Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

Early detection of uterine corpus endometrial carcinoma utilizing plasma cfDNA fragmentomics.

Author(s): Liu J, Hu D, Lin Y, Chen X, Yang R, Li L, Zhan Y, Bao H, Zang L, Zhu M, Zhu F, Yan J, Zhu D, Zhang H, Xu B, Xu Q

Publication: BMC Med, 2024, Vol. 22, Page 310

PubMed ID: 39075419 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   This study developed a prediction score for the detection of uterine corpus endometrial carcinoma (UCEC) based on the next-generation sequencing (NGS) fragmentomic profile of cell-free DNA (cfDNA)from plasma. The authors evaluated the sensitivity and specificity of the prediction score and investigated the potential effects of tumor fraction, cancer grade, specimen transport (on ice or at room temperature for 24, 48 or 72 h), duration of frozen storage (3 days, 7 days or 1 month) and the timing of collection relative to eating/exercise on the prediction score. To develop and test the model, EDTA blood was collected from 111 UCEC patients and 111 healthy volunteers. To test the effects of preanalytical variability (transport duration/temperature, timing of collection and frozen storage), EDTA blood was collected from 21 healthy volunteers and 4 patients with UCEC. Unless otherwise specified, plasma was separated by centrifugation at 16,000 g for 10 min within 4 h of collection. DNA was extracted from plasma using the QIAamp Circulating Nucleic Acid Kit without carrier RNA and quantified using the Qubit dsDNA HS Assay Kit. Sequencing libraries were prepared using the KAPA Hyper Prep Kit and pair-end sequenced using a NovaSeq instrument. Reads were trimmed using Trimmomatic and aligned to the human reference genome (GRCh37/UCSC hg19) using the Burrows-Wheeler Aligner and, when needed, down-sampled to 5X coverage. Copy number variation, fragment size distribution and GC-corrected nucleosome footprint profiles were used to construct a learning model to identify specimens from patients with UCEC versus healthy volunteers. Tumor fraction was calculated using ichorCNA. Plasma specimens collected from 21 healthy volunteers and 4 patients with UCEC were used to assess preanalytical effects on the prediction score. The prediction scores of plasma that was separated and analyzed within 2 h were compared to plasma that was (i) transported for 24 h, 48, or 72 h at room temperature or with an ice pack, or (ii) stored frozen at an unspecified temperature for 3 days, 7 days, 1 month, or 6 months. Plasma from the same 21 healthy volunteers and 4 UCEC patients were also collected multiple times to determine effects associated with the timing of blood collection relative to meals and exercise.

   Summary of Findings:

   The predictive power of the prediction score, which was calculated using the authors’ automated machine learning model for discrimination of UCEC from healthy specimens based on the cfDNA fragmentomic profile, was very high, with an area under the curve (AUC) of 0.991 in the training cohort (66 UCEC and 67 healthy) and 0.994 in the validation cohort (44 UCEC and 45 healthy).  Using a prediction score threshold of 0.423 for the identification of cancer resulted in a specificity and sensitivity of 95.5% and 98.5%, respectively, in the training cohort and 95.5% and 97.8%, respectively, in the validation cohort. As expected, the prediction score was significantly higher in specimens from patients with UCEC than those from healthy patients (P<0.001, both cohorts). The prediction score was modestly correlated with tumor fraction (R=0.4, P=1.5e^-5). Cancer score was non-significantly correlated with cancer stage and did not differ significantly among different tumor grades. Further analysis of the fragment size distribution identified chromosomes 1, 10, 6, 11, and 17 as having a high number of differences between specimens from healthy individuals and those from UCEC patients and the authors state these chromosomes are also reported to have the largest number of mutations associated with UCEC. Further analysis of the differences in nucleosome footprint features found enrichment of genes related to cancer and immune response.    

   The positive percent agreement (PPA) for predictionscore above the threshold was 100% between specimens collected and processed within 2 h and those transported on ice or at room temperature for 24, 48 or 72 h (95% confidence interval of 51.0-100.0%) or those stored frozen for 3 days, 7 days or 1 month (95% confidence interval of 56.6-100.0%). In contrast, the risk score was higher in specimens stored frozen for 6 months than in specimens processed within 2 h, resulting in a PPA of 60% (95% confidence interval of 23.1-88.2%). Importantly, there was 100% agreement in the risk score between specimens collected pre-meal, post-meal and after exercising (95% confidence interval of 51-100%).

   Biospecimens

   • Bodily Fluid - Plasma
   • Bodily Fluid - Blood

   Preservative Types

   • None (Fresh)
   • Frozen

   Diagnoses:

   • Normal
   • Neoplastic - Carcinoma

   Platform:

   Analyte	Technology Platform
   DNA	Next generation sequencing

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Preaquisition	Prognostic factor	Grade 1 Grade 2 Grade 3 Grade 4 Range of tumor fractions
   Preaquisition	Diagnosis/ patient condition	UCEC Healthy
   Biospecimen Preservation	Type of fixation/preservation	Frozen None (fresh)
   Storage	Specimen transport duration/condition	2 h (not transported) 24 h 48 h 72 h At room temperature With an ice pack
   Storage	Storage duration	2 h 3 days 7 days 1 month 3 months 6 months
   Biospecimen Acquisition	Time of biospecimen collection	Pre-meal Post-meal Post-exercise

   View more