*3.3. Correlation between Baseline CONUT Score and Response to TKI Treatment*

In order to evaluate a possible correlation between nutritional status and the response to first TKI treatment, patients were stratified into three groups according to CONUT score value. We observed that patients with a CONUT score 0–2 (Group A) had a significantly better PFS and OS than patients with a CONUT score 3–4 (Group B) (*p* = 0.001 for PFS and *p* = 0.006 for OS) and with a CONUT score 5–7 (Group C) (*p* = 0.02 for PFS, *p* = 0.002 for OS). Accordingly, we also observed that patients with a CONUT score 0–2 had a significantly better response, according to RECIST criteria, than the other two groups (*p* = 0.002 and *p* = 0.003, respectively). In contrast, PFS, OS, and radiological response were not significantly different between patients with a CONUT score 3–4 (Group B) and 5–6 (Group C) (Table 2).

**Table 2.** Response to TKIs according to baseline CONUT score (Group A score 0–2, Group B score 3–4, Group C score 5–7).


<sup>1</sup> BR and PFS were assessed in 40 patients. <sup>2</sup> OS was evaluated in 32 patients treated with only one TKI.

*3.4. ROC Analysis for 12 Months—PFS and OS*

By ROC curves analysis, we found that the best CONUT score cut-off able to predict the response to TKI treatment was 3 (Table S1). This cut-off had a specificity of 100% and 93.7% and a sensitivity of 70% and 68.8% for PFS and OS, respectively, with an Area Under the Curve (AUC) of 0.980 for mPFS, (*p* < 0.0001) and of 0.846 for mOS (*p* = 0.0001), as showed in Figure 1.

**Figure 1.** ROC curve analysis for 12-month progression free survival (PFS) (**a**) and 12-month overall survival (OS) (**b**).
