Figure 2 shows a flowchart showing the patient selection process. Of the 107 patients treated with ICRT, surgery was not performed in 10 patients. Two patients were excluded for having tumor types other than AD or SQ. Finally, 95 patients were enrolled in the study. Mean number of chemotherapy cycles for the ICRT was 2.1 ± 0.7. Chemotherapy regimens were carboplatin and paclitaxel in 46 patients, cisplatin and docetaxel in 46, and cisplatin and pemetrexed in 3. The radiation dose was 40 Gy for 49 patients, 46 Gy for 42, 50 Gy for 2, and 60 Gy for 2.
Table 1 shows the characteristics of the patients with AD (n = 55) versus those with SQ (n = 40) before ICRT. SQ showed a larger tumor size (5.8 ± 1.9 vs. 4.6 ± 1.9 cm) and higher SUV (14.6 ± 6.1 vs. 10.7 ± 6.0) than AD (p = 0.002 and 0.006, respectively). There were no significant differences in sex, age, comorbidity index, tumor location, clinical stage, number of chemotherapy cycles, radiation dose (40 Gy vs. ≥46 Gy), and surgical procedures between AD and SQ (p = 0.86, 0.62, 0.10, 0.07, 0.46, 0.32, 0.57, and 0.12, respectively).
Table 2 shows the clinical TNM stages before ICRT, which did not show a significant difference of N0 stage between the groups, i.e., 11 of 55 AD patients (20%) and 3 of 40 SQ patients (8%) (p = 0.14). There was also no significant difference of T3/T4 stages between the groups, i.e., 32 of 55 patients (58%) in AD and 30 of 40 (75%) in SQ (p = 0.12).
The tumor sizes of AD and SQ groups were significantly reduced after ICRT (p < 0.001), i.e., from 4.6 ± 1.9 cm before ICRT to 3.2 ± 1.6 cm after ICRT for AD, and from 5.8 ± 1.9 cm before ICRT to 3.4 ± 1.5 cm after ICRT for SQ. Figure 3 shows the change ratio in tumor size after ICRT, which was 0.74 ± 0.19 for AD and 0.61 ± 0.18 for SQ; SQ showed a significant decrease in tumor size compared to AD (p < 0.001). AD showed a partial response less frequently than SQ with significance (p = 0.016), i.e., 25/55 (46%) and 29/40 (73%), respectively.
All patients underwent PET before and after ICRT. The SUV of both AD and SQ significantly reduced after ICRT (p < 0.001), i.e., from 10.7 ± 6.0 before ICRT to 5.2 ± 4.0 after ICRT for AD, and from 14.6 ± 6.1 before ICRT to 4.2 ± 3.5 after ICRT for SQ. The change ratio of SUV after ICRT was significantly lower in the SQ group (0.32 ± 0.28) than in the AD group (0.53 ± 0.27) (p < 0.001) (Figure S1).
Lobectomy was performed in 88 patients, segmentectomy in 5, and pneumonectomy in 2. For all 4 cases with N3 stage before ICRT, the N3 stations were dissected; no metastases were revealed. Of the 88 patients who underwent lobectomy, 18 (20%) required bronchial reconstruction. Fifty-one of the 95 patients (54%) received the combined resection for T3/T4 disease. While 91 patients (96%) underwent complete resection (R0), the remaining 4 could not (R1 in 2 and R2 in 2) due to tumor remnants in the aorta, right main bronchus, vertebra, and esophagus, respectively. While 93 patients were discharged without major complications, the other 2 patients died of surgery-related complications.
Pathological CR was seen in 8 of the 55 AD cases (14%), which was less frequent than 22 of the 40 SQ cases (55%) (p < 0.001). However, the pathological N0 stage was not significantly different between AD and SQ, i.e., 33 of 55 (60%) and 32/40 (90%), respectively (p = 0.07) (Table S1).
External margins of residual tumors in AD were frequently seen at the periphery of tumor than those in SQ (Figs. 4 and 5). Figure 6 shows waterfall plots of the locations of the external margins of residual tumors (b/a). Of the 55 AD cases, 39 (71%) showed residual tumors at the periphery of the primary tumor (b/a = 1.0) in contrast to only 6 of 40 SQ cases (15%); the difference was significant (p < 0.001). The mean b/a value was 0.80 ± 0.37 in AD (median value, 1.0; interquartile range [IQR], 0.18), which was significantly higher than 0.33 ± 0.41 in SQ (median, 0; IQR, 0.71) (p < 0.001). Even in the 65 cases other than the pathological CR, AD still showed residual tumors at the periphery of the primary tumor (b/a = 1.0) more frequently than SQ (p < 0.001), i.e., 39 of the 47 (83%) and 6 of 18 (33%), respectively. The mean b/a value in the 65 patients other than pathological CR was 0.97 ± 0.17 (median value, 1.0; IQR, 0) in AD, which was significantly higher than 0.70 ± 0.29 (median value, 0.75; IQR, 0.48) in SQ (p < 0.001).
None of the patients were lost to follow-up. The median follow-up period was 35 months (range: 4–76 months). Forty-two patients (44%) received adjuvant postoperative chemotherapy: 26 of 55 AD patients (47%) and 16 of 40 SQ patients (40%), of which difference was not significant (p = 0.48). During the study period, 40 patients experienced recurrence (31 with AD, 9 with SQ) and 25 patients died (13 with AD, 12 with SQ). Among the 25 patients who died, 18 died of lung cancer and the other 7 died of other diseases, including surgery-related death in 2 patients.
Figure 7 shows the RFS, which was significantly worse in patients with AD than in those with SQ (p = 0.023, log-rank test), with 3-years RFS rates of 45% and 72%, respectively. For the 40 patients with recurrence, additional chemotherapy was administered to 26 of 31 patients with AD (84%) and 7 of 9 patients with SQ (78%), of which difference was not significant (p = 0.62).
Figure 8 shows the OS, which was not significantly different between AD and SQ (p = 0.45, log-rank test), with 3-years survival rates of 76% and 75%, respectively.