Tumor Genomic Profiles Associated with Clinical Benefit from Immune Checkpoint Inhibitor Therapy – An Exploratory Analysis of the INFINITY Precision Oncology Registry

Abstract

Background and Purpose

The multicenter precision oncology registry INFINITY investigated biomarker-driven therapy and outcomes in patients with advanced malignancies not eligible for standard therapy in routine clinical care in Germany. Although programmed cell death (ligand) 1 (PD-(L)1) antibodies have changed the treatment landscape of several malignancies and PD-(L)1 expression level is predictive and therefore often used for treatment selection, existing predictive biomarkers are insufficient to identify patients who will benefit from PD-(L)1 inhibitor-based therapy. This project aimed to identify genomic tumor profiles predicting benefit from anti-PD-(L)1 therapy.

Patients and Methods

Patients from the INFINITY registry treated with anti-PD-(L)1 monotherapy were stratified into two cohorts using a case–control design: clinical benefit (treatment duration >182 days) and no clinical benefit (treatment duration 22–63 days). Tumor tissue samples from the virtual biobank were requested from local pathologies and sent to the central pathology for next‑generation sequencing (NGS) using the Oncomine^TM Comprehensive Assay Plus panel (Thermo Fisher, NGS pan-cancer assay).

Results

Of the 26 patients with clinical benefit and 24 patients without clinical benefit, NGS-based genetic tumor profiles were available for 11 and 8 patients, respectively. Only patients with clinical benefit from anti-PD-(L)1 therapy showed genetic alterations in KEAP1, PIK3CA and MRE11.

Conclusion

Our analysis revealed a set of three genes (KEAP1, PIK3CA and MRE11) exclusively mutated in patients with clinical benefit from PD-(L)1-inhibitor therapy, possibly representing promising biomarker candidates. Given the small sample size and the heterogeneous tumor types included, these findings are exploratory, and validation in larger, preferably tumor-type-specific cohorts is required before any predictive value can be attributed to these genomic alterations.

Introduction

Immune checkpoint inhibitors (ICIs) have substantially altered cancer treatment, achieving significant clinical activity and long-lasting tumor responses in many cancer patients.¹ Among ICIs, inhibitors of the programmed cell death protein (ligand) 1 (PD-(L)1) have been shown to achieve reasonable efficacy across multiple tumor entities.^1–3 PD-L1 expression is known to be of predictive value for ICI therapy, but has also been discussed controversially.^4,5 Several biomarkers have been proposed to identify patients most likely to benefit from a PD-(L)1 inhibitor-based therapy: PD-L1 surface expression, tumor mutational burden (TMB), and microsatellite instability (MSI) status.^1,6,7 Additionally, specific patient characteristics, such as being male, aged under 65 years, absence of metastases in brain or liver, and no epidermal growth factor receptor (EGFR) mutation, have been associated with better clinical outcomes with anti-PD-(L)1 inhibitors.⁸ However, despite success of ICIs in several tumor types, most patients do not derive any long-term benefit, and it is of great interest to identify patients who benefit from and to understand the mechanism of resistance to ICIs.^1,9–11

Clinico-genomic databases have evolved as essential tools in the field of precision oncology, providing biological samples for biomarker analyses and linked with high-quality clinical outcomes data - especially in the real-world setting.^12,13 The multicenter precision oncology registry INFINITY investigated biomarker-driven treatment and clinical management of patients with advanced malignancies not eligible for standard treatment options in routine clinical care in Germany. Here we present an exploratory case–control study of the genomic tumor profiles of patients with clinical benefit from PD-(L)1 antibody treatment compared to those without. This analysis uses the virtual biobank infrastructure of a real-world precision oncology registry to link genomic tumor profiling with clinical outcomes data, demonstrating the feasibility and potential of such an approach for hypothesis-generating biomarker research.

Patients and Methods

INFINITY Precision Oncology Registry

The INFINITY precision oncology registry is a retrospective, non-interventional, multicenter cohort study with the aim of enrolling 500 patients with advanced solid tumors or hematologic malignancies who received a non-standard targeted therapy (NSTT) based on an actionable molecular alteration or biomarker from approximately 100 sites (office-based oncologists/hematologists and hospitals) across Germany. The overall goal is to provide insights into the implementation and the effectiveness of precision oncology in routine clinical practice. Automatic checks, manual central monitoring, as well as on-site monitoring are conducted to guarantee high‑quality, comprehensive clinical data. The INFINITY registry has been described in detail.^14,15 Besides collecting data on patient and disease characteristics as well as clinical outcomes, a virtual tissue sample biobank was set up for future translational research. The INFINITY registry was approved by the responsible ethics committee (Landesärztekammer Baden-Württemberg, reference number F-2019-088). Written informed consent was obtained from all patients alive at the time of registration. The study protocol of the INFINITY precision oncology registry has already been published.¹⁴ The INFINITY registry is registered at ClinicalTrials.gov (NCT04389541).

INFINITY Biobank: Research Question and Goals

The aim of this research project was to exploratively compare genomic signatures in tumor tissue samples from patients treated with a PD-(L)1-inhibitor-based therapy with or without clinical benefit, in order to generate hypotheses on potential biomarker candidates. The study was approved by the responsible ethics committee (Landesärztekammer Baden-Württemberg, reference number F-2021-094). For all patients participating in the INFINITY registry, who provided written informed consent for collection and analyses of their tumor tissue samples or who were deceased at the time of registration, tumor tissue samples collected during clinical routine were registered in the virtual biobank. Storage location (local pathology) of the samples, respective sample ID and key sample data (eg, sample date and type) were documented by the study site in an electronic data capture (edc) system. The logistics and distribution of tasks for this multidisciplinary biobank project are shown in Figure S1.

Patient Population and Sample Selection

Eligible patients for this research project were adults (≥18 years) with a solid tumor or hematologic malignancy identified from the INFINITY clinical database and who received a PD-(L)1 antibody monotherapy as NSTT. Patients were assigned to either one of the two groups: Patients with clinical benefit, defined as treatment duration of PD-(L)1 antibody monotherapy >182 days; and patients without clinical benefit, defined as treatment duration 22–63 days. Patients with a treatment duration <22 days were not considered, because these patients received only 1 cycle of therapy, and it is likely that their disease was in such an advanced stage that no treatment benefit could be expected. Patients with treatment duration between 63 days and 182 days were omitted as well to generate two distinctly separated groups in terms of treatment benefit and prognosis. We employed this case–control design based on clinical outcomes to maximize our ability to identify potential differences between responders and non-responders to PD-(L)1 antibody monotherapy. We deliberately selected patients with distinctly different outcomes to increase the contrast between groups. By focusing on these “extreme” cases - those with clear clinical benefit versus those without - we aimed to enhance our chances of uncovering meaningful differences in patient characteristics, biomarkers, or other factors that may be associated with treatment response. While we acknowledge that this approach may introduce selection bias and limit generalizability, we believe it serves as an efficient method for hypothesis generation. Essentially, we are “learning from the winners” (and the non-responders) by concentrating our analysis on the extremes of the outcome spectrum.

In case more than one sample per patient was documented in the virtual biobank, most recent samples were preferred, and resection samples were chosen over biopsy samples. All tissue samples were formalin-fixed and paraffin-embedded (FFPE). The samples were requested from local pathologies and shipped for next-generation sequencing (NGS) analysis to the central pathological institute (Institute of Pathology and Medical Genetics, University Hospital Basel, Switzerland).

Next Generation Sequencing (NGS)

The central pathology conducted quality control checks of the samples by histological review and determination of tumor cell content on a hematoxylin/eosin-stained tissue slide. If the sample qualified for analysis, nucleic acid (DNA and RNA) was extracted from the tumor cells micro-dissected from a tissue slide. Quality and quantity assessment of the nucleic acid was performed prior to sequencing.

NGS was performed using the Oncomine^TM Comprehensive Assay Plus panel (NGS pan-cancer assay), Ion 550^TM Chips and Ion 550^TM Chef Reagents (Thermo Fisher, Template preparation kit). Data analysis and variant calling was performed on Ion Reporter Software v5.18 using the Oncomine Comprehensive Plus - w2.3 - DNA - Single Sample workflow. Data preparation, annotation, and clinical interpretation of the sequencing results were conducted by Molecular Health GmbH using the MH VCF Adapter Suite software, version 2.1.0, and the IVDR-certified MH Guide software, version 5.3.0.

TMB values were reported as mut/Mb and categorized into three groups: low (TMB ≤5 mut/Mb), intermediate (5 mut/Mb < TMB < 20 mut/Mb) and high TMB (TMB ≥20 mut/Mb).¹⁶ Microsatellite status was categorized into high microsatellite instability (MSI-high) and microsatellite stable (MSS) based on the detection of microsatellite markers with the OCA Plus Assay according to the respective parameters of the Ion Reporter^TM Software.

Statistical Analysis

A descriptive analysis was performed. Statistical analysis was performed using SAS^® software (SAS Institute, Cary, NC, USA), version 9.4 of the SAS System for Windows. The statistical program R version 4.0.5 (www.r-project.org) was used for creating tables and figures.¹⁷

Results

Sample Selection, Patient and Tumor Characteristics

At the database cut for this research project on September 30, 2022, a total of 384 patients had been included in the INFINITY registry and had given their written informed consent to collect and analyze their respective tumor tissue sample(s) for future translational research. All of these tissue samples were primarily deposited and stored at 116 different local pathologies in Germany. The virtual biobank of the INFINITY registry captures data on tissue sample(s), names and addresses of the local pathologies in an electronic database (Figure S2).

For the present analysis, we identified a total of 50 eligible samples, 26 samples from patients with clinical benefit and 24 from patients without clinical benefit from PD-(L)1 antibody treatment (Figure 1). Upon shipment request, 15 of 31 local pathologies sent 25 samples to the central pathology laboratory in Basel. Following sample preparation and quality checks, we conducted NGS on 19 samples: 11 samples from patients with and 8 samples from patients without clinical benefit (Figure 1).

Figure 1.

Flow chart. Flow chart of all patients and samples of the INFINITY virtual biobank assessed for eligibility, requested from local pathologies, and finally analyzed in the full analysis set (FAS). * For this patient the treatment duration was corrected in retrospect, rendering the sample ineligible for data analysis.

Patients with clinical benefit (n=11) were older (median 69.1 vs 57.1 years), had a longer time from initial diagnosis to treatment start with PD-(L)1 antibody (median 39.9 vs 32.5 months), and had fewer prior therapy lines compared to patients without clinical benefit (n=8). Samples were derived from 12 different tumor types. Patient and tumor characteristics are shown in Table 1.

Table 1.

Patient and Tumor Characteristics at Time of Start of PD-(L)1 Antibody Treatment

Characteristics	Clinical Benefit (N=11)	No Clinical Benefit (N=8)
Sex
Female	5 (45.5%)	3 (37.5%)
Male	6 (54.5%)	5 (62.5%)
Age at start of PD-(L)1 antibody treatment
Median, years (range)	69.1 (55.3–82.7)	57.1 (31.3–83.3)
<70	6 (54.5%)	7 (87.0%)
≥70	5 (45.5%)	1 (12.5%)
ECOG performance status
0-1	10 (90.9%)	5 (62.5%)
2	1 (9.1%)	2 (25.0%)
3-4	0 (0.0%)	1 (12.5%)
Time from initial diagnosis to start of
PD-(L)1 antibody treatment
Median, months (range)	39.9 (8.5–72.8)	32.5 (9.5–94.9)
Number of prior palliative therapy lines
0	1 (9.1%)	1 (12.5%)
1	6 (54.5%)	1 (12.5%
2	2 (18.2%)	1 (12.5%)
3	2 (18.2%)	3 (37.5%)
4	0 (0.0%)	0 (0.0%)
≥5	0 (0.0%)	2 (25.0%)
Tumor entity
Colorectal cancer	1 (9.1%)	2 (25.0%)
Endometrial cancer	2 (18.2%)	0 (0.0%)
CUP – Squamous Cell carcinoma	1 (9.1%)	1 (12.5%)
Esophageal cancer	0 (0.0%)	2 (25.0%)
Penile cancer	2 (18.2%)	0 (0.0%)
Pleuramesothelioma	1 (9.1%)	0 (0.0%)
Biliary tract cancer	0 (0.0%)	1 (12.5%)
Cervix carcinoma	1 (9.1%)	0 (0.0%)
Gastric cancer	0 (0.0%)	1 (12.5%)
Head and neck cancer	1 (9.1%)	0 (0.0%)
Soft tissue sarcoma	1 (9.1%)	0 (0.0%)
Uterine cancer/Endometrial cancer	1 (9.1%)	0 (0.0%)
Vulva carcinoma	0 (0.0%)	1 (12.5%)

Note: Numbers are n (%), unless indicated otherwise.

Abbreviations: PD-(L)1, programmed cell death (ligand) 1; ECOG, Eastern Cooperative Oncology Group.

Patients who benefited from PD-(L)1 blockade most often achieved stable disease (SD) (45.5%) or partial response (PR) (27.3%), while the best overall response for patients without clinical benefit was mainly progressive disease (PD) (62.5%) (Table 2).

Table 2.

Best Overall Response to PD-(L)1 Antibody Treatment

	Clinical Benefit (N=11)	No Clinical Benefit (N=8)
Best overall response during first NSTT
Complete response (CR)	1 (9.1%)	0 (0.0%)
Partial response (PR)	3 (27.3%)	0 (0.0%)
Stable disease (SD)	5 (45.5%)	0 (0.0%)
Progressive disease (PD)	2 (18.2%)	5 (62.5%)
Not evaluable	0 (0.0%)	1 (12.5%)
Missing	0 (0.0%)	2 (25.0%)

Note: Numbers are n (%), unless indicated otherwise.

Abbreviations: PD-(L)1, programmed cell death (ligand) 1; NSTT, non-standard targeted therapy; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.

Genomic Profile – Results of Central NGS Testing

The interpreted NGS data have been sorted by clinical benefit status, showing only genes with alterations occurring in at least 5 patients and depicted as an oncoplot (Figure 2). The tumor mutational burden (TMB) differed between patients with vs without clinical benefit from PD-(L)1 antibody therapy: six vs two patients harbored low TMB (54.5% vs 25.0%), three vs six patients intermediate TMB (27.3% vs 75.0%) and two vs none patients (18.2%) high TMB (Table 3). High MSI was identified in two patients: one with colorectal cancer, high TMB and clinical benefit, and one with esophageal cancer, intermediate TMB, and no clinical benefit (Figure 2).

Figure 2.

Oncoplot. Oncoplot showing the genetic profile of the 19 analyzed samples. Samples from patients with clinical benefit to PD-(L)1 antibody treatment (green) are shown in the upper part (above the dashed line), those from patients without clinical benefit (red) in the lower part. Only genes with alterations in at least 5 samples are shown.

Table 3.

Results of NGS Testing – Summary of Specific Genetic Alterations

	Clinical Benefit (N=11)	No Clinical Benefit (N=8)
Tumor mutational burden (TMB)
TMB-high (TMB ≥20 mut/Mb)	2 (18.2%)	0 (0.0%)
TMB-intermediate (5 mut/Mb < TMB < 20 mut/Mb)	3 (27.3%)	6 (75.0%)
TMB-low (TMB ≤5 mut/Mb)	6 (54.5%)	2 (25.0%)
Microsatellite instability (MSI)
MSI-high	1 (9.1%)	1 (12.5%)
MSS	10 (90.9%)	7 (87.5%)
Genetic alterations unique to one group
KEAP1 mutation	5 (45.5%)	0 (0.0%)
PIK3CA mutation	5 (45.5%)	0 (0.0%)
MRE11 mutation	5 (45.5%)	0 (0.0%)

Note: Numbers are n (%), unless indicated otherwise.

Three genes were identified with alterations solely in the clinical benefit group: KEAP1, PIK3CA, and MRE11 harbored mutations in 5 patients each, all of them with clinical benefit from PD-(L)1 antibody therapy; such mutations were not found in patients without clinical benefit. Uniquely, one patient with penile cancer harbored all three of these alterations. This patient achieved a partial response as the best overall response (data not shown). While all mutations in KEAP1 and MRE11 were known to be of uncertain significance or unclassified, 3 of 5 mutations detected in PIK3CA were pathogenic or likely pathogenic (Figure 2).

Discussion

The biobank of the INFINITY precision oncology registry enables the setting up of translational research projects based on a clinico-genomic database of solid and hematologic malignancies in real-world Germany. In this exploratory case–control study, we aimed to identify differences in the genetic tumor profiles of patients with versus without clinical benefit from anti-PD-(L)1 therapy. We identified alterations present exclusively in patients with clinical benefit from anti-PD-(L)1 therapy, ie, mutations in the genes KEAP1, PIK3CA and MRE11. However, sample numbers were small, and these findings should be considered exploratory and hypothesis-generating rather than evidence for a predictive role of these genes.

We identified only 2 patients with high MSI status among 19 samples from 12 tumor types, one with colorectal cancer and one with esophageal cancer; both patients were also characterized by high or intermediate TMB, respectively. This is in line with previous reports, where high MSI generally occurred as a subset of high TMB.¹⁸ In this report, the vast majority (97%) of MSI-high samples was characterized by TMB ≥10 mut/Mb, but only 16% of samples with high TMB were classified as MSI-high.¹⁸ The co-existence of TMB and MSI-high status seems to depend on the cancer type: while MSI-high status was detected in 27 of 39 cancer types, the prevalence varied widely (endometrial 31.4%, colon 19.7%, stomach carcinoma 19.1%, and very rare or not detected at all in other entities, such as thyroid carcinoma or pancreatic adenocarcinoma).¹⁹ This shows a clear heterogeneity of MSI-high status across diverse tumor entities and is comparable to our results.

It has been demonstrated that cancers of diverse histology with documented MSI-high had an objective response rate of 53% and a disease control rate of 77% when treated with pembrolizumab, which lead to the first tumor-agnostic approval of pembrolizumab.²⁰ In our study, 1 of 2 patients with MSI-high status achieved clinical benefit (colorectal cancer patient).

The three genes with mutations exclusively in the sub-group of patients with clinical benefit from PD-(L)1 antibody treatment, KEAP1, PIK3CA and MRE11, warrant further investigations as to their predictive value. KEAP1, a ubiquitin ligase, plays a central role in the cell response to oxidative stress. Mutations in the KEAP1 gene, a tumor suppressor gene, and their association with response to PD-(L)1 antibody therapy have been studied especially in NSCLC; however, the results so far are controversial and KEAP1 mutations, for instance, may rather act as a prognostic but not as a predictive biomarker.^21–23 Our findings give supporting evidence to further investigate KEAP1 and the response to PD-(L)1 antibody treatment.

The PI3K/AKT pathway controls most hallmarks of cancer, including the immunosuppressive microenvironment.^24,25 Some studies have proposed a positive predictive value of PIK3CA mutations for PD-(L)1 antibody treatment, for example in cervical cancer,²⁶ TP53‑mutated gastric cancer,²⁷ colon adenocarcinoma²⁸ and microsatellite stable colorectal cancer.²⁹ Also in our study, alterations of PIK3CA were classified as pathogenic or likely pathogenic and detected only in the group of patients with clinical benefit to PD-(L)1 antibody treatment.

MRE11 is a double-strand break repair protein and is part of a complex which has a crucial role in DNA damage response.^30,31 The association between DNA damage response alterations and response to ICIs is of research interest in various cancer types.³² To our knowledge, there is no information on the role of MRE11 mutations as predictors of response to ICIs so far; thus, our results on MRE11 mutations warrant further investigations into their predictive value.

Taken together, precision medicine in real-world settings will be of high relevance in the future, and especially clinic-genomic databases will be of utmost importance to study the effectiveness of biomarker-driven therapies in terms of survival as well as predictors of clinical benefit and durable responses versus poor efficacy and resistance. Precision medicine programs such as the German national Network Genomic Medicine lung cancer (nNGM) for patients with advanced non-small cell lung cancer (NSCLC),³³ ie the large German cohort study to evaluate the effectiveness of a structured precision medicine program in patients with the initial diagnosis of advanced NSCLC,³³ or real-world databases such as the INFINITY registry will provide deep insights into precision oncology in routine care and will facilitate the implementation of precision medicine in the real-world setting with the goal to improve clinical outcomes.

Limitations

This is one of the first case–control studies conducted using the infrastructure of the INFINITY registry. First, it was a bit to our surprise that we were only able to collect a relatively small number of archived samples for NGS analyses, and we can only speculate on the reasons thereof. One explanation might be that the respective pathology units had no material left, because of use for other projects we are not aware of. This is certainly an aspect that needs to be improved for future projects. Second, we used a selected patient group based on pre-defined criteria for clinical benefit. Clinical benefit or lack thereof was defined based on PD-(L)1 antibody treatment duration. However, since patients with a wide range of tumor entities and varying numbers of prior therapy lines were included, treatment duration may be an imprecise parameter for determination of clinical benefit. Furthermore, this case–control design introduces an inherent selection bias. Due to these limitations, including tumor heterogeneity, the imprecision of treatment duration as a clinical benefit surrogate and selection bias, the results must be interpreted with caution. All analyses are purely descriptive, and hypothesis testing is not feasible. Finally, although the most recent samples available were chosen, the analyzed samples may not necessarily be the same samples that were relevant for therapy decision making.

Conclusion

With this research project we successfully collected tumor samples from local pathologies for central molecular testing and data analysis using the virtual biobank of the INFINITY precision oncology registry. 50% of requested samples were provided, 34% were ultimately included in the final analysis. Genetic profiles of known biomarkers for PD-(L)1 antibody treatment, TMB and MSI-high status were in line with published data. Three genes were identified with mutations exclusively occurring in the sub-group of patients with clinical benefit from PD-(L)1 antibody treatment, KEAP1, PIK3CA and MRE11. All three qualify as interesting leads for further potential investigations. Although sample numbers were too small to draw general conclusions, this exploratory, hypothesis-generating analysis provides valuable preliminary insights. The identified mutations require validation in larger, preferably tumor-type-specific cohorts before any predictive value can be attributed to these genomic alterations. The infrastructure we have established and tested during this study will certainly enable and foster other translational projects.

Funding Statement

Oncomine^TM Comprehensive Assay Plus (NGS pan-cancer assay), Ion 550^TM Chips, and Ion 550^TM Chef Reagents (Template preparation kit) were provided by Thermo Fisher Scientific Inc. Thermo Fisher Scientific Inc. The INFINITY clinical research platform was financially supported by Roche and BMS. None of the funders had any role in study design, data collection and analysis, interpretation of results, decision to publish, or preparation of the manuscript.

Abbreviations

CTLA-4, Cytotoxic T-lymphocyte-associated protein 4; CR, Complete response; ECOG, Eastern Cooperative Oncology Group; EDC, Electronic data capture; FFPE, Formalin-fixed paraffin-embedded; ICI, Immune checkpoint inhibitor; Mb, Megabase; MMRd, Mismatch repair deficiency; MSI, Microsatellite instability; MSS, Microsatellite stable; Mut, Mutations; NSTT, Non-standard targeted therapy; NGS, Next-generation sequencing; PD, Progressive disease; PD-1, Programmed cell death protein 1; PD-L1, Programmed cell death ligand 1; PR, Partial response; SD, Stable disease; TMB, Tumor mutational burden.

Data Sharing Statement

The data supporting the findings of the study are not openly available due to data privacy protection regulations. Enquiries regarding the data used for this study can be directed to the corresponding author.

Ethics

All experiments comply with the current laws in Germany and Switzerland, where they were performed. All procedures performed in studies involving human participants were in accordance with the ethical standards of the national research committee and with the 1964 Helsinki declaration and its later amendments. Written informed consent was obtained from all patients who were alive at the time of registration.

Disclosure

Outside of the published work, the institutions of L. Sellmann, J. Schröder, T. Decker, M. K. Schuler, D. Marschner, M. Zaiss, M. Reiser and M. Neise received remuneration for the documentation of patient data.

U. M. Martens received grants from the Dieter Schwarz Foundation, serves on advisory boards for Sanofi-Aventis, Novartis, Roche, and MSD, received payment for a presentation from BMS and travel support from Beigene, Ipsen, Pierre Fabre, Eli Lilly, Janssen, and holds stocks in Eli Lilly and Novo-Nordisk.

M.S. Matter has served as a consultant for Thermo Fisher, Merck, GlaxoSmithKline, Janssen-Cilag, Roche, Novartis, and received a speaker’s honorary from Incyte Biosciences and Astellas.

P.M. Jermann received grants from Thermo Fisher Scientific and Tecan, and was sponsored for NGS reagents, equipment and materials, payment for lectures, and support for attending meetings from Thermo Fisher Scientific.

H. Läubli has research collaborations with Ono Pharmaceuticals and Omnilinx Therapeutics and is co-founder and board member of Glycocalyx Therapeutics; received research support from Ono Pharmaceuticals and Omnilinx Therapeutics and honoraria for lectures from Novocure; receives royalties for a Siglec-9 antibody; submitted a patent for using fucosidase as a cancer treatment.

T. Decker received honoraria for lectures and advisory boards from AstraZeneca, Lilly, Novartis and Roche.

M. Zaiss received payments or honoraria for lectures and presentations from AstraZeneca, AbbVie, Roche, Lilly, Pfizer, and Takeda; participated on advisory boards for AstraZeneca, AbbVie, Novartis, Lilly, Esai, Roche, Pfizer, Gilead, Jannsen, and BMS.

M. A. Rauschendorf is an employee of Molecular Health GmbH.

B. Kasenda received consulting fees from Dayton Therapeutics and honoraria for lectures or presentations from Roche and Incyte.

N. Marschner is Chief Scientific Officer and shareholder of iOMEDICO AG.

The authors report no other conflicts of interest in this work.