT	HOCT	Length of follow-up (months)	Comments
31b	61	Female	L Retrocardiac lesionc	L Retrocardiac lesiond	ND	ND	NEG	NEG	48	Surgery not done because of eucortisolism
32b	45	Male	NEGc,e	NEGe	ND	ND	NEG	ND	61	Failed surgery for a FP RLL lesion identified by CT and MRI 4 yr before last evaluation
33	48	Male	Paraesophageal lesion,c,e retrocardiac nodule	Paraesophageal lesione	ND	ND	NEGf	NEGf	56
34	52	Male	LLL nodulec	LLL nodulee	ND	ND	NEGf	NEGf	67	FP LLL nodule surgically removed; pathological diagnosis is scar tissue
35	57	Male	NEGc	NEGe	ND	ND	NEGf	NEGf	58	FP CT-guided needle biopsy of subpectoral lesion identified by CT and MRI; pathological diagnosis is reactive LyN
36	82	Male	NEGc	NEG	ND	Tail of pancreas	LLL	LLL	198	Patient deceased; failed exploratory mediastinal, lung, pancreatic, and bowel exploration 16 yr before his final visit
37b	38	Male	R lung nodule	NEG	NEG	ND	NEG	NEG	16	Patient declines follow-up
38	69	Male	Three rib lesions	NEG	NEG	ND	NEG	NEG	0	Patient declines follow-up
39	64	Female	R lung nodulec	NEG	ND	NEG	NEG	LML	0	Patient deceased
40	46	Female	LLL nodule	LLL nodule	Uptake above urinary bladder	ND	NEG	ND	0	Patient deceased
41	33	Male	RML nodulec	RML nodulee	ND	ND	NEGf	ND	9

For chest MRI imaging, unless otherwise noted, 1.5-T MRI was obtained for all studies. FP, False positive; L, left; LLL, left lower lobe; LU, left upper; LUL, left upper lobe; LyN, lymph node; NA, not available/not applicable; ND, not done during the last patient evaluation; NEG, negative; POS, positive; Post, posterior; R, right; RLL, right lower lobe. 

^aAges of the patients were calculated based on the date of the most recent imaging. 

^bReported previously by Pacak et al. (13). 

^cCT was done by virtual bronchoscopy or high-resolution technique. 

^dSeen on both 1.5- and 3-T MRI. 

^eCardiac imaging (CT or MR) was performed. 

^fSPECT/CT obtained. 

Tumor-found group

Patients with initial positive imaging

Seventeen of 18 patients (Table 1, patients 2–18) underwent surgery during the initial visit with resection of tumors of 2.3 ± 2.0 (median 1.6, range 0.8–8) cm. Patient 1 had inoperable metastases.

Between 1 and 5 months before evaluation at NIH, eight patients had imaging of the actual tumor location. CT scan in two of eight patients (no. 14 and 15) did not report any lesion. In five others (no. 1, 4, 11, 12, and 16), lesions on CT were identified as nonspecific nodules or abnormalities rather than possible tumor. One patient (no. 7) had a pituitary MRI that did not report an adjacent neuroethesioblastoma.

At the NIH, CT identified at least one site of disease in all 18 patients. All except case 14 received contrast. Thoracic slice thickness of 5 mm (cases 3, 6, 8, and 12) detected tumors with a mean diameter of 3.3 ± 3.2 cm (median 1.9, range 1.3–8); slice thickness of 1.2 mm (cases 2, 9, 10, and 13–18) identified tumors with a significantly lower mean tumor diameter of 1.2 ± 0.4 (median 1.0, range 0.8–2) cm (P = 0.03).

MRI localized tumors in 16 of 17 patients; the size was 2.3 ± 2.1 (median 1.6, range 0.8–8) cm. All were bright on T2-weighted images. Only seven patients, including the one with the false-negative scan, received gadolinium. When patients with intrathoracic lesions had 1.5-T and cardiac or 3-T MRI of the chest (n = 2 and 6), both studies identified the same abnormalities.

A research radiologist identified lesions that were not reported by clinical staff radiologists on CT in cases 15 and 17 prospectively and in case 5 retrospectively. Research MRI review revealed tumor not reported by clinical staff in patient 15 prospectively and in patient 14 retrospectively.

LOCT localized tumor in 11 of 18 patients, including one case (no.14) that was read retrospectively as borderline positive after HOCT was clearly abnormal. The mean diameter of tumors identified by LOCT was 3.0 ± 2.4 (median 1.9, range 0.8–8) cm. HOCT identified tumor in four of eight patients with negative or equivocal LOCT. The diameter (1.1 ± 0.6; median 0.9, range 0.8–2 cm) was not significantly different from that detected by LOCT (P = 0.43). All 14 patients identified by LOCT/HOCT had corresponding lesions on CT, and 12 of 14 had corresponding lesions identified on MRI (Table 1).

All 18 patients in this group underwent either FDG- or F-DOPA-PET scans; tumors were identified in seven of 10 FDG-PET and five of eight F-DOPA scans. Lesions detected by FDG-PET had a mean diameter of 2.6 ± 1.9 (median 1.8, range 0.8–5.8) cm, and those detected by F-DOPA had a mean diameter of 2.6 ± 3.4 (median 1.6, range 0.9–8) cm. MRI and/or CT imaging also identified these lesions.

Tumor-found group with positive imaging on follow-up visit

After initially inconclusive imaging at NIH, 12 patients (Table 1, patients 19–30) had positive follow-up imaging and successful surgery a median of 28 (range 9–99) months after initial imaging. Two patients had mediastinal recurrence of pulmonary carcinoid. The mean tumor diameter in the other 10 was 1.3 ± 0.9 (median 0.8, range 0.65–3) cm. Although this was smaller than that of the tumors found on initial workup, the difference was not statistically significant (P = 0.08).

Conventional CT identified thoracic tumors in 10 of 11, using 1.2-mm (seven of eight tumors detected, diameter 0.65–3 cm) or 5 mm (three of three tumors detected, all 0.8 cm diameter) sections (Fig. 1). Patient 30 did not receive contrast. Cardiac CT in two patients confirmed the lesions seen on conventional scans. In case 26, mediastinal exploration for evaluation of lymphadenopathy seen on CT unexpectedly revealed a tumor at the inferior pulmonary ligament, whereas the lymph nodes seen on CT proved pathologically negative. In retrospect, the tumor was present on the CT scan.

Figure 1.

A, CT with 5-mm thoracic sections of patient 24, diagnosed with thymic neuroendocrine tumor, showing the lesion in the mediastinum (arrow); B, CT with 1.2-mm thoracic sections of patient 28, diagnosed with pulmonary carcinoid, showing the retrocardiac nodule (arrow).

When follow-up scans used thinner thoracic sections compared with earlier studies (cases 19, 22, 24, and 28), retrospective unblinded review of the images identified all tumors on the earlier images regardless of slice thickness. One of these had been reported as suspicious, whereas the other lesions had originally been described as nonspecific findings.

MRI identified tumor in 10 of 12 patients. Seven received gadolinium, including one with negative results. Intrathoracic tumor was detected in five of six who had 1.5- and 3-T MRI and in three who also had cardiac MRI. One patient had tumor identified by 3-T but not 1.5-T MRI.

Staff radiologists did not report culprit lesions on CT/MRI in three patients (no.19, 26, and 29). The research radiologist identified lesions on CT retrospectively in patient 26 and prospectively by MRI and CT in patients 19 and 29.

LOCT localized tumor in five of 10 patients, including one seen retrospectively after a positive high-dose scan in patient 27. HOCT was negative in two patients and positive in patients 27 and 29 (in whom HOCT was performed to evaluate lymph nodes).

DOPA-PET accurately localized tumor in one of three patients on follow-up before surgery. FDG-PET was performed in one patient and was negative.

False-positive studies

In the patients with surgically resected lesions, false-positive CT findings (Table 1, parentheses) were found in 15 of 30 patients (50%). In the thorax, these lesions were found at a similar frequency with 1- to 1.2-mm (eight of 17) and 5-mm (five of 11) sections. False-positive lesions included thyroid, pulmonary, or pleural nodules; enlarged lymph nodes; and pelvic and liver lesions. Lymph nodes were considered to be false positive if they were pathologically negative or not found in the region of the primary tumor.

With MRI, false positives were more common with 1.5-T (6 of 27 patients) than with the 3-T magnet (one of 13 patients); these were the same types of lesions seen on CT.

False-positive pulmonary results were found in four of 28 patients with LOCT and one of 12 with HOCT. Two of eleven patients undergoing FDG-PET had multiple foci of false-positive uptake in the mediastinum, lungs, hilum, and liver. No false-positive findings were noted with F-DOPA-PET.

Occult EAS patients (Table 2)

Eleven patients with EAS remained occult at last evaluation (Table 2). Of these, three (cases 38–40) had only one evaluation at NIH; two patients died within 6 months after discharge, and one was unwilling to return. One patient who died before readmission for HOCT had a 0.8-cm lung lesion on CT and MRI but not LOCT. The others had discordant or negative findings.

The remaining patients (cases 31–37 and 41) were followed for 9–198 months. Three (cases 32, 34, and 35) had failed surgery/negative biopsies during this follow-up period for concordant false-positive findings on CT and MRI (Table 2). Another patient (36) had failed exploratory surgery and subsequent inconclusive imaging for 16 yr. At their last visit, all eight patients had CT, MRI, and LOCT, and six of eight had HOCT. Six of eight had either FDG-PET or F-DOPA-PET imaging.

At last evaluation, CT and MRI identified subcentimetric concordant lesions in four of eight patients (31, 33, 34, and 41); OCT was negative. Patient 31 has an 8-mm retrocardiac mass but due to cyclic hypercortisolism and low ACTH levels has not undergone surgery. Case 33 has two small lesions on CT; one with poorly defined borders was also seen on MRI. Case 41 has atypical 8-mm elongated inhomogeneous lesion. In patient 34, a 7-mm lung nodule resected incidental to cardiac surgery was a scar.

Sensitivity, PPV, and false positives for individual and combined imaging modalities (Tables 3–5)

Table 3.

Sensitivity, PPV, and proportion of falsely positive lesions for each modality in subjects whose tumor was identified and in all subjects with EAS

	CT	MRI	LOCT	HOCT	FDG- PET	DOPA-PET
Tumor-identified patients (n = 30)
A. Sensitivity, % (95% CI) (B below × 100)	93 (79–98)	90 (74–96)	57 (39–73)	50 (25–75)	64 (35–85)	55 (28–79)
B. No. of patients with TP lesions/no. with imaging study	28/30	26/29	16/28	6/12	7/11	6/11
C. PPV, % (D below × 100)	66	74	79	89	53	100
D. No. of TP lesions/no. of total lesions on imaging	48/73	37/50	22/28	8/9	8/15	8/8
E. Proportion of FP lesions, %	34	26	21	11	47	0
All patients with EAS (n = 41)
A. Sensitivity, % (95% CI) (B below × 100)	68 (53–80)	65 (50–78)	41 (27–57)	30 (14–52)	50 (27–73)	46 (23–71)
B. No. of patients with TP lesions/no. with imaging study	28/41	26/40	16/39	6/20	7/14	6/13
C. PPV, % (D below × 100)	57	67	76	73	50	89
D. No. of TP lesions/no. of total lesions on imaging	48/84a	37/55a	22/29	8/11	8/16	8/9
E. Proportion of FP lesions, %	43	33	27	27	50	11

F, False positive; TP, true positive. 

^aPatient 33 had lesion detected by CT and MRI with inconclusive characteristics but in the same location; this was not considered to be true positive lesion. Patient 41 had elongated inhomogeneous lesion detected by CT and MRI in the same location; this was not considered to be true positive lesion. 

Table 4.

Sensitivity for the combinations of the anatomical and functional imaging modalities in subjects whose tumor was identified and in all subjects with EAS

	LOCT		LOCT and/or HOCT		FDG-PET		DOPA-PET
	Combined with CT	Combined with MRI	Combined with CT	Combined with MRI	Combined with CT	Combined with MRI	Combined with CT	Combined with MRI
Tumor identified patients (n = 30)
A. Sensitivity, % (95% CI) (B below × 100)	57 (39–73)	52 (34–69)	68 (49–82)	63 (44–78)	64 (35–85)	60 (31–83)	55 (28–79)	55 (28–79)
B. No. of patients with concordant TP lesions/no. with imaging study	16/28	14/27	19/28	17/27	7/11	6/10	6/11	6/11
C. PPV, % (D below × 100)	91	100	93	100	67	60	100	100
D. No. of concordant TP lesions/no. of total lesions on imaging	21/23	26/26	26/28	27/27	8/12	6/10	8/8	8/8
All patients with EAS (n = 41)
A. Sensitivity, % (95% CI) (B below × 100)	41 (27–57)	37 (23–53)	49 (34–64)	45 (30–60)	50 (27–73)	46 (23–71)	46 (23–71)	46 (23–71)
B. No. of patients with concordant TP lesions/no. with imaging study	16/39	14/38	19/39	17/38	7/14	6/13	6/13	6/13

The PPV is identical for both groups of patients. TP, True positive. 

Table 5.

Combinations of positive and negative imaging results in 30 patients with surgically identified ectopic ACTH-producing tumors

Unshaded (white background) boxes indicate that the test was done; those that are unfilled were positive, while those that were negative are indicated as such. Gray boxes indicate that the test was not done. Negative Octreotide scan doses are indicated as L (low dose, 6 mCi) or H (high dose, 18 mCi). When only L is indicated, the higher dose scan was not done. NEG, Negative; PT ID, patient identification number. 

For the tumor-found group, DOPA-PET and HOCT had the lowest sensitivity on a per-patient basis, and the other modalities had overlapping CI: CT 93% (95% CI = 79–98%), MRI 90% (95% CI = 74–96%), LOCT 57% (95% CI = 39–73%), HOCT 50% (95% CI = 25–75%), FDG 64% (95% CI = 35–85%), and F-DOPA 55% (95% CI = 28–79%). By modality, the fraction of patients with at least one false-positive finding was 50% by CT, 31% by MRI, 18% by L/HOCT, and 18% by FDG-PET.

On a per-lesion basis, PPV were 100% for DOPA-PET, 89% for HOCT, 79% for LOCT, 74% for MRI, 66% for CT, and 53% for FDG-PET. For FDG-PET, seven (47%) of 15 lesions proved false positive, whereas with F-DOPA-PET, there were no false-positive findings. When considering all patients (tumor-found and occult), per-patient sensitivities decreased and the percent of false-positive findings increased, as expected.

Combining functional with anatomic imaging by requiring abnormal CT or MRI and an abnormal L/HOCT increased PPV for individual lesions to 93% (CT) and 100% (MRI) (Tables 4 and 5). The combination of CT and MRI alone had a lower lesion PPV (80%) but higher sensitivity (83%) for individual patients. F-DOPA-PET similarly improved PPV of CT/MRI; FDG-PET did not.

Discussion

Imaging studies are the cornerstone of the management of EAS, because early localization and surgical removal of tumor constitutes optimal treatment. This study explored five questions related to localization of these tumors. 1) Which imaging modality or combination best detects tumors? 2) Is there any strategy to reduce the problem of false-positive results? 3) Does the technique used for any given modality influence the results? 4) What might explain negative results? 5) How does the method of interpretation affect the results?

Most tumors causing EAS are intrathoracic (83% in this series) (3,4,7,19). This suggests that initial imaging should focus on the chest. Of tumors identified at the first visit in this study, 14 of 18 were intrathoracic, whereas four were in the brain, abdomen, or neck. By contrast, in 12 patients whose tumor was identified on follow-up visits, 11 had intrathoracic tumors and one had pancreatic neuroendocrine tumor. These data support a strategy of performing thoracic studies first at all visits, with the understanding that additional scans may be required.

The conclusion of this study agrees with de Herder and Lambert’s editorial analysis (20) that no single imaging technique has optimal accuracy. The choice of imaging modalities is guided by the sensitivity of the procedure balanced by the risk of false-positive findings. One way to address the problem of false-positive results is to obtain two different types of imaging on the premise that false-positive lesions would not be concordant in both. In patients with identified tumors, the combination of two anatomic modalities, CT and MRI, had a per-patient sensitivity of 83% and a PPV per lesion of 80%. Perhaps more useful was the combination of anatomic and functional imaging. For instance, MRI and LOCT/HOCT resulted in a sensitivity of 63% and PPV of 100%. One patient with falsely negative LOCT/HOCT and pulmonary carcinoid had a positive F-DOPA PET, suggesting that the latter may also be a helpful adjunct. As expected, the combination of two positive imaging modalities decreases sensitivity but increases specificity, particularly with functional imaging.

Functional imaging reduces false-positive results because it relies on the specific properties of tumor cells, not just their anatomic characteristics. However, tumors lacking the relevant receptors (OCT), metabolic rate (FDG-PET), or amine precursor uptake (F-DOPA) have false-negative results (9). F-DOPA-PET had lower sensitivity (50%) in our study than in reports comprising mainly hindgut carcinoids (88–93%) (21,22). In EAS, FDG-PET is most likely to detect metabolically active tumors or adrenal pheochromocytomas (23). Because occult tumors may be metabolically less active, it is not surprising that FDG-PET best localizes overt and not occult disease (13).

Tsagarakis et al. (24) demonstrated a high diagnostic yield of LOCT in localizing EAS tumors; the modality identified tumor in all six patients with bronchial carcinoids, either initially or during follow-up evaluation. In contrast, Koopmans et al. (25) reported lower LOCT sensitivity compared with DOPA-PET (46 vs. 96%) in patients with metastatic gastrointestinal carcinoids. Our study suggests that use of a higher dose should be considered for OCT, because three of nine patients with negative LOCT scans were positive with HOCT, two others were retrospectively read as positive after identification by HOCT, and another was confirmed positive on HOCT. When new radioligands with increased affinity to all somatostatin subtype receptors become available, the sensitivity of OCT also should increase.

With improved imaging techniques over time, anatomical studies including CT (thinner 1.2-mm sections, chest) and MRI (3-T magnet) may better detect smaller tumors. This study did not address this question directly, because patients did not receive CT scans of different slice thickness at the same visit, and few had both 1.5- and 3-T MRI. Also, use of CT coregistration with PET may enhance its utility.

Cardiac imaging suppresses myocardial motion and may thus enhance detection of small tumors adjacent to the heart and/or great vessels. It can also demonstrate the extent of tumor involvement of cardiac structures to help direct surgery. However, because it is relatively new, few patients in our study underwent this imaging modality; the role of cardiac gating in tumor localization should be explored further.

Finally, PPV for each modality will differ among interpreting radiologists. In this study, routine clinical interpretations by staff radiologists did not identify all tumors. Compared with the clinical interpretations, the prospective research interpretation identified tumors smaller than 1 cm on CT and/or MRI in four patients, and the retrospective interpretation identified lesions in three others. Retrospective unblinded review of previous CT images with thicker thoracic sections identified tumors that had been identified prospectively on the clinical and/or research readings but were not reported as culprit lesions in four patients. Similarly, eight patients whose tumors were detected on initial evaluation at the NIH had recent outside imaging that did not identify the tumor(s). We were not able to clarify whether this represented differences in technique or interpretation or whether the 1- to 5-month gap between the studies might be important. Based on this experience, if tumors are not identified at initial evaluation, we recommend that the patient be referred to a highly specialized center to obtain additional imaging and interpretation by an experienced team of radiologists.

Tumors were best detected by correlating different imaging modalities. Knowledge of the facts that these tumors are often quite small and occur in locations that are unusual (e.g. epicardiac fat) or difficult to visualize or interpret (retrocardiac or pancreatic) may assist in their identification.

These results are potentially biased by our definition of false-positive lesions. Only suspicious lesions consistent with tumor, which could not be otherwise explained, were defined as positive. Inclusion of all lesions noted on reports would lead to even more false-positive findings and lower PPV.

This study has other limitations. PET examinations were performed without CT, which enhances accuracy. Some imaging techniques (e.g. F-DOPA-PET and HOCT) are not widely available, thus decreasing the translation of these results to practice. Also, not all imaging was performed at each visit. In addition, the study is small, and other tumor types might not be found with the same frequency.

In conclusion, we suggest that initial imaging of patients with presumed EAS include thoracic CT and MRI followed by LOCT. Further investigations in a larger population with different tumor types and amounts of tumor burden are necessary to confirm and extend these findings and determine the best imaging studies and/or their combinations for the detection of ectopic ACTH-producing tumors.