Table 1 Prognostic and predictive value of baseline AFP by Cox regression model.

From: Serum alpha-fetoprotein and clinical outcomes in patients with advanced hepatocellular carcinoma treated with ramucirumab

Prognostic scenario

Hazard ratio (95% CI)a

 

REACH AFP dichotomous

REACH AFP continuous

REACH-2

Pooled

Adjusting for baseline AFP and treatment onlyb

  AFP (≥400 vs. <400 ng/ml)

1.93 (1.58–2.35), p < 0.0001

NA

NA

NA

 AFP (ng/ml) log-transformed

NA

1.37 (1.27–1.47), p < 0.0001

1.59 (1.32–1.91), p < 0.0001

1.56 (1.37–1.76), p < 0.0001

Multivariate analysisc

  AFP (≥400 vs. <400 ng/ml)

1.82 (1.48–2.22), p < 0.0001

NA

NA

NA

  AFP (ng/ml) log-transformed

NA

1.34 (1.25–1.44), p < 0.0001

1.58 (1.31–1.91), p < 0.0001

1.53 (1.35–1.74), p < 0.0001

  ECOG PS (0 vs. 1)

0.77 (0.63–0.93), p = 0.0080

0.75 (0.62–0.91), p = 0.0040

0.68 (0.52–0.89), p = 0.0056

0.74 (0.61–0.90), p = 0.0020

  Macrovascular invasion (yes vs. no)

1.49 (1.20–1.85), p = 0.0003

1.47 (1.18–1.82), p = 0.0005

1.44 (1.09–1.89), p = 0.0100

1.42 (1.17–1.73), p = 0.0004

Predictive scenario

Hazard ratio (95% CI)

REACH AFP dichotomousb

    

  <400 ng/ml

1.06 (0.82–1.38), p = 0.664

   

 ≥400 ng/ml

0.65 (0.50–0.85), p = 0.002

   

AFP treatment interaction p value

0.008

   

REACH AFP continuousb

    

 AFP treatment interaction (SE), p value

−0.195 (0.068), 0.0042

   
  1. AFP alpha-fetoprotein, ECOG PS Eastern Cooperative Oncology Group performance status, NA not available, SE standard error.
  2. aAll p values in prognostic scenario are Wald’s p values.
  3. bUnstratified analyses.
  4. cAdjusting for baseline AFP, treatment, macrovascular invasion and ECOG PS.