Introduction
Immune checkpoint blockers (ICB), such as anti-PD-1/PD-L1, have become a standard treatment in several tumor locations. The benefit of immunotherapy has been largely driven by a subset of patients experiencing durable tumor responses: overall, about 15% to 60% of patients respond to immunotherapy-based approaches, depending on tumor type, tumor mutational burden features (e.g., DNA mismatch repair-deficient [dMMR]/microsatellite instability-high [MSI-H]) and PD-L1 expression levels [
1‐
3]. In several unselected tumor subtypes, such as in advanced colorectal cancer (CRC), studies evaluating ICB showed disappointing results, with median progression-free survival [PFS] and overall survival [OS] of 2.2 and 5.0 months, respectively, in a cohort with proficient MMR (pMMR) colorectal cancers treated with pembrolizumab [
3]. Similarly, within the MSI-H/dMMR subpopulation of patients with CRC (which accounts for less than 5% of all advanced CRC), median PFS ranged from 2.3 to 4.1 months according to the number of prior lines [
4].
An increasing number of preclinical [
5‐
7] and clinical [
8‐
11] data suggest that radiation therapy can enhance the immune anticancer response. Higher access to stereotactic body radiotherapy (SBRT) now allows to sharply target tumor lesions, reducing collateral damages to adjacent organs including lymph nodes [
6], while inducing immunogenic cell death, which promotes a T-cell-mediated immune response against antigens derived from dying tumor cells. Ionizing radiations also enhance the expression of MHC-I molecules favoring antigen presentation and activate the interferon (IFN) cGAS-STING DNA-sensing pathway, contributing to the amplification of a tumor-directed adaptive immune response.
Conversely, radiation therapy may also promote immunosuppressive effects including lymphocyte exhaustion and subsequent PD-L1 upregulation, attraction of immunosuppressive cells (e.g.: myeloid cells and regulatory T cells), immunosuppressive cytokines release, and/or radiation-induced lymphopenia (RIL) [
5‐
7]. RIL is frequent across all tumor types, often lasts several months after completion of radiotherapy, and has been shown to directly impact survival outcomes [
6,
12,
13]. For now, the best way to use radiation therapy to enhance immunotherapy efficacy with limited radio-induced immunosuppressive effects remains unclear.
More clinical and deep irradiated tumor data are needed for optimizing the combination regimens and the selection of patients who would best benefit from immuno-radiotherapy combinations. In this single-arm phase 2 study, we assessed the safety, efficacy and biological correlate analysis of SBRT in combination with the PD-L1 inhibitor atezolizumab in advanced pretreated cancer patients. Here, we report the results of the CRC cohort.
Patients and methods
Study design and participants
In this international, multi-center single-arm phase 2 trial, we enrolled different cohorts of patients including advanced CRC, renal cell carcinoma, non-small cell lung cancer (NSCLC) and sarcomas who had at least 1) one lesion eligible for SBRT and 2) one unirradiated lesion, both measurable by Response Evaluation Criteria in Solid Tumors, version 1.1 (RECIST v1.1). Additional key eligibility requirements included Eastern Cooperative Oncology Group Performance Status (PS) of 0 to 1, absence of autoimmune or immunodeficiency diseases, and adequate organ function. CRC patients should have been considered in treatment failure as per the current standard recommendation. We placed no limit on the number of prior therapies, although patients were not eligible if they had received prior PD-1 or PD-L1 inhibitor. Patients were eligible regardless of their PD-L1 or molecular target/MSI status. The trial was approved by the relevant ethics/institutional review board and was completed in accordance with international standards of good clinical practice. All patients provided written informed consent at the time of enrolment.
Procedures
Intravenous atezolizumab therapy, 1200 mg, was administered every 21 days (3 weeks). Hypofractionated SBRT was delivered concurrently with the 2nd cycle (week 6, Figure S
1) at an ablative dose, using 6MV photons with standard field encompassing tumor. SBRT was delivered at a recommended dose of 45 Gy in three fractions of 15 Gy (equivalent biologic dose (BED) > 100 Gy). The protocol allowed adapted doses based upon normal tissue tolerance constraints. The radiation dose was prescribed to the 90% isodose line in order to deliver 95% of the planned dose to 95% of the planned tumor volume (PTV). SBRT was applied to at least one tumor, and another untreated tumor site was required to be evaluable by RECIST 1.1. Central thoracic and brain lesions were not eligible for SBRT in this study but they could be considered as “not treated” evaluable metastases. Treatment with atezolizumab continued for up to two years in the absence of documented disease progression, unacceptable adverse events, intercurrent illness precluding further administration of treatment, the investigator’s decision to withdraw the participant, participant withdrawal of consent, pregnancy of the participant, or non-adherence with trial treatment.
Immune profiling
After having signed a dedicated informed consent for translational research purposes, patients underwent sequential tumor biopsies of the irradiated lesion at baseline, week 3 (pre-SBRT) and week 7 (post-SBRT) for biomarker analysis. In some cases, lesions strictly outside of the radiation field were also biopsied at the week 3 and week 7 timepoints to study potential abscopal impacts of SBRT. Tumor samples were both formalin fixed paraffin embedded (FFPE) and freshly frozen, respectively for immunohistochemistry (IHC) analysis and whole RNA extraction and sequencing. For each sample, tumor cellularity was assessed by a senior pathologist on a haematoxylin–eosin-saffron (HES)-stained slide from the FFPE-preserved biopsy. Samples with no tumor cells were excluded from the analyses.
FFPE blocs were sliced into 4-μm large sections to perform IHC multiplexing. An immune-targeting panel was developed at the experimental and translational pathology (PETRA) platform of Gustave Roussy according to the following (chromogenic library): 2Plex CD163/CD68 (anti-CD163 ref Mob460-05 clone 10D6 from DBS; anti-CD68 ref M0876 clone PG-M1 from DAKO), 4Plex CD8/PD-L1/FoxP3/cytokeratin (CK) (anti-CD8 ref 05937248001 clone SP57 from Roche; anti-PD-L1 ref 7994190001 clone SP263 from Roche; anti-FoxP3 ref ab99963 clone SP97 from Spring; anti-CK ref Mob190.05 clone AE1-AE3 from DBS) and 4Plex IRF1/CD20/CD3/CD68 (anti-IRF1 ref 8478 clone D5E4 from Cell Signaling; anti-CD20 ref M075501-2 clone L26 from DAKO; anti-CD3 ref A0452 polyclonal from DAKO; anti-CD68 ref M0876 clone PG-M1 from DAKO). Once stained, slides were digitalized at 20X using a VS120 scanner (Olympus Life Science). Finally, cellular densities of label-positive cells were automatically assessed using the HALO® image analysis software, in-situ hybridization module.
Tumor whole RNA was extracted retrospectively by batch using the AllPrep RNA Mini Kit (Qiagen) following the manufacturer’s instruction. RNA extracts were sent to Novogene to perform human mRNA sequencing after rRNA removal on an Illumina NovaSeq 6000 system, PE150, Q30 ≥ 85%. Raw data recorded as FASTQ files were processed by the bioinformatics platform of Gustave Roussy including quality control, preprocessing, aggregation/normalization steps and differential expression analyses. Unsupervised hierarchical clustering was performed on the log2-transformed TPM value differences between week 3 (pre-SBRT) and baseline samples considering 536 immune-related genes among the previously described LM22 gene set [
14]. Euclidean distances were used. Immune deconvolution from bulk RNA sequencing data was performed using the CIBERSORTx tool [
15]; “abs” option was used to obtain absolute scores compatible with inter-sample comparison.
Objectives and outcome measures
The primary efficacy endpoint was the one-year PFS rate. PFS was defined from the start of atezolizumab treatment to the first documented disease progression, death due to any cause. Response status was based on investigator assessment of scans using RECIST 1.1. Scans of all previously involved disease sites were performed at week 4, 7, 13, and then every 12 weeks or as clinically indicated. Secondary end-points included safety, as defined by Common Terminology Criteria for Adverse Events (CTCAE v4.03) toxicity profile, overall survival (OS), and objective response rates (ORRs). Patients who had a PFS of more than one year were described as “elite responders”.
Statistical analyses
Each cohort was analyzed separately. In each cohort, a Fleming 1-stage design was applied to demonstrate that the PFS rate at one year is not inferior to 15% but could reach 32%. To test in each cohort, the hypothesis that the PFS rate is greater than p0 = 15% with alpha = 0.033 and with a 90% power to detect activity greater than p1 = 32%, 54 evaluable patients should be enrolled. If 13 patients or more are alive and free of progression at one year in the cohort, the combined SBRT + atezolizumab is considered a success for the corresponding histology. To account for possibly 10% of non-evaluable patients, 60 patients were enrolled per cohort. P ≤ 0.05 was considered statistically significant. Statistical analyses were performed using SAS software, version 9.4 or R software, version 3.0.2 (R Foundation for Statistical Computing).
Translational data were analyzed using Prism V.9 (GraphPad, California, USA). Unless otherwise stated, multigroup comparisons were done according to an ordinary one-way ANOVA calculation followed by Tukey’s multiple comparisons test. For simple comparison analyses, an unpaired Student’s t-test with Welch’s correction was used to compare data when assuming Gaussian distributions with unequal standard deviation across groups, while the Mann–Whitney test was used for non-parametric testing.
Discussion
In this international single-arm phase II trial, we assessed the safety and the efficacy of SBRT combined with atezolizumab in unselected advanced CRC. Treatment was well tolerated with no unexpected toxicity (no grade 3 or more SBRT-related AE). The median PFS was modest (1.4 months [95% CI: 1.4- 2.6]; 6.3 months in MSI-H patients) in this unselected and pretreated population, as previously reported [
3,
4]. However, five (8%) “elite” patients had their disease controlled for longer than one year, two of whom were pMMR, which suggests that SBRT might have boosted the efficacy of atezolizumab in those patients.
Single or double agent ICB is considered an ineffective approach for pMMR CRC, with median PFS not exceeding 2.5 months in previous reports [
3,
16]. Presumed mechanisms include the low antigenicity [
17] and the high immunosuppression of those tumors, with a possible influence of liver metastases that have been shown to affect the overall ICB efficacy. For example, in a retrospective study performed on 95 patients with pre-treated pMMR CRC receiving an anti-PD1/PDL1 therapy, the overall response rate was of 19.5% in patients without liver metastases, and no response (0%) was observed among the 54 patients with liver metastases [
18]. The median PFS was also significantly higher in patients without liver lesions (4.0 months vs 1.5 months, respectively for patients without and with liver metastases). Patients with hepatic lesions may also experience an impaired tumor infiltration with (cytotoxic) T-cells, a phenomenon also called hepatic syphoning [
19]. One hypothesis underlying this observation is that the liver microenvironment could be directly immunosuppressive through the increase of Treg and myeloid-derived suppressive cells (MDSC) [
20] and subsequently, triggering the depletion of cytotoxic CD8
+ T and NK cells [
21]. Considering this, liver-directed ablative radiotherapy could represent a promising option to relaunch a systemic antitumor immunity in patients receiving ICB [
21].
Our study provides new data on the tumor infiltrate characterization and its dynamic throughout immuno-radiotherapy treatment, which may help selecting advanced CRC patients most likely to respond to immuno-radiotherapy regimens. First of all, immune-centric multiplex IHC and RNAseq suggested that SBRT redirected immune cells towards tumor lesions, even in the case of RIL (Fig.
3). This is consistent with what was observed in another phase II study in which patients with pretreated p-MMR advanced CRC received durvalumab-tremelimumab and radiotherapy. In this study, median PFS was of 1.8 months [95% CI 1.7–1.9 months],
n = 21/24 included patients had flow cytometry on peripheral blood mononuclear cells (PBMC) and authors could observe that CD8 + T lymphocytes were activated only among the two responders [
22]. In addition, in our analysis, we could notice that the change of RNAseq expression of immune-related genes between before and after the start of atezolizumab occurred in opposite directions according to the response group (Fig.
5A). This suggests that primarily-resistant immune systems may be identified as soon as the second cycle of atezolizumab. Importantly, T-cell interferon activation and increased expression of chemotaxis signals such as CXCL9 were observed in responding patients. Similar findings were shown in a SBRT-pembrolizumab prospective cohort of 68 patients with advanced tumors. The same work reported that elevated expression of
TGFb correlated with less tumor responses [
23]
. Other studies showed that radiation therapy and TGFβ inhibition could increase immune infiltration [
24] and clinical response [
25]. Interestingly, a recent trial assessing the efficacy of the bispecific antibody bintrafusp alfa (TGFβ-trap and anti-PD-L1) in patients with liver-limited MMRp mCRC was stopped early for loss of equipoise after 4 enrolled patients (out of 15 planned) [
26]. This suggests that the presumed deleterious impact of liver metastases may not be counteracted by the dual blockade of TGFβ and PD-L1.
This study has limitations. The true effect of SBRT is difficult to assess given the single arm design. The trial started in 2016 and the type of irradiation used (concomitant high dose-fractionation mostly delivered to a single site) might not be the more immunogenic regimen. Although debated [
27,
28], recent reports suggest that multisite irradiation delivered at a lower dose (e.g. 3 × 8 Gy) before ICB initiation could be more effective [
5‐
7]. Some other biomarkers such as circulating tumor DNA [
26], immunoscore [
29], aneuploidy [
30] or tumor mutational burden (TMB) were not assessed. In a phase II trial including 40 patients with MSS CRC, Parikh et al. did not find differences of TMB according to response after the combination of radiation therapy, ipilimumab and nivolumab [
31]. We could not integrate out of field tumor heterogeneity in the analysis due to the small size of most tumor biopsy samples that we retrieved. Alternatively, this could have been guided by imaging feature analysis using artificial intelligence-guided methods such as radiomics [
32], although this would have required a larger sample size to provide more tangible information. By comparison, a study in patients with HPV-unrelated head and neck squamous cell carcinomas described the high-dimensional multi-omics and spatial data analysis assessed on surgical specimens (
n = 21) in patients treated in a neoadjuvant phase I/II trial of SBRT with single-dose anti-PD-L1 durvalumab [
33]. Responders displayed an increase in post-treatment effector T cells and antigen presentation
.
Acknowledgements
The author would like to thank the promotion team at Gustave Roussy (Flora Ngadjeua Tchouatieu, Nadia Zaghdoud, Souad Cosse, Salim Laghouati and Geoffray Gengizalp). The authors thank collaborators from the histopathology department of Gustave Roussy and from the Experimental and Translational Pathology (PETRA, AMMICA, INSERM US23/CNRS) Platform at Gustave Roussy, including Catherine Genestie, Jean-Yves Scoazec, Laetitia Bordelet, Virginie Marty and Nicolas Signolle. The authors also wish to thank collaborators from the Bioinformatics core facility of Gustave Roussy, including Marc Deloger, Marine Aglave and Thibault Dayris.
Declarations
Competing interests
E.D. reports grants and personal fees from Roche-Genentech, grants and personal fees from AstraZeneca, grants and personal fees from Merck Serono, grants and personal fees from Boehringer, grants and personal fees from BMS, and grants and personal fees from MSD.
N.M. reports grants and personal fees from Merck Serono, grants and personal fees from Bayer, and grants and personal fees from MSD.
A.L. reports grants for academic research from PharMamar, Beigene, Roche, AstraZeneca and Amgen.
L.V. reports personal fees from Adaptherapy, is CEO of RESOLVED, has received non-personal fees from Pierre-Fabre and Servier, and a grant from Bristol-Myers Squibb, all outside the submitted work. Research Grants from Astrazeneca, BMS, Boehringer Ingelheim, Celsius, EIT Philips, GSK, INCA, IDERA, Janssen, Lombard, Merck, MedImmune, Pierre Fabre, Roche, Sanofi, Servier. Non-financial support (drug supplied) from Astrazeneca, BMS, Boringher Ingelheim, GSK, Idera, Medimmune, Merck, NH Theraguix, Roche.
M.R. reports receiving research funding from Roche and Highlight Therapeutics. She also has received speaker’s bureau honoraria from BMS and ROCHE.
RS received support from Fondation Bettencourt-Schueller (CCA Inserm-Bettencourt 2020).
C. Quevrin was funded by Ligue Contre le Cancer (Ref IP/SC #17563).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.