P-L13 Proposed sampling methodology for improved accuracy of hepatic burden assessment in neuroendocrine tumours

Abstract

Abstract Background Hepatic burden is a significant confounder in the assessment of impact of primary tumour resection in metastatic small bowel neuroendocrine tumours (SI-NET). For SI-NET metastatic hepatic burden >10% disease replacement or > 5 hepatic metastases are known prognostic markers, though nomograms and scores do not adequately account for this. Most trials do not adequately account for hepatic burden when assessing the survival difference between SI-NET primary tumour resection and no resection. We propose a sampling methodology to more accurately assess metastatic liver burden in SI-NET and correlate with delayed resection vs. upfront primary tumour resection at a specialist NET surgical unit. Methods Patients referred for metastatic SI-NET between January 2003 and February 2020 were identified from a prospective dataset. The earliest CT scan after diagnosis was used. The axial, coronal and sagittal slice position limits of the whole liver were recorded. These limits allowed equitable slice position of the liver, with 8 equally distributed axial, 4 equally distributed coronal and 4 equally distributed sagittal slices. Each slice was used to define the liver and metastatic area as assessed using liver CT windows. Liver burden was estimated as percentage total metastatic area summed from all 8 axial, 4 coronal and 4 sagittal slices. Results 157 total patients were on the collated data base and 46 patients were identified with an appropriate CT. Liver burden was positively skewed. Liver burden was significantly higher for delayed resection vs. upfront resection in all planes of assessment (axial: 11.61% vs. 0.14%, p = 0.003; coronal: 13.46% vs. 0.33%, p = 0.006; sagittal: 10.46% vs. 0.16%, p = 0.008). All planar assessments correlated well with one another (all Kendall’s tau ≥0.851, all p < 0.001). Liver metastatic burden correlated with total liver volume (Kendall’s tau 0.549-0.573, all p < 0.001). Conclusions Hepatic burden differs between resection groups in a small sample at our centre, highlighting the unmeasured confounders favouring primary tumour resection via positive bias. Therefore, hepatic burden needs quantifying in prospective studies that assess primary tumour resection in SI-NET. This is to ensure comparable groups after randomisation. Our method provides an assessment of this metastatic SI-NET liver burden.