Clearance of yttrium-90-labelled anti-tumour antibodies with antibodies raised against the 12N4 DOTA macrocycle.

Radioimmunotherapy (RIT) is currently limited by toxicity to normal tissues as a result of prolonged circulating radioantibody in the blood. In this study, the use of a clearing antibody was investigated (second antibody) in an attempt to reduce blood background levels of [90Y]A5B7 immunoglobulin G (IgG) activity, and, therefore, improve the therapeutic tumour-blood ratio in nude mice bearing human colorectal tumour xenografts. The second antibody was raised against the 12N4 macrocycle group used for chelation of 90Y, and is, thus, applicable to any anti-tumour antibody labelled with this methodology. Second antibody was administered 18, 24 or 48 h after radiolabelled antibody injection and produced up to a tenfold reduction in blood levels and a tenfold improvement in tumour-blood ratios. This has the advantage of reducing the risk of myelotoxicity caused by prolonged retention of activity in the blood. For all normal tissues, there was a similar or slightly lower uptake of [90Y]IgG with second antibody clearance, apart from a transient rise in liver activity due to complexes of primary and secondary antibody clearing via the liver. As a result of clearance of [90Y]IgG from the blood pool, there was an associated fall in the amount of antibody at the tumour site (up to 3.3-fold) at later time points for mice injected with second antibody. However, despite this, tumour-blood ratios remained superior to the control group at these later time points. Estimated dosimetry evaluation revealed that total dose to normal tissues, blood and tumour was lower than for the non-clearance group. Surprisingly, however, there was little improvement in total estimated tumour-blood dose ratio over the time period studied. This was probably because the majority of the dose was delivered to both the blood and tumour within the first 24 h after administration of [90Y]IgG, so that giving the clearing agent after this time did not produce a large difference in total estimated dose. The anti-macrocycle second antibody proved to be a successful clearing agent and could potentially be applied to any anti-tumour antibody coupled with the 12N4 macrocycle. In the light of the estimated dosimetry results described here, it would probably be most useful given at earlier time points (i.e. before 18 h after injection of primary antibody) to produce an improved tumour-blood ratio of total dose. Development of this strategy may allow higher levels of activity to be administered for RIT, and repeated dosing regimens.

tissues. The slow clearance of intact radiolabelled antibodies from the circulation and the time necessarv to achieve maximum uptake by the tumour has often resulted in large radiation doses to normal tissues. The amount of radiation that mav be administered for RIT is often limited by the potential damage to normal tissues. especially the bone marrow. caused by the persistence of radiolabelled antibodv in the circulation. Sexeral strategies hax-e been employed in an attempt to remove circulating antibody more rapidly. and produce high tumour-blood (therapeutic) ratios. These include the use of smaller. faster. clearing antibody fragments (Buchegger et al. 1990: Pedley et al. 1993. and specific in x-ivo or ex vivo clearing regimes (Begent et al. 1987: Norrgren et al. 1993 Prev-ious studies haxe demonstrated successful implementation of clearing acents. The use of a second antibody directed against the ( pnmary ) anti-tumour antibody has accelerated clearance and improved therapeutic ratios (Begent et al. 1987: Pedley et al. 1989). Other strategies have involved liposomally entrapped second antibodies (Keep et al. 1983) and axvidin-biotin systems (Marshall et al. 1995).
Most therapeutic studies to date haxe been carried out using the radionuclide iodine-131 (1""I). xhich is a medium range 3-emitter (0.6 MeV). However. there are problems associated with handlin2 large doses of ""I for RIT because of the high abundance of yenergy. More recently. sex eral inx estigators have suggested that alternative radionuclides such as yttrium-90 COY) or copper-67 C6-Cu) may be superior to ""I for RIT (Deshpande et al. 1988: King et al. 1994. Faxourable characteristics include higherenergy. shorter physical half-lives and stable chelation methods. Several studies hax e also reported higher tumour uptake and prolonged retention of 5'Y in tumour cells compared with I'l. e.g. Press et al (1996).
The murine antibodv A5B7. raised against carcinoembrxonic antigen (CEA). has been used for RIT labelled w-ith "I in nude mice bearing human colorectal tumour xenografts (Pedley et al. 1993) and in patients with colorectal cancer (Lane et al. 1994).
Good therapeutic responses haxe been demonstrated in xenograft models. but only a small number of responses hax e been produced clinicallI. In a recent studv. the 12N4 DOTA macrocvcle was conjugyated to A5B7 IgG and sucessfully radiolabelled wxith 41Y. 1307 The biodistribution of A5B7 IgG radiolabelled %vith either '"'I or 9'Y in the nude mouse xenograft model was compared and the results show ed that higher tumour uptake and retention of [3VY]IgG was observed (Casey et al. 1996). The aim in this studv was to investigate whether tumour-blood ratios could be further improved for [VY]IgG using a second antibody clearance system based on an antibodv raised against the 1 2N4 DOTA macrocycle.

Antibodies
The 12N4 DOTA macrocvcle %vas conjugated to bovine serum albumin for immunizations. usinc the method previouslI described by Turner et al ( 1994). The mouse IgG 1 anti-macrocycle antibody (1C) C -as raised using conventional hybridoma techniques and shown to be specific for the 12N4 macrocycle (Chaplin et al. in preparation). The mouse IgG1 anti-CEA antibody A5B7 was conjucated to the 1 2N4 macrocyvcle group for 'Y labelling, as described previously (Casev et al. 1996).
Biodistribution experiments A5B7 IgG was radiolabelled A-ith +'Y to a specific activity of 74 kBq g1-'. purified by high-performance liquid chromatography (HPLC) grel filtration and characterized as descnrbed previously (Casey et al. 1996). All the MF1 nude mice bearing LS174T human colorectal carcinoma xenografts (Pedlev et al. 1993) were injected by the tail v-ein intravenously (i.s-.) w-ith approximately 10 MBq of [90Y]IgG. Test animals s-ere subsequently injected intraperitoneally (i.p.) swith clearing antibody 1C2 at a fivefold molar excess (approximately 25 pa) oser the amount of labelled antibody oriainallv administered. Clearing was tested at 18. 24 or 48 h after injection of radiolabelled antibody. Test and control animals %vere bled and tissues removed for radioactivity assessment at various time inters als after injection. Bremstrahlunr radiation from 4`Y in tissues w as counted using a calibrated camma-counter (WNizard. Wallac. UK). The Mann-Whitnev nonparametric statistical test was used to compare data and results were considered to be sienificant w-hen P < 0.05.

Dosimetry
Estimated radiation doses ( ,) to blood. tumour and normal tissues per MBq of`Y injected were evaluated from the biodistribution data. Figures for percentage of injected actisity per gram of tissue (%c ia g-l) were decay corrected and the area under the %7 ia o-' over time curse w'as calculated using, the trapezoidal rule. Total estimated 0 dose to indi-idual tissues was assessed using the MIRD S factor of 1.93 for 90Y (MIRD pamphlet 11. 1975) to convert MBq c-' to cGy h-'. There A-as no correction for crossorgan doses.

RESULTS
To assess whether the anti-macrocycle clearing antibody (1 C2) could complex and clear 4'Y-labelled A5B7 IgG from the blood pool. a fivefold molar excess of unlabelled 1 C2 was administered at various time points after radiolabelled antibody injection. Previous experiments hase show-n that this level of second antibody in similar clearance systems was optimal: lower doses did

24-and 48-h clearance
The clearing antibodyxx as initiallv administered 24 h and 48 h after injection of 4'Y-labelled IgG. and the biodistribution was compared with the control group with no clearing antibody (FFigure 1). Administration of 1C2. 24 h after injection of [9''Y]IgG. produced a rapid decrease in the level of labelled antibodx in the circulation ( Figure 1A). By 1 hour after injection (25 h post antitumour antibody). the clearing, antibody had produced a significant reduction (9.7-fold) in blood radioactivitv level from 9.5% to 0.98%ia o-' (P < 0.05). This was accompanied by a large rise (fourfold) in liver activitv from 4.8%c to 19.0%7 ia a-' (P < 0.05) probably due to complexes of primary and secondary antibody clearingc via the liver. At later time points. 24 and 96 h after administration of clearing antibody (Figure I B and C.) there was an associated ninefold and 4.7-fold reduction in blood activitv.
respectively. compared with control animals (P < 0.05). Although liver activitv levels were hicher than for the control group at these later time points (1.8-fold and 1.42-fold respectively). the difference was not siganificant. For all normal tissues. there was similar or slightly lower uptake of [9"Y]IgG with second antibody clearance. in particular there was reduced splenic (6.0% reducing to 2.8% ia ca-' 24 h after clearance) and luna activitv (4.4%7 to 1.5% ia 24 h after clearance). The amount of radioactivitv retained in the tumour was not significantly different 1 h and 24 h after clearance. but at a later time (96 h after clearance) there was a large decrease in tumour activity (26%7 to 8.0% ia g-1). The large reduction in circulatina antibody after injection of clearinc antibodv and similar retention of activity in the tumour to the control grroup (to 48 h) produced up to a tenfold improvement in therapeutic tumour to blood ratios (Table 1). Clearance at 48 h produced a similar effect. whereby activity in the blood fell sharply by a factor of 8 (P < 0.05) within 1 hour of injection (5.5%7 to 0.69%7 ia gl-'). Aaain liver activitv was increased 1.7-fold. but this was less dramatic than for clearance at 24 h (fourfold). Normal tissue clearance was similar to the 24 h clearance group as described above. There was no significant difference in tumour activity when compared with either the control or the 24-h clearance group by 1 h after clearance. However. again at the latest time point (72 h after clearance) tumour levels fell from 26%7 to 11.1% ia g Tumour-blood ratios were significantly improved (7.1-fold) 1 h after clearance. Although tumour activitv was reduced at the later time point ( 120 h). the accompanied reduction in blood levels still created a 3.3-fold increase in tumour-blood ratio (33:1 compared A-ith 10:1 for the control group).  Table 2 shox s the estimated radiation dose to tumour and normal tissues per MBq of 4'Y administered for animals xxith and without second antibody clearance. Use of clearance antibody. civ-en either at 24 or 48 h. reduced the total radiation dose received by all organs. except the lixer. This included a 1.7-fold and 1.3fold reduction in blood dose for 24 and 48 h clearance groups respectively. The increase in liv-er activity at all time points in both clearing groups produced a tuxoto 2.4-fold higher estimated total dose than for the control group. There wvas an associated decrease in dose delivered to the tumour when the second antibodvx as given (1.6and 1.3-fold for 24 and 48 h clearance groups respectively). In spite of improved tumour-blood ratios at each time point shoxxrn in Table 1. the oxerall tumour-blood ratios of total estimated dose were surprisingly similar to the non-clearance group ( Table 2). The beneficial effect of second antibody clearance was obserxed within the first 24 h after administration of clearinr agent wxhich resulted in a tenfold decrease in total dose to blood during this period (data not shox-n). However. because the largest dose delivered to both the blood and tumour occurred within the first 24 h after administration of [VY]IgG and because the clearinc agent wxas gix en after this time. a large difference in total estimated dose was not produced.

18-h clearance
Further experiments were performed to inxestigate the effect of earlier clearance (before 24 h) and to cain more information on the first 24 h of [90Y]IgG administration. Ficure 2A illustrates a more detailed examination of the biodistribution of [9°Y]IgG (without clearance) to 72 h. Activitv localized to the tumour. and by 18 h there were higher levels of activity in the tumour than blood and all other normal tissues. Tumour activity accumulated over time duringr the 72-h period while radiolabelled antibody cleared from the blood. After this time. the actix itx in the tumour begyan to fall (Ficure 2C). Other normal tissues. wxith the exception of liver and spleen. showed a similar pattern of clearance related to the blood pool activitv.
there was a more marked difference in tumour activity. 18%c ia g'l remained in the tumour compared with 44%7c ia g-' of the control group (P < 0.05). However. in spite of this. because of the large reduction in blood activity. there was a large improvement (up to 3.3-fold) in tumour-blood ratios at both these time points on addition of clearing antibody (Table 3). Similarly. at the later time points there was a reduction in activity for all tissues. except the liver and femur. for the clearance group compared with control animals. It was predicted that levels of activity in the liver would rise in the first few hours following injection of clearing agent. as seen with the 24h clearance group. but it must be noted that this time point was not included in the biodistribution and dosimetry evaluations (1 h after injection of clearing agent). By 24 h after clearance. there was only a small rise in liver activity from 12%c to 16%7 ia g which was not significantly different to the control group. Estimated total doses were evaluated over the 72-h period with and without administration of clearing agent 18 h after injection (Table 4). Again. there was a reduction in total absorbed dose to normal tissues. blood and tumour for the clearance group.
However. despite the 1.6-fold reduction in blood dose tumour levels were also 1.5-fold lower. which did not improve the overall tumour-blood ratios (2.49:1 as opposed to 2.35:1 for the control group). Absorbed dose to the liver was also 1.2-fold higher. which produced a lower tumour-liver ratio.

DISCUSSION
In this studv. the first use of the anti-12N4 DOTA macrocvcle antibody (1C2) as an in vi-xo clearing agent is described. The most frequently reported dose-limiting toxicity in clinical RIT studies is myelosuppression (Bernstein et al. 1991: Lane et al. 1994). It has been established that a large part of the dose to bone marrow is delivered through high levels of circulating activity in the blood.
Therefore. if a reduction in blood levels of activity can be achieved usinc a clearing antibody. it would be expected that myelosuppression could be reduced. and larger doses could then be given with the prospect of potentially achieving higher doses to the tumour.
The biodistribution studies showed a substantial reduction in the level of [90Y]IgG in blood and normal tissues after the cleanrng antibody was given. although there was a transient rise in radioactivity in the liver due to rapid clearance of immune complexes via that organ. Most of the activity was cleared from the liver within 24 h of administration of clearing antibody. There was no associated increase in splenic activity using this clearino agent. which is an obvious advantage over the use of other second antibodies or clearing strategies which have demonstrated high accumulation of activity in the spleen (Pedley al. 1989: Goldenberg et al. 1987: Marshall et al. 1994). The spleen also catabolizes radiolabelled antibody complexes at a much slower rate and is more radiosensitive than the liver. which is disadvantageous in terms of an increased total and cross-organ absorbed dose.
Accelerated clearance of primary antibody produced higher tumour-blood ratios (up to tenfold). but unfortunately the levels of activity associated with the tumours of animals receiving second antibody were significantly lower at later time points. This is a common finding with most if not all other clearing systems and.
unfortunately. appears to be unavoidable (Sharkey et al. 1988: Pedley et al. 1989). Ideally. clearing agents or second antibody should be administered at a time when the antibody has reached its maximal value in the tumour so as to avoid excessive loss of activity. However. [9°Y]A5B7 IgG accumulates in the tumour up to 72 h (refer to Figure 2A). by which time blood levels have declined naturally and there would be little advantage in administering a clearing agent at this time. In previous studies. clearing agents have generally been administered 24-48 h after injection of pnmarv antibody. which is usually at the approximate peak level of tumour localization (Begent et al. 1987: Marshall et al. 1994). These studies have also reported improved tumour-blood ratios.
It is. however. important to consider not only the biodistribution data at various time points after clearance. but also area under the curve analysis of total dose when considering the potential merit of second antibody clearance. Although these dosimetry calculations can only be considered as estimates. they have previously proved useful predictors of toxicity and a guide to the levels of activitv that may be administered for RIT (Pedley et al. 1989. 1993. Surprisingly. although blood activity was significantly reduced after second antibody administration this only had a small effect on the cumulative radiation dose. most of which had already been given before injection of second antibody (i.e. before 24 h). Thus. these dosimetry estimates predicted no overall improvement in the tumour-blood dose ratio. Without considering these calculations. it would appear that the higher tumour-blood ratios observed at specific time points after clearance from the biodistribution data provide an advantage in terms of lower toxicitv.
The beneficial effect of second antibody clearance was observed in the first 24 h after administration of clearing agent. when a tenfold reduction in total dose to the blood occurred. However. this was not reflected in the overall dose estimates because the majority of the total dose to blood and tumour had occurred before the clearing antibody was administered. Even at the earliest second antibody clearance time point ( 18 h). the doses to blood and tumour were reduced by a similar amount. which resulted in a similar overall tumour-blood ratio. This indicates that there would be no improvement in terms of reduced toxicity on administration of second antibody given at  These studies were performed using ''"I-labelled antibodies compared with 9"Y used in this study. which may explain the discrepancy. 90Y is a higher energy [emitter. which will deliver a higher initial dose rate to the tumour than ''I. This results in the relatively high absorbed dose during the first 24 h after administration. If the anti-macrocycle clearing antibodvsAas administered at 6 h. then it is possible that an improved cumulative tumour-blood ratio could be generated. which may allow a larger amount of activity to be administered. In addition. this will produce a higher initial dose rate to the tumour which is known to be an important parameter for successful RIT (Fowler. 1990). Furthermore. this therapeutic strategy. which should ultimately reduce myelotoxicitv by the early removal of radiolabelled primary antibody. would favour repeated doses that could be administered at relatively frequent intervals. There are. however. two possible disadvantages to this therapeutic strategry. The immunogenicity of both murine antibodies and the high levels of administered activity. and therefore protein.
required for therapy that may result in a large amount of immune complex formation. which could saturate the reticuloendothelial system and lead to the circulation of excess immune complexes.
Technoloay is now available to produce humanized antibodies. and there is evidence to suggrest that the immune response for the limited number of humanized antibodies used clinically so far is largely reduced (Juweid et al.. 1995: Sharkey et al.. 1995: Stephens et al.. 1995. The liver receives the highest normal tissue absorbed dose with second antibody clearance. and. therefore. careful attention to possible hepatic toxicity and clearance of circulating immune complexes is required in a dose escalation manner if this strategy is to be used cliically. This clearance strategy is of great interest because of its potential universal application to any anti-tumour antibody. Conjugation of the 12N4 macrocycle is a mild procedure and does not generally affect the immunoreactivit-v. This labelling procedure has also been applied to cross-linked antibody fragments (Casey et al.. 1996). If used in combination with a metallic isotope such as indium-11 1 and the anti-macrocycle antibody. this may be advantageous for radioimmunodetection because improved images could be generated at earlier time points.
British Joumal of Cancer (1998) 78(10) The purpose of giving second antibodv >-as to investigate whether the radiation dose to bone marrow-could be reduced. permitting a higher tumour dose to be delivered. Although blood activity w-as sinificantlv reduced after second antibody administration (given at 18. 23 or 47 h after primary antibodv). this onl1 had a small effect on the cumulative radiation dose. most of which had been iyven before second antibody administration. If second antibody was administered at an earlier time (e.g. 6 h after primary antibod ). it is possible that the usefulness of this system could be improved by the deliverof a higher initial dose rate to the tumour to achieve a oreater cell kill. Further experiments are required to optimize this potentially new therapeutic strategy.