"Science" is overweight: Swedish scientists release the world's first cancer pathology map! Deciphering the effect of all protein coding genes in human body on the prognosis of 17 common cancers
Release date: 2017-08-25
In Friday's "Science" magazine, researchers from the Royal Institute of Technology in Sweden released a "Human Pathology Atlas", which analyzed the data of nearly 8,000 patients and concluded that Of the 17 common cancers, how the expression of all protein-coding genes affects the prognosis and survival of different cancer patients [1].
The map they built contains more than 900,000 predictable patient survival curves and 5 million pathological images. With existing data, they can generate personalized models for patients to predict changes in protein levels and tumor metabolism. How will affect the development of the tumor, to carry out targeted treatment, improve patient prognosis and prolong the survival of patients. Scientists have taken another important step in the “dream road†of personalized treatment for cancer patients.
The author of the study is Professor Mathias Uhlén of the Royal Institute of Technology in Sweden, who is the chairman of the European Biotechnology Federation and the leader of the Human Protein Atlas (HPA) [2]. Professor Uhlén put the data in this "cancer pathology map" on http:// for other researchers and patients.
Professor Mathias Uhlén
In the past ten years, scientists have done a lot of work, established the Cancer Genome Atlas (TCGA), Human Protein Atlas (HPA), etc., which has accelerated our overall understanding of cancer. Before the start of the study, the researchers first extracted transcriptome and clinical data from 96.6 of 11,000 cancer patients from the TCGA database, covering a total of 33 cancers. They analyzed the expression levels of all protein-coding genes (19,571) in these 33 cancers and also analyzed their expression in 37 normal tissues of 162 healthy volunteers in the HPA program.
The researchers found that more than 5,000 genes are expressed in cancer patients and normal tissues, and more than 2,000 genes are expressed in cancer patients but are restricted in normal tissues. Most of them can regulate DNA replication, apoptosis, and Mitosis.
Next, the researchers selected 17 common cancers among the 33 cancers, including 7932 tumor samples. They found that individuals with the same cancer had a large difference in gene expression, and overall, the gene expression of the 17 cancers overlaped. This means that for some patients, their gene expression may be closer to other types of cancer, rather than their own cancer, which has certain guiding significance for targeted therapy. But the only exception to liver cancer is that it has little overlap with other types of cancer and has a relatively unique expression.
The overlap of different cancer gene expressions, the upper left gray is liver cancer
The researchers performed Kaplan-Meier survival analysis on clinical data and genomic data such as the survival time of each patient, and obtained the relationship between the level of single or multiple gene expression levels and survival, in order to identify the "prognosis gene". . A total of more than 100 million Kaplan-Meier analysis maps were generated from 17 cancers and 19,571 protein-coding genes, and these genes were finally classified according to the results of the analysis to facilitate prognosis (high expression is associated with longer survival) and is not conducive to A gene with a prognosis (high expression associated with shorter survival).
The proportion of favorable and unfavorable genes varies among different cancer types. It is worth noting that 2375 genes have opposite effects on prognosis in different cancers, which also indicates that functional studies of proteins expressing prognosis genes are necessary. Of these prognostic genes, only a few have been shown to have predictive value in patient outcomes in past studies, such as the RBM3 gene associated with colorectal cancer survival [3].
However, most of the genes have not been studied or valued before. Through this achievement, researchers in related fields can also conduct research on some genes to explore whether they play a leading role or just follow other Changes in gene expression.
However, if you want to apply it to the clinic, it is not realistic to analyze more than 10,000 genes at a time. Therefore, the researchers suggest that you can choose the five most beneficial genes and five unfavorable genes in each cancer. Perform survival analysis and make clinical predictions.
Among the 17 cancers, the cancer with a high 3-year survival rate (more than 95% in testicular cancer and prostate cancer) has a relatively small number of prognostic genes, which may be due to differences in patient survival. Large, so a larger cohort of research is needed to identify more prognostic genes. Since there are few, there must be more, liver cancer and kidney cancer are among the "leaders", they have 2892 and 6070 prognostic genes, and most other cancers do not exceed 2000. Among the 2892 prognostic genes of kidney cancer, 2629 were unfavorable prognosis, and the proportion was high.
The number of different cancer prognosis genes (the difference between the horizontal and vertical lines)
The researchers also examined the overlap of prognostic genes between different cancer types. For most cancers, they have "promising" prognostic genes, and none of the genes are shared by more than seven cancers. Among the prognostic genes, only the liver, kidney, lung and pancreatic cancers overlap significantly, but in the prognostic genes, the overlap of kidney, breast, lung and pancreatic cancer is more obvious.
The researchers found that most of the genes that are not prone to prognosis are related to cell proliferation, including mitosis and cell cycle regulation. They independently studied 314 genes involved in cell cycle regulation, of which 194 (62%) were unfavorable genes, with an increase in expression in at least one cancer. However, these genes are not shared among different cancers, which means that although they are all genes with unfavorable prognosis, targeting the same one is different in different cancers.
Taking liver cancer as an example, sharing with other cancers (light orange) or unique (rose pink) genes that regulate cell metabolism with unfavorable prognosis
The researchers further analyzed high-expression genes associated with overall high survival, and most of these prognostic genes were thought to be highly expressed in normal tissues and decreased in tumor tissues. Still taking liver cancer as an example, the researchers found that 66% of genes that are highly expressed in normal liver tissue are down-regulated in liver cancer or liver cancer cell lines, and their expression levels are inversely related to tumor grade.
According to the commonly used three-level classification method, tumors are divided into G1-high differentiation, G2-medium differentiation and G3-low differentiation according to the degree of differentiation. The higher the grade of tumor, the lower the degree of differentiation, the higher the degree of malignancy, the faster the growth. The worse the prognosis. The classification results show that the higher the expression level of these tissues, the better the prognosis of patients, which is consistent with the known concept that the higher the tumor differentiation level, the worse the patient's prognosis and the lower the survival rate.
The researchers also did some antibody testing, and they found that the amount of these antibodies can also be used as a predictor of prognosis. For example, cancer testis antigens are expressed in many types of cancer, and under normal circumstances, it will only appear in immune-immunity sites, such as testes and placenta. Taking liver cancer as an example, the researchers tested the expression of testicular antigen in biopsy tissue and liver cancer cell lines of normal liver and liver cancer patients, and found that the expression of biopsy tissue and liver cancer cell line in liver cancer patients was much higher than that of normal liver. .
In addition to these, the researchers combined this data with the previously used genome-scale metabolic network model (GSMM, which includes metabolites, biochemical reactions in genes and metabolism, can analyze metabolism, predict growth) for this study. Each sample in the sample establishes an individualized GSMM. Based on this model, researchers can study the metabolism of tumors, analyze the growth rate of tumors under different metabolic conditions, find substances that increase growth rate, find the enzymes needed to produce it, and control the growth of tumors by controlling the genes encoding enzymes.
The idea of ​​establishing individualized GSMM for each sample
In the end, the researchers identified 2,553 essential genes that inhibit or kill tumors, 55 of which are common in patients. However, most of these essential genes also affect the metabolism of normal tissues. When performing targeted tests, normal tissues are also harmed a lot. If it is in the human body, it will produce corresponding side effects. Therefore, proteins corresponding to these essential genes are not suitable as targets for drug development.
However, the researchers also predict that 32 genes involved in the nucleotide metabolism of cancer cells will not have much toxic side effects on healthy tissues, so they may become potential therapeutic targets. The establishment and analysis of this model is a way for patients to personalize treatment.
The researchers tested whether the prognostic genes they found were "reliable" in a group of patients with non-small cell lung cancer (357 patients) and a group of colon cancer patients (60 patients). In lung cancer patients, the 100 most influential prognostic genes they identified were up to 74% overlapping from the TCGA database. The researchers further selected eight of them and analyzed their expression. The results showed that their expression levels were significantly correlated with the prognosis. Similar results were obtained in experiments in colon cancer patients, and the protein expression levels of the six genes screened by them were all related to prognosis.
Professor Uhlén said: "This study is different from previous similar studies in that it does not focus on cancer-related mutations, but on the downstream effects of all protein-coding gene mutations. We show for the first time in big data. At the level, how the impact of gene expression levels will change medical research. We are pleased to provide cancer researchers around the world with a free and open access resource pool, hopefully this will help accelerate the search for biomarkers needed for personalized cancer treatment. "[4]
Reference material
[1] http://science.sciencemag.org/content/357/6352/eaan2507.full
[2] https://
[3] Hjelm B, Brennan DJ, Zendehrokh N, et al. High nuclear RBM3 expression is associated with an improved prognosis in colorectal cancer [J]. PROTEOMICS-Clinical Applications, 2011, 5(11â€12): 624-635.
[4] https://
Source: Singularity Network (micro signal geekheal_com)
GMP Certificated Immune Globulin Injection Supplier in China
Hepatitis B Immunoglobulin,Hep B Immunoglobulin,Hepatitis B Immunoglobulin Vaccine,Hepatitis Immune Globulin
FOSHAN PHARMA CO., LTD. , https://www.fspharmamedicine.com