Metabolic symbiosis between oxygenated and hypoxic tumour cells: An agent-based modelling study.

Deregulated metabolism is one of the hallmarks of cancer. It is well-known that tumour cells tend to metabolize glucose via glycolysis even when oxygen is available and mitochondrial respiration is functional. However, the lower energy efficiency of aerobic glycolysis with respect to mitochondrial respiration makes this behaviour, namely the Warburg effect, counter-intuitive, although it has now been recognized as source of anabolic precursors. On the other hand, there is evidence that oxygenated tumour cells could be fuelled by exogenous lactate produced from glycolysis. We employed a multi-scale approach that integrates multi-agent modelling, diffusion-reaction, stoichiometric equations, and Boolean networks to study metabolic cooperation between hypoxic and oxygenated cells exposed to varying oxygen, nutrient, and inhibitor concentrations. The results show that the cooperation reduces the depletion of environmental glucose, resulting in an overall advantage of using aerobic glycolysis. In addition, the oxygen level was found to be decreased by symbiosis, promoting a further shift towards anaerobic glycolysis. However, the oxygenated and hypoxic populations may gradually reach quasi-equilibrium. A sensitivity analysis using Latin hypercube sampling and partial rank correlation shows that the symbiotic dynamics depends on properties of the specific cell such as the minimum glucose level needed for glycolysis. Our results suggest that strategies that block glucose transporters may be more effective to reduce tumour growth than those blocking lactate intake transporters.

New role of fat-free mass in cancer risk linked with genetic predisposition.

Cancer risk is associated with the widely debated measure body mass index (BMI). Fat mass and fat-free mass measurements from bioelectrical impedance may further clarify this association. The UK Biobank is a rare resource in which bioelectrical impedance and BMI data was collected on ~ 500,000 individuals. Using this dataset, a comprehensive analysis using regression, principal component and genome-wide genetic association, provided multiple levels of evidence that increasing whole body fat (WBFM) and fat-free mass (WBFFM) are both associated with increased post-menopausal breast cancer risk, and colorectal cancer risk in men. WBFM was inversely associated with prostate cancer. We also identified rs615029[T] and rs1485995[G] as associated in independent analyses with both PMBC (p = 1.56E-17 and 1.78E-11) and WBFFM (p = 2.88E-08 and 8.24E-12), highlighting splice variants of the intriguing long non-coding RNA CUPID1 (LINC01488) as a potential link between PMBC risk and fat-free mass.

Robustness and reproducibility for AI learning in biomedical sciences: RENOIR.

Artificial intelligence (AI) techniques are increasingly applied across various domains, favoured by the growing acquisition and public availability of large, complex datasets. Despite this trend, AI publications often suffer from lack of reproducibility and poor generalisation of findings, undermining scientific value and contributing to global research waste. To address these issues and focusing on the learning aspect of the AI field, we present RENOIR (REpeated random sampliNg fOr machIne leaRning), a modular open-source platform for robust and reproducible machine learning (ML) analysis. RENOIR adopts standardised pipelines for model training and testing, introducing elements of novelty, such as the dependence of the performance of the algorithm on the sample size. Additionally, RENOIR offers automated generation of transparent and usable reports, aiming to enhance the quality and reproducibility of AI studies. To demonstrate the versatility of our tool, we applied it to benchmark datasets from health, computer science, and STEM (Science, Technology, Engineering, and Mathematics) domains. Furthermore, we showcase RENOIR's successful application in recently published studies, where it identified classifiers for SET2D and TP53 mutation status in cancer. Finally, we present a use case where RENOIR was employed to address a significant pharmacological challenge-predicting drug efficacy. RENOIR is freely available at https://github.com/alebarberis/renoir .

Antitumour effect of the mitochondrial complex III inhibitor Atovaquone in combination with anti-PD-L1 therapy in mouse cancer models.

Immune checkpoint blockade (ICB) provides effective and durable responses for several tumour types by unleashing an immune response directed against cancer cells. However, a substantial number of patients treated with ICB develop relapse or do not respond, which has been partly attributed to the immune-suppressive effect of tumour hypoxia. We have previously demonstrated that the mitochondrial complex III inhibitor atovaquone alleviates tumour hypoxia both in human xenografts and in cancer patients by decreasing oxygen consumption and consequently increasing oxygen availability in the tumour. Here, we show that atovaquone alleviates hypoxia and synergises with the ICB antibody anti-PD-L1, significantly improving the rates of tumour eradication in the syngeneic CT26 model of colorectal cancer. The synergistic effect between atovaquone and anti-PD-L1 relied on CD8+ T cells, resulted in the establishment of a tumour-specific memory immune response, and was not associated with any toxicity. We also tested atovaquone in combination with anti-PD-L1 in the LLC (lung) and MC38 (colorectal) cancer syngeneic models but, despite causing a considerable reduction in tumour hypoxia, atovaquone did not add any therapeutic benefit to ICB in these models. These results suggest that atovaquone has the potential to improve the outcomes of patients treated with ICB, but predictive biomarkers are required to identify individuals likely to benefit from this intervention.

A machine learning and directed network optimization approach to uncover TP53 regulatory patterns.

TP53, the Guardian of the Genome, is the most frequently mutated gene in human cancers and the functional characterization of its regulation is fundamental. To address this we employ two strategies: machine learning to predict the mutation status of TP53from transcriptomic data, and directed regulatory networks to reconstruct the effect of mutations on the transcipt levels of TP53 targets. Using data from established databases (Cancer Cell Line Encyclopedia, The Cancer Genome Atlas), machine learning could predict the mutation status, but not resolve different mutations. On the contrary, directed network optimization allowed to infer the TP53 regulatory profile across: (1) mutations, (2) irradiation in lung cancer, and (3) hypoxia in breast cancer, and we could observe differential regulatory profiles dictated by (1) mutation type, (2) deleterious consequences of the mutation, (3) known hotspots, (4) protein changes, (5) stress condition (irradiation/hypoxia). This is an important first step toward using regulatory networks for the characterization of the functional consequences of mutations, and could be extended to other perturbations, with implications for drug design and precision medicine.

MAGNETO: Cell type marker panel generator from single-cell transcriptomic data.

Single-cell RNA sequencing experiments produce data useful to identify different cell types, including uncharacterized and rare ones. This enables us to study the specific functional roles of these cells in different microenvironments and contexts. After identifying a (novel) cell type of interest, it is essential to build succinct marker panels, composed of a few genes referring to cell surface proteins and clusters of differentiation molecules, able to discriminate the desired cells from the other cell populations. In this work, we propose a fully-automatic framework called MAGNETO, which can help construct optimal marker panels starting from a single-cell gene expression matrix and a cell type identity for each cell. MAGNETO builds effective marker panels solving a tailored bi-objective optimization problem, where the first objective regards the identification of the genes able to isolate a specific cell type, while the second conflicting objective concerns the minimization of the total number of genes included in the panel. Our results on three public datasets show that MAGNETO can identify marker panels that identify the cell populations of interest better than state-of-the-art approaches. Finally, by fine-tuning MAGNETO, our results demonstrate that it is possible to obtain marker panels with different specificity levels.

The impact of survivorship bias in glioblastoma research.

Despite advances in the therapy of Central Nervous System (CNS) malignancies, treatment of glioblastoma (GB) poses significant challenges due to GB resistance and high recurrence rates following post-operative radio-chemotherapy. The majority of prognostic and predictive GB biomarkers are currently developed using tumour samples obtained through surgical interventions. However, the selection criteria adopted by different neurosurgeons to determine which cases are suitable for surgery make operated patients not representative of all GB cases. Particularly, geriatric and frail individuals are excluded from surgical consideration in some cancer centers. Such selection generates a survival (or selection) bias that introduces limitations, rendering the patients or data chosen for downstream analyses not representative of the entire community. In this review, we discuss the implication of survivorship bias on current and novel biomarkers for patient selection, stratification, therapy, and outcome analyses.

WNT5A-ROR2 axis mediates VEGF dependence of BRAF mutant melanoma.

PURPOSE: Despite recent advances, approximately 50% of patient with metastatic melanoma eventually succumb to the disease. Patients with melanomas harboring a BRAF mutation (BRAFMut) have a worse prognosis than those with wildtype (BRAFWT) tumors. Unexpectedly, interim AVAST-M Phase III trial data reported benefit from adjuvant anti-VEGF bevacizumab only in the BRAFMut group. We sought to find mechanisms underpinning this sensitivity. METHODS: We investigated this finding in vitro and in vivo using melanoma cell lines and clones generated by BRAFV600E knock-in on a BRAFWT background. RESULTS: Compared with BRAFWT cells, isogenic BRAFV600E clones secreted more VEGF and exhibited accelerated growth rates as spheroids and xenografts, which were more vascular and proliferative. Recapitulating AVAST-M findings, bevacizumab affected only BRAFV600E xenografts, inducing significant tumor growth delay, reduced vascularity and increased necrosis. We identified 814 differentially expressed genes in isogenic BRAFV600E/BRAFWT clones. Of 61 genes concordantly deregulated in clinical melanomas ROR2 was one of the most upregulated by BRAFV600E. ROR2 was shown to be RAF-MEK regulated in BRAFV600E cells and its depletion suppressed VEGF secretion down to BRAFWT levels. The ROR2 ligand WNT5A was also overexpressed in BRAFMut melanomas, and in ROR2-overexpressing BRAFV600E cells MEK inhibition downregulated WNT5A and VEGF secretion. CONCLUSIONS: These data implicate WNT5A-ROR2 in VEGF secretion, vascularity, adverse outcomes and bevacizumab sensitivity of BRAFMut melanomas, suggesting that this axis has potential therapeutic relevance.

Machine Learning Using Gene-Sets to Infer miRNA Function.

miRNA are regulators of cell phenotype, and there is clear evidence that these small posttranscriptional modifiers of gene expression are involved in defining a cellular response across states of development and disease. Classical methods for elucidating the repressive effect of a miRNA on its targets involve controlling for the many factors influencing miRNA action, and this can be achieved in cell lines, but misses tissue and organism level context which are key to a miRNA function. Also, current technology to carry out this validation is limited in both generalizability and throughput. Methodologies with greater scalability and rapidity are required to better understand the function of these important species of RNA. To this end, there is an increasing store of RNA expression level data incorporating both miRNA and mRNA, and in this chapter, we describe how to use machine learning and gene-sets to translate the knowledge of phenotype defined by mRNA to putative roles for miRNA. We outline our approach to this process and highlight how it was done for our miRNA annotation of the hallmarks of cancer using the Cancer Genome Atlas (TCGA) dataset. The concepts we present are applicable across datasets and phenotypes, and we highlight potential pitfalls and challenges that may be faced as they are used.

New perspectives in liquid biopsy for glioma patients.

PURPOSE OF REVIEW: Gliomas are the most common primary tumors of the central nervous system. They are characterized by a disappointing prognosis and ineffective therapy that has shown no substantial improvements in the past 20 years. The lack of progress in treating gliomas is linked with the inadequacy of suitable tumor samples to plan translational studies and support laboratory developments. To overcome the use of tumor tissue, this commentary review aims to highlight the potential for the clinical application of liquid biopsy (intended as the study of circulating biomarkers in the blood), focusing on circulating tumor cells, circulating DNA and circulating noncoding RNA. RECENT FINDINGS: Thanks to the increasing sensitivity of sequencing techniques, it is now possible to analyze circulating nucleic acids and tumor cells (liquid biopsy). SUMMARY: Although studies on the use of liquid biopsy are still at an early stage, the potential clinical applications of liquid biopsy in the study of primary brain cancer are many and have the potential to revolutionize the approach to neuro-oncology, and importantly, they offer the possibility of gathering information on the disease at any time during its history.

Single cell RNA-sequencing: A powerful yet still challenging technology to study cellular heterogeneity.

Almost all biomedical research to date has relied upon mean measurements from cell populations, however it is well established that what it is observed at this macroscopic level can be the result of many interactions of several different single cells. Thus, the observable macroscopic 'average' cannot outright be used as representative of the 'average cell'. Rather, it is the resulting emerging behaviour of the actions and interactions of many different cells. Single-cell RNA sequencing (scRNA-Seq) enables the comparison of the transcriptomes of individual cells. This provides high-resolution maps of the dynamic cellular programmes allowing us to answer fundamental biological questions on their function and evolution. It also allows to address medical questions such as the role of rare cell populations contributing to disease progression and therapeutic resistance. Furthermore, it provides an understanding of context-specific dependencies, namely the behaviour and function that a cell has in a specific context, which can be crucial to understand some complex diseases, such as diabetes, cardiovascular disease and cancer. Here, we provide an overview of scRNA-Seq, including a comparative review of emerging technologies and computational pipelines. We discuss the current and emerging applications and focus on tumour heterogeneity a clear example of how scRNA-Seq can provide new understanding of a complex disease. Additionally, we review the limitations and highlight the need of powerful computational pipelines and reproducible protocols for the broader acceptance of this technique in basic and clinical research.

Obesity: a perfect storm for carcinogenesis.

Obesity-related cancers account for 40% of the cancer cases observed in the USA and obesity is overtaking smoking as the most widespread modifiable risk factor for carcinogenesis. Here, we use the hallmarks of cancer framework to delineate how obesity might influence the carcinogenic hallmarks in somatic cells. We discuss the effects of obesity on (a) sustaining proliferative signaling; (b) evading growth suppressors; (c) resisting cell death; (d) enabling replicative immortality; (e) inducing angiogenesis; (f) activating invasion and metastasis; (g) reprogramming energy metabolism; and (h) avoiding immune destruction, together with its effects on genome instability and tumour-promoting inflammation. We present the current understanding and controversies in this evolving field, and highlight some areas in need of further cross-disciplinary focus. For instance, the relative importance of the many potentially causative obesity-related factors is unclear for each type of malignancy. Even within a single tumour type, it is currently unknown whether one obesity-related factor consistently plays a predominant role, or if this varies between patients or, even in a single patient with time. Clarifying how the hallmarks are affected by obesity may lead to novel prevention and treatment strategies for the increasingly obese population.

The MITF regulatory network in melanoma.

Bidirectional interactions between plastic tumor cells and the microenvironment critically impact tumor evolution and metastatic dissemination by enabling cancer cells to adapt to microenvironmental stresses by switching phenotype. In melanoma, a key determinant of phenotypic identity is the microphthalmia-associated transcription factor MITF that promotes proliferation, suppresses senescence, and anticorrelates with immune infiltration and therapy resistance. What determines whether MITF can activate or repress genes associated with specific phenotypes, or how signaling regulating MITF might impact immune infiltration is poorly understood. Here, we find that MITF binding to genes associated with high MITF is via classical E/M-box motifs, but genes downregulated when MITF is high contain FOS/JUN/AP1/ATF3 sites. Significantly, the repertoire of MITF-interacting factors identified here includes JUN and ATF3 as well as many previously unidentified interactors. As high AP1 activity is a hallmark of MITFLow , invasive, slow-cycling, therapy resistant cells, the ability of MITF to repress AP1-regulated genes provides an insight into how MITF establishes and maintains a pro-proliferative phenotype. Moreover, although ß-catenin has been linked to immune exclusion, many Hallmark ß-catenin signaling genes are associated with immune infiltration. Instead, low MITF together with Notch signaling is linked to immune infiltration in both mouse and human melanoma tumors.

Liver glycogen phosphorylase is upregulated in glioblastoma and provides a metabolic vulnerability to high dose radiation.

Channelling of glucose via glycogen, known as the glycogen shunt, may play an important role in the metabolism of brain tumours, especially in hypoxic conditions. We aimed to dissect the role of glycogen degradation in glioblastoma (GBM) response to ionising radiation (IR). Knockdown of the glycogen phosphorylase liver isoform (PYGL), but not the brain isoform (PYGB), decreased clonogenic growth and survival of GBM cell lines and sensitised them to IR doses of 10-12 Gy. Two to five days after IR exposure of PYGL knockdown GBM cells, mitotic catastrophy and a giant multinucleated cell morphology with senescence-like phenotype developed. The basal levels of the lysosomal enzyme alpha-acid glucosidase (GAA), essential for autolysosomal glycogen degradation, and the lipidated forms of gamma-aminobutyric acid receptor-associated protein-like (GABARAPL1 and GABARAPL2) increased in shPYGL U87MG cells, suggesting a compensatory mechanism of glycogen degradation. In response to IR, dysregulation of autophagy was shown by accumulation of the p62 and the lipidated form of GABARAPL1 and GABARAPL2 in shPYGL U87MG cells. IR increased the mitochondrial mass and the colocalisation of mitochondria with lysosomes in shPYGL cells, thereby indicating reduced mitophagy. These changes coincided with increased phosphorylation of AMP-activated protein kinase and acetyl-CoA carboxylase 2, slower ATP generation in response to glucose loading and progressive loss of oxidative phosphorylation. The resulting metabolic deficiencies affected the availability of ATP required for mitosis, resulting in the mitotic catastrophy observed in shPYGL cells following IR. PYGL mRNA and protein levels were higher in human GBM than in normal human brain tissues and high PYGL mRNA expression in GBM correlated with poor patient survival. In conclusion, we show a major new role for glycogen metabolism in GBM cancer. Inhibition of glycogen degradation sensitises GBM cells to high-dose IR indicating that PYGL is a potential novel target for the treatment of GBMs.

Disease-associated KBTBD4 mutations in medulloblastoma elicit neomorphic ubiquitylation activity to promote CoREST degradation.

Medulloblastoma is the most common malignant brain tumour in children. Genomic studies have identified distinct disease subgroups: wnt/wingless (WNT), sonic hedgehog (SHH), and non-WNT/non-SHH, comprising group 3 and group 4. Alterations in WNT and SHH signalling form the pathogenetic basis for their subgroups, whereas those for non-WNT/non-SHH tumours remain largely elusive. Recent analyses have revealed recurrent in-frame insertions in the E3 ubiquitin ligase adaptor Kelch Repeat and BTB Domain Containing 4 (KBTBD4) in cases of group 3/4 medulloblastoma. Critically, group 3/4 tumours with KBTBD4 mutations typically lack other gene-specific alterations, such as MYC amplification, indicating KBTBD4 insertion mutations as the primary genetic driver. Delineating the role of KBTBD4 mutations thus offers significant opportunities to understand tumour pathogenesis and to exploit the underpinning mechanisms therapeutically. Here, we show a novel mechanism in cancer pathogenesis whereby indel mutations in KBTBD4 drive its recognition of neo-substrates for degradation. We observe that KBTBD4 mutants promote the recruitment and ubiquitylation of the REST Corepressor (CoREST), which forms a complex to modulate chromatin accessibility and transcriptional programmes. The degradation of CoREST promoted by KBTBD4 mutation diverts epigenetic programmes inducing significant alterations in transcription to promote increased stemness of cancer cells. Transcriptional analysis of >200 human group 3 and 4 medulloblastomas by RNA-seq, highlights the presence of CoREST and stem-like signatures in tumours with KBTBD4 mutations, which extend to a further sub-set of non-mutant tumours, suggesting CoREST alterations as a novel pathogenetic mechanism of wide relevance in groups 3 and 4. Our findings uncover KBTBD4 mutation as a novel driver of epigenetic reprogramming in non-WNT/non-SHH medulloblastoma, establish a novel mode of tumorigenesis through gain-of-function mutations in ubiquitin ligases (neo-substrate recruitment) and identify both mutant KBTBD4 and CoREST complexes as new druggable targets for improved tumour-specific therapies.

ADGRL4/ELTD1 Expression in Breast Cancer Cells Induces Vascular Normalization and Immune Suppression.

ELTD1/ADGRL4 expression is increased in the vasculature of a number of tumor types and this correlates with a good prognosis. Expression has also been reported in some tumor cells with high expression correlating with a good prognosis in hepatocellular carcinoma (HCC) and a poor prognosis in glioblastoma. Here we show that 35% of primary human breast tumors stain positively for ELTD1, with 9% having high expression that correlates with improved relapse-free survival. Using immunocompetent, syngeneic mouse breast cancer models we found that tumors expressing recombinant murine Eltd1 grew faster than controls, with an enhanced ability to metastasize and promote systemic immune effects. The Eltd1-expressing tumors had larger and better perfused vessels and tumor-endothelial cell interaction led to the release of proangiogenic and immune-modulating factors. M2-like macrophages increased in the stroma along with expression of programmed death-ligand 1 (PD-L1) on tumor and immune cells, to create an immunosuppressive microenvironment that allowed Eltd1-regulated tumor growth in the presence of an NY-ESO-1-specific immune response. Eltd1-positive tumors also responded better to chemotherapy which could explain the relationship to a good prognosis observed in primary human cases. Thus, ELTD1 expression may enhance delivery of therapeutic antibodies to reverse the immunosuppression and increase response to chemotherapy and radiotherapy in this subset of tumors. ELTD1 may be useful as a selection marker for such therapies. IMPLICATIONS: ELTD1 expression in mouse breast tumors creates an immunosuppressive microenvironment and increases vessel size and perfusion. Its expression may enhance the delivery of therapies targeting the immune system.

Replication catastrophe induced by cyclic hypoxia leads to increased APOBEC3B activity.

Tumor heterogeneity includes variable and fluctuating oxygen concentrations, which result in the accumulation of hypoxic regions in most solid tumors. Tumor hypoxia leads to increased therapy resistance and has been linked to genomic instability. Here, we tested the hypothesis that exposure to levels of hypoxia that cause replication stress could increase APOBEC activity and the accumulation of APOBEC-mediated mutations. APOBEC-dependent mutational signatures have been well-characterized, although the physiological conditions which underpin them have not been described. We demonstrate that fluctuating/cyclic hypoxic conditions which lead to replication catastrophe induce the expression and activity of APOBEC3B. In contrast, stable/chronic hypoxic conditions which induce replication stress in the absence of DNA damage are not sufficient to induce APOBEC3B. Most importantly, the number of APOBEC-mediated mutations in patient tumors correlated with a hypoxia signature. Together, our data support the conclusion that hypoxia-induced replication catastrophe drives genomic instability in tumors, specifically through increasing the activity of APOBEC3B.

Hypoxia-induced SETX links replication stress with the unfolded protein response.

Tumour hypoxia is associated with poor patient prognosis and therapy resistance. A unique transcriptional response is initiated by hypoxia which includes the rapid activation of numerous transcription factors in a background of reduced global transcription. Here, we show that the biological response to hypoxia includes the accumulation of R-loops and the induction of the RNA/DNA helicase SETX. In the absence of hypoxia-induced SETX, R-loop levels increase, DNA damage accumulates, and DNA replication rates decrease. Therefore, suggesting that, SETX plays a role in protecting cells from DNA damage induced during transcription in hypoxia. Importantly, we propose that the mechanism of SETX induction in hypoxia is reliant on the PERK/ATF4 arm of the unfolded protein response. These data not only highlight the unique cellular response to hypoxia, which includes both a replication stress-dependent DNA damage response and an unfolded protein response but uncover a novel link between these two distinct pathways.

Interferon- and STING-independent induction of type I interferon stimulated genes during fractionated irradiation.

BACKGROUND: Improvement of radiotherapy efficacy requires better insight in the dynamic responses that occur during irradiation. Here, we aimed to identify the molecular responses that are triggered during clinically applied fractionated irradiation. METHODS: Gene expression analysis was performed by RNAseq or microarray analysis of cancer cells or xenograft tumors, respectively, subjected to 3-5 weeks of 5 × 2 Gy/week. Validation of altered gene expression was performed by qPCR and/or ELISA in multiple cancer cell lines as well as in pre- and on-treatment biopsies from esophageal cancer patients ( NCT02072720 ). Targeted protein inhibition and CRISPR/Cas-induced gene knockout was used to analyze the role of type I interferons and cGAS/STING signaling pathway in the molecular and cellular response to fractionated irradiation. RESULTS: Gene expression analysis identified type I interferon signaling as the most significantly enriched biological process induced during fractionated irradiation. The commonality of this response was confirmed in all irradiated cell lines, the xenograft tumors and in biopsies from esophageal cancer patients. Time-course analyses demonstrated a peak in interferon-stimulated gene (ISG) expression within 2-3 weeks of treatment. The response was accompanied by a variable induction of predominantly interferon-beta and/or -lambda, but blocking these interferons did not affect ISG expression induction. The same was true for targeted inhibition of the upstream regulatory STING protein while knockout of STING expression only delayed the ISG expression induction. CONCLUSIONS: Collectively, the presented data show that clinically applied fractionated low-dose irradiation can induce a delayed type I interferon response that occurs independently of interferon expression or STING signaling. These findings have implications for current efforts that aim to target the type I interferon response for cancer treatment.

Mitochondrial Inhibitor Atovaquone Increases Tumor Oxygenation and Inhibits Hypoxic Gene Expression in Patients with Non-Small Cell Lung Cancer.

PURPOSE: Tumor hypoxia fuels an aggressive tumor phenotype and confers resistance to anticancer treatments. We conducted a clinical trial to determine whether the antimalarial drug atovaquone, a known mitochondrial inhibitor, reduces hypoxia in non-small cell lung cancer (NSCLC). PATIENTS AND METHODS: Patients with NSCLC scheduled for surgery were recruited sequentially into two cohorts: cohort 1 received oral atovaquone at the standard clinical dose of 750 mg twice daily, while cohort 2 did not. Primary imaging endpoint was change in tumor hypoxic volume (HV) measured by hypoxia PET-CT. Intercohort comparison of hypoxia gene expression signatures using RNA sequencing from resected tumors was performed. RESULTS: Thirty patients were evaluable for hypoxia PET-CT analysis, 15 per cohort. Median treatment duration was 12 days. Eleven (73.3%) atovaquone-treated patients had meaningful HV reduction, with median change -28% [95% confidence interval (CI), -58.2 to -4.4]. In contrast, median change in untreated patients was +15.5% (95% CI, -6.5 to 35.5). Linear regression estimated the expected mean HV was 55% (95% CI, 24%-74%) lower in cohort 1 compared with cohort 2 (P = 0.004), adjusting for cohort, tumor volume, and baseline HV. A key pharmacodynamics endpoint was reduction in hypoxia-regulated genes, which were significantly downregulated in atovaquone-treated tumors. Data from multiple additional measures of tumor hypoxia and perfusion are presented. No atovaquone-related adverse events were reported. CONCLUSIONS: This is the first clinical evidence that targeting tumor mitochondrial metabolism can reduce hypoxia and produce relevant antitumor effects at the mRNA level. Repurposing atovaquone for this purpose may improve treatment outcomes for NSCLC.