Can Ultra-low-pass Whole Genome Sequencing from Blood Plasma Detect Transformation to Malignant Peripheral Nerve Sheath Tumor in Patients with Neurofibromatosis Type I?

J. Szymanski,P. A. Jones,P. K. Harris,W. Feng,F. Qaium,C. I. Moon,Z. Zhou,T. A. Ball,A. Hirbe,A. A. Chaudhuri
DOI: https://doi.org/10.1016/j.ijrobp.2020.07.2095
2020-01-01
Abstract:Benign plexiform neurofibromas (PN) are common in patients with neurofibromatosis type I (NF1), and roughly half of these tumors will transition to malignant peripheral nerve sheath tumors (MPNST). Radiotherapy for MPNSTs can be effective, but currently imaging is often unable to distinguish MPNSTs from benign PNs. We hypothesize that tumor-derived copy number alterations are detectable in cell-free plasma DNA (cfDNA) from patients with NF1 using ultra-low-pass whole genome sequencing (ULP-WGS), and that these alterations can distinguish MPNST from its benign precursors. Blood was collected from NF1 patients with MPNST (n = 6) and NF1 patients with benign PN (n = 10) at several time points before and during treatment, and age-matched healthy donors at a single time point (n = 6). DNA was isolated from 2-8 mL plasma and sequenced on an Illumina Hiseq X10. Copy number alterations across the genome were quantified, and the proportion of tumor-derived cfDNA (tumor fraction) were calculated using ichorCNA (Broad). Additionally, genome-wide copy number alterations were segmented using CNVkit v0.9. MPNST and benign PN patients were monitored by CT or MRI at routine intervals and tumor sizes were quantified as area of greatest two dimensions. All samples had sequencing reads downsampled to 10 million and were informatically selected for fragment lengths between 95 and 150 bp. In downsampled reads, median sequencing depth was 0.6x; sequencing depth did not correlate with clinical status or tumor fraction (Fisher's p>0.1). Focal copy number loss in SUZ12, a mutation associated with MPNST, was observed in MPNST patient plasma (n = 2/6). As expected, focal copy number loss in NF1 was observed in all benign PN (n = 10) and MPNST (n = 6) patient plasma but not in any healthy donor plasma. Healthy donor, benign PN, and pre-treatment MPNST groups showed significant differences in tumor fraction by one-way ANOVA (p = 0.023), with MPNSTs harboring significantly higher levels of tumor-derived cfDNA compared to controls. Using a tumor fraction cutoff value of 2.5%, there was a significant relationship between tumor fraction and benign plexiform vs. MPNST status (chi-squared p = 0.05). On serial plasma sampling of MPNST patients, tumor fractions decreased significantly in response to external beam radiotherapy (EBRT), with post-EBRT levels being on average 2.4-fold lower than pre-EBRT (Student's T p<0.05). Alterations in copy number detectable in cfDNA by ULP-WGS can identify MPNST tumors in NF1 patients and to distinguish them from benign controls. This method has the potential to identify the transition from benign plexiform NF to MPNST and monitor response to radiotherapy, serving as an adjunct to imaging in this high-risk patient population.
What problem does this paper attempt to address?