Reply: Modeling concerns.

Abstract

We appreciate Dr. Tapper’s remarks on our paper1. As mentioned, due to the lack of longitudinal studies with MRE, modeling is a potent approach to connect disease progression with early detection and diagnosis. In addition, by raising questions and objections to the model, you allow us the opportunity to clarify and elaborate upon several points that deserve further attention. While the average age of the patient in the model is 50 years old, patient life span is modeled based on CDC, National Vital Statistics Reports life tables for the US population. The life expectancy for patients is based on age; hence the probability of a patient dying at any given age is predicated on the life tables. As ours is a lifetime model, the inclusion of life expectancy probabilities allows us to ensure that the population tapers off with age. In fact, the probability of dying at age 100 is 1.1 The incremental QALYs gained in the model are higher than the other estimates because we assumed that patients begin in the advanced fibrosis stage within the Markov model. This assumption leads to the effect of diagnosis and treatment to be greater than model stages based on the incidence at the outset. From the advanced fibrosis stage, the patients transition forward, or regress based on the transition probabilities referred in Table 2. As you stated, in practice, all patients may receive the lifestyle modification program at any stage. However, in the model, we assume that only patients identified as positive for advanced fibrosis will receive the lifestyle program as part of an intervention and regression of fibrosis was estimated at 19% per arm1 Lastly, within the MRE arm, the TP (83%) and FP (17%) received the lifestyle modification Within the VCTE arm, the TP (77%) and FP (22%) received the lifestyle modification.1 We agree that the assumption of prevalence of FIB4 > 2.67 to be 80% might be on the high side. However, there are no reliable data on this at the population level. Liver biopsy is required, which is considered as the gold standard for fibrosis staging, but is expensive and with risk to the patients. A study from the NIH NASH clinical Research Network, which consisted of subjects with histological proven NAFLD, evaluated the performance of noninvasive markers of fibrosis in patients with NAFLD. They reported that among their cohort of NAFLD patients, 41 (80%) of the 51 patients with FIB4 > 2.67 were found to have advanced fibrosis.2 The 2 strategies of MRE alone and VCTE+MRE have very similar costs and the difference is only in the magnitude of decimals (rounded in the table). We had hypothesized the VCTE+MRE arm to have higher costs because of including the additional scan. However, the proportion of patients being identified as positive for advanced fibrosis in the VCTE+MRE arm are lower than the MRE alone arm because only patients identified as positive by both are considered positive and move through the Markov stages and accrue costs. Essentially, it’s a proportionality difference that leads to the convergence in costs.