Non-invasive MRD monitoring and profiling of clonal evolution by ctDNA in patients with advanced cancers treated within molecular tumor boards.

3041 Background: Profiling of targetable genetic alterations within molecular tumor boards (MTB) guides for personalized treatment selection in patients with advanced cancers. During therapy, response is typically assessed by CT scans or MRI, which often have suboptimal sensitivity and specificity. Circulating tumor DNA (ctDNA) from blood plasma has emerged as a promising biomarker for noninvasive profiling of tumor mutational landscapes and disease monitoring. Here, we applied a pan-cancer next-generation sequencing (NGS) technology to assess the role of ctDNA for comprehensive tumor genotyping, early response prediction, and characterization clonal heterogeneity in patients receiving MTB recommended therapies. Methods: We developed and applied a custom targeted NGS approach (ExTARGET), which covers 266 genes across a 540 kb genomic region, to 157 plasma samples obtained at distinct milestones from 57 patients with diverse solid cancers. Plasma samples from healthy individuals ( n = 24) were used to determine the specificity of our technology. Results: We identified variants in 96% of baseline plasma samples by ctDNA profiling, with a median of 7 mutations per patient (range: 1-41). Most frequently mutated genes included KRAS (35%), BRAF (24%), ERBB2 (22%) and TP53 (22%). Targetable tumor variants that led to treatment recommendations within the MTB were found non-invasively in 69% of patients. Longitudinal monitoring of baseline ctDNA variants in on-treatment samples, obtained early during therapy ( n = 21), revealed that ctDNA dynamics were predictive of disease progression and preceded radiological/clinical progression in 8/19 (42%) patients. All patients with increasing ctDNA levels early during treatment showed radiologic disease progression in subsequent CT scans. On the other hand, an early decrease of ctDNA levels was associated with durable disease control in most patients and significantly favorable progression-free survival ( p = 0.008; HR = 0.1, 95%CI: 0.02-0.6). Next, we explored temporal clonal heterogeneity in plasma samples collected from 16 patients with disease progression following MTB-recommended therapies. We observed substantial clonal evolution over time, with all samples harboring at least one emerging variant. Among these emerging alterations, 19% were classified as ‘oncogenic’ and 5% were identified as potentially targetable. Conclusions: We here developed an NGS-based technology for ctDNA profiling in heavily pretreated patients receiving MTB-recommended therapies. Non-invasive genotyping from plasma robustly identifies targetable aberrations and allows comprehensive tumor genotyping. Monitoring of ctDNA during treatment and at disease progression facilitates early prediction of treatment response and profiling of temporal clonal heterogeneity that could enable subsequent treatment selection.