Abstract
Background/Aim: Extracellular vesicle DNA (EV-DNA) has emerged as a novel biomarker for tumor mutation detection using liquid biopsies, exhibiting biological advantages compared to cell-free DNA (cfDNA). This study assessed the feasibility of EV-DNA and cfDNA extraction and sequencing in old serum samples of patients with breast cancer (BC). Patients and Methods: A total of 28 serum samples of 27 patients with corresponding clinical information were collected between 1983 and 1991. EV-DNA was extracted using Exo-GAG kit (Nasabiotech) and cfDNA using QIAsymphony DSP Virus/Pathogen Midi Kit (Qiagen), respectively. Subsequently, 10 matched samples (EV-DNA n=5, cfDNA n=5) of five patients were subjected to sequencing using the Oncomine™ Breast cfDNA Research Assay v2 (Thermo Fisher Scientific). Results: Samples were collected on median 1.9 years after primary diagnosis [interquartile range (IQR)=0.2-7.2]. Median follow-up was 9.5 years (IQR=5.2-14.2). Median age of serum samples was 36.1 years (IQR=34.5-37.3). EV-DNA and cfDNA were extracted from 100% (28/28) of the included samples. Both, DNA quantity and concentration were comparable between EV-DNA and cfDNA. Sequencing was successfully performed in 100% (10/10) of the included samples. Two matched analyses yielded equivalent results in EV-DNA and cfDNA (no mutations, n=1; PIK3CA mutation, n=1), whilst in two analyses, PIK3CA mutation was only found in cfDNA, and in one analysis, a TP53 mutation was only found in EV-DNA. Conclusion: EV-DNA extraction and sequencing in old serum samples of patients with BC is feasible and has the potential to address clinically relevant questions in longitudinal studies.
Liquid biopsies offer the potential for non-invasive cancer detection, dynamic surveillance, as well as molecularly informed treatment adaptions and follow-up. The most widely used tools are circulating tumor cells (CTC) and cell-free DNA (cfDNA) (1-3). Extracellular vesicle (EV) DNA (EV-DNA) has recently emerged as informative cancer biomarker (4-8). EVs constitute a diverse and heterogeneous category of vesicles released by every cell type. They exhibit variable sizes and consist of a lipid bilayer membrane which protects against nucleases, enclosing nucleic acids, proteins, lipids, and metabolites (7). Numerous studies have demonstrated the existence of various DNA species within EVs, including single-stranded DNA (ssDNA)(9), double-stranded DNA (dsDNA) (8), and mitochondrial DNA (mtDNA) (10).
Several works have documented that DNA shed by EVs enables the detection of mutations, providing a reliable reflection of the mutational status in the originating tumor (4, 5, 11-15). Consequently, the utilization of EV-derived DNA for liquid biopsy purposes has started to be actively explored. Given that cfDNA and EV-DNA can be isolated from the same sample, it becomes pertinent to assess the advantages and disadvantages associated with each isolation method for liquid biopsy analysis. The utility of novel technologies, in this case EV-DNA sequencing, can be maximized if successful biobanking provides samples. These samples may offer the opportunity to conduct retrospective longitudinal analyses with long follow-up, allowing the evaluation of prognostic and predictive biomarkers based on the novel technology. We previously demonstrated that cfDNA can be extracted in sufficient quality and quantity to allow for DNA sequencing from old serum samples (16). However, it is still unclear whether EV-DNA can be extracted sufficiently and sequenced successfully. Moreover, limited studies provide insights into the comparison of both cfDNA and EV-DNA and no studies have so far investigated the utility of EV-DNA in breast cancer (BC).
Here, we assessed the feasibility of extracting EV-DNA and cfDNA with sufficient quality and quantity to perform sequencing in serum samples of patients with BC, which were obtained up to 38 years ago.
Patients and Methods
Serum samples were collected from patients with cancer in an oncological private practice in Basel, Switzerland. A total of 753 patients with cancer, of which 20.2% (152/753) were female patients with BC, had donated serum samples between 1983 to 1991. Patients with BC were enrolled after surgical removal of the primary tumor or after diagnosis of local/regional recurrences and/or distant metastases. A total of 10 ml of native venous blood were collected in a 10-ml BD Vacutainer blood collection tube (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and centrifuged in a Hettich centrifuge (Andreas Hettich GmbH & Co. KG, Tuttlingen, Germany) at 3075×g for 10 min. Next, the serum samples were transferred to three Nunc Cryogenic tubes (Thermo Fisher Scientific, Waltham, MA, USA) per patient and immediately frozen and stored at −70°C to −80°C. Primarily, the samples were stored at a private office for 9 years before being transferred to the Institute of Immunobiology in Freiburg, Germany until 1999, and subsequently relocated to the laboratory of Medical Genetics of the University of Basel, Switzerland where they have been stored until now. For each transfer, transportable refrigerating boxes were used to avoid thawing. DNA extraction for this analysis was performed from April 2021. The sample size was determined by the number of old samples with sufficient serum for both EV-DNA and cfDNA extraction. The number of matched samples for sequencing was deemed adequate for demonstrating the proof-of-principle of using targeted sequencing in these old samples. Clinicopathologic information was retrieved from individual patient files.
Informed consent was obtained from all participants included in the study before sample collection. This study was conducted in accordance with the Declaration of Helsinki and has been approved by the local ethics committee (Ethikkommission Nordwest- und Zentralschweiz, EKNZ 2018-00252).
Cell-free DNA extraction. Cell-free DNA was extracted from 28 serum samples of 27 randomly selected patients with BC using QIAsymphony DSP Virus/Pathogen Midi kit in a QIAsymphony robot (Qiagen, Venlo, the Netherlands) following the manufacturer’s instructions. The input volume was 1 ml and the final elution volume 30 μl. cfDNA concentration was determined using the Qubit dsDNA HS Assay kit (Invitrogen, Life Technologies, Carlsbad, CA, USA) on a Qubit fluorometer (Invitrogen). A volume of 2 μl of each cfDNA sample was diluted in 198 μl of Qubit working solution for the measurement.
Extracellular vesicle DNA extraction. EV-DNA was extracted from 28 serum samples of the same 27 randomly selected patients with BC, from whom cfDNA was extracted, using 1 to 2 ml of isolated serum. EVs were purified using the Exo-GAG kit (Nasabiotech, A Coruña, Spain), according to the manufacturer’s instructions. The serum sample was defrosted and centrifuged at 2,000×g for 5 min to remove cells and debris. Afterwards, the sample was centrifuged at 16,000×g for 15 min and DNA was extracted using the QIAamp DNA kit. DNA quantification and analysis were conducted as for cfDNA samples.
Somatic variants detection. The commercial panel Oncomine™ Breast cfDNA Research Assay v2 (Thermo Fisher Scientific) was used for the analysis. This panel is amplicon and unique molecular barcodes (UMIs)-based and covers relevant BC hotspot regions. The limit of detection (LOD) and quantification (LOQ) was defined for each genomic position using nine control data points from healthy donors. The extracted cfDNA and EV-DNA were amplified and tagged in triplicates. A single sequencing library for both fractions was prepared and sequencing was performed in an Ion S5 System (Thermo Fisher Scientific). The targeted coverage for each amplicon was 20,000×. Data analysis was performed as described earlier (17). A potential mutation was called if the variant read fraction (VRF) was higher than the LOD and the experimental sensitivity, whereas the latter was defined by the initial quantity of DNA (experimental sensitivity=2([DNA(ng)]/0.00649)−1 based on the assumption that a diploid cell can generate 0.00649 ng of cfDNA).
Statistical analysis. Descriptive statistics were used to assess patient, tumor, treatment, and survival characteristics using median and interquartile range (IQR) for continuous variables, and absolute frequencies and percentages for categorical variables. Median overall survival and invasive disease-free survival with corresponding 95% confidence intervals (CI) were established using the Kaplan-Meier method with right censoring at event, death, or the last follow-up. Statistical analyses were conducted using Microsoft Excel (Version 2308, Microsoft, Redmond, WA, USA) and R (version 4.2.2, R Core Team, Vienna, Austria) in the RStudio environment (18).
Results
For this study we randomly selected 28 serum samples of 27 patients with BC and a median follow-up of 9.5 years (IQR=5.2-14.2). Patient and tumor characteristics are summarized in Table I. Median age at diagnosis was 55.1 years (IQR=43.0-66.4). One patient had a bilateral cancer, with the majority of primary tumors being up to 5 cm in size (46.4%, 13/28) and node-positive (53.6%, 15/28). Treatment characteristics are summarized in Table II, with 7.4% (2/27) of patients having received neoadjuvant treatment, and 59.3% (16/27) adjuvant treatment. Recurrences occurred in 51.9% (14/27) of patients during the follow-up period and recurrence patterns are depicted in Table III. Median time to recurrence was 2.9 years (IQR=1.9-7.0), and median invasive disease-free survival 7.0 years (95%CI=4.3–not reached). Nine patients (33.3%) passed away during the follow-up period, with a median time to death of 12.8 years (IQR=9.4-14.4) and a median overall survival of 14.8 years (95%CI=12.8–not reached).
Patient and tumor characteristics at primary diagnosis.
Treatment characteristics at primary diagnosis.
Recurrence and survival characteristics.
Median time from primary diagnosis to collection of serum sample was 1.9 years (IQR=0.2-7.2). Serum samples were obtained from 29.6% (8/27) of patients at primary diagnosis, from 33.3% (9/27) of patients during or after adjuvant treatment, from 18.5% (5/27) of patients at or after locoregional recurrence, and from 18.5% (5/27) of patients at or after confirmation of distant metastasis (Table IV). Median age of serum samples for the present analysis was 36.1 years (IQR=34.5-37.3). cfDNA and EV-DNA were extracted from 100% (28/28) of the included matched patient samples. Both, DNA quantity and concentration were measured in cfDNA and EV-DNA (Figure 1). Median DNA quantity was 26.1 ng (IQR=15.2-52.8) for cfDNA, and 27.6 ng (IQR=19.2-67.1) for EV-DNA, whilst median DNA concentrations were 0.87 ng/μl (IQR=0.51-1.76) and 0.69 ng/μl (IQR=0.48-1.68), respectively (Table IV).
Twenty-eight serum samples with matched analysis of DNA concentration, DNA quantity, and sequencing using matched extraction of cell-free DNA and extracellular vesicles DNA.
Matched sample analysis of DNA concentration (A) and quantity (B) between cell-free DNA and extracellular vesicle DNA. DNA concentration and total DNA quantity showed no difference between cell-free DNA (cfDNA) and extracellular vesicles DNA (EV-DNA).
Subsequently, 10 matched samples of cfDNA (n=5) and EV-DNA (n=5) from five patients were randomly selected and subjected to sequencing. Median DNA quantity was 37.2 ng (IQR=26.5-63.0) for cfDNA samples, and 36.4 ng (IQR=32-66.4) for EV-DNA samples, whilst median DNA concentrations were 1.24 ng/μl (IQR=0.88-2.10) and 0.91 ng/μl (0.80-1.66), respectively (Table IV). All samples could be successfully sequenced, yielding a sequencing-success rate of 100%. Two matched analyses of patients with samples obtained at or after confirmed distant metastasis showed equivalent results, yielding no mutations with both techniques (patient #6) or PIK3CA mutations (patient #1) in cfDNA and EV-DNA (Table IV). In two cases of patients with samples taken during or after adjuvant treatment, PIK3CA mutations were found in cfDNA but not in EV-DNA (patients #9 and #27). Finally, in one patient who donated serum at primary diagnosis a TP53 mutation was only found in EV-DNA (patient #21).
Discussion
Successful biobanking offers the opportunity to subject old samples to analyses using novel technologies. Previously, we were able to show that cfDNA can be successfully extracted and cancer specific mutations to be detected in 30-year-old serum (16). In this study, we successfully performed EV-DNA and cfDNA extraction and sequencing in serum-samples, which were obtained up to 38 years ago. We found comparable DNA concentrations between EV-DNA and cfDNA, which were suitable to perform sequencing using Oncomine Breast cancer panel. We observed that PIK3CA mutations can be identified both in cfDNA and EV-DNA derived from these samples. Since the detection of PIK3CA mutations may be clinically relevant, our results indicate that the application of novel technologies, such as EV-DNA extraction and sequencing, on old samples is indeed feasible.
The role of biobanks in cancer research has been recognized decades ago (19). BC has become a malignancy with increased long-term survivorship, leading to an increased interest in long-term follow-up (20). However, recurrences may occur even after years of event-free survival, especially in HR+/HER2− BC (21). Therefore, biomarker-based longitudinal studies aim to address follow-up schemes in which recurrences may be detected before becoming clinically overt, identify treatment-resistance, and ultimately improve overall survival (22-24). Liquid biopsies offer a minimally invasive sample collection, with a magnitude of technologies aiming to address these goals (25). However, the applicability of novel technologies on samples stored over three decades needs testing, which is why we subjected old samples to EV-DNA isolation and sequencing.
Like cfDNA, EV-DNA can be obtained from bodily fluids and offers a non-invasive diagnostic option (26). Thereby analysis of EV-DNA can provide a comprehensive view on the genetic landscape of tumors, including tumor heterogeneity. EV-DNA analysis presents several advantages compared to cfDNA, as it exhibits greater stability compared to cfDNA, which has an estimated half-life of less than 2 hours (27). Moreover, DNA within EVs actively secreted from cells is protected from degradation within the intravesicular compartment (7). This could offer benefits in the analysis of archival samples (28-30). Nevertheless, EV-DNA tests lie in the absence of standardized isolation methods, in contrast to cfDNA. This deficiency necessitates additional steps for optimization, making the process technically more complex and requiring thorough clinical validation.
cfDNA represents DNA originating mainly from dying cells of different origin, including tumor cells (31). Circulating tumor cfDNA may represent as low as 0.01% of the total cfDNA. However, cancer cells may release more DNA than healthy cells, and targeted sequencing may still allow for a highly accurate mutation-detection (32, 33). However, in cases of limited levels of circulating tumor cfDNA, test accuracy may be low (34). It has been shown that tumor cells release more EVs compared to healthy cells (35). In light of the biological advantages of EV biomolecules as mentioned above, EV-DNA might therefore overcome the afore-mentioned limitations of cfDNA and increase the VRF and the number of mutations detected in liquid biopsies. Several approaches have combined both EV-RNA/DNA with cfDNA improving sensitivity and specificity compared to cfDNA alone (5, 12, 36). Interestingly, a non-oncological study demonstrated successful results analyzing exosomal DNA (Exo-DNA) and cfDNA from blood. Notably, while both approaches yielded positive outcomes, Exo-DNA exclusively achieved 100% sensitivity and specificity (37). In the oncological setting, EV nucleic acids have been shown to be superior to cfDNA if analyzed alone or in combination with cfDNA in non-small-cell lung cancer (4, 36), which was not found in pancreatic cancer (14), with conflicting results reported for patients with colorectal cancer (6, 15). Interestingly, both studies not showing beneficial results of EV-DNA analysis used plasma-extracted cfDNA, and showed lower concentrations of isolated EV-DNA compared to cfDNA (6, 14). To the authors’ knowledge, no study has so far focused on patients with BC. In our study, using old serum samples of patients with BC, we showed that DNA concentrations were similar for EV-DNA and cfDNA obtained from serum. These samples were successfully subjected to sequencing analysis using the Oncomine Breast panel. Whilst two PIK3CA mutations were only found in cfDNA, one TP53 mutation was only found in EV-DNA. Both, PIK3CA and TP53 mutations are amongst the most common somatic mutations across all BC subtypes. Whilst PIK3CA mutations are mainly present in luminal BC, TP53 mutations occur majorly in basal-like BC (38). Those samples harboring PIK3CA mutations were all obtained from patients with HR+ BC as was the sample harboring a TP53 mutation, whilst the sample without detected mutation was obtained from a patient with unknown HR status. It remains unclear whether mutations can be selectively detected in cfDNA or EV-DNA. Further studies should investigate whether both approaches could complement each other in detecting various mutations. If this is the case, the combined analysis of cfDNA and EV-DNA could prove more efficient than examining each fraction in isolation, as previously suggested by our research (13). Clinically, the detection of PIK3CA mutations may be relevant as treatment of patients with PIK3CA mutated, endocrine-resistant BC with Alpelisib showed an improved progression-free survival (39). Future endeavors may therefore focus on the combined analysis of serum contents, including CTCs, cfDNA, and EV-DNA as in the PROLIPSY study (NCT04556916), which was focused on the early detection of prostate cancer. This approach may be advantageous in cases where the quantity of circulating tumor biomass is low. In such cases, the use of EV-DNA has shown improved sensitivity and specificity for the detection of somatic mutations (40). Our results show that the application of novel technologies, in this case EV-DNA extraction and sequencing, on old samples is feasible.
Study limitations. This study has several limitations, including the limited sample size, of which only a subset was subjected to sequencing to show feasibility of detecting somatic mutations using both cfDNA and EV-DNA. This precluded any analyses of prognostic and predictive associations, which would have been of high clinical interest.
An important limitation of our study is the interpretation of the positivity in each fraction. While our data support the detection of mutations in both cfDNA and EV-DNA, at times, mutations were exclusively present in only one fraction. The significance behind this differential detection is not entirely clear and may be attributed to various factors. Moreover, it would be important to investigate whether positivity in each fraction could signify different prognoses. To draw solid conclusions, the application of this technology to larger cohorts and longitudinal follow-up studies are necessary. Additionally, we cannot exclude that some of the detected mutations may have been germline variants. However, the Oncomine panel covers only well-known somatic cancer specific hotspots, increasing the likelihood of detecting solely somatic mutations.
Finally, the isolation method employed in our study relies on the precipitation of EVs. Thus, the EV fraction is enriched, although cfDNA could potentially be trapped to a certain degree. Given the heterogeneity of particles and the absence of standardized methods for EV analysis and isolation, it becomes imperative to assess the specificity and sensitivity of our method in comparison to alternative approaches. This evaluation would provide a more comprehensive understanding of the reliability and efficiency of our isolation technique in the context of the diverse methods available for EV isolation.
Conclusion
Extracellular vesicle DNA extraction and sequencing, representing a novel technology in BC research, can be successfully employed on old samples of patients with BC. Therefore, EV-DNA analysis has the potential to address clinically relevant questions in longitudinal studies with long-term follow-up.
Acknowledgements
The Authors thank Michelle Attenhofer, Stefan Herms and Bettina Burger for their assistance with sample and data storage over the long period of time. Furthermore, the Authors thank the Basel Cancer League for paying the first freezer in 1983.
Footnotes
Authors’ Contributions
Study concept and design: Martin Heidinger, Salvatore Piscuoglio, Walter Weber, Walter P. Weber. Data acquisition: Martin Heidinger, Salvatore Piscuoglio, Miguel Ángel Navarro-Aguadero, Sara Sánchez, Marta Hergueta-Redondo, Miguel Gallardo, Santiago Barrio, Beatriz García-Peláez, Miguel Angel Molina-Vila, Ilaria Alborelli, Hector Peinado, Walter Weber, Walter P. Weber. Analysis and interpretation of data: Martin Heidinger, Daniel Egle, Salvatore Piscuoglio, Nadia Maggi, Marta Hergueta-Redondo, Ruth S. Eller, Sara Sánchez-Redondo, Julie M. Loesch, Santiago Barrio, Hector Peinado, Walter Weber, Walter P. Weber. Manuscript preparation: Martin Heidinger, Daniel Egle, Salvatore Piscuoglio, Ilaria Alborelli, Hector Peinado, Walter Weber, Walter P. Weber. All authors helped draft the work, revised it critically for important intellectual content, and read and approved the final version to be published.
Funding
This work was partially funded by the “OPO-Stiftung” and the “Friedrich Locher-Hofmann-Stiftung. None of the funders had any role in the design of the study; in the collection, analysis, and interpretation of the data; in writing the manuscript.
Conflicts of Interest
Walter P. Weber received research support from Agendia paid to the University Hospital Basel and honoraria for lectures from MSD. All other authors report no conflicts of interest.
- Received May 9, 2024.
- Revision received May 21, 2024.
- Accepted May 22, 2024.
- Copyright © 2024 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.