Reply

Abstract

We thank Cross et al.1 for their interest in our prospective validation of the Hospital Universitario La Fe index. We agree that the complexity of the calculation limits its utility in a busy clinic, but in the online version of the article,2 readers can find an Excel file with the mathematical formula as additional supporting information. Users can download our ready-to-use Excel file to a personal computer or a personal digital assistant. As for comparisons of the different staging methods for liver biopsy samples, Ishak (the pathologist who worked with Knodell) modified Knodell's score to create his own score, with only small differences between them. A recent work showed excellent concordance for fibrotic stages between the METAVIR and Ishak scoring systems in both hepatitis B virus and hepatitis C virus patients.3 Moreover, most publications that have used kappa statistics to evaluate scoring systems for inflammation and fibrosis in liver diseases have reported better levels of agreement for fibrosis than for inflammation.4 The approximate equivalence between the 3 staging systems is shown in Table 1. To try to understand how the pathological staging system influenced the results in our study, our pathologist also reviewed a number of biopsy samples in our center according to the Ishak score (data not shown). The same biopsy samples were scored according to the Knodell system, and then the results were compared. The agreement with respect to significant fibrosis was more than 90%; that is, a large majority of the biopsy samples with a Knodell score of ≥F3 were also reported to have an Ishak score of ≥F3. These results are consistent with those reported by other authors5 and allow the comparison of different staging systems with respect to significant fibrosis. One limitation of our study was the lack of standardization of the area under the receiver operating characteristic curves, as pointed out by Cross et al.1 We did not perform the standardization because the receiver operating characteristic curve adjustment proposed by Poynard et al.6 refers to patients with chronic hepatitis C (nontransplanted), so the formula used with the constant of 0.1056 and DANA (difference between the mean of advanced fibrosis stages and the mean of nonadvanced fibrosis stages) of 2.16 would not necessarily be extrapolated to the population of transplanted patients. Nevertheless, we sought to calculate the standardized area under the receiver operating characteristic curves of our previous studies2, 7 and to compare them to the results of the London Transplant Centres' score published by Cross et al.8 For the calculation of DANA, we took into account the equivalence between stages of different scales (Table 1) and the frequency distribution of METAVIR stages, as shown in Table 2. Knodell stage F3 overlaps with METAVIR stages F2 and F3, so we have multiplied by 2.5 (2 + 3/2) the prevalence of this group. As shown in Table 3, both models (the Hospital Universitario La Fe index and the London Transplant Centres' score) are quite similar with respect to the accuracy in predicting significant fibrosis in liver biopsy samples of patients who have undergone transplantation. Future, prospective studies should validate the use of the combination noninvasive scores. Salvador Benlloch*, Marina Berenguer*, Joaquin Primo , * Department of Hepatogastroenterology, Hospital Universitario La Fe, Valencia, Spain, Department of Internal Medicine, Hospital of Sagunto, Valencia, Spain.