Critical Reviews in Oncology/Hematology 113 (2017) 268–282
Contents lists available at ScienceDirect
Critical Reviews in Oncology/Hematology journal homepage: www.elsevier.com/locate/critrevonc
Prognostic and predictive biomarkers in neuroendocrine tumours David L. Chan a,b,c , Stephen J. Clarke a,b , Connie I. Diakos a,b , Paul J. Roach d , Dale L. Bailey b,d , Simron Singh c , Nick Pavlakis a,b,∗ a
Department of Medical Oncology, Royal North Shore Hospital, St Leonards, Australia Sydney Medical School, University of Sydney, Australia Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, Canada d Department of Nuclear Medicine, Royal North Shore Hospital, St Leonards, Australia b c
Contents 1. 2.
3.
4.
5. 6.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Tumour-based biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 2.1. Ki-67 index/mitotic count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 2.2. Genetic mutations in tumour as biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 2.3. MicroRNAs as biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 2.4. Epigenetic mutations and prognosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 2.5. Novel markers under investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 Blood-based and urine-based biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 3.1. Circulating hormones – urinary 5-HIAA, chromogranin a and derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 3.2. Other serum-based assays as biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 3.2.1. Neuron-specific enolase (NSE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 3.2.2. Plasma-based markers of the angiogenesis and mTOR pathways in NET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 3.2.3. Other plasma-based markers in NET. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .275 3.3. Circulating tumour cells (CTCs) in NET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 Structural and functional imaging-based biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 4.1. Structural imaging-based biomarkers – staging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 4.1.1. TNM in GEPNETs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 4.1.2. TNM in bronchial NETs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 4.1.3. TNM in other primary sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 4.2. Functional (nuclear) imaging as biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 4.2.1. Somatostatin receptor scintigraphy – 68 Ga-DOTATATE/TOC/NOC PET – and prognosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 4.2.2. FDG PET and prognosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 4.2.3. Dual 68 Ga/FDG PET imaging and prognosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 4.2.4. Nuclear imaging as predictive biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Predictive nomograms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
a r t i c l e
i n f o
Article history: Received 26 October 2016 Accepted 11 March 2017 Keywords: Neuroendocrine Prognosis
a b s t r a c t Neuroendocrine tumours are extremely heterogeneous malignancies. Despite marked heterogeneity in clinical course and prognosis, few biomarkers exist to help predict prognosis and guide treatment. Many tumour-based biomarkers (Ki-67, mitotic count, genetic/epigenetic changes and microRNAs) exist, but only Ki-67 and mitotic count have strong evidence to support their routine use. Blood-based markers are easily repeatable, but currently established biomarkers (chromogranin A and urinary 5-HIAA) are difficult to measure accurately in practice. Structural imaging is used routinely via the TNM system. Functional
∗ Corresponding author at: Department of Medical Oncology, Level 1, Royal North Shore Hospital, Reserve Road, St Leonards, NSW 2065, Australia. E-mail address:
[email protected] (N. Pavlakis). http://dx.doi.org/10.1016/j.critrevonc.2017.03.017 1040-8428/© 2017 Elsevier B.V. All rights reserved.
D.L. Chan et al. / Critical Reviews in Oncology/Hematology 113 (2017) 268–282 Biomarkers Review NETs Nomogram
269
imaging such as 68 Ga-based and FDG PET may become valuable biomarkers with their increasing availability, aided by ongoing quantitative research. Multiple nomograms have been proposed to integrate the above factors, but most have not been prospectively validated and are difficult to use in practice. Further research should aim to establish robust new biomarkers and integrate existing ones to help optimise NET treatment. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Neuroendocrine tumours were first described by Oberndorfer in 1907 (Oberndorfer, 1907) due to their potentially malignant character using the name “karzinoide”. As they arise from the enterochromaffin cells of the neuroendocrine system, they have more recently been named neuroendocrine tumours. They are most often found in the gastroenteropancreatic (GEP) system (57%) and the lungs (27%) (Yao et al., 2008a,b), and 50% of patients present with metastatic disease (Yao et al., 2008a). Whilst many therapeutic advances have been made in neuroendocrine tumours (NETs) (Lawrence et al., 2011a), 5 year overall survival (OS) for GEPNETs is only 61% (Hallet et al., 2015), and many patients with metastatic disease ultimately succumb to progressive disease. In contrast to other solid organ tumours, NETs may arise from a wide variety of primary organ sites – the whole gastrointestinal tract (particularly the small bowel), pancreas, and lungs as well as rarer sites such as the thymus and cervix. The primary site of origin appears to also influence prognosis, with colonic, gastric and hepatic NETs having worse overall survival in the metastatic setting (Yao et al., 2008a). Even metastatic NETs from the same primary site are heterogeneous in clinical presentation, aggressiveness and prognosis. For instance, patients with Grade 1 midgut NETs have a median survival of 16.6 years, compared to 1.1 years for Grade 3 NETs (Ahmed et al., 2009). In another study, patients with NET lesions which showed avidity on FDG PET had a median survival of 1.2 years, compared to 10 years for those who did not, indicating the FDG highlights patients with aggressive clinical behaviour and a high proliferative rate (Bahri et al., 2014). This heterogeneity presents a significant clinical challenge, as patients may either take false comfort from the misconception that NETs are benign when they have aggressive disease, or have an unduly pessimistic outlook when their outcome with a Grade 1 NET is projected to be excellent. Apart from the prognostic implications, heterogeneity also makes it difficult to optimise treatment in NETs. The conduct of clinical trialsis also hampered by the difficulty in selecting a homogenous patient cohort. Recent trials have focussed on specific NET subgroups (such as midgut, pancreatic, or GEPNETs) with potential inclusion of other subgroups (such as lung or hindgut NETs), but this specificity slows patient accrual to trials and makes any findings less generalizable. This clinical need has driven the search for accurate, affordable and repeatable biomarkers to help inform prognosis and predict response to treatment. If identified, these biomarkers would allow administration of the right treatment to the right patient at the appropriate time – avoiding unnecessary side effects from therapy yet administering effective clinical treatment before significant clinical deterioration occurs. A biomarker is “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” (Biomarkers Definitions Working Group, 2001), and guidelines exist to guide the conduct and reporting of biomarker studies (McShane and Hayes, 2012; Alonzo, 2005). Several reviews have focussed on blood-based biomarkers in NETs (Modlin et al.,
2014; Oberg et al., 2015). Serum-based assays offer advantages in terms of convenience and safety of collection (compared to repeat tissue biopsies). However, older studies have not demonstrated sufficient power to enable accurate prognostication, and have in general looked at individual measures in isolation. Newer studies with multiple analyses show promise, but have not been validated as yet. Molecular imaging has been increasingly employed in guiding management of NETs, with recent studies demonstrating potential prognostic significance. The optimal selection of therapies in a given NET patient at a given point in their clinical course remains an unanswered question. Guidelines (Pavel et al., 2016; Phan et al., 2010) tend to focus on the available evidence in trials conducted to date, based upon the inclusion criteria of those trials. As such, initial anti-proliferative therapy tends to be with somatostatin analogues, with other therapies (such as peptide receptor radionuclide therapy (PRRT), tyrosine kinase inhibitors, or chemotherapy) chosen upon failure of these therapies. The identification of biomarkers that could help predict prognosis and response to individual treatments would allow individual personalization of treatment for each patient to achieve optimal outcomes. This review was undertaken to evaluate the literature on tissue-based as well as molecular imaging derived biomarkers for neuroendocrine tumours, in order to identify areas where evidence of biomarker utility is robust and where gaps exist so as to direct further research.
2. Tumour-based biomarkers 2.1. Ki-67 index/mitotic count The Ki-67 index and mitotic count are markers of cell proliferation which have been increasingly utilised and reported since their adoption into the 2010 WHO histological grading system (Klimstra et al., 2010) for NETs of gastroenteropancreatic origin (GEPNETS) (Table 1). It is the most common tissue-based marker used in NETs worldwide. The mitotic count (MC) has been used as a biomarker for over 20 years. It is conventionally counted over at least 10 high-power fields, and has been shown to have prognostic significance in pancreatic (Pelosi et al., 1996; La Rosa et al., 1996; Hochwald et al., 2002; Ferrone et al., 2007), upper gastrointestinal (Pape et al., 2008a), and bronchial NETS (Beasley, 2000; Travis, 1998; Joseph et al., 2015). It is still used as the primary determinant of grade in bronchial NETs (Travis et al., 2004), along with the presence or absence of necrosis. However, the Ki-67 index often provides additional information which may affect management, particularly in GEPNETs. Ki-67 plays a very prominent role in NETs compared to other tumours, because of the wide disparity in biological behaviour between different grades of disease. It is present in cells undergoing all parts of the cell division cycle (G1, S, G2 and mitosis) but not in G0 (Scholzen and Gerdes, 2000). Therefore, the percentage cells which are Ki-67 antigen positive (otherwise known as the Ki-67 index, and occasionally shortened to just “Ki-67”) reflects the growth fraction of a cell population. Whilst its exact function is still unknown, the
270
D.L. Chan et al. / Critical Reviews in Oncology/Hematology 113 (2017) 268–282
Table 1 2010 World Health Organization (WHO) classification of gastroenteropancreatic neuroendocrine tumours (WHO, 2010). Grade
Mitotic count (mitoses per 10 high power fields)
Ki-67 index
Traditional nomenclature
WHO/ENETS nomenclature
Grade 1 Grade 2
20%
Carcinoid, islet cell tumour (Atypical) Carcinoid, islet cell tumour Small cell carcinoma, Large cell neuroendocrine carcinoma
Neuroendocrine carcinoma (large cell or small cell type)
Mixed adenoneuroendocrine carcinoma (MANEC) Hyperplastic and pre-neoplastic lesions Mit: Mitoses, HPF: High power fields, ENETS: European Neuroendocrine Tumour Society.
Ki-67 protein seems to play a functional role in cell proliferation, as demonstrated by the anti-proliferative properties of injected antibodies against the Ki-67 protein (Starborg et al., 1996). A recent study has suggested Ki-67 is a protein phosphatase-1 binding protein, involved in phosphor-regulation of the nucleolar protein B23 (also known as nucleophosmin) which is required for assembly of the peri-chromosomal compartment in human cells (Booth et al., 2014). The Ki-67 index has been shown to be a prognostic factor in other malignancies such as breast and lung cancer (Martin, 2004; Yerushalmi et al., 2010). It plays an even more critical role in NETs, as a higher Ki-67 index is one of the strongest indicators currently available for a more aggressive biological phenotype. A gastroenteropancreatic NET with a Ki-67 index of 2%; Comparison 2: = 5%
OS OS/PFS
La Rosa 1995 (La Rosa et al., 1996) Pelosi 1996 (Pelosi et al., 1996)
PNET (50) PNET (54)
2% 5%
OS OS
Gentil Perret 1998 (Gentil Perret, 1998) Clarke 1997 (Clarke et al., 1997) Hochwald 2002 (Hochwald et al., 2002) Rigaud 2001 (Rigaud, 2001) Rindi JNCI 2012 (Rindi, 2012) Panzuto 2011(Panzuto et al., 2011)
PNET (33) PNET (37) PNET (136) PNET (16) PNET (1072) Advanced PNET (202) Met GEPNET (63) NET (31) GEPNET (907) Met GEPNET (399) PNET (324) Resected PNET (57) PNET (274) G3 NEC (305) GEPNETS (32 GI, 5 pancreatic)
2000 cells 1000 cells − repeated*3 1000 cells 1000 cells Cells per HPF NA NA 2000 cells
Yes − UV/MV Comparison 1: Yes UV/MV for PFS, Yes UV (Not MV) for OS. Comparison 2: Yes for UV/MV for both OS and PFS Yes − UV, No − MV Yes − UV/MV
4% 10% 50 per 10HPF 2% 4.85% 5%
OS OS DFS/DSS OS OS PFS
2000 cells 1000 cells NA NA 500 cells 2000 cells 2000 cells NA Cells per mm2 (absolute)
15% 15% 20 10 2% 5% 20% 55% 1 cells/mm2
OS OS OS OS OS DFS OS OS/Response rate OS
Yes − UV Yes − UV Yes − UV, No − MV Yes − UV/MV Yes − UV/MV Yes − UV. Not significant at 2% Yes − UV/MV Yes − UV/MV Yes − UV/MV Yes − UV/MV Yes − UV/MV Yes − UV/MW Yes − UV/MV Yes − UV, No − MV Yes − UV
400 cells
2%
OS
No − UV
2000 cells
4%
DFS
Yes − UV/MV
2000 cells NA 400 cells
5.4% NA 0, 1%, 2%, 3%, 4%, 5–20%, 21–30%, 31–40%, 41–60%, >60% 5%
DFS OS OS
Yes − UV/MV Yes − UV Yes − UV, No − MV
OS
Yes − UV, No − MV
Continuous variable
OS
Yes − UV/MV
Hentic 2011 (Hentic et al., 2011) Faggiano 2008 (Faggiano et al., 2008) Garcia-Carbonero 2010 (Garcia-Carbonero et al., 2010) Pape 2008 (Pape et al., 2008a) Ekeblad 2008 Ekeblad et al., 2008 Boninsegna 2012 (Boninsegna et al., 2012) Scarpa 2010 (Scarpa et al., 2010) Sorbye 2013 (Sorbye et al., 2013) Chaudhry 1992 (Chaudhry et al., 1992) 2) Bronchial NETs Zahel 2012 (Zahel et al., 2012) Grimaldi 2011 (Grimaldi et al., 2011) Rugge 2008 (Rugge et al., 2008) Granberg 2000 (Granberg et al., 2000) Skov 2010 (Skov et al., 2010)
Walts 2012 (Walts et al., 2012) Liu 2014 (Liu et al., 2014)
Bronchial NET (200) Bronchial NET (106) Bronchial NET (67) Bronchial NET (43) Bronchial NET (238)
Bronchial NET (101) Bronchial NET (55)
8 HPF 2000 cells
UV = Univariate analysis; MV = Multivariate analysis; PFS = Progression-free survival, DFS = Disease-free survival, PNET = pancreatic NET OS = OVERALL SURVIVAL
even direct clinicians to the optimal site for biopsy to identify the highest grade disease. In summary, significant data exist to validate the prognostic value of Ki-67 in GEPNETs, with relatively less data in bronchial NETs. The current grading guidelines regarding grading of NETs, as shown above, suggest cut-offs of 3% and 20% for GEPNETs. Whilst Ki-67 has not been included in classification schema for bronchial NETs thus far, various guidelines recognise its value in addition to the mitotic count (Oberg et al., 2012). Ki-67 is still being investigated as a predictive biomarker. Studies for validation of predictive biomarkers should ideally be either prospective analyses, or failing that retrospective analyses of well-designed randomized trials (Mandrekar and Sargent, 2009). However, the bulk of studies investigating this issue have been retrospective cohort studies. Considering recent RCTs, PROMID was unable to evaluate Ki-67 in this context as 98% of patients had Ki–67 = 1 CTC was approximately 40%, compared to 68% for those with 3 was significantly associated with progression-free survival. Of note, Ki-67 and chromogranin A were not significant predictors of PFS on multivariate analysis whilst SUVmax > 3 remained significant. In a retrospective series of 37 patients with bronchial NETs, FDG PET SUVmax was shown to correlate to pathological aggressiveness, with a suggestion that SUVmax > 13.7 was associated with poorer prognosis in SCLC and LCNEC (Chong et al., 2007). However, this has not been prospectively validated. Future studies should investigate the role of FDG PET as a marker of response and as a predictive biomarker. There have been multiple sets of criteria proposed to evaluate FDG PET response (Choi, EORTC and PERCIST), none of which have been validated in NETs (Choi et al., 2007b; Wahl et al., 2009; Young et al., 1999). There is no agreement about which set should be employed, and no proposals as yet for application of these criteria to serial 68 Ga PET imaging. The above criteria will have to be unified before validation of their utility is able to take place. 4.2.3. Dual 68 Ga/FDG PET imaging and prognosis Dual 68 Ga/FDG PET imaging is of potential value in NET imaging because the two scans are complementary. 68 Ga PET indicates the presence of well-differentiated, G1/2 NET as these are the tumour cells that retain the somatostatin receptor targeted by 68 Ga. FDG PET, on the other hand, demonstrates the presence of highly metabolically active disease, which is more likely to be G3 NET which may be negative on 68 Ga PET (Binderup et al., 2010b). Dual
277
68 Ga/FDG
imaging has been investigated for its value in imaging both GEPNETs (Kayani et al., 2008) and bronchial NETs (Kayani et al., 2009; Jindal et al., 2010; Venkitaraman et al., 2014), but this has been hampered by the lack of an objective scoring criteria or schema. As noted above, criteria exist to determine response on FDG PET, but no similar criteria exist for 68 Ga PET. Indeed, application of the same criteria to 68 Ga PET would not be appropriate given that increasing intensity of a lesion on 68 Ga PET may either signify conversion of a signal well-differentiated tumour (with better prognosis) or alternatively tumour progression with increasing number of cells expressing the receptors. An additional challenge is the need for anatomical correlation between the two PET scans, as well as a reproducible way to interpret the relative intensity of a lesion on the two scans. Few studies have investigated the potential prognostic value of 68 Ga/FDG PET. The present authors have proposed a dual 68 Ga/FDG PET grading scheme (Chan et al., 2017) with patients with FDG + Ga68- lesions having poorer overall survival. This scheme (NETPET score) grades a pair of 68 Ga/FDG scans on a scale of 1–5, with 1 representing 68 Ga avid, FDG negative disease and 5 representing the presence of significant 68 Ga negative, FDG avid disease. The NETPET score stratifies patients into three groups – with scores of 1, 2–4 and 5 – with this stratification being significantly correlated with overall survival. Furthermore, significant numbers of patients with Grade 1 disease have NETPET scores of more than 1–indicating significant FDG avidity and potentially heterogeneous disease (Chan et al., 2016b). The area of dual 68 Ga/FDG imaging deserves further investigation as both a prognostic and predictive biomarker.
4.2.4. Nuclear imaging as predictive biomarkers Nuclear imaging can demonstrate the density of somatostatin receptor expression in NET cells throughout the body. Therefore, it could theoretically predict the uptake of somatostatin analogues in a particular patient and hence the likelihood of clinical benefit to that patient. Uptake on Octreoscan is associated with response to somatostatin analogues (Janson et al., 1994), and as a result CLARINET (Caplin et al., 2014) required patients to have positive somatostatin receptor scintigraphy (defined as Grade 2 on the Krenning scale – uptake more than the baseline liver level) for inclusion in the study. A similar logic applies to the use of nuclear imaging as a predictive biomarker for PRRT. PRRT – particularly lutetium-based therapies – will find an even larger place in NET treatment after preliminary reports from the randomized NETTER-1 trial (Strosberg et al., 2016) demonstrating an impressive improvement to PFS (HR 0.21) and potential improvement to OS in 230 patients with midgut NETs. Given that PRRT needs the presence of somatostatin receptors for internalization of the radionuclide, it is logical to use somatostatin imaging (whether Octreoscan or DOTA-based PET) to predict PRRT delivery (and hence efficacy) in the individual patient. A study of 33 patients receiving PRRT demonstrated that decrease in the tumour-to-spleen SUV ratio after cycle 1 (but not decrease in SUVmax) predicted for longer progression-free survival as well as clinical improvement (Kulkarni et al., 2011). The strongest signal for 68 Ga −DOTA PET/CT as a predictive biomarker comes from a study by Kratochwil et al. (Kratochwil et al., 2015) which analysed the baseline PET/CT scans of 30 patients with NETs undergoing PRRT. A lesion-level analysis was conducted looking for markers of response to PRRT. Baseline SUVmax was higher in responding lesions comparing to non-responding lesions, and SUVmax > 16.4 on DOTATOC-PET predicted for response on CT after 3 cycles of PRRT, with 95% sensitivity but only 60% specificity. Future research should investigate different data parameters to refine 68Ga-DOTA PET as a predictive biomarker for PRRT. Our
278
D.L. Chan et al. / Critical Reviews in Oncology/Hematology 113 (2017) 268–282
local institution considers the presence of significant FDG positive, Ga68 negative lesions as a relative contraindication to PRRT assuming that it indicates the presence of de-differentiated tumour that will not be treated efficaciously by PRRT. On the other hand, lesions which are both FDG and 68 Ga avid may be treated with PRRT, and patients with these findings experience a high disease control rate when treated with PRRT (Kashyap et al., 2015). 5. Predictive nomograms Several nomograms have been developed in an attempt to summarise the above risk factors into one predictive score. One such nomogram utilised the SEER database to assess survival in small intestinal NETs (Modlin et al., 2010), with one number derived from the factors of age, gender, ethnicity, symptoms, urinary 5HIAA, chromogranin A, liver function tests, tumour size, invasion, metastases, histology, Ki-67 index, presence/absence of carcinoid heart disease and therapy (surgery or SSA). However, despite a large derivation cohort of approximately 20,000 patients, validation was performed on only 33 patients. Another simpler nomogram, using grade, gender and age, has been developed for non-functional pancreatic neuroendocrine tumours (based on 326 patients) showing slightly improved discriminatory power compared to stage alone (Ellison et al., 2014). Other similar nomograms have recently been presented in abstract form for resected neuroendocrine tumours (Genc et al., 2016), NF-PNETs (Han et al., 2016; Jilesen et al., 2015), and resected neuroendocrine liver metastases (Ruzzenente et al., 2016). However, all of these nomograms were derived from relatively small cohorts (