We read with interest the News and Views article by Rindi and Wiedenmann (Neuroendocrine neoplasia goes molecular — time for a change. Nat. Rev. Clin. Oncol. 16, 149–150 (2018))1, two notable and eminent authorities in the field of genomic medicine. Despite the concerns raised by these experts, the registry study discussed in their article unequivocally demonstrated the clinical utility of the NETest liquid biopsy assay in the ‘real-world’ setting2. NETest scores enabled accurate prediction (100%) of disease progression and had a 98% correlation with response to therapy, reducing the requirement to perform an imaging test in 40% of patients.

NETest is the first blood-based multianalyte gene transcript (mRNA-based) test for neuroendocrine tumours (NETs). The genomic validity of this assay was independently confirmed in an NIH-funded study evaluating the gene expression profiles of >10,000 tumours from The Cancer Genome Atlas (TCGA) database, with the results indicating that the NETest gene signature accurately captured the NET phenotype2. Our registry study confirms the clinical utility of this signature.

The following factual inaccuracies were stated in the News and Views by Rindi and Wiedenmann1: that the Response Evaluation Criteria in Solid Tumors (RECIST) always enable an accurate assessment of NET progression; that our data should be viewed cautiously owing to tumour heterogeneity; that we provided no sensitivity data; that patient management depends on a multidisciplinary team (MDT) approach; and that the cost–benefit ratio of NETest is unfavourable.

First, RECIST criteria are of limited value in the assessment of indolent tumours, such as NETs, and have no predictive value in this disease3. The RECIST are a ‘trial tool’ not used in clinical practice. NETest, however, has clinical utility as the first blood-based objective genomic assay to facilitate identification of the molecular mechanisms that define individual tumour behaviour. NETest enables detection of 1 cell/ml of blood and is ~30,000–50,000 times more sensitive than imaging, which provides only volumetric data4.

Second, the real-world study captured the usual mix of NETs commonly encountered by physicians. We agree that tumour heterogeneity is an issue when performing tissue biopsies; hence, the limited value of NET tissue biopsy analysis as an accurate prognostic tool5. Using liquid biopsy approaches, however, all clonal components are sampled. NETest was effective as a diagnostic and prognostic tool in grade 1–2 tumours4, making it evident that heterogeneity is an issue derived from performing tissue biopsy.

Third, our clinical study was not designed to examine laboratory parameters because such data has already been extensively published. For example, the diagnostic accuracy of NETest for either lung or gastroenteropatic NETs is >90%4 — in the registry study2, it was 96% irrespective of tumour origin, with a coefficient of variation (CV) of <5%. Both the diagnostic and the laboratory analytic metrics exceeded those obtained using serum chromogranin A (CgA) assays. The latter biomarker exhibits a diagnostic accuracy of ~50–70% and a CV of <20%. NETest is specifically designed to be used for patient management and was 85% concordant with the disease status (for example, stable or progressive). This percentage surpasses the 80% metrics proposed by the NIH6 and exceeds the performance of CgA assays, which have a <50% concordance with disease status7.

Fourth, an MDT approach is important in the management of patients with all tumour types and not just those with NETs. The key issue with MDTs is that their validity in making consensus decisions is only as effective as the information they consider. MDTs evaluate imaging data, which can be interpreted subjectively and can be unreliable in guiding the management of patients with NETs8. In our study, a low NETest score enabled the identification of patients with molecularly stable disease. As a consequence, 40% of patients required fewer follow-up interventions than they would have received if they had been evaluated according to standard procedures. This result has obvious clinical and economic implications.

Fifth, blood biomarker information can provide an indication of disease status earlier than is possible with imaging and thereby facilitate earlier cessation of an ineffective therapy — an obvious cost benefit9. Economic modelling studies conclude that patient monitoring with NETest would substantially decrease the expenditure in dollars per year per patient. Imaging costs are US$6,000–10,000 per event and the cost of standard-of-care treatments (such as somatostatin analogues, peptide receptor radionuclide therapy or everolimus) are $50,000–100,000 per year per patient. Thus, a biomarker that enables decreased use of interventions in 40% of patients and alters the management of another 40% has cogent economic implications.

Iconoclastic views of novel sophisticated technologies has culminated in the management of NETs falling behind other oncological disciplines (such as the management of breast cancer)10, which ultimately is detrimental to patients. Rindi and Wiedenmann disappointingly reiterate the tautology “reaching perfection is almost impossible in the real world”1. On the contrary, only through the thoughtful adoption of novel advances rather than clinging to dogma can medicine reach for perfection and thereby provide the best care for our patients4.

There is a reply to this correspondence by Rindi, G. & Wiedenmann, B. Nat. Rev. Clin. Oncol. https://doi.org/10.1038/s41571-019-0186-4 (2019).