Diagnostic accuracy and safety of coaxial core-needle biopsy (CNB) system in Oncology patients treated in a specialist cancer centre with prospective validation within clinical trial data

Background: Image guided tissue biopsies are critically important in diagnosis and management of cancer patients. High yield samples are also vital for biomarker and resistance mechanism discovery through molecular/genomic analyses. Methods: All consecutive patients who underwent plugged image-guided biopsy at RM from June 2013 until September 2016 were included in the analysis. In second step, a second cohort of patients prospectively treated within two clinical trials (PROSPECT-C and R), were assessed for the DNA yield from biopsies required for complex genomic analysis. Results: A total of 522 plugged core biopsies were performed in 457 patients [52% men; median age 63 years (range 17-93)]. Histological diagnosis was achieved in 501/522 (96%) of performed biopsies. Age, gender, modality, metastatic site and seniority of the interventionist were not found to be significant factors associated with odds of failure on a logistic regression. Seventeen (3.3%) were admitted due to biopsy-related complications; 9, 3, 2, 1, 1,1 were admitted for grade I/II pain control, sepsis, vasovagal syncope, thrombosis, haematuria and deranged liver functions respectively; 2 patients with right upper quadrant pain after liver biopsy were found to have radiologically confirmed subcapsular haematoma requiring conservative treatment. One patient (0.2%) developed grade III haemorhage following biopsy of a gastric GIST tumour. Overall molecular analysis was successful in 89% (197/222 biopsies). Prospective validation in 82 biopsies revealed 92.06% and 87.3% success rate of DNA extraction and tumour content of >20% respectively. Conclusion: The probability of diagnostic success for complex molecular analysis is increased with plugged large co-axial needle biopsy technique, which also minimises complications and reduces hospital stay. High yield DNA acquisition allows genomic molecular characterisation for personalised medicine.


INTRODUCTION
Whilst cancer management and treatment options have significantly improved during the last few years, our knowledge and understanding about mechanisms of response and/or resistance to anti-cancer therapies remain relatively sparse. To date, this relative lack of understanding is partially due to difficulties in accessing prospectively collected tissue and studies showed that fresh frozen material from primary tumour resection specimens was associated with a tumour content of 60% [3]. Moreover, using FFPE DNA for large-scale genomic studies may demonstrate mutations that have occurred as a result of the fixation process, which All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint makes it difficult to distinguish real tumour variants from these fixation artefacts. Furthermore, low quality fragmented DNA can fail quality control in the pre-analytical stage impairing success rates.
Whilst a number of retrospective studies have demonstrated the safety and accuracy of diagnostic biopsies [4-6], data interpretation from such studies has often been hampered by small numbers, the lack of information on yield for molecular/genomic characterisation of tumours, and the lack of prospective validation. At Royal Marsden (RM) we have been using coaxial core-needle biopsy (CNB) system and a pre-formed gelatine sponge sealing device to conduct solid organ core biopsies in order to minimise the number of passes and reduce the risk of complications respectively. We present here the largest dataset demonstrating the safety and accuracy of this approach. Moreover, we took the opportunity to utilise a cohort of patients from two prospective clinical trials to validate tumour yields from biopsies in these translational studies.
deployed through the co-axial along the tract of the needle to facilitate haemostasis. The gelatin resorbs completely within 12 weeks.

Validation cohort
In the second step, a validation cohort of patients prospectively treated within two clinical trials was used to assess the DNA yield utilised for All participants in both studies were required to have mandatory pretreatment biopsies (6 cores), biopsies at partial response in PROSPECT-C and stable disease at 2 months in PROSPECT-R (6 cores) and at the time of progression (6-12 cores from two suitable progressing metastatic sites).

Prospective tissue collection procedures
Fresh frozen and FFPE tissue samples were obtained and plasma collection was conducted as per the study protocols at the clinically relevant defined time points. 16-gauge core biopsy was used to collect 3 or 4 fresh biopsy specimens and 1 or 2 specimens fixed in formalin and paraffin embedded. Within the trials, approximately 25% of the total length of a core was detached for primary culture and remaining ~75% of All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint the core was snap frozen and used for genomic analysis. One core was transported to establish tumour derived organoids and targeted panel validation [7]. One core was used for genomic analysis after being placed into cryovials and immediately snap frozen in liquid nitrogen. The remaining two cores were placed straight into formalin, and embedded in paraffin wax. Primary morphological and immunohistochemical analysis was performed by the histo-pathologist on the FFPE specimen, for confirmation of diagnosis. The samples were then stored in the GI & Lymphoma Research Bank of the RM, anonymised by trial number and time point.

Tissue sample processing
Biopsy cores were snap frozen in liquid nitrogen at the time of collection. Genomic (g)DNA and mRNA were co-extracted from cores using the Qiagen All-Prep kit. DNA was also isolated from whole blood samples using the Qiagen QIAamp DNA Blood Mini kit.

Whole exome sequencing
A minimum of 500 nanograms of gDNA was prepared for WES using the Agilent SureSelect Human All Exon v5 capture library, according to the manufacturers' protocol. The resulting libraries were sequenced to a mean depth of 100x using paired end 100 reads on an Illumina HiSeq 2500. High quality reads were aligned to the NCBI reference genome (hg19) using BWA (v0.7.12), and SAMtools (v0.1.19) to remove duplicates. Tumour content was estimated based on the CNVkit (v0.8.1) copy number profile. All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Table 1).

Statistical design
The success rate of biopsies was determined by the ability to perform standard molecular testing on tissue specimens, and safety determined by frequency of complications and extended hospital stay. Encrypted data was collected in a password protected Excel file and statistical analysis performed usingSTATA13. Chi-squared-analysis was undertaken to identify baseline characteristics that provided independent association with failure and success rates.  patients) and all other biopsies except liver, i.e. "others" (n=238 biopsies All rights reserved. No reuse allowed without permission.

Overall safety of image guided biopsies and cox regression analysis
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint from 228 patients). Success rates of 95.02% and 98.32% were observed in the two groups respectively (Supplementary Tables 3A and B).

Chi-squared tests to assess covariates of failure:
Results from chi-squared tests showed that the covariates of age category at earliest biopsy date, gender, modality of image guidance, metastatic site and seniority of the interventionist were not associated with the occurrence of failure. Association of site of biopsy (others vs. liver) however showed a significant trend in favour of others organs vs. liver, although the difference was not found to be numerically and clinically of significant impact  Table 4).

Validated genomic testing in patients with metastatic colorectal cancer (mCRC):
Given that mCRC patients underwent genomic profiling for clinically actionable mutations such as KRAS, NRAS, and BRAF analysis routinely with a clinically validated COBAS panel, we rationalized separating this cohort from the remaining patients. Of the total 144 patients with mCRC, 17 repeat charts and 38 patients who were referred from other hospitals were excluded. Of the remaining 89 patients, 2 (2.25%) had a failed molecular analysis due to insufficient DNA extraction-29 (32.58%), 6 (6.74%) and 3 (3.37%) were found to have KRAS exon 2-4, NRAS exons 2-4 and BRAF v600 mutations respectively.
Moreover, 36 patients were tested for TP53 and PIK3CA mutation; 26 (72.22%) and 5 (13.89%) were found to have these mutations respectively. These results All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17

gDNA extraction from biopsy: a cohort of PROSPECT-C and R studies
DNA was extracted from 62 biopsies taken from our prospective PROSPECT-C and R trials, and in 65% of cases; sufficient gDNA for WES was achieved from a single core. Two or three tissue cores were needed to yield sufficient DNA for 27% and 8% of the biopsy time points respectively (Supplementary Figures 1).
When required, utilising all available tissue cores allowed gDNA extraction rate of 100%. Tumour content was determined for 62 biopsies (75.61%) in the analysed cohort and was estimated as greater than 20% in 79.03% of cases ( Table 1).

Assessment of biopsy tumour content
All patients entering the PROSPECT trials were tested for KRAS/NRAS mutations in the archival tumour biopsy by standard COBAS methodology as this precluded entry into PROSPECT-C study [2]. As a result, all patients entering the PROSPECT-R trial had a catalogued KRAS/NRAS variant that could be used to investigate the tumour content of the respective biopsy samples [1]. Mutation profiles for KRAS/NRAS have previously been shown to be highly concordant between samples from the same colorectal tumour [8]. We therefore estimated the cancer cell content of biopsy samples using Sanger sequencing to detect the likely truncal KRAS/NRAS mutations identified previously by clinical sequencing assays. Samples were scored according to the following criteria: "high" tumour content if the variant base was detected at an intensity exceeding or equal to the wild type base; "medium" if the variant base was detected at more than 25% of the intensity of the wild type base; "low" if the variant base was clearly detected above background but at less than 25% of the All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint 1 4 intensity of the wild type base; "not detected" if the variant base could only be detected within the background noise or not at all (Table 2 and Figure 3).
Further cores were extracted and sequenced if the first had low or no detectable tumour content ( Table 2). In five cases, an additional core had medium tumour content where the first tested core was low/not detectable tumour content. Out of the 49 samples tested, 39 were scored as medium or high tumour content ( Table 2).
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

DISCUSSION
Tissue biopsies are often considered as the gold standard for diagnostic and research purposes; however, there are many logistical, technical and ethical challenges in the successful appliance of tissue biopsies in the clinic. To our knowledge, we present the largest dataset of tissue biopsies with a prospective validation cohort demonstrating high tumour yield and ability to perform genomic analysis via image-guided tissue sampling.
Biomarker discovery requires validation in prospective clinical trials; however, tissue collection procedures need to be optimised such that the valuable tissue obtained during trials is processed successfully [9, 10]. Moreover, even within a resourcefriendly environment, molecular profiling studies have often suffered due to inadequacy of samples; failure rates reportedly vary between 15-33% [11-14].
Keeping these issues in view, we ensured that pre-biopsy scans were discussed in person with a radiologist and only the most amenable lesions were chosen for pretreatment biopsies; experienced radiologists were then able to target multiple cores (6) from the periphery of the chosen lesions. The current study demonstrates that a strong infrastructure and good communication allows high quality tumour samples to be obtained in a time-efficient manner. The co-axial biopsy technique used at the Royal Marsden has the advantage of puncturing the capsule of solid organs (liver, kidney, spleen) only once, minimising the injury to normal tissues, improving patient experience and at the same time acquiring multiple large cores for diagnosis and molecular analysis. The application of pre-formed gelatin sponge sealing device at the biopsy tract provides a mechanical matrix that facilitates clotting. Gelfoam pledgets due their bulk, surface-acting hemostatic agents slow the flow of blood, protect the forming clot, and offer a framework for deposition of the cellular elements All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint of blood decreasing the risk of major bleeding [15]. The grade III hemorrhage in our series was only 0.2%, which compares favourably with the 0.5%-2% seen in large series in the literature using tru-cut needles with or without co-axial technique [16,17].
Common concerns about trials mandating research biopsies include the lack of patient understanding about the purpose of such studies, and the potential risks associated with additional interventional procedures within the research protocols [18][19][20][21]. In the case of our PROSPECT-C trial[1, 2], patients included could already access anti-EGFR antibody treatment via the cancer drug fund (CDF) independent of the research biopsy findings, which meant that the research biopsies were of no direct patient benefit. In order to ensure that patients clearly understood the purpose of their participation in PROSPECT-C and other research studies, a prospective patient based survey at the RM was performed. Remarkably, it showed that most patients who consented to a research biopsy gave an altruistic reason (e.g. to help research and/or others) as to why they agreed to participate [22]. A common concern regarding trial-related invasive intervention is procedure related complications.
Notably, the biopsy complication rates in more than 500 patients in our cohort (including patients on PROSPECT studies) were extremely low and compared favourably with published literature [23,24]. The technical reasons for success can be attributed to the use of large gauge co-axial needles, which enable multiple tissue cores to be sampled with a single pass. Subsequent application of gelatin foam pledgets via a co-axial cannula at withdrawal effectively seals the biopsy track and minimizes haemorrhage (<1%), thus enabling safe same-day discharge in the majority of patients. This technique however needs to be carefully considered in appropriate patients, for example, any attempt to biopsy lung paraenchyma would All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint carry a significant risk of the gelfoam pledget deploying in a pulmonary vein resulting in systemic embolus.
Following the safe acquisition of biopsy material, the processing of tumour samples has its own challenges. Firstly, the acquired sample contains a mixture of cancer cells and stroma (connective tissue, blood vessels and inflammatory cells). It is well established that stromal infiltration may lead to problems in interpreting genomic data [25,26]. In contemporaneous studies conducted at the RM (e.g. FOrMAT study), sample failure rates were high with only 16% of samples showing tumour content of more than 50% [27]. The FOrMAT study collected a range of GI tumour samples including pancreatic cancers, which are more likely to be dominated by inflammatory and stromal cells [28], but it relied on using only FFPE tissues. FFPE tissue has limitations for complex genomic studies as the DNA yield and quality is affected by the process of fixation and paraffin-embedding [29][30][31][32][33]. The PROSPECT studies benefited from parallel analysis using both FFPE and fresh frozen tissue, where the former was used for pathological assessment and the latter for molecular characterisation and genomic analysis. By utilising all available tissue cores as required, we achieved a gDNA extraction of >90% and an estimated tumour content of >20% in 87.27% of the cases. These data compare favourably with a recent large scale study comprising of >10,000 patients, who were subjected to a hybridizationbased NGS panel capable of detecting all-protein coding mutations, copy number alterations, and selected promoter mutations and structural re-arrangements[34].
We next took into account the limitations of tumour estimates generated by subjective pathological assessment of tumour morphology and cellularity estimates.
Cellularity can be estimated by quantifying the mutant alleles using technologies, such as Sanger or Ion Torrent sequencing, but this requires prior knowledge of the All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint 1 8 mutation [26,35]; in PROSPECT-R Sanger sequencing was used to assess tumour content, as RAS mutation was a pre-requisite for entry into the study. However, only patients with no known RAS pathway mutation could participate in PROSPECT-C, so alternative techniques were required for tumour cellularity estimates. An unbiased statistical approach that directly measures tumour content from the DNA sample was therefore, which allowed us to take into account factors such as tumour ploidy and ITH.
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

CONCLUSION:
Oncologic management and clinical trial participation require accurate histological and molecular characterisation. Image-guided biopsies using large gauge co-axial needles enable multiple tissue cores to be obtained with a single pass. This increases the probability of diagnostic success for complex molecular analysis.
Applying gelatin foam pledgets via the co-axial cannula following biopsy to seal the track reduces haemorrhagic risk and enables safe same day discharge in the majority of patients. By successfully obtaining sufficient number of tumour tissue samples within prospective trials, such studies can further understanding of tumour biology and help develop biomarkers of clinical and translational relevance.
Ultimately, this will enhance the application of personalised medicine in the clinic. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Science (New York, NY). 2018;359:920-6. All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Figure 1: Biopsy examples of patients within the study A)
A Computed tomography (CT) of a patient with a highly vascular retroperitoneal mass thought to be too high risk to biopsy at the local hospital. Surgery was also considered to be All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Figure 2: Maximum DNA yield of the whole analysed cohort from PROSPECT C and R patients
All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint 8 A) Cases with DNA yield >10ugr are plotted against right Y axis. B) cases divided according to their DNA concentration. Median value and with 95%CI represented in grey bars. In the small square cases with DNA concentration < 100 ng/ml are plotted against the left Y axis. All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint 0 author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint  author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Supplementary
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint 1 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.17.20065458 doi: medRxiv preprint Figure 1: Biopsy examples of patients within the study A) A Computed tomography (CT) of a patient with a highly vascular retroperitoneal mass thought to be too high risk to biopsy at the local hospital. Surgery was also All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint considered to be high risk of R1/R2 resection and a CT guided biopsy was recommended at our MDT. B) Biopsy was performed with a 15G/16G co-axial needle. The tract was plugged with16G Hunter plugs and there were no complications. The biopsy showed an Inflammatory Myofibroblastic Tumour which responded well on steroids and an operation was avoided. C) PET/CT of a 57yr old patient with relapsed Hodgkin's Lymphoma after 6 cycles of ABVD chemotherapy.
There was response in all sites of disease with the exception of a plaque of tissue behind the fundus of the stomach which appear FDG-avid on PET scan. A decision at the MDT was made to biopsy the lesion in order to exclude transformation of lymphoma. D) The 17G co-axial needle was placed medial to the left adrenal and above the splenic vessels adjacent to the lesion. Three cores were taken and the tract was plugged. There were no complications. The biopsy showed Hodgkin's Lymphoma which responded well to systemic therapy and consolidation RT. E) Coronal CT images of a 48yr old patient with a large tumour of the inferior vena cava (IVC) extending from the level of the renal veins to the right atrium. Occluded hepatic veins and ascites can be seen on the scan. F) The lesion was biopsied with a 15G/16G co-axial needle. G) The tract was plugged with three gelfoam pledgets. A diagnosis of leiomyosarcoma of the IVC was made and procedure had no complications.
All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint A) Cases with DNA yield >10ugr are plotted against right Y axis. B) cases divided according to their DNA concentration. Median value and with 95%CI represented in grey bars. In the small square cases with DNA concentration < 100 ng/ml are plotted against the left Y axis. All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04. 17.20065458 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.