Physical Limits of Proximal Tumor Detection via MAGE-A Extracellular Vesicles

TL;DR

This paper investigates the physical detection limits of tumor-derived extracellular vesicles near tumors, using a hybrid modeling approach to inform minimally invasive sensor design for early cancer detection.

Contribution

It introduces a theoretical framework combining stochastic and mean-field models to assess the feasibility of near-tumor EV detection with a smart-needle sensor.

Findings

Maximum detection radius of ~275 micrometers within 6000 seconds

Quantification of false negative probabilities due to stochastic EV transport

Guidelines for designing minimally invasive peri-tumoral sensors

Abstract

Early cancer detection relies on invasive tissue biopsies or liquid biopsies limited by biomarker dilution. In contrast, tumour-derived extracellular vesicles (EVs) carrying biomarkers like melanoma-associated antigen-A (MAGE-A) are highly concentrated in the peri-tumoral interstitial space, offering a promising near-field target. However, at micrometre scales, EV transport is governed by stochastic diffusion in a low copy number regime, increasing the risk of false negatives. We theoretically assess the feasibility of a smart-needle sensor detecting MAGE-A-positive microvesicles near a tumour. We use a hybrid framework combining particle-based Brownian dynamics (Smoldyn) to quantify stochastic arrival and false negative probabilities, and a reaction-diffusion PDE for mean concentration profiles. Formulating detection as a threshold-based binary hypothesis test, we find a maximum feasible detection radius of approximately 275 micrometers for a 6000 s sensing window. These results outline the physical limits of proximal EV-based detection and inform the design of minimally invasive peri-tumoral sensors.