Radioimmunotherapy of human head and neck squamous cell carcinoma xenografts with 131I-labelled monoclonal antibody E48 IgG.

Monoclonal antibody (MAb) E48 reacts with a 22 kD antigen exclusively expressed in squamous and transitional epithelia and their neoplastic counterparts. Radiolabelled with 99mTc, MAb E48 is capable of targeting metastatic and recurrent disease in patients with head and neck cancer. In this study, the capacity of 131I-labelled MAb E48 to eradicate xenografts of human squamous cell carcinoma of the head and neck (HNSCC) in nude mice was examined. Experimental groups received a single i.v. bolus injection of 400 microCi MAb E48 IgG (number of mice (n = 6, number of tumours (t) = 9) or 800 microCi MAb E48 IgG (n) = 5,t = 7), whereas control groups received either diluent (n = 3,t = 5), unlabelled MAb E48 IgG (n = 4,t = 5) or 800 microCi 131I-labelled isotype-matched control MAb (n = 6,t = 9). A 4.1-fold increase in the median tumour volume doubling time and regression of two out of ten tumours (20%) was observed in mice treated with 400 microCi. In mice treated with 800 microCi. In mice treated with 800 microCi, two out of seven tumours (29%) showed complete remission without regrowth during follow-up (greater than 3 months). Median tumour volume doubling time in the remaining five tumours was increased 7.8-fold. No antitumour effects were observed in mice injected with diluent, unlabelled MAb E48 or 131I-labelled control MAb. In the same xenograft model, chemotherapy with doxorubicin, 5-fluorouracil, cisplatin, bleomycin, methotrexate or 2',2'-difluorodeoxycytidine yielded a less profound effect on tumour volume doubling time. Increases in tumour volume doubling time with these chemotherapeutic agents were 4, 2.2, 2.1, 1.7, 0, and 2.6 respectively. Moreover, no cures were observed with any of these chemotherapeutic agents. From the tissue distribution of 800 microCi MAb E48, the absorbed cumulative radiation doses of tumour and various organs were calculated using the trapezoid integration method for the area under the curve. To tumour xenografts, 12,170 cGy was delivered, blood received 2,984 cGy, whereas in every other tissue the accumulated dose was less than 6% of the dose delivered to tumour. These data, describing the first radiolabelled MAb with therapeutic efficacy against HNSCC, suggest radioimmunotherapy with MAb E48 to be a potential therapeutic modality for the treatment of head and neck cancer.

Despite an increase in the locoregional control of head and neck squamous cell carcinoma (HNSCC), due to improved surgery and radiotherapy, current therapy regimens have failed as yet to increase the 5-year survival rate of patients with head and neck cancer (Choksi et al., 1988;Cognetti et al., 1988). Whereas fewer patients tend to die because of uncontrolled locoregional disease, there is an increase in the number of distant metastases and second primary tumours. The role of chemotherapy in these patients is limited. Responses are often observed but enhancement of survival is not obtained. These facts justify the search for more specific and effective therapeutical methods. Since HNSCC have an intrinsic sensitivity for radiation (Wessels et al., 1989a), we focus on the use of monoclonal antibodies labelled with radioisotopes for radioimmunotherapy (RIT). RIT of human tumours in experimental and/or clinical settings has already been described for various types of cancer, including colorectal carcinomas (Esteban et al., 1990;Lee et al., 1990;Schlom et al., 1991;Blumenthal et al., 1991), malignant gliomas (Colapinto et al., 1990;Lee et al., 1988a;Lee et al., 1988b;Williams et al., 1990), ovarian carcinoma (Stewart et al., 1989;Ward et al., 1988), small cell lung carcinoma (Smith et al., 1990;Smith et al., 1991;Beaumier et al., 1991), mammary carcinoma (Senekowitsch et al., 1989), renal cell carcinoma (Wessels et al., 1989b;Chiou et al., 1988), and cutaneous T cell lymphoma (Rosen et al., 1987;Mulshine et al., 1991). Thusfar however, no HNSCC-specific MAbs have been available to test the efficacy of RIT to eradicate HNSCC xenografts in an experimental setting. Therefore, we have developed a panel of MAbs, among which MAb E48, raised against HNSCC (Quak et al., 1990a;Quak et al., 1990b;Quak et al., 1992). MAb E48 recognises a 20-22kD antigen, on normal tissues selectively expressed on stratified squamous epithelia and transitional epithelium of the bladder. On tumours, reactivity is restricted to malignancies arising from these tissues. The MAb E48 defined antigen is involved in the structural organisation of squamous epithelia, possibly at the level of cell-cell adhesion (Schrijvers et al., 1991). Biodistribution and imaging studies with tracer amounts of '3'I-labelled E48 IgG and F(ab')2 fragments already demonstrated the capacity of MAb E48 for specific delivery of radioisotope to HNSCC xenografts (Quak et al., 1989;Gerretsen et al., 1991). Recent data from an ongoing phase I/II trial with intraveneously administered 99'Tclabelled MAb E48 F(ab')2 and IgG in patients with HNSCC indicate that MAb E48 is highly capable of detecting metastatic and recurrent disease (van Dongen et al., 1992). In the present study we demonstrate a dose dependent growth delay, regression and complete remission of established tumours by injection of single doses "1'1-labelled MAb E48 in nude mice bearing HNSCC xenografts. In this experimental model, the efficacy of RIT was compared to the antitumour activity of a number of clinically used or experimental chemotherapeutic agents (Braakhuis et al., 1991).

Material and methods
HN was established by subcutaneous implantation of tumour fragments measuring 3 x 3 x 1 mm, in the lateral thoracic region on both sides of nude mice. Thereafter, the xenograft line was maintained by serial transplantation . The tumour from which the HNX-HN line originates was a T4N2MO squamous cell carcinoma of the base of the tongue from a 54-year-old female patient. As determined by indirect immunoperoxidase staining, the expression pattern of the MAb E48 defined antigen in the HNX-HN line was comparable to the pattern of the majority of human HNSCC tumours . During experiments, food and water, with potassium iodide added to the water to prevent thyroid accumulation of 131i, were available ad libitum.
Iodine-131 labelling lodination of MAb IgG was performed essentially as described earlier . MAb IgG in phosphate buffered saline, pH 7.4, and "3'I were mixed in a ratio of approximately 1 mg MAb: I0mCi 31I in a vial coated with lodogen (Pierce). After 10min incubation at room temperature, a sample was removed to determine the amount of incorporated iodine by TCA precipitation. To the reaction mixture, 1 ml AG1-X-8 resin (BioRad) in PBS, 1% BSA was added to absorb free iodine. To remove the resin and to sterilize the product, the reaction mixture was filtered through a 0.22;tm filter. Quality control of '3'I-labelled MAb E48 IgG After labelling, the immunoreactive fraction was at least 85% in all experiments. Incorporated '3'I was higher than 90% in all experiments as determined by TCA precipitation. Specific activity of the radioimmunoconjugate varied between 5 and 0mCi mg -'.
MAb IgG in vitro binding assay The binding characteristics of radiolabelled MAb E48 IgG were analysed in an immunoreactivity assay, essentially as described earlier . In short, cells of the squamous cell carcinoma cell line UM-SCC 22B, a gift from Dr T.E. Carey (Ann Arbor. MI, USA), were fixed in 1% paraformaldehyde and five serial dilutions, ranging from 5 x 10' cells/tube to 3.1 x 105 cells/tube, were made with 1% bovine serum albumin (BSA) in 10mM phosphate-buffered saline (PBS). To the tubes, 10,000cpm of the labelled MAb IgG was added and incubated 120min at room temperature. To a duplicate of the last sample, excess unlabelled MAb IgG was added to determine non-specific binding. Cells were spun down and radioactivity in the pellet and supernatant was determined in a gamma counter and the percentage bound and free radiolabelled MAb was calculated (LKB-Wallac 1218 CompuGamma). Data were graphically analysed in a modified Lineweaver-Burke plot and the immunoreactive fraction was determined by linear extrapolation to conditions representing infinite antigen excess.

Toxicity studies
The maximum tolerated dose of each of the chemotherapeutic agents, corresponding to a weight loss between 5 and 15%, was determined as described earlier Braakhuis et al., 1991). In the same way, the maximum tolerated dose of '31I-labelled MAb E48 was determined. Nude mice without xenografts were injected with diluent (PBS) or with increasing doses of "3'I-labelled MAb E48 IgG. Total body dose was determined in a dose calibrator. Total accumulated radiation dose was calculated as described in the section 'Dosimetry calculations'. The weight of the mice was measured daily over a period of 4 weeks, at which timepoint no radioactivity could be detected.

In vivo biodistribution studies
Biodistribution studies with tracer dose '31I-labelled MAb E48 IgG in nude mice bearing HNX-HN xenografts have previously been described Quak et al., 1989). To compare the biodistribution of a therapeutical dose with a tracer dose, 28 mice bearing xenografts of a size comparable with the tracer dose study were injected i.v. with 80011Ci 'I-labelled MAb E48 IgG. At the time of injection the estimated xenograft volume was 323 ± 244mm3 as determined by measuring the tumour in three dimensions with calipers ((L x W x H)/2) (versus 352 ± 207.5 in the tracer dose study). Mice were bled, killed and dissected 2, 5 and 8h and 1, 3, 7, 10, 14, 21, 28 and 35 days after i.v. injection. Organs were immediately removed, placed in 5 ml plastic tubes and weighed. Samples were taken from blood, urine, tumour, liver, spleen, kidney, heart, stomach, ileum, colon, bladder, sternum, muscle, lung, skin and tongue. After weighing, radioactivity in all organs and tumours was counted in a gamma counter. The antibody uptake in the tumour and other tissues was calculated as the percentage of the injected dose per gram of tissue (% ID.g-1).

Dosimetry calculations
Dosimetry calculations were performed using the data of the biodistribution of 800jaCi '3ll-labelled MAb E48 IgG. The absorbed cumulative radiation dose for tumour and various organs was calculated using the trapezoid integration method for the area under the curve (Badger et al., 1986). Due to the therapeutic effect of the dose, tumours at day 35 had almost completely regressed and were thus not included in dosimetry calculations. The final segment of the area under the curve was calculated based on the biological halflife: dose of last segment = dose previous segment (day 21-day 28) x 0.693.(tj in previous segment)-'. cGy were further calculated by multiplying the #iCi.h.g-' by the g.cGy.(pCi.h)-factor published by the Medical Internal Radiation Dose committee for "'lI of 0.4313 (Dilman, 1969). Chemotherapy All drugs were injected at the maximum tolerated dose level (5-15% weight loss). Schedules were based on results of experiments performed in previous studies Braakhuis et al., 1991). Mean number of mice and tumours in all schedules was 5 and 7, respectively. The volume of the tumours at the time of injection ranged between 50-150mm3. The following doses and injection schedules were applied: Doxorubicin (DOX, Farmitalia, Bournonville-Pharma, Almere, The Netherlands) at 8mgkg-' i.v. at day 0 and 8; dFdC (2',2'difluorodeoxycytidin, Gemcitabine, LY 188011, Lilly Research, Windlesham, Surrey, United Kingdom) at 120mgkg-' i.p. at day 0, 3, 6 and 9; 5-FU (Fluorouracil Roche, Hoffman-La Roche, Mijdrecht, The Netherlands) at 125mgkg-' i.p. at day 0 and 8; CCDP (Platinol, Bristol Meyers, Weesp, The Netherlands) at 5mgkg-' i.v. at day 0, 8 and 15; BLEO (Bleomycin, Lundbeck, Amsterdam, The Netherlands) at 15mgkg-' i.p. at day 0, 1, 2, 3 and 4, and methotrexate (Ledertrexate, Lederle, Etten-Leur, The Netherlands) at 1.8mgkg-1 i.p. at day 0, 1, 2, 3 and 4.

Evaluation of therapeutic efficacy
Tumour bearing mice were treated with RIT or chemotherapy when most tumours reached a volume of at least 50mm3 (range 50-250mm3). Tumours smaller than 50mm3 at the time of injection were not included in the determination of the tumour volume doubling time because of inaccuracy in measuring these tumours. Tumour growth was expressed as the tumour volume at each timepoint relative to the tumour volume at day 0. Efficacy of RIT as well as chemotherapy was expressed by means of the tumour growth delay factor (GDF), defined as (TD,-TDC)/ TDC(TDt = median tumour volume doubling time of treated mice, TD, = median tumour volume doubling time of control mice). Prolonged survival (survival defined as the time period between day 0 and the timepoint of sacrifice, being when tumour size exceeded 1000mm3) was determined by comparing experimental groups with treatment groups using the Mann-Whitney U-test.

Toxicity studies
The total cumulative whole body radiation dose for 220ptCi, and reached a maximum of 19.4 ± 2.9% IDg-' after 3 days. Activity is retained in the tumour up to 8.1 ± 3.5% IDg-' at day 28. No specific accumulation is observed in any other tissue.

Dosimetry calculations
The absorbed cumulative radiation dose for tumour and various organs is shown in Figure 3. Based on the area under the curve of the biodistribution data of 800pCi 31I-labelled MAb E48 IgG the absorbed radiation dose to tumours in the group receiving 800 Ci was 12,170cGy, whereas blood received only 2,984cGy. Other tissues received the following dose: lung; 662cGy; kidney; 607cGy; spleen: 581cGy; bladder: 571cGy; heart: 543cGy; colon: 424cGy; ileum: 405cGy; sternum: 405cGy; liver: 403cGy; muscle: 276cGy; stomach: 251 cGy.
Evaluation of therapeutic efficacy Tumour growth expressed as the tumour volume at each timepoint relative to the tumour volume at day 0 for control and treatment groups is shown in Figure 4. Tumours in the groups receiving unlabelled MAb E48 IgG (Figure 4a Cumulative radiation dose in cGy x 1000 Figure 3 Total accumulated radiation dose in I1&cGy, calculated using the trapezoid integration method for the area under the curve. Tu, tumour; Blo, blood; Bla, bladder; Lu, lung; Ki, kidney; Sp, spleen; He, heart; Li, liver; Co, colon; Sto, stomach; I1, ileum; Ste, sternum; Mu, muscle. Days after injection Figure 4 Effects of unlabelled MAb E48, n = 4, t = 5 a, '3'labelled control MAb, n = 6, It = 9 b, diluent, n = 3, t = 5 c, 4001Ci '3'I-labelled MAb E48, n = 6, t = 9 d, and 800tCi '3'Ilabelled MAb E48, n = 5, t = 7 e on the growth of HNX-HN xenografts, expressed as the tumour volume during therapy relative to the tumour volume at the start of the therapy. Mice were sacrificed when tumours exceeded 1000mm3, n = number of animals, t = number of tumours. * = complete remission without regrowth during follow up (>3 months). longed survival, as determined by the Mann-Whitney U-test, was significant for both RIT groups as compared to control groups (P <0.01).

Discussion
Therapeutic efficacy of radiolabelled MAbs in the nude mouse model has been described for several tumour types.
Although clinical radioimmunoscintigraphy studies for the detection of HNSCC have been reported with "'In-labelled anti-epidermal growth factor receptor (Soo et al., 1987) and with "'In-labelled anti-carinoembryonic antigen (Kairemo & Hopsu, 1990;Kairemo & Hopsu, 1990), no reports are available on therapy experiments of HNSCC xenografts with radiolabelled MAbs. Here we present the first data on RIT of HNSCC. As a first approach to assess the potential of radiolabelled MAb E48 in eradicating HNSCC xenografts, therapy experiments, consisting of single bolus injection of two different doses, were designed in a straight forward manner. Dosimetry calculations were based on the biodistribution of a therapeutic dose, since continued tumour growth in biodistribution experiments with tracer dose may well result in underestimation of the radiation dose up to 35-52% (Lee et al., 1990;Badger et al., 1986). In our studies, no differences in biodistribution between tracer and therapeutic doses were observed. In tracer dose studies however, no data were available during the first 12h and after day 7, whereas in this study data were obtained from 2h p.i. up to 35 days, allowing more accurate dosimetry calculations. A remarkable good retention of MAb E48 was observed with 8.1 ± 3.5%IDg-' at day 28 after injection.  (Sharkey et al., 1987). Lee et al. obtained apparent cures of mice with intracranial glioma xenografts after a single injection of 1.25mCi MAb IgG (Lee et al., 1988). Complete ablation of highly radiation sensitive neuroblastoma xenografts was achieved with a single injection of 1 mCi IgG by the group of Cheung et al. (Cheung et al., 1986). Buchegger et al. completely eradicated xenografts of colon carcinomas with single injections of 2,200-2,800.tCi, but instead of IgG, pooled F(ab')2 fragments of three different anti-CEA MAbs were used (Buchegger et al., 1989). Other successful studies applied fractionated protocols (Smith et al., 1991;Senekowitsch et al., 1989;Schlom et al., 1990; Buchegger et  . al., 1989). In our study, single injections of 400 or 800yCi "3'1-labelled E48 MAb IgG showed pronounced anti-tumour effects, resulting in complete remissions of two out of seven tumours in the group receiving 800fiCi "3'I-labelled MAb E48 IgG. No remnant tumour could be detected when mice were sacrificed after 3 months follow-up. These cures might very well be due to the intrinsic sensitivity of head and neck tumours for radiation (Wessels et al., 1989a). In addition, the accumulated dose, 12,170cGy, in tumour tissue as a result of a single bolus injection 800yCi was very high, reflecting the excellent targetting and retention characteristics of MAb E48 in this experimental model. Therapeutic efficacy of RIT has been found to be inversely correlated with tumour size (Scholm et al., 1991;Lee et al., 1988;Sharkey et al., 1987). Accordingly, RIT has the potential to be the most useful in adjuvant therapy when minimal disease is present (Sharkey et al., 1987;Langmuir & Sutherland, 1988). In the case of head and neck cancer this would apply to patients with stage III and IV disease. In these patients local recurrences occur in 50-60%, while 15-25% develop distant metastases after surgery and/or radiotherapy (Choksi et al., 1988). Unfortunately, no relevant metastatic model for HNSCC is available. In our study, the correlation between tumour size and therapeutic effect could not be determined due to the selected size range.
In several studies, an increase in therapeutic efficacy combined with a decrease in toxicity has been observed when total dose was given in multiple fractions (Buchegger et al., 1990;Buchegger et al., 1989;Colapinto et al., 1990;Smith et al., 1991;Schlom et al., 1990). Therefore, the efficacy of RIT with MAb E48 with respect to growing, established HNX-HN xenografts will be further investigated comparing single injection regimen to multiple injection regimens. Furthermore, since 131I is not the isotope of choice in clinical applications because of the low percentage therapeutic P-emission (32%) and the high percentage damaging y-radiation (66%), and because of the rapid dehalogenation of "1'I-labelled conjugates, we have developed a MAb E48 radioimmunoconjugate labelled with "86Re, an isotope with a high percentage P-emission (90%) and low percentage y-emission (8%). MAbs labelled with this isotope have already been described in tumour localisation and tumour therapy studies (Beaumier et al., 1991;Goldrosen et al., 1991). MAb E48 labelled with this isotope will be tested in the HNX-HN xenograft model.
Thusfar, clinical results with chemotherapy have been disappointing with respect to the effect on 5-year survival of patients, despite the number of trials over the past 10 years (Choksi et al., 1988;Snow, 1991). In our HNX-HN xenograft model, a number of conventional drugs, known to produce remissions in patients with head and neck cancer, and one experimental chemotherapeutic agent have been evaluated (Braakhuis et al., 1991) unpublished data). In the dose schedules described, none of the chemotherapeutic agents caused tumour growth delay factors higher than those obtained with either 800yCi or 400fiCi 3'I-labelled MAb E48 IgG. Furthermore, no cures were observed with these chemotherapeutic agents.
One of the limitations of the nude mouse xenograft model for RIT studies with radiolabelled MAbs is the absence of antigen expression in normal tissues. The presence of the MAb E48 defined antigen in normal tissues in the clinical situation will obviously influence the pharmacokinetics and biodistribution of radiolabelled MAb E48. In clinical radioimmunoscintigraphy studies using 9'Tc-labelled MAb E48 (Fab')2 fragment we observed uptake of radioactivity in normal oral mucosa and adrenal glands (van Dongen et al., 1992). Uptake in these tissues seems to be diminished when using whole IgG. Most clinical trials with radiolabelled MAb for diagnosis or therapy of solid neoplasms have reported MAb uptake in large tumours in the range of 0.001-0.01 %IDg-l (Goldenberg, 1991;Epenetos & Kosmas, 1989). Preliminary data on the localisation of 'mTc-labelled MAb E48 IgG indicate accumulation of the conjugate in tumours of 0.5-4.0cm diameter up to a mean %IDg-' of 0.03 at 44h (range: 0.0143-0.0823, number of patients = 7). This looks very promising indeed, when taking into account the higher accumulation of MAbs in small tumour loads. Chatal et al. reported on the biodistribution of "'In-labelled MAb OC125 intraperitoneally injected into patients with ovarian carcinoma, demonstrating low accummulation in large tumours (0.0014-0.0032 %IDg-') but significantly higher accumulation in small tumour nodules (0.13 ± 0.08 %IDg-') and malignant cell clusters (median 0.33 with a maximum of 4.16 %IDg-') (Chatal et al., 1989). Assuming that this size correlation also applies for head and neck tumours and assuming that patients will tolerate a dose of lOOmCi of '311-labelled MAB E48 (or an equivalent dose of '86Re-labelled MAb E48) (Rosen et al., 1987;Ward et al., 1988), achieving radiation doses in tumour tissue enabling elimination of minimal disease lies within reach.
Our data, showing the capacity of a single bolus injection '3'I-labelled MAb E48 to eradicate HNSCC xenografts in nude mice, present the first successful RIT results for head and neck squamous cell carcinoma. Together with data from an ongoing phase I clinical trial in our hospital, showing the capacity of 99'Tc-labelled MAb E48 (F(ab')2 fragment and IgG in detecting metastatic and recurrent disease, this indicates the potential of radiolabelled MAb E48 for radioimmunotherapy of patients with head and neck cancer.
This work was supported in part by Centocor Europe Inc., Leiden, The Netherlands.