Results
The study decision tree model is presented in Figure 1 and the Markov models for possible outcomes in Figure 2. We found that the overall benefit, as measured by QALYs, varies according to BARFV600E status; for patients with BRAFV600E positive tumors, the QALYs are 38.58 and 36.56 for HT strategy and AS respectively, with a difference of 2.02, and for patients withBRAFV600E negative tumors, the QALYs are 39.43 and 37.77 for HT and AS strategy respectively, with a difference of 1.66.
Using one-way sensitivity analysis, we found that the main two variables that have a strong impact on the decision are the utility of AS health state and the utility of disease-free state after HT without complication, as presented in the tornado diagrams (Figure 3).
To assess the influence of the genetic status on these two variables, we conducted a two-way sensitivity analysis of the two utility variables in the scenarios ofBRAFV600E positive and negative tumors (Figure 1). The range of utility values in which HT will be the preferred strategy is wider in patients with positiveBRAFV600E compared to patients with negative toBRAFV600E (Figure 4).
The model predictions indicate that in order to determine the optimal strategy for a specific patient, the physician needs to first assess the patient’s perceived utility values for AS and post-HT without complication health states, and then to consider the need for genetic testing (see Figure 5 for the suggested management algorithm). According to this personalized approach, patients can fall into three groups with respect to their utility scores: (i) patients with high values for AS and low values for post-HT - would be recommended AS; (ii) patients with low values for AS and high values for post-HT - would be recommended surgery; (iii) patients with middle utility values - would be recommended genetic testing (Figure 4, grey area) to tailor each patient’s optimal personalized management plan.