Rituximab blocks binding of radiolabeled anti-CD20 antibodies (Ab) but not radiolabeled anti-CD45 Ab

The B-cell antigen CD20 is the target of 3 US Food and Drug Administration (FDA)–approved monoclonal antibodies (MoAbs), 1 unconjugated (rituximab), 2 radiolabeled (¹³¹iodine-tositumomab and ⁹⁰yttrium-ibritumomab tiuxetan), and at least 7 additional anti-CD20 antibodies that are in development (ofatumomab, ocrelizumab, PRO131921, TRU-015, veltuzumab, AME-133, and GA-101).^1-7  Unlike traditional chemotherapeutics, rituximab, like many other MoAbs, exhibits a long half-life, with mean serum concentrations of 82 μg/mL and 14 μg/mL at 3 and 6 months, respectively, after a standard 4-week course of weekly therapy at 375 mg/m².⁸  This long-term persistence of circulating rituximab is further increased by the frequent use of protracted maintenance regimens that may lead to high rituximab levels at the time of relapse.^9-11  It is in this situation that ¹³¹iodine-tositumomab and ⁹⁰yttrium-ibritumomab tiuxetan are approved and often used.^12,13  It is also anticipated that in this setting the multiple second-generation anti-CD20 MoAbs in development will also be used, even though little is known about the effect of circulating rituximab competing for the same target.⁷  Herein we evaluated the impact of prior rituximab on the binding of secondary anti-CD20 and anti-CD45 MoAbs to elucidate if antigen competition occurred, if it was clinically relevant, and if it could be circumvented by an alternate target.

The chimeric anti-CD20 MoAb C2B8 (rituximab; IgG₁) was purchased from Genentech (South San Francisco, CA). The anti-CD20 MoAb tositumomab (IgG_2a) was produced by GlaxoSmithKline (GSK; Philadelphia, PA). BC8 (IgG₁) and HB8181 (IgG_2a) were produced from the respective hybridomas using a hollow-fiber bioreactor system in the Biological Production Facility at the Fred Hutchinson Cancer Research Center (FHCRC; Seattle, WA). 1D8 (IgG₁) was produced in ascites generated by pristane-primed Balb/c mice and purified by protein G immunoabsorption column chromatography and used as an isotype-matched nonbinding control. For flow cytometry experiments, tositumomab and BC8 were conjugated to fluorescein isothiocyanate using the method provided (Sigma-Aldrich, St Louis, MO). Tositumomab and BC8 were iodinated with Na¹³¹I or Na¹²⁵I (Perkin Elmer, Boston, MA) by the chloramine T method as previously described.¹⁴ 

Ramos and Granta 519 cell lines were purchased from ATCC (Bethesda, MD). FL-18 was a gift from Dr Ron Levy (Stanford University, Palo Alto, CA). All cell lines were kept in log growth phase in RPMI-1640 supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 U/mL streptomycin (Invitrogen, Carlsbad, CA)_.

Human lymphoma cells and serum samples were provided by patients following human subject review board approval and informed consent obtained in accordance with the Declaration of Helsinki. The malignant lymphocytes were isolated from whole blood using Ficoll-Hypaque separation and stored at −80°C in 5% dimethyl sulfoxide before use. Serum rituximab concentrations were determined as previously published.⁵ 

Cells (10 × 101⁶/mL in 2% FBS–phosphate-buffered saline [PBS]) were incubated with the presumptive blocking (eg, rituximab) or control (HB8181) antibody for 30 minutes at 4°C and washed with 2% FBS-PBS. This was followed by a second incubation with the desired binding (BC8 fluorescein isothiocyanate conjugation [BC8-FITC] or tositumomab-FITC) or control (horse anti–mouse-FITC [for HB8181antibody]) antibody for 30 minutes at 4°C. After a second washing, the cells were put in PBS and immediately analyzed in a FACSCanto flow cytometer (Becton Dickinson, San Jose, CA). Fifty microliters containing peripheral blood lymphocytes were collected from rituximab-naive lymphoma patients with documented circulating tumor (20 × 10⁶ cells/mL in 2% FBS-PBS) were similarly incubated for 30 minutes. at 4°C with 50 μL serum from rituximab pretreated patients and washed with 2% FBS-PBS. Next, the cells were then incubated for 30 minutes. at 4°C with BC8-FITC or tositumomab-FITC. Ten microliters phycoerythrin (PE)-Cy-5–conjugated mouse anti–human CD-22 (BD Pharmingen, San Jose, CA) was included to identify the B cells from other CD45-bearing lymphocytes. The cells from each experiment were washed, put in PBS, and immediately analyzed in a FACSCanto flow cytometer (Becton Dickinson).

For all mouse studies, female BALB/c nu/nu mice, aged 6 to 8 weeks, were purchased from Harlan Sprague-Dawley (Indianapolis, IN) and were housed and cared for according to protocols approved by the FHCRC Institutional Animal Care and Use Committee. Ramos (10⁷), Granta (1.2 × 10⁷), or FL-18 (10⁷) cells were injected subcutaneously in the right flank, and palpable xenografts developed within approximately 10 days. Groups of 5 or more mice were injected intraperitoneally with 400 μg rituximab or an isotype-matched control (1D8) at 24 and 6 hours before the radioiodinated antibody administration. Next, 200 μg each ¹²⁵I-BC8 and ¹³¹I-tositumomab labeled with 0.37 MBq (10 μCi) of radioisotope were coinjected via the tail vein at 0 hours along with 400 μg nonspecific IgG_2a MoAb HB8181 and IgG₁ MoAb 1D8, respectively, to block nonspecific binding of the conjugates to Fc receptors in the liver, spleen, and bone marrow as previously published.^15,16  The mice were bled retroorbitally and humanely killed, and normal organs (lungs, liver, spleen, kidneys, stomach, large intestine, and small intestine) and tumors were harvested and weighed at 24 and 48 hours after the ¹²⁵I-BC8 and ¹³¹I-tositumomab were injected. Residual radioactivity in the harvested tissues was determined via gamma counter measurements. Using an aliquot of the injectate, the percentage of injected doses per gram (% ID/g) of ¹²⁵I and ¹³¹I in blood, tumor, and normal organs were calculated, and corrections for radioactive decay and crossover were made. Tumor-to-normal organ ratios of absorbed radioactivity were calculated from these data.

B-NHL xenografts were established as described. BC8 and tositumomab we labeled with fluoromer-680 as described by the manufacturer (Xenogen, Hopkinton, MA). When tumors were measurable (approximately 100 mm³), groups of at least 3 mice each were injected intraperitoneally with 400 μg rituximab or 1D8 24 hours before a tail-vein injection of 200 μg BC8-680 or tositumomab-680. Mice were then anesthetized with isofluorane, and images were captured using an IVIS 100 Imaging System (Xenogen). Imaging and quantification were done in accordance with the IVIS 100 Imaging System manual. Briefly, 2 images were captured, one captured with the emission set to Cy5, and the second image (background) captured using the emission set to Cy5 background. The Cy5 setting excites at 615 to 665 nm, and emission is between 695 to 770 nm. A region of interest was placed on the negative control mouse in the background image to allow determination of the total fluorescence for each image. The Cy5 background total was divided by the Cy5 total to calculate “K.” A new image was then generated using the formula a − b × K, where “a” equals the Cy5 image and “b” equals the Cy5 background image.

B-NHL xenografts were established as described. When tumors were measurable (approximately 100 mm³), groups of at least 9 mice each were injected intraperitoneally with 400 μg rituximab or 1D8 24 hours before a tail-vein injection of 200 μg ¹³¹I-BC8 or ¹³¹I-tositumomab each labeled with 14.8 MBq (400 μCi) ¹³¹I. Tumor size and body weight were measured on multiple days after injections. Mice were humanely killed when they experienced weight loss greater than 20% or tumor growth greater than 12 mm × 12 mm × 6 mm per animal health guidelines.

Continuous variables with normal distributions were compared using the Student t test. Simple proportions were evaluated using the chi-square or Fisher exact test as appropriate.

All studies were approved to collect patient samples for research purposes by the FHCRC International Review Board.

We first evaluated the effect of rituximab (anti-CD20) pretreatment on the ability of tositumomab (anti-CD20) and BC8 (anti-CD45) to bind their respective targets. Malignant B-NHL cells were preincubated with increasing concentrations of rituximab (0-2000 μg/mL), washed, and then incubated with fluoresceinated tositumomab (anti-CD20) or fluoresceinated BC8 (anti-CD45) and evaluated by flow cytometry for antibody binding. Fluoresceinated isotype–matched MoAb, HB8181, was used as a nonbinding negative control antibody. Prior rituximab exposure reduced binding of tositumomab to B-NHL cells (Ramos) by 86% at concentrations as low as 5 μg/mL (P < .001) and fully blocked ligation at concentrations of 50 μg/mL or greater (P < .001; Figure 1A). Increasing concentrations of rituximab had no impact on CD45 binding (P = .89; Figure 1B). Similar results were observed using the B-NHL lines FL-18 and Granta (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article).

To further evaluate the clinical relevance of these findings, we assessed whether this phenomenon could be replicated in patient-derived specimens. Serum was obtained from patients with B-NHL 4 weeks or more after the last rituximab therapy. The median rituximab concentration in these specimens was measured at 23.4 μg/mL (range, 14.6-82.2 μg/mL). Dilutions (1:1) of serum from 3 patients treated with rituximab were preincubated with malignant B-NHL cells obtained from a rituximab-naive patient, washed, incubated with fluoresceinated tositumomab (Figure S2) or fluoresceinated BC8 (Figure S3), then evaluated for the ability of the fluoresceinated MoAbs to bind to the CD20 and CD45 antigens, respectively, as assessed by flow cytometry. Figure 2 illustrates a direct comparison of the mean fluorescences of tositumomab and BC8 binding to patient-derived cells in the presence of serum from patients treated with rituximab. These data indicate that serum from patients with a remote exposure to rituximab fully blocks the binding of tositumomab to CD20 at concentrations at least as low as 7.3 μg/mL, but has no effect on CD45 ligation.

We then determined if the observed in vitro findings truly affected the potential to deliver radioimmunotherapy (RIT) to tumor sites in vivo. B-NHL–bearing mice were injected either with rituximab or nonbinding isotype-matched control antibody (1D8), followed by coinjection with ¹³¹I-tositumomab and ¹²⁵I-BC8. The retained radioactivity measured as a percentage injected dose per gram of normal tissue and tumor tissue was determined at 24 and 48 hours after RIT. Tumor sites in mice with human B-NHL (Ramos) xenografts received comparable radioisotope doses when treated with ¹²⁵I-BC8 in the presence or absence of prior rituximab (Figure 3A). In contrast, rituximab pretreatment reduced the tumor uptake of ¹³¹I-tositumomab by 55% compared with mice treated with a nonbinding control MoAb (P < .001; Figure 3B). Importantly, the estimated ratio of radiation delivered to tumor sites compared with nonhematopoietic tissues was greater than 1 with both ¹²⁵I-BC8 (Figure 3C) and ¹³¹I-tositumomab (Figure 3D) when no prior rituximab was used. However, the impact of rituximab pretreatment reduced the predicted tumor-to–nonhematopoietic target ratios of delivered radiation to 1:1 or less when CD20 was targeted, but had no appreciable effect on ratios when CD45 was targeted using the identical tumor models. For example, at 24 hours the mean predicted tumor-kidney ratio of delivered radiation using ¹²⁵I-BC8 was 2.7:1 after prior rituximab and 2.6:1 without prior rituximab (P = .7). In contrast, the predicted tumor-kidney ratio using ¹³¹I-tositumomab in the absence of rituximab was 2.5:1, but fell to 1.1:1 when mice were pretreated with rituximab (P < .001). In the absence of rituximab, tumor retention of ¹³¹I-tositumomab/g of tissue was 30% more than the lungs (the critical nonhematopoietic organ with the highest radiation exposure), whereas after rituximab pretreatment, ¹³¹I-tostumomab/g uptake in the lungs was double that of tumor sites (P = .007). These results were replicated with the Granta and FL-18 B-NHL lines (Figure S4).

Real-time imaging studies using fluoresceinated tositumomab and fluoresceinated BC8 corroborated the previously quantified biodistribution results (Figure 4). Mice pretreated with rituximab followed by fluoresceinated tositumomab demonstrated inferior tumor uptake and retention compared with those pretreated with a nonspecific binding control antibody (1D8). Rituximab pretreatment had no apparent impact on CD45 targeting by BC8.

Although the biodistribution studies indicate that circulating rituximab abrogates tumor targeting of CD20-directed therapy, it is unknown if this effect translates into reduced clinical efficacy. We thus determined if the inferior tumor targeting of the anti-CD20 radiolabeled MoAb due to rituximab pretreatment translated into inferior tumor control in mice bearing palpable human B-NHL xenografts. Mice received therapeutic doses (14.8 MBq [400 μCi]) of ¹³¹I-BC8 or ¹³¹I-tositumomab after being pretreated with either rituximab or an isotype-matched nonbinding control MoAb. Without rituximab pretreatment, tumor progression in both the ¹³¹I-BC8 or ¹³¹I-tositumomab groups was similarly delayed compared with untreated controls (Figure 5A). In contrast, after rituximab therapy, tumor control in the ¹³¹I-tositumomab group paralleled the control groups and was impaired relative to the ¹³¹I-BC8 cohort (Figure 5B).

In these series of experiments, we demonstrate that the sequential use of anti-CD20 MoAbs results in almost complete inhibition of binding of the second anti-CD20 MoAb, but does not affect targeting of alternate tumor antigens such as CD45. This effect was observed in vitro using both cell lines and patient-derived specimens and in vivo in human lymphoma xenograft models. Importantly, this phenonemenon directly translated into inferior tumor control when competing MoAbs blocked binding of the radioimmunoconjugates to target sites. Such findings should have an immediate clinical impact, as patients currently receiving the FDA-approved radiolabeled antibodies ¹³¹I-tositumomab and ⁹⁰Y-ibritumomab tiuxetan are often treated after progression on rituximab maintenance regimens or as consolidation after rituximab-chemotherapy combinations and undoubtedly have significantly elevated serum rituximab concentrations at the time of RIT administration.^9-11  In the clinic, however, it is unlikely that circulating rituximab completely abrogates the antitumor effect of RIT, because studies have shown efficacy in the setting of rituximab refractory disease where many patients likely had measurable serum rituximab just before RIT.^12,13,17  Furthermore, initial studies of anti-CD20 RIT in rituximab-naive patients with measurable B-NHL demonstrated that a limited dose of unlabeled antibody, delivered just before or concurrent with the radiolabeled dose, helped improve delivery of the conjugate to lymph nodes by blocking uptake in the spleen.^18-20  Nevertheless, Witzig and colleagues were able to demonstrate inferior response rates in patients with no detectable B cells (a potential surrogate for recent rituximab therapy; 47% vs 88%, P = .003) and a trend toward inferior responses in patients with measurable serum rituximab before RIT (66% vs 93%, P = .078).¹³  Taken together, these data suggest that there may be an “optimal” cold dose for every given patient depending on tumor burden, CD20 antigen density, splenomegaly, and serum rituximab concentrations. Ultimately, these hypotheses will need to be confirmed in the clinic with dosimetry studies paralleling our mouse models.

In our model, targeting an alternate antigen such as CD45 could circumvent this effect, but such strategies are not readily available in the clinic, leaving only a lengthy washout period from the last rituximab as an option. Furthermore, the panhematopoietic expression of CD45 would likely require that stem cell support be used after its administration.²¹  This same feature of CD45, however, could also be advantageous to enhance the crossfire targeting of minimal disease sites within hematopoietic tissues. Our future clinical trials plan to target CD45 with high-dose RIT in patients with lymphoid malignancies by building our group's experience using ¹³¹I-BC8 for the treatment of acute myeloid leukemia (AML).²² 

Our findings may also affect unconjugated antibodies. Due to the past success of CD20-directed therapies, extensive development of novel anti-CD20 MoAbs is ongoing, but these agents may also face the same limitations as we observed in our studies.^3,7  MoAb design using epitope-mapping strategies initially intended to optimize antibody function may be able to alleviate some of these concerns, though functional testing of these agents in clinically representative settings of circulating rituxumab remains critical.²³ 

Our results support the hypothesis that excess rituximab can impair the efficacy of CD20-directed RIT. Moreover, these findings also broadly illustrate that one can no longer develop an antigen-targeted therapy in isolation without considering the potential impact of competing MoAbs for the same target. Future preclinical and human evaluations of such drugs should reflect the likely clinical scenario whereby one or more agents may be vying for the identical epitope within a given patient.

We thank Kelli Boyle, Aimee Kenoyer, and Barbara Pender for laboratory expertise.

This work was supported by the Lymphoma Research Foundation Mantle Cell Lymphoma Research Initiative; National Institutes of Health grants P01CA44991, R01CA76287, R01CA109663, K23CA85479, and K08CA095448; SCOR grant 7040 from the Leukemia & Lymphoma Society; and a donation from Frank and Betty Vandermeer.

National Institutes of Health

Contribution: A.G. designed the study, analyzed the data, and wrote and edited the manuscript; O.P. provided guidance and technical support for the study design and edited the manuscript; S.W. assisted with study design, performed all experiments, analyzed the data, and edited the manuscript; D.M. provided rituximab concentrations and edited the manuscript; and J.P. provided technical support and guidance for the study design and edited the manuscript.

Conflict-of-interest disclosure: D.M. consults for Genentech, Biogen IDEC (which has an interest in rituximab), and GSK (maker of tositumomab); A.G. receives research support from GSK and Biogen-IDEC. The remaining authors declare no competing financial interests.

Correspondence: Ajay K. Gopal, University of Washington/Seattle Cancer Care Alliance, 825 Eastlake Ave E, G3-200, Seattle, WA 98195; e-mail: agopal@u.washington.edu.