Cancer-microbe associations have been explored for centuries, but cancer-associated fungi have rarely been examined. Here, we comprehensively characterize the cancer mycobiome within 17,401 patient tissue, blood, and plasma samples across 35 cancer types in four independent cohorts. We report fungal DNA and cells at low abundances across many major human cancers, with differences in community compositions that differ among cancer types, even when accounting for technical background. Fungal histological staining of tissue microarrays supported intratumoral presence and frequent spatial association with cancer cells and macrophages. Comparing intratumoral fungal communities with matched bacteriomes and immunomes re- vealed co-occurring bi-domain ecologies, often with permissive, rather than competitive, microenvironments and distinct immune responses. Clinically focused assessments suggested prognostic and diagnostic ca- pacities of the tissue and plasma mycobiomes, even in stage I cancers, and synergistic predictive perfor- mance with bacteriomes.

Fungal nucleic acids exist in many human cancer types

We profiled fungal DNA in two large cohorts of cancer sam- ples we previously examined for bacteria (Nejman et al., 2020; Poore et al., 2020). The first (Weizmann [WIS]) comprised 1,183 formalin-fixed paraffin-embedded (FFPE) or frozen samples of tumor and normal adjacent tissue (NAT; often paired) from eight tissue types (breast, lung, melanoma, ovary, colon, brain, bone, and pancreas) and non-cancer normal breast tissue. All samples were studied for fungal pres- ence using internal transcribed spacer 2 (ITS2) amplicon sequencing (Figure 1A; Tables S1 and S2). To account for po- tential contamination by environmental fungi or fungal DNA introduced via sample handling and processing, we included 104 paraffin-only and 191 DNA-extraction negative controls. These controls enabled detection and removal of fungal contaminants and separation of signal from noise in ITS2 data (STAR Methods).

The second cohort encompassed whole genome sequencing (WGS) and transcriptome sequencing (RNA-seq) data from The Cancer Genome Atlas (TCGA) (Figure 1A; Table S1). For quality control, we re-aligned all ($1011) unmapped DNA and RNA reads to a uniform human reference (GRCh38), then removed poor-qual- ity reads. Remaining reads were aligned to the RefSeq release 200 multi-domain database of 11,955 microbial (with 320 fungal) genomes (STAR Methods). 15,512 samples (WGS: 4,736; RNA- seq: 10,776) had non-zero microbial feature counts, of which 15,065 (97%) contained fungi. Of 6.06 3 1012 total reads, 7.3% did not map to the human genome: 98.8% of these unmapped reads mapped to no organism in our microbial database. Of the remaining 1.2% of non-human reads that mapped to our microbial database (0.08% of total reads), 80.2% (0.067% of total) were classified as bacterial, and 2.3% (0.002% of total) as fungal, providing 1.172 3 108 fungal reads for downstream analyses with an average read length of 57.4 bp (SD = 15.9; median = 51bp; methods enforced 45-bp minimum). Fungal-containing TCGA samples had an average of 7,780 (95% CI: [7,039, 8,521]) fungal reads/sample. Although TCGA lacked contamination controls, we implemented in silico decontamination based on sequencing plate and center (Poore et al., 2020) and cross-refer- enced all fungal species against the WIS decontaminated ampli- con cohort, the Human Microbiome Project (HMP)’s gut myco- biome cohort (Nash et al., 2017), and >100 other publications to obtain a final decontaminated list.

More Information: Here