Prognostic value of a three-scale grading system based on combining molecular imaging with 68Ga-DOTATATE and 18F-FDG PET/CT in patients with metastatic gastroenteropancreatic neuroendocrine neoplasias

We investigated on the added prognostic value of a three-scale combined molecular imaging with 68Ga-DOTATATE and 18F-FDG PET/CT, (compared to Ki-67 based histological grading), in gastroenteropancreatic neuroendocrine neoplasia patients. 85 patients with histologically proven metastatic gastroenteropancreatic neuroendocrine neoplasias, who underwent combined PET/CT imaging were retrospectively evaluated. Highest Ki-67 value available at time of 18F-FDG PET/CT was recorded. Patients were classified according to World Health Organization/European Neuroendocrine Tumor Society histological grades (G1, G2, G3) and into three distinct imaging categories (C1: all lesions are 18F-FDG negative/68Ga-DOTATATE positive, C2: patients with one or more 18F-FDG positive lesions, all of them 68Ga-DOTATATE positive, C3: patients with one or more 18F-FDG positive lesions, at least one of them 68Ga-DOTATATE negative). The primary endpoint of the study was Progression-Free Survival, assessed from the date of 18F-FDG PET/CT to the date of radiological progression according to Response Evaluation Criteria In Solid Tumors version 1.1. Classification according to histological grade did not show significant statistical difference in median Progression-Free Survival between G1 and G2 but was significant between G2 and G3 patients. In contrast, median Progression-Free Survival was significantly higher in C1 compared to C2 and in C2 compared to C3 patients, revealing three distinctive imaging categories, each with highly distinctive prognosis. Our three-scale combined 68Ga-DOTATATE/18F-FDG PET imaging classification holds high prognostic value in patients with metastatic gastroenteropancreatic neuroendocrine neoplasias.


INTRODUCTION
GastroEnteroPancreatic NeuroEndocrine Neoplasias (GEP NENs) represent a challenging clinical entity with marked heterogeneity in terms of prognosis, behavior and evolution over time. Approximately half of the GEP NEN patients have already developed distant metastases at the time of primary staging [1] with liver being the predominant localization of tumor spread [2]. Surgical resection of the primary and the metastases, when feasible, remains the only curative treatment in patients with GEP NENs. Accurate assessment of the disease extent and the prognosis are crucial in order to optimize disease management.
The choice between available therapeutic modalities mainly relies on the histological assessment of tissue samples. Anatomo-pathological reports provide information about tumor size, extent of local invasion, presence of nodal metastases, but also differentiation status and tumor grade [3]. The tumor grade relies on both, the percentage of neoplastic cells expressing the Ki-67 protein (during the active phases of the cell cycle) and the mitotic count and has been shown to provide significant prognostic information [1,4]. Current grading classifications endorsed by World Health Organization/European Neuroendocrine Tumor Society (2010 for all GEP NENs [5] and 2017 for pancreatic NENs [6]) rely on the use of Ki-67 labeling index and/ or the mitotic rate. NENs showing a well-differentiated histology and a low proliferation rate have been called Neuroendocrine Tumors (NETs). They often present with an indolent behavior and most express somatostatin receptors (SSTR) on their surface [7]. In contrast, NENs displaying a poorly differentiated histology and a high proliferation rate are called Neuroendocrine Carcinomas (NECs). Their natural history is aggressive and they fail to express SSTR. Several issues regarding Ki-67 assessment have been raised, concerning mainly sampling errors due to tumoral heterogeneity and interobserver variability, which may provide conflicting information about grading [8,9].
Positron Emission Tomography (PET) imaging, by means of 68 Ga-DOTA-agonists and 18 F-FDG, represent a powerful nuclear medicine tool for management of patients with GEP NENs, targeting differentiation status (via overexpression of trans-membrane SSTR, specific to the neuroendocrine phenotype) and tumoral aggressiveness (via glycolytic metabolism, a nonspecific energetic pathway of neuroendocrine neoplasias). Combined imaging by those two tracers, may highlight the intimate relationship between SSTR expression and metabolism, allowing an in vivo whole-body phenotypic characterization of the disease.
The primary objective of this study was to develop an easy-to-implement imaging score for combined molecular imaging ( 68 Ga-DOTATATE/ 18 F-FDG PET) reflecting tumoral heterogeneity in patients with metastatic GEP NENs and to assess its potential prognostic value. The secondary objective was to compare its prognostic value with that of the Ki-67 based histological grade (as available at the date of PET imaging).

RESULTS
124 in total GEP NENs patients who underwent combined 68 Ga-DOTATATE and 18 F-FDG PET/CT imaging, between January 2008 and December 2018 within a maximum time window of three months between them, were screened prior to inclusion. 39 patients were excluded (no morphologically measurable target lesions: 12 patients, unknown primary NENs: 3 patients, surgery with curative intent after the two PETs: 19 patients, and second primary malignancies: 5 patients) and 85 patients Figure 1: 124 GEP NENs patients who underwent combined 68 Ga-DOTATATE and 18 F-FDG PET/CT imaging in our Institute (within a maximum time window of three months between them), were screened prior to inclusion. 85 patients were finally included and evaluated on the study. www.oncotarget.com were finally included and evaluated (Figure 1), including 2 patients that were lost to follow up (for which however, survival data were available).
The cohort characteristics are summarized in Table 1 According to the histological classification, 21 patients were graded as G1, 46 as G2 and 18 as G3. According to the combined imaging classification, 28 patients were C1, 46 were C2 and 11 patients were C3 ( Table 2). The agreement between the two classification systems was poor with a Kendall's Rank Correlation Coefficient τ of -0.003 [38/85 (44.7%)].
The median Progression-Free Survival (mPFS) and the median Overall Survival (mOS) of the entire cohort was 12.9 months and 40.1 months respectively ( Figure 2).
Regarding histological grade classification, the mPFS values of G1, G2 and G3 patients were 21.2 months, 17.9 months and 5.9 months respectively. We did not observe any statistical difference between G1

DISCUSSION
One of the landmarks of Neuroendocrine Differentiation is the overexpression of SSTR on the cellular surface, a particularly attractive target for imaging and therapy (the "theragnostic" approach) [10]. Imaging with radiolabeled somatostatin analogs (agonists), should be PETbased (e. g. 68 Ga-DOTA-peptides), taking into account the numerous advantages over the classical somatostatin receptor scintigraphy [11]. Accurate staging of NENs is crucial, and the reported pooled sensitivity and specificity of 68 Ga-DOTA PET imaging is 96% and 100%, respectively [11,12]. It has been clearly documented that the expression of SSTR correlates strongly to the degree of differentiation of NENs [13] and the intensity of tracer uptake in 68 Ga-DOTA PET correlates with SSTR expression [14]. Therefore, uptake is higher in well-differentiated NETs compared to poorly-differentiated NECs. In addition, the uptake in 68 Ga-DOTA PET measured by maximun Standardized Uptake Value (SUVmax), holds a high prognostic value, as values of 19.3 or more, correlate with improved OS compared to those with poorer uptake [14]. Sharma et al [15], reported a lower cut-off in 68 Ga-DOTA PET uptake, still with similar prognostic significance. Finally and as a consequence, SSTR imaging can select patients that qualify for treatment with somatostatin analogues and Peptide Receptor Radionuclide Therapy (PRRT).
Traditionally, 18 F-FDG PET has been used for staging purposes in poorly-differentiated NECs or in case of negative SSTR imaging [16,17], as poorlydifferentiated NENs lose their ability to express SSTR. However, several prospective data underlined its strong independent prognostic value in NENs, even in those with low histological grade. In the pioneering study of Garin et al [18,19], 36 patients with documented welldifferentiated metastatic GEP NENs were offered a period of watch-and-wait before treatment initiation. Fifteen patients had a positive 18   Legend: The agreement between the two classification systems was poor results which were confirmed prospectively by Sansovini et al [21] and Nilica et al [22]. Finally, Johnbeck et al [23] assessed prospectively the prognostic value of 18 F-FDG PET in 100 NENs patients (77 with GEP NENs). Patients with a positive 18 F-FDG PET had a significantly worse prognosis (in terms of PFS and OS) than patients with a negative 18 F-FDG PET. Ezziddin et al [24] reviewed retrospectively data from 89 patients with metastatic GEP NENs and identified three distinctive prognostic groups based on the ratio of SUVmax of the lesion with the highest 18 F-FDG uptake to the SUVmean of the normal hepatic parenchyma (ratio≤1; 1<ratio<2.3; ratio≥2.3). These groups were associated with significant differences in OS (mOS not reached after 114 months of follow up versus 55 months versus 13 months). That being said, 18 F-FDG PET must be considered not only as a staging, but also (and perhaps most importantly), as a grading whole-body imaging tool, in which 18 F-FDG PET positive lesions had a significant impact on prognosis, regardless of the expression of SSTR.
One issue that should be validated in larger studies is the definition of 18 F-FDG positivity. In our cohort, we define as 18 F-FDG positive any lesion with tracer uptake superior to the uptake of local background. In the study of Garin et al [18,19], an SUVmax cutoff of 4.5 was used to define 18 F-FDG positivity, whereas Sansovini et al [21] applied an arbitrary SUVmax cutoff of 2.5 separating 18 F-FDG positive from 18 F-FDG negative lesions. Whether low 18 F-FDG uptake lesions represent an evolution step toward overt high uptake lesions or they just reflect a different pathophysiological underlying process (such as hypoxia [25]), remains to be investigated.
Combined molecular imaging with 68 Ga-DOTA PET and 18 F-FDG PET assesses simultaneously the tumor biology (SSTR expression and glycolytic metabolism). Different lesions can exhibit different degrees of tracer uptake and hence different degrees of differentiation and aggressiveness. These patterns of tumor heterogeneity reflect the precise phenotype of GEP NENs at any given moment in the course of the disease. Although both PETs provide complementary information, the major decision-making examination is 18 F-FDG PET, (alias the metabolic fingerprint of the disease) and chances to encounter 18 F-FDG-avid lesions or a mismatch in favor of 18

F-FDG, increase with tumor grade.
We developed an easy-to-implement three-scale grading score for combined molecular imaging ( 68 Ga-DOTATATE and 18 F-FDG PET) based on the spatial distribution of the lesions and the relative uptake of the respective tracers. We classified our cohort into three distinctive imaging categories, each with highly distinctive prognosis. This was in contrast to the classification based on the histological grade, obviously due to the size of our cohort, since high volume epidemiological studies have validated the actual histological grades. Ki-67 based histological grade plays a prominent role in NENs management as it is worldwide, the most common tissuebased prognostic biomarker and has become essential for guiding therapy [26,27]. However, if the biopsies leading to Ki-67 indexes determination are not PET-guided, they might be subject to sampling error, leading thus to underscores which are not reflecting necessarily the real tumor aggressiveness [28]. Thirteen patients in our cohort (15%), histologically labelled as G1, had 18 F-FDG-avid lesions, but this percentage in low grade NETs can be as high as 40% [20,22]. Moreover, Ki-67 index may evolve in the same patient over time and site [29,30], affecting not only treatment decision but also disease prognosis [30]. In some cases, Ki-67 indexes in our cohort were as old as 10 years. If dictated by clinical scenarios, combined PET imaging should be repeated over time, in order to detect and reveal aggressive 18 F-FDG-avid lesions. An accurate selection of (re-) biopsy sites can be the link between PETdriven biopsies and biopsy-driven treatments, optimizing NENs management and helping selection between different therapeutic options and strategies.
Chan et al [31], on their seminal study, classified 62 patients who performed 68 Ga-DOTATATE and 18 F-FDG PET within 31 days of each other, into five imaging categories (P1-P5), introducing the NETPET score. P1 indicated purely 68 DOTATATE-avid disease without 18 F-FDG uptake in any lesion (corresponding to our C1 category), while P5 indicated the exact opposite, pure 18  These two imaging classification systems (the NETPET and the one presented in our article) combine the biological information provided by the two PETs into a single classification scheme with high prognostic significance, with the ultimate purpose to identify mismatches in favor of 18 F-FDG, as they represent the most aggressive phenotype of the disease. Nevertheless, the three-scale classification is easier to implement, was tested into a larger cohort of patients and provided a complete survival information, not only in terms of PFS but also in terms of OS.
Besides their prognostic significance, these systems may have further implications on therapeutic management. C1 patients have a low-grade and indolent disease. In contrast, C3 patients have high-grade, metabolically active disease and therefore warrant aggressive treatment. PRRT is suitable for C1 patients and in C2 patients as well. In C3 patients, the 18 F-FDG positive/ 68 Ga-DOTATATE negative part of the disease will be less affected by beta-irradiation of the therapeutic radio-isotopes (for instance by cross-irradiation from neighbouring 68 Ga-DOTATATE positive lesions, if any), or not at all. PRRT therefore is not suitable for that patient category. Emerging data suggest the use of PRRT in combination with chemotherapeutic agents (such as temozolomide/capecitabine) in spatially concordant 18 F-FDG positive disease [32,33], in an attempt to boost the therapeutic efficacy of the former. Hence, Peptide Receptor Chemo-Radionuclide Therapy (PRCRT), can achieve unexpectedly long PFS, modifying the poor prognosis associated with 18 F-FDG-avidity [34].
In summary, the combined 68 Ga-DOTATATE/ 18 F-FDG PET imaging classification presented on this article represent the precise phenotype of GEP NENs at any given moment of the disease and holds high prognostic value, compared to classification based on the histological grade. Its value as prognostic imaging biomarker should be further confirmed within prospective trials and tested multi-centrically to establish inter-rater reliability.

Patient cohort and study design
Patients with histologically confirmed metastatic GEP NENs who underwent in our center combined imaging with 68 Ga-DOTATATE and 18 F-FDG PET/ CT within a maximum window of 3 months were retrospectively included on the study. Patients with no morphologically measurable target lesions according to Response Evaluation Criteria In Solid Tumors (RECIST version 1.1) [35], patients with unknown primary NENs,  patients who underwent surgery with curative intent after the two PETs and patients with second primary malignancies (or history of other active malignant disease) unless in remission for at least 5 years were excluded. Ki-67 index available at the day of 18 F-FDG PET/CT was considered for patient categorization. The highest value of Ki-67 was considered if several values were available. The oldest pair of PETs was considered if more than one imaging pair was available. The study was approved by the local Ethics Committee of Jules Bordet Institute (CE2531).
The primary endpoint of the study was Progression-Free Survival (PFS), defined as from the date of The secondary endpoint of the study was Overall Survival (OS), defined as from the date of 18 F-FDG PET/ CT to the date of death or the date of the last follow-up.

Imaging
All 18 F-FDG and 68 Ga-DOTATATE PET/CT images were acquired at the Nuclear Medicine Department of Jules Bordet Institute using a General Electric (GE-Healthcare) Discovery 690 Time of Flight (TOF) PET system. Before 18 F-FDG injection, patients had to fast for more than 6 hours and blood glucose level had to be lower than 150 mg/dL. Long acting somatostatin analogs were discontinued at least 4 weeks prior to 68 Ga-DOTATATE PET/CT acquisition.
Whole CT was performed with 64 slices helical scanner (VCT; GE Medical Systems). The tension was 120 kV, and the current was modulated by the Auto-mA software with a noise index of 30 (range: 30-200 mA) and the Adaptive Statistical Iterative Reconstruction (ASIR) algorithm. The other CT acquisition parameters were 0.5 s per CT rotation and a pitch of 0.98. The CT images were reconstructed with ASIR algorithm set at 40%, with a matrix of 512c×c512 (0.97c×c0.97 mm pixel size) and a slice thickness of 2.5cmm. The PET matrix was 192c×c192 pixels of 2.73c×c2.73 mm with a slice thickness of 3.27 mm.

Patient classification systems
Patients were divided according to histological grade [5,6] into three distinctive categories: G1, G2 and G3. The imaging classification was based upon the spatial distribution of the lesions and the relative uptake of the respective tracers. Anonymized PET image-sets were automatically co-registered anatomically, displayed simultaneously in transverse, sagittal and coronal planes and initially windowed with preset values for Standardized Uptake Value (SUV) of 0-15 for 68 Ga-DOTATATE PET and SUV of 0-7 for 18 F-FDG PET (AW Server 3.2, GE Healthcare). In both PETs, a lesion was considered as positive if tumoral uptake was superior to the local background. Patients were therefore divided into three distinct imaging categories: C1 (all lesions are 18 F-FDG negative and 68 Ga-DOTATATE positive, Figure 4), C2 (patients with one or more 18 F-FDG positive lesions, all of them 68 Ga-DOTATATE positive, Figure 5) and C3 (patients with one or more 18 F-FDG positive lesions, at least one of them 68 Ga-DOTATATE negative, Figure 6). Each pair of PETs was classified by two experienced nuclear medicine physicians into one of the three aforementioned categories (reporting was performed simultaneously). Radiological progression according to RECIST 1.1 was assessed without knowledge of the respective histological or imaging classification.

Statistical analysis
Assessments were performed using GraphPad Prism 7 (GraphPad Software Inc., California USA), with a statistical significance level at p ≤ 0.05. Kaplan-Meier survival curves were constructed and compared using the log-rank test. The agreement between histological grade and combined PET imaging was assessed using the Kendall's Rank Correlation Coefficient.