Standardized Uptake Values on SPECT/CT: A Promising Alternative Tool for Treatment Evaluation and Prognosis of Metastatic Neuroendocrine Tumours

(1) Background: The aim of our study was to assess the feasibility of 99mTcEDDA/HYNIC-TOC SPECT/CT quantitative analysis in evaluating treatment response and disease progression in patients with NETs. (2) Methods: This prospective monocentric study evaluated 35 SPECT/CT examinations performed on 14 patients with neuroendocrine tumours who underwent a baseline and at least one follow-up 99mTcEDDA/HYNIC-TOC scan as part of their clinical management. The examination protocol included a whole-body scan acquired 2 h after the radiotracer’s administration, with the SPECT/CT performed 4 h post-injection. Images were analyzed by two experienced physicians and patients were classified into response categories based on their changes in SUV values. (3) Results: We evaluated 14 baseline studies and 21 follow-up scans, accounting for 123 lesions. A statistically positive correlation has been found between the SUVmax and SUVpeak values in tumoral lesions (p < 0.05). No correlation has been found between the SUV values and the ki67 proliferation index. Finally, 64.29% patients were classified as SD at the end of the study, with only 14.29% of patients exhibiting PD and 21.43% patients with PR. (4) Conclusions: The quantitative analysis of 99mTcEDDA/HYNIC-TOC SPECT/CT data in patients with neuroendocrine tumours could represent an alternative to 68Ga-DOTA-peptides PET/CT for the monitoring and prognosis of NETs.


Introduction
Neuroendocrine neoplasms (NENs) are a group of heterogenous malignancies that originate from diffuse neuroendocrine cells dispersed throughout the body which possess secretory properties which lead to their specific syndromes of uncontrolled hormone hypersecretion [1,2]. NENs have been described in the nervous system, respiratory tract, gastrointestinal tract, thyroid, breast, skin and urogenital system, most commonly the gastro-entero-pancreatic (60% of the cases) and pulmonary (30% cases) [2,3]. Overall, considered as indolent tumours, their prognosis varies by morphology, proliferation rate, primary tumoral site and stage at diagnosis [3].
Bucharest, Romania, from November 2019 to October 2022. The reference population was represented by patients diagnosed with neuroendocrine tumours who underwent at least two consecutive 99m TcEDDA/HYNIC-TOC SRS in our centre.
The inclusion criteria for the study were the following: (1) a histopathologically confirmed neuroendocrine tumour (either from biopsy or surgical procedures), with clearly established values for the ki67 proliferation index; (2) at least two lesions presenting the uptake of 99m TcEDDA/HYNIC-TOC on the baseline SPECT/CT scan; (3) at least two consecutive SPECT/CT scans within a 4-12 month timeframe; (4) information on the acquisition protocol for both baseline and follow-up scans (measured prepared activity, time of measurement, injection time and residual syringe activity).
All patients were following treatment with long-acting release somatostatin analogues (SSA), either octreotide at a dosage of 20-40 mg or lanreotide in a dosage of 60-120 mg, both were administered subcutaneously at an interval of 28 days. Patients were requested to interrupt their administration at least 3-4 weeks prior to the investigation.
Considering the selection criteria and the relatively low incidence of these types of tumours, 14 patients eventually included in our study.

99m Tc-EDDA/HYNIC-TOC Radiochemical Purity
The radiochemical purity of the 99m Tc-EDDA/HYNIC-TOC was checked using radio thin-layer chromatography (radio-TLC) (POLATOM, Otwock, Poland) before each patient administration, as radio-TLC is preferred over radio-high-performance liquid chromatography due to its simplicity and rapidity [24]. The radiopharmaceutical was administered only if its measured radiochemical purity was higher than 95%.

Acquisition Technique and Reconstruction
Patient preparation included increased hydration, light food intake and mild laxatives at least 24 h before the examination.
For each scan, patients received an intravenous dose of 628-740 MBq (17-20 mCi) of 99m Tc-EDDA/HYNIC-TOC. The examination protocol included a "whole-body" (WB) scan performed 2 h after the radiotracer's administration and a SPECT/CT exam 4 h post-injection. The gamma camera used for the exams was a dual-head Discovery 670 DR SPECT/CT system (General Electric Healthcare, Chicago, Illinois, United States of America) with low energy high resolution (LEHR) collimators. The planar scan was acquired using the following parameters: a pallet speed of 10 cm/min, a 128 × 128 matrix size, and a 99m Tc energy window. SPECT/CT imaging (thoracic, abdominal and pelvic regions) was executed using a standard protocol: a 128 × 128 matrix size, a step and shoot rotation time of 30s/view (for a total of 120 views), and a dual energy window (140 keV ± 10% and 120 keV ± 5%). After the SPECT acquisition, a low-dose CT scan was performed maintaining the patient in the same position, at 120 keV, 40-340 mA. CT images were acquired using a standard 3.75 slice thickness, then reconstructed using the standard (slice thickness of 2.5 mm), bone plus (1.25 mm) and lung (1.25 mm) reconstruction filters provided by the vendor. The acquired SPECT/CT data were reconstructed using the Q.Volumetrix software provided by the manufacturer, using the ordered subset expectation maximization (OSEM) iterative reconstruction algorithm (8 subsets and 10 iterations), with attenuation correction, scatter correction and resolution recovery performed.

Scanner Calibration
Scanner calibration was performed to allow better and more precise measurements of 99m TcEDDA/HYNIC-TOC uptake in the targeted lesions. The SPECT/CT scanner was calibrated by scanning a uniform Jaszczak phantom without any inserts inside and a NEMA phantom with hot inserts, calculating the planar and SPECT sensitivity. The conversion factor for the sensitivity calculated between the two aquisitions showed a deviation in the range of 0-5%. According to GE Healthcare, accuracy is calculated based on the following equation: where A c,sph,j is the true activity concentration in the spheres and a c,shp,j is the activity concentration measured in the reconstructed spheres. The measurement accuracy for 10 iterations and 8 subsets was higher than the 90% value recommended by the camera vendor (General Electric Healthcare).

Image Analysis
The images were interpreted independently by two experienced physicians on a dedicated workstation for imaging diagnosis (GE Xeleris V, General Electric Healthcare). Normal physiological 99m Tc-EDDA/HYNIC-TOC uptake was visually established in organs, using the CT images and clinical correlations. Pathological uptake was assessed in the primary lesions and the sites of the metastases. A quantitative analysis was performed in the reconstructed images using the software tool provided by the camera's vendor (GE Q.Volumetrix, GE Xeleris V, General Electric Healthcare). The delineation of the volumes of interest (VOI) were drawn semiautomatically, and SUV max and SUV peak based on lean body mass (SUV lbm ) were the elected methods in calculating the radiotracer's uptake. SUV max values for normal tissue (liver, spleen and bone) were obtained by positioning a sphere of 10 mm radius inside imagistically normal tissue. We have chosen to utilise SUV lbm as it has long been recommended in practical use because it excludes the percentage of a patient's body fat from its equation, which would raise the possibility of SUV calculation errors [25].
Quantitative analysis statistics are calculated using the following formulas [26]: Measure Activity = preinjection activity Measured Time = preinjection time As precise SPECT/CT criteria for the characterization of treatment response have yet to be developed, we decided to adapt the PERCIST 1.0 criteria for our study [27]. Thus, treatment efficacy or disease progression were evaluated by comparing the summed SUV max values obtained from up to five lesions between the baseline and the follow-up examinations, the results being classified as: complete response (CR)-no uptake in any of the baseline lesions and no new lesions on the follow-up study; partial response (PR)-a decrease >30% in the summed SUV max value on the follow-up study; stable disease (SD)-a decrease or increase in the summed SUV max value <30%; and progressive disease (PD)new tumoral foci or an increase >30% in the summed SUV max value on the follow-up study. For a better assessment of our results, we used radiology reports (either contrast-enhanced CT or magnetic resonance imaging) as a standard of reference for each patient, in correlation to both the baseline and the follow-up scans.

Statistical Analysis
Data analysis was performed using IBM SPSS Statistics Version 26 (IBM, SPSS, Inc., Chicago, IL, USA, 2019). Data were expressed as mean ± standard deviation (SD) and percentages. A paired samples T-test was used to compare the results. Correlations between ki67 and SUV values, as well as between SUV max and SUV peak , were made using a Pearson correlation test. A p value < 0.05 was considered as statistically significant.

Results
We evaluated 35 studies (14 baseline scans and 21 follow-up scans) from 14 patients with a mean age 48.36 ± 16.93 years old (range 18-69) who had undergone a neuroendocrine scintigraphy with SPECT/CT in our clinic. Most of our patients were males (64.3%), while the most frequent neuroendocrine tumour type was pancreatic NET (42.9%), followed by NETs of the small bowel (21.4%) and pulmonary NETs (21.4%). All of our patients had metastatic disease at the inclusion in the study, the most frequent site for metastases being the liver (n = 11, 78.6%), followed closely by the lymphatic system (n = 10, 71.4%). The main patient characteristics are presented in Table 1. All patients underwent treatment with long-acting somatostatin analogues between the baseline and the follow-up scan. The mean number of treatment cycles received by our patients between two consecutive SPECT scans was 7.76 ± 3.08 (range 3-11 cycles).
All of our patients had well differentiated tumours (NET G1-42.9%, NET G2-50%), with the exception of one, who presented a poorly differentiated NET (G3), with a ki67 index of 25%. The mean ki67 proliferation index in our patient group was 7 ± 6.25. The main neuroendocrine immunohistochemical and biochemical markers are presented in Table 2. A total of 123 lesions were analysed, out of which 64 liver metastases (52%), 25 metastatic lymph nodes (20.3%), 7 bone metastases (5.7%) and 27 (22%) were lesions with other origins (primary tumours, peritoneal metastases, pleural metastases). SUV max and SUV peak were extracted for each lesion.
We performed a separate analysis regarding the 99m Tc-EDDA/HYNIC-TOC uptake in the most frequent metastatic sites, comparing them to the uptake normal tissue (Table 3). Mean SUV max values were higher in liver and lymph nodes metastases (12.44 ± 7.76 g/mL, respectively 11.98 ± 10.45 g/mL) compared to the bone lesions, which presented a mean SUV max of 5.90 ± 3.68 g/mL, probably correlated to the lower SSTR expression in osseous tissue. SSTR-expressing malignant and non-malignant tissues showed statistically significant differences for all three types of the analysed metastases: SUV max in metastases were higher than SUV max in health tissue (p < 0.05 for all). Target lesion to normal tissue (TL/N) ratios of liver and bone metastases were generally high: 4.20 ± 2.67 for liver lesions and 9.37 ± 2.91 for bone lesions (with normal vertebrae as healthy tissue). The TL/N ratio for lymph metastases was relatively low compared to the other two (1.84 ± 1.53), as spleen was used as a reference for normal tissue and the physiological uptake of 99m Tc-EDDA/HYNIC-TOC is relatively high (the mean SUV max in spleen was 8.01 ± 5.31). The calculation of TL/N ratios in lymph node metastases was done in nine patients, as one of them underwent a splenectomy as part of their surgical treatment.
The mean SUV peak was in accordance with the SUV max values, observing higher SUV peak values in hepatic and lymphatic lesions (10.15 ± 6.79 g/mL, respectively 8.81 ± 7.99 g/mL), than in bone metastases (5.55 ± 2.47 g/mL) ( Table 4). Abbreviations: SUV-standardized uptake value; r-a correlation coefficient. p < 0.05 was considered as statistically significant.
SUV max and SUV peak of the three metastatic sites showed a statistically significant positive correlation (p < 0.05 in all three cases), meaning that SUV peak increases similarly to the elevation of SUV max in NET lesions (Figure 1).
EDDA/HYNIC-TOC is relatively high (the mean SUVmax in spleen was 8.01 ± 5.31). The calculation of TL/N ratios in lymph node metastases was done in nine patients, as one of them underwent a splenectomy as part of their surgical treatment.
SUVmax and SUVpeak of the three metastatic sites showed a statistically significant positive correlation (p < 0.05 in all three cases), meaning that SUVpeak increases similarly to the elevation of SUVmax in NET lesions (Figure 1). Abbreviations: SUV-standardized uptake value; r-a correlation coefficient. p < 0.05 was considered as statistically significant. A correlation analysis between the SUVmax and SUVpeak and the ki67 proliferation index was performed in order to determine whether 99m Tc-EDDA/HYNIC-TOC uptake varies with the tumoral differentiation grade, but our results invalidated this hypothesis, as no statistically significant correlation was found between the three parameters. The results can be observed in Table 5. Table 5. Correlation between ki67, SUVmax and SUVpeak in different tumoral sites. A correlation analysis between the SUV max and SUV peak and the ki67 proliferation index was performed in order to determine whether 99m Tc-EDDA/HYNIC-TOC uptake varies with the tumoral differentiation grade, but our results invalidated this hypothesis, as no statistically significant correlation was found between the three parameters. The results can be observed in Table 5. For the assessment of the utility of quantification of SPECT/CT data in monitoring the patients diagnosed with NETs, we evaluated a baseline scan and at least one follow-up scan for each patient, resulting in a total of 35 examinations that were finally included in our study. The mean clinical follow-up period was 8.76 ± 3.08 months (range 4-12 months). Four of our patients had more than just one follow-up scan, at the recommendation of their clinician; however, for their inclusion in a response category, only the conclusion of the last performed scan was counted. The patient distribution after follow-up evaluation is shown in Figure 2.
For the assessment of the utility of quantification of SPECT/CT data in monitoring the patients diagnosed with NETs, we evaluated a baseline scan and at least one followup scan for each patient, resulting in a total of 35 examinations that were finally included in our study. The mean clinical follow-up period was 8.76 ± 3.08 months (range 4-12 months). Four of our patients had more than just one follow-up scan, at the recommendation of their clinician; however, for their inclusion in a response category, only the conclusion of the last performed scan was counted. The patient distribution after follow-up evaluation is shown in Figure 2.   Out of the four patients that had multiple follow-up scans, one had three follow-up examinations, with a 6-month timeframe for each of them, and presented continuous PD on all of them. The second patient, who also had three follow-up scans, taken 6 months apart each, exhibited SD on the first two scans, but was finally classified as PR on the final examination. The third patient, who also had three follow-up SPECT/CTs, but at irregular intervals (6-11 months), was classified as PD on the first follow up scan, reclassified as PR on the second follow-up examination, and was finally assessed as SD on the last one. The fourth and last patient had two follow-up examinations performed, the first one 12 months after the baseline scan, being assessed as PD, and the second one 16 months after the first scan, where they were classified as SD. Out of the four patients that had multiple follow-up scans, one had three follow-up examinations, with a 6-month timeframe for each of them, and presented continuous PD on all of them. The second patient, who also had three follow-up scans, taken 6 months apart each, exhibited SD on the first two scans, but was finally classified as PR on the final examination. The third patient, who also had three follow-up SPECT/CTs, but at irregular intervals (6-11 months), was classified as PD on the first follow up scan, reclassified as PR on the second follow-up examination, and was finally assessed as SD on the last one. The fourth and last patient had two follow-up examinations performed, the first one 12 months after the baseline scan, being assessed as PD, and the second one 16 months after the first scan, where they were classified as SD.

Discussion
Radiolabeled somatostatin analogues are valuable tools for both the diagnosis and therapy of NETs because of their capacity to bind to SSTR, which are usually overexpressed on the tumoral cellular surface [28]. Although 68 Ga-DOTA-PET/CT currently represents the gold standard in the diagnosis and monitoring of patients with NETs, SPECT/CT using 99m Tc-labelled somatostatin analogues could represent a good alternative in centres that do not have the benefit of PET/CT equipment or 68 Ge/ 68 Ga generators [21].
Quantifying SPECT/CT data has proven useful in other nuclear medicine branches, as with bone tumours or cardiac amyloidosis [29][30][31]. Our study aims to outline its utility in neuroendocrine patient workup, highlighting that, with a precise scanner calibration, it can be a useful tool in monitoring treatment response and disease progression.
The literature regarding quantitative SPECT/CT in NETs is relatively scarce, with few studies published to this point. Reilly et al. [14] conducted a study regarding the characterisation of physiological and neuroendocrine tumour uptake of 99m Tc-EDDA/HYNIC-TOC using SPECT/CT SUVs. They observed that SUVmax is higher in liver and lymph node metastases (21.8 ± 13.3, respectively 16.3 ± 7.6) than in bone lesions (12.9 ± 7), and certified that SUVmax in secondary lesions is notably higher than in normal tissue. Their results are in agreement with ours, as both the SUVmax and the SUVpeak obtained in our patient group for liver and lymphatic metastases (SUVmax 12.44 ± 7.76, respectively 11.98 ± 10.45; SUVpeak 10.15 ± 6.79, respectively 8.81 ± 7.99) were much more elevated than the values obtained for bone metastases (SUVmax 5.90 ± 3.68; SUVpeak 5.55 ± 2.47), and all SUVs for metastatic lesions were higher than the SUVs for normal tissues.

Discussion
Radiolabeled somatostatin analogues are valuable tools for both the diagnosis and therapy of NETs because of their capacity to bind to SSTR, which are usually overexpressed on the tumoral cellular surface [28]. Although 68 Ga-DOTA-PET/CT currently represents the gold standard in the diagnosis and monitoring of patients with NETs, SPECT/CT using 99m Tc-labelled somatostatin analogues could represent a good alternative in centres that do not have the benefit of PET/CT equipment or 68 Ge/ 68 Ga generators [21].
Quantifying SPECT/CT data has proven useful in other nuclear medicine branches, as with bone tumours or cardiac amyloidosis [29][30][31]. Our study aims to outline its utility in neuroendocrine patient workup, highlighting that, with a precise scanner calibration, it can be a useful tool in monitoring treatment response and disease progression.
The literature regarding quantitative SPECT/CT in NETs is relatively scarce, with few studies published to this point. Reilly et al. [14] conducted a study regarding the characterisation of physiological and neuroendocrine tumour uptake of 99m Tc-EDDA/HYNIC-TOC using SPECT/CT SUVs. They observed that SUV max is higher in liver and lymph node metastases (21.8 ± 13.3, respectively 16.3 ± 7.6) than in bone lesions (12.9 ± 7), and certified that SUV max in secondary lesions is notably higher than in normal tissue. Their results are in agreement with ours, as both the SUV max and the SUV peak obtained in our patient group for liver and lymphatic metastases (SUV max 12.44 ± 7.76, respectively 11.98 ± 10.45; SUV peak 10.15 ± 6.79, respectively 8.81 ± 7.99) were much more elevated than the values obtained for bone metastases (SUV max 5.90 ± 3.68; SUV peak 5.55 ± 2.47), and all SUVs for metastatic lesions were higher than the SUVs for normal tissues.
Piwowarska-Bilska et al. [32] researched multiple methods of calculating SUV max and SUV mean values in normal liver and liver metastases in SPECT/CT images with 99m Tc-EDDA/HYNIC-TOC. They obtained a mean SUV max value for liver metastases, corrected to lean body mass (lbm), of 27.7 ± 20.3, while the SUV max for normal tissue, calculated in small liver VOI, was 5.7 ± 1.4. Their values were higher than the ones we obtained, but still in agreement with our results, the differences being partly attributed to the dependency of SUV values on scanner calibration and image reconstruction, and also to the different segmentation methods used in their study.
To the best of our knowledge, this is the first study to evaluate the potential utility of quantitative SPECT/CT data in evaluating treatment response and disease progression in patients with neuroendocrine tumours. At present, 68 Ga-DOTA-PET/CT is considered the standard prognostic tool in NET patients, its data quantification being proven to predict response in PRRT [33][34][35]. A recent study, published by Lee et al. [36], evaluated the utility of SUV measurements on 68 Ga-DOTATATE PET/CT for predicting progressionfree survival on patients diagnosed with gastro-entero-pancreatic NETs treated with SSA. They concluded that SUV max on 68 Ga-DOTATATE PET/CT could predict the outcome of the treatment independently on the pathologic variables (tumoral grade, ki67 index and mitotic count), as they observed that 68 Ga-DOTATATE, as a specific ligand for SSTR2, is strongly correlated with a tumours' SSTR2 expression [36]. The same correlation was found by Miederer et al. [37], who correlated the SUV max of 68 Ga-DOTATOC PET/CT with the cell membrane-based SSTR2 immunohistochemistry score in 17 patients with NETs, and discovered a strong correlation between the membranous SSTR2 expression and 68 Ga-DOTATOC uptake. As 99m Tc-EDDA/HYNIC-TOC is also known to possess a high affinity for SSTR 2 [19,21], and considering our results, where no correlation between ki67 index and the values of SUV max and SUV peak was found, we can also confirm that SUV values calculated on 99m Tc-EDDA/HYNIC-TOC SPECT/CT can work as independent predictors of disease progression and treatment response.
The uptake pattern of 99m Tc-EDDA/HYNIC-TOC also correlates with the one observed on 68 Ga-DOTA-PET/CT. A complex study by Kroiss et al. [38] related that the highest physiological uptake of 68 Ga-DOTATOC can be observed in spleen (SUV max 32.6 ± 11.8), with lower values for liver (SUV max 12.5 ± 4) and a considerably low uptake for bones (SUV max 1.9 ± 0.8), which is in agreement with the numbers derived from our research: the highest SUV max , of 8.01 ± 5.31, for spleen, and a low SUV max in bone tissue of 0.66 ± 0.37. Furthermore, the values of the TL/N ratios we obtained (which correspond to the T/NT ratios on 68 Ga-DOTA-PET/CT), of 4.20 ± 2.67 for hepatic lesions and of 9.37 ± 2.91 for bone lesions, are close to the results published in their paper, a T/NT ratio for liver of 2.8 ± 1.6 and a T/NT ratio for bones of 10.50 ± 14.2.
In our research, we also analysed the effectiveness of using peak SUV as a diagnostic and prognostic tool. Various studies suggested that SUV peak measured on PET/CT could represent a better alternative to SUV max because it is less affected by image noise [39,40]. SUV peak is defined as the average SUV within a small, fixed size region of interest centred on a high uptake part of the tumour [40]. Regarding the SUV peak measured on SPECT/CT, little information has been documented. The only two studies found, those of Tzortzakakis et al. [41] and Ogura et al. [42] defined SUV peak as the maximum activity concentration in a 1cm 3 volume of a larger VOI drawn, and described its correlation to the other SUVs measured, SUV max and SUV mean . Our research also highlighted the strong positive correlation between the SUV max and SUV peak values (p < 0.05), suggesting that a change in SUV peak might also translate into disease progression or treatment response. We chose not to evaluate the SUV mean of the lesions in our studies because of the lack of accuracy in correctly delineating the volume of interest on native CT.
There are, however, some limitations to our study. Firstly, our patient group is comprised of only 14 individuals, which might affect the applicability of our results on a larger scale. Secondly, a standardization of quantitative SPECT/CT has not been well established yet, and although we have performed a rigorous scanner calibration according to the international norms and to the camera vendor's instructions, further studies need to be performed with regard to this matter. Finally, the variable follow-up period (a range of 4-16 months) might alter the accuracy of patient classification into PERCIST-based groups. Nevertheless, this one last point should have minor implications for our study, as our intention was not to assess the treatment response of our patients, but rather to evaluate the practicability of using quantitative SPECT/CT as a tool in patient monitorization.

Conclusions
The quantitative analysis of 99m TcEDDA/HYNIC-TOC SPECT/CT data in patients with neuroendocrine tumours for evaluating the treatment response could represent an alternative to 68 Ga-DOTA-peptides PET/CT for patient management, allowing the monitoring of disease evolution and guiding for treatment adjustment. Although the usage of PERCIST-based criteria showed good results, a standardization of the method based on more clinical studies is needed for applicability in clinical practice.