Association of tertiary lymphoid structures with outcomes from PD-1 inhibition in advanced malignant pleural mesothelioma.

Abstract

e20545 Background: Predictive biomarkers of benefit form immunotherapy in malignant mesothelioma patients remain poorly defined. The 9-15 PROMISE-MESO study (NCT02991482) of the European Thoracic Oncology Platform (ETOP) evaluated second-line pembrolizumab against chemotherapy. Pembrolizumab led to an increased response rate compared to chemotherapy, which did not translate to survival benefits. We evaluated if the presence of tertiary lymphoid structures (TLSs) with different compositions or LRRC15+ immune suppressive fibroblasts could predict the clinical benefit from pembrolizumab. Methods: We performed multiplex immune histochemistry analysis on tumor tissues. 144 patients were included in the clinical trial, from whom, 112 had available tumor tissue for analysis: 55 patients in the chemotherapy arm, and 57 patients in the immune therapy arm. We performed the following panel: calretinin or WT1 (mesothelioma cells), CD20, CD23, CD11c, LRRC15, aSMA, and DAPI. We used the Akoya Vectra Polaris platform for analysis. The IFQuant analytical pipeline developed at the Swiss Institute for Bioinformatics (SIB) was used for analyses. Results: We defined three categories of TLSs: general CD20+ TLSs independent of other markers. CD20/CD23 double-positive as mature TLSs (mTLS), and TLSs with dendritic cells (CD11c+, cTLS). Immune suppressive CAFs were defined as LRRC15+ cells/mm2. CD20+ TLSs were identified in 84% of patients' stroma, with a median of 21 TLSs/tumor tissue and 2 TLSs/mm2. mTLSs were rare, only detected in 10 of the 112 patients (9%). cTLSs with dendritic cells were more frequent and present in 45% of all patients (50/112), and roughly half of the TLS+ patients (53%) had CD11c+ cells detected in TLSs. LRRC15+ cancer-associated fibroblasts (CAFs) were detected in 85% of patients, with a median density of 61 cells/mm2. LRRC15- tumors also contained TLSs but had significantly fewer than LRRC15+ tumors (p = 0.0079). We did not find any significant association between the responses or PFS with TLS, mTLS, and LRRC15+ CAFs. We found inverse association between cTLS levels and response to immune therapy (p = 0.037) but not for PFS. Conclusions: Our data shows that advanced malignant mesotheliomas frequently harbor TLSs, but these TLSs are mostly not activated/mature. The absence of antigen-presenting cells in TLSs, like CD11c dendritic cells, is unexpectedly associated with increased response to pembrolizumab. The high prevalence of immune suppressive LRRC15+ CAFs might explain the low response rate to pembrolizumab and could be linked to the absence of maturation of TLSs. Clinical trial information: EudraCT number: 2016-002062-31 .