Ronan M.
Valentine
*a,
Sally H.
Ibbotson
a,
Kenny
Wood
b,
C. Tom A.
Brown
b and
Harry
Moseley
a
aPhotobiology Unit, Ninewells Hospital and Medical School, University of Dundee, Dundee, DD1 9SY, UK. E-mail: r.y.valentine@dundee.ac.uk
bSUPA, School of Physics and Astronomy, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, UK
First published on 25th October 2012
Understanding the interactions of non-ionizing radiation with living organisms has been the focus of much research over recent decades. The complex nature of these interactions warrants development of theoretical and experimental studies to gain an insight into predicting and monitoring the success of photodynamic therapy (PDT) protocols. There is a major impetus towards evidence-based recommendations for patient diagnosis, treatment and management. Knowledge of the biophysical aspects of PDT is important for improving dosimetry protocols. Fluorescence in clinical PDT may be used to detect and diagnose pre-malignant and malignant conditions, while photobleaching can monitor changes in fluorescence during treatment. Combining empirical fluorescence photobleaching clinical data with computational modelling enables clinical PDT dosimetry protocols to be investigated with a view to optimising treatment regimes. We will discuss how Monte Carlo radiation transfer (MCRT) modelling has been intercalated in the field of fluorescence detection and PDT. In this paper we highlight important aspects of basic research in PDT by reporting on the current utilisation of fluorescence in clinical PDT from both a clinical and theoretical perspective. Understanding and knowledge of light propagation in biological tissue from these perspectives should have a positive impact on treatment planning.
PDT dates back as far as 1900, when Oscar Raab was the first to accidently demonstrate that the combination of acridine red and light elicited a detrimental effect on in vitro paramecium.4 Lipson et al. were largely responsible for initiating the modern era of PDT when studies involving hematoporphyrin derivative (HpD) were performed in the 1960s.5,6 In 1978, Dougherty first demonstrated the clinical effectiveness of PDT by performing treatments on 25 patients presenting with a total of 113 skin tumours.7 Kennedy et al. introduced the concept of topical PDT employing 5-aminolaevulinic acid (5-ALA) and reported his clinical experience in 1990.8 As a result, PDT in dermatology typically employs photosensitiser pro-drugs, usually 5-aminolaevulinic acid (5-ALA) or its methyl ester, methyl aminolevulinate (MAL).
Following administration of these pro-drugs to skin lesions, uptake and conversion to protoporphyrin IX (PpIX) occurs via the haem biosynthetic pathway resulting in the accumulation of PpIX in lesional cells. PDT occurs when PpIX is activated with wavelengths around 630 nm and this is chosen due to its depth of tissue penetration. Fluorescence detection (FD) takes place when PpIX is excited at around 405 nm, which then emits a characteristic red fluorescence peak at approximately 635 nm. FD is widely used as an adjunct to PDT. In 1924, Policard reported on the fluorescence of tumours under illumination with UV/violet light.9 Since then there have been many fluorescence studies pertaining to PDT in dermatology.10–13 The observation of surface PpIX fluorescence is typically used as a detector of pre-malignant and malignant diseases. Also, the decrease in surface PpIX fluorescence during PDT (photobleaching) serves to monitor the treatment progress. An optical technique, known as fluorescence spectroscopy, may be employed to detect this surface PpIX fluorescence by using light–tissue interactions. The diagnostic potential of fluorescence spectroscopy lies in its sensitivity to various biochemical and structural changes in tissue that accompany the developments from normal tissue to cancer.14,15
Implicit dosimetry is a technique, which uses fluorescence photobleaching as a dose metric and a surrogate for measuring the generation of singlet oxygen.16 Therefore, monitoring in vivo PpIX fluorescence and photobleaching during PDT may provide information about the amount of photosensitiser in the tissue that has become photobleached during treatment, allowing singlet oxygen production to be inferred.17 While surface PpIX fluorescence is diminished at the end of standard treatment, PpIX fluorescence may still be present at depth in the tumour. Modelling fluorescence can extract pertinent photosensitiser and singlet oxygen information at depth, which should optimise clinical PDT treatment times.
Fluorescence signals depend on the characteristics of the fluorophore in question but are also strongly influenced by tissue optical properties.18 Further optimisation warrants investigation into the propagation of light through biological tissue. In order to gain more of an understanding of fluorescence in clinical PDT and to fully optimise clinical PDT treatments, radiation transfer simulations may be used. One approach is to use Monte Carlo radiation transfer (MCRT) modelling. Our three-dimensional (3D) MCRT model can tag PpIX fluorescence photons enabling us to infer where exactly they originate from within the tumour and subsequently track them to the tumour surface. Also, the position of the singlet oxygen generated within the tumour can be recorded. Monte Carlo radiation transfer (MCRT) modelling can enable predictions to be made about the efficacy of clinical PDT treatments.19
Based on the assumption of implicit dosimetry and a clinically derived photobleaching dose constant, β, as a model input parameter, we have used our MCRT model to address the question of whether it is advisable to increase treatment times beyond the disappearance of surface PpIX fluorescence. We believe there is an opportunity to provide more effective PDT at depth in the tumour by delivering a larger treatment light dose.
Fig. 1 Light–tissue interactions upon which optical diagnostic techniques operate. The chromophore is used to imply that a molecule absorbs light, while the fluorophore indicates that a molecule emits light. |
Here we focus our attention on fluorescence signals obtained from tissue. There are many clinical advantages associated with fluorescence spectroscopy such as earlier diagnosis and immediate treatment. Also, it can assist in pharmacokinetic studies of drug-induced fluorescence. Fluorescence detection may potentially provide a higher predictive accuracy in locating optimal biopsy sites and assist in defining surgical margins for tumour excision.21 Importantly, changes in fluorescence spectra pertains to biochemical changes in tissue that precede morphological changes in tissue, which may potentially offer earlier detection of disease.22
Fig. 2 Broad PpIX absorption spectrum together with characteristic PpIX fluorescence emission with a dominant peak at 635 nm. |
Skin chromophore | Absorption spectral range | Fluorescence | Absorption maxima (nm) | Emission maxima (nm) |
---|---|---|---|---|
Oxyhaemoglobin | UV-visible | No | 412, 542, 577 | — |
Deoxyhaemoglobin | UV-visible | No | 430, 555, 760 | — |
Melanin | UV-visible | No | Increases with decreasing wavelengths | — |
Water | IR-long visible | No | 760, 900, 1250, 1400 | — |
Porphyrins | Visible | Yes | Ex: ∼405 nm | Em: 630 nm |
NADH | UV | Yes | Ex: ∼350 nm | Em: 460 nm |
Tryptophan | UV | Yes | Ex: 295 nm | Em: 340–350 nm |
Collagen | UV | Yes | Ex: 335, 370 nm | Em: 380, 460 nm |
Elastin | UV-visible | Yes | Ex: 420, 460 nm | Em: 500, 540 nm |
Prodrug-induced PpIX generation depends on the enhanced retention at the site of application, the augmented rate of pro-drug uptake and also on the rate of enzymatic conversion of the pro-drug into PpIX.26 The success of PDT may be limited by the transport and distribution of the pro-drug in the tumour.27
Fluorescence spectroscopy commonly employs fibre optic probes – placed in contact with the tissue surface – which deliver excitation light to a tissue site.28 Also, studies involving fluorescence imaging have been carried out, allowing larger suspicious areas to be inspected.29,30 We have used an optical biopsy system (OBS) operating on the principle of fluorescence spectroscopy to investigate and compare ALA and MAL-induced PpIX fluorescence in vivo in patients presenting with Bowen's Disease (BD) and superficial basal cell carcinoma (sBCC) receiving PDT.31Fig. 3 illustrates differences in PpIX fluorescence from a sBCC and surrounding normal skin and Fig. 4 depicts mean PpIX fluorescence intensity spectra and photobleaching of PpIX recorded from patients presenting with BD and sBCC and with varying cream application times. In all cases, the mean PpIX fluorescence intensities recorded during treatment and after treatment are less than 10% and 5%, respectively of their corresponding mean PpIX fluorescence intensity recorded before treatment.
Fig. 3 Fluorescence intensity spectra illustrating the differences in PpIX fluorescence, from a sBCC lesion (-) and surrounding normal skin tissue () 6 h after ALA application. Normal skin tissue that has not been incubated with ALA is also shown (). Reduced autofluorescence from the lesion (-) is typical when compared to the surrounding normal skin tissue.31 |
Fig. 4 Mean PpIX fluorescence intensity spectra from patients presenting with (a) sBCC after ALA application (6 h) (n = 10); (b) sBCC after MAL application (3 h) (n = 10); (c) BD after ALA application (4 h) (n = 10); and (d) BD after MAL application (3 h) (n = 10), recorded immediately before PDT (-), mid-irradiation () and immediately post-PDT ().31 |
β = τ (Ψ) |
Fig. 5 Reduction in mean normalised PpIX fluorescence intensity during PDT recorded from patients presenting with (a) sBCC after ALA application (6 h) (n = 10); (b) sBCC after MAL application (3 h) (n = 10); (c) BD after ALA application (4 h) (n = 10); and (d) BD after MAL application (3 h) (n = 10), allows for the monitoring of in vivo photobleaching. Diamonds (◊) are representative of data from individual lesions. A best-fit exponential is shown, and the time for fluorescence reduction to 50% is indicated.31 |
β is defined as the light dose that causes a 37% reduction in the photosensitiser fluorescence signal.32 PpIX fluorescence photobleaching occurs during treatment leading to a decrease in the observed PpIX fluorescence as the photosensitiser is photochemically destroyed by light. During PDT, singlet oxygen evolves from the interaction between excited triplet state photosensitiser molecules in tumour cells and molecular oxygen where an energy transfer occurs leading to singlet oxygen production3 (Fig. 6). Singlet oxygen is regarded as the main PDT-induced cytotoxic agent.33 The photobleaching mechanism is assumed to rely on singlet oxygen16 where reactions occur between the ground state photosensitiser and singlet oxygen causing irreversible photosensitiser destruction.
Fig. 6 Jablonski diagram illustrating PDT reactions. |
PpIX fluorescence photobleaching may indirectly predict singlet oxygen generation and thus be indicative of the photodynamic dose (PDD) delivered during PDT treatment.49 The PDD may be defined as the local yield of singlet oxygen generation per unit volume of tumour tissue and is assumed to be proportional to the number of photons absorbed by the photosensitiser per unit volume of tumour tissue.34
PpIX fluorescence photobleaching may be used as a dose metric and has previously been modelled using the following equation
The PDD is directly related to the photosensitiser concentration and the light fluence rate as shown in
(1) Direct dosimetry depends on the detection and measurement of singlet oxygen during treatment. This may be achieved through singlet oxygen luminescence dosimetry (SOLD), which detects singlet oxygen luminescence at 1270 nm and in doing so can quantify the amount of singlet oxygen generated.36
(2) Explicit dosimetry refers to predicting the singlet oxygen generation by measuring the interdependent quantities, light, photosensitiser and oxygen individually during a PDT treatment. This dosimetry approach is challenging as it requires the measurement of the different treatment factors, which through their interdependency will impact on each other.
(a) Photosensitiser dosimetry refers to determining the photosensitiser concentration in the target tissue, which may be quantified through optical spectroscopy techniques, such as fluorescence and absorption.37
(b) Oxygen dosimetry is the least evolved in PDT. However, there have been several interesting studies performed.38,39
(c) Light dosimetry is described by calculations of the light dose, which can be practically measured and clinically implemented into routine PDT treatment protocols. The light dose is calculated by integrating the light irradiance over the time of exposure.
(3) Implicit dosimetry uses a surrogate measure – fluorescence photobleaching of the photosensitiser – during treatment to predict the generation of singlet oxygen and therefore the biological response. In vivo studies have shown a positive correlation between PpIX photobleaching and tissue damage.32
Tissue optical properties are wavelength dependent and therefore this dependency strongly affects how deep the light may penetrate into the tissue. The therapeutic window between 600 and 1000 nm is the spectral region ideal for diagnostic and therapeutic medical applications such as PDT and fluorescence spectroscopy.23 In the visible region, the depth of light penetration increases as the wavelength increases.44 Wavelengths within this range penetrate the deepest into tissue, partly due to lower scattering but mainly because of lower absorption.45 Haemoglobin absorption greatly affects the attenuation of 405 nm light more so than 630 nm light.46 Tissue optical properties are influenced by the photosensitiser concentration and blood oxygenation, and during PDT the distribution of photosensitiser and oxygen may be altered due to photobleaching.47 The fluence rate has a dominant effect in PDT and this has been shown through several studies.32,48,49 It is important to note that the optical properties affect the fluence rate, which in turn has an impact on the PDD.
The fluence rate will increase sharply at the air-tissue boundary and then exponentially decrease with increasing depth. This sub-surface peak is attributed to the loss of backscattered light through the tissue surface.16 The penetration depth describes the optical transparency of tissue and is defined as the depth in the tissue at which the intensity of the propagating light falls to approximately 37% (1/e) of the incident value.46 The penetration depth is given by
δ = 1/μeff |
Wilson and Patterson50 have reported the depth of a PDT treatment, dt with typical clinical photosensitisers as
dt ∼ 3−5δ |
A comprehensive review of optical properties has been carried out by Cheong et al.40 and Sandell and Zhu.47 Non-invasive optical properties measurements may be obtained by absorption spectroscopy and theoretically using the diffusion approximation47 and Monte Carlo modelling.
One of the most common techniques used to simulate light propagation through tissue is the Monte Carlo radiation transfer (MCRT) method.52–54 The MCRT method is a technique that solves the RTE in order to compute the distribution of light in tissue. MCRT modelling is the most flexible approach to describing photon transport in biological tissue offering information about the trajectories of individual photons through a medium, which builds up an accurate depiction of where and how the light traversed the medium.55 It is often chosen to simulate light transport in tissue as it can handle anisotropic scattering, complicated geometries, acquire tissue optical properties and compute the light dose administered to structures similar to those observed in PDT.52
It is statistical in nature and describes the propagation of light by utilising probability distribution functions (PDFs). By sampling randomly from PDFs using cumulative distribution functions (CDFs), variables such as the optical depth, τ, the albedo, a, and photon scattering angles; cosine of the deflection angle, cos θ and the azimuthal angle, ϕ, may be randomly chosen at interaction sites, enabling the position, direction and path of a photon to be determined. A computer generated pseudo-random number is used to represent a variable that is to be determined. Random numbers are the crux of the MCRT method.
The scattering of a photon is described by two directional angles; the deflection angle, θ and the azimuthal angle, ϕ. The Henyey–Greenstein scattering phase function56 is adopted to approximate scattering in tissue and is expressed mathematically in the form
g is defined as
Jacques et al.57 have shown that scattering within skin tissue can be represented quite well by the Henyey–Greenstein scattering phase function. Typical values for g are in the region of 0.7 to 0.95.
Furthermore, the azimuthal angle is taken into account by the following equation
where P(ϕ) is the probability distribution function and ϕ is the azimuthal angle, which is sampled over the interval 0 to 2π thus
ϕ = 2πξ |
where ξ is a random number in the range 0 to 1.
Fig. 7 Three-dimensional (3D) MCRT model geometry, where x, y are mutually orthogonal axes in the skin surface and z represents depth within the skin tissue.19 |
A photobleaching dose constant, β, was obtained from in vivo ALA-induced PpIX fluorescence photobleaching data recorded from patients presenting with sBCC (Fig. 5(a)) This clinically derived value was calculated and found to be 14 J cm−2. A range of values for β have been reported in the literature ranging from 1.8 J cm−2–33 J cm−2.17,18,32,49 Therefore, from an implicit dosimetry perspective, this key parameter determines the rate at which singlet oxygen is generated in the tumour during treatment. A surface irradiance, of 82 mW cm−2 was delivered over a simulated treatment time of approximately 30 minutes, thus administrating a simulated total treatment light dose of 150 J cm−2.
PpIX fluorescence detection was simulated at particular time points in the treatment in the presence of photobleaching. PpIX fluorescence photons were tagged in our model enabling us to ascertain exactly where they originated from within the tumour. Fig. 8(a) illustrates the simulated PpIX fluorescence detected at the tumour surface, which has originated from varying tumour depths. At the beginning of the treatment simulation, PpIX fluorescence photons generated from superficial regions were high, which decreased rapidly with depth. PpIX fluorescence photons generated at shallower tumour depths have a higher probability of escape than those photons generated at deeper tumour depths. As the treatment simulation progressed and photobleaching occurred, the PpIX fluorescence photons generated near the surface decreased with more PpIX fluorescence photons originating from deeper within the tumour, contributing more to the PpIX fluorescence detected at the tumour surface.
Fig. 8 (a) PpIX fluorescence detected at the surface that has originated from varying depths in the tumour with increasing light dose in the treatment and fluorescence simulations. (b) Total PpIX fluorescence detected at the surface of the tumour with increasing light dose in the treatment and fluorescence simulations.19 |
Fig. 8(b) depicts the total PpIX fluorescence detected at the tumour surface with increasing treatment light dose.
We also recorded the position of the singlet oxygen generated within the tumour during the treatment simulation in the presence of photobleaching. Fig. 9 depicts the singlet oxygen production in units of 1O2 cm−3 generated in the tumour by the photosensitiser as a function of treatment light dose at varying tumour depths (1, 2, 3 and 4 mm) for a set of optical properties.61 In our model, the threshold photodynamic dose, PDT, assumed to be 8.60 × 1017 1O2 cm−3 generated by the photosensitiser34 was signified by a horizontal line. Tissue necrosis was assumed when PDD > PDT (Fig. 9).
Fig. 9 Singlet oxygen production in units of 1O2 cm−3 generated by the photosensitiser in the tumour, as a function of light dose at varying tumour depths (1, 2, 3 and 4 mm) with chosen optical properties.19 |
MCRT modelling of fluorescence and photobleaching enables predictions to be made relating to the origin of fluorescence and generation of the singlet oxygen.19 Light dosimetry planning in PDT involves estimating the fluence rate in the tissue, which is used to determine the depth of light penetration along with the value and position of the maximum light dose delivered within the tissue.62 Although accurate determination of optical properties in vivo is difficult, new technologies are paving the way for larger data sets from a range of human tissues.47
Our simulated data (Fig. 8 and 9) should facilitate predictions for PDT dosimetry pertaining to the light dose and photodynamic dose. While our model incorporates some simplifications such as assuming a constant β, unlimited oxygen availability and uniform PpIX distribution throughout the tumour, our results suggest that delivering a larger treatment light dose beyond the disappearance of surface PpIX fluorescence may continue to provide effective treatment at deeper locations in the tumour. Oseroff et al.63 previously reported on the need for a treatment light dose of at least 100 J cm−2 for effective PDT. Optimal PDT treatment regimes would provide minimal damage to normal skin while delivering an effective therapeutic result to both superficial and thicker NMSCs. Further in vivo studies are required and new provisions undertaken in order to further knowledge pertaining to PpIX fluorescence photobleaching in clinical PDT of NMSC. Also, further modelling of photon transport in tissue and calculations of photon distribution based on the RTE, which require knowledge of the absorption and scattering coefficients and the anisotropy factor can assist in optimising PDT treatment parameters, dose metrics and treatment regimes.
Footnote |
† This article is published as part of a themed issue on current topics in photodermatology. |
This journal is © The Royal Society of Chemistry and Owner Societies 2013 |