Tumor Uptake of Anti-CD20 Fabs Depends on Tumor Perfusion

DISCUSSION

Tumor heterogeneity is one of the main challenges during clinical antibody diagnosis and therapy. Despite prevalent target expression, subgroups of patients do not properly respond to antibody treatment, and the uneven distribution of most antibodies in tumor tissue, including metastases, leaves whole tumor areas or lesions untargeted. Numerous clinical and experimental studies have shown that tumor perfusion has a strong impact on both radiotracer uptake and drug delivery. Early detection of changes in tumor permeability and perfusion during treatment would therefore be of great benefit for response prediction. To study the underlying mechanisms of tumor heterogeneity and the role of tumor permeability and perfusion, a well-defined preclinical animal model is highly desired. To this end, we characterized two NHL xenograft models, Granta-519 and SUDHL-4, with apparently similar tumor biology regarding morphology, metabolic activity, and CD20 target expression but overt differences in perfusion and hemodynamic properties.

Similar autoradiography distribution patterns were found for the small-molecule nontargeted radioactive tracer ¹⁸F-FDG in both xenograft models. These result agree with a previous study that investigated SUDHL-4 tumors using ¹⁸F-FDG and the well-established small-molecule metabolic tracer 3′-deoxy-3′-¹⁸F-fluorothymidine (7). Likewise, comprehensive investigation of Granta-519 tumors using the CD20-targeting radiolabeled antibody Fab fragments applied here (Fab-PAS₂₀₀ and Fab-ABD) revealed strong uptake, similar to an analogous Fab tracer derived from trastuzumab that was previously investigated in a HER2-positive xenograft model (8).

In contrast, almost no specific tumor uptake and a highly unusual rimlike accumulation pattern were observed in SUDHL-4 tumors for both PASylated and ABD-fused αCD20 Fab versions, and a lower intratumoral gadolinium-DTPA concentration was detected by DCE MRI analysis. Furthermore, investigation of tracer localization at a cellular resolution level using light-sheet fluorescence microscopy revealed broad transcapillary diffusion of the tracer into the surrounding tissue of Granta-519 tumors, whereas the Fab tracer remained mainly intravascular in the SUDHL-4 tumors. H33342 perfusion studies confirmed these results, showing reduced extravasation and tissue penetration of the fluorescent dye in the SUDHL-4 tumors. Finally, the superior perfusion properties of the Granta-519 tumor were in line with the immunohistochemical quantification of CD31-positive tumor vessels, which indicated a significantly higher density of tumor vasculature in Granta-519 than in SUDHL-4.

The distinct behavior of the CD20-positive SUDHL-4 and Granta-519 tumors with respect to the various tracers investigated here can be explained in part by their differing biomolecular properties. In comparison to the rapidly excreted small-molecule imaging compounds ¹⁸F-FDG, gadolinium-DTPA, and H33342, with a molecular size of less than 1 kDa, both versions of the recombinant αCD20 antibody fragment, Fab-ABD and Fab-PAS₂₀₀, exhibit much larger apparent molecular sizes of 40 kDa (≥100 kDa after association with albumin) and 219 kDa, respectively, and, consequently, much longer plasma half-lives of 29 and 5 h, although having the same antigen affinity (18). Low-molecular-weight compounds usually can rapidly cross the capillary wall, whereas large molecules such as monoclonal antibodies show slower transcapillary and interstitial transport. Therefore, the influence of tumor vessel architecture and physiology has a greater impact on macromolecular tracers such as antibodies and Fab fragments.

In general, the existence of different types of blood vessels is known from the literature. In healthy tissue, a regularly patterned and functional vascularization is found, with a normal vessel wall and endothelium. These semipermeable vessels, with a pore size of 5 nm, allow the efflux of both small molecules and proteins up to around 60 kDa but hinder transcapillary transport of larger proteins, particularly antibodies (19). For tumors, on the other hand, vessels with a leaky, tortuous, dilated, and saccular phenotype and a random pattern of interconnections are described, most likely caused by an imbalance of pro- and antiangiogenic signaling (20,21). This perforated phenotype should permit a similar distribution of macromolecular tracers and small-molecule reagents outside the tumor vessels. Indeed, in the Granta-519 tumors we have seen high extravasation for the Fab and all 3 small-molecule compounds. Nevertheless, some vessels did not show this leaky phenotype and were impermeable for the αCD20 Fab as observed by light-sheet fluorescence microscopy.

Interestingly, the SUDHL-4 tumor did not match this phenotype, since accumulation of both the Fab-ABD and the small-molecule gadolinium-DTPA and H33342 was clearly reduced. Hence, there seems to be a second type of distorted tumor vasculature that is characterized by a more closed, less permeable vasculature. This prevents extravasation of macromolecules and also some small molecules (depending on size, hydrophobicity, or charge), whereas low-molecular-weight metabolites, in particular amino acids and glucose, may still reach the tumor. In fact, such a less permeable vasculature is in accordance with the known low or heterogeneous tumor penetration of antibodies or drugs previously observed in certain other xenograft models (22) and, even more importantly, in patients (3).

Interstitial hypertension is another factor in tumor physiology that may play a role in tumor targeting and tissue penetration (19,22) and, in principle, could offer an explanation for the low tracer uptake seen here in the SUDHL-4 tumors. Indeed, in the biomedical literature high interstitial fluid pressure is assumed to prevent efficient tumor penetration of reagents for imaging and therapy (23). However, although a positive pressure within the tumor tissue would clearly counteract convective molecular transport, it is a misconception that interstitial fluid pressure should slow down intratumor molecular diffusion.

Interstitial fluid pressure is likely caused by local colloid osmotic pressure due to a higher concentration of macromolecules such as polypeptides or proteoglycans in the tumor interstitium. Notably, this phenomenon somewhat contradicts the concept of generally permeable tumor vasculature mentioned above and its consequences for macromolecular accumulation discussed further below. Even in biologic systems, Newton’s law of actio est reactio applies, which means that under circumstances of tissue homeostasis the difference in osmotic pressure between capillary lumen and (tumor) interstitium must equal the static pressure between both compartments. In contrast, diffusion is driven by individual concentration differences—that is, the chemical potential for each molecular species—which is only marginally affected by static pressure (at least at the moderate values prevailing in a living organism).

Apart from that, there are a couple of other aspects that govern efficient tumor targeting, especially for protein-based tracers or drugs (24). First, there is the general aspect of vascular permeability already discussed above. Importantly, most tumors lack lymph drainage, which prevents a convective force that drives permeation of macromolecules through healthy tissues. However, the defective and leaky blood vessels seen in many tumors are considered to cause the so-called enhanced permeability and retention effect. This phenomenon describes the extensive leakage of blood plasma components, in particular large proteins and other macromolecules larger than 40 kDa, into the tumor tissue and their nonspecific accumulation there (25). Although the enhanced permeability and retention effect is in fact assumed to promote tumor targeting by antibodies (and liposomes, viruslike particles, and so forth), it does not guarantee proper distribution within the tumor tissue, beyond the capillary vessel.

On the other hand, the most obvious diffusional barrier within the tumor interstitium is the extracellular matrix—especially in the context of a usually crowded tumor parenchyma—which hampers intratumoral distribution of macromolecules in a size-dependent manner (26). Although this aspect deserves more detailed investigation, it has become clear that not only glycosaminoglycans but also various collagens form part of the gellike structure that causes an interstitial transport resistance and slows down diffusion of larger molecules. Taken together, the situation is complex, and well-defined models are needed. The two distinct xenograft models characterized in the present study should prove useful for dissecting the effects of tumor vessel permeability, extracellular matrix composition, and, possibly, interstitial fluid pressure in dedicated experimental setups.

In the clinical situation, detection and treatment of tumors with a complex and heterogeneous physiology is difficult, and patients often face inefficient diagnostic or therapeutic efficacy. Although, for example, the combination of the CD20-specific antibody rituximab with standard chemotherapy results in higher overall survival and complete response rates in many subtypes of NHL, a significant proportion of patients relapse with incurable disease (27). Therefore, assessment of tumor perfusion and permeability in addition to common ¹⁸F-FDG PET staging could considerably improve tumor diagnosis and, eventually, therapy. Indeed, a recently published clinical study showed that adverse findings on perfusion and diffusion MRI correlated with adverse prognosis in immunocompetent patients with primary central nervous system lymphoma (28).

On the basis of our present investigation, the SUDHL-4 xenograft is a suitable preclinical model for investigating the mechanisms of hampered tumor perfusion and permeability for protein therapeutics or tracers, including strategies for tumor vessel normalization or other approaches to improve extravasation and tissue distribution. Several studies have shown that normalization of leaky, abnormal tumor vessels is possible by applying antiangiogenic therapies such as bevacizumab (29). One study found a decrease in interstitial fluid pressure by more than 70% and augmented tumor radiation response in two different mouse models bearing human tumor xenografts after anti–vascular endothelial growth factor treatment (30). During this normalization process, immature vessels were pruned and the integrity and function of the vasculature was improved by enhancing the perivascular cell and basement membrane coverage, eventually resulting in a deeper penetration of molecules into the tumor tissue. Another study demonstrated improvement of specific antibody uptake by targeting the tumor stromal microenvironment with hyaluronidase, thus altering the tumor extracellular matrix (31). Consequently, the SUDHL-4 xenograft offers a model to study alternative CD20-specific tracers with smaller size and faster diffusion than antibodies, such as Fab fragments, anticalins, Affibody molecules, nanobodies, or even peptides.