Summary

Aldehyde dehydrogenases (ALDHs) are a family of NAD(P)-dependant enzymes that catalyse the oxidation of aldehydes to carboxylic acids. Endogenous aldehydes can be produced as a result of metabolism of amino acids, alcohols, lipids and vitamins, while exogenous aldehydes can derive from the metabolism of cytotoxic drugs and environmental factors. Currently 19 ALDH isoforms have been characterised in humans, and ALDH1 isoforms (ALDH1A1, ALDH1A2, ALDH1A3) are considered to be of particular interest due to their role in the detoxification of anti-cancer drugs.
ALDH1A1 expression has been correlated with poor overall survival in a range of cancers, including leukemia, lung, liver, pancreatic, breast, colorectal and ovarian cancer. Resistance to chemotherapy represents a significant challenge in the treatment of ovarian cancer, with a high proportion of women ultimately succumbing to recurrent disease. Tumour recurrence arises from drug-resistant tumour cells able to repopulate the tumour niche and seed new lesions. These cells have been termed cancer stem cells (CSCs), and are characterised by high ALDH1A1 expression, alongside other biomarkers such as CD133 and CD44. Extensive effort has focused on the creation of ALDH1A1-targeted therapies, with some showing success in reducing proliferation and expression of stemness markers in CSC-enriched ovarian cancer in vitro models, as well as synergising with cisplatin treatment to improve efficacy. Moreover, inhibition of ALDH1A1 in animal models of ovarian cancer has proven an attractive therapeutic strategy for CSC depletion and the prevention of recurrence.
Given the causal link between ALDH1A1 expression and chemoresistance in some cancers, the identification of high ALDH1A1-expressing tumour represents a clinical challenge that if solved, could significantly improve patient outcomes. Imaging these tumours using ALDH1A1 expression as a marker of chemoresistance could inform therapeutic interventions and offer the chance to tailor bespoke patient treatments. Despite the commercially-available fluorescent imaging agents being widely-adopted for the isolation of ALDH-positive cells in cell culture, the poor tissue penetration of the fluorescent signal and requirement for direct tumour injection currently limits their in vivo utility. In order to circumvent these inherent limitations, the authors and others have proposed the use of positron emission tomography (PET) as an alternative to fluorescence-based imaging. The authors have recently reported that [¹⁸F]N-ethyl-6-(fluoro)-N-(4-formylbenzyl)nicotinamide – [¹⁸F]1 – has excellent affinity and isozyme selectivity for ALDH1A1 in cancer cells but is rapidly oxidised to its corresponding carboxylic acid in the blood when administered at tracer doses. Here, the authors employed an enzyme-cleavable prodrug strategy to increase the blood stability of our ALDH1A1-targeted radiotracer. Protecting the metabolically labile aldehyde from oxidation maintained ALDH1A1 specificity, producing positive tumour-to-background images in mouse models of ovarian cancer.

Results from nanoScan PET/CT

For the imaging studies, dynamic PET scans were acquired on the nanoScan PET/CT system (1–5 coincidence mode; 3D reconstruction; CT attenuation-corrected; scatter corrected). Mice received a bolus intravenous injection of approximately 3.0 MBq of [¹⁸F]2 through a tail vein cannular on a four mouse imaging bed. A 60 min PET scan was acquired immediately after the injection of the radiotracer. Animals were maintained under isoflurane anaesthesia (1.5–2% in oxygen) at 37 °C during and after radiotracer administration, and throughout the scan. CT images were acquired for anatomical visualisation and for CT attenuation correction (480 projections; helical acquisition; 50 kVp; 300 ms exposure time).
The acquired data were sorted into 19 time frames of 4 × 15 seconds, 4 × 60 seconds, and 11 × 300 seconds for image reconstruction (Tera-Tomo 3D reconstructed algorithm; 4 iterations; 6 subsets; 400–600 keV; 0.3 mm 3 voxel size). VivoQuant software (v.2.5, Invicro Ltd.) was used to analyse reconstructed images. Regions of interest were drawn manually using CT images and 50–60 min summed dynamic PET images as reference. Time versus radioactivity curves (TACs) were generated using the percentage injected dose per mL (% ID mL⁻¹) and from this the area under the time versus radioactivity curve (AUC) was generated.
The dynamic [¹⁸F]2 PET/CT imaging was performed at low and high molar activities (MAs) to understand the effect of radiotracer metabolism on its pharmacokinetics and tumour retention. [¹⁸F]2 was characterised by rapid extraction from the blood and excretion through the urinary tract (Fig. 4a). At early time points, high [¹⁸F]2 uptake was evident in the lung and liver, with liver uptake peaking at 3 and 5 min for low and high MAs, respectively (18.2 ± 2.9% ID mL⁻¹ and 19.8 ± 2.4% ID mL⁻¹; Fig. 4b). Radiotracer retention in the brain was significantly higher with low MA [¹⁸F]2 compared to high MA, shown through the area under the time activity curve (AUC; 117.1 ± 31.3% ID h g⁻¹ and 35.4 ± 6.0% ID h g⁻¹, respectively; p = 0.006; n = 3–7; Fig. 4c). High brain uptake at low MA is likely a product of increased blood concentrations of a blood brain barrier (BBB)-permeable [¹⁸F]1, or a consequence of efflux transporter (e.g. P-glycoprotein) saturation on the BBB, facilitating transient accumulation in the brain. Hepatobiliary excretion was evident by ∼20 min p.i., suggesting possible metabolism of the parent compound at this time point (Fig. 4a).
Representative axial PET/CT images of SKOV3-ip1 and SKOV3-TRip2 tumours at both high and low MAs are shown in Fig. 4d. Importantly, 20 min post-injection of [¹⁸F]2, positive SKOV3-ip1 tumour-to-muscle ratios >1.2 were observed, that continued to increase to 2.5 and 2.1 in SKOV3-ip1 tumours 50–60 min post-injection for low and high MA scans, respectively (Fig. S6a and b, ESI). In high MA scans, 50–60 min after administration, retention of [¹⁸F]2 reached 1.1 ± 0.3% ID mL⁻¹ in SKOV3-ip1 tumours compared to 0.5 ± 0.2% ID mL⁻¹ in the muscle (p < 0.001; n = 7; Fig. 4e). There was no difference in retention of [¹⁸F]2 in SKOV3-ip1 and SKOV3-TRip2 tumours either at low (1.0 ± 0.1% ID mL⁻¹ and 0.9 ± 0.5% ID mL⁻¹, respectively; p = 0.8; n = 3) or high MAs (1.1 ± 0.3% ID mL⁻¹ and 0.8 ± 0.2% ID mL⁻¹, respectively; p = 0.06; n = 3–7). At high MAs, it is likely that the geminal diacetate only offers protection against blood oxidation during the first pass, limiting the ability of [¹⁸F]2 to differentiate between high and low ALDH-expressing tumours. Conversely, whilst addition of carrier-added 2 reduces ALDH-mediated blood oxidation, the unlabelled compound also competes with [¹⁸F]2 for tumour ALDH, thereby blocking production of [¹⁸F]3.

• The authors have developed a prodrug strategy for the non-invasive imaging of ALDH activity in animal models of human ovarian cancer.
• In vivo, the acylal group at trace levels was rapidly cleaved in the blood, the extent of which was substantially reduced through co-injection of the isotopologue [¹⁹F]2.
• PET imaging of [¹⁸F]2 in ovarian tumour-bearing mice at high and low molar activities revealed tumour-specific uptake, but was unable to distinguish between high and low ALDH-expressing tumours. 

Full article on pubs.rsc.org