That, more or less, is the problem in head and neck squamous cell carcinoma, or HNSCC: the cancer is not one thing, and its immune neighborhood is not one thing either. Which means treating every case like it came off the same factory line was always a bit optimistic in the way protocol amendments are "minor."
Four tumor moods, and yes, they matter
A new 2026 study by Khatri and colleagues took a very large swing at this problem. They analyzed 1,149 tumors from 1,102 patients with HNSCC using transcriptomic and genomic data, then used a multi-omics approach called Similarity Network Fusion to sort tumors by immune behavior rather than by vibes alone [1]. The result was four immune subtype clusters.
In plain English, the tumors fell into four camps:
- ISC1: immune-cold and EMT-enriched
- ISC2: immune-activated
- ISC3: mixed immune-regulatory with stromal enrichment
- ISC4: immunosuppressed [1]
That may sound like alphabet soup served by a bioinformatics committee, but the idea is actually pretty intuitive. Some tumors are like badly lit warehouses with no security in sight. Some are crawling with immune cells that look ready to throw punches. Some have plenty of cells around but also a lot of stroma - the structural tissue that can act like a bureaucratic maze for T cells. And some are actively suppressing the immune response, which is the cellular equivalent of cutting the fire alarm and bribing the night guard.
That matters because HNSCC has been a poster child for heterogeneity. Reviews over the past two years keep landing on the same point: the tumor microenvironment strongly shapes prognosis and immunotherapy response, yet our current biomarkers still miss plenty of the plot [2,3].
Your immune system is not failing. It is getting outmaneuvered.
If you want the bar-stool version, your immune system is running private security for your body 24/7. Tumors survive by learning how to dodge the patrol, hide in the basement, or convince the guards to frisk each other instead.
This paper suggests we should stop asking only, "Does this tumor express marker X?" and start asking, "What kind of immune ecosystem are we dealing with?" That is a smarter question. A single PD-L1 score can be useful, but it can also flatten a very weird landscape into one number, which is how you end up making clinical decisions with the emotional texture of an airport kiosk.
The authors also linked these immune subtypes to progression-free survival and to different responses across checkpoint inhibitors, chemotherapy, and combination treatments [1]. That fits with the broader HNSCC literature: only a minority of patients get major benefit from immunotherapy, and multi-omics approaches may help explain why [3,4].
The really interesting part: tumors can switch costumes
Here is the plot twist that makes trialists sit forward in their chairs.
The study did not just classify tumors once and call it a day. In 44 patients with matched pre- and post-treatment tumors, the researchers found that immune states could change after therapy. Checkpoint inhibition tended to push tumors toward more immune activation, while chemotherapy tended to enrich immunosuppressive signaling [1].
That is a big deal.
It suggests a tumor’s immune subtype is not a permanent zodiac sign. It is more like a moving weather system. Sunny with CD8s on Monday, stromal nonsense by Thursday. If that holds up, then a biopsy from months ago may not fully represent the battlefield you are treating now.
Other recent work in HNSCC points in the same direction. Spatial and topographic studies show that not just the presence of immune cells, but where they sit relative to tumor cells and stroma, can shape response [5]. Recent long-term follow-up from KEYNOTE-048 also reinforces that immunotherapy can deliver durable benefit for some patients with recurrent or metastatic disease, but clearly not for everyone [6]. The field has been telling us for years that timing, context, and patient selection matter. This paper gives that argument a better map.
Why this could matter outside the lab
If these immune subtypes prove reproducible in prospective studies, they could help with three very practical problems.
First, they might improve treatment selection. An immune-activated tumor may not need the same playbook as an immune-cold, EMT-heavy one.
Second, they could help explain why resistance happens. Sometimes the drug did not fail because it was "bad." Sometimes the tumor moved the goalposts.
Third, they raise the possibility of serial immune profiling - checking how the tumor changes during treatment rather than assuming it stays frozen in time. Which, to be fair, is how biology has spent decades humiliating our love of tidy baseline categories.
No, this is not ready to replace clinical judgment tomorrow morning. It is still a framework, not a bedside test. But it is the kind of framework that feels useful because it respects the mess instead of pretending the mess is noise.
References
-
Khatri I, de Souza T, van Asten SD, et al. Integrative immune subtyping of HNSCC reveals clinically relevant phenotypes and treatment-associated transitions. npj Precision Oncology. 2026. DOI: https://doi.org/10.1038/s41698-026-01440-y
-
Liu XH, Wang GR, Zhong NN, et al. Multi-omics in immunotherapy research for HNSCC: present situation and future perspectives. npj Precision Oncology. 2025;9(1):93. DOI: https://doi.org/10.1038/s41698-025-00886-w. PMCID: https://pmc.ncbi.nlm.nih.gov/articles/PMC11954913/
-
Gatfield ER, Tadross J, Ince W. Immune checkpoint inhibitor use in head and neck squamous cell carcinoma: the current landscape and future perspectives. Future Oncology. 2024;20(23):1695-1711. DOI: https://doi.org/10.1080/14796694.2024.2362612. PMCID: https://pmc.ncbi.nlm.nih.gov/articles/PMC11485897/
-
Diao P, Dai Y, Wang A, et al. Integrative Multiomics Analyses Identify Molecular Subtypes of Head and Neck Squamous Cell Carcinoma with Distinct Therapeutic Vulnerabilities. Cancer Research. 2024;84(18):3101-3117. DOI: https://doi.org/10.1158/0008-5472.CAN-23-3594
-
Muijlwijk T, Nijenhuis DNLM, Ganzevles SH, et al. Immune cell topography of head and neck cancer. Journal for ImmunoTherapy of Cancer. 2024;12(7):e009550. DOI: https://doi.org/10.1136/jitc-2024-009550
-
Rischin D, Tahara M, Greil R, et al. Pembrolizumab with or without chemotherapy in recurrent or metastatic head and neck squamous cell carcinoma: 5-year follow-up from the randomized phase III KEYNOTE-048 study. European Journal of Cancer. 2025;221:115395. DOI: https://doi.org/10.1016/j.ejca.2025.115395
Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.