Circulating Tumor DNA as Marker for Response to Antineoplastic Treatment of Metastatic Cancer (FLUIDO)

Patients Patients with metastatic cancer who are treated with antineoplastic treatment are eligible if a molecular marker is either known due to the type of cancer (Case A) or if a molecular marker is known due to routine assessment (Case B). An example for "Case A" is pancreatic cancer, which is known to harbor KRAS-mutations as part of tumorigenesis in more than 90% (8), or multiple cancers harboring methylated WIF(9). A typical example for "Case B" is colorectal or gastric cancer. Systemic treatment requires the knowledge of the mutational status of RAS, EGFR and BRAF, which is assessed routinely from tumor tissue. If a mutation is found the patient qualifies for participation in the project. We finally plan to include at least 40 patients with mPDAC, another 40 patients with mCRC and 20 patients with mGC (100 patients in total at least).

• The inclusion criteria therefore are:
• Metastatic cancer (mPDAC, mGC, mCRC)
• Known mutation of the cancer
• Signed informed consent
• At least 18 years
• Eligible for antineoplastic treatment
• Patient treated at the Ordensklinikum Linz
• Exclusion criteria o Inclusion criteria not met

Treatment monitoring Circulating tumor DNA is analysed from peripheral blood. For this purpose, 30 ml blood is taken at start and during treatment additionally to the blood volume required for analyses in the frame of routine. Therefore, there are no extra blood sampling time points additionally to these required for routine care. Due to that, the time points depend on the cancer type investigated as treatment schedules are different. However, within an entity the time points are homogenous.

Analysis of ctDNA The preparation of ctDNA is done in the Laboratory for Molecular Biology and Tumor Cytogenetics at the Ordensklinikum Linz (Dr. Gerald Webersinke) and digital droplet PCR is performed by the Department of Genetics at the Medical University of Innsbruck (Prof. Johannes Zschoke). The analysis is performed in a batch and not in real time. In detail, ctDNA detection is based on KRAS-screening in mPDAC or by patient specific ddPCR. The latter patient specific ddPCR is based on the mutations found in the tumor by NGS-screening (ARCHER or Truesight 170 or whole-genome sequencing if panels are negative) at the Laboratory in Linz. This is the case for mGC and mCRC, where multiple mutations are possible (KRAS, NRAS, BRAF, TP53). In patients suffering from mCRC, this analysis in performed within routine procedures as such mutations are crucial for treatment decisions. In case of mGC mutations screening is done within the proposed project.

Radiology assessment Assessment of treatment response is performed as part of the routine treatment after two to three months of treatment at the department of Radiology of the Ordensklinikum Linz.

Statistics Data are analyzed in order to detect a dependency between ctDNA dynamics and response to treatment. Method for example is chi-square test. Software used is RStudio Version 1.2.5019.

Descriptive Statistics Mean, standard deviation, minimum and maximum for every concentration of ctDNA at each time point.

ROC-AUC analysis for every time point

• Dynamics of ctDNA concentration at every time point (metric variable)
• Outcome after 12 weeks as a dichotome variable (responder [SD/PR/CR], non-responder)
• Calculation of AUC with 95% CI (all time points)
• ROC-curve (all time points)
• Calculation of sensitivity and specificity for all time points with 95% CI
• Determination of the optimal threshold-value upon sensitivity and specificity for every time point
• Comparison of predictive value of each time point based on sensitivity/specificity, AUC by CI-values

Power-analysis Based on the current patient recruitment we assume to end up with a final number of 100 patients.

Assumptions:

• n = 100
• Power 80 %
• Significance-niveau 5 %, Confidence-Intervall (CI) 95 %
• Proportion of responders (estimated prevalence): 0.2, 0.4 and 0.6
• Sensitivity and specificity: 0.6, 0.7 and 0.8
• Dropout-rate: 15 %

Plan for analysis We first compare Roc curves at the different timepoints to identify the timepoint that as the best tradeoff between performance and gain of time (lack of power to provide statistical comparisons). In a second step, we identify the best cut point of ctDNA at the chosen timepoint based on youden index for instance and give sensitivity/specitivity and also NPV/PPV at this cut-point.