Table 1.
A table to demonstrate patient demographic differences in training and validation data.
| Characteristic | Training, N = 1951 | Validation, N = 761 | p-value2 |
| Tumor volume (cc) | 4 (0 - 31) | 14 (0 - 158) | <0.001 |
| Tumor motion (cm) | 0.56 (0.00 - 3.43) | 0.36 (0.00 - 2.73) | 0.016 |
| Tumor lobe location | 0.7 | ||
| Lower | 69 (35%) | 29 (38%) | |
| Upper | 126 (65%) | 47 (62%) | |
| Age (years) | 75 (45 - 92) | 78 (52 - 92) | 0.036 |
| Biological sex | 0.3 | ||
| Female | 99 (51%) | 44 (58%) | |
| Male | 96 (49%) | 32 (42%) | |
| Histological subtype | <0.001 | ||
| Adenocarcinoma, NOS | 35 (37%) | 45 (59%) | |
| Squamous cell carcinoma | 37 (39%) | 24 (32%) | |
| Carcinoma, NOS | 14 (15%) | 0 (0%) | |
| Other | 8 (8.5%) | 7 (9.2%) | |
| Unknown | 101 | 0 | |
| Performance status (ECOG) | <0.001 | ||
| 0 | 3 (1.8%) | 29 (38%) | |
| 1 | 64 (37%) | 43 (57%) | |
| 2 | 83 (49%) | 4 (5.3%) | |
| 3 | 21 (12%) | 0 (0%) | |
| Unknown | 24 | 0 | |
| Local relapse | 13 (6.7%) | 12 (16%) | 0.020 |
| Regional failure | 15 (7.7%) | 5 (6.6%) | 0.8 |
1Statistics presented: n (%); Median (range).
2Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test.
Tumours were larger and less mobile in the validation data compared to training. Worse performance status was reported in the training set. Differences were also identified in histological sub-type with a larger proportion of adenocarcinoma in the validation data, but this could be influenced by the missing data reported for training. In the validation set there were significantly more local relapses but a similar percentage of regional failures.