High-resolution positron emission microscopy of patient-derived tumor organoids

Abstract

Tumor organoids offer new opportunities for translational cancer research, but unlike animal models, their broader use is hindered by the lack of clinically relevant imaging endpoints. Here, we present a positron-emission microscopy method for imaging clinical radiotracers in patient-derived tumor organoids with spatial resolution 100-fold better than clinical positron emission tomography (PET). Using this method, we quantify ¹⁸F-fluorodeoxyglucose influx to show that patient-derived tumor organoids recapitulate the glycolytic activity of the tumor of origin, and thus, could be used to predict therapeutic response in vitro. Similarly, we measure sodium-iodine symporter activity using ^99mTc- pertechnetate and find that the iodine uptake pathway is functionally conserved in organoids derived from thyroid carcinomas. In conclusion, organoids can be imaged using clinical radiotracers, which opens new possibilities for identifying promising drug candidates and radiotracers, personalizing treatment regimens, and incorporating clinical imaging biomarkers in organoid-based co-clinical trials.

Fig. 1. Schematic representation of the oPEM workflow.

Tumor organoids were cultured for 2–3 weeks in a submerged basement membrane matrix culture system. Patient-derived organoids (right) were imaged with high resolution using FDG, a radiotracer commonly used in the clinic for diagnosis and staging of head-and-neck cancer patients (left). The organoids were placed on thin inorganic scintillators and the resulting scintillation light was imaged with a highly sensitive microscope through a ×20 objective lens. The scale bars of PET and oPEM images highlight the large difference in image resolution. The color bar shows radioactivity (Bq/pixel) inside tumor organoids.

Fig. 2. Characterization of head and neck tumor organoids grown from tumor tissue.

a, b Fluorescence immunohistochemistry comparing (a) fresh tumor sample and (b) tumor organoid grown from head-and-neck adenoid cystic carcinoma sample labeled with four markers: blue showing Hoechst 33342 (HOE, nuclei); red, E-cadherin (E-cad, tumor epithelial cells); green, vimentin (VIM, tumor-associated fibroblasts); and yellow, CD3 (tumor-infiltrating T cells). c, d Co-registration of hematoxylin & eosin staining (H&E) and RNA in situ hybridization (RNAscope duplex assay) detecting expression of the glucose transporter GLUT1 (blue), stem-cell marker CD44 (blue), and cell proliferation marker Ki67 (red) in c fresh human oral squamous cell carcinoma tissue and d corresponding tumor organoids. Scale bar: 0.2 mm.

Fig. 3. Positron emission microscopy of tumor organoids.

a Brightfield image (left), radioluminescence image (middle), and overlay (right) show the correlation between FDG uptake distribution and tissue structure. FDG uptake is elevated in most of the organoids. b Comparison between fluorescence imaging with 2-NBDG, a fluorescent glucose analog (left), and oPEM imaging with FDG (middle) indicates inconsistent co-localization (right) of the two probes. c Tumor organoid labeled with live/dead fluorescent stains (three left panels) shows that FDG uptake (right panel) is associated with tissue viability. Blue: Hoechst 33342 (all nuclei), Green: SYTOX green (dead nuclei). d FDG-oPEM of human oral squamous cell carcinoma organoids (whole mount) and co-registration with H&E staining of organoid sections. e Two different regions of an H&E section and its corresponding whole mount FDG-oPEM/BF (top panels) co-registered with in situ hybridizations (bottom panels) showing expression of the glucose transporter GLUT1 (blue) and cell proliferation marker Ki67 (red). The color bar shows radioactivity (Bq/pixel). Scale bar: 0.5 mm.

Fig. 4. Comparison of PET vs oPEM imaging.

a PET/CT scan of patient T3. Two papillary thyroid carcinoma nodules (white arrows) show contrasting levels of FDG uptake. b oPEM images of tumor organoids derived from these nodules. The organoids retained the contrasting metabolic identity of the original two nodules. The dotted lines show the spatial extent of the individual organoids. The color bar shows radioactivity (Bq/pixel). Scale bar: 1 mm. c Scatter plot of FDG influx constant (K_i) for organoids (n = 5) derived from left and right nodules. The median uptake is shown as a black line. The organoids derived from the left nodule took up >10 times more FDG on average than those from the right nodule. An unpaired two-tailed t test was applied for significance testing. d Scatter plot showing the correlation between the K_i of organo_ids (n = 13) and tumors of origin (n = 4, shown in four different colors; Pearson’s $r$= 0.756, $P$ = 0.0032).

Fig. 5. oPEM & fluorescence co-imaging of metabolic activity after cisplatin treatment.

a Untreated organoid shows a spatial pattern of cellular viability (fluorescence stain; left column) consistent with the FDG uptake profile (right column). Organoids treated with increasing doses of cisplatin experience a decrease in their metabolic activity. Blue: Hoechst 33342 (all nuclei), Green: SYTOX green (dead nuclei). The color bar shows radioactivity (Bq/pixel). b Quantitative FDG uptake inside tumor organoids treated with 0 or 10 µM dose of cisplatin (n = 2 organoids from two independent experiments). c The intensity profile of the FDG signal and green channel along the dotted line (inset) shows a spatial anti-correlation.

Fig. 6. Functional characterization of NIS activity in tumor organoids derived from papillary thyroid carcinoma patients.

a oPEM imaging of pertechnetate (^99mTcO⁻₄) uptake by tumor organoids with (left) or without (right) competitive inhibition by 1 mM potassium iodide (KI). Scale bar: 200 µm. b A nearly 10-fold difference in ^99mTcO⁻₄ uptake is observed between the two experimental conditions (n = 6 independent tumor organoids). Data are shown as mean ± S.E.M. (box), median (horizontal line), and 95% confidence interval (CI; whiskers). c ^99mTcO⁻₄ uptake kinetics of NIS-expressing and wild-type MDA-MB-231 cells. Error bars correspond to the standard error from n = 3 independent measurements. d Gamma counting and oPEM images of ^99mTcO⁻₄ uptake by 3D multicellular spheroids (10,000 cells/spheroid) made from NIS-expressing and wild-type MDA-MB-231 cells (n = 3 independent spheroids). Data are shown as mean ± S.E.M. (box), median (horizontal line), and 95% CI (whiskers). Statistical significance was assessed using a two-tailed unpaired t test (b, d). CPM: counts per minute. The color bar shows radioactivity (Bq/pixel).