Fluorescence Lifetime Imaging of Experimental Tumors in Hematoporphyrin Derivative-Sensitized Mice

Photochemistry and Photobiology, 1997, 66(2): 229-236

Fluorescence Lifetime Imaging of Experimental Tumors in Hematoporphyrin Derivative-Sensitized Mice

Rinaldo Cubeddu*’, Gianfranco Canti2, Antonio Pifferil, Paola Taronil and Gianluca Valentinil ’Dipartimento di Fisica, Politecnico di Milano, Milano, Italy and 2Dipartimento di Farmacologia, Universita di Milano, Milano, Italy

Received 27 December 1996; accepted 28 April 1997

ABSTRACT

Tumor detection has been carried out in mice sensitized with hematoporphyrin derivative (HpD) by measuring the spatial distribution of the fluorescence lifetime of the exogenous compound. This result has been achieved using a time-gated video camera and a suitable mathematical processing that led to the so-called “lifetime images.” Extensive experimental tests have been performed on mice bearing the MS-2 fibrosarcoma or the L1210 leukemia. Lifetime images of mice show that the fluorescence decay of HpD is appreciably slower in the tumor than in healthy tissues nearby, allowing a reliable detection of the neoplasia. The lengthening of the lifetime in tumors depends little on the drug dose, which in our experiments could be lowered down to 0.1 mgkg body weight, still allowing a definite tumor detection. In order to ascertain the results achieved with the imaging apparatus, high-resolution spectroscopy, based on a time-cor- related single photon counting system, has also been performed to measure the fluorescence lifetime of the drug inside the tumor and outside. The outcomes obtained with two techniques are in good agreement.

INTRODUCTION

The development of a reliable system for early cancer diagnosis is a major challenge for the scientific community. For such parts of the human body that can be efficiently reached by light, directly or by means of an endoscope, optical techniques are very promising because they are effective, noninvasive and relatively low cost. Among them, fluorescence has been used for many years for medical diagnosis. Recently, it has received renewed attention because high-sensitivity and high-speed video cameras have become available.

Unfortunately, a broadband fluorescence emission is in general characteristic of both cancerous and noncancerous tissues. Therefore, the mere presence of a fluorescence signal does not provide any diagnostic aid. Nevertheless, high-per-

*To whom correspondence should be addressed at: C.E.Q.S.E.- C.N.R., Dipartimento di Fisica, Politecnico di Milano, Piazza L. da Vinci 32, 20133 Milano, Italy. Fax: +39 2 23996126; e-mail: [email protected]

0 1997 American Society for Photobiology 0031-8655/97 $5.00+0.00

formance video cameras and fast image grabbers allow the acquisition and processing of high-quality fluorescence images, leading to spatial maps of some fluorescence-related properties (intensity at a specific wavelength, intensity ratio at different wavelengths, time decay and so on). Such maps, in turn, can be plotted on a video monitor in gray shades or as pseudocolor images, which can show the localization of the disease in a form that physicians are more familiar with.

In order to find a selectivity criterion to discriminate the cancerous tissue from the healthy one, two main approaches have been followed: either the endogenous fluorescence of the tissue or the exogenous fluorescence emission of a tumor-specific marker have been considered.

The first approach relies on natural fluorophores that are always present in tissues and on the severe morphological and biochemical alteration that takes place during tumor growth. In fact, it has been demonstrated that in many situations the fluorescence of tumors differs from the one of healthy tissues both in intensity and spectral shape ( 1 4 ) .

On the other hand, for several years clinicians have been using photosensitizing drugs in combination with a suitable excitation light for the photodynamic therapy (PDT)? of tumors (5,6). Photodynamic therapy relies on the selective localization of the drug in the tumor and on its cytotoxic effect upon activation by red or near infrared light (7,s). Because photosensitizers are usually fluorescent, they can also be used as diagnostic markers (9-1 l), taking advantage of their compatibility for human administration. This is certainly true for the purified version of the hematoporphyrin derivative (HpD): Photofrina, porfimer sodium, which has been recently approved in some countries for the treatment of specific types of malignant tumors.

Whichever approach is chosen for fluorescence imaging (exogenous or endogenous fluorescence), the selectivity cri- tenon can be found either in the spectral domain or in the time domain.

In the spectral domain, double wavelength excitation or acquisition techniques followed by image subtraction can be used to detect the neoplasia (12-16).

Alternatively, some research groups, including the authors, have been following the time-domain approach (17- 19). Actually, the fluorescence decay of HpD in hydrophobic

?Abbreviations: bw, Body weight; CCD, charge-coupled device; HpD, hematoporphyrin derivative; PDT, photodynamic therapy.

229

230 Rinaldo Cubeddu et a/.

environments is typically multiexponential, with a long-lived component having a time constant of about 15 ns (20), whereas the natural fluorescence mainly extinguishes within 5 ns. Therefore, using a nanosecond-pulsed excitation and acquiring the fluorescence signal after a 20-30 ns delay, it is mainly the long-lived emission of the sensitizer that con- tributes to the image. Working with experimental tumors injected intradermally on the lower dorsum of mice, the authors proved that this technique effectively suppresses the natural fluorescence background and allows reliable tumor detection under proper conditions of drug dose and uptake time (21). The optimal drug dose was found to be 5-10 mgkg body weight (bw), which is only 2.5-5 times lower than the therapeutic dose for this tumor model (22). A noninvasive diagnostic system to be used in humans would re- quire a significant reduction in the drug dose because cuta- neous sensitization, which can be a severe porphyrin side effect (23), is acceptable for the treatment of a definite malignant tumor but cannot be afforded for a mere diagnostic trial. Unfortunately, the HpD fluorescence contrast between tumor and healthy tissues disappears at very low dose, at least in our tumor model. Moreover, fluorescence intensity images are affected by eventual unevenness in the excitation beam and the negative effects of the noise become dominant as the drug dose is reduced.

On the other hand, the photophysical properties of HpD, in particular the fluorescence time decay, are known to depend on the environment (24,25). Accordingly, the study of the HpD average fluorescence decay time in tumor-bearing animals could lead to an alternative way to detect the neoplasia. Pursuing this aim, we recently set up an innovative system to measure the spatial distribution of the fluorescence decay time in animals sensitized with HpD (26). Using such a system, the tumors could be detected in mice bearing the MS-2 fibrosarcoma and administered with a sensitizer dose wen 250 times lower than the therapeutic one.

This work is intended to better investigate the diagnostic potential of that technique on two different tumor models and to compare the results of the imaging system to high 1 ime-resolution spectroscopic measurements performed on the same samples.

MATERIALS AND METHODS Animals and tumors. All experiments were carried out in accordance with protocols approved by the local experimental animal welfare committee and conformed to national regulations for animal exper- imentation.

The MS-2 fibrosarcoma was originally induced by the Moloney rnurine sarcoma virus and maintained by weekly intramuscular pas- sage of tumor cell homogenate into the right hind leg of inbred €3alb/C mice (Charles River, Calco, Italy). The chemically induced 1,121 0 lymphoid leukemia was maintained by weekly intraperitoneal injection in inbred Balb/C mice (Charles River, Calco, Italy).

For both the fibrosarcoma and the leukemia, 4-5 days before the measurements, loh tumor cells were injected intradermally in the lower dorsum of hybrid DBM2 X Balb/C mice (Charles River, Cal- co, Italy). The fluorescence studies were performed when the tumors had reached a mean diameter of 6-8 mm. At the measurement time, the tumors were free of evident necrosis.

Chemicals. Hematoporphyrin derivative was kindly provided by Prof. T. G . Truscott (Department of Chemistry, Keele University, Keele, UK). For the experiments, it was dissolved in saline at a ccincentration of 5 mg/mL.

Time-resolved Juorescence imaging. Mice bearing the MS-2 fi-

01 Gate and Delay

Figure 1. Set-up for fluorescence lifetime imaging.

brosarcoma were injected intraperitoneally with the following HpD doses: 0.1, 0.25, 1 or 2.5 mgkg bw. Six mice for each dose were examined 12 h after the drug injection. The mice bearing the L1210 leukemia were handled according to the same protocol, but only two drug doses (0.25 and 2.5 mgkg bw) were considered. Control measurements were carried out by recording the natural fluorescence of tumor-bearing mice (six animals per tumor type) not injected with any drug.

The experiment was performed twice. The experimental set-up used for this study is displayed in Fig. 1 and is similar to the one already described in detail in Cubeddu et al. (27). The excitation source was a 405 nm dye (DPS, Exciton, Dayton, OH) laser pumped by a subnanosecond nitrogen laser (LN203C, Laser Photonics, Or- lando, FL) working at 50 Hz. The laser beam was coupled to an optical fiber. For the measurements, the distal end of the fiber was placed at = I 5 cm from the animal; in such a way a region of ~ 3 . 5 cm in diameter was illuminated with an average irradiance of 75 FW/cm2. Care was taken to preserve the irradiance constant all through the experiment. The mice were sacrificed immediately before the image acquisition to prevent movements, which may ham- per the measurements. All the images were acquired within 1 min after the death of the animal. In some cases, for both sensitized and control mice, the images were acquired when the animals were alive and anesthetized with ethyl urethane. Images were acquired by means of an intensified charge-coupled device (CCD) video camera (ICCD225, Photek, St. Leonards-on-Sea, England), whose light am- plifier can be gated for times as short as 5 ns. The width and the leading edge of the acquisition window were set by means of a delay/pulse generator (DG535, Stanford Research Systems, Sunny- vale, CA) triggered by the laser pulses. An orange cut-off filter (Wratten no. 22, Kodak, Rochester, NY, 19% transmisssion at 560 nm and 89% at 610 nm) was placed in front of the video camera to cut the scattered laser light and short-wavelength visible fluorescence. Three images were acquired from each mouse at delays of 10, 20 and 30 ns after the excitation pulses. The acquisition gate was always 100 ns wide. The gain of the image intensifier was set so as to fully exploit the dynamic range of the CCD camera and was carefully monitored. Image acquisition and on-line processing were performed by means of two boards (DT2861 DT2858, Data Translation, Marlboro, MA) plugged into a personal computer. In particular, on-line frame averaging allowed us to reduce the salt- and-pepper noise introduced by the light intensifier. Each image was digitized on 256 gray shades and converted to a 5 12 X 5 12 X 8 bits matrix. Then, the images were recorded on the computer hard disk.

A mathematics software (Matlab, Mathworks, Natick, MA) run- ning on a Risc computer (Alpha 7310, DEC, Maynard, MA) was used off-line to calculate the spatial distribution of the decay time and to display the results in gray shade or pseudocolor images. Be- fore the mathematical processing, each 512 X 512 pixel matrix was

Photochemistry and Photobiology, 1997, 66(2) 231

Figure 2. Lifetime (a) and 20 ns delayed (b) images of a tumor on the lower dorsum of a mouse bearing the MS-2 fibrosarcoma and sensitized with 2.5 mgk-g bw of HpD. The rectangles show the regions where the average value T was calculated. Correspondence between gray levels and lifetimes in nanoseconds (a) and between gray levels and fluorescence intensity in counts (b) is reported on the right.

shrunk by a factor of 16 in order to reduce the computing effort. The whole processing required about 2 s.

The theoretical principles of this technique can be briefly summarized as follows. We are interested in the long-lived component of HpD; therefore, we acquire the fluorescence images after three delays greater than or equal to 10 ns, which is when the natural fluorescence of the tissue and the short-lived components of HpD are mainly extinguished. Consequently, the time behavior of the fluorescence in each point of the sample can be roughly approximated by a monoexponential function. Using the standard notation where the subscripts i and j identify the image pixels, we get three matrices Fl,,(dk), whose elements depend on as many decay times T,,, as the pixel number:

F,,,(dk) C~,.,I,.,(O)exP (- dk/T,,,) (k = 1 , 2, 3 ) where dk is the acquisition delay (ie. 10, 20 or 30 ns), I,,,(O) is the matrix of the fluorescence intensity of the long-lived component of HpD immediately after the excitation pulse and C is a constant that depends on the sensitivity of the video camera and on the gathering power of the collecting optics.

By means of a simple linear regression, the three previous equa- tions lead to the matrix T,~ , , which is the result of our measurements and can be plotted in a gray shade or pseudocolor image.

Calibration of the method for the evaluation of fluorescence decay times was performed with dyes of known lifetimes, showing an ac- curacy better than 1 ns (28).

To appraise the results quantitatively, for each image the average fluorescence decay time T was calculated over two rectangles corresponding, respectively, to the tumor and to a portion of healthy tissue nearby (e .g . see Figs. 2-6), and the ratio R, of the two values was taken. The average value of R, was then evaluated over all mice administered with the same drug dose and over all control (ie. not sensitized) animals. A one-tailed unpaired Student's t-test was em- ployed to determine whether the differences in the R, values for different groups were statistically significant.

Time-resolved fluorescence spectroscopy. Time-resolved fluorescence spectroscopy was carried out immediately after the imaging study on the same animals, bearing either the fibrosarcoma or the leukemia and sensitized with 0.25 mg/kg bw of HpD.

Figure 3. Lifetime (a) and 20 ns delayed (b) images of a tumor on the lower dorsum of a mouse bearing the MS-2 fibrosarcoma and sensitized with 0.25 mgkg bw of HpD.

The measurements were performed with a home-built computer- controlled apparatus, described in detail in Cubeddu et al. (29). A mode-locked krypton ion laser (1-200-K, Coherent, Palo Alto, CA), tuned at 413 nm was used as the excitation source. The laser light was coupled to a 600 pm quartz fiber (SFS600, Fiberguide, Stirling, NJ) for the excitation. The laser irradiance on the animal skin never exceeded 1.5 mW/cm* to prevent the bleaching of the drug. Atten- tion was paid to illuminate and collect from the same region, limited to the tumor area in a set of measurements and to the surrounding healthy tissue in a separate set. Two measurements were performed on each area. The axes of the excitation fiber and the collection bundle were fixed at an angle of about 15" to each other. The fluorescence light was collected with a 2 mm quartz fiber bundle (K0218, EG&G PAR, Princeton, NJ), dispersed through a mono- chromator set at 630 nm and detected by a double microchannel plate photomultiplier (R1.564-01, Hamamatsu, Hamamatsu City, Ja- pan). The signal was handled by an electronic chain for time-cor- related single photon counting. The above-described set-up has an overall time resolution of about 90 ps and a spectral resolution of 2 nm.

For the evaluation of the fluorescence lifetimes, linear and nonlinear curve-fit procedures (30) were used to fit a multiexponential function to the experimental data. The quality of the fi t was judged evaluating the reduced x2, the weighted residuals and their autocor- relation function.

RESULTS The lifetime images indicate that the decay time of the long- lived component of the HpD fluorescence in correspondence with the tumor is systematically longer than in the surrounding healthy tissue, for both the MS-2 fibrosarcoma and the L 1210 leukemia. This result can be immediately appreciated by observing Figs. 2a, 3a, 4 and 5, which show the plot of T,,, for selected mice, bearing either of the tumor types and treated with different drug doses (either 2.5 or 0.25 mgkg bw). The figures have been obtained by expanding the range of measured values for T,., on a full 64 level gray scale. In such a way, black pixels correspond to the shortest fluorescence lifetime, while white ones refer to the longest lifetime.


Figure 4. Lifetime image of a tumor on the lower dorsum of a mouse bearing the L1210 leukemia and sensitized with 2.5 mgkg bw of HpD.

In order to keep, as far as possible, the representation of the experimental results unaltered, gray scale images have been considered, instead of pseudocolor ones, even though the latter ones are more effective for tumor detection. The stri- ation visible in the images is due to the razor blades used to shave the animals. Figure 2a shows the dorsum of a mouse bearing an MS-2 tumor and sensitized with 2.5 mgkg bw of HpD. The tumor location is clearly visible and the borders of the lesion are sharply defined. For comparison, Fig. 2b shows a fluorescence intensity image of the same mouse, acquired after a delay of 20 ns with respect to the excitation pulses. Even in this image the tumor can be easily identified. For a better comparison between the two techniques, the contrast between neoplasia and healthy tissue was evaluated for both image types, according to the following procedure. The average fluorescence decay time T or the average intensity I, in Fig. 2a and b respectively, were calculated in two rectangles (shown in the images) corresponding to the tumor and to a portion of normal tissue nearby, and the ratios R, and R, were taken. For the mouse shown in Fig. 2 the values of R, and RI are 1.23 and 2.05, respectively.

The images reported in Fig. 3 refer again to the fibrosarcoma, but in this case the animal was sensitized with a significantly lower drug dose (0.25 instead of 2.5 mgkg bw). The lifetime image (Fig. 3a) still allows a clear detection of the neoplasia, while a fluorescence intensity image (Fig. 3b) fails, even though it was acquired 20 ns after the excitation pulses, when the natural fluorescence background had already mostly died away. The ratios R, and R,, which are equal to 1.19 and 0.83, respectively, confirm this statement. It is worth noting that in Fig. 3b the highest fluorescence signal is located in correspondence with a region of inflammation caused by the blade used to shave the animal before the experiment.

A similar approach has been followed for the mice bearing the L1210 leukemia, and analogous results have been achieved. As an example, Figs. 4 and 5 show the lifetime images for two mice sensitized with 2.5 or 0.25 mgkg bw of HpD, respectively. The corresponding values of the ratio K, are 1.25 and 1.22, respectively. Measurements performed on the leukemia confirmed that fluorescence intensity images are effective only at high doses, allowing tumor detection after the administration of 2.5 mgkg bw of HpD but not after the injection of 0.25 mgkg bw (data not shown).

Figure 5. Lifetime image of a tumor on the lower dorsum of a mouse bearing the L1210 leukemia and sensitized with 0.25 mgkg bw of HpD.

healthy surrounding tissues are observed either in the fluorescence intensity or in the average lifetime. Only in some cases is it possible to recognize the tumor mass, and even then it can barely be done. An example is shown in Fig. 6, which reports the lifetime image obtained from a control mouse bearing the fibrosarcoma. The ratio R, for the image displayed in Fig. 6 is 1.06, one of the highest values obtained from control animals.

No significant differences either in fluorescence intensity or lifetime were found between live (anesthetized) mice and dead animals, either sensitized or not, provided that the measurements on dead animals were performed immediately after the animal was sacrificed (i.e. within 1 min).

The outcomes of the imaging experiment are summarized in Table 1, which shows, as a function of the drug dose, the mean values and the standard deviations of T inside and outside the tumor and of their ratio 5. The values of T relative to tumor and healthy tissue in each mouse were calculated as described before (ie. average over different pixels in the same image). Then a further average was taken over all mice sensitized with the same drug dose. The ratio R, for any of the drug doses is statistically higher than for the control animals (P < 0.001 for the fibrosarcoma and P < 0.02 for the leukemia). On the other hand, even though the mean value of R, seems to increase with the HpD dose, the differences are not significant ( P > 0.1 for both tumor models).

In all the experimental conditions considered for spectroscopic point measurements (i.e. pathologic and healthy areas, in both tumor models), the decay curves were best fitted by the sum of four exponential functions with the lifetimes and relative amplitudes reported in Table 2. However, attention should be paid mainly to the values of the longest life-

~ 14.0 13.5 13.0 12.5 12.0 11.5 11.0 10.5 10.0 9.5 9.0

In general, for control (not sensitized) mice no or only very minor differences between the tumor area and the

Figure 6. Lifetime image of a tumor on the lower dorsum of a control (not sensitized) mouse bearing the MS-2 fibrosarcoma.


Table 1. of the HpD dose (mgkg bw), as calculated from imaging data*

Fluorescence lifetimes T (ns) in the tumor and in the healthy tissue and their ratio R, (mean 2 standard deviation) as a function

Dose (mdke bw)

2.5 1 0.25 0.1 0 (control)

MS-2 fibrosarcoma 14.27 2 1.63 13.84 f 0.80 12.20 i 1.14 T (tumor) (ns)

T (healthy) (ns) 11.89 f 0.88 11.63 2 0.91 10.43 5 0.80 R, 1.20 2 0.08 1.19 % 0.06 1.17 ir 0.06

L1210 leukemia T (tumor) (ns) 12.76 ? 0.85 10.75 i 1.75 T (healthy) (ns) 10.23 t 0.50 9.12 5 1.05 R, 1.24 i 0.02 1.17 % 0.09

*For each drug dose and tumor type six mice were measured.

time T , , which has to be compared with the values measured with the imaging technique. The average T, is significantly longer for the fluorescence coming from the tumor area than for the one detected in the surrounding region. This is true for both the fibrosarcoma and the leukemia, characterized by an average value of the ratio R, (between the T , value in the tumor and in the surrounding tissue) equal to 1.18 and 1.12, respectively. On the other hand, for what concerns the three shorter lifetimes, differences are observed in some cases between the value in neoplastic and healthy tissues, but such differences are either not statistically significant or not com- mon to both the tumor types.

DISCUSSION In all the cases considered in this study it appears that the average fluorescence lifetime in the tumor is significantly longer than in the healthy tissue, allowing the reliable detection of the neoplasia in lifetime images. Instead, the usual intensity-based images fail to reveal the tumor at low doses, for both the MS-2 fibrosarcoma and the L1210 leukemia, even though they have been acquired 20 ns after the excitation to get rid of the short-lived natural background. Im- ages taken synchronously with the excitation are even more blurred because the natural tissue fluorescence completely masks the drug signal (17,21,27).

Moreover, the diagnostic potential of the lifetime imaging technique is rather independent of the drug dose and of the tumor model. The values reported in Table 1 deserve some comments. The mean decay time T either in the tumor or in the healthy tissue decreases with the drug dose, but the ratio k, which is the actual diagnostic figure, stays essentially constant for both the fibrosarcoma and the leukemia. The shortening of the lifetimes at low drug doses is likely due to the contribution of the short-lived natural fluorescence, which becomes more relevant as compared to the exogenous long-lived signal. The measured lifetime, which is only an approximate estimate, is therefore more influenced by the autofluorescence at low HpD doses. However, being that the broadband tissue fluorescence is not selective for tumors in the models considered, its presence leads only to a uniform bias in time-gated images, which results in a moderate shift of the lifetimes but does not significantly affect the ratio R,.

The standard deviations of T inside and outside the tumor mainly account for both animal inhomogeneities and instru-

1.74% 1.37 11.16 f 0.97 0.15 2 0.79 10.71 f 0.68 1.16 50 .08 I .04 Z 0.04

9.79 f 0.48 9.37 t 0.15 1.05 f 0.04

mental errors and are lower than 16% of the mean values for all the experimental conditions considered in this study. The standard deviations of the ratio R, are remarkably lower (57%). This can be explained by taking into account that the main instrumental error, i.e. the shift of the time refer- ence, affects the absolute value of the measured lifetime but not significantly the relative differences between the values evaluated in tumor and healthy tissues. Moreover, at least part of the biological diversity cancels out when one takes the ratio of quantities measured on the same animal.

According to the above considerations, the lifetime imaging technique proved very reliable: in almost all of the examined mice the ratio R, was appreciably greater than 1. Therefore, the sensitivity of this techniques approaches loo%, at least in the situations considered up to now. This result seems very promising, even though it has been presently achieved only with a double set of experimental tumors. Moreover, a positive indication in terms of specificity can be drawn from the clear detection of the neoplasia even in the presence of regions of strong inflammation, which blind the traditional fluorescence images, as can be observed in Fig. 3b. A further aspect is much in favor of this technique: it is rather insensitive to unevenness in the excitation beam and vignetting of the collecting optics. In fact such spurious effects tend to cancel one another out in mathematical computations that lead to the lifetime matrices. This feature is particularly important in view of the possible em- ployment of this technique in conjunction with an endoscope, which is usually significantly affected by the above- mentioned drawbacks.

The average ratio R, for control mice is 1.04 and 1.05, for the fibrosarcoma and the leukemia, respectively, whereas one would expect an average value equal to 1. Some elements could be considered to justify this discrepancy. First of all, the presence of endogenous porphyrins in the neoplastic tissues has already been suggested by some authors (31,32). Furthermore an artifact, which can be ascribed to the morphology of the tumor, cannot be completely ruled out. In fact the swelling of the tissue might alter the pene- tration of the excitation light and the reabsorption of the fluorescence signal. Nevertheless, the visual comparison of lifetime images of control mice and sensitized mice clearly shows the relevance of the drug for the successful employ- ment of this diagnostic technique. As an example, the ratios


R, for the images in Figs. 3a (0.25 mgkg bw) and 6 (control) are 1.19 and 1.06, respectively, which are indeed not very different values, yet it is hard to locate the tumor in the latter image, while it is straightforward to do it in the former one.

As a matter of fact, the imaging study indicates that the long-lived component of HpD is characterized by a longer lifetime when incorporated into neoplastic tissues. This is in agreement with the dependence on the environment typically observed for various photophysical properties of HpD. How- ever, it should be kept in mind that the lifetime matrices considered so far do not represent an accurate measure of the long decay time of the HpD fluorescence, because they result from a forced monoexponential fit on the tail of the fluorescence decay, which is unrealistic.

In a preliminary study, fluorescence images were acquired even with delays shorter than 10 ns (i.e. 3, 5, 8 ns). The corresponding lifetime images are more sensitive to the presence of short-lived signals, which can be due to both the endogenous emission and the faster HpD components. When these images are compared to the ones reported above (obtained with longer delays), it appears that the localization of tumor areas is more difficult in the former ones. This seems to indicate that, at least in the experimental conditions considered, overall the short-lived species, both endogenous and exogenous, give no significant contribution to tumor detection in the time domain. Hence, either the fast components of HpD are not characterized by different lifetimes when incorporated in tumor and healthy surrounding tissues or this difference is masked by the natural fluorescence signal.

To confirm the results of the imaging experiments and to better investigate the eventual influence of the pathologic substrate, the fluorescence lifetimes were accurately measured with the high-resolution spectroscopy apparatus. The choice of the excitation and detection wavelengths, at 413 and 630 nm, respectively, has been made according to the following considerations. The long-lived emission of HpD, usually attributed to free chromophores such as monomeric species and end-rings of polymeric chains, is centered around 630 nm in hydrophobic environments, such as cells and tissues (25) . Moreover, its absorption maximum is red- shifted toward 400 nm, as compared to the one of species characterized by a faster fluorescence decay.

Fluorescence lifetimes measurements performed on mice sensitized with 0.25 mgkg bw HpD basically confirm the main result obtained from image analysis. In fact, for all the animals considered, the measured value of the longest lifetime T, was higher in the tumor than in the healthy peritu- moral area. As already observed for imaging data, the difference is small (between 10% and 20% of the average value of T,), but significant, as proved by the limited standard deviation (<lo%). None of the three faster fluorescence components showed a reliable diagnostic potential. In agreement with the imaging results, this seems to suggest that the biological substrate has no clear influence on the lifetime of short-lived exogenous species. However, it should also be taken into account that the natural fluorescence species are mainly short-lived and contribute non-negligibly to the lifetimes T ~ . T~ and T~ reported in Table 2. As suggested above, this effect might mask a difference between tumor and healthy tissue possibly present even in the fast components of HpD.


Further studies are definitely required to investigate the mechanisms that lead to the lengthening of the HpD decay time in the tumor. However, its diagnostic meaning and the effectiveness of the lifetime-imaging technique seem clearly stated, at least for the tumor models considered.

The lowest drug dose considered in this study (0.1 mgkg bw), scaled to humans, corresponds to the administration of a very low quantity of porphyrins, which probably does not give any significant skin sensitization. If the positive results in terms of sensitivity and specificity achieved with experimental tumors are confirmed also with spontaneous tumors, this technique may become a powerful aid for endoscopic examinations.

Presently, the imaging apparatus requires a rather long acquisition time and an off-line processing; however, there are no technological difficulties that prevent a real-time diagnostic system to be set up. The instruments, including the video camera, can easily be computer controlled. Even in the present implementation, the dye laser could be replaced by a more powerful solid-state source (optical parametric laser oscillator) improving the S/N ratio and reducing the need for a time-consuming frame averaging, and the matrix calculations could be performed on-line (in some millisec- onds) by the fast processor present in the most updated image boards provided by several manufacturers. The development of a system suitable for a clinical trial is in progress.

AcknowledRements-This work has been partially supported by the National Council for Research (Italy), under the Special Project “Applicazioni Cliniche della Ricerca Oncologica.”

REFERENCES 1. Tang, G. C., A. Pradhan and R. R. Alfano (1989) Spectroscopic

differences between human cancer and normal lung and breast tissues. Lasers Surg. Med. 9, 290-295.

2. Marchesini, R., M. Brambilla, E. Pignoli, G. Bottiroli, A, C. Croce, M. Dal Fante, P. Spinelli and S. Di Palma (1992) Light- induced fluorescence spectroscopy of adenomas, adenocarcino- mas and non-neoplastic mucosa in human colon I. In vitro measurements. J . Photochem. Photnbiol. B 14, 21 9-230.

3. Schomacker, K. T., J. K. Frisoli, C. C. Compton, T. J. Flotte, J. M. Richter, N. S. Nisioka and T. F. Deutsch (1992) Ultravi- olet laser induced fluorescence of colonic tissue: basic biology and diagnostic potential. Lasers Surg. Med. 12, 63-78.

4. Ramanujam, N., M. F. Mitchell, A. Mahadevan, S. Wairen, S. Thomsen, E. Silva and R. Richards-Kortum (1994) In vivo diagnosis of cervical intraepithelial neoplasia using 337-nm-e~- cited laser-induced fluorescence. Proc. Natl. Acad. Sci. USA 91,

5. Fisher, A. M. R., A. L. Murphree and C. J. Comer (1995) Clin- ical and preclinical photodynamic therapy. Lasers Surg. Med.

6. Schuitmaker, 1. J., H. L. L. M. van Leengoed, F. W. van der Meulen, W. M. Star and N. van Zandwijk (1996) Photodynamic therapy: a promising modality for the treatment of cancer. J. Photochem. Photobiol. B 34, 3-12.

7. Dougherty, T. J. (1987) Photosensitizers: therapy and detection of malignant tumors. Photochem. Photobiol. 45, 879-889.

8. Henderson, B. W. and T. J. Dougherty (1992) How does photodynamic therapy work? Photochem. Photohiol. 55, 145-1 57.

9. Kato, H., K. Aizawa, J. Ono, C. Konaka, N. Kawate, K. Yoney- ama, K. Kinoshita, K. Nishimiya, H. Sakai, M. Noguchi, T. Tomono, S. Kawasaki, Y. Tokuda and Y. Hayata (1984) Clin- ical measurements of tumor fluorescence using a new diagnostic system with hematoporphyrin derivative, laser photoradiation, and a spectroscope. Lasers Surg. Med. 4, 49-58.

10. Profio, A. E. (1984) Laser excited fluorescence of hematopor-

10193-1 0 1 97.

17, 2-31.

phyrin derivative for diagnosis of cancer. IEEE J. Quunf. Elec-

1 1 . Kessel, D. (1987) Tumor localization and photosensitization by derivatives of hematoporphyrin: a review. IEEE J . Quant. Elec- tron. 23, 1718-1720.

12. Balchum, 0. J., A. E. Profio and N. Razum (1987) Ratioing Auorometer probe for localizing carcinoma in situ in bronchi of the lung. Photochem. Phorobiol. 46, 887-891.

3. Baumgartner, R., H. Fisslinger, D. Jocham, H. Lenz, L. Ru- precht, H. Stepp and E. Unsold (1987) A fluorescence imaging device for endoscopic detection of early stage cancer-instrumental and experimental studies. Photochem. Photohiol. 46,

4. Brodbeck, K. J., A. E. Protio, T. Frewin and 0. J. Balchum (1987) A system for real time fluorescence imaging in color for tumor diagnosis. Mecl. Phys. 14, 637-639.

5. Monnier, P., M. Savary, C. Fontolliet, C. Wagnieres, A. Chat- elain, P. Cornaz, C. Depeursinge and H. Van Den Bergh ( 1 990) Photodetection and photodynamic therapy of ‘early‘ squamous cell carcinoma of the pharynx, oesophagus and tracheo-bron- chid tree. Lasers Med. Sci. 5, 149-169.

16. Audersson-Engels, S., I. Johansson, K. Svanberg and S. Svan- berg (1991) Fluorescence imaging and point measurements of tissue: applications to the demarcation of malignant tumors and atherosclerotic lesions from normal tissue. Photochem. Photo- biol. 53, 807-814.

17. Cuheddu, R., G. Canti, P. Taroni and G. Valentini (1991) Use of time-gated fluorescence imaging for diagnosis in biomedi- cine. J. Photochem. Photobiol. B 12, 109-1 13.

18. Kohl, M., J. Neukammer, U. Sukowski, H. Rinneberg, H. J. Sinn, E. A. Friedrich, G. Graschew, P. Schlag and D. Wohrle (I99 I ) Imaging of tumors by time-delayed laser-induced fluorescence. In Future Trends in Biomedicul Applications of La- .sers, Vol. 1525 (Edited by L. 0. Svaasand), pp. 26-34. SPIE Press, Bellingham.

19. Schneckenburger. H., K. Konig, K. Kunzi-Rapp, C. Westphal- Frosch and A. Ruck (1993) Time-resolved in-vivo fluorescence of photosensitizing porphyrins. J . Photochem. Photobiol. B 21, 143-147.

20. Cubeddu, R., R. Ramponi and G. Bottiroli (1986) Time-resolved fluorescence spectroscopy of the tumor localizing fraction of heinatoporphyrin derivative in micelles. Chem. Phys. Lett. 128, 439442.

21. Cubeddu, R., G. Canti, P. Taroni and G. Valentini ( 1 993) Time- gated fluorescence imaging for the diagnosis of tumors in a murine model. Photochem. Photobiol. 57, 480485.

22. Canti, G., L. Ricci, V. Cantone, P. Franco, 0. Marelli, A. An- dreoni, R. Cubeddu and A. Nicolin (198:) Hematoporphyrin derivative photoradiation therapy in niurine solids tumors. Cun- cer Lett. 21, 233-237.

23. Razum, N., 0. J. Balchum, A. E. Profio and F. Carstens ( 1 987) Skin photosensitivity: duration and intensity following intrave- nous hematoporphyrin derivates, HpD and DHE. P hotochem.

24. Roeder, B. and H. Wabnitz (1987) Time-resolved fluorescence spectroscopy of hematoporphyrin, mesoporphyrin, pheophor- bide a and chlorine e, in ethanol and aqueous solution. J. Pho- tochem. Photobiol. B 1, 103-1 13.

25. Cubeddu, R., R. Ramponi, W.-Q. Liu and F. Docchio (1989) Time-gated fluorescence spectroscopy of the tumor localizing fraction of HpD in the presence of cationic surfactant. Photo- chem. Photobid. 50, 157-163.

26. Cubeddu, R., A. Pifferi, P. Taroni, 0. Valentini and G. Canti (1995) Tumor detection in mice by measurement of fluorescence decay time matrices. Opt. Lett. 20, 2553-2555.

27. Cubeddu, R., P. Taroni and G. Valentini (1993) Time-gated imaging system for tumor diagnosis. Opt. Eng. 32, 320-325.

28. Cubeddu, R., G. Canti, A. Pifferi, P. Taroni and G. Valentini (1997) A real time system for fluorescence lifetime imaging. Proc. Soc. Photo-Opt. Instrum. Eng. 2976. (In press)

29. Cubeddu, R., F. Docchio, W.-Q. Liu, R. Ramponi and P. Taroni (1988) A system for time-resolved laser fluorescence spectroscopy with multiple picosecond gating. Rev. Sci. Instrum. 59, 2254-2258.

~ K W Z . 20, 1502-1 507.

759-763.

Photobiol. 46, 925-928.

236 Rinaldo Cubeddu eta/.

30. Marquardt, D. W. (1969) An algorithm for least-squares esti- mation of nonlinear parameters. J. Soc. Znd. Appl. Math. 11, 431441.

31. Harris, D. M. and J. Weekhaven (1987) Endogenous porphyrin fluorescence in tumors. Lasers Surg. Med. 7 , 467472.

32. Bottiroli, G., A. C. Croce, D. Locatelli, R. Marchesini, E. Pig- noli, S . Tomatis, C. Cuzzoni, S . Di Palma, M. Dal Fante and P. Spinelli (1995) Natural fluorescence of normal and neoplastic human colon: a comprehensive “ex uivo” study. Lasers Surg. Med. 16, 48-60.

Fluorescence Lifetime Imaging of Experimental Tumors in Hematoporphyrin Derivative-Sensitized Mice

Documents