Journal of Immunotherapy
`15:42-52 © 1994 Raven Press, Ltd., New York
`
`Human Immune Response to Monoclonal Antibodies
`
`M. B. Khazaeli, Robert M. Conry, and Albert F. LoBuglio
`
`Comprehensive Cancer Center, Department of Medicine, University of Alabama at Birmingham,
`Birmingham, Alabama, U.S.A.
`
`Summary: Monoclonal antibodies either of mouse, mouse/human chimeric, or
`human origin have been used safely in human trials for a decade. Considerable
`effort has been committed in investigations of manipulating endogenous im(cid:173)
`munological activity against tumors and targeting various cytotoxic agents to
`cancers. These studies have identified several problems that need to be re(cid:173)
`solved before any such reagents can be used routinely in patients. One of these
`problems has been the immunogenicity of these monoclonal antibodies. This
`review article discusses what is known regarding human immune response to
`monoclonal antibodies and their clinical consequences. Key Words: HAMA(cid:173)
`Monoclonal antibodies-Chimeric monoclonal antibodies-Human monoclo(cid:173)
`nal antibodies.
`
`There has been and continues to be considerable
`optimism regarding the clinical application of mono(cid:173)
`clonal antibodies in human disease. Several re(cid:173)
`agents have been tested and approved for clinical
`use in the United States, Europe, Japan, and else(cid:173)
`where (1-3). However, it is also clear that initial clin(cid:173)
`ical investigations of monoclonal antibodies in vivo
`have identified several problems that hinder the ulti(cid:173)
`mate utility of such reagents. One of these problems
`has been the immunogenicity of these laboratory(cid:173)
`generated proteins, resulting in considerable investi(cid:173)
`gative efforts to characterize the immune response
`and seek strategies that reduce, delay, or obviate this
`problem. This article will review current knowledge
`regarding human immune response to monoclonal an(cid:173)
`tibodies as well as clinical consequences of human
`anti-mouse antibody (HAMA) development.
`
`IMMUNOGENICITY OF PROTEINS
`
`Immunogenicity refers to the ability of a molecule
`to induce an immune response in a particular host
`
`Received April 30, 1993; accepted June I, 1993.
`Address correspondence and reprint requests to Dr. M. B.
`Khazaeli at Comprehensive Cancer Center, University of Ala(cid:173)
`bama at Birmingham, 1824-6th Avenue South WTI 263, Birming(cid:173)
`ham, AL 35294-3300, U.S.A.
`
`or recipient. This is different than antigenicity,
`which refers to a molecule having antigen sites
`(epitopes), which can be demonstrated by one or
`another in vitro assay, e.g., antibody binding or
`lymphocyte transformation. Murine monoclonal an(cid:173)
`tibodies are generally classified as homologous pro(cid:173)
`teins, i.e., the immunoglobulin molecules of mice
`share considerable homology with those of humans.
`Immunogenicity of homologous proteins has been
`the subject of consi,detable research, utilizing such
`models as myoglobin, cytochrome C, etc. (4). The
`host immune response to such proteins has been the
`classic example of multicellular cooperation in the
`immune system and is outlined schematically in
`Fig. 1. This figure depicts the several cell types and
`steps involved in generating an antibody response
`to such homologous proteins. The first cell type has
`been termed the antigen presenting cell (APC),
`which may be a macrophage, B-lymphocyte, or
`other cell type, and which is required to internalize
`the large molecular weight protein, digest the pro(cid:173)
`tein into peptide fragments of 6-15 amino acid
`lengths, and present such peptides on the cell sur(cid:173)
`face in association with major histocompatibility
`(MHC) class II molecules for interaction with a sec(cid:173)
`ond cell population, i.e., T-helper cells. Recogni(cid:173)
`tion of peptides displayed on the surface of APCs
`
`42
`
`

`

`HAMA
`
`43
`
`FIG. 1. Schematic representation of antigen
`processing and T-cell help necessary for anti(cid:173)
`body production.
`
`requires MHC complementarity (MHC restricted)
`and involves specific recognition by T-cell recep(cid:173)
`tors reinforced by several other complementary
`T-cell surface molecules and APC surface recep(cid:173)
`tors. Such T-helper cell binding triggers a response
`that includes T-cell proliferation, release of inter(cid:173)
`leukin (IL)-2 and other lymphokines, appearance of
`IL-2 receptors, and interaction with naive B-cells to
`generate multiple B-cell clones that proliferate and
`differentiate into plasma cells that produce antibody
`with specificity to a variety of epitopes on the orig(cid:173)
`inal protein immunogen.
`It is clear that the epitopes that interact with the
`T-cell receptor are few in number and represent
`short peptide segments composed of a linear se(cid:173)
`quence of amino acids in the proteins. Epitopes that
`are targets of specific antibodies are many in num(cid:173)
`ber and usually represent topographic areas on the
`surface of the intact protein molecule, representing
`amino acids that are adjacent to each other on the
`intact "folded" protein, but not adjacent to each
`other in the amino acid sequence of the protein (to(cid:173)
`pographic epitopes). This distinction is important
`since it appears that the immunogenic potential of a
`protein is determined by a relatively few T-cell spe(cid:173)
`cific epitopes that then trigger a broad B-cell anti(cid:173)
`body response. Thus, alteration of a few amino ac(cid:173)
`ids can dramatically alter the immunogenicity of a
`homologous protein (reviewed in 4). This obviously
`has important implications as we enter the era of
`genetically designed molecules.
`
`METHODOLOGY USED TO DETECT
`IMMUNE RESPONSE
`
`Studies to characterize the human immune re(cid:173)
`sponse to monoclonal antibodies have primarily in(cid:173)
`volved analysis of patients' postexposure serum
`
`:~:: _J
`
`~ - - . , IL-5
`IL-6
`IFN-Y
`
`for the presence of antibody, i.e., human anti(cid:173)
`monoclonal antibody (HAMA) response. Consider(cid:173)
`ably less information is available regarding T-cell
`immune response or lgE/allergic response.
`Figure 2 shows the various HAMA assay strate(cid:173)
`gies used to detect antibody response to monoclo(cid:173)
`nal reagents. Solid phase assays (Fig. 2A) are com(cid:173)
`posed of the monoclonal reagent bound to a solid
`phase, exposed to test sera, washed, and exposed
`to an enzyme linked (ELISA) or radiolabeled (ra(cid:173)
`diometric) anti-human immunoglobulin reagent.
`This approach has several advantages, including
`commercial availability of reagents and equipment
`(ELISA format) high affinity of antibody to solid
`phase antigen, and ability to characterize immuno(cid:173)
`globulin class and subclass (5,6). Its drawbacks are
`that high affinity binding of bound antigen (mono(cid:173)
`clonal antibody) results in high "background" bind(cid:173)
`ing with normal sera requiring dilution of serum
`samples ranging from 1: 10 to 1 :500 for standard as(cid:173)
`says or use of multiple dilutions of each sera with
`construction of complicated manipulations of data
`for readout values. The specificity and affinity of
`
`Detection System
`
`Anti-Human lgG61
`
`Detection System
`
`>(cid:173)
`>(cid:173)
`>-
`
`Detection System
`
`+ sera
`
`HPLC
`SPA
`
`C. Soluble phase
`
`>(cid:173)
`>(cid:173)
`>-
`
`~ =enzyme or isotope
`}- = monoclonial antibody
`FIG. 2. Schematic representation of HAMA assays. A: Solid
`phase assays. B: Double antigen assays. C: Liquid-phase assays.
`
`J Immunother, Vol. 15, No. 1, 1994
`
`

`

`44
`
`M. B. KHAZAELI ET AL.
`
`the anti-human immunoglobulin reagent represents
`a second variable that influences assay results, and
`the format is not readily applied to antichimeric,
`humanized, or human monoclonal antibodies. The
`second assay format utilizes a "double antigen" de(cid:173)
`sign (Fig. 2B), in which serum antibodies not only
`initially bind solid phase antigen (monoclonal anti(cid:173)
`body) but also are required to bind soluble antigen
`(radiolabeled or enzyme linked) in order to be de(cid:173)
`tected (7). This requirement for binding of soluble
`antigen (monoclonal antibody) excludes detection
`of low affinity or nonspecific human lg binding to
`the solid phase and allows the assay to be used with
`undiluted serum (8). An additional advantage of this
`approach is that it can be used with sera from any
`species, with chimeric or human monoclonal anti(cid:173)
`bodies, and allows for convenient study of epitope
`specificity in competitive inhibition studies (9). The
`disadvantage of this approach is that reagents are
`not commercially available, so that preparation and
`quality control of enzyme-linked or radiolabeled
`monoclonal antibodies are required, and the system
`does not provide lg class or subclass data. A com(cid:173)
`mercial kit using this format is available, but has the
`limitation that it does not detect anti-V-region (anti(cid:173)
`id) response (10). The third assay format (Fig. 2C) is
`the least frequently reported and utilizes human an(cid:173)
`tibody binding of soluble antigen (monoclonal anti(cid:173)
`body) with subsequent partition of antigen-antibody
`complexes by high-performance liquid chromatog(cid:173)
`raphy (HPLC) (11, 12) or Staphlylococcus protein A
`(SPA) binding (13). The lack of popularity reflects
`the need for expensive equipment (HPLC) or non(cid:173)
`specificity of the SPA system. These three formats
`account for almost all the HAMA studies in the
`literature, although a few other procedures have
`been used, such as passive hemagglutination test
`(14, 15).
`Assay for cellular immune response to monoclo(cid:173)
`nal antibodies have utilized lymphoblastic transfor(cid:173)
`mation assays using whole mononuclear cells
`(16, 17) or cloned T-cell (16). Detection of T-helper
`binding and response can also be reflected in ap(cid:173)
`pearance of IL-2 receptors or their release into the
`supernate (see Table 1).
`The most problematic area of human response to
`monoclonal antibodies has been the determination
`of allergic (lgE) responses. There has been no se(cid:173)
`rologic assay for detection of lgE response to
`monoclonal antibodies that has withstood critical
`analysis. A major technical problem is the presence
`of lgG and lgM human antibodies to mouse lg; re-
`
`J Immunother, Vol. 15, No . 1, 1994
`
`TABLE 1. T-cell responses to monoclonal antibodya
`
`Patientb
`
`3H-thymidine
`proliferation
`
`1251 IL-2
`binding
`
`A
`B
`C
`
`9
`lO
`37
`
`7
`12
`24
`
`Soluble IL-2
`receptor
`ELISA
`
`11.
`6
`Not done
`
`a All values represent the stimulation ratio, (stimulated cells)/
`(control cells), using l00-300 µg/ml of monoclonal antibody
`CB72.3.
`b Three representative patients 6-9 months following treat(cid:173)
`ment with CB72.3.
`
`suiting in false-positive assays that utilize murine
`anti-lgE reagents or compete for antigen binding
`with lgE antibodies. Many studies have utilized
`skin testing (18,19) or intravenous test doses (20) to
`screen for the allergic state, but anaphylactic reac(cid:173)
`tions (readily reversible) have occurred despite
`such screening efforts (20-22).
`
`ASSAY STANDARDIZATION
`AND SENSITIVITY
`
`The issues of standardization and sensitivity of
`HAMA assays have been problematic and contro(cid:173)
`versial. These issues readily apply to quantitation of
`a specific substance or class of molecules, e.g., glu(cid:173)
`cose, cholesterol, carcinoembryonic antigen (CEA),
`or immunoglobulin G or even a specific monoclonal
`antibody (23), where a standard curve of the sub(cid:173)
`stance can be used to quantitate the same substance
`in an unknown sample. However, it is clear that
`each animal ( or person) responds to an immuno(cid:173)
`genic challenge with a unique heterogeneous poly(cid:173)
`clonal response composed of various immunoglob(cid:173)
`ulins, subclasses, and antigen-combining sites. This
`results in sera having varying strengths of antibody
`binding to the target molecule, i.e., varying anti(cid:173)
`body affinity. It is also clear that individual varia(cid:173)
`tion in binding affinity can vary over 2-3 logs, so
`that the ability of each antibody response to pro(cid:173)
`duce a biologic effect (e.g., binding to solid phase or
`soluble antigen) could vary by several logs. This
`variation is influenced by mode of immunization,
`sampling time postexposure, use of booster doses,
`or adjuvants, and the unique immunoregulatory
`makeup of the individual host. Obviously, all of the
`assays described above measure a biologic activity,
`and use of a standard, affinity-purified antibody
`(immunoglobulin) from human (6,24), primate (10),
`goat (25), or rabbit (26) for a standard curve simply
`
`

`

`HAMA
`
`45
`
`tells you the amount of immunoglobulin in that in(cid:173)
`dividual animal that would produce the biologic ac(cid:173)
`tivity measured, but does not quantitate the amount
`of immunoglobulin producing the same effect in
`other animals of the same or different species.
`Thus, using such standards to measure "sensitiv(cid:173)
`ity" of an assay reflects the binding affinity of the
`standard chosen and will vary with any particular
`standard in the same assay. The use of such stan(cid:173)
`dard curves has led to considerable confusion in the
`literature. For example, several laboratories (6,24)
`have used human antibody or immunoglobulin to
`construct standard curves to quantitate HAMA.
`They have found that sera from normal individuals
`contain 1-10 µg/ml of HAMA and that following
`exposure to mouse monoclonal antibodies, the re(cid:173)
`sponse increases to 20-1,000 µg/ml. Other groups
`have used affinity-purified antibody for standard
`curve with a "sensitivity" of 0.015-.0001 µg/ml
`(26,27). It is unlikely that either observation reflects
`reality. Values read from such standard curves
`should be called "antibody equivalents" and inter(cid:173)
`preted as arbitrary units of activity. The assays
`used to measure HAMA in humans actually mea(cid:173)
`sure the amount of human immunoglobulin binding
`to an antigen under a certain set of experimental
`conditions. Appropriate standardization requires
`that a known positive control be run with each as(cid:173)
`say to be sure that the assay is performing as ex(cid:173)
`pected and providing reproducibility data. "Sensi(cid:173)
`tivity" of an assay for human immune response is
`best related to the frequency of detection of an im-
`
`mune response with a specific assay design in clin(cid:173)
`ical trials rather than to the minimum amount of an
`arbitrary antibody preparation able to be detected.
`
`IMMUNE RESPONSE TO MURINE
`MONOCLONAL ANTIBODIES
`
`Administration of xenogeneic murine monoclonal
`antibodies in humans has not, unexpectedly, re(cid:173)
`sulted in the frequent development of HAMA.
`HAMA response has occurred across a wide dose
`range from < 1 mg to > 1 g of monoclonal antibody
`(8,28) and appears to be more likely to occur with
`repeated injection (29). The F Ab fragment of murine
`monoclonal reagents appears to be less immuno(cid:173)
`genic than the intact molecule presumably due to its
`short plasma half-life and absence of immunogenic
`epitopes on the C-2 and C-3 components of the con(cid:173)
`stant regions (28).
`As listed in Table 2, patients with solid neoplasms
`generally have a 50-75% incidence of HAMA re(cid:173)
`sponse when exposed to monoclonal antibodies
`(30-49). It is interesting that all 56 melanoma pa(cid:173)
`tients from three trials developed HAMA to R24, a
`monoclonal antibody to the glycolipid GD3 (42-44)
`and 30 of 30 melanoma patients treated with two
`different anti-GD2 monoclonal antibodies devel(cid:173)
`oped HAMA (45,47). This suggests that GD2 and
`GD3 specificity may enhance immunogenicity and
`it may be pertinent that R24 (anti-GD3) appears to
`bind and promote proliferation of a subpopulation
`of T-cells (50). Concurrent administration ofIL-2 or
`
`TABLE 2. HAMA response in patients with solid neoplasms receiving mouse monoclonal antibodies
`
`MoAb
`
`17-lA
`17-lA
`17-lA
`17-lA
`17-lA
`17-lA
`B72.3
`B72.3
`B72.3
`B72.3
`B72.3
`HFMG-1
`R24
`R24
`R24
`14O2a
`96.5; 48.7
`3F8
`NP-2
`L6
`
`Cancer
`
`No. patients
`
`% Positive
`
`Reference
`
`GI
`GI
`GI
`GI
`GI
`GI
`GI
`GI
`GI
`GI
`GI
`Ovarian
`Melanoma
`Melanoma
`Melanoma
`Melanoma
`Melanoma
`Melanoma
`CEA-producing carcinoma
`Breast
`
`65
`25
`20
`71
`14
`161
`110
`28
`35
`30
`97
`10
`12
`15
`21
`13
`4
`17
`29
`19
`
`54
`92
`85
`77
`93
`63
`25
`16
`52
`50
`49
`50
`100
`100
`100
`100
`100
`100
`86
`68
`
`31
`32
`33
`34
`35
`36
`37
`38
`39
`40
`41
`56
`42
`43
`44
`45
`46
`47
`48
`49
`
`J lmmunother, Vol. 15, No. 1, 1994
`
`

`

`46
`
`M. B. KHAZAELI ET AL.
`
`interferon does not appear to alter the incidence of
`HAMA response (35,44). Table 3 reviews the
`HAMA response in trials of patients with hemato(cid:173)
`logic malignancies. In general, patients with chronic
`lymphocytic leukemia or B-cell lymphoma appear
`to have reduced frequency of HAMA response
`while patients with acute myeloid leukemia and cu(cid:173)
`taneous T-cell lymphoma have a frequency similar
`to solid tumor patients (6,24,51-56).
`The HAMA response is not detected prior to day
`8, usually reaches a peak at 2-6 weeks, and may
`persist for many months. Upon initial exposure to
`murine monoclonal reagents, a typical primary im(cid:173)
`mune response is noted with an initial IgM antibody
`response and subsequent peaking of the IgG re(cid:173)
`sponse including all IgG subclasses (57). Anamnes(cid:173)
`tic responses with large amounts of HAMA may
`follow second exposure to antibody usually of the
`IgG class of immunoglobulins (57 ,58).
`The specificity of the HAMA response in regard
`to reactivity with different regions of the murine
`immunoglobulin molecule has been variable. In
`general, a pattern of early reactivity with murine
`constant regions followed by increasing specificity
`for V-regions (anti-idiotypic), particularly after re(cid:173)
`peated infusions, has been noted (6,24,35,40). How(cid:173)
`ever, we have shown that murine B72.3 can induce
`a rapid and frequent response to B72.3 V-regions
`(9), and a similar finding was reported in transplant
`patients receiving OKT3 (59). HAMA specificity
`for the complimentary determining region (CDR) on
`murine monoclonals has been reported as result of
`inhibition of antigen binding (60) or inhibition of
`anti-idiotypic antibody binding to the murine mono(cid:173)
`clonal antibody (61).
`The HAMA response appears to involve the clas(cid:173)
`sic characteristics of an immune response to protein
`
`TABLE 3. HAMA response in patients with
`hematologic malignancies
`
`MoAb
`
`Cancer
`
`No.
`patients
`
`%
`Positive
`
`Reference
`
`TIO!
`TIO!
`TIO!
`TIO!
`TIO!
`TIO!
`Anti-idio
`OKB7
`MB-I
`Ml95
`
`CTCL
`CTCL
`CTCL
`CTCL
`CLL
`CTCL
`NHL
`NHL
`NHL
`AML
`
`5
`6
`4
`10
`6
`24
`11
`18
`10
`6
`
`100
`100
`0
`50
`0
`29
`45
`28
`0
`67
`
`51
`52
`53
`6
`6
`24
`95
`54
`55
`56
`
`CTCL, cutaneous T-cell lymphoma; CLL, chronic lymphocytic leuke(cid:173)
`mia; NHL, non Hodgkins lymphoma; and AML, acute myeloid leukemia.
`
`J Immunother, Vol. 15, No . /, 1994
`
`antigens. Lanzavecchia et al. (16) demonstrated
`that the immune response to several murine mono(cid:173)
`clonal reagents induced classic human leukocyte
`antigen-restricted T-helper cell binding and prolif(cid:173)
`eration, dependent upon antigen-presenting cells.
`They also showed that the T-cell response included
`T-cell cytolytic activity to murine monoclonal(cid:173)
`coated tumor cells. More recently, these observa(cid:173)
`tions were extended by Kosmas (17), who demon(cid:173)
`strated T-cell lymphoblastic transformation to
`HMFG I monoclonal antibody in 11 of 13 patients
`with ovarian cancer with evidence of memory cells
`present as long as 1 year after therapy. The lym(cid:173)
`phocyte response included increased expression of
`IL-2 receptor on responding lymphocytes. Obser(cid:173)
`vations in our laboratory (Table 1) have shown that
`the in vitro lymphocyte response to monoclonal an(cid:173)
`tibodies includes proliferation, increased IL-2 bind(cid:173)
`ing (receptor activity), and release of soluble IL-2
`receptor into the supernate (62).
`
`IMMUNE RESPONSE TO
`GENETICALLY-ENGINEERED MOUSE/HUMAN
`MONOCLONAL ANTIBODIES
`
`In an attempt to alter the immunogenicity of mu(cid:173)
`rine monoclonal antibodies, techniques have been
`developed using recombinant DNA technology to
`utilize antigen-combining sites derived from murine
`hybridoma technology, with human constructs for
`the remainder of the molecule. In general, two
`types of reagents have reached clinical trials.
`The first genetically-engineered molecules de(cid:173)
`signed and tested were termed "chimeric" (ch)
`monoclonal antibodies, which are composed of mu(cid:173)
`rine V-regions combined with human heavy and
`light chain constant regions (63). Table 4 provides a
`summary of immune response to chimeric antibod(cid:173)
`ies (published or presented to date). All studies uti(cid:173)
`lized human IgG 1 and k constant regions, except for
`ch B72.3, which was IgG4, k. The two trials with ch
`17-lA demonstrated a very low frequency of HAMA
`even with repeated exposure over a wide dose
`range (64,65). In contrast, our studies (58,66,67) and
`those of Baker et al. (68) both demonstrated a fre(cid:173)
`quency and amount of HAMA to ch B72.3 similar to
`that seen with murine B72.3. Trials of ch 14.18 in
`adults with melanoma (69) and children with neuro(cid:173)
`blastoma (70) demonstrated a relatively high fre(cid:173)
`quency of HAMA but the amounts of antibody ap(cid:173)
`peared to be 1-2 logs less than the HAMA response
`to murine 14G2a, the murine source of V-regionfor
`
`

`

`HAMA
`
`47
`
`Antibody
`
`ch17-1A
`ch17-1A
`chB72.3
`chB72.3
`chB72.3
`chl4.18
`ch14.18
`chMT412
`ch L6
`chCD4
`ch NRL-13
`
`TABLE 4. HAMA response to mouse/human chimeric monoclonal antibodies
`
`Cancer
`
`No. patients
`
`% Positive
`
`Reference
`
`Colorectal
`Colorectal
`Colorectal
`Colorectal
`Colorectal
`Melanoma
`Melanoma
`RRA
`Breast
`Mycosisfungoides
`Colon
`
`10
`6
`12
`12
`6
`13
`7
`25
`18a
`7
`9
`
`10
`0
`58
`75
`50
`54
`71
`4
`22a
`29
`
`64
`65
`67
`66
`68
`69
`70
`73
`
`72
`71
`
`RRA, refractory rheumatoid arthritis.
`a Goodman G. E., Hellstrom I., Yelton D. E., et al., unpublished data.
`
`ch 14.18. Trials with ch L6 and ch NRLU13 in solid
`tumor patients appeared to have a lower frequency
`as well as lower amount of HAMA compared to
`trials with their murine counterparts (Goodman
`G. E., Hellstrom, I., Yelton, D. E., et al., unpub(cid:173)
`lished data; 71). A chimeric anti-CD4 trial in pa~
`tients with mycosis fungoides noted two of seven
`patients developing HAMA to the murine V-region
`(72), while a second study with a different chimeric
`anti-CD4 reagent (ch MT412), in patients with rheu(cid:173)
`matoid arthritis, had only one of 25 patients devel(cid:173)
`oping antibodies to the murine V-region (73).
`In general, these immune responses have been
`directed to the murine V-region with characteristics
`of an anti-idiotypic response. No alloantibodies to
`human constant region have been documented to
`date. We have described a novel specificity to
`epitopes that appears to require components of both
`murine V-region and human constant region in our
`trial with ch B72.3 (67). These studies indicate that
`chimeric monoclonal antibodies are, in general, less
`immunogenic than murine monoclonal reagents, but
`that their degree of immunogenicity can vary
`greatly due to yet to be defined mechanisms.
`The second type of murine/human constructs are
`the so-called CDR grafted or "humanized" mono(cid:173)
`clonal antibodies, which utilize not only human
`constant regions but also human framework of the
`V H and V L regions. This reduces the murine com(cid:173)
`ponent to the CDR regions and selects other vital
`residues required for CDR orientation (74). Scant
`information on the immunogenicity of these mole(cid:173)
`cules is currently available. Campath-lH, a human(cid:173)
`ized anti-lymphocyte antibody, has been given to
`two patients with B-cell lymphoma (75) and one pa(cid:173)
`tient with systemic vasculitis (76) with excellent
`clinical responses and no HAMA. However, these
`
`patients undoubtedly had impaired immune respon(cid:173)
`siveness. A preliminary report of a humanized anti(cid:173)
`placental alkaline phosphatase (anti-PLAP) conju(cid:173)
`gated to 1111n via a 1,4,7,10-tetra azacyclodode(cid:173)
`cane-N,N1 ,N2 ,N3-tetraacetic acid (DOT A) linker
`demonstrated an anti-DOTA response without
`HAMA (77). The apparent dosage reported in this
`article were a few hundred micrograms with rela(cid:173)
`tively brief plasma circulation, so that the immuno(cid:173)
`genicity of this reagent will require further study. A
`variety of humanized monoclonal antibodies are ap(cid:173)
`proaching clinical trials and a clearer picture of their
`immunogenicity will be forthcoming.
`
`HUMAN MONOCLONAL ANTIBODIES
`
`Clinical trials with analysis of immune response
`to human monoclonal antibodies have been re(cid:173)
`ported for four lgM and one lgG monoclonal re(cid:173)
`agents. The first report involved a human lgM
`monoclonal antibody to GD2 ganglioside called
`L72, which was injected directly (intralesionally)
`into cutaneous metastatic melanoma nodules in
`eight patients (78). Total doses administered varied
`from 3-15 mg subdivided into two-four treatments.
`They utilized a passive sheep erythrocyte aggluti(cid:173)
`nation assay for detection of antibody response and
`reported that five of eight patients gave either mild
`or strong agglutination during the course of treat(cid:173)
`ment. Steis et al. reported on two human IgM
`monoclonal antibodies (16.88 and 28A32) directed
`to colon cancer-associated antigens in phase I trials
`with metastatic colon cancer patients (79). A latex
`agglutination assay was used to monitor antibody
`response with no evidence of antibody response to
`MAb 16.88 in 12 patients receiving repeated injec(cid:173)
`tions. The latex agglutination assay, however, had a
`
`J Immunother, Vol. 15, No. 1, 1994
`
`

`

`48
`
`M. B. KHAZAELI ET AL.
`
`high degree of "nonspecific" positivity with MAb
`28A32, 5/12 normal sera, 7/23 colon cancer patients
`prior to receiving antibody and in 8/14 patients at
`some time after multiple infusions of 28A32. These
`were all of "low titer," i.e., 1/10 dilution.
`Clinical trials with human monoclonal reagents
`for infectious diseases have included phase I trials
`with an IgM antibody to endotoxin (HA-lA) in 15
`nonseptic patients with cancer (80) and 34 patients
`with sepsis (81) using single infusions of 50 µg-250
`mg. A "double antigen" radiometric assay detected
`no evidence of antibody response to HA- lA. More
`recently, a large trial of HA-lA in septic patients
`randomized patients to a single dose of 100 mg of
`HA-lA or placebo (1). Using the same assay sys(cid:173)
`tem, no antibody was detected in 262 patients re(cid:173)
`ceiving HA-lA or in 281 patients receiving placebo.
`Azuma et al. reported on a phase I trial of a human
`IgG monoclonal antibody to cytomegalovirus (26) in
`which healthy volunteers received either single (n
`= 16) or three repeated doses (n = 4) of 10-100 mg.
`A "double antigen" ELISA assay was utilized with
`no evidence of antibody response in any patient.
`These studies suggest that human monoclonal an(cid:173)
`tibodies have low immunogenicity. Potentially,
`anti-idiotypic antibody responses may well occur,
`but will probably require repeated doses and/or sub(cid:173)
`cutaneous/intratumor injections.
`
`IMMUNE RESPONSE TO MAb-CONJUGATES
`
`Conjugation of various toxic moieties to mono(cid:173)
`clonal antibodies has been one strategy to enhance
`antitumor effects. As regards radioactive isotope
`conjugates, no immune response to them has been
`reported. However, Epenetos' group reported im(cid:173)
`mune responses to the macrocycle chelating agent,
`DOTA, when used to bind 90Y to murine or CDR(cid:173)
`grafted monoclonal antibodies (77,82). This was not
`seen in prior studies with the less stable chelating
`agent diethylene triaminepentaacetic acid (DTPA).
`The immune response to DOT A is presumably an
`example of response to a hapten, with the monoclo(cid:173)
`nal antibody acting as the carrier molecule.
`A similar mechanism is probably responsible for
`the antibody response to the drug moiety in mono(cid:173)
`clonal antibody-drug conjugates. The best demon(cid:173)
`stration of this response was reported by Peterson
`et al. (83). In their study, patients received single or
`multiple doses of two vinca alkaloid conjugates of
`the MoAb KS 1/4. Immune responses to KS 1/4 were
`present in 32 of 44 (73%) patients, while 22 of 44
`
`J Immunother, Vol. 15, No. 1, 1994
`
`(50%) also had antibody specific for the vinca alka(cid:173)
`loid.
`The immune response to toxins in immunoconju(cid:173)
`gates of ricin A chain and Pseudomonas exotoxin
`(PE) was expected, since these molecules are quite
`immunogenic by themselves (84). Phase I trials of
`ricin A chain conjugates in patients with melanoma
`(85) or colon cancer (86) produced high titers of
`human antibody to ricin A chain and murine immu(cid:173)
`noglobulin in almost all patients, and this immune
`response was not inhibited by concomitant immu(cid:173)
`nosuppressive drug regimens (18,87). Immune re(cid:173)
`sponse was also noted in trials with recombinant
`ricin A chain conjugated to monoclonal antibodies
`in breast cancer patients (88,89). Human immune
`response to unmodified ricin A chain conjugates in
`patients with chronic lymphatic leukemia was less
`frequently a problem (90), reflective of the immu(cid:173)
`nosuppressed patient population. Similarly, trials
`with deglycosylated ricin A conjugates (91) or
`blocked ricin conjugates (92) in patients with non(cid:173)
`Hodgkin's lymphoma produced less of an immune
`response than in those with solid tumors, but still
`represented a limiting factor with regard to repeated
`cycles of therapy in some patients. Experience with
`monoclonal antibody conjugates of PE is more lim(cid:173)
`ited but has a similarly high frequency of antibody
`response in solid tumor (ovarian cancer) patients
`(93).
`
`CONSEQUENCES OF IMMUNE RESPONSE
`
`Although there was considerable concern regard(cid:173)
`ing adverse effects or symptoms that might occur
`with infusion of monoclonal antibodies in patients
`with prior immune response, experience has shown
`that the vast majority of such patients have no
`symptoms. In our own experience (Table 5), a total
`of 42 patients received a repeat infusion of mono(cid:173)
`clonal antibody at a time when they had demon(cid:173)
`strated human antibody to the infused reagent. We
`observed five episodes of allergic reactions, includ(cid:173)
`ing two that were early and readily reversible ana(cid:173)
`phylactic reactions (20). There are only a few other
`anaphylactic (or anaphylactoid) reactions reported
`in the literature (21,22), suggesting that this is quite
`uncommon and, to date, readily reversible. The in(cid:173)
`cidence of mild allergic phenomena (rash, hives,
`flushing, etc.) seems to be ~5% (20,94-96).
`A second type of adverse event, serum sickness
`secondary to human antibody/monoclonal antibody
`complexes, has not been seen in our clinical trials
`
`

`

`HAMA
`
`TABLE S. Adverse effects of monoclonal antibodies in patients with prior immune response
`
`Mm:ine
`17-IA
`
`Murine
`14G2a
`
`Murine
`xomazyme-Mel
`
`Chimeric
`B72.3
`
`Murine 96.5 +
`ZME 018
`
`No. infusions in
`RAMA-positive patients
`Anaphylaxis
`Mild allergic phenomena
`(rash, hives,
`flushing , etc.)
`Serum sickness
`Reference
`
`12
`2
`
`I
`0
`8
`
`II
`0
`
`0
`0
`45
`
`6
`0
`
`0
`0
`87
`
`8
`0
`
`0
`0
`67, 97
`
`5
`0
`
`0
`0
`46
`
`49
`
`Total
`
`42
`2 (5%)
`
`2 (7%)
`0
`
`and has only rarely been reported in the literature
`(95). In addition, Epenetos' group noted the devel(cid:173)
`opment of a "vasculitis-like" rash 1-2 weeks after
`exposure to murine or humanized antibodies (77)
`utilizing a DOTA linking agent for isotope studies.
`They demonstrated antibody to DOT A and sug(cid:173)
`gested that the rash was secondary to antibody re(cid:173)
`sponse. We considered a similar effect of antibody
`response to 14G2a in two patients who developed
`transient peripheral motor neuropathy (45).
`Finally, a highly reproducible and often dramatic
`alteration of pharmacokinetics can occur due to
`preexisting HAMA, which mediates rapid clearance
`of infused monoclonal reagent (87 ,95 ,97). This ef(cid:173)
`fect appears to be dependent both on amount of
`human antibody response and dose of monoclonal
`antibody infused. In our trial with murine 17-lA,
`infusion of 400 mg 17-lA was unaffected by the
`presence of anti-17-lA (20). Conversely, in our
`B72.3 trial using 1-4 mg infusions, anti-ch B72.3
`responses produced dramatic shorteni