Unsupervised clustering reveals cell types in 145 specimen melanoma PBMC cytometry dataset

Poster Highlights

• In this poster, presented at the SITC Spring Scientific 2025, we used unsupervised clustering to analyze a 29M cell cytometry dataset to identify immune cell populations and subsets that correlate with response outcomes in melanoma patients treated with anti-PD1 therapy.
• The clustering revealed a CD161+ positive subset of CD4+ T Cells that were associated with response.
  • A statistically significant increase in TBET and decrease in CCR7 expression were associated with response.
  • This confirms literature findings that these trends in marker expression were associated with higher survival.
• Unsupervised clustering, paired with advanced quality control methods, can overcome the inherent challenges of high-dimensional data analysis, paving the way for deeper insights into immunological complexity.