The validation cohort was composed of 1,055 patients with HCV-associated cirrhosis from the other 21 centers involved in the RESIST-HCV network.
There were no differences with respect to patient characteristics between the training and validation cohorts, except for the percentage of patients with EV (see Table 1, Supplementary Digital Content 1, http://links.lww.com/AJG/A183). The overall EV prevalence in the validation cohort was lower than that in the training cohort (42.4% vs 71.5%; P < 0.001), but the rates of large EV were similar between the groups (9.5% and 8.9%; P = 0.88).
In the validation cohort, the RESIST-HCV criteria were met by 315 patients (30%); among them, 75 (23.8%) had grade 1 varices and 6 (1.9%) had grade 2 or 3 EV (Figure 5).
Assessment of TE utility in the algorithm for noninvasive prediction of large varices
To assess the role of TE in predicting large EV, we first performed a pairwise comparison of AUROC for the 2 parameters of PLT count and serum albumin level included in the Baveno VI and Expanded Baveno VI criteria. We found that PLT count had superior performance in predicting large EV compared with TE (AUROC PLT = 0.77 vs AUROC LSM = 0.61; P = .0132; see Figure 1, Supplementary Digital Content 2, http://links.lww.com/AJG/A184). Indeed, on attempting to verify the ability of a single PLT count to predict large EV, we found that 83 of 326 patients (25.5%) had PLT counts higher than 150 × 109 cell/L. Results were similar to those found with the Baveno VI and Expanded Baveno VI criteria, as 45 patients (54.2%) had small EV and only 2 patients (2.4%) had large EV (see Figure 2, Supplementary Digital Content 3, http://links.lww.com/AJG/A185). For these 2 cases, the TE showed pressures of 21.2 and 16.0 kPa, whereas PLT counts were 173 × 109 cell/L and 194 × 109 cell/L. Both cases had grade 2 EV and compensated cirrhosis. To the PLT count and serum albumin level validated in our validation cohort, we added a TE cut-off of 25 kPa, as proposed by the Expanded Baveno VI criteria.
The percentage of cases that met the RESIST-HCV criteria plus TE (27.6% and 22.7% among training and validation cohorts, respectively) was lower than the percentage that met the RESIST-HCV criteria. As with the RESIST-HCV criteria without TE, 1 of 90 patients (1.1%) from the training cohort and 5 of 240 patients (2.1%) from the validation cohort had large EV who met the RESIST-HCV criteria with TE (Figure 5).
Analysis of the avoidance of endoscopy based on the Baveno VI, Expanded Baveno VI, and RESIST-HCV criteria in the entire RESIST cohort
Overall, 15.6% of patients met criteria for Baveno VI, 36% for Expanded Baveno VI, and 31.4% for RESIST-HCV. False-negative rates and NPVs for the entire RESIST-HCV cohort, the training cohort, and the validation cohort are shown in Table 3 and Supplementary Table 2 (Supplementary Digital Content 1, http://links.lww.com/AJG/A183).
The rates of endoscopy avoidance were similar using the RESIST-HCV and Expanded Baveno VI criteria; the NPVs were also similar (98.4% and 97.4%). Adding TE to the RESIST-HCV algorithm reduced the number of EGDs that would have been performed without increasing the NPV (98.2%). Moreover, the RESIST-HCV criteria do not include the liver stiffness evaluation, so they may be used in centers that lack this capability. Further diagnostic performance characteristics of noninvasive criteria used to rule out medium/large EV are shown in Supplementary Table 3 (Supplementary Digital Content 1, http://links.lww.com/AJG/A183).
In this study, we have shown that the original Baveno VI criteria (6) and the Expanded Baveno VI criteria (9) to screen for EV requiring prophylactic therapy in patients with cirrhosis can be simplified and safely replaced with the RESIST-HCV algorithm that includes only serum albumin level and PLT count. The use of the RESIST-HCV criteria can reduce the number of endoscopies needed by more than 30% while minimizing the risk of missing varices needing treatment.
We have also validated in our cohort both the Baveno VI and the Expanded Baveno VI recommendations. In contrast to previous studies (7–9), however, our study also aimed to validate the criteria using a large cohort of patients with a single etiology and included patients with Child-Pugh B cirrhosis. This approach was informed by the issues highlighted in the meta-analysis of Marot and colleagues (13), who stated that these validations have relevant bias related to the stage of advanced liver disease, EV prevalence, and cirrhosis etiology. Previously, Maurice et al. (7) analyzed data for 310 patients with cirrhosis (any etiology) who had TE measurements of 10 kPa or greater and had non-decompensated disease. A TE cut-off value of 10 kPa is associated with a high rate of patient misclassification, with a large portion of patients not having a clear diagnosis of cirrhosis and portal hypertension (14). Despite selection rules, the Baveno VI criteria incorrectly classified 2% of patients in an English cohort (7). A study by Ding et al. (15) excluded patients with a history of decompensation and analyzed 2 cohorts of less than 100 patients with any cirrhosis etiology.
The Expanded Baveno VI criteria (9) were met by 36% of patients in our cohort who would have been able to forego EGD. This reduction in invasive testing occurred without an increase in the false-negative results. It should be noted that using these criteria, 13 patients had medium/large EV for whom it would have been unsafe to forego EGD.
The main objectives of our study were to simplify the assessment of portal hypertension in patients with HCV and advanced liver disease and to identify the role of TE in predicting the presence of EV requiring prophylactic therapy to assess the advantage of using a noninvasive algorithm including TE in this setting. Although TE has now been adopted broadly by hepatology centers, it is not immediately available in all liver units. Thus, hepatologists have asked patients to go to other centers if TE is needed to establish the diagnosis of cirrhosis.
Among the 22 centers in the RESIST-HCV network, TE was available in 12. All 12 of these facilities offered TE for patients referred from outside centers that lacked access to that testing. However, the main limitation of TE unavailability is in the need for an annual re-screening, as suggested by the Baveno VI criteria, to determine whether there are TE changes that would prompt an EGD evaluation. An even greater challenge is the use of TE for at-risk persons with poor adherence who have a history of drug use, HIV, or incarceration, who may not have access to TE.
To simplify the criteria for the diagnosis of medium/large EV, we have proposed an alternative, noninvasive algorithm using only routinely performed laboratory tests. Indeed, multivariate logistic analysis showed that PLT count and serum albumin level were independently associated with medium/large EV and that a PLT count cut-off of greater than 120,000 cells/μL and a serum albumin level greater than 3.6 g/dL were able to identify patients without medium/large EV with an NPV slightly higher than that for the Baveno VI and Expanded Baveno VI criteria. Nevertheless, the great advantage of the RESIST-HCV criteria was the ability to identify more than 30% of cases for which EGD could be avoided. Adding TE to the RESIST-HCV algorithm did not further improve the performance of the criteria.
Our study has some limitations that deserve mention. First, the data analysis was retrospective. Second, a minority of patients with decompensated cirrhosis was included; for these patients, the EV size is less applicable as a predictor of bleeding. Third, the evaluation of EV size was performed by several endoscopists, as cohorts composed of patients from more than 20 liver centers throughout Sicily. To investigate the possibility of bias, we selected patients attending a referral liver centre for the training cohort and included all other patients attending other centers in the validation cohort. Our data show the cohorts to have similar prevalences of EV requiring prophylaxis and similar diagnostic results for the Baveno VI, Expanded Baveno VI, and RESIST-HCV criteria.
Another relevant point that needs to be addressed is that the AUROC curves for all noninvasive tools analyzed were low. This phenomenon reflects the fact that for all noninvasive criteria used to diagnose medium/large EV, researchers have given priority to sensitivity over specificity. Therefore, the AUROCs are often suboptimal.
We maintain that at least 3 issues still need to be addressed. First, noninvasive testing does not allow for the detection of small EV, which are used to define the cirrhosis stage and have prognostic relevance after sustained virological response achievement (1–3). Second, for patients with HCV, modifications induced by sustained virological response in liver stiffness and in blood tests may hamper the ability of noninvasive strategies in the setting of long-term follow-up of patients with cirrhosis (16,17). Third, further studies are needed to validate our new criteria in the growing population of patients with cirrhosis caused by alcoholic and nonalcoholic steatohepatitis.
In conclusion, the use of our new noninvasive strategy can avoid more than 30% of EGDs by using only routine blood test results (PLT count and serum albumin level) while maintaining a very high NPV and low false-negative rate. A noninvasive strategy is needed to rule out also the presence of small EV, and the RESIST-HCV criteria require validation in other larger cohorts of persons with cirrhosis of different etiologies.
CONFLICTS OF INTEREST
Guarantor of the article: Vincenza Calvaruso, MD, PhD.
Specific author contributions: V.C.: analysis and interpretation of data, drafting of the manuscript, statistical analysis, critical revision of the manuscript for important intellectual content. I.C., A.L., S.M., R.B.: acquisition of data. S.P.: acquisition of data, critical revision of the manuscript for important intellectual content. F.B., E.C., G.M., G.B., M.D., A.M., A.D., G.A., B.C., A.D., L.G., I.S., P.C., F.C., V.P., T.P., A.A., C.I., L.M., G. Scifo, M.R., G. Squadrito: acquisition of data. G.R.: study concept and design, critical revision of the manuscript for important intellectual content. C.C.: study concept and design, analysis and interpretation of data, statistical analysis, critical revision of the manuscript for important intellectual content. A.C.: study concept and design, analysis and interpretation of data, critical revision of the manuscript for important intellectual content. V.D.M.: study concept and design, analysis and interpretation of data, drafting of the manuscript, study supervision. All authors approved the final version of the manuscript.
Financial support: No financial support was received in relation to this manuscript. The RESIST-HCV is funded by unrestricted grants from Gilead, MSD, Abbvie, and BMS.
Potential competing interests: V.C., S.P., M.D., G. Scifo: participated in the advisory board for Abbvie. G.R.: Participated in the advisory boards for Abbvie, BMS, Gilead, MSD/Merck. C.C.: participated in the advisory board for MSD/Merck. A.C.: research support from Abbvie, BMS, Gilead, Merck/MSD, Intercept; provided consultancy, speakers bureau, and participated in the advisory boards for Abbvie, BMS, Gilead, MSD/Merck. V.D.M.: research support from Abbvie, BMS, Gilead, Merck/MSD; participated in the advisory boards for Abbvie, BMS, MSD/Merck. The other authors have no disclosures to declare.
WHAT IS KNOWN
- ✓ Patients with cirrhosis should undergo EGD to screen for the presence of EV.
- ✓ The Baveno VI consensus and an expanded algorithm use TE and PLT count to identify patients with cirrhosis who can avoid EGD.
WHAT IS NEW HERE
- ✓ RESIST-HCV criteria using only routine blood tests (PLTs, albumin) can avoid more than 30% of EGDs used to screen for EV.
- ✓ The use of RESIST-HCV criteria simplifies the diagnosis of portal hypertension.
1. D'Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: A systematic review of 118 studies. J Hepatol 2006;44(1):217–31.
2. Garcia-Tsao G, Abraldes J, Berzigotti A, et al. Portal hypertensive bleeding in cirrhosis: Risk stratification, diagnosis and management 2016 practice guidance by the American Association for the Study of Liver Diseases. Hepatology 2017;65:310–335.
3. Di Marco V, Calvaruso V, Ferraro D, Bavetta MG, et al. Effects of viral eradication in patients with hepatitis C virus and cirrhosis differ with stage of portal hypertension. Gastroenterology 2016;151:130–139.e2.
4. de Franchis R. Revising consensus in portal hypertension: Report of the Baveno V Consensus Workshop on methodology of diagnosis and therapy in portal hypertension. J Hepatol 2010;53:762–8.
5. Procopet B, Berzigotti A, Abraldes JG, et al. Real-time shear-wave elastography: Applicability, reliability and accuracy for clinically significant portal hypertension. J Hepatol 2015;62(5):1068–75.
6. de Franchis R. Expanding consensus in portal hypertension: Report of the Baveno VI Consensus Workshop: Stratifying risk and individualizing care for portal hypertension. J Hepatol 2015;63:743–52.
7. Maurice JB, Brodkin E, Arnold F, et al. Validation of the Baveno VI criteria to identify low risk cirrhotic patients not requiring endoscopic surveillance for varices. J Hepatol 2016;65(5):899–905.
8. Augustin S, Pons M, Genesca J. Validating the Baveno VI recommendations for screening varices. J Hepatol 2017;66:459–60.
9. Augustin S, Pons M, Maurice JB, et al. Expanding the Baveno VI criteria for the screening of varices in patients with compensated advanced chronic liver disease. Hepatology 2017;66(6):1980–8.
10. Calvaruso V, Cabibbo G, Cacciola I, et al. Incidence of hepatocellular carcinoma in patients with HCV associated cirrhosis treated with direct-acting antiviral agents. Gastroenterology 2018;155(2):411–21.e4.
11. North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. N Engl J Med 1988;319(15):983–9.
12. Schepis F, Cammà C, Niceforo D, et al. Which patients with cirrhosis should undergo endoscopic screening for esophageal varices detection? Hepatology 2001;33:.
13. Marot A, Trépo E, Doerig C, et al. Liver stiffness and platelet count for identifying patients with compensated liver disease at low risk of variceal bleeding. Liver Int 2017;37(5):707–16.
14. Calvaruso V, Bronte F, Magro B, et al. High rate of misclassification of fibrosis stage using transient elastography thresholds to prioritize HCV patients for antiviral treatment. Dig Liver Dis 2017;49(1):e65–6.
15. Ding NS, Nguyen T, Iser DM, et al. Liver stiffness plus platelet count can be used to exclude high-risk oesophageal varices. Liver Int 2016;36(2):240–5.
16. Tada T, Kumada T, Toyoda H, et al. Improvement of liver stiffness in patients with hepatitis C virus infection who received direct-acting antiviral therapy and achieved sustained virological response. J Gastroenterol Hepatol 2017;32(12):1982–8.
17. Deterding K, Höner Zu Siederdissen C, Port K, et al. Improvement of liver function parameters in advanced HCV-associated liver cirrhosis by IFN-free antiviral therapies. Aliment Pharmacol Ther 2015;42(7):889–901.
Supplemental Digital Content
© The American College of Gastroenterology 2019. All Rights Reserved.