Background: Comprehensive Genomic Profiling (CGP) liquid biopsy tests are vital tools for oncologists that enable the non-invasive detection of diverse somatic mutations, many of which correspond to targeted therapies, in circulating tumor DNA (ctDNA). However, these tests have less utility when ctDNA signal is low, which is common even in late-stage cancer patients. BillionToOne’s Northstar (NS) Select assay is a novel CGP test that seeks to leverage proprietary technology and novel calling algorithms to address this limitation by achieving higher sensitivity. Methods: Clinical samples for 182 unique patients with stage III or IV solid tumors were obtained via a prospective validation study assessing head-to-head concordance with widely adopted, commercially available ‘comparator’ liquid biopsy CGP assays. The study cohort was recruited from 1 large hospital center and 6 community clinics across the US and was diverse in patient demographics and clinical presentations. Over 15 different cancer types were represented in the cohort, with the majority of patients having either Lung, Breast, Colorectal, or Prostate cancer, all of which are commonly tested using liquid biopsies in the US. Pathogenic (including likely pathogenic) variants called by NS Select and/or the comparator assay were analyzed for concordance. For a subset of samples (N=28), white blood cell-derived genomic DNA was also processed to identify which variants were detected due to clonal hematopoiesis of indeterminate potential (CHIP). Results: In the head-to-head comparison, a large number of variants (N=380) were concordant between the two assays. In addition, NS Select detected 43% (N=147) more pathogenic variants than comparators when controlling for matched coverage regions. The majority of these variants were detected below the reported limit of detection of the comparator assays, but above the limit of detection of NS Select. The rate of CHIP mutations was not significantly different in the variants detected by NS Select (19.0% ± 7.6%) vs. comparators (17.2% ± 9.2%). Finally, the proportion of patients with no pathogenic variants detected was shown to be lower by 45% when using NS Select vs. the comparator assays (20/182 vs. 36/182). Conclusions: NS Select reported clinically useful information about the tumor for most patients. For patients in which ctDNA signal was high, NS Select had very high concordance with comparator tests. Furthermore, additional low-VAF pathogenic mutations identified by NS Select translated to a higher diagnostic yield compared to other CGP assays, and CHIP was ruled out as an explanation for these additional detections. This suggests that some patients with low levels of ctDNA would derive clinical benefit from testing with NS Select due to its high sensitivity.