A Drosophila platform identifies a novel, personalized therapy for a patient with adenoid cystic carcinoma

Abstract

Adenoid cystic carcinoma (ACC) is a rare cancer type that originates in the salivary glands. Tumors commonly invade along nerve tracks in the head and neck, making surgery challenging. Follow-up treatments for recurrence or metastasis including chemotherapy and targeted therapies have shown limited efficacy, emphasizing the need for new therapies. Here, we report a Drosophila-based therapeutic approach for a patient with advanced ACC disease. A patient-specific Drosophila transgenic line was developed to model the five major variants associated with the patient's disease. Robotics-based screening identified a three-drug cocktail-vorinostat, pindolol, tofacitinib-that rescued transgene-mediated lethality in the Drosophila patient-specific line. Patient treatment led to a sustained stabilization and a partial metabolic response of 12 months. Subsequent resistance was associated with new genomic amplifications and deletions. Given the lack of options for patients with ACC, our data suggest that this approach may prove useful for identifying novel therapeutic candidates.

Keywords: biotechnology; cancer systems biology; genetics; molecular physiology.

Graphical abstract

Figure 1

Developing a personalized Drosophila avatar screening platform (A) Overview of personalized approach. Genomic analysis of the patient's tumor identified predicted tumor drivers used to develop a personalized fly avatar. Robotics-based drug screening identified a three-drug cocktail that was vetted for safety by a tumor board and internal review board. (B) Prioritized oncogenes and tumor suppressors that emerged from our genomic analysis. FAT4, ERCC2, and FAT1/FAT3 were heterozygous. See also supplemental figure, tables. (C) Immunohistochemistry (brown) identified high levels of plasma membrane and nuclear NOTCH1, indicating elevated NOTCH1 protein and activity in patient tumor sections obtained prior to treatment. Similar immunohistochemical assays failed to validate elevated MAP2K2 activity (pERK) or loss of MAX, and neither were included in the final avatar model. (D) Schematic of transformation vector used to target 4 of 5 cancer genes to different Drosophila tissues. Inducible Notch overexpression (UAS-Notch) was introduced by standard genetic crosses. (E) Small hairpins targeting xpd, ft, and kug in CPCT012.2 led to a ~50% reduction in expression as assessed with qPCR. We used this line as the best model of heterozygosity. (F) Quantifying results of directing ptc > CPCT012 expression on the wing's ptc domain, which led to expansion of the domain including a loss of the sharp boundary. Results are represented as the ratio of the ptc domain area to total wing disc area. (G) Example of ptc > CPCT012-mediated expansion. The ptc domain was visualized with an included UAS-GFP marker (green). Insets highlight expansion; dotted lines indicate added black background to square images. Error bars represent standard error of the mean.

Figure 2

Screen for candidate combinations of FDA-approved drugs (A) Flowchart of multi-step drug screen. An initial screen of the Focused FDA Library yielded tofacitinib and docetaxel as weak single agent hits. Subsequent screens identified tofacitinib, vorinostat, and pindolol as an effective 3-drug combination. (B) Data demonstrating initial rescue by docetaxel and tofacitinib as single agents. (C) Data demonstrating CPCT012 rescue to adulthood by gemcitabine plus tofacitinib, an effective two-drug combination. Tofacitinib was used at a dose below that required for significant rescue. (D) Data demonstrating CPCT012 rescue to adulthood by tofacitinib, vorinostat, and pindolol. The 3-drug combination proved the most effective at rescuing CPCT012 to adulthood. Asterisks (∗) in panels (B–D) indicate p < 0.05 as assessed by Student's t-test. Error bars represent standard error of the mean.

Figure 3

Body PET scans from baseline and after 6 months of treatment (A) Prior to the start of treatment, tumor volume (upper panel) and standardized uptake value (SUV; lower panel) of the 2-deoxy-2-[¹⁸F]fluoro-D-glucose (FDG) tracer were increasing over time, indicating progressive disease. Initiation of treatment led to stabilization of total tumor volume and reduction of lung SUV. (B) Control scans just prior to treatment highlight extensive tumor metastases in the bone and lung. (C) Glucose tracer (FDG) uptake in the lung and bone metastases was substantially reduced after 6 months of therapy in imaged sites; further, no new lesions appeared. These data indicate clinical benefit from the drug treatment, manifested as reduced FDG uptake and absence of progression. h = heart, r = renal tubules, b = bladder.

Figure 4

Patient somatic genomic profiles The patient tumor samples from 2016 to 2019 exhibited significant genomic differences. (A and B) (A) Somatic protein-altering molecular variants (SNVs and indels with AF ≥ 0.05) and (B) somatic copy number variant (sCNV) profiles of the four tumor samples are summarized, as assessed with saasCNV (Zhang and Hao, 2015). The 2019 specimens contained de novo variants and more unstable sCNV profiles.