Localized In Vivo Prodrug Activation Using Radionuclides

Abstract

Radionuclides used for imaging and therapy can show high molecular specificity in the body with appropriate targeting ligands. We hypothesized that local energy delivered by molecularly targeted radionuclides could chemically activate prodrugs at disease sites while avoiding activation in off-target sites of toxicity. As proof of principle, we tested whether this strategy of radionuclide-induced drug engagement for release (RAiDER) could locally deliver combined radiation and chemotherapy to maximize tumor cytotoxicity while minimizing off-target exposure to activated chemotherapy. Methods: We screened the ability of radionuclides to chemically activate a model radiation-activated prodrug consisting of the microtubule-destabilizing monomethyl auristatin E (MMAE) caged by a radiation-responsive phenyl azide, and we interpreted experimental results using the radiobiology computational simulation suite TOPAS-nBio. RAiDER was evaluated in syngeneic mouse models of cancer using the fibroblast activation protein inhibitor (FAPI) agents [^99mTc]Tc-FAPI-34 and [¹⁷⁷Lu]Lu-FAPI-04 and the prostate-specific membrane antigen (PSMA) agent [¹⁷⁷Lu]Lu-PSMA-617, combined with caged MMAE or caged exatecan. Biodistribution in mice, combined with clinical dosimetry, estimated the relationship between radiopharmaceutical uptake in patients and anticipated concentrations of activated prodrug using RAiDER. Results: RAiDER efficiency varied by 70-fold across radionuclides (^99mTc > ¹¹¹In > ¹⁷⁷Lu > ⁶⁴Cu > ³²P > ⁶⁸Ga > ²²³Ra > ¹⁸F), yielding up to 320 nM prodrug activation/Gy of exposure from ^99mTc. Computational simulations implicated low-energy electron-mediated free radical formation as driving prodrug activation. Radionuclide-activated caged MMAE restored the prodrug's ability to destabilize microtubules and increased its cytotoxicity by up to 2,600-fold that of the nonactivated prodrug. Mice treated with [^99mTc]Tc-FAPI-34 and caged MMAE accumulated concentrations of activated MMAE that were up to 3,000 times greater in tumors than in other tissues. RAiDER with [^99mTc]Tc-FAPI-34 or [¹⁷⁷Lu]Lu-FAPI-04 delayed tumor growth, whereas monotherapies did not (P < 0.003). Clinically guided dosimetry suggests sufficient radiation doses can be delivered to activate therapeutically meaningful levels of prodrug. Conclusion: This proof-of-concept study shows that RAiDER is compatible with multiple radionuclides commonly used in nuclear medicine and can potentially improve the efficacy of radiopharmaceutical therapies to treat cancer safely. RAiDER thus shows promise as an effective strategy to treat disseminated malignancies and broadens the capability of radiopharmaceuticals to trigger diverse biologic and therapeutic responses.

Keywords: FAPI; combination chemoradiotherapy; radionuclide therapy; theranostics.

Graphical abstract

FIGURE 1.

RAiDER concept. Targeted radionuclides accumulate in tumor tissues, locally delivering radiation to chemically activated caged prodrugs. Radionuclide-mediated prodrug activation occurs through reduction of phenyl azide caging moiety (orange), leading to linker self-immolation and release of active drug payload (purple) from its drug delivery vehicle (blue), which in this work is long-circulating serum albumin. Consequently, locally activated therapeutic payloads combine with ionizing radiation to maximize tumor cytotoxicity while sparing off-target tissues. Image created with BioRender.com.

FIGURE 2.

Radionuclide-mediated drug release from caged MMAE. (A) Caged MMAE (10 µM) was exposed to different radionuclides with varying activities. Corresponding total radiation dose release was estimated using TOPAS-nBio (26) and compared with prodrug-activated drug release, indicating varying radionuclide efficiencies (µM/Gy). (B) Drug release efficiencies (nM/Gy) across radionuclides and external beam modalities. Highest release efficiency (^99mTc) was compared with other nuclides or modalities (Violin plots: black line, median; dotted color lines, quartiles, 1-way ANOVA with Tukey multiple comparison test). Some values for ¹³⁷Cs irradiation were previously presented (18) and reproduced here to facilitate comparison. Pearson correlation coefficient R was calculated across all data points. linac = linear particle accelerator.

FIGURE 3.

RAiDER restores biologic prodrug activities in vitro. (A) Cytotoxicity of albumin (Alb)–conjugated caged MMAE; ^99mTc-activated, Alb-conjugated caged MMAE; and free MMAE on TBP3743 anaplastic thyroid cancer, measured 72 h after treatment by resazurin-based assay (n = 3, mean ± SE). (B) Representative images and quantification of TBP3743 colony formation (mean ± SE, 2-way ANOVA with Geisser–Greenhouse correction). (C) Representative images of microtubule immunofluorescence (top) and apoptosis by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL; bottom) 24 h after treatment in TBP3743 cells (mean ± SE; 2-way ANOVA with Geisser–Greenhouse correction; scale bars, 200 µm). DAPI = 4′,6-diamidino-2-phenylindole; frac. max = fraction of maximal count/signal as observed in control conditions.

FIGURE 4.

Computational modeling of radionuclide-mediated drug release using TOPAS-nBio. (A) Conceptual mechanism to explain radionuclide-dependent RAiDER efficiencies. Red shading illustrates spatial distribution and frequency of free radical–generating ionization events from given radionuclide within vial. Ionization clouds interact with prodrug molecules most efficiently for nuclides generating lower-energy electrons. (B) Correlation between observed drug release across radionuclides and their estimated number of electrons generated within 100- to 110-keV energy window. (C) Comparison of drug release efficiency with dose imparted by low-energy electrons (LEEs) across radionuclides. Pearson correlation coefficient R was calculated across all data points. LET = linear energy transfer.

FIGURE 5.

In vivo RAiDER biodistribution and prodrug activation. (A) Biodistribution of fluorescent albumin (Alb)–conjugated caged MMAE, measured by tissue Cy5 fluorescence, and [^99mTc]Tc-FAPI-34, measured by γ-scintillation counting in B6129SF1/J mice bearing syngeneic TBP3743 tumors. (B) Biodistribution of activated Alb-conjugated caged MMAE (with [^99mTc]Tc-FAPI-34) compared with nonactivated Alb-conjugated caged MMAE, as measured by liquid chromatography–mass spectrometry or high-performance liquid chromatography. Data are mean ± SE (n = 3–4 mice per condition, 2-tailed t test shown). %ID/g = percentage injected dose per gram of tissue.

FIGURE 6.

RAiDER enhances ability of targeted radionuclides to block tumor growth. (A and B) Mice bearing TBP3743 tumors were treated with albumin (Alb)–conjugated caged MMAE (5 mg/kg/dose) and either [^99mTc]Tc-FAPI-34 (A; 18.5 MBq/dose; n = 18 total mice, 36 total tumors) or [¹⁷⁷Lu]Lu-FAPI-34 (B; 18.5 MBq; n = 20 total mice, 40 total tumors). Caliper measurements of tumor growth are shown over time (left) and with individual tumor sizes on indicated day (middle; 2-tailed Mann–Whitney tests compare combination and pooled control or monotherapies). Mouse mass was measured during treatment time course (right). Data are displayed as mean ± SE.

FIGURE 7.

Estimating tumoral drug release in patients using clinical dosimetry. Lesion SUVs derived from [⁶⁸Ga]Ga-PSMA PET and [⁶⁸Ga]Ga-DOTATATE PET in prostate and neuroendocrine cancer patients receiving [¹⁷⁷Lu]Lu-PSMA-617 (38) and [¹⁷⁷Lu]Lu-DOTATATE (39), respectively, were compared with expected intratumoral active drug release (calculated based on release efficiencies in Fig. 2). Absorbed dose within each lesion was calculated in those previous studies using ¹⁷⁷Lu SPECT/CT, demonstrating good correlation with PET SUV. Red line denotes inhibitory concentration of 50% estimated for aggressive cancers in this study (Supplemental Table 2).