Treatment planning evolution: Comparing approaches in photodynamic and radiation therapies

Abstract

The setups of previous and ongoing clinical trials are based on prescribed PDT doses, PS concentration, and light intensity derived from averages of previous clinical or study populations. It is understood that monitoring of personalized PS and light dose is needed to improve PDT outcomes. Monitoring of photophysical, photochemical, or cytotoxic moieties is common, representing concepts of delivered, absorbed, or equivalent doses similar to those used in radiation therapy (RT). Unlike RT, these dose concepts are not equally well developed and standardized across the PDT clinical indications; however, there is potential to improve PDT treatment setup, planning, and delivery by leveraging methodologies from RT. This review summarizes dose definitions and advancements in RT treatment planning and presents the equivalent dose concepts for PDT, particularly how these concepts can expand on the existing methods for PDT treatment planning. By identifying the major limitations and areas for improvement in PDT planning, the hope is to stimulate preclinical and clinical research studies that can enhance the efficacy of PDT, improving patient outcomes.

Oncological therapy based on radiation therapy (RT) is well‐developed, with widely accepted therapeutic dose concepts based on its mechanisms of action, which are applied to personalized treatment planning using sophisticated programs and tools. While the mechanisms of action are well understood for photodynamic therapy (PDT), universal dose definitions are not generally accepted. Dose concepts range from the administered drug dose and light energy to the generated singlet oxygen concentration. Here, we outline similarities between RT and PDT dose metrics and discuss how these dosimetry concepts are currently, and could be, applied to PDT treatment planning.

Abbreviations

ART

adaptive radiation therapy

CAD

computer‐aided design

CT

computed tomography

CTV

clinical target volume

DFT

diffuse fluorescence tomography

DNA

deoxyribonucleic acid

GTV

gross tumor volume

Gy

Gray

IMRT

intensity‐modulated radiotherapy

LET

linear energy transfer

MLC

multileaf collimators

MRI

magnetic resonance imaging

OAR

organs at risk

OCT

optical coherence tomography

PDT

photodynamic therapy

PET

positron emission tomography

PS

photosensitizer

PTV

planning treatment volume

RBE

relative biological effectiveness

ROS

reactive oxygen species

RT

radiation therapy

SOLD

singlet oxygen luminescence dosimetry

SV

Sievert

US

ultrasound

VMAT

volumetric‐modulated arc therapy

INTRODUCTION

A patient's cancer treatment is determined based on the malignancy's origin, progression, and genetic composition. Standard treatment methods for solid tumors include surgery, radiation therapy, chemotherapy, and, more recently, targeted biologics and immunotherapies. ¹ Surgery remains the primary approach for reducing the imminent cancer burden. In the 1890s, radiation therapy (RT) was introduced to manage localized diseases. Over time, it was found that combining surgery and radiation was more effective than using either as a stand‐alone therapy for cancer control. ¹ Chemotherapy is a systemic treatment often used to manage distant and metastatic diseases, including microscopic diseases. Cancer chemotherapy has advanced considerably beyond the initial nitrogen mustard in the 1920s, providing an ever‐increasing number of available drugs. ¹ Recently, small or smart targeting molecules, known as biologics (ie, kinase inhibitors or various enzyme inhibitors), have been developed and integrated into clinical practice. ² The results are promising when combined with radiotherapy or traditional drugs such as cisplatin. However, radiotherapy and surgery still play the primary role in achieving long‐term disease control in solid tumors. ¹

Novel physical cancer treatment, including photodynamic therapy, photothermal therapy, focused ultrasound, and irreversible electroporation, provides spatial/temporal confinement of energy deposition, reducing the cytotoxic load and potentially lowering the patient's risk of morbidity. The development of these techniques as stand‐alone or adjuvant therapy options is often motivated by their potential to reduce morbidity/toxicity, treatment duration, and, ultimately, the cost of treatment associated with traditional therapies. ³

Given the lack of sophisticated personalized treatment planning in emerging physical cancer therapies, it is instructive to review the evolution of treatment planning in RT, including external beam and brachytherapy, as they can provide a guideline to develop robust planning procedures for these novel physical therapeutic approaches specifically focused on PDT.

Overview of cancer radiobiology

In radiotherapy, ionizing energy encompasses various types, including electromagnetic particles like photons and subatomic particles such as electrons, protons, neutrons, or other charged heavy ions. The capacity of these particles to transfer energy to living tissue is quantified by linear energy transfer (LET). ⁴ Different forms of ionizing energy exhibit distinct LET levels, influenced by their interactions with body tissues. Charged particles—such as protons, electrons, and heavy ions—directly ionize molecules within cells. In contrast, neutral particles like photons and neutrons induce indirect ionization by interacting with atoms, ultimately producing charged electron particles. Photons are the most common type of radiation used to treat patients. ⁴ The effectiveness of diverse radiotherapy techniques is standardized using a parameter called relative biological effectiveness (RBE), a function of LET. ⁴

Particles charged from the radiation beam, either directly or indirectly, damage cells by inducing DNA damage, leading to disruptions in the DNA structure. This damage can occur through two primary mechanisms: direct ionization leading to the breakage of the phosphodiester backbone or the formation of free radical species that cause similar damage. ⁵ Double‐strand breaks (DSBs) induced by radiation are the most harmful form of DNA damage, which can lead to cell death if not repaired. ⁵ Nevertheless, mechanisms responding to DNA damage serve as a crucial defense against damage caused by radiation.

In radiation modalities with LET values comparable to photons, about two‐thirds of the DNA damage is caused by the indirect generation of free radical species, while the remaining one‐third is due to direct DNA damage. ⁵ Low LET radiation deposits a relatively small amount of energy, while radiation particles (positive or negatively charged) deposit a larger energy quantity in each interaction, resulting in more significant biological effects. ⁵

Radiation therapy can be administered externally or internally. The most common method, external beam radiation, involves directing energy from a radioactive source outside the body, focusing and shaping it to target the tumor. ⁶ Conversely, brachytherapy involves placing radioactive sources directly into or near the tumor, delivering high doses of radiation to a localized area while minimizing damage to surrounding healthy tissue. While external beam radiation therapy is applicable for various internal organs, brachytherapy is utilized primarily for treating cancers such as cervical, eye, and prostate cancer through intracavitary and interstitial procedures. ⁷

Ionizing radiation dose definitions and dose confinement

Understanding the correlation between delivered power or energy dose and the resulting cytotoxic effect (due to the deposited energy) is paramount to predicting biological and clinical outcomes for all physical cancer therapies. In ionizing radiation therapy, one encounters: (1) The absorbed dose, measured in Gray (Gy or J∙kg⁻¹), signifies the energy density delivered by the ionizing radiation and transferred to the target per unit mass of material. (2) Equivalent dose (Sv) considers the relative biological effect for the specific kind of radiation, denoted as 1 Sv = 1 Gy W _R, where W _R is the radiation weighting factor that accounts for the type of radiation and its biological effectiveness, as described in reference. ⁸ (3) The effective or cytotoxic dose to an organ within the irradiation field is the tissue‐weighted sum of the equivalent doses. It is obtained by multiplying the equivalent dose with the tissue weighting factor, denoted as W _T, as detailed in reference. ⁹ This multiplication accounts for the radiosensitivity of a particular tissue, providing a more nuanced understanding of the therapy's impact. The equivalent and effective doses represent the stochastic health risk to the whole body, which is the probability of cancer induction and genetic effects of low levels of ionizing radiation. ⁹ Both concepts are predominantly used in radiation protection during RT or environmental exposure to ionizing radiation. ¹⁰

Based on these dose definitions, treatment planning in radiation therapy aims to improve the spatial precision and accuracy of the delivered cytotoxic dose. The goal is to achieve complete tumor volume coverage with the minimum cytotoxic dose required to cause complete cell death without exceeding the maximum permissible dose for the target and, more importantly, the healthy surrounding tissue. This is generally defined by an acceptable therapeutic ratio—given by the probability of tumor control versus the probability of unacceptable toxicity. Excessive toxicity could also result in morbidity due to acute radiation poisoning. ⁹

Evolution of treatment planning in radiation therapy

The first mathematical models for RT dosimetry were developed in the 1920s to calculate the delivered radiation dose [Gy] in the planning volume to generate “isodose” distribution diagrams. ¹¹ These models used simple geometry and assumed a uniform radiation dose distribution within the tissue, independent of the radiation weighting factor. Advances in computer technology in the 1970s led to the development of imaging modalities such as computed tomography (CT), which improved patient anatomy description and positioning for dose optimization. ¹¹ The use of conformal radiotherapy with advanced 3D planning techniques enhanced the understanding of the relationship between the ionizing radiation dose and toxicity, resulting in the introduction of the equivalent and effective dose in the planning process. ¹¹ Traditionally, treatment utilized naturally occurring radioactive elements such as cobalt 60, which emitted photons during decay. However, in contemporary practice, photons are produced by linear accelerators. These machines accelerate a stream of electrons towards a target, causing atomic interactions that generate photons. The photons are then directed at the patient via a mobile gantry, and motorized collimators shape the radiation beam as it exits the gantry head. ¹² The multileaf collimators (MLC), introduced in the 1980s, permitted photon‐beam shaping and intensity control, permitting full use of the equivalent and effective dose paradigms. ¹¹ Contemporary radiotherapy relies on pre‐therapy imaging of a patient's anatomy to identify the gross tumor volume (GTV), specifying the visible tumor, and to conduct dose deposition simulations throughout the planning treatment volume (PTV), using the established and widely recognized weighting factors W _R and W _T. The PTV encompasses the clinical target volume (CTV), which includes tissue that may contain microscopic disease and a margin for uncertainties in patient setup and physiological motion and changes. This is known as internal target volume (ITV). PTV can also include the organs at risk (OAR). ¹

The term “3DCRT”, three‐dimensional conformal radiation therapy, is associated with “forward” treatment planning, in which the planner chooses the suitable field sizes, directions, and other relevant parameters for the calculation and the optimization process, avoiding critical structures, such as the central nervous system and a significant fraction of the bone marrow. Conversely, intensity‐modulated radiotherapy (IMRT) and volumetric‐modulated arc therapy (VMAT) are modern treatments that offer greater versatility than 3DCRT. An “inverse” planning strategy is employed for each, meaning that the planner defines the dose for target and OAR limitations while assigning weight prioritization. ¹² This advancement allows personalized treatment plans considering the tumor's location, size, shape, and radiation beams required to exceed the minimum radiation dose across the tumor volume. ¹¹ , ¹²

Monitoring the cytotoxic dose administered per radiation fraction is restricted in accommodating expected variations in a patient's positioning or anatomy during ionizing radiation delivery. Providing large margins around the tumor is one approach to ensure adequate CTV exposure; however, increasing tumor margins can lead to higher doses in the OARs. Using bolus materials, such as tissue‐equivalent substances or specialized gels, is another common technique in radiotherapy to shape the radiation beam more precisely. ¹³ Breath‐hold techniques and compression belts are also utilized to minimize motion artifacts. ¹⁴

More recently, adaptive radiation therapy (ART) has enabled the adjustment of a treatment plan to enhance radiation dose confinement due to anatomical or physiological variations that may have occurred since the initial simulation. Modern ART can be categorized into three main types: offline, online, and real‐time adaptation. Offline adaptation involves adjusting a patient's treatment plan after delivering one or more treatment fractions, with activities like re‐simulation, re‐contouring, and re‐planning, similar to the initial plan created. Online adaptation focuses on plan adjustments immediately before delivering each treatment fraction, typically through re‐contouring and re‐planning using data from image‐guided radiation therapy (IGRT). Real‐time adaptation automatically modifies the treatment plan during the actual delivery of each treatment fraction. It relies on real‐time imaging to either gate the treatment beam or track the target using the MLC, allowing adjustments based on the dynamic changes due to breathing, positioning, and organ movement. ¹⁵ Furthermore, 4D CT (4D‐CBCT) or 4D MRI are sometimes employed for motion correction during treatment, ensuring precise dose delivery. ¹⁶ , ¹⁷

Current RT limitations

Significant progress has been made in RT cancer treatment, but its primary limitation remains its impact on healthy organs. The side effects, particularly long‐lasting and occasionally severe, present considerable challenges in expanding this approach to cancer therapy. Such side effects are common in deep‐seated tumors, where a substantial amount of healthy tissue unavoidably receives radiation to treat the cancerous area effectively. Increased awareness of these harms in the 1980s and beyond led to reduced radiotherapy use in pediatric and hematologic malignancies, where chemotherapy became the preferred treatment. ¹⁸ , ¹⁹ Despite the utilization of advanced radiotherapy techniques, deep‐seated malignancies like prostate cancer may still exhibit a considerable 20% incidence of substantial long‐term side effects. ²⁰ Even advanced equipment like the Gamma Knife (Elekta) allows for the use of PTV on sub‐mm‐scales for treating deep‐seated intracranial structures, ²¹ but still presents a notable risk of side effects, including unintended damage and brain tissue radionecrosis, occurring at a rate of 5–10%. ²² On the other hand, brachytherapy faces unique challenges in modern radiotherapy. These include dosimetry challenges related to the precise placement of radioactive materials, their invasive nature, potential radiation exposure to staff, and complications arising from proximity to critical organs. ²³ Patient‐specific factors and the availability of specialized equipment further limit its accessibility, with a notable disparity in the global distribution of brachytherapy facilities, making it less available in certain regions. ²³

Developing treatment planning for novel physical therapies, such as photodynamic therapy, photothermal therapy, focused ultrasound, and irreversible electroporation, introduces distinct hurdles. These therapies generally require only one session, eliminating the need for re‐planning; however, precision in personalized treatment planning remains essential. For focused ultrasound, the physical energy is delivered interstitially directly to the tumor and is highly confined, removing the need to consider motion artifacts. ²⁴ However, for modalities like photodynamic therapy and photothermal therapy, the energy confinement, and hence the deposition of a cytotoxic dose, is governed by diffusion processes for photons, electrons, and heat, respectively. ²⁵ Unlike conventional therapies, concepts such as deposited energy density and weighting factors cannot be directly applied to these novel physical therapies, necessitating a thorough understanding of the target's physical properties for accurate treatment planning. The following section focuses on PDT as an example of novel physical therapy, defining the dose concept, dosimetry, treatment planning (including pretreatment planning status), current limitations, how PDT can learn from radiation treatment planning and future directions.

Photodynamic therapy

Photodynamic therapy (PDT) uses light‐active drugs called photosensitizers (PS) to generate spatial and temporal controlled cytotoxicity, leading to cell death. The basic concept of PDT dates back a hundred years. From the 1970s to 1990, initial clinical therapies emerged for treating skin and bladder tumors. ²⁶ , ²⁷ Photofrin was the first PS to receive FDA approval in 1995 for esophageal cancer treatment. ²⁶ , ²⁸ Compared to traditional cancer treatment methods like radiotherapy and chemotherapy, PDT is less invasive, has shorter side effect durations, and is often performed as an outpatient procedure, providing advantages for healthcare providers and patients. ²⁹ , ³⁰ The therapy's confinement is due to selective accumulation and retention of the PS in highly proliferating tissues and the confinement of the activation light photons due to high scattering and absorption within biological tissues. PDT can destroy tumor cells directly or via occlusion of tumor‐associated vasculature in the CTV, depriving the cells of oxygen and nutrients. PDT is typically administered in a single session for most conditions; however, unlike chemotherapy or radiation therapy, there are no restrictions on repeating treatments at the same site. PDT can also stimulate the host's immune system against tumor cells by inducing inflammation and recruiting leukocytes to the target area. ³¹ , ³² In contrast, conventional anticancer treatments, particularly surgical interventions, are often immune suppressants. ³³

PDT mechanism of action

Photodynamic therapy has three principal efficacy‐determining parameters: first the optical energy either transmitted via a tissue surface, defined as radiant exposure (H) when quantified as energy traversing a unit area from one hemisphere only, or fluence (ϕ) for interstitial light delivery and inside light scattering tissues when quantified as traversing a unit area from all directions. Both are given in [J∙cm⁻²]. Second, the PS concentration ([PS_S0] [μM]), and third, the partial oxygen pressure (pO₂[%]) or ground state oxygen concentration ([³O₂]), whereby O₂ (μM) = 1.295 μM/mmHg × pO₂, where the constant 1.295 μM/mmHg is independent of temperature. ²⁸ , ³⁴ , ³⁵ , ³⁶ , ³⁷ In contrast to photothermal and irreversible electroporation, PDT requires an exogenous mediator, the PS, to absorb the light photon's quantum energy. The PS is administered below systemic toxicity; however, a minimum concentration is required to achieve toxic levels upon light irradiation. Upon light absorption, the PS transitions from its singlet ground state (S ₀) into an electronically excited singlet state (S ₁) with a nanosecond lifetime. While the PS can lose its excess energy through the emission of light (fluorescence) and production of heat (internal conversion) as most dyes, they are designed to undergo intersystem crossing to a longer life excited triplet state, T ₁, by spin conversion of the highest energy state electron. ³⁸ T ₁ can decay by phosphorescence to S ₀ or undergo charge or energy transfer reactions with other nearby molecules, called type I and type II photoreactions. In type I reaction, the PS transfers an electron to a reducing agent, generating free radicals that can react with molecular oxygen, generating oxygen radicals which result in oxidized biomolecules. In type II reaction, the PS exchanges spin and energy with ground state oxygen (³O₂), forming singlet oxygen (¹O₂). Concurrently, other ROS, including hydrogen peroxide (H₂O₂), superoxide ion (O ₂ ^−•), and hydroxyl radical (OH ^• ), are generated, see Figure 1. The lifetime of the reactive oxygen species depends on the present microenvironment and is mostly in the 10⁻⁹ to 10⁻⁶ second range in biological media such as blood plasma, ³⁹ resulting in limited diffusion distances (<1 μm). While H₂O₂ is relatively stable, allowing for greater diffusion distances, its breakdown product, O ₂ ^−•, is highly reactive, with half‐lives of 10⁻⁹. Other type I products, such as peroxides, also show high reactivity, with half‐lives of 0.1–50 ms, allowing for greater diffusion distances. Damage caused by these species occurs locally, with diffusion distances of about 0.001–0.1 mm. ⁴⁰ Once the concentration of ROS and, in particular, ¹O₂ reaches μM concentrations, growth arrest and cell death by necrotic or apoptotic pathways occur due to indiscriminate oxidation of lipids, proteins, and nucleic acids. ¹⁹ , ⁴¹ , ⁴² Although type I and type II reactions co‐occur and are responsible for cellular toxicity and PDT's therapeutic efficacy, type II reaction‐based PDT is dominant with most clinically used and investigational PSs, making them the primary cytotoxic agents responsible for the biological effects. At low oxygen concentrations, type I reactions may contribute more to PDT‐induced damages. ⁴⁰ Figure 1 compares the mechanism leading to cell death following RT and PDT.

FIGURE 1.

Comparison of cell death mechanisms in radiation therapy (RT) (A) and photodynamic therapy (PDT) (B). In RT (A), cell damage is induced directly through ionization effects on DNA and indirectly via reactive oxygen species (ROS) generated by water radiolysis. In PDT (B), the photosensitizer (PS) absorbs light and transitions via the singlet excited state to the lowest triplet state and can produce ROS through Type I and Type II mechanisms, which then result in lipid peroxidation and oxidation to other biomolecules, destroying their biological functionality.

PDT limitations

Photodynamic therapy faces several constraints, including challenges in treating disseminated metastases and reduced effectiveness in tumors encircled by necrotic tissue due to the essential role of oxygenation in the photodynamic process. Additionally, delivering precise light to reach deep‐seated tumors is a significant challenge. Superficial and well‐oxygenated tumors allow for a higher rate of reactive oxygen species (ROS) production. This results in more effective PDT treatment than deeply seated tumors, leading to variability in treatment response among patients. ⁴² Variability in PS uptake across different tissues can lead to inconsistent treatment outcomes, further compounded by the variation in PS half‐lives, which can prolong patient photosensitivity post‐treatment. Additionally, precise dosimetry is crucial to ensure effective treatment while minimizing damage to surrounding healthy tissue, which is often difficult to achieve due to the lack of standardized protocols. Patient‐specific factors, such as anatomical and physiological differences, further complicate treatment planning. Ensuring adequate and uniform light delivery, particularly for deep‐seated or complex tumors, presents another significant hurdle. Moreover, PDT is primarily effective for certain early‐stage or superficial cancers, limiting its broader applicability compared to other treatment modalities like chemotherapy and radiation therapy. The need for specialized equipment and expertise also limits the availability of PDT in many treatment centers. Despite its efficacy for certain cancers, these challenges constrain PDT's broader clinical use. ²⁶

PDT dose concepts

Any treatment plan must be based on a dose concept and the known tissue response as a function of a given dose. In PDT, concepts analogous or similar to ionizing radiation, including “absorbed dose,” “equivalent dose,” and “effective dose,” have been developed. The absorbed dose rate can be defined as the product of the fluence rate, ϕ, in [mW∙cm⁻²], the PS concentration [PS_S0] in [mM], and its molar absorption coefficient ε(λ) [L∙M⁻¹∙cm⁻¹], assuming ubiquitous availability of ground state oxygen. The absorbed dose is influenced by the PS's pharmacokinetics, including its localization at various scales (mesoscopic, microscopic, and subcellular) relative to critical structures, such as the mitochondria, and photon diffusion through tissue. The equivalent dose depends on the PS's ability to generate ¹O₂ and other ROS, as determined by their respective quantum yields and the microenvironment, particularly a minimal pO₂ to favor ROS generation. Similar to RT, the effective dose depends on the biology of the target tissues and a cell's ability to mitigate the cytotoxic ¹O₂. For instance, a study by Lilge et al. ⁴³ demonstrated that intracranial tumors could tolerate high cytotoxic ROS loads much better than the normal brain cortex, resembling the response observed in cases of ionizing radiation. While in RT, the equivalent and effective dose are averaged over a particular tissue or organ and are predominantly used in ionizing radiation protection, in PDT, they are applied spatially resolved and used for both exceeding the minimal dose to be delivered to the target tissue volume and limit the maximum permissible dose to the surrounding non‐target tissues and OAR.

To create PDT dose concepts comparable to the RT's “equivalent” and “effective” therapeutic dose concepts, it is essential to review the photophysical and photochemical processes driving PDT, which lead to the reduction and inactivation of biomolecules, denoted as “A.” The three efficacy‐determining parameters are also linked by photophysical (photobleaching and self‐shielding) and photochemical processes (photobleaching, oxygen consumption). These higher‐order dependencies can be expressed through the following equations.

$$\begin{array}{c}\frac{\mathrm{d}\left[{\mathrm{PS}}_{\mathrm{S}0}\right]}{\mathrm{d}t}=-\frac{\epsilon \left(\lambda \right)}{\mathit{h\nu }}\varphi \left[{\mathrm{PS}}_{\mathrm{S}0}\right]+\left(k_1+k_2\right)\left[{\mathrm{PS}}_{\mathrm{S}1}\right]\hfill \\ +k_5\left[{\mathrm{PS}}_{\mathrm{T}1}\right]+k_4\left[{\mathrm{PS}}_{\mathrm{T}1}\right]\left[{}^3{\mathrm{O}}_2\right]-k_8\left[{\mathrm{PS}}_{\mathrm{S}0}\right]\hfill \end{array}$$ (1)

$$\frac{\left[{\mathrm{PS}}_{\mathrm{S}1}\right]}{\mathrm{d}t}=\frac{\epsilon \left(\lambda \right)}{\mathit{h\nu }}\varphi \left[{\mathrm{PS}}_{\mathrm{S}0}\right]-k_2\left[{\mathrm{PS}}_{\mathrm{S}1}\right]-k_3\left[{\mathrm{PS}}_{\mathrm{S}1}\right]$$ (2)

$$\frac{\left[{\mathrm{PS}}_{\mathrm{T}1}\right]}{\mathrm{d}t}=k_3\left[{\mathrm{PS}}_{\mathrm{S}1}\right]-k_3\left[{\mathrm{PS}}_{\mathrm{T}1}\right]-k_7\left[{\mathrm{PS}}_{\mathrm{T}1}\right]-k_9\left[{\mathrm{PS}}_{\mathrm{T}1}\right]\left[A\right]$$ (3)

$$\begin{array}{c}\frac{\mathrm{d}\left[{}^3{\mathrm{O}}_2\right]}{\mathrm{d}t}=-S{k}_4\left[{\mathrm{PS}}_{\mathrm{T}1}\right]\left[{}^3{\mathrm{O}}_2\right]\hfill \\ +\left(k_5\right)\left[{}^1{\mathrm{O}}_2\right]+P\left(1-\frac{\left[{}^3{\mathrm{O}}_2\right]}{{\left[{}^3{\mathrm{O}}_2\right]}_0}\right)\hfill \end{array}$$ (4)

$$\begin{array}{c}\frac{\mathrm{d}\left[{}^1{\mathrm{O}}_2\right]}{\mathrm{d}t}=-S{k}_4\left[{\mathrm{PS}}_{\mathrm{T}1}\right]\left[{}^3{\mathrm{O}}_2\right]\hfill \\ -\left(k_5\right)\left[{}^1{\mathrm{O}}_2\right]-k_6\left[{}^1{\mathrm{O}}_2\right]\left[A\right]-k_4\left[{}^1{\mathrm{O}}_2\right]\left[{\mathrm{PS}}_{\mathrm{S}0}\right]\hfill \end{array}$$ (5)

$$\begin{array}{c}\frac{\mathrm{d}\left[A\right]}{\mathrm{d}t}=-k_9\left[A\right]\left[{}^1{\mathrm{O}}_2\right]-{\mathrm{k}}_6\left[A\right]\left[{\mathrm{PS}}_{\mathrm{T}1}\right]\hfill \end{array}$$ (6)

Whereby [PS_S0], [PS_S1], and [PS_T1] denote the concentration of the PS in the singlet ground, singlet excited, and triplet excited state, respectively. [A] represents the concentration of biological targets, which act as reducing agents that photochemically neutralize ¹O₂, limiting its biological functionality. ⁴⁴ Once the rate of damage exceeds the cell's ability to compensate through de novo synthesis, this leads to cell death. P refers to the oxygen perfusion rate of the tissue, and S refers to the probability of a [PS_T1] encountering a ³O₂. The rate constants are given by k ₁ = decay of the singlet excited state by internal conversion [s⁻¹], k ₂ = decay of the singlet excited state by fluorescence [s⁻¹], k ₃ = the intersystem crossing rate constant [s⁻¹], k ₄ = energy transfer rate to ³O₂ [M⁻¹∙s⁻¹], k ₅ = decay rate of the triplet state by phosphorescence [s⁻¹], k ₆ = reaction rate with reducing biomolecules A [M⁻¹∙s⁻¹], k ₇ = phosphorescence decay rate for ¹O₂ [M⁻¹∙s⁻¹], k ₈ = photobleaching rate of PS [M⁻¹∙s⁻¹], k ₉ = energy or charge transfer to biomolecules A from [PS_T1] [M⁻¹∙s⁻¹] for Type I and Type II reactions, [A] represents sinks for the cytotoxic ¹O₂; ϕ is the fluence rate [W∙cm⁻²], h = 6626 10⁻³⁴ J∙s is the Planck's constant and ν = c ₀ ∙λ ⁻¹ , the excitation wavelength, and c ₀ = 299,792,458 m∙s⁻¹, the speed of light. Figure 2. visually describes Protoporphyrin IX's energy and spin transitions, corresponding to Equations (1), (2), (3), (4), (5), (6) above.

FIGURE 2.

Shows the energy and spin transitions for Protoporphyrin IX, according to Equations (1), (2), (3), (4), (5), (6) above. The fluorescence and singlet oxygen quantum yields are for MeOH as a solvent, according to. ⁴⁷

In contrast to the somewhat stable states of PS_S0, ³O₂ and A, the PS_S1, PS_T1 and ¹O₂ are short‐lived entities due to rapid intersystem crossing, phosphorescence, and fluorescence decay events. Hence, they quickly reach their equilibrium concentrations, as demonstrated in studies by Finlay et al., making them suitable for monitoring the rate of cytotoxic moiety generation. Values for k ₁ to k ₈ have been determined for certain PSs, including Photofrin, as indicated by Zhu et al. and references. ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ Other research groups have determined these rate constants, as recently summarized by Kim et al. Table 5 and citations within. ⁴⁵ , ⁴⁶

Unlike the time‐invariant dose‐determining parameters for IMRT and IGRT, Equations (1), (2), (3), (4), (5), (6) demonstrate that the PDT dose is time‐dependent. In particular, variables like k ₈, P, and ϕ, through changes in the effective attenuation coefficient, defined by ${{\mathrm{\mu }}_{\mathrm{eff}}}^{-1}={\left(\sqrt{3{\mathrm{\mu }}_{\mathit{a}}\left({\mathrm{\mu }}_{\mathit{a}}+{{\mathrm{\mu }}_{\mathit{s}}}^{\prime }\right)}\right)}^{-1}$ can result in a highly temporally dynamic $\frac{\mathrm{d}\left[{}^1{\mathrm{O}}_2\right]}{\mathrm{dt}}$. This necessitates continuous monitoring of the spatial–temporal distribution of the PDT reaction educts PS_S0, ³O₂, ϕ, and A or one of the other transient products. These educts can still vary significantly between patients and their tissues in the PTV, given the local variation in blood perfusion, adipose tissue, and resulting microenvironment. Hence, some form of online monitoring of these educts or the PDT resulting products during photon delivery will be required.

Spatial and temporal limitations in critical reaction educts or efficacy‐determining parameters will saturate the attainable [¹O₂] for Type II PDT. In particular, the photochemical reaction can be stalled for PS with high k ₇ or in tissues with low P. If the [PS_S0] in normal tissue surrounding the CTV is initially low or decreases rapidly due to a high k ₈; a high fluence can be applied to the PTV without compromising the therapeutic selectivity, as the PS is bleached prior to exceeding a cell's protein and lipid replacement capacity.

To clarify the dose concepts, Table 1 summarizes the definitions and concepts of administered dose (exposure), absorbed dose, equivalent dose, and effective dose in both RT and PDT.

TABLE 1.

Comparison of dose concepts in RT and PDT.

	RT	PDT
Administered dose (exposure)	Coulomb kg−1	Photosensitizer (mg kg−1) × light energy (J) or (J cm−2)
Absorbed dose	Energy absorbed by tissue D (J kg−1) or (Gy)	Photons absorbed by photosensitizer D PDT = [PS] × ε × (J × hν λ−1) (hv cm−3)
Equivalent dose	HT = D × W R (J kg−1) or (Sv)	[ROS] = DPDT × ϕ Δ (μM)
Effective dose	HTT = HT × W T (J kg−1) or (Sv)	DPDT > T a (hv cm−3)	[ROS] > [ROS] threshold dose (μM)

^a The term T refers to the photon‐based threshold dose for PDT, which is the number of photons absorbed by the photosensitizer per volume tissue to achieve cytotoxicity, whereas [ROS] threshold dose more accurately referred the minimum reactive oxygen species concentration required to induce cytotoxicity.

PDT dosimetry

The definitions of PDT doses are primarily based on the fluence rate or radiant exposure, H [J∙cm⁻²]. While this may seem straightforward, it represents the most reductionist approach to defining the delivered dose. This approach requires assumptions related to the variability of the PS pharmacokinetics and tissue optical properties between patients. As illustrated in Figure 3 for superficial PDT, the PDT dose follows surface proximal a 1D gradient, given that the radiance exposure, topically applied photosensitizer, and oxygen diffusion via the surface can all be described by 1D diffusion processes. Hence, using large‐area sources negates the need for detailed treatment planning. In PDT for skin malignancies, the PS is administered topically to the tumor surface, reducing variation in PS uptake, and the dermis' optical properties are well documented. ⁴⁸ Irradiances avoiding thermal tissue damage are often employed, assuming homogeneous PS distribution across the target depth, resulting in good treatment outcomes for standardized delivered PDT doses. ⁴⁹

FIGURE 3.

Illustration of the fluence or photon density gradients for (A) surface illumination, causing a predominantly 1D PDT dose gradient, and (B) for interstitial applications for a 2D PDT dose gradient.

Similar considerations apply to bladder PDT given the ease of PS instillation, high selectivity provided by an intact urothelium excluding the PS from the normal bladder wall, optical access via endoscopy, and limited thickness of tumors (<2 mm) simultaneously. The generation of a steep PDT dose gradient achieved by installation and short wavelength PS excitation limits or prevents morbidity within the muscle layer, a higher risk in patients who have undergone extensive prior treatments. ⁴¹ , ⁵⁰ , ⁵¹

Dosimetry for esophageal tissue is complicated due to esophageal folds and the need to keep the light sources in the esophagus' center during movement caused by peristalsis and breathing, potentially resulting in uneven light distribution. Using elastic catheter balloons improves these conditions and the homogeneity of photon dose distribution. However, the PDT efficacy might decrease due to reduced blood flow to the esophageal lining and diminished oxygen availability to the target, caused by the pressure exerted by the balloon onto the mucosa and submucosa and preventing oxygen access via the lumen surface. This requires reduced irradiances and, thus, longer illumination times to allow oxygenation during the treatment. ⁵¹ , ⁵²

These PDT dosimetry definitions are based on the concept of delivered doses and rely on empirical observations to predict therapeutic responses. This PDT dose concept is also evident in the critical fluence model by Jankun et al., ⁵³ where the minimum dose to the target must be determined for each clinical PS dose and the target tissue. This involves assuming the validity of population averages for PS pharmacokinetics, resulting in an average [PS_S0], well‐perfused tissues for a minimum pO₂ and equal dose responses for the target and surrounding normal organs.

Nevertheless, this delivered dose concept has proven successful in surface treatment, like the skin or the bladder wall, when the target depth is shallow compared to the inverse of the effective attenuation coefficient μ_eff, [mm⁻¹]. In superficial PDT indications, such as the skin or bladder, the delivered dose input parameters include the administered PS, either systemically in units of [mg∙kg⁻¹] or topically based on surface area [g∙m⁻²]. The delivered optical energy density is measured in [J] for cavities (eg, the bladder) or [J∙cm⁻²] for surfaces (eg, the skin).

For other indications, empirical treatment plans have also been developed. Betrouni et al. ⁵⁴ work focused on the effects of light in vascular WST11 (Tookad) mediated PDT based on data from phase I and phase II clinical trials. The model establishes a relationship between the tissue volume illuminated by the light diffusers and the MRI‐determined necrosis volumes. The resulting empirical model does not consider the tissue optical properties and is valid only for WST11 [PS] 4 mg∙kg⁻¹ and H 200 J∙cm⁻¹. ⁵⁴

Next to the empirically determined source placement density and interstitial indications, monitoring PDT treatment delivery requires an online monitoring concept for at least one parameter listed in Equations (1), (2), (3), (4), (5), (6). This ensures the correct absorbed dose is delivered to achieve the desired clinical outcome, generally necrosis, within the framework of a PDT dose concept.

Wilson et al. ³⁴ proposed multiple dose concepts based on spatially resolved measurable events categorized into explicit and implicit factors. Explicit factors include light fluence rate at the necrosis boundary, PS concentration, and tissue oxygenation. Implicit factors involve directly monitoring and quantifying the photobiological response, including singlet oxygen, the PS's excited triplet, or singlet excited state.

Dosimetry, based on the quantification of photobiological effects, requires these effects to fully develop, which takes hours or days post‐PDT illumination onset. Hence, it is presently unsuitable for PDT progress monitoring. Future development of imaging technologies or procedures may allow for rapid quantification of modified or expressed biomarkers correlated with tissue outcomes, as in the case of immune‐activated biomarkers, including NF‐kappa B. ⁵⁵ For treatment monitoring, techniques such as alterations in blood oxygen saturation (StO ₂) ⁵⁶ and tissue changes detectable by ultrasound or in photoacoustic imaging, as reviewed by Hester et al., ⁵⁷ can be considered. Other tissue response surrogates discussed by Pogue et al., ⁵⁸ including MRI, PET, CT, OCT, and US, detectable vascular volume and flow changes if they contribute significantly to the tissue response. Volume‐averaged measurements using Electrical Impedance Spectroscopy can also predict PDT outcomes. ⁵⁹ , ⁶⁰

Biomarkers for tissue response evolving post‐PDT cannot be employed for PDT treatment planning or monitoring of the PDT dose rate.

Monitoring [¹O₂] or [PS_T1], quantities directly or closely related to the biological outcome provide the most direct prediction of treatment outcomes with the fewest remaining assumptions.

Monitoring of [¹O₂] is readily achieved by detecting its near‐infrared luminescence emission at 1270 nm. As it is the dominant cytotoxic moiety, predicting cell death only requires knowledge of its subcellular ¹O₂ location and the target cells' responsivity. While spectroscopic monitoring of [PS_T1], is equally possible, the local pO₂ must be sufficiently high to become a direct PDT dose metric. For superficial PDT applications, implementations have been demonstrated for phosphorescence‐based [¹O₂] monitoring, sometimes referred to as singlet oxygen luminescence dosimetry (SOLD) ⁶¹ and for transient [PS_T1] absorption spectroscopy. ⁶² While the technical implementations are complex and generally under‐sample the PTV, particularly for interstitial applications, progress in monitoring these short‐lived transients was shown by Takakura et al. ⁶³ Transient spectroscopy of these two quantities and [PS_S1] mirrors the “effective dose” concept from ionizing radiation therapy.

For interstitial and cavity‐targeted PDT, implicit dosimetry concepts are more suitable to monitor treatment progress, given the lower technical implementation barriers; however, a signal loss does not allow identification of the cause as only one parameter is monitored. In the case of no detectable 1270 nm emission for ¹O₂ monitoring, the lack of excitation photons, PS, or ³O₂ can be the root cause, but the physician can take no corrective action.

Implicit dosimetry based on fluorescence monitoring of [PS_S1] and the known photobleaching rate k ₈ or direct dosimetry that quantifies [¹O₂] resembles the ionizing radiation's “effective dose” concept more closely. The latter quantifies the cytotoxic moiety, while the former is a surrogate for [¹O₂], as ¹O₂ denatures the PS by photobleaching.

Quantifying the rate and magnitude of PS photobleaching as a PDT response biomarker remains subject to the measurement of [PS_S1], which is typically achieved through fluorescence quantification at specific emission wavelength or by full spectroscopic approaches, including hyperspectral imaging. ⁶⁴ , ⁶⁵ , ⁶⁶

Valentine et al. ⁶⁷ employed Monte Carlo simulations to clinically investigate PpIX fluorescence monitoring during PDT in non‐melanoma skin cancer. Primarily aimed at enhancing the sensitivity and specificity of fluorescence‐based prediction of the fluorescence origin and the PDT generation of singlet oxygen. The results also suggested that higher irradiance (ϕ), causing the loss of surface PpIX fluorescence, can be advantageous for reaching deeper tumors.

It has been shown that single wavelength fluorometric assessments have limitations in quantifying PS concentrations and changes thereof during PDT. ⁶⁸ Hence, this seemingly inexpensive approach can benefit from further technical developments.

Similar to PS photobleaching monitoring, singlet oxygen monitoring ⁶⁹ aims to quantify the temporal cytotoxic dose and includes implicit PDT dose metrics. Generation and consumption of ¹O₂ is given by.

$$\frac{\mathrm{d}\left[{}^1{\mathrm{O}}_2\right]}{\mathrm{d}t}=S_{∆}{k}_4\left[{}^3{\mathrm{O}}_2\right]\left[{\mathrm{PS}}_{\mathrm{T}1}\right]-k_5\left[{}^1{\mathrm{O}}_2\right]-k_6\left[A\right]\left[{}^1{\mathrm{O}}_2\right]$$ (7)

Utilizing k ₇, the ¹O₂ phosphorescence decay rate (as described in Equation 5), which is in a constant relationship with k ₆, the rate of lipid and protein oxidation and denaturation, D _PDT can be written as:

$$D_{\mathrm{PDT}}={\int }_0^T\varphi \left(t\right)\frac{ϵ\left(\lambda \right){\left[{\mathrm{PS}}_{\mathrm{S}0}\right]}_{t=0}}{\mathit{h\nu }}\times \frac{1}{k_7}\left(1-{\mathrm{e}}^{-k_8\varphi \left(t\right)t}\right)\mathrm{d}t$$ (8)

This simplified intrinsic approach assumes that the concentration of biological targets [A] is constant over the treatment duration. This is reasonable given the infinite concentration of potential targets for all practical purposes. ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶

Quantifying ¹ O ₂ is technologically difficult, given its short lifetime, ~100 ns, related to k ₆ [s⁻¹]. Conversely, the quantification of ³ O ₂ is well established. Γ(t) in Equation 9 denotes the time‐dependent consumption of ³ O ₂ by normal metabolism, Γ _meth, and PDT Γ _PDT (t), and equally depends on ϕ(t), [PS_S0], and ε(λ), modulated by the probabilities to consume oxygen through the processes listed in Equation 7 results in the oxygen consumption therapy.

$$\begin{array}{c}\Gamma \left(t\right)={\Gamma }_{\mathrm{PDT}}\left(t\right)+{\Gamma }_{\text{meth}}\hfill \\ =S_{∆}{\varphi }_t\left(t\right)\epsilon \left[{\mathrm{PS}}_{\mathrm{S}0}\right]\left(\frac{k_6\left[A\right]}{k_5+k_6\left[A\right]}\right)\left(\frac{k_4\left[{}^3{\mathrm{O}}_2\right]}{k_4\left[{}^3{\mathrm{O}}_2\right]+k_9}\right)\hfill \end{array}$$ (9)

where k ₉ represents the non‐chemical decay of [PS_T1] in the oxygen depletion zone around the capillaries ⁷⁵ demonstrated experimentally for spheroids, ⁶⁹ , ⁷⁵ , ⁷⁶ , ⁷⁷ and numerical solutions were presented for a microscopic and macroscopic scale. ⁷⁸ , ⁷⁹ , ⁸⁰

Online monitoring of the effective PDT dose, based on intrinsic parameters [PS_S1], [PS_T1], and [¹O₂] enables the determination of the PDT dose rate. However, the root cause of local excessive or insufficient PDT dose rates remains unspecified, making corrective actions ambiguous.

Foster et al. developed a model to assess oxygen consumption during Photofrin‐mediated PDT to predict therapeutic outcomes, highlighting the importance of oxygen consumption. The [PS], 4–10 μg∙ml⁻¹, and ϕ were measured volumetrically on excised tumors to establish an average optical density (OD) of 0.4 cm⁻¹. To account for potential experimental outcomes, PDT dose calculations were performed using irradiances in the 50–200 mW∙cm⁻² range. ⁷⁶

All the above approaches employed monitoring one or two dose metrics with an applicable PDT dose model given an empirically determined fixed source placement derived from population‐averaged optical properties of the target tissues.

Gregor et al. ⁸¹ have recently proposed a framework to overcome the rigid interstitial source spacing to account for the number of cylindrical light diffusers and their insertion pattern, showing that the inter‐diffuser distance can be increased monotonically with the number of diffusers.

Explicit dosimetry, based on the product of ϕ(r) and [PS_S1], mirrors the “absorbed dose” concept. However, the correlation to an effective dose must be determined for each PS, the excitation wavelength, and the target tissues.

The underlying assumption for explicit dosimetry is spatial and temporal invariant photochemical and photobiological rate constants k ₁ to k ₈, including the triplet and singlet oxygen quantum yields, denoted as ϕ _T and ϕ _Δ, given by ${\varphi }_T=\frac{k_3}{k_1+k_2+k_3}$ and ${\varphi }_{\mathrm{\Delta }}=\frac{k_4}{k_1+k_2+k_5+k_{9.}}$ respectively. That is to say, the quantum yields are independent of the biological microenvironment.

The local PDT dose, D _PDT (r), for explicit dosimetry, can be defined as. ⁷³

$$D_{\mathrm{PDT}}\left(r\right)={\int }_0^T\frac{\varphi \left(t\right)}{\left[A\right]\mathit{h\nu }}\epsilon \left(\lambda \right){\varphi }_{\Delta }\left[{\mathrm{PS}}_{\mathrm{S}0}\right]\left(t\right)\mathrm{d}t$$ (10)

Hence, acquiring information about ϕ(t,r) and [PS_S0] (t,r) throughout the PTV is feasible with spatial resolution better than ½ μ_eff. The ubiquitous availability of oxygen and constant [A] are assumed. For instance, this approach is employed in the photodynamic threshold model developed by Farrell et al., ⁷³ which states that necrosis occurs once D_PDT exceeds the critical tissue‐dependent value. This zero‐order dosimetry approach, which relies on the administered doses, the PS's singlet oxygen quantum yield (ϕ _Δ), and a minimum [³O₂] calculated by multiplying the pO₂ with the oxygen solubility in tissue at 1.295 [μM∙mHg⁻¹], along with empirically determined CTV tissue necrosis extent can work well, ⁴⁵ , ⁸² given adequate selectivity of the PS towards the tumor.

An advantage of PDT monitoring by explicit quantities is that it allows for real‐time dose adjustment, as at least two efficacy‐determining parameters are monitored. Particularly for the risk of undertreatment, the fluence H(r) or fluence rate ϕ(r) can be adjusted to overcome limitations in either [³O₂] or [PS_S0]. ⁸³

Online monitoring of the absorbed PDT dose by integrating multiple efficacy‐determining parameters, ϕ(t), [PS_S0] and [³O₂], allows for correcting factors such as photobleaching, oxygen reperfusion, or excessive PS concentration. This approach accounts for phenomena like self‐shielding of the fluence rate ³⁴ , ⁸⁴ as the PDT reaction educts are quantified.

Quantification of [PS_S1] is typically performed by fluorescence via an established k ₆ for the PS in question, as determined by various groups in preclinical and clinical studies. ⁴⁴ , ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ⁹⁰ , ⁹¹ , ⁹² , ⁹³ For surface proximal targets, imaging of [PS_S1] by spatial frequency domain imaging (SFDI) for surface and diffuse fluorescence tomography (DFT) by Dan et al. ⁹⁴ is feasible, but knowledge of the tissue's optical properties is required.

Explicit monitoring of three PDT efficacy‐determining parameters was demonstrated by Sun et al. ⁹⁵ in a mouse model based on Equation 11:

$$D_{\mathrm{PDT}}={\int }_0^T\xi \frac{\left[{}^3{\mathrm{O}}_2\right]\left(t\right)}{\left[{}^3{\mathrm{O}}_2\right]\left(t\right)+\beta }\frac{\varphi \left(t\right)}{\mathit{h\nu }}\epsilon \left(\lambda \right){\varphi }_{\Delta }\left[{\mathrm{PS}}_{\mathrm{S}0}\right]\left(t\right)\mathrm{d}t$$ (11)

Whereby D_PDT is given by the temporal integral of the ¹O₂ generation. The photophysical parameter 𝜉 = 3.7 × 10⁻³ cm²∙s⁻¹∙mW⁻¹ represents the specific oxygen consumption rate, and β = 11.9 μM is the oxygen quenching threshold concentration for Photofrin‐mediated PDT using 630 nm excitation light. For the PDT dose to become independent of the treatment wavelength, ϕ(r) is converted into a photon density [hν∙cm⁻²] and [PS_S1] represented as molar concentrations. In this form, an “effective dose” can be determined for tissue and PS combinations when the PSs show comparable subcellular accumulation patterns and ROS quantum yields. The latter allows the determination of differential W _T values for different tissues, similar to IR therapy.

Sun et al. ⁹⁵ measured the fluence rate by a single isotropic fiber optic detector inserted 3 mm deep into a 4–5 mm diameter tumor. This measurement was augmented with a phosphorescence‐based pO₂ probe (OxyLite). Zhu et al. ⁹⁶ demonstrated the use of multiple isotropic detectors for the prostate and the thorax. ⁹⁷ , ⁹⁸ Additionally, Lilge et al. demonstrated using irradiance sensors for the bladder. ⁴¹ , ⁹⁹

The explicit PDT dose concept based on monitoring the PS and oxygen concentrations was expanded by Zhu et al. ⁴⁶ and Simphotek Inc. with their Dosie™ simulation device.

$$D_{\mathrm{PDT}}={\int }_0^T\xi \frac{\left[{}^3{\mathrm{O}}_2\right]\left(t\right)}{\left[{}^3{\mathrm{O}}_2\right]\left(t\right)+\beta }\frac{\varphi \left(t\right)}{\mathit{h\nu }}\epsilon \left(\lambda \right)\left[{\mathrm{PS}}_{\mathrm{S}0}\right]\left(t\right)\mathrm{d}t$$ (12)

Beeson et al. ¹⁰⁰ have determined 𝜉 = 7.15 × 10⁻² cm²∙s⁻¹∙mW⁻¹ for HPPH‐mediated PDT, and Kim et al. ⁴⁶ determined 𝜉 = 55 × 10⁻³ cm²∙s⁻¹∙mW⁻¹ for BPD‐MA‐mediated PDT. The oxygen quenching threshold concentration appears to be PS‐independent—however, the study by Kim et al. ⁴⁶ highlighted that while there were consistent temporal changes in [³O₂] during PDT, measurements of molecular oxygen [³O₂] may not accurately predict [¹O₂].

To predict the cumulative ¹O₂‐mediated vascular response, Lui et al. ¹⁰¹ developed a one‐dimensional dose model for ALA‐induced PpIX‐mediated PDT in normal human skin incorporating reaction kinetics, with ϕ and [PS_S0], [³O₂]. Using input parameters from literature, the predicted responses align well with clinical data and provide insights into the distribution of PS, ³O₂, and ¹O₂ as a function of depth from the surface, accounting for oxygen diffusion from the skin surface.

Ong et al. ⁸⁶ developed a four‐channel PDT dose dosimetry system that simultaneously monitors ϕ(t) and [PS_S1] fluorescence at four distinct locations during Photofrin‐mediated PDT within the pleural cavity. The influence of tissue optical properties was mitigated based on empirical correction factors established through Monte Carlo simulations of the fluorescence over the physiologically relevant tissue optical property ranges. The study of 22 sites within the pleural cavity of eight patients exhibited a substantial intra‐ and inter‐patient difference in the Photofrin concentrations at 2.9 and 8.3 times, respectively. This technique was further expanded by Beeson et al., ¹⁰² albeit assuming homogeneous PS distributed within the tumor and ubiquitous pO₂ availability.

Vaupel et al. ¹⁰³ demonstrated that tumors have extensive hypoxic regions, given that the rate of tumor volume expansion outpaced the development of a supporting vasculature. A lack of supporting vasculature was shown to limit not only the penetration of PS ¹⁰⁴ but also of other chemotherapeutic drugs ¹⁰⁵ and oxygen. ¹⁰⁶ The combination of hyperbaric oxygen therapy did lead to improved PDT efficacy ¹⁰⁷ , ¹⁰⁸ ; however, its technical complexity and the requirement for post‐therapy decompression hindered its translation into clinical practice. While the local pO₂ is critical for the therapeutic efficacy (see Equation 4), the rate of photochemical consumption and, consequently, the generation rate of ¹O₂ can be modulated via the fluence rate, ϕ(r). ¹⁰⁹ Given the nature of rapid changes in the local pO₂, it is essential to map and monitor this parameter throughout the CTV.

PDT treatment planning

Photodynamic therapy treatment planning should optimize physical parameters under the direct control of the clinician. Due to the inability to control the PS's spatial accumulation, retention, and vascular perfusion of the CTV, treatment planning needs to focus on the number, density, and emission power density of implanted diffusing tip fibers and the emitted photon density. This approach aims to achieve 100% of the minimum dose throughout the CTV while minimizing or adverting high PDT doses to the adjacent tissues and OAR within the PTV.

In a recent study by Gregor et al., ⁸¹ Monte Carlo simulations were utilized to optimize the distances between light sources in PDT using a tool called “TracePro”, developed by Lambda Research Corporation, to simulate light propagation in tissue. ¹¹⁰ The research aimed to determine the optimal positioning of cylindrical light diffusers (CLDs) for treating head and neck squamous cell carcinoma (HNSCC) with PDT and photoimmunotherapy. This study conducted Monte Carlo simulations to model light propagation around CLDs inserted perpendicularly into a semi‐infinite tumor, aiming to determine the volume receiving a fluence above the therapeutic threshold. An optimization algorithm was then used to calculate and maximize the volumes of necrosed tumors.

Davidson et al. ¹¹¹ developed treatment‐planning software for a Phase II clinical trial of Tookad™‐mediated PDT for persistent prostate carcinoma following radiation therapy. This software allowed personalized treatment planning based on the predicted light distribution in the prostate and nearby tissues based on MRI imaging. The photon dose was calculated using a finite element‐based diffusion equation solution. Using online monitoring, the optical properties were also measured in vivo. The plan and execution were verified only by comparing the predicted light distribution with physical measurements during treatment.

Swartling et al. ¹¹² expanded the concept for Temoporfin‐mediated PDT for early‐stage (T1c) primary prostate cancer using Simphotek, Inc.'s commercial treatment planning and monitoring software called “Interactive Dosimetry by Sequential Evaluation” (iDOSE®) paired with online dosimetry software and multi‐parameter monitoring. ⁴⁶ The 3‐D pretreatment plan was based on ultrasound images of all optical fiber positions and their emission power, whereas the tissue optical properties between fiber pairs were monitored online. This monitoring allowed adjustments to the optical powers to target the entire prostate while minimizing exposure to nearby organs. While the online dosimetry setup monitored the tissue optical properties, PS fluorescence, and NIR spectroscopy, the PS fluorescence and pO₂ data were not factored into the dose calculations, citing the lack of suitable dosimetry models.

Currently, the iDOSE hardware and software provide light dose plans based on the generated ¹O₂ concentration to direct optical fiber positions and the laser powers at each position in the CTV. ¹¹³ , ¹¹⁴ , ¹¹⁵ Thus, an approach that overcomes the predominantly used empirical photon source density in the planning process is presented. This is also now employed for different indications, such as the pleural cavity.

Simphotek developed “INTELLI,” another innovative clinical and operating room system designed to enhance iPDT. This system combines light‐based medical device hardware with proprietary software to provide a comprehensive solution for dosimetry, delivery, feedback, and treatment planning, combining concepts of delivered and absorbed PDT dose metrics. INTELLI will utilize laser‐coupled fibers to target and treat deep‐seated tumors, adjusting the light dose based on real‐time simulations and monitoring PS concentration and distribution. Initial prototypes have demonstrated promising results in early clinical trials, with further trials planned to validate the system's real‐time PDT dose monitoring and adjustment capabilities. ⁴⁶

In a study by Altschuler et al., ¹¹⁶ a systematic search procedure using the Cimmino feasibility algorithm is described and evaluated to optimize the location, length, and strength of light sources for photodynamic therapy. The Cimmino feasibility algorithm, an iterative linear method, was initially applied to radiotherapy inverse problems by Censor et al. ¹¹⁷ This algorithm is deemed safer than other optimization techniques because it guarantees convergence. It defaults to the least‐squares solution if all prescribed dose constraints are unmet.

Yassine et al. ¹¹⁹ developed a tool for PDT pretreatment planning called PDT‐SPACE, an open‐source software for automatic iPDT planning. PDT‐SPACE employs advanced computer‐aided design (CAD) algorithms to allocate power to light diffusers, determining their optimal location, length, and emission power to minimize damage to surrounding OAR while maintaining the delivered PDT dose throughout the tumor to achieve necrosis. The software includes a power allocation algorithm that creates customized power profiles for cylindrical diffusers, ensuring the light distribution conforms precisely to the tumor shape, further reducing damage to healthy tissue. ¹¹⁸ This process begins with an initial heuristic source placement, then calculating the resulting photon distribution and PDT dose throughout the PTV. Optimization requires random or guided perturbations of the source position, length, and power to achieve the optimal source distribution and position. ¹¹⁹ , ¹²⁰ , ¹²¹ , ¹²² , ¹²³ , ¹²⁴ Wang et al. have extended PDT‐SPACE with two significant enhancements. The first improvement allows for the specification of clinical access constraints for light source insertion to avoid penetration of critical structures and minimize surgical complexity. The second enhancement generates an initial placement of light sources as a starting point for refinement, eliminating the need for clinicians to provide a starting solution. ¹²⁵ Figure 4 shows the evolution of PDT treatment planning, from early empirical models and basic dosimetry to advanced computational approaches, patient‐specific modeling, and recent AI‐driven and multimodal therapy integration.

FIGURE 4.

Evolution of PDT treatment planning from empirical dosimetry models in the 1970s to advanced, AI‐driven, and multimodal therapy integration in the 2020s. ¹²⁶ , ¹²⁷ , ¹²⁸ , ¹²⁹ , ¹³⁰

Current limitations in PDT treatment planning

Photodynamic therapy planning is challenging since predicting treatment outcomes depends on knowing the actual tissue optical properties, PS, and ³O₂ availability throughout the PTV, preferably spatially resolved. ¹³¹ , ¹³²

Unlike brachytherapy, where cytotoxic dose delivery is time‐invariant, PDT dosimetry will always require some online monitoring, as described above. It has been shown that optimizing the treatment fiber position can result in complete tumor coverage while minimizing morbidity. However, using catheters or needles to implant the photon source deeply often leads to displacement from the intended location, resulting in a position uncertainty of 2–3 mm. ¹³³ , ¹³⁴ This also necessitates real‐time tracking and updating of their position using X‐ray or ultrasound scanners. However, this lengthens the surgical procedure in the prostate by approximately 30 mins. ¹¹¹

The Achilles heels of PDT treatment planning are the wide variety in the photon transport coefficients, spatial changes in the vascular perfusion throughout the PTV and local PS accumulation, and, unlike ionizing radiation, a multitude of cell damage and tissue response pathways.

The photosensitizers accumulation variations intra‐ and inter‐patient have been shown by various groups. In a preclinical study, Penjweini et al. found that Photofrin‐mediated PDT could not produce significant complete responses or long‐term tumor control in all mice with radiation‐induced fibrosarcoma tumors. This was evidenced by tumor regrowth and size increase in some instances. The variability was considered to be due to spatial variations in interstitial Photofrin concentration, oxygen levels [³O₂](r), and photochemical oxygen consumption, resulting in an insufficient amount of reactive singlet oxygen [¹O₂] _rx (r) for effective tumor destruction. ¹³⁵ Johansson et al. ¹³⁶ observed that higher PPIX fluorescence in brain tumor rims strongly correlated with long‐term survival, with three out of five patients surviving over 3 years, while those with lower or variable levels survived less than a year. In another study, Zhou et al. ⁸⁵ suggested that measuring [PS] within the target volume and adjusting light doses accordingly could minimize subject variation and enhance the consistency of treatment outcomes.

Current online monitoring tools for [PS] measurements in PDT face several limitations that impact their effectiveness in clinical settings. Zhou et al. ⁸⁵ demonstrated that microprobes could detect PS variation within a tumor, but their application is typically limited to superficial regions of the CTV, which may not accurately represent PS heterogeneity in larger tumors with complex tissue structures. Additionally, SpectraCure's high‐resolution approach using cut‐end optical fibers in the prostate, as reported by Elliott et al., requires a large number of catheters to be inserted, which can be impractical for organs with heterogeneous tissues and varying vascular structures, such as the brain. ¹³⁷ , ¹³⁸

Predicting an organ's tissue optical properties prior to inserting optical fibers for therapy remains very difficult or impossible unless new technologies are developed. MRI imaging sequences allow us to differentiate between water, fat, and vasculature in combination with their known absorption spectra, as proposed by Jacques. ⁴⁸ Aumiller et al. ¹³⁹ demonstrated that, for porcine brain tissue, the MRI relaxation increased with increasing water concentration in the tissue, which correlated with an μ_a decreased and μ_s' increased for 635 nm radiation. However, these correlations are currently too weak to quantify a tissue's optical properties accurately. Hence, demonstrating a treatment plan's robustness within the range of anticipated tissue optical properties requires the perturbation of the optical input parameters for a given plan and demonstration of the plan's performance recovery while only permitting power and total energy delivery to be adjusted.

These limitations underscore the need for more advanced monitoring tools to provide detailed, spatially resolved PS measurements to enhance PDT treatment planning and efficacy.

The vast majority of current treatment planning platforms prioritize maximizing immediate tissue destruction within the target volume, typically the malignancy outlined by the oncologist, using tissue necrosis as the biological endpoint. However, it remains unclear whether simultaneous planning for other cell death mechanisms, including apoptosis, autophagy, or ferroptosis, should be considered to maximize the body's immune response against the aberrant tumor cells. ¹⁴⁰ , ¹⁴¹ , ¹⁴² , ¹⁴³ While various strategies to improve the PDT‐induced immune response have been published, ¹⁴⁴ , ¹⁴⁵ their inclusion in PDT treatment planning has not been reported.

Future directions

Photodynamic therapy's main limitation lies in treating deep‐seated tumors effectively due to light delivery and dosimetry challenges. A novel radiation‐induced photodynamic therapy (radioPDT) shows a promising solution using X‐ray energy to activate a novel nanoparticle containing a PS for a PDT effect. This mechanism enables deeper tissue penetration than traditional PDT and uses the current radiation treatment planning methods. By integrating the mechanisms of DNA damage from radiotherapy with PDT‐induced cell and organelle membrane damage, vascular damage, and immune‐priming effects, radioPDT could significantly enhance the efficacy of the treatment while minimizing toxicities. Currently, many different types of nanoparticles are being engineered and studied for the use of this treatment ¹⁴⁶ , ¹⁴⁷ , ¹⁴⁸ , ¹⁴⁹ ; however, this therapy is still in the early stages, and additional preclinical studies are necessary to gain a deeper understanding of the therapeutic mechanism and limitations, including the minimum drug dose required for effective radioPDT and the minimum X‐ray absorption needed for activation.

SUMMARY

This review explores how PDT treatment planning can learn from the established methodologies of RT treatment planning. While both therapies share common optimizing therapeutic goals towards the malignancies while minimizing damage to surrounding healthy tissues, they face unique challenges due to their underlying mechanisms and dosimetry requirements. Towards these goals, radiation therapy treatment planning has evolved significantly, encompassing personalized planning, whereas iPDT treatment planning is still in its developmental phase.

Building on the similarities in delivered, absorbed, and effective doses between RT and PDT treatments, future developments in PDT treatment planning can benefit from the well‐standardized methodologies established in RT.

For RT, minimizing damage to healthy tissues while maximizing tumor control remains a priority, necessitating ongoing advancements in imaging, dose calculation, and delivery techniques. In particular, damages in normal tissue can sometimes manifest only weeks or months later, as in oral mucositis ¹⁵⁰ or dermatitis. ¹⁵¹ While pretreatment planning is now often combined with monitoring of the primary radiation, for example, via the Cherenkov effect in real‐time, to date, they are not predictive of the observed morbidities. ¹⁵²

Current PDT practices rely heavily on population average data related to PS pharmacokinetics, vascular perfusion, and tissue optical properties, which does not account for individual variations in treatment efficacy parameters. However, there is progress: online monitoring of PDT efficacy parameters and the absorbed dose has become feasible, as seen in work by SpectraCure and Simphotek. ⁴⁶ Extension to direct quantification of the [PS] and light fluence, indirect measurements based on Blood flow (BF) and blood volume (BV) images for PS and oxygen biodistribution prediction in tissues are ongoing, as shown in a rabbit pancreas tumor model by Elliott et al. ¹³⁷ showing a linear correlation between PS fluorescence, BF, BV, and the permeability surface‐area product. Therefore, innovative surrogate techniques to predict [PS] and oxygen levels, as proposed by Østergaard et al. ¹⁵³ could be developed using MTT and incorporating available clinical MRI imaging with applicable sequences such as FAIR‐FISP, ASL, CASL, and IVIM. Alternative CT or photoacoustic imaging methods could also be employed.

To further enhance PDT treatment planning, personalization must focus on optimizing efficacy, determining parameters within the clinician's control, including the post‐PS administration time point for maximum local PS uptake and selectivity, as well as the number, density, position, and power of implanted light sources. Advances in simulation tools and algorithms, including Monte Carlo simulations and optimization algorithms, have enabled customizing PDT treatment plans based on a patient's anatomy and tumor target. Pretreatment planning using currently available tools must become a standardized part of the treatment protocol to achieve optimal results. This involves accurately quantifying the inter‐ and intra‐patient variability in PS accumulation, tissue optical properties, and oxygen for the planning process.

Integrating these measurements into personalized PDT pretreatment planning software approaches is still in its early stages. As a community, we must discuss approaches for standardizing primary input data to improve outcomes across all applicable indications. Conducting PDT treatment planning within the context of these uncertainties can also enhance the evaluation of patient eligibility before treatment initiation.

The applicability of these dosimetry, dose prediction, and monitoring approaches to new acoustic and X‐ray sensitizer activation techniques remains to be determined.

Establishing standardized minimal PDT dosimetry requirements for iPDT should be an immediate goal for the PDT community, followed by developing improved treatment planning systems applicable to a wide range of clinical indications.

Biographies

Tina Saeidi is a PhD candidate in Medical Biophysics at the University of Toronto, specializing in photodynamic therapy (PDT) research and treatment planning. She holds a B.Sc. in Electrical Engineering from Iran and an M.Sc. in Laser and Photonics from Ruhr University Bochum, Germany. Tina's research focuses on developing computational models to predict photosensitizer distribution in tumors, utilizing parameters such as blood flow and transit time. Her work aims to improve pre‐treatment planning for PDT and enhance patient selection by increasing the precision and efficacy of treatment planning.

Azin Mirzajavadkhan is a PhD candidate in Biomedical Engineering at the University of Toronto, specializing in preclinical cancer research. She earned her B.Eng in Biomedical Engineering from Toronto Metropolitan University. Currently her research focuses on using experimental models and image‐based methods to assess the impact of cancer treatments on metastatic bone quality. Her work includes establishing a novel treatment approach combining nanoparticles and radiation therapy (radiation‐induced photodynamic therapy) in a preclinical model of mixed metastases derived from prostate cancer cells. Through her multidisciplinary research, Azin has developed an appreciation for innovative techniques and their potential to advance cancer treatment strategies.

Lothar Lilge obtained his M.Sc. in Experimental Physics from the Johann Wolfgang Goethe University in Frankfurt, Germany in 1987 and his Ph.D. in biophysics from the Westfaelische Wilhelms University in Muenster, Germany, in 1992, both with Prof. F. Hillenkamp as supervisor. His Ph.D. studies included 3 years at the Massachusetts General Hospital in Boston, USA. In 1992 he came first to McMaster University Hamilton and then the Ontario Cancer Institute for his post‐doctoral training. From 1995 to 2002, he was employed by Photonics Research Ontario, a Centre of Excellence of the Province of Ontario, serving in various positions from Staff Scientist to Director of Operations for the Biophotonics User Facility. He holds a Professor appointment in the Department of Medical Biophysics at the University of Toronto. He is a Senior Scientist at the Princess Margaret Cancer Centre at University Health Network. His expertise includes photodynamic therapy, covering topics from photosensitizer evaluation to treatment planning and clinical trial support. Indications are for a range of cancers and antimicrobial infection control. Optical diagnostics for disease detection or quantification of disease prevalence, particularly for Breast Cancer. Currently, Dr. Lilge is a board member of the International Photodynamic Association (IPA) and Optics Within Life Science (OWLS).

Saeidi T, Mirzajavadkhan A, Lilge L. Treatment planning evolution: Comparing approaches in photodynamic and radiation therapies. Photochem Photobiol. 2025;101:1100‐1119. doi: 10.1111/php.14071