Table 2. Cox proportional regression analysis of the optimal cutoff value calculated from the exploratory dataset.
| Number of patients | PFS | ||||||
|---|---|---|---|---|---|---|---|
| Optimal cutoff | Above cutoff | Below cutoff | Uni-variate analysis | Multi-variate analysisc) | |||
| n (%) | n (%) | HR (95% CI) |
P | aHR (95% CI) |
P | ||
| Co-occurence homogeneity | 6930 | 17 (10.6) | 144 (89.4) | 3.85 (2.14–6.93) |
<0.01a) | 3.19 (1.47–6.92) |
<0.01a) |
| Voxel-alignment intensity variabillity | 692 | 16 (9.9) | 145 (90.1) | 4.57 (2.39–8.72) |
<0.01a) | 3.66 (1.65–8.10) |
<0.01a) |
| Neighborhood-intensity difference busyness | 1.97 | 15 (9.3) | 146 (90.7) | 4.39 (2.30–8.38) |
<0.01a) | 3.33 (1.49–7.42) |
<0.01a) |
|
Intensity size zone intensity variability |
98 | 10 (6.2) | 151 (93.8) | 6.20 (2.82–13.65) |
<0.01a) | 4.27 (1.78–10.30) |
<0.01a) |
| Co-occurence entropy | -173000 | 153 (95.0) | 8 (5.0) | 6.41 (2.80–14.68) |
<0.01b) | 4.86 (1.97–11.98) |
<0.01b) |
| Voxel-alignment short zone emphasis | 0.43 | 140 (87.0) | 21 (13.0) | 4.50 (2.42–8.39) |
<0.01b) | 3.95 (1.77–8.81) |
<0.01b) |
| Neighborhood-intensity difference contrast | 0.01192 | 131 (81.4) | 30 (18.6) | 3.09 (1.24–7.71) |
0.02b) | 2.37 (0.92–6.13) |
0.07b) |
| Intensity size zone high intensity zone emphasis | 418 | 146 (90.7) | 15 (9.3) | 3.34 (1.75–6.34) |
<0.01b) | 3.18 (1.53–6.55) |
<0.01b) |
| SUV(max) | 16.8 | 26 (16.1) | 135 (83.9) | 2.77 (1.58–4.87) |
<0.01a) | 2.23 (1.17–4.24) |
0.01a) |
| SUV(average) | 6.45 | 112 (69.6) | 49 (30.4) | 1.64 (1.01–2.70) |
0.05a) | 1.43 (0.84–2.41) |
0.19a) |
| SUV(metabolic volume) | 109 | 24 (14.8) | 137 (85.2) | 2.89 (1.65–5.05) |
<0.01a) | - | - |
| SUV(metabolic volume) categorized by 45cm3 | 45 | 51 (31.7) | 110 (68.3) | 1.88 (1.15–3.10) |
0.01a) | - | - |
| Initial disease status | - | - | - | 2.01 (0.86–4.65) |
0.11 | - | - |
| ECOG PS | - | - | - | 2.28 (1.24–4.20) |
0.01 | - | - |
| Type of TKI | - | - | - | 0.72 (0.10–5.23) |
0.74 | - | - |
| Type of EGFR mutations | - | - | - | 0.94 (0.74–1.20) |
0.62 | - | - |
Initial disease status was divided into two groups: recurred and metastatic.
ECOG PS was divided into two groups: subjects with ECOG PS 0 and 1 vs. subjects with ECOG PS 2, 3, and 4.
a) P calculated by Cox proportional regression analysis compared subjects above the optimal cutoff value to those below the optimal cutoff value
b) P calculated by Cox-proportional regression analysis compared to subjects with below optimal cutoff value to subjects with above optimal cutoff value
c) Multivariate analyses were conducted for each parameter adjusted for ECOG PS, SUV metabolic volume (categorized) and initial disease status
Abbreviations: PFS = progression free survival, HR = hazard ratio, CI = confidential interval, aHR = adjusted hazard ratio, SUV = standardized uptake value