Proteomics Predicts Radiotherapy Toxicity in Cancer

Roughly six in ten patients with localized solid tumors get radiotherapy with curative intent, and many carry bowel or urinary side effects for years afterward. The problem is timing: clinicians cannot tell before the first dose who will tolerate treatment. A 2026 study in Communications Medicine applied longitudinal plasma proteomics to weekly blood from patients with prostate, bladder, or head and neck cancer, charting how radiation reshapes the circulating proteome and which baseline proteins flag later toxicity.

How Plasma Proteomics Maps the Radiotherapy Response

The team profiled three cohorts: 26 prostate, 23 bladder, and 11 head and neck cancer patients. Rather than depleting albumin the usual way, they used a nanoparticle protein-corona method, incubating liposomes with plasma so low-abundance proteins adsorb onto the particle surface and rise above the albumin background before LC-MS/MS. This protein-corona approach, first detailed by Hadjidemetriou and colleagues (2019), drives the depth reported here. Differential analysis moved fast: 93 differentially abundant proteins in prostate cancer one week in, 142 in bladder, and 141 in head and neck, with most changes inside the first two weeks. Principal component analysis kept the three tumor types separated at every timepoint.

Key Findings

• Changes appear within a week: differentially abundant proteins surfaced seven days after the first fraction, 93 to 142 per cohort, concentrated in the opening fortnight.

• A shared inflammatory-to-repair arc: across all three cancers, mass spectrometry pathway analysis traced an early rise in complement and neutrophil signaling that later shifted to extracellular matrix remodeling.

• Ficolin-1 moved in one direction everywhere: among hundreds of cohort-specific proteins, FCN1 was the single marker consistently downregulated across all three cohorts.

• Baseline blood carried predictive signal: pre-treatment protein profiling separated prostate patients by later toxicity (Fisher exact p = 0.028), and one early factor reached an AUC of 0.93.

Figure 1. Study design and longitudinal sampling. Panel a summarizes the three clinical cohorts (prostate adenocarcinoma, urothelial bladder carcinoma, and oropharyngeal head and neck carcinoma), with patient numbers, sex, age, radiation dose and fractions, and concurrent therapies. Panel b shows the plasma collection timeline, sampled at baseline (t0) and weekly until treatment end. Adapted from Abumanhal-Masarweh et al. (2026), Communications Medicine.

A Systemic Response Hiding in the Blood

Radiation hits a defined target, but its fingerprints spread through the whole circulation. Early in treatment the plasma leaned toward lipid-metabolism pathways tied to membrane damage and oxidative stress. By the middle weeks, complement activation through the lectin pathway and neutrophil-driven immunity dominated, then the late signal turned to matrix organization and apoptotic-cell clearance. Five pathways stayed active throughout, yet the proteins driving them differed by cancer type, which complicates any hope of one universal radiotherapy blood test.

From Baseline Blood to a Toxicity Prediction

The prostate cohort doubled as the proof of concept. Of 26 patients, 17 developed late bowel or urinary toxicity more than three months out. Using Multi-Omics Factor Analysis on baseline plasma, the authors found a single latent factor that split patients into two groups matching toxicity status (Fisher exact p = 0.028). One on-treatment factor measured a week in reached an AUC of 0.93, with roughly 28 baseline proteins carrying weight. With only nine toxicity-free patients, the effect needs replication before anyone builds a clinical assay on it.

In Practice: Deep Plasma Proteomics Workflows

Getting past albumin and the other high-abundance carriers is the hard part of any blood biomarker hunt, and the corona-based enrichment here is one route to the low-abundance proteins that actually move with disease. We chase the same depth in deep plasma work at Dalton, and a study like this one is a useful reminder that batch correction across plates and locked-down baseline sampling often decide whether a toxicity signature survives validation. The weekly cadence also shows how much the pre-analytical choices matter long before the statistics start.

Frequently Asked Questions

What is plasma proteomics?

Plasma proteomics is the large-scale measurement of proteins circulating in blood, usually by mass spectrometry. It lets researchers compare thousands of proteins between patient groups or over time to find biomarkers of treatment response or side effects.

Can plasma proteomics predict radiotherapy toxicity?

In this study, baseline and early-treatment plasma protein signatures separated prostate cancer patients who later developed toxicity from those who did not, with one factor reaching an AUC of 0.93. The cohort was small, so the finding is promising rather than clinic-ready and needs larger validation.

How much blood does a clinical proteomics study need?

Methods like the protein-corona workflow used here run on standard plasma draws, often under a milliliter per timepoint. The main cost driver is mass spectrometry instrument time and careful sample handling, not blood volume.

Conclusion

What holds up here is the timing: the circulating proteome reacts within days, and the first two weeks look like the window where biomarkers live. What is not yet settled is whether a single protein panel can predict toxicity well enough to change a treatment plan, given the small cohorts and cohort-specific drivers. For trial designers, the practical move is to bank weekly plasma early, because the most informative samples may be the ones drawn before symptoms appear.

Citation

Abumanhal-Masarweh, H., Assi, S. A., Liu, X., Guerrero Quiles, C., Lodhi, T., Williams, K. J., Cheadle, E. J., Kostarelos, K., Choudhury, A., Wedge, D. C., West, C. M. L., & Hadjidemetriou, M. (2026). Longitudinal plasma nano-proteomics reveals acute systemic responses to radiotherapy and predictive biomarkers of late toxicity. Communications Medicine. https://doi.org/10.1038/s43856-026-01552-3

Note

This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.

By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.