If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Liver Disease and Digestive Interventional Radiology, National Clinical Research Center for Digestive Diseases and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an 710032, Shaanxi Province, China
Corresponding author. Address: Department of Liver Diseases and Interventional Radiology, Xi'an International Medical Center Hospital, Northwestern University, Xi'an, China. Tel.: +86-29-84771537, fax: +86-29-82539041.
Transjugular intrahepatic portosystemic shunt (TIPS) implantation is an effective and safe treatment for complications of portal hypertension. Survival prediction is important in these patients as they constitute a high-risk population. Therefore, the aim of our study was to develop an alternative prognostic model for accurate survival prediction after planned TIPS implantation.
With great interest, we have read the letters from Kraglund et al.1 and Wang et al.2 who provided external validation of the FIPS score3 and also proposed a detailed risk stratification combining the Child-Pugh score and the FIPS score.
We have read with great interest the impressive study conducted by Bettinger et al. about the development of the Freiburg index of post-TIPS survival (FIPS) score.
The authors established a prognostic model to achieve both individual outcome prediction (with the formula and the online calculator) and risk stratification (with a cut-off point), which had superior performance compared to previous prognostic models. Moreover, its ability to stratify patients at high and low risk remained robust in subgroups with different indications for transjugular intrahepatic portosystemic shunt (TIPS), different types of TIPS stents, well-preserved liver function, and impaired renal function.
Since the FIPS score is developed on the basis of a training cohort which consisted of German patients, we hoped to externally validate it and to assess its prognostic value within a Chinese TIPS cohort of 536 patients. In this cohort, the indication of TIPS included secondary prophylaxis of variceal rebleeding (n = 468, 87.3%), refractory ascites (n = 60, 11.2%) and early TIPS (n = 8, 1.5%). Among these patients, 109 (20.3%), 350 (65.3%), and 77 (14.4%) patients were graded as Child-Pugh A, B, and C, respectively, and the median model for end-stage liver disease (MELD) score was 11.5 (IQR 9.7–141). Major etiologies of chronic liver disease included HBV infection (292, 54.5%), autoimmune liver disease (53, 9.9%), and HCV infection (7.8%). Median follow-up was 23.7 (IQR 16.8–36.2) months.
We first tested the performance of FIPS score, Child-Pugh score,
Regarding the time-dependent AUROC for 3-month, 6-month, 1-year, 2-year, and 3-year death, Child-Pugh score had the best performance at all time points, whereas corresponding AUROCs of FIPS score were slightly lower (Fig. 1A, Table S1). Similarly, the C-indices for Child-Pugh score and FIPS score were 0.68 and 0.66, respectively (Table S1).
Fig. 1Performance and risk stratification of prognostic models.
(A) Time-dependent AUROC of the FIPS score, Child-Pugh score, MELD score, and CLIF C-AD score.
(B) Risk stratification of the FIPS score with 0 point being the cut-off. Survival was significantly different between the 2 strata with log-rank p <0.01. (C) Risk stratification of Child-Pugh score. Survival was significantly different between grades A, B, and C with log-rank p <0.01. (D) Risk stratification of the FIPS score in patients with Child-Pugh A. Survival was significantly different between the 2 strata with log-rank p <0.01. (E) Risk stratification of the FIPS score in patients with Child-Pugh B. Survival was significantly different between the 2 strata with log-rank p <0.01. (F) Risk stratification of the FIPS score in patients with Child-Pugh C. There was no statistical difference between the 2 strata with log-rank p = 0.10. (G) Risk stratification of Child-Pugh score in low-risk patients defined by FIPS score. Survival was significantly different between different grades with log-rank p <0.01. (H) Risk stratification of Child-Pugh score in high-risk patients defined by FIPS score. There was no statistical difference between the 3 grades with log-rank p = 0.56. FIPS, Freiburg index of post-TIPS survival; MELD, model for end-stage liver disease; TIPS, transjugular intrahepatic portosystemic shunt.
Subsequently, we validated the discriminative ability of FIPS score and Child-Pugh score. According to the original cut-off point of the FIPS score (0.92), only 7 patients were identified as high-risk patients, with significantly higher mortality than those with low risk (log-rank p <0.01, Fig. S1). Given the extremely low number of high-risk patients, we replaced the original cut-off with -0.006, the 85th percentile in the current cohort. For ease of use, we further modified it to 0 since only 1 patient had a FIPS score falling between -0.006 and 0, and the capability of risk stratification remained robust (log-rank p <0.01, Fig. 1B). Similarly, Child-Pugh score could also achieve adequate risk stratification with its grading system (log-rank p <0.01, Fig. 1C).
Finally, we investigated the risk stratification effect of the FIPS score in different Child-Pugh grades and vice versa. The FIPS score could significantly discriminate patients with high and low risk in Child-Pugh A and B subgroups (both with log-rank p <0.01, Fig. 1D, E), but not in the Child-Pugh C subgroup (log-rank p = 0.10, Fig. 1F). Inversely, in the low-risk subgroup defined by FIPS score, Child-Pugh score could stratify patients into grades A, B, and C (log-rank p <0.01, Fig. 1G). While in high-risk subgroup, Child-Pugh score did not reach statistical significance for risk stratification (log-rank p = 0.56, Fig. 1H).
In the original study, Bettinger et al. focused on the performance of FIPS score in predicting 3-month and 6-month survival, which proved satisfactory. Accordingly, we confirmed that FIPS score has similar AUROC for predicting 3-month and 6-month survival in this external validation cohort, and that patients with high and low risk of death could be well-stratified. Indeed, FIPS score might be a landmark since it is the first model to achieve individual outcome prediction for patients receiving TIPS. However, in our cohort, Child-Pugh score appeared to be a more favourable choice with better performance (Fig. 1A, Table S1) if a single prognostic model is to be independently used for risk stratification, since it had the best performance. Interestingly, in subgroup analyses, FIPS score could further stratify risk levels in patients even if they were classified as low and intermediate risk groups according to Child-Pugh score (i.e., grade A and B), and Child-Pugh score could also stratify low-risk patients defined by FIPS score. These results might be caused by the different variables used in these 2 scoring systems, and consequently indicated that FIPS score and Child-Pugh score are to some degree complementary in identifying high-risk patients. Therefore, when a more accurate and detailed risk stratification is required, the FIPS risk stratification based on Child-Pugh grading system could be a new solution.
In summary, for individual outcome prediction of survival after TIPS, FIPS score is the best option; whereas for risk stratification, Child-Pugh score appeared to be more favorable for Chinese patients, while FIPS score could provide a more detailed and accurate "secondary" risk stratification on the basis of Child-Pugh grade.
Financial support
This study was supported by grants from National Natural Science Foundation of China (81420108020).
Authors’ contributions
Study concept and study design: Qiuhe Wang, and Guohong Han. Follow-up and data collection: Wei Bai. Statistical analyses: Qiuhe Wang. Drafting and revision of the manuscript: Qiuhe Wang, Wei Bai, and Guohong Han.
Data availability statement
The deidentified data can be made available upon request for non-commercial purposes and after approval of a study proposal through a signed data access agreement. Proposals should be directed to the corresponding author ([email protected]).
Conflict of interest
Prof. Han is listed as a co-author in the FIPS study by Bettinger et al. due to his role of data acquisition.
Supplementary data
The following are the supplementary data to this article:
00434-7&id=gr1.jpg){kind=link}