Multispectral singlet oxygen luminescent dosimetry (MSOLD) for Photofrin-mediated Photodynamic Therapy

Abstract

Direct detection of singlet-state oxygen ([¹O₂]) constitutes the holy grail dosimetric method for type II PDT, a goal that can be quantified using multispectral singlet oxygen dosimetry (MSOLD). However, the short lifetime and extremely weak nature of the singlet oxygen signal produced has given rise to a need to improve MSOLD signal-to-noise ratio. This study examines methods for optimizing MSOLD signal acquisition, specifically employing an orthogonal arrangement between detection and PDT treatment light, consisting of two fiber optics - connected to a 632-nm laser and an InGaAs detector respectively. Light collected by the InGaAs detector is then passed through a filter wheel, where spectral emission measurements are taken at 1200 nm, 1240 nm, 1250 nm, 1270 nm, and 1300 nm. The data, after fitting to the fluorescence background and a gaussian-fit for the singlet oxygen peak, is established for the background-subtracted singlet oxygen emission signal. The MSOLD signal is then compared with the singlet oxygen explicit dosimetry (SOED) results, based on direct measurements of in-vivo light fluence (rate), in-vivo Photofrin concentration, and tissue oxygenation concentration. This study focuses on validating the sensitivity and minimum detectability of MSOLD signal in various in-vitro conditions. Finally, the MSOLD device will be tested in Photofrin-mediated PDT for mice bearing Radiation-Induced Fibrosarcoma (RIF) tumors.

1. INTRODUCTION

Multispectral singlet oxygen luminescence dosimetry (MSOLD) is a promising technique for use as a possible in-vivo, direct method of photodynamic therapy (PDT) dosimetry [1][2]. Unlike singlet oxygen explicit dosimetry (SOED) determining the singlet oxygen by theoretical calculation based on explicit measurement of light fluence, PS uptake, and tissue oxygen concentrations, MSOLD directly measures fluorescence emission for singlet oxygen [¹O₂] reactive species during PDT treatment therefore providing a possible simple solution for PDT dosimetry. It is well known that [¹O₂] fluorescence signal in the region of 1270 nm is extremely weak and buried under a strong luminescence background signal, which results in technical challenge of detecting [¹O₂] signal directly. Direct singlet oxygen luminescence dosimetry (SOLD) could resolve the weak short-lived [¹O₂] signal using high repetition pulsed laser and near-infrared photomultiplier tubes (PMTs) [3]. However, the use of pulsed laser and complex technique makes it incompatible with most of existing clinical PDT studies using continuous-wave (CW) light sources. With the recent advent of commercially available InGaAs detector that can capture near-infrared (NIR) spectral signal more efficiently, number of studies have adopted this new technology and successfully detected the weak [¹O₂] signal during photodynamic therapy[4]-[6]. Ryan et al [7]also captured singlet oxygen signal in Photofrin phantom using multichannel InGaAs detector and demonstrated the MSOLD signal was linearly proportional to the SOED-calculated [¹O₂]_rx. However, the signal-to-noise ratio of MSOLD still needs significant improvement; there is also a need to demonstrate the sensitivity of MSOLD signal at low Photonfrin concentration that is closer to clinical application.

In this study, we have performed a series of measurements in vitro in phantoms with Photofrin concentration from 5 to 400 mg/kg in methanol and 2% intralipid. Strong correlation between MSOLD signal and SOED were found confirmed previous results and expanded the sensitivity for Photofrin down to 5 mg/kg with 20% uncertainty. We also tested the MSOLD device in the Photofrin-mediated PDT for mice bearing radiation-Induced Fibrosarcoma (RIF) tumors.

2. METHODS

2.1. MSOLD spectral measurement setup

Figure 1(a) shows a schematic experimental setup with laser pointing along x-axis, and fiber detector placing along y-axis. Tissue-simulating liquid phantoms in disposable cuvettes were prepared in Methanol at photofrin concentrations of 5 mg/kg, 10 mg/kg, 50 mg/kg, 100 mg/kg, 200 mg/kg, 300 mg/kg, 400 mg/kg as well as a 0 mg/kg control. Two prototype MSOLD instruments using InGaAs detectors were employed to detect the [¹O₂] signal, one is a lab-made multichannel dosimeter system with narrow beam filters using a 1 mm fibers, the other one is a commercial InGaAs spectrometer (Avantes, Apeldoorn, Netherlands). A 633 nm CW laser (6 W max output) is used to excite photosensitizer via a fiber with micro-lens tip (Pioneer Optics Company, Bloomfield, CT, USA), with illumination spot size around 1 cm in diameter. The 1 mm cable (Thorlabs, Newton, NJ, USA) orthogonal to incident laser was used to collect photoemission in a cuvette phantom with laser illumination at a fluence rate of 500 mW·cm⁻². A filter wheel with a range of narrow bandpass filters (1200 nm, 1240 nm, 1250 nm, 1270 nm and 1300 nm) are used to obtain signal at those wavelengths as illustrated Figure 1(b). The filtered signal was measured by the lab-made multichannel dosimeter system with mounted InGaAs photodiode (Thorlabs, Newton, NJ, USA) and a custom photodiode amplifier with 5000× gain. A much stronger [¹O₂] signal than previous study was achieved by using such a high gain. The measurements made by Avantes spectrometer can capture the spectrum from 1200 nm to 1600 nm directly from phantom without using the filter wheel.

Figure 1.

(a) Schematic diagram of experimental setup; (b) Actual experimental setup with laser fiber, detector and filter wheel illustrated.

A similar experimental setup described above was used for the in-vivo mouse study, the only difference is that the angle between laser and detector is approximately 50 degrees.

2.2. MSOLD spectral analysis and calibration curve

To fit MSOLD spectrum captured by multichannel dosimeter system, an exponential fit was applied to the points at 1200 nm, 1240 nm, and 1300 nm to obtain the background function. A Gaussian fit was then established for the background-subtracted spectrum to get the signal from singlet oxygen emission. The amplitude of the Gaussian peak was used to establish the calibration curve that relates the known Photofrin concentration to the normalized MSOLD signal.

A linear background was used for the spectrum measured by the Avantes spectrometer. The spectrum ranging from 1240 nm to 1340 nm was fitted with the linear background and Gaussian functions at 1274 nm with fixed width. Shorter spectral range was used due to the complexity of the strong luminescence background.

2.3. SOED data analysis

Singlet-oxygen explicit dosimetry determines the reacted singlet oxygen, [¹O₂]_rx, based explicit measurement of light fluence rate ($\varphi$), photosensitizer concentration [S₀], and oxygen concentration [³O₂] during PDT. They follow the following equations [6]:

$$\frac{d\left[S_0\right]}{dt}=-\xi \frac{\left[{{}^{3}O}_2\right]}{\left[{{}^{3}O}_2\right]+\beta }\varphi \left[S_0\right]\sigma \left(\left[S_0\right]+\delta \right),$$ (1)

$$\frac{d\left[{{}^{3}O}_2\right]}{dt}=-\left(\xi \frac{\varphi \left[S_0\right]}{\left[{{}^{3}O}_2\right]+\beta }\right)\left[{{}^{3}O}_2\right]+g\left(1-\frac{\left[{{}^{3}O}_2\right]}{\left[{{}^{3}O}_2\right](t=0)}\right),$$ (2)

$${\left[{{}^{1}O}_2\right]}_{rx}=\int \xi \frac{\left[{{}^{3}O}_2\right]}{\left[{{}^{3}O}_2\right]+\beta }\varphi \left[S_0\right]\frac{1}{\left(\sigma \left(\left[S_0\right]+\delta \right)+1\right)}dt.$$ (3)

Where the parameters used are listed in Table 1 for Photofrin [6].

Table 1.

Summary of photophysical parameters for Photofrin in vitro

Parameter	Definition	in vitro
ε (cm−1μM−1)	Photosensitizer extinction coefficient	0.0035 @ 632nm
β (μM)	Oxygen quenching threshold concentration	11.9
δ (μM)	Low concentration correction	25 ± 4.3
ξ (cm2mW−1s−1)	Specific oxygen consumption rate	10.3×10−3 @ 632 nm
σ (μM−1)	Specific photobleaching ratio	(6.6±7)×10−5

3. RESULTS

3.1. MSOLD spectral measurements of Photofrin in MeOH

The MSOLD spectral signal was measured for a series of PH concentrations in MeOH in cuvette phantoms. Both Avantes spectrometer and multichannel dosimeter were used to detect the singlet oxygen signal. Liquid phantom with Photofrin concentration 0, 100, 200, 300 and 400 mg/kg were prepared and measured using multichannel dosimeter system with the results shown in Figure 2. (a). The details of data processing for MSOLD spectra measured by multichannel dosimeter can be found in Ref. [7]. Signal of wavelength at 1200 nm, 1240 nm, 1250 nm, 1270 nm and 1300 nm were recorded using InGaAs detectors with 5000× gain. The symbols represent the experimental data after applying correction factor for the flited signal. A linear background and Gaussian function are used to fit the experimental data shown as solid line in Fig. 2(a). The Gaussian function for each concentration is plotted as dashed line in Fig. 2(a). The data points and curves with same color represent data measured from same PH concentration. The MSOLD signal increases as the PS concentration increases. The peak at 1270 nm wavelength indicates the signal from singlet oxygen. It is worth to mention that we have achieved stronger signal at 1270 nm under same Photofrin concentration compared to the work in Ref. [7] as higher gain was used in this study. Figure 2(b) shows the integrated area of SO peak as function of PH concentration, a second order polynomial function is used to establish the calibration curve.

Figure 2.

(a) MSOLD spectral measurements for PH concentration 0, 100, 200, 300, 400 mg/kg in MeOH, the dots, solid line and dash line with each color represents raw data, fitting and Gaussian singlet oxygen signal respectively. (b) The integrated area of singlet oxygen peak as function of PH concentration from fitting of Fig. 1(a), a second order polynomial function is used to fit the data.

MSOLD signal was also measured by Avantes spectrometer with range from 1200 nm to 1600 nm. The full range spectrum reveals more details about the singlet oxygen and background. The measurements were carried out for a series of Photonfrin in MeOH with PH concentration 0, 5, 10, 50, 100, 200, 300 and 400 mg/kg. The measured spectra were fitted with a linear background and Gaussian function as presented in Figure 3(a)-3(h). The intensity of SO signal as function of PH concentration is plotted in Figure 3(i) and fitted with a second order polynomial function. The trend of SO signal versus PH concentration in Fig. 3(i) is like the trend in Fig. 2(b) proving the validity of measurements using multichannel dosimeter system.

Figure 3.

(a)-(h) Smoothed MSOLD spectrum measured by spectrometer for liquid phantom with PH concentration 5, 10, 30, 50, 100, 200, 300 and 400 mg/kg in pure methanol (MeOH), linear background and gaussian function are used; (i) Calibration curve that relates amplitude of singlet oxygen peak to the known Photofrin concentration.

3.2. MSOLD of Photofrin in MeOH with 2% intralipid

We also prepared a variety of Photofrin concentrations (100, 200, 300 and 400 mg/kg) in methanol with 2% intralipid to increase the scattering of light so the optical properties of liquid phantom are closer to real tissue. MSOLD signals were captured by using the Avantes spectrometer as shown in Figure 4(a)–(e), which was fitted using a linear background and a Gaussian function with a fixed width. Strong laser background signal appears at ~1230 nm that is due to the scattering of incident laser by intralipid. The amplitude of the whole spectrum as well as SO signal was suppressed significantly compared to the study in methanol only phantom. The results again demonstrate why it is so difficult to detect SO signal during patient PDT treatment. Figure 4(f) shows a calibration curve for normalized SO peak amplitude and PS concentration that is similar to the measurements without intralipid. The amplitude of SO signal is only about 12 counts for the Photofrin concentration of 100 mg/kg that places the fitting of the spectrum in an important position and stresses the necessity of optimizing the signal detection.

Figure 4.

(a)-(e) MSOLD spectrum and the fitting using Gaussian function and linear background for Photofrin with various concentrations (0, 100, 200, 300, 400 mg/kg) in 2% Intralipid, (f) calibration curve that relates amplitude of normalized SO peak and PS concentration.

3.3. MSOLD and SOED comparison

Singlet oxygen spectra were measured for samples with PH concentration 100, 200, 300 and 400 mg/kg and collected over 900 seconds with rate of 1 spectrum per second. All the spectra were smoothed and then fitted with a linear background and Gaussian function using MATLAB. Fig. 5(a)-5(d) show the intensity at 1274 nm for the raw spectrum and fitted spectrum and SO spectrum over 900 seconds. The decrease of SO signal is due to the bleaching of photosensitizer during the light illuminance. Figure 5(e)-4(h) present the fittings for MSOLD spectrum at each PH concentration at 0 second. Gaussian function at 1274 nm with fixed width was used to fit the raw spectrum from 1240 nm to 1340 nm. The SO signal is about constant for PH concentration 100 mg/kg and 200 mg/kg, while SO signal obviously decreases for PH concentration 300 mg/kg and 400 mg/kg. It is probably because the photosensitizer bleaches at higher rate for higher concentration.

Figure 5.

(a)-(d) Change of MSOLD signal as function of time up to. 900 seconds, the blue, red and yellow dots are the intensity for raw data, fitting curve and singlet oxygen peak at 1274 nm respectively. (e) –(h). The spectrum and the fitting at zero second for samples with PH concentration 100, 200, 300 and 400 mg/kg in MeOH.

To compare MSOLD signal to SOED-calculated [¹O₂]_rx, tissue simulating phantoms of Photofrin in MeOH were prepared at PH concentrations of 50, 100, 200, 300 and 400 mg/kg. The singlet oxygen signal, ground state oxygen and Photofrin concentration were measured simultaneously over 900 seconds with rate of 1 spectrum per second while the sample was illuminated by the CW laser. Full instantaneous MSOLD signal from 1200 nm to 1600 nm was recorded by Avantes spectrometer using the identical experimental setup as illustrated in Figure 1. Singlet oxygen signal for each spectrum was extracted by fitting the spectrum using the same method above. A custom contact probe [8] was used to measure PS uptake. An oxylite oxygen probe (Oxford Optronix, Abingdon, OX, UK) was used to measure oxygen ([³O₂]) concentration. The reacted singlet oxygen, [¹O₂]_rx, was calculated based on the method in Ref. [7]. PS concentration and oxygen partial pressure was measured every second over 900 seconds treatment time. Oxygen partial pressure in mmHg was converted to μM by multiplication of a factor of 1.3 [6]. Fluence rate (mW) was measured with a laser power meter (Coherent, Santa Clara, CA, US).

Figure 6(a)(d) and 6(b)(e) shows a strong correlation between instantaneous and cumulative SOED and MSOLD signal, Figure 6(c) shows the linear correlation between instantaneous SOED and MSOLD for pre-, during- and post-PDT. The cumulative SOED and MSOLD is also linearly correlated as shown in Figure 6(f). The strong linear correlation for both cumulative and instantaneous SOED and MSOLD demonstrates the effectiveness of MSOLD for PDT.

Figure 6.

(a) (d) Instantaneous and cumulative SOED for samples with Photofrin concentration 50, 100, 200, 300 and 400 mg/kg in MeOH. (b) (e) The corresponding instantaneous and cumulative MSOLD. (c) Correlation between instantaneous MSOLD and SOED for pre-, during- and post PDT. (f) Correlation between cumulative MSOLD and SOED.

3.4. MSOLD for in-vivo mice study

MSOLD device was tested in Photofrin-mediated PDT for mice bearing Radiation-Induced Fibrosarcoma (RIF) tumors injected 5mg/kg Photofrin at 24 hours drug-light interval (DLI). The angle between incident laser and detection fiber is approximately 50 degrees that was 90 degrees in the case of phantom experiments (see Fig. 7). Singlet oxygen spectrum was measured with Avantes spectrometer, smoothed SO spectrum at zero second is shown in Figure 8(c) and (d). The SO peak is shifting to about 1280 nm which is probably due to the change of angle between laser and detecting fiber. A Gaussian function with same width as phantom study and linear background are used to fit the SO peak for both mouse studies. Figures 8(a) and 8(b) show the amplitude of SO signal, raw SO peak and fitted SO peak over 900 treatment time. The amplitude of SO signal varies from 10 to 15 with 5 mg/kg Photofrin concentration, which is consistent to the value for 5 mg/kg Photofrin in methanol in Figure 3(a). The MSOLD spectrum from the mouse study is much noisier than spectrum in phantom study, the singlet oxygen peak appears in a broader shape making the fitting using Gaussian function to be more difficult. The noise and peak broadening from measurements in mouse is probably due to the more complicated tissue environment than simple liquid phantom.

Figure 7.

(a) Schematic diagram of experimental setup for MSOLD measurements with mouse. (b) Picture during the measurements with mouse.

Figure 8.

(a) (b) Trend of singlet oxygen amplitude for raw data, fitted spectrum and normalized singlet oxygen peak over 460 seconds for the two in-vivo mice study, (c) (d) Singlet oxygen spectrum and the fitting using linear background and Gaussian function at zero second.

4. DISCUSSION

MSOLD signal from both multichannel dosimeter and InGaAs spectrometer shows increasing signal for increasing concentration for singlet oxygen peak-intensity allowing the calculation of a calibration curve relating MSOLD signal and photofrin concentration in liquid phantoms. The wavelength of singlet oxygen for the liquid phantom with 2% intralipid is about 1274 nm that is a little larger than the case without intralipid. The slightly shift of singlet oxygen peak could be due to the change of solvent. The MSOLD signal is increasing with Photofrin concentration in a manner of second order polynomial is likely due to the absorption by Photofrin that will need further investigation. 2% intralipid was used in the liquid phantom to simulate real tissue, significant suppression of MSOLD and a strong background signal was observed in the intralipid liquid phantom. The MSOLD signal from the in-vivo Photofrin-mediated PDT with 5 mg/kg PS concentration is comparable to measurement from the liquid phantom without intralipid. The MSOLD signal from low PS concentration is still noisy that is revealed by full range spectrum. More efforts should be devoted to increase the amplitude of singlet oxygen signal and improve the signal to noise ratio for clinical applications.

5. CONCLUSIONS

Both multichannel dosimeter system and an Avantes spectrometer were used to detect MSOLD signal for a series of Photofrin concentrations ranging from 5 mg/kg to 400 mg/kg. Calibration curves relating the intensity of singlet oxygen signal and Photofrin concentration was established for measurements by both detectors demonstrating that real-time MSOLD monitoring using either multichannel dosimeter system or InGaAs spectrometer is feasible to detect singlet oxygen signal in Photofrin mediated PDT. MSOLD signal is compared to SOED dosimetry and found to be proportional to each other. We also showed that the InGaAs spectrometer can detect singlet oxygen signal with Photofrin concentration as low as 5 mg/kg in Photofrin mediated PDT for both in-vitro phantom and in-vivo mice study.