Background: About 20 000 patients are diagnosed with muscle-invasive UC annually, where the five-year survival rate is approximately 5% in metastatic cases. The transcription factor PPARG is associated with the luminal lineage subtype reflecting ∼65% of all advanced UC patients. Recurrent genetic alterations in PPARG, including focal amplification, missense mutations, and fusions, as well as hotspot mutations in its obligate heterodimer, retinoid X receptor alpha (RXRA) are characteristic of this molecular subtype. Materials and

Methods: We analyzed tumor tissue collected from over 3000 patients with muscle-invasive UC sequenced using the Tempus xT assay (DNA-seq of 648 genes at 500x coverage; RNA-seq; 3262pt DNA, 2685pt both). Molecular classification as luminal (luminal papillary, luminal, and luminal infiltrated subtypes) or non-luminal (basal-squamous and neuronal subtypes) was performed using non-negative matrix factorization (NMF) rank 5 following the Robertson method, excluding PPARG from the gene set to eliminate the inference of bias. Mean expression was reported as Log2[TPM+1]. Mutations were reported as single nucleotide variants or insertion/deletions (SnvIndel).

Results: We determined that luminal-papillary (669pt), luminal (449pt), and luminal-infiltrated (621pt) UC subtypes comprised 65% of the real-world cohort, consistent with subtype distribution as reported in the Robertson TCGA data set. Higher PPARG expression was associated with luminal subtypes (7.78 Log2[TPM+1], p < 0.00001) compared to non-luminal subtypes (5.47 Log2[TPM+1]). The prevalence of mutations in PPARG, RXRA, and fibroblast growth factor receptor 3 (FGFR3) was also significantly higher in the luminal subtype, where the frequency of PPARG amp (copy number 4–7) was 10% (p < 0.05), PPARG SnvIndel was 2.0% (p = 0.8),RXRA hotspot SnvIndel was 3.3% (p < 0.00005) and FGFR3 SnvIndel was 14% (p < 0.005). Collectively 1 in 3 advanced UC patients in a real-world population harbored mutations in PPARG, RXRA or FGFR3. A PPARG expression threshold (7.75 Log2[TPM+1]) was derived from logistic regression comparing amplified and diploid patients, where gain of even one copy of PPARG is associated with overexpression. Many diploid patients exhibit PPARG expression (7.49 Log2[TPM+1]) comparable to amplified patients. High PPARG expression was sustained in metastatic samples (7.55 Log2[TPM+1]), and expression did not significantly vary by metastatic tissue site (p > 0.05).

Conclusion: To the best of our knowledge, this represents the largest realworld dataset from patients with advanced and metastatic UC evaluated to date. Stratification of advanced UC patients based upon PPARG status as a defining feature of the luminal phenotype supports targeting of this pathway with novel agents under development.

VIEW THE PUBLICATION

VIEW THE POSTER