Table 2.
Patients’ characteristics in the training and test datasets.
| Variables | Training dataset (n=194) | Test dataset (n=85) | |||||
|---|---|---|---|---|---|---|---|
| PNI- (n=144) | PNI+ (n=50) | Pintra | PNI- (n=63) | PNI+ (n=22) | Pintra | Pinter | |
| Age (years, SD) | 66.21 (9.8) | 64.16 (9.7) | 0.203 | 63.35 (10.7) | 63.59 (7.8) | 0.923 | 0.078 |
| Gender (N, %) | |||||||
| Male | 103 (71.5) | 31 (62.0) | 0.209 | 47 (74.6) | 18 (81.8) | 0.492 | 0.208 |
| Female | 41 (28.5) | 19 (38.0) | 16 (25.4) | 4 (18.2) | |||
| CEA (N, %) | |||||||
| Abnormal | 45 (31.3) | 27 (54.0) | 0.004* | 20 (31.7) | 13 (59.1) | 0.023* | 0.786 |
| Normal | 99 (68.7) | 23 (46.0) | 43 (68.3) | 9 (40.9) | |||
| CA19-9 (N, %) | |||||||
| Abnormal | 11 (7.6) | 8 (16.0) | 0.151 | 4 (6.3) | 6 (27.3) | 0.025* | 0.620 |
| Normal | 133 (92.4) | 42 (84.0) | 59 (93.7) | 16 (72.7) | |||
| DIS (cm, SD) | 8.11 (3.3) | 7.15 (4.3) | 0.151 | 8.20 (4.0) | 7.86 (4.7) | 0.740 | 0.608 |
| CRM status (N, %) | |||||||
| Positive | 29 (20.1) | 20 (40.0) | 0.005* | 7 (11.1) | 9 (40.9) | 0.006* | 0.242 |
| Negative | 115 (79.9) | 30 (60.0) | 56 (88.9) | 13 (59.1) | |||
| mrEMVI status (N, %) | |||||||
| Positive | 28 (19.4) | 16 (32.0) | 0.068 | 14 (22.2) | 8 (36.4) | 0.192 | 0.562 |
| Negative | 116 (80.6) | 34 (68.0) | 49 (77.8) | 14 (63.6) | |||
| Tumor stage (N, %) | |||||||
| T1-2 | 48 (33.3) | 3 (6.0) | 0.000* | 22 (34.9) | 2 (9.1) | 0.020* | 0.736 |
| T3-4 | 96 (66.7) | 47 (94.0) | 41 (65.1) | 20 (90.9) | |||
| Lymph node (N, %) | |||||||
| N0 | 56 (38.9) | 11 (22.0) | 0.030* | 29 (46.0) | 6 (27.3) | 0.124 | 0.289 |
| N1-2 | 88 (61.1) | 39 (78.0) | 34 (54.0) | 16 (72.7) | |||
PNI-, patients without pathologic perineural invasion; PNI+, patients with pathologic perineural invasion; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; DIS, the distance from the end of the convex edge of the tumor to the edge of the anus; CRM, circumferential resection margin; mrEMVI, MRI-based extramural vascular invasion. Data are presented as counts or means (standard deviations in parentheses). Pintra is the result of univariate analyses between the PNI+ and PNI- groups while Pinter represents whether a significant difference exists between the training and test datasets. *P < 0.05.