An interim report on the investigator-initiated phase 2 study of pembrolizumab immunological response evaluation (INSPIRE)

Derek Clouthier, Scott Lien, S.y. Cindy Yang, Linh T. Nguyen, Venkata Manem, Diana Gray, Michael Ryczko, Albiruni R. Abdul Razak, Jeremy Lewin, Stéphanie Lheureux, Ilaria Colombo, Philippe L. Bédard, David W. Cescon, Anna Spreafico, Marcus O. Butler, Aaron R. Hansen, Raymond Jang, Sangeet Ghai, Ilan Weinreb, Valentin Sotov, Ramy Gadalla, Babak Noamani, Mengdi Guo, Sawako Elston, Amanda Giesler, Sevan Hakgor, Haiyan Jiang, Tracy L. McGaha, David G. Brooks, Benjamin Haibe‐Kains, Trevor J. Pugh, Pamela S. Ohashi, Lillian L. SiuJournal for ImmunoTherapy of Cancer 2019 · DOI

Abstract: This study provides technical feasibility data for correlative studies. Tissue biopsies provide distinct data from the blood and may predict response to pembrolizumab.

Introduction

Immune checkpoint inhibitors (ICIs), specifically those targeting the PD-1/PD-L1 axis like pembrolizumab, have significantly advanced the treatment of advanced solid tumors. However, objective response rates remain modest, typically around 20% across various indications. Current clinical practice often relies on static biomarkers such as PD-L1 expression or Tumor Mutational Burden (TMB) derived from archival tissue. These samples may not accurately represent the dynamic immune environment at the time of treatment. The INSPIRE study (Pembrolizumab Immunological Response Evaluation) was designed as a phase 2 "basket" trial to address this gap by performing longitudinal, multi-omic analysis of both peripheral blood and fresh tumor biopsies.

Methods and Study Design

The INSPIRE trial enrolled patients across five distinct cohorts: squamous cell carcinoma of the head and neck (SCCHN), triple-negative breast cancer (TNBC), high-grade serous ovarian cancer (HGSC), metastatic melanoma (MM), and other advanced solid tumors (MST). Patients received 200 mg of pembrolizumab every three weeks.

A central feature of the study is the mandatory collection of paired fresh tumor biopsies at baseline and during the early stages of treatment (weeks 6 or 9). These samples, along with serial blood draws, underwent comprehensive analysis including Whole Exome Sequencing (WES), RNA sequencing, and multi-parameter flow cytometry. Target occupancy was specifically assessed using a flow cytometry-based assay to determine the extent of receptor blockade in both compartments.

Study workflow and feasibility of multi-omic analysis from fresh tumor biopsies.

Results

Technical Feasibility and Target Blockade

The interim analysis of the first 80 patients demonstrates the technical feasibility of this intensive sampling approach, though success rates varied by biopsy site. Liver and lymph node biopsies generally yielded higher cell counts for flow cytometry compared to lung or chest wall samples.

A critical finding was the kinetic discordance between peripheral blood and the tumor microenvironment. While PD-1 receptor blockade in the blood reached ≥98% by week 3 in nearly all patients, intratumoral target occupancy was significantly delayed. Only 45% of patients achieved full blockade in the tumor by week 6, increasing to 79% by week 9. This suggests that peripheral saturation does not necessarily equate to therapeutic drug levels within the tumor.

Biomarkers of Clinical Response

Clinical response rates varied significantly by cohort, with the highest response in melanoma (60%) and no responses observed in the ovarian cancer cohort. Analysis of immune phenotypes revealed that the baseline frequency of intratumoral 4-1BB+ PD-1+ CD8 T cells was a significant predictor of clinical response (P < 0.05). The presence of these "primed" T cells, alongside the fold expansion of CD8 T cells during treatment, distinguished responders from non-responders.

Comparison of immune checkpoint expression (PD-1, TIGIT, 4-1BB) in peripheral blood vs. tumor tissue across clinical response groups.

Furthermore, genomic analysis indicated that while TMB was not a significant predictor in this specific cohort, the Percent Genome Altered (PGA)—a measure of large-scale chromosomal instability—correlated with response.

Discussion and Limitations

The INSPIRE study underscores the necessity of analyzing fresh tumor tissue to understand the mechanisms of ICI response and resistance. The disconnect between blood and tumor kinetics highlights the limitations of peripheral biomarkers for predicting intratumoral drug activity.

As an interim report, this study has limitations, including small sample sizes in certain cohorts and relatively short follow-up (median 4.1 months), which precludes definitive survival analysis. Future work will focus on validating the 4-1BB signature through immunohistochemistry and investigating the impact of the gut microbiome on treatment outcomes.

Summary updated

Abstract

This study provides technical feasibility data for correlative studies. Tissue biopsies provide distinct data from the blood and may predict response to pembrolizumab.
BibTeX
@article{clouthier2019interim,
  author = {Derek Clouthier and Scott Lien and S.Y. Cindy Yang and Linh T. Nguyen and Venkata Manem and Diana Gray and Michael Ryczko and Albiruni R. Abdul Razak and Jeremy Lewin and Stéphanie Lheureux and Ilaria Colombo and Philippe L. Bédard and David W. Cescon and Anna Spreafico and Marcus O. Butler and Aaron R. Hansen and Raymond Jang and Sangeet Ghai and Ilan Weinreb and Valentin Sotov and Ramy Gadalla and Babak Noamani and Mengdi Guo and Sawako Elston and Amanda Giesler and Sevan Hakgor and Haiyan Jiang and Tracy L. McGaha and David G. Brooks and Benjamin Haibe‐Kains and Trevor J. Pugh and Pamela S. Ohashi and Lillian L. Siu},
  title = {{An interim report on the investigator-initiated phase 2 study of pembrolizumab immunological response evaluation (INSPIRE)}},
  journal = {Journal for ImmunoTherapy of Cancer},
  year = {2019},
  doi = {10.1186/s40425-019-0541-0},
}
Figure