Chapter 3 100 PD-1T TIL numbers and TIS scores for the training cohort are presented in Figure S3G and H. A cut-off value of 90 PD-1T TILs per mm2 and a TIS score of 6.65 demonstrated similar predictive accuracy as observed in the training cohort for DC 6m (Fig. 3C,D and Table 3). In the validation cohort, PD-1T TIL numbers were significantly higher in patients with DC 12m versus PD (P<0.001) (Fig. 3E). PD-1T TILs maintained a consistently high AUC of 0.80 (95% CI: 0.65-0.94) and demonstrated good performance (sensitivity: 86%, specificity: 74%, NPV: 95% and PPV: 50%) (Fig. 3C and Table 3). A subgroup analysis revealed an enrichment of patients with DC 12m in the PD-1T TILs ≥90 group and patients with PD in the PD-1T TILs <90 subgroup (Fig. 3E). TIS scores did not significantly differ between the two groups (P=0.31) and showed a low AUC of 0.63 (95% CI 0.43-0.82) (Fig. 3D,F). Nevertheless, a cut-off score of 6.65 reached a sensitivity of 100%, specificity of 39%, NPV of 100% and PPV of 26% (Table 3). Although these findings did not meet the prespecified ≥50% specificity criterium, they accurately identified all patients with DC 12m including 39% of patients with PD. Taken together, PD-1T TILs and TIS, as individual biomarkers, showed superior predictive accuracy for DC 12m compared to CD8+IT-CD8 and CD8+TIS. Notably, PD-1T TILs alone demonstrated greater performance than TIS alone, particularly in terms of specificity and PPV. Figure 3. Performance of selected composite and individual biomarkers to predict DC at 12 months in NSCLC patients treated with PD-1 blockade. (A) Receiver operating characteristic (ROC) curve for predictive value of CD8+IT-CD8 for disease control at 12 months (DC 12m) in the training cohort (n=55) (AUC 0.85; 95% CI: 0.73-0.96) and validation cohort (n=77) (AUC 0.67; 95% CI: 0.53-0.81). (B) ROC curve for predictive value of CD8+TIS for DC 12m in the training cohort (n=28) (AUC 0.91; 95% CI: 0.79-1.00) and validation cohort (n=38) (AUC 0.59; 95% CI: 0.360.82). (C) ROC curve for predictive value of PD-1T TILs for DC 12m in the training cohort (n=42) (AUC 0.82; 95% CI: 0.70-0.94) and validation cohort (n=61) (AUC 0.80; 95% CI: 0.65-0.94). (D) ROC curve for predictive value of TIS for DC 12m in the training cohort (n=28) (AUC 0.77; 95% CI: 0.58-0.96) and validation cohort (n=40) (AUC 0.63; 95% CI: 0.43-0.82). (E) PD-1T TILs per mm2 in pretreatment samples from patients with DC 12m (n=14) and PD (n=47) in the validation cohort (n=61). Dashed line indicates a cut-off of 90 PD-1T TILs per mm2. Medians, interquartile ranges and minimum/maximum shown in boxplots, **P<0.01 by Mann Whitney U-test. (F) TIS scores in pretreatment samples from patients with DC 12m (n=7) and PD (n=33) in the validation cohort (n=40). Dashed line indicates a cut-off score of 6.65. Medians, interquartile ranges and minimum/maximum shown in boxplots, P=0.31 by Mann Whitney U-test. ▶
RkJQdWJsaXNoZXIy MTk4NDMw