Table 2.
Baseline matching characteristics used in the primary MAIC analysis, before and after matchinga
| Baseline characteristics | CELESTIAL 2L population with AFP ≥ 400 ng/mL | REACH-2 population | |
|---|---|---|---|
| Unmatched (N = 202) | Matching-adjusted (N = 105) | Unmatched (N = 292) | |
| Age under 65 years, % | 57.87 | 51.71 | 51.71 |
| Female, % | 21.91 | 20.21 | 20.21 |
| Duration of prior sorafenib treatment < 5 months, %b | 44.94 | 57.19 | 57.19 |
| Extrahepatic disease, % | 76.40 | 72.26 | 72.26 |
| Macrovascular invasion, % | 34.83 | 35.27 | 35.27 |
| Etiology, % | |||
| Hepatitis B | 44.38 | 36.64 | 36.64 |
| Hepatitis C | 22.47 | 26.03 | 26.03 |
| Nonviralc | 30.34 | 31.51 | 31.51 |
| AFP, median log10(AFP)d,e | 3.94 | 3.00 | 3.53 |
| ALBI grade 1, % | 34.83 | 48.97 | 48.97 |
| BCLC stage B, % | 8.99 | 18.49 | 18.49 |
2L second line, AFP alpha-fetoprotein, ALBI, albumin–bilirubin, BCLC Barcelona Clinic Liver Cancer, HCC hepatocellular carcinoma, IQR interquartile range, MAIC matching-adjusted indirect comparison
aMatching variables selected as potential effect modifiers by expert clinical panel
bCategorization of prior sorafenib treatment use in (and published for) the REACH-2 trial
cHCC of nonviral etiology was not recorded directly in the REACH-2 trial. Estimate is derived from the sum of patients with etiology of alcohol use plus nonalcoholic steatohepatitis fatty liver, using the total REACH-2 population size as the denominator. There might be overlap in patients between these two etiology categories
dReported as median of log10(AFP) because of the magnitude of difference in median (IQR) AFP for the pre-matched CELESTIAL and REACH-2 populations: median (IQR) values of 8813 (1648, 30,751) ng/mL for the 2L CELESTIAL population with AFP ≥ 400 ng/mL versus 3394 (1177, 16,812) ng/mL for REACH-2
eMedian baseline serum AFP level was available from the REACH-2 publication but, owing to the way that weights are calculated (using centering moments based on mean rather than median values) and because baseline AFP values for the eligible CELESTIAL subpopulation did not follow a Normal distribution, median log10(AFP) values for the REACH-2 population and for the 2L CELESTIAL subgroup with AFP ≥ 400 ng/mL could not be exactly matched