Threading light directly into the tumor when the surface can't reach deep enough
Surface-illuminated photodynamic therapy works beautifully for thin, accessible lesions — skin, esophageal mucosa, bladder wall. But red light at the PDT-relevant wavelengths (630–753 nm) only penetrates 5–10 mm of tissue before scattering and absorption extinguish the fluence below the cytotoxic threshold. Bulky or deep tumors — prostate, pancreas, deep head-and-neck masses — are simply out of reach from the surface. Interstitial PDT solves this by bringing the light source inside the tumor itself, and it starts with precise 3D imaging.
Photodynamic therapy depends on three components meeting simultaneously inside the target tissue: a photosensitizer, molecular oxygen, and light of the correct wavelength. Surface-applied light — from a diffuse fiber laid on the skin, an endoscopic probe, or an interstitial balloon in a hollow organ — attenuates exponentially with depth as it scatters off tissue microstructure and is absorbed by hemoglobin, melanin, and the photosensitizer itself.
At the red-shifted wavelengths PDT typically uses (630 nm for porfimer sodium, 660–690 nm for many second-generation sensitizers, 753 nm for the vascular-targeted agent padeliporfin), the 1/e optical penetration depth in typical soft tissue is on the order of 2–3 mm, meaning clinically useful fluence rarely persists past 5–10 mm from the irradiated surface. A tumor with a 3 cm radius has a core that is simply never reached by external or endoluminal light, no matter how long the exposure or how bright the source — the light is gone before it gets there.
Interstitial PDT reframes the geometry problem: instead of trying to push light further through tissue, place many small light sources directly inside the tumor volume, so that no point in the tumor is ever more than the effective penetration radius away from a source.
A single interstitial fiber has an effective cytotoxic radius of only about 5–10 mm — comparable to the light's own penetration depth. Treating a bulky tumor is therefore a coverage problem: enough fibers, correctly spaced, so their individual "hot zones" tile the entire volume with no gaps.
Just as in radiotherapy and brachytherapy planning, the interstitial PDT workflow begins with volumetric imaging — typically contrast-enhanced MRI, sometimes fused with CT or intraoperative ultrasound. A radiation oncologist or interventional radiologist contours:
• Gross Tumor Volume (GTV): the visible/measurable tumor extent on imaging • Clinical Target Volume (CTV): GTV plus a margin for microscopic extension, where applicable • Organs at risk (OARs): urethra and rectum (prostate), duodenum and major vessels (pancreas), major nerves and carotid vessels (head and neck)
The contoured volumes are exported as a 3D mesh into treatment planning software, which becomes the geometric substrate for every subsequent step — fiber count, trajectory, spacing, and the dose calculation that verifies the plan before a single fiber is inserted.
Once the tumor volume and critical structures are mapped, planning software must answer a deceptively hard geometric question: how many cylindrical-diffusing fibers, placed where, at what spacing, will deliver fluence above the cytotoxic threshold everywhere inside the tumor while staying a safe distance from everything outside it? The methodology is a direct descendant of interstitial brachytherapy seed and needle planning.
Interstitial PDT planning software mirrors the logic of interstitial brachytherapy (e.g., prostate seed implants or gynecologic HDR): the tumor volume is discretized into a fine 3D grid, and for every candidate fiber arrangement the software computes the predicted fluence (light dose) at every grid point by summing the contribution of each fiber's cylindrical diffusing source.
A cylindrical diffusing fiber does not emit from a single point — light exits uniformly along the active diffusing segment (1–5 cm, chosen to match the tumor's longest axis in that trajectory), producing a fluence field shaped like a fat cylinder with rounded ends rather than a sphere. The planning algorithm models each fiber as a line source and integrates the resulting fluence-versus-distance falloff (dominated by tissue scattering and absorption coefficients, μs' and μa) across the whole grid.
The optimizer then searches over fiber count, trajectory angle, insertion depth, and spacing to find the arrangement that (a) keeps ≥90–95% of the tumor volume above the cytotoxic fluence threshold, while (b) keeping fluence at every organ-at-risk surface below its damage threshold.
Because tissue optical properties are highly heterogeneous and light transport is a stochastic scattering process, simple exponential (Beer-Lambert) falloff models are only a first approximation. Rigorous interstitial PDT plans instead run Monte Carlo light transport simulations — the same class of algorithm used in radiotherapy dose calculation — in which millions of simulated photon packets are launched from each fiber's diffusing segment and tracked through a voxelized tissue model with per-voxel absorption and scattering coefficients until they are absorbed or exit the volume.
Summing the absorbed-energy maps from every fiber (respecting superposition, since light fields simply add) yields a full predicted 3D fluence-rate map for the entire multi-fiber array — directly analogous to how brachytherapy software sums dose contributions from multiple radioactive seeds. This overlap map is what planning software actually optimizes against: it reveals the "gaps" between fibers where fluence dips below threshold before a single needle is ever placed in the patient.
Rule of thumb used in clinical prostate interstitial PDT planning: fibers spaced 1–1.5 cm apart give adequate field overlap, because each fiber's own effective cytotoxic radius is only about 5–10 mm — space them much wider and the fields simply do not touch, leaving an under-treated corridor between fibers.
A plan is only as good as its execution. Translating a computed trajectory map into physical fiber placement requires both a purpose-built optical device — the cylindrical diffusing fiber — and a precise, image-guided percutaneous or catheter-based insertion technique, borrowed directly from interventional radiology and brachytherapy afterloading.
A plain cleaved optical fiber emits light from a small spot at its tip — a point source, useful for surface contact PDT but poorly suited to a long, cigar-shaped tumor volume. Interstitial PDT instead uses cylindrical diffusing fibers, in which the distal segment of the fiber core (typically 1–5 cm, selected per-patient to match the tumor's long axis along that trajectory) has its cladding modified — by roughening, doping, or etching a diffusing microstructure — so that light guided down the fiber leaks out radially and roughly uniformly along the entire active length, rather than concentrating at the tip.
The result is a fluence field shaped like an elongated cylinder with rounded ends, closely matching the geometry planning software modeled in the previous stage. Diffuser uniformity (typically specified as ±20–30% variation along the active length) is a key manufacturing quality metric, since hot or cold spots along the fiber translate directly into over- or under-treated tissue.
Fibers are packaged inside thin flexible catheters (~1–1.8 mm outer diameter) that can be pre-placed like brachytherapy needles, allowing the diffusing fiber itself to be threaded in, treatment delivered, and the fiber withdrawn while the catheter position is preserved for verification or re-treatment.
Fibers are inserted using the same stereotactic principles as brachytherapy or biopsy-needle placement: a template grid or robotic guidance array registers the planned 3D trajectories to the patient's anatomy under real-time MRI, CT fluoroscopy, or transrectal ultrasound (prostate). Each catheter is advanced along its planned trajectory to the planned insertion depth, and its final position is confirmed against the plan before treatment begins — any deviation greater than a few millimeters can be corrected by adjusting fiber depth or triggering a re-plan.
For prostate vascular-targeted PDT (padeliporfin/TOOKAD), fibers are placed transperineally through a brachytherapy-style template under general or spinal anesthesia, typically as a single same-day outpatient procedure. For pancreatic and deep head-and-neck tumors, fibers are more often placed percutaneously under CT or intraoperatively under direct/laparoscopic visualization, given the proximity of bowel, major vessels, and other radiosensitive structures.
Because interstitial PDT fibers are non-radioactive and produce no ionizing radiation, the insertion and dwell workflow avoids the radiation-safety shielding and afterloading logistics required for HDR brachytherapy — while still reusing its stereotactic placement accuracy, typically within 1–3 mm of the planned trajectory.
With every fiber positioned at its planned coordinates, treatment begins: all diffusing fibers are activated together, each pumping light from an external laser source down its own strand. The therapeutic effect emerges not from any single fiber but from the superposition of every fiber's radial field — the same physical principle the planning software modeled, now playing out in the actual tumor.
Light fluence, like radiation dose, is additive: at any point inside the tumor, the total fluence delivered is simply the sum of the contributions from every fiber whose field reaches that point. A lone fiber creates a well-covered cylindrical "hot zone" immediately around itself that fades below the cytotoxic threshold within roughly 5–10 mm — but the region exactly midway between two fibers, spaced at the planned 1–1.5 cm interval, receives meaningful contributions from both, and the sum clears the threshold even though neither fiber alone would cover that point.
This is precisely why fiber spacing is the critical planning variable: too sparse, and gaps persist between fibers where fluence stays sub-threshold (visible as grey underdosed pockets in the coverage map); too dense, and fibers are wasted, insertion trauma increases, and fluence near critical structures can spike above safe limits. All fibers illuminating simultaneously — rather than sequentially — also matters biologically: PDT consumes tissue oxygen as it generates cytotoxic singlet oxygen, and simultaneous multi-fiber illumination distributes that oxygen demand more evenly than treating one region at a time, reducing the risk of localized hypoxia blunting the effect.
Photobleaching and photosensitizer/oxygen depletion mean the relationship between light dose and cytotoxic effect is not perfectly linear — real-time or model-based dosimetry during illumination (rather than a pre-treatment plan alone) is increasingly used clinically to catch under-dosed regions while the fibers are still in place.
Unlike traditional photosensitizers that accumulate inside tumor cells over hours to days, padeliporfin (a water-soluble bacteriochlorophyll derivative) stays confined to the vasculature and is activated within minutes of intravenous infusion. Illuminating the tumor with 753 nm light while the drug is still circulating triggers rapid, localized thrombosis and occlusion of the tumor microvasculature — the tumor is destroyed by acute vascular shutdown and ischemic necrosis rather than direct cytotoxicity to individual cells.
This vascular-targeted approach is why the fiber layout must so precisely track the tumor volume's actual blood supply geometry, and why real-time confirmation that every planned voxel received adequate fluence matters clinically — a missed pocket of tumor vasculature can remain perfused and viable even when the surrounding tissue is thoroughly treated.
The final step closes the loop opened at imaging: confirm, quantitatively, that the delivered fluence actually covered the tumor volume above the cytotoxic threshold, and that critical structures stayed within their safety margin. This verification step — plus a growing evidence base in prostate and pancreatic cancer — is what has moved interstitial PDT from an experimental concept to an approved clinical therapy.
Two complementary verification strategies are used clinically. Real-time light dosimetry places small optical probes (isotropic detectors on thin catheters, positioned during planning at representative points — tumor center, near-margin, and adjacent-to-OAR locations) that measure actual delivered fluence rate throughout illumination, allowing the treating team to extend exposure time or flag under-dosed regions before the fibers are withdrawn.
Post-hoc verification uses the same Monte Carlo/analytic light-transport model from the planning stage, now fed with the as-placed fiber coordinates (confirmed by post-insertion imaging) rather than the planned ones, to recompute the actual delivered fluence map. Comparing this "as-treated" map against the cytotoxic threshold across the full tumor volume gives a coverage percentage analogous to a brachytherapy post-implant dosimetry report — the interstitial PDT equivalent of confirming D90 coverage in a seed implant.
This planning-to-verification loop — image-based target definition, dose-modeled trajectory optimization, image-guided placement, and post-treatment coverage confirmation — is structurally identical to modern interstitial brachytherapy workflow, just substituting photodynamic fluence for ionizing radiation dose.
Vascular-targeted interstitial PDT with padeliporfin (TOOKAD Soluble) is the most clinically mature application, approved in Europe for low-risk localized prostate cancer as an alternative to active surveillance or radical treatment. In the pivotal PCM301 phase III trial, roughly 58% of VTP-treated patients converted to a negative repeat biopsy at 24 months versus about 28% on active surveillance, with significantly reduced rates of disease progression to radical therapy — while preserving continence and erectile function far better than surgery or radiotherapy, since the vascular-targeted mechanism spares the urethra and neurovascular bundles when fibers are planned with adequate margin.
Interstitial PDT for pancreatic cancer — where surgical resection is frequently impossible and the tumor sits adjacent to major vessels and the duodenum — has been explored since early feasibility studies (e.g., verteporfin-based interstitial PDT, Bown et al.) demonstrated that CT/ultrasound-guided fiber placement could produce localized, controllable zones of tumor necrosis in locally advanced disease. It remains an active area of early-phase clinical investigation, with the same core planning-and-verification methodology described here being adapted to the pancreas's tighter safety margins and more complex vascular anatomy.