TP53 Mutation in Circulating exoDNA Correlates With Outcomes of Patients With Hepatocellular Carcinoma

Background: Detection of tumor-specic mutations in exosomal DNA (exoDNA), a promising liquid biopsy material, has been used to assess the prognosis of hepatocellular carcinoma (HCC) patients. This study was the rst to use a droplet digital PCR (ddPCR) platform to detect tumor-specic mutations in circulating exoDNA and to evaluate the prognosis of HCC patients. Methods: Blood samples from 40 HCC patients were obtained between 2018 and 2019, with clinically annotated follow-up until 2020. A ddPCR platform was used to detect an HCC tumor-specic mutation in the TP53 gene. We analyzed the correlation between TP53 mutation detected in circulating exoDNA and patient clinical data. The ratio of mutant droplets/total droplets (MD/TD) was calculated according to ddPCR results. Results: TP53 mutations in circulating exoDNA were detected in 33 of the 40 patients (82.5%). Patients with high MD/TD (>62.5%) were more likely to show microvascular invasion (P=0.028) and high MD/TD predicted a shorter recurrence-free survival time (P<0.001). Conclusions: High MD/TD of TP53 detected in serum was associated with microvascular invasion and might be used to predict the prognosis of HCC patients. The diagnostic performance of detecting exosome-derived tumor DNA will likely improve when more mutations in other tumor-specic genes are combined. DdPCR: Droplet digital Polymerase Chain Reaction (PCR); MD/TD: The ratio of mutant droplets/total droplets; AFP: Alpha-fetoprotein; CfDNA: Circulating free DNA; MVI: Microvascular invasion; RFS: Recurrence-free survival.

patients as well as prognosis assessment. Genetic mutations have been traditionally identi ed by genome sequencing of biopsy tumor fragments or cfDNA, but this represents only a small fraction of tumor heterogeneity because cfDNA is released by dead tumor cells [6]. ExoDNA is more stable than cfDNA because of the lipid bilayer structure of exosomes [13,14], and exosomes contain larger DNA fragments, which is conducive to the detection of mutations. Additionally, because exosomes can be released from various cells in tumors, exoDNA in plasma may re ect tumor heterogeneity. Based on these advantages, increasing studies have been reported on tumor-speci c mutations in circulating exoDNA, such as EGFR mutations in lung adenocarcinoma patients and KRAS mutations in pancreatic cancer patients [9,11,15]. However, no studies have been reported on the detection of tumor-speci c mutations in circulating exoDNA in HCC patients.
In this study, here we report the rst investigation of the value and signi cance of detection of tumorspeci c mutations in circulating exoDNA in HCC patients using droplet digital PCR (ddPCR).

Patient characteristics
This study included 40 patients with primary HCC who were enrolled between October 2018 and August 2019 at the Second A liated Hospital of Nanchang University. The clinical characteristics of the 40 patients are listed in Table 1, and the preoperative information of the 40 patients is provided in Supplementary Table 1. Correlation between TP53 mutation in circulating exoDNA and clinicopathologic characteristics The c.747G>T mutation in TP53 has been identi ed as one of the hottest mutants in HCC patients [16][17][18].
And our previous studies have con rmed the presence of TP53 mutations in circulating tumor DNA in HCC patients [16]. Thus, these mutations might be detectable in exoDNA, which also contains abundant tumor cell-derived genetic information. We therefore selected the TP53 gene for analysis and detected TP53 gene mutation status in circulating exoDNA. The copy number of wild-type and mutant TP53 genes in all patients is shown in Supplementary Table 2. TP53 mutations were identi ed in 33 of 40 patients (sensitivity, 82.5%). The high sensitivity may be due to the wide source and stability of exosomes. We examined the correlation between TP53 mutation status and patient clinical characteristics but found no correlation with clinical characteristics.
The ddPCR analysis results are displayed in Supplementary Figure 1. According to the results of ddPCR, we calculated and analyzed the ratio of mutant droplets/total droplets (MD/TD). Notably, the MD/TD was associated with microvascular invasion (MVI) (P=0.035). ROC curve analysis showed that the best cut-off value of MD/TD to predict MVI was >62.5% (sensitivity 45%, speci city 95%, 95%CI [0.524-0.826], P=0.024); the detailed results are shown in Supplementary Figure 2. Multi-factor logistics regression analysis also demonstrated that high MD/TD was an independent risk factor for MVI (p=0.028). Thus, high MD/TD can predict MVI to a certain extent. However, other clinicopathological features such as cirrhosis, cancer thrombin, tumor size, and tumor number were not signi cantly associated with TP53 mutation status and MD/TD. The detailed results of these analyses are shown in Table 1.
Correlation between TP53 mutation rate in circulating exoDNA and survival We also analyzed whether TP53 mutation status in exoDNA could predict the prognosis of HCC patients who had undergone surgical treatment. We developed a regular and rigorous follow-up strategy of the 40 HCC patients from the day of operation to the date of tumor recurrence or up to 576 days. We rst compared the survival curves of patients with and without TP53 mutation in exoDNA and found no signi cant difference between groups (P=0.329). We analyzed the survival curves of patients with high and low MD/TD and found that the median recurrence-free survival was 61 days (range, 53-202 days) for patients with high MD/TD and 368 days (range, 51-576 days) for patients with low MD/TD (P<0.001, log-rank test, Figure 1A). This result showed that patients in the group with high MD/TD were more likely to relapse than those in the group with low MD/TD. We also performed survival analysis for patients with MVI according to MD/TD, and the results also showed that patients with high MD/TD were more likely to relapse than patients with low MD/TD (P=0.035, log-rank test, Figure 1B). Thirteen patients did not relapse during follow-up, and all of these patients showed low MD/TD.

Discussion
In this study, TP53 mutation and MD/TD in circulating exoDNA were detected to evaluate their diagnostic and prognostic value in HCC. Exosomes are secreted by a wide range of sources, including tumor cells [10], which is why exosome DNA is easier to detect than cfDNA. This study demonstrated that exosomes and puri ed DNA can be extracted from serum volumes as low as 250μL for the detection of TP53 mutations.
Our results showed that TP53 mutations were detected in 82.5% of patients, which is immensely higher than the previous detection rate for tumor-speci c mutations in cfDNA. We previously detected TERT, TP53 and CTNNB1 mutations in cfDNA of HCC patients, with a sensitivity of only 19.5% [16]. Other studies have shown that the different stages of cfDNA at between 30.8% and 57.9% [19]. In addition, a meta-analysis showed that cfDNA lacks robustness in the diagnosis of HCC [20]. This nding may be explained by the single source of cfDNA and the small number of circulating gene fragments produced from tumor tissues, with the total number of circulating DNA fragments <1.0% [21,22], and it makes traditional technology di cult to detect. No studies thus far have described the detection of TP53 mutations in circulating exoDNA in patients with HCC. Therefore, we used ddPCR to detect TP53 mutations in circulating exoDNA of HCC patients and explored the possibility of circulating exoDNA as a new non-invasive liquid biopsy method in the diagnosis and prognosis of HCC. Krimmel et al reported that the detection of TP53 mutations was related to age [23]. However, our study did not nd an association between age and TP53 mutation status.
Our results showed that patients with high MD/TD were more likely to show MVI. HCC patients with vascular invasion have a poor prognosis [24], and our study con rmed that patients with high MD/TD were more likely to relapse compared with patients with low MD/TD (P<0.001). Wang et al. reported that circulating tumor DNA correlates with MVI and predicts tumor recurrence of HCC [17]. These results are consistent with our ndings. We speculate that the association between MD/TD and vascular invasion may be because MVI makes it easier for nucleic acid fragments from the tumor to enter the circulation. Our ndings suggest that HCC patients with high MD/TD are more likely to show MVI and a poor prognosis. Therefore, MD/TD may be a candidate factor to predict the occurrence of MVI in HCC patients and thus to assess patient prognosis.
Like most tumors, HCC tumors show heterogeneity [25,26]. After the multiple stages of cell division and proliferation during tumor development, tumor cells undergo various genetic changes, resulting in aberrant growth rate, invasion ability, sensitivity to drugs, and other activities [27]. This may also contribute to the limitations of traditional biopsy. Notably, the detection of tumor-speci c mutations in circulating exoDNA overcomes obstacles associated with tumor heterogeneity. Detection of mutations in exoDNA is therefore a promising non-invasive liquid biopsy approach for early diagnosis, e cacy assessment, follow-up monitoring, and prognosis assessment in HCC patients. locally advanced and metastatic pancreatic cancer, respectively [9], with rates similar to our sensitivity detection results of 82.5%. In addition, Yang et al. demonstrated the value of circulating exoDNA as a rapid and low-cost method for identi cation of cancer-driving mutations in circulating exoDNA such as KRAS and TP53 mutations in pancreatic cancer patients and healthy individuals, respectively [11]. In contrast to these studies, we did not detect TP53 gene mutations in circulating exoDNA in healthy individuals, which also led to the lack of speci city in our experimental results. Uchiyama et al. reported that due to low exosome DNA concentrations, these samples may be in a "gray area" for detecting [28], which may be a potential limitation of circulating exoDNA as a liquid biopsy. Moreover, compared with other reports, we detected the TP53 mutation with a coverage area of only 87 bp, which made the detection of circulating exoDNA less effective as a non-invasive liquid biopsy.
To increase the value of circulating exoDNA detection in HCC application, we are currently planning a prospective study that will cover almost all HCC-speci c mutations. We will also improve the exoDNA extraction technology to improve the sensitivity of ddPCR, so as to improve the detection of low exosome DNA concentrations. We anticipate that this approach will provide bene cial information for the molecular evaluation of personalized HCC therapy. The detection of circulating exoDNA may enable further development of precision medical technologies to realize early diagnosis of HCC and to customize personalized treatment strategies.

Conclusions
At present, the diagnosis, treatment and follow-up monitoring of HCC are changing in the world. It is the diversi ed diagnosis and prognostic monitoring that makes it possible for patients with HCC to receive timely and effective treatment. In our study we proposed for the rst time a framework for uid biopsy of HCC patients by detecting tumor-speci c mutations in circulating exoDNA. Our results provided strong evidence that the mutation rate of the TP53 gene in circulating exoDNA was associated with MVI and might be used to predict patients with shorter recurrence-free survival. We also demonstrated that the detection of circulating exoDNA could overcome the limitations caused by tumor heterogeneity. The detection of tumor-speci c mutations in circulating exoDNA may be a reliable method for the diagnosis of HCC, and simultaneously detection of multiple tumor-speci c genes will improve the diagnostic performance. A future prospective study is required to validate the results of our preliminary study.

Sample collection
A total of 40 patients with primary HCC at the Second A liated Hospital of Nanchang University were enrolled between October 2018 and August 2019. All patients agreed to the analysis of exoDNA mutations in collected preoperative blood. Postoperative pathology revealed HCC. HCC patients were eligible if they agree to have their blood collected for detection of exoDNA before surgery. None of the HCC patients received preoperative cancer-related therapies. Written informed consent was obtained from all patients enrolled in the study. Blood samples were collected in EDTA tubes, and within 1 h, samples were centrifuged at 3000×g for 20 min. Serum samples were stored at -80ºC. This study was approved by the Ethics Review committee of the Second A liated Hospital of Nanchang University.

Exosome extraction and exoDNA extraction
Exosomes were extracted from serum samples using the ExoQuick™ kit (System Bioscience, CA, USA) following the manufacturer's instructions. We examined the exosomes by electron microscopy (Supplementary Figure 3A) and detection of the exosome-speci c protein CD63. Protein expression of CD63 was assessed by western blotting and three experiments were repeated to verify the presence of CD63 (Supplementary Figure 3B). ExoDNA was extracted from exosomes using the ExoDNA™ Extraction kit (BioVision Inc, SF, USA) following the manufacturer's instructions. We used this method to extract exosome DNA and the presence of DNA was con rmed in all samples.

Study design and droplet digital PCR (ddPCR)
DdPCR (also known as single-molecule PCR) generally consists of two steps: PCR ampli cation and uorescence signal analysis. Our experimental procedures followed the guidelines proposed for reporting digital PCR data [29]. The PCR primers and probe sequences were designed using Primer Premier 5.0 software. Each PCR reaction contained 10μl ddPCR Supermix for Probes (Bio-Rad, USA), 3.6μl of primer (Sangon Company, China), 1μl of probe (Sangon Company), 2μl of template DNA from exoDNA and 3.4μl of ddH2O to give a total volume of 20μl. The PCR conditions were 96°C for 10 min; 40 cycles of 94°C for 30 s and 60°C for 60 s; with a nal incubation at 98°C for 10 min.

Preparation and analysis of droplets
We added a new DG8 Cartridge (BioAssay Systems, CA, USA) into the holder and added 20μL of the samples into eight holes in the middle row. Next, 70μL drop-generated Oil (DG Oil) was added into the bottom row of eight cartridges in the same chamber. The gasket was covered and the holes on both sides were rmly hooked. The holder was placed in the QX200 droplet generator (Bio-Rad, CA, USA) for generation of droplets, which generally took about 2 min to complete. Droplets were produced in the top row of cartridges and subsequently transferred into a 96-well plate. The plate was placed on a heatsealing instrument and covered with a heat-sealing membrane for sealing. The QX200 Droplet analyzer was used for droplet analysis and detection, and the data were uploaded to the computer for nal analysis. After the reactions were completed, the threshold line was adjusted to the appropriate position according to the speci c reaction conditions for result interpretation. The 5' primer ends of mutant and wild-type genes were labeled with the uorescent dyes FAM (blue) and VIC (green), respectively, to assess the mutation status of the genotyped PCR product with the QX200 Droplet analyzer.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported by grant from the key research projects of Jiangxi Province, China (20181BBG70010). It contributed to study design and data collection, analysis and interpretation, and it also contributed to authorization of software and paper written.
Availability of data and materials The clinical data generated during the current study are available in medical system of The Second A liated Hospital of Nanchang University. The experimental data of ddPCR are from Sangon Biotech Company (Shanghai, China).