Molecular characterization of residual disease post-neoadjuvant sacituzumab govitecan (SG), pembrolizumab, or their combination in patients with muscle-invasive bladder cancer (MIBC).

4617 Background: In patients (pts) with MIBC, we tested the activity of neoadjuvant SG (SURE-01, NCT05226117), SG + pembrolizumab (SURE-02, NCT05535218) and pembrolizumab alone (PURE-01, NCT02736266), followed by surgery. We aimed to molecularly characterize the residual post-neoadjuvant therapy (NAT) MIBC and its changes from baseline samples. Methods: Post-NAT samples (ypT≥2) were available for transcriptome analyses from 12 pts in SURE-01, 24 in SURE-02, and 26 post-pembrolizumab alone. Decipher bladder genomic subtyping classifier (GSC) expression profiling assay (Veracyte) was used to compare molecular subtyping distributions and genes/signatures expressions between matched pre-post NAT samples, including 192 locked transcriptomic signatures in GRID (v3.1) and pre-specified therapeutic target genes. FGFR3-active tumors were determined using a previously-developed long non-coding RNA (lncRNA)-based classifier. The clinical primary endpoint was event-free survival (EFS). Results: Post-NAT molecular subtyping identified Luminal (L) vs non-Luminal (NL) tumors in: 17% (2/12) vs 83% (10/12) (SURE-01), 21% (5/24) versus 79% (19/24) (SURE-02), and 8% (2/26) versus 92% (24/26) (PURE-01). In total, 21 pts from SURE studies and 22 from PURE-01 had matched pre-post NAT samples. Consensus molecular subtyping revealed a concordance in 86% (6/7) of pts in SURE-01, 57% (8/14) in SURE-02 and 36% (8/22) in PURE-01. Stroma-rich tumors were found in 17% (2/12), 29% (7/24) and 50% (13/26) of post-NAT samples. After NAT, we found a decrease in L subtypes in SURE-01 (57% → 29%) and in SURE-02 (36% → 14%) vs an increase in stroma-rich (Consensus) subtype in SURE-02 (7% → 36%) and in PURE-01 (14% → 50%). Elevation of TGF-beta signaling hallmark (p=0.02) and immune exclusion (p=0.02) scores characterized post-NAT samples in SURE-01, while in SURE-02 we found elevated stromal gene scores (p=0.004) and lower DNA repair hallmark scores (p=0.03) after NAT. Comparing matched samples from SURE-01 and 02, we found significantly decreased TROP2 expression scores (p=0.03) and lower HER2 (p=0.06) and NECTIN4 expression (p=0.07). Application of the lncRNA-based FGFR3-activity classifier revealed a total of 10/36 (28%) FGFR3-active tumors post-NAT in SURE-01 and SURE-02, all without EFS events during follow-up. For SURE-02, we found 4/24 (17%) pts harboring claudin-low tumors post-NAT, who did not experience EFS events following adjuvant pembrolizumab, similar to the low EFS rate observed in PURE-01 (N=7, 22%). Conclusions: This study expands our knowledge on residual MIBC after various NAT, displaying a noteworthy plasticity and significant changes post-NAT that are important to unveil the biology of these tumors and putative biomarkers for sequential therapies.