Characteristics of Current and Previous Studies
| Author | Study design | No. of patients | FIGO stage | Primary treatment | Imaging | Training/validation set | Outcomes | Follow-up (mo) | Feature type | Best AUC |
|---|---|---|---|---|---|---|---|---|---|---|
| Yang et al. (13) | Retrospective | 20 | IB1-IVA | CCRT + BT | PET | No | RTT | — | 6 texture features | NA |
| Yang et al. (14) | Retrospective | 90 | IB2-IVA | CCRT + BT | PET | No | RTT | — | 4 texture features | RTT (energyGLCM) = 0.659 (0.534–0.756) |
| Ho et al. (15) | Prospective | 44 | IB-IIIB | CCRT + BT | PET | No | RTT; OS | 56 (10–83) | 4 texture features | RTT (GLNUGLRLM) = 0.76 (0.56–0.88) |
| Reuzé et al. (16) | Retrospective | 118 | IB1-IVB | CCRT + BT | PET | Group 1 = 79; group 2 = 39 | LR | 36 (median) | Mixed* | LR = 0.86 (0.75–0.97) |
| Altazi et al. (17) | Retrospective | 80 | IB-IVA | CCRT + BT | PET | TS = 56; VS = 24 | DM; LRR | 19 (mean) | Mixed* | DM = 0.92 (SE 0.07); LRR = 0.92 (SE 0.05) |
| Chen et al. (18) | Retrospective | 142 | IB-IIIB | CCRT + BT | PET | TS = 77; VS = 65 | OS; PFS; DMFS; PRFS | 40 (7–84) | 7 texture features | PRFS (HGREGLRLM) = 0.31; PFS = 0.36 |
| Schernberg et al. (19) | Retrospective | 108 | IB1-IVA | CCRT + BT | PET | TS = 69; VS = 39 | LRC; OS | TS = 57.0 (6.8–100.6); VS = 30.8 (5.0–60.0) | Mixed* | NA |
| Lucia et al. (20) | Retrospective | 102 | IB1-IVA | CCRT + BT | PET; MRI | TS = 69; VS = 33 | DFS; LRC | TS = 36 (6–79);VS = 17 (6–30) | 1 texture feature | LRC (GLNUGLRLM) = 0.95 |
| Lucia et al. (21) | Retrospective multicenter | 190 | IB1-IVA | CCRT + BT | PET; MRI | TS = 112; VS = 78 | DFS; LRC | TS = 24.3 (6.0–83.0); VS1 = 33 (14–50); VS2 = 26 (7–86) | Mixed* | LRC = 0.93 |
| Mu et al. (22) | Retrospective multicenter | 154 | IIB-IVA | CCRT + BT | PET; CT | TS = 78; VS = 76 | PFS; OS | TS = 28 (15–39); VS = 25 (15–36) | Mixed* | PFS = 0.72OS = 0.79 |
| Ferreira et al. (23) | Retrospective multicenter | 158 | IB1-IVA | CCRT + BT | PET | TS 80%; VS 20% | DFS | 23 (4–84) | Mixed* | DFS = 0.78 (0.67–0.88) |
| de Alencar et al. (24) | Retrospective | 47 | IB2-IVA | CCRT + BT | PET | No | OS | 23.5 (3.7–39.0) | 3 texture features | OS (LRLGEGLRLM) = 0.74 |
| Cho et al. (25) | Retrospective multicenter | 68 | IIB-IVA | CCRT + BT | PET | No | OS; PFS | 49 (12–139) | 1 texture feature | PFS (GLNUGLRLM) = 0.85 (0.74–0.92) |
| Current study | Retrospective | 195 | IB2-IIIB | CRT + surgery | PET | No | pR; DFS; OS | 76.0 (70.7–78.7) | 1 shape and 3 texture features | pR = 0.65 (0.57–0.73); OS = 0.63 (0.54–0.73) |
↵* Radiomic signature models with or without clinical, CT, or MRI features.
FIGO = International Federation of Gynecology and Obstetrics; CCRT = concurrent chemoradiotherapy; BT = brachytherapy; RTT = response to treatment; NA = not available; energyGLCM = energy derived from gray-level cooccurrence matrix; OS = overall survival; GLNU = gray-level nonuniformity; GLRLM = gray-level run-length matrix; LR = local recurrence; TS = training set; VS = validation set; DM = distant metastasis; LRR = locoregional recurrence; OS = overall survival; PFS = progression-free survival; DMFS = distant metastasis-free survival; PRFS = pelvic relapse-free survival; HGRE = high gray-level run emphasis; LRC = locoregional control; DFS = disease-free survival; LRLGE = long runs gray-level emphasis.