This retrospective analysis included a total of 32 patients with histologically confirmed, unresectable, metastatic gastroenteropancreatic neuroendocrine tumors (GEP-NET) who underwent treatment with [¹⁷⁷Lu]Lu-octreotate (17 men, 15 women; age range: 40–90 years; mean age: 67.8 years; median age: 70 years). Prior to PRRT, patients underwent various pre-treatments, including surgical resection, somatostatin analog (SSA) therapy, targeted molecular therapies (e.g., everolimus, sunitinib), or chemotherapy, depending on tumor burden, progression status, and individual patient characteristics. All patients met the general inclusion criteria for peptide receptor radionuclide therapy (PRRT), including sufficient tumor uptake (i.e., uptake ≥ liver uptake) on baseline [⁶⁸Ga]Ga-DOTATOC-PET/CT (24–26). Within the cohort, 12 patients had pancreatic NET, while 20 patients had non-pancreatic GEP-NET. The study was conducted in accordance with the Declaration of Helsinki and national regulations. Written informed consent was obtained from all participants for the scientific analysis of their data.

PRRT was administered with a mean activity of 7.9 GBq (216 mCi) [¹⁷⁷Lu]Lu-octreotate per treatment cycle, targeting a total of four cycles at standard intervals of 3 months (10–14 weeks). The ¹⁷⁷Lu (IDB Holland, Baarle-Nassau, Netherlands) had a specific activity ranging from approximately 100 to 160 GBq/μmol at the time of administration. Peptide labeling was conducted to achieve an apparent specific activity of approximately 54 GBq/μmol, defined as the ratio of activity to the total peptide amount (27, 28). Nephroprotection was provided through standard amino acid co-infusion following the Rotterdam protocol, consisting of lysine (2.5%) and arginine (2.5%) in 1 L of 0.9% NaCl, administered at an infusion rate of 250 mL/h (29, 30).

Baseline [⁶⁸Ga]Ga-DOTATOC PET/CT was performed 2 to 7 days prior to the first PRRT cycle. Long-acting somatostatin analogs were discontinued for at least 4 weeks, while short-acting analogs were paused for at least 1 day before imaging. DOTATOC labeling was conducted using ⁶⁸Ga eluted from an in-house ⁶⁸Ge/⁶⁸Ga generator, following the procedure described by Zhernosekov et al. (31). The PET/CT scans covered the area from the base of the skull to the upper thighs, with five to seven bed positions, and were acquired 30 min after the intravenous injection of 200 MBq [⁶⁸Ga] Ga-DOTATOC. Imaging was performed using a hybrid PET/CT scanner (Biograph 2, Siemens Medical Solutions Inc., Hoffman Estates, Illinois, United States), which consisted of a dual-detector helical CT and a high-resolution PET scanner with a 16.2 cm axial field of view and lutetium oxyorthosilicate (LSO) crystal detectors (6.45 × 6.45 × 25 mm). CT imaging was performed for attenuation correction and anatomical localization, with acquisition parameters set to a tube current of 60 mAs, a tube voltage of 130 kV, a rotation time of 0.8 s, a slice thickness of 5 mm, a slice width of 5 mm, and a table feed of 8 mm per s. To enhance vascular and parenchymal delineation, 140 mL of iodinated contrast material (Ultravist 300; Schering, Berlin, Germany) was administered via an automated injector (XD 5500; Ulrich Medical Systems, Ulm, Germany) with a start delay of 50 s. Following CT image acquisition, PET data were collected for 5 min per bed position (total duration: approximately 35 min). The PET scanner had a coincidence time resolution of 500 ps, a coincidence window of 4.5 ns, and a sensitivity of 5.7 cps/kBq at 400 keV. Attenuation-corrected PET data were reconstructed using a standardized ordered-subset expectation maximization (OSEM) iterative reconstruction algorithm with two iterations, eight subsets, and a 5 mm Gaussian filter.

For each patient, two to three tumor lesions were selected as target lesions, specifically those that were well-demarcated. Irregular regions of interest (ROIs) with a threshold of 50% of the maximum DOTATOC uptake were drawn on the transverse PET slices. The standardized uptake values (SUV), including SUV_mean and SUV_max, were calculated for each lesion using the standard formula that accounts for the measured activity concentration, corrected for body weight and injected activity. To normalize tumor SUV values, normal liver parenchyma was used as the background reference, and the SUV ratios of target lesions to the liver (SUV_meanT/L and SUV_maxT/L) were derived. Functional tumor volume (TV_PET) was also determined for each lesion using the same threshold. Restaging with [⁶⁸Ga]Ga-DOTATOC PET/CT was performed 3 months after the completion of PRRT, following the same imaging protocol as at baseline. The response of each tumor lesion was assessed based on the percentage change in functional tumor volume (%∆TV_PET). In the case of ∆TV_PET, variations are expressed as absolute values, whereas for %∆TV_PET, variations are presented as percentages.

The CT-based tumor volume (TV_CT) was manually segmented and measured using Sectra IDS7 PACS (Version 24.2). ΔTV_CT was defined as the absolute change in tumor volume between baseline and post-PRRT imaging, while %ΔTV_CT represents the relative volume change normalized to baseline volume. In the case of ∆TV_CT, variations are expressed as absolute values, whereas for %∆TV_CT, variations are presented as percentages.

Data are presented using descriptive statistics, including median (minimum–maximum), mean ± standard deviation, and count (percentage). Chi-squared tests or Fisher’s exact tests (as appropriate) were used to compare the proportions of patient groups dichotomized based on baseline characteristics. Mann–Whitney U tests were applied to compare quantitative tumor parameters (SUV_max, SUV_mean, SUV_meanT/L, and SUV_maxT/L) across different groups. The association between tumor parameters in baseline [⁶⁸Ga]Ga-DOTATOC PET/CT and the respective response to PRRT (%∆TV_PET) was assessed using Spearman’s rank correlation analysis. All tests were two-sided, and a p-value <0.05 was considered statistically significant. Statistical analyses were conducted using SPSS (version 20.0; SPSS Inc., Chicago, IL, United States) and GraphPad Prism (version 10.2.3).

A total of 121 PRRT cycles with [¹⁷⁷Lu]Lu-octreotate were administered to 32 patients. The mean age of the cohort was 67.8 years (range 40–90 years, median 70 years). Patients received up to four PRRT cycles, with a mean of 3.8 ± 0.7 cycles. The mean cumulative activity of [¹⁷⁷Lu]Lu-octreotate was 29.3 ± 0.7 GBq. Treatment response, assessed according to the modified SWOG criteria (32), included partial response (PR) in 12 patients (37.5%), minimal response (MR) in eight patients (25%), stable disease (SD) in eight patients (25%), and progressive disease (PD) in four patients (12.5%). Therefore, we used post-PRRT PET/CT as the gold standard for response assessment, as it provides functional information on tumor activity. The mean progression-free survival (PFS) was 28.6 ± 15 months. No carcinoid crises were observed.

Lesion-based response analysis following PRRT was conducted for 66 lesions. At baseline, the mean SUV_max was 28.1 ± 16 (range: 3.0–91.2), SUV_mean was 13.3 ± 5.1 (range: 2.4–27.1), SUV_meanT/L was 3.6 ± 1.7 (range: 0.96–10.45), and SUV_maxT/L was 7.7 ± 5.6 (range: 1.3–30.9). The functional tumor volume at baseline (TV_PET) was 53.1 ± 12.2 mm³. Moreover, 3 months after PRRT completion, the absolute functional tumor volume change (∆TV_PET) was 25.5 ± 9.3 mm³, while the percentage change in functional tumor volume (%∆TV_PET) was 40.2 ± 49.7%.

Some discrepancies between contrast-enhanced CT and [⁶⁸Ga]Ga-DOTATOC PET/CT were observed, particularly regarding detectability and tumor size assessment. Figure 1 illustrates an example of a patient with a pancreatic neuroendocrine tumor (P-NET) before and 3 months after the completion of PRRT. In the lesion-based analysis, neither metastatic site (hepatic vs. extrahepatic, p = 0.702) nor baseline lesion volume (p = 0.480) significantly influenced lesion response. However, lesions originating from the pancreas showed a significantly greater response compared to non-pancreatic lesions (%∆TV_PET 56.8 ± 39.8 vs. 28.4 ± 50.1; p = 0.020). Pretreatment SUV-derived values and treatment-induced volumetric changes, stratified by baseline patient characteristics, are summarized in Table 1 (∆TV_CT, %∆TV_CT) and Table 2 (∆TV_PET, %∆TV_PET). Since Ki-67 index data were not available for all patients and FDG-PET/CT follow-up data were missing for two patients, 26 and 24 patients, respectively, were analyzed in the tables. No significant difference in CT-derived tumor volume change (∆TV_CT) was observed based on tumor type (GEP-NET vs. P-NET), overall response (responders vs. non-responders), Ki-67 status, or metastatic location (liver vs. other sites). However, %∆TV_CT differed significantly between responders (66.7 ± 39.2) and non-responders (12.7 ± 32.4), while no significant difference was found for other parameters.

Similarly, the change in PET-derived tumor volume (∆TV_PET) did not show significant differences based on metastatic location or tumor type. Patients with a high Ki-67 index (>2%) exhibited a significantly greater ∆TV_PET (54.1 ± 106.5 mL) compared to those with low Ki-67 (<2%) (0.84 ± 28.8 mL). Furthermore, ∆TV_PET was significantly higher in overall responders (47.7 ± 95.5 mL vs. 10.5 ± 32.6 mL). Regarding tumor type, %∆TV_PET was significantly greater in P-NETs (70.1 ± 29.5) than in GEP-NETs (27.1 ± 54.3). Additionally, overall responders exhibited a significantly higher %∆TV_PET (59.1 ± 35.9 vs. 12.3 ± 53.7). However, no significant differences in %∆TV_PET were observed based on Ki-67 status or metastatic location.

Baseline SUV_mean (p = 0.071) and SUV_max (p = 0.706) of tumor lesions showed no significant association with lesion-based response. In contrast, higher tumor-to-background ratios at baseline, specifically SUV_meanT/L (p = 0.011, r = 0.381) and SUV_maxT/L (p = 0.004, r = 0.435), were associated with more pronounced changes in functional tumor volume, as illustrated in Figure 2. For marker lesions, higher SUV_meanT/L and SUV_maxT/L were predictive of therapy response, defined as a >50% decrease in %∆TV_PET (p = 0.0027 and p = 0.0001, respectively), as shown in Figure 3; however, a considerable overlap was observed.

Given the higher likelihood of complete remission in smaller lesions, the association between SUV-derived values and lesion-based response was further evaluated for lesions >10 mL (n = 44). In this subgroup, the correlation between pretreatment SUV_maxT/L (p < 0.001) and SUV_meanT/L (p < 0.001) with functional tumor volume response remained statistically significant (data not shown).