Background: The J0509 (phase III study for chemotherapy-naive ED-SCLC) demonstrated amrubicin plus cisplatin (AP) was inferior to irinotecan plus cisplatin (IP). However, median overall survival (OS) of both AP and IP (15 and 17 mo) was more favorable than those of previous trials (9-12 mo), probably because switching to different topo-I or topo-II in the second-line therapy, especially the use of topo-II in IP arm, was frequent. This analysis aimed to investigate whether observed survival benefit of IP arm can be explained by the treatment switching, and how post-protocol chemotherapy affected the result of J0509. Methods: Two analysis sets from J0509 were used: all randomized 283 pts and 250 pts who received post-protocol chemotherapy. One pt without initiation date of second-line therapy was excluded. A rank-preserving structural failure time (RPSFT) model was used to estimate “causal survival benefit” that would have been observed if all pts had been followed with the same type of regimen as randomized throughout the follow-up period. Additionally, to assess the survival impact of second-line use of topo-II, OS after initiating second-line therapy (OS2) was analyzed by multivariate Cox models. Results: %treatment switching in IP arm and AP arm was 65.2% (92/141) and 43.7% (62/142). By RPSFT model, estimated OS excluding the effect of the treatment switching was 2.7-fold longer in IP (topo-I) arm than AP (topo-II) arm. This causal survival benefit was stronger than the original report of J0509 (nearly 1.4-fold extension by Cox model), indicating that re-challenging topo-I in IP arm appeared beneficial. The multivariate Cox analysis for OS2 (n = 250) revealed second-line use of topo-II was detrimental (hazard ratio, 1.5; 95%CI, 1.1-2.1). Among sensitive relapsed pts in IP arm, OS2 was favorable in the following order: irinotecan-based regimen > the other topo-I > topo-II. Conclusions: IP remains the standard therapy. Re-challenging topo-I, especially irinotecan-based topo-I, seemed beneficial for IP-sensitive pts. This result should be confirmed in further investigations with large sample size. Clinical trial information: 000000720.