- Results: Using targeted sequencing of cfDNA in a large cohort of solid tumor patient, sequencing reads harboring tumor-specific somatic mutations were isolated to examine the exact size distribution of tumor cfDNA. - For the majority of studied cases, 166 bp remained as the peak size of tumor cfDNA, with tumor cfDNA showing an increased proportion of short fragments (100-150 bp). - Less than 1% of cfDNA samples were found to be peaked at 134/144 bp and independent of tumor cfDNA purity. - Tumor cfDNA shortening occurred simultaneously at both 5 ′ and 3 ′ ends of the DNA wrapped around nucleosomes.. - Conclusions: Tumor cfDNA shortening exhibited two distinctive modes. - Tumor cfDNA purity and chromatin inaccessibility were contributing factors but insufficient to trigger a global transition from 166 bp dominant to 134/. - 144 bp dominant phenotype.. - The length of cfDNA is typically peaked at 166 bp, which is reminis- cent of the size of DNA wrapped around a nucleosome plus the linker [3, 4]. - Full list of author information is available at the end of the article. - cfDNA usually represents only a small proportion of the total cfDNA [14, 15], the greatest challenge to study the fragment size of tumor cfDNA is to differentiate tumor cfDNA in the presence of cfDNA shed from non- neoplastic sources (e.g., from hematopoietic cells). - Although a consensus is forming toward that tumor cfDNA is shorter than the healthy counterpart, little ef- fort was invested to address the exact extent of the shortening. - To our knowledge, the majority of publica- tions reported shortened tumor cfDNA fragment, but still retaining the peak at 166 bp within the size distribu- tion [17]. - However, it is still generally accepted that the modal size of tumor cfDNA is indeed between 130 bp and 150 bp, while the overall cfDNA size distribution peaking at 166 bp is the consequence of low tumor cfDNA purity in the abun- dance of cfDNA from non-neoplastic origin [11].. - Isolating sequencing reads carrying tumor-specific somatic mutations permitted us to separ- ate tumor cfDNA and to observe shortened but still 166 bp-peaked size distribution. - We also identified a small fraction of cases whose overall cfDNA size distribution displayed peaks at 134/144 bp instead of 166 bp. - These cases displayed higher-than-normal tumor cfDNA pur- ity, but tumor cfDNA purity was not a determinant for the occurrence of the extremely short cfDNA. - factor for tumor cfDNA shortening, but failed to estab- lish its correlation with the occurrence of the overall shift to 134/144 bp size distribution.. - Presence of 134/144 bp dominant samples was not determined by tumor cfDNA purity. - “166 bp dominant” as observed in Fig. - description of the fragment size distribution. - At first glance, the MSAF of the 134/144 bp dominant samples were significantly higher than that of the remaining 5573. - “166 bp dominant” samples (p-value <. - A closer investigation, however, revealed a fraction of 166 bp dominant samples presenting similar MSAF as 134/144 bp dominant samples (Fig. - To avoid bias toward either 134/144 bp dominant sam- ples or 166 bp dominant samples, we randomly selected 35,166 bp dominant samples with matching MSAF com- pared to the bp dominant samples, and found similar gender composition, age distribution, pri- mary tumor site, disease stage, metastatic status, muta- tion profile, and previous treatment history (not significant, Fisher’s exact test) (Table 1). - 2b, a size ratio threshold of 4 can be set to distin- guish 134/144 bp dominant samples from the 166 bp dominant samples. - We also estimated the tumor purity within the bp dominant samples and the randomly selected 35,166 bp dominant samples using a publicly. - Provided with the fact that the NGS enrichment panel interrogates only a limited fraction of the human haploid genome, tumor purity estimation based on VAF is prone to interference due to CNV and loss of heterozygosity (LOH). - Taken together, although 134/144 bp dominant cfDNA samples displayed signifi- cantly higher tumor purity than most of the routine cfDNA samples, counter examples with comparable tumor purity are readily identifiable within the 166 bp dominant cfDNA samples. - We thus conclude that tumor cfDNA purity is not the determinant factor for the oc- currence of 134/144 bp dominant cfDNA samples.. - chromatin inaccessibility in the 166 bp dominant samples Because the tumor cfDNA purity poorly explained the dominance by 134/144 bp fragments, we started to. - 1 Fragment size distribution of tumor cfDNA. - b-d Size distribution of cfDNAs harboring the BRAF V600E b, KRAS G12D c, or EGFR T790M d mutations from the 605 cancer patients compared to the size distribution of the corresponding WT cfDNA from the same patients or from the healthy controls. - 2 The presence of 134/144 bp dominant samples was independent of the tumor cfDNA purity. - a Pooled insert size distribution of 134/144 bp dominant cfDNA samples from solid tumor patients ( n bp dominant cfDNA samples from solid tumor patients (n = 35, randomly selected from 5608), and cfDNA samples from healthy controls ( n = 5). - b cfDNA size ratios bp/163 – 169 bp) showed poor correlation with the MSAF within the bp dominant and the 35,166 bp dominant cfDNA samples. - Spearman correlation ρ was labelled on the top right corner of the Fig. - c No statistically significant difference (t-test, two-tailed) was observed between the tumor cfDNA purity of bp dominant and the 35,166 bp dominant cfDNA samples estimated based on the CNV profile. - d Tumor cfDNA estimation based on MSAF and CNV profile showed minimal correlation. - Similar relationship between the DHS scores and the cfDNA size was also observed in the validation dataset, in which the WGS results of each cancer pa- tient sample was compared against the DHS signal of a cell line representing the patient’s primary tumor site, or against that of GM12878 when the DNase-seq results of the corresponding cell line were unavailable (Supplementary Figure S4). - We characterized the fragment size distributions of cfDNA from cancer patients based on their H3K9me3 and H3K4me3 modifications (using the ChIP-seq data of the A549 cell line [22. - Although the fragment size of cfDNA positively corre- lated with chromatin inaccessibility, it is also realized that none of the above tested conditions resulted in a 134/144 bp dominant size distribution or a size ratio higher than 4. - Intriguingly, the publicly available WGS result of a stage IV lung adenocarcinoma patient dis- played 134/144 bp dominant phenotype [25]. - In addition, tumor genomic regions with copy number loss were associated with larger cfDNA frag- ments while tumor genomic regions with copy number gain were concordant with smaller cfDNA fragments (Supplementary Figure S7), indicating that the cfDNA originated from the healthy cells within this patient remained at 166 bp dominant size distribution. - Taken together, we conclude that chromatin inaccessibility, as Table 1 Clinical demographics of the 134/144 bp dominant and the 166 bp dominant cfDNA samples. - 134/144 bp Dominant ( n = 35) 166 bp Dominant ( n = 35) p -value. - Fragmentation patterns of cfDNA with different sizes We then tried to determine which part of the DNA mol- ecule wrapped around the nucleosomes was cleaved to produce the 134/144 bp fragments. - We thus examined the positions of the 5′. - and 3′ fragmentation endpoints of cfDNA mapped near the first nucleosome immediately downstream of TSS for the WGS results of both 134/144 bp dominant and 166 bp dominant samples [25]. - neither the nucleosome positioning nor the shortening pattern was different between the 134/144 bp dominant sample and the 166 bp dominant samples (Fig. - As the cfDNA fragment shortened, the 5′ and the 3′ bound- ary simultaneously receded towards the center of the nu- cleosome at comparable rates for both groups of samples (Fig. - It is speculated that the shortened nucleic acids in 134/144 bp dominant cfDNA corresponded to histone H1.0-protected segments. - at 134/144 bp. - The 134/144 bp dominant cfDNA sam- ples exhibited higher tumor cfDNA purity and MSAF compared to the general cfDNA samples, but tumor cfDNA purity alone could not explain the size. - The 5 ′ and 3 ′ boundary of each sequencing reads of different lengths bp bp bp,151 – 160 bp, and 163 – 169 bp) were separately tallied in consecutive non-overlapping 5 bp windows within − 300 to 500 bp of the TSS, and normalized to the mean sequencing depth of the region. - fragmentation frequency in a four 166 bp dominant cancer patient cfDNA samples (SRX1921676, SRX1921677, SRX1921678, and SRX1921680) and b one 134/144 bp dominant cancer patient cfDNA sample (SRX1921679). - (C and D) Normalized 3 ′ fragmentation frequency in c four 166 bp dominant cancer patient cfDNA samples (SRX1921676, SRX1921677, SRX1921678, and SRX1921680) and d one 134/144 bp dominant cancer patient cfDNA sample (SRX1921679). - In each graph, vertical dashed lines marked the positions of the first peak of the curves downstream of TSS as fitted using locally weighted scatterplot smoothing (LOWESS). - distribution shift from 166 bp dominant to 134/144 bp dominant. - 144 bp dominant and 166 bp dominant samples in the clinical demographics or the mutational profiles within a panel of 422 solid-tumor-related genes, either. - With lim- ited availability of cases, we showed that the transition from 166 bp dominant to 134/144 bp dominant was glo- bal across the genome instead of being limited to spe- cific genomic regions.. - Given that the main focus of this study is to characterize the 166 bp dominant versus the 134/144 bp dominant phenotype within tumor cfDNA samples ra- ther than to investigate the size difference between tumor and healthy cfDNA, and the fact that the size dis- tribution of the healthy cfDNA was consistent with the literature, we decided to settle on the imbalanced cohort as part of this study.. - Although exploring the short size of cfDNA and con- sequently enriching tumor cfDNA has great clinical im- plications, the mechanism of the shortened tumor cfDNA was still elusive. - We therefore boldly postulate that the transition from 166 bp dominant phenotype to 134/. - 144 bp dominant phenotype requires the involvement of the host immune system recognizing foreign genetic ma- terial. - This hypothesis also explains why tumor cfDNA is processed to 134/144 bp within only a small fraction of cancer patients. - Overall, this large scale NGS analysis and characterization of cfDNA in solid tumor patients was the first recognition of two distinctive modes of tumor cfDNA shortening. - Sequencing reads of cfDNA carrying tumor specific somatic variants exhibited increased frac- tion of short fragment sizes, but the peak size was still 166 bp, identical to that of the cfDNA collected from healthy donors. - Within over 99% of the investigated solid tumor patients, the principal size of the total cfDNA size distribution was 166 bp with slight enrich- ment of short fragments of less than 150 bp. - of the examined 5608 cases, tumor cfDNA displayed a principal size at 134 bp or 144 bp. - Tumor cfDNA purity and chromatin inaccessibility were contributing factors toward shorter cfDNA distribution in cancer patients, but could not be established as the determinant for the occurrence of 134/144 bp dominant phenotype. - Further studies on the involvement of host immune system in processing 166 bp dominant cfDNA into 134/144 bp dominant cfDNA may provide insights in the etiology of cfDNA shortening.. - This study was approved by the Ethic Committee of Xiangya Hospital of the Central South University. - A sample size of 605 cancer patients and 5 healthy controls were used to study the size distribution of tumor cfDNA. - An additional 70 samples were used to study the shortening of tumor cfDNA. - The log2 fold change ratio of each 100 k bp window was calculated by dividing normalized read count of the tumor sample by that of its matched whole blood control.. - CNV analysis of targeted sequencing was performed through a customized algorithm, which is a part of the LDT. - The log2 fold change ratio of each target interval was calculated by dividing normalized read count of the cfDNA sample by that of its matched whole blood con- trol. - The 134/144 bp dominant cfDNA sample generally display a size ratio larger than 4.. - Estimation of tumor cfDNA purity using ABSOLUTE Segmentation information of the CNV profiles from the copy number analysis was processed by ABSOLUTE to estimate the fraction of tumor cfDNA within each sam- ple [19]. - The orientation of each read was ad- justed according to the orientation of the transcript.. - List of the gene symbols of the 422 genes targetable by the enrichment panel. - The presence of 134/144 bp dominant samples was independent of the tumor cfDNA purity. - (A) cfDNA size ratios bp/163 – 169 bp) showed poor correlation with the second high MSAF within the bp dominant and the 35,166 bp dominant cfDNA samples. - 163 – 169 bp) showed poor correlation with the third high MSAF within the bp dominant and the 35,166 bp dominant cfDNA samples. - The left panel showed the boxplot of the size ratio bp reads/163 – 169 bp reads) of cfDNA (lung cancer patient P2, SRX1921680) mapped to the 2000 bp region centered at the TSSs of genes with different expression levels according to lung cancer cell line A549. - The right panel showed the boxplot of the size ratio of cfDNA (a healthy control, SRX mapped to the 2000 bp region centered on the TSSs of genes with different expression levels according to bone marrow tissue (downloaded from http://www.proteinatlas.org/about/. - This graph shows a boxplot of the size ratio of cfDNA based on WGS data obtained from a publicly available dataset [1].. - cfDNA shortening to the 134/144 bp domin- ant state was a global event. - Red data points highlighted the regions of the top 100 highest size ratios. - The grey horizon- tal dashed line marked size ratio = 4, which is a threshold to separate the 134/144 bp dominant samples from 166 bp dominant samples. - (A) Fre- quencies of 5 ′ endpoint of the fragments (51 – 60 bp, 61 – 70 bp, 71 – 80 bp, 81 – 90 bp, and 91 – 100 bp) in consecutive non-overlapping 5 bp windows within − 300 to 500 bp of the TSS. - (B) Frequencies of 3 ′ endpoint of the fragments (51 – 60 bp, 61 – 70 bp, 71 – 80 bp, 81 – 90 bp, and 91 – 100 bp) in consecutive non-overlapping 5 bp windows within − 300 to 500 bp of the TSS. - In each graph, vertical dashed lines marked the positions of the first peak downstream of TSS of the curves as fitted using locally weighted scatterplot smoothing (LOWESS).. - All authors intellectually contributed to the study design and provided constructive editing advices to the first draft of the manuscript. - This study was funded by No.2017KTSCX046 of Characteristic Innovation Project of the Department of Education of Guangdong Province of China.. - The funding body played no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.. - Maternal plasma DNA sequencing reveals the genome-wide genetic and mutational profile of the fetus. - Cell-free DNA provides a good representation of the tumor genome despite its biased fragmentation patterns. - Analysis of the size distributions of fetal and maternal cell-free DNA by paired-end sequencing.. - Tissue-based map of the human proteome
Xem thử không khả dụng, vui lòng xem tại trang nguồn hoặc xem
Tóm tắt