The hepatocellular carcinoma Early Recurrence Score (ERS)—Ready for clinical implementation?

Surgical resection for patients with hepatocellular carcinoma (HCC) can be curative but is not without risks and recurrence rates are high. Clinical risk prediction tools use patient and disease factors to estimate the probability of an outcome and can in turn assist clinical decision-making.1 Over the last 20 years, there has been a surge in published articles centred on risk prediction for HCC (Figure 1). The availability of large data sets incorporating clinical and molecular information, together with more advanced data analytical models, has provided opportunities to improve the strength of these risk prediction tools.1 Developing risk prediction tools, however, is not without challenges. Guidelines for the development and reporting of clinical prediction research have been established to support this.2, 3 Published in this journal of Liver International, Consentin et al. (In Press) have followed these guidelines, retrospectively analysing available data from a cohort of 2539 patients with Barcelona Clinic for Liver Cancer (BCLC) 0/A stage HCC, undergoing surgical resection in 17 centres across Europe and Asia, with the aim of developing a score to predict early HCC recurrence. Having the ability to predict the risk of recurrence for patients being considered for liver resection for HCC using pre- or post-operative data can potentially inform clinical decision-making, including which patients are more likely to benefit from surgery, structuring surveillance following surgery, as well as identifying those that may benefit from future neoadjuvant or adjuvant therapies. Risk prediction tools may also aid in comparing patients within different trials. Consentin et al. (In Press) have proposed an Early Recurrence Score (ERS), using 6 variables that are established as independent predictors of early recurrence, accumulating points associated with risk features, up to a maximum of 11. These included alpha-fetoprotein, scoring 0–3 points based on levels of <10 mg/mL, 10–100 ng/mL and >100 ng/mL; size of largest nodule >40 mm (1 point); the presence of multifocality (2 points); the presence of satellite nodules (2 points); the presence of vascular invasion (1 point) and surgical margin positive (2 points). The ERS facilitated stratification of cases into four categories of 2-year recurrence risk. The ERS performed better than any of its individual components and outperformed existing post-operative risk prediction models4 when predicting time to recurrence and recurrence-free survival. The absence of other factors known to predict recurrence after HCC resection (portal hypertension, liver synthetic function or operative characteristics) from the ERS model does not appear to have disadvantaged its performance. In creating the ERS, the authors have used a complex cohort, divided into a development and validation groups using random sampling of participating centres, before comparing the performance to different predictive scoring systems and clinically relevant endpoints. As discussed by the authors, these are challenging analyses,4, 5 particularly given the heterogeneity of the cohort on multiple levels, with available data sets varying in completeness in different institutions. The parameters included were influenced to a degree by the data available, but as such it has the potential to be widely applicable. None the less, independent validation, preferably in a prospective fashion, will be important to confirm its superiority to other tools before it can be adopted into clinical practice. It is also important to consider that future tools may be considerably enhanced by the inclusion of data centred on the molecular characteristics of HCC, the application of machine learning and artificial intelligence, as these become more available. But the use of parameters that are readily available to surgical teams globally, make this a simple, noteworthy and attractive tool, with potential clinical relevance, paralleling the established models for the prediction of recurrence after liver resection for colorectal liver metastases6, 7 that have become embedded in practice within the liver surgical community. Data sharing is not applicable to this article as no data sets were generated or analysed during the current study.