(CANCER RESEARCH 49. 4062-4067. July 15. 1989]

Phase I Evaluation of an Anti-Breast Carcinoma Monoclonal Antibody 260F9-Recombinant Ricin A Chain Immunoconjugate1

Louis M. Weiner,2 Joyce O'Dwyer, Joanne Kitson, Robert L. Comis, Arthur E. Frankel, Robert J. Bauer, Michael S.

Konrad, and Eric S. GrovesDepartment of Medical Oncology, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111 ¡L.M. H'., J. O., J. A'., R. L, C.J; Duke University Medical Center,Durham, \orth Carolina 27710¡A. E. F.]; and Cetus Corporation. Emeryville, California 94608 /R. J. B.. M. S. A'., E. S. G.J

ABSTRACT

Four women with metastatic breast cancer were treated with monoclonal antibody 260K9-recombinant ricin A chain, a ricin A chain immunoconjugate (1C) which targets a M, 55,000 antigen expressed by humanmammary carcinomas. Patients were treated by daily, 1-h i.v. injectionsfor 6 to 8 consecutive days. Two patients were treated with 10 MR/kgdaily and the two others were treated with 50 «ig/kgdaily. The trial wassuspended after four patients had been treated because patients treatedwith a continuous infusion schedule with this 1C had developed significantneurological toxicity at doses similar to those used in this study. Thehalf-life of the 1C showed a t.,a of approximately 1.8 h, a t.,/3 ofapproximately 8.3 h, and a peak concentration of about 200 ng/ml, at thelower dose level, and showed a f.,or of approximately 2.5 h, t, p of about10.4 h, and a peak concentration of 500 and 850 ng/ml for the twopatients at the higher dose level. All four patients developed evidence ofa human anticonjugate antibody response within 16 days of the onset oftherapy. The treatment was associated with significant toxicity, manifested by a syndrome consisting of weight gain, edema, hypoalbuminemia,and dyspnea. Similar symptoms were observed in patients treated bycontinuous infusions of the 1C. This clinical syndrome, seen at doses of1C which were insufficient to saturate antigen-expressing malignanttumor deposits in this trial, has been seen in other 1C therapy trials andin clinical trials using the cytokine interleukin 2. To investigate a possiblemechanism responsible for this toxicity, human monocytes were incubated with varying concentrations of 1C. There was detectable binding of1C to human monocytes at 1C concentrations which were achieved clinically in this trial. Furthermore, the binding appeared to be abrogated bypreincubation of the monocytes with pooled human immunoglobulin, thussuggesting that binding occurs via Fey receptor-mediated mechanisms.Binding was not affected if different linkers between recombinant ricin Achain and the antibody were used or if a different antibody moiety wasused in the 1C preparation. Chemically linked dimers of MOPC-21 boundto human monocytes at least as well as the ICs; this binding was notabrogated by preincubation with pooled human immunoglobulin. Sincethe 1C preparations used in this clinical trial contained small percentagesof dimers and/or multimers, the clinical toxicity syndromes which weobserved may be related to this series of observations. A more completeunderstanding of the relationship of this previously unreported mechanism of 1C binding to human cells expressing Fc*yreceptors with the

clinical manifestations of the capillary leak syndrome will await production and testing of Hal)').. ICs or highly purified whole antibody 1C

preparations which contain only monomers. Further investigations intothe mechanisms by which 1C binding to FCT receptor-bearing cells maylead to disruption of endothelial cell integrity may provide importantclues to the pathogenesis of the capillary leak syndrome seen with avariety of biological therapies.

INTRODUCTION

The availability of monoclonal antibody technology has madeit possible to develop antibody-toxin conjugates that can deliver

Received 10/31/88; revised 4/6/89; accepted 4/14/89.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1This study was supported in part by National Cancer Institute Grant P30

CA 06927.2Recipient of a National Cancer Institute Clinical Investigator Award. To

whom address requests for reprints should be addressed, at the Department ofMedical Oncology, Fox Chase Cancer Center. Central Avenue and ShelmireStreet. Philadelphia, PA 19111.

the toxin moiety to malignant cells in a highly specific manner.A variety of toxins have been tested; among the most potent ofthese is ricin. This toxin catalytically inhibits protein synthesisby binding to and cleaving the A'-glycosidic bond of position/44324in 28S RNA of 60S subunits of mammalian ribosomes(1-3). The native toxin is composed of two chains, A and B,linked by a disulfide bond. The B chain is responsible forbinding to cell surfaces, while the A chain contains the enzymatic activity. ICs' containing purified A chains have been

tested extensively in preclinical models and clinical trials(4-14).

The IgGl murine monoclonal antibody 260F9 is specific fora A/r 55,000 antigen expressed by approximately 50% of breastcancer cells (15). This antibody has been conjugated to recombinant ricin A chain to form the immunoconjugate 260F9 MAb-rRA (16). Following extensive preclinical testing which confirmed the selectivity, biological activity, and acceptable toxicityprofiles of the 1C, Phase I evaluations of this compound werebegun using two different administration schedules. This communication describes our observations of a daily injectionschedule and focuses on the capillary leak syndrome seen withthis 1C. We show that this and other ICs bind to human cellsthat do not express the antigen identified by 260F9, via uptakeby cellular Fc> receptors.

MATERIALS AND METHODS

Clinical Study Design. Four patients with metastatic breast cancerwere treated with this immunoconjugate between May and August1987. Their clinical characteristics are summarized in Table 1. Patients1 and 2 were assigned to receive 10 ug/kg daily of immunoconjugate,administered by a 1-h infusion for 8 consecutive days. Patient 3 wasscheduled to receive 50 Mg/kg daily by the same route for 8 consecutivedays, while patient 4 was assigned to receive 50 /¿g/kgdaily for 6 daysbased upon pharmacokinetic studies in the first three patients. Thepatients were hospitalized to receive all treatments in the clinical studyunit at Fox Chase Cancer Center.

The patients were evaluated for toxicity on a daily basis whilereceiving treatment and at approximately weekly intervals thereafter.Unless specified otherwise, all toxicity is graded according to therecently developed National Cancer Institute common toxicity criteria.

1C Preparation. 260F9 MAb-rRA 1C was prepared by the CetusCorporation. Briefly, purified 260F9 MAb was linked to rRA througha linker containing a disulfide bond. Ricin A chain was produced inEscherichia coli Kl 2 from the gene sequence of ricin A chain clonedfrom the complementary DNA of the castor bean. The final 1C waspurified free of unconjugated rRA and conjugates containing multimersof MAb and rRA; however, a small fraction of contaminating MAb-MAb-rRA persisted in the final product. Clinically used 1C preparationscontained a small concentration of dimers and multimers, less than10% as assessed by nonreducing gel electrophoresis.

ICs 260F9 MAb-Ll-rRA and 260F9-L2-rRA were prepared at Cetususing two thioether linkers, LI and L2, to aid in determining the role

1The abbreviations used are: 1C, immunoconjugate: 260F9 MAb-rRA. murinemonoclonal antibody 260F9-recombinant ricin A chain immunoconjugate: rRA.rccombinant ricin A chain; SPDP, yV-succinimidyl-3-(2-pyridylthio)propionate.

4062

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

PHASE l TRIAL OF IMMUNOCONJUGATE 260F9-MAh-rRA

Table l Patient characteristics

Patient1234Age(yr)35713354PS(%)"909010080Priortherapy*Melph.

5-FUra, CTX.MTX.DXR.sternalXRT5-FUra.

MTX, VLB,CTX.DXR.MMCMTX,

5-FUra, Pred.ChlorMTX.MMC, TAM,CTX,VLB.

5-FUra. ehestwallXRTMctastatic

sitesChest

andabdominalwallBoneSCNBoneLungsChest

wall, leftSCNpleura,lungs

°PS, Karnofsky performance status; Melph, melphalan: 5-FUra, 5-fluoroura-

cil: CTX. cyclophosphamide: MTX. methotrexate: DXR. doxorubicin: XRT,radiation therapy; VLB. vinblastinc; MMC, mitomycin C; Pred. prednisone:Chlor, ehlorambucil: Tam. tamoxifen: SCN, supraclavicular lymph node.

* Excluding surgery.

of the disii Hide bond in the binding of the clinically tested 1C to humanmonocytes.

MOPC-21 MAb-MAb ICs were produced at Cetus using SPDPlinker technology. The final mixture was purified of free MAb andhigher order multi-MAb conjugates by a column sizing step.

Pharmacokinetics and Anti-Conjugate Antibody Determinations. Samples were obtained at baseline; at 1.0, 1.5, 2.0, 3.0, 4.0, 6.0, 12.0, 18.0,and 24.0 h; and then daily prior to treatment. On the last day oftherapy, blood was sampled prior to and just following dosing, and 24h after this last injection. Serum levels of 260F9 MAb-rRA weredetermined using a double sandwich enzyme-linked immunosorbentassay. The sensitivity of this assay is less than 1.0 ng/ml. One component of the sandwich was directed at the rRA and the other at themurine MAb component; thus the assay determined the serum concentration of conjugates composed of rRA and murine MAb. Analysis wasperformed using a nonlinear, least squares fit of the data using a twocompartment model, which fit the data better than a one compartmentmodel. Data were weighted inversely and the fit program used wasSimusolv (Dow Chemical Company). All serum sample times for eachpatient were incorporated into the data set used for the fitting of thepresented parameters.

Anti-drug and anti-drug component antibody titers were determinedusing an enzyme-linked immunosorbent assay. Briefly, the drug or oneof its components (260F9 MAb or rRA) was coated on plates. Appropriate dilutions of patient serum were incubated in the plate, unboundmaterial was washed free, and the plate was developed with peroxidaseanti-human IgG or peroxidase anti-human IgM. The isotype-matchedmurine antibody 520C9 was used as a plate coat to distinguish betweenresponses directed at the binding specificity of 260F9 (antiidiotype)and those directed at the constant regions of murine IgG I antibodies.

Cell Phenotype Studies. A variety of antibodies and ICs were usedfor in vitro analysis of binding to monocytes. Pooled human IgG,MOPC-21, and UPC-10 myeloma proteins were obtained from SigmaChemical Co., St. Louis, MO. Monoclonal antibody 260F9 was generously provided by Dr. David Ring (Cetus Corporation). 260F9 MAb-rRA, MOPC-21 MAb-rRA, 260F9 MAb-Ll-rRA, and 260F9 MAb-L2-rRA were all prepared by Cetus Corporation. F(ab')2 fragments ofgoat anti-mouse immunoglobulin conjugated with fiuorescein isothio-cyanate were purchased from Cappel, Malvern, PA.

Peripheral blood was obtained from normal donors and peripheralblood mononuclear cells were purified on Ficoll-Isopaque gradients, asdescribed (17). In some experiments purified monocytes were obtainedby reversible binding to gelatin and autologous plasma-coated plasticflasks, as described (18). To evaluate binding of various proteins tomonocytes, purified monocytes or peripheral blood mononuclear cellsuspensions were incubated with various concentrations of the bindingproteins for 30 min at 4°C.Cells were washed and F(ab')? fragments

of goat anti-mouse IgG labeled with fiuorescein isothiocyanate wereadded. Following another 30-min incubation at 4°Cthe cells were

washed and fluorescence was evaluated by flow cytometry using aFACScan flow cytometer (Becton Dickinson, Mountain View, CA).Gates were set on monocytes by forward and side scatter criteria; greaterthan 95% of the cells in the monocyte gates expressed Leu-M3 (datanot shown). Results are expressed as percentage of cells with more

fluorescent intensity than 95-98% of the negative control cell population. Statistics were performed using paired t test analysis, as described(19).

RESULTS

Clinical Course. The courses of treatment are summarized inTable 2. Patients 1 and 2 received their planned therapy at fulldoses for all 8 days. Patient 3 required a 1-day treatment breakand 50% dosage reductions thereafter, so that she receivedtherapy for 7 full days, with 62.5% of her planned dose actuallyadministered. Patient 4 received full doses of a 6-day plannedtreatment course. The shorter treatment course was undertakenbecause of the demonstration of anticonjugate antibody responses observed in the previous patients (see below).

One patient, #3, had a clinical response, with the disappearance of her sole site of disease, a 1-cm' lung nodule identifiedby chest computer-assisted tomographic scan. However, shehad other lung abnormalities, also seen on computer-assistedtomographic scan, of uncertain disease significance, which didnot change with therapy. She had recurrence of her tumor at achest wall/skin site 3.5 months after the initiation of hertherapy.

A number of toxicities were observed with this therapy, evenat doses which were too low to yield detectable binding of theimmunoconjugate to the target antigen of 260F9 (Table 3).Malaise and other constitutional symptoms such as fever ormyalgias were commonly seen but were not dose limiting. Whileall patients experienced performance status deterioration, onlypatient 3 declined significantly. Peripheral blood monocytecounts did not change with therapy. Anemia was universal andcould not be totally accounted for by multiple phlebotomies.Weight gain averaging 4.2 kg was observed as was edema,hypoalbuminemia, hypoproteinemia, and eosinophilia. Patient

Table 2 Treatment summary

Duration of therapy (days)

Assigned daily Planned total Total dosePatient dose (*ig/kg) dose (nig) Planned Actual received (nig)

1234101050506.85.520.717.78886887"66.85.513.017.7

" This patient received full doses for 3 days. Treatment was held for 1 day andthen resumed at 50rr doses for the remaining 4 days.

Table 3 Adverse events during therapy

Patient

ToxicityMaximum

Karnofskyperformancestatusdrop(%)Highestfever(grade)Most

wt gain(kg)Hypotension(grade)Tachycardia(grade)Edema(grade)Largest

hemoglobin drop (g/100ml)Eosinophilia(%)Serum

albumin (cl ofbaseline)Scrumtotal protein (''< ofbaseline)Rash

(grade)Dyspnea(grade)Neurological(grade)Nausea,

vomiting(grade)Flu-likesymptoms (grade)110050003.14697100000210050012.410778302a00034015.23021.686069101»2I41021.70011.73798600Ie11

°Delayed onset of dyspnea in absence of pulmonary tumor, edema, or further

weight gain.* Delayed onset of fingertip parcsthesias (see text).' Left arm weakness and diminished sensation. Tumor in brachial plexus found

on magnetic resonance imaging scan.

4063

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

PHASE I TRIAL OF IMMUNOCONJUGATE 260F9-MAb-rRA

3 experienced grade 3 orthostatic hypotension which was associated with weight gain, edema, hypoalbuminemia, and hy-poproteinemia. Her clinical picture closely matched that of thecapillary leak syndrome which has been observed in patientsreceiving high dose interleukin 2 therapy (20, 21). In addition,this patient experienced a grade 2 nonpruritic rash on her trunkand extremities following the last dose of immunoconjugate.

Three of the four patients experienced delayed symptomswhich were possibly attributable to the immunoconjugate therapy. Patient 2 experienced grade 2 dyspnea 6 days after theconclusion of therapy in the absence of any demonstratedpulmonary tumor, edema, or weight gain. She was thought tohave congestive heart failure by her local treating physicians,although this was never documented by noninvasive or invasivestudies. Patient 3 developed the delayed onset of fingertipparesthesias in the arm contralateral to her prior mastectomyincision. These symptoms resolved completely within 6 months.She has since developed evidence of local skin recurrence buthas had no evidence of axillary or brachial plexus involvement.Finally, patient 4 experienced progressive left arm weaknessand diminished sensation in that arm following therapy. However, extensive tumor was found in the left brachial plexus uponmagnetic resonance imaging scanning.

Pharmacokinetics and Anti-Drug Antibodies. Pharmacoki-netics are displayed in Table 4A. The two patients treated with10 Mg/kg/day developed peak serum concentrations of approximately 200 ng/ml. tv,a was approximately 1.8 h and tv,ßwasapproximately 8.3 h for these two patients. For the two patientstreated with doses of 50 ^g/kg/day, tv,a was approximately 2.5h, Õ././3was was approximately 10.4 h and the peak serum levelswere approximately 500 and 850 ng/ml, respectively. Troughlevels of 1C were noted to decline in patient 1 during the last 2days of therapy (from 25 to 10 ng/ml). The role of anti-drugantibodies in mediating this decrease is unclear. Peak serumlevels did not change in this patient from days 1 to 8 (264 to248 ng/ml).

As shown in Table 4B, all four patients developed anticon-jugate antibody titers, which in all cases exceeded the titers toeither the antibody or ricin A chain components of the conjugate. Significant IgG titers were noted against 260F9 antibody;

Table 4 Pharmacokinetics: one and two compartment Jit models

A. Pharmacokinetics

Patient Patient 2 Patient 3 Patient 4

Daily dose (pg/kg) (1-h in 10 10 50 SOfusion)Peak

serumconcentration(ng/ml)Half-life

(h)Volumeof distributionofcentral

compartment(ml/kg)'*«

(h)<»0

(h)AmplitudetofH0Area

under curve(ng/ml-h)Clearance

(ml/kg-h)K„(ml/kg)2604.9

±1.339.1±0.21.1

±0.047.9±0.10.49

±0.0116206.262.71704.4

±1.769.1±0.52.6

±0.38.7±1.30.32±0.0996010.495.58505.0

±1.163.7±0.13.0

±0.110.9±0.50.15

±0.01480810.491.95007.1

±1.990.7±0.22.0

±0.049.9+0.10.42±0.01423711.81

39.0

B. Human anti-conjugate antibody responses

IgG (peak tiler) IgM (peak titer)

First dayPatient of rise

Conjugate 260F9 rRA 520C9

Conjugate 260F9 rRA 520C9

1 8 80 40 35 3 15 7 <1 42 16 20 555 4 2 <1 43 10 300 40 30 30 30 20 4 204 15 10 917 2 <1 <1 <1

Fig. 1. The binding of the IgGl murine antibody 260F9 (•)and 260F9 MAb-rRA 1C (O) to human monocytes was assessed by flow cytometry as described in"Materials and Methods." At protein concentrations exceeding 1 x 10"* M, the1C bound significantly more than did native antibody. *, P < 0.01; **, P < 0.05;***,/>< 0.001.

in patient 1, this appeared to be largely an antiidiotype responsesince the human anti-mouse antibody response against the IgGlmurine antibody 520C9 was substantially less than that of260F9. The development of an antiidiotype response distinguishable from a human anti-mouse antibody response was notclear in the other patients tested. In general, the IgG peak titerssubstantially exceeded the IgM peak titers. In all patients, ahuman anticonjugate response was noted with 16 days of theonset of treatment and as early as 8 days following the onset oftherapy in patients 1 and 3.

Patient Tumor Studies. In patient 1 an abdominal wall metastasis was biopsied and found to demonstrate the antigenrecognized by antibody 260F9. A chest wall metastasis wasbiopsied at the conclusion of 8 days of treatment and analyzedfor antigen expression and persistence of the immunoconjugate.While the malignant cells in this lesion expressed antigen, asmeasured by the method of immunoperoxidase staining offrozen lesion sections with 260F9 antibody, no persistent immunoconjugate was detectable in the lesion (data not shown).This finding is not unexpected since higher doses have beenrequired to achieve saturation of tumor binding sites in clinicaltrials with other antibodies (22).

Toxicity Mechanism Studies. The identification of significantcapillary leak syndrome toxicities in association with immunoconjugate therapy suggested the possibility that the immunoconjugate was delivering recombinant ricin A chain to unintended targets and that this delivery might be inducing the toxicsyndrome. We chose to investigate the possibility that theaddition of the ricin A chain had altered the structure of theIgGl murine antibody 260F9 so that the modified protein wascapable of binding to human Fc7 receptors. To test this, humanmonocytes were incubated in vitro with varying concentrationsof immunoconjugate or with 260F9. As shown in Fig. 1, thenative antibody did not bind to monocytes over a broad rangeof concentrations. In contrast, the immunoconjugate exhibiteda concentration-dependent increase in binding so that detectable binding was noted at concentrations as low as 10~8 M. It

should be noted that the peak serum 1C concentrations even atthe lower dose level (Table 4A) were associated with substantialbinding to monocytes in vitro. This enhanced binding wasobserved not only when the monocytes were incubated withthese proteins in small volumes for immunophenotypic analysesbut also when the monocytes were incubated in tissue culturemedia at 1 x IO6cells/ml in the appropriate concentrations of

immunoconjugate (data not shown).4064

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

PHASE I TRIAL OF IMMUNOCONJUGATE 260F9-MAb-rRA

To investigate the mechanisms by which this binding mayoccur, monocytes were incubated with a variety of ICs and otherproteins as shown in Table 5. To evaluate the role of the 260F9-binding portion on monocyte binding, monocytes were exposedto an 1C identical to 260F9 MAb-rRA except that the antibodyportion of the 1C was the MOPC-21 myeloma protein. Toevaluate the possibility that the specific linker used to conjugatethe antibody to ricin A chain was responsible for the observedresults (through either disulfide exchange or another linker-specific mechanism), two ricin A chain 260F9 ICs preparedusing different thioether linker technologies, designated 260F9MAb-Ll-rRA and 260F9 MAb-L2-rRA, were tested as well.The myeloma protein UPC-10, an IgG2a murine protein whichinteracts with the Fc7 receptor expressed by human monocytes,was tested as well. Finally, the ability of preincubation with 50Mg/ml pooled human IgG to inhibit 1C binding via saturationof monocyte Fc7 receptors was evaluated as well. As shown bythe representative experiment depicted on Table 5, all of theICs bound well to monocytes. Furthermore, the binding of each1C was inhibited to control levels when the monocytes werepreincubated with human immunoglobulin. The ICs did notbind to lymphocytes. To further evaluate the Fc7 receptor-mediated uptake of the immunoconjugate, a number of experiments were performed in which human monocytes were preincubated with human immunoglobulin prior to the addition ofthe 1C. As shown in Fig. 2, preincubation with pooled humanIgG clearly suppressed binding of the 1C to human monocytes.

The clinically used 1C preparations contained less than 10%dimcrs and multimers by nonreducing gel electrophoresis. Tofurther investigate the mechanisms by which this 1C may bindto the Fc7 receptor expressed by human monocytes, the abilityof chemically constructed SPDP-linked dimcrs of MOPC-21to bind to human monocytes was investigated. Fig. 3 demonstrates the results of a series of experiments in which monocyteswere incubated with either MOPC-21, 260F9 MAb-rRA, orMOPC-21 dimers at varying concentrations. As expected, theimmunoconjugate bound to human monocytes; this bindingwas inhibited by preincubation with human immunoglobulin.The MOPC-21 dimer bound to monocytes as least as well asthe immunoconjugate. Furthermore, this binding was only partially reversed by preincubation with human immunoglobulin.Since the preparation used may have contained higher multimers of MOPC-21, it is possible that these immunoglobulinaggregates possessed substantially higher affinity for Fc7 receptors than did the dimers. These data suggest one possibleexplanation for the binding of this 1C to human Fc7 receptor-bearing cells.

DISCUSSION

This dose escalation phase I clinical trial of bolus 1C administration was suspended after four patients had been accrued

Table 5 Ricin A chain immunoconjugaies bina to human monocytes via Fey

Preincubation with humanimmunoglobulin

Monocytes(%positive)Tested

protein (25>ig/ml)Human

immunoglobulin260F9MOPC-21UPC-10260F9

MAb-rRAMOPC-21-MAb-rRA260F9-Ll-rRA260F9-L2-rRANo9.79.33.691.448.242.329.064.4Yes_"—__—7.48.16.45.7Lymphocytes(%

positive)No8.110.910.412.012.912.011.68.2Yes—__—8.48.28.18.5

50^

40LUW90)coaiB-

30ooE§|O).E

20.Cu>"àio"oCog

10£--ITTHumanigI1260F9Ii260F9-

Human Ig +MAb-rRA 260F9-MAb-rRA

1—¿�,not done.

Fig. 2. Purified human monocytes were incubated with 10 * M 260F9 or260F9 MAb-rRA as described in the text. Cells were preincubated with eithermedium or 50 »<g/mlof pooled human immunoglobulin (Human Ig) for 30-45min. These conditions saturate human monocyte Fc-yreceptors (data not shown).The binding of human immunoglobulin to monocytes could not be detected inthis assay since l-'(ali').. fragments of goat anti-mouse immunoglobulin were used

to detect the presence of adhered immunoglobulin.

because patients treated by continuous infusion of the 1C inanother phase I trial developed severe and unexpected neurologic toxicity.4 No serious neurotoxicity was observed with the

bolus schedule, and we speculate that the continuous infusionschedule may have resulted in an increased area under thecurve, permitting observation of the neurotoxicity at equivalentdaily 1C doses. We found two additional problems which maylimit the likelihood of reaching therapeutically useful dosagesof this 1C. The first, not unexpectedly, is the development ofanticonjugate antibodies. This response was seen as early as 8days following the onset of therapy in our series. This responsedoes not seem to be trivial in that patient 3 experienced a rashand in patient 1, who developed anticonjugate antibodies byday 8, peak serum levels of 1C decreased late in treatment. Thesecond problem was the toxic syndrome consisting of edema,hypoalbuminemia, weight gain, and, occasionally, congestiveheart failure or pulmonary edema. This toxicity syndrome wasnot related to the schedule of administration of 260F9 MAb-rRA because similar toxicity was noted in patients treated witha continuous infusion schedule at doses of either 50 or 100 ^g/kg daily for 8 days.

This symptom complex has been reported in clinical trialsusing other ICs incorporating the ricin A chain. Spitler et al.

* B. J. Gould, M. J. Horowitz, P. W. Carter, J. Moore. E. Groves, D. Anthony,L. M. Weiner, and A. Frankel. A phase I study of a continuous infusion anti-breast cancer immunotoxin: observation of a targeted toxicity not predicted byanimal studies, submitted for publication.

4065

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

PHASE l TRIAL OF IMMUNOCONJUGATE 2601 <) MAb-rRA

co7°+1Cco

0)60(01"o

E40C

CO1

30O).C1

20ut"

10Percento0--".-__mì%W Ti

rh*MOPC-21260F9-MAb-rRA MOPC-21 MOPC-21 MOPC-21

5x10'8M 10'7M DIMER DIMER DIMER5X10-8M 5X10-9M SxlO-'OM

Fig. 3. In this set of four experiments, monocytcs were incubated with MOPC-21, 260F9 MAb-rRA. or SPDP-linked dimers of MOPC-21. The cells wereprcincubated without (D) or with (D) pooled human immunoglobulin. MOPC-21dimers bound to monocytes at least as well as did 260F9 MAb-rRA. Furthermore,the binding of the dimers was not reversible by preincubation with pooled humanimmunoglobulin (P = 0.26). Since the dimers may have contained multimers ofimmunoglobulin. the enhanced affinity of these complexes for Fc-yreceptors maybe responsible for the relative inability of human immunoglobulin to abrogatebinding.

(14) treated 22 patients with an 1C composed of an antimela-noma antibody and highly purified native ricin A. Weight gain(>5%) was observed in 6 of 22 patients, while hypoalbuminemiawas noted in 20 of 22. Doses of at least 0.5 mg/kg/day resultedin fluid retention and orthostatic hypotension, requiring interruption of therapy. A similar syndrome has been observed inan animal model using the same conjugate (23). If this syndrome is found to be dose related, then it may prove quitedifficult to administer tumor toxic doses to all parts of tumortissue prior to the development of disabling toxicity.

When this syndrome was observed with 1C therapy, it wasimportant to investigate the possibility that this modified protein bound to Fc7 receptors, even though murine IgGl antibodies do not ordinarily bind well to the Fc7 receptor expressed bymonocytes (24). We have found that the 1C binds to humanmonocytes at concentrations which were achieved clinically andthat this binding is dependent upon availability of monocyteFc7 receptors. This binding is not dependent upon the antibody-binding site specificity or to the method of linkage used, so thatFc7 portion modifications are involved, either by the ricin Achain modification of the monomer or by increase in Fc7valency caused by creation of 1C dimers or multimers (Fig. 3).Preparation of ICs which contain only monomers will help toanswer this question. If no Fc7 receptor binding is observedwith 1C monomers, then these preparations may be adequateto permit treatment of patients with doses which achieve tumorsaturation. However, if the 1C monomers bind to monocytes,then ICs which incorporate F(ab) or F(ab'): fragments of anti

body will require testing (25). Preincubation of purified monocytes with pooled human immunoglobulin at concentrationsless than those in serum abrogated detectable binding of the 1Cto human monocytes. However, this lack of detectability maysignificantly underestimate the physiological impact of prolonged exposure to low receptor occupancy levels of 1C. Furtherstudies are therefore required to evaluate the binding and stimulation or cytotoxicity of the 1C to monocytes in the presence

of normal human serum so that the physiological relevance ofthis in vitro model may be assessed more fully. The relationshipof these in vitro observations with the capillary leak syndromewill be clarified when agents which do not bind to monocytesare evaluated clinically.

An additional mechanism for some of the reported 1C toxicityin other trials is hepatic uptake due to recognition of mannose-containing oligosaccharide chains of native ricin A chain byspecific surface receptors expressed by nonparenchymal hepaticcells. Byers et al. (26) have shown that mannose-containingblocking agents can prolong blood circulation of a native ricinA chain-containing immunotoxin and decrease liver uptake aswell in a murine model. Dose-dependent binding of native ricinA chain ICs to rabbit alveolar macrophages following i.v. infusion has been observed (27). In that study, in vitro binding couldbe inhibited by preincubation of alveolar macrophages withvarious mannose-containing compounds, although the role ofpossible Fc7 receptor-mediated uptake was not evaluated. However, the 80% inhibition of binding by mannose argues for theprimacy of mannose receptor-mediated uptake of 1C by rabbitalveolar macrophages. In another study the ability of ricin Achain to act directly as a ligand for the mannose receptorexpressed by rat bone marrow macrophages was demonstratedwith corresponding effects on cellular protein synthesis (28). Inthis latter study, antibody-ricin A conjugates were not evaluated. Since our 1C preparations utilized recombinant nongly-cosylated ricin A chain, we do not believe that mannose receptor-mediated uptake is an important mechanism for the toxic-ities we observed.

Capillary leak syndrome is commonly seen in patients treatedwith high doses of the cytokine interleukin 2, particularly whenthe treatment is combined with infusions of autologous interleukin 2-activated lymphocytes (20, 21). The clinical observation that this syndrome worsens when autologous activatedlymphocytes are added suggests that these cells may be involvedin the pathogenesis of the syndrome. Damle et al. (29) havedemonstrated that these lymphokine-activated killer cells bindto endothelial cell surfaces, further supporting the associationof these cells with the syndrome. Recently it has been demonstrated that endothelial cells from interleukin 2-treated patientsexpress activation antigens, although Fc-y receptors were notevaluated in that study (30). The mechanisms by which suchactivated cells may induce this syndrome remain unknown,although lymphokine-activated killer cells mediate detachmentand lysis of human umbilical vein endothelial cells (29, 31).Although we observed little binding of 260F9 MAb-rRA tohuman lymphocytes (Table 5), studies which evaluate the interactions of ricin A chain ICs with interleukin 2-activated largegranular lymphocytes are required. The recent demonstrationthat human lymphokine-activated killer cells lyse autologousmonocytes (32) suggests that monocyte lysis may be a commonmechanism for disparate etiologies and manifestations of thecapillary leak syndrome.

The mechanisms by which 1C therapy may lead to capillaryleak syndrome require further investigation. Understanding thepathogenesis of this IC-induced toxicity may lead to a betterappreciation of the mechanisms which induce this syndromewhen it is associated with other therapeutic agents.

ACKNOWLEDGMENTS

The authors are grateful to HélèneZintak for assistance in managingthe patients, to Isabella Day for expert manuscript preparation, to Dr.Lou Houston for preparation of 260F9 MAb-Ll-rRA and 260F9 MAb-

4066

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

PHASE I TRIAL OF IMMUNOCONJUGATE 260F9-MAb-rRA

L2-rRA, and to Jeffrey Kassel for preparation of the MOPC-21, SPDP-

linked conjugate.

REFERENCES

1. Olsnes, S. Closing in on ricin action. Nature (Lond.), 328:474-475, 1987.2. Endo, Y., Mitsui, K., Motizuki, M., and Tsurugi, K. The mechanism of

action of ricin and related toxic lectins on eukaryotic ribosomes. J. Biol.Chem., 262: 5908-5912, 1987.

3. Endo, Y., and Tsurugi, K. RNA jV-glycosidase activity of ricin A-chain. J.Biol. Chem., 262: 8128-8130, 1987.

4. Martin, P. J., Hansen, J. A., and Vitella, E. S. A ricin A chain-conlainingimmunoloxin lhal kills human T lymphocytes in vitro. Blood, 66: 908-912,1985.

5. Leonard, J. E., Wang, Q., Kaplan, N. O., and Royston, I. Kinetics of proteinsynthesis in human T-lymphocytes by seleclive monoclonal antibody-ricinconjugates. Cancer Res., 45: 5263-5269, 1985.

6. Bregni, M., DeFabritis, P., Raso, V., Greenberger, J., Lipton, J., Nadler, L.,Rothstein, L., Ritz, J., and Bast, R. C., Jr. Elimination of clonogenic tumorcells from human bone marrow using a combination of monoclonal anli-body:ricin A chain conjugales. Cancer Res., 46: 1208-1213, 1986.

7. Embleton, M. J., Byers, V. S., Lee, H. M., Scannen, P., Blackball, N. W.,and Baldwin, R. W. Sensitivily and seleclivity of ricin toxin A chain-monoclonal antibody 79IT/36 conjugales againsl human lumor cell lines.Cancer Res., 46: 5524-5528, 1986.

8. Byers, V. S., Pimm, M. V., Scannen, P. J., Pawluczyk, I., and Baldwin, R.W. Inhibition of growth of human tumor xenografls in athymic mice trealedwith ricin toxin A chain-monoclonal antibody 79IT/36 conjugates. CancerRes., 47: 5042-5046, 1987.

9. Pirker, R., FitzGerald, J. P., Hamilton. T. C., Ozols, R. F., Laird, W.,Frankel, A. E., Willingham, M. C., and Paslan, I. Characlerizalion ofimmunoloxins aclive against ovarian cancer cell lines. J. Clin. Invest., 76:1261-1267, 1985.

10. Canevari, S., Orlandi, R., Ripamonli, M., Tagliabue, E., Aguanno, S., Miotli,S., Menard, S., and Colnaghi, M. I. Ricin A chain conjugated with monoclonal antibodies seleclively killing human carcinoma cell in vitro. 3. Nati.Cancer Inst., 75: 831-839, 1985.

11. Laurenl, G., Kuhlein, E., Casellas, P., Canal, X., Carayon, P., Poncelel, P.,Correli, S., Rigai, F., and Jansen, F. K. Delerminalion of sensitivity of freshleukemia cells to immunotoxins. Cancer Res., 46: 2289-2294, 1986.

12. Griffin, T. W., Richardson, C., Houslon, L. L., LePage, D., Bogden, A., andRaso, V. Amilumor aclivily of inlraperiloneal immunoloxins in a nude mousemodel of human malignanl mesolhelioma. Cancer Res., 47: 4266-4270,1987.

13. Laurenl, G., Pris, J., Parcel, J. P., Carayon, P., Blylhman, H., Casellas, P.,Poncelet, P., and Jansen, F. K. Effects of therapy with T101 ricin A-chainimmunotoxin in iwo leukemia palienls. Blood, 67: 1680-1687, 1986.

14. Spiller, L. E., de Rio, M., Khenligan, A., Wedel, N. L., Brophy, N. A.,Miller, L. L., Harkonen, W. S., Rosendorf, L. L., Lee, H. M., Mischak, R.P., Kawahala, R. T., Sloudemire, J. B., Fradkin, L. B., Baulisla, E. E., andScannen, P. J. Therapy of palienls with malignant melanoma using amonoclonal antimelanoma antibody-ricin A chain immunotoxin. CancerRes., 47: 1717-1723, 1987.

15. Frankel, A. E., Ring, D. B., Tringale, F., and Hsieh-MA, S. T. Tissue

distribution of breast cancer-associated anligens defined by monoclonal antibodies. J. Biol. Response Modif, 4: 273-286, 1985.

16. Björn,M. J., Ring, D., and Frankel, A. Evaluation of monoclonal anybodiesfor the developmenl of breasl cancer immunoloxins. Cancer Res., 45: 1214-1221, 1985.

17. Boyum, A. Separalion of leukocyles from blood and bone marrow. Scand. J.Clin. Lab. Invesl., 21 (Suppl. 97): 7. 1968.

18. Freundlich, B., and Avdolovic, N. Use of gelalin/plasma coaled flasks forisolatili); human peripheral blood monocyles. J. Iinmuno]. Methods, 62: 31-37, 1983.

19. Matlhews, D. E., and Farewell, V. Analyzing normally distributed data. In:Using and Understanding Medical Statislics, pp. 103-122. Basel: S. KargerAG, 1985.

20. Rosenberg, S. A., Lotze, M. T., Muul, L. M., el a!. Observations on Ihesystemic administration of aulologous lymphokine-aclivaled killer cells andrecombinanl inlerleukin-2 to palienls wilh melaslalic cancer. N. Engl. J.Med., 313: 1485-1492, 1985.

21. Rosenberg, S. A., Lolze, M. T., Muul, L. M., et al. A progress reporl on Ihetreatment of 157 palients wilh advanced cancer using h mpliokinc activatedkiller cells and interleukin-2 or high-dose interIeukin-2 alone. N. Engl. J.Med., 316: 889-897, 1987.

22. Schroff, R. W., Alton, C. M., Woodhouse, C. S., Arams, P. G., Farrell, M.M., Carpenter, B. E., Oldham, R. D., and Foon, K. A. Monoclonal anlibodytherapy in malignanl melanoma: factors effecling in vivo localizalion. J. Biol.Response Modif., 6:457-472, 1987.

23. Harkonen, S., Sloudemire, J., Mischak, R.. Spiller, L. E., Lopez, H., andScannon, P. Toxicily and immunogenicily of monoclonal antimelanomaanlibody-ricin A chain immunoloxin in rals. Cancer Res., 47: 1377-1382,1987.

24. Lübeck.M. D., Sleplewski, Z., Baglia, F., Klein, M. H., Dorringlon, R. J.,and Koprowski, H. The inleraclion of murine IgG subclass proteins wilhhuman monocyle Fc receplors. J. Immunol., 135: 1299-1304, 1985.

25. Slreel, N. E., Fullon, R. J., Sanders, V. M., and Vilelta, E. S. Inhibition ofthe helper function of murine T cells with FAB'-anli-L3T4 ricin A chainimmunoloxin. J. Immunol., 139: 1734-1738, 1987.

26. Byers, V. S., Pimm, M. V., Pawluczyk, 1. Z. A., Lee, H. M., Scannon, P. J.,and Baldwin, R. W. Biodislribulion of ricin loxin A chain-monoclonalanlibody 791T/36 immunoloxin and influence of hepalic blocking agenls.Cancer Res.. 47: 5277-5283, 1987.

27. Bourrie, B. J. P., Casellas, P., Blylhman, H. E., and Jansen, F. K. Study ofthe plasma clearance of anlibody-ricin-A-chain immunoloxins. Eur. J.Biochem., 155: 1-10, 1986.

28. Simmons, B. M.. Stahl. P. D., and Russell. J. H. Mannose receplor-mediateduptake of ricin loxin and ricin A chain by macrophages. J. Biol. Chem., 267:7912-7920, 1986.

29. Damle, N. K., Doyle, L. V., Bender, J. R., and Bradley, E. C. Inlerleukin-2aclivaled human lymphocyles exhibil enhanced adhesion lo normal vascularendolhelial cells and cause Iheir lysis. J. Immunol., 138: 1779-1785, 1987.

30. Colran, R. S., Pober, J. S.. Gimbrone, M. A. Jr., Springer, T. A., Wiebke,E. A., Gaspari, A. A., Rosenberg, S. A., and Lolze, M. T. Endolhelialactivation during interleukin 2 immunotherapy. J. Immunol., 139: 1883-1888, 1987.

31. Miltenburg, A. M. M., Meijer-Paape, M. E., Daha, M. R., and Paul, L. C.Endothelial cell lysis induced by lymphokine-activaled human peripheralblood mononuclear cells. Eur. J. Immunol., 17: 1383-1386, 1987.

32. Djeu, J. Y., and Blanchard. D. K. Lysis of human monocytes by lymphokine-aclivated killer cells. Cell. Immun., Ill: 55-65, 1988.

4067

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from

1989;49:4062-4067. Cancer Res   Louis M. Winer, Joyce O'Dwyer, Joanne Kitson, et al.   Antibody 260F9-Recombinant Ricin A Chain ImmunoconjugatePhase I Evaluation of an Anti-Breast Carcinoma Monoclonal

  Updated version

  http://cancerres.aacrjournals.org/content/49/14/4062

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/49/14/4062To request permission to re-use all or part of this article, use this link

Research. on December 16, 2020. © 1989 American Association for Cancercancerres.aacrjournals.org Downloaded from