Cohort | Event | Prol model | Multivariable HR | 95% CI | p-value |
|---|
TCGA | OS | Median | 1.76 | 1.19–2.61 | 4.77 × 10−3 |
TCGA | OS | Median adj. stage | 1.52 | 1.02–2.26 | 4.20 × 10−2 |
TCGA | OS | Quartile | 3.25 | 1.84–5.72 | 4.62 × 10−5 |
TCGA | PFI | Median | 1.89 | 1.37–2.63 | 1.25 × 10−4 |
TCGA | PFI | Median adj. stage | 1.73 | 1.24–2.41 | 1.14 × 10−3 |
TCGA | PFI | Quartile | 2.85 | 1.76–4.63 | 2.14 × 10−5 |
LCI | OS | Median | 1.79 | 1.16–2.76 | 8.79 × 10−3 |
LCI | OS | Median adj. stage | 1.67 | 1.07–2.60 | 2.34 × 10−2 |
- Hazard ratios of overall survival and progression free interval based on the Prol cell-type proportion in the TCGA HCC cases (n = 361) and hazard ratios of overall survival in the LCI HCC cases (n = 221) show that an increased abundance of Prol is associated with decreased survival. Cox proportional hazard regression was performed for the event and model indicated. The Prol model indicates the predictor tested. The median model stratifies the cases into low and high abundance groups based on whether the individual’s estimated Prol proportion was below or above the median, respectively. The median adjusted (adj.) stage results are obtained by including in the median model a covariate for the low and high AJCC tumor stage status, where stage I and II form the low stage and stage III and IV form the high stage. The quartile model tests low and high abundance groups by splitting participants below and above the 25th and 75th percentile of Prol proportion estimates, respectively. All tests in TCGA were adjusted for age, sex, and ethnicity. All tests in LCI were adjusted for age and sex. Unadjusted p-values are shown. HR indicates hazard ratio, CI confidence interval, OS overall survival, PFI progression free interval
