Table 2.

Correlations between patient characteristics and the number of alterationsa

All alterations considered (including VUS)
No. of alterationsUnivariableMultivariableb
Variablesmedian (95% CI)P-valueB coefficientt-StatisticcP-value
Overall (N = 670)1 (1–1)
Gender
 Women (N = 336)1 (1–1)0.318
 Men (N = 334)1 (1–1)
Age at diagnosisd<0.0010.0341.1930.233
 ≤62 years old (N = 349)1 (0–1)
 >62 years old (N = 321)2 (1–2)
Tumor type
 Gastrointestinal vs. not (N = 213)2 (1–2) vs. 1 (1–1)0.011−0.030−0.8150.415
 Brain vs. not (N = 152)0 (0–0) vs. 2 (1–2)<0.001−0.084−2.3870.017
 Lung vs. not (N = 139)2 (2–3) vs. 1 (1–1)<0.001−0.012−0.3260.745
 Breast vs. not (N = 55)1 (1–2) vs. 1 (1–1)0.147
 Head and neck vs. not (N = 25)3 (1–6) vs. 1 (1–1)0.0010.0692.3590.019
Alterations detected
TP53 vs. not (N = 218)3 (3–4) vs. 0 (0–1)<0.0010.2688.868<0.001
EGFR vs. not (N = 87)4 (3–5) vs. 1 (1–1)<0.0010.2217.746<0.001
KRAS vs. not (N = 84)4 (3–5) vs. 1 (1–1)<0.0010.1665.494<0.001
PIK3CA vs. not (N = 61)5 (4–6) vs. 1 (1–1)<0.0010.2498.682<0.001
≥1 gene altered withe
 ctDNA ≥ 10% (N = 101)5 (4–6)<0.0010.2167.035<0.001
 ctDNA < 10% (N = 569)1 (1–1)
Characterized alterations considered (excluding VUS)
Overall (N = 670)0 (0–1)
Gender
 Women (N = 336)0.5 (0–1)0.619
 Men (N = 334)0 (0–1)
Age at diagnosisd0.0010.0150.6730.501
 ≤62 years old (N = 349)0 (0–0)
 >62 years old (N = 321)1 (0.5–1)
Tumor type
 Gastrointestinal vs. not (N = 213)1 (0–1) vs. 0 (0–0)0.001−0.004−0.1190.905
 Brain vs. not (N = 152)0 (0–0) vs. 1 (1–1)<0.001−0.037−1.1770.240
 Lung vs. not (N = 139)1 (1–2) vs. 0 (0–0)<0.0010.0180.5820.561
 Breast vs. not (N = 55)1 (1–1) vs. 0 (0–1)0.048−0.003−0.1110.911
 Head and neck vs. not (N = 25)2 (1–4) vs. 0 (0–1)<0.0010.0642.6680.008
Characterized alterations detected
TP53 vs. not (N = 206)2 (2–2) vs. 0 (0–0)<0.0010.35815.163<0.001
EGFR vs. not (N = 51)3 (2–4) vs. 0 (0–0)<0.0010.27912.267<0.001
KRAS vs. not (N = 82)3 (2–3) vs. 0 (0–0)<0.0010.24110.127<0.001
PIK3CA vs. not (N = 48)4 (3–5) vs. 0 (0–0)<0.0010.29913.350<0.001
≥1 gene altered withe
 ctDNA ≥ 10% (N = 92)3 (3–4)<0.0010.1837.473<0.001
 ctDNA < 10% (N = 578)0 (0–0)
  • aOnly tumor types with >25 patients were included; only the most prevalent genes were included (genes altered in >50 patients were included for the top table and >40 patients for the bottom table).

  • bVariables with P ≤ 0.05 in the univariable analysis were included in the multivariable model. P values were calculated using the Mann–Whitney tests for univariable analyses and a linear regression model was computed for the multivariable analysis.

  • cCut-off of 62 years old corresponds to the median (Table 1).

  • dThe t-statistic is the ratio of the B coefficient and the SE; the higher the value, the greater is the importance of the variable in the model.

  • eThe cut-off of 10% corresponds to ten times the median of percentage of ctDNA detected in all the alterations.