Deciphering the implications of mitophagy-related signatures in clinical outcomes and microenvironment heterogeneity of clear cell renal cell carcinoma

Abstract

Background

The role of mitophagy in various cancer-associated biological processes is well recognized. Nonetheless, the comprehensive implications of mitophagy in clear cell renal cell carcinoma (ccRCC) necessitate further exploration.

Methods

Based on the transcriptomic data encompassing 25 mitophagy-related genes (MRGs), we identified the distinct mitophage patterns in 763 ccRCC samples. Subsequently, a mitophage-related predictive signature with machine learning algorithms was constructed, designated as RiskScore, to quantify the individual mitophagy status in ccRCC patients. Employing multispectral immunofluorescence (mIF) and immunohistochemistry (IHC) staining, we detected the effect of PTEN-induced putative kinase 1 (PINK1) in the prognosis and immune microenvironment of ccRCC.

Results

Our analysis initially encompassed a comprehensive assessment of the expression profiling, genomic variations, and interactions among the 25 MRGs in ccRCC. Subsequently, the consensus clustering algorithm was applied to stratify ccRCC patients into three clusters with distinct prognostic outcomes, tumor microenvironment (TME) characteristics, and underlying biological pathways. We screened eight pivotal genes (CLIC4, PTPRB, SLC16A12, ENPP5, FLRT3, HRH2, PDK4, and SCD5) to construct a mitophagy-related predictive signature, which showed excellent prognostic value for ccRCC patients. Moreover, patient subgroups divided by the RiskScore showed contrasting expression levels of immune checkpoints (ICPs), abundance of immune cells, and immunotherapy response. Additionally, a nomogram was established with robust predictive power integrating the RiskScore and clinical features. Notably, we observed that PINK1 expression markedly correlated with favorable treatment response and advanced maturation stages of tertiary lymphoid structures, which potentially shed light on enhancing anti-tumor immunity of ccRCC.

Conclusion

Collectively, this study initially developed a signature associated with mitophagy, which demonstrated an excellent ability to predict the clinical prognosis, TME characterization, and responsiveness to targeted therapy and immunotherapy for ccRCC patients. Of particular note is the pivotal role of PINK1 in mediating the treatment response and immune microenvironment for ccRCC patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00432-023-05349-y.

Introduction

Renal cell carcinoma (RCC) is the most common type of kidney cancer, accounting for approximately 90% of cases worldwide. Its incidence varies globally, with higher rates reported in developed countries (Zheng et al. 2016; Siegel et al. 2023). It is estimated that there are around 403,000 new cases of RCC diagnosed worldwide each year. In the US, RCC represents approximately 2–3% of all adult malignancies (Siegel et al. 2023). The disease bears significant responsibility for worldwide cancer-related mortality, with roughly 175,000 annual deaths attributed to RCC (Braun et al. 2021).

Renal cell carcinoma has several histological subtypes, including clear cell renal cell carcinoma (ccRCC), papillary renal cell carcinoma, chromophobe renal cell carcinoma, and other rare subtypes (Li 2022a). Predominant among these is clear cell carcinoma, encompassing 70–80% of RCC occurrences. Treatment of ccRCC depends on various factors, including the clinical staging, pathological grading, and histological subtype of the tumor. Despite therapeutic advancements, resistance to therapy and disease relapse remain challenges in managing advanced ccRCC (Chevrier et al. 2017). Molecular biomarkers, such as von Hippel–Lindau (VHL) gene mutation status, loss of heterozygosity (LOH) in specific chromosomal regions, and gene expression levels (e.g., PBRM1, BAP1, and SETD2), have exhibited associations with patient outcomes and responsiveness to targeted treatments (Xu et al. 2022; Xu, et al. 2022a; Zaffuto et al. 2016). The integration of these molecular biomarkers into prognostic models holds potential for precise risk stratification and personalized therapeutic strategies (Peter et al. 2021; Senbabaoglu et al. 2016). Therefore, the identification of dependable biomarkers or models for early diagnosis, prognostic assessment, and treatment response evaluation constitutes a persistent challenge in RCC research. Individualized precision medicine could help improve diagnostic accuracy, guide treatment decisions, and monitor disease progression of ccRCC (Vesely et al. 2022; Xu et al. 2019a).

Mitophagy, which is the selective degradation of damaged or dysfunctional mitochondria, plays a pivotal role in maintaining cellular homeostasis and preventing the accumulation of defective mitochondria (Doblado et al. 2021). In the context of carcinoma, arising from epithelial cells, mitophagy's impact on tumor genesis and progression embodies dual facets of benefit and detriment (Poole and Macleod 2021). In its normative role, mitophagy serves as a tumor suppressive mechanism, eliminating compromised mitochondria that amass reactive oxygen species (ROS) and incite genomic instability, precipitating carcinogenesis (Panigrahi et al. 2020; Korolchuk et al. 2017). Through the removal of dysfunctional mitochondria, mitophagy preserves mitochondrial integrity, counteracting mutational buildup that might foment cancer emergence.

Nonetheless, excess or dysregulated mitophagy can promote tumor advancement, counteracting its initial tumor suppressive role. In some cases, enhanced mitophagy can contribute to the survival and growth of cancer cells by eliminating damaged mitochondria that trigger programmed or immunogenic cell death (Poole and Macleod 2021; Xu 2023). This allows cancer cells to evade cell death signals and proliferate uncontrollably. In cancer cells, mitophagy can also support metabolic adaptation by selectively removing damaged mitochondria that produce less ATP and replacing them with functional mitochondria that enhance energy production. This metabolic rewiring empowers cancer cells to meet augmented energy demands and support tumor growth. Some studies suggest mitophagy's involvement in conferring resistance to therapies in carcinoma (Luo 2023). Cancer cells with elevated mitophagy level may be better equipped to survive harsh conditions, such as nutrient deprivation or exposure to chemotherapy and radiation (Poole and Macleod 2021). By eliminating damaged mitochondria and maintaining mitochondrial function, cancer cells augment survival and withstand the impacts of anti-cancer interventions.

RCC is characterized by hypoxic conditions resulting from aberrant tumor vascularization (Xu et al. 2021a; Qu et al. 2022). Within a hypoxic milieu, low oxygen concentrations stimulate mitophagy as an adaptive response to maintain mitochondrial quality and function, and mitophagy can help to preserve cellular energy and promote cell survival (Gilkes et al. 2014). However, excessive or prolonged mitophagy can induce mitochondrial dysfunction and promote tumor aggressiveness. In this study, we aim to explore the effect of mitophagy on ccRCC malignancy in a large multi-ethnic population and establish a novel approach for facilitating individualized ccRCC diagnosis and treatment through the integration of multi-omics data and machine learning algorithms.

Materials and methods

Patients collection and data normalization

We utilized the publicly available transcriptome data and clinical information of 763 ccRCC patients from TCGA database (https://portal.gdc.cancer.gov/repository), CPTAC database (https://cptac-data-portal.georgetown.edu/studysummary/S050), ICGC database (https://icgc.org/), and EMBL database (https://www.ebi.ac.uk/arrayexpress/), including TCGA-KIRC, CPTAC-3, ICGC RECA-EU, and E-MTAB-3267 cohorts (Clark et al. 2019; Marquardt et al. 2021). The expression levels of the RNA-seq samples were transformed from fragments per kilobase of transcript per million mapped reads (FPKM) to transcripts per million (TPM), and log2(TPM + 1) was taken. Batch effects were corrected using the combat algorithm of the “sva” package in R software (Leek et al. 2012). In addition, copy number variations (CNVs) files, somatic mutation data, and tumor mutation burden (TMB) data were obtained from the TCGA database. The detailed baseline clinical data of ccRCC patients are summarized in Table S1.

This study included 94 Chinese patients with advanced ccRCC, who underwent first-line sunitinib at the Department of Urology of Fudan University Shanghai Cancer Center, during the period from January 2008 to December 2019. Pathological examinations identified tumor and tumor-adjacent tissue regions, with clinicopathological characteristics documented in Table S2.

Collection of mitophagy-related genes (MRGs)

MRGs set was required from the Molecular Signatures Database (MSigDB; https://www.gsea-msigdb.org/gsea/msigdb/human/search.jsp). We selected “REACTOME_MITOPHAGY” gene set and intersected the same genes between the gene set and the transcriptome data of above integrated cohort for further analyses (Table S2).

Unsupervised clustering analysis

Unsupervised clustering was executed to stratify patients into distinct molecular subtypes with the “ConsensusClusterPlus” package in R software (Hu et al. 2022). The consensus matrix, cumulative distribution function (CDF), and relative change in area under the CDF curve determined the optimal cluster number.

Exploration of the immune landscape in distinct subgroups

The CIBERSORT algorithm quantified immune cell proportions within the tumor microenvironment (TME) of each ccRCC sample (Hu et al. 2022). Expression levels of immune checkpoints (ICPs) were evaluated and contrasted among distinct clusters. Additionally, stromal and immune compositions in ccRCC samples were determined using the ESTIMATE algorithm (Xu et al. 2019b).

Identificationof differential expression genes (DEGs) and functional enrichment analysis

DEGs amid diverse mitophagy-related clusters were identified through the “limma” package in R software, applying criteria of adjusted P value < 0.05 and |log2-fold change (FC)|> 0.8. Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were carried out via the “clusterProfiler” package in R software, with thresholds of p value and q value < 0.05. Gene set variation analysis (GSVA) using the KEGG gene set (c2.cp.kegg.v7.5.1) was executed across clusters, employing criteria of |log2-fold change (FC)|> 0.15 and adjusted P value < 0.001 (Xu et al. 2021b; Hanzelmann et al. 2013).

Construction and validation of a mitophagy-related predictive signature

Univariate Cox regression analysis with the DEGs identified the prognosis-related genes. Statistically significant DEGs (P < 0.001) were then used for the subsequent least absolute shrinkage and selector operation (LASSO) analysis and multivariate Cox analysis to construct the predictive signature (RiskScore) (Xu et al. 2019b). The RiskScore was computed with the RiskGenes as described: RiskScore = h₀(t)* exp (∑expression of RiskGene* corresponding coefficient). Patients were stratified into low or high RiskGroup based on the median RiskScore.

Prognostic significance of the signature was assessed with Kaplan–Meier analysis, and stratified survival analyses for the patients in different clinical subgroups were conducted. Tumor immune dysfunction and exclusion (TIDE) analysis (http://tide.dfci.harvard.edu/) predicted patients' immunotherapy response, with lower TIDE scores indicating better response (Xu et al. 2021c; Jiang et al. 2018). Validation of immunotherapy response using the mitophagy-related model was performed on the immune checkpoint inhibitors (ICIs) received David Liu cohort, comprising 121 metastatic melanoma patients treated with nivolumab or pembrolizumab.

Development and validation of a nomogram

A nomogram was constructed combining the RiskScore and clinical traits to predict the 1-, 3-, and 5-year overall survival (OS) for ccRCC patients using the “rms,” “regplot,” and “survival” packages in R software. The reliability of the nomogram was assessed through ROC and calibration curves.

Immunohistochemistry (IHC) assays

Histopathological assessment of PINK1 expression on ccRCC slides followed established protocols, with independent evaluation by experienced pathologists. The slides of ccRCC tissue or adjacent normal kidney tissues were selected for staining, and the slides were placed on a baking machine at 56 ℃ for about 1 h. Immunohistochemical staining goes through a series of steps, including dewaxing, removal of endogenous peroxidase, antigen repair, washing, sealing, primary antibody incubation, secondary antibody incubation, stained with diaminobenzidine solution (DAB), dehydration, neutral gum sealing, and photography. The primary PINK1 polyclonal antibody (PA5-13402, Invitrogen, USA) applied to the tissue section and incubated overnight or for a designated period at an appropriate temperature. The specific implementation methods are as previously described (Zheng et al. 2021).

Multispectral immunofluorescence (mIF) staining assays

Tissue slides were prepared with primary and secondary antibodies, while excluding fluorophores, to establish negative controls for autofluorescence evaluation. Multiplex stained slides were scanned using a Vectra Polaris Quantitative Pathology Imaging System (Akoya Biosciences) at 20-nm wavelength intervals from 440 to 780 nm with a fixed exposure time and an absolute magnification of ✕200. All scans for each slide were then superimposed to obtain a single image. Multilayer images were imported to inForm v.2.4.8 (Akoya Biosciences) for quantitative image analysis. For the analysis of ccRCC immune microenvironment, specifically the context of tertiary lymphoid structures (TLS) maturation, pathological slides were costained for CD20, CD23, CD3, CD68, CDC163, pan-CK, and DAPI.

Statistical analysis

All statistical analyses and graph visualizations were implemented in R v4.1.3. The cutoff value was defined via “survminer” R package or median threshold. Statistical significance was set at p < 0.05 (two sides) unless otherwise indicated.

Results

Differential expression and genomic mutations of mitophagy-related genes (MRGs) in ccRCC

In our study, we analyzed the mRNA expression levels of 25 MRGs across ccRCC and normal tissues sourced from TCGA database (Fig. 1A). A total of 22 MRGs were differentially expressed at mRNA level, among which 14 MRGs were upregulated and eight MGRs were downregulated in ccRCC. Only seven in 336 ccRCC samples had mutations of the 25 MGRs, and UBC and ULK1 had the highest mutations frequencies (Fig. 1B). Additionally, we ascertained a profile of CNVs associated with the 25 MRGs, revealing noteworthy amplification tendencies in genes such as SQSTM1, VDAC1, ATG12, and MFN1. Conversely, genes such as TOMM20, ATG5, MFN2, PRKN, and PINK1 exhibited a greater frequency of deletion within CNVs (Fig. 1C). The chromosomal distribution of these 25 MRGs is displayed in Fig. 1D. Besides, an encompassing visualization of the interactions, correlations, and prognostic implications of the 25 MRGs is depicted in Fig. 1E.

Fig. 1.

Landscape of expression and genetic variations of MRGs in ccRCC. A Expression levels of MRGs in tumor and adjacent tissues of ccRCC patients from the TCGA-KIRC cohort. B Mutation frequency and types of MRGs in ccRCC from the TCGA-KIRC cohort. C CNV frequency of MRGs in ccRCC from the TCGA-KIRC cohort. D Location of CNV alteration of MRGs on chromosomes in ccRCC from the TCGA-KIRC cohort. E The interactions of MRGs in ccRCC from the TCGA-KIRC cohort

Identification of mitophagy-related clusters and characterization of distinct clusters

To further explore the role of mitophagy in ccRCC, we applied the unsupervised clustering algorithm to stratify ccRCC patients based on the mRNA expressions of the 25 MRGs. The optimal number of the identified clusters was 3 (370 patients in MitophagyCluster A, 305 patients in MitophagyCluster B, and 88 patients in MitophagyCluster C) (Fig. 2A and Table S3). The result was defined by the least crossover of the consensus matrix, the smooth trend of the CDF, and no significant shift of the area under the curve (Figure S2). Principal component analysis (PCA) further corroborated the discernible divergence among the three MitophagyClusters (Fig. 2B). Survival analysis revealed that patients in MitophagyCluster A had best prognosis (Fig. 2C). The varying clinicopathological features and expression profiles of the 25 MRGs among MitophagyClusters were captured within a heatmap, with patients in MitophagyCluster A displaying heightened MRG expression levels (Fig. 2D). Functional enrichment analysis of GSVA results unveiled distinct pathway enrichments in MitophagyClusters A, B, and C, elucidating diverse biological contexts (Fig. 2E–G). These pathways encompassed mTOR signaling, ubiquitin-mediated proteolysis, renal cell carcinoma, aminoacyl tRNA biosynthesis within MitophagyCluster A, and underscoring the intricate interplay of molecular processes. Moreover, an in-depth exploration of the TME features, particularly the immune landscape, was undertaken across the different MitophagyClusters. It emerged that MitophagyCluster A exhibited elevated stromal scores alongside reduced immune scores (Fig. 2H and Table S4). Patients in MitophagyCluster A had lowest expression of CTLA4, PDCD1, TMIGD2, CD70, and TNFRSF18 and highest expression of CD274, TNFSF18, CD200, CD40, PDCD1LG2, NRP1, HAVCR2, CD160, TNFSF4, ADORA2A, TNFRSF14, and VTCN1 (Fig. 2I). The CIBERSORT analysis showed that the abundance of immune cells was varied among distinct MitophagyClusters (Fig. 2J and Table S5). Collectively, the MitophagyClusters unveiled distinct immunophenotypes with pronounced implications for the immune microenvironment.

Fig. 2.

Clinical and TME characteristics of mitophagy-related clusters. A Consensus matrix heatmap of unsupervised clustering analysis. B PCA analysis of MRGs expression difference among three MitophagyClusters. C Survival analysis of three MitophagyClusters. D Heatmap of clinicopathological features and expressions of MRGs among MitophagyClusters. E–G GSVA of enriching biological pathways among three MitophagyClusters. H The differences in estimate, stromal, and immune scores of MitophagyClusters. I The differences in expressions of ICPs among three MitophagyClusters. J The difference in infiltrating abundance of immune cells among three MitophagyClusters

Functional enrichment and identification of gene subgroups based on the DEGs

To further explore the biological characteristics of the MitophagyClusters, we identified 112 MitophagyClusters-related DEGs through the “limma” package (Table S6). Elucidating the interconnectivity among these DEGs, a protein–protein interaction (PPI) network was constructed utilizing the STRING database (Fig. 3A). Subsequent GO enrichment analysis highlighted enrichment within biological processes such as vascular process in circulatory system, small-molecule catabolic process, apical part of cell, apical plasma membrane, anion transmembrane transporter activity, and growth factor binding, etc. (Fig. 3B). Following this, uniCox analysis unveiled a pool of 110 prognosis-related DEGs, offering a comprehensive perspective on the clinical relevance of these genes.

Fig. 3.

Functional enrichment analysis and identification of GeneClusters. A PPI network of the DEGs from STRING database. B GO enrichment analysis of DEGs between two GeneClusters. C Consensus matrix heatmap of unsupervised clustering analysis. D PCA analysis of expression difference in prognosis-related genes between two GeneClusters. E Survival analysis of two GeneClusters. F Heatmap of clinicopathological features and expression of prognosis-related genes between two GeneClusters. E–G GSVA of enriching biological pathways between two GeneClusters. H The differences in estimate, stromal, and immune scores of GeneClusters. I The differences in expression of ICPs between two GeneClusters. J The difference in infiltrating abundance of immune cells between two GeneClusters

Based on the 110 prognosis-related DEGs, we conducted the unsupervised clustering algorithm again to stratify the patients into different subgroups, termed as GeneCluster (Fig. 3C and Table S3). We displayed the CDF plot, delta plot, and tracking plot in Figure S3. PCA analysis underscored substantial transcriptome distinctions between GeneCluster A and GeneCluster B (Fig. 3D). Furthermore, superior clinical prognosis was demonstrated among patients situated within GeneCluster A (Fig. 3E). We found that patients in GeneCluster A were tended to be stagesI–II, T1–T2, and in MitophagyCluster A (Fig. 3F). Pathway enrichment analysis through GSVA illuminated an array of enriched pathways, serving to delineate the functional contexts of GeneCluster A and GeneCluster B (Fig. 3G). Furthermore, we analyzed the immune TME features between the GeneClusters A and B. We found that patients in GeneCluster A had lower immune score, implying lower abundance of immune fraction (Fig. 3H). Besides, we compared the expressions of ICPs and abundance of infiltrating immune cells between the two GeneClusters. Patients in GeneCluster A had low expressed of CD44 and TNFRSF18 and high expressed of CD86, CD274, TNFSF18, CD200, CD40LG, CD40, PDCD1LG2, TNFRSF4, NRP1, HAVCR2, CD160, CD48, TNFSF4, etc. (Fig. 3I). Patients in GeneCluster B showed higher infiltrating level of T cells CD4 memory activated, Tregs, macrophages M0, and neutrophils and lower abundance of T cells gamma delta, monocytes, macrophages M1, and macrophages M2 (Fig. 3J).

Development of the mitophagy-related prognostic signature

To comprehensively quantify the mitophagy-related pattern in ccRCC, we applied multivariate Cox and LASSO analyses to construct a prognostic predictive signature, termed as RiskScore (Fig. 4A, B and Table S7). We identified eight risk genes with most prognostic significance, including CLIC4, PTPRB, SLC16A12, ENPP5, FLRT3, HRH2, PDK4, and SCD5. The patients were divided into high RiskGroup and low RiskGroup by the median RiskScore (Table S3). Survival analysis revealed that patients in high RiskGroup suffered worse prognosis (Fig. 4C). The AUCs of 1, 3, and 5 years of ROC curves were 0.733, 0.688, and 0.705, implying excellent predictive power of the model (Fig. 4D). The distribution plot showed that as the RiskScore increased, the mortality rate of ccRCC patients went up (Fig. 4E). Figure 4F displays the distribution of patients across MitophagyCluster, GeneCluster, and RiskGroup. We found that patients in MitophagyCluster A and GeneCluster A had lower RiskScore, consistent with the improved clinical prognosis (Fig. 4 G-H). To further evaluate the predictive ability of the prognostic signature, we stratified patients into two subgroups according to different clinical characteristics and analyzed the difference in prognosis between the two subgroups. We found that male patients with advanced T stage, TNM stage, and grade had higher RiskScore, and no significant difference was found for age (Figs. 5A–F and S4A-D). Besides, across all subgroups except stagesI–II, patients in the high RiskGroup consistently exhibited diminished prognoses (Figs. 5G–L and S4E-H).

Fig. 4.

Construction of the mitophagy-related predictive signature. A-B Coefficient profiles of the prognosis-related genes and identification of the best parameter in LASSO analysis. C Kaplan–Meier analysis of the OS between two RiskGroups. D ROCs analysis of RiskScore in predicting the 1-, 3-, and 5-year OS. E Distribution plot of RiskScore, OS status, and expression of risk genes. F Alluvial diagram of patients distributions in MitophagyClusters, GeneClusters, and RiskGroups. G Differences in RiskScore among three MitophagyClusters. H Differences in RiskScore between two GeneClusters

Fig. 5.

Correlation between RiskGroup and clinical characteristics. A-F Proportion of clinical features (T stage, TNM stage, and grade) in low and high RiskGroup. G-L Survival analysis of ccRCC patients between two RiskGroups in different subgroups stratified by clinical features (T stage, TNM stage, and grade)

Construction and validation of a prognostic nomogram

Subsequent to integrating the RiskScore alongside salient clinical attributes, an insightful nomogram was developed to predict the 1-, 3-, and 5-year overall survival (OS) for ccRCC patients. We found that age, grade, TNM stage, and RiskGroup were significant predictive features of the nomogram (Fig. 6A). The calibration plot effectively demonstrated an excellent alignment between actual OS and the nomogram-predicted OS for ccRCC patients (Fig. 6B). Moreover, the predictive potency of the nomogram was further assessed through the utilization of ROC curves. Impressively, the respective AUCs for 1-, 3-, and 5-year OS were 0.848, 0.799, and 0.764, thereby underscoring the nomogram's robust predictive efficacy (Fig. 6C–E).

Fig. 6.

Establishment and validation of a nomogram. A The nomogram for predicting the 1-, 3-, and 5-year OS based on the RiskGroup and clinicopathological features. B Calibration curves for validation of the nomogram. C-E ROCs of the nomogram for predicting the 1-, 3-, and 5-year OS

Exploration the immune microenvironment and immunotherapy response of the riskgroups

We first analyzed the abundance of infiltrating immune cells between the high and low RiskGroups. This analysis illuminated the heightened infiltration of Tregs, macrophages M0, and T cells CD4 memory activated within the high RiskGroup, demonstrating a positive correlation with RiskScore. Conversely, mast cells resting, macrophages M1, T cells gamma delta, and macrophages M2 exhibited reduced infiltration within the high RiskGroup, correlating negatively with the RiskScore (Fig. 7A, B). Considering the significance of ICIs in the clinical treatment for ccRCC, we compared the difference in expression levels of ICPs between the two RiskGroups. TNFRSF18 and CD44 were high expressed in high RiskGroup and positively correlated with the RiskScore (Fig. 7C, D). In contrast, NRP1, ADORA2A, CD200, CD274, and PDCD1LG2 exhibited reduced expression within the high RiskGroup, coupled with a negative correlation to RiskScore (Fig. 7C, D). Employing the ESTIMATE algorithm, we computed stromal, immune, and estimate scores, which indicated higher immune scores within the high RiskGroup, indicative of elevated immune fractions (Fig. 7E). Subsequently, we conducted TIDE algorithm to assess the immunotherapy response of the RiskScore. Patients in low RiskGroup had lower TIDE score, indicative of heightened sensitivity to immunotherapy (Fig. 7F). To rigorously evaluate the predictive utility of the RiskScore within the context of immunotherapy response, external validation was performed within the David Liu cohort. This validation corroborated that the high RiskGroup was associated with poorer prognosis and a diminished rate of complete or partial response (CR/PR), thereby affirming the predictive power of the RiskScore in immunotherapy response (Fig. 7G, H).

Fig. 7.

Evaluation of immune TME and immunotherapy response of RiskScore. A The difference in infiltrating abundance of immune cells between two RiskGroups. B The correlation between RiskScore and infiltrating abundance of immune cells. C The differences in expression of ICPs between two RiskGroups. D The correlation between RiskScore and expression of ICPs. E The differences in estimate, stromal, and immune scores of RiskGroups. F The differences in TIDE score between two RiskGroups. G Survival analysis of patients stratified by RiskScore in David Liu cohort. H The differences in immunotherapy response between two RiskGroups in David Liu cohort

Role of PTEN-induced putative kinase 1 (PINK1) in clinical prognosis and immune microenvironment of ccRCC

PTEN-induced putative kinase 1 (PINK1)-mediated mitophagy is a specific pathway involved in the selective removal of damaged or dysfunctional mitochondria. While PINK1-mediated mitophagy has been extensively studied in the context of neurodegenerative diseases, such as Parkinson's disease. However, its role within the realm of cancer, including ccRCC, remains an area of ongoing elucidation. In light of this, we endeavored to elucidate the clinical ramifications of PINK1-mediated mitophagy within ccRCC, offering potential insights into therapeutic strategies. Thus, we assessed expression of PINK1 in ccRCC slides and displayed the images in both high- and low-magnification fields (Fig. 8A). To evaluate the maturation dynamics of tertiary lymphoid structures (TLS), co-staining of slides for CD3, CD20, CD21, and CD23 was performed (Helmink et al. 2020; Horeweg et al. 2022; Xu et al. 2022b). CD68 and CD163 signals were also detected on the relationship between PINK1 expression and M2 macrophages infiltration. All dense lymphocytic aggregates irrespective of CD45/CD20/CD21/CD23 signals were imaged as multispectral high-power field (HPF) on the Vectra 3.0 imaging system (PerkinElmer). Each HPF image was assessed for the TLS maturation stage as follows: early TLS (E-TLS), dense lymphocytic aggregates with mixed B and T cells, lacking CD23 signals; follicle-like TLS (FL-TLS), dense lymphocytic aggregates with CD23 signals, indicating the presence of a germinal center (both FDC and GC reaction) (Fig. 8B) (Silina et al. 2018). Subsequently, we validated our observations within an external cohort enrolling a total of 94 Chinese patients with ccRCC receiving tyrosine kinase inhibitors (TKIs), particularly sunitinib, post-partial, or radical nephrectomy in the FUSCC. The results suggested that ccRCC patients with higher expression of PINK1 exhibited significantly enhanced PFS (P = 0.039, HR = 0.561) and OS (P = 0.018, HR = 0.565) (Fig. 8C). In more mature TLS, immune cells become better equipped to recognize and attack tumor cells. B cells can produce antibodies targeting tumor antigens, T cells can be activated to directly attack cancer cells, and dendritic cells can present tumor antigens to other immune cells, initiating a coordinated immune response. Interestingly, we found a trend that elevated PINK1 expression markedly correlated with advanced maturation stages of TLS, which may shed light on augmented anti-tumor immune milieu of the ccRCC sample (Fig. 8D).

Fig. 8.

Effects of PINK1 in the clinical prognosis and immune microenvironment of ccRCC. A The images of PINK1 expression in both high- and low-magnification fields using IHC staining analysis. B Evaluation of the maturation stage of tertiary lymphoid structures (TLS) by multispectral immunofluorescence analysis. C Survival analysis of patients stratified by PINK1 in FUSCC cohort receiving TKIs. D Differential PINK1 expression with advanced maturation stages of TLS

Discussion

With rapid advancement in precision medicine and high-throughput approaches, several prognostic models incorporating various clinical and pathological factors, such as tumor stage, grade, size, and lymph node involvement, have been developed to assess the risk and estimate prognosis in ccRCC patients (Qu et al. 2022; Chen et al. 2016). Well-established prognostic models include the Memorial Sloan Kettering Cancer Center (MSKCC), the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) scores, and other multi-omics models (Li 2022a; Yang et al. 2019). In our study, we estimated the predictive potency and biological ramifications of mitophagy in ccRCC with multi-omics data from large-scale populations, which facilitates patient prognosis estimation and guides therapeutic decision-making.

Targeted therapies have revolutionized the landscape of ccRCC therapeutics. Inhibitors targeting the vascular endothelial growth factor (VEGF) pathway, such as sunitinib, pazopanib, and axitinib, have demonstrated efficacy in advanced and metastatic ccRCC. Additionally, TKIs, such as everolimus and temsirolimus, have shown remarkable efficacy in certain patient populations (Capitanio et al. 2021). These targeted therapies improve progression-free survival and overall survival rates compared to traditional chemotherapy. In this study, we incorporated a validation cohort enrolling 94 Chinese ccRCC patients, all recipients of TKIs, specifically sunitinib, following partial or radical nephrectomy in FUSCC. ccRCC patients exhibiting elevated PINK1 expression experience notably favorable clinical outcomes in contrast with their counterparts demonstrating subdued PINK1 expression. This discernment underscores the potential of PINK1-mediated mitophagy as a predictive marker for positive responses and extended survival within the ambit of adjuvant targeted therapies.

Mitophagy in ccRCC can play a role in metabolic adaptation within the TME. Dysfunctional mitochondria generate reactive oxygen species (ROS) and contribute to metabolic perturbations (Xu et al. 2022c; Li et al. 2023). Mitophagy orchestrates the targeted elimination of damaged mitochondria, thereby sustaining mitochondrial integrity and cellular energy homeostasis. This process allows cancer cells to selectively eliminate mitochondria with impaired function and replace them with functional mitochondria, thereby supporting the increased energy demands of tumor cells in the TME (Geissler et al. 2015). Dysregulated mitophagy, which results in the accumulation of damaged mitochondria, increased levels of reactive oxygen species (ROS), and cellular stress, has been implicated in promoting tumor progression and aggressiveness in cancers (Siska et al. 2017). While dysregulated mitophagy has been linked to promoting tumor progression and aggressiveness in certain cancers like ccRCC, it is not a universal predictor for all cancer patients. The interplay between mitophagy and tumor behavior is highly complex and influenced by multiple factors. Therefore, while understanding mitophagy's role can provide valuable insights into cancer biology, it cannot serve as a definitive predictor of tumor behavior for all individuals with cancer. Further research is needed to explore the specifics of this relationship in various cancer types and to develop a deeper understanding of how mitophagy contributes to cancer progression in different contexts.

In ccRCC, dysregulated mitophagy can contribute to the altered metabolic state of tumor cells and provide a growth advantage. Mitophagy may influence the redox balance and oxidative stress within the TME of ccRCC. Dysfunctional mitochondria contribute to the accumulation of ROS, which, in turn, can promote tumorigenesis and tumor progression (Luo 2023). By selectively eliminating damaged mitochondria through mitophagy, ccRCC cells can reduce ROS levels and maintain redox homeostasis. This can impact the oxidative stress levels in the TME and potentially influence tumor cell proliferation and survival.

Mitochondria are critical organelles involved in various cellular processes, including energy production and programmed cell death. If the mitochondria are dysfunctional or not functioning optimally due to abnormal mitophagy, cancer cells might become less susceptible to treatment-induced apoptosis. This cellular heterogeneity could contribute to variable responses to treatment and the emergence of drug-resistant populations within the tumor (Siska et al. 2017). Understanding these mechanisms is crucial for developing effective treatment strategies that can overcome or exploit this potential resistance mechanism. The impact of PINK1-mediated mitophagy within ccRCC extends to the realm of immune responses within the TME (Xu et al. 2022c; Wang et al. 2018). Mitochondria interface with immune signaling pathways, capable of releasing danger-associated molecular patterns (DAMPs) that activate immune responses. Mitophagy may modulate the release of DAMPs alongside metabolic dysregulation, thereby influencing immune responses and interactions between tumor and stromal components within the TME (Zheng et al. 2021; Tang et al. 2023; Anwaier et al. 2023). The regulation of mitophagy in ccRCC cells can mediate immune surveillance, immune cell infiltration, and potentially impact the response to immunotherapies (Hunter et al. 2021; Tian et al. 2023).

Within the realm of ccRCC treatment, ICIs, particularly those targeting the programmed cell death protein 1 (PD-1) pathway, have demonstrated remarkable efficacy (Braun et al. 2021). Agents such as nivolumab and pembrolizumab have demonstrated durable responses and improved survival outcomes, especially in patients with advanced and metastatic disease. Combination therapies involving ICIs and targeted agents are being explored to enhance therapeutic efficacy (Cai et al. 2019). Our utilization of the TIDE algorithm revealed that patients in the low RiskGroup exhibited lower TIDE scores, implying heightened sensitivity to immunotherapy (Jiang et al. 2018). Furthermore, external validation within the David Liu cohort reaffirmed the correlation between high RiskGroup categorization and adverse prognostic outcomes. This underscores the significance of the mitophagy-related signature in predicting responses to standard adjuvant therapies, such as the combination of nivolumab and ipilimumab. Besides, the treatment of ccRCC is increasingly moving toward combination therapies, leveraging the benefits of different treatment modalities, and the PINK1-mediated mitophagy contributed to elevated frequency and advanced maturation of TLS (Meylan et al. 2022; Munoz-Erazo et al. 2020). It is mutually confirmed that, by eliminating damaged mitochondria, ccRCC cells can remove sources of cellular stress and enhance their survival in the face of anticancer treatments (Kim et al. 2021). This resilience to therapeutic interventions holds potential implications for the efficacy of conventional treatments, inclusive of targeted therapies and immunotherapies (Vesely et al. 2022; Li 2022b).

Conclusion

Collectively, this study first comprehensively demonstrated the effects of mitophagy on the TME characteristics, clinical prognosis, and treatments response in ccRCC patients, which may contribute to more precise treatment strategies for ccRCC patients. Besides, we found that PINK1 markedly correlated with treatment response and reshaped immune TME in congruence with advanced maturation stages of tumor-associated tertiary lymphoid structures in ccRCC.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ccRCC

Clear cell renal cell carcinoma

CDF

Cumulative distribution function

CNVs

Copy number variations

CPTAC

Clinical Proteomic Tumor Analysis Consortium

DAMPs

Danger-associated molecular patterns

DCA

Decision curve analysis

DEGs

Differential expression genes

EMBL

European Molecular Biology Laboratory

FPKM

Fragments per kilobase of transcript per million

FUSCC

Fudan University Shanghai Cancer Center

GO

Gene Ontology

GSVA

Gene set variation analysis

ICGC

International Cancer Genome Consortium

ICIs

Immune checkpoint inhibitors

ICPs

Immune checkpoints

IHC

Immunohistochemistry

IMDC

International Metastatic Renal Cell Carcinoma Database Consortium

KEGG

Kyoto Encyclopedia of Genes and Genomes

LASSO

Least absolute shrinkage and selector operation

LOH

Loss of heterozygosity

mIF

Multispectral immunofluorescence

MRGs

Mitophagy-related genes

MSKCC

Memorial Sloan Kettering Cancer Center

OS

Overall survival

PCA

Principal component analysis

PD-1

Programmed cell death protein 1

PINK1

PTEN-induced putative kinase 1

PPI

Protein–protein interaction

RCC

Renal cell carcinoma

ROC

Receiver operating characteristic

ROS

Reactive oxygen species

ssGSEA

Single-sample gene set enrichment analysis

TCGA

The cancer genome atlas

TIDE

Tumor immune dysfunction and exclusion

TKI

Tyrosine kinase inhibitors

TMB

Tumor mutation burden

TME

Tumor microenvironment

TPM

Transcripts per million

VEGF

Vascular endothelial growth factor

VHL

Von Hippel–Lindau

Author contributions

JY, JX, and WL: helped in conceptualization. JX, WL, and SL: contributed to data curation and formal analysis. JX, WL, and HT: worked in investigation and methodology. HT and TW: worked in resources and software. JY, HT, and TW: worked in supervision. JX, JY, and WL: contributed to original draft. JY, HT, and TW: helped in editing.

Funding

This study was partially supported by the grant of Longhua Hospital of Shanghai University of Traditional Chinese Medicine (YW002.017).

Data availability

The datasets used and/or analyzed in the study are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

The authors report no potential conflicts of interest.

Ethical approval and consent to participate

The study design and test procedures followed the Helsinki Declaration II. The ethics approval and consent to participate of the Urology and Pathology departments in this study were approved by the ethics committee of Renji Hospital, Shanghai Jiao Tong University School of Medicine.