Non-specific irreversible ⁸⁹Zr-mAb uptake in tumours: evidence from biopsy-proven target-negative tumours using ⁸⁹Zr-immuno-PET

Positron emission tomography (PET) with zirconium-89-labelled monoclonal antibodies (⁸⁹Zr-mAbs), known as ⁸⁹Zr-immuno-PET, is a powerful tool in precision medicine and drug development. It enables the visualization and quantification of ⁸⁹Zr-mAbs tumour uptake in vivo [1, 2]. These ⁸⁹Zr-immuno-PET studies are generally proof of concept and phase 1 studies, focusing on tracer pharmacokinetics and dose optimization [1]. An important step in the interpretation of PET imaging is understanding the molecular processes underlying the ⁸⁹Zr-mAb uptake on PET.

Distribution of mAbs into tissue depends on the structure of the capillary endothelium [3] and can occur via different processes. Firstly, mAbs distribute across the vascular endothelium into the interstitial space, and leave the interstitial space with lymph fluid, both via convective transport [3, 4] (see Fig. 1). Secondly, transport may take place via diffusion. Diffusion is limited in healthy tissue due to the relatively large size of mAbs [4], but it may play a more significant role in tumour tissue. Due to rapid cell division, the endothelial cells in tumour vasculature are often disorganized leading to leakiness [3]. The increased permeability allows diffusion of mAbs across the tumour endothelium and, together with the poor lymph drainage in tumour tissue, leads to accumulation inside the interstitial space (i.e. the enhanced permeability and retention (EPR) effect) [5]. Thirdly, mAbs may enter endothelial cells via receptor-mediated endocytosis where they can bind to the neonatal Fc-receptor (FcRn) [4] (see Fig. 1). Subsequently, FcRn-bound mAbs are brought back into the blood circulation or interstitial space, while FcRn-unbound mAbs are catabolized by lysosomes [4]. This catabolism is the main elimination pathway for mAbs [4, 6]. Literature is unequivocal about the role of FcRn in tumour tissue. FcRn has been shown to be present at very low levels in most human tumour cell lines [7] and was described to be downregulated and dysregulated in various cancer types [8, 9]. Both dysregulation and upregulation of FcRn were described to be associated with tumour growth [8, 10]. Moreover, mAbs may bind to Fcγ receptors on immune cells [11]. Lastly, mAbs bind to the specific target receptor present within the tissue (see Fig. 1).

The different distribution processes may contribute to the ⁸⁹Zr-mAb uptake measured on PET [12]. Firstly, the distribution of mAbs into the interstitial space (via convection, diffusion or FcRn binding) and subsequent re-circulation via the lymphatic system results in non-specific reversible ⁸⁹Zr-mAb uptake. Secondly, FcRn-unbound mAbs are catabolized within endothelial cells leading to residualization and off-target accumulation of ⁸⁹Zr [13], referred to as non-specific irreversible uptake [12]. Binding of mAbs to Fcγ receptors may also contribute to non-specific irreversible uptake, as studied previously [11]. Thirdly, actual target-specific binding takes place, which leads to target-specific irreversible ⁸⁹Zr-mab uptake for irreversibly binding or internalizing tracers. As the non-specific irreversible ⁸⁹Zr-mAb uptake is inversely related to FcRn binding, the presence of this uptake mechanism may provide insight into the role of FcRn in tumour tissue.

The reason for our interest in non-specific irreversible uptake in tumours is twofold. Firstly, uptake on PET is commonly quantified using the standardized uptake value (SUV) [14]. For ⁸⁹Zr-immuno-PET, there is an increased interest in reporting the tumour-to-plasma ratio (TPR), which provides more valid results in quantifying ⁸⁹Zr-mAb uptake when the mass dose is varied as compared to SUV [15]. Target-specific uptake is of main interest, but both measures quantify the total uptake, in which non-specific irreversible and reversible uptake may be present as well. Secondly, a more comprehensive approach to describe the distribution of mAbs is with the use of physiologically based pharmacokinetic (PBPK) models. Some models did incorporate FcRn binding within the tumour compartment [16‐18], while in other models, the distribution of mAbs into the tumour interstitial space is determined by the EPR effect [19‐23]. PBPK models potentially improve our understanding of ⁸⁹Zr-immuno-PET studies; however, an in-depth understanding the role of FcRn in tumour tissue is lacking but desired [8, 10].

The different contributions of ⁸⁹Zr-mAb uptake to the PET signal have been separated previously using Patlak linearization [12, 24‐32]. This method enables separation of PET signal resulting from reversible and irreversible uptake processes [33]. It is assumed that uptake of ⁸⁹Zr-mAbs can be separated in these two components, where irreversible uptake includes both target-specific and non-specific uptake. Patlak linearization was applied to ⁸⁹Zr-immuno-PET data of healthy non-target-expressing organ tissue to establish baseline uptake values representing the non-specific irreversible uptake. As shown by Jauw et al., comparison of uptake in target-expressing organs with these baseline values allows quantification of actual target-specific uptake [12]. Additionally, when applying the co-infusion of additional doses of unlabelled mAbs, the competition in binding between ⁸⁹Zr-labelled and unlabelled mAbs has shown to lead to a decreased uptake of ⁸⁹Zr-labelled mAb. This enables assessment of target engagement in tumour tissue [24].

A similar approach of measuring uptake of ⁸⁹Zr-mAbs in non-target-expressing tissue was applied here to tumour tissue to study the role of non-specific irreversible uptake. The current study aimed to investigate the presence of non-specific irreversible uptake in tumour tissue using Patlak linearization on ⁸⁹Zr-immuno-PET data of biopsy-proven target-negative tumours. Such a dataset is quite unique and is based on incidental findings, because applying unspecific antibody in a clinical setting is generally not ethically justified. An estimation of the contribution of non-specific irreversible uptake in tumour tissue will enable the quantification of actual target-specific ⁸⁹Zr-mAb uptake.

For five patients who received ⁸⁹Zr-anti-CEA-IL2v, one PET scan was missing, and for three patients, one blood sample was missing. Therefore, for obtaining the Patlak K_i values, three imaging time points were included in five cases, and two time points in twelve cases.

Of the patients who received ⁸⁹Zr-anti-CD20, two tumours were classified as CD20-negative and two were CD20-positive according to IHC. Of the patients who received ⁸⁹Zr-anti-CEA-IL2v, four tumours were classified as CEA-negative and nine tumours were CEA-positive.

Individual Patlak plots are presented in Fig. 2. From these Patlak plots, the K_i values (i.e. the slopes) were obtained. K_i values of both CD20-negative (0.43 µL/g/h and 0.92 µL/g/h) and CEA-negative tumours (mdn = 1.97 µL/g/h, interquartile range (IQR) = 1.50–2.39) were higher than zero (see Fig. 3). The median K_i values of CD20-positive tumours (1.87 µL/g/h and 1.90 µL/g/h) and CEA-positive tumours (mdn = 2.77 µL/g/h, IQR = 2.11–3.65) were higher than those of target-negative tumours (see Fig. 3). A complete overview of the data is shown in Additional file 1: Tables S1 and S2.

The presence of non-specific irreversible uptake in tumour tissue was studied by applying Patlak linearization on ⁸⁹Zr-immuno-PET data of biopsy-proven target-negative tumours. Biopsy-proven target-negative tumours showed positive K_i values suggesting the presence of non-specific uptake of ⁸⁹Zr-mAbs measured in vivo using ⁸⁹Zr-immuno-PET. Additionally, as expected, the irreversible uptake in target-positive tumours was higher than in target-negative tumours. Moreover, irreversible uptake of ⁸⁹Zr-CEA-IL2v was higher than for ⁸⁹Zr-anti-CD20.

Despite the absence of target expression in the target-negative tumours, there was still irreversible uptake found, i.e. a K_i value above zero. This non-specific irreversible uptake in tumours is most likely due to the degradation of ⁸⁹Zr-mAbs that are not protected by FcRn. Also, binding of mAbs to Fcγ receptors on immune cells may play a role in irreversible ⁸⁹Zr-mAb uptake [11]. Previous literature, however, describes that extravasation of mAbs into the tumour interstitial space occurs primarily via diffusion [20]. The increased permeability and poor lymphatic drainage in tumour tissue lead to the EPR effect, resulting in accumulation of mAbs within the interstitial space [5, 40]. Nonetheless, this diffusion process would result in reversible uptake, as it does not include degradation of mAbs. The current study suggests the presence of non-specific irreversible uptake, which contradicts the conception that extravasation occurs solely via diffusion. Both protection by FcRn and the EPR effect are thought to contribute to ⁸⁹Zr-mAbs extravasation into tumour tissue; however, the ratio between them may differ considering the high intra-tumour variability in characteristics such as vascularization, morphology, level of necrosis, size, acidity and immunogenic potential [41‐43].

PET imaging has been used previously to provide a measurement for non-specific ⁸⁹Zr-mAb uptake in healthy tissue. Jauw et al. analysed ⁸⁹Zr-immuno-PET data of four different ⁸⁹Zr-mAbs in organs (i.e. lungs, liver, kidney and spleen) in which target expression was absent. K_i values ranged from 0.2 µL/g/h for lung to 1.1 µL/g/h for liver, representing the rate of catabolism of ⁸⁹Zr-mAbs not protected by FcRn [12]. These values are comparable to the values for non-specific irreversible uptake in CD20-negative tumours found in this study. A K_i value of zero would be expected in tissue where neither target-specific nor non-specific uptake is present. For example in brain tissue, where the blood–brain barrier hampers ⁸⁹Zr-mAb distribution, a Patlak K_i value of 0.0 to 0.1 µL/g/h was recently found [28], which suggests that the values found in the current study are not merely variability in data.

Both the target-positive and target-negative tumours of the ⁸⁹Zr-CEA-IL2v data showed higher irreversible ⁸⁹Zr-mAb uptake than the ⁸⁹Zr-anti-CD20 data. Additionally, the K_i value in CEA-negative tumours was also slightly higher than previously found in healthy organs without target expression [12]. This is likely the result of binding of ⁸⁹Zr-CEA-IL2v to IL-2Rβγ receptors on immune cells [38]. CEA-IL2v has shown to bind to CD8 + T and natural killer cells and leads to the expansion of these cells in the blood, lymphoid tissue and tumours [38]. Binding to these tumour-infiltrating immune cells would result in increased K_i values when these ⁸⁹Zr-CEA-IL2v-bound immune cells remain inside the tumour tissue during the course of the study. On the other hand, in a previous study, the concentrations of CEA-IL2v in blood and the uptake in tumour were predicted using a mathematical model [44]. In this model, only the unbound CEA-IL2v was used as input for tumour uptake, and not the complex of CEA-IL2v with IL-2Rβγ expressed on immune cells [44]. Nonetheless, the binding of ⁸⁹Zr-CEA-IL2v to immune cells would explain the difference in irreversible binding with the ⁸⁹Zr-anti-CD20 study.

In conventional PET pharmacokinetic modelling, dynamic PET imaging data are obtained and fitted to different one- and two-tissue models [45]. The best model fit indicates the pharmacokinetic characteristics of the tracer. For a tracer that is best described by a two-tissue irreversible model, Patlak linearization is a simplified approach to obtain the pharmacokinetic constant for irreversible uptake, K_i. In ⁸⁹Zr-immuno-PET, static PET images are acquired at multiple days p.i. because of the long half-life of mAbs. This leads to a limited amount of data for which conventional modelling is not possible. Instead, Patlak linearization is applied to ⁸⁹Zr-immuno-PET studies, assuming a two-tissue irreversible model. PBPK modelling is a different type of modelling that considers multiple tissues represented by different compartment. Contrary to conventional pharmacokinetic modelling, PBPK modelling is not data-driven, but allows a priori predictions of tracer concentrations based on physiological information [16]. This allows evaluation of the separate processes that are involved in mAb pharmacokinetics. All quantification approaches assume that the processes between compartments occur at a constant rate during the course of the study. However, migration of immune cells might potentially influence quantification of PET imaging, which should be kept in mind for the ⁸⁹Zr-CEA-IL2v study, as discussed previously.

There are some limitations to the study. Ideally, Patlak linearization would be applied to three imaging time points with one time point at 24 h p.i, as this improves the accuracy and precision of the method [35]. It is important to note that in this study, there were only two imaging time points available in twelve out of seventeen cases. Additionally, the ⁸⁹Zr-anti-CD20 data did not include an imaging time point around 24 h p.i. From simulation studies, it can be seen that the K_i value is always slightly underestimated, even more so when there is no 24 h p.i. time point included. Therefore, an underestimation of the K_i value of about 8% may be expected [35]. Additionally, the current study evaluates ⁸⁹Zr-mAb uptake in target-negative tumours, which is a unique dataset, but consequently also a limited amount of data. In future studies, a K_i value for non-specific irreversible uptake in tumours may be determined more reliably, and its translation to other ⁸⁹Zr-mAbs may be evaluated.

⁸⁹Zr-mAb tumour uptake on PET was interpreted in the context of the target presence measured with IHC. There are aspects to consider when relating PET uptake to IHC. IHC is currently the standard procedure in the clinic for measuring target expression. However, the biopsy obtained for IHC assessment is a small part of the tumour and sampling errors may occur, while tumour uptake on PET is obtained from the whole tumour. Heterogeneity in target expression within the whole tumour will therefore not be detected with IHC. However, CEA-negative tumours are thought to be highly unlikely to have expressed any CEA [37, 46]. Additionally, one of the CD20-negative tumours was biopsied twice, both showing no target expression.

The current study was performed using data of patients with two different types of cancer. DLBCL develops from the B cells in the lymphatic system and is characterized by CD20 expression, which is a transmembrane protein [34]. The anti-CD20 mAbs target the malignant B cells and trigger cell death. For the majority of solid malignancies, amongst which also colorectal, non-small cell lung and gastric cancer, CEA is overexpressed [37, 38]. The CEA-IL2v bispecific mAbs induce the local immune response by binding to the CEA in tumour tissue while also binding to immune cells. The difference between these two mAbs is in their variable region; however, their constant region is similar. Since the constant region binds to FcRn [4], no differences in uptake of target-negative tumours may be expected.

During the two clinical studies, different mass doses were applied which is known to influence target-specific ⁸⁹Zr-mAb uptake. Administration of additional unlabelled mAbs results in competition between ⁸⁹Zr-labelled and unlabelled mAbs for target binding [24]. Because of this competition and the limited number of receptors, there is less target-specific uptake resulting in lower K_i values. Three different mass doses of ⁸⁹Zr-CEA-IL2v were administered to the patients: 6, 20 or 30 mg. The mass dose of 30 mg indeed resulted in lower K_i values in target-positive tumours as compared to the 20 mg mass dose. However, the mass dose of 6 mg resulted in lower K_i values compared to the higher mass doses. The reason for this is unknown, a possible explanation might be the spleen acting as a IL-2Rβγ sink, as has been previously hypothesized by [37]. This results in low plasma activity concentrations and less supply for the tumour. Nonetheless, non-specific uptake is not influenced by differences in mass dose, because it results from degradation of mAbs not protected by FcRn, a process that is not saturable with these relatively low mass doses [4]. Since the irreversible uptake in target-negative tumours only results from this non-specific mAb degradation, the K_i values in target-negative tumours are not influenced by differences in mass dose.

The current study shows that non-specific irreversible uptake is present in the absence of specific targets. This contribution may interfere with the quantification of actual target-specific irreversible uptake, which is of interest. If a baseline value for non-specific irreversible tumour uptake could be established in future, it may be used as a correction when quantifying ⁸⁹Zr-mAb uptake in tumours, similar to what was previously proposed by Jauw et al. to use baseline K_i values in healthy tissue to measure target engagement [12]. It is, however, more common to report SUV or TPR. Since the K_i value represents the rate of irreversible uptake, a positive K_i value means that a TPR always increases over time. The Patlak linearization described in this manuscript can be used to estimate the contribution of non-specific irreversible uptake to a SUV or TPR (see Additional file 1: Figure S2). For ⁸⁹Zr-anti-CD20 uptake at the latest imaging time point, the total uptake would consist of 71% irreversible uptake, of which the larger part is CD20-specific uptake (46%) and the smaller part is non-specific uptake (25%). For the ⁸⁹CEA-IL2v study, the total uptake at the latest imaging time point would be dominated by irreversible uptake (94%). This is separated into CEA-specific uptake (27% of total uptake) and non-specific uptake (67% of total uptake), which most likely includes irreversible binding to IL-2βγ receptors [38]. Consequently, both specific and non-specific irreversible uptakes are highest at the latest imaging time point. PET scans should be scheduled as late as possible, while still acquiring reliable PET scans considering the decreasing number of counts over time due to radioactive decay. Because of this trade-off, PET scans are typically acquired up to seven days p.i. [2].

The aim of this study was to investigate the presence of non-specific irreversible ⁸⁹Zr-mAbs uptake in tumours, as it may interfere with quantification of target-specific uptake. Patlak linearization was retrospectively applied to ⁸⁹Zr-immuno-PET data of biopsy-proven target-negative tumours to measure the irreversible uptake (expressed in a K_i value). Biopsy-proven target-negative tumours showed Patlak K_i values above zero, which suggests the presence of non-specific irreversible uptake. The Patlak K_i values in the current study are consistent with non-specific irreversible uptake in healthy organs. Consequently, a tumour-to-plasma ratio always increases over time for ⁸⁹Zr-mAbs uptake in tumours, also in absence of the target.