This article is from the WeChat public account: Academic Headline (ID: SciTouTiao) , author: academic Jun, title figure from: IC photo
According to a study published today in the journal Nature, Stanford University researchers have developed a machine learning method that enables differential screening for patients with early lung cancer This method is based on the detection of tumor-derived DNA in a blood sample. Investigation, early and non-invasive .
Traditional lung cancer screening is generally recommended for high-risk groups for CT scans. This model has been shown to reduce lung cancer-related deaths. However, due to high costs, fewer screening programs, and concerns about false positives, the use of such screening is not high. Even in the United States, only about 5% of eligible individuals are screened for lung cancer by CT scan.
Blood testing based on liquid biopsy technology is a popular new cancer detection method, but Most liquid biopsies are often applied to patients with advanced cancer . After all, these groups Have higher levels of tumor-associated DNA markers in their blood than in early patients.
In the latest paper published by Nature today, Maximilian Diehn and colleagues from Stanford University, optimized an existing assessment of circulating tumor DNA (ctDNA ) sequencing method .
They improved DNA extraction and identified changes that are expected to be effective markers of disease. The researchers used this method to show that although ctDNA levels are low in patients with early-stage lung cancer, they are a powerful prognostic indicator.
Researchers then used this data to improve a machine learning method to predict lung cancer-derived DNA present in blood samples. This method can distinguish early lung cancer patients from risk-matched controls in an initial sample of 104 patients with early non-small cell lung cancer and 56 matched controls; in another independent validation consisting of 46 cases and 48 controls In the cohort, the researchers confirmed the above results.
Highly concerned liquid biopsy
In recent years, the performance of liquid cancer biopsies has attracted particular attention. As a branch of in vitro diagnostics, liquid biopsy can reduce detection hazards through non-invasive sampling, and it is efficient, accurate, and cost effective.
Untreated, cancer cells continue to divide and die under normal circumstances. When cancer cells die, they release DNA fragments into the blood. Learning to read this information can enable clinicians to quickly and non-invasively monitor the presence and size of tumors, the patient’s response to treatment, and the tumor’s response to treatment over time. development changes.
Currently, liquid tumor biopsy targets are circulating tumor cells (CTCs) , circulating tumor DNA (ctDNA) , circulating RNA (circulating RNA) and exosomes. Among them, ctDNA has attracted more and more attention due to its promising research prospects. ctDNA (circulating tumor DNA) is one of the free DNA (cell-free DNA, cfDNA) Class with characteristic markers, which can be identified, quantified and tracked by high-throughput sequencing technology.
Characteristics of ctDNA that have been discovered include site mutations, nucleosome occupancy rates, and differences in methylation modifications. Early detection of tumors, dynamic monitoring of tumor development and efficacy, and tolerance can be performed based on differences in these indicators. Drug testing, recurrence risk assessment and prognosis prediction.
Prof. Maximilian Diehn of Stanford University in the United States has stated that ctDNA can not only diagnose solid tumors, but also monitor treatment response, detect small residual lesions, and target treatment-resistant mutations, which may be the preferred noninvasive tumor screening method. “One of the exciting events in this area is that circulating tumor DNA can be applied to many different clinical situations.”
Combination of molecular technology and machine learning
In this latest study, researchers describe a method for analyzing circulating tumor DNA by deep sequencing (CAPP-Seq) To better enable early cancer screening and personalized analysis.
Researchers have found that although the level of ctDNA in early lung cancer is low, ctDNA already exists before most patients receive treatment, and its presence has a strong prognostic significance.
The team led by Maximilian Diehn and Ash Alizadeh conducted this study
< / p>
Researchers also found that most somatic mutations in free DNA (cfDNA) in lung cancer patients reflect clonal hematopoietic mutations (mutations From white blood cells) and is non-recurrent. Compared with tumor-derived mutations, clonal hematopoietic mutations occur on longer cfDNA fragments and lack the mutational characteristics associated with smoking.
Combining these findings with other molecular features, Researchers have developed and prospectively validated what is known as “lung cancer in plasma” (Lung cancer likelihood in plasma, lung-clip) ‘s machine learning algorithm can distinguish early lung cancer patients from risk-matched control groups .
Schematic diagram of the possibility of lung cancer in plasma (Lung-CLiP)
< / p>
ResearchPeople said that this non-invasive lung cancer screening method combines improved molecular technology with machine learning to detect the presence of lung cancer cell-derived cfDNA in blood samples, and can use plasma to detect a significant portion of early lung cancer.
Also unlike previous liquid biopsy studies that attempted to develop pan-carcinoma screening and analysis, the researchers this time focused on non-small cell lung cancer, using lung cancer-specific features to reduce unidentified confounding factors Impact on test results.
In addition, unlike previous studies that did not validate or use cross-cohort validation, the calcium study uses independent validation, which avoids the possibility of over-fitting the model and leading to overly optimistic results.
Researchers believe that a potential application of Lung-CLiP is as a preliminary screening for a high-risk population. Positive patients can be further tested and confirmed, which may increase the number of people who are screened for lung cancer each year and save more s life.
Paper title:
Integrating genomic features for noninvasive early lung cancer detection
Abstract of the paper:
Paper address:
https: //doi.org/10.1038/s41586-020-2140-0
This article is from WeChat public account: Academic Headline (ID: SciTouTiao) , author: academic Jun