0360.3016190 $3.00 + .OO Copyright 0 1990 Pergamon Press plc
In1 J Radmmn Onmlo~y Biol Phys Vol. 19,pp. 633-642 Pnnted I” the U.S.A. All rights reserved.
??Original Contribution
TARGETING AND THERAPY OF HUMAN GLIOMA XENOGRAFTS IN VZVO USING RADIOLABELED ANTIBODIES JEFFERY A. WILLIAMS, LARRY E. DILLEHAY,
M.D.,’ JAMES A. EDWARDS, B.S.,’ BARRY W. WESSELS, PH.D.,~ PH.D.,’ PHILIP M. WANEK,~ J. KENNETH
MOODY D. WHARAM,
M.D.,’
STANLEY E. ORDER, M.D.,
POGGENBURG, Sc.D.,
PH.D.,~
F.A.C.R.’
AND JERRY L. KLEIN, PH.D.’ ‘Division of Radiation Oncology, The Johns Hopkins Oncology Center, 600 North Wolfe Street, Baltimore, MD 2 1205; *Division of Radiation Oncology and Biophysics, The George Washington University Medical Center, 901 23rd Street, N.W., Washington, D.C. 20037; and 3Hybritech, Incorporated, San Diego, CA 92121 Radiolabeled antibodies provide a potential basis for selective radiotherapy of human gliomas. Monoclonal F’96.5, a mouse IgGta immunoglobulin, defines an epitope of a human melanoma cell surface protein and specifically binds the U-251 human glioma as measured by immunoperoxidase histochemistry. “‘In-radiolabeled P96.5 specifically targets the U-251 human glioma xenograft and yields 87.0 microcuries (uCi) of tumor activity per gram per 100 uCi injected activity compared to 4.5 uCi following administration of radiolabeled irrelevant monoclonal antibody. Calculations of targeting ratios demonstrate deposited dose to be 11.6 times greater with raciiolabeled F’96.5 administration compared to irrelevant monoclonal antibody. Tumor dose found in normal organs is less than 20% of the tumor dose, further supporting specific targeting of the human glioma xenograft by this antibody. Monoclonal antibodies QCIO54 and ZME018, which define a tumor-associated and a second melanoma-associated antigen, respectively, demonstrate positive immunoperoxidase staining of the tumor, but comparatively decreased targeting. To test the therapeutic potential of “Y-radiolabeled P96.5, QCIO54, and ZMEO18, tumors and normal sites were implanted with miniature thermoluminescent dosimeters (TLD). Average absorbed doses of 3770 + 445 (mean + SEM), 2043 + 134, and 645 + 48 cGy in tumor, 353 + 41,243 k 22, and 222 f 13 cGy in a contralateral control intramuscular site, 980 * 127,815 k 41, and 651 k 63 cGy in liver, and 275 + 14,263 + 11, and 256 + 18 cGy in total body were observed 7 days following administration of 100 uCi g”Y-radiolabeIed P96.5, QCIO54, and ZMEOl8, respectively. To test the therapeutic potential, tumor-bearing nude mice were given intracardiac injections of either buffer or WY-radiolabeled P96.5, QCI054, or ZME018. Striking tumor regression and prolonged survival were measured following administration of goY-Iabeled P96.5. Average maximal decreases in tumor volume were 42.7 it 11.9 and 94.2 -t 3.3 percent 28 and 58 days following 100 and 200 uCi WY-radioIabeIed P96.5 administration, respectively. The time required to achieve four times the initial tumor volume was 6.1 _+0.9 days for buffer; 43 t 12 and 63 * 10 days for 50 and 100 uCi 90Y-radioIabeled P96.5; 7 + 2, 20 f 1, and 53 + 4 for 50, 100, and 200 uCi 90Y-radioIabeIed QCIO54, and 9 ? 1, 13 + 1, and 29 + 3 days for 50, 100, and 200 uCi “Y-radiolabeled ZMEOl8, respectively. Average tumor regrowth failed to occur 180 days following administration of 200 uCi 9oYlabeled P96.5. Targeting specificity via shared neuroectodermal antigens provides a basis for human glioma ra-
dioimmunotherapy.
Human glioma, Radioimmunotherapy,
Tumor localization, Thermoluminescent
INTRODUCI’ION
(44).
dosimetry.
The tolerance of adjacent normal brain limits the
total radiation dose that can be safely administered (38). Radiolabeled antibodies offer potential selectivity in the delivery of radiation to human neoplasms (11, l&29,30, 33, 37, 39-41). The role of radioimmunotherapy in the treatment of human gliomas, however, remains largely
Human malignant gliomas are rarely cured and conventional postoperative external beam radiotherapy only extends median survival (5 1). With higher radiation doses, median survival, but not long-term survival, is improved
Supported by The Preuss Foundation for Brain Tumor Research, Inc., Hybritech, Inc., The Children’s Cancer Foundation, Inc., and NIH Grant CA4379 1. Presented at the ASTRO 1989 meeting in San Francisco, CA. Accepted for publication 29 March 1990.
Reprint requests to: J. A. Williams, Division of Neurosurgery, The University of Oklahoma, Health Sciences Center South Pavilion, Room 4SP200, Oklahoma City, OK 73 190. Acknowledgements-We thank Ms. Alverta E. Fields for her assistance in the preparation of this manuscript and Ms. Marianna Zahurak for statistical assistance. 633
634
I. J. Radiation Oncology 0 Biology 0 Physics
unexplored (7, 11, 13, 19,20, 3 1, 32,48,49, 58, 59). The results of many studies using both monoclonal and polyclonal antibodies indicate shared antigenicity among neuroectodermally derived tissues including glioma, melanoma, neuroblastoma, and fetal brain (6, 8, 9, 14, 15, 19, 34, 45, 48, 54, 55). Radiolabeled antibodies directed against neuroectodermal antigens may provide differential delivery of radiation to gliomas and sparing of surrounding normal brain. P96.5, a murine monoclonal antibody of the lgG2a subclass, targets P97, a cell surface glycoprotein of molecular weight 97,000. Initially described in high concentrations in human melanomas (25), P97 has subsequently been found in a variety of cultured cells of neuroectoderma1 origin including glioblastoma, neuroblastoma, and retinoblastoma (9,26, 34, 58). Similarly, monoclonal antibody ZMEO 18 targets GP240, a second melanoma cell surface glycoprotein of molecular weight 240,000 (24). To the present, GP240 has not been assessed in human glial neoplasms. Monoclonal antibody QC10.54 targets ferritin, a tumor-associated protein found in several human neoplasms (3, 17, 23, 37, 43, 56) including neuroblastoma (24) a neuroectodermally derived tumor. Hence, a shared antigenic profile including P97, GP240, and ferritin may exist among neuroectodermally derived neoplasms. Radiolabeled P96.5 (42) ZME018 (50), and QC1054 (Order, S. E., personal communication Sept, 1988) have successfully targeted disparate human neoplasms in vivo. Despite the possible cross-reactivity with neuroectodermally derived neoplasms, the systematic analysis of these antibodies for possible human glioma immunotherapy has not been performed. We therefore tested histochemical binding, in vivo targeting, and therapeutic potential of these antibodies in a human glioma xenograft-nude mouse model.
METHODS
AND MATERIALS
Tumor cell line Human Grade IV glioma cell line, U-25 1, was obtained from the DCT Tumor Repository, National Cancer lnstitute, Frederick Cancer Research Facility, Frederick, Maryland. Cells were cultured in Dulbecco’s modified minimal essential medium/Ham’s F- 12 nutrient mixture containing 10% fetal bovine serum* and antibiotics (penicillin and streptomycin) in 12 X 80 mm plastic culture dishes.+ Cells were grown in a humidified incubator at 37°C and gassed with a mixture of 5% carbon dioxide and 95% air. Media was changed twice weekly and cells passaged at confluence with 0.5% trypsin.*
* GIBCO, Grand Island, NY. + Falcon Plastics, Cockeysville,
September 1990, Volume 19, Number 3
Nude mice Six-week-old male athymic nude mice were injected subcutaneously in the left posterior thigh with 5 X lo6 glioma cells. Tumors were visible 1 week after injection. All experiments were performed when tumors grew to 0.3-0.4 g. Immunoperoxidase staining lmmunoperoxidase staining of U-25 1 human glioma xenografts was performed using monoclonal antibodies P96.5, ZME018, QClO54, and control PAY276 as previously described (57). Antibody radiolabeling Monoclonal antibodies were labeled with “‘In (P96.5, ZME018, QClO54, PAY276) or 9oY (P96.5, QCl054, and ZMEO 18) by chelation using a proprietary procedure developed by Hybritech, Inc., at a labeling ratio of 5 mCi per mg of protein. Percent binding of isotope to antibody was determined by thin layer chromatography and exceeded 95% in each experiment. Assay of radiolabeled antibody distribution Tumor-bearing nude mice were injected with 100 uCi of radiolabeled antibody in a final volume of 0.2 ml by intracardiac injection and sacrificed 1, 2, 5, and 7 days following injection in groups of at least two animals per time point. Calibration of well counter. A known quantity of 9 or “‘In was placed in a radioisotopic calibrator (sensitivity to 1 uCi). Counts per minute from the well counter were plotted versus known activity (mCi) allowing direct conversion of experimental CPM to mCi. Expression of accumulated activity in organs and ratio calculation. The tumor and all organs were individually weighed (Metler Balance Model H5 1AR with 0.01 mg sensitivity).* Counts per minute were obtained from the well counter and then converted to uCi by a conversion factor as determined above. The counts per minute per gram of each organ and tumor were recorded. The resulting activity in the tumor and organs was then expressed as uCi/gm of tissue as a function of time. A single exponential fit to the curve of activity/gram versus time was calculated and a best fit determined by the method of least squares for each organ and each experiment. The effective half-life (Teff) and initial concentration (Ao) of antibody in tumor or organ were calculated. The total dose deposited in tissue is equal to the sum of the penetrating and nonpenetrating radiations. The majority of dose is deposited by the nonpenetrating component. Assuming a uniform isotropic model for the dis-
t: Metler Instrument MD.
Carp, Hightstown,
NJ.
635
Targeting and therapy of human glioma xenografts 0 J. A. WILLIAMSet al.
tribution of radiolabeled antibody, the total deposited dose for any organ resulting from the sum of nonpenetrating radiations (NP) accumulated over time is given by: D=
~.~~XAOXT,~XNP(CG~)
where D = Ao = Teff = NP =
total mean dose (cGy) initial concentration of antibody in tumor or organ effective half-life (days) sum of nonpenetrating radiations.
Although the NP component is difficult to assess in small organs, expressing the results as a dose ratio of specific/ nonspecific antibody cancels this term in the equation. Thus, the targeting ratio equals: Ao X T,R (specific) Ao X Tetf (nonspecific) in the tumor or normal organ in question. The percentage of the tumor dose found in normal organs was determined by normalizing the product of initial activity and effective half-life of accumulated radioactivity (Ao X Teff) in each normal organ to that in the tumor. Absorbed dose calculations I. In vivo thermoluminescent dosimetry. Paired miniature teflon-embedded CaS04:Dy thermoluminescent dosimeters (TLD) were implanted directly into the human glioma xenografts and dorsal interscapular subcutaneous sites, while individual TLD were implanted in liver and a contralateral control subcutaneous site via a 20 gauge needle. Each TLD was first cross-calibrated with known activities of 9oY as previously described (52). Mice bearing 0.3-0.4 gram tumors and dosimeters were injected with 100 uCi ‘oY-labeled P96.5, QCI054, or ZMEO 18 antibody as described above and sacrificed 7 days following injection. TLD were removed, rinsed in Radiac, and measured for absorbed radiation dose as follows: the TLD glow curve peak was integrated over 50 seconds with T, = 115°C and T2 = 275°C. Temperature ramping was 3.5”C per second with prereadout annealing cycle of 5 seconds. The major integration peak occurs at 220-240°C with a minor peak (5%) at 120°C for this TLD. 2. MZRD method. Calculations of absorbed dose (cGy) in tumor and liver were also obtained using the Medical Internal Radiation Dose (MIRD) method (16). The method uses maximal initial activity (Ao:(uCi/gm) and effective radiolabeled antibody half-life (Tef) (days) calculated as described above, an equilibrium dose-rate constant of 1.98 (gm-cGy)/(uCi)/hr) for 9v’6 and an infinite
volume boundary correction factor of 0.4 (35, 53). Thus: Dose (cGy) = Ao (uCi/gm) X 1.44 X 24 X Teff (gm-cGy)/(uCi-hr)
X 0.4.
9oY-radiolabeled antibody therapy Groups of at least 10 tumor-bearing animals received either 2% bovine serum albumin in phosphate-buffered saline (PBS), unlabeled P96.5, or 50, 100, or 200 uCi 9oYradiolabeled P96.5 (therapy experiment l), or 50, 100, or 200 uCi “Y-radiolabeled QCI054 (therapy experiment 2) or 50, 100, or 200 uCi “Y-radiolabeled ZMEO18 (therapy experiment 3) in a final volume of 0.2 ml by intracardiac injection. Tumor volume was calculated by the formula: Tumor Volume = 4/3 X Pi X (L X W X H)/2 where L, W, and H are tumor length, width, and height, respectively. The change in tumor volume with time was calculated as the ratio of tumor volume (V) to its initial volume (V,). The logarithm of the ratio of average tumor volume to average initial tumor volume was plotted versus time. Initially, this ratio is unity and the logarithm of this ratio is zero. Increases or decreases above or below this baseline value indicate tumor growth or regression over time. Tumor regression. Tumor regression is defined as any decrease in an individual tumor volume below its initial value (V/V, < I). Tumor maximal volume reduction. The maximal reduction (percent) in tumor volume is the average V/V, X 100 at the nadir of the log V/V0 versus time plot. Tumor growth delay. Tumor growth delay is the time (days) required for a four-fold increase in average tumor volume (V/V, > 4) following a given treatment. Survival. Tumor-bearing animals were checked daily for survival. Kaplan-Meier (28) survival curves were plotted to demonstrate observed survival differences among groups of treated and control tumor-bearing animals. The survival difference between the buffer control group and the unlabeled (cold) P96.5 control group in Therapy Experiment 1 was tested using the log-rank statistic (36). To evaluate group survival differences within each therapy experiment and across both therapy experiments, the Cox regression method ( 12) was used. RESULTS Immunoperoxidase staining Positive immunoperoxidase staining of fixed U-25 1 tumors was demonstrated with monoclonal antibodies P96.5, ZMEO 18, and QCI054. No staining was seen with irrelevant monoclonal antibody PAY276. These results demonstrated the presence of the antigens P97, GP240,
I. J. Radiation Oncology 0 Biology0 Physics
636
September 1990, Volume 19, Number 3 Table 2. Absorbed dose-MIRD calculation vs. TLD injected activity 100 uCi “Y-radiolabeled antibody
I
P9s.5
Es3
mEO18
m
cc1054
m
PAY276
Tumor P96.5 QUO54 ZME018 Liver P96.5 QC1054 ZMEO18 AOR BRAN
LIVER
SFIXEN KDNY
LUNG
FEMUR HEART MUSCLE
Organ
Fig. 1. Comparison of maximum activity (Ao) (uCi/gram) deposited in the tumor and normal organs by “‘In-radiolabeled monoclonal antibodies following injection of 100 uCi of radiolabeled antibody: Percent injected dose (% I.D.) per gram equals Ao (uCi/gram)/(injected dose (uCi)) X 100.
and ferritin in the U-251 human glioma xenograft and provided a basis for radiolabeled antibody therapy studies. U-251 human glioma initial accumulated activity (Ao) The initial accumulated activity (Ao) in the glioma xenograft is 87.0 uCi/gm following ’’‘In-radiolabeled monoclonal antibody P96.5 administration (Fig. 1). This value exceeds initial accumulated activities of control PAY276 monoclonal antibody (4.5 uCi/gm) and demonstrates greater accumulation than both ZMEO 18 (22.0 uCi/gm) and QCI054 (5.1 uCi/gm) monoclonal antibodies. U-251 human glioma targeting ratios The targeting ratio is the quotient of absorbed dose (cGy) in tumor or normal organs following specific antibody administration, and the absorbed dose (cGy) in tumor or normal organs following nonspecific antibody administration. The dose deposited in tumor by “‘Inradiolabeled P96.5 is 11.6 times greater than control PAY276, 10.4 times greater than QC1054, and 5.7 times greater than ZMEO 18.
Table 1. Thermoluminescent dosimetry absorbed dose (cGy) following 100 &I “‘Y-labeled P96.5, QCI054, or ZMEO 18 administration Average dose* P96.5
Site Tumor Liver Contralateral Total body
control
3770 980 353 275
f 445 f 127 + 41 + 14
QCI054
ZMEO 18
2043 + 134 815 f 41 243+ 22 263+ 11
645 f 48 651 +63 222+ I3 256 + 18
* Absorbed dose (cGy) mean + S.E.M.
Ao (uCi/gm)
T1/2 elf(D)
MIRD* (rad)
TLD (rad)+
37.8 16.9 8.4
3.6 2.8 2.4
3725 1303 548
3770 + 445 2043 k 134 645 f 48
16.2 12.9 12.2
1.9 2.2 2.2
844 776 745
980 + 127 815 f 41 651 f 63
* MIRD formalism of absorbed dose calculation as described in text. + Mean f S.E.M.
U-251 human glioma percent dose (tumor) Dose to normal organs, when calculated as a percentage of the dose to the tumor, defines the therapeutic ratio. With P96.5 administration, the doses to normal organs
are 20% or less of that in tumor. In contrast, doses to spleen and all remaining normal organs are a higher percentage of tumor dose following ZMEO18 and QC1054 administration. The superior maximal initial activity in tumor (Ao) and tumor targeting ratio of “‘In-radiolabeled P96.5, and the low percentages of absorbed dose in normal organs suggested comparison of the therapeutic potentials of 9%radiolabeled P96.5, QC1054, and ZMEO 18. In vivo thermoluminescent dosimetry Direct measurement of absorbed radiation dose by microthermoluminescent dosimetry revealed increased average absorbed doses in tumor, and moderate absorbed doses in liver, in contralateral intramuscular control sites, and in total body sites following administration of 100 uCi “Y-radiolabeled P96.5 compared to QC1054 and ZME018 (Table 1). MIRD dosimetry Thermoluminescent dosimeter readings were confirmed by the Medical Internal Radiation Dose formalism (Table 2). Human glioma xenograft therapy Following intracardiac administration of buffer or unlabeled P96.5, no tumors regressed and the average growth delays (days) to four times initial tumor volume were 6.1 f 0.9 and 6.9 f 0.6, respectively (Table 3 and Fig. 2). Following administration of 50 or 100 uCi “Y-radiolabeled P96.5, the proportion of tumor regressing was l/ 12 and 9/10, the average percentage decrease in tumor volume was 0 and 43 -t 12, and the average growth delay to four times initial tumor volume was 42.7 f 11.9 and 62.6
637
Targeting and therapy of human glioma xenografts 0 J. A. WILLIAMS etal. Table 3. “Y-radiolabeled
P96.5, QCI054, and ZMEO18 human glioma xenograft therapy
Administered activity (uCi)
Proportion of tumors regressing
Average decrease in tumor volume (%)
Time to 4X tumor volume (days)
Median survival (days)
0 0 50 100 200
o/11 o/11 l/2 9110 12/12
0 0 0 43 f 12* 94k 3
6.1 k 0.9 6.9 f 0.6 42.9 f 11.7 62.6 + 9.7 NR+
46 46 81 90 NR+
50 100 200
o/10 o/10 7110
0 0
50 100 200
o/10 l/10 2110
Therapy experiment 1 Buffer Unlabeled P96.5 “Y-labeled P96.5
Therapy experiment 2 “Y-labeled QC1054
Therapy experiment 3 “Y-labeled ZMEO 18
37+
13 0 0 0
7.0 + 2 20.0 rt 1 53.0 t 4
48 66 80
9.0 -t 13.0 f 29.0 f
55 59 72
1.0 1.0 3.0
* Mean -t standard error of mean. ’ NR 4X original tumor volume (4 X Vo) and median survival not reached 180 days after radiolabeled antibody administration.
+ 9.7 days, respectively. Following administration of 200 uCi of “Y-radiolabeled P96.5, 12/ 12 tumors regressed an average of 94 + 3% in volume, and regrowth to initial average volume was not achieved 180 days after injection. Following administration of 50, 100, and 200 uCi 9oYradiolabeled QC1054 O/ 10, O/10, and 7/ 10 tumors regressed 0,0,and 37 + 13% in average tumor volume, and the times to four times initial tumor volume were 7 + 2, 20 + 1, and 53 + 4 days, respectively (Table 3, Fig. 3). In contrast, following administration of 50, 100, or 200 uCi “Y-radiolabeled ZMEO 18, O/IO, 1/ 10, and 2/ 10 tumors regressed, no average decrease in tumor volume was measured, and the times to four times initial tumor volume were 9 t 1, 13 -+ 1, and 29 & 3 days, respectively (Table 3 and Fig. 4).
1.6 ^ 1.4 -
.
12I08-
B $
;.06
.
Control
..mm.
50 uCi ,,,,++,+t* * -1"
.. ,,+'*
??
.
B04m i
. .
*t
,’
,,,."00@~~ 100 uCi
Survival in the two control groups (buffer and unlabeled P96.5) in Therapy Experiment 1 was not significantly different (p = .933) and for the subsequent Cox regression model within this experiment they were combined. Results of this model in Therapy Experiment 1 are shown in Figure 5. Increasing administered activities of “Y-radiolabeled P96.5 result in proportionately and significantly improved survival. The probability of survival 177 days following administration of 200 uCi “Y-radiolabeled P96.5 is 65%. The relative risk is the risk of death in a treated group when compared to the risk of death in the reference control group. Following 200 uCi “Y-radiolabeled P96.5 treatment, the relative risk is 0.018. In Therapy Experiment 2, administered activities of 50 uCi “Y-radiolabeled QC1054 did not cause significant
5 ? > g
“-
”
In1 J Radmmn Onmlo~y Biol Phys Vol. 19,pp. 633-642 Pnnted I” the U.S.A. All rights reserved.
??Original Contribution
TARGETING AND THERAPY OF HUMAN GLIOMA XENOGRAFTS IN VZVO USING RADIOLABELED ANTIBODIES JEFFERY A. WILLIAMS, LARRY E. DILLEHAY,
M.D.,’ JAMES A. EDWARDS, B.S.,’ BARRY W. WESSELS, PH.D.,~ PH.D.,’ PHILIP M. WANEK,~ J. KENNETH
MOODY D. WHARAM,
M.D.,’
STANLEY E. ORDER, M.D.,
POGGENBURG, Sc.D.,
PH.D.,~
F.A.C.R.’
AND JERRY L. KLEIN, PH.D.’ ‘Division of Radiation Oncology, The Johns Hopkins Oncology Center, 600 North Wolfe Street, Baltimore, MD 2 1205; *Division of Radiation Oncology and Biophysics, The George Washington University Medical Center, 901 23rd Street, N.W., Washington, D.C. 20037; and 3Hybritech, Incorporated, San Diego, CA 92121 Radiolabeled antibodies provide a potential basis for selective radiotherapy of human gliomas. Monoclonal F’96.5, a mouse IgGta immunoglobulin, defines an epitope of a human melanoma cell surface protein and specifically binds the U-251 human glioma as measured by immunoperoxidase histochemistry. “‘In-radiolabeled P96.5 specifically targets the U-251 human glioma xenograft and yields 87.0 microcuries (uCi) of tumor activity per gram per 100 uCi injected activity compared to 4.5 uCi following administration of radiolabeled irrelevant monoclonal antibody. Calculations of targeting ratios demonstrate deposited dose to be 11.6 times greater with raciiolabeled F’96.5 administration compared to irrelevant monoclonal antibody. Tumor dose found in normal organs is less than 20% of the tumor dose, further supporting specific targeting of the human glioma xenograft by this antibody. Monoclonal antibodies QCIO54 and ZME018, which define a tumor-associated and a second melanoma-associated antigen, respectively, demonstrate positive immunoperoxidase staining of the tumor, but comparatively decreased targeting. To test the therapeutic potential of “Y-radiolabeled P96.5, QCIO54, and ZMEO18, tumors and normal sites were implanted with miniature thermoluminescent dosimeters (TLD). Average absorbed doses of 3770 + 445 (mean + SEM), 2043 + 134, and 645 + 48 cGy in tumor, 353 + 41,243 k 22, and 222 f 13 cGy in a contralateral control intramuscular site, 980 * 127,815 k 41, and 651 k 63 cGy in liver, and 275 + 14,263 + 11, and 256 + 18 cGy in total body were observed 7 days following administration of 100 uCi g”Y-radiolabeIed P96.5, QCIO54, and ZMEOl8, respectively. To test the therapeutic potential, tumor-bearing nude mice were given intracardiac injections of either buffer or WY-radiolabeled P96.5, QCI054, or ZME018. Striking tumor regression and prolonged survival were measured following administration of goY-Iabeled P96.5. Average maximal decreases in tumor volume were 42.7 it 11.9 and 94.2 -t 3.3 percent 28 and 58 days following 100 and 200 uCi WY-radioIabeIed P96.5 administration, respectively. The time required to achieve four times the initial tumor volume was 6.1 _+0.9 days for buffer; 43 t 12 and 63 * 10 days for 50 and 100 uCi 90Y-radioIabeled P96.5; 7 + 2, 20 f 1, and 53 + 4 for 50, 100, and 200 uCi 90Y-radioIabeIed QCIO54, and 9 ? 1, 13 + 1, and 29 + 3 days for 50, 100, and 200 uCi “Y-radiolabeled ZMEOl8, respectively. Average tumor regrowth failed to occur 180 days following administration of 200 uCi 9oYlabeled P96.5. Targeting specificity via shared neuroectodermal antigens provides a basis for human glioma ra-
dioimmunotherapy.
Human glioma, Radioimmunotherapy,
Tumor localization, Thermoluminescent
INTRODUCI’ION
(44).
dosimetry.
The tolerance of adjacent normal brain limits the
total radiation dose that can be safely administered (38). Radiolabeled antibodies offer potential selectivity in the delivery of radiation to human neoplasms (11, l&29,30, 33, 37, 39-41). The role of radioimmunotherapy in the treatment of human gliomas, however, remains largely
Human malignant gliomas are rarely cured and conventional postoperative external beam radiotherapy only extends median survival (5 1). With higher radiation doses, median survival, but not long-term survival, is improved
Supported by The Preuss Foundation for Brain Tumor Research, Inc., Hybritech, Inc., The Children’s Cancer Foundation, Inc., and NIH Grant CA4379 1. Presented at the ASTRO 1989 meeting in San Francisco, CA. Accepted for publication 29 March 1990.
Reprint requests to: J. A. Williams, Division of Neurosurgery, The University of Oklahoma, Health Sciences Center South Pavilion, Room 4SP200, Oklahoma City, OK 73 190. Acknowledgements-We thank Ms. Alverta E. Fields for her assistance in the preparation of this manuscript and Ms. Marianna Zahurak for statistical assistance. 633
634
I. J. Radiation Oncology 0 Biology 0 Physics
unexplored (7, 11, 13, 19,20, 3 1, 32,48,49, 58, 59). The results of many studies using both monoclonal and polyclonal antibodies indicate shared antigenicity among neuroectodermally derived tissues including glioma, melanoma, neuroblastoma, and fetal brain (6, 8, 9, 14, 15, 19, 34, 45, 48, 54, 55). Radiolabeled antibodies directed against neuroectodermal antigens may provide differential delivery of radiation to gliomas and sparing of surrounding normal brain. P96.5, a murine monoclonal antibody of the lgG2a subclass, targets P97, a cell surface glycoprotein of molecular weight 97,000. Initially described in high concentrations in human melanomas (25), P97 has subsequently been found in a variety of cultured cells of neuroectoderma1 origin including glioblastoma, neuroblastoma, and retinoblastoma (9,26, 34, 58). Similarly, monoclonal antibody ZMEO 18 targets GP240, a second melanoma cell surface glycoprotein of molecular weight 240,000 (24). To the present, GP240 has not been assessed in human glial neoplasms. Monoclonal antibody QC10.54 targets ferritin, a tumor-associated protein found in several human neoplasms (3, 17, 23, 37, 43, 56) including neuroblastoma (24) a neuroectodermally derived tumor. Hence, a shared antigenic profile including P97, GP240, and ferritin may exist among neuroectodermally derived neoplasms. Radiolabeled P96.5 (42) ZME018 (50), and QC1054 (Order, S. E., personal communication Sept, 1988) have successfully targeted disparate human neoplasms in vivo. Despite the possible cross-reactivity with neuroectodermally derived neoplasms, the systematic analysis of these antibodies for possible human glioma immunotherapy has not been performed. We therefore tested histochemical binding, in vivo targeting, and therapeutic potential of these antibodies in a human glioma xenograft-nude mouse model.
METHODS
AND MATERIALS
Tumor cell line Human Grade IV glioma cell line, U-25 1, was obtained from the DCT Tumor Repository, National Cancer lnstitute, Frederick Cancer Research Facility, Frederick, Maryland. Cells were cultured in Dulbecco’s modified minimal essential medium/Ham’s F- 12 nutrient mixture containing 10% fetal bovine serum* and antibiotics (penicillin and streptomycin) in 12 X 80 mm plastic culture dishes.+ Cells were grown in a humidified incubator at 37°C and gassed with a mixture of 5% carbon dioxide and 95% air. Media was changed twice weekly and cells passaged at confluence with 0.5% trypsin.*
* GIBCO, Grand Island, NY. + Falcon Plastics, Cockeysville,
September 1990, Volume 19, Number 3
Nude mice Six-week-old male athymic nude mice were injected subcutaneously in the left posterior thigh with 5 X lo6 glioma cells. Tumors were visible 1 week after injection. All experiments were performed when tumors grew to 0.3-0.4 g. Immunoperoxidase staining lmmunoperoxidase staining of U-25 1 human glioma xenografts was performed using monoclonal antibodies P96.5, ZME018, QClO54, and control PAY276 as previously described (57). Antibody radiolabeling Monoclonal antibodies were labeled with “‘In (P96.5, ZME018, QClO54, PAY276) or 9oY (P96.5, QCl054, and ZMEO 18) by chelation using a proprietary procedure developed by Hybritech, Inc., at a labeling ratio of 5 mCi per mg of protein. Percent binding of isotope to antibody was determined by thin layer chromatography and exceeded 95% in each experiment. Assay of radiolabeled antibody distribution Tumor-bearing nude mice were injected with 100 uCi of radiolabeled antibody in a final volume of 0.2 ml by intracardiac injection and sacrificed 1, 2, 5, and 7 days following injection in groups of at least two animals per time point. Calibration of well counter. A known quantity of 9 or “‘In was placed in a radioisotopic calibrator (sensitivity to 1 uCi). Counts per minute from the well counter were plotted versus known activity (mCi) allowing direct conversion of experimental CPM to mCi. Expression of accumulated activity in organs and ratio calculation. The tumor and all organs were individually weighed (Metler Balance Model H5 1AR with 0.01 mg sensitivity).* Counts per minute were obtained from the well counter and then converted to uCi by a conversion factor as determined above. The counts per minute per gram of each organ and tumor were recorded. The resulting activity in the tumor and organs was then expressed as uCi/gm of tissue as a function of time. A single exponential fit to the curve of activity/gram versus time was calculated and a best fit determined by the method of least squares for each organ and each experiment. The effective half-life (Teff) and initial concentration (Ao) of antibody in tumor or organ were calculated. The total dose deposited in tissue is equal to the sum of the penetrating and nonpenetrating radiations. The majority of dose is deposited by the nonpenetrating component. Assuming a uniform isotropic model for the dis-
t: Metler Instrument MD.
Carp, Hightstown,
NJ.
635
Targeting and therapy of human glioma xenografts 0 J. A. WILLIAMSet al.
tribution of radiolabeled antibody, the total deposited dose for any organ resulting from the sum of nonpenetrating radiations (NP) accumulated over time is given by: D=
~.~~XAOXT,~XNP(CG~)
where D = Ao = Teff = NP =
total mean dose (cGy) initial concentration of antibody in tumor or organ effective half-life (days) sum of nonpenetrating radiations.
Although the NP component is difficult to assess in small organs, expressing the results as a dose ratio of specific/ nonspecific antibody cancels this term in the equation. Thus, the targeting ratio equals: Ao X T,R (specific) Ao X Tetf (nonspecific) in the tumor or normal organ in question. The percentage of the tumor dose found in normal organs was determined by normalizing the product of initial activity and effective half-life of accumulated radioactivity (Ao X Teff) in each normal organ to that in the tumor. Absorbed dose calculations I. In vivo thermoluminescent dosimetry. Paired miniature teflon-embedded CaS04:Dy thermoluminescent dosimeters (TLD) were implanted directly into the human glioma xenografts and dorsal interscapular subcutaneous sites, while individual TLD were implanted in liver and a contralateral control subcutaneous site via a 20 gauge needle. Each TLD was first cross-calibrated with known activities of 9oY as previously described (52). Mice bearing 0.3-0.4 gram tumors and dosimeters were injected with 100 uCi ‘oY-labeled P96.5, QCI054, or ZMEO 18 antibody as described above and sacrificed 7 days following injection. TLD were removed, rinsed in Radiac, and measured for absorbed radiation dose as follows: the TLD glow curve peak was integrated over 50 seconds with T, = 115°C and T2 = 275°C. Temperature ramping was 3.5”C per second with prereadout annealing cycle of 5 seconds. The major integration peak occurs at 220-240°C with a minor peak (5%) at 120°C for this TLD. 2. MZRD method. Calculations of absorbed dose (cGy) in tumor and liver were also obtained using the Medical Internal Radiation Dose (MIRD) method (16). The method uses maximal initial activity (Ao:(uCi/gm) and effective radiolabeled antibody half-life (Tef) (days) calculated as described above, an equilibrium dose-rate constant of 1.98 (gm-cGy)/(uCi)/hr) for 9v’6 and an infinite
volume boundary correction factor of 0.4 (35, 53). Thus: Dose (cGy) = Ao (uCi/gm) X 1.44 X 24 X Teff (gm-cGy)/(uCi-hr)
X 0.4.
9oY-radiolabeled antibody therapy Groups of at least 10 tumor-bearing animals received either 2% bovine serum albumin in phosphate-buffered saline (PBS), unlabeled P96.5, or 50, 100, or 200 uCi 9oYradiolabeled P96.5 (therapy experiment l), or 50, 100, or 200 uCi “Y-radiolabeled QCI054 (therapy experiment 2) or 50, 100, or 200 uCi “Y-radiolabeled ZMEO18 (therapy experiment 3) in a final volume of 0.2 ml by intracardiac injection. Tumor volume was calculated by the formula: Tumor Volume = 4/3 X Pi X (L X W X H)/2 where L, W, and H are tumor length, width, and height, respectively. The change in tumor volume with time was calculated as the ratio of tumor volume (V) to its initial volume (V,). The logarithm of the ratio of average tumor volume to average initial tumor volume was plotted versus time. Initially, this ratio is unity and the logarithm of this ratio is zero. Increases or decreases above or below this baseline value indicate tumor growth or regression over time. Tumor regression. Tumor regression is defined as any decrease in an individual tumor volume below its initial value (V/V, < I). Tumor maximal volume reduction. The maximal reduction (percent) in tumor volume is the average V/V, X 100 at the nadir of the log V/V0 versus time plot. Tumor growth delay. Tumor growth delay is the time (days) required for a four-fold increase in average tumor volume (V/V, > 4) following a given treatment. Survival. Tumor-bearing animals were checked daily for survival. Kaplan-Meier (28) survival curves were plotted to demonstrate observed survival differences among groups of treated and control tumor-bearing animals. The survival difference between the buffer control group and the unlabeled (cold) P96.5 control group in Therapy Experiment 1 was tested using the log-rank statistic (36). To evaluate group survival differences within each therapy experiment and across both therapy experiments, the Cox regression method ( 12) was used. RESULTS Immunoperoxidase staining Positive immunoperoxidase staining of fixed U-25 1 tumors was demonstrated with monoclonal antibodies P96.5, ZMEO 18, and QCI054. No staining was seen with irrelevant monoclonal antibody PAY276. These results demonstrated the presence of the antigens P97, GP240,
I. J. Radiation Oncology 0 Biology0 Physics
636
September 1990, Volume 19, Number 3 Table 2. Absorbed dose-MIRD calculation vs. TLD injected activity 100 uCi “Y-radiolabeled antibody
I
P9s.5
Es3
mEO18
m
cc1054
m
PAY276
Tumor P96.5 QUO54 ZME018 Liver P96.5 QC1054 ZMEO18 AOR BRAN
LIVER
SFIXEN KDNY
LUNG
FEMUR HEART MUSCLE
Organ
Fig. 1. Comparison of maximum activity (Ao) (uCi/gram) deposited in the tumor and normal organs by “‘In-radiolabeled monoclonal antibodies following injection of 100 uCi of radiolabeled antibody: Percent injected dose (% I.D.) per gram equals Ao (uCi/gram)/(injected dose (uCi)) X 100.
and ferritin in the U-251 human glioma xenograft and provided a basis for radiolabeled antibody therapy studies. U-251 human glioma initial accumulated activity (Ao) The initial accumulated activity (Ao) in the glioma xenograft is 87.0 uCi/gm following ’’‘In-radiolabeled monoclonal antibody P96.5 administration (Fig. 1). This value exceeds initial accumulated activities of control PAY276 monoclonal antibody (4.5 uCi/gm) and demonstrates greater accumulation than both ZMEO 18 (22.0 uCi/gm) and QCI054 (5.1 uCi/gm) monoclonal antibodies. U-251 human glioma targeting ratios The targeting ratio is the quotient of absorbed dose (cGy) in tumor or normal organs following specific antibody administration, and the absorbed dose (cGy) in tumor or normal organs following nonspecific antibody administration. The dose deposited in tumor by “‘Inradiolabeled P96.5 is 11.6 times greater than control PAY276, 10.4 times greater than QC1054, and 5.7 times greater than ZMEO 18.
Table 1. Thermoluminescent dosimetry absorbed dose (cGy) following 100 &I “‘Y-labeled P96.5, QCI054, or ZMEO 18 administration Average dose* P96.5
Site Tumor Liver Contralateral Total body
control
3770 980 353 275
f 445 f 127 + 41 + 14
QCI054
ZMEO 18
2043 + 134 815 f 41 243+ 22 263+ 11
645 f 48 651 +63 222+ I3 256 + 18
* Absorbed dose (cGy) mean + S.E.M.
Ao (uCi/gm)
T1/2 elf(D)
MIRD* (rad)
TLD (rad)+
37.8 16.9 8.4
3.6 2.8 2.4
3725 1303 548
3770 + 445 2043 k 134 645 f 48
16.2 12.9 12.2
1.9 2.2 2.2
844 776 745
980 + 127 815 f 41 651 f 63
* MIRD formalism of absorbed dose calculation as described in text. + Mean f S.E.M.
U-251 human glioma percent dose (tumor) Dose to normal organs, when calculated as a percentage of the dose to the tumor, defines the therapeutic ratio. With P96.5 administration, the doses to normal organs
are 20% or less of that in tumor. In contrast, doses to spleen and all remaining normal organs are a higher percentage of tumor dose following ZMEO18 and QC1054 administration. The superior maximal initial activity in tumor (Ao) and tumor targeting ratio of “‘In-radiolabeled P96.5, and the low percentages of absorbed dose in normal organs suggested comparison of the therapeutic potentials of 9%radiolabeled P96.5, QC1054, and ZMEO 18. In vivo thermoluminescent dosimetry Direct measurement of absorbed radiation dose by microthermoluminescent dosimetry revealed increased average absorbed doses in tumor, and moderate absorbed doses in liver, in contralateral intramuscular control sites, and in total body sites following administration of 100 uCi “Y-radiolabeled P96.5 compared to QC1054 and ZME018 (Table 1). MIRD dosimetry Thermoluminescent dosimeter readings were confirmed by the Medical Internal Radiation Dose formalism (Table 2). Human glioma xenograft therapy Following intracardiac administration of buffer or unlabeled P96.5, no tumors regressed and the average growth delays (days) to four times initial tumor volume were 6.1 f 0.9 and 6.9 f 0.6, respectively (Table 3 and Fig. 2). Following administration of 50 or 100 uCi “Y-radiolabeled P96.5, the proportion of tumor regressing was l/ 12 and 9/10, the average percentage decrease in tumor volume was 0 and 43 -t 12, and the average growth delay to four times initial tumor volume was 42.7 f 11.9 and 62.6
637
Targeting and therapy of human glioma xenografts 0 J. A. WILLIAMS etal. Table 3. “Y-radiolabeled
P96.5, QCI054, and ZMEO18 human glioma xenograft therapy
Administered activity (uCi)
Proportion of tumors regressing
Average decrease in tumor volume (%)
Time to 4X tumor volume (days)
Median survival (days)
0 0 50 100 200
o/11 o/11 l/2 9110 12/12
0 0 0 43 f 12* 94k 3
6.1 k 0.9 6.9 f 0.6 42.9 f 11.7 62.6 + 9.7 NR+
46 46 81 90 NR+
50 100 200
o/10 o/10 7110
0 0
50 100 200
o/10 l/10 2110
Therapy experiment 1 Buffer Unlabeled P96.5 “Y-labeled P96.5
Therapy experiment 2 “Y-labeled QC1054
Therapy experiment 3 “Y-labeled ZMEO 18
37+
13 0 0 0
7.0 + 2 20.0 rt 1 53.0 t 4
48 66 80
9.0 -t 13.0 f 29.0 f
55 59 72
1.0 1.0 3.0
* Mean -t standard error of mean. ’ NR 4X original tumor volume (4 X Vo) and median survival not reached 180 days after radiolabeled antibody administration.
+ 9.7 days, respectively. Following administration of 200 uCi of “Y-radiolabeled P96.5, 12/ 12 tumors regressed an average of 94 + 3% in volume, and regrowth to initial average volume was not achieved 180 days after injection. Following administration of 50, 100, and 200 uCi 9oYradiolabeled QC1054 O/ 10, O/10, and 7/ 10 tumors regressed 0,0,and 37 + 13% in average tumor volume, and the times to four times initial tumor volume were 7 + 2, 20 + 1, and 53 + 4 days, respectively (Table 3, Fig. 3). In contrast, following administration of 50, 100, or 200 uCi “Y-radiolabeled ZMEO 18, O/IO, 1/ 10, and 2/ 10 tumors regressed, no average decrease in tumor volume was measured, and the times to four times initial tumor volume were 9 t 1, 13 -+ 1, and 29 & 3 days, respectively (Table 3 and Fig. 4).
1.6 ^ 1.4 -
.
12I08-
B $
;.06
.
Control
..mm.
50 uCi ,,,,++,+t* * -1"
.. ,,+'*
??
.
B04m i
. .
*t
,’
,,,."00@~~ 100 uCi
Survival in the two control groups (buffer and unlabeled P96.5) in Therapy Experiment 1 was not significantly different (p = .933) and for the subsequent Cox regression model within this experiment they were combined. Results of this model in Therapy Experiment 1 are shown in Figure 5. Increasing administered activities of “Y-radiolabeled P96.5 result in proportionately and significantly improved survival. The probability of survival 177 days following administration of 200 uCi “Y-radiolabeled P96.5 is 65%. The relative risk is the risk of death in a treated group when compared to the risk of death in the reference control group. Following 200 uCi “Y-radiolabeled P96.5 treatment, the relative risk is 0.018. In Therapy Experiment 2, administered activities of 50 uCi “Y-radiolabeled QC1054 did not cause significant
5 ? > g
“-
”
