⁶⁸Ga-DOTA-Tyr³-Octreotide PET in Neuroendocrine Tumors: Comparison with Somatostatin Receptor Scintigraphy and CT

Patients

From September 2004 to April 2006, 84 consecutive patients (48 men, 36 women; age range, 28–79 y; mean age ± SD, 58.2 ± 12.2 y) were enrolled in this prospective phase IIb study. For analysis, the patients who were investigated were divided into 3 groups (14): The first group consisted of patients who underwent imaging for the initial detection and localization of suspected NET and potential metastases in the presence of clinical or biochemical suspicion (detection; n = 13). Patients with histologically proven neuroendocrine malignancies were enrolled for staging purposes in the second group (staging; n = 36). In the third group, patients were referred during posttherapy follow-up (follow-up; n = 35) to exclude or to detect tumor recurrence. In the last 2 groups, at least 1 tumor manifestation was histologically confirmed in patients with multiple metastases. The patient characteristics are summarized in Table 1.

Four patients had hypoglycemia with symptoms of both neuroglycopenia and catecholamine response. During symptoms, blood glucose levels were <40 mg/dL. Therefore, a pancreatic islet cell tumor was considered in these patients.

NET are generally differentiated between those producing hormone-related symptoms (e.g., flush or diarrhea) and those presenting without any hormonal symptoms. Accordingly, 27 patients with clinical and biochemical signs for a secreting tumor and 57 patients with a nonfunctional tumor were included.

In patients who were referred for restaging during follow-up (n = 35), various therapeutic procedures were performed before inclusion. Most of these patients were treated by surgery (n = 29), some of them without further treatment (n = 6). Additional drug therapy was administered to 23 patients. Seven patients were treated with chemotherapy, and 16 patients were treated with long-acting SST analogs alone or in combination with interferon-α.

Written informed consent was obtained from all patients before being included in the study, and the study was approved by the local ethics committee.

PET

SRS

^99mTc-Labeled hydrazinonicotinyl-Tyr³-octreotide (^99mTc-HYNIC-TOC) was prepared using a kit formulation as recently described (16). Each patient received a mean activity of 400 MBq (intravenously) of the tracer. Whole-body imaging was performed at 2 and 4 h after injection using a dual-detector VertexPlus scintillation camera (Philips), which was followed by SPECT (6). The scan speed for whole-body imaging was 10 cm/min when using the ^99mTc-labeled derivative. The camera was equipped with a low-energy, all-purpose, parallel-hole collimator (window setting, 140 keV; width, 10%; 180° rotation detector head; 64 projections; 128 × 128 matrix; 40-s acquisition time per projection). The SPECT image data were reconstructed by standard filtered backprojection using a Butterworth filter.

DOTA-TOC was radiolabeled as reported elsewhere (17). ¹¹¹In-DOTA-TOC whole-body scintigraphy in anterior and posterior views was performed at 4, 24, and 48 h after a single-dose injection of 150 MBq of the ¹¹¹In-labeled radiopharmaceutical. Scintigraphic acquisitions were obtained with the same double-head γ-camera as described (ADAC; VertexPlus), equipped with a medium-energy, parallel-hole collimator (window setting, 172 and 246 keV; window width, 20%). The scan speed for whole-body imaging was 5 cm/min using the ¹¹¹In-labeled radiopharmaceutical. SPECT was acquired after 24-h whole-body imaging using the same reconstruction algorithm as mentioned earlier.

Sixty-six patients were investigated with only 1 tracer: 33 patients with ^99mTc-HYNIC-TOC and 33 with ¹¹¹In-DOTA-TOC. In 18 patients, both radiopharmaceuticals were used for comparison with PET and CT.

CT and Image Fusion Procedure

Helical CT scans of the thorax and the abdomen with a slice thickness of 2.5 mm were obtained with the HiSpeed CT/i Advantage scanner (GE Healthcare). Typically, 150 mL (twice the weight of the patient in kilograms) iopromidum contrast media (Ultravist 370; Schering) were administered at 5 mL/s, with scan delays of approximately 30 s for the late arterial phase and 70 s for the portal phase.

For image fusion, the PET, SPECT, and CT scans were performed sequentially using an individualized vacuum mattress with external markers attached to it. For every image acquisition (PET, SPECT, and CT), the patient was repositioned into the vacuum mattress (11). The image fusion procedure was used for anatomic delineation of abnormal findings in SPECT and PET.

Interpretation and Data Evaluation

PET and SPECT studies were interpreted independently by 2 experienced nuclear medicine physicians. Corresponding studies were compared lesion by lesion for final analysis and ruled as matching or mismatching by the 2 nuclear medicine specialists. If the result of the 2 viewers was discordant, a third reader—who acted as referee—was consulted. They were aware of the patients' clinical history, which was provided by the referring physician but were unaware of any result of other imaging modalities. The criteria for image interpretation of PET and SPECT are summarized in Table 2. As a measure for diagnostic yield, the number of lesions that could be identified clearly as single foci was determined. Lesions within the liver were rated as 1 organ metastasis, considering the irregular configuration and confluence of some lesions, so that an individual metastasis frequently was not delineated. A lesion-by-lesion analysis was performed for all other tumor foci. Concordant findings on nuclear medicine techniques (PET and SPECT) and CT meant that both techniques (PET or SPECT and CT) were consistent with malignancy. In the case of discrepancies with regard to nuclear medicine and CT findings, further assessment of abnormal foci was mandatory—that is, by histologic proof or follow-up controls with CT or MRI after 3 mo and, if necessary, after 6 mo. If malignant evolution on follow-up or progression on therapy was observed, these suggestive findings were considered malignant for the final decision. Those patients with no abnormal findings were monitored over a period of at least 6 mo with repeated CT or MR scans before a scan result was considered true-negative (TN).

CT and, if necessary, MR scans were interpreted by experienced radiologists who were unaware of the scintigraphic results. A positive diagnosis was based on the specific appearance of malignant disease derived from NET as reported elsewhere (18). Thus far, abnormal findings were assessed histologically in 13 patients. Repeated clinical examinations with CT were performed in 73 cases and with MR in 22 cases. More details are given in Table 1. Attention has been directed toward unproven findings of the scans.

Statistical Analysis

The results of the 3 imaging modalities (PET, SPECT, and CT) were classified as true-positive (TP), TN, false-positive (FP), or false-negative (FN) according to the reference standard, as described earlier. The χ² test for independence, or the Fisher exact test when appropriate, was used to evaluate differences in lesion detectability when subgroups of the patients being investigated were statistically compared (¹¹¹In-DOTA-TOC vs. ^99mTc-HYNIC-TOC and secreting vs. nonsecreting tumors). The McNemar test of correlated properties was used to statistically compare the imaging results of ⁶⁸Ga-DOTA-TOC PET with SPECT and diagnostic CT. Analysis was done on a lesion basis and on a patient basis. All P values < 0.05 were considered significant. Cohen's κ-statistic with 95% confidence intervals was calculated to show the degree of association between the techniques. The function of uptake over time was assessed using linear regression analysis.

For evaluation of the clinical value of ⁶⁸Ga-DOTA-TOC PET in comparison with the other imaging modalities, organ systems were assessed for recognition of any lesion in the tissue. The focus of interest was related to unknown tumor lesions arising in organ systems, unaware of malignant involvement from other imaging techniques (e.g., bone), with clinically relevant information in terms of further patient management.

Biodistribution of ⁶⁸Ga-DOTA-TOC

Uptake of ⁶⁸Ga-DOTA-TOC was routinely found in neuroendocrine tissue (pituitary and adrenal glands) and in 57 patients (67.8%) in the pancreatic head without known pathology. Additionally, the spleen and the urinary excretion system also showed enhanced tracer accumulation. Homogeneous uptake was observed in the liver and in 34 patients (40.4%) in the thyroid gland. Because of the excellent tumor-to-organ contrast, NET lesions could be easily identified by visual analysis (Fig. 1). However, anatomic delineation of abnormal findings was difficult without image fusion because of highly specific tracer uptake. Intravenous injection of ⁶⁸Ga-DOTA-TOC was well tolerated in all patients, and no side effects were observed in any patients after tracer injection. On the basis of an initial evaluation of 8 patients, the optimal time of acquisition was determined to be 100 min after injection. In 1 of these 8 patients, 2 additional thoracic findings were observed when comparing late acquisition 100 min after injection with the earlier acquisition (patient 12). The acquisition protocol was adapted after this initial series with only 1 single whole-body scanning at 100 min after injection. Linear regression analysis of tumor uptake values shows a significant increase in SUVs of 14% (60 min vs. 20 min, R² = 0.96), 9% (100 vs. 60 min, R² = 0.97), and 24% (100 vs. 20 min, R² = 0.89). Interpatient SUVs at different times are given in Table 3.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

A 28-y-old female was referred for primary diagnosis of a NET because of elevated tumor markers in serum. PET (A) clearly depicted an abnormal focus in upper abdomen (arrow). This lesion could be delineated in the pancreas after image fusion with CT (B). There was also increased contrast medium enhancement in the margin when using helical CT (C). SPECT with ^99mTc-HYNIC-TOC was also positive for this tumor in upper abdomen (D). This positive finding was confirmed by histopathology revealing a NET with 1 cm in diameter. (Top) Coronal views; (bottom) axial views.

Analysis on Patient Basis

Results from all 84 patients studied are summarized in Table 4. Among the 84 patients, ⁶⁸Ga-DOTA-TOC PET was TP in 69 (82.1%), TN in 12 (14.3%), FP in 1 (1.2%), and FN in 2 (2.4%) patients, indicating a sensitivity of 97% (69/71 patients), a specificity of 92% (12/13 patients), and an accuracy of 96% (81/84 patients) on a patient basis. An analysis per patient comparing the scan results of PET with SPECT and with diagnostic CT emphasizes the improved diagnostic efficacy of ⁶⁸Ga-DOTA-TOC, with a P value of < 0.001 using the McNemar test (Table 5). Cohen's κ-statistic of 0.3 showed only fair association between the techniques.

Analysis on Lesion Basis

⁶⁸Ga-DOTA-TOC PET studies detected 375 abnormal findings in 70 patients, of which 374 were TP and 1 was FP. This FP finding was found in patient 30, who had clinical symptoms suggestive of a secreting NET and elevated tumor markers (chromogranin A [CgA] of 34.6 U/L and neuron-specific enolase [NSE] of 26.7 μg/L) and who presented with enhanced tracer uptake in the pancreatic head. Surgical exploration was negative, and histology and further follow-up controls did not confirm this finding to be malignant. Overall, 23 abnormal findings in 22 patients were considered malignant in the pancreas. Fourteen of those were found in the pancreatic head, with 1 FP finding.

⁶⁸Ga-DOTA-TOC was FN in 2 patients. A 47-y-old woman was referred for initial staging of a NET unknown primary (patient 36). PET and SPECT were negative for multiple liver metastases that were histologically confirmed by biopsy. Both nuclear medicine techniques were also negative for histologically confirmed small liver metastases in the other patient, a 67-y-old man (patient 78). This patient was referred for follow-up after surgery and chemotherapy of a rectal tumor. In both patients, diagnostic CT revealed a TP scan result.

⁶⁸Ga-DOTA-TOC and Functional Status of NET

The fraction of patients with clinical and biochemical features of a NET consisted of 18 TP, 8 TN, and 1 FP results, whereas in the group of patients with nonfunctioning tumors, 51 TP, 4 TN, and 2 FN results were observed. When comparing both groups, no statistically significant difference was found for PET (P = 0.96). Both patients with the FN scan result did not show any functional activity of the tumor, whereas the FP result was observed in a patient with elevated CgA level and persisting diarrhea, suggestive of a hormone-active tumor, as mentioned earlier.

PET Versus Scintigraphy (SPECT) and Diagnostic CT

All 3 modalities (PET, SPECT, and CT) showed an equivalent scan result in 39 patients (46.4%), including 27 TP and 12 TN results (Fig. 1).

Discrepancies between PET and SPECT were found in 32 patients (38%), all of whom were TP with PET and FN with SPECT. In this patient group, liver metastases were missed in 10 cases. Twenty-two additional small lymph node metastases also were not detected with SPECT in 15 patients. In 2 patients with carcinoid tumors, small peritoneal deposits escaped detection with SPECT. Furthermore, 32 bone metastases were not delineated by conventional scintigraphy but were positive with ⁶⁸Ga-DOTA-TOC PET. Discrepancies between PET and CT were found in 34 patients (40.5%), of whom there were 2 TP, 1 TN, 5 FP, and 26 FN findings with CT. FP findings with CT were caused by suggestive small nodular lung lesions in 2 patients and in 2 additional cases by enlarged lymph nodes. One 55-y-old male patient was referred for initial detection of a NET in the case of elevated CgA levels (patient 52). Abdominal CT visualized a lesion in the wall of the jejunum with a diameter of 1.4 cm. The contrast medium showed enhanced uptake of a primary NET. However, PET and SPECT were negative. Surgical exploration revealed a benign leiomyoma, which was proven by histology. Site-related differences are illustrated in Table 6.

Eighteen patients were investigated with both SPECT tracers, yielding a comparable scan result. The ^99mTc-labeled compound was TP in 18 patients, TN in 11, FP in 1, and FN in 21 patients. When using ¹¹¹In-DOTA-TOC, the scan result was TP in 29 patients, TN in 1, and FN in 21 patients. No statistically significant difference was observed between the 2 groups (P = 0.84).

Clinically Valuable Information Obtained by PET

In 18 patients (21.4%), ⁶⁸Ga-DOTA-TOC provided further clinically relevant information in comparison with diagnostic CT alone, including 9 patients with unknown bone metastases (Fig. 2). The primary tumor or residual tumor at the primary site was demonstrated in 5 patients with ⁶⁸Ga-DOTA-TOC PET but escaped detection by CT. A 61-y-old woman (patient 65) was referred after treatment of a pulmonary carcinoid tumor for follow-up. Diagnostic CT was negative, but PET revealed small metastatic lesions in the myocardium and in the pancreas, with focally enhanced tracer accumulation. Multiple liver metastases were known in a 47-y-old woman (patient 80) who was investigated during follow-up after surgical resection of a small bowel carcinoid. ⁶⁸Ga-DOTA-TOC additionally showed a small lesion in the right breast initially not found with the other 2 modalities (Figure 3). This lesion with a diameter of 7–4 mm and 3 other metastases in the liver were surgically removed. In 2 patients, small liver metastases were not shown with diagnostic CT and SPECT (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

A 56-y-old woman with multiple liver and lymph node metastases was referred for restaging after surgery and chemotherapy. CT presented these tumor lesions; however, it was negative for bone lesions. Beside the visceral metastases, some additional osteoblastic and osteolytic bone metastases were clearly depicted with ⁶⁸Ga-DOTA-TOC (A). Only some of these bone metastases were delineated by conventional scintigraphy (B, anterior view; C, posterior view). Osteoblastic bone lesions were confirmed by ¹⁸F-Na-fluoride PET (D). Retrospective CT analysis after image fusion revealed some of these bone metastases.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

A 47-y-old female patient was referred for scanning after resection of a carcinoid of the ileum. Multiple liver metastases were known (A). Additionally, ⁶⁸Ga-DOTA-TOC showed a small lesion in right breast (arrows) (B). This finding was initially not detected with CT or scintigraphy (C). Ultrasound-guided fine-needle biopsy confirmed a metastasis in soft tissue derived from the NET with 7- to 4-mm diameter (D). This tumor lesion and 3 liver metastases were consecutively surgically removed.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

A 62-y-old male patient was investigated after resection of a small bowel carcinoid. ⁶⁸Ga-DOTA-TOC PET displayed multiple small liver metastases (A). These liver lesions were negative with the other 2 modalities, CT and scintigraphy (B) including SPECT (C). Ultrasonography (D) and further follow-up controls confirmed these lesions. Diameters of metastases were in the range of 1 cm. Positive PET finding initiated treatment with [¹⁷⁷Lu-DOTA⁰,Tyr³,Thr⁸]octreotide (¹⁷⁷Lu-DOTA-TATE).

Compared with scintigraphy, ⁶⁸Ga-DOTA-TOC PET provided further valuable clinical information in 12 patients (14.3%). Three patients have just been mentioned. Unknown bone metastases were shown in 5 patients. Surgical intervention was omitted in 3 patients because widespread disease was detected by ⁶⁸Ga-DOTA-TOC, showing additional unknown distant tumor lesions. One 34-y-old male patient was investigated after chemotherapy and chemoembolization of metastatic lesions in the liver of a NET unknown origin. PET additionally showed the primary tumor in the pancreatic head and local lymph node metastases (patient 34).