Monoclonal antibodies in the detection and
`therapy of micrometastatic epithelial cancers
`
`Gert Riethmuller and Judith P. Johnson
`
`Institute of Immunology, University of Munich, Munich, Germany
`
`The initial promise of monoclonal antibodies as major therapeutic agents
`in human epithelial cancer has not been realized. Inaccessibility of cells
`in solid tumors due to factors such as
`the nature of the vascular
`endothelia and high pressure in the tumor are primarily responsible
`for the failure of antibody therapy. Although new strategies employing
`recombinant antibodies and immunoglobulins designed to actively engage
`the immune system may prove beneficial, micrometastatic tumor cells (at
`the stage of minimal residual disease) are likely to be the only suitable
`targets for antibody therapy. The diagnostic approaches to identify and
`characterize these cells and their use for prognosis and monitoring
`adjuvant immunotherapy is discussed.
`
`Current Opinion in Immunology 1992, 4:647--655
`
`Introduction
`
`The hope that monoclonal antibodies (mAbs), with
`their exquisite specificity and multiple effector func(cid:173)
`tions, would revolutionize the diagnosis and treatment
`of cancer, has failed to materialize. While mAbs have
`successfully replaced polyclonal antisera for the detec(cid:173)
`tion of tumor markers in the serum, their therapeutic
`use in patients with solid tumors can only be viewed as
`a disappointment, even with improvements in the ability
`of antibodies to target and destroy tumor cells in model
`systems. Thus, fundamental changes in the strategies of
`using mAbs to treat tumors are required. In this review
`we focus on several such strategies, on the one hand,
`involving new developments in antibody production and
`use, and on the other, focusing on the target of the ther(cid:173)
`apy itself. As more has been learned about the biology
`of the interaction of tumors with the vascular system and
`the extracellular matrix, it has become clear that much
`of the difficulty with passive antibody therapy is related
`to the accessibility of the tumor cells to these reagents.
`In light of this, micrometastasis, or the stage of minimal
`residual disease, may not be only an important field for
`diagnostic studies but also the best target for antibody
`mediated therapy.
`
`Monoclonal antibodies in diagnosis:
`determination of minimal residual disease
`
`The metastatic spread of small localized primary tumors
`in the absence of clinical signs of the disease in distant
`
`parts of the body can, at present, be diagnosed only in
`retrospect when the true extent of the disease manifests
`itself by an overt clinical relapse. For example, nearly 25%
`of breast cancer patients who do not show signs of dis(cid:173)
`ease in the regional lymph nodes, do in fact suffer from
`disseminated disease, borne out by the occurrence, often
`many years later, of distant metastases [ 1] . The concen(cid:173)
`tration of tumor antigens shed from a few tumor cells is
`too low for detection by the most sensitive immunoassays
`and even the most sophisticated physical diagnostic pro(cid:173)
`cedures available, including nuclear magnetic resonance
`and photon emission tomography, are not specific and
`sensitive enough to detect individual tumor cells dis(cid:173)
`seminated to distant organs. Thus, in the absence of the
`diagnosis of disseminated disease, the development of ef(cid:173)
`fective antibody-mediated (adjuvant) therapies proceeds
`at a scandalously slow pace. As with all adjuvant thera(cid:173)
`pies they must first show some efficacy in 5 year trials
`on terminal patients with a heavy tumor burden before
`they can be tested on earlier stages. It is evident that in
`this dilemma, a firm diagnosis of minimal residual disease
`would be of invaluable help to identify patients actually
`in need of adjuvant therapy. Furthermore, the monitor(cid:173)
`ing of therapeutic effects would enormously benefit the
`development of adjuvant therapies.
`Since the early years of the last decade there have
`been several attempts to use mAbs to distinguish in(cid:173)
`filtrating tumor cells from hematopoietic bone marrow
`cells [2] , an approach which was greatly expanded by
`Neville's group at the Ludwig Cancer Institute in Lon(cid:173)
`don [3]. The general feasibility of this method was
`
`Abbreviations
`ADCC-antibody-dependent cell-mediated cytotoxicity; APAAP-alkaline phosphatase-anti-alkaline phosphatase;
`CH-cytotoxic T lymphocyte; Fv-single chain monoclonal antibody; H-heavy; HLA-human leukocyte antigen; L-light;
`mAb-------monoclonal antibody; MHC- major histocompatibility complex; PCR-polymerase chain reaction;
`SEA-staphylococcal enterotoxin A; TCR-T-cell receptor; V-variable.
`
`© Current Biology Ltd ISSN 0952-7915
`
`647
`
`Celltrion v. Genentech
`IPR2017-01374
`Genentech Exhibit 2059
`
`

`

`648 Cancer
`
`soon demonstrated using different mAbs in a variety
`of epithelial tumors (Table 1). Because of the gener(cid:173)
`ally low concentration and considerable heterogeneity
`of expression of tumor-associated membrane antigens,
`we chose to use the abundant intracellular cytokeratin
`proteins [ 4,5] as markers for detecting epithelial cancer
`cells in the bone marrow [ 6•• J. Using an immunocyto(cid:173)
`chemical alkaline phosphatase-anti-alkaline phosphatase
`(AP MP) detection system it is possible to detect epithe(cid:173)
`lial cells at a fequency of 10- 5 to 1 o- 6 bone marrow
`cells [7,8••,9•,10] . Patients without epithelial malignan(cid:173)
`cies essentially lack cytokeratin-positive cells [7,11], while
`between 10 and 40% of these samples exhibit staining
`with antibodies directed against the membrane antigens
`epithelial membrane antigen and human milk fat glob(cid:173)
`ule mucin [6••,12,13]. Further evidence for the specificity
`of this method was obtained from double marker stud(cid:173)
`ies using a combination of immunoautoradiography and
`APMP staining. In the bone marrow from carcinoma pa(cid:173)
`tients, the expression of the leukocyte common antigen,
`CD45 and cytokeratin 18 were never found on the same
`cell [7).
`
`Although this method appears to have a high level of
`specificity, the low frequency of tumor cells in the bone
`marrow makes it likely that they are not uniformly dis(cid:173)
`tributed and suggests that a considerable sampling error
`occurs when only one site is sampled [ 14]. In fact, triple
`site aspiration (left and right iliac crest plus sternum) in(cid:173)
`creased the percentage of positive stage M0 breast cancer
`patients from 10.7% to 28.2% [15] (M denotes distant
`metastases and M0 refers to tumors where there is no
`evidence of distant metastases). For routine purposes, a
`two-sided aspiration of the pelvic crest is currently per(cid:173)
`formed on the operating table, immediately prior to the
`operation.
`
`The judicious use of double staining techniques allows
`one to define additional markers on the individual dis(cid:173)
`seminated cells. Thus cytokeratin-positive cells in the
`bone marrow of tumor patients were shown to ex(cid:173)
`press proliferation-associated molecules such as the the
`Ki 67 nuclear antigen and receptors for transferrin and
`epidermal growth factor [ 15]. Similar to cells in the
`metastatic lesions themselves, c.ytokeratin-positive bone
`marrow cells also often lacked HIA class I expression. In
`one study, 50% of the breast cancer patients with positive
`tumor cells in the bone marrow had only class I nega(cid:173)
`tive epithelial cells in the bone marrow [ 8••]. These cells
`not only express proliferation-assrx:iated markers but are
`able to grow in vitro. Epithelial cells could be expanded
`from bone marrow aspirates by the use of various combi(cid:173)
`nations of growth factors [16] and the rate of successful
`expansions was substantially increased when the culture
`vessels were precoated with extracellular matrix proteins.
`Using a serum free culture system, Hay et al [ 17) re(cid:173)
`cently reported the growth of small-cell lung cancer cells
`in 100% of positive bone marrow samples.
`
`What is the clinical significance of the presence of epi(cid:173)
`thelial cells in the bone marrow? In a number of studies
`on breast carcinoma patients [ 14,18••] as well as in colo(cid:173)
`rectal cancer [19] and neuroblastoma [20], the presence
`of micrometastases in the bone marrow has now been
`shown to be associated with a shorter disease-free in(cid:173)
`terval. Surprisingly, while bone marrow micrometastatic
`cells were found in 33% of the colorectal tumor pa(cid:173)
`tients (Dukes C stage, i.e. a trans-serosal tumor with
`regional lymph node involvement), most of the man(cid:173)
`ifest metastases involved the liver, indicating that the
`presence of tumor cells in the bone marrow is associ(cid:173)
`ated with the probability of metastases development in
`general, and not necessarily with the manifestation of
`
`Table 1. Detection of bone marrow micrometastasis in various types of epithelial cancer.
`
`Origin of tumor
`
`Marker antigens
`
`Correlation with established
`risk factors
`
`Prognostic
`value
`
`References
`
`Breast
`
`Colorectum
`
`Stomach
`Prostate
`
`Small-cell lung
`cancer
`Non-small-cell
`lung cancer
`Bladder
`Renal cell
`carcinoma
`
`Epithelial membrane antigen (EMA)
`Cytokeratin
`MBr1
`Cytokeratin
`17.1A
`Cytokeratin
`Cytokeratin
`PSA + Cytokeratin + EMA
`Cytokeratin
`SM1
`LCA1, LCA2, and LCA3
`Cytokeratin
`
`Cytokeratin
`Cytokeratin
`
`'(Unpublished data); n.t., not tested.
`
`+
`+
`-
`+
`
`+
`+
`+
`+
`+
`+
`+
`
`+
`+
`
`+
`n.t.
`n.t.
`n.t.
`
`+
`+
`n.t.
`+
`n.t.
`n.t.
`+
`
`n.t.
`n.t .
`
`[14]
`[71
`157]
`118••]
`
`[19)
`16 .. ]
`172]
`Oberneder'
`[58]
`159]
`Pantel'
`
`Oberneder'
`Oberneder'
`
`

`

`Monoclonal antibodies and epithelial cancer Riethmuller and Johnson
`
`649
`
`skeletal metastasis. Taking all of the published studies
`together, a good correlation exists between the presence
`of epithelial cells in bone marrow and the conventional
`risk factors based mainly on the extent of tumor dissem(cid:173)
`ination. Furthermore, the total tumor burden may be es(cid:173)
`timated from the number of epithelial cells in the bone
`marrow [18 00,20]. The evaluation of tumor cells in the
`bone marrow is, therefore, a potentially powerful diag(cid:173)
`nostic tool since it allows one to look at a part of the
`tumor burden which the surgeon has left behind. Not
`only could this provide a way to monitor the effective(cid:173)
`ness of various therapies, but with a further characteri(cid:173)
`zation of these cells ( oncogene, tumor suppressor gene
`expression, etc) it may eventually be possible to more
`accurately predict the metastatic potential of these cells.
`
`Monoclonal antibodies in therapy: new
`strategies for production and use
`
`Although passive antibody therapies have been effective
`against established tumors in many experimental sys(cid:173)
`tems, the therapy of solid tumors with mAbs in the
`clinic has had a history of failure. A brief look at the
`literature covering the years of 1984-1991 provides no
`hint of a consistent therapeutic efficacy of mAbs. In a re
`view of 12 studies comprising 196 patients with a variety
`of solid tumors, Gisler lists only two complete remissions
`(Table 2) and the number of partial remissions remains
`within the realm of the anecdotal (R Grisler, in Can(cid:173)
`cer and the Immune System, Proceedings of European
`School of Oncology, Venice 1990). The general conclu(cid:173)
`sion that has been drawn from this conspicuous failure
`is that antibodies as naked mouse immunoglobulins are
`
`inefficient, and a number of new strategies are being un(cid:173)
`dertaken to create new antibodies and to use mAbs to
`actively engage the host's own immune response against
`the autologous tumor.
`
`The development of new reagents that react with tumors
`through recombinant DNA technology
`While emphasis in recent years has been on using genetic
`engineering to 'humanize' and 'Fe customize' rodent an(cid:173)
`tibodies, the future almost certainly lies in the production
`of human antigen-binding molecules from combinatorial
`libraries in Eschericha coli [21]. In this method, indepen(cid:173)
`dent cDNA libraries are created from the mRNA'> encod(cid:173)
`ing variable (V) light (L) and heavy (H) chain antibody re
`gions, using the polymerase chain reaction (PCR). These
`two libraries are then randomly combined and cloned
`into a phage vector which directs the expression of one
`VL and one VH polypeptide chain. The use of vectors
`which allow the Fab fragments to be expressed as in(cid:173)
`tact antigen-binding sites on the phage surface, either
`in monomeric or multimeric form, have greatly facili(cid:173)
`tated screening and selection of desired antibodies [ 22•].
`In addition, the most recent developments indicate that
`high affinity antibodies of virtually any specificity can be
`obtained by applying rounds of random mutagenesis and
`selection to the Fabs produced from the cDNA of naive
`IgM + B lymphocytes, a situation which should obviate
`the need for immune donor cells [23••].
`Once the desired antigen-binding moieties have been se(cid:173)
`lected, the VL and VH regions can be produced as a sin(cid:173)
`gle polypeptide ( with the regions separated by a flexible
`linker) to generate single chain mAbs or Fv molecules.
`These can be further engineered to contain toxins or
`
`Table 2. Mouse monoclonal antibodies in clinical trials on solid tumors.
`
`Tumor
`
`Mouse antibody
`
`lsotype
`
`Number of
`patients
`
`Effect
`
`References
`
`Gastrointenstinal
`tumors
`
`Breast, colon,
`ovarian and
`lung cancer
`
`Pancreatic
`adenocarcinoma
`
`Neuroblastoma,
`melanoma
`
`Melanoma
`
`Melanoma
`
`C017-1A
`
`lgG2a
`
`L6
`
`lgG2a
`
`BW494
`C017-1A
`
`3F8
`anti-GD2
`
`9.2.27
`
`R24 (anti-GD3)
`
`lgG1
`lgG2a
`
`lgG3
`
`lgG2a
`
`lgG3
`
`20
`22
`8
`20
`19
`
`18
`19
`
`17
`
`20
`
`21
`12
`
`20±
`22±
`8-
`20±
`1++,18?
`
`18-
`4+, 15-
`
`1++,6+,10-
`
`20-
`
`4+, 17
`3+, 2±, 7-
`
`[60]
`[61]
`[62]
`[63]
`[64]
`
`[65]
`[66]
`
`[67]
`
`[68,69]
`
`[70]
`[71]
`
`+ +, Complete response; +, partial response; ±, minor response; -, no response.
`
`

`

`650 Cancer
`
`other effector domains [ 24••]. Fvs frequently refold into
`a conformation that retains the binding characteristics of
`the divalent mAb and may penetrate tissues much more
`efficiently than intact immunoglobulins. Such characteris(cid:173)
`tics can be used to predict whether they will be important
`anti-tumor reagents and their preliminary use in mouse
`models confirms this [25] .
`
`Use of monoclonal antibodies to engage the
`host's immune system
`
`The passive administration of tumor-reactive mAbs
`is increasingly being replaced by immunotherapeutic
`schemes designed to engage the host's own immune sys(cid:173)
`tem in the destruction of the autologous tumor. Among
`the most widely used strategies at present is the attempt
`to exploit the idiotypic network [ 26 .. ,27• ,28• J. This ap(cid:173)
`proach is based on the theory that certain anti-idiotypic
`antibodies (the Ab2Ps) will express the internal image
`of the original immunizing epitope [29]. Treatment with
`such antibodies is, therefore, equivalent to immunizing
`patients with the tumor antigen, or rather with a sin(cid:173)
`gle epitope of this antigen. In point of fact, patients
`treated with Ab2s frequently produce Ab3s which re(cid:173)
`act with the original tumor antigen [30,31,32•] and in
`some cases, even appear to develop a cellular reactivity
`against this antigen although the nature of this reactiv(cid:173)
`ity remains poorly characterized [30,33,34] .The clinical
`consequences of active immunization in those instances
`where the tumor associated antigen, e.g. the 17.1 A colo(cid:173)
`rectal carcinoma epithelial antigen, is also widely ex(cid:173)
`pressed on normal epithelia are unclear.
`
`Despite the common production of Ab3s that react with
`tumor cells in the anti-idiotype treated patients, these
`clinical trials have generally not been any more successful
`than those using Abls, antibodies to the original epitope,
`[27• ,28• ] . However, Mittelman et al. [32•], recently re(cid:173)
`ported a significant increase in survival time in melanoma
`patients producing Ab3s against the chondroitin sul(cid:173)
`phate proteoglycan, a surface antigen of melanoma cells.
`This result could not be accounted for by higher gen(cid:173)
`eral immune reactivity (i.e. performance status) of the
`patients. While such discrepant results may reflect, in
`part, the fact that immunization with these reagents in(cid:173)
`duces an immune response against a single epitope that
`may have varying functional and structural characteristics
`(e.g. density, sensitivity to modulation, functional signif(cid:173)
`icance), studies in a mouse anti-tumor model (reviewed
`in [26 .. J) indicate that selection of the Ab2, used in such
`trials is of critical importance. In this system, four differ(cid:173)
`ent Ab2s, which were very similar in their interaction with
`the Abl and in their ability to induce anti-tumor antigen
`humoral and cellular responses, were examined for their
`ability to vaccinate against tumor growth. Only one Ab2
`was able to induce protective immunity; more disturb(cid:173)
`ing was the observation that one actually appeared to
`potentiate tumor growth (35]. The finding of a linear
`seyuence homology between the second hypervariable
`region of the protective Ab2 L chain and the tumor anti·
`
`gen [ 26 .. J, not only suggests a molecular basis for the
`similarity between the internal image and the tumor anti(cid:173)
`gen, but means that selection of appropriate Ab2s may
`eventually he possible through sequence comparison of
`the antibody variable regions with the sequence of the
`antigen. However, this is not possible for carbohydrate
`antigens that are among the best candidates for tumor
`surrogate antigen vaccines given their effectiveness as tu(cid:173)
`mor target structures [ 36• J . This would be cumbersome
`enough for protein antigens, but it would be even more
`difficult for carbohydrate antigens because of the diffi(cid:173)
`culty of producing them in large quantities; despite this,
`the effectiveness of carbohydrate antigens as tumor target
`structures means that they are among the best candidates
`for surrogate antigen vaccines [36•].
`
`A conceptually simpler approach to engage the host
`immune system is the use of bi-specific antibodies to
`target host effector cells to the tumor. These reagents
`combine, in a single molecule, specificity for a tumor
`associated antigen with specificity for an effector cell
`molecule, most commonly the T-cell receptor (TCR)·
`associated CD3 complex (37,38]. Addition of such con(cid:173)
`jugates to interleukin-2- or anti-CD3-activated leukocytes
`prior to their injection reduces the growth of established
`tumors in nude mice more effectively than the activated
`cells alone [39] , a result which was also obtained in a
`small clinical trial [ 40 J. Most cells that are targeted to the
`tumor by this approach are not directed against tumor
`antigens. But bi-specific heteroconjugates can also in(cid:173)
`crease the killing of autologous tumors by clones of
`tumor-specific tumor-infiltrating lymphocytes and cyto(cid:173)
`toxic T lymphocytes (CTL<i) provided the tumor cells
`express the target antigens [ 41 •]. However, if the tu·
`mars are negative or heterogeneous for target-antigen
`expression, the presence of heteroconjugates can inhibit
`tumor-specific cytotoxicity, presumably by blocking the
`TCR. Given the universality of tumor cell heterogeneity,
`this could he a serious drawback to the successful use
`of such reagents.
`
`..
`An approach which might overcome this problem is one
`which takes advantage of the characteristics of bacterial
`toxin superantigens [ 42]. These molecules bind with
`high affinity to MHC class II antigens and once bound,
`demonstrate a specificity for particular families of TCR vp
`chains. Reactivity with the TCR leads to T-cell activation,
`cytokine release and killing of the cell expressing MHC
`class II antigen. By coupling a colon carcinoma reactive
`mAb to staphylococcal enterotoxin A (SEA), Dohlstein et
`al, [ 43•• J have been ahle to replace class II dependent
`T°-cell activation with a tumor antigen specific T-cell acti(cid:173)
`vaJ:ion and have achieved killing of carcinoma cell lines by
`CTLs bearing the appropriate V~ chains. Since SEA will
`not engage the T cell until it is 'presented' by mAb bind(cid:173)
`ing, activation of the T cells and cytokine release should
`occur only locally, in the immediate environment of the
`tumor cells. Thus, like the use of cytokine transfected tu(cid:173)
`mor cell vaccines [ 44] , which also lead to spatially and
`temporally concordant expression of tumor antigens and
`cytokines, SEA- mAh conjugates may also trigger develop·
`ment of the host's own anti-tumor response.
`
`

`

`Monoclonal antibodies and epithelial cancer Riethmuller and Johnson
`
`651
`
`Micrometastatic tumor cells as targets for
`immunotherapy: the way to secondary
`prevention
`
`Even in the face of major improvements in the ability of
`antibodies to bind to and kill tumor cells, the problem
`of the accessibility of the tumor cells to these reagents
`remains unresolved. In fact, recent data on the microcir(cid:173)
`culation and extracellular matrix of tumors indicate that
`this may in fact be the most important impediment to
`successful antibody treatment of solid tumors [ 45,46].
`
`The organization of the tumor's vascular system is largely
`determined by the histogenic type of tumor (i.e. tissue
`of origin), its differentiation state, proliferation rate and
`location. The non-fenestrated capillaries often found in
`epithelial tumors prevent antibodies from entering the tu(cid:173)
`mor tissue. Another major factor obviating the transcapil(cid:173)
`lary transport of macromolecules is the tumor interstitial
`pressure which is generally much higher than the sur(cid:173)
`rounding normal tissue and which increases with tumor
`size [ 47]. The pressure gradient from the center to the
`periphery of a tumor leads to an outward convection of
`the interstitial fluid so that extravasated mAb will be trans(cid:173)
`ported to the periphery of the tumor. In the interstitium
`itself, the dense network of glycosaminoglycans presents
`a formidable obstacle for free diffusion of antibody. In
`addition, the antigens shed from the tumor, when immo(cid:173)
`bilized in this interstitium, may neutralize antibodies and
`lead to precipitation of antigen-antibody complexes. A fi(cid:173)
`nal barrier that must be overcome in differentiated metas(cid:173)
`tases of adenocarcinomas is the shielding of target anti(cid:173)
`gens from diffusing antibodies by the intercellular tight
`junctions.
`
`While the clinical trials have only resulted in a very few
`well documented antibody-induced regressions ( which
`may more easily be explained by derangements in the
`tumor vasculature or extracellular matrix rather than by
`the inherent characteristics of the antibody), they have
`clearly shown that antibody therapy is remarkably free of
`toxicity and serious side effects. Thus, the stage would
`now seem to be set for applying passive antibody ther(cid:173)
`apy to ils most logical target - minimal residual disease
`(or micrometastasis), a state which applies to 1/3 to 1/2
`of patients with epithelial cancer (the most common type
`of cancer), following (curative) radical surgery of the pri(cid:173)
`mary lesion.
`
`One of the first hints that micrometastatic cells might be
`particularly suitable targets for passively administered an(cid:173)
`tibody was provided by Schlimok et al. [7] who demon(cid:173)
`strated that tumor cells in bone marrow of patients
`with colorectal cancer can be labelled with mouse im(cid:173)
`munoglobulin in vivo. In a later report [8 .. ], the ther(cid:173)
`apeutic effects of antibody infusions over a period of
`several months up to 2 years were monitored by im(cid:173)
`munocytochemical analysis of bone marrow. In 12 of
`the 23 Dukes C patients with clinically manifest metas(cid:173)
`tases, micrometastatic cells were repeatedly identified and
`clinical relapse occurred in nine of these patients.
`
`In order to monitor directly the therapeutic elimina(cid:173)
`tion of micrometastatic cells, patients with high num-
`
`bers of micrometastatic cells in the bone marrow were
`treated with an IgG3 anti-Lewis Y antibody displaying
`high complement-dependent cytotoxicity [ 48•]. (This an(cid:173)
`tibody was selected because IgG3 mouse antibodies are
`particularly effective in the activation of human comple(cid:173)
`ment and ADCC). While no regressions were observed
`in these patients with extensive metastatic disease, a dra(cid:173)
`matic decrease in the number of tumor cells in bone
`marrow was observed, an effect not seen in placebo
`treated patients ( G Schlimok, G Riethmi.iller, K Pante! et
`al., abstract 709, ProcAm Soc Clin One 1991, 10:212). To
`analyze the efficacy of antibodies against micrometastatic
`cells in colorectal cancer a prospective randomized trial
`was initiated into which only patients with regional dis(cid:173)
`ease (Dukes Cl and C2; metastases in 1-3 regional lymph
`nodes and 4 or more, respectively) were recruited. The
`trial comprising a total of 176 patients in a treatment and
`control arm will be closed at the end of 1992 with a me(cid:173)
`dian observation time of 5 years. A preliminary analysis at
`a median observation time 4.5 years indicates differences
`between treated and control patients in terms of overall
`survival and recurrence rate of distant metastasis ( G Ri
`ethmi.iller, G Schlimok et al., unpublished data). Although
`another prospective randomized trial on pancreatic can(cid:173)
`cer patients treated with an anti-carbohydrate antibody
`failed to show therapeutic effects [ 49••], the state of
`minimal residual disease in these patients can be dis(cid:173)
`puted as the partial pancreato-duodenectomy (removal
`of the pancreas and duodenum) employed for tumor re(cid:173)
`section may not have been complete in many of them.
`
`Conclusions and perspectives
`
`While it is clear that a major consideration for the suc(cid:173)
`cessful application of antibody therapy in solid tumors
`is the accessibility of the tumor cells, other aspects also
`need to be taken into account. Heterogeneity of tumor
`cells is the major obstacle of any antigen-targeting ther(cid:173)
`apy. To cope with antigen heterogeneity, which increases
`with tumor progression, a carefully selected combination
`of antibodies directed to independently controlled mem(cid:173)
`brane antigens should be applied as early as possible,
`i.e. immediately after the removal of the primary tumor.
`Published data demonstrate that selected pairs of anti(cid:173)
`bodies directed against different epitopes of a particular
`antigen exhibit up to a 100-fold synergistic increase in
`cell killing in vitro [ 50] or in vivo [ 51] or result in
`augmented binding [52-54]. A<; passive antibody ther(cid:173)
`apy imitates a natural effector mechanism, a well defined
`oligoclonal combination of antibodies, in which each has
`been shown to eliminate micrometastatic cells, should be
`as acceptable for clinical use as current registered poly(cid:173)
`clonal anti-thymocyte globulin preparations.
`
`The optimal combination of such antibodies would con(cid:173)
`sist of human or 'humanized' immunoglobulins. Since
`normal epithelia are less accessible to cytotoxic anti(cid:173)
`bodies (they are located behind a dense basal mem(cid:173)
`brane which is quasi-impermeable to non-secretory im(cid:173)
`munoglobulin), an absolute specificity for therapeutic an(cid:173)
`tibodies appears less critical. Indeed, toxicity to normal
`epithelia has been negligible in the trials reported thus
`far. Although the inactivation of complement by homo!-
`
`

`

`652 Cancer
`
`ogous restriction factors may contribute to the resistance
`of the tumor cells to lysis, it may be possible to eventually
`inhibit these factors (e.g., CS-binding protein or CD59
`membrane glycoprotein) [55). As demonstrated by sev(cid:173)
`eral studies [7] micrometastatic cells can be reproducibly
`identified in individual patients over time, thus offering
`a new way to monitor micrometastasis. Such monitoring
`will be mandatory for the expeditious development of
`an effective passive antibody therapy. Furthermore, new
`techniques such as PCR may replace the more laborious
`immunocytological methods [ 56] .
`
`In conclusion, it would be a major mistake to let anti(cid:173)
`bodies, exquisitely specific and versatile tools, and the
`powerful natural effector mechanisms they can harness
`be misspent on the unsuitable target of advanced solid
`tumors
`
`Acknowledgements
`
`The authors are supported by Dr. Mildred Scheel Stiftung, Deutsche
`Krebshilfe, Bonn, Germany. The authors would like to thank Dr.
`Klaus Pante! for helpful discussion.
`
`References and recommended reading
`
`Papers of particular interest, published within the annual period of re(cid:173)
`view, have been highlighted as:
`of special interest
`of outstanding interest
`
`1.
`
`2.
`
`3.
`
`4.
`
`5.
`
`BAUM M: The Curability of Breast Cancer. In Breast Can(cid:173)
`cer Management. Edited by Stall BA. London: Heinemann;
`1977:3-13.
`
`KEMSHEAD JT, WAI.'iCH F, PRITCHARD J, GREAVES M: Mono(cid:173)
`clonal Antibody to Ganglioside GQ Discriminates between
`Haemopoietic Cells and Infiltrating Neuroblastoma Cells in
`Bone Marrow. Int J Cancer 1981, 27:447--452.
`
`DEARNAI.EY DP, SLOANE JP, ORMEROD MG, STEEI.E K, COOMBES
`RO, CUNK HMO, POWLES TJ, FORD HT, GAZET JC, NEV1I.LE AM:
`Increased Detection of Mammary Carcinoma Cells in Mar(cid:173)
`row Smears Using Antisera to Epithelial Membrane Antigen.
`Br J Cancer 1983, 44:85-90.
`
`Mou R, FRANKE WW, SCHIil.ER DL, GEIGER B, KREPLER R: The
`Catalogue of Human Cytokeratins: Pattern of Expression
`in Normal Epithelia, Tumors and Cultured Cells. Cell 1982,
`31:11-24.
`
`DEBUS E, WEBER K, OSBORN M: Monoclonal Cytokeratin An(cid:173)
`tibodies that Distinguish Simple from Stratified Squamous
`Epithelia: Characterization on Human Tissues. EMBO J 1982,
`1:1641-1647
`
`6.
`••
`
`SCHLIMOK G, FUNKE I, PANTEL K, STROBEL F, LINDEMANN F,
`WITTE J, RIETHMOI.I.ER G: Micrometastatic Tumor Cells in
`Bone Marrow of Patients with Gastric Cancer: Methodologi(cid:173)
`cal Aspects of Detection and Prognostic Significance. Eur
`] Cancer 1991, 27:1461-1465.
`By comparing the specificity and sensitivity of antibodies to various
`membrane and cytoplasmic (cytokeratin) antigens the superiority of
`anti-cytokeratin antibodies in distinguishing infiltrating epithelial can(cid:173)
`cer cells from hematopoietic bone marrow cells is demonstrated. In
`gastric cancer, the incidence of micrometastatic cells correlates with
`conventional risk parameters such as the tumor size, the stage of dif(cid:173)
`ferentiation and the presence or absence of metastases, and thus also
`with disea5e-free survival time.
`
`7.
`
`SCHLIMOK G, FUNKE I, HOU.MANN B, G0TTLINGER H, SCHMIDT
`G, HAUSER H, SWIERKOT S, WARNECKE HH, SCIINEIDER B,
`KOPROWSKI H, RIETHM0Il.ER G: Micrometastatic Cancer Cells
`
`· in Bone Marrow: in Vitro Detection with Anti-cytokeratin
`and in Vivo Labelling with Anti-17- lA Monoclonal Antibod(cid:173)
`ies. Proc Natl Acad Sci USA 1987, 84:8672-8676.
`
`8.
`
`PANTEL K, SCHUMOK G, KUTTER D, SCHAUER G, GENZ T,
`WIEBECKE B, BACKMANN R, FUNKE I, RIETHMOU.ER G: Frequent
`Down Regulation of Major Histocompatibility Class I Anti(cid:173)
`gen Expression on Individual Micrometastatic Carcinoma
`Cells. Cancer Res 1991, 51:4712-4715.
`This study of HIA class I expression reveals an unexpectedly frequent
`loss of class I antigen by individual micrometastatic cells in breast, colo(cid:173)
`rectal and gastric cancer. For future therapies aimed at recruitment of
`CTI.s this is an important observation to consider.
`
`9.
`
`MOLINO A, COLOMBATTI M, BONETTI F, 2.ARDINI M, PASINI F,
`PERINI A, PELOSI G, TTIDENTE G, VENERI D, CETTO GL: A
`Comparative Analysis of Three Different Techniques for the
`Detection of Breast Cancer Cells in Bone Marrow. Cancer
`1991, 67:1033-1036
`A methodical paper demonstrating the higher sensitivity of the APAAP(cid:173)
`immunocytochemical method over imrnunofluorescence and conven(cid:173)
`tional immunocytological methods by detecting epithelial cells in nu(cid:173)
`cleated bone marrow cells.
`OSBORNE MP, WONG GY, ASINA S, OW LJ, COTE RJ, ROSEN
`P: Sensitivity of Immunocytochemical Detection of Breast
`Cancer Cells in Human Bone Marrow. Cancer Res 1991,
`51 :270&--2709.
`
`10.
`
`11.
`
`12.
`
`13.
`
`14.
`
`15.
`
`16.
`
`SCHUMOK G, SWIERKOT S, HAUSER H, JUNKER A, WARNECKE HH,
`RIETHMDU.ER G: Detection of Bone-marrow Micrometastases
`Using Monoclonal Antibodies. Blut 1985, 51:203.
`
`DEh'iOL G,