Patients With Chronic HCV Who Need Early Treatment
Patients With Chronic HCV Who Need Early Treatment
After removal of duplicate entries, 2257 unique articles were identified by our systematic literature search (Figure 1). On the basis of abstract review, 69 were selected for full-text review. Two study authors classified 29 articles as meeting the predefined criteria for analysis. In total, these 29 studies included 5817 unique patients from 20 separate patient cohorts. Sixteen of these studies investigated predictors of histological progression, eight studies evaluated predictors of clinical outcomes, and the remaining five studies investigated both histological and clinical outcomes. Fourteen studies included treatment-naïve patients only, five included both treatment-naïve and treatment-experienced patients, eight included treatment-experienced patients only, and two studies did not describe the treatment status of the patients. We contacted four authors to obtain additional unpublished data.
(Enlarge Image)
Figure 1.
Flow diagram of studies included in the systematic review. Many studies met multiple exclusion criteria. Each study was coded under a single criterion only. Includes animal models, paediatric populations, patients who had previously undergone liver transplant, patients with chronic liver disease other than HCV monoinfection, evaluation of only specific subsets of populations with CHC. Includes studies that were descriptive papers only, studies that did not specifically evaluate for predictors of histological or clinical progression, and studies that evaluated predictors that are not readily clinically available. Includes studies that focused on risk factors for the development of HCC only, and studies where some patients achieved SVR and the results were not stratified based on response to treatment.
A total of 21 studies evaluated predictors of histological progression. The studies included populations from Europe (n = 10), Asia (n = 2), and North (n = 8) and South America (n = 1). Only one study was prospective with the remaining 20 being retrospective analyses of previously collected data. The sample size for included studies varied (range 36–622 patients) with the majority having <200 patients (n = 14). A number of studies had overlapping cohorts. Four studies were derived from the Hepatitis C Anti-viral Long-term Treatment Against Cirrhosis (HALT-C) cohort, a US multi-centre randomised controlled trial to evaluate the safety and efficacy of low dose pegylated IFN in CHC patients with advanced fibrosis who failed to respond to prior IFN therapy. Four other pairs of studies drew from the same cohort of patients. These studies were included in the review despite overlapping cohorts given differences in predictors examined, outcomes evaluated and criteria for selection of subsets of patients analysed within the overall larger cohort. The average duration of follow-up ranged from a median of 2.5–10 years.
The studies had varied inclusion and exclusion criteria as detailed in Table 1. Among the non-HALT-C studies, 11 studies had explicit requirements for baseline Ishak/METAVIR fibrosis stage. Five studies required minimal or no fibrosis at baseline and the remaining six studies required lack of cirrhosis on initial biopsy. Only 14 studies described criteria used to determine adequacy of biopsy specimens. The majority of the studies had a single pathologist blinded to clinical data score the biopsies while the HALT-C study had a panel of pathologists review the biopsies and consensus staging was recorded. Exclusionary alcohol intake was described in nine studies though the cut-off amounts and methods for ascertaining alcohol intake varied across the studies. The studies were predominately comprised of male patients in their late 30s to early 50s.
A total of 13 studies evaluated predictors of clinical outcomes. Six studies were conducted in the US (including 5 HALT-C studies), five in Europe and two in Asia. Only two studies were prospective with the remaining 11 being retrospective analyses. Sample size in each study varied from 52 to 1457 patients. Apart from the HALT-C studies, there was only one additional overlapping cohort. The average duration of follow-up ranged from a median of 2.3 to a maximum of 14.4 years. Compared to studies on histological progression, the studies on clinical outcomes consisted of patients who were older, had more advanced fibrosis at baseline, and were more likely to be treatment experienced.
A summary of the specific outcomes evaluated and incidence of these outcomes in each study is displayed in Table 2, Table 3 and Table 4. For studies where the outcome was defined as ≥1 fibrosis stage increase on follow-up biopsy (n = 13), the incidence of that outcome ranged from 21–61% over a range of follow-up of 2.5–10 years. Studies applying a stricter definition of fibrosis progression (≥2 stage increase on follow-up biopsy, n = 3) had less variability in range of incidence of outcome, reporting 22–34% over a range of follow-up of 3.5–5.8 years. Studies with higher rates of fibrosis progression tended to have longer follow-up durations (>6 years), though there were several studies with follow-up of ≥6 years that had low rates of fibrosis progression. No identifiable differences in patient characteristics between studies with high vs. low incidence of fibrosis progression were noted.
Studies assessing risk factors for clinical progression (n = 13) included several distinct outcomes. Four studies evaluating progression from compensated to decompensated cirrhosis reported an incidence between 13% and 40% over a range of follow-up of 2.3–14.4 years. No clear pattern was identified between length of follow-up or patient characteristics and rate of outcomes. Notably, the definition of decompensation varied across studies. Four studies evaluating the incidence of overall mortality reported incidences between 8% and 47%. The range of follow-up for these studies was 3.9–14.4 years, with a higher rate of outcomes reported in studies with longer duration of follow-up. The remaining studies used an aggregate outcome encompassing a broad range of clinical end points including decompensation, increase in Child–Turcotte–Pugh score, development of HCC, liver transplant and liver related as well as overall mortality. The reported incidence of this aggregate outcome was 13–31% over a range of follow-up of 3.5–6.3 years.
A detailed list of the predictors evaluated and the results of univariate analysis is provided in Tables S3–S5 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f. For each study, the predictor variables were categorised as follows: (i) baseline clinical characteristics including demographics and relevant co-morbidities; (ii) baseline laboratory results; (iii) baseline histological features or (iv) longitudinal laboratory and histology results.
All studies investigating predictors of histological progression evaluated baseline clinical characteristics, baseline laboratory results and baseline histology results except for Tamaki et al. who did not evaluate baseline histological features. Only half of the studies evaluated longitudinal variables which were predominantly serial aminotransferase levels. Longitudinal biopsy results such as changes in steatosis score or histological activity index (HAI) were assessed in only five studies. The predictors that were most consistently evaluated are listed in Figure 2a. The most common clinical characteristics assessed were age, gender, HCV genotype, alcohol intake, body mass index (BMI) and biopsy interval, and the most common laboratory values evaluated were platelet count and ALT levels. Baseline histological features were also frequently investigated predictors and were included in >70% of studies.
(Enlarge Image)
Figure 2.
List of variables identified to have significant predictive value for (a) histological and (b) clinical progression.
Multivariable analysis was performed in all but two studies. Variables found to be independently predictive of histological progression are listed in Table 2 and Table 4. Among all the variables assessed, baseline steatosis was most consistently reported as independently predictive of subsequent fibrosis progression (significant on multivariate analysis in 6 of 21 studies) with an odds ratio (OR) [(95% confidence interval (CI)] of 4.8 (1.3–18.3) to 14.3 (2.1–111.1). Notably, one study found that effect of baseline steatosis on fibrosis progression was dependent on baseline fibrosis stage. Baseline Ishak/METAVIR fibrosis stage was the next most consistently identified independent predictor of histological progression (significant on multivariable analyses in five of 21 studies). Only one of these studies reported the effect size, with adjusted relative risk of 1.93 (95% CI 1.3–9.0). Figure 2a depicts the number of studies in which individual variables were significantly or not significantly predictive of histological progression on multivariate analyses.
All 13 studies examining predictors of clinical outcomes included baseline clinical characteristics and laboratory results (Tables S4 and S5 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f). Baseline histology was assessed in only eight studies though biopsies were performed in every study. Only three studies incorporated longitudinal data which consisted of serial laboratory values only. The predictors that were most consistently evaluated are listed in Figure 2B. The most common clinical characteristics assessed were age, gender and BMI; the most common laboratory values evaluated were platelet count and ALT level.
Multivariable analysis was performed in all but two studies. The variables found to be independently predictive of clinical progression are listed in Table 3 and Table 4. Among the variables assessed, baseline platelet count was the most consistent independent predictor of clinical outcomes (significant on multivariate analysis in six of 13 studies) followed by age, baseline AST/ALT ratio, albumin and bilirubin (each significant in four studies). Figure 2B depicts the number of studies in which individual variables were significantly or not significantly predictive of clinical outcomes in multivariate analyses.
Five studies provided prediction models, three for fibrosis progression and four for clinical outcomes (Table S6 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f). Four of the models were derived from the HALT-C study. All the prediction models are primarily comprised of baseline laboratory results. Only one of the models incorporated longitudinal data. None of the models had been validated in external CHC cohorts and only two models reported the associated area under the receiver operating characteristic curve.
Studies evaluating predictors of histological progression were of varying quality, whereas studies investigating predictors of clinical outcomes or studies investigating combined outcomes were all of high quality except for one study. Six studies on histological progression included a small number of patients with advanced fibrosis or cirrhosis on initial biopsy who were not able to progress according to the author's definition.[17][18, 25, 28, 33, 38] Two studies evaluated select cohorts (Levine et al. evaluated untreated Irish women who acquired HCV infection during pregnancy only, and Livingston et al. evaluated only treatment naïve Alaska Native and American Indian persons) and were scored as having limited representativeness. The remaining studies were scored as being at least somewhat representative of the average patient with CHC in the community (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f).
Results
Studies Included in the Systematic Review
After removal of duplicate entries, 2257 unique articles were identified by our systematic literature search (Figure 1). On the basis of abstract review, 69 were selected for full-text review. Two study authors classified 29 articles as meeting the predefined criteria for analysis. In total, these 29 studies included 5817 unique patients from 20 separate patient cohorts. Sixteen of these studies investigated predictors of histological progression, eight studies evaluated predictors of clinical outcomes, and the remaining five studies investigated both histological and clinical outcomes. Fourteen studies included treatment-naïve patients only, five included both treatment-naïve and treatment-experienced patients, eight included treatment-experienced patients only, and two studies did not describe the treatment status of the patients. We contacted four authors to obtain additional unpublished data.
(Enlarge Image)
Figure 1.
Flow diagram of studies included in the systematic review. Many studies met multiple exclusion criteria. Each study was coded under a single criterion only. Includes animal models, paediatric populations, patients who had previously undergone liver transplant, patients with chronic liver disease other than HCV monoinfection, evaluation of only specific subsets of populations with CHC. Includes studies that were descriptive papers only, studies that did not specifically evaluate for predictors of histological or clinical progression, and studies that evaluated predictors that are not readily clinically available. Includes studies that focused on risk factors for the development of HCC only, and studies where some patients achieved SVR and the results were not stratified based on response to treatment.
Characteristics of Studies on Histological Progression
A total of 21 studies evaluated predictors of histological progression. The studies included populations from Europe (n = 10), Asia (n = 2), and North (n = 8) and South America (n = 1). Only one study was prospective with the remaining 20 being retrospective analyses of previously collected data. The sample size for included studies varied (range 36–622 patients) with the majority having <200 patients (n = 14). A number of studies had overlapping cohorts. Four studies were derived from the Hepatitis C Anti-viral Long-term Treatment Against Cirrhosis (HALT-C) cohort, a US multi-centre randomised controlled trial to evaluate the safety and efficacy of low dose pegylated IFN in CHC patients with advanced fibrosis who failed to respond to prior IFN therapy. Four other pairs of studies drew from the same cohort of patients. These studies were included in the review despite overlapping cohorts given differences in predictors examined, outcomes evaluated and criteria for selection of subsets of patients analysed within the overall larger cohort. The average duration of follow-up ranged from a median of 2.5–10 years.
The studies had varied inclusion and exclusion criteria as detailed in Table 1. Among the non-HALT-C studies, 11 studies had explicit requirements for baseline Ishak/METAVIR fibrosis stage. Five studies required minimal or no fibrosis at baseline and the remaining six studies required lack of cirrhosis on initial biopsy. Only 14 studies described criteria used to determine adequacy of biopsy specimens. The majority of the studies had a single pathologist blinded to clinical data score the biopsies while the HALT-C study had a panel of pathologists review the biopsies and consensus staging was recorded. Exclusionary alcohol intake was described in nine studies though the cut-off amounts and methods for ascertaining alcohol intake varied across the studies. The studies were predominately comprised of male patients in their late 30s to early 50s.
Characteristics of Studies of Clinical Outcomes
A total of 13 studies evaluated predictors of clinical outcomes. Six studies were conducted in the US (including 5 HALT-C studies), five in Europe and two in Asia. Only two studies were prospective with the remaining 11 being retrospective analyses. Sample size in each study varied from 52 to 1457 patients. Apart from the HALT-C studies, there was only one additional overlapping cohort. The average duration of follow-up ranged from a median of 2.3 to a maximum of 14.4 years. Compared to studies on histological progression, the studies on clinical outcomes consisted of patients who were older, had more advanced fibrosis at baseline, and were more likely to be treatment experienced.
Incidence of Histological Progression
A summary of the specific outcomes evaluated and incidence of these outcomes in each study is displayed in Table 2, Table 3 and Table 4. For studies where the outcome was defined as ≥1 fibrosis stage increase on follow-up biopsy (n = 13), the incidence of that outcome ranged from 21–61% over a range of follow-up of 2.5–10 years. Studies applying a stricter definition of fibrosis progression (≥2 stage increase on follow-up biopsy, n = 3) had less variability in range of incidence of outcome, reporting 22–34% over a range of follow-up of 3.5–5.8 years. Studies with higher rates of fibrosis progression tended to have longer follow-up durations (>6 years), though there were several studies with follow-up of ≥6 years that had low rates of fibrosis progression. No identifiable differences in patient characteristics between studies with high vs. low incidence of fibrosis progression were noted.
Incidence of Clinical Progression
Studies assessing risk factors for clinical progression (n = 13) included several distinct outcomes. Four studies evaluating progression from compensated to decompensated cirrhosis reported an incidence between 13% and 40% over a range of follow-up of 2.3–14.4 years. No clear pattern was identified between length of follow-up or patient characteristics and rate of outcomes. Notably, the definition of decompensation varied across studies. Four studies evaluating the incidence of overall mortality reported incidences between 8% and 47%. The range of follow-up for these studies was 3.9–14.4 years, with a higher rate of outcomes reported in studies with longer duration of follow-up. The remaining studies used an aggregate outcome encompassing a broad range of clinical end points including decompensation, increase in Child–Turcotte–Pugh score, development of HCC, liver transplant and liver related as well as overall mortality. The reported incidence of this aggregate outcome was 13–31% over a range of follow-up of 3.5–6.3 years.
Predictors of Histological Progression
A detailed list of the predictors evaluated and the results of univariate analysis is provided in Tables S3–S5 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f. For each study, the predictor variables were categorised as follows: (i) baseline clinical characteristics including demographics and relevant co-morbidities; (ii) baseline laboratory results; (iii) baseline histological features or (iv) longitudinal laboratory and histology results.
All studies investigating predictors of histological progression evaluated baseline clinical characteristics, baseline laboratory results and baseline histology results except for Tamaki et al. who did not evaluate baseline histological features. Only half of the studies evaluated longitudinal variables which were predominantly serial aminotransferase levels. Longitudinal biopsy results such as changes in steatosis score or histological activity index (HAI) were assessed in only five studies. The predictors that were most consistently evaluated are listed in Figure 2a. The most common clinical characteristics assessed were age, gender, HCV genotype, alcohol intake, body mass index (BMI) and biopsy interval, and the most common laboratory values evaluated were platelet count and ALT levels. Baseline histological features were also frequently investigated predictors and were included in >70% of studies.
(Enlarge Image)
Figure 2.
List of variables identified to have significant predictive value for (a) histological and (b) clinical progression.
Multivariable analysis was performed in all but two studies. Variables found to be independently predictive of histological progression are listed in Table 2 and Table 4. Among all the variables assessed, baseline steatosis was most consistently reported as independently predictive of subsequent fibrosis progression (significant on multivariate analysis in 6 of 21 studies) with an odds ratio (OR) [(95% confidence interval (CI)] of 4.8 (1.3–18.3) to 14.3 (2.1–111.1). Notably, one study found that effect of baseline steatosis on fibrosis progression was dependent on baseline fibrosis stage. Baseline Ishak/METAVIR fibrosis stage was the next most consistently identified independent predictor of histological progression (significant on multivariable analyses in five of 21 studies). Only one of these studies reported the effect size, with adjusted relative risk of 1.93 (95% CI 1.3–9.0). Figure 2a depicts the number of studies in which individual variables were significantly or not significantly predictive of histological progression on multivariate analyses.
Predictors of Clinical Outcomes
All 13 studies examining predictors of clinical outcomes included baseline clinical characteristics and laboratory results (Tables S4 and S5 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f). Baseline histology was assessed in only eight studies though biopsies were performed in every study. Only three studies incorporated longitudinal data which consisted of serial laboratory values only. The predictors that were most consistently evaluated are listed in Figure 2B. The most common clinical characteristics assessed were age, gender and BMI; the most common laboratory values evaluated were platelet count and ALT level.
Multivariable analysis was performed in all but two studies. The variables found to be independently predictive of clinical progression are listed in Table 3 and Table 4. Among the variables assessed, baseline platelet count was the most consistent independent predictor of clinical outcomes (significant on multivariate analysis in six of 13 studies) followed by age, baseline AST/ALT ratio, albumin and bilirubin (each significant in four studies). Figure 2B depicts the number of studies in which individual variables were significantly or not significantly predictive of clinical outcomes in multivariate analyses.
Mathematical Prediction Models
Five studies provided prediction models, three for fibrosis progression and four for clinical outcomes (Table S6 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f). Four of the models were derived from the HALT-C study. All the prediction models are primarily comprised of baseline laboratory results. Only one of the models incorporated longitudinal data. None of the models had been validated in external CHC cohorts and only two models reported the associated area under the receiver operating characteristic curve.
Quality Assessment and Risk of Bias
Studies evaluating predictors of histological progression were of varying quality, whereas studies investigating predictors of clinical outcomes or studies investigating combined outcomes were all of high quality except for one study. Six studies on histological progression included a small number of patients with advanced fibrosis or cirrhosis on initial biopsy who were not able to progress according to the author's definition.[17][18, 25, 28, 33, 38] Two studies evaluated select cohorts (Levine et al. evaluated untreated Irish women who acquired HCV infection during pregnancy only, and Livingston et al. evaluated only treatment naïve Alaska Native and American Indian persons) and were scored as having limited representativeness. The remaining studies were scored as being at least somewhat representative of the average patient with CHC in the community (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.12921/asset/supinfo/apt12921-sup-0001-TableS1-S6.docx?v=1&s=130e5a4e31352228b5691bb0dcf446af48e7576f).