`
`Development of human anti-murine antibody (HAMA) response
`in patients
`
`Joe J. Tjandra, Lanny Ramadi and Ian F. C. McKenzie
`
`Research Centre for Cancer and Transplantation. Department of Patho/og)'. Un11•ersity of
`l\felboume, Par/...'11ille, Victoria 3052, Australia
`
`(Submitted 1 June 1990. Accepted for publication 9 No1•ember 1990.)
`
`lluman anti-mouse antibody (HAMA) response was detcmuned in the scrum of 67
`Summary
`patients who received subcutaneously administered radiolabellcd murine monoclonal antibodies
`(MoAb) (50 µg-3 mg) for immunolymphoscintigraphy and of JO patients with advanced colorectal
`cancer who received munnc MoAb-N-acetyl melphalan (MoAb-N-AcMEL) conjugates (amount of
`MoAb ranged from 120 mgtm2 body surface area to 1000 mg!m2 body surface area) ai. therapy. A
`pre-existing low level of apparent human anti-mouse antibody reactivity could be detected in the
`serum of normal subjects and patients prior to administration of murine MoAb. Subcutaneous
`administration of low doses of murine MoAb, as used in 1mmunolymphoscintigraphy, was associated
`"-ith a low incidence (4/67 or 6%) of elevated IIAMA response; the use of F(ab'h fragments was
`associated wnh the development of elevated HAMA response in one of three patients. By contrast,
`therapy with hepatic artery infusion of murine MoAb-N-AcMEL con1ugatcs in three repetitive daily
`doses (each infusion la su ng 2 h) elicited elevated llAMA responses in I Oil 0 (I 00%) patients, usually
`1-3 weet...s after the start of therapy. The HAMA response ofpauents in the therapy group was higher
`than those in the immunolymphoscmt1graphy study and the use of steroids did not prevent the
`development of the HAMA response. Further administration of MoAb-N-AcMEL conjugates to a
`patient, who had already developed HAMA, led to 'scrum sickncss._typc symptoms and a transient
`reduction in the HAMA titres. The elevated HAMA response was polyclonal, containing increased
`levels of both immunoglobulin M and G (lgM and lgG) and was directed against mouse-specific
`detcm1inants, the isotype (presumed to be the Fe portion), the F(ab'h and the 'idiotypc' of mouse
`immunoglobuhns.
`
`INTRODUCTION
`
`Murine monoclonal antibodies (MoAb) with
`specificity for tumour-associated antigens are
`increasingly being used as carrier molecules for
`radio-imaging agents such as t3q, 11 t In, 1231 ( l-
`4) and for therapeutic agents such as cytotoxic
`drugs (5-7). One problem with the use ofmurine
`MoAb in humans has been the development of
`the human anti-murine antibody (HAMA)
`response (8-11) which restricts repetitive dos(cid:173)
`ing. However, reports differ on the incidence
`and nature of the development of HAMA
`response; the differences were probably related
`
`I. F. C. McKenzie, Research
`Correspondence:
`Centre for Cancer and Transplantation, Department
`of Pathology, University of Melbourne, Parkville, Vic.
`3052, Australia.
`Abbreviations used in this paper: 131 I, iodine-131;
`HAMA, human anti-murine antibody; 1.v., intra(cid:173)
`venous; w/w, weight :wcight; lg, immunoglobulin;
`MoAb, monoclonal antibody/ies; PBS, phosphate buf(cid:173)
`fered saline; s.c., subcutaneous(ly).
`
`to the different clinical programmes (amount of
`foreign protein administered. the route and
`number of treatments and the time interval
`between treatments), the immune status of the
`patients and the methodological differences in
`assay techniques for HAMA. In addition, the
`natu1e of HAMA response in cancer patients
`may be different from that elicited in transplant
`patients by mouse MoAb OKT3 and this is not
`surprising as OKT3 has both stimulatory and
`suppressive effects on T cells ( 12).
`Radio-iodinated murine MoAb RCC-1 (reac(cid:173)
`tive with breast cancer) had been administered
`subcutaneously (s.c.) for immunolymphoscinti(cid:173)
`graphy. and when used together with cold iodine
`labelled 'blocking' Ly2· l antibody (non-reactive
`with breast cancer), successful localization of
`axillary lymph node metastases from breast can(cid:173)
`cer occurred in about 90% of cases (3). Various
`MoAb have also been conjugated to the N-acetyl
`derivative ofmelphalan (N-AcMEL) and shown
`to have in vitro and in vivo specificity and cyto(cid:173)
`toxicity and specifically inhibit the growth of
`
`1 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`368
`
`J. J. TJANDRA ET Al.
`
`human colon carcinomas xenografted in athy(cid:173)
`mic mice {13,14). The MoAb-N-AcMEL conju(cid:173)
`gates have also been administered by means of
`hepatic artery
`infusion
`to patients with
`advanced colorectal cancer and have achieved
`biological responses in some patients (7).
`The aim of the study was to determine the nat(cid:173)
`ure of the HAMA response in patients after
`regional administration of murine MoAb for
`immunolymphoscintigraphy (s.c. injection) or
`therapy (hepatic artery infusion).
`
`MATERIALS AND METHODS
`
`Murine monoclonal an1ibodies
`For imaging studies. murine MoAb 3El ·2 immuno(cid:173)
`globulin (lgM), RCC-1 (formerly called 17· l; fgG 2a)
`and Ly2·1 (IgG2a) were used. Antibody 3El·2 was
`raised against fresh human breast carcinoma ( 15).
`RCC-1 was raised by immunizing inbred Biozzi mice
`with the MCF-7 breast cancer cell line (16). The Ly2· l
`antibody is reactive with the murine Ly2· l specificity
`but not with human breast cancer ( 17). The Ly2· I anti(cid:173)
`body was used in the study as a 'blocker' to reduce
`background non-specific uptake of 1311-RCC-1 (3).
`Murine MoAb used for therapy included 30·6 (IgG2b),
`which is reactive against a large number of colon car(cid:173)
`cinoma ~ell lines (18). 1-1 and JGT (lgGI), which are
`both anti-CEA (carcinoembryonic antigen) ( 19). Irrel(cid:173)
`evant murine MoAb used to evaluate the HAMA
`response included BCI (lgG3), BC2 (IgGI), and BC3
`(lgM) which were directed against mucin-like glyco(cid:173)
`proteins of breast (20); and polyclonal antibodies of
`monkey and sheep origin (produced in our labora(cid:173)
`tory). MoAb 3El ·2 was purified from ascitic fluid (21)
`and the antibodies RCC-1 and Ly-2· l were purified
`on Protein A-Scpharose (Pharmacia fnc., Piscataway,
`New Jersey, USA as described previously (22); anti(cid:173)
`bodies 30·6, 1-1 and JGT by Protein A-Su pa rose (Phar(cid:173)
`macia Inc., Piscataway, NJ). Purity was assessed by
`sodium dodccyl sulfate-polyacrylamide gel electro(cid:173)
`phoresis. The F(ab'h fragments were obtained from
`purified MoAb RCC-1 by pepsin digestion (23). Anti(cid:173)
`bodies were aliquoted and stored at - 70°C until
`used.
`
`lodinalion of monoclonal an1ibodies
`Purified antibodies (3E1 ·2, RCC-1 or RCC-1
`F(ab'h) were radiolabelled withl31I (Amersham Int.
`UK) using the iodobead (24) or enzymobead reagent
`(Bio-Rad, Richmond, California, USA) (25). The puri(cid:173)
`fied Ly2· l antibody was labelled with non-radioactive
`sodium iodide or 1311 using the chloramine T method
`(26). Preparation of the iodinated antibodies has been
`previously described (3,2 1 ).
`
`Preparation of dn1g-an1ibody conjugates
`The N-acetyl derivative of melphalan (N-AcMEL)
`was prepared and conjugated to the antibodies (30·6,
`1-1 and JGT) as described previously (14). The anti(cid:173)
`body activity and cytotoxicity of the immunocon-
`
`~ug~t~s. were ensured by resetting assay (27) and
`mh1b1t1.on of DNA synthesis using [3H]-thymidmc,
`respcct1 vely ( 14).
`
`Patient studies
`For diagnostic studies (immunolymphoscintigra(cid:173)
`phy) to localize axillary lymph node metastases, 67
`patients with various breast conditions (benign and
`malignant) (Table I) received a variety of iodinated
`preparations s.c. They included 13 IJ-labelled 3E I· 2
`(50-200 µg), l31f-labelled RCC-1 (50-400 µg) , 1311.
`labelled Ly-2· 1 (0·4-2 mg), 1311-labelled RCC-1 (0·4-1
`mg) together with 'blocker' antibody Ly2· 1 (2 mg) iod(cid:173)
`inated with non-radioactive sodium iodide and 131[.
`labelled RCC-1 F(ab')i. To prevent thyroid uptake of
`free radioiodine, patients received potassium iodide (5
`mL of 16·54% w/v) and sodium perchlorate (400 mg)
`orally, I h before the s.c. injection; the potassium iod(cid:173)
`ide was continued for 5 days after the injection.
`For therapy, nine patients with extensive colorectal
`hepatic metastases (2/9 also had pulmonary metas(cid:173)
`tases) and one patient who had a curative resection of
`the Duke's C colon cancer received between 120
`mg!m2 and 1000 mg!m2 body surface area of MoAb
`(274-1696 mg) (1-1 and/or JGT and/or 30·6) conju(cid:173)
`gated with between 5 mg!m2 body surface area to 20
`mg!m2 body surface area of N-AcMEL (Table 2). The
`MoAb selected for conjugation with N-AcMEL were,
`where possible, individually chosen for each patient,
`based on the binding of the particular antibody (I-1,
`JGT, 30·6) to sections of the primary colon cancer
`tissue as assessed by immunoperoxidase staining. Jn
`general, MoAb was selected only ifit stained >50% of
`the carcinoma cells on the sections. None of the
`patients received any other form of treatment for 4
`weeks before and 8 weeks after treatment with the
`immunoconjugates.
`The immunoconjugate was administered via hep(cid:173)
`atic artery infusion over 2 h per day for 2 days (7). All
`patients received the immunoconjugates in three
`equal doses (t - 0 h, t - 24 h, l - 48 h). Patients had
`prophylactic intravenous (i. v.) dexamethasone 8 mg
`just before each infusion of the immunoconjugate and
`oral prednisolone I 0 mg daily for 7 days after com(cid:173)
`pletion of infusion.
`the Medical
`Both studies were approved by
`Research Board of the Royal Melbourne Hospital and
`wri~ten informed consent was obtained from every
`patient. Blood samples were obtained from patients
`before, during and after the injection of MoAb conju(cid:173)
`gates to detect human anti-murine immunoglobulin
`(HA.MA) response. Normal scrum samples (n-20)
`were also obtained from apparently healthy blood
`donors. All samples were aliquoted and stored at
`-1o·c.
`
`Human anti-murine amibody (HAMA) response
`Human antibodies against the murine MoAb were
`measured by an enzyme-linked immunosorbcnt assay
`(E.LISA) •. modified fro~ that described previously (9).
`Nmety-s1x well
`flexible polyvinylchloride (PVC)
`plates (Costar, Cambridge, Massachusetts, USA) were
`
`2 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`HUMAN ANTl-MURINE IMMUNOGLOBULIN RESPONSES
`
`369
`
`coated with 100 µUwell of various purified murine
`MoAb (Sµg/mL) in a O·l M carbonate buffer, pH 9·6
`and non-specific binding blocked with I% bovine
`serum albumm/PBS (phosphate buffered saline) pH
`7·6. Serial dilutions of patients' sera and pooled nor(cid:173)
`mal human serum (50 µUwell) in PBS/0·05% Tween
`20 were added to the antibody coated wells and incu(cid:173)
`bated for 16 h at 4°C. Plates were then washed with
`PBS/0·05% Tween 20 and then reacted with 50
`µUwcll of a I :600 dilution of sheep anti-human
`immunoglobulin conjugated to horseradish peroxi(cid:173)
`dase (Amersham International, UK) for 3 h at 37°C.
`The colour reaction was developed using 50 µL of
`0·03% ABTS [2.2'-azino-di- (3-ethylbenthiazoline)
`sulfonate] (Amersham International, UK) and 0·02%
`h}'drogen peroxide (BDH Chemicals, Poole, UK) and
`read with an ELISA plate reader (Titertek Multiscan
`MC) at a wavelength of405 nm. In some cases, the IgM
`and lgG components of the HAMA response were sep(cid:173)
`arately measured by using phosphatase labelled affin(cid:173)
`ity purified goat anti-human lgM or lgG (Kirkegaard
`and Parry, Maryland) respectively and the colour reac(cid:173)
`tion developed with alkaline phosphatase substrate.
`Results were expressed as the absorbance value of
`patient serum compared with control serum (pooled
`normal human serum from 20 apparently healthy
`blood donors) and a positive test was defined as one in
`which the absorbance was equal to or greater than
`twice the absorbance of pooled normal human serum.
`The HAMA titre was determined by obtaining the
`inverse of the highest serum dilution which gave a
`positive test result, and was arbitrarily graded as weak
`(titre 100-< 1600), moderate (titre 1600-6400) or
`strong (titre > 6400). The background was too high at
`serum dilution less than 11100; thus, within the sensi(cid:173)
`tivity of the assay, HAMA titre ofless than I 00 cannot
`be determined accurately.
`
`RESULTS
`
`Imaging studies
`A low level of apparent anti-murine immunoglo(cid:173)
`bulin activity can be measured in the serum of
`normal individuals (NHS) (Fig. 1, NHS) and
`also in the pre-immune serum of patients. These
`levels were detected both in patients who later
`produced an elevated HAMA response and in
`those who did not produce such a response. A
`HAMA response was considered positive when
`the absorbance of the serum was equal to or
`greater than twice the absorbance of pooled
`normal sera (Fig. 1 ). Table l summarizes the
`measurements of antibody to murine antibodies
`in 67 patients with various breast conditions
`(benign and malignant) prior to subcutaneous
`injection of murine antibodies for imaging
`studies and 2 weeks-2 months after injection.
`Four of 67 patients (6%) showed a positive
`HAMA response (Table I), which developed
`
`3
`
`2
`
`e c
`"' 0
`:!..
`0
`0
`
`0-t-~~-r~~---..~~~..--:;;;;;"'--~
`102
`105
`
`Serum dilullon- 1
`
`Fig. I. Level of pre-existing anti-murine antibody in
`a representative pooled normal human serum (El)
`compared with the elevated HAMA response m a
`patient(• ) (Table 2; Patient 7) who had received thera(cid:173)
`peutic admmistration of murine MoAb-N-AcMEL
`conjugates 4 weeks previously.
`
`within 2 months of exposure. None of the
`patients had a strong response (titre > 6400) and
`in the limited number of patients studied, the
`intensity of the response did not relate to the
`amount (50 µg-3 mg) of murine antibody
`injected. In addition, none of the patients
`developed any clinical side effects from the
`injection that could be related to the develop(cid:173)
`ment of a positive HAMA response. Thus doses
`of up to 3 mg of murine MoAb given s.c. did not
`give rise to an elevated HAMA response in most
`patients.
`
`Therapy studies
`Table 2 summarizes the HAMA response in ten
`patients with advanced colorectal cancer prior
`to and after therapy with MoAb-N-acetyl mel(cid:173)
`phalan conjugates (amount of antibody: 120
`mg!mL 1000 mg!m2)
`via hepatic artery
`infusion. All the patients had three doses of the
`injection, 24 h apart and received prophylactic
`steroid therapy I h before and for 7 days after
`therapy. All the ten patients (I 00%) developed
`elevated HAMA responses; the intensity of the
`response did not correlate with the amount of
`murine antibodies received (Table 2) or with the
`pre-immune level ofanti-murine immunoglobu(cid:173)
`lin activity. The elevated HAMA response
`
`3 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`Table 2. Summary of therapy studies.
`
`Patient
`
`MoAb administered
`
`1-1
`
`Amount of Ab•
`120 mg!m2 (R X I)
`
`1-1 , JGT
`
`160 mg!m2 (RX 2)
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`30·6, 1- 1, JGT
`
`980mg!m2
`
`1-1
`
`250 mglm2
`
`30·6, l-1 , JGT
`
`340 mg!m2
`
`1-1
`
`380 mg!m2
`
`1-1 , JGT
`
`500 mg!m2
`
`30·6
`
`440mg!m2
`
`30·6, 1-1 , JGT
`
`1000 mg!m2
`
`1-1 , JGT
`
`820 mglm2
`
`10
`
`1-1. JGT
`
`1000 mg!m2
`
`Time from administration
`of antibody (weeks)
`
`HAMA response
`(titre)
`
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`4
`8
`0
`3
`0
`4
`8
`0
`4
`
`< 100
`1·5X 10s
`l ·OX lOS
`l ·OX IOS
`1·9X 10s
`1·5 X IOS
`< 100
`5·0 X 104
`2·5X 104
`< 100
`2·0 X 104
`1·5 X 104
`< 100
`< 100
`3·2X 103
`< 100
`l·O X I05
`5·0 X 104
`< 100
`5·0 X 104
`5·0X 104
`< 100
`4·0 X 10s
`2·0 X 105
`< 100
`l ·OX 10s
`< 100
`3·2X 103
`3·2 X 103
`< 100
`5·0 X 104
`
`•When multiple antibodies were used for drug conjugation, the final preparation of the immunoconjugates had,
`where possible, equal proportions of each MoAb. Doses were expressed m amount of MoAb/surface area of the
`patient. R X 1, first course of treatment; R X 2, second course of treatment, given to Patient I.
`
`4 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`HUMAN ANTl-MURINE IMMUNOGLOBULIN RESPONSES
`
`371
`
`developed in l/ l 0 patients as early as Day 5 after
`the first administration of infusion, but most
`patients (7110) developed elevated HAMA
`response (as defined earlier) between 1 and 3
`weeks after the exposure to the foreign protein
`(Fig. 2). The peak response occurred 14-28 days
`following exposure in 7/8 cases; in two other
`patients, serum beyond this period was not
`available. In 2/3 patients who were followed for
`up to 9 months, the elevation of HAMA
`response persisted but there was a gradual fall in
`the titres. Five of the patients (Table 2; Patient 1
`RX2. patients 7. 8. 9 and 10) developed fever
`and this appeared to correlate with the amount
`of the immunoconjugates received and was not
`related to the intensity of the HAMA response.
`One patient (Patient 1) received 120 mglm2 of
`MoAb and developed an elevated HAMA
`response (Fig. 3). Two months later a further
`dose ( 160 mglm2 MoAb) was given and
`symptoms suggestive of a Type III hypersensi(cid:173)
`tivity 'serum sickness' reaction developed but
`was self limiting. During the second course of
`treatment (Fig. 3, R X2), there was an initial
`reduction of HAMA titre which might indicate
`immune complex formation, followed by a
`boost of the HAMA response, as from Day 10
`onwards.
`
`Specificity of HAMA response
`To determine whether the HAMA response was
`directed against common determinants on the
`constant domain of mouse immunoglobulins or
`whether it was specific for each MoAb. the bind(cid:173)
`ing of patients' sera to microtitre wells contain(cid:173)
`ing relevant MoAb (i.e .. the same as that injected
`into patients), or other murine MoAb ('irrel(cid:173)
`evant') of the same or different isotypes as that
`of the relevant antibody were assayed as before.
`The results of a representative patient (Table 2;
`Patient 1) are shown. As shown in Fig. 4(a). sera
`from Patient 1 in the therapy studies, who had a
`positive HAMA response. contained antibodies
`that bound well to both the relevant (injected:
`1-1) and the irrelevant murine MoAb (3El·2,
`BC2), irrespective of antibody subclass. Similar
`results were obtained when the serum samples of
`other patients who developed a positive HAMA
`response were tested.
`This indicates that there is a major reaction to
`common determinants of all mouse immunoglo(cid:173)
`bulins; such a reaction could be with the kappa
`light chain, but this was not specifically meas(cid:173)
`ured.
`
`.,
`c
`!!!
`iii
`Q.
`ci z
`
`3
`
`2
`
`.,
`~
`< ::=;
`<
`J:
`
`200 000
`
`100 000
`
`2
`
`3
`
`4
`
`6
`
`Time (weeks)
`
`Fig. 2. Time in weeks after initiation of MoAb-N(cid:173)
`AeMEL treatment when elevated HAMA responses
`were ftr.;t detected. Weck I refers to days 1-7 after the
`start of treatment; Week 2 refers to days 8-14. and so
`on.
`
`Rx 1 W 2 W 4 W 8 Rx 2 W 2 W 4
`
`Time (weeks)
`
`Fig. 3. Development of elevated HAMA response in
`Patient I (Table 2) after the first treatment (RX I) of
`MoAb-N-AeMEL with time. Changes in levels of free
`circulating HAMA during (RX 2) and after second
`administration of MoAb-N-AcMEL were also illus(cid:173)
`trated. The time was expressed m weeks (W) from the
`time of respective treatment.
`
`5 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`372
`
`J. J. TJANDRA ET AL.
`
`The binding of sera of patients, before and
`after immunization, to polyclonal antibodies of
`monkey and sheep origin coated on the plate was
`also assessed in a direct binding assay. Data
`from a representative patient (Table 2, Patient
`
`l) are presented in Fig. 4(a). There was detect(cid:173)
`able antibody response towards immunoglobu(cid:173)
`lin preparations of monkey and sheep in the
`'pre-immune' serum but the elevated antibody
`response after therapy was directed predomi(cid:173)
`nantly against determinants unique to murine
`immunoglobulin preparations. Similar resulls
`were obtained when the sera of other patients
`who had a positive HAMA response were
`tested.
`A competitive inhibition ELISA, performed
`on immune serum from Patient I showed that
`MoAb 1-1 and JGT (relevant MoAb) competed
`for 1-1 binding sites more efficiently than the
`irrelevant murine MoAb of the same (lgG I:
`BC2) or different (lgG2a: Ly-2· l: lgG2b:30·6:
`lgM: BC3) isotypes (Fig. 4b). However. all the
`murine MoAb inhibited the binding of human
`anti-murine antibody
`to murine antibody
`(coated on the plate) to some extent. Similar
`results were obtained using sera from other
`patients with positive HAMA response in both
`the therapy and the imaging studies (data not
`shown). In these inhibition studies, which are
`analogous to an absorption test, it was clear that
`the best inhibition by irrelevant antibody in the
`case of Patient I (Fig. 4b) was by antibody BC2
`that was of the lgG 1 isotype (same isotypc as the
`administered antibodies) and this inhibited
`better than irrelevant lgG2a, IgG2b and IgM
`antibodies. It was concluded that part of the
`HAMA response was anti-isotypc specific, pre(cid:173)
`sumably to the Fe region of the lgG I immuno(cid:173)
`globulin. The fact that inhibition was best using
`the relevant (administered) antibody 1-1 or JGT.
`indicated that someoftheimmune response was
`specific to the administered antibodies. This lat(cid:173)
`ter response is called, in various studies, 'anti(cid:173)
`idiotypic' ( l 0,28) but we have not conducted the
`appropriate studies to determine whether this
`component of the HAMA response is to the anti(cid:173)
`gen binding site or to other sites on the murine
`antibody molecules. Thus the HAMA response
`in the patients studied had three components: (i)
`anti-mouse
`immunoglobulin
`(mouse-specific
`determinants: heavy and/or light chains); (ii)
`anti-isotype; and (iii) antibody response specific
`to the administered mouse immunoglobulin
`(loosely termed 'anti-idiotype' in this study).
`In addition, the lgM and lgG components of
`the anti-murine immunoglobulin activity could
`be separately measured in the 'pre-immune'
`human serum (Fig. 5) and each component was
`increased with the development of elevated
`HAMA response. It is clear that the HAMA
`response was essentially polyclonal and con(cid:173)
`tained increased levels of lgM as well as lgG.
`
`• Pre-treatment
`D Post-treatment
`
`Sheep
`
`3Et .2
`
`BC2
`
`Monk
`Target preparations
`
`1-1
`
`• 1 mg/ml
`• Control
`
`la 250 µgimL
`
`3
`
`2
`
`0
`
`2
`
`(a)
`
`E' c
`"' 0 ...
`
`0
`0
`
`(b)
`
`E' c
`"' 0
`::!.
`0
`0
`
`0
`
`1-1
`
`JGT
`
`BC2
`30. 6
`Inhibitor
`
`BC3
`
`Ly2. 1
`
`Fig. 4. (a) Reaction of HAMA in the serum of a
`representative patient (Table 2: Patient 1) with various
`murine MoAb, relevant (1-1 :lgG I) and irrelevant
`(JEI ·2: lgM, BC2: lgG I) and to polyclonal antibodies
`from monkey (Monk) and sheep. The various anti(cid:173)
`bodies were coated on the plate and solid phase bind(cid:173)
`ing assays performed similar to HAMA assay; a
`1 :2000dilution of serum was used. The binding of pre(cid:173)
`existing HAMA in the same patient to the various anti(cid:173)
`bodies was also shown.
`(b)
`Inhibition ofHAMA activity by different murine
`MoAb preparations. Serum of Patient I was incubated
`with relevant MoAb (1-1 , JGT; both lgG 1) or irrel(cid:173)
`evant murine MoAb of different isotypes (30·6:
`lgG2b, BC2: lgG I, BC3: lgM, Ly 2· I: lgG2a) in dif(cid:173)
`ferent concentrations (250 µg/mL or I µg/mL) and
`tested in HAMA assays by measuring the binding of
`the reaction mixture to MoAb 1-1 coated plate. Only
`the results of a representative patient (Patient 1) are
`shown, for clarity. Results were compared to the bind(cid:173)
`ing of the same respective serum sample in HAMA
`assay in the absence of a blocking antibody (con(cid:173)
`trol).
`
`6 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`HUMAN ANTl-MURlNF. IMMUNOGLOBULIN RESPONSES
`
`373
`
`e c
`"' 0
`:!.
`0
`0
`
`e c
`"' 0
`:!.
`0
`0
`
`l•l
`
`2
`
`0
`102
`
`(b)
`
`06
`
`05
`
`0 4
`
`03
`
`02
`
`0 1
`
`00
`1 o~
`
`103
`
`10•
`
`106
`
`Serum d1lu11on· 1
`
`(•)
`
`(b)
`
`3
`
`2
`
`e c
`"' 0
`:!.
`0
`0
`
`e
`c
`"' 0
`:!.
`0
`0
`
`Serum d1lu11on· •
`
`10 3
`
`10•
`
`106
`
`Serum d1lu1ton-1
`
`10•
`
`Serum d1lu11on -1
`
`Fig. 5. Levels of (a) lgG and (b) lgM HAMA detected
`m the serum of a reprc~cntative patient (Table 2;
`Patient 6) before (a ) and after(• ) therapeutic mject1on
`of MoAb-l\-AcMEL conjuµtes
`
`ll1IM1I respOlll<' to Fe and F(ab'), of mouse
`i m111 unoglobu/i11
`in
`the relative contribution
`To determine
`patients' sera of antibodies to Fe and F(ab'h
`portions of the injected murine 1mmunoglobu(cid:173)
`lin. ELISA plates were set up with whole IgG
`MoAb, or with equimolar amounts of the F(ab'h
`fragments of the same immunoglobulin; and the
`binding of patients' sera to these different prep(cid:173)
`arations was assayed using serum samples with
`an elevated HAMA response. In a representative
`patient (Fig.6). the scrum displayed elevated
`responses to both the whole munne 1mmuno(cid:173)
`globulins and the F(ab'h fragments. The reac(cid:173)
`tions to the whole lgG of both relevant (Fig. 6a)
`and irrelevant (F1g. 6b) antibodies were greater
`than to F(ab')i fragments. although this could
`
`Fig. 6. Elevated HAMA responses made by a repre(cid:173)
`sentat1 vc patient (from imaging studies) to whole lgG
`(• )and f(ab')i (• ) of(a) relevant (RC'C-1) and (b) irrel(cid:173)
`evant (Ly-2 I) murine MoAb; both were of lgG2a sub(cid:173)
`class Relevant an11bod\ refers to the administered
`·
`antibody.
`
`reflect differences in the amount of antigen
`bound. The pre-immune sera abo displayed a
`similar pattcm of reaction. although the binding
`was less. Thus the HA.MA response was directed
`against both the Fe and F(ab'h of mouse
`immunoglobuhn.
`
`DI CUSSION
`The HAMA responses in patients who received
`munne MoAb for diagnostic stud ies or for ther(cid:173)
`apy were examined in this stud) . lmmunolym(cid:173)
`phoseint1graphy, using subcutaneously adminis(cid:173)
`tered antibodies in amounts ranging from SO
`µg 3 mg was clinically safe (3.2 t) and was
`incidence (4/67) of
`low
`associated with a
`
`7 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`J. J TJANDRA f.T .H .
`
`HAMA, as measured by the ELISA system used
`1n the study. In addition. the degree of response
`tends to be mild to moderate CHAMA titres I 00-
`6400). as defined in the study This is in contrast
`to the higher incidence ofHAMA observed with
`some studies of immunoscintigraphy which
`employed larger doses (0· 16-20 mg) of munne
`MoAb injected intravenously (29) or indeed
`with the larger amounts of antibody given intra(cid:173)
`arterially in this study. The difference in im(cid:173)
`mune response may be because the dose used
`was small (50 µg-3 mg) and was injected subcu(cid:173)
`taneously rather than intravenously. A similar
`lack of elevated HAMA response had been
`reported by others who used small doses (250
`µg) of murine antibodies injected intravenously
`(8). However. there arc other possible expla(cid:173)
`nations, such as 131 J-MoAb being taken up in B
`cells which may be inactivated by the 'hot ' anti(cid:173)
`gen, although such an inactivation did not occur
`after intravenous use of radiolabelled ant1bod1es
`(29). Another major difference for consideration
`is that the melphalan-antibody conjugate could
`be more immunogenic than either MoAb or t lt ((cid:173)
`MoAb used in the imaging procedure: we have
`no information on whether this is the case.
`However. the low incidence of HAMA response
`1n imaging is of practical importance as it indi(cid:173)
`cates that such diagnostic procedures can be
`repeated in most patients and do not preclude
`future therapy with murine MoAb.
`By contrast. hepattc artery infusion of murme
`MoAb-.\-acetyl melphalan conJugates m three
`repetitive doses over 48 h elicited prompt and
`dramatic immune response
`to the murine
`immunoglobulins mall ten patients. An elevated
`HAMA response was noted m a patient as early
`as the fifth day followmg exposure to the
`foreign protein: however. the peak values of
`HAMA response usually occurred 14-28 days
`following exposure. The degree of HAMA
`response did not correlate with the amount of
`murine antibodies received within the dose
`range of 120-1000 mg!m2 (Table 2). and there
`was no correlation between the degree of
`immune response in the patients and the clinical
`response (7). The high mc1dence of elevated
`HAMA response in this group of patients could
`be related to the larger doses used. its repeated
`(three doses) daily administration and the route
`of adminstration by hepatic artery infusion.
`This is in accord with the high incidence of
`HAMA ( 17/ 18 patients) that was detected after
`repeated daily i.v. mfusions of an lgG I antibody
`m patients with pancreatic cancer (28). It is of
`interest that the frequency of HAMA had been
`noted to be lower (50%) \\.ith single mfus1ons of
`
`murine antibody 17-IA (lgG2a) (30). Although
`others had reported a reduction of this anti(cid:173)
`murine response when paltcnts received high
`mitial doses of MoAb 17-1 A (700 mg) by 1. v
`infusions. and had attributed that to induction
`of immune tolerance (31 ). this was not the
`experience of this study. The repeated daily
`administrations of the immunoconJugatcs via
`hepatic artery infusion before the development
`of detectable elevated HAMA response were
`well tolerated: by contrast. injection of the
`immunoconjugates m the presence of an ele(cid:173)
`vated HAMA response was assoc1ate<l with an
`initial reduction in the llAMA titres which
`probably corresponded to immune complc' for(cid:173)
`mation and with clinical symptoms suggestive of
`a Type III hypcrscnsittv1ty 'scrum sickness'
`reaction. followed by a boost m the I !AM\
`response (Fig. 3). The presence of an elevated
`HAM'A response therefore precludes any further
`exposure to murinc 1mmunoglobulms. Where
`increased anti-munne antibody levels were
`detected after exposure to murinc ant1bod1cs
`either for imaging or for therap). the HAMA
`response was polyclonal. containing increased
`levels of both IgM and lgG and was directed
`against mouse-specific determinant!.. the iso(cid:173)
`type (presumed to he the Fe pon1on). the
`f(ab'h, as well as the '1d1otypc' (1.e. anti-indi(cid:173)
`vidual injected antibody) of mouse 1mmuno(cid:173)
`globulin. The contradictory reports on the nat(cid:173)
`ure of the HAMA response. especially with
`regard to the ·anti-1d1otypic' component could
`be related to the different assay systems used to
`evaluate the response (8. 10.32.31).
`The low level of human ant1-munnc antibody
`response detected m the serum of normal sub(cid:173)
`JCCts and 10 patients prior to administration of
`murinc MoAb. is in accord with other studies
`(8-10) and there is e\ 1dence that this at least in
`part reflects rheumatoid factor activity in nor(cid:173)
`mal human sera and not a prc·.cxisttng specific
`antibody for murine immunoglobultn (34). 1t
`appeared that a significant component of the
`elevated HAMA response in patients gi,en
`murine MoAb represented secondary response
`to antigenic determinants common to mouse
`1mmunoglobulins and some unknown 1mmu(cid:173)
`nogen. to which low levels of sensitization had
`already occurred. However, the predominant
`component of the elevated HAMA response was
`directed against mouse-specific determinants
`(Fig. 4a) and it is generally conceded (35) that
`the antibody response developed by pat icnts was
`specifically related to the species from which the
`administered antibody was derived.
`The findings of this study have several clinical
`
`8 of 11
`
`Celltrion, Inc., Exhibit 1065
`
`
`
`HUMAN ANTl-MURINE IMMUNOGLOBULIN RESPONSES
`
`375
`
`implications. First. as the human anti-murine
`antibody response was polyclonal and was
`directed against the antigenic determinants on
`the constant. Fe. as well as the F(ab'h portion of
`the mouse immunoglobulin molecule. the use of
`F(ab'h fragments would not confer a real advan(cid:173)
`tage over intact lgG, as illustrated by the ele(cid:173)
`vated HAMA response in 1/3 patients given
`F(ab'h in imaging studies (Table I). Second. the
`administration of murine antibodies subcu(cid:173)
`taneously in doses up to 3 mg was associated
`with a low incidence (4/67 or 6%) or elevated
`HAMA response and thus would not, in most
`cases. preclude subsequent exposure to murine
`
`antibodies. Third, repeated (three doses) daily
`administrations of murine antibodies over 48 h
`in doses up to I 000 mglm2 by hepatic artery
`infusion was not associated with untoward com(cid:173)
`plication, provided there was no prior immu(cid:173)
`nization
`to murine
`immunoglobulins
`(7).
`Finally, the use of steroids in the peri-therapy
`period did not prevent the development of'posi(cid:173)
`tive' HAMA response.
`
`Acknowledgements The authors thank Marie Pic