Fig. 4

Effect of the number of ‘methylated’ CpG sites on the hazard of mortality: the patients’ ‘methylated’ CpG sites were plotted against the patients’ martingale residuals of the model without the variable “methylated CpG sites”. The red line shows the corresponding LOWESS (locally weighted scatterplot smoothing) curve and gives a hint for the true effect of the number of ‘methylated’ CpG sites on the hazard of mortality for glioblastoma patients with survival of at least three months and (a) at least one (n = 170) or (b) no (n = 33) adjuvant TMZ cycle after XRT/TMZ