Background: ctDNA levels in plasma samples permits temporal assessment of tumor mutational status and tumor burden during therapy. Poziotinib is an oral HER2 TKI in development for NSCLC patients harboring HER2 exon 20 insertion mutations. We assessed serial plasma samples for changes in HER2 exon 20 insertion mutations and other driver mutations in first- and second-line patients comparing to clinical response per RECIST1.1. Methods: NSCLC patients harboring HER2 exon 20 insertion mutations were enrolled into the poziotinib ZENITH20 using tumor tissue based NGS. Serial plasma samples were collected at baseline, at C3D1, at Day 1 of every other cycle until disease progression. The Guardant360^â 74-gene liquid biopsy assay was used to assess changes in tumor-associated somatic variants including the target variant HER2 exon20 insertion as well as other emergent driver mutations in ctDNA as expressed as percent variant allele frequency (%VAF). Results: 23 first- and second-line NSCLC patients were evaluable with tumor tissue confirmation of HER2 exon 20 insertion mutations. 22 of 23 (96%) had baseline plasma samples with detectable ctDNA. 21 of 22 samples had detectable HER2 exon 20 insertion mutations (mean % VAF 20±5) resulting in a concordance of 95% versus tissue based NGS. 7 patients had serial testing through C7D1 permitting assessment of ctDNA dynamics and comparison to clinical responses. 5 of 7 (71%) serially tested patients treated with poziotinib at 16mg QD had undetectable HER2 exon 20 insertion at C3D1 which was associated with a tumor response PR. Tumor escape (PD) was observed in 2 of the 5 patients which correlated with increases in target HER2 exon 20 insertion VAF in the plasma with the remaining 3 patients ≥PR. Notably, the rise in HER2 exon 20 in ctDNA occurred prior to tumor escape. In one patient treated with poziotinib at 16 mg QD we observed undetectable levels of the HER2 exon 20 insertion in ctDNA at C3D1 which continued through C16. This patient’s responses correlated with patient tumor response of SD at C2 which then became PR through C9 and CR through C17. Conclusions: Poziotinib treatment resulted in reductions in HER2 exon 20 insertion mutations in ctDNA preceded and correlated with the clinical tumor response. Increases in ctDNA HER2 exon 20 insertion mutations were observed prior to confirmation of tumor escape. Serial monitoring of ctDNA is a potential predictive biomarker for treatment response and disease progression. Future evaluation in a larger population is required to confirm the impact of these findings.