Fraternal but not identical twins: HLH and ICANS look a lot alike Free

CD19 CAR-T therapy has improved treatment for refractory/relapsed B-ALL, however, toxicities like cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS) and immune-effector cell associated hemophagocytic lymphohistiocytosis (HLH) syndrome (IEC-HS) can occur. There is currently no biomarker to predict patients who will develop toxicity and mechanisms are not fully elucidated. We have previously demonstrated an overlap in biological markers of HLH and ICANS, suggesting that while their clinical presentation is different, they may share similar biological mechanisms. Here we utilized scRNAseq to profile PBMC from B-ALL patients treated with CAR-T cells to understand whether there are immunophenotypes that correlate with severe ICANS.

Longitudinal PBMC samples from patients enrolled in two trials for pediatric B-ALL PLAT02 (CD19 CAR-T, NCT02028455) and PLAT05 (CD19 CD22 CAR-T NCT03330691) were analyzed by scRNAseq. Patients included 2 non-responders and 12 complete responders, of whom 6 developed severe neurotoxicity (grade 3 or higher neurotoxicity/ICANS). One of the non-responders was diagnosed with secondary HLH attributed to the ALL but neither of the non-responders developed ICANS.

We assessed for markers that predict toxicity by evaluating the samples taken prior to apheresis for CAR-T manufacturing (preA samples). We found marked inter-patient variation and no significant differences in transcriptional profiles within cell types when comparing patients with mild versus severe toxicity. When comparing preA PBMC to post CAR-T PBMC, we observed an ablation of all CD19+ cells in responders. However, surprisingly, we observed a high degree of correlation (R2>0.96) in gene expression within each PBMC cell type when we compared preA to Day 28 for all groups of patients. These data suggest that despite undergoing an immunologically significant event, patient baseline phenotype is maintained.

The patient's immunological set-point is driven by multiple factors, including their cancer. Three of the mild toxicity patients had detectable circulating blasts. We observed that the blasts from one patient had IL18 transcripts and the blasts from all three had IL2RA transcripts. These HLH markers were also expressed in monocytes and T-cells, respectively, as expected. These data suggest that the inflammatory milieu can be a product of patient immune as well as leukemia biology. Notably, our data highlights the use of scRNAseq to understand patient heterogeneity in cellular sources of inflammatory cytokines. We hypothesize that the source of inflammation may impact toxicity. Perhaps the patients with circulating blasts experienced only mild toxicity because their inflammation was driven by their cancer, which was eliminated by the CAR-T cells, while patients with severe toxicity may have had additional inflammation driven by their immune phenotype.

To this end, we evaluated a timepoint (Day 7) aligning with toxicity. Patients with mild toxicity had more monocytes than T cells and patients with severe toxicity had more T cells than monocytes, which was corroborated by complete blood counts. This is consistent with traditional mechanisms of HLH, where CD8+/NK cells over-expand. IL-10 is highly upregulated during HLH and during toxicity post CAR-T. IL10 was expressed within the monocytic compartment as expected, but also within the T cell compartment. Given the skew in T cells/monocytes, this indicates that patients with severe toxicity had more IL10 producing T cells. One patient with severe toxicity appeared to be an exception to the T cell/monocyte skew, but we found that their monocytes uniquely expressed an HLH marker (CD163, which enables heme scavenging) suggesting an HLH-like pathological mechanism. These data suggest that, like HLH, neurotoxicity can be driven by different mechanisms, an abundance of activated T cells and/or pathologically activated monocytes.

This study demonstrates the utility of scRNAseq in discovering contributors to inflammation, highlighting that causation is a function of complex biology and patient heterogeneity. These results suggest that, while clinically distinct, HLH and ICANS may in fact share both circulating immune markers and underlying biological mechanisms- albeit at different sites of action. Therefore, we hypothesize that the treatment regimens successfully used for HLH could serve as inspiration for therapies to treat ICANS.