Results and conclusion
| First author [ref.] | ITT | PP | Benefit on another outcome | Conclusion | Conclusion consistent with results? |
| First-line setting | |||||
| Negoro [13] | Hazard ratio 0.83, 95% CI 0.64 to 1.09 | NR | No | E not inferior to C | Yes |
| Rosell [14] | Difference in rates (E−C) 3%, 90% CI NR | Difference in rates (E−C) 3%, 90% CI −10 to 3.4% | No | E not inferior to C for response but C superior to E for OS | Yes |
| Fossella [15] | C versus E1: hazard ratio 1.18; 97.2% CI 0.99 to 1.42 | NR | Yes | Superiority of E1 (abstract) NI of E1 (discussion) | No for E1 (superiority cannot be claimed) |
| C versus E2: hazard ratio 1.05, 97.2% CI 0.88 to 1.25 | NR | Failure of NI for E2 | Yes for E2 | ||
| Ohe [16] | NR | Difference in rates (E−C) −8.2%, 95% CI −19.6 to 3.3% | No | The four regimens have similar efficacy | No |
| NR | Difference in rates (E−C) 0.4%, 95% CI −10.0 to 11.7% | ||||
| NR | Difference in rates (E−C) −10.9%, 95% CI −22.3 to 0.5% | ||||
| Novello [17] | Hazard ratio 1.21, 90% CI 0.97 to 1.51 | NR | Yes | Failure of NI for E | Yes |
| Park [18] | Difference in rates (E−C) 3.4%, 95% CI −8 to 14% | NR | Yes | E not inferior to C but TTP better with C | Yes |
| Scagliotti [19] | Hazard ratio 0.94, 95% CI 0.84 to 1.05 | NR | Yes | E provides similar efficacy with better tolerability | No, only NI can be claimed |
| Heymach [20] | Hazard ratio 1.26, 95% CI 0.83 to 1.91 | NR | NA | Early closure: n=40 (C), n=73 (E); failure to detect NI | Yes |
| Mok [21] | Hazard ratio 0.74, 95% CI 0.65 to 0.85 | NR | Yes | E superior to C | Yes (closed testing procedure) |
| Okamoto [22] | Hazard ratio 0.928, 99.2% CI 0.67 to 1.28 at IA | NR | Yes | E not inferior to C | Yes |
| Lee [23] | Difference in rates (E−C) 1.7%, 95% CI −10 to 13.3% | NR | No | E well tolerated and not inferior | Yes (higher neutropenia with E, despite wrong statement about 95% CI) |
| Second-line setting | |||||
| Schuette [24] | NR | NR | Yes | E provides similar efficacy and better tolerability and is a feasible alternative | No, NR as an equivalence trial |
| Hanna [25] | Hazard ratio 0.99, 95% CI 0.80 to 1.20 | NR | Yes | E equivalent to C, should be a new standard | No, neither equivalence nor NI can be claimed |
| Ramlau [26] | Difference in rates (E−C) −3.6%, 95% CI −9.6 to 2.5% | NR | No | E provides activity and may be an option | No, no reference in the abstract to NI (indeed demonstrated) |
| Krzakowski [27] | Hazard ratio 1.004, 95% CI 0.84 to 1.12 | NR | No | NI demonstrated, E is an option | No, as higher toxicities with E |
| Maruyama [28] | Hazard ratio 1.12, 95% CI 0.89 to 1.40 | NR | Yes | NI not demonstrated; however, no differences | No, trial not designed to show a difference between arms |
| Kim [29] | Hazard ratio 1.02, 96% CI 0.901 to 1.143 | Hazard ratio 1.02, 96% CI 0.905 to 1.150 | Yes | NI demonstrated, E is a valid option | Yes |
| Li [30] | NR | Difference in rates (E−C) 4.5%, 95% CI −2.5 to NR | Yes | Equivalent efficacy outcomes and better safety profile for E | No |
| Natale [31] | Hazard ratio 0.98, 95.22% CI 0.87 to 1.10 | NR | No | No efficacy advantage and higher toxicity | No, as NI is demonstrated by design |
| Shi [32] | Hazard ratio 0.84, 95% CI 0.67 to 1.05 | NR | Yes | Icotinib could be a new treatment option | Yes, although reference to NI missing |
ITT: intention to treat; PP: per protocol; NR: not reported;. E: experimental arm; C: control arm; OS: overall survival; NI: noninferiority; TTP: time to progression; IA: interim analysis.