Table 4.
Thromboembolism risk classification, estimated cumulative incidence of thromboembolism, and predictive ability of newly derived and established thromboembolism risk models in the NSCLC cohort.
| Risk Model | Cohort | High Risk | Low Risk | Prediction of Thromboembolism | Prediction of Mortality | Prediction of Progression | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. (%) | Cum. TE % a | No. (%) | Cum. TE %a | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | sHR b (95% CI), p-Value |
BIC c | AUC d (95% CI) | HR e (95% CI), p-Value f |
HR e (95% CI), p-Value f |
||
| Model 1 g,i | CHT | 60 (72) | 26.5 | 23 (28) | 0.0 | 100 (79–100) |
34 (23–47) |
27 (16–40) |
100 (85–100) |
All TE in high risk group | 127 | 0.67 (0.61–0.73) |
2.2 (1.2–4.2), 0.015 |
1.9 (1.2–3.2), p = 0.011 |
| Model 1 g,i | CHT/ RT | 60 (51) | 26.5 | 57 (49) | 3.1 | 100 (81–100) |
27 (19–37) |
19 (11–29) |
100 (87–100) |
All TE in high risk group | NR e | 0.64 (0.59–0.68) |
1.5 (0.9–2.3), 0.096 |
1.6 (1.1–2.4), p = 0.018 |
| Model 2 g,j | CHT | 56 (67) | 26.6 | 27 (33) | 3.7 | 94 (70–100) |
39 (27–52) |
27 (16–40) |
96 (81–100) |
8.2 (1.1–61.6), p = 0.04 |
135 | 0.66 (0.58–0.75) |
1.6 (0.9–2.9), 0.078 |
1.5 (0.9–2.4), p = 0.081 |
| Model 3 g,k | CHT | 49 (59) | 30.6 | 34 (41) | 2.9 | 94 (70–100) |
49 (37–62) |
31 (18–45) |
97 (85–100) |
12.3 (1.7–91.8), p = 0.01 |
131 | 0.72 (0.62–0.80) |
2.4 (1.4–4.2), 0.0001 |
2.2 (1.5–3.4), p = 0.002 |
| Khorana g,l | CHT | 20 (24) | 20.2 | 63 (76) | 18.9 | 25 (7–52) |
76 (64–86) |
20 (6–44) |
81 (69–90) |
1.1 (0.4–3.3), p = 0.89 |
143 | 0.51 (0.39–0.63) |
2.01 (1.2–3.4), 0.005 |
1.9 (1.2–3.2), p = 0.002 |
| PROTECHT g,m | CHT | 53 (64) | 20.9 | 30 (36) | 16.3 | 69 (41–89) |
37 (26–50) |
21 (11–34) |
83 (65–94) |
1.3 (0.5–3.7), p = 0.60 |
142 | 0.53 (0.40–0.66) |
1.8 (1.1–2.9), 0.013 |
1.9 (1.2–2.8), 0.002 |
| CONKO g,n | CHT | 40 (48) | 25.0 | 43 (52) | 13.8 | 63 (35–85) |
55 (43–67) |
25 (13–41) |
86 (72–95) |
1.9 (0.7–5.3), p = 0.71 |
141 | 0.59 (0.45–0.73) |
2.6 (1.7–4.2), <0.001 |
2.8 (1.9–4.2), <0.001 |
| CATS h,o | CHT | - | - | - | - | 64 (NR) | 82 (NR) | 20 (NR) | 97 (NR) | - | - | - | - | - |
(a) Estimated TE incidence at six months from cumulative incidence function for TE with death as a competing risk; (b) Univariable sHR; (c) lowest BIC reflects most efficient model; (d) Area under ROC curve; (e) Stage-adjusted (stage IIIB/IV vs. stage I-IIIA) Cox proportional hazards regression; (f) p-value for mortality and progression among patients with high vs. low TE risk; (g) Model applied to NSCLC study population; (h) Model applied to mixed cancer population (published data); (i) Model 1: high risk if: (i) receiving chemotherapy and ((ii) baseline d-dimer ≥ 1.5 mg/L or (iii) month one d-dimer ≥ 1.5 mg/L or (iv) baseline fibrinogen ≥ 4 g/L and d-dimer ≥ 0.5 mg/L); (j) Model 2: high risk if: (i) and ((ii) or (iv)); (k) Model 3: high risk if: (i) and ((ii) or (iii) or (iv)) and ECOG PS ≥ 2; (l) Khorana: 1 point each for lung cancer, BMI ≥ 35 kg/m2, haemoglobin < 100 g/L, white cell count >11 × 109/L and platelet count >350 × 109/L (baseline biomarkers); (m) PROTECHT: as for Khorana plus 1 point each for gemcitabine chemotherapy and platinum-based chemotherapy; (n) CONKO: as for Khorana excluding BMI and adding 1 point for ECOG PS ≥ 2; (o) CATS: as for Khorana plus 1 point each for sP-selectin ≥ 53.1 ng/mL, D-dimer ≥ 1.44 µg/mL (baseline). BIC, Bayesian Information Criterion; NPV, negative predictive value; NR, not reported; PPV, positive predictive value; ROC, receiver operating characteristic; sHR, sub-distribution hazard ratio.