1174 Utility of immune cell receptor repertoires as predictors of development and mechanism of immune checkpoint inhibitor induced pneumonitis

<h3>Background</h3> Pneumonitis is an immune-related adverse event (irAE) associated with immune checkpoint inhibitor (ICI) therapy that is increasing in incidence and carries a risk of hospitalisation, treatment complexity and mortality. Biomarker monitoring, early identification and novel, mechanistically driven treatment options would improve outcomes. <h3>Methods</h3> Patients enrolled on the translational HYST study (IRAS 105915) who developed pneumonitis and other irAEs were identified. Samples taken at the onset of post-ICI irAEs were evaluated along with patient-matched baseline samples and tolerant controls. Following PBMC gDNA and RNA extraction, high throughput sequencing of the TCR and BCR adaptome was conducted using iR-RepSeq+.¹ RNA also underwent immunophenotyping. Libraries were multiplexed and sequenced on the Illumina NextSeq 1000 P1 600-cycle. iRepertoire IR-Map software and Kruskal-Wallis testing were used for analysis. <h3>Results</h3> Samples were grouped: baseline ‘no-tox’ (n = 5), baseline ‘tox’ (n = 10), post-treatment ‘no-tox’ (n = 5), and post-treatment ‘tox’ (n = 10). Within the two ‘tox’ groups there were pneumonitis (n=5) and hepatitis (‘other-tox’) (n=5). Several significant differences between those with and without toxicity were observed including the largest clone percentage in both TRD and IGK and several class-switch recombination indices on IgA, IgD, IgG and IgM (p &lt; 0.05). Compared to ‘no-tox’, both ‘tox’ groups showed a significant subset of expanding and contracting CDR3 peptide sequences for receptor chains TRA, TRB, IGK, and IGL (edgeR, FDR &lt; 0.05) (figures 1,2). The largest clone percentage of TRG showed significant baseline differences. Pneumonitis generally showed a trend towards lower immune repertoire diversity at baseline compared to controls (figure 3). Unsupervised learning of 146 immunophenotyping genes revealed strong two-cluster sample separation with 49 genes showing inter-cluster significance (FDR &lt; 0.05). 5/5 pneumonitis, 4/5 hepatitis and 3/5 baseline pneumonitis samples fell into a single cluster. The most impactful genes included S100A8, ITGAL, CD247, and RUNX3 (FDR = 3.64E-04). <h3>Conclusions</h3> Comparing samples across different timepoints/toxicity status’ regarding pneumonitis/irAEs illustrated four main findings in this exploratory cohort. There was a baseline difference in the largest clone percentage of TRG suggesting the potential for prediction; there were differentiating TCR and BCR repertoire cohorts between ‘tox’ and ‘non-tox’ groups; only the ‘tox’ group illustrated repertoire time-dynamic changes and delineating signals in immunophenotyping were suggestive of mechanistic features including S100A8, which is infection-associated and hypothetically, would link pre-infection and pneumonitis. Whilst this is a small sample size there are several significant findings that may be factors in determining predisposition and mechanistic propagation of pneumonitis/irAEs which warrant further investigation. <h3>Acknowledgements</h3> This analysis was funded by F. Hoffmann-La Roche Ltd and Biotech Support provided by iRepertoire Inc. <h3>Reference</h3> Han J, Lotze MT. The adaptome as a biomarker for assessing cancer immunity and immunotherapy. <i>In Methods in Molecular Biology</i> 2020;<b>2055</b>:369–397. Springer. https://doi.org/10.1007/978-1-4939-9773-2_2. <h3>Ethics Approval</h3> Ethic approval within A Mechanistic Investigation into Drug and Chemical Induced Hypersensitivity Reactions (IRAS 105915) with UK Ethics COmmittee. All samples were obtained from patients following informed, written consent.