The initiative led to the cooperation of investigators from four countries and included data previously published from five patient cohorts. These are the CLIP1 and CLIP 3 cohorts from Italy, a cohort from Spain, a cohort from Hong-Kong, China and an Israeli cohort. These five original data sets were assembled into one metadatabase including 548 patients: CLIP1 with 80 patients (14.6%), CLIP3 with 75 patients (13.7%), the Spanish cohort with 94 patients (17.2%), a Chinese cohort with 197 patients (36%) and an Israeli cohort of 102 patients (18.6%). There were 159 patients below the age of 65 (29%) at diagnosis, 269 (49%) patients diagnosed between the ages of 65 and 75, and 120 (22%) patients diagnosed at age 75 and older. There were 47 (8.6% of the entire cohort) patients who were diagnosed at age 80 or more. The studies were conducted in different eras with the CLIP studies occurring in the 1990's and the Spanish and Israeli studies conducted later. The Chinese study was the only one that spanned more than 20 years, although most of the recruitment was done during the same period as the Israeli and Spanish studies.
Patient characteristics are presented in Table 1. In the combined data set, females account for 24% of the patients, and 325 subjects (93%) were cirrhotic at diagnosis. The Chinese cohort included older patients, none under the age of 65. Among the non-Chinese countries, HBV was found in 21%% of the patients (range 15–27%), in contrast, in the Chinese cohort, 63% of the patients were HBV carriers. Tables S1–S3 https://onlinelibrary.wiley.com/store/10.1111/liv.12486/asset/supinfo/liv12486-sup-0001_TableS1-S5_FigS1-13.doc?v=1&s=21d001f249a038a798ae831e3f1eede420bed412 present the distributions of HCC score/stages in the three age groups.
Age at diagnosis was positively associated with the era when this study was conducted (r = 0.15, P = 0.001), patients diagnosed at late dates were on average older. There was a negative association between date of diagnosis and disease stage: patients with Okuda I comprised 17%, 23%, 46% and 62% of the included patient cohorts during 1988–1992, 1993–1997, 1998–2002, 2003–2008 respectively. Okuda II patients comprised 79%, 70%, 51% and 36% of the included respectively (P < 0.001). Figure 1 depicts dates of diagnosis, age at diagnosis and overall survival histograms, showing the individual contribution of each data set. Survival probabilities are presented in Table S4 https://onlinelibrary.wiley.com/store/10.1111/liv.12486/asset/supinfo/liv12486-sup-0001_TableS1-S5_FigS1-13.doc?v=1&s=21d001f249a038a798ae831e3f1eede420bed412.
Time of diagnosis, extent of survival and age at diagnosis of included patients. Composite histograms representing distribution of diagnosis date (top panel), survival time (middle panel) and age at diagnosis (lower panel). In all panels, the contribution of each data set is presented in a unique colour.
Crude survival rates were similar between patients aged younger than 65, 65–75 and 75 and older (Fig. 2A). Median (95% CI) survival estimates were 23 (17–28), 21 (17–26) and 19 (15–23) months (P = 0.14) in the different age groups respectively. Stratified survival analysis per study group did not demonstrate significant survival difference between the age groups (all P-values between 0.1 and 0.9). Limiting the analysis to patients diagnosed since 2000 had no effect on the results of the latter years. Because our Chinese cohort included only patients older than 65, we stratified this age group to three strata 65–70, 70–75 and >75. Using univariate analysis, we saw that younger age was associated with better survival (Fig. S1 https://onlinelibrary.wiley.com/store/10.1111/liv.12486/asset/supinfo/liv12486-sup-0001_TableS1-S5_FigS1-13.doc?v=1&s=21d001f249a038a798ae831e3f1eede420bed412). However, age was no longer a significant factor in the multivariate analysis, and disease stage remained the only independent prognostic predictor. Figure 2B depicts study-specific survival curves per each study group. Multivariate analysis models generated non-significant hazard ratios associating age and mortality. Disease stage was the dominant predictor of mortality (Table 2). These results were consistent when analyses were not stratified for study and when the CLIP staging system or the Child score was used instead of the Okuda score. Figures S2–S11 https://onlinelibrary.wiley.com/store/10.1111/liv.12486/asset/supinfo/liv12486-sup-0001_TableS1-S5_FigS1-13.doc?v=1&s=21d001f249a038a798ae831e3f1eede420bed412 depict crude survival curves in each separate HCC score/stage. Survival periods prior to and after 2000, as well as separate crude survival curves, comparing the three age groups, prior to and after 2000 are also presented in Table S5 and Figures S12 and S13 https://onlinelibrary.wiley.com/store/10.1111/liv.12486/asset/supinfo/liv12486-sup-0001_TableS1-S5_FigS1-13.doc?v=1&s=21d001f249a038a798ae831e3f1eede420bed412.
Survival curves per age groups. (A) cumulative survival per age group; (B) Study-specific survival curves per age group.
Two data sets had additional data regarding adverse events (the Chinese and Israeli cohorts). A total of 43/299 (14.4%) patients had documented acute renal failure and 122/299 (41%) patients had evidence of deterioration in liver function (DLF). It should be noted that patients younger than 65 were all from the Israeli cohort and all had evidence of DLF following TACE (>50% INR and hepatocellular enzyme increases above the upper limit of normal). Age group rates of these complications are presented in Table 3. Rare complications included vascular complications 4/299(1.3%) and significant haemorrhage (blood loss, either into the gastrointentinal tract or through vessel puncture, requiring transfusion of at least two blood units) 15/299 (5%).
Liver International. 2014;34(7):1109-1117. © 2014 Blackwell Publishing