Thrilled to see a basic mechanistic discovery translate to meaningful clinical benefit! aacrjournals.org/cancerdiscov...

This week, that idea reached patients: Viral-mimicry priming + anti-PD-1 in relapsed/refractory NK/T-cell lymphoma delivered 47.6% complete responses, 66.7% ORR, and ~50% 2-year OS, even after prior PD-1 failure (Huang, Cancer Discovery 2025).

It was great to collaborate with brilliant colleagues and friends on this multidisciplinary project, combining evolutionary modeling, statistical physics, immunology, and wet-lab validation

Today’s paper reframes repeat activation from "noise" to a conserved immunogenic signal, explaining why therapies derepressing repeats prime anti-tumor immunity and why tumors develop countermeasures. We believe viral mimicry escape is necessary during cancer initiation and thus a cancer hallmark

Cancers usually adapt this viral mimicry pressure. p53 loss leads to chronic mimicry and tolerance in HGSOC (Ishak, Cancer Discovery 2025), PDAC relies on LINE-1 ORF1p subverting sensing (Sun, Immunity 2024), while other use ADAR1/XRN1 (Mehdipour, Nature 2020; Hosseini, Cell Reports 2024).

A decade ago, we showed that DNA demethylation induces repeat transcripts, triggering an antiviral state that selectively kills cancer cells, termed viral mimicry (Roulois, Cell 2015). Today's work helps to explain why repeats often appear "viral."

2) Organism-centric: PAMPs act as a cell-intrinsic alarm for transcriptional dysregulation, retained to help organisms detect and correct errors. We call this ‘the fire alarm hypothesis’ (Lindholm, Trends in Cancer 2023)

This enabled us to score the immunogenicity of different repeats across species and led to two non-exclusive evolutionary hypotheses: 1) Repeat-centric: PAMP-like motifs support the repeat lifecycle.

We found that many transposons resemble pathogens to innate immune sensors, a property widespread across eukaryotes. Here, we formalized "viral mimicry" with a quantitative statistical physics framework, estimating selective forces that enrich pathogen-associated motifs beyond the genome background

read more here: www.science.org/content/arti...

They’re not just passive genomic elements—they’re active players that may hold the key to new therapies.

Exciting to see the field gaining attention and momentum.

Over the past decade, retroelements have become one of the hottest topics in biomedical science, shedding light on evolution, aging, cancer, autoimmunity, neurodegeneration, and more.

We observers that we can interfere with this process, for example by blocking the reverse transcriptase activity of LINE1, suggesting a promising avenue for early cancer interception.