Unbiased screen for pathogens in human paraffin-embedded tissue samples by whole genome sequencing and metagenomics

Abstract

Identification of bacterial pathogens in formalin fixed, paraffin embedded (FFPE) tissue samples is limited to targeted and resource-intensive methods such as sequential PCR analyses. To enable unbiased screening for pathogens in FFPE tissue samples, we established a whole genome sequencing (WGS) method that combines shotgun sequencing and metagenomics for taxonomic identification of bacterial pathogens after subtraction of human genomic reads. To validate the assay, we analyzed more than 100 samples of known composition as well as FFPE lung autopsy tissues with and without histological signs of infections. Metagenomics analysis confirmed the pathogenic species that were previously identified by species-specific PCR in 62% of samples, showing that metagenomics is less sensitive than species-specific PCR. On the other hand, metagenomics analysis identified pathogens in samples, which had been tested negative for multiple common microorganisms and showed histological signs of infection. This highlights the ability of this assay to screen for unknown pathogens and detect multi-microbial infections which is not possible by histomorphology and species-specific PCR alone.

Keywords: infection; metagenomics; next-generation sequencing; pathogen identification; pathology.

Figure 1

Schematic overview of metagenomics analysis of DNA extracted from FFPE tissue samples. FFPE, formalin-fixed paraffin-embedded; Next gen. sequencing, Next-generation sequencing.

Figure 2

Sensitivity of metagenomics analysis on test samples of known composition. (A) Schematic workflow for generating serial dilutions of known bacterial pathogens (mock communities), admixed with human genomic DNA. (B) Detection of known bacterial pathogens and associated phages in diluted mock communities and detection of reference strains in a microbial community standard by quantitative PCR (qPCR, left column of panels), taxonomic profiling for bacteria (middle column of panels) and viruses (right column of panels). Each row of panels represents results of an individual mock community represented by several dilutions. Within a row of panels, line color and symbol shape indicate data originating from a specific isolate or reference species. Background of bacterial/viral analysis: The most abundant background signal detected for a bacterial and a viral species is indicated. WGS, whole genome sequencing; Ct, cycle threshold.

Figure 3

Bacterial and viral reference species identified by metagenomics analysis in round robin tests. (A) Genetic origin of sequence information interpreted from read mapping to the human genome GRCh37 (hg19) or taxonomic profiling of non-human reads. Reads that could not be assigned to human, bacteria or viruses were labeled as unidentified. (B) Different species identified in bacterial and viral reads in individual round robin test samples.

Figure 4

Metagenomics analysis of FFPE lung tissue samples. (A) Detection of bacterial pathogens in human FFPE lung tissue samples by quantitative PCR (qPCR) and by metagenomics analysis. (B) Detection of S. aureus by qPCR and by metagenomics analysis in 16 samples. Box shows range from first to third quartile and median, whiskers display minimum and maximum. Statistical analysis: T-test; ***, p<0.001. (C) Grade of neutrophilic infiltration in human FFPE lung autopsy tissues. (D) Number of samples positive for the indicated species, as determined by metagenomics analysis. (E) Fraction of samples with one or more bacterial species identified by metagenomics analysis, sorted by grade of neutrophilic infiltration. Statistical analysis: χ-square-test; ****, p<0.0001. (F) Fraction of reads representing individual species for each sample presented in E. Colors (e.g. red, green, blue, purple, orange) represent individual genera and shades of the same color (e.g. dark red, red, light red, rose, faint rose) represent individual species of the same genus, sorted from most frequent to least frequent. According to their rank of abundance within an individual sample, genus and species are highlighted by colors and shades: red represents the most frequent genus; green the 2^nd, blue the 3^rd, purple the 4^th and orange the 5^th most frequent genus. Grey represents all genera with rank 6 or higher. FFPE, formalin-fixed paraffin-embedded; Ct, cycle threshold.