Immunogenicity of the Hu14.18-IL2 Immunocytokine Molecule in Adults with Melanoma and Children with Neuroblastoma Jacquelyn A

Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963 Published Online First on September 8, 2009 as 10.1158/1078-0432.CCR-08-2963 Cancer Therapy: Clinical

Immunogenicity of the Hu14.18-IL2 Immunocytokine Molecule in Adults With Melanoma and Children With Neuroblastoma Jacquelyn A. Hank,1 Jacek Gan,1 Hyunji Ryu,1 Amy Ostendorf,1 Michael C. Stauder,1 Adam Sternberg,1 Mark Albertini,1 Kin-Ming Lo,2,3 Stephen D. Gillies,2,3 Jens Eickhoff,1 and Paul M. Sondel1

Abstract Purpose: Immunocytokine (IC) hu14.18-IL2 is a fusion protein of humanized antidisialoganglioside (GD2) antibody (hu14.18) and interleukin (IL)-2. Sixty-one melanoma and neuroblastoma patients received IC in phase I/Ib studies. Patient sera were examined in ELISA to determine if an anti-IC antibody response occurred during treatment. Experimental Design: Serum was assayed for anti-idiotypic antibody (anti-id Ab) based on ability to bridge biotinylated hu14.18 to plate-bound hu14.18 and ability to inhibit binding of hu14.18 to GD2 antigen and/or murine anti-idiotypic antibody. ELISA was also used to detect antibodies to the Fc-IL2 end of hu14.18-IL2. Results: Thirty-two patients (52%) developed an anti-idiotypic antibody response (absorbance, >0.7) in the bridge ELISA. Twelve patients (20%) had an intermediate response, whereas 17 patients (28%) were negative (adsorbance, <0.3). The development of antibody to hu14.18-IL2 detected in the bridge ELISA was not related to the dose of hu14.18- IL2. Twenty of 33 adult patients (61%) demonstrated an anti-idiotypic antibody response based on binding inhibition ELISA. The anti-idiotypic response was inversely correlated (P < 0.002) with IC measured during the second course of treatment, indicating that development of anti-idiotypic antibodies interfered with detection of circulating hu14.18-IL2. All patients developed some inhibitory activity in the binding inhibition assay designed to detect antibodies to the Fc-IL2 region of the IC. There was a positive correlation between the peak serum level of IC in course 1 and the anti–Fc-IL2 response. Conclusions: Patients treated with hu14.18-IL2 developed anti-idiotypic antibodies and anti Fc-IL2 antibodies. No association was seen between development of anti-IC antibodies and clinical toxicity. (Clin Cancer Res 2009;15(18):5923–30)

In an effort to improve antitumor effects with interleukin 2 each IgG1 heavy chain,was created. The proposed mechanism (IL-2; ref. 1) or monoclonal antibody (mAb; ref. 2) alone,or of action is localization to tumor via recognition of tumor- combined treatment with the individual components (3–7), associated GD2 disialoganglioside (10–13). Localization of IC an immunocytokine (IC; refs. 8,9),which contains the tumor facilitates activation of natural killer cells through Fc and IL-2 reactive 14.18 mAb linked to IL-2 at the carboxy terminus of receptors (14) and activation of T cells through IL-2 receptors (15). Natural killer cells mediate cytolytic activity via antibody- dependent cellular cytotoxicity and non–MHC-restricted cytotoxicity (9). In some preclinical models,tumor antigen –specific Authors' Affiliations: 1The University of Wisconsin Paul P. Carbone Comprehensive Cancer Center, Madison, Wisconsin; 2The EMD-Lexigen T-cell memory is also induced (15,16). Clinical reports for Research Center, Billerica, Massachusetts; and 3EMD Pharmaceuticals, separate phase I studies treating melanoma (MEL) and neu- Inc., (now EMD Serono, Inc.), Durham, North Carolina roblastoma (NBL) patients with this IC were recently published Received 12/19/08; revised 5/26/09; accepted 5/29/09; published OnlineFirst (14,17). The present study was designed to determine if pa- 9/8/09. Grant support: NIH CA32685, CA14520, CA87025, CA81403, RR03186, and a tients receiving the IC developed an immune response to the grant from the Midwest Athletes for Childhood Cancer Fund. The pediatric IC. We monitored patients for development of antibody to phase I Study (COGADVL0018) was supported by the Children's Oncology the IC. Adult MEL patients with responding or stable disease Group, and performed by COG phase I institutions. were eligible to receive a second course of IC (14). Pediatric The costs of publication of this article were defrayed in part by the payment NBL patients with stable or responding disease were eligible of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. toreceiveuptofourorsixcoursesofIC,respectively(17). Note: Stephen D. Gillies is currently employed by Provenance Biopharma- We established ELISAs to detect antibodies specific for the ceuticals Corp. two separate functional ends of the IC. Antibodies against the Requests for reprints: Jacquelyn A. Hank, University of Wisconsin, K4/454, idiotypic determinant (18) and against the carboxy terminus of 600 Highland Avenue, Madison, WI 53792. Phone: 608-263-7262; Fax: 608- 263-4226. E-mail: [email protected]. the IgG heavy chain where IL-2 is linked (Fc-IL2 end) were de- F 2009 American Association for Cancer Research. tected. These antibodies could potentially interfere with the doi:10.1158/1078-0432.CCR-08-2963 proposed functions of the IC. An anti-idiotypic antibody might

www.aacrjournals.org 5923 Clin Cancer Res 2009;15(18) September 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Cancer Therapy: Clinical

polystyrene plates (Falcon #3915; Fisher Scientific) at a concentra- Translational Relevance tion of 250 pmol/L. Hu14.18-IL2 was diluted to 250 ng/mL and further diluted 1:2 with patient serum. This mixture was incubat- This article presents the results of immunologic ed at room temperature for 1/2 h and 100 μL per well added to monitoring of melanoma and neuroblastoma pa- duplicate wells and incubated for 2 h at 37°C. Hu14.18-IL2 bound tients receiving hu14.18-IL2 as part of phase I clinical to GD2 was detected using biotinylated goat-anti-human-IL2-anti- trials. The discovery that some of these patients de- body (8.33 ng/mL; R&D Systems),followed by ExtrAvidin –horse- velop antibodies to the hu14.18-IL2 molecule will radish peroxidase (HRP) conjugate (2 μg/mL; SIGMA-Aldrich help us in the design of future clinical trials with this Chemicals) and addition of TMB substrate (DAKO Chemicals). μ molecule. The reaction was stopped using 50 Lperwellof2NH2SO4. The absorbance (OD 450/570) was used to determine the amount of hu14.18-IL2 detected interpolating to a standard curve. The % inhibition was then calculated as follows: prevent the IC from targeting to tumor (18). An antibody against the Fc-IL2 end of the IC (anti–Fc-IL2) might interfere with immune activation facilitated through IL-2. We report %=100-100 × (hu14.18 detected with post-treatment serum) here on the occurrence,frequency,and potential immunologic (hu14.18 detected with pre-treatment serum) effects of the antibody response to hu14.18-IL2.

Materials and Methods The specificity of the anti-IC antibodies detected in the binding inhibition assays was assured by comparing IC detected using pretreatment Hu14.18-IL2. IC (EMD 273063) was provided by EMD Pharma- serum to IC detected using serum obtained at times following treatment. ceuticals,Inc.,(now EMD Serono,Inc.). One miligram of IC contains • Anti-IC binding inhibition (1A7) for detection of anti-IC antibo- 3×106 IU of IL-2 (19) and 0.8 mg of the hu14.18 antibody. dies using 1A7-coated plates (Fig. 1C). This assay is identical to the Study design. These phase I trials were nonrandomized dose esca- anti-IC binding inhibition assay described in the paragraph above lation studies. Initial clinical and immunologic results were previously (for Fig. 1B) with the exception that the capture molecule is the reported (14,17). Briefly,hu14.18-IL2 was given as a 4-h i.v. infusion murine 1A7 antibody. This antibody,specific for the idiotypic de- on days 1,2,and 3 of each 28-d treatment course. Adult patients re- terminant of the 14.18 anti-GD2 antibody,binds to the14.18 ceived up to two courses and pediatric patients received up to six mAb,and thus serves as a mimic for GD2 (24). courses of IC. Unless otherwise indicated,serum samples were taken • Anti-idiotypic binding inhibition for detection of anti-idiotypic with morning blood draws,before administration of IC. The day and antibody by using 1A7 mAb–coated plates and biotinylated course for blood samples are identified as follows: C1D1,course 1,day hu14.18 mAb (Fig. 1D). The 1A7 mAb resembles the GD2 antigen 1; C3D8,course 3,day 8. Peak IC serum levels were determined from and will capture or bind biotinylated hu14.18 antibody. The blood obtained within 1/2 h of completing the IC infusion. bound biotinylated 14.18 can then be quantified using the same Cell lines. M21 (GD-2–positive melanoma; refs. 14,17) and IL-2 biotin-avidin enzyme system used in the bridging assay described receptor–positive RL-12 (subline of NKL-human leukemia obtained above (Fig. 1A). Briefly,C8 Maxisorp microtiter plates (Nunc) are from Dr. Paul Leibson of the Mayo Clinic,Rochester MN; ref. 20) were coated overnight at 4°C with 120 μLof2μg/mL 1A7. Serum sam- maintained as previously described. ples are diluted 1:5 with a solution of 3.1 ng/mL biotinylated hu14.18 and added to plates (100 μL/well). The plates are incubated ELISAs overnight at 4°C and bound 14.18 is determined using the same SIL-2Rα. SIL-2Rα was measured by (Immunotech) ELISA kit. biotin-ExtrAvidin-HRP enzyme system used above. Results are pre- Detection of IC. Measurement of IC in patients' sera by ELISA was sented as “% inhibition,” where the amount of hu14.18 detected in done as previously described (18,21,22). pretreatment serum is defined as 0% inhibition for that patient, Detection of anti-IC antibodies. The humanized 14.18-IL2 has two and % inhibition is calculated as for the assay presented in Fig. 1B. types of immunogenic epitopes recognized as foreign by some patients. • Anti–Fc-IL2 binding inhibition for detection of antibody to the Fc- The anti-GD2 idiotypic determinant and the determinants created IL2 region of the IC using anti-IL2–coated plates and biotinylated where IL-2 is directly linked to the carboxy terminus of IgG heavy Fc-IL2 (Fig. 1E). Briefly,microtiter plates (Nunc) are coated over- chains,called Fc-IL2. We have created ELISAs that allow the detection night at 4°C with 120 μLof2μg/mL neutralizing rat monoclonal of antibodies to these determinants (Fig. 1). anti-human IL-2 antibody MQ1-17H12 (BD Pharmingen). Serum • Anti–Id-Bridge to detect anti-14.18 Idiotypic antibody is similar to samples are diluted 1:5 with a solution of 5 ng/mL biotinylated our previously published “bridging” method detecting anti-14.18 hu14.18 Fc-IL2 fragment and added to plates at 100 μL per well. chimeric antibody (Fig. 1A; ref. 23). Serum samples obtained before The plates are incubated overnight at 4°C and bound Fc-IL2 deter- therapy from all patients were negative in this assay [adsorbance mined using enzyme system described above. Any decrease in the (OD),<0.3]. detected amount of Fc-IL2 from that expected reflects the presence • Anti–IC-binding inhibition (GD2) for detection of anti-IC anti- of inhibitory activity (anti–Fc-IL2 antibody). Results are presented bodies using GD2-coated plates (Fig. 1B; ref. 21). Hu14.18-IL2 as % inhibition,where the amount of Fc-IL2 detected in pretreat- binds to GD2 and is detected using biotinylated anti-IL2 antibo- ment serum is defined as 0% inhibition for that patient,and % in- dies. When patient serum containing anti-idiotypic antibody is hibition is calculated as for assays B and D above. combined with IC before adding to GD2-coated plates,the anti- Statistical analysis. The association between anti-idiotypic activity idiotypic antibody inhibits the 14.18 antibody component of the and IC dose used Pearson's correlation analysis. The comparison of peak IC from binding to GD2. In addition,when patient serum con- serum levels of hu14.18-IL2 between courses used Wilcoxon Signed Rank tains antibodies specific for the Fc-IL2 determinant,this antibody test. An exact McNemar's test was used to compare anti-idiotypic and binds to the IC and inhibits the biotinylated anti-IL2 antibody anti–Fc-IL2 response rates between courses. Logistic regression analysis from detecting the IC. Thus,this ELISA detects anti-IC antibodies was done to assess the dose response relationship between IC dose and against both idiotypic and Fc-IL2 determinants. Briefly,purified anti–FC-IL2 response. χ2 analysis was used to evaluate the association disialoganglioside GD2 (Advanced ImmunoChemical,Inc.) sus- between anti-IC antibodies and clinical toxicities. A two-sided signifi- pended in 100% ethanol was absorbed to 96-well flat-bottom cance level of 0.05 was used for all statistical tests.

Clin Cancer Res 2009;15(18) September 15, 20095924 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Immunogenicity of the Hu14.18-IL2 Immunocytokine

Results idiotypic Ab response,14 had their highest OD value during course 1,whereas only 7 had their highest value after the last Detecting and quantifying anti-idiotypic responses. The anti– course. This indicates that,for most patients,continued exposure id-bridge assay,based on our previously developed human to IC does not continue to “boost” the anti-idiotypic response anti-chimeric antibody assay (23),was established to specifical- detected in this bridge assay. ly see how well patients' serum specimens (obtained at greater The influence of anti-IC antibody on IC binding to GD-2. We than or equal to thrice per 4-week treatment course) could tested whether the serum samples could inhibit hu14.18-IL2 IC bridge biotinylated hu14.18 mAb to hu14.18 bound to the from binding to GD2. We devised a GD2-based anti-IC ELISA plate (Fig. 1A). (Fig. 1B; ref. 21) where serum from a patient is incubated with a Of 61 adult and pediatric patients evaluated in these phase I fixed amount of IC to determine if anti-idiotypic antibody in- dose escalation studies,reflecting 126 courses of hu14.18-IL2 hibits binding of hu14.18-IL2 to the plate-bound GD-2. This treatment,28% were negative,20% showed a minimal re- assay was used for the first 18 MEL patients accrued to the sponse,and 52% had a positive anti-idiotypic response with phase I study. Six patients showed measurable activity in at least one specimen showing an OD reading of >0.7 No pre- the anti–IC-GD2 binding inhibition ELISA but no activity in treatment samples were positive and there was no correlation the bridge assay (data not shown; Fig. 1A). This suggests that of anti-idiotypic activity and IC dose. The “anti-idiotypic” this binding inhibition assay (Fig. 1B) is more sensitive than specificity of this assay was confirmed by showing that serum samples from patients that were able to “bridge” biotinylated the bridge assay,or that it may detect antibody not detectable hu14.18 mAb to hu14.18 bound to the plate could also in the bridge assay. It is possible that some high affinity and/or bridge biotinylated hu14.18 mAb to a plate coated with low level of anti-idiotypic antibody that can be detected in this ch14.18 mAb,but not to plates coated with Rituxan,Hercep- GD2 binding inhibition assay may not be detected in the bridg- tin,KS1/4,R24,or UPC-10 humanized or murine mAbs (data ing ELISA,indicating a potential distinction in sensitivity not shown). for anti-idiotypic detection. The IC structure is such that two We next determined when the peak anti-idiotypic responses distinct types of anti-IC antibody might be detected in this were detected. The adult study included 19 patients who received GD2-based IC-binding inhibition assay. First,anti-idiotypic two courses of IC. Seven patients did not develop any “bridging antibodies could prevent the IC from binding to the plate- antibody,” whereas 12 showed a detectable anti-idiotypic bound GD2. Second,an antibody against the Fc-IL2 linkage response. Five of these 12 showed their highest anti-idiotypic region might interfere with detection of IC when using an activity in course 1,3 in course 2,and 4 had similar levels of anti-IL2 detection antibody (shown schematically in Fig. 1B). anti-idiotypic bridging antibodies during course 1 and 2. In the To address this,we developed two additional ELISA assays pediatric study,there were 20 patients who received more than that independently detected anti-idiotypic antibody or anti- one treatment course. Three of these patients did not develop a body against the Fc-IL2 linkage component. These are the bridging antibody (one patient with three courses and two anti-idiotypic binding inhibition (Fig. 1D) and anti–Fc-IL2 patients with four courses). Seventeen developed detectable binding inhibition (Fig. 1E) assays. anti-idiotypic Ab,9 had their highest OD value in their first course, Anti-idiotypic antibodies. The anti-idiotypic binding inhibi- 4 during their final course,and 4 during a course between their first tion assay was designed to detect an antibody specific for the and final courses. Thus,it seems that for this bridging anti- idiotypic determinant. Patients' serum was added to biotinylated idiotypic assay,the highest values are more often observed after hu14.18 mAb,before the ELISA capture step (Fig. 1D). The the first course than the last. Of 29 patients in the two studies capture antibody is the murine 1A7 mAb (24),specific for that received more than one course and developed an anti- the 14.18 antibody idiotypic determinant. When patients have

Fig. 1. ELISA methods to detect antibodies to the hu14.18-IL2 IC. A, anti–Id-Bridge detects anti-idiotypic antibody in patient serum bridging biotinylated hu14.18 antibody to plate-bound 14.18 antibody. B, anti-IC binding inhibition (GD2) detects anti-idiotypic antibody in patient serum interfering with hu14.18-IL2 IC binding to GD2-coated ELISA plates and also detects anti Fc-IL2 antibody in patient serum that interferes with anti-IL2 detection of IC bound to GD2-coated plates. C, anti-IC binding inhibition (1A7) detects both anti-idiotypic and anti–Fc-IL2 antibodies, as in B, with the exception that the capture reagent is plate-bound murine anti-idiotypic 1A7, which specifically recognizes the 14.18 idiotypic determinant. D, anti–Id binding inhibition detects anti-idiotypic antibody in patient serum interfering with binding of biotinylated hu14.18 to plate-bound 1A7. E, anti–Fc-IL2 Binding inhibition detects antibody in the patient serum to a determinant on the IC where the IL-2 is genetically linked to the carboxyl terminus of the IgG1 heavy chain (anti Fc-IL2 antibody). This antibody in patient serum blocks the binding of the biotinylated Fc-IL2 fragment to plate bound anti-IL2 antibody.

www.aacrjournals.org 5925 Clin Cancer Res 2009;15(18) September 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Cancer Therapy: Clinical an anti-idiotypic antibody,it binds to the idiotypic site of the Although the magnitude of the anti-idiotypic response de- hu14.18 mAb and may compete for the binding of the tected in this binding inhibition assay was not influenced by hu14.18 by the 1A7 plate-bound “capture” mAb,thereby in- the dose of hu14.18 administered,it was influenced by the hibiting the binding of the hu14.18-biotin to the plate (as de- number of courses the individual patients received. This was tected by the ExtrAvidin-HRP system). This assay has been noted in the pediatric study where 28 patients received from done on serum from all 61 patients in these phase I trials. one to four courses of hu14.18-IL-2. The percentage of patients The results from course 1 for all adult MEL patients are shown developing >50% inhibition in the anti-idiotypic binding inhi- in Fig. 2A. Results are plotted as % inhibition,based on how bition assay increased with the number of courses of IC admin- serum obtained after IC treatment compares to pretreatment istered (Table 1). This change in reactivity was significant serum. Each line reflects data from a separate patient. When between course 1 and subsequent courses (course 1 versus data from both course 1 and 2 for these 33 adults are com- course 2,3,or 4, P = 0.002, P = 0.026, P = 0.007,respectively), bined,26 patients (79%) showed a statistically significant an- as well as when comparing course 1 to courses 2,3,and 4 com- ti-idiotypic response. Serum specimens showing >28% bined (P ≤ 0.001). This contrasts with the results of the bridging inhibition were considered to show positive responses (above anti-idiotypic ELISA,where more patients developed their high- background),based on being greater than the mean inhibition est bridging activity during their first course. plus 2 SD for the pretreatment serum samples from these We next asked whether the induction of anti-idiotypic anti- same patients in this assay. When results in the anti-idiotypic body as detected by this anti-idiotypic binding inhibition ELISA bridge and anti-idiotypic binding inhibition assays are com- (Fig. 1D) would also influence the serum levels of hu14.18-IL2 pared,for all 126 treatment courses for all 61 patients,4 dis- measured during a second course of treatment. When data from tinct reactivity patterns are observed. Nine patients (15%) 11 evaluable adult melanoma patients,who received two were negative in both assays. Thirty-two patients (52%) devel- courses of IC at the same dose for each course were analyzed, oped anti-idiotypic activity detected in both bridging and those patients that showed greater inhibitory activity in this binding inhibition assays. Eight of the 61 patients (13%) de- binding inhibition assay also showed a corresponding decrease veloped reactivity in the binding inhibition assay but not in in their peak IC serum level measured immediately after finish- the bridge assay. Twelve patients (20%) were positive in the ing the 4-hour infusion on C2D1 compared with their peak lev- bridge assay and negative in the binding inhibition assay. el obtained immediately after the 4-hour infusion of the same IC dose on C1D1 (R = -0.83; P = 0.002; data not shown). This indicates that the patients with the strongest anti-idiotypic antibody responses show the lowest detectable serum IC levels during the first day of their second course of treatment. Anti–Fc-IL2 antibodies. A separate “anti–Fc-IL2 binding inhibition” assay was designed to detect antibody specific for the Fc-IL2 linkage component of the IC (Fig. 1E). Serum specimens were added to biotinylated Fc-IL2 fragments before their capture on the ELISA plate coated with an anti-IL2 antibody,followed by detection with ExtrAvidin-HRP. If the patient made an antibody that bound to this fragment,it might interfere with the binding of the Fc-IL2 to the capture anti-IL2 antibody. Data from all 33 adult MEL patients are shown for course 1 in Fig. 2B. C1D8 serum from 32 of 33 patients showed development of detectable inhibitory activity. Table 2 shows representative anti-IC values in the three binding inhibition assays (Fig. 1B-D) for two patients with strong anti-idiotypic responses. Separate assays (data not shown) evaluate the functional effects,and the specificity,of the anti IC antibodies detected in these ELISA assays,and will be presented in a subsequent report.4 There was no relationship between the magnitude of the anti–Fc-IL2 response and the level of hu14.18-IL2 detected in the serum during course 2 (R = 0.272; P = 0.41). A dose-response relationship was identified for this anti– Fc-IL2 response; patients receiving higher doses of IC show stronger anti–Fc-IL2 responses measured on day 8 and day Fig. 2. Anti-ID (Fig. 1D) and anti–Fc-IL2 (Fig. 1E) results for C1D0, D8, and 15 in course 1 (P = 0.005; data not shown). Similarly there D15 for all 33 adult patients in course 1. Results shown are % inhibition, was a significant positive correlation between the area under based on the change in concentration of biotinylated hu14.18 (A)or biotinylated Fc-IL2 (B) detected when combined with serum obtained on the curve for serum concentration of hu14.18-IL2 on C1D1 day 8 and day 15 compared with serum obtained before treatment. A, the and the development of anti–Fc-IL2 antibodies. The biolog- anti-idiotypic results are shown. The mean ± S.D values are 29.6 ± ical mechanism for these results requires further evaluation, 22.0% for d8 and 27.9 ± 23.7% for d15. B, the anti Fc-IL2 results are shown. The mean ± S.D values are 37.1% ± 12.9% for d8 and 30.8 ± 15.9% for but might suggest that greater doses of hu14.18-IL2 or greater d15. Serum specimens showing >28% inhibition were considered to show positive responses (above background), based on being outside of 2 SDs from the mean concentration of hu14.18-IL2 or Fc-IL2 detected when using pretreatment serum samples from these same patients in these assays. 4 Hank, et al., in preparation.

Clin Cancer Res 2009;15(18) September 15, 20095926 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Immunogenicity of the Hu14.18-IL2 Immunocytokine

Table 1. Time course for development of anti-Id and anti-Fc-IL2 antibodies in pediatric patients

Anti-ID inhibition Anti–Fc-IL2 inhibition

Course 1 2 3 41234 N 2820151126181511 (++) 25% 50% 53% 64% 31% 22% 47% 55%

NOTE: The % of pediatric patients with anti-idiotypic and anti-FcIL2 antibodies seen in the binding inhibition ELISAs following each of four courses of IC. ++, >50% inhibition in the anti-idiotypic and >40% inhibition in the anti-FcIL2 ELISA, respectively. exposures to hu14.18-IL2 were more potent at inducing the an- atric trial (Fig. 3) shows that the anti-idiotypic response may be ti–Fc-IL2 response. preventing IC detection,as well as removing some IC from the Clinical correlations with anti-IC antibody. Data from two circulation. The serum levels of hu14.18-IL2 immediately after representative NBL patients receiving four courses of hu14.18- the 4-hour IC infusion (Fig. 3) have been measured on days 1 IL2,one with and one without an antibody response to the and 3 for each of the four courses. IC,are presented in Fig. 3. Patient 10 generated very little The levels of anti-idiotypic and anti-FcIL2 antibody were anti-idiotypic antibody (Fig. 3A) and the amount of IC detected evaluated in the MEL study for any relationship with the previ- in the serum was similar for all four courses (serum IC levels ously published hu14.18-IL2 induced changes: (a)changein were measured on days 1 and 3 of each treatment course. In lymphocyte counts,(b ) change in lymphocyte phenotype contrast,patient 19 developed anti-IC antibodies during the (increase in % CD 16+,and CD56 + and decrease in % CD3+ first course of treatment,which were evident in the binding in- peripheral blood mononuclear cells),and ( c) increase in serum hibition assay by day 8 of course 1 (Fig. 3B). Anti-idiotypic anti- sIL2R level (14). No significant correlations were seen between bodies were also noted in the bridging assay following courses the anti-idiotypic response and these three parameters. The an- 2,3,and 4. The serum IC levels detected during courses 2,3, ti–Fc-IL2 response did not correlate with lymphocyte numbers and 4 were markedly lower than the levels seen in course 1, or phenotype,but patients that induced the strongest anti –Fc- indicating that the anti-IC response was associated with lower IL2 antibody were also those that showed a greater increase in detectable IC. The lymphocyte count and serum level of sIL2Rα their sIL2Rα levels in course 2 than in course 1 (P < 0.01). The are both indicators of immune activation induced by IL2 (25). mechanism for this relationship remains uncertain. Although the level of IC detected in serum from patient 19 was lower in courses 3 and 4 than in course 1 (presumably due to Discussion the anti-IC response),the lymphocytosis and increase in sIL2- Rα in courses 3 and 4 were comparable with that seen in course Treatment of adult and pediatric patients with IC induces an- 1 (Fig. 3D). Thus,administration of IC in courses 3 and 4 to tibody responses to IC in some but not all patients. Some pa- patient 19 still resulted in comparable IL2-induced in vivo tients initially make a low level of antibody that does not activation (lymphocytosis and increase in sIL2Rα) despite the increase progressively with increasing number of courses. relative inability to detect circulating IC in the ELISA. Patient Others make antibody responses that are more detectible with 10 generated very little anti-idiotypic antibody and values subsequent courses of treatment. This phenomenon was previ- for lymphocyte counts,and sIL2R α increased and remained ously reported by Welt et al. (26) who treated 11 patients with elevated throughout the four courses of therapy (Fig. 3C). advanced colorectal cancer with the humanized IgG1 A33 anti- As noted above,greater anti-idiotypic responses were associ- body. Of these 11 patients,3 did not develop an antibody ated with an inhibition in the detected levels of IC on C2D1 to A33 antibody. The eight patients developing a “human (P < 0.002). Either the circulating anti-idiotypic Ab in these pa- anti-humanized antibody” (HAHA) response,were divided into tients is removing the IC from the serum or preventing it from two groups. Five patients developed a type I HAHA response, being detected in the ELISA. Data from patient 19 in the pedi- which generally developed early,within 2 weeks of treatment

Table 2. Anti–hu14.18-IL2 activity (% inhibition) detected in binding inhibition assays

Day of serum collection Patient 5 Patient 9 C1D8 C1D15 C2D8 C2D15 C1D8 C1D15 C2D8 C2D15

Anti-idiotypic binding inhibition 93 90 83 85 642 94 90 Anti-Fc-IL2 binding inhibition 43 34 57 72 42 2653 51

Anti-IC binding inhibition (GD2)8795989570509999

NOTE: The % inhibition in the ability to detect a standard amount of biotinylated hu14.18 (anti-idiotypic, as shown in Fig. 1D), Fc-IL2 fragment (anti–Fc-IL2, as shown in Fig. 1E), or IC (anti-IC, as shown in Fig. 1C) in ELISA when combined with serum samples obtained on day 8 or 15 of the first or second course of therapy with hu14.18-IL2 IC. All % inhibition values are based on the pretreatment serum sample for each patient (defined as 0% inhibition for that patient). Values shown are % inhibition for two patients, treated with two courses of hu14.18-IL2 in the adult MEL study that showed greater than average inhibitory activities in these three assays.

www.aacrjournals.org 5927 Clin Cancer Res 2009;15(18) September 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Cancer Therapy: Clinical

Fig. 3. Serum IC level measured at multiple time points on days 1 and 3 of each 3-d course of IC for two patients who received four courses of hu14.18-IL2. Solid lines, the IC level with each of 10 determinations per infusion indicated by an open circle (A and B). Black bars, anti-IC antibodies detected in the anti-idiotypic binding inhibition assay (Fig. 1D); white bars, the anti-IC antibodies detected in the anti-idiotypic bridge assay (Fig. 1A; A and B). C and D, the lymphocyte count (black bars) and sIL-2 receptor (white bars), surrogate markers for immune activation. and usually resolved within 7 weeks of treatment and did not hu14.18-IL2 infused daily on days 1,2,and 3 of that course progress with the number of courses. Three patients developed a may have “complexed” with the majority of the circulating type II HAHA response,characterized by progressively increasing anti-idiotypic antibody,thereby limiting the amount of anti- titers of anti-huA33 antibodies. We noted similar trends in anti- idiotypic antibody that could be detected on day 4. The peak IC responses. Some patients developed antibodies that increased IC serum levels during courses 1 and 2 are ∼4 μg/mL,whereas in magnitude with the number of courses (i.e.,patient 19; Fig. 3). during courses 3 and 4,serum IC was virtually undetectable. Some patients developed detectable levels of anti-IC antibody, This absence of detectable IC following an IV hu14.18-IL2 infu- but the strength of the response did not continue to increase, sion suggests that the anti-IC response was able to “neutralize” or did not seem to affect the level of IC detected in the patient virtually all of the hu14.18-IL2 given to this patient at that time. serum following subsequent IC infusions (i.e.,patient 10; Similarly,the anti-idiotypic response measured with the bridge Fig. 3). Although it seems that patient 19 developed a response ELISA on day 4 of courses 3 and 4 is only ∼20% of that seen on that is similar to the type II response described by Welt et al. (26), day 1 of these same two courses,suggesting that infusing the significance of this response in these IC-treated patients is not hu14.18-IL2 that generated a peak serum level of 4 μg/mL dur- yet known,as the generation of an antibody was not associated ing course 1 and 2 was able to “neutralize” most of the circulat- with increased toxicity upon subsequent infusions of IC. ing anti-IC antibody. This suggests that the serum level of In Fig. 3 (and in data not shown),anti-IC antibodies that are functional anti-IC antibody might also be near 4 μg/mL. Because detected in the binding inhibition ELISA limit the detection of our other methods of anti-idiotypic detection require binding IC in patient serum. Of interest are the anti-idiotypic antibodies inhibition assays,where the inhibiting anti-IC antibody is in ex- measured by the bridging assay on days immediately following cess,it is impossible to determine the actual concentration (i.e., each 3-day course of IC. In Fig. 3B,patient 19 shows high OD μg/mL) of anti-IC antibody. The antibody to the Fc-IL2 determi- values corresponding to the levels of anti-idiotypic bridging an- nant does not completely inhibit Fc-IL2 binding to anti-IL2 mAb tibody at C2D8 and all subsequent time points,except for the in the ELISA. Unlike the anti-idiotypic assays,where % inhibi- striking drop in activity seen on day 4 of courses 3 and 4. These tion values were occasionally approaching 100%,% inhibition serum samples are obtained 20 hours following the completion values in the anti–Fc-IL2 assay rarely exceeded 90% (see Table 2). of the third IC infusion. As the half-life of the IC is 3.1 to 3.7 Furthermore anti–Fc-IL2 inhibitory activity detected in these hours (14,17),circulating IC should not be influencing the re- ELISA is readily lost with a 10-fold dilution of the serum sample sults of the course 3 and 4 day 4 bridging assay. The major drop (data not shown). In contrast,anti-idiotypic Ab can completely in the anti-idiotypic antibody seen in courses 3 and 4 on day 4 inhibit specific binding of the hu14.18 antibody to GD2-coated compared with the day 1 value suggests that the 10 mg/M2 of (Fig. 1B) and 1A7-coated (Fig. 1C) ELISA plates. In some

Clin Cancer Res 2009;15(18) September 15, 20095928 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Immunogenicity of the Hu14.18-IL2 Immunocytokine patients,serum samples continue to inhibit at dilutions as great remains controversial whether an anti-idiotypic response might as 1:100. However,in two pediatric patients with considerable be beneficial or harmful for the therapeutic effect desired in trea- inhibition in the anti-idiotypic binding inhibition assay ted patients. In some clinical studies of mAb used as cancer throughout four courses of treatment,the serum sample with therapy,the presence of an anti-idiotypic antibody response,par- the most potent inhibition (and the ability to retain inhibitory ticularly if transient,has been associated with an improved anti- activity with 1:100 dilution) was noted in the second,rather tumor effect (27–29). This improved effect has been postulated than third or fourth course (data not shown),suggesting some to result from the induction of an “anti–anti-idiotypic” response, waning of the anti-idiotypic response. designated an “antibody-3 response,” which may have direct The clinical significance of the anti-idiotypic antibodies is yet antitumor activity. In fact,clinical efforts to induce a potent anti- to be determined. Of these 61 melanoma and neuroblastoma tumor response have used immunization with anti-idiotype patients,52 developed antibodies detected in either the anti- antibodies as vaccines,designed to activate a tumor reactive an- idiotypic bridge or anti-idiotypic binding inhibition assays. Of tibody-3 response (24,29). Our initial efforts to detect antibody these 52,32 patients were positive in both assays,8 were posi- 3 responses in selected patients showing strong anti-idiotypic tive only in the binding inhibition assay,and 12 were positive responses have not revealed antibody 3 activity (data not only in the bridge assay. These anti-idiotypic antibodies were shown). It remains uncertain whether the induction of the not associated with an increase in toxicity or allergic phenome- anti-idiotypic and anti–Fc-IL2 antibodies might help or interfere na (14,17). Nor were these anti-idiotypic antibodies associated with the desired in vivo clinical effect. As there were no clinical with changes in clinical laboratory markers or markers associat- responses noted in these phase I studies,it is impossible to ed with IL-2 induced immune activation (sIL-2Rα and increase predict the impact of the generation of antibody response to in the lymphocyte count). It is not clear if the amount of anti-IC the IC upon clinical outcome. We have just completed phase II antibody present in the serum would be sufficient to limit the trials of this agent for 14 adults with MEL and 39 children with ability of the IC to target GD2-positive tumors in vivo and,if so, NBL; antitumor activity was seen in each study (30,31). We are what levels of reactivity in vitro would correspond to meaningful now performing the in vitro analyses of anti-IC responses for in vivo inhibition of function. Our preliminary data indicate that these 53 patients. levels of anti-IC antibody detected for most patients in the In summary,anti-IC antibodies were formed in most pa- ELISA reported here do not necessarily correspond to levels that tients. These included HAHA responses to the idiotypic deter- inhibit IC binding to GD2 or IL2R in conditions designed to minant and antibodies specific for a neoantigen created at the simulate the in vivo setting.4 The development of anti–Fc-IL2 an- Fc-IL2 junction site of the IC molecule. These antibodies do not tibody also was not associated with an increase in toxicity or seem to induce allergic reactions or increase the toxicity of the change in the lymphocyte count,but there was a significant IC. The in vivo functional significance of this antibody response hu14.18-IL2 dose effect and correlation of anti–Fc-IL2 response is yet to be determined. The lymphocytosis and sIL2R levels with the AUC for hu14.18-IL2. This suggests that greater expo- seen in the face of highly reactive anti–Fc-IL2 antibody,suggest sure to hu14.18-IL2 induces more anti–Fc-IL2 antibodies. In ad- that these anti-IC antibodies do not inhibit IL-2–induced im- dition,patients that induced the strongest anti– Fc-IL2 antibody mune activation in vivo. Additional studies are under way to were those that showed a greater increase in their sIL2R levels in better characterize the in vivo effects of these anti-idiotypic course 2 than in course 1. This indicates an association between and anti–Fc-IL2 antibodies. stronger activation of sIL2R in a patient's second course with induction of a strong anti–Fc-IL2 antibody. The mechanism of this requires further study,as do its clinical implications. Disclosure of Potential Conflicts of Interest Although anti-idiotypic antibodies can interfere with the desired antigen binding function of the mAb,at least in vitro,it No potential conflicts of interest were disclosed.

References 1. Rosenberg SA. Progress in human tumour im- refractory follicular non-Hodgkin's lymphoma. cytolysis and suppression of tumor growth. Cancer munology and immunotherapy. Nature 2001; Br J Haematol 2002;117:828–34. Res 1987;47:1098–104. 411:380–4. 6. Fleming GF, Meropol NJ, Rosner GL, et al. A 11. Schulz G, Cheresh DA, Varki NM, et al. Detec- 2. Yu AL, Uttenreuther-Fischer MM, Huang CS, phase I trial of escalating doses of trastuzumab tion of ganglioside GD2 in tumor tissues and et al. Phase I trial of a human-mouse chimeric combined with daily subcutaneous interleukin 2: sera of neuroblastoma patients. Cancer Res antidisialoganglioside monoclonal antibody report of cancer and leukemia group B 9661. Clin 1984;44:5914–20. ch14.18 in patients with refractory neuroblasto- Cancer Res 2001;8:3718–27. 12. Cheung NK, Lazarus H, Miraldi FD, et al. Gangli- ma and osteosarcoma. J Clinical Oncol 1998;16: 7. Fiedler W, Kruger W, Laack E, Mende T, oside GD2 specific monoclonal antibody 3F8: a 2169–80. Vohwinkel G, Hossfeld DK. A clinical trial of phase I study in patients with neuroblastoma and 3. Hank JA, Robinson RR, Surfus J, et al. Aug- edrecolomab, IL2 and GM-CSF in patients with malignant melanoma. J Clin Oncol 1987;5:1430–40. mentation of antibody dependent cell mediated advanced colon cancer. Oncol Rep 2001;8: 13. Cheung NK, Kushner BH, Cheung IY, et al. cytotoxicity following in vivo therapy with 225–31. Anti-G(D2) antibody treatment of minimal resid- recombinant Interleukin-2. Cancer Res 1990;50: 8. Gillies SD, Reilly EB, Lo KM, Reisfeld RA. Anti- ual stage 4 neuroblastoma diagnosed at more 5234–9. body-targeted interleukin 2 stimulates T-cell kill- than 1 year of age. J Clin Oncol 1998;16:3053–60. 4. Bajorin DF, Chapman PB, Wong G, et al. Phase ing of autologous tumor cells. Proc Natl Acad Sci 14. King DM, Albertini MR, Schalch H, et al. A I evaluation of a combination of monoclonal an- U S A 1992;89:1428–32. phase i clinical trial of the immunocytokine tibody R24 and interleukin 2 in patients with 9. Reisfeld RA, Gillies SD. Antibody-interleukin-2 EMD 273063 (hu14.18-IL2) in patients with mela- metastatic melanoma. Cancer Res 1990;50: fusion proteins: a new approach to cancer ther- noma. J Clin Oncol 2004;22:4463–73. – 7490–5. apy. J Clin LabAnal 1996;10:160 6. 15. Becker JC, Pancook JD, Gillies SD, et al. T cell 5. Friedberg JW, Neuberg D, Gribben JG, et al. 10. Mujoo K, Cheresh D, Yang HM, et al. Disialo- mediated eradication of murine metastatic mel- Combination immunotherapy with rituximab ganglioside GD2 on human neuroblastoma: tar- anoma induced by targeted interleukin-2 and interleukin 2 in patients with relapsed or get antigen for monoclonal antibody mediated therapy. J Exp Med 1996;183:2361.

www.aacrjournals.org 5929 Clin Cancer Res 2009;15(18) September 15, 2009 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Cancer Therapy: Clinical

16. Becker JC, Varki N, Gillies SD, et al. An anti- anti-ganglioside antibody-Interleukin 2 immuno- cancer patients treated with repeated doses of body-interleukin-2 fusion protein overcomes tu- cytokine. In: Buolamwini JK, Adjei AA, editors. humanized monoclonal antibody A33. Cancer mor heterogeneity by induction of a cellular Methods in Molecular Medicine-Novel Antican- Res 2001;61:6851–9. immune response. Proc Natl Acad Sci U S A cer Drug Protocols. Totowa (NJ): Humana Press; 27. Cheung NK, Guo HF, Heller G, et al. Induction 1996;93:7826–31. 2003, p. 123–31. of Ab3 and Ab3′ antibody was associated with 17. Osenga KL, Hank JA, Albertini MR, et al. A 22. Carpenter BA. Enzyme-linked immunoassays. long-term survival after anti-G(D2) antibody Phase I Clinical Trial of the hu14.18-IL2 (EMD In: Rose N, deMacario EC, Fahey JL, Friedman H, therapy of stage 4 neuroblastoma. Clin Cancer 273063) as a treatment for children with refracto- Penn GM, editors. Manual of Clinical Laboratory Res 2000;2653–60. ry or recurrent neuroblastoma and melanoma: a Immunology. 4th ed.: American Society for Mi- 28. Cheung N-KV, Cheung IY, Canete A. Antibody study of the Children's Oncology Group. Clin crobiology Press; 1992, p. 2–9. response to murine anti-GD2 monoclonal anti- Cancer Res 2006;12:1750–9. 23. Albertini MR, Gan J, Jaeger P, et al. Systemic bodies: correlation with patient survival. Cancer 18. Gan J, Kendra K, Ricci M, Hank JA, Gillies SD, Interleukin-2 modulates the anti-idiotypic re- Res 2001;61:6851–9. Sondel PM. Specific ELISA systems for quantita- sponse to chimeric anti-GD2 antibody in patients 29. Foon KA, Lutzky J, Baral RN, et al. Clinical and tion of antibody-cytokine fusion proteins. Clin Di- with melanoma. J Immunother 1996;19:278–95. immune responses in advanced melanoma agn Immunol 1999;6:236–42. 24. Foon KA, Sen G, Hutchins L, et al. Antibody re- patients immunized with an anti-idiotype 19. Hank JA, Surfus J, Gan J, et al. Distinct clinical sponses in melanoma patients immunized with antibody mimicking disialoganglioside GD2. and laboratory activity of two recombinant inter- an anti-idiotype antibody mimicking disialogan- J Clin Oncol 2000;18:376–84. leukin-2 preparations. Clin Cancer Res 1999;5:281–9. glioside GD2. Clin Cancer Res 1998;4:1117–24. 30. Shusterman S, London WB, Gillies SD, et al. 20. Robertson MJ, Cochran KJ, Cameron C, Le 25. Bogner MP, Voss SD, Bechhofer R, et al. Anti-neuroblastoma activity of hu14.18-IL2 JM, Tantravahi R, Ritz J. Characterization of a Serum CD25 levels during Interleukin-2 therapy: against minimal residual disease in a Children's cell line, NKL, derived from an aggressive hu- Dose dependence and correlations with clinical Oncology Group (COG) phase II study. J Clin Oncol man natural killer cell leukemia. Exp Hematol toxicity and lymphocyte surface sCD25 expres- 2008;26, (May 20 suppl; abstr 3002). 1996;24:406–15. sion. J Immunother 1992;11:111–8. 31. Albertini MR, Hank JA, Schalch H, et al. Phase 21. Hank JA, Surfus JE, Gan J, Ostendorf A, Gillies 26. Ritter G, Cohen LS, Williams C, Jr., Richards II trial of hu14.18-IL2 (EMD 273063) for patients SD, Sondel PM. Determination of peak serum EC, Old LJ, Welt S. Serological analysis of hu- with metastatic melanoma. J Clin Oncol 2008; levels and immune response to the humanized man anti-human antibody responses in colon 26, (May 20 suppl; abstr 9039).

Clin Cancer Res 2009;15(18) September 15, 20095930 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst September 8, 2009; DOI: 10.1158/1078-0432.CCR-08-2963

Immunogenicity of the Hu14.18-IL2 Immunocytokine Molecule in Adults With Melanoma and Children With Neuroblastoma

Jacquelyn A. Hank, Jacek Gan, Hyunji Ryu, et al.

Clin Cancer Res Published OnlineFirst September 8, 2009.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-08-2963

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2009/09/04/1078-0432.CCR-08-2963. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.