Cancer cell states predict response, and in mice they can be bent toward it
Preprints link NK-cell states, a bladder-cancer marker and breast-cancer signatures to outcomes, while matrix softening and radiation injury change immune response in mice.
- In a preprint, a tool called NKceiver predicts COVID-19 severity and response to immune checkpoint inhibitors from NK-cell states. Self-reported m:…7102
- High SETDB1 expression is an independent predictor of poor overall survival in bladder cancer patients in a preprint. Self-reported m:…0341
- In a preprint, models distinguished invasive from non-invasive triple-negative breast cancer cells with an AUC of 0.92 on sample-level data. Self-reported m:…3222
- Matrix softening enhances the efficacy of PD-1 blockade in syngeneic mouse models in a preprint, while also inducing adaptive immune resistance. Self-reported m:…4669
- Vaccination with astatine-211-injured melanoma cells suppresses pulmonary metastasis in syngeneic mice in a preprint, and the effect requires STING signaling. Self-reported m:…0351
Reading the state
Five preprints posted to bioRxiv share a move: they read the state of tumor or immune cells and use it to predict what happens next, then ask whether that state can be changed. The studies cover natural killer cells, bladder cancer, triple-negative breast cancer, gastric cancer and melanoma. All five are preprints, so every result below is unreviewed work from the authors themselves, not an established finding. m:…7102 m:…0341 m:…3222
The natural killer cell preprint introduces HNKA, a method that identifies disease-biased redistribution versus transcriptional displacement in NK-cell states. CD56dim NK ISG15 states are associated with infection and autoimmunity, while CD56bright NK ITM2C states are associated with tumors. The preprint's tool, NKceiver, predicts COVID-19 severity and response to immune checkpoint inhibitors, which would make a cell-state readout useful both for prognosis and for anticipating immunotherapy response. m:…7102
The bladder cancer preprint is narrower and more direct. High SETDB1 expression is an independent predictor of poor overall survival in bladder cancer patients, and SETDB1-high patients are more likely to present with squamous and sarcomatoid differentiation features. The authors present SETDB1 as a potential therapeutic target for bladder cancer due to its association with tumorigenesis and chemotherapy resistance, which turns a prognostic marker into a candidate for intervention. m:…0341
The triple-negative breast cancer preprint adds a machine-learning version of the same idea. XGBoost models can distinguish invasive from non-invasive MDA-MB-231 cells with an AUC of 0.92 when trained on sample-level data. Invasive cells are characterized by increased KRAS signaling and E2F target expression and decreased interferon-alpha and gamma responses, and the top-ranked transcriptional markers of invasive potential are PHLDA2, ANTXR2, VIM and HMGA1. The marker list is the practical output: a short signature that separates invasive from non-invasive cells in this cell line. m:…3222
The state as a lever
The gastric cancer preprint shows the state is not just a readout. High matrix stiffness in gastric cancer is associated with collagen accumulation and immune exclusion, and stiff matrices promote F-actin remodeling and YAP nuclear enrichment, driving LDHA-dependent glycolytic reprogramming in gastric cancer cells. In other words, the physical stiffness of the tumor environment reshapes both metabolism and immune access, which gives researchers a mechanical handle on a biological problem. m:…4669
Softening the matrix changes immunotherapy response, but in both directions at once. Matrix softening enhances the efficacy of PD-1 blockade in syngeneic mouse models, while also inducing adaptive immune resistance via IFN-gamma-responsive PD-L1 expression. That double effect is the kind of result that matters for trial design: a combination that helps on one axis can create resistance on another, and the preprint names both. m:…4669
The melanoma preprint takes the lever further, using radiation injury itself. Astatine-211 injury induces an immunostimulatory melanoma-cell state characterized by DNA damage and STING-dependent inflammatory signaling. Vaccination with astatine-211-injured B16F10 cells suppresses pulmonary metastasis in syngeneic C57BL/6J mice, and STING signaling is required for the antimetastatic effect of vaccination with the injured tumor cells. The result is in mice, not patients, but it frames injured tumor cells as a potential vaccine input rather than just debris. m:…0351
What to take from the set
What the five preprints suggest, taken together, is that cell-state readouts are converging on clinical use cases: predicting survival, predicting immunotherapy response, and separating invasive from non-invasive cells. The levers are earlier and more uncertain, confined to mouse models and mechanical or radiation manipulations. The gap between the two halves is the story: prediction is getting specific, with named markers and measured accuracy, while intervention is still showing that changing a state creates new resistance as well as new response. m:…0341 m:…7102 m:…3222 m:…4669
What is still unresolved
- All five studies are preprints and have not been peer-reviewed.
- The matrix-softening and astatine-211 results are in mouse models, not patients.
- Whether the NK-cell, SETDB1 and breast-cancer signatures validate in independent patient cohorts.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.