Table 4.
Multivariate LASSO analysis.
| Type | Item | All treatments | Oxaliplatin containing | PLAT/PYR | Target-mono | TOP/PYR | |
|---|---|---|---|---|---|---|---|
| Clinical | Prior Treatment | 0.57 | |||||
| Gender | 0.77 | −1.14 | −1.10 | ||||
| Counts | nr of Tandem Duplications | 1.19 | |||||
| nr of 10kb–1Mb deletions | 0.01 | ||||||
| Mutational Signatures | DBS2 | −0.02 | |||||
| DBS5 | 0.13 | ||||||
| DBS11 | −0.03 | ||||||
| SBS17b | 0.07 | 0.13 | |||||
| SBS39 | 0.04 | ||||||
| SBS41 | −0.21 | ||||||
| Driver Genes | APC | 0.23 | |||||
| KRAS | 0.78 | ||||||
| PIK3CA | 0.22 | ||||||
| FBXW7 | 17.65 | ||||||
| Non-coding | LINC00672 | 0.90 | |||||
| GISTIC Regions | Gain 17q12 | (ERBB2*) | 0.44 | ||||
| Gain 18p11.32 | (CETN1*) | 0.59 | |||||
| Gain 20q11.1 | (BCL2L1*) | 0.12 | |||||
| Gain 8p11.21 | (KAT6A) | −0.78 | −1.56 | ||||
| Gain 7p21.3 | (VWDE) | 3.68 | |||||
| Gain 7q31.2 | (MET*) | 3.59 | |||||
| Gain 7p12.3 | (PKD1L1) | 3.04 | |||||
| Gain 7q34 | (no genes in peak) | 3.59 | |||||
| Gain 14q23.1 | (no genes in peak) | 1.64 | |||||
| Loss 18q12.2 | (hsa-mir-924*) | −1.52 | −3.30 | 2.79 | |||
| Loss 6q26 | (PARK2) | −1.38 | |||||
| Loss 9p21.3 | (CDKN2A*) | −1.88 | |||||
| Loss 16q23.1 | (WWOX) | −1.70 | |||||
| Loss 4q22.1 | (CCSER1) | 2.66 | |||||
| Loss 4q35.1 | (IRF2) | 1.78 | |||||
| Loss 18q21.2 | (SMAD4) | −1.35 | −3.30 | 2.79 | |||
| Loss 18q23 | (NFATC1*) | −1.14 | −3.30 | ||||
| Loss Xp22.31 | (STS*) | −1.40 | 1.78 | ||||
| Loss 14q23.3 | (GPHN) | −2.09 | |||||
Items that reached univariate statistical significance (p < 0.05) were used in a multivariable penalized ordinal regression model for treatment response. Univariate regression was performed for genomic features (Supplementary Data 3) using the ‘polr’ function from the MASS R package (v7.3-51.4) and subsequently those with a univariate two-sided p-value <0.05 were selected for multivariable ordered LASSO regression using the ordinalNet R package (v2.7). Regression coefficients are shown for features that remained significant in the multivariable model.
*Multiple genes present in region; bold: known Fragile Site region