lack of reciprocity in drug and light dose dependence of ... · photodynamic therapy if differences...

[CANCER RESEARCH 55. 3078-3084. July 15. 1995]

Lack of Reciprocity in Drug and Light Dose Dependence of Photodynamic Therapyof Pancreatic Adenocarcinoma in Vitro1

K. Thomas Moesta, William R. Greco, Stephen O. Nurse-Finlay, John C. Parsons, and Thomas S. Mang2

Great Lakes BiomÃ©dical Laser Research Center. Department of Oral anil Masillo/acial Surgen; State University of New York at Buffalo. Buffalo. New York 14214-3008Â¡T.5. M./: Robert-Roessle-Clinic. Robert-Roessle-Slrasse 10. O-1II5 Berlin-Buch. Germany IK. T. MJ; and Departments of Radiation Biology Â¡S.O. N-F.I and BionwthemalicsÂ¡W.R. G.. 1. C. P.I and Grace Cancer Drug Center Â¡W.R. C.}. Roswell Park Cancer Institute. New York State Department of Health. Buffalo, New York 1426Ã•-0001

ABSTRACT

Two human pancreatic cell lines, MIA PaCa 2 and Capan 2, weretreated by photodynamic therapy in vitro with Phntophrin (0.01-25 Mg/ml;24 In and then light (1-50 .l/cnr: A = 630 nm). The following model was

fit to 6 dataseis with weighted nonlinear regression:

_ [Econ - BlfFDLf ,C, â€” ~ ' p riÂ«-Â» r IH. ' **

where F = I or

ci-*-*}F =

kL

1 +[yFDLf

F =k,L k,L

The symbols are: /â€¢.'.cell growth: Econ, control growth in the absence of

the combination; />'.background signal; m, slope parameter; y, interactionparameter; /'. concentration of Photofrin; /.. light dose; /â€¢'.fraction of

Photofrin not photobleached by the light dose; A.A., Aâ€žbleaching parameters; A, distribution parameter for biexponential bleaching equation.Simple reciprocity of photosensitizer concentration and light dose was notfound; compensation for photobleaching was critical. MIA PaCa2 required the monoexponential bleaching factor, whereas Capan 2 requiredthe biexponential bleaching factor. The greater photosensitivity of MIAPaCa2 over Capan 2 can be best explained not by differences in theinteraction parameter but rather by differences in the photobleachingpattern and rate. It may be possible to further enhance the selectivity ofphotodynamic therapy if differences in photobleaching between differentcell types can be exploited by adequate dosimetry.

planning (7). Whether the fluorescing components of Photofrin areidentical with the ones that are photodynamically active is controversial. Thus, the photobleaching rates measured by a decrease in Photofrin fluorescence might not necessarily reflect an equal reduction inphotodynamic effect.

PDT has demonstrated a highly selective effect on experimentalpancreatic tumors in rodents (11). Investigations show a significantaccumulation of photosensitizer in chemically induced animal pancreatic adenocarcinoma tissue as compared with normal pancreas(12). Malignant:normal pancreatic tissue photosensitizer ratios rangefrom 2.2:1 for Photofrin to 3.0:1 for aluminum-sulfonated phthalo-

cyanine (13). Another notable finding in these studies was that normalpancreatic tissue shows a higher resistance to PDT, even at similartissue concentrations of photosensitizer. The latter phenomenon isunique for the pancreas and has been attributed to a preferentiallocalization of the photosensitizer within the lymphatics. However,these studies also demonstrated a dramatic differential in the photo-

bleaching kinetics in the malignant versus the normal pancreatictissues. Normal pancreas demonstrated a slower photobleaching rate,which was consistent with the physiological resistance to PDT (11).

In this study we report the effective photobleaching rates of Photofrin in two pancreatic tumor cell lines, estimated from fitting athree-dimensional dose-response model, which includes bleaching

parameters, to data from in vitro PDT studies of Photofrin and light.Data were obtained over wide ranges of light and Photofrin doses.

INTRODUCTION

In the process of tumor destruction, PDT' uses light to activate an

intracellularly bound photosensitizing drug. The mechanisms of cellular and tumor destruction are not entirely understood (1), but theeffects are generally accepted to be proportional to the product of drugand light dose. This relationship has been generally referred to asreciprocity of drug and light dose, and has been shown by severalauthors both for in vitro (2-4) and for in vivo (5, 6) systems [for in

vivo systems, a linear relationship is assumed between injected doseand accumulated tissue concentration of photosensitizer (7, 8)]. However, this theory does not include the effect of photobleaching, thedestruction of the photosensitizer by light. Photobleaching has beenobserved in vitro (9, 10) and in vivo (7) by a light-induced decrease in

fluorescence and is of considerable importance for clinical dosimetry

Received 8/2/94; accepted 5/16/95.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate (his fact.

' Supported by National Cancer Institute Grants CA47299, CA55791, CA46732.

RR10742 and CA16056. K. T. M. received support from Deutsche Forschungsgemeinschaft (Bad Godesberg. Germany).

2 To whom requests for reprints should be addressed, a! Department of Oral and

Maxillofacial Surgery. 112 Squire Hall, State University of New York at Buffalo, 3435Main Street. Buffalo. NY 14214-3(X)8.

3 The abbreviations used are: PDT. photodynamic therapy; Â£,cell growth measured by

the thymidine uptake assay; Econ, control growth in the absence of the combination ofphotosensitizer and light; B, background signal; m, slope parameter; y. interaction parameter; D, concentration of Photofrin; L, light dose; F. fraction of Photofrin not bleachedby Ihe light dose; k, k,. As, bleaching parameters; A, distribution parameter for biexponential bleaching equation.

MATERIALS AND METHODS

Human Cell Lines. MIA PaCa 2 was obtained from American TypeCulture Collection (CRL 1420). The cell line grows with a doubling time of 26h under our conditions and is known to form undifferentiated tumors in nudemice (14). Capan 2 was obtained from American Type Culture Collection(HTB 80) and forms highly differentiated tumors in nude mice (15). Thedoubling time was 48 h. Both cell lines were maintained in Iscove's modified

DMEM (GIBCO-BRL, Grand Island, NY), substituted with 10% certified

PCS (GIBCO). These pancreatic carcinoma lines were chosen for their significant differences in PDT sensitivity (16) and for the fact that in vivo photo-

bleaching rates vary considerably between normal pancreas and pancreaticadenocarcinoma (11).

Photosensitizer. Porfimer sodium (Photofrin) was obtained from QuadraLogic Technologies (Vancouver, British Columbia, Canada). The freeze-dried

powder was reconstituted with the use of 5% dextrose and was then diluted toa stock concentration of 2(X) fig/ml and frozen in uliquots until use. Photofrinis a complex mixture of porphyrin compounds; characterization of the mixturehas been described previously (for example, see Ref. 17).

Light System. Light at a wavelength of 630 nm was delivered by a tunabledye laser with the use of Kilon Red dye pumped by an argon ion laser (SpectraPhysics Models 375 and 171, respectively). Wavelength accuracy was checkedfor each experiment with the use of an Oriel monochromator (Model 7240).Total light intensity was measured with the use of an United Detector Technologies Integrating Sphere. The power density was then calculated based onthe diameter of the treatment field. Homogeneity of the light field was verifiedwith a Yellowstone Instruments Photometer (Yellowstone Instruments, Inc.).

PDT Treatment. The cells were plated in 96-well plates at 3000 cells/well.Forty-eight h later, the cells were exposed for 24 h to various concentrations

of Photofrin in full medium, containing 10% FCS. The medium was then

3078

Research. on February 27, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

PHOTODYNAMIC THERAPY OF PANCREATIC ADENOCARCINOMA

removed, the cells were washed once with HBSS, and again covered withdrug-free medium. After a "leak-off time of 45 min, the medium was changed

again. After temperature adjustment of the new medium (15 min), the entireplate was exposed to laser light at 630 nm from below by an index grated lens.The uniformity of light application was within Â±10%. The power density wasconstant with 10 mW/cnr. Light doses from 1 to 50 J/citr were produced byvariation of the exposure time from 100 to 5000 seconds. At each light dose,a complete 96-weII plate was evaluated, containing a 100-fold range of

Photofrin concentrations, each in heptuplicate. All experiments were performed three times. The entire treatment setup was situated within an incubatorproviding stable temperature, humidity, and 5% CO2 atmospheric content. Allmanipulations of cells containing photosensitizer were carried out in a darkroom under a light source specifically filtered to remove Photofrin-activating

light wavelengths.Thymidine Uptake Assessment. Twenty-four h after treatment, the cells

were incubated with 1 jiCi ['H]thymidine per well (100 jj.1)for 6 h. The plates

were then rapidly frozen on dry ice and stored at -20Â°C until harvesting. After

thawing in a waterbath, the cells were harvested with the use of a CambridgeTechnology PHD Harvester with Whatman glass microfibre filter 934AH. Celllysis was induced by exposure to distilled water. Dried filters were dissolvedin Liquiscint and counted in a Beckman Liquid Scintillation Counter ModelLS5810.

Data Analyses and Modeling. Equations 1-4 were used to model the

interaction of Photofrin and light, and were derived from concepts of agentinteraction reported previously (18-20). Equation 1 isa model of coalism (21);

i.e., the situation in which neither of two agents is effective individually, butin which the combination of the two agents (e.g., Photofrin and light) iseffective in eliciting a response (e.g., growth inhibition). Equations 2-4 are

models which describe the photobleaching of Photofrin by light.

^ _ [Econ - BlyFDL]'"t, T~7 .-r-, w in. + "\+[yFDL]â„¢

where (a) F = 1

(b)

(c)

F =

_

kL

- A)(\ - A(\ - (

k,L

(1)

(2)

(3)

(4)

Equation 1 is in the form of the ubiquitious Hill equation (22, 23), which isfrequently used to model concentration-effect phenomena. Note that D (the

Photofrin concentration) and L (the light dose) are multiplied inside thebracket. This models reciprocity. The y interaction parameter models theintensity of the interaction between Photofrin and light. A larger value for yimplies a more intense interaction. The magnitude of y is algebraically related

to the degree of bowing of isobols; a larger y will result in a larger degree ofbowing. The effective IC5(, for Photofrin at a particular light dose is [ 1/yL], andthe effective \C5I, for light at a particular Photofrin concentration is [l/yD].When F is equal to 1 (Equation 2), this implies strict reciprocity of photosensitizer and light. Deviations from strict reciprocity are modeled with morecomplex expressions for F (Equations 3 and 4). and with the estimation ofbleaching parameters, k, kÂ¡,k2, and A. When the slope parameter, m, isnegative, the concentration-effect curves tend downward with increasing agent

concentrations; when m is positive, the curves tend upward. The absolutemagnitude of m indicates the steepness of the curves; a high value of m impliesa steep curve, a low value, a shallow curve. As a point of reference, note thatenzyme inhibitors usually have m values of 1 or -1 (depending on the

definition of the end point). Note that when bleaching models. Equations 3 and4, are inserted into Equation 1, then the light dose, L, can be canceled in severalplaces, leaving a simplified composite equation.

In vitro PDT experiments were analyzed by fitting the set, Equations 1-4,

to data with iteratively reweighted nonlinear regression with a custom softwarepackage (SYNFIT; Ref. 20), which was written in the computer languageMicrosoft FORTRAN (Microsoft Corp., Bellevue, WA). The data from each of6 experiments were analyzed separately (3 for Capan 2, 3 for MIA PaCa 2).Each experiment typically included a wide range of Photofrin concentrationsand light doses, and the numbers of data points ranged from 452 to 575. Datawere weighted by the reciprocal of the square of the predicted response, unlessthe predicted response was less than 10% of Econ. Then the weighting factorwas 1/(0. lEcon)2. This weighting scheme avoided overweighting data near the

limit of detection. All weights were normalized such that the sum of the finalweights for one experiment equaled the number of data points.

RESULTS

The best-fit parameter estimates from the fits of the modelingsystem, Equations 1-4, to the data from 6 experiments are listed in

Table 1. With few exceptions, the 95% confidence intervals in Table1 did not encompass zero, and the standard errors were typically lessthan 10% of the parameter estimates. This implies that the parameterswere well estimated and that the final fits of the model to the datawere good. The 95% confidence intervals for the bleaching parameters were used to decide among the hiearchical Equations 2-4 for the

most appropriate model for photobleaching. If the 95% confidenceintervals for either kÂ¡or k2 from a fit of Equation 1 plus Equation 4to the data encompassed zero, then the fit was rejected, and then a fitof Equation 1 plus Equation 3 was tried. This was true of the first twoexperiments with MIA PaCa 2. Then, if the 95% confidence intervalfor k from a fit of Equation 1 plus Equation 3 to the data would have

Table 1 Parameter estimates (Â±S.E.)for the fit of Equations ÃŒ-4to data from 6 experiments in which cells were exposed to Phott

Beneath each SE are listed the lower and upper 95% confidence intervals for the parameters. Experimental and statistical details aregivenCell

lineCapan

2Capan

2Capan

2MIA

PaCa2MIA

PaCa2MIA

PaCa 2Experiment

no.123123No.of

wells496547572575452574ficai1360137

129014302060

Â±35199021308380

Â±1208140

862033300

Â±45032400342IKI12900

Â±290123001350024400

Â±3902360025200B95.8

Â±6.982.310997.1

Â±1469.7125110Â±

1972.814797.7

Â±65-29.7

225246

Â±37173319141

Â±5043.0239m-2.69

Â±0.18-3.04-2.34-3.57

Â±0.23-4.02-3.12-3.20

Â±0.11-3.42-2.98-3.57

Â±0.10-3.77

-3.37-5.

17 Â±0.31-5.78-4.56-4.20

Â±0.16-4.51-3.89y0.498

Â±0.0520.3960.6000.328

Â±0.0190.2910.3650.605

Â±0.0370.5320.6780.518

+ 0.0083

0.5020.5340.694

+ 0.017

0.6610.7270.464

Â±0.00870.4470.481k

ort,0.0229

Â±0.013-0.00258

0.04840.00552

Â±0.0064-0.00702

0.01810.0191

Â±0.00410.01110.02710.01

34 Â±0.000880.011680.01510.135

+ 0.00580.1240.1460.0160

+ 0.0011

0.01380.0182ifrin

pÃ¬nslÃ¬^htat 630 nm

in thetext.*20.557

Â±0.230.1061.010.401

Â±0.120.1660.6360.855

+ 0.13

0.6001.11A0.694

Â±0.5690.8190.652

Â±0.5520.7520.788

Â±0.7610.8150.0640.0510.014

3079



PHOIODYNAMIC THERAPY OF PANCREATIC ADI NIK AKCINOMA

encompassed zero, the fit would have been rejected and we wouldhave concluded that photobleaching of Photofrin by light was notimportant. However, such a rejection did not occur. The data from all3 Capan 2 experiments were fit better by the more complex photo-

bleaching model. Equation 4. In contrast, the data from the first 2 MIAPaCa 2 experiments were fit better by the simpler photobleachingmodel, Equation 3. Although the data from the third MIA PaCa 2experiment were fit better by Equation 4, the kt estimate was negativeand unrealistic: therefore, this fit was rejected on mechanisticgrounds. The fit of Equation 3 to the data from the third MIA PaCa 2experiment is reported in Table 1.

The control [3H]thymidine incorporation (Econ parameter) was 3-9

times (Table 1) higher for MIA PaCa 2 than for Capan 2. This reflectsthe faster growth rate of MIA PaCa 2. Another source of variation inEcon is the use of different y counters for different experiments,depending on availability. The slope parameter, m, tended to be higherin absolute magnitude for MIA PaCa 2 than for Capan 2, indicatingsomewhat steeper concentration-effect curves for MIA PaCA 2. The

combined action parameter, y, tended to be similar for the 2 cell lines.The photobleaching parameters k or k, are similar for 5 of theexperiments (0.00552-0.0229), with the exception of the second MIAPaCa 2 experiment (k = 0.135). It should be noted that in the secondMIA PaCa 2 experiment, the data at the highest light dose, 50 J/cm2,

could not be evaluated due to technical problems, and that in the firstand third MIA PaCa 2 experiments, the data measured at this lightdose was highly important for the estimation of k. Therefore, the highestimate of k from the second MIA PaCa 2 experiment is somewhatsuspect, and in future repititions of experiments, it will always beimportant to include sufficient design points at high light doses to wellcharacterize k. For Capan 2, the majority of the Photofrin (A distribution parameter, 0.652-0.788) undergoes very fast photobleaching

(k2 parameter). It should be noted that y and k2 were consistentlyhighly correlated, with = 0.82, 0.76, and 0.89 for the 3 respective

Capan 2 experiments. Therefore, problems in the estimation of y willtend to greatly influence the estimate of k2. and vice versa.

Results are shown graphically for the first MIA PaCa 2 experimentin Figs. 1-4. and for the third Capan 2 experiment in Figs. 5-8. In Fig.

IA, the best fit of Equations 1 plus 3 to the data is shown as athree-dimensional concentration-effect surface superimposed on the

raw data. The surface starts at Econ at the [0,0] point, and remains atEcon for the entire ranges of concentrations of Photofrin and doses oflight for which only one agent is present. The surface then fallssharply towards background (B) in all regions in which both Photofrinand light are applied. The surface is scooped out; this indication ofcoalism can be best seen in the tightening of the fishnet into thebottom far left corner of the cube. Fig. \B was constructed the sameas Fig. L4, except that k was set to O cm2/J. A comparison of the two

panels in Fig. 1 provides a visual indication of the effect of photo-

bleaching of Photofrin. The change in the surface is very subtle,because the bleaching factor results in only a slight decrease in theproduct of Photofrin concentration and light dose. The surface in Fig.\A falls somewhat steeper than the surface in Fig. IÃŸ.Because all datapoints include a projection line to the surface, the better fit in Fig. \Acan be appreciated by the shorter projection lines.

A two-dimensional representation of the same concentration-effect

surface in Fig. \A is shown in Fig. 2. The isobols in Fig. 2 wereformed by the intersection of the surface shown in Fig. 1/4with planesat 5, 10, 25, 50, 75, 90, and 95% of control. Equation 5, the model foreach isobol. was derived from Equation 1, the model for the entiresurface. The degree of bowing of the isobols indicates the degree ofinteraction of Photofrin and light and is mathematically related to themagnitude of the interaction parameter, y. As y increases and as the %

k = 0.0134 cm'/J= 0 cm2/J

"*

Fig. 1. A, three-dimensional concentration-effect surface for the exposure of MIA PaCa 2 human pancreatic cells to Photofrin (24-h exposure, medium change, wash, 45-min efflux.medium change) and then laser light at 630 nm, with exposure times ranging from 100 to 5000 s to achieve light doses from 1 to 50 J/cm2. Grid, predicted co nee nirat ion -effect surfaceestimated from fitting Equation 1 to the dala with nonlinear regression as described in the text; points, measured [3H]thymidine uptake from single wells of a 96-well plate. The setof best-fit parameters from experiment 1 for MIA PaCa 2 cells, listed in Table 1, including the bleaching constant, k = 0.0 134 cnr/J. were used to simulate the three-dimensional surface.Both fitted and raw data are expressed as a percentage of the estimated Econ parameter. â€¢,above the surface; O, below the surface. Vertical projection lines are drawn to indicate thedistance of the points to the surface. Concentration scales are linear. B, same as A except that the bleaching parameter, it, was set to 0, to illustrate its influence on the fittedthree-dimensional surface.

3080




Light (J/cm2)

Fig. 2. Families of Iwo-dimcnsional isobols for the exposure of MIA PaCa 2 human

pancreatic cells to Photofrin and then laser light, as described in the legend to Fig. 1. Thecurves were calculated as described in Fig. \A and in the text. The set of contours is atwo-dimensional representation of the three-dimensional surface in Fig. \A. The levels ofpredicted percentage control, ElEcon (5-95), are listed to the right of the isobols. Note that

the isobols do not reach the axes.

control increases, the degree of bowing increases. Note that theisobols in Fig. 2 will never reach the X- and X-axes. However, theisobol for 100% control will coincide with the X- and K-axes.

D =[% control/000 - % control)]1

(5)

Figs. 3 and 4 are concentration-effect curves (logarithmic concen

tration/dose scales) for Photofrin at different light doses, and for lightat different Photofrin concentrations, respectively, using estimatedparameters from the first MIA PaCa 2 experiment from Table 1. Thecurves are intersections of the surface shown in Fig. \A with verticalplanes at the doses of light (Fig. 3) and concentrations of Photofrin(Fig. 4) listed. Actual data points are included to provide a pictoralrepresentation of the experimental design and a visual analysis of thegoodness of fit. Although there is some scatter of the points about thefitted curves, there is no overall indication of a lack of fit of Equation1 plus Equation 3 to the data. Although not shown, the fit of Equation1 plus Equation 2 (F = 1) results in systematic shifts of the curves

away from the data.

Fig. 3. Families of two-dimensional concentration-effect curves for

the exposure of MIA PaCa 2 human pancreatic cells to Photofrin andthen laser light, as described in the legend to Fig. 1. Ordinate. Photofrinconcentration on a log scale; abscissa, percentage control (ElEcon);each curve is for a different light dose. The curves are predictedconcentration-effect curves, as described in Fig. 1 and in the text. Thisset of curves is a two-dimensional representation of the three-dimen

sional surface in Fig. 3.

100

80

OÂ£60Oo* 40

20

Light (J/cm2)

0.01 0.1

[Photofri100

100

80

o 60

OÃ¼

40

20

[Photofrin](Â¡ig/mL)

10

Light (J/cm2)

Fig. 4. Families of two-dimensional concentration-effect curves for theexposure of MIA PaCa 2 human pancreatic cells to Photofrin and thenlaser light. Abscissa, light dose (log scale).

100

3081



I'llOTODYNAMIC THERAPY OF PANCREATIC ADENCK ARC INOMA

ki = 0.0191 crrWJk2= 0.855 crrf/J

75

1 M

2E

Ã¯/Jj\ â€¢'//mm\//!& â€¢â€¢â€¢â€¢â€¢*Â»-â€¢mÃ•Â»10

= k2= 0 cm2/J

30

2.8 42 VF ! 50

/

X.

Fig. 5. A, three-dimensional concentration-effect surface for the exposure of Capan 2 human pancreatic cells to Photofrin. The experimental and graphical details are as in the legendto Fig. 1. The set of best-fit parameters from experiment 3 for Capan 2 cells, listed in Table 1, including the bleaching constants, k, = 0.0191 cnr/J and k2 = 0.855 cnr/J, were usedto simulate the three-dimensional surface. B, same as A except that the bleaching parameters, k, and *,. were set to 0. to illustrate its influence on the fitted three-dimensional surface.

Figs. 5-8 for the third Capan 2 experiment are analogous to Figs.1-4 for the first MIA PaCa 2 experiment. Several differences between

respective figures are noteworthy. The effect of photobleaching ismuch more evident for Capan 2 than for MIA PaCa 2. This can beseen in the striking difference between Fig. 5A (with fitted bleachingparameters) and Fig. 5fi (with kÂ¡= k2 = 0) for Capan 2, in contrast

to the subtle difference between Fig. 1, A and fi. The bowing of theisobols for Capan 2 in Fig. 6 is less than for MIA, indicating less ofan interaction between Photofrin and light. The fit of the model(Equation 1 plus Equation 4) to the data is quite good; this is visuallyindicated in Figs. 7 and 8.

Fig. 9 includes simulations of the photobleaching factors, F, for all6 experiments. The subtle photobleaching for the first and third MIAPa Ca 2 experiments seem to be typical for MIA PaCa 2. As mentioned above, the photobleaching factor, F, for the second MIA PaCa2 experiment may have been unduly influenced by the absence ofreliable measurements at the highest light dose, 50 J/cm2. The 3

50

759095

012345Light (J/cm2)

Fig. 6. Families of two-dimensional isobols for the exposure of Capan 2 humanpancreatic cells to Photofrin and then laser light, as described in the legend to Fig. 1. Thecurves were estimated as described in Fig. \A and in the text.

experiments for Capan 2 all show similar patterns and degrees ofphotobleaching. It is clear from Fig. 9 that the degree of photobleaching of Photofrin by light is substantial, especially for Capan 2, withthe fraction of effective Photofrin reduced to 50% of the appliedconcentration at light doses ranging from about 2 to 11 J/cm2. These

results are similar to those reported (10) for the reduction of Photofrinfluorescence in an in vitro cell system (human carcinoma NHIK 3025cells). Moan et ai. (10) reported a 50% reduction of fluorescence froma fluence from a fluorescent lamp, corresponding to about 10 J/cm2 of

monochromatic light at 625-630 nm.

DISCUSSION

The reported experiments and data analyses accent the usefulnessof a response surface approach to the study of the combined-action ofmultiple agents. Questions about the empirical nature of photody-

namic therapy, such as the strength of interaction of photosensitizerand light or the importance of photobleaching, can be asked andanswered. Empirical results can lead to mechanistic insights. In addition, models such as Equations 1-4 can be expanded in a hiearchical

manner to characterize more complex phenomena, such as the use ofseveral photosensitizers and/or other agents in in vitro systems, anddose-response experiments in in vivo and clinical systems. Uncer

tainty estimates that accompany the estimated parameters provide ahigh degree of statistical rigor. Finally, estimates of parameters, suchas y, m, k, kÂ¡,and k2, may be used to establish structure-activity

relationships among sets of photosensitizer analogues.We chose the thymidine uptake technique to determine PDT effects

because the low cloning effeciency of Capan 2 (8%) would have madea cloning assay impractical. The thymidine uptake technique has beenextensively validated for other cell lines in comparison with cloningassays and the MTT-assay (25-26). There remains a potential for

underestimation of cell growth when conducting the thymidine exposure too early, or of overestimating cell proliferation from the pres-

3082




, 5 a o , . . , ,

ft

100

80

Fig. 7. Families of two-dimensional concentration-effect curves forthe exposure of Capan 2 human pancreatic cells to Photofrin and thenlaser light, as described in the legend to Fig. 1. Abscissa, Photofrinconcentration on a log scale: oridnate, percentage control (E/Econ);each curve is for a different light dose.

S 60

OO

40

20

Light (J/cm2)

0.1 1

[Photofrin] (/ig/mL)100

enee of cells that have stopped growing due to PDT, but which areable to resume the incorporation of [3H]thymidine after a short re

covery time. However, we previously validated our experimentalsystem by showing a good relationship between [3H]thymidine incor

poration and cloning efficiency (27) at several light doses.The first conclusion that can be drawn from the analysis of the data

is the lack of simple reciprocity of photosensitizer concentration anddelivered light dose. Good fits to the data from each of the 6 experiments are obtained only if a term for photobleaching is introducedinto the fitted model. This is consistent with the observation ofphotobleaching by fluoresence decrease in in vitro (10), in vivo (11),and clinical systems (7).

The second conclusion of this study is that the greater photosensi-

tivity of MIA PaCa 2 over Capan 2 can be best explained not bydifferences in the interaction parameter but rather by differences in thephotobleaching pattern and rate of bleaching of the Photofrin incorporated by the cells. The explanation for this may be manyfold. Capan2 cells, when grown in high inoculated cell numbers, demonstratedistinctly different growth patterns than MIA PaCa 2. Capan 2 forms

distinct clusters with microscopically indiscernible intercellular limits.MIA PaCa 2 cells grow dispersed on the plate without establishingintercellular contacts. With increasing cell density, Photofrin uptakeand PDT sensitivity of Capan 2 diminishes considerably. Abolishmentof these intercellular contacts by methods such as incompletetrypsinization reverses the effects of cluster density. Therefore, cellsurface availability plays a role in the uptake and consequent sensitivity of these cells. In this study we attempted to eliminate differencesbetween the two cell lines in cell cluster formation by inoculating cellsat low cell numbers. Previous studies have demonstrated that thistechnique helps to equalize, although not entirely, the uptake levels ofPhotofrin in Capan 2 to those in MIA PaCa 2 cells grown undersimilar conditions (16).

Differences observed in the two cell lines have been directly relatedto the level of differentiation of the lines (28). This difference indifferentiation may result in the uptake of different components ofPhotofrin by the two cell lines, resulting in varying photobleachingrates. Likewise, cellular compartmentalization of the photosensitizermay also result in differences in both PDT sensitivity and photo-

100

80

feoO

O

vP*" 40

20

[Photofrin] (/i g/mL)

0003005007O.100.15020030

10

Fig. 8. Families of two-dimensional concentration-effect curves for the

exposure of Capan 2 human pancreatic cells to Photofrin and then laserlight. Abscissa, light dose (log scale).

100Light (J/cm2

3083




1.00

0.75ml

F 0.50 -

0.25 -

Light (J/cm2)

Fig. 9. Simulations of the bleaching equations 2 and 3. with the best-fit parametersfrom Table 1. Each curve is labeled with the cell line (m. MIA PaCa 2; c. Capan 2) andIhc experiment number (1-3).

bleaching rates. It is hypothesized from previous investigations thatCapan 2 cells may only bind the photosensitizer on the outer cellmembrane (27). The lack of internalization of the photosensitizerwithin the cell could account for both increased photobleaching anddecreased photosensitivity.

Differences in photobleaching have also been observed in vivobetween normal and malignant pancreatic tissues; photobleaching isfaster in tumor tissue than in normal tissue (11). This observationcould be interpreted to be contrary to the results of our study becausethe more differentiated Capan 2 might be expected to act more likenormal tissue than MIA PaCa 2. However, comparisons between invitro and in vivo results are difficult and should be made with cautionbecause the former is obtained at the cellular level and the latter at thetissue level of organization. Differences in photobleaching results maybe due to critical factors which may differ dramatically between invitro and in vivo systems, such as the localization of photosensitizerand the possible existence of a photosensitizer triplet state or singletoxygen quencher.

These studies demonstrate that it may be possible to further enhance the selectivity of PDT by using three-dimensional dose-re

sponse models to characterize differences in photobleaching betweendifferent cell types, and then exploiting these differences with accurate dosimetry.

REFERENCES

1. Henderson. B. W.. and Dougherty. T. J. How does phoiodynamic therapy work?Photochem. Photobiol., 55: 145-157, 1992.

2. Henderson. B. W.. Bellnier, D. A.. Ziring, B.. and Dougherty, T. J. Aspects of thecellular uptake and retention of hematoporphyrin derivative and their correlationwith the biological response to PRT in vitro. Adv. Exp. Med. Biol., 160: 129-138,1983.

3. Gibson, S. L., and Hilf. R. Interdependence of fluence, drug dose and oxygen onhematoporphyrin derivative induced photosensitization of tumor mitochondria.Photochem. Photobiol.. 42: 367-373, 1985.

4. Pingar. V. H., Potter, W. R., and Henderson, B. W. Drug and light dose dependenceof photodynamic therapy: a study of tumor cell clonogenicity and histologie changes.Photochem. Photobiol., 45: 643-650, 1987.

5. Cowled, P. A., and Forbes. I. J. Photocytotoxicity in vivo of haemaloporphyrinderivative components. Cancer Lett.. 28: 111-118, 1985.

6. Pingar, V. H., and Henderson, B. W. Drug and Light dose dependence of photodynamic therapy: a study of tumor and normal tissue response. Photochem. Photobiol.,46: 837-841, 1987.

7. Mang, T. S., Dougherty, T. J., Potter, W. R., Boyle, D. G., Somer, S., and Moan, J.Photobleaching of porphyrins used in phoiodynamic therapy and implications fortherapy. Photochem. Photobiol., 45: 501-506, 1987.

8. Bellnier, D. A., and Dougherty, T. J. Membrane lysis in Chinese hamster ovarycells treated with hematoporphyrin derivative plus light. Photochem. Pholobiol., 36:43-47, 1982.

9. Moan, J., Christensen. T., and Jacobsen, P. B. Porphvrin-sensitized photoinactivationof cells in viiro. Prog. Clin. Biol. Res., 170: 419-442, 1984.

10. Moan. J.. Christensen, T., and Jacobsen, P. B. Porphyrin-sensitized photoinactivation

of cells in vitro. In: D. R. Doiron and C. J. Gomer (eds.). Porphyrin Localization andTreatment of Tumors, pp. 419-442. New York: Alan R. Liss, 1984.

11. Mang, T. S., and Wieman. T. J. Photodynamic therapy in the treatment of pancreaticcarcinoma: dihematoporphyrin ether uptake and photobleaching kinetics. Photochem.Photobiol.. 46: 853-858, 1987.

12. Schroder, T., Chen, I. W., Sperling, M., Bell, R., Brackelt, K., and Joffee, S. N.Hematoporphyrin derivative uptake and photodynamic therapy in pancreatic carcinoma. J. Surg. Oncol., 38: 4-9. 1988.

13. Chatlani, P. T., Toda, N., Ban. H., Bedwell. J.. Krasner, N., and Bown, S. G. Phoiodynamic therapy of normal and neoplastic rodent pancreas. Abstracts of the 3rd biennialmeeting of the Internationa] Photodynamic Association, p. 38. Buffalo, NY, 1990.

14. Marincola, F. M.. Drucker. B. J., Siao. D. Y., Hough, K. L.. and Holder. W. D. J. Thenude mouse as a model for the study of human pancreatic cancer. J. Surg. Res., 47:520-529, 1989.

15. Kyriazis, A. A., Kyriazis, A. P., Sternberg, C. N.. Sloane, N. H., and Loveless, J. D.Morphological, biological, biochemical, and kariotypic characteristics of humanpancreatic ductal adenocarcinoma Capan 2 in tissue culture and the nude mouse.Cancer Res., 46: 5810-5815, 1986.

16. Moesta, K. T., Hurley, E. L., and Mang, T. S. Intercellular organization determinesthe differential PDT-sensitivities of two human pancreatic cancer lines in v/rro. In: P.

Spinelli. M. Dal Fonte, and R. Marchesini (eds.). Photodynamic Therapy and BiomÃ©dicalLasers, pp. 292-296, Amsterdam: Eisevier Science Publishers. 1992.

17. Dougherty, T. J., Potter, W. R., and Bellnier, D. Photodynamic therapy for thetreatment of cancer: status and advances.. In: D. Kessel (ed.), Photodynamic Therapyof Neoplastic Disease, pp. 1-19. Boca Raton. PL: CRC Press, 1990.

18. Greco, W. R.. Park, H. S.. and Rustum, Y. M. An application of a new approach forthe quantitation of drug synergism to the combination of dj-diamminedichloroplati-num and 1-ÃŸ-D-arabinofuranosylcytosine. Cancer Res., 50: 5318-5327, 1990.

19. Greco, W. R., and Rustum. Y. M. Reply to letters by Berenbaum and SÃ¼hne!concerning Greco et al. Cancer Res., 50: 5318-5327, 1990. Cancer Res., 52:4561-4565. 1992.

20. Gaumont. Y.. Kisliuk, R. L.. Parsons, J. C., and Greco, W. R. QuantitÃ¤ten of folieacid enhancement of antifolate synergism. Cancer Res., 52: 2228-2235, 1992.

21. Greco, W. R.. Unkelbach H. D., Poch. G.. SÃ¹hnel, J.. Kundi. M., and BÃ¶deker,W.Consensus on concepts and terminology for interaction assessment: the SaaresilkÃ¤Agreement. Arch. Complex Environ. Studies, 4: 65â€”69,1992.

22. Hill, A. V. The possible effects of the aggregation of the molecules of haemoglobinon its dissociation curves. J. Physiol., 40: iv-vii, 1910.

23. Holford. N. H. G.. and Scheiner, L. B. Understanding the dose-effect relationship:clinical applications of pharmacokinetic-pharmacodynamic models. Clin. Pharmaco-kinet., 6: 429-453, 1981.

24. Metzler, C. M. Estimation of pharmacokinetic parameters: statistical considerations.Pharmacol. Ther., 13: 543-556, 1981.

25. Richter, A. M., Waterfield, E., Jain, A. K., Sternberg, E. D.. Dolphin, D.. and Levy,J. G. In vitro evaluation of phototoxic properties of four structurally related benzo-porphyrin derivatives. Photochem. Photobiol., 52: 495-500, 1990.

26. Richter. A. M.. Kelly, B., Chow, J., Liu, D. J., Towers, G. H. N., Dolphin, D., andLevy, J. G. Preliminary studies on a more effective phototoxic agent than hematoporphyrin. J. Nail. Cancer Inst., 79: 1327-1332, 1987.

27. Moesta, K. T., Dmytriuk, A., Schlag, P.. and Mang. T. S. Individual in vitro sensitivitiesof human pancreatic carcinoma cell lines to photodynamic therapy., ht: T. J. Dougherty(ed.), Optical Methods for Tumor Treatment and Detection: Mechanisms and Techniquesin Photodynamic Therapy. Proc. Int. Soc. Optical Eng., I645: 43-51, 1992.

28. Fowler, G. J., Rees, R. C., and Devonshire, R. The photokilling of bladder carcinomacells in vitro by phenothiazine dyes. Photochem. Photobiol., 52: 489-494, 1990.

3084



1995;55:3078-3084. Cancer Res K. Thomas Moesta, William R. Greco, Stephen O. Nurse-Finlay, et al.

in VitroPhotodynamic Therapy of Pancreatic Adenocarcinoma Lack of Reciprocity in Drug and Light Dose Dependence of

Updated version

http://cancerres.aacrjournals.org/content/55/14/3078

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/55/14/3078To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



lack of reciprocity in drug and light dose dependence of ... · photodynamic therapy if differences...

Documents