Table 1.
Overview of model development studies for the prognosis of patients with colorectal liver metastases.
| First Author (Year) | Data Collection | Model Type | Univariate Screening of Predictors | Outcome(s) | Patients (n) | Internal/External Validation | Missing Data | Risk Groups |
|---|---|---|---|---|---|---|---|---|
| Buisman (2022) [20] | retrospective | Cox regression | no | OS | 4112 | Cross-validation | Multiple imputation | yes (4 groups) |
| Bertsimas (2022) [21] | retrospective | RF, OPT | no | OS and intrahepatic recurrence | 761 | IV: Split sample/external validation cohort | Complete case analysis | no |
| Bao (2021) [22] | retrospective | NGS, Cox and LASSO regression | yes | OS | 144 | External validation (gene signature only) | No information | yes (2 groups) |
| Lam (2023) [23] | retrospective | Cox and LASSO regression | yes | OS and RFS | 572 | Split sample | Multiple imputation | no |
| Reijonen (2023) [24] | retrospective | Cox regression | yes | OS and DFS | 816 | Not performed | The final sum of their risk score points was estimated using the mean of the evaluable predictors | yes (3 groups) |
| Margonis (2018) [25] | retrospective | Cox regression | yes | OS | 502 (development), 747 (validation) | External validation | No information | yes (3 groups) |
| Paredes (2020) [26] | retrospective | Mixed-effects logistic regression | no | Recurrence | 703 (development), 703 (validation) | Split sample, bootstrapping | Multiple imputation | yes (3 groups) |
| Fruhling (2021) [27] | retrospective | Cox regression | yes | OS | 1212 | Cross-validation | Multiple imputation | yes (3 groups) |
| Taghavi (2021) [28] | retrospective | RF | no | Development of metachronous metastases | 70 (development), 21 (validation) | Split sample, cross-validation | Single imputation | no |
| Brudvik (2019) [29] | retrospective | Cox regression | no | OS, RFS | 564 (development), 608 (validation) | External validation | Complete case analysis | no |
| Moaven (2023) [30] | retrospective | GBT and LRB in a leave-one-out cross-validation | no | OS, recurrence | 1004 | Cross-validation, bootstrapping | Variables with more than 20% missing data were eliminated from the model | yes (3 groups) |
| Villard (2022) [31] | retrospective | Cox regression | no | OS | 1013 (development), 391 (validation) | External validation | Multiple imputation | yes (4 groups) |
| Chen (2020) [32] | retrospective | Cox regression | no | RFS | 787 (cohort 1), 162 (cohort 2) | IV: Bootstrapping/temporal validation | Complete case analysis | yes (3 groups) |
| Chen (2022) [33] | retrospective | Cox regression | yes | OS | 1095 | Not performed | Multiple imputation | no |
| Dai (2021) [34] | retrospective | Logistic regression | yes | Early recurrence within 6 months | 150 (development), 52 (validation) | Split sample | Complete case analysis | no |
| Liu (2021) [35] | retrospective | Cox regression | yes | OS after recurrence | 867 | Bootstrapping | No information | yes (2 groups) |
| Liang (2021) [36] | retrospective | Cox regression | yes | Post-recurrence survival | 251 (development), 125 (validation) | Split sample, bootstrapping | Complete case analysis | yes (3 groups) |
| Wu (2021) [37] | retrospective | Cox regression | yes | Recurrence, PFS | 229 (development), 128 (validation) | Temporal validation | Complete case analysis | yes (3 groups) |
| Sasaki (2022) [38] | prospective | Cox regression | yes | OS | 1205 (development), 1307 + 1058 (validation) | External validation | No information | yes (3 groups) |
| Huiskens (2019) [39] | retrospective | Logistic regression | yes | 90-day mortality (after stage 2) | 486 | Not performed | Complete case analysis | yes (3 groups) |
| Bai (2022) [40] | retrospective | Cox regression | yes | OS and RFS | 341 (development), 325 (validation) | External validation | Complete case analysis | yes (3 groups) |
| Fang (2022) [41] | retrospective | Cox regression | yes | OS | 237 | Not performed | Complete case analysis | yes (3 groups) |
| Qin (2022) [42] | prospective | Cox regression | yes | ihPFS | 314 | Not performed | No information | yes (3 groups) |
| Kawaguchi (2021) [43] | prospective | Cox regression | yes | OS | 810 (development), 673 (validation) | External validation | Complete case analysis | no |
| Zhang (2023) [44] | retrospective | Cox and LASSO regression | yes | OS | 415 (development), 207 (validation) | IV: Split sample/External validation cohort | No information | yes (2 groups) |
| Chen (2021) [45] | retrospective | Logistic and Cox regression | yes | Postoperative complications, PFS, OS | 380 | Not performed | Complete case analysis | yes (3 groups) |
| Jin (2022) [46] | retrospective | Cox regression | yes | CSS | 881 (development), 169 (validation) | IV: Split sample/External validation cohort | Complete case analysis | yes (2 groups) |
| Zhai (2022) [47] | retrospective | Cox regression | yes | Liver RFS | 147 | Not performed | Complete case analysis | yes (3 groups) |
| Liu (2021) [48] | retrospective | Cox regression | yes | PFS | 532 (development), 237 (validation) | External validation | No information | yes (2 groups) |
| Moro (2020) [49] | retrospective | CART analysis | no | OS | 1123 | Bootstrapping | Multiple imputation | yes (4 groups) |
| Chen (2021) [50] | retrospective | Logistic and Cox regression | yes | Complications, PFS, OS | 169 | Not performed | Complete case analysis | yes (3 groups) |
| Yao (2021) [51] | retrospective | Logistic and Cox regression | yes | Presence of LN metastases, PFS | 241 | Not performed | Complete case analysis | no |
| Kazi (2023) [52] | retrospective | Logistic and Cox regression | yes | Serious complications | 92 | Bootstrapping | No information | yes (4 groups) |
| Meng (2021) [53] | retrospective | Cox regression | yes | OS | 174 (development), 60 (validation) | Split sample | Complete case analysis | yes (2 groups) |
| Imai (2016) [54] | prospective | Cox regression | yes | OS | 439 | Not performed | No information | yes (4 groups) |
| Chen (2022) [55] | retrospective | Logistic regression | yes | Early recurrence (<11 months) | 144 (development), 40 (validation) | Another cohort from the same hospital | Complete case analysis | no |
| Cheng (2022) [56] | retrospective | Cox regression | yes | CSS | 1314 (development), 560 (validation) | Split sample | Complete case analysis | yes (2 groups) |
| Kulik (2018) [57] | retrospective | Logistic regression | yes | OS | 965 | Not performed | Complete case analysis | no |
| Bai (2021) [58] | retrospective | Cox regression | yes | OS | 490 | Not performed | Complete case analysis | yes (7 and 6 groups) |
| Wang (2021) [59] | retrospective | Cox and LASSO regression | no | OS | 113 (development), 114 (validation), 168 (external validation) | IV: Split sample/external validation cohort | Complete case analysis | yes (2 groups) |
| Xu (2021) [60] | retrospective | Logistic regression | yes | Major pathologic response to chemotherapy | 241 (development), 241 (validation) | Split sample | Complete case analysis | yes (2 groups) |
| Sasaki (2018) [61] | retrospective | A priori selection of predictors and interactions | no | OS | 604 (development) | External validation | No information | yes (3 groups) |
| Wada (2022) [62] | retrospective | Cox and LASSO regression | no | Recurrence | 169 (development), 151 (validation) | External validation | No information | yes (2 groups) |
| Kim (2020) [63] | retrospective | Cox regression | yes | Recurrence | 197 (development), 98 (validation) | Split sample | No information | yes (2 groups) |
| Dupre (2019) [64] | prospective | Cox regression | yes | OS | 364 (development), 219 (validation) | External validation | No information | yes (2 groups) |
| Qi (2023) [65] | retrospective | Automated tissue classification and quantification of CRLM SOFs derived from histology images with deep learning and Cox regression | yes | OS | 433 (development), 403 (validation) | External validation | Complete case analysis | yes (SOF scoring system 2 groups, SOF-CRS 3 groups) |
| Wu (2021) [66] | retrospective | Cox regression | yes | PFS | 158 | Not performed | Complete case analysis | yes (3 groups) |
| Dasari (2023) [67] | retrospective | Cox and LASSO regression | yes | OS | 927 (development), 309 (validation) | Split sample | Complete case analysis | yes (5 groups) |
| Liu (2023) [68] | retrospective | Cox and LASSO regression | yes | OS | 295 (development), 295 (validation) | Split sample | Complete case analysis | yes (2 groups) |
| Amygdalos (2023) [69] | retrospective | GBT with the Top6 selected predictors | no | OS | 389 (development), 98 (validation) | Split sample | Complete case analysis | yes (2 groups) |
| Chen (2023) [70] | retrospective | Cox regression | yes | OS | 85 | Not performed | Complete case analysis | yes (3 groups) |
| Wu (2018) [71] | retrospective | Cox regression | yes | OS and CSS | 4825 (development), 4790 (validation) | Split sample | Complete case analysis | no |
| Deng (2023) [72] | retrospective | Logistic regression | yes | Early recurrence (<13 months) | 323 (development), 71 (validation) | External validation | Complete case analysis | no |
| Berardi (2023) [73] | prospective | Logistic regression | yes | Treatment failure (recurrence or death within 12 months) | 535 (development), 248 (validation) | Split sample | No information | yes (2 groups) |
| Liu (2019) [74] | retrospective | Cox regression | yes | DFS | 447 (development), 117 (validation) | External validation | No information | yes (3 groups) |
| Welsh (2008) [75] | prospective | Logistic regression | yes | R1 resection margin | 911 | Bootstrapping | Single (median) imputation | no |
| Famularo (2023) [76] | prospective | Survival RF to estimate the best possible treatment, then CART was used to develop a decision tree | no | OS | 448 | Cross-validation | Multiple imputation | yes (7 groups) |
| He (2023) [77] | retrospective | Logistic regression | yes | Benefit from upfront surgery (survival > 15 months) | 572 (development), 242 (validation) | Split sample | Complete case analysis | no |
| Kattan (2008) [78] | retrospective | Cox regression | yes | DSS | 1477 | Bootstrapping | No information | no |
| Wensink (2023) [79] | retrospective | Cox regression | no | Early extrahepatic recurrence (at 6 and 12 months) | 1077 | Bootstrapping and internal–external cross-validation | Multiple imputation | yes (4 groups) |
| Fendler (2015) [80] | retrospective | Cox regression | yes | OS | 100 (development), 25 (validation) | IV: Bootstrapping/external validation cohort | No information | no |
| Marfa (2016) [81] | prospective | CART analysis | no | OS | 57 (development), 28 (validation) | Split sample | No information | yes (2 groups) |
| Jiang (2023) [82] | retrospective | Cox regression | yes | OSS and CSS | 225 (development), 180 (validation) | External validation | Complete case analysis | no |
| Endo (2023) [83] | retrospective | OPT analysis | no | OS and RFS | 679(development), 679 (validation) | Split sample | Multiple imputation | yes (multiple nodes) |
| Rees (2008) [84] | prospective | Cox regression | yes | CSS | 929 | Bootstrapping | Single (median) imputation | yes (5 groups) |
| Zakaria (2007) [85] | retrospective | Cox regression | yes | DFS, recurrence | 662 | Not performed | Complete case analysis | yes (3 groups) |
| Tan (2008) [86] | retrospective | Cox regression | yes | OS | 296 | Not performed | Multiple imputation | yes (3 groups) |
| Hill (2012) [87] | retrospective | Cox regression | yes | Survival following resection for recurrence | 280 | Bootstrapping | No information | yes (3 groups) |
| Takeda (2021) [88] | retrospective | Cox regression | yes | OS | 341 (development), 309 (validation) | External validation | Complete case analysis | yes (4 groups) |
| Wang (2017) [89] | retrospective | Cox regression | yes | OS | 300 | Not performed | No information | yes (4 groups) |
| Spelt (2013) [90] | retrospective | ANN and Cox regression | yes | OS | 241 | Cross-validation | Multiple imputation | no |
OS: overall survival, RF: random forest, OPT: optimal policy tree, IV: internal validation, NGS: next-generation sequencing, LASSO: least absolute shrinkage and selection operator, RFS: recurrence-free survival, DFS: disease-free survival, GBT: gradient-boosted trees, LRB: logistic regression with bootstrapping, ihPFS: intrahepatic progression-free survival, CART: classification and regression tree, LN: lymph node, CSS: cancer-specific survival, CRLM: colorectal liver metastasis, SOFs: spatial organization features, CRS: clinical risk score, ANN: artificial neural network.