Machine learning algorithm improved automated droplet classification of ddPCR for detection of BRAF V600E in paraffin-embedded samples

Gabriel A Colozza-Gama,Nikola Bešič,Janete M Cerutti,Fabiano Callegari,Ana Carolina De J Paniza

doi:10.1038/s41598-021-92014-4

Gabriel A Colozza-Gama, Nikola Bešič + Show 3 more

Open Access

https://doi.org/10.1038/s41598-021-92014-4

Copy DOI

Abstract

Somatic mutations in cancer driver genes can help diagnosis, prognosis and treatment decisions. Formalin-fixed paraffin-embedded (FFPE) specimen is the main source of DNA for somatic mutation detection. To overcome constraints of DNA isolated from FFPE, we compared pyrosequencing and ddPCR analysis for absolute quantification of BRAF V600E mutation in the DNA extracted from FFPE specimens and compared the results to the qualitative detection information obtained by Sanger Sequencing. Sanger sequencing was able to detect BRAF V600E mutation only when it was present in more than 15% total alleles. Although the sensitivity of ddPCR is higher than that observed for Sanger, it was less consistent than pyrosequencing, likely due to droplet classification bias of FFPE-derived DNA. To address the droplet allocation bias in ddPCR analysis, we have compared different algorithms for automated droplet classification and next correlated these findings with those obtained from pyrosequencing. By examining the addition of non-classifiable droplets (rain) in ddPCR, it was possible to obtain better qualitative classification of droplets and better quantitative classification compared to no rain droplets, when considering pyrosequencing results. Notable, only the Machine learning k-NN algorithm was able to automatically classify the samples, surpassing manual classification based on no-template controls, which shows promise in clinical practice.

Highlights

Sanger sequencing is still considered the gold standard for detecting mutations when small fragments of DNA are analyzed for specific single nucleotide variants (SNVs)
We used DNA isolated from Formalin-fixed paraffin-embedded (FFPE) sections of a papillary thyroid carcinoma (PTC) and a follicular thyroid adenoma (FTA) known as positive and negative for BRAF V600E, respectively
Our hypothesis is that because of the high variability of the technique, the information about biological populations of tumors with different variant allele frequency (VAF) is smoothed and information is lost. This is the first study that compared Sanger, Pyrosequencing and droplet digital polymerase chain reaction (PCR) (ddPCR) with a real-life dataset consisting of tumor samples derived from FFPE with variable DNA quality

Summary

Introduction

Sanger sequencing is still considered the gold standard for detecting mutations when small fragments of DNA are analyzed for specific single nucleotide variants (SNVs). This procedure compromises the quality of genomic DNA and, the PCR-based analyses of DNA isolated from FFPE such as Sanger Sequencing It does not allow quantitative evaluation of mutated alleles, as well as presents low sensitivity for detecting somatic cancer mutations present at very low (< 20%) variant allele frequency (VAF). One such methodology is pyrosequencing, a synthesis-based sequencing method that uses small fragments of PCR to initiate the synthesis of a new strand and detect incorporated bases by fluorescence This methodology has vastly improved detection of SNVs, especially in highly degraded material derived from FFPE. One advantage of ddPCR is that it is an old technology based on new chemistry and, the platform shows great potential for advancements, Scientific Reports | (2021) 11:12648 Such as the possibility to perform higher multiplexing and the use of machine learning methods to implement a more accurate automatic classification of droplets. The limiting dilution strategy and the Poisson distribution allows to determine the absolute count of target DNA copies per s ample[3,4]

Objectives

Methods

Results

Conclusion