Trial design and data reporting
The primary objective of this open-label, first-in-human, phase 1, three-arm umbrella trial (ClinicalTrials.gov: NCT02316457) was to separately assess the feasibility, safety, and tolerability profile of two different mRNA–LPX-based vaccine types: an off-the-shelf warehouse vaccine composed of non-mutated TAAs and an on-demand manufactured individualized neoantigen vaccine. Vaccine-induced antigen-specific immune responses were investigated (secondary endpoint). Here, we report findings related to the individualized neoantigen vaccine arm of this trial. The trial was carried out in Germany and Sweden in accordance with the Declaration of Helsinki and Good Clinical Practice Guidelines, and with approval by the independent ethics committees (Ethik-Kommission of the Landesärztekammer Rheinland Pfalz, Mainz, Germany and Regionala Etikprövningsnämnden, Uppsala, Sweden) and the competent regulatory authority (Paul-Ehrlich Institute, Langen, Germany and Medical Products Agency, Uppsala, Sweden). All patients provided written informed consent.
Eligibility criteria were: histologically confirmed invasive adenocarcinoma TNBC (pT1cN0M0–[any]T[any]NM0); previous standard of care treatment (that is, neoadjuvant chemotherapy of the primary tumour followed by surgery or surgery and adjuvant chemotherapy); prior radiotherapy was allowed; at least 18 years of age; adequate haematopoietic, hepatic and renal function; tumours expressing at least 5 neoantigens. Patients were eligible for enrolment within one year after completion of standard of care therapy per local policy (for example, surgery and/or chemotherapy and/or radiotherapy). Key exclusion criteria were the recurrence of breast cancer prior to the start of trial treatment and presence of clinically relevant autoimmune disease or active viral infections. The individualized neoantigen vaccine consisted of two single-stranded RNA molecules each encoding up to ten neoantigen vaccine targets selected based on somatic mutation analysis of each patient’s tumour. RNAs were liposomally formulated into RNA–LPX for intravenous administration as described11. For patients treated with neoadjuvant chemotherapy, the FFPE tumour sample from the diagnostic core biopsy was used for analysis of tumour antigen expression. In exceptional cases (for example, low sample quality), the resected FFPE tumour tissue from surgery could be used for analysis of antigen expression instead. agCapture 3.4.2.6 (ArisGlobal) was used for electronic data capture.
The clinical trial report of the primary and secondary endpoints assessed in the main study phase until end of treatment and a 56-day follow-up period in 2020 was submitted to health authorities in spring 2021. Beyond this follow-up, the three patients pretreated with bridging TAA vaccine were followed up passively; one patient consented to provide blood samples via a research project. The following 11 patients (only treated with neoantigen vaccine) participated in an active long-term follow-up for 3 years until 2023. The data generated in the long-term follow-up have been summarized in an addendum to the clinical trial report and submitted to health authorities in the spring of 2024. Following this follow-up period, these patients were then followed up passively.
Next generation sequencing
Tumour DNA was extracted from three 10 μm curls of FFPE tumour tissue in duplicates using a modified version of QIAamp DNA FFPE Tissue kit (Qiagen). RNA extraction was done in duplicates using the ExpressArt Clear FFPE RNAready from AmpTec. For DNA extractions from PBMC cells, the DNeasy Blood and Tissue Kit (Qiagen) was used. Extracted nucleic acids were used for generation of various NGS libraries. Targeted RNA-seq libraries were constructed in duplicate from FFPE tumour tissue RNA using the NEBNext RNA First Strand Synthesis Module and NEBNext Ultra Directional RNA Second Strand Synthesis Module for cDNA syntheses and a modified version of SureSelect XT V6 Human All Exon (Agilent) using 100 ng total RNA input. DNA whole-exome libraries were constructed in duplicates from 100 ng of FFPE tumour DNA and matching PBMC DNA using a modified version of SureSelect XT V6 Human All Exon (Agilent). NGS libraries for whole-exome sequencing of the tumour and matching PBMCs were prepared by fragmenting 100 ng genomic DNA in a total volume of 15 μl using microTUBE-15 AFA Beads Screw-Cap (Covaris) to an average fragment length of approximately 150 bp. For NGS, the libraries were diluted to 3 nM and clustered at 10 pM using the Illumina HiSeq 3000/4000 PE Cluster Kit. Pooled exome library from FFPE and PBMC DNA were sequenced as 4-plexes on three lanes, whereas the RNA library replicates were sequenced as 2-plexes in one lane. All libraries were sequenced paired-end 50 nt on an Illumina HiSeq 4000 platform using two HiSeq 3000/4000 SBS 50 cycles kits (Illumina). For P12, post-vaccination samples were paired-end (100 nt) sequenced on an Illumina NovaSeq 6000 platform using an Illumina NovaSeq 6000 S2 Flow Cell and an Illumina NovaSeq 6000 S2 Reagent Kit v.1.5 (100 Cycles) instead.
Bioinformatics and mutation discovery
All genomics-related data analysis steps were coordinated by proprietary bioinformatic pipeline (BioNTech) implemented in the Python programming language and described in brief below. For each of the replicate of the DNA libraries, at least 180 × 106 paired-end 50 nt reads were available covering ≥70% of targeted bases with ≥100× coverage. For the RNA libraries a minimum of 75 × 106 paired-end 50 nt reads were required.
For mutation detection, DNA reads were aligned to the reference genome hg19 with bwa (v.0.7.10)32. The resulting alignment files were converted to BAM format using SAMtools (v.0.1.19)33. Somatic SNVs were called using an in-house mutation caller4 and short insertions and deletions were called using Strelka34, by comparing the aligned reads of the tumour DNA to those of the matching PBMC DNA.
Genomic coordinates of identified somatic variants were compared with the UCSC Known Genes transcript coordinates to associate the variants with genes, transcripts, and potential amino acid sequence changes. Synonymous and non-sense mutations were filtered. For verified and non-synonymous cancer mutations that were selected as potential neoantigen vaccine targets, a mutated peptide sequence (MPS) was determined based on the mutated transcript sequence for contribution to design of a MPS concatemerized vaccine. In the case of SNVs, these were the 27mer peptide regions with the changed amino acid in the centre. For insertions and deletions resulting in a frameshift, the MPS featured the sequence from the changed amino acid to the next stop codon (maximum 50 amino acids). Germline variants in the region of the mutated peptides were identified using SAMtools based on the matching PBMC DNA. Protein changing germline variants were first phased based on the RNA-seq reads and, if in-phase with the somatic mutations, included in the patient-specific neoantigen vaccine target.
... continue reading