Finally, while he's not on Bluesky, I want to highlight that this work wouldn't be possible without truly special mentorship from my PI, Ben Izar, who has continuously motivated me to pursue exciting, challenging, and clinically meaningful projects like this. A true MD/PhD role model and mentor.

This work would not have been possible without generous funding to the Izar Lab, from @columbiacancer.bsky.social Human Tissue Immunology and Immunotherapy Initiative (HTI3), the Pharming Group, the NIH and NCI, and @cancerresearchinst.bsky.social

This work was an incredible team effort with many folks from the Izar Lab: Chris Frangieh, Neeha Kothapalli, Johannes Melms, Clarissa Heck, and many more. We are so grateful for wonderful collaboration and guidance from disease experts Josh Milner and Dusan Bogunovic @bogunoviclab.bsky.social

This work has already impacted care: a patient was treated with leniolisib based directly on our variant classification. And it serves as a useful blueprint for other IEIs and beyond ... Stay tuned!

How relevant are these variants? By integrating our screen with All of Us (>600K genomes) and an IEI cohort (>8K pts), we found 100+ carriers enriched for immune disease phenotypes. So the disease could be markedly common than we thought...

Some PIK3R1 variants were less responsive to leniolisib. We identified combo therapies (e.g. leniolisib + mTORi) that restored suppression in both edited cells and patient samples.

We validated screen results and phenotypes using T cells from two APDS patients with variants mapping to structural GOF hotspots flagged by our screen, directly linking screen data to real disease.

Otherwise-healthy T cells engineered these newly identified GOF variants showed hyperactivation, exhaustion & transcriptional changes—mirroring APDS patient cells. Like patients, they also responded to leniolisib, the FDA-approved targeted inhibitor.

But, more importantly the result: reliable recovery of known pathogenic variants, along with dozens of novel GOF variants previously unstudied or classified as VUS

Side note... anyone who has done phosflow and/or CRISPR screens will appreciate the insanity of doing phosflow on hundreds of millions of primary T cells at once for a pooled screen 🫠

Usually these are tested one at a time... Instead we used CRISPR base editing to engineer thousands of PIK3CD/PIK3R1 variants in primary T cells, and mapped these to pAKT/pS6—a readout used to guide variant classification in real patients.

NGS often reveals variants of uncertain significance (VUS), stalling care and precision therapy. In Activated PI3Kδ Syndrome (APDS), an inborn error of immunity, PI3Kδ inhibitor leniolisib is effective but only approved for pts w/ proven pathogenic GOF variants...