Medical imaging can assess the tumor and its environment in their entirety, which makes it suitable for monitoring the temporal and spatial characteristics of the tumor. Progress in computational methods, especially in artificial intelligence for medical image process and analysis, has converted these images into quantitative and minable data associated with clinical events in oncology management. This concept was first described as radiomics in 2012. Since then, computer scientists, radiologists, and oncologists have gravitated towards this new tool and exploited advanced methodologies to mine the information behind medical images. On the basis of a great quantity of radiographic images and novel computational technologies, researchers developed and validated radiomic models that may improve the accuracy of diagnoses and therapy response assessments. Here, we review the recent methodological developments in radiomics, including data acquisition, tumor segmentation, feature extraction, and modelling, as well as the rapidly developing deep learning technology. Moreover, we outline the main applications of radiomics in diagnosis, treatment planning and evaluations in the field of oncology with the aim of developing quantitative and personalized medicine. Finally, we discuss the challenges in the field of radiomics and the scope and clinical applicability of these methods.

Fructose-1,6-bisphosphatase is shown to be depleted in clear cell renal cell carcinoma (ccRCC) and inhibits ccRCC progression by antagonizing glycolytic flux in renal tubular epithelial cells and by restraining cell proliferation, glycolysis, and the pentose phosphate pathway in von Hippel–Lindau-protein-deficient ccRCC cells by blocking hypoxia-inducible factor function. von Hippel–Lindau mutations occur in the vast majority of clear cell renal cell carcinoma (ccRCC) tumours, and these mutations result in stabilization of hypoxia-inducible factors. But this is not sufficient to cause the typical metabolic alterations found in ccRCC, nor sufficient for tumour formation. This paper reports that fructose-1,6-bisphosphatase (FBP1) was uniformly depleted in all of more than six hundred ccRCC tumours examined. FBP1 is shown to inhibit renal carcinoma progression through two different mechanisms. First, the enzyme antagonizes glycolytic flux in renal tubular epithelial cells, the presumptive ccRCC cell of origin, and this inhibits any potential 'Warburg effect'. Second, FBP1 restrains cell proliferation, glycolysis and the pentose phosphate pathway in ccRCC cells deficient in the von Hippel–Lindau protein (pVHL) by blocking nuclear hypoxia-inducible factor function. Clear cell renal cell carcinoma (ccRCC), the most common form of kidney cancer1, is characterized by elevated glycogen levels and fat deposition2. These consistent metabolic alterations are associated with normoxic stabilization of hypoxia-inducible factors (HIFs)3 secondary to von Hippel–Lindau (VHL) mutations that occur in over 90% of ccRCC tumours4. However, kidney-specific VHL deletion in mice fails to elicit ccRCC-specific metabolic phenotypes and tumour formation5, suggesting that additional mechanisms are essential. Recent large-scale sequencing analyses revealed the loss of several chromatin remodelling enzymes in a subset of ccRCC (these included polybromo-1, SET domain containing 2 and BRCA1-associated protein-1, among others)6,7,8,9, indicating that epigenetic perturbations are probably important contributors to the natural history of this disease. Here we used an integrative approach comprising pan-metabolomic profiling and metabolic gene set analysis and determined that the gluconeogenic enzyme fructose-1,6-bisphosphatase 1 (FBP1)10 is uniformly depleted in over six hundred ccRCC tumours examined. Notably, the human FBP1 locus resides on chromosome 9q22, the loss of which is associated with poor prognosis for ccRCC patients11. Our data further indicate that FBP1 inhibits ccRCC progression through two distinct mechanisms. First, FBP1 antagonizes glycolytic flux in renal tubular epithelial cells, the presumptive ccRCC cell of origin12, thereby inhibiting a potential Warburg effect13,14. Second, in pVHL (the protein encoded by the VHL gene)-deficient ccRCC cells, FBP1 restrains cell proliferation, glycolysis and the pentose phosphate pathway in a catalytic-activity-independent manner, by inhibiting nuclear HIF function via direct interaction with the HIF inhibitory domain. This unique dual function of the FBP1 protein explains its ubiquitous loss in ccRCC, distinguishing FBP1 from previously identified tumour suppressors that are not consistently mutated in all tumours6,7,15.

Tumor progression involves the ability of cancer cells to communicate with each other and with neighboring normal cells in their microenvironment. Microvesicles (MV) derived from human cancer cells have received a good deal of attention because of their ability to participate in the horizontal transfer of signaling proteins between cancer cells and to contribute to their invasive activity. Here we show that MV may play another important role in oncogenesis. In particular, we demonstrate that MV shed by two different human cancer cells, MDAMB231 breast carcinoma cells and U87 glioma cells, are capable of conferring onto normal fibroblasts and epithelial cells the transformed characteristics of cancer cells (e.g., anchorage-independent growth and enhanced survival capability) and that this effect requires the transfer of the protein cross-linking enzyme tissue transglutaminase (tTG). We further demonstrate that tTG is not sufficient to transform fibroblasts but rather that it must collaborate with another protein to mediate the transforming actions of the cancer cell-derived MV. Proteomic analyses of the MV derived from MDAMB231 and U87 cells indicated that both these vesicle preparations contained the tTG-binding partner and cross-inking substrate fibronectin (FN). Moreover, we found that tTG cross-links FN in MV from cancer cells and that the ensuing MV-mediated transfers of cross-linked FN and tTG to recipient fibroblasts function cooperatively to activate mitogenic signaling activities and to induce their transformation. These findings highlight a role for MV in the induction of cellular transformation and identify tTG and FN as essential participants in this process.

Vesicular structures called microvesicles (MVs) that are shed from the surfaces of cancer cells are capable of transferring oncogenic cargo to recipient cancer cells, as well as to normal cells, sending mitogenic signals that greatly enhance tumor growth. Because MVs are stable in the circulation, they also may have a key role in secondary colonization and metastasis. Thus, understanding how MVs are generated could have important consequences for interfering with cancer progression. Here we report that the small GTPase RhoA triggers a specific signaling pathway essential for MV biogenesis in various human cancer cells. Inhibiting the activity of different proteins comprising this pathway blocks MV biogenesis in the donor cancer cells and prevents oncogenic transformation in cell culture as well as tumor growth in mice. Although RhoA has often been implicated in human cancer, these findings now highlight a previously unappreciated role for this GTPase in malignant transformation, and demonstrate that blocking MV biogenesis may offer novel approaches for interfering with malignant transformation.

The heterogeneous nature of tumour microenvironment (TME) underlying diverse treatment responses remains unclear in nasopharyngeal carcinoma (NPC). Here, we profile 176,447 cells from 10 NPC tumour-blood pairs, using single-cell transcriptome coupled with T cell receptor sequencing. Our analyses reveal 53 cell subtypes, including tumour-infiltrating CD8⁺T, regulatory T (Treg), and dendritic cells (DCs), as well as malignant cells with different Epstein-Barr virus infection status. Trajectory analyses reveal exhausted CD8⁺T and immune-suppressive TNFRSF4⁺Treg cells in tumours might derive from peripheral CX3CR1⁺CD8⁺T and naïve Treg cells, respectively. Moreover, we identify immune-regulatory and tolerogenic LAMP3⁺DCs. Noteworthily, we observe intensive inter-cell interactions among LAMP3⁺DCs, Treg, exhausted CD8⁺T, and malignant cells, suggesting potential cross-talks to foster an immune-suppressive niche for the TME. Collectively, our study uncovers the heterogeneity and interacting molecules of the TME in NPC at single-cell resolution, which provide insights into the mechanisms underlying NPC progression and the development of precise therapies for NPC.