Assessment of target-mediated uptake with immuno-PET: analysis of a phase I clinical trial with an anti-CD44 antibody

Treatment of cancer has improved as a result of immunotherapy with monoclonal antibodies (mAbs). Ideally, mAbs selectively target tumour cells, resulting in limited toxicity compared to classical chemotherapy.

However, lack of mAb selectivity may result in significant toxicity and/or suboptimal tumour targeting, leading to therapy failure. Therefore, it is important to confirm tumour selectivity of a novel candidate mAb to minimize toxicity and maximize efficacy, preferably in early stages of drug development. Currently, toxicity is assessed by dose escalation in traditional phase I trials, using dose-limiting toxicity and a maximum tolerated dose to establish the therapeutic dose for the next stages of drug development (phase II and III trials). For the ideal mAb, selective tumour targeting is expected, with limited target antigen-mediated specific uptake in normal tissues.

Recently, there is increasing interest in the use of imaging techniques to measure the mAb biodistribution in vivo without requiring blood or tissue samples [1]. After inert and stable radiolabelling, the radioactive mAb can be used to study the biodistribution of the non-radioactive mAb. According to this principle, positron emission tomography (PET) with ⁸⁹Zr-labelled mAbs provides a non-invasive tool for in vivo visualization and quantification of mAbs [2‐4]. Quantification of antibody accumulation in normal tissues and tumour using PET imaging can be an important non-invasive tool to evaluate the therapeutic potential of antibodies and antibody conjugates. For this purpose, target-mediated specific uptake is of interest. However, the measured PET signal comprises non-specific uptake (dependent on the tissue blood volume fraction, as well as other tissue characteristics, for example, size of endothelial fenestrae by which the antibody passes through the capillary wall) and potentially target antigen-mediated specific uptake. Differentiation between these specific and non-specific contributions to the PET signal is possible, if we assume that they are dose-dependent and dose-independent, respectively (Fig. 1). In this paper, we present an experimental approach to assess specific uptake with immuno-PET in a dose escalation study, using RG7356 as an example.

Investigational RG7356 is an anti-CD44 recombinant humanized mAb, which targets the constant region of the extracellular domain of CD44 and provides antibody-dependent cellular phagocytosis of the malignant cells by macrophages [5]. CD44 is a human cell-surface glycoprotein, which is expressed by several solid tumours as well as cancer stem cells and has a role in cell proliferation, migration and angiogenesis. This target antigen has been considered attractive for immunotherapy [6], as blocking inhibits tumour growth and metastatic potential [7, 8]. A preclinical dose escalation study with ⁸⁹Zr-labelled RG7356 confirmed tumour targeting of CD44+ tumours in xenograft-bearing mice. Since RG7356 is not cross-reactive with murine CD44, studies in mice do not provide any information regarding accessible binding sites in physiologically normal organs. Assessment of biodistribution in cynomolgus monkeys showed uptake in normal CD44+ tissues, for example, spleen and bone marrow [9]. Subsequently, a first-in-human phase I dose escalation trial was performed in patients with advanced, CD44-expressing solid tumours, showing that RG7356 was well tolerated with modest clinical efficacy [10]. PET imaging with ⁸⁹Zr-labelled RG7356 was performed to assess biodistribution and tumour uptake of RG7356 in a subgroup (13 of 65 patients) of this phase I trial. The previous publication on the main trial [10] included the background of the antigen and antibody as well as a brief summary of visual assessment of the imaging subgroup. ⁸⁹Zr-RG7356 was localized to the spleen, bone marrow and liver, while tumour accumulation of ⁸⁹Zr-RG7356 was observed with co-administration of unlabelled antibody (≥ 199 mg), suggesting that the modest efficacy was not related to poor drug delivery to the tumour per se [10]. In the current study, we separately report on the exploratory imaging sub-study with ⁸⁹Zr-labelled RG7356 by performing quantitative analysis of the imaging data.

The aim of this study was to quantify biodistribution and tumour uptake of ⁸⁹Zr-RG7356 and to evaluate whether uptake of ⁸⁹Zr-RG7356 in normal tissues was dose-dependent as indication of target-mediated specific uptake. With this example, we aim to provide a general approach for application of immuno-PET to evaluate new drug-target-combinations in first-in-human studies.

In this study, we assessed dose-dependent and dose-independent uptake of the ⁸⁹Zr-labelled anti-CD44 antibody RG7356 in normal tissues to identify specific, target-mediated uptake on immuno-PET in a dose escalation phase 1 study.

Both dose-dependent and dose-independent uptake were observed, reflecting specific as well as non-specific uptake of RG7356. For tissues without antigen expression, a linear increase in antibody concentrations can be expected for increasing antibody doses, driven by perfusion and blood volume of the tissue. However, our results suggest a mechanism that extracts antibody from the blood pool to tissues, in addition to the non-specific uptake mechanisms (Fig. 3c). Therefore, tissue-to-blood AUC ratios were used to evaluate dose-dependent uptake of RG7356 for the following tissues: liver, spleen, bone marrow, kidney, lung and brain. For the brain, a constant low tissue-to-blood AUC ratio was observed for all dose cohorts. Assuming that RG7356 does not cross the blood–brain barrier, this value is determined by the blood volume fraction of the brain. For the spleen, liver, bone marrow, kidney and lung, dose-dependent uptake of ⁸⁹Zr-RG7356 was observed, indicating target antigen-mediated specific uptake in these tissues. A very similar pattern of dose-dependent uptake in the spleen, liver and bone marrow has been reported previously in the preclinical study with ⁸⁹Zr-RG7356 in cynomolgus monkeys, indicating that such preclinical immuno-PET studies can be predictive with respect to normal tissue uptake in human [9].

Target antigen expression in these tissues is a plausible explanation for dose-dependent uptake, as protein expression of CD44 has been reported for normal bone marrow, spleen, lung, kidney and liver (bile ducts) [16, 17]. Although dose-dependent uptake in tissues was observed, a constant tissue-to-blood AUC ratio was reached at 450 mg for all tissues, indicating target antigen saturation.

In addition, dose-independent uptake of the tracer in the liver, spleen, bone marrow, kidney and lung was observed, indicating non-specific uptake. For the liver, based on a 30% blood volume fraction [18], a liver-to-blood AUC ratio of 0.3 would be expected. However, we observed a liver-to-blood AUC ratio of 0.85 ± 0.08 for the 675 mg dose cohort. The difference between the tissue-to-blood AUC ratio and blood volume fraction represents an additional accumulation mechanism in the liver, for example, the large endothelial fenestrae or antibody catabolism. Stability of ⁸⁹Zr-labelled antibodies, with minimal release of ⁸⁹Zr, has been demonstrated in many in vitro and in vivo preclinical as well as clinical studies [11, 19, 20]. There are no experimental data supporting accumulation of free ⁸⁹Zr in normal tissues, except for the observation that free ⁸⁹Zr, arising after internalisation and intracellular catabolism of the conjugate, may accumulate in bone tissue (not bone marrow) [19]. However, in our study, we did not observe ⁸⁹Zr accumulation in the bone (Fig. 2).

Although dose-dependent, as well as dose-independent, uptake in normal tissues was found in this imaging study, there were no safety concerns in the corresponding phase I dose escalation study, with treatment doses up to 1500 mg biweekly/2250 mg weekly. The overall safety profile of RG7356 was acceptable. Dose-limiting toxicities included febrile neutropenia and aseptic meningitis [10]. However, this phase I study was terminated at an early stage due to the lack of evidence of a clinical and/or pharmacodynamic (PD) dose–response relationship with RG7356.

We observed tumour uptake in all patients receiving ≥ 450 mg, with an extremely high tumour blood volume fraction estimated for the 675 mg cohort. Although this might suggest target-mediated specific tumour uptake, another study design would have been more informative on this point [21]. This requires measurement of the same tumour lesion after administration of different antibody doses to exclude differences in tumour characteristics, for example, blood volume fraction. Learning from the present study, we recently demonstrated target-mediated specific tumour uptake in a PET imaging study with an anti-HER3 mAb in which the ⁸⁹Zr-labelled antibody was administered twice (with a variable dose of unlabelled antibody) to a single patient [21]. Although biopsies taken after the immuno-PET could have provided additional confirmation with immunohistochemistry, this was not included in the study design due to the fragile patient population.

No focal tumour uptake was visualized in the lowest dose cohorts (1–200 mg). This observation cannot be explained by the level of CD44 expression or the percentage of CD44-positive tumour cells (Table 1). A probable explanation why tumour visualization is hampered for the lowest dose cohorts is that dose-dependent uptake in normal tissues leads to lower visual tumour contrast, assuming similar binding constants and accessibilities of the target antigen in both normal tissues and tumour. However, in the higher dose cohorts, differences between binding constant became apparent where the dose-dependent tracer uptake in normal tissues does not significantly contribute to the imaging signal anymore (Fig. 1); target antigen-mediated tracer uptake will result in sufficient visual contrast to allow identification and quantification of tumour targeting.

In this study, PET imaging with the novel anti-CD44 monoclonal antibody RG7356 confirmed tumour uptake for patients receiving ≥ 450 mg. However, dose-dependent uptake of RG7356 in normal tissues indicates target antigen expression, limiting the use of RG7356 for targeting toxic payloads to the tumour like in antibody–drug conjugate (ADC) approaches.

This exploratory imaging study demonstrates how immuno-PET with a ⁸⁹Zr-labelled mAb can be used as a general method during phase I dose escalation studies to evaluate the therapeutic potential of an antibody. Evaluation of dose-dependent and dose-independent normal tissue uptake with immuno-PET reflects specific and non-specific uptake. Antibody quantification obtained by molecular imaging provides an additional, non-invasive method to study in vivo mAb biodistribution, besides traditional PK obtained by blood sampling. Especially for a candidate mAb with a potential future as ADC, the resulting information (prevention of potential toxicity/additional development costs) may justify the patient burden and cost for additional scans in a limited number of patients in a phase I setting.

For a mAb which continues in further stages of drug development, in vivo measurements of antibody concentrations in tissue and tumour can be of value for PK/PD modelling/dose optimization and response prediction to guide individualized treatment.

This study demonstrates how immuno-PET in early clinical drug development provides a non-invasive technique to quantify dose-dependent uptake in normal tissues, indicative of specific, target-mediated uptake.