In this study, we provide comprehensive DNA methylation profiles of UTUC and UCB. Overall, UTUC and UCB showed very similar DNA methylation landscapes. Considering the smaller UCB sample size, we further validated using public cohorts, and the results were as expected (Additional file
1: Fig. S7A). There seems to be a common epigenetic mechanism of positive selection in urothelial carcinogenesis, indicating a potential opportunity for similar management strategies for UC. Of note, by supervised analysis, we also identified differentially methylated regions between UTUC and UCB. Promoter hypermethylated genes in UTUC were enriched in the signaling pathways (Additional file
1: Fig. S7B) that were crucial for cancer development, such as regulation of tumor necrosis factor production (TNF) and T cell activation. Consistently, it is reported that the genes of the TNF pathway were lower expressed in UTUC compared to UCB [
25] and the majority of UTUC has a T cell depleted immune contexture [
26]. We also found that promoter regions of some EMT-related genes were more hypomethylated in UTUC, such as
ZEB1 and
SLC38A1 (Additional file
1: Fig. S7C&D)
. These results may partially explain the aggressive clinical behavior and a more advanced presentation of UTUC. Several previous publications have established a prognostic influence of methylation in UC [
27,
28], and we found that the methylation levels of some CpG sites were consistent in high-risk patients and in the Methy-High subgroup of UC patients (Additional file
1: Fig. S7E). Notably, with a sufficient number of samples, we identified that DNA methylation subtypes were an independent prognostic marker for muscle-invasive UCB patients (Fig.
3H). The prognostic value of pathological factors for UCs has been discussed elsewhere, and there is a high prevalence of patients with TaT1 due to their often long-term survival and lower risk of cancer-specific mortality compared to patients with T2-4 tumors [
29]. More research is needed to determine the role of DNA methylation classification in improving the predictive accuracy of currently available risk tables [
30,
31]. Further prospective studies including a larger number of UC patients will be required to validate the prognostic capacity of DNA methylation subtype classification.
Mechanistically, we identified that Methy-High UCs have a predominantly basal expression pattern with high immune and stromal scores and are linked to poor prognosis. This classification, if validated, might be used in the future to predict outcomes of resected UC patients and help identify those who may benefit from chemotherapy. Cisplatin-based combination chemotherapy is the standard-of-care therapy for UC, both as radical perioperative treatment and as palliative first-line treatment for advanced disease [
32‐
34]. Notably, preclinical data indicate that DNMTis will circumvent cisplatin resistance in various cancers, including UCs [
35,
36]. Recently, phase I trials of SGI-110 combined with cisplatin and gemcitabine for solid malignancies, including urothelial carcinoma (SPIRE), have been performed, and the addition of SGI-110 to gemcitabine and cisplatin was tolerable, despite some additional myelosuppression, and such combinations warrant further investigation to assess efficacy [
37].
More importantly, we identified that Methy-High UCs had a T cell inflamed and immunosuppressive environment, showing therapeutically efficacy to SGI-110 through enhancing antitumor immunity. These findings were consistent with recent investigations suggesting that upregulation of CTAs, activation of endogenous retroviral elements (ERVs), and an antiviral defense response contribute to the clinical efficacy of DNMTis [
38,
39]. The 2021 updated European Association of Urology guidelines recommend that metastatic UC patients positive for programmed death ligand 1 (PD-L1) and ineligible for cisplatin receive immunotherapy (atezolizumab or pembrolizumab) [
40]. Since the DNA methylation profile is cancer-specific and varies greatly between cancer types, the antitumor activities of DNMTis may be specific for individual cancer types. Our integrative analysis successfully linked the antitumor effects of SGI-110 to detailed epigenetic alterations in UC cells, identified potential therapeutic targets, and provided a rationale for SGI-110 combination with immune checkpoint therapies. Further confirmation of our findings in clinical trials of combination therapy is warranted, with the goal of future implementation in clinical practice.