Circulating tumor DNA (ctDNA) in metastatic melanoma (MM) patients (pts) with brain metastases (mets).

9581 Background: Brain mets are common in pts with MM, occurring in up 50% of pts. Serial ctDNA levels at baseline and early on therapy predict objective response (OR) and overall survival (OS) in MM pts without brain mets receiving immune checkpoint inhibitors (ICI). The predictive utility in pts with brain mets is unknown. Methods: BRAF/NRAS ctDNA at baseline and early on ICI therapy (week 3-8) was analysed using mutation specific digital droplet PCR (detection threshold 2.5 copies/ml plasma) in MM pts with brain mets. Intracranial (IC) and extracranial (EC) disease volume (sum of product of diameters of ALL mets [SPD]) and OR, progression-free survival (PFS) and OS were analyzed. Results: 48 pts with brain mets were studied, 40 with concurrent EC disease (SPD EC med 892mm 2 [25-22417mm 2 ], IC med 200mm 2 [9-1487mm 2 ]) and 8 with isolated IC mets (SPD med 150mm 2 [66-1035mm 2 ]). At baseline, ctDNA was detectable in 28 (58%) pts, 28/40 (70%) with IC + EC mets and 0/8 with isolated IC mets. Baseline ctDNA reflected EC disease volume (Pearson’s r = 0.4, p = 0.01) but not IC volume (r = 0.1, p = 0.5). Baseline ctDNA did not associate with IC or EC OR or PFS, however, undetectable ctDNA was associated with superior OS (HR 0.4, p = 0.01). Early on therapy, ctDNA predicted EC response but not IC response; EC OR was 65% if ctDNA undetectable and 6% if detectable (p < 0.01), IC OR was 35% if undetectable and 12% if detectable (p = 0.1). Nevertheless, undetectable ctDNA early on therapy was associated with a superior PFS (HR 0.3, p < 0.01) and OS (HR 0.2, p < 0.01), indicating survival is largely determined by extracranial disease activity. In the 8 pts with IC disease only, 7/8 had disease progressive disease as best response. 1/8 subsequently developed detectable ctDNA at week 12, with multiple new EC mets seen at first restaging scan at this time. Conclusions: ctDNA does not appear to be a useful biomarker for detecting brain mets nor monitoring brain response in melanoma pts receiving ICI. This has important implications when using ctDNA in the setting of surveillance in the metastatic and adjuvant setting.