Integrating Functional Imaging and Molecular Profiling for Optimal Treatment Selection in Neuroendocrine Neoplasms (NEN)

Gastroenteropancreatic NEN (GEP-NEN) are group of malignancies with significant clinical, anatomical and molecular heterogeneity. High-grade GEP-NEN in particular present unique management challenges. In the current era, multidisciplinary management with access to a combination of functional imaging and targeted molecular profiling can provide important disease characterisation, guide individualised management and improve patient outcome. Multiple treatment options are now available, and combination and novel therapies are being explored in clinical trials. Precision medicine is highly relevant for a heterogenous disease like NEN. The integration of dual-tracer functional PET/CT imaging, molecular histopathology and genomic data has the potential to be used to gain a more comprehensive understanding of an individual patient’s disease biology for precision diagnosis, prognostication and optimal treatment allocation.

Due to non-specific hormone secretory syndromes or symptomatology, NEN is often identified late: 60-85% of patients have incurable metastatic disease at diagnosis [2 ••, 6]. It is at this advanced stage that patients are typically Grace Kong and Emma Boehm contributed equally to this work. 1 3 referred for multidisciplinary assessment. Initial workup of NEN typically involves conventional radiology and histopathology assessments but these alone are inadequate to provide full characterisation for this complex heterogenous disease. This review will focus on the importance of precision evaluation and the need to improve and develop diagnostic paradigms to guide personalised therapeutic treatment of GEP-NEN. Access to molecular imaging and molecular testing can resolve diagnostic uncertainty, aid prognostication and guide therapeutic selection particularly for patients with higher-grade disease where disease heterogeneity is common. We will discuss the important role of molecular imaging with positron emission tomography (PET) using somatostatin receptor (SSTR) tracers, integrated with metabolic imaging using 2-[18F]fluoro-2-deoxy-d-glucose [ 18 F]FDG (FDG) to non-invasively assess disease biology and heterogeneity. In addition, the development and integration of molecular testing with pathway-focussed histopathological analysis and both germline and tumour somatic mutational analysis can provide further important diagnostic insights, as well as treatment stratification for selected patients with GEP-NEN [2••].

Molecular Imaging: a Non-invasive Way to Understand Whole Body Disease Biology and Guide Treatment Selection
Radiology using computed tomography (CT) and magnetic resonance (MRI) remain the cornerstone of NEN imaging and are widely available for detecting and monitoring sites of disease. However, recognised limitations of CT include the inability to identify small malignant primary NEN lesions, lymph nodes or bone metastases which are prevalent for metastatic NEN [7••, 8, 9••]. The sensitivity and specificity for NEN detection or restaging may be reduced if serial scanning is performed using non-uniform protocols [10]. It is now well established that molecular PET/CT imaging using SSTR and FDG radiotracers play essential incremental roles in the staging, restaging and theranostics selection for patients with NEN, by characterising specific disease biology.

SSTR PET/CT Imaging
SSTR (particularly subtype 2) is commonly overexpressed on well-differentiated NEN and represents a useful molecular imaging and therapeutic target [11, 12•]. The initial approved modality [ 111 In]In-DTPA-octreotide single-photon emission computed tomography (SPECT)/CT has become superseded by PET/CT imaging due to its superior imaging resolution, diagnostic performance and quantitation [13][14][15]. Even sub-centimetre lesions with high SSTR expression can be visualised with a high target-to-background ratio.

FDG PET/CT Imaging
FDG is the most used oncological PET imaging agent. Uptake of this radiolabeled glucose analogue correlates with tissue metabolism and proliferation, where uptake is typically high in rapidly growing tumours or tumours with metabolic reprogramming favouring glycolysis. FDG is not a NEN-specific tracer, but FDG positivity is closely correlated with higher NEN tumour grade (typically G2 or G3 NET and NEC), poor differentiation and worse prognosis [16•, 17•, 18, 19]. Studies have established an inverse relationship between proliferation rate and SSTR positivity [20,21]. A higher proliferation rate is expected for higher-grade disease in approximately 75% of G3 NET and around 90% of NEC cases [22,23].

Dual-Tracer Imaging (SSTR and FDG Tracers)
This combined imaging approach can provide powerful complementary information to characterize NEN biology. It is well recognized that significant heterogeneity can exist within an individual patient, such that well-differentiated lesions (SSTR-expressing) can co-exist with higher-grade components (often FDG-avid) [24,25]. SSTR imaging positivity is a marker of well-differentiated NEN. FDG positivity is a marker of disease metabolic activity and NEN aggressiveness. The use of dual-tracer imaging can assess the heterogeneity of disease biology within a patient, impacting on prognostication and management.

As a Prognostic Biomarker
Typically, patients with FDG positive/SSTR negative disease have a poor prognosis and shorter overall survival compared to patients with FDG positive/SSTR positive, or solely SSTR positive disease respectively (latter with best prognosis) [26•, 27]. Earlier institution of more aggressive treatments and frequent monitoring is warranted for patients with highly FDG-avid disease.

To Guide Biopsy Site
Tumour grading based on ease of access or location alone may not be representative of the true highest-grade disease given the potential disease heterogeneity. Dual-tracer imaging phenotype can guide the site for targeted biopsy. Typically, the lesion with the most intense FDG uptake is likely to represent a disease of the highest proliferative activity and grade [28].

To Guide Patient Management and Therapeutic Approach
Molecular imaging phenotype guides selection for PRRT and stratifies other systemic therapies. High SSTR expression at all disease sites is the main prerequisite for PRRT. PRRT can be effective even if lesions show FDG uptake provided that all these lesions also have high SSTR expression to allow therapeutic targeting [29,30]. Spatially discordant (FDG positive/SSTR negative) disease cannot be targeted with PRRT alone, and in this case, other systemic or combination options should be considered [26 •, 31]. Given the poorer prognosis, patients with highly FDG-avid disease (concordant or discordant) should be followed up more frequently following therapy.
The dual-tracer molecular imaging approach is therefore highly recommended for patients with (1) higher-grade disease including G2 and 3 NEN; (2) patients with presumed G1 disease but with non-SSTR-avid suspicious lesions on radiological imaging; (3) at the time of more rapid progression than expected for the grade (i.e. initial pathological sampling error or transformation to higher grade); (4) to assess heterogeneity and guide biopsy site; and (5) for theranostic selection and to guide therapeutic options [12•].
Whilst SSTR imaging is now widely considered the standard of care for NEN, the combined use with FDG PET/ CT is yet to be universally applied due to geographical differences in resources and regulatory limitations. Its benefits warrant further prospective validation to enable integration in NEN management.

Histopathology: Defining Morphology and Protein Expression for Diagnosis and Prognostication
Histopathological evaluation of tumour morphology, proliferative index and immunohistochemical (IHC) biomarker expression is the foundation of NEN diagnosis and grading [1, 2••, 6]. As discussed, the use of molecular imaging phenotype will guide the biopsy site to ensure sampling representative of the highest-grade lesion. Guidelines specify a minimum requirement for structured reporting of morphology, immunostaining for expression of standard neuroendocrine differentiation markers (chromogranin A, synaptophysin and CD56 or INSM1), as well as proliferation markers (Ki-67/MIB1) (1). GEP-NENs are almost always pan-cytokeratin-positive, but CK7/CK20negative. The use of morphology and proliferative index to stratify GEP-NENs into NETs (G1-3) or NECs has prognostic and therapeutic implications; however, the optimal parameters remain controversial, and predictors of treatment response are lacking [32]. Importantly, the assessment of the Ki-67 index may be limited by sample error due to inadequate sample size or scoring methodology and should be performed by pathologists with experience in NENs to ensure accuracy and reproducibility. Patients with GEP-NET G3 have better overall survival (OS) than patients with NEC at 43.6 vs 5.3 months [33]. Patients with NEC have been reported to have a better response to platinum-based chemotherapy than NET-G3, although overall survival remains lower [33]. It is important to recognise however that classification based on morphology alone may be challenging and molecular analysis is an essential adjunct.
IHC markers of neuroendocrine cell-of-origin and differentiation are essential to resolve the common diagnostic uncertainty around defining G3 NET versus NEC. Additional IHC markers of NET differentiation include somatostatin receptor type 2 (SSTR2), which can also be used to infer somatostatin analogue (SSA) sensitivity and utility of SSTR functional imaging and is reduced in poorly differentiated cancers [34,35]. Nuclear staining for the neuroendocrine transcription factor Insulinoma-associated protein-1 (INSM1) has very high sensitivity and specificity (99 and 96% respectively) for GEP-NET, and 100% positive and negative predictive value for differentiating pancreatic NET from other pancreatic differentials including ductal adenocarcinoma, solid pseudopapillary neoplasm and acinar cell carcinoma [36][37][38]. Loss of immunostaining for alpha-thalassemia/mental retardation X-linked (ATRX) and Death Domain Associated Protein (DAXX, pancreatic NET) correlates with loss of function mutations and is associated with well-differentiated disease and may have prognostic value [34, 39 ••]. Retained expression of ATRX and DAXX, but the loss of expression of retinoblastoma 1 (RB1) and SMAD4, and altered p53 expression are typical of GEP-NECs [2••, 32, 34, 39••, 40,41]. Glucose Transporter-1 (GLUT1) positivity is a marker of aggressive behaviour and poor prognosis in GEP-NET [42][43][44], and a potential surrogate for FDG PET/CT positivity. SSTR2, INSM1, ATRX, DAXX, RB1 and p53 IHC assessments are now more frequently available in anatomical pathology departments and should be incorporated as part of standard care for complex cases unable to be resolved by routine histological examination.

Genomics of GEP-NEN: a Nuanced Understanding of Individual Disease Biology Has the Potential to Inform Patient-Specific Treatment Strategies
Whilst the integration of molecular imaging and histopathology/IHC techniques have improved NEN characterisation and patient care, a precision medicine approach is needed to manage such complex heterogenous disease and improve individualised outcome. NETs and NECs have distinct genomic profiles and gene drivers (some can be inferred via IHC as in the previous section) such that the role of genomic analysis in GEP-NEN should extend beyond the consideration of germline testing for risk management alone. Rather, genomics can aid in diagnosis, prognosis, treatment selection and trial design.

Germline Testing
Germline testing is currently only recommended for GEP-NET patients with features of clinical endocrine tumour syndromes [45][46][47][48]. It has long been known that approximately 10% of GEP-NEN is associated with germline mutations driving the classical syndromes of multiple endocrine neoplasia type 1 (MEN1, encoding the histone modifying Menin 1 protein), as well as neurofibromatosis 1 (NF1), von Hippel Lindau (VHL) and tuberous sclerosis (TSC1/TSC2). To challenge this paradigm, the seminal International Cancer Genome Consortium study involving whole genome sequencing of 98 apparently sporadic pancreatic NETs revealed previously unknown germline alterations in up to 17% of patients including homologous recombination DNA repair genes (BRCA2 and CHEK2) as well as the base-excision DNA repair gene MUTYH [49••, 50]. For patients with small intestinal GEP-NET (SI-NET), long been considered a sporadic disease notorious for a paucity of recurrent driver genes (with the exception of somatic CDKN1B in a minor fraction), germline mutations in IMPK, OGG1 and DNA repair-associated genes including CHEK2, RAD51C, ATM and MUTYH have recently also been identified [50][51][52]. The pathogenicity and clinical significance of these defects in SI-NET are at present unclear [53, 54••].
Recognising the cohort of patients with GEP-NEN who harbour DNA repair defects and have SSTR-expressing disease on molecular imaging could inform the rational allocation to combination PRRT and drugs that inhibit alternative/rescue DNA repair pathways, such as Poly-ADP Ribose (PARP) inhibitors to maximise radiosensitivity. Such a therapeutic strategy is under active investigation in the PARLuNET trial (NCT05053854), and NCT04086485. Patients with tumours driven by DNA repair defects might also plausibly benefit from a combination of radionuclide therapy and DNA-damaging agents used in the treatment of advanced NET including the antimetabolite capecitabine and the alkylating agent temozolomide [31,55].

Somatic Profiling
Genomic profiling of NEN reveals recurrent features and has a clear diagnostic application. NETs typically have few driver mutations [56••, 57••]. Sporadic NETs frequently harbour somatic mutations in MEN1 but also VHL and TSC2 [58,59]. Loss of function mutations in chromatin-modifying genes ATRX/DAXX corresponds to alternative lengthening of telomeres (ALT), chromosomal instability and recurrent genome-wide patterns of chromosomal loss [ Somatic testing can potentially lead to targeted treatment or trial allocation in NEN, and comprehensive genomic profiling is endorsed at clinical discretion in NEN NCCN guidelines [45]. The NCI-MATCH study found that 10% of patients with unspecified subtypes of "neuroendocrine cancer" who underwent tumour panel gene testing were allocated to trials [

Liquid Biopsy
The detection and analysis of circulating tumour DNA (ctDNA) from blood sampling are a non-invasive method to overcome procedural risks and the issues of undersampling of disease heterogeneity inherent in tissue biopsy. Given the limitations in sensitivity and specificity of current markers such as chromogranin A for diagnosis/prognostication in NEN, novel non-invasive biomarkers are sorely needed.

Treatment Selection for GEP NEN: Current Approach and Future Perspectives for Precision Therapy
The selection of therapy for NENs is currently primarily based on histology (grade), primary site, structural/functional imaging, IHC and clinical behaviour. As described, molecular characterisation (e.g. TMB status) may have a role in future treatment decision-making. Regarding MTAs, everolimus has demonstrated increased PFS relative to placebo in pancreatic NETs (HR = 035, P < 0.001) [92], non-functional pulmonary NETs and GEP NETs (HR = 0.48, P < 0.00001) [93]. Sunitinib has a PFS advantage relative to placebo for pancreatic NETs: HR = 0.42, P < 0.001 [94].

First-line Therapy
In terms of chemotherapy, modern phase III trials are lacking. Patients selected are those with progression post-SSA, PRRT or MTAs, if unsuitable for PRRT, or those with large volume or rapidly progressive disease. The integration of dual-tracer molecular imaging plays an important role in identifying patients with these poor prognostic features. Chemotherapy is more active in patients with pancreatic NETs, with ORR from 31 to 70% and OS exceeding 40 months [95•]. Regimens include capecitabine plus temozolomide (CapTem), temzolomide, FOLFOX, capecitabineoxaliplatin (CapOx) and streptozotocin-5FU. The activity of CapTem was confirmed by the randomised phase II E211 trial [96••].
The optimal therapy sequencing of the available options, however, has not been validated. The SEQTOR study (GETNE 1206) randomised patients with progressive pancreatic NET to everolimus followed by streptozotocin-5FU upon progression (arm A), or the reverse sequence (arm B). On initial analysis, both sequential strategies showed similar efficacy and PFS [97••].

Grade 3 NETs and NECs
The treatment approach for patients with G3 NETs and NECs differs substantially given their histopathology, imaging characteristics and genomics (see Table 1). Given the more aggressive nature of the disease, early institution of therapy is important to optimise patient outcome: the integration of molecular imaging and molecular profiling could play an important role for these patients.

First-line Therapy
The clinical behaviour of NECs is similar to extensivestage small cell lung cancer (SCLC) [22].  [102,107] and a Ki-67 ≤ 60% predict less benefit from platin-based chemotherapy [100]. In G3 NET, the ORR to platin-based regimens is < 5%, with PFS < 3 months, but prolonged OS [18,100,108,109]. Hence, patients with G3 NETs benefit from similar therapies used in G2 NETs [18,22]. Several heterogeneous retrospective series have indicated activity for CapTem in G3 NETs: ORR varies from 30 to 51%, median PFS of 9 to 15.3 months and OS from 19 to 29.3 months [110][111][112][113][114]. The optimal threshold for higher ORR is a Ki-67 from 10 to 40% [115]. Data for other therapies in G3 NET is limited. The pivotal SSA phase III trials had not included G3NET [83,116] and so their use should be limited to patients with confirmed SSTR expression, no FDG discordance (this should be closely monitored), or for management of secretory syndromes [117]. The data on MTAs in G3 NET is sparse. Everolimus has been evaluated in patients with G3NETs (Ki-67 20-55%) in the first/second-line setting (N = 15), with a median PFS of 6 months, and OS of 28 months [118]. A completed German study (EVINEC) has evaluated everolimus as a second-line treatment for G3 NET and G3 NEC (NCT02113800). Sunitinib was evaluated in 31 patients with pancreatic grade 3 NET/NECs: with partial response in 4 and stable disease seen in 14 patients [119]. A completed Nordic phase II study has evaluated temozolomide and everolimus as first-line treatment in metastatic G3NET (Ki-67 21-55%) (NCT02248012).
ICI is also promising in progressive high-grade NET and NEC, based on their higher TMB; the latter is greater in NECs and with microsatellite instability noted in 14% of NECs [132]. A meta-analysis of 10 heterogenous, singlearm studies of ICI in NEN (N = 464) found a pooled ORR of 15.5% [133). The response was based on primary site: with thoracic NEN being more likely to respond than GEP-NEN (ORR 24.7% vs 9.5% respectively) and well-differentiated tumours having a lower response rate than NECs (ORR 10.4% vs 22.7% respectively) [133]. Very limited activity has been observed with single-agent immunotherapy [134,135], relative to combined PD1 and CTL4 blockade. From the CA209-538 study, 29 patients with heavily pre-treated NETs were treated with a combination of ipilimumab and nivolumab. Overall, in the 13 (45%) with high-grade disease, the ORR was 24% and a DCR of 72% [136••]. The SWOG S1609 DART trial reported the results of the high-grade G3 NET/NEC cohort (N = 19) with a median Ki-67 value of 80%. The ORR was 26% and the clinical benefit rate (stable disease for ≥ 6 months plus PR and CR) was 32% [137••]. Other trials are yet to be reported, including a phase II trial of PDR001 (PD-L1 inhibitor) (NCT02955069), Nivolumab combined with EP (NCT03980925) and toripalimab in pancreatic NEN (NCT03043664, NCT02939651 and NCT03147404). Even within TMB-high NENs, however, there is a heterogeneous response to ICI highlighting the need for further biomarkers for stratification.

Perspectives for Precision Therapy Utilizing Multidisciplinary Diagnostic Approaches
NEN is a challenging, heterogenous disease with different clinical, imaging, pathological and genomic complexities to consider in each patient. Multiple treatment options are now available, and combination and novel therapies are being explored in clinical trials. However, clinical treatment selection and sequencing are still mainly based on disease grade, primary site, agent availability and local protocols, without personalisation. Precision medicine is highly relevant for a Fig. 1 A case example of a 54-year-old female, with a previous history of treated localised breast cancer, and previously resected grade 1 (Ki-67 2%) pancreatic NET. She presented with new, multiple hepatic (A) and mesenteric nodal metastases (B). A Ga-68 DOTATATE PET/CT (C) showed metastatic disease in the liver, nodes and bones with high SSTR expression. FDG PET/CT (D) showed some lesions with concordant FDG avidity. The lesion with the highest metabolic activity (mesenteric node, E) was targeted for biopsy and diagnosis. Histopathology (F) showed monotonous cuboidal cells with granular eosinophilic cytoplasm, ovoid nuclei and fine chromatin. By IHC Ki-67 labelling index was 25% (G) and DAXX expression was lost (H). Other IHC (not shown) demonstrated expression of SSTR2 and synaptophysin, retained ATRX and Rb, a p53 wild-type pattern, and no staining for chromogranin or multiple breast markers. Overall, the features were supportive of a G3 NET and not breast carcinoma or NEC. Genomic sequencing confirmed DAXX mutation and MEN1 mutation, typical for NET. The patient proceeded to receive PRRT treatment for metastatic G3 NET heterogenous disease like NEN. In the current era, the integration of molecular imaging (SSTR and FDG PET/ CT) and molecular profiling (IHC profile and genomic analyses) can provide important disease characterisation, to guide precision management and individualised treatment selection/sequencing (see Fig. 1). This is particularly crucial for patients with advanced high-grade NENs and to resolve G3 NET vs NEC disease biology, as clinical behaviour and treatment options can differ significantly. It is also imperative to focus on incorporating prospective serial translational genomic analysis of tissue and blood, to develop novel liquid biopsy and tumour testing methodologies to understand NEN pathogenesis, discover predictive and prognostic biomarkers to explain the differential response to therapy and subsequently guide future trial design for rational treatment allocation. Using multidisciplinary diagnostic approaches should be the focus of future development to improve individualised therapy and patient outcomes.

Conclusion
We are in an exciting era for the biological interrogation of neuroendocrine neoplasms to guide precision management by incorporating molecular imaging assessment with clinically relevant molecular pathology pathway and genomic evaluation. Our technological capability for precision diagnosis needs to be developed in parallel with therapy advancements in patients with advanced-stage highergrade NEN and globally is only a reality for patients who have geographical or financial access to major NEN referral centres [138]. It is therefore imperative not only to place molecular imaging and genomics at the centre of NEN patient management but to also show the symptomatic, survival and health economic benefits of doing so through high-quality research such that these technologies are widely supported by guidelines and imbursed by regulatory bodies.
Acknowledgements Dr. Catherine Mitchell reported the histopathology displayed in Fig. 1.

Declarations
Conflict of Interest All authors declare no conflict of interest. GK has received research funding from Pfizer and Cyclotek, had consulting or advisory role for ITM, and received a Clinical Fellowship Award from the Peter MacCallum Foundation.

Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.