Table 1 Studies of potential radiomic biomarkers for PDAC patients considering radiotherapy
From: Selection of patients with pancreatic adenocarcinoma who may benefit from radiotherapy
Study | Study type | Pt No. | PDAC stage | Image | Timing | Radiomic parameter | Endpoints | Conclusion | Significance |
|---|---|---|---|---|---|---|---|---|---|
2019 Khalvati F [32] | Retrospective | 98 | Resectable | Baseline CT | Pre-op | Sum entropy, cluster tendency features | OS | May stratify patients for NAT or alternatives | HR 1.56, 1.35 p = 0.005; 0.022 for two readers |
2022 Elsherif SB [33] | Retrospective | 54 | Resectable/ BR | Baseline Dual energy CT | Pre-NAT | ∫ T (HU·mL) (PPP) = 507.85 | OS, PFS | Predict pathologic lymph node status | P = 0.006 P = 0.001 |
2020 Parr E [34] | Retrospective | 74 | Localized | Contrast- enhance CT | RT planning CT | 6 or 7 radio-genomic features | OS Loco-regional recurrence | OS and local recurrence better predict by radiomic features than clinical features | 6 features for OS: p < 0.0001; 7 features for loco-regional recurrence: p < 0.0001 |
2022 Wang CX [35] | Retrospective | 297 | BR & LA | Pre, arterial, venous phase contrast-enhanced CT | Before CRT | Normalized area under the curve (nAUC) | OS | nAUC, correspond to tumor cell death validated by histo-pathology, predict OS | P < 0.0001 |
2022 Koay EJ [36] | Prospective Nab-Cape with concomitant RT after ICT | 23 | BR & LA | Baseline (post- ICT) and follow-up (post-CRT) CT | Post-ICT, post-CRT | Type I (remains or sharper) and II (blurring) interface response | OS, PFS | Type I interface response associate with better OS and PFS | p = 0.004, p = 0.03 |
2022 Rossi G [37] | Retrospective | 71 | LA | Contrast-enhanced planning CT for RT | Post-ICT | 4/1655 radiomic features | Resectability | 4 features-model predict resectability after NAT | AUC: 0.944 |
2019 Cozzi L [38] | Retrospective | 100 | LA for SBRT | Contrast free planning CT | Before SBRT | 9 features for OS 4 features for local control | OS Local control | Low risk group had OS 15.1 and local control 28.6 months | P = 0.05 P = 0.004 |
2018 Cheng Z [39] | Retrospective | 191 | BR & LA | Planning CT for SBRT | Before SBRT | Overlap-volume histograms of GTV to key arterial structure | Margin negative resection | Tumor involved > 1 cm key arterial structure, less likely to have margin negative resection (23% vs. 77%) | P < 0.01 |
2019 Nasief H [40] | Retrospective | 24 | Resectable or BR | Daily non-contrast CTs during CT-guided CRT | During CRT | 73 Delta radiomic features/ >1300 radiomic features | Response and OS | Decreased delta radiomic features and CA19-9 predict better OS | P = 0.001 P = 0.031 |
2019 Yamamoto KN [41] | Retrospective | 1089 | LA | Three sequential CT during ICT | FFX, GEM or Gem + Nab | Time series tumor volume data derived Local/metastatic advancement index (LAI/MAI) | Primary tumor size, metastatic number, OS | 1. RT after ICT improves OS in larger LAI 2. CRT leads to a significant survival benefit when FFX but not GEM or GEM + Nab | 1.P = 0.0547,0.0429,0.0379 for FFX, Gem, Gem + Nab; 2. p = 0.008, 0.236, 0.253 for FFX, GEM, GEM + Nab |
2018 Bali MA [42] | Prospective | 24 | PDAC receiving chemotherapy | DW-MR | Baseline, wk2, wk8 post-chemotherapy | ROI-ADC, DW-volume, diffusion parameters | Response | At wk2, 25th percentile of H-D and H-PF change correctly classified response in 20/24 pts; at wk8, DW-volume change correctly classified 22/24 pts | P = 0.003 P < 0.0001 |
2020 Itchins M [44] | Retrospective | 115 | Resectable & BR | PET | Baseline and pre-op | SUV(max) | OS | Pre-op SUV(max) < 5 after NAT predict improved OS (42.95 vs. 26.05 months) | P = 0.02 |
2022 Abdelrahman AM [45] | Retrospective | 202 | Resected BR/LA | PET | Before and after NAT | Metabolic response | Pathologic response, OS | Metabolic response predict pathologic response and OS | P < 0.001 P < 0.001 |
2021 Panda A [46] | Retrospective | 44 | BR & LA | PET | Before and after NAT | Complete metabolic response, mean change in SUVmax | Pathologic response OS | Change in SUVmax and complete metabolic response were associated with OS | P < 0.05 |
2021 Zimmermann C [47] | Prospective | 25 | Resectable BR & LA | PET/CT | Before and after NAT | Decreased SUVmax ≥ 30% | Response | Median SUVmax decreased after NAT (8.29 and 3.83) | p < 0.001 |
2017 Sakane M [48] | Retrospective | 25 | Resectable & BR | PET/CT | Before and after CRT | SUVpeak, MTV, TLG | Pathologic response | Higher post-CRT SUVpeak, positive MTV/TLG predict unfavorable pathologic effects of CRT | P = 0.013 P = 0.014 |