Nanopore-based random genomic sampling for intraoperative molecular diagnosis

Abstract

Background

Central nervous system tumors are among the most lethal types of cancer. A critical factor for tailored neurosurgical resection strategies depends on specific tumor types. However, it is uncommon to have a preoperative tumor diagnosis, and intraoperative morphology-based diagnosis remains challenging. Despite recent advances in intraoperative methylation classifications of brain tumors, accuracy may be compromised by low tumor purity. Copy number variations (CNVs), which are almost ubiquitous in cancer, offer highly sensitive molecular biomarkers for diagnosis. These quantitative genomic alterations provide insight into dysregulated oncogenic pathways and can reveal potential targets for molecular therapies.

Methods

We develop iSCORED, a one-step random genomic DNA reconstruction method that enables efficient, unbiased quantification of genome-wide CNVs. By concatenating multiple genomic fragments into long reads, the method leverages low-pass sequencing to generate approximately 1–2 million genomic fragments within 1 h. This approach allows for ultrafast high-resolution CNV analysis at a genomic resolution of 50 kb. In addition, concurrent methylation profiling enables brain tumor methylation classification and identifies promoter methylation in amplified oncogenes, providing an integrated diagnostic approach.

Results

In our retrospective cohort of 26 malignant brain tumors, iSCORED demonstrated 100% concordance in CNV detection, including chromosomal alterations and oncogene amplifications, when compared to clinically validated assays such as Next-Generation Sequencing and Chromosomal Microarray. Furthermore, we validated iSCORED’s real-time applicability in 15 diagnostically challenging primary brain tumors, achieving 100% concordance in detecting aberrant CNV detection, including diagnostic chromosomal gains/losses and oncogene amplifications (10/10). Of these, 14 out of 15 brain tumor methylation classifications aligned with final pathological diagnoses. This streamlined workflow—from tissue arrival to automatic generation of CNV and methylation reports—can be completed within 105 min.

Conclusions

The iSCORED pipeline represents the first method capable of high-resolution CNV detection within the intraoperative timeframe. By combining CNV detection and methylation classification, iSCORED provides a rapid and comprehensive molecular diagnostic tool that can inform rapid clinical decision. The integrated approach not only enhances the accuracy of tumor diagnosis but also optimizes surgical planning and identifies potential molecular therapies, all within the critical intraoperative timeframe.

Background

Maximal safe surgical resection is the primary treatment for central nervous system (CNS) tumors. Surgical resection reduces tumor burden, and the obtained diagnostic specimen guides subsequent treatment decisions [1, 2]. A major factor to determine the aggressiveness of resection depends on the tumor type. For instance, gross total resection enhances prognostic outcomes in atypical teratoid rhabdoid tumor [3], isocitrate dehydrogenase (IDH)-wildtype glioblastoma of receptor tyrosine kinase (RTK) I and RTK II [4], and IDH-mutant astrocytoma [5], but not in others like diffuse midline gliomas [6], WNT, SHH and group 3 medulloblastoma [7]. Therefore, accurate diagnosis is crucial for guiding neurosurgical decisions. Preoperative imaging modalities can not reliably distinguish between glioma entities. Current clinical practice relies on intraoperative examination of frozen specimens. However, this approach is limited by overlapping tumor morphology, inter-observer variability, and the absence of molecular information, resulting in non-specific or inaccurate provisional diagnoses [8]. Recently, tumor diagnosis has been advanced by the characterization of altered genome-wide DNA methylation signatures [9,10,11]. In conjunction with machine-learning methods, ultrafast methylation analysis using sequencing-based techniques is emerging as a powerful tool for intraoperative brain tumor classification [12]. However, the accuracy and concordance with the final pathological diagnosis are frequently hindered by low tumor purity [10, 12], which is a common phenomenon of gliomas due to their infiltrative growth patterns [13].

Copy number variation is a predominant type of quantitative genomic alterations involved in a wide range of biological processes, including human evolution [14,15,16], neurodegeneration [17, 18], and developmental disorders [14, 19, 20]. In particular, CNVs contribute to cancer development and progression by altering gene dosage, thereby activating oncogenes and inactivating tumor suppressor genes [21,22,23]. Molecular profiling of CNVs provides insight into tumor aggressiveness and is integral to pathological diagnosis and grading [13]. For instance, the presence of 1p/19q co-deletion is required for the diagnosis of oligodendroglioma according to WHO classification [13]. High CNVs adversely impact overall survival in IDH-mutant astrocytoma with high copy number variation load, CDKN2A/B homozygous deletion, RB1 homozygous deletion, MYCN amplification, and PDGFRA amplification [13, 24, 25]. Moreover, CNVs precede histological findings in cases of discrepancy: for instance, grade 4 glioblastoma can be diagnosed when there is combined chromosomal 7 gain/10 loss or EGFR amplification, even in the absence of typical high-grade histological features [13, 26]. Finally, oncogene amplification suggests potential therapeutic strategies, such as targeting tyrosine kinase receptors and related pathways in EGFR-amplified glioblastomas [27, 28] (ClinicalTrial.gov). Thus, the importance of characterizing CNV extends beyond tumor diagnosis to inform therapeutic interventions.

Current common CNV methodologies include fluorescence in situ hybridization (FISH) [29], nucleotide hybridization [15, 17, 19, 30,31,32], and next-generation sequencing [18, 21,22,23]. These techniques necessitate high-complexity centralized laboratories with a turnaround time of several days to weeks. In addition to the high costs, the prolonged process could delay clinical molecular diagnosis and therapeutic plans [33, 34]. Nanopore sequencing (Oxford Nanopore Technologies, ONT) is an emerging sequencing technology enabling real-time interpretation of long-read nucleotide sequences. Its advantages include low setup fee, compact size, and straightforward library preparation. Taking advantage of real-time data availability during sequencing, there have been some successes in ultrafast CNV diagnostics [35, 36] using sparse sequencing to analyze short fragmented DNA or native long DNA. However, the intraoperative application is limited by the low genomic resolution, which restricts detection to large-scale chromosomal alterations (e.g., 10 Mb in STORK [36]). The major challenge is that very low quantities of aligned DNA fragments can be obtained within a short sequencing timeframe.

Our approach to increase the quantity of genomic fragments and to achieve high-resolution CNV, without simply extending sequencing duration, involves randomly concatenation of fragmented genomic DNA into long reads. This approach enables the identification of multiple mappable DNA fragments in one sequencing read, thus optimizing sequencing efficacy. Through sequencing a subset of randomly assembled genomic fragments (termed genomic sampling), the genome-wide chromosomal integrity can be quantitatively assessed. While previous attempts using long concatenated reads for quantitative genomic analysis have been introduced, the methods are lengthy and require sequential mechanical shearing (SMASH [37]) or enzymatic digestion (SMURF [38, 39]) followed by DNA purification, and ligation. The multi-step preparation requires a high input DNA (micrograms) and takes several hours to process, thus prohibiting clinical intraoperative application. An important rate-limiting factor has been the lack of a highly efficient method to reliably process genomic DNA in one reaction.

Here we develop a novel one-step random genomic DNA concatenation method, named irreversible Sticking Compatible Overhangs to Reconstruct DNA (iSCORED), which enables efficient and unbiased CNV assessment using a real-time Nanopore sequencer. The simple one-step reaction requires 200–400 ng of gDNA and achieves high-resolution (50–60 kb) whole-genome CNV analysis by sequencing the long concatenated reads (~ 1–2 kb) assembled from short DNA fragments (~ 100–150 bp). Furthermore, by utilizing the 5-methylcytosine data available through Nanopore sequencing, we demonstrated the feasibility of concurrent methylation classification in primary CNS tumors and promoter methylation analysis of amplified oncogenes. The pipeline was first applied to study a cohort of 26 intracranial neoplasms, consisting of 17 primary CNS tumors and 9 metastatic tumors. The results were compared with those from next-generation sequencing (TruSight® Tumor 170 and whole exome sequencing) and chromosomal microarray (Affymetrix OncoScan®) [40] generated in a Clinical Laboratory Improvement Amendments (CLIA) certified pathology laboratory at Dartmouth-Hitchcock Medical Center (DHMC). Following validation of high accuracy and rapid molecular results, we subsequently applied our real-time pipeline to investigate 15 diagnostically challenging primary brain tumors during surgery.

Methods

Patients and clinical samples in the archival and prospective cohorts

We retrieved 26 banked malignant brain tumor specimens from the institutional biorepository at the Dartmouth-Hitchcock Medical Center (DHMC). The specimens were procured from patients who underwent craniotomy for the resection of malignant brain tumors between December 2021 and April 2023 at DHMC. The prospective study of 15 additional primary brain tumors was acquired during the standard intraoperative morphology-based diagnosis. All diagnoses are confirmed by two board-certified neuropathologists (CCL and GJZ; Additional file 1: Table S1). All clinical samples were analyzed by iSCORED sequencing, Next-Generation Sequencing, and Chromosomal Microarray (Affymetrix OncoScan®). A subset of samples (n = 10) were also processed with transposase-based preparation (RAD114) for direct gDNA sequencing. Four control gDNA were acquired from Coriell Institute (NA20967, NA12878, NA24385, and NA24631) and processed with iSCORED sequencing to establish the control iSCORED datasets.

DNA extraction from frozen tissue

We incorporated the genomic DNA extraction process into the standard morphology-based intraoperative diagnosis. Briefly, 5–10 scrolls of tissue (> 5 mm × 5 mm) were sectioned at 5 μm thickness onto blank slides in a cryostat machine. Using the DNeasy Blood & Tissue Kit (Qiagen #69504) with minor modifications for ultrafast extraction, a premixed solution containing 180 μl of tissue lysis buffer (buffer ATL) and 2 μl of RNase A (Thermo Scientific #01236994) was added onto the slide and the tissue was scrapped off, then transferred to a 1.5 ml Eppendorf tube. The tube was incubated at 37 °C for 2 min, after which 20 μl of protease K was added to the reaction, followed by an additional incubation at 56 °C for 8 min. Buffer AL (200 μl) and pure ethanol (200 μl) were then subsequently added to the reactions. The final reaction was mixed thoroughly by vortexing and added to the spin column inserted to the vacuum to facilitate the extraction procedures. Sequential buffer AW1 (500 μl) and AW2 (500 μl) were added to wash the column, which was centrifuged at 20,000 × g for 30 s for final cleanup. The DNA was eluted with 50–75 μl of AE buffer and its quantity and quality were checked using a Nanodrop Spectrophotometer (ThermoFisher Scientific). The entire DNA extraction process from tissue collection in a cryostat machine to Nanodrop DNA measurement is finished within approximately 20 min (Fig. 4a, b).

iSCORED reaction

Approximately 200–400 ng of input genomic DNA (gDNA) was used for the iSCORED reaction, followed by bead purification for Nanopore sequencing. The reaction mixture comprised 15 μl in total, which included quick ligase buffer (4.5 μl), quick ligase (1 μl, NEB E6056), and NTAN cocktail mix (1 μl; equal amounts of MseI, BfaI, CviQI and NdeI from New England Biolabs). The reaction was incubated at 37 °C for 30 min with intermittent cooling/agitation to enhance ligation. Specifically, the reaction was agitated at 900 rpm on the 18 °C pad at reaction time points of 9–10, 14–15, 19–20, and 24–25 min. End repair/dA tailing buffer (1.75 μl) and enzyme mix (0.75 μl, NEB E7546) were added to the mixture, which was then incubated at 20 °C for 5 min and 65 °C for 5 min. For final ligation to the Nanopore motor proteins, a freshly made premix of ligation buffer (LNB, 4 μl), adaptor protein (AMF, 1.5 μl, ONT LSK110 for R9.4.1 flowcells; LA 1.5 μl ONT LSK114 for R10.4.1 flowcells) and quick ligase (1.5 μl) were added to the reaction solution and incubated for 10 min at room temperature (20–22 °C) with intermittent mixing. Finally, 10 μl of ddH₂O and 14 μl of PEG-deprived AMPure XP beads (Beckman Coulter A63881) were added for standard magnetic bead purification. Of note, the AMPure XP beads were re-suspended in 2.5 M NaCl to eliminate PEG 8000 in the original solution. The beads were washed twice with long fragment buffer (LFB, 80 μl) before eluting into 12 μl elution buffer. Approximately 30–50 ng of ligated DNA were loaded for flongle flowcells (R9.4.1), 100 ng were used for MinION flowcells (R9.4.1), and 50–75 ng for PromethION flowcells (R10.4.1).

Transposase-based rapid sequencing

The library construction follows the SQK-RAD114 protocol. Briefly, for each reaction, approximately 200–300 ng of extracted gDNA were adjusted to the total volume of 10 μl with ddH₂O and mixed with 1 μl of fragmentation mix (FRA). The reaction is incubated at 30 ºC for 2 min and then at 80 ºC for 2 min. The rapid adaptor (RA) is diluted by combining 1.5 μl of RA with 3.5 μl of ABD and then added to the reaction. The reaction is gently mixed by flicking and incubated for 5 min at room temperature.

Normalization of intrinsic regional variability

Control human genomic DNA (gDNA; NA12878 from Coriell Institute) was first processed with iSCORED and sequenced extensively (16,564,873 mapped fragments) to establish a reference dataset for normalization to equivalent bins. The bins with counts below 0.2th percentile were removed to eliminate false positive bins as tested in four control datasets (NA20967, NA12878, NA24385, and NA24631 from Coriell Institute; Additional file 1: Fig. S4). An array of the proportion was segmented at the chromosomal level to optimize the resolution of the normalization and increase the sensitivity at which outliers were detected. A normalized vector of ratios (r1), identical in size to the reference array, was created. The distribution of this vector, at the chromosomal level, was employed to detect outliers using three components: (i) a threshold of 5 to filter the elements in r1 that were not greater than a specific threshold = 5, (ii) Z scores to determine the statistical significance of the deviation from the distribution, and (iii) the presence of surrounding outlier bins (a minimum of two consecutive bins must be present for a set of datapoints to be considered as outliers).

Coefficient of variation calculation across genome

Similar to what was described above, a normalized vector was created using NA12878 as a reference. For this approach, we used three other commonly used control gDNAs (NA20296, NA24385, and NA24631 from Coriell Institute). Each one of these datasets was segmented into independent data subsets (with no overlapping fragments) that vary in the number of mapped fragments. The number of fragments in these datasets were 70 k, 200 k, 300 k, 400 k, 500 k, 600 k, 700 k, 800 k, 900 k, 1 M, 1.25 M, 1.75 M, 2 M. For each dataset, the coefficient of variation (CoV) across the genome was calculated to assess the variability. In addition, the behavior of the CoV as the number of fragments change was assessed with the first order derivative of the CoV function.

Basecalling and read filtering

Fast5 files were converted to pod5 with ONT’s pod5-file-format (https://github.com/nanoporetech/pod5-file-format) and basecalled with ONT’s Dorado v0.5.2 (https://github.com/nanoporetech/dorado) using dna_r9.4.1_e8_fast@v3.3 for r9.4.1, or dna_r10.4.1_e8.2_400bps_sup@v4.1.0 for r10.4.1 with the following setting: –modified-bases 5mCG –emit-moves. The resulting unmapped SAM (uSAM) files were converted to fastq using samtools FASTQ -TMM, ML to carry the methylation information forward into the FASTQ header. The resulting FASTQs were first processed with Porechop (v0.2.1, https://github.com/rrwick/Porechop) to trim adapter sequences and split reads with internal adapters. Filtered with NanoFilt (2.8.0) to remove rare reads greater than 15 kb which represented native genomic reads that did not contribute to our analysis [41].

Read processing into aligned fragments

The following was adapted from Prabakar et al. [38]. Briefly, filtered reads were aligned to GRCh37/hg19, GRCh38/hg38, or hs1/T2T using BWA-MEM [38] (v0.7.17) with the following settings: -x ont2d -k 12 -W 12 -A 4 -B 10 -O 6 -E 3 -T 120. These settings allowed the segmentation of the concatenated reads into individual fragments that were aligned to their respective genomic regions. To ensure accurate quantitative CNV analysis, the duplex reads were identified and excluded from the original uSAM with ONT-Duplex Tools (https://github.com/nanoporetech/duplex-tools). A single member of each pair was then removed from the SAM file using a list of readIDs and Picard (https://github.com/broadinstitute/picard). The genome was then subdivided into either 5,000 or 50,000 genomic bins for CNV and amplification analysis, respectively, and mapped fragments per bin were calculated [42].

CNV analysis

Two approaches to CNV analysis were utilized. In the SMURF-seq analysis pipeline [38], counts of uniquely mapped fragments to 5,000 or 50,000 bins in the hg19 genome reference were normalized for biases in GC content and plotted with an implementation of DNAcopy [43] (v1.74.1) using circular binary segmentation identified breakpoints in bin counts. The ichorCNA [44] (v0.2.0) pipeline was similarly used for the T2T CNV analysis, uniquely mapped reads were similarly aligned to bins 50 kb or 500 kb in length, which were normalized for GC bias, mappability scores, and a panel of control genomic DNA (NA12878, NA20296, NA24385, and NA24631) [42].

Output table and graph

The output table contained a list of at least two consecutive statistically significant outliers (i.e., bins) to minimize the potential of identifying isolated/noisy outliers due to individual genome variation. The table displayed the corresponding position, ratio, Z score at the chromosomal level of each sample, along with commonly amplified gene(s) found in these bins of interest. All annotated genes from the hg19 or T2T reference genome were included in the table with 75 commonly amplified genes highlighted in red. The graph was automatically generated if there were significant bins in the sample of interest (with the Figeno package in the T2T reference genome) [42].

Methylation calling and tumor classification

The following were adapted from Rapid-CNS² pipeline [45]. The filtered reads were aligned to GRCh38/hg38 using BWA-MEM [46] (v0.7.17) with the following settings: -x ont2d -k 12 -W 12 -A 4 -B 10 -O 6 -E 3 -T 120 -C -Y to map the individual fragments within each read. The -C command allowed the SAM tags for methylation to be moved from the FASTQ header back into the SAM file. The -Y command turned off soft-clipping which would otherwise de-couple the methylation tag information. The per-site methylation is extracted using mbtools (https://github.com/jts/mbtools). A custom python script converted the bedfile to make it compatible with Rapid-CNS² which processes the methylation information using a random forest classifier trained on Illumina BeadChip 450 K methylation array from the Heidelberg reference cohort of brain tumor methylation profiles [47]. For methylation classification with Sturgeon the methylation data from reads aligned to either hg38 or T2T were extracted with Modkit (v0.2.7, https://github.com/nanoporetech/modkit) with the command “modkit extract –allow-non-primary” to account for our chimeric reads. The Sturgeon pipeline was then run with the corresponding genome probe bedfile [42].

To determine the minimum time required for methylation classification, we simulated the collection of methylation data over time using samples that had been sequenced for > 60 min. The data were subdivided into several bins (10, 20, 30, 45, 60 min). Sequencing start time was recovered from the uSAM read header and the aligned SAM was filtered accordingly. The data were then processed using our standard analysis pipeline as described above to extract the number of detected methylation features, the methylation classification, and the calibrated scores at each time point.

To study the role of tumor percentage in methylation classification, gDNA from the control human frontal lobe was processed with iSCORED. The resulting reads were in silico admixed with datasets from a medulloblastoma, an oligodendroglioma, or a glioblastoma, all of which had > 90% tumor percentage and calibrated scores of ~ 0.99 in methylation classification. Reads equivalent to 1 hour of sequencing on the MinION at different ratios of tumor to control (tumor percentages of: 0–100 in intervals of 10) were used for methylation classification using the Rapid-CNS² pipeline.

Computer specifications

Our computer system comprises an Alienware desktop computer (Intel® Core™ i9-12900 K Processor, 24 cores, 64 GB of RAM, 2 Tb of storage, RTX 3090Ti) and a custom-made workstation (AMD Ryzen Threadripper PRO 5995WX 2.7 GHz 64-Core sWRX8 Processor, RTX 4090, 512 Gb of RAM and 16 Tb of storage).

Flowcell reuse for cost-effectiveness

The MinION (R9.4.1) and PromethION (R10.4.1) flowcells were washed for sequential runs by using the flowcell wash kit (WSH004-XL). Briefly, 400 μl of flowcell wash mix (398 μl of wash diluent and 2 μl of wash mix) were loaded to the priming port to allow a DNase I reaction for 60 min at room temperature. The reaction solution was removed from the waste port and the storage buffer (500 μl) was loaded into the priming port before storing at 4 °C for next use. A minimum of 800 active pores in MinION and 3,000 active pores in PromethION, as verified by flowcell check, were required for a successful run. Following the protocol, MinION and PromethION flowcells could be re-used up to 7 times. The lack of library cleanup in the rapid transposase-based method (SQK-RAD114) seems to negatively impact the health of PromethION flowcells. Using the same protocol, we were only able to use the flowcells 2–3 times before the active pores count fall below 2,000.

By employing the wash protocol, our results demonstrated 100% consistency with independent experiments using new flongle flowcells, indicating the absence of detectable carryover between experiments (Additional file 2: Fig. S9). The ability to reuse flow cells resulted in a reduced sequencing cost of $140–185 per sample (Additional file 1: Table S5).

Timeframe for iSCORED pipeline

To enable intraoperative molecular diagnosis, our pipeline includes a real-time basecalling process along with periodical filtering and alignment of samples for CNV, amplification, and methylation analysis. Upon completion of the defined sequencing period, the separate files are merged to finalize the analysis (Additional file 1: Fig. S8). The pipeline is designed to operate on standard computers, circumventing the need for complex and expensive infrastructure such as Cloud computing systems (see computer specifications).

MinION sequencing generates 344 ± 24 Mb of data (SEM) within 60 min. With the introduction of P2 Solo (ONT), four of our archival specimens were analyzed using PromethION technology (R10.4.1). Using identical controlled library preparation methods, PromethION sequencing yielded an average of 395 ± 55 Mb (SEM) of data in 25 min, corresponding to an average of 1.69 ± 0.37 million (SEM) mapped fragments (Additional file 1: Fig. S6a).

Chromosomal microarray (Affymetrix OncoScan®)

DNA from FFPE samples was isolated using the QIAGEN QIAamp FFPE Tissue Kit (Qiagen, Valencia, CA). DNA quantity was measured with the Qubit Fluorometer 3.0 and Qubit dsDNAHigh-Sensitivity assay kit (Thermo Fisher Scientific company, Waltham, MA). Samples were then subjected to CMA following the protocol of the OncoScan FFPE Assay Kit (Affymetrix, Santa Clara, CA) described previously [40].

Next-generation sequencing

Tissue samples used in this study had been previously sequenced using either the Illumina TruSight® Tumor 170 assay or the newer DHCancerSeq whole exome sequencing assay. Both tests had been clinically validated in the Center for Clinical Genomics and Advanced Technology at the Dartmouth Hitchcock Medical Center Department of Pathology and Laboratory Medicine for routine clinical use. Each assay was validated to perform NGS using DNA and RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tissue samples for somatic analysis with an integrated bioinformatics pipeline for sequencing analysis, variant calling, and interpretation. The same set of 170 genes was analyzed regardless of which assay was used on the samples for this study. For the DHCancerSeq, the Agilent SureSelect Human All Exon V8 was used to provide a comprehensive and most up-to-date coverage of protein coding regions from RefSeq, CCDS, and GENCODE. The SureSelect Human All Exon V8 spans a 35.1 Mb target region of the human genome with an end-to-end design size of 41.6 Mb. The V8 exome workflow is automated with the Magnis NGS Prep System. The Illumina TST170 workflow was automated using the Beckman Coulter Biomek NX^P robotic workstations. Each assay was designed to examine single nucleotide variants, small deletions, small insertions, amplifications, fusions, and splice site variants to obtain a comprehensive somatic molecular profile for diagnosis, prognosis, and prediction of therapeutic response. TST170 libraries were sequenced on the Illumina NexSeq500 and DHCancerSeq libraries sequenced on the Illumina NovaSeq 6000 System. RNA library preparations and target enrichment were performed using the Illumina TST-170 sequencing assay. Routine clinical NGS was performed by either the Illumina TST170 assay or the DHCancerSeq which utilizes the Agilent SureSelect Human Exome V8. The same 170 gene targets were evaluated in each assay.

TST 170 amplification targets:

AKT2, ALK, AR, ATM, BRAF, BRCA1, BRCA2, CCND1, CCND3, CCNE1, CDK4, CDK6, CHEK1, CHEK2, EGFR, ERBB2,ERBB3, ERCC1, ERCC2, ESR1, FGF1, FGF10, FGF14, FGF19, FGF2, FGF23, FGF3, FGF4, FGF5, FGF6, FGF7, FGF8, FGF9, FGFR1, FGFR2, FGFR3, FGFR4, JAK2, KIT, KRAS, LAMP1, MDM2, MDM4, MET, MYC, MYCL1, MYCN, NRAS, NRG1, PDGFRA, PDGFRB, PIK3CA, PIK3CB, PTEN, RAF1, RET, RICTOR, RPS6KB1, TFRC.

Methylation profiling of amplified oncogenes

Methylation information from the bam files was plotted using MethylArtst [48] using the “region” function with a setting of 1,000 windows and a smoothing window of 4. Gene and promoter locations for important oncogenes were extracted from Ensembl for GRCh38. A region spanning 0.8 × of the length of the gene upstream and downstream was plotted, highlighting the promoter region.

Results

Concurrent fragmentation and concatenation of genomic DNA

The central concept of iSCORED is simultaneous digestion and ligation of DNA molecules by utilizing a panel of restriction endonucleases (REs) capable of generating compatible cohesive ends. Within the same reaction, DNA ligase catalyzes random re-ligation of the digested fragments to form long concatemers. Irreversible ligation products are generated when cohesive ends produced by different restriction enzymes are ligated together. This unidirectional reaction is possible because of the staggered nature of the DNA recognition sequences and actual phosphodiester bond breakage sites (Fig. 1a, b, Additional file 1: Fig. S1). The likelihood of forming such irreversible ligations increases with the number of different restriction enzymes producing compatible ends. Using CTAG overhangs as an example, the digested fragments were concatenated to larger chimeric molecules in the presence of DNA ligase (Additional file 1: Fig. S1).

Fig. 1

Proposed iSCORED method for rapid copy number analysis. a iSCORED schematic showing simultaneous compatible end ligation with TA enzyme cocktail (T^TAA, C^TAG, G^TAC, and CA^TATG by MseI, BfaI, CviQI, and NdeI, respectively). b Long stochastically concatenated DNA molecules are analyzed with a Nanopore device and aligned to the reference for genome-wide quantitative measurement. c The reconstruction efficiency of four iSCORED cocktail combinations is compared. The reaction is incubated at 37 °C for 30 min. CATG cocktail: NcoI (C^CATGG), PciI (A^CATGT), BspHI (T^CATGA). CTAG cocktail: NheI (G^CTAGC), SpeI (A^CTAGT), AvrII (C^CTAGG), XbaI (T^CTAGA). CG cocktail: MspI (C^CGG), HinP1I (G^CGC), HpyCH4IV (A^CGT), and TaqI-V2 (T^CGA). EcoRV is employed as a control since it generates blunt ends upon restriction digestion. d Optimization of iSCORED reaction by adjusting various experimental parameters, such as incubation periods, DNA ligases, and intermittent mixing and cooling. e An oligodendroglioma sample was processed either sequentially (digestion, purification, and ligation), with iSCORED, or sequenced as native gDNA. Samples were normalized to contain the same amount of sequencing data. The number of unique fragments mapped per genomic bin are shown for each sample (left panels). The resulting CNV plots are shown in the right panels (resolution = 600 kb per bin). CoV for sequential approach, iSCORED, and native gDNA sequencing are 0.57, 0.54, and 3.3, respectively. f Comparison of library preparation times across three methods. The sequential DNA digestion and ligation required 150 min, while the iSCORED required 75 min and the native DNA method required 45 min. The goal of intraoperative molecular diagnosis is achieved within 120–150 min of receiving the resected specimen

Systematic analysis and optimization of all overhang candidates

We next examined all existing 4-mer and 6-mer Type IIP REs capable of generating 2-nucleotide and 4-nucleotide overhangs (Additional file 1: Fig. S2). Given the palindromic nature of Type IIP REs, there are 16 (= 4²) and 4 (= 4¹) possible combinations for 4-nucleotide and 2-nucleotide overhangs, respectively. Depending on the RE recognition sequence (4 or 6 bp), the same overhang could be generated by 4¹ or 4² different enzymes; however, some of the theoretical combinations do not exist, and some are partially or completely blocked by DNA methylation (Additional file 1: Fig. S2; New England Biolabs).

We tested the top four overhang candidates that had the highest number of RE combinations while exhibiting the least possibility of methylation inhibition (Fig. 1c). To quantitatively measure the reconstruction efficiency (i.e., the number of uniquely mapped fragments per sequencing read), we compared combinations generating 4-nt overhangs with those generating 2-nt overhangs. Surprisingly, we found that the efficiency of 4-nt overhang combinations is not superior to that of 2-nt overhang combinations (Fig. 1c, Additional file 1: Fig. S3). This is presumably due to the significantly higher number of generated fragments by REs with 2-nt overhangs. The most efficient combination was the TA overhang cocktail mix that consisted of MseI, BfaI, CviQI, and NdeI, resulting in a mean reconstruction efficiency of 4.6 and mapped fragment of 120 bp. To further optimize the iSCORED reaction, we tested various incubation periods and DNA ligases (Fig. 1d). Our experiments revealed that an incubation period of 30 min at 37 °C with intermittent agitation at 18 °C (900 rpm) yielded the highest mean reconstruction efficiency of 8.7 (Fig. 1d). This experimental condition was thus utilized for the remainder of the study.

Genomewide aneuploidy detection in tumors

To detect large-scale CNV (> 10 Mb) and aneuploidy, the sequenced reads were first segmented into individual fragments by identifying short matches when mapping to the GRCh37/hg19 reference genome [38]. These uniquely mapped fragments were then filtered for quality (alignment scores ≥ 120, see the “Methods” section for details) and assigned to predefined genomic bins (600 kb) for quantitative analysis. High numbers of mapped fragments per bin generated low variability between bins and this helped ensure high confidence in the resulting CNV plot. Finally, circular binary segmentation [49] through DNACopy [50] was employed to identify copy number alterations across genomic bins.

The performance of the iSCORED pipeline was compared to the conventional sequential SMURF approach [38] (i.e., digestion, purification, and ligation) and unprocessed native gDNA. By normalizing the datasets to the same amount of total DNA sequence, both the iSCORED and SMURF methods exhibited an over 16-fold increase in the number of fragments compared to native gDNA sequencing (Fig. 1e). The significant increase in fragment count resulted in low variability, which was critical for detecting copy number changes with high confidence. Specifically, the coefficient of variations (CoV) for the SMURF, iSCORED, and native gDNA sequencing were 0.54, 0.57, and 3.3, respectively. Detection of large CNV and aneuploidy by iSCORED showed 100% concordance with clinically validated chromosomal microarray data (Additional file 2: Fig. S9) and also demonstrated a much higher resolution than short-read based analysis [36] within a comparable timeframe of 2 h (Fig. 1f, Additional file 1: Table S2) [51]. While the SMURF approach demonstrates a slightly better reconstruction efficiency (14.5), its sequential approach requires a substantial input DNA (2–3 μg) [38] and an extended preparation time (90–120 min) [38, 39]. This is in contrast to the iSCORED method, which requires only 200–400 ng of input gDNA and a preparation time of 30 min. The prolonged process prohibits its intraoperative application in standard CNS surgeries (typically 3–4 h; Additional file 1: Table S3). The molecular information is crucial for accurate pathology diagnosis as many primary CNS tumors are defined by molecular alterations. For instance, chromosomal 7 gains and 10 losses are characteristic of glioblastoma, while the presence of 1p/19q codeletion is required for oligodendroglioma diagnosis according to the 2021 WHO Classification [13].

Refinement of quantitative measures to detect copy number variations

While large 600-kb bins are effective in detecting aneuploidy, most clinically-relevant gene amplifications occur within a range of hundreds of kilobases to a few megabases [52, 53]. In such cases, using a large bin could result in averaging out the dose change, leading to decreased accuracy of small amplifications. Thus, we refined the bin size to 60 kb, which was similar to the highest genomic resolution of clinically validated chromosomal microarray analysis (CMA). When using the control human genome (NA12878) to quantify the total mapped fragments in refined 60-kb bins, the numbers mapped fragments fluctuated substantially across the predefined bins (Fig. 2a). Since the iSCORED is a restriction enzyme-based method, this finding was presumably due to variations in the density and distribution of restriction enzymes’ cutting sites across the human genome. We have termed the phenomenon intrinsic regional variability (IRV) (Fig. 2a, Additional file 1: Fig. S4). The background fluctuations might allow for tolerating outliers driven by true copy number changes, impacting the detection accuracy. In addition, this fluctuation behavior was inversely related to the amount of the data acquired (Fig. 2b). Hence, this finding was characterized in the context of the corresponding number of fragments in the genome.

Fig. 2

Normalization of variable mapped fragments in predefined bins for accurate copy number detection. a The number of mapped fragments per bin fluctuates across the wild-type genome (intrinsic regional variability, IRV), yielding a relatively high coefficient of variation (CoV) of 0.68 and hampering detection of true outliers. b Extensive sequencing does not address the fluctuation due to IRV (left panel). Normalizing the samples with the control wild-type dataset, the CoV dramatically drops and stabilizes at ~ 1 million mapped fragments (right panel). c The control genome data displayed CoV of 0.09 after normalization (upper panel). Application of this approach allows for detecting regions of amplification in both chromosome 2 and chromosome 19 (defined as copy number > 10). d Mixture of tumor with wild-type gDNAs shows that the amplified copies increase as the tumor percentage increases (Pearson correlation coefficient of 0.99). Using Z values of 10 as cutoff, the genetic amplification CCNE1 in chromosome 19 could be reliably detected at 5% tumor purity with 500,000 mapped fragments

For this purpose, we utilized the coefficient of variation (CoV) [54] as a quantitative index and performed a time-lapsed analysis of the sequenced control samples. After comparing the sub-datasets with varying numbers of mapped fragments, we found that significant fluctuations reached a plateau around a CoV of 0.68 at one million mapped fragments, regardless of extensive sequencing (Fig. 2b, left panel). To address this issue, we performed bin-specific normalization by calculating a ratio of the mapped fragments in the sample of interest to those in the commonly used control reference genome (NA12878). This normalization significantly reduced the observed genomic fluctuations by approximately fourfold (Fig. 2b, right panel). The CoV of the normalized data was substantially reduced to 0.09 down from 0.68 in the corresponding non-normalized data.

This normalization process also allowed us to infer the required number of total mapped fragments to reliably identify regions of copy number change. When investigating the slope [55] of CoV as a function of the acquired fragments, the inflection point was at a datapoint with mapped fragments of < 500 k, while the first order derivative function approached a value of zero at about one million mapped fragments (Fig. 2b, right panel). Thus, we determined that acquiring approximately one million mapped fragments was sufficient to reliably detect genomic dosage changes. It is worth noting that pre-defined bins with inherently low counts can lead to high sampling variability, resulting in false positive detections. Using normal well-characterized control gDNAs (NA20967, NA12878, NA24385, and NA24631 from Coriell Institute), we established that excluding bins within the lowest 0.2% genomic counts ensures reliable genomic dosage assessment (Additional file 1: Fig. S4). Bin-specific normalization and lowest 0.2% bin exclusion helped effectively detect true copy number variations by minimizing the effects of intrinsic regional variability. Finally, with the advent of a complete human genome reference that includes pericentromeric and subtelomeric regions [56, 57], we have aligned our iSCORED datasets to the T2T reference genome (Additional file 2: Fig. S9, see the “Methods” section for details) [42].

Gene amplification across various tumor purity levels

To demonstrate the effectiveness of the iSCORED procedure and its analysis pipeline, we first analyzed a metastatic adenosquamous carcinoma from the esophagus (Case M9 in Table 1). A CCNE1 gene amplification was detected in chromosome 19 (140 copies, inset in Fig. 2c), consistent with clinically validated next-generation sequencing. We further determined the minimum tumor percentage to reliably detect CCNE1 amplification by assessing a range of mixtures comprising control gDNA (NA12878) and tumor gDNA. This revealed a positive correlation between increasing amplification and rising tumor percentage (Pearson r = 0.99). By utilizing a Z-score cutoff of 10 to establish detection confidence, we were able to detect CCNE1 amplification in samples with as low as 5% tumor percentage using only 500 k mapped fragments (Fig. 2d). Additionally, low copy number gain (22 copies) was also reliably detected with the same parameter, albeit at a higher tumor percentage and with more fragments (20% and 1.5 million fragments, respectively, Additional file 1: Fig. S5). Overall, our results demonstrate the effectiveness of the iSCORED pipeline in detecting gene amplifications, even in samples with low tumor purity. The detection resolution outperforms the tumor percentage thresholds employed by clinical next-generation sequencing platforms, which are generally set at 15–20%.

Table 1 Comparison of oncogene amplification results between iSCORED and NGS

Rapid molecular analysis of the brain tumor cohort

To validate the iSCORED pipeline, we performed blind testing of a cohort of 26 intracranial neoplasms, including 17 primary CNS tumors and 9 metastatic tumors. Taking advantage of the mechanical destruction of frozen tissue sectioned at 5 μm thickness using a cryostat, high-quality gDNA is extracted within 15 min and processed through the iSCORED pipeline (see methods for details). The performance was timed and the findings were compared to the results from clinically validated next-generation sequencing (TruSight® Tumor 170 and whole exome sequencing) and chromosomal microarray analysis (Affymetrix OncoScan®) [40].

Within 1-h of MinION sequencing, an average 344 ± 24 Mb of data were generated, corresponding to 1.38 ± 0.08 million mapped fragments (SEM, Additional file 1: Fig. S6) [51]. This is higher than the predetermined required data quantity for confident CNV detection (0.5–1 × 10⁶ mapped fragments, Fig. 2b). Furthermore, the output is approximately threefold the data volume of the recently published SMURF-based method (nCNV-seq) within the same 50–60 min sequencing window [39]. Across the 26 investigated samples, the diagnostic accuracy of the iSCORED platform was 100% in detecting gene amplification of more than 10 copies (95% confidence interval [58]: 91–100%, Pearson r = 0.81 by comparing to the NGS results; Table 1). One sample was detected to have MYB amplification (21 copies; case M8) by the iSCORED pipeline, a finding that was not originally uncovered by TST 170 panel but was later verified by a whole exome NGS study (13 copies).

The output genomic graph from the iSCORED pipeline provided precise information on amplified regions and the confidence of detected outliers (Fig. 2c and Additional file 2: Fig. S9). EGFR amplification is a molecular defining alteration in glioblastomas, typically occurring as extrachromosomal DNA ranging from 1–3 megabases (Mb) in size [52, 59]. In our cohort of six EGFR-amplified glioblastomas, the average amplification regions spanned 1.66 ± 0.44 Mb (SEM) with an average copy number of 150.5 ± 47 (SEM). These samples also exhibited diverse regions and degrees of amplification, which is consistent with the known heterogeneity of glioblastoma [60, 61] (Fig. 3d).

Fig. 3

Concurrent methylation analysis of primary CNV tumors. a Acquired methylation classification features with iSCORED-processed MinION sequencing over time using the Sturgeon and Rapid CNS. b Comprehensive comparison of Rapid CNS and Sturgeon methylation classification for primary brain tumors across multiple time points from the initiation of sequencing. c In silico mixture of glioblastoma, medulloblastoma, and oligodendrogliomas with control brain tissue dataset at various ratios using the Sturgeon (T2T) and Rapid CNS (hg38) (total data quantity after 1 h of sequencing). d Exact amplified regions covering EGFR oncogene in glioblastoma samples. e Methylation characterization of amplified EGFR oncogene reveals promoter hypomethylation. Subep = subependymoma, oligo = oligodendroglioma, medullo = medulloblastoma, GBM = glioblastoma

Concurrent tumor methylation classification using iSCORED-generated dataset

Methylation classification of tumor types has emerged as an important diagnostic tool in clinical practice, particularly in brain tumors [9, 12]. The Heidelberg methylation classifier, for instance, successfully classified 91 tumors from 2801 formalin-fixed paraffin-embedded (FFPE) archival tissue [9]. Leveraging the capability of Nanopore sequencing to identify 5-methycytosine (5mC) from native DNA without additional sample preparation, we extracted methylation information from our sequencing data and classified it using Rapid-CNS² and Sturgeon, both of which are machine learning classification systems trained on the Heidelberg dataset [10].

To evaluate the reliability of methylation classification over time, we first processed MinION data at five timepoints (10, 20, 30, 45, and 60 min). RapidCNS² extracted the number of methylation features that overlapped with the 100 k most variable features from the Heidelberg dataset (Fig. 3a). Within 45 min, all samples had identified more than 1,000 CpG features, reaching the cut-off determined by the RapidCNS² authors [62], and 10 out of 14 classification results aligned with the final pathological diagnoses (Fig. 3a, b and Additional file 1: Fig. S7). While Sturgeon utilizes all 450 k features from the Heidelberg dataset, leading to a larger overall number of features, these samples also reached sufficient CpG sampling within 45 min, and achieved concordant diagnoses in 10 out of 14 cases. In comparing the concordant cases, Sturgeon (T2T and hg38 reference) achieved a passing score (0.8) in 9 out of 10 cases, whereas Rapid-CNS² (hg38 reference) achieved a passing score (0.6) in 8 out of 10 cases. Among the 4 inconclusive cases, Sturgeon classified two cases (GBM3 and GBM4) as control inflammation with high calibrated scores (0.97–1.0), whereas Rapid-CNS² assigned lower scores to the same cases (0.28, 0.27 for GBM3 and GBM4, respectively) (Fig. 3a and b, Additional file 3: Table S6).

Rapid-CNS² and Sturgeon utilized datasets derived from native DNA Nanopore sequencing, which are long and unfragmented. To validate that our short concatenated fragments generated via iSCORED did not affect methylation classification, we compared the data of three oligodendroglioma samples processed with iSCORED to those from native DNA sequencing using both Rapid CNS² and Sturgeon systems [12, 45]. The results revealed comparable classification scores, suggesting that fragmentation in iSCORED did not appear to affect the accuracy of methylation classification. Of note, while the oligodendroglioma with a lower tumor percentage (80.5% in oligo_1, Additional file 1: Fig. S7a) consistently fell below the 0.6 threshold in Rapid CNS², all three samples passed the required threshold (0.8) in the Sturgeon (T2T) system after 30 min of MinION sequencing (Additional file 1: Fig. S7c).

Tumor purity can significantly affect methylation classification results [9, 12]. This effect is particularly pronounced in glioma due to their infiltrative growth pattern, resulting in a mixture of neoplastic cells, normal brain parenchyma, and inflammatory cells. By comparing to final pathological diagnoses, our data is consistent with this observation: all four samples showing inconsistent classification at 60 min of sequencing were gliomas with low tumor percentages (20–60%). To further assess the impact of tumor purity on classification accuracy, we in silico admixed data from three tumors with the highest classification scores with control CNS tissue (frontal cortex, Fig. 3d). Our analysis revealed a drop in classification scores as tumor purity decreased, with Sturgeon analysis being more resistant to low tumor purity (Fig. 3c). However, both RapidCNS² and Sturgeon either fell below the classification threshold (0.6 and 0.8, respectively) or did not yield consistent classification results when the tumor purity was 60% or lower. Overall, Sturgeon analysis (T2T reference) exhibits a superior calibrated classification score over the course of sequencing (Fig. 3b) and higher resistance to tumor purity issue when compared to Rapid-CNS² (hg38). We have thus selected Sturgeon (T2T) as the methylation classification neural network for intraoperative molecular diagnosis.

Promoter hypomethylation in the amplified oncogenes

The epigenetic landscapes of amplified oncogenes offer mechanistic insights into transcriptional regulations [52, 63]. Despite the inherent low-pass nature of iSCORED, gene amplification ensures sufficient coverage for methylation profiling in the defined regions. Using glioblastoma as a proof of principle, within 1 h of MinION sequencing, we consistently detected hypomethylation across approximately 285 CpG sites within the promoters of the amplified EGFR (n = 6, coverage depth of 7.6 ± 2.7 (SEM), Fig. 3e, f). Such promoter hypomethylation is not exclusive to amplified EGFR of glioblastoma. Low 5mC percentages at CpG islands of oncogene promoters were detected in two other amplified oncogenes in glioblastomas (MYCN in GBM1 and MDM2 in GBM6), as well as five oncogenes amplified in metastatic tumors (FGFR1 and CCND1 of breast cancer in M1, ERBB2 of lung cancer in M6, MYB of esophageal cancer in M8 and CCNE1 of esophageal cancer in M9; Additional file 2: Fig. S9).

Intraoperative validation of 15 diagnostically challenging CNS tumors

We employed the iSCORED pipeline for rapid molecular characterization of a prospective cohort of diagnostically challenging primary brain tumors during craniectomy at DHMC. The study included 8 high-grade gliomas, 4 low-grade gliomas, and 3 spindle cell neoplasms (Fig. 4b). The average time from specimen arrival to completion of DNA extraction was 20 ± 0.2 min (SEM). The iSCORED procedure consistently took 30 min, and subsequent library preparation before loading to PromethION flowcells took 31.8 ± 0.5 min (SEM). Since the first 5–6 min of Nanopore PromethION sequencing is used for flowcell check rather than library sequencing [12], we coordinated with library preparation to load the DNA library at the end of the flowcell check to avoid delays in data generation. The bioinformatics pipeline was programmed to analyze the initial 18 min of generated data [42], aiming to achieve 0.5–1.0 × 10⁶ mapped fragments. Final data analysis and integration, including methylation classification and copy number plot generation, took 4.9 ± 0.4 min (SEM). The entire workflow, specimen arrival to output graph generation, was accomplished within 104.7 ± 0.7 min (SEM). The PromethION flowcells were used up to 7 times and generated 168 ± 15 Mb of data within 18 min, which corresponded to an average of 708,163 ± 77,700 (SEM) mapped fragments and identified 19,296 ± 1571 methylation features (SEM) [42, 51] (Fig. 4c).

Fig. 4

Prospective molecular analysis of diagnostically challenging brain tumors with iSCORED pipeline. a Shown is the incorporated iSCORED workflow applied during intraoperative morphology-based diagnosis. Additional 10–15 scrolls of tissue sections, each 5 μm thick, are prepared to extract gDNA for subsequent iSCORED library preparation. Either MinION or PromethION sequencing could be utilized (both with concurrent analysis during sequencing). The final output graphs comprise whole genome CNV, gene amplification regions, and methylation classification with quantitative confidence scores (Z scores for gene amplification and calibrated scores for methylation classification). b Real-time intraoperative molecular diagnosis with precise timestamps recorded from tissue arrival to final reports in 15 diagnostically challenging brain tumors. The entire workflow could be completed within ~ 105 min. The morphology-based intraoperative diagnosis was compared to generated molecular results, including methylation classification and CNV results [51]. The numbers within the brackets of methylation classification and oncogene amplification denote the calibrated scores of corresponding diagnoses and detected copy number, respectively. * Scores of different glioblastoma subtypes. PXA = pleomorphic xanthoastrocytoma c Sequencing data, including mapped fragments and identified CpG sites, obtained within the initial 18 min (PromethION flowcells). d A comparative analysis of Nanopore-based molecular assays, including genomic detection resolution, library preparation time, input genomic DNA quantity, and required sequencing duration, reveals the iSCORED-based assay as the only method to achieve genome-wide high-resolution CNV detection within the surgical window. SMURF = sampling molecules using re-ligated fragments [38]. STORK = short-read transpore rapid karyotyping [36]. WGS = whole genome sequencing [35]

In the cohort, the cases were selected due to the pressing need to define glioblastoma within the high-grade glioma category. Among the high-grade gliomas, methylation profiling based on the Sturgeon (T2T reference) method accurately classified 7/8 as glioblastoma, with calibrated scores of 0.94 ± 0.02 (SEM). Sample IO_7 did not align with the final pathological diagnosis by methylation classification and showed the classification of control inflammation with a calibrated score of 0.9. The result was presumably due to the low tumor percentage of 60% in the submitted specimen. Nevertheless, the CNV plot [51], in line with chromosomal microarray [64], accurately identified key chromosomal aberrations detected in glioblastomas, including chr 7 gain and chr 10 loss (Additional file 2: Fig. S9). Finally, the CNV plots, in accordance of clinically validated NGS, accurately detected all oncogene amplifications with both hg19 and T2T reference genomes [44, 56, 57] (n = 10/10, including EGFR, PDGFRA, CDK4, MDM4, MDM2, KIT) (Additional file 2: Fig. S9 and Additional file 3: Tables S7 and S8).

Distinguishing between astrocytoma, oligodendroglioma, and other lower-grade gliomas via conventional morphology-based intraoperative diagnosis is challenging to due to freeze artifacts. In addition, some glial neoplasms are defined by specific molecular alterations that could not be acquired during traditional intraoperative diagnosis. In our cohort, methylation profiling accurately classified astrocytoma, oligodendroglioma, and pleomorphic xanthoastrocytoma with a calibrated score of 0.99 ± 0.004 (mean ± SEM; n = 4). According to WHO classification, a required chromosomal aberration for oligodendroglioma diagnosis is 1p/19q codeletion which was observed in the oligodendroglioma (IO_6) by both our generated CNV plots [51] and chromosomal microarray [64] (Additional file 2: Fig. S9).

Finally, the three spindle cells neoplasms exhibited a range of unusual phenotypes, leading to a wide range of differential diagnoses: chordoid neoplasm in IO_3, rhabdoid neoplasm in IO_4 and clear cell neoplasm in IO_12. These three cases were classified as meningioma by methylation profiling with high calibrated scores (0.99, 0.99, and 0.92, respectively). The final diagnoses aligned with rare and higher-grade subtypes of meningiomas: chordoid, rhabdoid, and clear cell meningioma, respectively [13].

Other studies have utilized the transposase-based DNA sequencing method (RAD114), which enables rapid library preparation and initiation of sequencing. With a 10-min tagmentation and adaptor ligation step, RAD114 allows sequencing be extended to 80 min, maintaining a similar total turnaround time of 104.7 ± 0.7 min. We analyzed 10 representative cases using this rapid sequencing kit, which generated approximately 12.5% of the total mapped fragment counts compared to those produced by 18 min of iSCORED sequencing (88,670 ± 18,356 versus 708,163 ± 77,700 in iSCORED, p < 0.005; Additional file 2: Fig. S10a). While large-scale chromosomal abnormalities could be seen on the CNV plot, the 8 known amplified oncogenes in the 10 analyzed tumors could not be reliably detected by the automatic pipeline due to low signal and high noise levels (Additional file 2: Fig. S10b) [51]. The lack of fragmentation and concatenation, as described in iSCORED, resulted in longer fragments but a limited number of mapped fragments, which negatively impacted accurate detection of oncogene amplification. The results are consistent with published datasets from other Nanopore-based methodologies. Using 60 min of MinION sequencing, other approaches yielded low mapped fragment counts that limit high-resolution CNV analysis: short read approach for aneuploidy analysis (0.08 million in STORK [36]; genomic resolution of 10 Mb), long native DNA for intraoperative methylation profiling (0.013–0.06 million [11, 12]; genomic resolution of 2 Mb in Sturgeon) and sequential concatenation (~ 0.5 million reads in SMURF [38]; genomic resolution of 600 kb) (Additional file 1: Table S2-4).

Discussion

In this study, we present iSCORED, a novel library preparation method capable of rapidly and cost-effectively generating high-resolution CNV profiles, detecting gene amplifications, and classifying tumors by their methylation profiles. Our prospective study demonstrates the effectiveness of the iSCORED pipeline in providing accurate and critical molecular characterization of primary brain tumors during surgery. While intraoperative Nanopore sequencing of native genomic DNA could identify large-scale chromosomal alterations, the output is suboptimal for high-resolution CNV detection [11, 12]. The limitation stems from insufficient genomic sampling caused by low mapped read counts, making it challenging to detect small oncogene amplifications (Additional file 2: Fig. S10). We overcome this limitation by employing the iSCORED method to efficiently concatenate small digested genomic fragments in one reaction. The approach requires low DNA quantity and significantly enhances genomic sampling efficiency to achieve high-resolution CNV and concurrent methylation classification in 105 min, well within craniotomy surgical windows of 3–4 h. Finally, compared to NGS sequencing platforms [34] and hybridization-based arrays [40, 65], the low cost per sample, a mere $140–185 USD, and the ease of setting up the infrastructure with a budget of $6,000–8,000 USD for MinION and $14,000–16,000 USD for PromethION make it an economical option for clinical applications (Additional file 1: Tables S2 and S5).

The 2021 WHO classification of the Central Nervous System Tumors has incorporated a substantial amount of molecular diagnostic alterations for accurate classification of CNS tumors [13] Genome-wide copy number analysis and methylation profiling are imperative for proper tumor grading and integrated diagnoses. The iSCORED platform demonstrates a high accuracy in detecting chromosomal imbalances and gene amplifications with the tumor purity thresholds set at 5% and 20% for high and low copy number amplifications, respectively. Concurrent 5mC information also informs tumor methylation classification, though this analysis is more sensitive to tumor purity. Using a calibrated score of 0.8 as the cutoff, accurate classification typically requires a tumor percentage of more than 60%. The phenomenon is not unique to iSCORED platform but also observed in other DNA methylation arrays [9, 12]. Thus, the combined characterization of CNV and methylation can synergistically improve molecular diagnosis, particularly in cases with sub-optimal tumor contents. For instance, the presence of both whole chromosome 7 gain and whole chromosomal 10 loss (+ 7/ − 10) is a dominant and specific molecular signature of IDH-wildtype glioblastoma with a sensitivity of 59% and specificity of 98% [13, 26]. In our cohort of four retrospective glioblastoma samples and one prospective glioblastoma case, methylation classification was hindered by low tumor contents. Nevertheless, specific chromosomal imbalances (+ 7/ − 10 in GBM 3, 4, 9 and IO_7; + 7/ − 10q in GBM 8) were identified (Additional file 2: Fig. S9), aligning with WHO criteria for glioblastoma [13, 51]. Therefore, in cases where tumor methylation classification is limited by low tumor purity, CNV analysis can serve as a highly sensitive molecular marker for integrated diagnosis.

Non-sequencing-based technologies have recently been utilized for intraoperative diagnosis, including ultrasound imaging [66], mass spectrometry (MS) [67, 68], Raman spectroscopy (RS), and artificial intelligence (AI) [69]. These methodologies detect genetic mutations indirectly: MS measures 2-hydroxyglutarate as a surrogate marker for IDH mutation, RS assesses vibrational energy modes to determine IDH, ATRX and 1p/19q status and AI-based diagnostic screening combines neural network and simulated Raman histology to predict IDH, ATRX and 1p/19q status. Ultrasound imaging utilizes radio-frequency signals to identify IDH, TERTp, and 1p/19q mutations. The primary advantage of these techniques lies in their ultra-rapid interpretation that could provide results within seconds to a few minutes. This capability offers surgeons “real-time” information on key mutations and may help assess tumor percentage at resected margins, significantly enhancing intraoperative decision-making. However, these methods are fundamentally indirect tools specifically trained to predict a limited panel of genetic alterations based on abnormal metabolism, histological characteristics, and ultrasound signals. Despite highly accurate in appropriate contexts, this specificity poses challenges in applying these methodologies to other types of mutations and cancers.

In contrast, the iSCORED platform utilizes Nanopore sequencing to directly identify genetic and epigenetic information from native DNA. The iSCORED intraoperative pipeline uses a more potent sequencer (PromethION) to compensate for a longer library preparation protocol (~ 62.8 min), thereby shortening the overall workflow to 105 min (Fig. 4b). While other intraoperative pipelines using transposase-based method for direct gDNA sequencing could be completed within 90 min, iSCORED generated approximately 10–50 times more mapped fragments across the genome (0.71 ± 0.08 million (SEM) in iSCORED [51] versus 0.06 million in Vermeulen et al. [12] and ~ 0.013 million in Djirackor et al. [11]) (Additional file 1: Table S4). The low mapped fragment counts fall below the minimum threshold of 0.07 million and result in more than 15 false positive hits across the genome (Additional file 1: Fig. S4b). This limitation is exemplified by the unreliable detection of chromosome 19q deletion in Sturgeon after 50 min of sequencing [12]. Finally, the generated methylation dataset from iSCORED is compatible with those trained in Sturgeon and Rapid-CNS², upgrading the pipeline’s capability for CNS tumor methylation classification [11, 12, 45, 70]. While the 18-min iSCORED sequencing covered fewer CpG sites compared to direct long-read genomic DNA sequencing (~ 20,000 CpG sites in 18 min for iSCORED versus ~ 40,000 CpG sites in 60 min for Sturgeon; Fig. 4c, Additional file 1: Table S4), the quantity become comparable when using the MinION flowcells and a 60-min sequencing period (~ 40,000 CpG sites; Fig. 3a). Therefore, our method represents a rapid unified sequencing-based platform capable of identifying combinations of chromosomal alterations, oncogene amplifications and methylation classification within the craniectomy timeframe.

The identification of key single nucleotide variations, including IDH1, IDH2, and TERTp mutations is essential for integrated diagnosis according to current WHO classification [13]. Similar to other sparse sequencing methods, the iSCORED pipeline is limited by low genomic coverage and not suitable to reliably detect point mutations within the given time constraints. Recent development of ddPCR and qPCR enable rapid detection of a selected panel of point mutations within 100 min and can be conducted in parallel to identify these mutations [71, 72]. Alternatively, our sequencing pipeline could potentially incorporate amplicon-based sequencing from ultra-fast PCR, which could enhance our ability to identify these targeted point mutations.

We have shown that CpG methylation can be measured within the amplified regions. Previous studies involving cell lines indicated that oncogene promoters, when presented as extrachromosomal DNA, exhibited hypomethylation compared to the chromosomal DNA of the same gene loci [52]. In our retrospective cohort of 26 tumors encompassing primary brain tumors and metastatic cancers, we identified eight distinct oncogene amplifications, all characterized by hypomethylated promoters (Fig. 3f and Additional file 2: Fig. S9) [51]. The pattern suggests epigenetic modifications could serve as a fundamental mechanism for active transcription in the amplified oncogenes. As a result, a combined approach integrating both genetic (amplification) and epigenetic (promoter methylation) data could potentially be a better parameter for predicting protein expression, tumor behavior, and clinical outcomes.

While our study focuses on the intraoperative diagnosis of primary brain tumors, it accurately detects large-scale chromosomal alterations and oncogene amplifications in various metastatic neoplasms. Despite potential differences in time constraints compared to craniectomy, our findings indicate that the pipeline could be broadly applied to other types of tumors and different settings. Specifically, when the intraoperative molecular diagnosis is not required, achieving next-day turnaround time (< 24 h) could be feasible by multiplexing samples and simultaneously sequencing up to 24 samples in parallel (LSK114.24, ONT), thereby reducing overall sequencing costs by 3.4 times. This technology could enhance accurate cancer diagnosis, particularly in challenging cases like tumors of unknown origin and poorly differentiated neoplasms.

Conclusions

Somatic copy number alterations represent a major type of genetic mutations in cancer initiation, progression, and treatment resistance [21,22,23, 25]. Among the various cancer types, ovarian carcinoma and sarcoma bear the highest burden of CNVs, accounting for approximately 80% of cases. Following closely are uterine carcinosarcoma and esophageal carcinomas, with approximately 75% exhibiting CNVs [73]. Oncogene amplification serves as the molecular hallmark in specific cancer diagnoses. For example, amplification of MDM2 and/or CDK4 is almost always present in well-differentiated liposarcoma, while MYC amplification is detected in > 90% of radiation-induced angiosarcoma [74]. The total number of copy number alterations could predict tumor aggressiveness and be utilized for tumor grading [13, 24, 25]. Accurate identification and comprehensive understanding of CNVs in cancers are pivotal for advancing diagnostic accuracy. Moreover, the detection of oncogene amplification would augment molecular-targeted therapeutics. For instance, Trastuzumab (Herceptin) is a monoclonal antibody that binds to the HER2 receptor and is employed to treat ERBB2-amplified breast and gastric cancers [75,76,77,78,79]. Several EGFR inhibitors are used to treat EGFR-amplified cancers, such as gefitinib and erlotinib for non-small cell lung cancer [80, 81], and cetuximab for colorectal cancer [82]. These findings underscore the importance of CNV analysis across diverse cancer types and the potential applicability of the iSCORED platform, particularly in situations when time-sensitive molecular information is critical. With the advancements in methylation classification across cancers, our integrated molecular characterization, combining CNV and methylation profiling, will benefit all tumor types [83]. Therefore, iSCORED-based rapid molecular diagnosis has the potential to unlock a new frontier in personalized medicine.