The current study is a multicentre international collaboration combining cohorts of patients with metastatic GEPNEN who had undergone dual PET imaging on comparable current generation PET/CT scanners and populations. GEPNEN patient data from individual centres has been reported in-part previously [10,11,12], but the analysis of combined multicentre, international data makes this the largest study to date to investigate the prognostic value of dual PET imaging. We conclude that the NETPET score is significantly prognostic for OS and TTP on both univariate and multivariate analyses.

The NETPET score stratifies patients with metastatic GEPNENs into three prognostic classes, which strongly predict for OS and TTP after accounting for histological grade and other known prognostic factors. The NETPET score is correlated with histological grade but may also highlight situations where historically “under-grading” has occurred (such as reliance on a biopsy from a single site for prognostication). Whilst dual PET imaging can be considered for all patients with a diagnosis of metastatic GEPNEN, it may be of particular utility for grade 2–3 GEPNENs which have wide biological heterogeneity, where the results of a single biopsy site may not reflect the overall biology of the disease in individual patients.

Previous studies from the three participating centres have investigated the prognostic value of the NETPET score in smaller groups of patients [10, 11, 13, 14], with Chan et al. [10] including patients with primaries of any site. Similar smaller studies have also been conducted in bronchial NENs; and whilst the prognostic impact of the NETPET score was preserved [15], a significant minority of these patients may be non-avid on both [68Ga]DOTATATE and [18F]FDG PET, a finding that deserves further investigation [16]. Dual PET imaging has also been investigated in other settings, such as in directing management of GEPNENs [17] and in diagnostic NEN workup [18], and prior to PRRT [19]. The current study performed a repeat interrogation of relevant databases to identify eligible patients imaged since the completion of the original studies. The current study is one of the few to investigate the presence/absence of both [68Ga]DOTATATE and [18F]FDG avidity in the same tumour using spatial correlation, rather than just determining overall avidity on each scan individually. A significant correlation was noted in this study between histological grade and the NETPET score. A patient with P5 classification in the current cohort was extremely unlikely to have low-grade disease; additionally, a patient with grade 1 disease was unlikely to have a P5 classification (i.e. [18F]FDG avid, [68Ga]DOTATATE non-avid disease).

The strengths of this study include the large number of patients with an uncommon disease from three independent centres in an international collaboration, and the use of harmonised acquisition protocols in all three PET systems. The same objective and reproducible imaging analysis protocol was also applied across all centres by nuclear medicine physicians with expertise in NENs, with high inter-rater and intra-rater reliability. A large number of potential prognostic covariates were collected and allowed for multivariate analysis, which is important in a heterogeneous disease entity such as NENs. We acknowledge that the retrospective nature of this study makes it susceptible to bias, although consecutive series of patients were identified at each participating centre. Furthermore, the interval between PET scans (maximum 90 days) allows the potential for lesions to shift or grow, impeding accurate scan comparison. However, this is unlikely to significantly impact our data given the median interval between the two scans was 9 days, while the median cohort TTP was 17.1 months. We have not reported the impact of different treatments received by individual patients, potentially introducing bias in the TTP and OS analysis. Particularly, the commencement of PRRT may present a survival advantage in the P1–4 cohort, which is currently being investigated by the authors. We also acknowledge the non-standardised approach to determine progression as a limitation of this study. The above findings should ideally be confirmed in a prospective clinical trial, using the NETPET score as an exploratory/novel endpoint, although the uncommon nature of NENs would make such a study difficult to accrue.

Whilst the current study is a retrospective analysis, several changes to clinical practice in NENs should be considered as a result. The use of the NETPET score as a prognostic biomarker is supported by its significance on univariate and multivariate analysis, and it may serve to identify patients with a potentially aggressive disease course. Furthermore, the utility of dual PET imaging is highlighted by 59% of grade 1 patients (a group typically seen as ‘low risk’) demonstrating [18F]FDG avidity. These patients may be considered for closer follow-up or more aggressive choices of systemic treatment, perhaps pending a dual PET guided biopsy to identify potential development of high-grade disease (i.e. the site of greatest [18F]FDG avidity). Importantly, this study highlights the potential contribution of dual PET imaging in directing the care of patients with advanced NENs. Further research is needed to determine if the NETPET score can also predict response to treatment on repeat imaging.

Histological grade and differentiation are well-established markers for prognosis, and remain so in our overall cohort (Supplementary Fig. S2). Preliminary results (Supplementary Fig. S1) suggest that the NETPET score further stratifies traditional histological grading, however, larger numbers and longer follow-up is required to validate this small subgroup analysis. Similarly, accrual of larger numbers of grade 3 patients is required to assess whether the NETPET score is prognostic within the well-differentiated and poorly differentiated subgroups. This is important for future evaluation as the NETPET score potentially highlights areas of more aggressive disease for monitoring and/or additional treatment (such as radiotherapy), where traditional histology does not.

We recommend dual PET imaging for patients with rapidly progressing disease (irrespective of original biopsy findings), and those with grade 2 or well-differentiated grade 3 disease, in order to identify areas of potentially [18F]FDG avid, non-[68Ga]DOTATATE avid (i.e. discordant) disease. Such sites may not respond to SSTR-dependent therapies, thus influencing treatment selection. However, we acknowledge that economic considerations such as funding for dual tracers may limit the widespread adoption of routine dual PET imaging into clinical practice.

Several avenues of exploration are suggested by the findings presented here. Quantitative analysis of both [18F]FDG and [68Ga]DOTATATE PET scans is underway in the current cohort, in order to compare the prognostic value of PET-volumetric data based on each scan, an area previously investigated in smaller cohorts [10, 20, 21]. Lesion-based analysis and comparison of histopathological characteristics to PET findings may provide unequivocal evidence that PET imaging can act as a “virtual biopsy” to predict disease biology and evolution. It remains unknown whether the NETPET score may predict for the efficacy of PRRT, chemotherapy, and other systemic treatments such as molecular targeted therapy and SSAs. Therefore, validation of the NETPET score to predict treatment response and inform clinical management is of high interest. Finally, translational research into the cellular and molecular underpinnings of each [18F]FDG/[68Ga]DOTATATE phenotype will enable further insight into the biological behaviour predicted by dual PET imaging, ultimately leading to better therapy selection and superior patient outcomes.

Our large multicentre study validates the NETPET score as a robust prognostic biomarker of OS and TTP in patients with metastatic GEPNEN, and it represents a valuable complement to the prognostic algorithm. Dual PET imaging should be considered in all patients with a diagnosis of metastatic GEPNEN to guide the most optimal site for biopsy and inform the management approach.