Integrated multi-omics profiling of a Milu deer facial tumor

A spontaneous facial tumor from a Milu deer, also known as Père David’s deer, has provided the first integrated genomic, transcriptomic and pathological profile of neoplasia in the species. Researchers at the Beijing Milu Ecological Research Center examined a mass measuring approximately 3.5 by 2.0 by 1.5 centimetres and nearby macroscopically normal tissue from a single deceased animal. Histology identified an infiltrative epithelial neoplasm with mesenchymal features, while sequencing detected 6,537 differentially expressed genes and 114 high-confidence, tumor-specific non-synonymous somatic mutations. The exploratory case report establishes a baseline for wildlife disease surveillance, but its single-case design means the results cannot yet establish species-wide cancer mechanisms or identify proven therapeutic targets.

The Milu deer is a cervid species that has survived severe population bottlenecks and is now maintained through captive breeding and reintroduction. Although the species is known for annual antler regeneration without neoplastic transformation, the genetic and molecular basis of spontaneous tumors in Milu deer has remained largely uncharacterized. The study addressed that gap by combining frozen tumor and adjacent normal tissue, with the normal sample collected approximately two centimetres from the tumor margin. Researchers used hematoxylin and eosin staining, immunohistochemistry, RNA sequencing, quantitative real-time PCR and whole-genome resequencing to characterize the lesion from cellular morphology through to DNA and RNA variation. The investigators described the case as exploratory and hypothesis-generating because it involved one animal and did not include biological replicates.

Under the microscope, the lesion showed disorganized nests of cells, adenoid structures and pronounced nuclear pleomorphism, features the researchers interpreted as consistent with an infiltrative epithelial neoplasm. Immunohistochemistry supported that classification: tumor cells stained strongly for the epithelial marker pan-cytokeratin and widely expressed vimentin, a marker associated with mesenchymal features. Together, the observations were consistent with epithelial-to-mesenchymal transition, a state in which epithelial cells acquire traits often associated with migration and invasion. The cells were negative for SOX10 and S100, which the team said excluded a neural crest origin such as melanoma or schwannoma. The tumor also lacked detectable CD3-positive T cells and CD68-positive macrophages, leading the authors to describe it as an “immune-cold” epithelial tumor with significant mesenchymal features.

RNA sequencing produced a broad contrast between the tumor and adjacent tissue. Among 11,680 expressed genes, the analysis classified 6,537 as differentially expressed, including 5,423 with increased and 1,114 with decreased expression in the tumor. Because the study included only one tumor and one matched non-tumor sample, the researchers used NOISeq, a non-parametric method designed for single-sample or low-replicate RNA sequencing data, and treated the results as individual observational trends rather than population-level biomarkers. The analysis also detected differences in alternative splicing, with skipped exons forming a major category. Quantitative PCR for CHAF1A, LGALS3, DMBT1 and COG2 produced expression trends consistent with the RNA sequencing results, supporting the reproducibility of those trends. The sequencing library had an RNA integrity value of 8.5, and reads were aligned to the red deer reference genome because a completed Milu deer assembly was unavailable.

Functional enrichment pointed to a network of biological processes altered in the tumor. Genes were enriched for protein kinase activity, small GTPase binding, phosphorylation, actin-cytoskeleton organization, focal adhesion and extracellular matrix-receptor interaction. Other enriched pathways involved cell-cycle control, apoptosis, MAPK and Wnt-related signaling, chemokine signaling and mTOR. Together with pan-cytokeratin and vimentin staining, these patterns were interpreted as evidence of coordinated cytoskeletal remodeling and epithelial-to-mesenchymal transition. The results also indicated altered pathways associated with tumor-microenvironment remodeling, including interactions between cells and the extracellular matrix. These are pathway-level associations from one specimen, however, rather than direct demonstrations that any single signaling route caused the tumor. The analysis was designed to characterize the molecular landscape of this case and identify hypotheses for later study, not to establish a causal hierarchy among the altered processes.

A conspicuous enrichment result concerned the human papillomavirus infection pathway. The authors were explicit that this did not establish viral involvement: no direct viral detection was performed. Instead, the pathway consists largely of conserved host processes governing cell division, apoptosis and immunity, which can appear altered in cancers whether or not a virus is present. The finding matters because papillomavirus-associated fibropapillomas are documented in other cervid species, including roe deer and red deer. The authors argue that systematic viral surveillance would be needed to distinguish this sporadic, non-communicable lesion from a possible outbreak, but the sequencing data alone cannot resolve that question. Their interpretation was that the enrichment reflects a molecular pattern resembling viral oncogenesis without evidence of active viral replication, rather than evidence that HPV itself caused the lesion.

Whole-genome sequencing was performed on the matched tumor and normal tissues at average depths of 54.95-fold and 26.41-fold, respectively. Initial comparison with the red deer reference genome found approximately 1.05 million single-nucleotide variants in each sample. Most represented shared germline differences between the Milu deer individual and the reference rather than tumor-specific mutations, and they were distributed mainly in intergenic and intronic regions. The tumor carried a slightly higher proportion of exonic non-synonymous variants, at 35.04% compared with 34.78% in the control, and a slightly higher proportion of stop-gain variants, at 0.29% compared with 0.28%. Applying paired tumor-normal calling and conservative filters produced 114 coding, non-synonymous variants detected in tumor tissue but absent from the adjacent normal sample. The paired design allowed fixed species-level differences shared by both samples to be treated as background, while requiring at least 10 alternative reads and a tumor variant allele frequency above 0.10 for the selected somatic calls.

Intersecting the 114 genes carrying high-confidence somatic mutations with the differentially expressed set yielded ten candidate genes: GON4L, DUSP28, BBS9, CHAF1A, LGALS3, DMBT1, DDX21, COG2, SIRPA and TSKU. DMBT1 and LGALS3 showed the strongest apparent increases in expression, with log2 M-values of 14.51 and 10.31, respectively. TSKU, CHAF1A and BBS9 also appeared upregulated, while DUSP28 was downregulated. Comparisons with research on human and other mammalian cancers suggest possible relationships between the candidates and extracellular matrix remodeling, chromatin assembly, MAPK regulation, immune signaling, proliferation and metastasis. For example, elevated LGALS3 and SIRPA could be consistent with an immunosuppressive tumor niche, while DMBT1 has been linked in other studies to metastatic behavior. These functions remain hypotheses in the Milu deer case because the genes were not experimentally manipulated or shown to cause the tumor.

Several limitations constrain the conclusions. The report concerns one sporadic case, and the lack of biological replicates prevents conventional replicated differential-expression analysis from establishing statistically generalizable biomarkers. Individual variation and technical noise therefore cannot be fully separated from disease-associated expression changes. Both sequencing datasets were aligned to the red deer genome because a completed, annotated Milu deer reference assembly was unavailable. That substitution may introduce mapping and annotation biases, although the paired design helps remove fixed differences shared by the tumor and normal tissues. The study also did not directly test for viruses or experimentally validate the prioritized genes in cells or animals. The findings are consequently best treated as a provisional molecular and histopathological baseline for the endangered species rather than a complete account of Milu deer cancer biology.

Even with those constraints, the work shows how histopathology, immunohistochemistry and two sequencing platforms can be combined in a protected species where paired biological material is scarce and routine surveillance is limited. The case also highlights the need to interpret pathway annotations carefully: an apparent viral signature can emerge from profound disruption of host networks without an infectious agent, and shared cancer genes can illuminate plausible mechanisms without proving their role. The investigators identify a native Milu deer reference genome, additional independent tumors, direct pathogen testing and functional validation of candidate drivers as priorities for future work. Building that evidence base incrementally would help distinguish recurring mechanisms from features peculiar to this individual and provide a stronger foundation for disease monitoring in captive and reintroduced Milu deer populations.

Subject of Research: Integrated genomic, transcriptomic and histopathological analysis of a spontaneous facial tumor in a Milu deer

Article Title: An exploratory Case Report: integrated genomic and transcriptomic profiling of a spontaneous facial tumor in the endangered Milu deer (Elaphurus davidianus)

Article References: Zhang, Y., Cheng, Z., Shan, Y., Zhang, Q., Zhang, P., Bai, J., Guo, Q., & Zhong, Z. (2026). An exploratory Case Report: integrated genomic and transcriptomic profiling of a spontaneous facial tumor in the endangered Milu deer (Elaphurus davidianus). Frontiers in Veterinary Science, 13, Article 1933048. https://doi.org/10.3389/fvets.2026.1933048

Image Credits: AI Generated

DOI: 10.3389/fvets.2026.1933048

Keywords: Milu deer, facial tumor, cancer genomics, RNA sequencing, somatic mutations, wildlife oncology, epithelial-mesenchymal transition, comparative oncology, wildlife conservation, veterinary pathology, exploratory, Case