Targeting Tropomyosin Receptor Kinase in Cutaneous CYLD Defective Tumors With Pegcantratinib

Key Points Question Can targeting tropomyosin receptor kinase with an existing topical kinase inhibitor, pegcantratinib, 0.5% (wt/wt), reduce cutaneous cylindroma tumor volume more than placebo? Findings In this phase 2 clinical trial that included 150 tumors from 15 patients with CYLD cutaneous syndrome, pegcantratinib-treated tumors did not achieve the primary outcome of response. Molecular analyses of biopsy material demonstrated drug penetration; however, drug concentrations achieved were inadequate to abrogate tropomyosin receptor kinase signaling in CYLD cutaneous syndrome tumors. Meaning These findings indicate that further studies should examine dose-escalation of pegcantratinib in these patients.

In two patients, clinical criteria for diagnosis of CCS was confirmed, but no gene mutation was detected, a feature found in 15% or more patients with CCS. 4

Tissue analysis
Tumor biopsy material was used to determine drug penetration at 3 levels (detailed below) into the punch biopsy. We used a novel method for obtaining material for drug assay from skin tumor cells that were adjacent to sections where we obtained RNA for transcriptomics. We also obtained histology at each of the 3 levels of the tumor to determine how far into the tumor the sample was being obtained from.

Drug penetration assay
Skin punch biopsy samples were serially sectioned such that tissue could be obtained at 3 defined levels within the tumor for drug measurement. The top level contained tissue from approximately 0-300 µm from the surface, the middle level approximately 624-924 µm from the surface and the bottom level approximately 1248-1548 µm from the surface. Samples were analysed using a recently developed and fully validated method for the quantitation of pegcantratinib in human tumor material collected by punch biopsy. 5 The method involves HPLC separation of pegcantratinib followed by in-source fragmentation and detection of a small molecule pegcantratinib-specific fragment by LC-MS analysis. The assay was linear over a concentration range of 1-500 ng/mL, with a lower limit of quantification (LLOQ) of 30ng/mL exhibited.

Cell viability assay
Primary cells from patient tumors were cultured on three-dimensional (3D) tissue cullture scaffolds for 28 days and then treated with a range of concentrations of pegcantratinib for 14 days, as previously described. 6 Matched cells were grown on standard tissue culture plastic and treated with pegcantratinib for 3-14 days. Cell viability was assessed using an ATP dependent luminescent assay (CellTitre Glo -Promega UK).

Transcriptomic analyses
RNA was extracted from trial samples and control samples and stranded preparation was performed using the Illumina stranded mRNA kit. Libraries were prepared and sequenced using an Illumina Hiseq 2500, giving 45 million paired end reads per sample which were 100 bp in length. FASTQ files were aligned using the splice aware aligner program STAR to generate alignment files. 7 The read counts for each sample file were counted using the R package Subread. 8 Differential gene expression analysis was carried out using the package DeSeq2. 9 Hierarchical clustering analysis was done employing an Euclidian distance method and a "complete" agglomerative method, using the R package hclust. Assessment of the robustness of clustering was carried out using the R package pvclust, which calculates p-values for hierarchical clustering via multiscale bootstrap resampling. 10 Variant analysis was carried out on the RNAseq data using The Genome Analysis Toolkit (GATK) 11 following GATK best practices guidelines for variant calling on RNAseq. Raw RNAseq reads were mapped against the human reference genome version hg19 using the STAR aligner. 7 The GATK tool SplitNCigarReads (specially used for RNAseq) was used to split reads into exon segments and hard-clip any sequences overhanging into the intronic regions. Indel realignment and base recalibration was carried out before variant calling. Variants were screened against the 100K Human Genome Project database. After variant calling, variant filtering was then applied using hard filters.

Quantitative PCR
Quantitative PCR was carried out to validate the expression of NTRK1, NTRK2 and NTRK3 using established methods. cDNA was generated from an input of 200ng of RNA from snap frozen samples, using the High capacity cDNA Reverse Transcription kit (Applied Biosystems) according to the manufacturer's instructions. PCR reactions were set-up in 25µl reactions using 1µl of input cDNA, Taqman probes (Supplementary Table 5

) and Taqman Gene Expression Master Mix (both Applied
Biosystems) with qPCR carried out on a 7500 FAST thermocycler (Applied Biosystems) according to the manufacturer's protocol. All samples were run in triplicate and results were analysed using the 2 -ΔΔCt method. Water was used as a no template control for each Taqman probe.

Protein expression analyses
Immunohistochemistry (IHC) staining for BCL2 and neurofilament expression and immunoblotting was carried out as described previously. 6 Tissue sections from snap frozen skin tumor biopsies were fixed, blocked and then probed overnight at 4°C with primary antibody against BCL2. Secondary HRP antibodies were applied the following day and visualised with DAB as per manufacturer's instructions (Dako-K4009 Envision). Antibodies against pERK, ERK and BCL2 were obtained from Cell Signalling, USA. Slides were scanned on a Leica slide scanner, and images captured 4 from representative areas of each tumor. Image intensity was assessed using Image J calibrated in conjunction with a consultant dermatopathologist.

Statistical Considerations
Phase 1b. Given previous published work, 12,13 it was anticipated that there would be low numbers of treatment related adverse events in phase 1b. It was therefore felt reasonable to recruit only eight patients. Lack of adverse skin reactions in at least five of eight treated patients allowed the trial to progress to phase 2a, following recommendation from an independent Data Monitoring Committee. Escalation of dose was not pursued due to lack of safety data and manufacturing limitations in humans in this exploratory study in CCS.

Phase 2a.
It was anticipated that 75 tumors could be measured in each treatment arm (150 tumors in total) in a single centre. As the first exploratory trial, the design parameters are provided as an exemplar of the size of errors that may be anticipated with these patient numbers. Tumors receiving placebo treatment were not expected to respond to treatment, set to be very small at 5% (p0). Using Fleming A'Herns early phase methodology 14 any response on the experimental treatment <5% (p0) would not indicate a treatment worthy of further investigation. A level of efficacy of >15% (p1) would indicate a treatment that warranted further investigation. This level of efficacy was clinically plausible and relevant given that there is no current medical treatment or existing data for this intervention for these patients. 75 tumors recruited in the experimental arm would provide associated error levels of 3.4% type I error (alpha) and 10.8% type II error (beta), deemed acceptable in the early phase setting. The justification to investigate pegcantratinib further is based on observing a minimum number of responses. The trial recruited an equal number of placebo-treated tumors to provide an unbiased benchmark. As an early phase trial, statistical analyses were exploratory (not based on hypothesis testing), based on descriptive data presented by treatment group. Small sample trials based on early phase methodologies usually do not have sufficient statistical power for hypothesis driven statistical comparisons and as such are discouraged in favour of point and confidence interval estimation, 15 the strategy followed in this trial.