Monocyte-lineage tumor infiltration predicts immunoradiotherapy response in advanced pretreated soft-tissue sarcoma: phase 2 trial results

Abstract

Immunoradiotherapy holds promise for improving outcomes in patients with advanced solid tumors, including in soft-tissue sarcoma (STS). However, the ideal combination of treatment modalities remains to be determined, and reliable biomarkers to predict which patients will benefit are lacking. Here, we report the results of the STS cohort of the SABR-PDL1 phase II trial that evaluated the anti-PDL1 atezolizumab combined with stereotactic body radiation therapy (SBRT) delivered concurrently with the 2nd cycle to at least one tumor site. Eligible patients received atezolizumab until progression or unmanageable toxicity, with SBRT at 45 Gy in 3 fractions). The primary endpoint was one-year progression-free survival (PFS) rate with success defined as 13 patients achieving 1-year PFS. Sixty-one heavily pretreated patients with STS (median 5 prior lines; 52% men; median age 54 years; 28% leiomyosarcoma) were enrolled across two centers (France, Spain). SBRT was delivered to 55 patients (90%), with the lung being the most commonly irradiated site (50%). After a median follow-up of 45 months, the one-year PFS rate was 8.3% [95% CI: 3.6-18.1]. Median PFS and overall survival were 2.5 and 8.6 months, respectively. Best responses included partial responses (5%) and stable disease (60%). Immune profiling revealed increased immunosuppressive tumor-associated macrophages (e.g., IL4I1, HES1) and monocyte-recruiting chemokines in non-responders. Higher monocyte/lymphocyte ratios (MonoLR) in tumor and blood correlated with progression. PD-L1 status, lymphoid infiltration, and tertiary-lymphoid structures were not predictive. Although the primary endpoint was not met, this study highlights MonoLR imbalance as a potential biomarker to identify STS patients likely to benefit from immunoradiotherapy. EudraCT No. 2015-005464-42; Clinicaltrial.gov number: NCT02992912.

Fig. 1

SABR-PDL1 trial design (NCT02992912)

Fig. 2

CONSORT diagram

Fig. 3

Clinical outcomes: a Overall survival. b Progression-free survival, and c Waterfall plot showing the best response and best change from baseline in size of target lesions in evaluable patients

Fig. 4

Tumor infiltration with immunosuppressive cells of the monocytic lineage after SBRT associates with progression. a and b CD68/CD163 co-labeling assessed by multiplexed immunohistochemistry on tumor FFPE biopsies. CD68+/CD163− (a) and immunosuppressive CD68+/CD163+ (b) cell densities were automatically quantified by the HALO® software on digitalized slides. Are displayed: cell density at baseline according to treatment response (left panel), cell density according to biopsy timepoint among patients with progressive disease (middle panel), and among patients with stable or responsive disease (right panel). Blue and yellow crosses refer, respectively, to patients with Progressive and Stable+Elite disease. Mean ± 95%CI. Kruskal–Wallis with Dunn’s multiple comparisons test. c Differential gene expression analysis of RNA-seq data comparing transcription profiles of tumor biopsies from patients who progressed with patients who showed stable disease or tumor response among 1775 selected genes. Genes differentially expressed between groups with adjusted P-value < 0.05 are indicated by yellow or blue crosses, respectively when downregulated or upregulated in progressive patients. Relevant genes implicated in immune regulation are depicted. d–f Heatmap depicting scores per immune cell type (y-axis) according to subgroups classified according to treatment response and biopsy timepoint (x-axis) using two independent published deconvolution tools, CIBERSORTabs (f) and xCELL (e). Scores are normalized per cell type based on the calculation: mean of the subgroup/total mean. Blue indicates higher values than the mean of samples, while pink indicates lower values. Stars indicate significant P-values obtained from the comparison of ’Progressive’ versus ‘Stable+Elite’ among samples obtained at the indicated timepoint, Kolmogorov–Smirnov test. f Idem, deconvolution of bulk RNA-seq data using the MoMac-VERSE obtained from single-cell RNA-seq description of 17 mononuclear cell types. Scores are normalized per cell type based on the calculation: mean of the subgroup/total mean. Blue indicates higher values than the mean of samples, while pink indicates lower values. Stars indicate significant P-values obtained from the comparison of ’Progressive’ versus ‘Stable+Elite’ among samples obtained at the indicated timepoint, Kolmogorov–Smirnov test. Crosses indicate missing values. *P-value ≤ 0.05; **P-value ≤ 0.01; ***P-value ≤ 0.001

Fig. 5

Higher tumor and blood monocyte/lymphocyte ratio in non-responders. a Peripheral lymphocyte counts from left to right at baseline, cycle 1, cycle 2, and cycle 3, according to the response group (blue: ‘Progressive’ group; yellow: ‘Stable+Elite’ group). Mean ± 95%CI, Kolmogorov–Smirnov test. b Monocyte/lymphocyte ratio (MonoLR) computed on blood values (monocyte absolute count/lymphocyte absolute count) from left to right at baseline, cycle 1, cycle 2, and cycle 3, according to the response group. Mean ± 95%CI, Kolmogorov–Smirnov test. c Tumor monocytic-lineage/lymphocyte ratio (MonoLR_tumor) obtained from RNAseq data with CIBERSORTabs and EPIC, according to the response group. Mean ± 95%CI, Kolmogorov–Smirnov test. d Correlation plot between blood MonoLR (x-axis) and tumor MonoLR values by EPIC (y-axis). Blue crosses refer to progressive patients while yellow crosses refer to patients with stable or responsive disease. The dotted line fits with the simple linear regression of equation Y = 0.3401*X−0.07988. e and f Kaplan–Meier curves of progression-free survival (PFS, e) and overall survival (f) from left to right: using baseline blood MonoLR in the training cohort, using baseline blood MonoLR in the validation cohort, and using baseline tumor MonoLR as estimated by the EPIC algorithm. Black lines refer to patients with baseline MonoLR < indicated threshold whereas green lines refer to patients with baseline MonoLR ≥ threshold. Logrank hazard ratio and P-values are depicted