Immune checkpoint inhibitors work in soft tissue sarcomas rich in tertiary lymphoid structures, but TLS-negative "cold" tumors rarely respond. The PEMBROSARC basket study tested whether intra-tumoral injection of the TLR4 agonist G100, added to low-dose cyclophosphamide, could sensitize cold STS to pembrolizumab. Explicyte's multiplex immunofluorescence profiling of sequential biopsies showed CD4 and CD8 T-cell infiltration rose in more than half of patients — yet the CD8-to-Treg ratio fell in most cases, only 2 of 17 patients stayed progression-free at 6 months, and inflaming the tumor did not translate into clinical benefit.
This cohort of the PEMBROSARC basket study, run by the Early Phase Trials Unit at Institut Bergonié and led by Prof. Antoine Italiano, asked whether a toll-like receptor 4 (TLR4) agonist could make immunologically “cold” soft tissue sarcoma responsive to PD-1 blockade. Patients received intra-tumoral G100, low-dose oral cyclophosphamide, and intravenous pembrolizumab (NCT02406781). The work was funded by MSD, the French Ministry of Health (PHRC), Institut National du Cancer (INCa), the Association pour la Recherche contre le Cancer (ARC), and the Agence Nationale de la Recherche (RHU CONDOR); G100 was supplied by ImmuneDesign. Explicyte contributed the multiplex immunofluorescence immune-contexture profiling of the paired baseline and on-treatment biopsies and ran plasma proteomics (Olink panel).
Twenty patients with metastatic, pretreated soft tissue sarcoma and a superficial injectable lesion were enrolled between February 2019 and December 2020. All had TLS-negative (“cold”) disease. Treatment paired 50 mg cyclophosphamide orally twice daily (one week on, one week off), 200 mg pembrolizumab IV on day 8 of a 21-day cycle, and weekly 20 µg intra-tumoral G100 for 6 to 12 weeks, with the first G100 injection one week before cyclophosphamide.
Of 17 patients assessable for efficacy, only 2 were progression-free at 6 months, giving a 6-month non-progression rate of 11.8% (95% CI: 1.5–36.4) — below the study’s bar. Five patients (29.4%) showed tumor shrinkage: 1 partial response and 4 stable disease. Median PFS was 1.8 months and median overall survival 10.6 months.
Explicyte ran multiplex immunofluorescence on paired baseline and cycle 2 day 1 biopsies from 14 patients. Intra-tumoral CD8+ T-cell infiltration increased in 7 patients (50%) and CD4 T-cell infiltration in 8 (57%) after G100, confirming the agonist could push a subset of cold tumors toward inflammation.
Despite rising T-cell density, the CD8/FoxP3+CD4 ratio fell in 11 of 14 patients on treatment, pointing to predominant regulatory T-cell induction rather than a productive effector response. Separately, high circulating soluble PD-L1 at baseline was the only protein significantly associated with worse PFS and overall survival (Olink panel).
The cohort shows that inflaming a cold sarcoma is achievable but not sufficient: what matters is whether the incoming T cells are effector- or regulatory-skewed.
For developers pursuing “cold-to-hot” immunotherapy strategies, this cohort is a cautionary data point: TLR4 agonism is double-edged, and a measurable rise in tumor T cells can mask a Treg-dominated response that yields no clinical benefit. Immune-monitoring endpoints for such trials should quantify effector-to-regulatory balance, not infiltration alone. Baseline soluble PD-L1 also emerges as a candidate prognostic biomarker worth prospective evaluation in sarcoma immunotherapy.
