Characterization of evolution trajectory and immune profiling of brain metastasis in lung adenocarcinoma

Characterizing the evolutionary trajectory and immune profiling of brain metastasis (BM) may provide insights in the development of novel therapeutic strategies. Here, we performed whole-exome sequencing and multiplex immunofluorescence (MIF) of 40 samples from 12 lung adenocarcinoma (LUAD) patients with BM and compared to their paired primary tumors. We observed significantly higher intertumor heterogeneity between paired primary tumors and BMs, with only a median of 8.3% of genetic mutations identified as shared. Phylogenetic analysis revealed that BM-competent clones genetically diverged from their primary tumors at relatively early stage, suggesting that the parallel progression model is dominant. In cases with synchronous lymph node metastasis (LNM), phylogenetic analysis suggested that BM is a later event than LNM. MIF analysis found that BMs exhibited significantly lower CD8+ T cell infiltration (P = 0.048), and elevated CD4+Foxp3+ T cell infiltration (P = 0.036) and PD-1 expression (P = 0.047) in comparison to the matched primary tumors, indicating an immunosuppressive microenvironment in BMs. The current study revealed the discrepancy of mutational landscape as well as tumor immune microenvironment between BM and its primary tumor – such findings shall help us better understand the unique biological features of BM and develop innovative strategies accordingly for our patients with LUAD.


Patients' selection
Surgical specimens and biopsy tissues were snap-frozen in liquid nitrogen within 30 minutes of harvest.Then 8-10 μm fresh frozen or formalin-fixed paraffin-embedded (FFPE) sections were made.Hematoxylin-eosin stained slides were reviewed by experienced pathologists to exam the histological subtype and the proportion of malignant cells relative to nonmalignant stromal cells.The criteria for further sequencing analysis of each sample was that the presence of at least 50% tumor nuclei and less than 20% necrosis presented on histological review of each sample.

DNA extraction and library construction
After DNA extraction, the quality check was conducted using Agilent 2100 Bioanalyzer (Life Technologies, USA) per manufacturer's recommended protocol.
Fragmented DNAs were constructed via KAPA Hyper Prep Kit (Illumina platforms, KAPA Biosystems, Massachusetts, USA) according to the manufacturer's protocol.
Multiple indexing adaptors were ligated to the ends of the DNA fragments to prepare them for hybridization onto a flow cell.Purification and size selection of the library were performed using AMPure XP magnetic beads (Beckman Coulter, Brea, CA, USA).The concentration and quality of the library was determined using the Qubit 3.0 system (Invitrogen) and Bioanalyzer 2100 (Agilent, Agilent HS DNA Reagent, 5067-4627).

Whole-exome sequencing
DNA libraries were subjected to whole-exome capture with xGen Exome Research Panel v1.0 (Integrated DNA Technologies), which spans a 39 Mb target region (19,396 genes) of the human genome and covers 51 Mb of end-to-end tiled space.
Human Cot-1 DNA (Life Technologies) and xGen universal blocking oligos (Integrated DNA Technologies) were added as blocking reagents to reduce nonspecific hybridization.The capture reaction was performed using NimbleGen SeqCap EZ Hybridization and Wash Kit (Roche) and Dynabeads M-270 (Life Technologies) according to manufacturers' protocols.The captured samples were sequenced on an Illumina HiSeq X-TEN platform with a paired-end run of 2 × 150 bp.The quality of each read was initially verified using the software embedded in the HiSeq X-TEN sequence.The sequencing depth was more than 150×.A FASTQ file was generated for each tested sample for sequence alignment and converted to a BAM file for further analysis.

Data filtering and variant calling
Tumor-normal paired sample calling was processed during the mutation calling procedure, in order to filter out individual's private germline mutations.The resulted somatic mutations above variant allele frequency (VAF) of 2% were selected for annotation.ANNOVAR was run to screen the nonsynonymous mutations in the exonic region for further study.The candidate variants with a minor allele frequency above 0.2% as recorded in either the population database EXAC (The Exome Aggregation Consortium) v.03 or the Genome Aggregation Database (gnomAD) were filtered out for further identification of the somatic mutations.

Copy number profiling
This CNVkit v0.8.5 takes into consideration of both on-and off-target sequencing reads and applies a series of corrections for the extraneous variability sources like GC content, target footprint size and spacing to improve accuracy in copy number calling.To more precisely infer allele-specific tumor copy number profiles and mutations profiles of each sample, we used Sequenza to evaluate tumor cellularity, ploidy and copy number states among all whole-exome sequenced sample.Briefly, it calculates GC content and processes the sequencing data from the tumor and normal specimens.It calculates the variant alleles and allelic frequency from the tumor specimen via using python script (sequenza-utils).

Phylogenetic tree construction
LICHeE reconstructs the lineage trees and sample heterogeneity by evaluating it on simulated trees of heterogeneous cancer cell lineage evolution.Briefly, identified somatic SNVs from sequencing data was firstly grouped and clustered.Then an evolutionary constraint network was constructed to capture valid evolutionary timing relationships between the mutations of each cluster pair.

Multiplex immunofluorescence staining
Protein blocking was performed using Antibody Diluent / Block (72424205, PerkinElmer) for 10 min at room temperature.Then primary Abs were incubated for 1h at 37℃ or 12 hours at 4℃.The primary Abs were grouped into two 5-antibody panels: panel 1 consisted of CD3 (ZM-0417, clone LN10; Zsbio; dilution 1:50), CD4 MWT was performed to remove the Ab TSA complex with Tris-EDTA buffer (pH 9).TSA single stain slides were finished with MWT and counterstained with DAPI for 5 min and were enclosed in Antifade Mounting Medium (I0052; NobleRyder).