Journal of Photochemistry
B: Biology, 5 (1990)
69 - 84
EXPERIMENTAL TESTS OF THE FEASIBILITY OF SINGLET OXYGEN LUMINESCENCE MONITORING in vivo DURING PHOTODYNAMIC THERAPY MICHAEL S. PATTERSON,
STEEN J. MADSEN and BRIAN C. WILSON
Hamilton Regional Cancer Centre and McMaster University, 711 Concession Street, Hamilton, Ontario, L&V 1 C3 (Canada) (Received April 28, 1989; accepted August 22, 1989)
Singlet oxygen, luminescence, lifetime, photodynamic therapy.
Summary Singlet oxygen (‘0,) is thought to be the cytotoxic agent in photodynamic therapy (PDT) with current photosensitizers. Direct monitoring of ‘02 concentration in uiuo would be a valuable tool in studying biological response. Attempts were made to measure ‘02 IR luminescence during PDT of cell suspensions and two murine tumour models using the photosensitizers Photofrin II and aluminium chlorosulphonated phthalocyanine. Instrumentation was virtually identical to that devised by Parker in the one positive report of in uiuo luminescence detection in the literature. Despite the fact that our treatments caused cell killing and tissue necrosis, we were unable to observe ‘0, emission under any conditions. We attribute this negative result to a reduction in ‘02 lifetime in the cellular environment. Quantitative calibration of our system allowed us to estimate that the singlet oxygen lifetime in tissue is less than 0.5 ps. Some technical improvements are suggested which would improve detector performance and perhaps make such measurements feasible .
1. Introduction Photodynamic therapy (PDT) is a form of local cancer treatment in which cell death is caused by photochemical reactions involving an exogenous photosensitizer. For photosensitizers in current use, such as haematoporphyrin derivative (HPD), these reactions are thought to be type II [l] in which singlet excited molecular oxygen is produced as follows hv %I--+
0 Elsevier Sequoia/Printed
in The Netherlands
T1 + 30,-S0+
where So, S1 and Ti are the ground, excited singlet and excited triplet states of the photosensitizer molecule, and 302 and ‘02 are the ground triplet and excited singlet states of oxygen. Once ‘02 is produced it may react with other molecules, be physically quenched, or return to the ground state with emission of a near-IR (1270 nm) photon. This latter transition occurs with low probability but, since its first observation [2, 31, it has been widely used to measure singlet oxygen yields and lifetimes in solution. The detection of this luminescence in uiuo during PDT would provide a means of directly monitoring the cytotoxic agent and would greatly aid the study of mechanisms of damage and response to treatment. To date only Parker 143 has reported the successful detection of ‘02 luminescence in uiuo. As discussed by Rodgers [ 51, other attempts have been made with systems of inadequate sensitivity or inadequate rejection of background fluorescence. The main purpose of this work was to make a detailed assessment of the feasibility of monitoring singlet oxygen luminescence during PDT using Parker’s technique and the best available equipment; a number of biological model systems were studied. Unfortunately, the results were negative, since, in experiments involving cell suspensions, two mouse tumour models and two photosensitizers, we were unable to detect luminescence in any instance, even though the PDT treatments caused considerable biological damage. There are a number of possible reasons for this failure, but we believe that the most likely explanation is a reduction in singlet oxygen lifetime in the cellular environment. Quantitative calibration of our system gave an estimate of the upper limit for the singlet oxygen lifetime of about 0.5 p.s. If luminescence is to be measured in uiuo, more sensitive detectors are required; we conclude this paper by describing techniques which may yet make this feasible.
2. Theory From eqns. (1) - (3) we can write three coupled differential equations describing the kinetics of the reactions
= MS,1 - ; [%I
_ % y- [%I -
@“, [T 1 D
where local concentrations are expressed in molecules per cubic centimetre and 4 (photons cm-2 s-l) is the local fluence rate, CJ(cm2) is the photo-
sensitizer ground state absorption cross-section, T, (s) is the photosensitizer excited singlet lifetime, @‘T is the triplet quantum yield (per absorbed photon), rT (s) is the photosensitizer triplet lifetime, CI?nis the singlet oxygen quantum yield (per absorbed photon) and rn (s) is the overall singlet oxygen lifetime. We have implicitly assumed that [S,] and [ 302] are independent of time (i.e. they are not depleted during the reaction). We will solve for [ ‘OJ under two conditions: (i) where the local fluence is a very short pulse and (ii) where the local fluence rate is sinusoidal. Let us first consider the pulsed case where the excitation is an infinitesimally short pulse of iV photons cm -* at t= 0 (i.e. 4 =A%(t)). Since 7,isof the order of nanoseconds we will assume that all the triplet photosensitizer molecules are created instantaneously so that [T,] = NcJ[S,]@++‘~T
A straightforward solution for [ ‘02] now follows [‘O,]
similar to the result reported by Parker and Stanbro . We also note that [S,] will be given by [S,] = Nu[S,]e-f17s
Let us now consider the case of sinusoidal excitation where 4 = N cos wt. The solutions can be directly derived by Fourier transformation of eqns. (8) and (9) so that NQ[%[email protected]
w(TD + TT) sin at}
(16) and ww, [%I
= NU[S()]T, cos
sin wt) for wr, < 1
The sinusoidally varying [‘O,] can be considered as the sum of a component in phase with the excitation (cos wt term) and a component 90” out of phase (sin wt term). The local singlet oxygen luminescent emission (photons cmV3 s-l) in the region of 1270 nm will be (13) where rL is the luminescence lifetime. In general, $ and hence [ ‘O,] will vary with position in the irradiated sample due to optical attenuation. The detector response and attenuation of the emitted IR radiation will also be
position dependent, as expressed by a factor R,,,,(r). The total observed luminescence signal (photons s-l) will therefore be given by the volume integral
1°21R [email protected]
We will also observe photosensitizer fluorescence emission in the same wavelength interval. This signal can be expressed as F 1270
where Cl270 is the fraction of fluorescence photons emitted in the wavelength region of interest and TF is the fluorescence lifetime. As will be shown experimentally, F1270 is usually comparable with, or even much greater than L1270, so that some method must be used to discriminate between the two types of emission. From in vitro experiments we expect that ?-, Q rT and r,4 rn. For pulsed excitation, therefore, eqns. (8) and (9) predict that, for a sufficiently long time after the pulse, the fluorescence background (including that due to tissue autofluorescence) will be negligible. Hence time-resolved emission measurements have the potential to detect singlet oxygen luminescence in the presence of a fluorescence background provided that the detector response time is short enough. This method was not used in this study but, as shown later, may offer advantages in some conditions. If the excitation light is chopped to form a square wave and the first harmonic of the emission is detected, this is equivalent to sinusoidal excitation. Equations (10) and (11) indicate that the fluorescence emission (including tissue autofluorescence) will be in phase with the excitation, whereas at least part of the luminescence signal will be 90” out of phase. Thus, as shown below, quadrature detection can be used to reject the fluorescence background to give a signal proportional only to the luminescence. The fraction of the luminescence signal in the quadrature component will depend on the chopping frequency, as indicated in eqn. (10). A final result of interest is the ratio of the amplitude of the quadrature component (Q1270) to the amplitude of the fluorescence signal in a wavelength band centred at 1200 nm (F,,,,), well outside the singlet oxygen luminescence peak. From eqns. (lo), (12), (14) and (15) we can write [email protected]
s NSoIR1270(4 v
where we have assumed that all quantities within the integral signs are potentially position dependent and that all quantities outside are spatially invariant. Since the absorption of light in tissue in the near IR is dominated by water, and since the absorption coefficient of water changes by less than 10% from 1200 to 1270 nm , it is reasonable to assume that R,,,,(r) = R,,,,(r) and that the integrals in eqns. (16) and (17) are, in fact, identical. Hence the ratio of these two amplitudes is
(18) and is also independent of the frequency response of the detector. This relationship will be used in deriving an upper limit for rn in tissue from the in uiuo measurements.
3. Materials and methods 3.1. The detection system The system configuration was similar to that described by Parker  and is illustrated in Fig. 1. The output of an argon-pumped dye laser (Cooper Medical, Stanford, CT, model 770) was chopped using an acousto-optic modulator (Intra Action Bellwood, IL, model ADM 40) and focused into an optical fibre (core, 400 pm). The output face of the fibre was imaged to a
SLIDING FILTER MECHANISM
Fig. 1. Schematic diagram of apparatus for the detection of singlet oxygen luminescence in vivo. The sliding filter mechanism facilitates the quadrature detection method and allows rough verification of the luminescence spectrum. For in vitro experiments a quartz cuvette (1 cm) was used and samples were irradiated through the side face.
beam (diameter, 5 mm) incident on the sample surface. For all experiments the incident power was switched between zero and approximately 50 mW so that the mean incident fluence rate was 130 mW cme2. IR light emitted by the sample was collected by an f/l lens and focused onto a liquid-nitrogen-cooled germanium detector (North Coast Optical, Santa Rosa, CA, model 817s). A 1000 nm long-pass filter (Oriel, Stratford, CT, model 58867, blocking 0.1%) was permanently fixed over the detector window and a sliding mechanism allowed one of three bandpass filters to be added to this. These were manufactured by Andover Corporation (Lawrence, MA) and had centre wavelengths of 1201 nm (bandpass 9.4 nm, blocking O.OOl%), 1272 nm (bandpass 18 nm, blocking 0.001%) and 1303 nm (bandpass 8.5 nm, blocking 0.001%). For consistency with the description above, these will be referred to as 1200, 1270 and 1300 nm filters. The output of the integral detector preamplifier was measured by a lock-in amplifier (Stanford Research Systems, Sunnyvale, CA, model SR510). The experiments consisted of measuring, at each of the three wavelengths, the signal in phase and the quadrature signal 90” out of phase with the excitation for a given chopping frequency. To do this, at each frequency the 1200 nm filter was positioned in front of the detector and the phase of the lock-in amplifier reference was adjusted to zero its output. The 1270 nm filter was then selected and the quadrature component was measured. As described above, this signal should be due to singlet oxygen luminescence only. The quadrature component was also measured at 1300 nm as a crude check on the luminescence spectrum, which should peak near 1270 nm and be negligible at both 1200 and 1300 nm . The phase was switched by 90” and the in-phase component was measured at the three wavelengths. Finally, with the 1200 nm filter in place, the phase was switched back 90” to check that the lock-in output remained at zero, thus demonstrating that the phase did not drift significantly over the time of approximately 1 min required to obtain the six readings at a given chopping frequency. The whole procedure was repeated at chopping frequencies in the range 1 - 100 kHz appropriate for each sample.
3.2. In vitro experiments All in vitro experiments were performed using a quartz cuvette (1 cm X 1 cm X 4 cm). For convenience, the samples were irradiated through the side face of the cuvette and the emission was detected through the front face. Experiments were performed with two photosensitizers: (i) Photofrin II, a purified haematoporphyrin derivative, was obtained from Photomedica (Raritan, NJ) or Quadralogic Technologies (Vancouver, BC); (ii) aluminium chlorosulphonated phthalocyanine (AlSPC) of mixed sulphonation was purchased from Porphyrin Products (Logan, UT). Attempts were made to measure the singlet oxygen luminescence for each sensitizer in methanol solution, aqueous solution and aqueous solutions with varying amounts of foetal calf serum (FCS).
Attempts were also made to measure the luminescence from suspensions of P388 cells derived from a lymphoid tumour in DBA-2 mice [ 91. The cells were grown in suspension in Opti-Mem and FCS (Gibco Laboratories, Grand Island, NY) and incubated at 37 “C for 24 h with either Photofrin II or AlSPC at a concentration of 50 E.cgml-‘. The cells were spun down, washed and then resuspended in fresh photosensitizer-free medium at concentrations ranging from lo6 to 10’ viable cells ml-‘. A search was made for the quadrature component as a function of chopping frequency in the same way as for the photosensitizer solutions. The total fluence incident on the face of the cuvette was approximately 180 J cm-’ at 624 nm for Photofrin II and 670 nm for AlSPC. For Photofrin II, 624 nm was used as this corresponds to the wavelength of maximum ‘02 yield in methanol and is also close to the 625 nm peak in the action spectrum reported by Versteeg et al. [lo]. The absorption peak for AlSPC is actually at 672 nm but 670 nm was the highest wavelength attainable with our laser. After irradiation, the percentage of viable cells was assessed by a Trypan Blue exclusion assay. 3.3. In viva experiments We attempted to measure singlet oxygen luminescence in two tumour systems: (i) the methylchloranthrene-induced fibrosarcoma (MIFS) tumour [ll] implanted in the footpad of C57/BLJ6 mice; (ii) the radiation-induced fibrosarcoma (RIF) tumour  implanted in the flank of C3H/HeJ mice. In both cases the tumours were allowed to grow until they attained a diameter of about 1 cm. For each tumour model six mice were injected intraperitoneally 24 h prior to treatment. Two were injected with Photofrin II (50 mg kg-’ body weight), two with AlSPC (50 mg kg-‘) and two with an equivalent volume of saline. Prior to irradiation, the detection system was aligned for maximum response using a cuvette containing a standard solution of Photofrin II in methanol. A fixed plate with a hole (8 mm) was used to mark the position of the cuvette and the mice were placed with the tumours in this aperture. During irradiation the mice were anaesthetized with sodium pentobarbitol (0.01 ml g-’ Somnotol, MTC Pharmaceuticals, Mississauga, ON). The excitation wavelength was 624 nm for the mice injected with Photofrin II and the corresponding control, and 670 nm for the mice injected with AlSPC and the corresponding control. The total incident fluence at the tumour surface for each treatment was approximately 200 J cm-*. The response of the tumours to treatment was visually monitored for several days following irradiation.
4. Results 4.1. Experiments in solution In Fig. 2 we show a plot of the ratio of the quadrature signal at 1270 nm to the fluorescence signal at 1200 nm (Q12,a/F,2,,) us. chopping frequency for 50 mg ml-’ Photofrin II in methanol. (For all the experiments
1.2 1.0 0.8
d 0.4 0.2
CHOPPING FREQUENCY (kHz)
Fig. 2. R e s u l t s f o r a s o l u t i o n o f 50 # g m1-1 P h o t o f r i n II in m e t h a n o l . T h e r a t i o o f t h e q u a d r a t u r e c o m p o n e n t at 1 2 7 0 n m (Q1270) t o t h e i n - p h a s e f l u o r e s c e n c e c o m p o n e n t at 1 2 0 0 n m (F120o) is p l o t t e d a g a i n s t c h o p p i n g f r e q u e n c y . T h e s m o o t h c u r v e is t h e b e s t l e a s t - s q u a r e s fit o f e q n . ( 1 8 ) t o t h e d a t a , f r o m w h i c h t h e p h o t o s e n s i t i z e r t r i p l e t l i f e t i m e a n d t h e singlet o x y g e n l i f e t i m e m a y b e d e t e r m i n e d . T h e s i g n a l - t o - n o i s e r a t i o at 15 k H z was 210:1.
r e p o r t e d here, the fluorescence signal per n a n o m e t r e interval at 1270 nm is a b o u t half th at at 1200 nm, in agreement with the results of Parker .) Also shown on this graph is the best least-squares fit of eqn. (18) to the data, where the free parameters o f the fit were the singlet oxygen lifetime TD, the triplet lifetime TT and an overall multiplicative constant. An excellent fit was obtained for rD = 9.0 + 0.3 pS and Tw 0.3 + 0.2 ~S, which are in good agreement with the literature values of 6 - 10.4 #s [13, 14] and 0.25/~s . The quadrature c o m p o n e n t also displayed the expect ed spectral behaviour, although the signal at 1300 nm was a b o u t 10% of t hat at 1270 nm rather than zero. Although the manufacturer's data showed insignificant overlap of the transmission band of the 1300 nm filter with the singlet oxygen emission spectrum, the large acceptance angle of the optics m ay have caused sufficient broadening o f the bandpass to explain this observation. Since this signal was thus conf i r m e d to be 102 luminescence, its magnitude can be used to calculate the sensitivity o f our system and compare it with the ex p ecte d value. Equation ( 1 0 ) w a s used to calculate  with the following additional data. (i) The singlet oxygen yield f or Photofrin II in m e t h a n o l was taken as 0.25, as r e p o r t e d by Keir e t al. . (ii) The fluence of exciting light in the solution decreased with d e p t h × as e -~ax where pa was measured as 0.074 cm -1 at 624 nm in a conventional spectrophotometer. (iii) The emi t t e d IR radiation was a t t e n u a t e d by 17% on passing through 0.5 cm o f m e t h a n o l . =
CHOPPING FREQUENCY (kHz)
Fig. 3. Plot of the absolute overall system sensitivity (measured in microvolts per molecule of 102) us. chopping frequency. Also shown is the system noise (/.JV HZ-‘/~) as a function of frequency.
The results of this calculation are shown in Fig. 3 as a plot of the overall system sensitivity (microvolts per molecule of ‘0,) us. chopping frequency. The shape of this curve is determined by the frequency response of the detector and preamplifier. The system noise measured directly using the lock-in amplifier is also shown. The best signal-to-noise ratio (SNR) is obtained at about 10 kHz, but the SNR stays within a factor of two of the peak value for frequencies in the range 2 - 100 kHz. If we assume that the luminescence is detectable at an SNR of unity, and the detection bandwidth is 1 Hz, then the system can detect 1.0 X 10’ molecules of ‘02 at a chopping frequency of 10 kHz. Using the manufacturer’s data for the noise equivalent power (NEP) of the detector and the additional information that the geometrical collection efficiency is 0.022, transmission through the optical components is 0.32 and the singlet oxygen luminescence lifetime rL is 4 s [ 41, we predict a detection limit of 0.7 X lo* molecules; considering the many approximations, this value is in good agreement with the observed value. This comparison demonstrates that the system is detector noise limited. Having verified the operation of the system, we then attempted to observe the singlet oxygen luminescence from a solution of 50 pg ml-’ Photofrin II in phosphate-buffered saline (PBS) (pH 7.2). No quadrature signal could be observed. Using the data of Fig. 3 and the literature values rT = 2.4 /Js and rn = 3.2 /AS, we can place an upper limit on the quantum yield of ‘0, from Photofrin II in aqueous solution. Our value of 0.0030 is
six times less than the yield reported by Keir et al. [ 151. This large reduction in @n has been observed by a number of investigators [17,18] and has been attributed to aggregation of the photosensitizer molecules in aqueous solution. This phenomenon is also known to reduce visible fluorescence but, interestingly, the in-phase fluorescence components at 1200, 1270 and 1300 nm were all comparable with those observed in the methanol solutions. Similar experiments were performed for 2 pg ml-’ AlSPC in methanol and PBS. The ratio Q12,0/F1200is plotted us. chopping frequency in Fig. 4, together with the best least-squares fits to eqn. (18). For methanol we obtained a value of rn = 8.5 + 0.5 ps, which is close to the literature values [13,14] and the value of 9.0 ~.lsobtained with Photofrin II. Our estimate of the triplet lifetime in methanol was 0.3 f 0.3 ps. A literature search failed to yield a value of rr for comparison, but our result is the same as that reported by Parker  for Photofrin II. For AlSPC, a measurable luminescence signal was obtained in aqueous solution although it was considerably weaker than in methanol. (The fluorescence emission F,,,, was comparable for AlSPC in methanol and water solutions.) Our analysis gave a value of rn = 3.2 + 0.2 E.CS for water, which is in agreement with the value reported by Parker and Stanbro  and close to the 4.2 /.LSobserved by Rodgers . Again, there is no literature report of rr for comparison, but our value of 2.4 + 0.5 /.u is comparable with that found for other photosensitizers in air-saturated solutions . Since the in viuo environment of the photosensitizer is rich in proteins, and because it has been reported that rn is reduced in protein solutions , we also measured the luminescence from aqueous solutions of AISPC to which varying amounts of foetal calf serum had been added. (No luminescence was observed from aqueous solutions of Photofrin II to which FCS
CHOPPING FREQUENCY (kHz)
Fig. 4. Ratio Q1270/F1200 for solutions of 2 pg ml-’ AlSPC in methanol (open symbols) and saline (filled symbols). The smooth curves are the best least-squares fits of eqn. (18) to the data. For the methanol experiment, the signal-to-noise ratio at 15 kHz was 130:1, and for the water experiment it was 9:l at 30 kHz.
had been added.) As the concentration of FCS was increased, we noted a reduction in the observed luminescence and a shift in the peak of emission to lower frequencies. The reduced emission is probably due to a reduction in rn as expected. The shift in the peak implies an increase in rT which has also been observed by Parker and Stanbro , and attributed to oxygen consumption exceeding oxygen diffusion. No luminescence could be observed when the FCS concentration exceeded 7% by volume. At this concentration we estimate that rT had increased from 2.4 to about 40 ~.tsand that rn had been reduced from 3.2 to approximately 1.5 ps. There is a large uncertainty associated with these estimates due to the poor signal-to-noise ratio of the measurements. 4.2. Cell suspensions The results for the in vitro cell suspension experiments can be summarized very simply by stating that under none of the conditions used could a quadrature component be observed for either photosensitizer, even when the treatment resulted in a 100% cell kill. The possible reasons for this negative result are discussed below. 4.3. In viva experiments In a total of 12 mice (eight photosensitized, four control), no singlet oxygen luminescence was observed for either photosensitizer. Despite this, all of the photosensitized mice showed marked tissue necrosis in the treated area in the days following the irradiation received during the measurements, whereas no change other than tumour progression was evident in the control mice. This tumour response is expected from the photosensitizer and light doses used (50 mg kg-‘, 200 J cms2). Photosensitizer IR fluorescence was readily observed, and the magnitude of this signal was, in all cases, about 100 I.~V with the 1200 nm filter in place. It should be noted that a small quadrature component, typically around 1 PV and comparable with the noise in amplitude, was observed at 1270 nm for all tumours, whether photosensitizer was present or not. However, since an equally large signal was seen at 1300 nm, this component cannot be due to singlet oxygen, and may be caused by fluorescence from the optics excited by scattered light. In earlier experiments reported by us [ 201 in which a different long-pass filter was used, this effect was about an order of magnitude larger and could be observed when any strong light-scattering sample was used. 5. Discussion From the results of the experiments for photosensitizers in solution it can be concluded that the experimental system operated as expected from the equipment specifications and was capable of detecting luminescence from singlet oxygen even when generated in an aqueous environment. There are a number of possible reasons why singlet oxygen luminescence was not detected from tumours in uiuo or in cell suspensions.
(i) The singlet oxygen yield in viuo may be much lower than in solution and the biological damage in photodynamic therapy may be due to some other mechanism. While possible, this alternative is unlikely, since Moan et al.  have shown that ‘0, is generated during PDT of cell suspensions with haematoporphyrin. These workers also showed an enhancement of killing when cells were suspended in D,O rather than H,O - an observation consistent with type II reactions. Moreover, Weishaupt et al.  noted that photodynamic inactivation of cells by PDT with HPD was completely inhibited when a singlet oxygen trap was present. (ii) As suggested by Moan and Sommer , a large fraction of the tumour volume may be hypoxic due to poor vascularization. The yield of singlet oxygen in such regions would be low and the total luminescence signal would be much less than expected. The blood vessels themselves could still be damaged and this would cause the observed tumour necrosis as a result of induced ischaemia . To test this hypothesis we attempted to observe luminescence in uiuo from surgically exposed livers of two normal rats injected with 50 mg kg-i AlSPC. Despite the fact that the concentration of AlSPC should be very high in the liver  so that most of the light is absorbed by the photosensitizer, and that the organ is well oxygenated, no luminescence was observed. (The observed fluorescence signal was about five times that measured for the turnours.) We conclude that this explanation is also unlikely. (iii) The most probable explanation is that the luminescence signal is quenched by a reduction in singlet oxygen lifetime in the cellular environment. We can estimate an upper limit to the singlet oxygen lifetime in viuo in the following manner. In eqn. (18), the ratio of the quadrature component of luminescence (Q1& to the m-phase fluorescence at 1200 nm (F,,& is given by the product of two terms. The first term depends on the singlet oxygen yield @n, the luminescence lifetime rL, the fluorescence spectrum C12-,0and the quantum yield of fluorescence 7&r. The second term depends on the chopping frequency o, the triplet lifetime rr and the singlet oxygen lifetime are known for the experiments in soluTD. S&e u9 TD7 TT and &d~1200 tion, these data can be used to calculate the unknown first term. If we assume that this term has approximately the same value in uiuo as in solution, then eqn. (18) can be used to calculate an upper limit for rn. (It is possible to proceed without this assumption using only eqn. (16), but this requires knowledge of the optical properties of the tumour and the uptake of photosensitizer. It was felt that the use of the internal “fluorescence standard” involved fewer assumptions.) We first set Qi2,,, equal to the measured electronic noise (cf. Fig. 3, assuming Q12,0 would be detectable at a SNR of unity) and F,,, e qual to the observed fluorescence signal less the fluorescence observed from an uninjected control animal. (This latter factor compensates for the tissue autofluorescence, which is typically 20% of the photosensitizer fluorescence.) Equation (13) reduces to a quadratic equation in rn which we can now solve if the know rr. Parker  has
suggested a value of 9.6 I.CS corresponding to an aqueous environment with physiological oxygenation. Truscott et al.  have reported a value of rT = 7.7 ps for HPD in suspensions of fibroblasts measured by transient absorption spectroscopy. Similar experiments by Firey et al.  suggest a triplet lifetime of 8 I.CSfor zinc phthalocyanine in mouse myeloma cells. We have calculated rn for values of rT = 1 ps and 7T = 10 ps as the actual V6lUe is probably in this range. The results of these calculations are shown in Fig. 5 where, for each chopping frequency and tissue type, the upper limit of Tn is represented by a bar. The lower limit of the bar is based on rr = 1 E.CS and the upper end on rr = 10 /.u and the range of results calculated for both photosensitizers is shown. It can be seen that for rr = 10 j.~s,the measurements at 10 kHz provide most information about Tn. As there appears to be little variation among the three tissue types, we can conclude that, in general, singlet oxygen lifetimes in uiuo are probably less than 0.5 ps. Based on the data of Matheson et al. [ 191 for chemical reaction rates, Moan et al.  have estimated that the lifetime of singlet oxygen in cells should be less than 1 ps. This estimate is compatible with the lack of an observable singlet oxygen luminescence signal and shows that more sensitive detection techniques will be necessary if this emission is to be measured in uiuo. It should be emphasized that there is no need for faster detectors or operation at higher chopping frequencies. As shown by Parker [ 41, when rn Q rr, the quadrature component is at a maximum at a chopping frequency equal to 1/2xrr. Reduction in rn causes a reduction in the magnitude of emission, but the
Fig. 5. Estimates of the upper limit of singlet oxygen lifetimes us. chopping frequency for two murine tumour models and normal rat liver tissue. The estimates are based on the observed photosensitizer fluorescence signal and the lack of a detectable luminescence signal. The large range shown is the result of uncertainty in the triplet lifetime in tissue. The lower end of the range corresponds to a calculation based on a triplet lifetime of 1 j.rs and the upper end to a triplet lifetime of 10 1-1s.
optimum frequency is still 10 - 20 kHz. We will conclude by considering possible future approaches to this problem of limited sensitivity. Phosphors have been developed  which, after a “charging” exposure to visible or UV light, respond to the absorption of near-IR light by the emission of photons at visible wavelengths, which can be detected with a photomultiplier tube operated in either the current or photon-counting mode. The quantum efficiency of conversion from IR to visible is about 0.02, so that an overall system quantum efficiency of approximately 10e3 is attainable. Using a cooled photomultiplier in photon-counting mode, the ultimate detection limit is determined by statistical fluctuation in the detected photon number. In this case, an NEP of about 2 X lo-l6 W is possible in principle, giving an improvement of two orders of magnitude over our present detector. In practice, the detection limit may be determined by visible light emitted from the phosphor due to thermal excitation, and with current phosphors it appears that quadrature methods using such detectors offer no advantage over the present system. However, if the excitation light is delivered in pulses short compared with rr, this limitation can be overcome. The phosphorescence lifetime is only about 5 ns , so that the ‘0, luminescence photons are emitted for only a few microseconds after the excitation pulse. Hence the duty cycle of detection (and hence the dark counts) can be considerably reduced. For example, by cooling the phosphor and photomultiplier tube, the dark count rate with current phosphors can be reduced to 300 counts s-l . If we excite the sample with pulses at 100 Hz, and count for 10 ~.lsafter each pulse, the detected dark count rate is 0.3 counts s-i. At useful signal count rates this will be insignificant. If the same average power is incident on the sample as was used in the experiments reported here (25 mW), we require pulses of 0.4 mJ. We calculate that this pulse energy is low enough to avoid significant ground state depletion of the photosensitizer assuming typical clinical concentrations. Calculations using eqns. (8) and (10) suggest that IO2 should be detectable in viuo using such a pulsed system, even if the lifetime of ‘02 is as short as 10 ns. We are currently assembling a system to test this hypothesis.
6. Conclusions We have demonstrated that a sensitive system designed specifically for the detection of singlet oxygen luminescence in uiuo is unable to detect this emission from murine tumours or cell suspensions during photodynamic therapy. We were unable to reproduce the results of Parker [ 41 who reported the detection of luminescence in one mouse tumour under similar conditions and with virtually identical instrumentation. Parker used a different tumour model and an unanaesthetized mouse; however, since we observed a large tumour response from the treatment, we do not believe that these factors can explain the different results, so that this discrepancy remains unresolved.
We ascribe the lack of observed luminescence in our experiments to a shortened lifetime for singlet oxygen in uivo. Quantitative calibration of our system leads to an estimate for this lifetime of less than 0.5 ps, which is consistent with the estimate of less than 1 E.CS made by Moan et al. . Finally, we suggest that a new strategy, based on pulsed excitation and phosphor-photomultiplier detectors, may offer a considerable improvement in the detection of weak luminescence. This method of monitoring singlet oxygen during PDT may yet prove to be feasible.
Acknowledgments This work was supported by the National Cancer Institute of Canada. The authors are grateful to Dr. John Parker of the Applied Physics Laboratory, Johns Hopkins University, for several helpful discussions.
References 1 C. S. Foote, Mechanistic characterization of photosensitized reactions, in Photosensitization - Molecular, Cellular and Medical Aspects, Springer, Berlin, 1988, pp. 125 - 144. 2 A. A. Krasnovskii, Photosensitized luminescence of singlet oxygen in solution, Biofizika, 21 (1976) 748 - 749. 3 A. N. Khan and M. Kasha, Direct spectroscopic observation of singlet oxygen emission at 1268 nm excited by sensitizing dyes of biological interest in liquid solution, Proc. Natl. Acad. Sci., India, 76 (1979) 6047 - 6049. 4 J. G. Parker, Optical monitoring of singlet oxygen during photodynamic treatment of tumors, IEEE Circuits and Devices Magazine, January (1987) 10 - 21. 5 M. A. J. Rodgers, On the problems involved in detecting luminescence from singlet oxygen in biological specimens, J. Photo&em. Photobiol., B: Biol., 1 (1988) 371 373. 6 J. G. Parker and W. D. Stanbro, Dependence of photosensitized singlet oxygen production on porphyrin structure and solvent, in D. R. Doiron and C. J. Gomer (eds.), Porphyrin Localization and Treatment of Tumors, Alan R. Liss, New York, 1984, pp. 259 - 284. 7 W. G. Driscoll and W. Vaughan (eds.), Handbook of Optics, McGraw-Hill, New York, 1978, pp. 15 - 28. 8 S. H. Whitlow and F. D. Findlay, Single and double electronic transitions in molecular oxygen, Can. J. Chem., 45 (1966) 2087 - 2091. 9 American Type Culture Collection Catalogue of Strains, II, 1981. 10 J. Versteeg, W. Star, W. van Putten and H. Marijnissen, Wavelength dependence of HPD-PDT effects on rat ears, Lasers Med. Sci., 3 (Abstracts Issue 210) (1988). 11 J. Varani, W. Orr and P. A. Ward, A comparison of the migration patterns of normal and malignant cells in two assay systems, Am. J. Pathol., 90 (1978) 159 - 171. 12 P. R. Twentyman, J. M. Brown, J. W. Gray, A. J. Frank0 and R. F. Kallman, A new mouse tumor model system (RIF-1) for comparison of end-point studies, J. Natl. Cancer Inst., 64 (1980) 595 - 604. 13 J. G. Parker and W. D. Stanbro, Energy transfer processes accompanying laser excitation of hematoporphyrin in various solvents, Johns Hopkins APL Tech. Digest, 2 (1981) 196 - 199.
84 14 M. A. J. Rodgers, Activated oxygen, in R. V. Bensasson, G. Jori, E. J. Land and T. G. Truscott (eds.), Primary Photo-Processes in Biology and Medicine, Plenum Press, New York, 1985, pp. 181 - 195. 15 W. F. Keir, E. J. Land, A. H. MacLennan, D. J. McGarvey and T. G. Truscott, Pulsed radiation studies of photodynamic sensitizers: the nature of DHE, Photochem. Photobiol., 46 (1987) 587 - 589. 16 H. Morita and S. Nagakura, Hydrogen-bonded O-H and O-D overtone bands and potential energy curve of methanol, J. Mol. Spectrosc., 49 (1974) 401 - 413. 1’7 R. Redmond, E. J. Land and T. G. Truscott, A comparison of the photophysical properties of porphyrins used in cancer phototherapy, in R. V. Bensasson, G. Jori, E. J. Land and T. G. Truscott (eds.), Primary Photo-Processes in Biology and Medicine, Plenum Press, New York, 1985, pp. 181 - 195. 18 J. Moan, The photochemical yield of singlet oxygen from porphyrins in different states of aggregation, Photochem. Photobiol., 39 (1984) 445 - 449. 19 I. B. C. Matheson, R. D. Etheridge, N. R. Kratowich and J. Lee, The quenching of singlet oxygen by amino acids and proteins, Photochem. Photobiol., 21 (1975) 165 171. 20 M. S. Patterson, B. C. Wilson and S. J. Madsen, Optical methods for the measurement of photosensitizer concentration and singlet oxygen production in photodynamic therapy, Lasers Med. Sci., 3 (Abstracts Issue 207) (1988). 21 J. Moan, E. 0. Pettersen and T. Christensen, The mechanism of photodynamic inactivation of human cells in vitro in the presence of haematoporphyrin, Br. J. Cancer, 39 (1979) 398 - 407. 22 K. R. Weishaupt, C. J. Gomer and T. J. Dougherty, Identification of singlet oxygen as the cytotoxic agent in photo-inactivation of a murine tumor, Cancer Res., 36 (1976) 2325 - 2329. 23 J. Moan and S. Sommer, Oxygen dependence of the photosensitizing effect of hematoporphyrin derivative in NHIK 3025 cells, Cancer Res., 45 (1985) 1608 - 1610. 24 B. W. Henderson, S. M. Waldow, T. S. Mang, W. R. Potter, R. B. Malone and T. J. Dougherty, Tumor destruction and kinetics of tumor cell death in two experimental mouse tumors following photodynamic therapy, Cancer Res., 45 (1985) 572 - 576. 25 S. G. Bown, C. J. Tralau, P. D. Coleridge Smith, D. Akdemir and T. J. Wieman, Photodynamic therapy with porphyrin and phthalocyanine sensitization - quantitative studies in normal rat liver, Br. J. Cancer, 54 (1986) 43 - 52. 26 T. G. Truscott, A. J. McLean, A. M. R. Phillips and W. S. Foulds, Detection of haematoporphyrin derivative and haematoporphyrin excited states in cell environments, Cancer Lett., 41 (1988) 31 - 35. 27 P. A. Firey, T. W. Jones, G. Jori and M. A. J. Rodgers, Photoexcitation of zinc phthalocyanine in mouse myeloma cells: the observation of triplet states but not of singlet oxygen, Photochem. Photobiol., 48 (1988) 357 - 360. 28 Quantex Photonic Products, Rockville, MD, personal communication, 1989.