IDO1 degrades tryptophan into immunosuppressive kynurenines and has long been cast as a tumor-cell escape mechanism — yet IDO1 inhibitor trials have largely failed, often in unselected patients. Analyzing 891 lung tumors from two randomized atezolizumab trials plus multiplex imaging and spatial transcriptomics, this study flips the story: in NSCLC, high IDO1 marks an inflamed, T-cell-rich tumor and predicts better response to PD-L1 blockade, not worse. IDO1 is driven by the immune system, expressed by tumor cells and by follicular dendritic cells inside tertiary lymphoid structures, where its metabolite blunts B-cell antibody production. The implication: IDO1 inhibitors may work only in inflamed, TLS-bearing tumors.
This translational study in Clinical Cancer Research, first-authored by Dr. Alban Bessede (Explicyte) and led by Prof. Antoine Italiano, set out to resolve a paradox: whether IDO1 expression helps or hurts patients with non–small cell lung cancer (NSCLC) on PD-1/PD-L1 blockade. The team reanalyzed tumor transcriptomes from the randomized POPLAR (phase 2) and OAK (phase 3) atezolizumab-versus-docetaxel trials, then validated findings at the protein and functional level. The work was funded by the Conseil Régional de Nouvelle-Aquitaine (BIP AIRIC grant). Explicyte ran essentially the full experimental program — RNA-seq reanalysis and immune deconvolution, GeoMx DSP spatial transcriptomics, multiplex immunofluorescence and digital-pathology analysis on the PhenoImager HT, single-cell data reanalysis, and the ex vivo functional coculture assays.
Reanalyzing baseline transcriptomes of 891 NSCLC tumors from the POPLAR and OAK trials, Explicyte found that high IDO1 expression predicted longer PFS (median 3.8 vs 1.89 months, P < 0.001), longer overall survival (HR 1.73), and higher response — but only in the atezolizumab arm, not the docetaxel arm. IDO1 remained independently associated with outcome on multivariate analysis, marking it as predictive rather than merely prognostic.
IDO1 correlated positively with PD-L1, and IDO1-high tumors were enriched for IFN-γ-producing Th1 cells, CD8+ T cells, NK cells, and B cells. On a 7-color multiplex-IF panel across 53 ICI-treated NSCLC tumors, tumor-cell IDO1 tracked with more activated CD8+/PD-1+ T cells and a shorter distance between tumor cells and the nearest CD8+ cell — direct spatial evidence that IDO1 flags a T-cell-inflamed, “hot” tumor.
Using a 9-color multiplex-IF panel and confirming with single-cell data, Explicyte showed IDO1 is expressed within tertiary lymphoid structures specifically by CD21+/CD23+ mature follicular dendritic cells — the first report of IDO1 in TLS. GeoMx DSP spatial transcriptomics of TLS segmented into CD20+ and CD20− regions linked high intra-TLS IDO1 to more plasmacytoid dendritic cells and regulatory T cells, and TLS with high proliferating Tregs tracked with worse response and survival.
To test function, the team used ex vivo cocultures of tonsil-derived follicular helper T cells and memory B cells. L-kynurenine — IDO1’s main metabolite — dose-dependently suppressed the Tfh-driven IgG secretion that is a hallmark of productive TLS, and separately blunted T-cell-mediated killing of H1299 lung cancer cells and IFN-γ release. This positions IDO1 as a negative-feedback brake on TLS-driven antitumor immunity.
The study reframes IDO1 in NSCLC from a tumor-intrinsic villain to an immune-driven marker of inflammation — reconciling why past IDO1 inhibitor trials failed and pointing to the patients who might actually benefit.
For drug developers, this is a patient-selection thesis for a stalled drug class: IDO1 inhibitors have disappointed in all-comer trials, and this work argues their benefit should be sought specifically in T-cell-inflamed, TLS-bearing NSCLC rather than cold tumors. It also nominates intra-TLS IDO1 and proliferating-Treg content as candidate biomarkers to enrich such trials, and makes the mechanistic case for combining IDO1 blockade with PD-1/PD-L1 therapy to reinvigorate TLS-driven immunity. The spatial and functional toolkit here — GeoMx DSP, multiplex IF, Tfh/B-cell coculture — is directly reusable for that patient-selection work.
