“…After being excluded by titles and abstracts, 29 full-text articles were assessed for eligibility. 18 records didn't meet the inclusion criteria and were discarded: 2 involving other kinase inhibitors (Nault et al, 2013[ 28 ], 2015[ 29 ]), 9 with overlapped patient data (Antonelli et al, 2018[ 2 ]; Gigante et al, 2015[ 13 ]; Hoshino et al, 2015[ 16 ]; Imai et al, 2015[ 17 ], 2019[ 19 ]; Labeur et al, 2018[ 23 ][ 24 ]; Okada et al, 2019[ 33 ]; Saeki et al, 2019[ 34 ]), 5 lacking HR and 95 % CI (Mir et al, 2012[ 26 ]; Okada et al, 2020[ 32 ]; Saeki et al, 2018[ 35 ]; Uchikawa et al, 2020[ 39 ]; Ueki et al, 2016[ 40 ]), 1 without any interesting outcome (Cheng et al, 2019[ 7 ]) and 1 with low quality and faulty data (Nishikawa et al, 2017[ 30 ]). Thus, 11 retrospective studies (Antonelli et al, 2018[ 3 ]; Endo et al, 2020[ 10 ]; Hiraoka et al, 2017[ 15 ]; Imai et al, 2020[ 18 ]; Labeur et al, 2019[ 25 ]; Naganuma et al, 2017[ 27 ]; Sawada et al, 2019[ 36 ]; Takada et al, 2018[ 38 ]; Uojima et al, 2020[ 41 ]; Wu et al, 2021[ 43 ]; Yamashima et al, 2017[ 45 ]) were included in this meta-analysis, comprising 1148 patients.…”