Written by | Zheng Shiwei is responsible for editing | Xu Feng
Liquid biopsy has become an effective cancer detection tool due to its fast sampling and low invasiveness. The free DNA of cancer patients contains circulating tumor DNA (ctDNA), and by analyzing the free DNA of cancer patients, it is possible to detect genetic and apparent changes related to tumors. Pediatric oncology patients with large amounts of free DNA often have suboptimal clinical outcomes, and some earlier studies have further elucidated the value of liquid biopsy in disease surveillance.
Current analysis of free DNA in childhood cancer focuses on the detection of gene markers for tumors, such as fusion genes and copy number variants due to chromosomal translocations. The data, often from droplet-based digital PCR or targeted second-generation sequencing, is less sensitive and typically focuses on only a few gene fragments, and most childhood cancer samples have no repetitive sequence abnormalities in their genomes. But the researchers found that free DNA fragment patterns can provide very useful information, including the source of the target DNA, the structure of chromatin, and the apparent state of tumor cells. Therefore, analyzing the information contained in free DNA fragments can complement traditional genetic marker-based detection methods and help medical staff better develop personalized treatment plans.
On May 28, 2021, the Bock research group of the CeMM Research Center in Austria, together with the Tomazuou research group of the Child Cancer Institute in Santa Ana, Austria, published a paper entitled Multimodal analysis of cell-free DNA whole-genome sequencing for pediatric cancers with low in Nature Communications Mutational burden article. The researchers conducted deep sequencing of a total of 241 free DNA samples collected from 95 Patients with Ewing Sarcoma, 31 patients with other types of sarcoma, plus 22 healthy control groups, and performed a detailed analysis of the distribution of DNA fragments. Using bioinformatics and machine learning algorithms, the researchers summarized the distribution of free DNA fragments and related epigenetic characteristics associated with Juve sarcoma, and also established the clinical association between free DNA fragments and children's solid tumors.
The researchers first macroscopically counted the length distribution of DNA fragments. They found that the free DNA fragments in the Juve sarcoma sample were significantly shorter than those in the healthy control group, with a focus on 167 bp — which is similar to the length of DNA orbiting the unit nucleosomes. The researchers also found that the phenomenon of shorter lengths of free DNA fragments is not unique to Juve sarcoma, but is also present in many adult tumors (lung and colon cancer) and other pediatric sarcoma samples.
In addition, in many Patients with Juve sarcoma whose tumor markers were not detected, the free DNA fragment length was still significantly shorter than that of the healthy control group. In addition, the researchers also used the short length feature to find that the length filtering of free DNA fragments can improve the sensitivity of gene copy number variation detection in Juve sarcoma samples. Since the phenomenon of gene copy number variation in children's tumors is relatively uncommon, free DNA fragment length analysis may complement traditional detection methods and may be of great help for future clinical research and detection.
Proportional distribution of short free DNA fragments (20-150 bp) in children sarcoma and adult cancer samples (compared to healthy human control group)
In addition to macroscopic statistics, the researchers also focused on analyzing differences in enrichment of the length of free DNA fragments at different genomic loci. The researchers first divided the genome into consecutive intervals of 100 kb and then calculated the ratio of short fragments (100 to 150 bp) and long fragments (151 to 220 bp) within each interval (S/L ratio) and compared them to the results of the healthy control group. They found that in Juve sarcoma and other pediatric sarcomas, the S/L ratio of tumor DNA fragments was not evenly distributed across the entire chromosome, and that this differential distribution persisted after removing the interval with copy number variation.
To explore the biological significance of this differential distribution, the researchers then enriched the sequences of gene ranges with differential distributions with LOLA software, and the results showed that the sites with high S/L ratios (short fragment concentrations) matched with the H3K27ac ChIP-seq of the Juve sarcoma sample and the DNase I susceptibility sites specific to Juve sarcoma, showing that they were in an open state of chromatin, and this particular phenomenon was not detected in other types of tumor samples.
Contrast control group log2 (S/L) values in different chromosomal arms in Juve sarcoma, other pediatric sarcomas and healthy samples
Given that the uneven distribution of free DNA fragmentation throughout the genome reflects the chromatin structure of Juve sarcoma tumor cells, the researchers further explored the feasibility of detecting free DNA tumor sources under unknown genetic alterations. To do this, the researchers developed a software called LIQUORICE (liquid biopsy regions-of-interest coverage estimation) that compares free DNA fragments and target gene regions associated with a variety of different cancers and calculates free DNA coverage at different loci. When applying LIQUORICE to a sample of Juve sarcoma, the researchers calculated fragment coverage for the following four gene regions of interest: 1) DNase I chromatin open site specific to Juve sarcoma; 2) EWS-FLI1 binding site; 3) EWS-FLI1 positive correlation enhancer; and 4) EWS-FLI1 negative correlation enhancer.
In patients with Juve sarcoma, especially in samples where tumor markers can be detected, liquorice results showed a significant decrease in the coverage of free DNA at class 1-3 loci compared to other sarcomas and healthy controls. To confirm that the reduction in free DNA in the EWS-regulated region reflects the apparent state of the tumor, the researchers also performed methyl sequencing on a genome-wide scale on Juve sarcoma samples, and the results showed that the degree of methylation at these sites was significantly reduced in the sites where coverage decreased, confirming that the decrease in fragment coverage was associated with changes in chromatin structure.
Coverage of free DNA fragments from control groups and tumor samples in different gene ranges of interest
Finally, the researchers used machine learning methods to explore whether it was possible to predict Juve sarcoma cases from information hidden in free DNA fragments. Using the quantitative statistical methods of the three fragments mentioned in the article (macroscopic fragment size, fragment-wide range distribution S/L ratio, target genome interval fragment coverage), plus copy number variation as input data into the classifier, they found that the trained classifier model could accurately distinguish the free DNA of Juve sarcoma patients and healthy people (AUC value = 0.97). In addition, the sensitivity of post-training model classification is much higher than that of traditional genetics-based classification methods (e.g., detection of EWS-Ets fusion genes by whole genome sequencing), and the classifier can accurately distinguish Juve sarcoma sample information even at a sequencing depth of 0.1x.
The classifier predicts the ROC plot of the Juve sarcoma and healthy control group samples
Taken together, the study demonstrates new ideas for using information from free DNA fragments in liquid biopsies. Different from traditional tumor marker detection, the size of free DNA fragments, the differential distribution of fragment lengths across the genome, and the different coverage of the target gene range can effectively label Juve sarcoma samples. Through bioinformatics analysis, the researchers also fully demonstrated that free DNA fragments contain important epigenetic information. Since the gene mutation rate of many childhood tumors is lower than that of adult cancers, the analysis of free DNA fragment patterns proposed in this study has the potential to improve the efficiency of clinical detection and monitoring of children's tumors.
Writing
Editor-in-charge
make
Typography | Che Jie proofreads | uu