Table 2.
Landscape of clinical prediction models in metastatic non-small-cell lung cancer
| Author (year) | Study design | Sample size | Stage and type of lung cancer | Type of model | Outcome(s) investigated | Type of prognostic features included in the model(s) | Predictive features included in the final model(s) | Variables significantly associated with outcome | Model discrimination | Guideline approval |
|---|---|---|---|---|---|---|---|---|---|---|
| Banna et al. (2021)12 | Retrospective | 784 | mNSCLC with PD-L1 of at least 50% treated with first-line immunotherapy | Multivariate Cox regression analysis | Overall survival | Clinicopathological data | Performance status, use of steroids, NLR | Performance status, use of steroids, NLR | 1-year overall survival Favourable risk = 78.2% Intermediate risk = 53.8% Poor risk = 10.7% |
No—needs further validation with RCTs |
| Banna et al. (2022)13 | Retrospective | 308 | mNSCLC | Multivariate stepwise Cox regression analysis | Overall survival, progression-free survival, NLR, and SII | Clinicopathological data | Number of metastatic sites, squamous histology, SII | Number of metastatic sites, squamous histology, SII | Median overall survival Favourable risk = 20.3 months Intermediate risk = 12.4 months Poor risk = 8.4 months C-index = 0.623 Progression-free survival: Favourable risk = 11.3 months Intermediate risk = 7.9 months Poor risk = 5.7 months C-index = 0.613 |
No—needs further validation with RCTs |
| Chen et al. (2019)25 | Retrospective | N/A | NSCLC | N/A | Hypoxic effect on cells and their transcription profile, epigenetic effects | Genetic data | 17 prognostic genes | 17 prognostic genes | Three datasets showed significance in survival time between high- and low-risk groups: 0.0049, 0.04, and 0.025, respectively | N/A—pre-clinical |
| Ding et al. (2022)26 | Retrospective | 764 samples (31 non-metastatic, 733 metastatic) | NSCLC | Multivariate Cox regression and LASSO | Association between prognostic risk model and TNM stage, immune microenvironment, mechanisms | Genetic data | 6 mRNAs | 6 mRNAs | N/A | N/A—pre-clinical |
| Wang et al. (2023)27 | Retrospective | 599 samples | mNSCLC with brain metastases | Multivariate Cox regression analysis and LASSO analysis | Tumour mutational signature, immune signature, sensitivity to treatment | Genetic data | 11 prognostic genes | 11 prognostic genes | N/A | N/A—pre-clinical |
| Liu et al. (2021)28 | Retrospective | 243 | m NSCLC treated with chemotherapy and radiation | Multivariate Cox regression analysis | Overall survival | Clinicopathological features | Sex, KPS score, number of chemotherapy cycles, Hb level, neutrophil count, platelet count | Sex, KPS score, number of chemotherapy cycles, Hb level, neutrophil count, platelet count | Median overall survival (months) Low risk = 17.0 Moderate risk = 14.0 High risk = 8.0 1-year OS rate: Low risk = 67.7% Moderate risk = 59.6% High risk = 26.2% 2-year OS rate: Low risk = 32.1% Moderate risk = 18.0% High risk = 7.9% 3-year OS rate: Low risk = 19.3% Moderate risk = 7.9% High risk = 0% |
No—does not account for molecular therapy |
| Ganti et al. (2019)29 | Retrospective | 1467 | mNSCLC | Multivariable Cox proportional hazards model | Overall survival/progression-free survival | Clinical data | Sex, performance status, tumour stage, weight loss | Sex, performance status, tumour stage, weight loss | Training cohort Good prognosis = 11.77 months Poor prognosis = 6.21 months Area under 1-year ROC = 0.61 Area under 2-year ROC = 0.62 C-index = 0.57 Testing cohort: Good prognosis = 13.15 months Poor prognosis = 8.52 months Area under 1-year ROC = 0.60 Area under 2-year ROC = 0.65 C-index = 0.57 |
No—needs further validation with RCTs |
| Navani et al. (2023)30 | Observational cohort | 495 | mNSCLC treated with at least 1 dose of ICI monotherapy | LASSO Cox regression | Overall survival | Clinicopathological data | ECOG, LDH level, NLR | ECOG, LDH level, NLR | Median overall survival Favourable risk = 28.3 months Intermediate risk = 9.1 months Poor risk = 2.1 months Intermediate versus favourable HR 2.08 Poor versus favourable HR 5.21 |
No—needs further validation with RCTs |
| Mezquita et al. (2018)33 | Retrospective | 466 | mNSCLC | Multivariate Cox proportional hazards regression model | Overall survival, progression-free survival, disease control rate | Pathological data | dNLR, LDH | dNLR, LDH | Median overall survival: Good risk = 34 months Intermediate risk = 10 months Poor risk = 3 months Progression-free survival Good risk = 6.3 months Intermediate risk = 3.7 months Poor risk = 2.0 months |
No—needs further validation with RCTs |
| Chen et al (2023)34 | Retrospective | 194 | mNSCLC | Linear regression | Overall survival | Pathological data | 15 radiomic feature, fractal dimensions, texture features | PD-L1 status, treatment response | Moderate discrimination PD-L1 expression AUC = 0.66-0.72 Treatment response AUC = 0.68 |
No—needs further validation with RCTs |
dNLR, derived neutrophil-to-lymphocyte ratio; ECOG, Eastern Cooperative Oncology Group; Hb, haemoglobin; ICI, immune checkpoint inhibitor; KPS, Karnofsky Performance Status; LASSO, least absolute shrinkage and selection operator; LDH, lactate dehydrogenase; mNSCLC, metastatic non-small-cell lung cancer; N/A, not applicable; NLR, neutrophil-to-lymphocyte ratio; PD-L1, programmed death-ligand 1; RCTs, randomised control trials; ROC, receiver operating characteristic; SII, systemic immune-inflammatory index; TNM, tumour–node–metastasis.