63Plasma proteomic profiling reveals distinct mitochondrial signatures in chromophobe renal cell carcinoma

Abstract Background Chromophobe renal cell carcinoma (chRCC) is a rare kidney cancer subtype that can be challenging to differentiate histologically. Improved molecular classification is essential for accurate diagnosis and potential therapeutic targeting. SomaScan, a high-throughput aptamer-based proteomic platform, enables quantification of thousands of proteins in plasma and offers a promising approach to identify clinically relevant biomarkers for chRCC. Methods We analyzed plasma proteomic profiles from 215 patients with renal cell carcinoma using SomaScan, including 18 chRCC and 197 clear cell renal cell carcinoma (ccRCC) samples. After identifying proteins significantly upregulated in chRCC (log2 fold change 1 and adjusted p-value 0.05), a reproducibility-focused feature selection strategy was employed to prioritize proteins most consistently associated with chRCC. These upregulated proteins were used as input features for the Least Absolute Shrinkage and Selection Operator (LASSO) machine learning algorithm. To enhance the stability and reproducibility of feature selection, we implemented a bootstrap strategy: 100 bootstrap datasets were created by randomly sampling the rows of the upregulated protein matrix with replacement. In each of 100 bootstrap replicates, we randomly sampled with replacement to create a training set, performed LASSO logistic regression for feature selection and model fitting, and evaluated the resulting model on the out-of-bag (OOB) test samples. Results Of 7289 proteins measured, 215 were significantly upregulated in chRCC compared to ccRCC. Unsupervised clustering revealed that the majority of chRCC samples formed a distinct group, suggesting a unique proteomic signature. Notably, chRCC plasma showed strong enrichment of mitochondrial and metabolic proteins, reflecting the tumor’s distinct bioenergetic profile. Several mitochondrial enzymes were among the most upregulated proteins in chRCC. These included ECH1 (log2FC = 1.81) and ECI1 (log2FC = 1.66), involved in fatty acid β-oxidation, as well as ALDH6A1 (log2FC = 2.33), which participates in valine and pyrimidine catabolism. Elevated expression of PCK2 (log2FC = 1.06), a gluconeogenesis enzyme, and creatine metabolism enzymes such as CKMT1A (log2FC = 1.07) and GATM (log2FC = 2.24) could further highlight the distinctive metabolic profile of chRCC. Upregulation of MRPL14 (log2FC = 1.06), a mitochondrial ribosomal protein, suggests increased mitochondrial biogenesis and translational activity. These findings underscore a bioenergetic program in chRCC that contrasts with the glycolytic phenotype characteristic of ccRCC. In addition to mitochondrial proteins, we observed differential expression of proteins involved in intracellular trafficking and extracellular matrix remodeling. ARL15 (log2FC = 1.24), a GTPase involved in vesicle transport and metabolic regulation, was upregulated in chRCC. KLK15 (log2FC = 1.52), a kallikrein family protease, may play a role in extracellular matrix remodeling. Conclusions Plasma proteomic profiling revealed a distinct mitochondrial-based signature in chRCC, offering novel insights into its biology and supporting accurate diagnostic classification. The identified proteins reflect key biological processes such as oxidative metabolism and metabolic adaptation—features that distinguish chRCC from ccRCC and may inform future diagnostic and therapeutic strategies.