Table 1.

Summary of main findings in longitudinal studies

	Study	Paired samples (n)	Experimental methods	Main findings
Selected markers	Stark et al. 2003 [3]	27	IHC	- MLH1, MSH2, TP53 expression significantly lower in recurrent tumors
	Shinsato et al. 2013 [33]	11	IHC	- MLH1, PMS2 protein levels are reduced in TMZ-resistant cells - MLH1 induced PMS2 protein instability confers TMZ resistance to GBM cells
Genomic and transcriptomic analyses	Kim J. et al. 2015 [4]	38	WES, RNA-seq	- Distally recurred tumors, in contrast to locally recurred ones, share only a minority of initial tumor mutations - Primary GBMs rarely develop hypermutation after TMZ treatment
	Wang J. et al. 2016 [19]	114	WES, Transcriptome analysis	- 2/3 of primary GBMs switched subtype upon recurrence - Hypermutation preferentially targets highly expressed genes - Novel mutation in LTBP4 found in 11% of recurrent tumors - TGF-β pathway is a potential therapeutic target
	Martinez et al. 2009 [7]	20	Semiquantitative PCR, LOH analysis	- Initial tumors without P53 and EGFR mutations acquired new EGFR amplification upon recurrence - Recurrences display two distinct patterns depending on the profile of the original tumor
	Sottoriva et al. 2018 [43]	11	WES, TES	- Multi-regional WES revealed extensive ITH involving EGRF amplification and the loss of chromosome 10 containing PTEN and CDKN2A
	Kim H. et al. 2019	23	WGS, WES, PyClone clustering	- Mutation clustering seen as clonal (67.9%), subclonal (29.8%) - 90% of TP53 and PIK3CA/PIK3R1 mutations are clonal - TP53 mutational status has influence on clonal tumor progression
	Muscat et al. 2018 [21]	21	WGS, WES, Targeted deep sequencing	- Variant burden reduced in recurrent tumors - Neutral tumor evolution in untreated GBM shifted towards non-neutral evolutionary dynamics in recurrent GBM after treatment - Increased mutation rate occurred in one recurrent tumor, attributable to TZM-induced hypermutator phenotype
	Wang Q. et al. 2017 [10]	8	Transcriptome profiling	- Macrophage/microglia-rich microenvironment shapes the MES phenotype - NF1 deactivation results in macrophage/microglia attraction - Gene-expression subtype retained in 55% of the cases
Epigenomic analyses	Hegi et al. 2005 [22]	206	Methylation-specific polymerase-chain-reaction	- MGMT promoter was methylated in 45% of cases - MGMT promoter methylation was an independent favorable prognostic factor - MGMT promoter methylation results in survival benefit
	de Souza et al. 2018 [24]	77	Comprehensive DNA-methylation analysis	- Classification of diffuse IDH mutant and IDH-wt gliomas - G-CIMP-high subgroup identified with worst prognosis, and the capability to recur as a more aggressive tumor - Predictive biomarkers for assessing the risk of recurrence identified
	Klughammer et al. 2018 [25]	112	RRBS, RNA-seq	- Optimized RRBS can be used to infer transcriptional subtypes - DNA methylation can be predictive of immune cell infiltration, the extent of necrosis and subcellular tumor cell morphology - Recurrent progression-associated demethylation of Wnt promoters in association with worse prognosis

The table highlights observations from analyses with selected markers and from genomic, transcriptomic and epigenomic studies on sequential glioblastoma specimens