We included a total of 38,359 patients from the SEER database, including 22,132(56.7%) in the surgical group and 16,227(43.3%) in the non-surgical group. We matched 7 potential confounders at a 1:1 ratio by using PSM, and then 18,839 patients were included in our final study. Some factors (Age, Sex, Racehistological type, stage, radiotherapy, chemotherapy, year of diagnosis) differed significantly before matching, There was no difference in baseline characteristics after matching except for stage, chemotherapy, and year of diagnosi. Baseline levels before and after matching are shown in Table 1.
Table 1
Baseline characteristics of all patients before and after matching
Variables | Before matching | p-value | After matching | p-value |
Surgery (n = 22132) | No surgery (n = 16227) | Surgery (n = 9439) | No surgery (n = 9439) |
Age, No.(%) | | | ༜0.001 | | | 0.899 |
| 70–74 years | 7766 (35.09%) | 7385(45.51%) | | 3861 (40.9%) | 3907 (41.39%) | |
| 75–79 years | 6728 (30.4%) | 5509 (33.95%) | | 3155 (33.43%) | 3136 (33.22%) | |
| 80–84 years | 4915 (22.21%) | 2667 (16.44%) | | 1819 (19.27%) | 1789 (18.95%) | |
| 85 + years | 2723(12.3%) | 666 (4.1%) | | 604 (6.4%) | 607 (6.43%) | |
Sex, No.(%) | | | ༜0.001 | | | 0.635 |
| Male | 14857(67.13%) | 9696 (59.75%) | | 6219 (65.89%) | 6188(65.56%) | |
| Female | 7275 (32.87%) | 6531 (40.25%) | | 3220 (34.11%) | 3251 (34.44%) | |
Year of diagnosis, No.(%) | | | 0.464 | | | ༜0.001 |
| 2010–2012 | 10983(49.62%) | 8114 (50%) | | 4849 (51.37%) | 5800 (61.45%) | |
| 2013–2015 | 11149(50.38%) | 8113 (50%) | | 4590 (48.63%) | 3639 (38.55%) | |
Race, No.(%) | | | ༜0.001 | | | 0.959 |
| White | 18338(82.86%) | 14191 (87.45%) | | 8042 (85.2%) | 8029(85.06%) | |
| Black | 1958 (8.85%) | 868(5.35%) | | 643 (6.81%) | 646(6.84%) | |
| Others༊ | 1836 (8.3%) | 1168 (7.2%) | | 754 (7.99%) | 764 (8.09%) | |
Grade, No.(%) | | | ༜0.001 | | | ༜0.001 |
| Well differentiated (Grade I) | 2029 (9.17%) | 3068 (18.91%) | | 941(9.97%) | 896(9.49%) | |
| Moderately differentiated (Grade II) | 6832 (30.87%) | 7476 (46.07%) | | 3427 (36.31%) | 2942 (31.17%) | |
| Poorly differentiated (Grade III) | 12670(57.25%) | 5449 (33.58%) | | 4871 (51.61%) | 5382(57.02%) | |
| Undifferentiated (Grade IV) | 601 (2.72%) | 234 (1.44%) | | 200 (2.12%) | 219(2.32%) | |
Histology, No.(%) | | | ༜0.001 | | | 0.977 |
| Non-small cell - adenocarcinoma | 10643(48.09%) | 10479 (64.58%) | | 5150 (54.56%) | 5148 (54.54%) | |
| Non-small cell - Others༊ | 11489(51.91%) | 5748 (35.42%) | | 4289 (45.44%) | 4291(45.46%) | |
Radiotherapy, No.(%) | | | ༜0.001 | | | 0.907 |
| Yes | 11260(50.88%) | 1588 (9.79%) | | 1584 (79.9%) | 1588 (79.4%) | |
| None/Unknown | 10872(49.12%) | 14639 (90.21%) | | 7845 (20.1%) | 7851 (20.6%) | |
Chemotherapy, No.(%) | | | ༜0.001 | | | ༜0.001 |
| Yes | 9511 (42.97%) | 3180 (19.6%) | | 3112 (32.97%) | 2617 (27.73%) | |
| None/Unknown | 12621(57.03%) | 13047 (80.4%) | | 6327 (67.03%) | 6822(72.27%) | |
Others༊:Asian, Pacific Islander, American Indian or Alaska Native;Non-small cell - Others༊:Non-small cell NET,Non-small cell NEC,Non-small cell Others |
In order to compare the survival of surgical patients and non-surgical patients of different ages, 18,839 were divided into four stages according to age: 70–74, 75–79, 80–84, and ≥ 85, and the 5-year overall survival rates were 48.9%, 41.3%, 39.5% and 30.4% respectively, the 5-year css rates of patients at four stages were 60.7%, 55.3%, 56.1%, and 49.4% respectively, we can figure that out OS and CSS were significantly higher in the surgery group than in the non-surgery group at all ages(all p < 0.05) (Fig. 2).
Univariate and multivariate Cox regression analyses were applied to matched patients,The results of multivariate Cox analysis showed that Surgery (OR = 4.381, 95%CI 4.225–4.544, P < 0.001) was an independent predictor of OS (supplemental Table 1). Similarly, multivariate Cox analysis showed that surgery (OR = 5.216, 95%CI 5.001–5.440, P < 0.001) was significantly associated with CSS (supplemental Table 2). Surgery had the highest OR value, so surgical treatment was the most important protective factor for OS and CSS in older CRC patients. In addition, we further compared the survival of patients with and without surgery at different stages, and the results showed that surgery can significantly improve patient survival(all p < 0.05)(Fig. 3).
We further performed univariate multivariate Cox regression analysis on postoperative patients. Multivariate analysis showed that multivariate Cox analysis showed CT (OR = 1.187, 95%CI 1.082–1.301, P < 0.001) and RT (OR = 1.187, 95%CI 1.082–1.301, P < 0.001) correlated with OS(supplemental Table 3). Similarly, CT (OR = 1.187, 95%CI 1.082–1.301, P < 0.001) and RT (OR = 1.187, 95%CI 1.082–1.301, P < 0.001) were correlated with CSS. however, postoperative CT and RT are risk factors for elderly patients(supplemental Table 4).
To further analyze the impact of postoperative chemotherapy on patients, we divided postoperative patients into chemotherapy and non-chemotherapy group, and used PSM to balance baseline levels(supplemental Table 5).Compared with surgery alone and surgery + chemotherapy, the 5-year OS rate of patients with stage I lung cancer was 61.7% and 45.9%, respectively. (P = 0.0011, Fig. 4A).The 5-year CSS rates of surgery alone and surgery + chemotherapy were 79.0% and 51.5% respectively (P < 0.0001, Fig. 4E). In patients with stage 2 lung cancer, 5-year OS rates were 47.7% for surgery alone and 38.5% for surgery + chemotherapy (P = 0.0131, Fig. 4B). The 5-year CSS rates of surgery alone and surgery + chemotherapy were 64% and 47.5%,respectively (P = 0.0002, Fig. 4F). In patients with stage 3 lung cancer, 5-year OS rates were 46.2% for surgery alone and 42.3% for surgery + chemotherapy. (P = 0.0266, Fig. 4C); The 5-year CSS rates of surgery alone and surgery + chemotherapy were 61.3% and 50.5% respectively(P < 0.0001, Fig. 4G). In patients with stage IV lung cancer, the 5-year OS rate and CSS rate of surgery + chemotherapy show no significant difference compared with surgery alone (all P > 0.05, Fig. 4D,H).
In order to further analyze the impact of postoperative radiotherapy on patients, we divided postoperative patients into radiotherapy and non radiotherapy group, and used PSM to balance baseline levels(supplemental Table 6). In patients with stage I lung cancer, the 5-year OS rates of surgery alone and surgery + radiotherapy were 53.6% and 52%, respectively (P = 0.4303, Fig. 5A); The 5-year CSS rates of surgery alone and surgery + radiotherapy were 73.1% and 65.3% respectively(P = 0.1879, Fig. 5E). In patients with stage II lung cancer, the 5-year OS rates of surgery alone and surgery + radiotherapy were 51.4% and 33.6% respectively. (P < 0.001, Fig. 5B); The 5-year CSS rates of surgery alone and surgery + radiotherapy were 69.0% and 44.8% respectively (P < 0.0001, Fig. 5F). In patients with stage III lung cancer, the 5-year OS rates of surgery alone and surgery + radiotherapy were 42.1% and 19.8% respectively (P < 0.0001, Fig. 5C); The 5-year CSS rates of surgery alone and surgery + radiotherapy were 56.8% and 29.4% respectively(P < 0.0001, Fig. 5G). In patients with stage IV lung cancer, there was no significant difference in 5-year CSS rate between surgery + radiotherapy and surgery alone (P > 0.05, Fig. 5H).