Fractionated photoimmunotherapy stimulates an anti-tumour immune response: an integrated mathematical and in vitro study

Epithelial ovarian cancer (EOC) often presents as disseminated peritoneal disease with occult micrometastases and exhibits high recurrence despite initial remission rates. Conventional immunotherapies have shown limited benefit in EOC, likely due to an immunologically cold tumour microenvironment and the immunosuppressive effects of high-intensity chemotherapy. Photodynamic therapy (PDT) and photoimmunotherapy (PIT) offer intraperitoneal, spatially selective cytotoxicity and can induce immunogenic cell death. Low-irradiance, low-dose PDT can stimulate antitumour immunity, but PIT’s dose-dependent immunological effects remain less defined. This study asks whether PIT dosing can be rationally designed to exploit immune stimulation, using integrated in vitro measurements and a mathematical model of tumour–immune dynamics to test whether fractionated, low-dose PIT can convert tumour escape into immune-mediated control in EOC.

Prior clinical trials combining immunotherapy with chemotherapy in EOC have largely failed to improve survival, reinforcing the need for strategies that modulate EOC’s immunologically cold microenvironment. PDT is known to induce immunogenic cell death and stimulate local/systemic inflammatory responses, though intraperitoneal toxicity has limited traditional PDT. EGFR-targeted PIT improves tumour selectivity and safety and EGFR is commonly overexpressed in ovarian cancer. Low irradiance and low, non-ablative PDT doses can conserve oxygen and promote immune activation, whereas ablative regimens do not. Mathematical oncology provides tools to study complex, non-linear tumour–immune interactions; established models (e.g., Kuznetsov et al.) have been used to explore therapy-perturbed dynamics. However, the dose-dependent immunomodulation of PIT and its rational fractionation to harness immunity had not been systematically evaluated, motivating the present integration of experiments and modelling.

- Cell models: Human EOC cell lines Ovcar5 (EGFP-tagged) and Ovcar3 (mCherry-tagged) were used. OT-1 mouse T cells (DsRed-expressing) and human Jurkat-GFP T cells were employed in 3D co-culture systems. Cultures were mycoplasma-free and used for ≤28 passages. - Treatments: PDT used unconjugated benzoporphyrin derivative (BPD, 1 µM). PIT used cetuximab–BPD immunoconjugates at BPD-equivalent 1 µM. Photosensitizer incubation: 90 min (PDT) or 24 h (PIT), with media refresh before illumination. Illumination: 690 nm diode laser at 150 mW/cm², with fluences 0, 1, 3, 10, 30, 60 J/cm² (n=3 biological replicates). Chemotherapy: Cisplatin at 1, 3, 10, 30, 100, 300 µM in Ovcar3–Jurkat 3D co-culture. - Readouts: After 3 days, plates were imaged by high-content microscopy. Custom MATLAB processing produced background-corrected fluorescence intensity metrics proportional to cell viability for tumour and T cells. - Dose–response fitting: Survival fractions S(d) for tumour (C) and T cells (E) were fit by least-squares to a sigmoidal function S(d)=S₀+(S∞−S₀)/[1+exp(μ_IC50·log d − log IC50)], estimating S₀, S∞, μ_IC50, and IC50 for each modality and cell type. - Tumour–immune model: An ordinary differential equation model (Kuznetsov et al.) of tumour–effector interactions with logistic tumour growth, immune influx, recruitment via immunogenic cell death, and effector clearance was used. Parameters (non-dimensional) largely followed the original: α=1.636, β=0.002, γ=1, σ=0.118, ρ=0.95, η=20.19, δ=0.374, μ=0.00311. Numerical integration used MATLAB ode45. State-space resolution spanned E∈(0,3.5)×10⁶ and C∈(0,450)×10⁶ with 30-day time steps. - Mapping treatment effects: For selected initial conditions (e.g., C₀=29×10⁶, E₀=4.1×10⁶) and later for all initial conditions leading to tumour escape, dose–response SFs were applied instantaneously to C and E to generate post-treatment states, then simulated forward to visualize trajectories on the C–E phase plane across dose ranges (chemo 1–100 µM; PDT/PIT 1–60 J/cm²). - Fractionated PIT simulations: For a given total dose, PIT was delivered as multiple fractions (e.g., 3 J/cm²×5; 6 J/cm²×10). Each fraction applied S(d) instantaneously to C and E, followed by ODE simulation for 60 days between fractions, proceeding until equilibrium for the final fraction. The minimum number of fractions (n_fx,min) required for tumour control was computed for fraction sizes (1, 3, 5, 7 J/cm², etc.) across initial conditions on the tumour-escape side of the separatrix.

- In vitro dose responses: • Chemotherapy (cisplatin) decreased T cell survival more than tumour cell survival across doses, implying limited ability to leverage immune-mediated control. • PDT produced similar reductions in tumour and T cells, with greater marginal effects at mid-range doses. • PIT exhibited two regimes: (i) low light doses (≤10 J/cm²) led to minimal tumour cytotoxicity but increased T cell numbers above control (SF_E>1), indicating immune stimulation; (ii) higher doses (>10 J/cm²) were cytotoxic to both, with T cell survival exceeding tumour survival. - Phase-plane mapping showed for representative initial conditions that single, unfractionated treatments (chemo, PDT, or PIT across their tested dose ranges) did not shift the system to immune control; trajectories remained in tumour-escape basins. - Fractionated PIT: • A single 15 J/cm² PIT fraction reduced tumour cells transiently but led to eventual escape. Delivering the same total dose as 3 J/cm²×5 increased effector cells after each fraction, crossed the separatrix, and achieved immune-mediated tumour control. • Similarly, 60 J/cm²×1 failed to control tumour, whereas 6 J/cm²×10 fractions led to cumulative effector stimulation and durable control. • The minimum number of fractions n_fx,min to achieve control was lowest near the separatrix and increased with (a) greater distance from the separatrix and (b) larger per-fraction doses (as immune stimulation decreased at higher doses).

The study integrates empirical dose–response data with a validated tumour–immune ODE model to evaluate how PIT dosing can reshape tumour–effector dynamics. Low-dose PIT uniquely stimulates effector (T) cells while sparing them relative to tumour cells, unlike chemotherapy (which is more toxic to T cells) or PDT (which reduces both comparably). Simulations demonstrate that fractionating a fixed total PIT dose into multiple low-dose fractions can cumulatively increase effector cells, shifting the system across the separatrix into the tumour-control basin, thereby enabling immune-mediated control without increasing total dose. This mechanistic rationale contrasts with traditional PDT fractionation aimed at tissue reoxygenation; here, the objective is immunostimulation. The findings suggest potential applications in advanced/recurrent EOC for managing disseminated disease and for priming tumours for checkpoint blockade. Practical light-delivery strategies (e.g., implantable or wearable low-irradiance sources) may facilitate fractionated regimens. However, translation requires careful calibration of the model to human EOC tumour–immune data and validation in prospective studies.

Fractionated, low-dose PIT can harness immunostimulatory effects to convert tumour escape into immune-mediated control in EOC models, as supported by integrated in vitro data and mathematical simulations. Delivering the same total light dose in multiple small fractions increases effector cell numbers and improves the likelihood of durable control. Future work should (1) calibrate and validate the model with patient-derived and syngeneic data, (2) extend the model to include immunosuppressive populations and oxygen/ROS dynamics, (3) elucidate mechanisms of T cell stimulation by low-dose PIT, (4) explore combinatorial strategies (e.g., EGFR-targeted PIT with checkpoint blockade or multi-antigen targeting), and (5) develop practical, low-irradiance light-delivery platforms for fractionated intraperitoneal therapy.

- In vitro co-cultures lacked cognate antigen-specific cytotoxicity; observed bystander effects may not fully represent in vivo T cell dynamics. - Fluorescent protein expression (EGFP/mCherry) could influence immune responses after phototherapy. - The tumour–immune ODE model parameters were calibrated from prior murine systems, not directly from human EOC; applicability requires re-calibration. - The model includes only effector cells and omits immunosuppressive populations and their modulation by PIT. - Oxygen consumption and ROS dosimetry were not modelled; immune-focused fractionation differs from oxygen-repletion strategies in PDT. - Phase-plane conclusions depend on chosen initial conditions and fitted dose–response parameters from specific cell lines and experimental conditions.