Clear cell renal cell carcinoma (ccRCC) is the most common renal cancer type (90%), with nearly 300,000 (2.6%) new cases yearly and higher incidence in developed countries [1, 2]. CcRCC is heterogeneous and characterized by clear or eosinophilic cells arranged in trabecular cords or nest architecture in a delicate stroma richly vascularized. The risk of recurrence post-nephrectomy for locally advanced kidney cancer is notably variable, with key factors including T stage, ISUP (International Society of Urological Pathology) grade, and the presence of necrosis or symptoms at diagnosis [3–5]. Various prognostic nomograms, such as UISS (University of California Los Angeles Integrated Staging System), ASSURE, and Leibovitch, have incorporated these clinical and pathological data [3, 6–10]. It is noteworthy that, as of now, no biological markers are employed in clinical practice to augment the prognostic precision of existing models.

As metabolic reprogramming is one of the hallmarks of cancer and mitochondria play a central role in metabolism, their role in carcinogenesis has been evaluated for years [11, 12]. CcRCC is characterized by a metabolic shift from mitochondrial oxidation to anaerobic glycolysis in the presence of oxygen, known as the Warburg effect [13–15], driven by the loss of function of the Von Hippel-Lindau gene (VHL) and the Hypoxia-inducible Factor (HIF) stabilization [16, 17].

In ccRCC, somatic variants in the nuclear genome accumulate according to the phylogenetic clone model, retaining a driver strain defect: the bi-allelic inactivation of VHL [18]. Other mutated genes are numerous and most often affect polybromo-1 (PBRM1), SET domain containing 2 (SETD2) or BRCA1-associated protein 1 (BAP1) genes [18–20], which are involved in methylation or chromatin compaction phenomena. Mitochondrial DNA (mtDNA) is a 16,569 bp circular double-stranded DNA molecule, maternally inherited. It encodes 13 polypeptides involved in oxidative phosphorylation (OXPHOS), and regulates reactive oxygen species (ROS) [21]. The accumulation of qualitative alterations and/or quantitative variation of mtDNA has been found to play a crucial role in carcinogenesis and metastasis, among other elements, by modulating cellular metabolism [11, 12]. Variations in mtDNA and their consequences on survival and disease progression are variable among cancers. In cancer subtypes of renal origin (chromophobe carcinoma) or non-renal origin (adrenocortical carcinoma or glioblastoma), high mtDNA copy number (mtDNAcn) is associated with better survival [22–24]. Conversely, a significant increase in mtDNAcn is associated with poorer prognosis in melanoma or breast carcinoma [22, 25]. In ccRCC, the impact of mtDNAcn on prognosis is conflicting. Some studies have shown that an increase in mtDNAcn relative to normal matched normal tissue is associated with a worse prognosis [22]. Still, other studies have suggested that a decrease in mtDNAcn compared to normal paired tissue and mitochondrial respiratory chain activity in vitro is associated with increased tumor growth, invasion capacity, and drug resistance mechanisms [26].

The first aim of this study was to assess the influence of mtDNA genetics (a.k.a mitochondrial copy number variation, deletions, and mutations) on the prognosis in patients with ccRCC and to explore histologic correlation. As an exploratory second aim, we looked at the impact of mtDNA depletion on cell proliferation in a ccRCC cell line.

Mitochondria play a major role in multiple cellular functions, such as energy production, formation of reactive oxygen species [30, 31], and initiation of apoptosis [30, 32]. Mitochondrial dysfunctions are involved in tumorigenesis [33] through mechanisms of epithelial-mesenchymal transition [34] and metabolic changes (i.e., Warburg Effect) [13–15].

MtDNAcn variations during carcinogenesis are highly dependent on cancer subtypes [22–24, 35]. As previously described [22, 26], we observed that most ccRCCs showed an overall mtDNA depletion compared to the normal kidney. This finding is probably due to the loss of function of the VHL complex secondary to a biallelic inactivation of VHL [35, 36], which is present in more than 90% of ccRCCs. Indeed, VHL loss of function induces an abnormal activation of HIF, responsible for ccRCC’s distinctive metabolic reprogramming [36–38], the “Warburg effect” [16, 17], which promotes the shift of glucose use toward the lactate pathway instead of Oxidative Phosphorylation (OXPHOS) [13, 15]. In addition, through the inhibition of carnitine palmitoyltransferase 1A (CPT1A), HIF is responsible for reducing fatty acid transport into the mitochondria and compels the storage of fatty acids as lipid droplets in the cytoplasm [39]. Those mechanisms may partially explain why ccRCCs are naturally depleted in mitochondria, displaying minimal respiratory capacity [40] and harbouring a clear cell appearance [39]. In this context, as tumor cells appear to depend little on mitochondria (OXPHOS) for their metabolism, they may downregulate the level of mtDNAcn.

The effects of mtDNA depletion in in vitro models depend on the cell line origin and types of cancer. In breast cancer cell lines, for example, the mtDNAcn reduction slows down the cell proliferation rate and diminishes the tumorigenic phenotype [41, 42]. In contrast, glioblastoma cell lines maintain a tumorigenic phenotype through a low mtDNAcn as a result of a hypermethylation process as previously shown [43]. In vitro, we showed that mtDNA depletion decreases cell proliferation in the 786-O cell line, which is derived from a primary renal clear cell carcinoma, after being treated with BET. These results support our clinical data, suggesting that a lower mtDNAcn in the tumor is associated with a better prognosis in ccRCC, similar to breast cancer [25]. The opposite data were observed in chromophobe renal cell carcinoma or glioblastoma in young subjects under 40 years old, where an increase of mtDNAcn was associated with better prognosis and survival [22–26]. One limitation is that our in vitro study was designed to be exploratory and basic. It could be worthwhile to complete it with an additional essay exploring glucose or glutamine metabolites, for example.

We showed that the tumor mtDNAcn relative to that of healthy tissue was an independent prognostic factor in ccRCC patients. Indeed, the HIGH group ratio was associated with worse overall survival, with a 5-year survival rate of 78.7% compared to 95.5% in the LOW group.

Since 2021, pembrolizumab has exhibited its efficiency in reducing the likelihood of recurrence and mortality following nephrectomy for locally advanced kidney cancer [44, 45]. Despite its positive impact, this treatment induces many side effects, underscoring the importance of identifying patients at the highest risk of recurrence for optimal use in the years ahead. In our cohort, very few patients received adjuvant therapy such as targeted therapy (VEGF inhibitors) or an immune checkpoint inhibitor. The effect of adjuvant treatment did not influence our data. In the same way, there was no statistical difference in mtDNAcn in renal failure patients (n = 5) compared to others, so we believe that despite the morphologic control of frozen healthy renal tissue, it did not cause a bias.

Our results show that, within mitochondrial genetics, mtDNA copy number variation appears to play a more critical role in the physiopathology of ccRCC than the accumulation of somatic mutations. In our cohort, as previously described in renal cancer [11, 35, 46], few mtDNA somatic mutations were identified, mainly carrying low interspecies conservation scores and with a low to moderate impact on mitochondrial function, suggesting a transient rather than a driving role of those variants. However, NGS was performed on a small cohort, which may have limited the detection of certain anomalies due to a lack of power.

CcRCCs are heterogeneous tumors that can display clear or eosinophilic cytoplasm features in the same tumor. As explained above, the clear cell appearance of ccRCC is likely due to multiple mechanisms related to VHL inactivation. Moreover, ultrastructural and immunohistochemical studies have shown a higher number of mitochondria in eosinophilic ccRCC than in its clear cell counterpart [47–50]. Interestingly studies of intratumoral heterogeneity showed that the eosinophilic components of ccRCC are associated with poor prognosis outcomes [51, 52] compared to the clear cell components despite a common genetic background with VHL inactivation [18, 53]. In our study, although the diagnosis of ccRCC was done by an experienced pathologist who ruled out differential diagnosis bearing clear cell morphology, we did not sequence the VHL gene, which remains a limitation. This limitation is somewhat limited because VHL inactivation is present in almost all ccRCC. Indeed recent advances have identified new entities in renal carcinoma with clear cell morphology and VHL wild type phenotype such as TFE3-rearranged RCC, ELOC (formely TCEB1)-mutated RCC or clear cell renal papillary tumor [54–59].

Our data showed that eosinophilic morphology is associated with increased mtDNAcn and worse DFS and CSS. A recent study [52] has demonstrated the value of distinguishing ccRCCs’ clear cells from eosinophilic cells for the prognostic and therapeutic strategy. Indeed, eosinophilic ccRCCs, due to an abundant lymphocytic inflammatory infiltrate, are good responders to immunotherapy, whereas clear cells ccRCC, characterised by an abundant vascular stroma, have a higher sensitivity to targeted therapy [52]. Nilsson et al. [51] performed RNA sequencing on eosinophilic and clear cell RCC. They showed an overexpression in the eosinophilic component of the DNA polymerase subunit gamma (POLG) encoding for the POLG catalytic subunit, solely responsible for the mtDNA replication. The POLG induction may explain the increase of mtDNAcn in this contingent even though we currently do not know the mechanism responsible for POLG overexpression. Hence, the accumulation of mtDNAcn in eosinophilic areas is more likely due to an increase in mitochondrial biogenesis rather than a defect in the mitophagy process. Interestingly, in this study, the TFAM gene (Transcription Factor A, Mitochondrial) was downregulated, which could be surprising considering its function as an mtDNA transcription factor. However, TFAM is also identified as the significant mtDNA packaging protein, constituting the core component of the mitochondrial nucleoid [60, 61]. Hence, an increased TFAM expression results in stronger mtDNA compaction, making it less permissive to replication and mitochondrial genome accessibility [62]. As a result, TFAM underexpression enables replication consistent with the mtDNAcn increase.

It is important to better understand why and how the increase in mtDNAcn could benefit to ccRCC since it is naturally deprived (see above). One hypothesis is the metabolism of glutamine particularly because this one is favoured by the metabolic acidosis due to the Warburg effect [63, 64]. Glutamine is a non-essential amino acid provided by the diet and synthesised endogenously. Glutamine is further metabolised into glutamate supporting the biosynthesis of nucleotides and amino acids and the energy metabolism throughout the Tricarboxylic Acid Cycle (TCA), the OXPHOS via the production of mitochondrial citrate, and fatty acid beta-oxidation [32, 65, 66]. Glutamine would act as a proper energy substrate in this context while the glycolysis and beta-oxidation pathways are reduced [37, 65]. An increase in the number of mitochondria could be associated with an increase in glutaminolysis metabolism, providing the energy and nucleotide substrates necessary for tumor cell proliferation. Evidence supporting this hypothesis relies on the increased isocitrate dehydrogenase IDH1 gene expression in eosinophilic cells [51]. Hakimi et al [38] reported that tumor progression and metastasis were associated with increased metabolites in the glutathione pathway. Yet, glutathione is a tripeptide made of glutamate, cysteine, and glycine. Moreover, through the formation of glutathione and glutamate, glutamine enables tumor cells to resist to ROS overproduction and anoikis via autophagy phenomena [65, 67] and promotes the formation of metastases [65]. Moreover, it has recently been shown that the Solute Carrier Family 1 Member 5 (SLC1A5) transporter [68], which allows the entry of glutamine into the mitochondria, had an oncogenic role since its in vitro inactivation was associated with a reduction of tumor growth [68]. Further experiments should be done to investigate the glutamine pathway and metabolites in clear and eosinophilic cells in ccRCC. This hypothesis is supported by several in vitro and phase 1-2 studies performed in kidney cancer (papillary or clear cells). These studies showed that glutaminase inhibitors decrease cell growth proliferation and tumor size and potentiate the effects of targeted therapies, especially immunotherapies [69, 70]. Glutamine inhibitors act dually by inhibiting cell proliferation intrinsically (downstream signalling pathways) and by restoring the activation of effector T cells by increasing the availability of cytoplasmic glutamine. As we have seen above, eosinophilic cells are enriched in lymphocytic inflammatory infiltrate and mitochondria, so we would expect immunotherapies combined with glutaminase inhibitors to be beneficial in eosinophilic rather than clear cells ccRCC in future clinical trials.

In conclusion, high mtDNAcn in tumors compared to healthy kidneys is an independent factor of poor prognosis in ccRCC and is associated with an eosinophilic morphology which highlights the significance of identifying this pattern on pathological examination. The benefit of mtDNAcn increase for the tumor cell could be mediated through the glutamine metabolic pathway. If confirmed, this hypothesis may provide a new perspective on the pathological diagnosis and therapy of ccRCC, especially concerning the synergistic combination of immunotherapy and glutaminase inhibitors.