1.

Introduction

Photodynamic therapy (PDT) is a cancer treatment modality that utilizes light energy to activate a preadministered photosensitizer drug to achieve localized tumor control.^{1, 2, 3} It is well established that PDT cytotoxicity is mainly mediated via singlet oxygen produced during the photochemical reactions.^{4, 5, 6, 7} Similar to other types of radiation therapies, proper dosimetry is critical to the success of a PDT treatment. Given a target with its intrinsic sensitivity to PDT, the effectiveness of the treatment depends on the interplay of three basic factors: The concentration of the photosensitizer in the target, the light absorption by the photosensitizer, and the presence of molecular oxygen.^{1, 3} There are other factors, such as bleaching of the photosensitizer and oxygen concentration decrease, that can affect the treatment outcome.^{8, 9} Current clinical PDT dosimetry is largely based on two parameters, delivered optical and drug “doses.” The optical dose is often described as the energy fluence and fluence rate per unit area (for superficial irradiation) or per unit length (for interstitial irradiation). The parameters do not consider the large variations in optical inhomogeneities in each specific target. The photosensitizer concentration in the target tissue is determined by the pharmacokinetics of the drug and, similar to the optical properties, varies greatly both intrapatient and interpatient. In the current PDT dosimetry, these variations are neglected, largely due to the complexity of the issue.

It is known that monitoring the production of singlet oxygen by measuring its luminescence at $1270\mathrm{nm}$ would provide the ultimate marker for PDT cytotoxicity and the ideal dosimetry technique,^{10, 11} yet it is hampered by several issues. The luminescence is extremely weak and is interfered with by the infrared light of the body, it has a half-life on the order of ns.^{12, 13} and its wavelength is beyond most conventional optical detectors. Although technically it is possible to overcome these disadvantages, these characteristics of singlet oxygen luminescence do make it difficult and too expensive as a method for practical applications.

Reactive oxygen species (ROS)–specific chemiluminescence (CL) probes have been used to evaluate ROS production. A probe molecule can interact with ROS and, after a series of reactions that result in the release of additional chemical energy, release photons.^{14, 15} Depending on the probe used, the photons are typically in the visible wavelength range and are easily measurable with conventional optical detectors such as a photon multiplier tube (PMT). We have previously reported that, in an in vitro system, (FCLA-CL) is directly related to PDT treatment parameters.^{16, 17} More importantly, without differentiating the treatment protocols, the total CL observed during a PDT treatment is a direct marker for the cell viability. This establishes the base of the current study, that ROS-induced CL may be used as a marker for real-time PDT dosimetry monitoring.

There are many issues that need to be addressed before this technique can have any practical application. Most important is the issue of whether it be used in vivo, as that is the ultimate goal of PDT dosimetry. Compared to an in vitro study in which the experimental conditions and optical properties of a sample can be well controlled, a real tissue sample poses several problems. PDT irradiation light, scattered by the target sample, is an overwhelming source to diminish the signal-to-noise ratio in the measurement. In the in vitro study, the samples were nearly optically transparent, and the excitation light was introduced to the sample in a perpendicular way to the optical axis of the detector system, yet there was still a significant amount of light reaching the PMT that required the use of band-limiting filters for both background suppression and PMT protection. This is a far more severe problem for in vivo measurement, as most tissues are highly scattering at wavelengths used for PDT.

The solution to this problem lies in the mechanism of CL photon production. Because CL photons are produced as a result of chemical reactions, the intensity decay upon termination of the singlet oxygen source (namely, turning off the PDT excitation) is governed by the chemical reaction speed. We thus hypothesize that CL produced in PDT has a long lifetime. It can be gated for in vivo measurement without the interference of the excitation light source.

2.

Materials and Methods

2.2.

Apparatus for Chemiluminescence Detection

The schematic for chemiluminescence detection is shown in Fig. 1 . The irradiation light source for the photosensitization reaction is a custom-built gated diode laser system (maximum power $100\mathrm{mW}$ , $635\mathrm{nm}$ laser diode controller, LDC 2000, ThorLab, USA). The chemiluminescence was measured using a photon multiplier tube (PMT, Model MP 952, PerkinElmer Optoelectronics, Wiesbaden, Germany) with a counter (PCL-836, Advantech Co., Ltd., Taiwan). The irradiation and CL measurement system is synchronized and controlled by LabVIEW (LabVIEW version 6.1, National Instruments, USA). The time sequence and length of the irradiation and data acquisition are both controlled by the program. To avoid damage and saturation of the measurement system, the PMT is turned off during light irradiation. A pair of band filters [ $590\text{‐}\mathrm{nm}$ short-pass (SP) and a $530\text{‐}\mathrm{nm}$ bandpass (BP) filter] is used to further protect the PMT from scattered irradiation light.

Fig. 1

Schematic of the experimental system for chemiluminescence measurements.

3.

Results

3.1.

Delayed Chemiluminescence and CL Quenching Experiment

The temporal profiles of CL were analyzed for its delayed luminescence property. Figure 2 shows the CL decays, and the curves are fitted with the single exponential component. The lifetimes are $0.18\mathrm{s}$ and $0.07\mathrm{s}$ for samples with and without HSA, respectively. The results also indicate that in solution of PBS∕HSA, the intensity of CL is higher²³ and the lifetime is longer than in PBS. The t-test about lifetime between PBS∕HSA and PBS was made, and $\mathrm{p}⩽0.01$ .

Fig. 2

The decay CL and the single-exponential fit. In PBS and PBS∕HSA solution (PpIX $10\mu \mathrm{M}$ , FCLA $5\mu \mathrm{M}$ , HSA $5\mu \mathrm{M}$ ), the photosensitizer was excited by laser ( $20\mathrm{mW}∕{\mathrm{cm}}^2$ , $635\mathrm{nm}$ ). The irradiation and detection times are $1\mathrm{s}$ and $2\mathrm{s}$ , respectively. The correlation coefficient is A: ${\mathrm{R}}^{\wedge }2$ 0.95101; B: ${\mathrm{R}}^{\wedge }2$ 0.95965.

Overall, the CL decay is considerably slower compared to that of ${}^1\mathrm{O}_2$ fluorescence at $1270\mathrm{nm}$ . It has been reported that PpIX PDT in PBS is a Type II reaction.^{24, 25} To confirm that ${}^1\mathrm{O}_2$ is the main ROS product in our experiment, a well-established ROS quencher $\mathrm{Na}{\mathrm{N}}_3$ was used to selectively quench the ${}^1\mathrm{O}_2$ in both PBS and PBS∕HSA solution. With $\mathrm{Na}{\mathrm{N}}_3$ added to the samples, CL production was significantly quenched by over 70%, with or without the presence of the HSA (Fig. 3 ). This proved that the dominant CL in the study was due to ${}^1\mathrm{O}_2$ .

Fig. 3

Effect of $20\text{‐}\mu \mathrm{l}$ $(0.05\mathrm{mol}∕\mathrm{l})$ sodium azide $\left(\mathrm{Na}{\mathrm{N}}_3\right)$ on fluoresceinyl cypridina luciferin analog chemiluminescence (FCLA-CL) in PBS solution and PBS∕HSA solution, respectively. The PDT parameters are PpIX, $10\mu \mathrm{M}$ ; FCLA, $5\mu \mathrm{M}$ ; laser, $20\mathrm{mW}∕{\mathrm{cm}}^2$ ; irradiation and detection time $1\mathrm{s}$ and $2\mathrm{s}$ ). ${}^{*}$ indicates $\mathrm{p}<0.01$ .

3.2.

Relationship Between CL Rate and Light Fluence Rate

By evaluating the maximum CL production rate immediately after each irradiation period, the relationship between the irradiation light fluence rate and CL production rate was compared. The result is shown in Fig. 4 . There is a strong linear relationship between the light fluence rate and the CL rate.

Fig. 4

The relationship between CL rate and laser fluence rate in PBS∕HSA solution The fluence rate is 5, 10, 15, 20, 25, 30, 35, and $40\mathrm{mW}∕{\mathrm{cm}}^2$ . The linear fit was done, and the correlation coefficient is 0.98998.

3.3.

Experiment of Changing Irradiating Time

As in any photochemical reaction, CL production requires a certain time periods to reach its dynamic equilibrium upon the onset of light irradiation. To establish a reasonable irradiation time (pulse width), CL was measured immediately after various light irradiation periods. The results show that, at $20\mathrm{mW}∕{\mathrm{cm}}^2$ , it takes approximately $1\mathrm{s}$ of irradiation for the CL production to reach its dynamic equilibrium (Fig. 5 ). The CL lifetime was also analyzed for the effect of irradiation time. The results show that the irradiation pulse width had little effect on the CL lifetime (Fig. 6 ).

Fig. 5

Cumulative CL in PBS∕HSA solution after different laser excitation times. The irradiation time is 0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, 12.8, and $25.6\mathrm{s}$ . The other PDT parameter is the same as shown in Fig. 3.

Fig. 6

In PBS∕HSA solutions, the delayed CL lifetime with different excitation times. The samples were the same batch treated with PDT shown in Fig. 5.

3.4.

Experiment In Vivo

The preliminary toxicity study results show that mouse can easily tolerate the subcutaneous FCLA injection. After injection of saline either alone or with FCLA, the skins responded identically: each injection caused a skin lump that was absorbed completely within an hour. Throughout the experiment, there was no erythema due to the injection. The injection alone, with or without FCLA, caused absolutely no observable behavioral change in the animals. CL signals can be readily detected from in vivo animal skin during PDT with excellent signal-to-noise ratio. As shown in Fig. 7 , the CL signal obtained from the in vivo animal has a very similar temporal profile to that from PBS∕HSA solution. Figure 8 shows the integrated CL from the in vivo skin as a function of PDT irradiation fluence (time). The curve shows that the total amount of CL does not linearly increase with the irradiation time, but rather shows a trend of gradually slowing down.

Fig. 7

The decay CL of the sensitized mouse. PpIX $\left(0.02\mu \mathrm{mol}\right)$ and FCLA $\left(0.01\mu \mathrm{mol}\right)$ mixture $200\mu \mathrm{l}$ (NaCl, 0.9%) was transfused into mouse skin. About $1\mathrm{h}$ later, the mouse was fixed and irradiated by laser $(20\mathrm{mW}∕{\mathrm{cm}}^2$ , $635\mathrm{nm}$ ; irradiation and detection times, $3\mathrm{s}$ and $1\mathrm{s}$ ).The lifetime is about $0.14\mathrm{s}$ .

Fig. 8

The course of cumulative CL on mouse skin. All the CL data for the mouse from Fig. 7 were accumulated. The total laser fluence is about $10\mathrm{J}∕{\mathrm{cm}}^2$ .

4.

Discussion

Controlling the dose is key during PDT. Many methods have been reported to estimate the PDT dose, for example, to predict the PDT outcome by monitoring the oxygen concentration in tissue or its change induced by PDT.^{26, 27} The fluorescence of the photosensitizer was also detected to estimate the dosimetry of PDT.²⁸ But to estimate the subsequent effect of PDT by monitoring the oxygen concentration or the flourescence of the photosensitizer is indirect. It is known that singlet oxygen is the main reactant in PDT, and the detection of singlet oxygen is key to predicting therapeutic effect. Many methods of detection of ${}^1\mathrm{O}_2$ have been explored. Chemiluminescence has been reported as a sensitive means to detect singlet oxygen. We have previously reported using FCLA-CL as an alternative dosimetry technique to evaluate ${}^1\mathrm{O}_2$ production in PDT in vitro. The results indicate that FCLA-CL is a sensitive marker for PDT-induced biological outcome, regardless of the treatment protocols. With in vivo PDT applications, Sakurai have proposed a detection and imaging method in vivo by a sensitive and specific chemiluminescence probe.²⁹ However, the interference from the scattered excitation light source inevitably would become a major deterrent for in situ treatment monitoring, causing a major decrease in the signal-to-noise ratio. The current study has developed a new approach of using CL techniques as a means to measure PDT ${}^1\mathrm{O}_2$ production in vivo.

It has been observed that FCLA-CL has a long lifetime. By gating the excitation light source, CL measurement can be made in a “dark” environment with minimum background interference. The results show that, with the new gated technique, the delayed CL intensity is proportional to the irradiation light fluence rate. The finding is similar to that observed in vitro. Yet a simple relationship between CL intensity and PDT treatment protocol is not a full guarantee of the validation of the technique for in vivo applications. To evaluate the feasibility of the technique, we have investigated various factors that may impact the CL measurement results.

The single-exponential fit was made and the half lifetime was given to indicate the decay rate in Fig. 2. Since the back of the fit curve was not well fitted, a double-exponential fit was done. The analysis indicates that the average lifetime of the slow components of the fit is about $2\mathrm{s}$ , and the initial intensity is only about 10% of the fast components. The selectivity of the delayed CL technique was evaluated by using a highly selective ${}^1\mathrm{O}_2$ quencher, $\mathrm{Na}{\mathrm{N}}_3$ , in the PDT-CL process. The results clearly demonstrate that the majority of the CL signal was due to ${}^1\mathrm{O}_2$ , the well-known dominant agent that mediates PDT biological effect. But the double-exponential fit and the proportion of the CL quenching also reveal that superoxide may contribute, to a degree, to the CL production, and the fast component must come from singlet oxygen. Nevertheless, as shown in our previous in vitro study, integrated CL provides a marker directly linked to the PDT-induced biological effect.

In a practical PDT treatment, it is likely that the changes in singlet oxygen production occur in a gradual way that does not require continuous monitoring that in turn requires frequent interruption to the treatment process. With minimal interruption of the treatment, the CL measurement can be made periodically every few seconds or minutes, depending on the treatment protocol and progress. The impact of the irradiation length to the subsequent delayed CL signal was studied. The CL intensity increased gradually with the irradiation light period and eventually reached a plateau (approximately after $1\mathrm{s}$ irradiation with our experimental protocol). Variation in the irradiation time has minimum effect on the delayed CL lifetime. These findings are true both in vitro and in vivo. It is likely due to the fact that CL is a continuous excitation and de-excitation process that takes a certain time to reach its dynamic balance. The results suggest that the delayed CL measurement is a reliable monitoring of the PDT process, given that each measurement follows an adequate irradiation light period.

Our results show that the delayed CL monitoring technique can be adopted easily for in vivo study. Direct ${}^1\mathrm{O}_2$ fluorescence measurement has been reported recently in vivo using a specialized PMT and pulse laser system.²⁵ In comparison, the CL technique can be realized with a conventional optical system with minimum modification to the existing PDT system. It provides a new approach for in vivo ${}^1\mathrm{O}_2$ monitoring during PDT with significantly improved signal-to-noise ratio at a lower cost. As shown in the results, CL did not increase linearly with increased irradiation light fluence. This may not indicate a true decrease in ${}^1\mathrm{O}_2$ production rate but is likely due to combined local chemical consumption∕redistribution of FCLA and PpIX and photo-bleaching of PpIX. Among these factors, only changes in local PpIX concentration and photo-bleaching would lead to true changes in ${}^1\mathrm{O}_2$ production. More detailed study is necessary to identify the contributions of each factor. The major disadvantage of the new technique is that the ${}^1\mathrm{O}_2$ measurement is made through the FCLA probe that itself has to be introduced into the target. It is obvious that quantitative measurement of FCLA-CL is more complex compared to that of direct ${}^1\mathrm{O}_2$ fluorescence, with the added pharmacokinetics and irreversible consumption of the FCLA. Our results clearly show that the technique is feasible in ${}^1\mathrm{O}_2$ detection in vivo. It is thus critical to understand the characteristics of FCLA. One possible approach to resolve this issue is to use the fluorescence signal of FCLA that is readily detectable both in vivo and in vitro as an internal marker for the probe. This is currently under investigation.

In summary, this work has shown the feasibility of using delayed CL to detect singlet oxygen in vivo and also to obtain some useful data about detection in vivo. Based on these data, we will study the relation between CL and the curative tumor response.