Introduction
Testing strategies
Commercial panels | Custom panels | SNV | CNV | Indels | Fusions | DNA input | FFPE suitability | Comments | ||
---|---|---|---|---|---|---|---|---|---|---|
Whole genome | Y | Y | Y | Y | Y | 1μg | Good-quality FFPE samples required to obtain DNA suitable for WGS | Hypothesis-free scanning of whole genome. Less sensitive for subclonal variation studies | ||
Exome | y | y | Small | Good-quality FFPE samples required to obtain DNA suitable for exome sequencing | ||||||
Illumina Truseq cancer panels | Y | Y | Y | N | Small | N | 10 ng +, more for FFPE | Yes, but requires good-quality material | Requires higher quality DNA than other equivalent technologies. Deep sequencing in a cost-effective manner. Identification of rare mutations andsubclonal variation detection in heterogenous tumour samples. | |
Illumina RNASeq | Y | Y | Small | Y | 20 ng + | Yes | ||||
Ampliseq cancer panels—DNA | Y | Y | Y | Small | N | 10 ng +, more for FFPE | Yes | Larger panels require multiple oligos pool, homopolymer errors | ||
Ampliseq cancer panels—RNA | Y | Y | Y | N | Y | Y | 10 ng+ | Yes |
Implementation of NGS in a diagnostic laboratory
Validation and verification
Validation
Test performance characteristic | TPC applied to NGS | Metrics and notes on assessment |
---|---|---|
Reportable range | The region of the genome in which a sequence of acceptable quality can be derived (may not be a contiguous region) | Reporting range must be confirmed during test validation |
Reference range | The spectrum of normal variation of sequence within the population that the assay is designed to detect. | Test results outside this range may be clinically significant and require additional investigation. |
Limit of detection (LOD) | The lowest allele frequency to which the assay can detect with an acceptable quality to enable confidence in a result i.e. the LOD is within the reporting range, (establishes the detection limit for sequence variants) | Minimum and maximum amount of DNA for 95 % test runs with adequate “no call” rate Allelic read percentage Sensitivity of the assay must be defined within the reporting range of allele frequencies and amount on input DNA for which the LOD was defined. |
Repeatability | Concordance of variant detection between runs from the same sample under the same conditions e.g. prepare different libraries from the same samples run at the same time with the same operator and same instrument (within-run or intra-batch variability) | Analyse adequate number of runs depth of coverage Uniformity of coverage Transition/transversion ratio Pair-wise agreement |
Reproducibility | Consistency of results from the same sample under different variations in conditions e.g. between different runs, different sample/library preparations, by different operators, or using different instruments (between-run or inter-batch variability). | Analyse adequate number of runs Depth of coverage Uniformity of coverage Transition/transversion ratio Pair-wise agreement |
Accuracy (if reporting VAF) | The degree of agreement between the nucleic acid sequences derived from the NGS assay and a reference sequence (a measure of sequencing accuracy and error rates) | Adequate depth of coverage Uniformity of coverage Positive percent agreement Negative percent agreement Technical positive predictive value Rate of “no call” Allelic read fraction (number of independent reads assessed when calling a variant) |
Precision (if reporting VAF) | The degree of agreement between replicate measurements of the same material across users and runs (a combination of reproducibility and repeatability) | Analyse adequate number of samples Depth of coverage Uniformity of coverage Transition/transversion ratio |
Analytic sensitivity | The proportion of samples that test positive for a sequence variation and are correctly classified as positive (=TP / (TP + FN) (false-negative rate) | Depth of coverage Number of independent reads used to make a base call. This is dependent upon the amplifiability of the template DNA in the assay. Evaluation of base quality scores and signal-to-noise ratios |
Analytic specificity | The proportion of samples that test negative for a sequence variation and are correctly identified as negative (=TN / (TN + FP) (false-positive rate) Some laboratories establish specificity by calculating the number of false positives per assay run. | Coverage (read depth and completeness) Number of independent reads assessed when making a base call Evaluation of base quality scores and signal-to-noise ratios. Potential for cross-reactivity and interfering substances Cross-contamination |
Sequencing depth and allelic frequency cut-offs | The minimum sequencing coverage necessary for confident detection and variant calling (established for different variants) |
Verification
Sample handling
Sample transportation, receipt, handling and storing
Morphological assessment and choice of region of interest
Nucleic acid extraction, quantification and storage
Sample misidentification
Library preparation
Quality assessment
General considerations
Next-generation sequencing quality scores
Phred quality score | Probability of incorrect base call | Base call accuracy |
---|---|---|
10 | 1 in 10 | 90 % |
20 | 1 in 100 | 99 % |
30 | 1 in 1000 | 99.9 % |
40 | 1 in 10,000 | 99.99 % |
50 | 1 in 100,000 | 99.999 % |
60 | 1 in 1,000,000 | 99.9999 % |
Quality assurance for NGS laboratory testing and bioinformatics
Checklist item | QC metric(s) | Conditions when recommended to repeat or recheck before release | Consequences | Troubleshooting suggestions |
---|---|---|---|---|
Tissue sample acceptance criteria | Establish and document criteria for accepting or rejecting specimens. Minimum neoplastic cell content met? Sufficient amount of material provided? Suitable material for assay (FF or FFPE)? Tissue fixed appropriately? Transport conditions met? | If tissue fails to meet minimum neoplastic cell content or the test has not been validated for the type of tissue provided | Can lead to false-negative test results (failure to detect mutations present) and therefore sub-optimal treatment May result in re-testing that depletes available tissue sample for other tests | Requesting additional tissue samples is invasive for the patient and may not be possible. However, this should be highlighted in the report and a further sample requested (if clinically relevant). |
Nucleic acid preparation | DNA quality (concentration, quality and volume) i.e. sufficient amount of amplifiable DNA? DNA quantity (double-stranded DNA content) | If the DNA falls below the minimum quantity or quality range that the test has been validated for | DNA that is not clean (e.g. trace EDTA or other inhibitors) can inhibit reactions. Can result in poor quality sequencing data, lower number of reads with adequate quality and variable recovery of sequence reads with high GC-content May lead to false-negative results and result in sub-optimal treatment. Tissue material may be depleted by multiple preparations. | Consider using a different extraction method: it is often helpful to have more than one extraction method validated for clinical use within the laboratory. Failed samples should be reported as such and further material requested, if clinically appropriate. |
Sample identification | Sample is clearly identifiable | If there is not complete traceability of the sample identity | Misidentification of samples could lead to administering the incorrect treatment to a patient (exposing a patient to unneeded treatment or accidentally withholding a treatment that is required) | If it is not possible to trace a sample and there is concern that a sample may not come from the patient, then a further sample may be needed. Use of SNP or other identification methods to link the patient to the sample is now available in cases where this is a major issue and clinically indicated. |
Library preparation | Minimum library concentration satisfied? Library preparation QC? Negative and positive controls in library preparation successful? | Check library QC for each sample and repeat or discard results if not acceptable. | May lead to poor or inaccurate coverage, especially in GC-rich regions. Library may have poor representation (lack the complexity) or bias not in the original sample. Can result in accuracies and failure to detect mutations present, which may result in sub-optimal treatment. There is also the possibility of false-positive results due to low allele frequency. | Consider new library preparation and/or use of an alternative method, such as PCR to verify uncertain results that are of clinical significance. |
Sequencing | Minimum sequencing depth satisfied? Minimum sequencing quality metrics satisfied? Verify the quality of base calling and sequencing accuracy (e.g. minimum base quality Q score). Adequate average read length Coverage reviewed per sample, gene, exon and mutation to be reported | If fails to meet sequencing acceptance criteria then either limited reporting on genes reaching criteria or samples should be repeated | Less depth or coverage means more uncertainty in the integrity of the data. Limited reports may be acceptable if all actionable mutations are covered at sufficient depth, but consider repeating assay. | Repeat sequencing with existing library, or if indicated, start afresh with new sample. Verification of uncertain results with another method (e.g.) targeted PCR may be helpful if a clinically relevant finding is involved. |
Variant detection and reporting | Variant present at allelic frequency and at adequate sequencing depth and quality score? % Reads mapped to reference genome, target region and % target covered Confirm presence of variant in forward and reverse strands. Check for non-specific mapping. Review mutation calls in the background of sequencing artefacts | Check presence and interpretation of variants found using validated software tools. | Variants of clinical significance may be missed or called incorrectly. | The best option here is probably to verify the finding with another method, if available, or to repeat sequencing. |
Exceptions to standard protocols | Review impact of any documented exceptions to standard pperating procedure | Quality management issue: highlight non-compliances and eradicate from practice where possible. | Variants of clinical significance may be missed or called incorrectly. | Repeat the test using the existing library or stored DNA depending upon where in the process the error occurred. |
Equipment and reagents | Reagents all within expiry and stored appropriately? Equipment appropriately maintained and calibrated? | Quality management issue: highlight non-compliances and eradicate from practice where possible. SOPs for reagent storage and regular servicing of equipment. | Poor reagent storage can lead to poor enzyme activity and poor ligation. Contamination can modify the end structure of the DNA and inhibit ligation. Variants of clinical significance may be missed or called incorrectly. | Order new reagents, and ensure expiry of consumables is checked before each run. Checklists can help ensure that this issue is avoided. For affected samples, repeat the test using the existing library or stored DNA depending upon where in the process the error occurred. |
Bioinformatics | Correct pipeline and version run? Appropriate reference sequence used? Successful calling of any controls? Potential cross-contamination? Any tumour-normal pairs are appropriately matched? | Check presence and interpretation of variants found using validated software tools. | Newer analysis software may be better optimised. Using outdated or incorrect software can lead to variants of clinical significance which may be missed or called incorrectly. | Ensure software is updated regularly and that servicing agreements are in place for sequencers and associated equipment. |
Results sign off | Result signed off by competent individual? | Ensure that all those signing off reports have evidence of adequacy of training and competence | Variants of clinical significance may be interpreted or presented incorrectly, leading to incorrect interpretation and treatment. | Ensure competence of individuals analysing and reporting NGS results: training records should be maintained regularly. |
Data analysis of sequence reads
Read processing/read quality
Alignment
Variant calling
INsertion/DELetion (indels)
Variant annotation and filtering
Variant interpretation
Reporting of results
Content of test report
• Patient identifiers | |
• Sample type e.g. FFPE, fresh frozen | |
• Tissue/tumour type e.g. lung, colorectal, melanoma | |
• Tissue sample identification e.g. unique block number | |
• Restatement of the clinical question | |
• Percentage of neoplastic content of sample used for NGS | |
• Extent of testing performed i.e. the genes and more specifically which regions tested e.g. exons, introns or hot spots analysed | |
• NGS method used e.g. platform, type of panel (amplicon, hybridisation), exome, WGS | |
• Sensitivity of the method i.e. percentage variant alleles detectable in a background of wild-type DNA | |
• Reference sequences for genes testeda
| |
• Results (using HGVS mutation nomenclature)a
| |
• How/where additional information about the analysis can be obtainedb
| |
• Interpretation and conclusion (see “Interpretation” and “Other reporting considerations” sections) |