A locally treated tumor can teach the immune system to hunt down its distant twin
Photodynamic therapy (PDT) combines a light-activatable photosensitizer, tissue-penetrating light of a matched wavelength, and molecular oxygen to generate cytotoxic reactive oxygen species (ROS) inside tumor cells. Unlike the comparatively immunologically silent apoptosis induced by many conventional chemotherapies, PDT inflicts an abrupt burst of oxidative and endoplasmic reticulum (ER) stress — a form of cellular injury that the immune system is primed to interpret as danger.
PDT works through a three-part reaction: a photosensitizer accumulates preferentially in tumor tissue (often localizing to the ER, mitochondria, or plasma membrane), light of a specific wavelength excites it to a reactive triplet state, and the excited photosensitizer transfers energy to molecular oxygen — generating singlet oxygen (¹O₂) and other reactive oxygen species within nanometers of where the photosensitizer sits.
Because this ROS burst is generated intracellularly and instantaneously upon irradiation, it produces a much more acute, spatially concentrated stress event than most chemotherapies, which act over hours-to-days through slower biochemical mechanisms (DNA damage, mitotic arrest). Photosensitizers that localize to the ER — such as hypericin — are especially potent inducers of ER stress, because ROS generated there directly disrupts protein folding machinery.
The subcellular localization of the photosensitizer matters enormously: ER-targeted photosensitizers reliably trigger the unfolded protein response and immunogenic cell death, while photosensitizers that localize only to lysosomes or the plasma membrane are far less immunogenic — the ER stress pathway is the switch that converts cell death into a vaccine-like event.
Not all dying cells alert the immune system. Billions of cells die every day in the body through routine turnover, and this "tolerogenic" apoptosis is actively anti-inflammatory: dying cells display "eat-me" signals like phosphatidylserine that prompt phagocytes to clear them quietly, releasing anti-inflammatory cytokines (IL-10, TGF-β) and maintaining self-tolerance.
Immunogenic cell death (ICD) — a concept formalized by Guido Kroemer and Laurence Zitvogel's laboratories — is a distinct death modality in which dying cells actively emit a coordinated sequence of damage-associated molecular patterns (DAMPs) that flip this default: calreticulin exposure recruits phagocytes for engulfment rather than tolerance, HMGB1 engages TLR4 on dendritic cells to drive maturation, and extracellular ATP acts as a "find-me" signal recruiting immune cells and activating the NLRP3 inflammasome. The specific combination and kinetics of these signals — not cell death alone — is what determines whether a dying tumor cell becomes immunologically invisible or becomes an adjuvant that trains the immune system against it.
Radiotherapy, surgery, and many cytotoxic chemotherapies (e.g., 5-fluorouracil, gemcitabine) predominantly induce classical apoptosis with minimal DAMP exposure — a largely tolerogenic form of cell death that leaves the immune system unengaged. Only a small subset of chemotherapeutics (anthracyclines like doxorubicin, oxaliplatin) have been shown to reliably induce bona fide ICD.
PDT joins this select group of ICD inducers, and arguably induces it more reliably and acutely than most drugs, because the oxidative/ER stress mechanism is intrinsic to its mode of action rather than an incidental side effect. This has made PDT a subject of active study — pioneered by researchers like Pawel Mroz, Michael Korbelik, and Patrizia Agostinis since the early-to-mid 2000s — as a localized "in-situ vaccination" strategy rather than merely a tumor-ablation technique.
As PDT-stressed tumor cells progress toward death, they execute a choreographed sequence of damage-associated molecular pattern (DAMP) exposure and release. These three signals — surface calreticulin, extracellular HMGB1, and extracellular ATP — form the canonical molecular signature that the Kroemer and Zitvogel laboratories established as the defining hallmarks of immunogenic cell death.
Under ER stress, calreticulin (CRT) — normally a resident ER chaperone protein — is translocated to the outer leaflet of the plasma membrane in a process requiring PERK-mediated eIF2α phosphorylation, caspase-8 activity, and SNARE-dependent exocytosis. This "ecto-CRT" is exposed on the cell surface while the cell is still alive or in early apoptosis, well before membrane integrity is lost.
Ecto-CRT binds LRP1/CD91 on phagocytes and dendritic cells, providing a potent pro-phagocytic "eat-me" signal that overrides the "don't-eat-me" signals (like CD47) that would otherwise protect the cell from engulfment. Because PDT generates ER stress so directly and acutely, it is an unusually efficient inducer of calreticulin translocation compared to death stimuli that bypass the ER.
High-mobility group box 1 (HMGB1) is a nuclear DNA-binding protein released passively once plasma membrane integrity is lost during late apoptosis/necrosis. Once extracellular, HMGB1 binds Toll-like receptor 4 (TLR4) on dendritic cells, providing a critical maturation signal that licenses dendritic cells to efficiently cross-present captured antigens to T cells.
Extracellular ATP is actively secreted by dying cells via lysosomal exocytosis and pannexin-1 channels prior to full membrane rupture. ATP acts as a chemotactic "find-me" signal that recruits monocytes and dendritic cell precursors to the site of cell death, and engages the P2X7 purinergic receptor on immune cells to trigger NLRP3 inflammasome assembly and IL-1β secretion — providing an additional pro-inflammatory signal that reinforces immune activation.
Together, calreticulin (recognize), ATP (recruit), and HMGB1 (activate) form a temporally layered signaling code: exposure before death, secretion during death, and passive release after death — each step recruiting and instructing a different arm of the innate immune response.
A tumor cell dying with all three DAMPs present in the correct sequence is treated by the immune system as equivalent to a vaccine adjuvant — this is the molecular basis for describing PDT-treated tumors as generating an "in-situ vaccination" effect.
The magnitude of DAMP release scales with PDT dose intensity (photosensitizer concentration × light fluence). Sub-lethal or very low-dose PDT can leave cells alive without triggering meaningful ER stress; excessively high doses can drive rapid necrosis that bypasses the regulated calreticulin translocation pathway and instead causes uncontrolled, less-organized DAMP spillage. An intermediate, carefully titrated PDT dose that drives cells through the regulated apoptotic-ICD pathway — with calreticulin surface exposure preceding membrane rupture — appears to generate the most reliable, high-quality immunogenic signal for downstream dendritic cell activation.
DAMP gradients emanating from the PDT-treated tumor act as a homing beacon for immature dendritic cells (DCs) patrolling nearby tissue and blood vessels. Once at the site, calreticulin exposure on dying tumor cells converts DCs from passive bystanders into active phagocytes that engulf tumor debris — capturing the tumor-specific antigen cargo needed to launch an adaptive immune response.
Immature dendritic cells reside in a surveillance state throughout peripheral tissues, sampling their environment for danger signals. Extracellular ATP released from PDT-stressed cells diffuses outward, forming a concentration gradient that immature DCs sense via P2Y2 purinergic receptors, driving directed migration toward the tumor bed. HMGB1 contributes complementary chemotactic and activating cues.
This recruitment step is critical: without sufficient DC infiltration into the irradiated tumor microenvironment, even abundant DAMP signaling and antigen availability cannot be converted into an adaptive immune response, because there are no antigen-presenting cells present to capture it.
Once at the tumor site, dendritic cells encounter PDT-stressed cells displaying surface calreticulin. Engagement of CRT by the DC receptor LRP1/CD91 triggers active phagocytic engulfment of the dying or dead tumor cell material — including cytoplasmic and nuclear contents carrying tumor-specific antigens (neoantigens arising from tumor mutations, along with tumor-associated self-antigens).
This internalization step is the physical bridge between innate danger sensing and adaptive immune priming: the DC is no longer simply detecting that something died nearby, it is now physically carrying molecular fragments of the tumor that it will later present to T cells. The cDC1 subset of dendritic cells (dependent on the transcription factor BATF3) is particularly important here, as this subset specializes in cross-presentation — the ability to route exogenously captured antigen into the MHC class I pathway needed to prime CD8+ cytotoxic T cells.
Cross-presentation by cDC1 dendritic cells is the linchpin step connecting a local photochemical reaction to a systemic cytotoxic T-cell response — without it, engulfed tumor antigen would only ever generate a CD4+ helper response, not the CD8+ killer cells needed for direct tumor cell elimination.
The local cytokine milieu generated alongside DAMP release — including type I interferons, IL-1β (downstream of the ATP/P2X7/NLRP3 axis), and pro-inflammatory chemokines — further conditions recruited DCs toward an immunostimulatory rather than tolerogenic phenotype. This is important because DCs are context-dependent: the same phagocytic capture event that occurs during routine tissue turnover typically produces tolerogenic, not immunogenic, DC programming. It is specifically the co-incidence of antigen uptake with DAMP and cytokine signaling that redirects DC differentiation toward a mature, immunostimulatory state in the next stage.
Antigen-loaded dendritic cells, having received DAMP and cytokine signals at the tumor site, undergo a maturation program: they upregulate co-stimulatory molecules and chemokine receptors, migrate to the nearest tumor-draining lymph node, and present processed tumor antigen on MHC class I to naive CD8+ T cells — priming a clonal, tumor-specific cytotoxic T-cell response.
Naive T-cell activation classically requires three coordinated signals from the dendritic cell: Signal 1 is antigen-specific — a peptide derived from the engulfed tumor material, presented on MHC class I (for CD8+ T cells) or MHC class II (for CD4+ T cells), recognized by the T-cell receptor. Signal 2 is co-stimulation — CD80/CD86 on the mature DC engaging CD28 on the T cell, without which antigen recognition alone leads to T-cell anergy rather than activation. Signal 3 is cytokine polarization — inflammatory cytokines such as IL-12 and type I interferons, produced by DCs that received strong danger signals, that shape the T cell toward an effector cytotoxic phenotype rather than a regulatory one.
Because PDT-induced ICD supplies robust DAMP and cytokine signaling upstream, dendritic cells arriving at the lymph node are far more likely to deliver all three signals in a coordinated, activating configuration — rather than the incomplete signaling that leads to T-cell tolerance.
The dendritic cell cross-presentation pathway allows exogenously captured antigen (engulfed tumor cell debris) to be routed into the MHC class I presentation pathway normally reserved for endogenously synthesized proteins. Captured antigen is processed in endosomal compartments and either transported into the cytosol for proteasomal degradation and TAP-mediated loading onto MHC-I, or loaded directly within specialized endosomal compartments — both routes converge on the same outcome: peptide-MHC-I complexes displayed on the DC surface.
Within the draining lymph node, mature DCs present these tumor peptide-MHC-I complexes to a repertoire of circulating naive CD8+ T cells. T cells bearing a T-cell receptor that recognizes the specific tumor peptide are triggered to proliferate clonally, differentiate into effector cytotoxic T lymphocytes (CTLs) equipped with perforin and granzyme killing machinery, and exit the lymph node into circulation.
Because the antigens being cross-presented are tumor-specific (mutation-derived neoantigens or overexpressed tumor-associated antigens), the resulting CD8+ T-cell clones are specifically armed to recognize tumor cells anywhere in the body that share the same antigen profile — this is the cellular basis for the abscopal effect in the next stage.
Primed, tumor-antigen-specific CD8+ T cells generated in the draining lymph node exit into systemic circulation. Because these T cells recognize antigens shared between the treated and untreated lesions, they can traffic to and attack distant, previously untreated tumor deposits — a phenomenon termed the abscopal effect, transforming a spatially localized ablative therapy into a systemic cancer immunotherapy.
The term "abscopal effect" (from Latin ab scopus, "away from the target") was first coined in 1953 by R.H. Mole to describe regression of tumors outside a radiotherapy field. For decades this was considered a rare curiosity of radiation oncology. As the mechanistic understanding of immunogenic cell death matured through the 2000s and 2010s, it became clear that the abscopal effect is fundamentally an immune-mediated phenomenon: any local therapy capable of triggering sufficient ICD and antigen release can, in principle, generate a systemic T-cell response that reaches untreated lesions.
PDT is a compelling ICD-driving modality for eliciting abscopal responses because — unlike radiotherapy, which can also be locally immunosuppressive at high doses through lymphocyte depletion — PDT's cytotoxic mechanism is largely confined to light-exposed, photosensitizer-laden tissue, sparing circulating lymphocytes and preserving the peripheral immune compartment needed to mount a robust systemic response.
In both radiotherapy and PDT literature, abscopal regression of untreated lesions remains a relatively infrequent, hard-to-reliably-reproduce outcome when the local therapy is used alone. Tumors employ multiple immune evasion strategies — upregulating checkpoint ligands like PD-L1, recruiting immunosuppressive regulatory T cells and myeloid-derived suppressor cells, and depleting the tumor microenvironment of infiltrating effector T cells — that can neutralize an otherwise well-primed systemic T-cell response before it can eliminate distant disease.
This has motivated a wave of clinical and preclinical trials combining PDT with immune checkpoint inhibitors — antibodies blocking PD-1/PD-L1 or CTLA-4 — under the rationale that PDT supplies the "priming" arm (antigen release, DC activation, T-cell generation) while checkpoint blockade removes the "brakes" that would otherwise blunt the T-cell attack on distant tumor sites. Early trials combining PDT with anti-PD-1 therapy have reported encouraging increases in both local and distant tumor control compared to either modality alone.
The conceptual framing that has emerged is PDT as an in-situ tumor vaccine: rather than manufacturing a vaccine from isolated tumor antigens ex vivo, PDT generates the vaccine directly inside the patient's own tumor, using the tumor's own antigenic repertoire, and checkpoint blockade acts as an adjuvant that unlocks the full potential of the resulting T-cell response.
The overall arc from Stage 1 to Stage 5 illustrates a broader conceptual shift in oncology: from viewing local therapies (surgery, radiation, ablation, PDT) purely as means of physically destroying a tumor mass, to viewing them as potential triggers of systemic antitumor immune memory. A tumor destroyed by PDT is not just gone — if it dies immunogenically, it leaves behind a trained population of tumor-specific T cells patrolling the body, capable of recognizing and eliminating micrometastatic disease and distant lesions that were never directly treated.
This reframing is part of the rationale driving a broader wave of combination strategies pairing local ablative or cytotoxic therapies with systemic immunotherapy — with PDT positioned as an especially attractive partner given its favorable immunogenic cell death profile, minimal systemic toxicity, and repeatability at accessible tumor sites.