PET Criteria by Cancer Type from Imaging Interpretation to Treatment Response Assessment: Beyond FDG PET Score

Background: in recent years, the role of positron emission tomography (PET) and PET/computed tomography (PET/CT) has emerged as a reliable diagnostic tool in a wide variety of pathological conditions. This review aims to collect and review PET criteria developed for interpretation and treatment response assessment in cases of non-[18F]fluorodeoxyglucose ([18F]FDG) imaging in oncology. Methods: A wide literature search of the PubMed/MEDLINE, Scopus and Google Scholar databases was made to find relevant published articles about non-[18F]FDG PET response criteria. Results: The comprehensive computer literature search revealed 183 articles. On reviewing the titles and abstracts, 149 articles were excluded because the reported data were not within the field of interest. Finally, 34 articles were selected and retrieved in full-text versions. Conclusions: available criteria are a promising tool for the interpretation of non-FDG PET scans, but also to assess the response to therapy and therefore to predict the prognosis. However, oriented clinical trials are needed to clearly evaluate their impact on patient management.


Introduction
Positron emission tomography (PET) and PET/computed tomography (PET/CT) have emerged in recent years as pivotal tools for the non-invasive assessment of a high number of pathological conditions. The most used tracer to perform this imaging modality is [ 18 F]fluorodeoxyglucose ([ 18 F]FDG) which can evaluate the glycolytic activity of cells [1,2], but other radiopharmaceuticals that explore different metabolic pathways or localize to particular targets because of specific binding interactions are available nowadays. The role of PET/CT imaging for the assessment of neoplasia has emerged in several steps of oncological patients' work-up, such as in the diagnosis, staging, re-staging after therapy and follow-up. During the last 10 years, the use of many criteria has emerged in order to standardize the initial interpretation of oncological disease or their response after or during therapy. Imaging criteria can help in the interpretation of the scan, guiding the reader to better define its status or diagnosis. Moreover, since PET/CT is able to reflect different metabolic pathways of the tissues, functional changes can occur early in the course of therapy, preceding reduction in the size of tumors. Therefore, PET/CT can also be useful for the assessment of response to a specific therapy, thus guiding the future diagnostic and therapeutic work-up of the patient [3]. Different approaches measuring the response rate of neoplasms with morphological imaging modalities have been classically developed, such as the "Response Evaluation Criteria in Solid Tumors" (RECIST) [4]. Pointing our attention to [ 18 F]FDG PET/CT, some specific response criteria have been formulated to improve its diagnostic predictive value, such as the EORTC, "PET/CT Criteria for early prediction of Response to Immune checkpoint inhibitor Therapy" (PECRIT), the "PET Response Evaluation Criteria for ImmunoTherapy" (PERCIMT) or the "PET Response Criteria in Solid Tumors" (PERCIST) [4][5][6][7]. In this scenario most of the PET criteria that have been proposed in the literature focus on [ 18 F]FDG PET/CT, while only a small number of them were thought suitable for other radiopharmaceuticals. The aim of this systematic review is, therefore, to collect and discuss PET criteria developed for non-[ 18 F]FDG probes in the oncological setting both for imaging interpretation and treatment response assessment.

Research Strategy
A bibliographic literature search up to 30 November 2022 was performed on three electronic databases (PubMed/MEDLINE, Scopus, Google Scholar) in order to find articles concerning the use of non-[ 18 F]FDG PET criteria for the assessment of solid tumors. The search algorithms were different combinations of the following terms: "PET criteria", "PET score", "positron emission tomography", "imaging interpretation", "response", "post-treatment evaluation criteria", "osteosarcoma", "Ewing sarcoma", "bone tumor", "brain", "glioma", "meningioma", "breast cancer", "pancreatic carcinoma", "gastric cancer", "esophageal cancer", "anal cancer", "GIST", "cervical cancer", "bladder cancer", "head and neck squamous cell carcinoma", "head and neck carcinoma", "lung cancer", "NSCLC", "small cell lung cancer", "mesothelioma", "pleural mesothelioma", "medullary thyroid cancer", "thyroid cancer", "anaplastic thyroid cancer", "papillary thyroid cancer", "adrenal gland tumor", "neuroendocrine tumor", "neuroendocrine", "paraganglioma", "prostate", "prostate cancer", " 18 F", " 68 Ga", " 11 C", "MET", "FET", "DOPA", "FCH", "choline", "DOTA", "DOTA-TOC" "DOTATATE", "DOTANOC", "PSMA" and "NaF". To identify supplementary eligible articles, the references of the retrieved articles were also screened for additional papers. Two reviewers (A.L. and J.G.) screened, retrieved, and selected data from each report. Original articles edited in the English language and performed on humans which evaluated non-[ 18 F]FDG PET criteria for the evaluation of solid cancers were finally included in this review. Exclusion criteria were the assessment of [ 18 F]FDG PET criteria and the assessment of semiquantitative parameters without the production of criteria. Review, meta-analysis, conference proceedings, case reports and case series were also excluded from the present analysis. The included studies were finally divided per cancer type classification, and they were also distinguished, if necessary, between interpretation and response assessment criteria in order to compare and evaluate mean differences between studies. From the included studies, the following information was extracted: characteristics of the studies (first author, year of publication), non-[ 18

Research Strategy Results
The Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) flowchart of research strategy and studies selection is summarized in Figure 1. The literature search revealed 183 articles; among them, a total of 149 were excluded after reviewing titles, abstracts, and full texts because the reported data were not within the field of interest of this review. Therefore, 34 studies were considered suitable for the analysis and subsequently divided following different cancer types: Prostate Cancer (PCa) in 21 studies, Neuroendocrine tumors (NET) in 10, primary bone cancer in 2, and brain tumors in 1.
The included studies were finally divided per cancer type classification, and they were also distinguished, if necessary, between interpretation and response assessment criteria in order to compare and evaluate mean differences between studies. From the included studies, the following information was extracted: characteristics of the studies (first author, year of publication), non-[ 18 F]FDG PET criteria/score name, description of non-[ 18 F]FDG PET criteria/score, type of cancer, radiopharmaceutical, sample size, and main findings.

Research Strategy Results
The Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) flowchart of research strategy and studies selection is summarized in Figure 1. The literature search revealed 183 articles; among them, a total of 149 were excluded after reviewing titles, abstracts, and full texts because the reported data were not within the field of interest of this review. Therefore, 34 studies were considered suitable for the analysis and subsequently divided following different cancer types: Prostate Cancer (PCa) in 21 studies, Neuroendocrine tumors (NET) in 10, primary bone cancer in 2, and brain tumors in 1. Finally, the research revealed 15 different non-[ 18 F]FDG PET criteria: the European Association of Nuclear Medicine (EANM) criteria, the Prostate-Specific Membrane Antigen-Reporting and Data Systems (PSMA-RADS), the PROMISE (miTNM) criteria, the E-PSMA reporting system, the Pro-PET score, and the PRIMARY for PSMA PET imaging interpretation of PCa; PSMA PET Progression (PPP) and RECIP 1.0 as treatment response evaluation criteria for PCa. For NET, the PET-based Krenning score (KS), the somatostatin receptor (SSTR)-RADS and the NETPET grade were proposed for SSTR PETbased imaging interpretation and the selection of eligible patients for [ 177 Lu]Lu-DOTA radioligand therapy (RLT), whereas MORE and ZP score were recently proposed for the assessment of response to RLT in NET patients. In primary bone cancer, the NAFCIST Finally, the research revealed 15 different non-[ 18 F]FDG PET criteria: the European Association of Nuclear Medicine (EANM) criteria, the Prostate-Specific Membrane Antigen-Reporting and Data Systems (PSMA-RADS), the PROMISE (miTNM) criteria, the E-PSMA reporting system, the Pro-PET score, and the PRIMARY for PSMA PET imaging interpretation of PCa; PSMA PET Progression (PPP) and RECIP 1.0 as treatment response evaluation criteria for PCa. For NET, the PET-based Krenning score (KS), the somatostatin receptor (SSTR)-RADS and the NETPET grade were proposed for SSTR PET-based imaging interpretation and the selection of eligible patients for [ 177 Lu]Lu-DOTA radioligand therapy (RLT), whereas MORE and ZP score were recently proposed for the assessment of response to RLT in NET patients. In primary bone cancer, the NAFCIST emerged as a response assessment criterion with [ 18 F]sodium-fluoride ([ 18 F]NaF) PET in patients affected by osteosarcoma or osteosarcoma-like tumor who underwent Radium 223 dichloride ( 223 RaCl 2 ) therapy. Finally, the FuMeGa score was proposed to evaluate post-surgery response in high-grade glioma (HGG) patients with [ 18 F]Fluorocholine.
The proposed criteria per cancer type are reported in Figure 2 and detailed in Table 1. Data about the studies included in the review and their main findings are summarized in Table 2. emerged as a response assessment criterion with [ F]sodium-fluoride ([ F]NaF) P patients affected by osteosarcoma or osteosarcoma-like tumor who underwent Rad dichloride ( 223 RaCl2) therapy. Finally, the FuMeGa score was proposed to evaluate surgery response in high-grade glioma (HGG) patients with [ 18 F]Fluorocholine.
The proposed criteria per cancer type are reported in Figure 2 and detailed in 1. Data about the studies included in the review and their main findings are summa in Table 2.   The criteria are structured as follows: first, "anomalous" findings, defined as suggestive radiotracer uptake above physiologic background, are recorded. Thus, all these sites are classified as "pathologic" for PCa, unless another explanation is possible. Third, the anatomic localization is considered (up to 5 lesions). However, in the same document, the authors observed a moderate interobserver agreement among multiple readers. Namely, inter-reader agreement for the presence of anomalous findings was 0.47 but became substantial when readers judged the anomalous findings as suggestive for a pathologic, uncertain, or non-pathologic image (K's alpha: 0.64) [8].
The second proposed interpretation criterion, termed PSMA-RADS Version 1.0, was introduced by Rowe et al. in 2018 [9]. The PSMA-RADS can be applied on both individual target lesions (maximum of 5 per-scan) or on the overall impression of the imaging study. As shown in Table 3, PSMA-RADS is based on a 5-point scale that reflects the confidence of the interpreting imaging specialist that a given lesion represents a site of PCa, scoring from PSMA-RADS-1 (=definitively benign) to PSMA-RADS-5 (=high degree of certainty that PCa is present).       Lesion targetable: biopsy is suggested. Alternatively, follow-up imaging (either anatomic or PSMA-targeted PET/CT) after 3-6 months is recommended.

PSMA-RADS-3B
Equivocal uptake in bone lesion not definitive but also not atypical of PCa on anatomic imaging.
Comparison to other imaging modalities (bone scan, [ 18 F]NaF PET, or tumor-protocol MRI images) or bone biopsy. Alternatively, follow-up imaging (either anatomic or PSMA-targeted PET/CT) after 3-6 months is recommended.

PSMA-RADS-3C
Lesions that would be atypical for PCa but have high levels of uptake. The likelihood of non-prostatic malignancy or another benign tumor is high.
Biopsy to histologically confirm the diagnosis is often preferred. Alternatively, organ-specific follow-up imaging may be done (e.g., liver-protocol MRI to evaluate possible primary hepatocellular carcinoma).

PSMA-RADS-3D
Lesion suggestive of malignancy on anatomic imaging but lacking uptake.
Consider non-prostatic malignancy, neuroendocrine PCa, and an uncommon case of prostate adenocarcinoma that fails to express PSMA. Biopsy to histologically confirm the diagnosis is often preferred; alternatively, organ-specific follow-up imaging may be done.

PSMA-RADS-4 (Pca highly likely)
Intense uptake in sites typical of PCa but lacking definitive findings on conventional imaging.
No biopsy confirmation will be needed, although tissue for genomic analysis (or other purposes) may be useful. PSMA-RADS-5 (Pca almost certainly present) Intense uptake in sites typical of PCand having corresponding findings on conventional imaging.
No biopsy confirmation will be needed, although tissue for genomic analysis (or other purposes) may be useful.  58.7% of these lesions resulted true-positive for PCa. Moreover, the authors showed that PSMA-RADS-3A lesions were more likely to represent PCa than PSMA-RADS-3B ones [25]. Consistent with this evidence, another study reported that debatable lesions proved to have no clinical relevance in 84.6% of cases, and only 11% of equivocal PSMA-RADS-3B bone lesions were true positive [26]. In addition, Bhoil et al. reported that in the skeletal system only a minority of equivocal lesions findings will have characteristic changes in PCa involvement on follow-up imaging [27]. In the context of indeterminate lesions, an elevated prostate-specific antigen (PSA) value was shown in 81.9% of patients with true positive PSMA-RADS-3A lesions [28]. In this setting, the point-spread function (PSF) reconstruction, a modern reconstruction model with higher spatial resolution, was able to solve the diagnostic uncertainty of PSMA-RADS-3A lesions in 7.6% of cases, re-categorized as PSMA-RADS-4 [29]. PSMA-RADS categories were also correlated to maximum standardized uptake value (SUV max ) by Mihatsch and colleagues, who reported a significantly higher SUV max in PSMA-RADS-5 lesions. Furthermore, the challenging PSMA-RADS-3A lesions showed significantly lower SUV max and SUV peak compared to the PSMA-RADS-4 or -5 [30].
In 2019, Eiber et al. published the Prostate Cancer Molecular Imaging Standardized Evaluation, so-called PROMISE criteria. This molecular imaging TNM system (miTNM, version 1.0) consists of a standardized reporting framework for PSMA-ligand PET/CT or PET/magnetic resonance imaging (MRI) and provides a standardized system for the presence, location, and extent of local PCa and its pelvic and extrapelvic spread (Table 4) [10].
As suggested by the authors, the diagnosis of a PCa lesion should be assessed considering PSMA-uptake, location, and CT or MRI findings, as well as the specific clinical scenario. For this purpose, a standardized method for the PSMA expression level of tumor lesions was introduced (Table 5). Notably, the molecular imaging PSMA (miPSMA) score described the PSMA expression in relation to the mean uptake in the blood pool, liver (or spleen as reference organ instead of liver, for PSMA ligands with liver-dominant excretion e.g., [ 18 F]PSMA-1007), and parotid gland. Scores 2 and 3 are considered typical for PCa lesions and favorable for PSMA-directed RLT [10].
In this setting, a prospective study determined the intra-and inter-observer agreement of [ 68 Ga]Ga-PSMA-I&T PET/CT in 80 patients according to miTNM, reporting great interobserver agreement in the miN (Cohen's K = 0.74) and miM (Cohen's K = 0.84) categories. Differently, a lower interobserver agreement in the miT (Cohen's K = 0.52) category was observed, especially in patients who had bladder or surrounding soft-tissue invasion [31]. Recently, Wang et al. evaluated the predictive role of preoperative miTNM from [ 68 Ga]Ga-PSMA-11 PET in 187 patients with primary PCa who underwent radical prostatectomy. They showed that miTNM correlated with postoperative Gleason score, surgical margin status and time to biochemical recurrence [32]. Comparing [ 68 Ga]Ga-PSMA-11 to [ 18 F]PSMA-1007 PET in terms of miTNM staging, the miM staging was shown to be more concordant than the miT and miN; however, both tracers appeared widely exchangeable [33]. Koehler et al. evaluated the influence of a late scan of the pelvis at [ 68 Ga]Ga-PSMA-I&T PET/CT, reporting a change in 19.5% of the cases in miTNM in comparison with the early scan, and a change in staging in a not negligible number of subjects [34]. The inter-and intra-observer agreements were also evaluated according to the miTNM and PSMA-RADS in the study by Demirici et al. A substantial agreement was reported for miTNM system, while the PSMA-RADS showed almost-perfect agreement among readers. However, in the case of benign lesions authors observed more discordant results for PSMA-RADS than miTNM [35]. Finally, the 3 PSMA PET interpretation criteria (EANM, PSMA-RADS and PROMISE) demonstrated substantial to almost perfect interreader, intrareader, and intercriteria agreement in most situations as described by the group of Stanford [36].  In 2019, a consensus statement for standardized reporting of the PSMA-ligand PET was proposed by a panel of worldwide experts. The E-PSMA provides a structured report including what needs to be included in a report, considering different clinical settings, and including elements from the PROMISE and RADS systems [11].
More recently a dual-tracers scoring system combining [ 68 Ga]Ga-PSMA and FDG was proposed for patients referred to [ 177 Lu]Lu-PSMA RLT, termed Pro-PET score ( Table 6). The scoring scheme, a 5-point categorical scale, is based on the single lesion that was the most FDG avid relative to its [ 68 Ga]Ga-PSMA uptake (most discordant lesion). The concept proposal study retrospectively recruited 47 patients and showed that Pro-PET significantly correlated with the symptomatic response (p = 0.05), biochemical tumor marker response (p = 0.05), metabolic response (p = 0.001), anatomical response (p = 0.012), PFS (p = 0.03) and OS (p = 0.027). The trend observed showed unfavorable outcome when the disease shifted more towards the high grade of the Pro-PET scoring system [12]. In November 2022, a new score for primary PCa diagnosis based on intraprostatic PSMA-uptake pattern was published by Emmett and his group. The PRIMARY combined different pattern information and SUV max assigned to each patient ( Table 7). The specific patterns were described as follows: Pattern A, defined as diffuse transition zone (TZ) activity if centrally placed within the prostate, with no PSMA activity extending to the edge of the prostate margin; Pattern B, symmetrical central zone (CZ) activity with no PSMA activity extending to the prostate margin; Pattern C, focal TZ activity defined visually as more than twice background TZ activity; Pattern D, focal peripheral zone (PZ) activity. Table 7 shows the 5-point PRIMARY score classification [13].  [14] that defined progressive disease (PD) according to three different classes based on the appearance of new PSMApositive lesions, clinical or laboratory data and, eventually, biopsy or imaging confirmation (Table 8).  The authors reported that inter-observer agreement was substantial, and that progression of disease evaluated by score was a significant prognostic marker for overall survival (OS) [37].

PPP Criteria Category Progression Criterion Description and Recommendations
Considering the same scenario, Gafita et al. proposed a PSMA-PET response evaluation criteria for metastatic castration-resistant PCa (mCRPCa) patients treated with [ 177 Lu]Lu-PSMA therapy, termed RECIP. Namely, the appearance of new lesions (=any new focal uptake of PSMA-ligand higher than surrounding background, and each tumor SUV max > mean SUV mean ) and changes in PSMA-VOL (the total positive PSMA volume) were combined to develop RECIP 1.0, which included the classifications of response to therapy presented in Table 9. The authors reported that RECIP-PD had a prognostic impact and the combination of PSA values and RECIP 1.0 criteria may result in a more reliable prognostic evaluation [15]. Table 9. Standardized framework for response evaluation criteria in PSMA PET/CT (RECIP) in mCRPCa patients. More recently, the same group compared the available response criteria for PCa, both specific (aPCWG3, PPP and RECIP 1.0) and non-specific (RECIST 1.1, aPERCIST), in the response evaluation of patients treated with [ 177 Lu]Lu-PSMA. As a result, a better accuracy and inter-reader agreement were obtained with PSMA-specific criteria (PPP and RECIP 1.0). Moreover, a significant lower percentage of patients was defined as having PD according to RECIP 1.0 and these subjects had a higher risk of death in comparison with other response criteria [38].  (Table 10) [16]. However, the PET-based KS showed a higher sensitivity compared to [ 111 In]Pentetreotide one, applied both to planar and single emission tomography (SPECT) scintigraphy [49][50][51].  [39], reporting a detection rate of SSTR-positive disease (KS 2-4) of 23%, 38%, and 72% with [ 111 In]Pentetreotide planar scintigraphy, SPECT and PET, respectively. Moreover, an influence of the size of the lesion on the KS for the three modalities and a correlation between SUV max and KS were reported. These results imply that patients with lesions <2 cm would not have qualified for RLT based on [ 111 In]Pentetreotide but appear as candidates on SSTR PET. The predictive role of KS from [ 68 Ga]Ga-DOTA-NOC PET/CT was also demonstrated in lung carcinoids, with high sensitivity [40]. Finally, Menon and colleagues explored a dual-time-point [ 68 Ga]Ga-DOTA-TATE PET/CT imaging protocol, but no significant changes in KS were reported [41]. The KS was created starting from [ 111 In]Pentetreotide scintigraphy; therefore, reliable standards and criteria for SSTR PET are still lacking since many pitfalls can influence PET scans [52][53][54][55][56].

Neuroendocrine Tumors
In this scenario, Werner et al. proposed a structured reporting system based on a 5-point scale for SSTR PET imaging, named SSTR-RADS Version 1.0, which might serve as a standardized assessment for both diagnosis and treatment planning in NET. The uptake levels for SSTR-RADS were established by a three-point qualitative assessment as shown in Table 11. For the clinical report, they suggested an overall interpretation of the SSTR PET scan with a minimum of clinical and imaging acquisition information and the number of lesions. Moreover, RLT may be considered in the case of an overall SSTR-RADS score of 4 or 5 (Table 12). The interobserver reliability of SSTR-RADS in [ 68 Ga]Ga-DOTA-TOC PET/CT was evaluated by the same group, observing a high agreement in establishing the level of uptake and in the appropriateness of choosing RLT for both inexperienced and experienced readers [17]. Table 11. A three-point visual score for defining the uptake level of SSTR-avid lesion.

Level 1 ≤bloodpool
Level 2 Uptake > bloodpool, but ≤physiological liver uptake Level 3 uptake > physiological liver uptake Suggestive but not definitive for NET. Equivocal uptake in soft-tissue sites typical for NET metastases (e.g., regional lymph nodes).
Not to be considered

SSTR-RADS 3B
Suggestive but not definitive for NET. Uptake in bone lesions not atypical for NET.
Single lesions: locoregional procedure. Increased number of lesions: RLT

SSTR-RADS 3C
Intense uptake in a site highly atypical for NET. Suggestive of an SSTR-expressing, non-NET benign tumor or malignant process.

3
Tissue confirmation of tumor histology should be considered.
Not to be considered

SSTR-RADS 3D
High likelihood of malignant NET lesion, but negative on an SSTR-PET scan (de-differentiated NET or another type of malignancy grading from P1 (=purely SSTR-avid disease without FDG uptake in any lesions), to P5 (=significant FDG-positive/SSTR-negative disease) [18]. The score is presented in Table 13. The authors found a statistically significant correlation between NETPET grade and OS at the univariate analysis. Furthermore, the score correlated with WHO 2010 histological grade (p < 0.00001) [18]. Some years later, the same group of authors validated the grading system in a bigger cohort of 319 metastatic/unresectable gastroenteropancreatic (GEP) NET patients, confirming its prognostic value in terms of OS and time-to-progression as well as with the histological grade [43]. The impact of NETPET was subsequently evaluated by other studies, and its potential use as a prognostic marker was confirmed [44,45] even in patients with bronchial neuroendocrine neoplasms (NENs) [46].

Response Assessment: ZP and MORE
A standardized and reliable response assessment system after RLT is an unmet clinical need, as both morphological and functional imaging have shown limitations [57,58]. Recently, Zwirtz et al. compared the response evaluation with respect to OS in patients treated with at least two cycles of RLT introducing a new metabolic criterion, based on modified EORTC, so-called MORE criteria (Table 14). In addition, to identify other possible predictors of response, they generated two new combined parameters named ZP and ZPnormalized, using baseline CT and SSTR-PET data and summarized as follow [19]: P (Target) = SUV mean (Target) × Hounsfield Unit (HU) (Target) ZPnormalized (Target) = normalized SUV mean (Target) × HU (Target) The concept proposal study, including 34 GEPNET patients, demonstrated that baseline ZP and ZPnormalized with overlapping sensitivity and specificity were the only predictive parameters of lesion progression after three RLT cycles. Moreover, patients who presented a progressive disease after the second cycle of RLT according to MORE criteria showed a significantly shorter OS [19,59].  [20]. NAFCIST is based on the measurement of the mean SUV peak in 1 cm 3 . The single most active bone avid lesion on each scan and the summed activity of up to the five most active ones (no more than two per organ) were considered for the category's determination (Table 15). A significant correlation between the changes in NAFCIST and in alkaline phosphatase and bone alkaline phosphatase was shown, as well as a negative correlation between the cumulative dose of [ 223 Ra]Cl 2 and changes in NAFCIST, thus demonstrating that the more [ 223 Ra]Cl 2 administered, the more NAFCIST value decreased. Moreover, NAFCIST correlated with OS. Interestingly, in the same sample, PERCIST applied to [ 18 F]FDG PET/CT did not show any significant correlations with outcomes [20]. In a subsequent work, the same authors applied radiomics to [ 18 F]NaF PET imaging. A decrease in 18 Fconcentration in metastatic areas was associated with an increase in intra-metastatic disorder after [ 223 Ra]Cl 2 treatment.
Even with a small sample, these findings suggested that the more cycles of treatment were administered, the more radiotracer concentration and entropy decreased [47]. Analyzing the postoperative score, significant differences were found for progression free survival (PFS) and OS for incomplete versus complete metabolic resections, respectively. The authors found that postoperative positive [ 18 F]Fluorocholine PET/CT localizations correlated with the sites of tumor recurrence. Furthermore, on preoperative PET/CT, they observed that lesions with higher tracer uptake were followed by higher metabolic residual lesions after surgery [48].

Discussion
From its introduction in the 2000s, PET/CT gained space for the molecular imagingbased assessment of tumors in clinical practice [60,61]. In this scenario, many efforts have been made to develop standardized image analyses through the introduction of reliable, easy-to-use, and practical criteria. The existing scores were introduced mainly to assess treatment response, but the advent of new specific radiopharmaceuticals (e.g., PSMAligands compounds) underlined the need for a standardized method for imaging interpretation too. In this setting, PSMA showed to be the most promising tracer for PCa with high sensitivity and specificity, even if its role in different pathological conditions has been reported [62][63][64][65][66][67][68]. In this scenario, several criteria for interpreting PSMA-ligand PET were proposed. The first introduced EANM criteria demonstrated moderate consensus among readers, probably due to the absence of a scale, thus underlining the importance of categories in a structured reporting system. As consequence, the PSMA-RADS and the miTNM were proposed: the first, a 5-point visual scale, demonstrated a high interobserver agreement, even with different levels of experience; however, it lacks a reference uptake scale as introduced by the miTNM. The latter is the only score endorsed by the EANM guidelines and seems to pave the way for its introduction in large clinical trials [11]. However, considering the recent approvement of [ 177 Lu]Lu-PSMA RLT for PSMA-positive mCRPCa patients [69], neither PSMA-RADS nor miTNM include treatment recommendations for this therapy needing warranting improvements. In this scenario, the recently proposed Pro-PET score combined both PSMA-ligand and [ 18 F]FDG PET/CT performed before RLT to improve patients' selection and serving as a prognostic marker. Among interpretation criteria, PRIMARY score is lastly emerged in literature as a promising criterion for the use of PSMA-ligand PET/CT also in the diagnosis of primary tumor. Considering instead the evaluation of response to treatment in PCa patients, it is usually performed by applying RECIST version 1.1 to CT, alongside the assessment of the PSA trend. However, these criteria have several limitations and are not enough to fully evaluate PCa patients [70,71]. [ 18 F]Fluorocholine PET/CT or PSMA-ligands PET/CT are not currently recommended by most updated guidelines for this purpose, mainly due to the absence of prospective, randomized, large sample size trials [72][73][74][75]. In this scenario, Fanti et al. introduced the PPP score, based on PCWG3 guidelines principles, for the evaluation of progression in PCa patients. However, this criterion missed the definition of complete, partial, and stable metabolic patterns and does not consider changes in PSMA expression (as neuroendocrine differentiation) [76,77]. The traditional four response categories were re-introduced with RECIP 1.0 used to assess the response to [ 177 Lu]Lu-PSMA. According to preliminary data, both criteria seem to correlate with OS and appear superior to assess responses compared to their non-specific counterpart (RECIST 1.1, aPERCIST).
Nowadays, theragnostic represents the driving force to introduce generalizable framework systems for standardized reporting as well as for treatment response assessment both for PCa and NET patients. In the last few years, the SSTR-RADS, structured in a reciprocal fashion of PSMA-RADS, were introduced to convey to the nuclear medicine scan reader the level of certainty that an equivocal finding is a site of disease, avoiding common pitfalls in interpreting SSTR imaging [52][53][54][55]. Such a framework also allowed identifying appropriate candidates for treatment with [ 177 Lu]Lu-DOTA compounds. Our systematic review pointed out the few data available for the SSRT-RADS, probably due to the absence of dual-tracer incorporation into the score. Namely, dual PET imaging should be considered for all patients with a diagnosis of metastatic GEPNET (grade 2-3). In this regard, Chan and colleagues devised a novel scheme for dual SSTR/FDG grading, termed NETPET grade, which showed a significant correlation with OS, time to progression, and histological grade, serving as a predictor of outcome. Despite the introduction of RLT many years ago, no reliable treatment response assessment criteria were introduced [78,79]. In this setting, some authors evaluated different PET-derived parameters for response evaluation, with contradictory results [80,81]. The newly introduced scores that emerged in our review (MORE, ZP) need to be validated, but preliminary results demonstrated their possible role in prognostication and in prediction of response to RLT.
Other minor investigated non-[ 18 F]FDG PET criteria deserve mention. The NAFCIST criteria, introduced to assess treatment response in patients undergoing [ 223 Ra]Cl 2 therapy on [ 18 F]NaF PET imaging, seem to better correlated with outcome than PERCIST. Moreover, compared to RECIST, known to be suboptimal for the evaluation of osteosarcoma with calcified bone-forming tumors often not shrinking even if responding, NAFCIST could represent a more accurate method of categorizing osteosarcoma owing to its better ability in reflecting bone-forming component [20]. Finally, the FuMeGa criteria emerged as the first metabolic criteria introduced for post-operative PET interpretation in HGG patients. It is known that postoperative assessment in glioma patients is crucial in the imaging follow-up and for prognostic considerations [82]. Even for the evaluation of recurrence, PET imaging demonstrated its added value over MRI, given its ability to assess tumor metabolism and reduce pseudo-progression pitfalls [83]. However, [ 18 F]Fluorocholine is not the standard of choice because amino acid tracers (such as [ 11 C]MET, [ 18 F]FET) demonstrated better diagnostic performance. However due to the limited availability of the amino acid PET tracers, [ 18 F]Fluorocholine PET could be useful in blood-brain alterations, namely HGG.
To sum up, this systematic review showed a trend to standardization particularly evident in PCa because of its incidence, with promising evidence for SSTR-PET imaging and preliminary experiences for other oncological scenarios which warrant further and larger applications.

Conclusions
In conclusion, many criteria regarding the use of non-[ 18 F]FDG PET imaging in oncology have been proposed in the literature, but the majority of them are not integrated into clinical practice. Even if more data are needed to clearly evaluate their impact on the management of patients, these criteria represent promising tools for the interpretation of PET scans and standardization of reporting, but also to assess the response to therapy and, therefore, to guide the prognosis.