@phdthesis{8720734,
  abstract     = {{Liquid biopsies are becoming more widely used for guiding clinical care of patients with cancer. With these liquid biopsies, information about the make-up of the tumor is obtained with a blood draw, urine collection or obtained from other biofluids, rather than by performing a classical and more invasive tissue biopsy. The information obtained from these biofluids can then be used to address multiple clinical questions, for example, is a cancer present? How is the cancer evolving under treatment? Is it possible that the cancer has relapsed? In this thesis, I investigated whether this approach can be beneficial for children with cancer. More specifically, I focused on DNA that is present in the blood of children with cancer (cell-free DNA, cfDNA including cell-free tumor derived DNA, ctDNA).
In a first phase, we performed a thorough review of the literature to assess the current status of the liquid biopsy field in pediatrics and what most important hurdles are before this technique can be used for daily clinical management. One of the key findings of this study was that most of the studies that report on liquid biopsies in children have a small sample size and are retrospective, with a lack of prospective studies.
Next, we investigated to what extent the copy number landscape of the tumor is represented in the cfDNA for multiple pediatric cancer tumor entities. Shallow whole-genome sequencing was performed on both the tissue DNA and the cfDNA, and agreement between the two techniques was evaluated. One of the strongest factors influencing the agreement between the liquid biopsy and the tumor was observed to be the cfDNA quality.
A second aim was to examine the potential of cfDNA in improving the diagnostic process of pediatric cancer patients. In pediatric oncology, there are several diagnostic challenges that are still unresolved. To address this question, we focused on methylation of cfDNA instead. However, up until recently, inferring the methylation pattern from cfDNA was either prohibitively expensive, or required extensive amount of cfDNA. We therefore applied a novel and affordable technique, termed "cell-free reduced representation sequencing" (cf-RRBS).
In a first step, an analytical computational workflow was designed to enhance the reproducibil- ity and the speed of the data-analysis of the cf-RRBS technique. Afterwards, we applied the cf-RRBS technique on 60 liquid biopsy samples from pediatric cancer patients, and we were successful in inferring the histopathological diagnosis from the cfDNA methylation pattern in 80% of the cases. Again, we observed a correlation between the cfDNA quality and the (mis)classification rate.
Further exploration of the pre-analytical variables revealed that the time between blood draw and centrifugation is the most important parameter that determines the cfDNA quality. Furthermore, we were able to show that if the blood sample cannot be processed within 24 hours, blood collection tubes, designed specifically for liquid biopsies, are a suitable alternative.
In ongoing research we are evaluating and benchmarking algorithms that allow the detection of the tumor fraction and tumor origin, with the goal of increasing the sensitivity and obtaining a robust classification even if the tumor fraction is below 10%. Finally, we are applying the methods and workflow that has been developed in this work to address a specific diagnostic challenge in pediatric oncology: obtaining a definitive diagnosis at the start of the renal tumor treatment.}},
  author       = {{Van Paemel, Ruben}},
  keywords     = {{cfDNA,liquid biopsy,pediatric cancer,methylation,cf-RRBS}},
  language     = {{eng}},
  pages        = {{VI, 200}},
  publisher    = {{Ghent University. Faculty of Medicine and Health Sciences}},
  school       = {{Ghent University}},
  title        = {{Expanding the diagnostic toolbox in pediatric oncology : exploration of the value of the cell-free DNA methylome}},
  url          = {{https://github.com/rmvpaeme}},
  year         = {{2021}},
}