Extended Data Fig. 3: Associations between Factor 4 and demographic and clinical characteristics.
a, Association of F4 to age. P values are from two-sided Pearson’s correlation test. (n = 217 samples) b and c, Associations of F4 to sex and pretreatment status. P values are from two-sided t-tests. d and e, Kaplan-Meier plots for showing the associations between F4 and TTT or OS in patients without previous treatment. The P-values were assessed by Cox regression models with F4 as a continuous variable. For visualization purposes only, optimal cutoffs to separate patients into high and low CLL–PD groups were estimated by the maximally selected rank test implemented in the R/CRAN package maxstat (v0.7). f and g, Forest plots showing the hazard ratios with 95% confidence intervals and P values from multivariate Cox models that include known demographic and genomic risk factors, for TTT and OS in patients without previous treatment. F4 remained significantly associated with TTT in multivariate analysis. In multivariate analysis for OS, none of the risk factors except for age were significant, however, the hazard ratio showed the same trend for F4 as in the full dataset analysis, consistent with the reduced statistical power of the subset analysis. (n = 154 patients) h, Correlation between F4 and lymphocyte doubling time (LDT) in previously untreated patients. P values and coefficients are from two-sided Pearson’s correlation tests. i, Correlation between F4 and lymphocyte doubling time (LDT) in M/U-CLL separately. P values and coefficients are from two-sided Pearson’s correlation tests. (n = 43 and 40 samples for M-CLL and U-CLL, respectively).
