Reply: From Mice to Humans: The Exocrine Pancreas Does Not Matter in Human GLP-1 Receptor Imaging

CHALLENGES

β-cells present an extremely difficult imaging target in the clinic, in part because of their low fraction (∼1%) in the pancreas and the small size of islets, resulting in significant volume averaging with surrounding exocrine tissue. In addition, as Willekens et al. have shown (1), the expression of GLP-1R has significant interspecies variability. The important question for the clinic is if and how much GLP-1R protein expression exists in off-target cells in humans. Specifically, expression of GLP-1R on exocrine cells is critical, since even expression levels 100-fold lower than β-cells can still be significant given their abundance relative to β-cells, and estimates have indicated that 1,000-fold lower levels are needed for β-cell mass quantification (2).

Unfortunately, many studies cited previously (3) indicate measurable expression of GLP1-R protein in human exocrine cells (4). The importance of quantifying target expression cannot be overstated, and this has to be done at the protein level. In this case, the transcriptional data are not representative of protein levels, which are the relevant metric for molecular imaging. For example, the messenger RNA expression data from Willekens et al. show similar mouse and rat messenger RNA endocrine-to-exocrine ratios despite very different protein levels (1). Absolute quantification (i.e., number of receptors per cell) is critically important. The “small differences in exocrine pancreatic uptake between wild-type and GLP-1R knockout mice” on a per-cell basis translate into over half of the total signal in the pancreas because of their 100-fold higher prevalence. For mice, we quantified 54,000 GLP-1R per β-cell and approximately 50-fold lower levels for exocrine cells (∼40-fold lower based on single-cell flow cytometry measurements (3) and 60-fold lower based on bulk percentage injected dose/g (5)). Indeed, 1,000 receptors per cell is below the limit of detection for most single-cell and tissue fluorescent methods and can easily be overlooked compared with the intense β-cell staining. Despite this difficulty in detection, a robust signal unfortunately remains, and this level in mice is well above the limits necessary to cause problems in whole-body imaging.

RELEVANCE

To clarify, we do not claim that “the lack of clinical distinction between healthy volunteers and subjects with long-term diabetes” was caused by exocrine uptake, but rather the full sentence indicates this is a possible result consistent with the evidence. To rigorously and unambiguously identify residual β-cell mass in patients, uptake from the exocrine pancreas has to be discounted.

Even more problematic, Waser and Reubi demonstrated that only about half of human samples (3/5) appear to express detectable exocrine signal (6). If it were uniform among patients, we would agree that some long-standing type 1 diabetic patients showing background levels would support a lack of exocrine uptake. However, since some patients appear to lack exocrine expression, the background levels could simply be the patients lacking exocrine expression. Now in contrast, the preliminary results reported by Gotthardt et al. indicate less variability in the 7 patients they studied. Here, the endocrine-to-exocrine ratio was 3.9 ± 0.5, with little variation. However, this number is similar to the ratios they reported in mice (4.11 ± 0.9 and 4.56 ± 0.9) rather than rats (44–106). These clinical data appear consistent with histology reports of significant human exocrine GLP-1R expression (7,8).

The fact that these clinical data are similar to mice does not that mean that the absolute expression levels are the same in mice and humans. At the tracer doses used, the similarity is more likely a reflection of delivery, that is, higher vascularization in endocrine tissue than in exocrine tissue. This is why the mouse autoradiography ratio is only approximately 4 at tracer doses whereas the absolute expression differences we measured are close to 50 at saturating doses. It is currently unclear what this ratio is in humans.

WHERE DO WE GO FROM HERE?

The most recent data reported by Gotthardt and colleagues are exactly the type of data that need to be collected if we want to determine whether the exocrine uptake can be selectively blocked. I commend them for their significant efforts to collect these important data, since pancreatic samples are difficult to obtain yet exactly the type of data the scientific community needs. Similar to our collaboration using intraoperative imaging agents to look at antibody distribution in tumors and associated healthy tissue (9), ex vivo analysis can enable absolute probe uptake after in vivo administration.

If their preliminary results continue to mimic the mouse pancreas, selective blocking may be needed to suppress exocrine uptake and reliably detect β-cells as we propose. The method we used in mice is not perfect, and the absolute endocrine-to-exocrine expression ratio in humans likely needs to be greater than in mice to practically work in humans. However, potential blocking agents are available that could be used in a similar study. Although we used a lipophilic fluorescent dye to simultaneously bind albumin and facilitate imaging, a lipophilic conjugate such as liraglutide would be a potential Food and Drug Administration–approved blocking agent.

The path is difficult, but it is an exciting time to investigate β-cell biology. Importantly, these results do not discount the potential presence of β-cells within the diabetic pancreas. Rather, they indicate that the exocrine expression level has to be addressed to definitively image residual β-cell mass within the human pancreas. The method outlined by Khera et al. (3) is one approach that can be pursued in humans to address this issue. The work by the Gotthardt lab and others staining for β-cell markers such as GLP-1R in patients with long-standing diabetes raises the possibility of reversing this disease. To interpret these imaging results, we need to definitively know the endocrine-versus-exocrine uptake of these molecular probes. Exendin has many attributes of an ideal imaging agent—tight binding, high retention from metabolic trapping, low nonspecific sticking, and rapid clearance. We just need to ensure the cellular specificity to be able to use this agent in the clinic.