Our recent collaboration with the Weizmann Institute of Science and the University of California San Diego found that fungi are ubiquitous in human cancer types and often interact with bacteria in synergistic, rather than competitive ways. This is the latest chapter in an ever-expanding book of knowledge by Micronoma colleagues that characterizes the existence of multi-domain microbiomes in cancer and their roles in early-cancer detection.

The international work, Pan-cancer analyses reveal cancer type-specific fungal ecologies and bacteriome, published on the cover of the September 29, 2022 issue of Cell, systematically profiled fungal communities—the mycobiome—in 35 types of cancer and more than 17,000 tissue and blood samples from four independent cohorts, refined with hundreds of experimental contamination controls. Cell co-published this story alongside data from an independent group that made similar findings strengthening the validity of the results. 1

Recent papers, including the initial discovery of pan-cancer bacteriomes by our co- founder and Chief Analytics Officer, Gregory Sepich-Poore, published in Nature in 2020, significantly advanced the study of the presence of microbes in tumor tissues and blood by characterizing microbes among more than 30 cancer types. But this new paper—led by Greg at Micronoma and co-first authors Lian Narunsky-Haziza and Ilana Livyatan at the Weizmann—was the first to comprehensively explore the presence of fungi in dozens of cancer types.

It’s understandable why it had not been done before. Besides having genomes that are more similar to humans compared to bacteria, fungi are also present in extremely low quantities. In comparison to bacteria, which comprise approximately 1% of cells in tumors fungi are somewhere between 10 fold to 100-fold less abundant, making them particularly difficult to identify and quantify. 2

Nonetheless, always willing to accept the challenge of looking for microscopic needles in giant haystacks, our international group of researchers used diverse, state-of-the-art methods to identify cancer fungi and calculate their abundances: PCR to measure fungal (and bacterial) load, histological imaging to quantify their prevalence, sequencing to estimate their relative abundances, and metagenomic analyses to estimate their diversities (e.g., the unique number of fungal species detected per sample).

Remarkably, when tumors had more fungi, they frequently had more bacteria (using relative abundances or loads), and when they had more diverse kinds of fungi, they also had more diverse kinds of bacteria. These positive correlations between tumor-derived
fungal and bacterial communities suggested a form of synergy, wherein they both may
interact with and benefit from each other.

That finding was counterintuitive to what the researchers expected, since in environments with limited resources, fungi and bacteria often compete with each other. Yet, instead, they appeared to be acting in concert within tumors. From the host side, the tumor microenvironment also appeared to be providing a “permissive” context for both groups to concurrently thrive.

That permissive vs. competitive relationship and context was of particular interest because it showed that fungi and bacteria can coexist, and even thrive, together in the same tumors. It also raises the question of how frequently their interactions are taking place given the relatively low abundances of tumor-derived fungi and bacteria. That question could be answered in the next step of microbiome research, which is expected to unveil the spatial distribution of such microorganisms within tumors, (meaning, looking at a map of the cancer within the tissue, indicating the margins, nearby vessels, etc.). Showing spatial co-locations between fungi and bacteria would provide strong evidence they are interacting to each other’s benefit rather than simply coexisting in a tumor, which is large with respect to their individual sizes, even on the scale of millimeters.

After identifying fungi in dozens of cancer types, a natural next step was to relate them to clinically-relevant metadata and to see if the fungi provided diagnostically useful information. Since bacteria data existed on these same patients, they were also able to combine fungal and bacterial abundances to see if or how much doing so improved the results.

For the diagnostic studies, the team compared how well microbial information could distinguish between cancer types, between tumor and adjacent normal tissue, between cancer stages, and between cancer-bearing versus healthy individuals using either tumor tissue, blood, or plasma samples from multiple independent cohorts. Notably, one of the plasma cohorts they explored for microbial data had previously been used to benchmark the utility of another novel biomarker of cancer, human-centric fragmentomic diagnostics, which observes the size of the fragments of degraded circulating DNA to attempt detection of cancer. 3 This enabled the direct comparison between the diagnostic utility of microbial information versus human fragmentomic  method.

Surprisingly, in that plasma dataset, the microbial data provided better diagnostic performance than the human-centric fragmentomic information on the same patients using the same machine learning approach. Moreover, combining the fungal and bacterial information led to the best diagnostic performances.

It is worth noting that this research stemmed from a serendipitous conversation between Ravid Straussman and Greg, after both discovered they were independently working on the same project to profile fungi in cancer-bearing patients. Analogous to synergies between fungi and bacteria, their choice to collaborate rather than compete proved fruitful, with potential major implications for future cancer diagnostics, Greg highlighted.

“In every case we studied across different cohorts, the addition of fungal information to bacterial information led to synergistic and significant [diagnostic] performance increases, which suggests that combining the two can lead to better diagnostics for patients in the future,” he said.

The paper also identified prognostically-relevant associations, including tumor-derived fungi that correlated with patient immunotherapy responses and survival. However, at this point, more work is needed to see if those fungi are mechanistically involved in shaping antitumor responses.

Among many other questions this study raises, a key one is determining how fungi find their way into tumors, and whether they take the same paths as bacteria. Another is to look at how the combination of fungi and bacteria can predict and/or elicit unique immune responses in their hosts, particularly in the setting of immunotherapy response. Despite all of these new arising questions, one of the biggest immediate benefits of the research is that it provides the first atlas of fungi in multiple cancer types, thereby laying the groundwork for such mechanistic studies that could benefit future therapies.

1. Anders B. Dohlman, et al. (2022), A pan-cancer mycobiome analysis reveals fungal
involvement in gastrointestinal and lung tumors,Cell, 185 (20),
2. Sepich-Poore, et al. (2022). Cancer's second genome: Microbial cancer diagnostics and
redefining clonal evolution as a multispecies process. BioEssays, 44, e2100252.
3. Cristiano, et al. (2019) Genome-wide cell-free DNA fragmentation in patients with
cancer. Nature 570.