IMG-36. AI-guided retrospective imaging analysis of treatment response to targeted therapy in BRAF V600K metastatic melanoma

Artificial intelligence (AI) is transforming neuro-oncology, especially in imaging, by offering innovate ways to track disease progression and treatment response. We present the case of a 41-year-old man with metastatic BRAF V600K melanoma and innumerable brain metastases who showed a remarkable response to BRAF (BRAFi) and MEK (MEKi) inhibitors, avoiding whole-brain radiation therapy (WBRT). A convolutional neural network algorithm was used to quantify metastatic lesion count and volume before and after therapy. Initially diagnosed in December 2021 with ulcerated melanoma and positive sentinel lymph nodes, he underwent resection and completed nine cycles of adjuvant pembrolizumab. Next-generation sequencing revealed a BRAF V600K mutation and high tumor mutational burden. After loss to follow-up, he re-presented in June 2024 with cutaneous recurrence, widespread metastases, and over 30 brain lesions on MRI. He began BRAFi/MEKi with initial clinical improvement but was transitioned to ipilimumab/nivolumab by his medical oncologist. No follow up brain imaging was obtained. By cycle four of immunotherapy, he developed seizures. MRI in February 2025 showed disease progression with hundreds of enhancing lesions. A multidisciplinary team recommended restarting targeted therapy and deferring WBRT to preserve cognition. By March 2025, MRI and clinical status improved, with resolution of seizures and return to baseline cognition. AI-based retrospective analysis revealed: 109 lesions totaling 7,720 mm³ in July 2024; 424 lesions measuring 78,581 mm³ in February 2025; 582 fragmented lesions measuring 40,660 mm³ in March; and 552 lesions totaling 38,325 mm³ in May. The paradox of increasing lesion count and decreasing volume reflects fragmentation of previously confluent lesions. This case demonstrates how AI can complement diagnostics and support a personalized treatment approach. It also underscores the utility of BRAFi/MEKi in acutely managing high-burden melanoma brain metastases and potentially avoiding WBRT.