Abstract
Background/Aim: Gastrointestinal stromal tumours (GISTs) harbour genetic aberrations in receptor tyrosine kinase KIT (KIT) or platelet-derived growth factor receptor A (PDGFRA) in 85-90% of the patients. Circulating tumour DNA (ctDNA) is a potential biomarker in patients with GIST. Previous studies investigating ctDNA around surgery in patients with GIST presented divergent results regarding the proportion of patients with detectable ctDNA. This study aimed to 1) investigate the feasibility of detecting and monitoring ctDNA pre-and postoperative, 2) compare two different circulating free DNA (cfDNA) extraction methods, and validate results obtained by next-generation sequencing (NGS) using Real-Time PCR technology. Patients and Methods: Eight patients planned for immediate surgery or surgery after neoadjuvant oncological treatment were included in the study, from whom blood collection was performed pre- and postoperatively for ctDNA analysis. Furthermore, blood samples from six patients with GIST harbouring a point mutation in KIT or PDGFRA in tissues from primary tumours were used for comparison and validation sub-study. Results: In this explorative study, none of the patients with very low to intermediate risk GIST harboured KIT, or PDGFRA mutated ctDNA in pre-or postoperative blood samples. The methods used for cfDNA extraction gave similar output, and the two methods for ctDNA analysis gave identical results. Conclusion: There is no benefit in analysing ctDNA around surgery in very low to intermediate-risk GIST patients. Larger studies investigating ctDNA in patients with high-risk GIST around surgery are warranted.
Gastrointestinal stromal tumour (GIST) is the most common mesenchymal tumour of the gastrointestinal tract (1), with an annual incidence of 10-15 per million inhabitants (2). The majority of GISTs present a genetic aberration in receptor tyrosine kinase KIT (KIT) (81%) or platelet-derived growth factor receptor A (PDGFRA) (7%) (3).
Surgery is the primary and only curative treatment for patients with local GIST (4). Adjuvant treatment is offered to patients at high risk of progression (5) according to the modified National Institute of Health (NIH) consensus criteria (6). Imatinib, a tyrosine kinase inhibitor, is approved as an adjuvant treatment based on the results from a phase III, randomised, placebo-controlled, double-blinded trial investigating 713 patients resected for local GIST (7). The 1-year recurrence-free survival (RFS) was 98% for patients treated with adjuvant imatinib versus 83% for the placebo group of patients (p<0.0001). After surgery and during adjuvant treatment, patients with GIST are monitored with computed tomography (CT) scans every three to six months for about three to five years and subsequently annually depending on the risk status of the tumour (4).
CT scans have a limit of detection of 3 mm (8). A tumour of 1 cm3 contains 1×109 cells (9). Therefore, when the tumour has a size detectable on a CT scan, it contains a very large number of tumour cells, and a biomarker could therefore be important for earlier detection of relapse.
Liquid biopsy is emerging with the advantages of being less invasive than tissue biopsy, with the possibility for longitudinal and real-time monitoring of cancer (10) and independent of the anatomic location of the tumour. Furthermore, a liquid biopsy may be superior in showing tumour heterogeneity when compared to a tissue biopsy (10). The liquid biopsy includes the following groups of material with tumour origin: circulating tumour DNA (ctDNA), circulating tumour cells, and extracellular vesicles (10).
Circulating free DNA (cfDNA) is released into the blood through cell death (11). However, patients with cancer have been shown to have a higher cfDNA level compared to subjects without evidence of cancer (12). ctDNA constitutes a fraction of the cfDNA (13), is harboured solely by cancer patients, and therefore ctDNA may be a potential future biomarker for cancer diagnostics and treatment monitoring. However, how ctDNA can change the clinical practice in diagnostics and treatment monitoring has not been clarified.
To our knowledge, two studies (14, 15) have investigated ctDNA at the time of surgery of GIST with divergent results regarding the proportion of patients having detectable ctDNA. The usefulness of ctDNA in this setting is still unknown, and further clarification is warranted.
The aim of this study was to 1) investigate the feasibility of detecting and monitoring ctDNA pre- and postoperative, 2) to validate the cfDNA extraction using QIAamp® Circulating Nucleic Acid kit compared to Chemagic cfDNA 5k kit H24, and 3) to validate the next-generation sequencing (NGS) ctDNA results using Real-Time PCR, in patients with GIST.
Patients and Methods
A prospective, non-randomised, non-interventional, explorative study aiming to detect mutations in KIT or PDGFRA in DNA extracted from plasma samples was performed.
Patients. A total of eight patients with a GIST measuring ≥4 cm, planned for primary surgery or neoadjuvant treatment before the surgery at the Department of Surgery and Transplantation at Rigshospitalet, Copenhagen, Denmark, in the period from July 2019 to July 2021 were offered inclusion in the study.
Primary tumour size was measured on the CT scan and tumour specimens received during surgery. The only exception to this was the patients receiving neoadjuvant treatment (n=1), where the primary tumour size was measured on a CT scan only. Blood samples were collected preoperative and one day postoperative. Demographic information, date of surgery, histological examination, metastatic disease, and the date of the follow-up CT scans, including information regarding recurrent disease, were documented retrospectively from medical records. For quality assessment of the methods used for cfDNA extraction and the sequencing ctDNA analysis, blood samples from six patients with GIST harbouring a point mutation in KIT or PDGFRA in tissues from primary tumours were used.
The study protocol was approved by the Regional Ethics Committee (H-18029854) and the Head of Knowledge Centre on Data Protection Compliance (P-2019-706). The study was performed with the Good Clinical Practice standard, according to the latest revised Helsinki declaration, and according to national laws. All patients provided informed, written consent before inclusion. The date for data cut-off was the 31st of May 2022. The median follow-up time was 2.55 years (1.32-2.76 years). No patient experienced a relapse of GIST during the follow-up period.
Mutation analysis of DNA from tumour tissue. Mutation analyses on diagnostic tissue specimens were performed as part of a clinical routine using NGS with a primer panel covering all exons of the following genes: KIT, PDGFRA, NF1, BRAF, KRAS, NRAS, PIK3CA, PTEN, SDHA, SDHB, SDHC, and SDHD at the Department of Pathology at Rigshospitalet or Herlev Hospital, Copenhagen, Denmark.
Blood sampling. Thirty-six ml of peripheral blood was collected from patients for ctDNA analysis in ethylenediamine tetraacetic acid (EDTA)-coated blood tubes. According to Nationwide standard operating procedures, all blood samples were collected and handled through the Danish CancerBiobank, Bio- and GenomeBank, Denmark. The blood samples were processed and stored in a freezer within three hours from the collection of the blood samples. While the blood samples were still warm, they were centrifuged at 2,000 × g or 2,500 × g at 4°C for 10 min. The centrifugation was stopped slowly for 45 s. The fraction of plasma was transferred to another tube and plasma was stored at −80°C until analyses.
cfDNA extraction from plasma. The cfDNA extraction from the study patients’ plasma samples (study patient 1-8: 16 ml plasma) was performed at Herlev Hospital, Denmark, Figure 1. The cfDNA extraction from plasma samples from control patients 1-6 (control patients 1-3: 8 ml plasma, control patient 4-6: 16 ml plasma) was performed at Herlev Hospital and from plasma samples from control patients 1-3 (8 ml plasma) also at Naestved Hospital, Denmark.
Flow chart of included patients in the study of pre-and postoperative ctDNA and the quality assessment of the methods used for cfDNA extraction and ctDNA analysis. The cfDNA extraction and ctDNA analyses have been performed at the Department of Pathology at Herlev Hospital, Denmark (orange), and the Department of Pathology at Naestved Hospital, Denmark (blue). ctDNA: Circulating tumour DNA; GIST: gastrointestinal stromal tumour; cfDNA: circulating free DNA; QIAamp®: Circulating Nucleic Acid kit from QiaGen, Aarhus, Denmark; NGS: next-generation sequencing; Chemagic: cfDNA 5k kit H24 from PerkinElmer, Odder, Denmark.
At Herlev Hospital, plasma was thawed at room temperature, and cfDNA was extracted using the QIAamp® Circulating Nucleic Acid kit (Qiagen, Aarhus, Denmark) according to the manufacturer’s instructions. The cfDNA concentration was quantified using Qubit™ dsDNA HS and BR Assay Kit (Thermo Fisher Scientific, Roskilde, Denmark) on the Qubit fluorometer (Thermo Fisher Scientific) and stored at −80°C until analysis.
At Naestved Hospital, plasma was thawed at room temperature, and cfDNA was extracted using the Chemagic cfDNA 5k kit H24 (PerkinElmer, Ballerup, Denmark) according to the manufacturer’s instructions. The cfDNA concentration was quantified using Qubit™ dsDNA HS and BR Assay Kit (Thermo Fisher Scientific) on the Qubit fluorometer (Thermo Fisher Scientific) and stored at −80°C until analysis.
cfDNA library preparation and sequencing. GIST ctDNA libraries using 1-10 ng DNA were prepared manually using Ion AmpliSeq™ Library Kit 2.0 following the manufacturer’s protocol (MAN0006735 F.0, Thermo Fischer Scientific) with a GIST panel containing primers designed at AmpliSeq.com (Thermo Fisher Scientific) covering all exons of BRAF, KIT, KRAS, NF1, NRAS, PDGFRA, PIK3CA, PTEN, SDHA, SDHB, SDHC, and SDHD. Each library was uniquely barcoded with Ion Xpress™ Barcode Adapters 1-96 Kit (Thermo Fischer Scientific) and normalised to library concentration at 100 pM using the Ion Library Equalizer™ Kit (Thermo Fischer Scientific). Template preparation and chip loading were performed on the Ion Chef™ system using the Ion 550™ kit- Chef and loaded onto Ion 550™ Chips (Thermo Fisher Scientific) diluted to a final concentration of 50 pM. Sequencing was performed using the Ion S5XL™ Sequencer (Thermo Fisher Scientific).
NGS data analysis and variant classification. Sequencing data from the S5XL runs were initially processed using Ion Torrent Suite™ v5.16.1 (Thermo Fisher Scientific) and data quality-verified using CoverageAnalysis v5.16. Variant calling from the sequencing data was generated using Ion Reporter™ v5.16 with a custom workflow GIST_RH v5.16 and hg19 as the reference genome. To eliminate erroneous variant calls, a sorting filter was set with the following parameters: allele frequency >0.5%; alternate allele count >4; homopolymer length <5; locations in exonic, splicesite_3, splicesite_5; variant effect in missense, non-frameshift, frameshift, nonsense, stopless; and not in UCSC common SNPs. Each variant was visually examined using the software Integrative Genomics Viewer (IGV) (16). Variants at KIT or PDGFRA were reported.
Real-time polymerase chain reaction (PCR). Real-Time PCR was performed using the c-KIT mutation detection kit for Real-Time PCR (Entrogen/Triolab, Broendby, Denmark). The analysis was set up according to kit specifications, except for cfDNA template concentration. Some samples had below the recommended yield of DNA after extraction, therefore, the analysis was performed with 45 cycles instead of 40 cycles.
Results
Patient characteristics. Of the 16 included patients, the study ended up with eight GIST patients with complete biological material (Figure 1). Patient and tumour characteristics of study patients and patients included in the quality assessment are shown in Table I. Regarding the study patients, the median age at diagnosis was 78 years (range=52-92 years), and there was a slight predominance of women (62.5%). Only one patient (12.5%) had a tumour outside the stomach. The majority of the patients had a very low or low risk (75%) tumour, according to Armed Force Institute of Pathology (AFIP) criteria (17). One of the patients had undergone neoadjuvant treatment before surgery, and the same patient was offered adjuvant treatment after the surgery. The seven patients excluded from the study due to missing blood samples were classified as being at no risk (n=1), very low risk- (n=4), low risk (n=1), and intermediate-risk (n=1) GIST, according to the AFIP criteria (17). One patient ended with another diagnosis than GIST and was therefore excluded.
Patient and tumour characteristics at the time of inclusion in the study*.
The patients included in the quality assessment (n=6) harboured a known point mutation in KIT or PDGFRA in the tumour tissue and either locally advanced (n=2) or metastatic GIST (n=4).
The mutation analyses of DNA extracted from tissue specimens showed that six patients (75%) harboured a gene aberration in KIT exon 11, while the two remaining patients harboured a mutation in KIT exon 13 or PDGFRA exon 14, respectively.
cfDNA extraction. The mean cfDNA extracted pre- and postoperatively from plasma samples was 5,079.3 ng/ml and 1,987.0 ng/ml, respectively. Five patients (62.5%) had a lower cfDNA concentration postoperative than preoperative (Table II).
cfDNA concentration after extraction from plasma.
NGS analysis. The NGS performed on cfDNA from blood samples collected pre-and postoperatively detected no mutations in the patients’ tumours. One patient in this study received neoadjuvant treatment with imatinib. ctDNA was also negative in this patient before treatment started.
Quality assessment. As a quality assessment of both the cfDNA extraction method and the NGS results, blood samples from six patients with GIST harbouring a known point mutation in KIT or PDGFRA in the tumour tissue were used. At the time of blood collection from these patients, two were undergoing neoadjuvant treatment, while four had metastatic disease.
To validate the cfDNA extraction method, eight ml of plasma from three patients (control patients 1-3) was extracted at Herlev Hospital and the Department of Pathology at Naestved Hospital.
To confirm the results obtained using NGS, a Real-Time PCR was performed on nine samples: the extracted cfDNA from both sites (control patients 1-3) and three plasma samples solely undergoing cfDNA extraction at Herlev Hospital (control patients 4-6). The flow of the quality assessment is shown in Figure 1.
The results of the quality assessment performed on plasma samples from six patients with GIST are shown in Table III. The cfDNA extraction on control patients 1-3, performed at two different sites with different methods, showed a mean cfDNA at the Department of Pathology at Herlev Hospital of 377.3 ng/ml (262.0 ng/ml-550.0 ng/ml) and at the Department of Pathology at Naestved Hospital of 510.7 ng/ml (465.0 ng/ml-583.0 ng/ml).
Results from the quality assessment.
According to the manufacturer, the optimal cfDNA concentration for Real-Time PCR was 10 ng/μl. That was impossible to achieve with our samples, since the concentration ranged between 0.11 ng/μl and 10.4 ng/μl. Only one sample had a concentration >10 ng/μl. In the NGS, only one sample was ctDNA positive with mutations in KIT exon 11 (p.Leu576Pro) and KIT exon 17 (p.Asp816Val). In the Real-Time PCR, the negative controls did not show increase in fluorescence signal, and the positive controls showed increased fluorescent signal around cycle 30. Only one of the nine samples showed fluorescence at cycle 33 indicating a mutation in KIT exon 11 (p.Leu576Pro) and KIT exon 17 (p.Asp816Val), which fully agreed with the NGS results.
Discussion
ctDNA has potential clinical utility in diagnostics, prognosis, treatment selection, and monitoring during treatment (11). It is possible to monitor the disease in real-time due to the ctDNA’s short half-life of 35-114 min (18, 19). In this study, we investigated the feasibility of detection and monitoring ctDNA pre-and postoperative in patients with GIST and validated the methods used for both cfDNA extraction and ctDNA analysis.
The present study found a lower mean cfDNA concentration postoperative compared to preoperative. These results follow previous findings of patients with malignant gastrointestinal tract disease having more cfDNA than patients with benign disease (12). However, no patient in this study was ctDNA positive preoperative, postoperative, or before starting neoadjuvant treatment. This study’s follow-up time was 2.55 years, and no patient experienced a relapse. Thereby, no analyses of relapse have been performed. The patients’ risk statuses ranged from very low to intermediate risk-GIST; therefore, it could be proposed that patients with very low to intermediate risk GIST excrete no or very little amounts of ctDNA, and thereby mutations are undetectable with the used technologies.
There are several essential pre-analytical factors when analysing ctDNA; the blood collection tubes (20), time from blood collection to processing (20), storage temperature of the tubes until processing (21), centrifugation regime (22, 23), storage temperature after processing (24), number of freeze and thawing cycles (25), and method used for cfDNA extraction (26).
Regarding the blood collection tubes, two types exist: non-preservative and preservative tubes, where the major difference is the maximum tolerable time from blood collection to processing of the blood samples (20). The blood from the non-preservative EDTA tubes should be processed within four hours from blood collection (27), while blood from preservative tubes such as Streck tubes can await processing significantly longer (28). When processing immediately after the collection of blood samples, no significant difference in the amount of cfDNA extracted from EDTA and Streck tubes has been proven (29). Our study used non-preservative EDTA tubes, and all samples were processed and stored at −80°C within three hours of collection.
Regarding the storage temperature of the tubes until processing, the cfDNA extracted from plasma from EDTA tubes increases 24 h after blood collection at room temperature (21). At this time point, the amount of cfDNA extracted from Streck tubes is stable. However, if the blood samples are stored at 4°C, the amount of cfDNA is equal in the blood collected in EDTA tubes compared to that in Streck tubes for up to three days. The amount of cfDNA is stable in Streck tubes at room temperature for up to seven days (28). Our study stored the blood samples at room temperature (about 21°C) until processing. The time from blood collection to processing followed by storage in a freezer was ≤3 h.
Regarding centrifugation, there is no consensus on how to handle the blood samples. The recommendations tend towards double centrifugation, since the second centrifugation removes the cells that contaminate the plasma from the buffy coat, which may not be removed by the first centrifugation (22, 23). Our study centrifuged the blood samples at 2,000 × g or 2,500 × g at 4°C for 10 min, as at the time of our study, this was accepted as standard procedure.
Regarding the storage temperature after processing the blood samples, cfDNA is best preserved when plasma is stored at −80°C (24). However, if the plasma is stored for more than three years, the cfDNA amount decreases (30). In this study, the plasma samples were stored at −80°C and all analysed before three years.
Regarding the number of freeze and thawing cycles, it is recommended to store plasma in smaller aliquots to avoid multiple freezes and thawing cycles (25). Multiple freeze-thaw cycles of plasma lead to increased fragmentation of cfDNA, while extracted cfDNA does not show the same fragmentation when exposed to multiple freeze-thaw cycles (25). In our study, the plasma-EDTA was stored at −80°C until cfDNA extraction and following analyses.
Regarding the method used for cfDNA extraction, cfDNA can be extracted using silica membrane-based spin columns or magnetic beads, and the extraction can be performed either manually or automatically (26). The methods differ in recovery efficiency and the sizes of DNA fragments obtained (31). QIAamp® circulating nucleic acid kit, a silica membrane-based spin column purification kit, has performed best in several tests comparing different cfDNA extraction kits regarding the yield of cfDNA (31-33) and has also shown a high recovery efficiency of small cfDNA fragments (>75 bp) (31, 33). We assessed the quality of the method used for cfDNA extraction in this study through extraction with two different methods: a magnetic bead-based method at Naestved Hospital and a silica membrane-based method at Herlev Hospital. The amount of cfDNA obtained by the two methods was in accordance (Table III).
NGS makes it possible to do analyses of several areas of interest in one workflow. We selected this method for our study as it is routine at Herlev Hospital. To validate the results of the NGS analyses, we performed a Real-Time PCR known to have a higher sensitivity (down to 0.01%) compared to NGS (down to 0.5%) (34). The Real-Time PCR confirmed the results from the NGS analyses. In the quality assessment, one of the cfDNA samples had identical mutations detected both using Real-Time PCR and NGS, confirming that the NGS can be used for the detection of mutations if several mutation analyses should be performed in one workflow. The quality assessment performed on blood samples collected from other patients with GIST than the study patients confirm the reliability of the methods used for cfDNA extraction and ctDNA mutational analysis.
The amount of cfDNA is largely variable and can blur the picture when analysing ctDNA since ctDNA only constitutes a fraction of the total amount of cfDNA, sometimes even below 0.01% (13). The amount of cfDNA depends on several factors such as the patient’s age (35), exercise (36), infection (37), weight (38), and traumatic injury (39), which make a comparison of results across studies difficult.
In other types of cancer than GIST, detectable ctDNA has been shown to correlate to the tumour size (40-42) and disease stage (43). Other factors theoretically influencing the amount of ctDNA secreted from a tumour that have not been adequately investigated include tumour vascularisation, the primary tumour site, and mitotic rate (44).
Several studies have investigated ctDNA in patients with GIST. However, to the authors’ knowledge, only two studies, one by Kang et al. (14) and one by Johansson et al. (15), investigated ctDNA in patients with GIST using collected blood samples around surgery (Table IV). Both studies found a higher rate of ctDNA positive samples than our study. Our study included patients with local disease unlike the study by Johansson et al. (15), with primary tumour sizes ranging from 3.7 to 23.8 cm. Why the study by Kang et al. found a significantly higher rate of ctDNA positive samples cannot be explained.
Our study does not support a benefit of analysing ctDNA around surgery in patients with very low to intermediate risk GIST. The relevance of ctDNA in patients with high-risk GIST remains unclarified and future studies are warranted. However, before ctDNA can be used in clinical practice within any type of cancer, several issues need to be addressed, such as whether a detectable ctDNA/increase in ctDNA without CT-verified relapse or progression should lead to initiating oncological treatment or to a change of treatment. The clinical decision would in this case be solely based on ctDNA, excluding the possibility of evaluating the tumour on a CT scan during treatment.
Conclusion
Our results using blood and tissue samples from GIST patients indicate that patients with localised very low to intermediate risk GIST excrete no or undetectable amounts of ctDNA. The methods used for cfDNA extraction (QIAamp® Circulating Nucleic Acid kit and Chemagic cfDNA 5k kit H24) performed equally and ctDNA results obtained from NGS analyses were validated using Real-Time PCR assay with concordant results. Therefore, our results do not support analysing ctDNA around surgery for patients with very low to intermediate risk GIST. Future, more extensive studies investigating ctDNA in patients with high-risk GIST around surgery are warranted.
Acknowledgements
The Authors would like to thank the nurses at the Department of Surgery and Transplantation at Rigshospitalet, Copenhagen, Denmark, for handling the logistics during the study. The Danish CancerBiobank is acknowledged for biological material and data regarding handling and storage. The Authors would like to express their gratitude to Professor Niels Pallisgaard at the Department of Pathology at Naestved Hospital, Denmark, for participating in the quality assessment of the cfDNA extraction by extracting cfDNA at their department.
Footnotes
↵* These Authors contributed equally to this study.
Authors’ Contributions
Conceptualization, C.M.B., A.K.H., N.A.P., and E.H.; methodology, C.M.B., N.A.P., A.K.H., and E.H.; validation, C.M.B., E.H., T.S.P., and W.S.R.; formal analysis, C.M.B.; investigation, C.M.B., N.A.P., A.K.H., P.D.H., and L.P.; resources, A.K.H., N.A.P., P.D.H., and L.P.; data curation, E.H., and T.S.P.; writing – original draft preparation, C.M.B.; writing – review and editing, C.M.B., N.A.P., A.K.H., E.H., P.D.H., L.P., T.S.P., and W.S.R.; visualization, C.M.B., N.A.P. A.K.H., and E.H.; supervision, N.A.P., A.K.H., and E.H.; project administration, C.M.B. N.A.P., A.K.H., and E.H.; funding acquisition, C.M.B., and A.K.H. All Authors have read and agreed to the published version of the manuscript.
Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.
Funding
Candys Foundation (grant number 2019-332), the Fund for the Promotion of Medical Science (grant number 18-L-0161), Harboe Foundation (grant number 18250), Herlev & Gentofte’s Research Council, Beckett Foundation (grant number 19-2-3833), Agnes and Poul Friis Foundation (grant number 81008-003), Else and Mogens Wedells Foundation (grant number 11-20-1), Aase and Ejnar Danielsen’s Foundation (grant number 20-10-0045), and Toemrermester Joergen Holm and hustru Elisa f. Hansens Memorial Scholarship (grant number 20010) funded this research.
- Received August 23, 2022.
- Revision received September 7, 2022.
- Accepted September 8, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.