Translational Research

A data handling system dedicated to biobanking and biosample tracking will be developed. This database will be manufactured by xailabs GmbH in close cooperation with Charité – Universitätsmedizin Berlin and Medizinische Universität Insbruck.

The data handling systems will be accessible from virtually any internet-ready device, allowing the users to access the information they need and to which they have rights anytime, from anywhere, with no plug-ins, downloads or applets. Systems will fully respect the Good Clinical Practice (GCP) and national regulations.

Data security and storage: All data communication is encrypted, using advanced encryption technologies [Secure Socket Layer (SSL) for secure transmission, digital certificates for server identification]. All data is secured against hardware failures and systematic back-ups are made at defined intervals (daily, weekly and monthly back-ups).

Archival FFPE tumour tissues For all 133 patients included in the Phase II GANNET53 archival FFPE tumour tissue is available and centrally stored at Charité. All clinical partners have sent FFPE samples to P3 to allow Central Histopathological Review prior to study inclusion (FFPE samples were mandatory according to study protocol). Charité (P3) further prepared the samples for immunohistochemical staining (e.g., cutting of FFPE blocks). FFPE slides were sent to Medical University of Innsbruck and Medical University of Vienna, for preliminary companion diagnostics analysis. Core biopsies were performed from FFPE blocks for the development of Tissue Microarrays (TMAs) and sent to Medical University of Vienna for further analyses.

Blood fraction samples (plasma, serum, cell pellets): Blood samples (for blood fraction isolation) have been collected before treatment start and at different time-points during treatment in the Phase II GANNET53 study.

Time-points of blood sample collection (according to trial protocol): Blood collection was performed before and 24 hours after (+/- 3 hours) the administration of study drug on day 1 of cycle 1. On day 1 of cycles 2 and 3 blood collection was performed exclusively before study drug administration. After cycle 3, blood collection took place on day 1 of every other cycle (i.e. cycles 5, 7, 9 etc.) prior to administration of the study drug.

Impressively, in 103/133 (77%) enrolled patients a complete set of sequential blood samples per patient has been successfully collected (at all pre-specified time-points according the trial protocol and biomaterial collection manual). In only 30 patients one or two sequential blood samples are missing. In 68% of patients, sequential blood samples from at least 4 different time points are available.

More than half of included patients (51%, 68/133) had 4 to 5 sequential blood-samples taken during study treatment. Another 15% (20/133) had more > 5 sequential blood-samples collected over time. The distribution of patient numbers with respect to the numbers of sequential blood samples taken over time is shown below (one patient with screening-failure included, thus, num n=134):

Circulating tumour cells (CTCs) in the blood: Sequential blood samples of 128/133 patients of the Phase II GANNET53 trial were shipped Medical University of Vienna by the collecting clinical sites for further processing and CTC analysis. Samples were received from 11 different clinical centres in Austria, Belgium France and Germany. The average number of CTS shipments (2 Streck tubes per shipment) per clinical centre was 47. A total of 521 blood samples for CTC analysis were received. Impressively, an average of 4 sequential CTC blood samples per patient (range: 1 - 9) were successfully collected. All samples were processed at the partner site Medical University of Vienna, which involved enrichment for CTCs using the Parsortix technology. Cytospins and lysates of each sample were prepared for further analysis.

Biopsies of the actual relapse: Biopsies at time of study inclusion were taken in 29/133 (22%) patients. In 20 cases, 2 to 4 biopsies per patients are available; in 9 cases 1 biopsy is available. In 24 of 29 patients, biopsies were stored as fresh-frozen samples, in 5 of 29 patients biopsies were stored as FFPE samples. The figure below shows the rate of tumour biopsies taken from the actual relapse in patients enrolled in the GANNET53 Phase II clinical trial. The number of tumour biopsies taken is also depicted in the graph.

Ascites and pleura effusions: Ascites and pleura effusions were shipped from clinical partners directly to Medical University of Vienna. Fourteen ascites samples of eight patients, as well as eight pleura effusion samples of two patients were collected during the Second Reporting Period. All samples were processed at Medical University of Vienna: ascites cells were frozen in liquid nitrogen, ascites supernatant was stored, as well as cytospins. Furthermore, there was the attempt to cultivate all samples to create immortalized tumour cell lines for further experimental testing. Seven cell lines from 6 patients could be established. The number of patients from whom ascites was successfully collected was low. Indeed, this was expected, as most of the included platinum-resistant patients do not present with a high volume of ascites at the time of relapse, requiring paracentesis.

Literature research was carried out to gain information on assays applying the PLA principle. It was decided to use the Duolink Assay (Sigma-Aldrich). Further research was necessary to decide on the antibodies to be applied. We used information regarding already established and described assays, specificity of the antibodies and experience with them in immunofluorescence staining protocols. HSP90 antibody H-114 (Santa Cruz) and p53 antibody DO-1 (Merck Millipore) were chosen. Using suitable cell lines as positive control, antibodies were tested for their specificity and concentrations to be applied. This experience was then applied to the PLA technology. The assay was established using again different ovarian cancer cell lines. Antibody concentrations, incubation times, washing steps were optimized until sufficient sensitivity and specificity was achieved. Subsequently, the assay was applied on ovarian cancer cell lines with different p53 status, in parallel with staining for the targets HSP90, p53. Examples are shown in Figure 1.

Figure 1: On the top: Cell line CaOV4 showing a strong HSP90, p53 staining and a high number of PLA signals; On the bottom: Cell line A2780ADR showing a strong, but heterogenous HPS90 staining, no staining for p53, and only rare PLA signals;

During the GANNET53 study this assay should be applied on FFPE samples of primary tumor tissue as well as ascites samples. The assay was tested, optimized and could successfully applied at all samples types used in the GANNET study. Figure 2 shows the application of the assay on FFPE tissue (left) and ascites cells (right). Each red dot represents an event, indicative of an interaction of p53 and Hsp90.

Figure 2: PLA detecting HSP90, p53 interaction applied on; on the left: FFPE tumor tissue; on the right: an ascites sample

The methodology applied for p53 mutation analysis was established as part of task 6.1. We made use of a capture-based SureSeq procedure, offered by the company Oxford Gene Technology. This technology allowed determining the mutation status of p53, but additionally, also BRCA1, BRCA2, PTEN, ATM, ATR and NF1 were analysed.

Of 133 patients included in the trial, we received FFPE tissue from 131. Of 2 patients, the FFPE material was not sufficient or not available for shipment from Charité Berlin (P3) to the Medical University of Vienna (P5). DNA was isolated from all 131 available samples using the QIAgen FFPE DNA kit (Qiagen). In cases where the DNA quantity was not sufficient, additional FFPE sections were used for DNA isolation via Gene Read FFPE kit (Qiagen). The DNA concentration was determined by Qubit (Thermo Fisher Scientific) quantification, and the quality was assessed by fragment analyser (Advanced Analytical). Subsequently, the samples were divided into three categories depending on the DNAs quality and quantity. Samples with an available input amount of >500 ng and an average fragment length of >1.000 bp, were considered good quality. Samples only fulfilling one of those requirements were considered intermediate quality, and samples failing both were rated poor quality. The input amount of DNA for next generation sequencing and the number of samples per sequencing lane was adapted accordingly.

For seven samples, it was not possible to isolate DNA with sufficient quantity and quality, despite increasing the amount of FFPE material and using a kit specifically for FFPE DNA purification for subsequent next-generation sequencing. The average fragment length of those samples was 100bp and lower and the available DNA amount was less than 100ng.

In 115/124 (92.7%) of patients a TP53 mutation was detected, whereas 9/124 (7.3%) were found to carry TP53 wild type alleles. Of the detected mutations, 5 were single nucleotide variants affecting splice acceptor or donor sites, 21 were insertions or deletions resulting in a frameshift. Furthermore, we detected 5 in frame deletions, 4 synonymous mutations, and in the majority of cases (n=85) a single nucleotide variation resulting in a missense variant. In 4 tumours, more than one mutation was detected. These results fit well into reports from current literature. The Cancer Genome Atlas (TCGA) Research Network reported TP53 mutations in 96% of high-grade serous carcinoma specimens.