CLONOSEQ® ASSAY TECHNICAL INFORMATION
Download the Full Technical SummaryDownload
The clonoSEQ Assay is an in vitro diagnostic that uses multiplex polymerase chain reaction (PCR) and next-generation sequencing (NGS) to identify and quantify rearranged IgH (VDJ), IgH (DJ), IgK, and IgL receptor gene sequences, as well as translocated BCL1/IgH (J) and BCL2/IgH (J) sequences in DNA extracted from bone marrow from patients with B-Cell acute lymphoblastic leukemia (ALL) or multiple myeloma (MM).
The clonoSEQ Assay measures minimal residual disease (MRD) to monitor changes in burden of disease during and after treatment. The test is indicated for use by qualified healthcare professionals in accordance with professional guidelines for clinical decision-making and in conjunction with other clinicopathological features.
The clonoSEQ Assay is a single-site assay performed at Adaptive Biotechnologies Corporation.
There are no known contraindications.
SPECIAL CONDITIONS FOR USE
- For in vitro diagnostic use.
- For prescription use only (Rx only).
SUMMARY AND EXPLANATION
The clonoSEQ Assay is an in vitro diagnostic assay that utilizes NGS to identify frequency and distribution of clonal sequences consistent with a malignant lymphocyte population in a sample.
Minimal Residual Disease (MRD) refers to the number of cancer cells that remain in a person during and following treatment. Clinical practice guidelines in hematological malignancies recognize that MRD status is a reliable indicator of clinical outcome and response to therapy. Studies in hematological malignancies have demonstrated the strong correlation between MRD and risks for relapse, as well as the prognostic significance of MRD measurements during and after therapy. In acute lymphoblastic leukemia MRD assessment has been established as an essential component in clinical management and is recommended to occur upon completion of initial induction and at additional time points based on the regimen used.3
In multiple myeloma MRD assessment after each treatment stage is recommended (e.g., after induction,
therapy/ASCT, consolidation, maintenance). MRD tests may also be initiated at the time of suspected
- MRD values obtained with different assay methods may not be interchangeable due to differences in assay methods and reagent specificity.
- The results obtained from this assay should always be used in combination with the clinical examination, patient medical history, and other findings.
- The clonoSEQ Assay is for use with bone marrow specimens collected in EDTA tubes.
- Results may vary according to sample time within the course of disease or by sampling site location.
- The assay may overestimate MRD frequencies near the limit of detection (LoD).
- The MRD frequency LoD varies based on the amount of DNA that is tested and using lower DNA input may prevent MRD detection at low frequencies.
- Sample processing and cell enrichment strategies may affect the measured MRD frequency.
- The volume and cellularity of sampled input material may affect the ability to detect low levels of disease.
- False positive or false negative results may occur for reasons including, but not limited to:
technical and/or biological factors.
The clonoSEQ Assay is a next-generation sequencing (NGS) based assay that identifies rearranged IgH (VDJ), IgH (DJ), IgK, and IgL receptor gene sequences, as well as translocated BCL1/IgH (J) and BCL2/IgH (J) sequences. The assay also includes primers that amplify specific genomic regions present as diploid copies in normal genomic DNA (gDNA) to allow determination of total nucleated cell content.
Testing begins with gDNA extracted from bone marrow (Figure 1). Extracted gDNA quality is assessed and rearranged immune receptors are amplified using a multiplex PCR. Reaction-specific index barcode sequences for sample identification are added to the amplified receptor sequences by PCR. Sequencing libraries are prepared from barcoded amplified DNA, which are then sequenced by synthesis using NGS. Raw sequence data are uploaded from the sequencing instrument to the Adaptive analysis pipeline. These sequence data are analyzed in a multi-step process: first, a sample’s sequence data are identified using the sample index sequences. Next, data are processed using a proprietary algorithm with in-line controls to remove amplification bias. When the clonoSEQ Clonality (ID) assessment is conducted, the immune repertoire of the sample is checked for the presence of DNA sequences specific to “dominant” clone(s) consistent with the presence of a lymphoid malignancy. Each sequence that is being considered for MRD tracking is compared against a B cell repertoire database and assigned a uniqueness value that, together with its abundance relative to other sequences, is used to assign the sequence to a sensitivity bin which will be used in the estimation of the reported LoD and LoQ on the patient report. During clonoSEQ Tracking (MRD) assessment, the complete immunoglobulin receptor repertoire is again assessed, and the previously identified dominant clonotype sequence(s) are detected and quantified to determine the sample MRD level. The clonoSEQ assay MRD assessment measures residual disease in a biologic sample.
Figure 1: clonoSEQ Assay Workflow
Following completion of these data processing steps, a report is issued. A Clonality (ID) report indicates the presence of dominant sequences residing within a presumed malignant lymphocyte clonal population, as identified in the baseline (diagnostic or high disease burden) sample from a patient. After one or more dominant sequence(s) have been identified in a baseline sample, subsequent samples from the same patient can be assessed for MRD after which a Tracking (MRD) report is generated. The MRD is expressed as a frequency that quantifies the level of residual disease based on the number of remaining copies of the initially dominant sequence(s) relative to the total number of nucleated cells in the sample.
The minimum DNA sample input requirement is 500ng. Shipment of 1ml of bone marrow aspirate (BMA) is requested to obtain sufficient extracted gDNA and extraction methods have been validated using volume ranging from 250 µL to 1 mL. The extraction method isolates DNA by first lysing cells and denaturing proteins after which the DNA is bound to a substrate. Once the DNA is bound, a series of wash steps removes impurities. Following the wash steps the DNA is eluted from the substrate. DNA is quantified using a spectrophotometer; the measured DNA concentration is used to add up to 20 µg of gDNA to the assay. The MRD test can be performed with 500 ng – 20 µg gDNA. Internal controls in the PCR and sequencing steps are used to confirm that sufficient gDNA has been amplified and that amplification was successful.
Genomic DNA is amplified using locus-specific multiplex PCR using V, D and J gene primers containing molecular barcodes to amplify IgH (VDJ), IgH (DJ), IgK, IgL, BCL1/IgH (J), BCL2/IgH (J) and housekeeping gene (HKG) sequences. Reaction-specific index barcodes for sample identification are added to the amplified receptor sequences by PCR. Sequencing libraries are prepared by pooling barcoded amplified DNA. qPCR is used to verify the adequacy of the pooled amplified DNA library concentration.
Sequencing and Data Analysis
Sequencing is conducted with the Illumina NextSeqTM 500 System. The sequencing process incorporates multiple quality checks. Sequencing output is then processed by the bioinformatics pipeline software as follows:
Flowcell Level Metrics
The analysis pipeline performs quality control (QC) checks on the flowcell data. The pipeline evaluates the percentage of reads that pass the Illumina quality filter (%PF), which must be greater than 70% of reads. The system uses spike-in PhiX templates to evaluate the error rate. The pipeline evaluates the proportion of PhiX reads, which must be greater than 2%, and the associated error rate as computed by the Illumina RTA software, which must be less than 3%.
Demultiplexing and FASTQ Generation
The pipeline uses Illumina software to demultiplex reads from the instrument output run folder. The analysis pipeline performs a QC check to evaluate whether unexpected barcodes are observed and raises a flag if more than 30,000 reads carry a barcode not specified in the input sample sheet.
Read Assignment to Receptors
The pipeline assigns reads to rearranged receptors for each sample after demultiplexing.
Clonal Sequence Determination
After assigning reads to receptor loci, the pipeline then clusters reads into clonal receptor sequences.
Sample Level QC Checks
The pipeline performs a series of sample level QC checks: assessment that sequencing data is sufficient and acceptable based on amplification of sets of internal synthetic controls, assessment that sufficient gDNA is sampled, and a final screen of the calculated values for biologic relevance. One set of internal synthetic controls are evaluated for sufficient read quantity per molecule and read coverage across receptor loci. Another set of internal controls’ presence or absence is used to screen for the expected degradation of residual primers. The estimated mass of input gDNA based on an optical density measurement and the estimated number of sampled nucleated cells based on amplification of a set of internal reference gene are used as metrics to check if sufficient material is sampled. The pipeline also checks that the detected numbers of total and B cells are within a biologically relevant range, and screens for clone sharing by evaluating if sequences are shared across samples that are processed together.
Clonal sequences are assessed for their suitability as ID sequences (to be used for subsequent tracking) by first aggregating highly similar sequences and requiring that the frequency of the sequence is at least 3% as a percentage of all sequences in the locus. The clone must also have a frequency of at least 0.2% of all nucleated cells in the sample and must have sufficient abundance and differentiation from a polyclonal background. Each sequence that is being considered for MRD tracking is compared against a B cell repertoire database and assigned a uniqueness value that, together with its abundance relative to other sequences, is used to assign the sequence to a sensitivity bin which will be used in the estimation of the reported limit of detection (LoD) and limit of quantitation (LoQ).
When a previous calibration test has identified suitable ID sequences for tracking, they are compared to sequences in the most recent tracking sample in order to assess residual disease. After approximate matching, which allows for mutations in the sample clones as compared to the ID sequences, sequence proportions in the sample are assessed and compared to the LoD and LoQ values. The analysis pipeline then reports whether ID sequences were detected above the LoQ, above LoD but below LoQ, below LoD, or not detected.
The following controls are used to measure the success of DNA extraction, PCR amplification and sequencing:
Synthetic Internal Controls
Each sample includes two sets of internal synthetic controls. The controls are panels of synthetic analogues of somatically rearranged B-cell receptor (BCR) immune receptor molecules. The composition of the reference template pools before and after amplification is measured and used for QC. One set of synthetic templates is added to every pre-amp PCR well as a positive control; these synthetic templates are used to measure primer performance, including identification and correction of amplification bias, and to screen for sufficient sequencing coverage. Another set is added after a step used to remove residual primers; the lack of amplification of these molecules is used to confirm the success of primer removal.
DNA Extraction Process Controls
Each extraction is performed with Positive and Negative Extraction Controls. The Extraction Negative Control is used to confirm lack of contamination during the extraction process. The Extraction Negative Control is subsequently amplified and sequenced in the same fashion as test samples. The Extraction Positive Control is included to assess effectiveness of the extraction process (it is required to be above a pre-set threshold for DNA recovery). If readily available, source material for Extraction Positive Controls is matched to the specimen source type. Exception: The Extraction Positive Control for bone marrow specimens consists of frozen human whole blood.
PCR Amplification Process Controls
Each PCR amplification is performed with an Amplification Positive and Negative Control and subsequently sequenced in the same manner as test samples. The Amplification Positive Control consists of gDNA derived from peripheral blood mononuclear cells (PBMCs) and serves as an additional check to confirm successful product amplification. Buffer (1xTE) is used as the negative control.
Sequencing Process Controls
To every sequencing flow cell, two sequencing controls are added. Both a PhiX control purchased from Illumina and a well-characterized amplified library (Sequencing Positive Control) are loaded with test samples.
The pipeline renders results into a PDF-formatted patient report. The report displays any ID sequences identified in the sample that can be used for tracking with their quantitation and sample-level metrics. For tracking tests, the report includes a result (ID sequences detected above LoD, below LoD, or not detected) and quantitation for the tracked sequences within the most recent sample.
Standards/Guidance Documents Referenced
CLSI guideline EP06-A Evaluation of the Linearity of Quantitative Measurement Procedures- A Statistical Approach.
All reagents, materials, and equipment needed to perform the assay, with the exception of sample collection materials, are used exclusively at the Adaptive Biotechnologies single laboratory site. The clonoSEQ Assay is intended to be performed with serial number-controlled instruments.
An ambient temperature sample shipper kit is optional for use and is available through Adaptive Clinical Services Team if requested by the ordering healthcare provider.
SAMPLE COLLECTION AND TEST ORDERING
For clonoSEQ sample collection requirements and ordering information please visit: www.adaptivebiotech.com/clonoseq/ordering
The clonoSEQ Assay is intended to be performed using the:
- Illumina NextSeqTM 500 Series System (qualified by Adaptive Biotechnologies)
2. PERFORMANCE CHARACTERISTICS
2.1 Sample Stability
2.1.1 Frozen Bone Marrow Stability at -15° C to -25° C
To demonstrate frozen bone marrow stability, four bone marrow samples from donors were aliquoted and stored frozen (-15° C to -25° C). Samples were tested after freezing and the frequencies of predefined clones within the specimen were compared to baseline. Frozen bone marrow was stable for 12 months at -15°C to -25°C based on all tested clone frequencies in all samples being within the prespecified ± 30% MRD frequency variation.
2.1.2 Bone Marrow Stability at Room Temperature and
To demonstrate bone marrow stability, four bone marrow samples from donors were aliquoted and stored at room temperature (19° C to 25° C) or refrigerated (2° C to 8° C) for up to seven days.
Bone marrow samples stored at room temperature (19° C to 25° C), remained stable for three days as established by MRD measurements remaining within the allowable ± 30%. For bone marrow samples stored refrigerated (2° C to 8° C), MRD measurements remained within the prespecified ± 30% frequency variation for at least seven days establishing seven days of stability.
2.1.3 Shipping Stability
Sample stability of bone marrow samples stored in Adaptive shipping containers at ambient temperature was tested after exposure to simulated summer and winter shipping conditions. Study results demonstrated that samples are stable for up to 4 days (96 hours), based on MRD frequencies remaining equivalent within the prespecified ± 30% under ambient shipping conditions.
2.1.4 Freeze/Thaw Stability of Bone Marrow Samples
The stability of bone marrow samples was evaluated using four bone marrow samples from donors split into aliquots (0.25ml) with one aliquot extracted upon receipt. The remaining aliquots were subjected to freeze/thaw cycles. gDNA was extracted and the concentration was determined using a spectrophotometer. Each sample was processed using the clonoSEQ Assay. Bone marrow samples subjected to three freeze/thaw cycles continued to report acceptable sample MRD frequencies remaining within the prespecified ± 30%.
2.2 Specimen Characterization
A panel of clinical specimens from 23 patients with MM and 21 patients with ALL was used for precision, quantitation accuracy and linearity studies. An additional 22 specimens from patients with other lymphoid malignancies were used to supplement the analytical sensitivity studies. Sample types included bone marrow (BMA), BMMCs, CD138+ bone marrow cells, peripheral blood, and PBMCs. gDNA was isolated from these clinical samples and blended with gDNA isolated from bone marrow to contrive specific MRD levels for the analytical studies.
A study was performed to evaluate whether MRD estimates from blended gDNA were equivalent to MRD estimates from blended cells at known concentrations. The accuracy and linearity of sample MRD frequency was assessed at zero and across 11 MRD frequency levels ranging from 3.3x10-7 to 3.0x10-3 for both blended gDNA and DNA extracted from blended cells. These dilutions included levels below LoD and spanned the range of reportable MRD levels. The MRD estimates on gDNA blends were comparable to the MRD estimates of the blended cells they were intended to mimic. Therefore, the blended gDNA from the clinical samples were determined to be functionally equivalent to clinical specimens for use in specific analytical studies.
Methods to isolate gDNA from BMA, bone marrow mononuclear cells (BMMCs), formalin fixed paraffin embedded (FFPE) bone marrow clot slides, and BMA smear slides were evaluated for performance in the clonoSEQ Assay. Studies were performed to determine extraction equivalence across four extraction runs with three variables (automated extraction instrument, operator, and reagent lot). Based on the results of all extractions, the tested gDNA extraction variables (operator, instrument, extraction reagent lot, and extraction run) met acceptance criteria. DNA isolated from FFPE bone marrow clot slides and BMA smear slides was only assessed for utility in identifying sequences and not MRD tracking.
Precision studies tested gDNA extracted from clinical specimens from 23 patients with MM and 21 patients with ALL. The gDNA from these specimens were used to contrive specific MRD levels by pooling and blending them into gDNA extracted from the BMA of healthy donors. The study included three DNA inputs (500 ng, 2 μg, 20 μg) and six MRD levels were tested at each DNA input for each patient sample. The studies were designed to test the MRD levels of 2.8x10-5, 8.0x10-5, 2.8x10-4, 8.0x10-4, 2.8x10-3 and 8.0x10-3 at 500 ng DNA input; 7.0x10-6, 2.0x10-5, 7.0x10-5, 2.0x10-4, 7.0x10-4 and 2.0x10-3 at 2 µg DNA input; and 7.0x10-7, 2.0x10-6, 7.0x10-6, 2.0x10-5, 7.0x10-5 and 2.0x10-4 at 20 µg DNA input. These frequencies correspond to an estimated 2.14, 6.13, 21.44, 61.26, 214.40 and 612.56 malignant cells tested at each DNA dilution.
The precision study used a main effects screening design over 21 calendar days. This study used ten runs, with two PCR plates each run, using three operator sets, four reagent lots, and four instrument sets (two thermal cycler/liquid handlers and two NextSeqTM instruments). The study design is summarized in Figure 2.
Figure 2: Precision Study Design Schematic
Each run of the assay tested 18 combinations of DNA input and MRD frequency in duplicate. In all, 360 contrived samples were tested. Of these, one plate with 18 samples was invalid due to sample QC failures; the plate-level failure rate was therefore 1 / 20 = 0.05. An additional two contrived samples (88 MRD measurements) failed sample QC due to insufficient sequencing coverage. While normal operating procedures permit re-sequencing, for this analysis these two samples were classified as failures. The analysis used the remaining 340 contrived samples with up to 44 MRD measurements per sample, for a total of 14,744 MRD measurements.
The precision of the clonoSEQ Assay is largely dependent upon the number of malignant cells that are being evaluated rather than the MRD frequency. Consequently, the same MRD frequency is expected to have lower precision at lower DNA inputs. For this study, precision estimates were first calculated based on the MRD frequency per DNA input followed by estimates of imprecision of the absolute number of malignant cells detected per reaction.
Precision of MRD Frequency for MM and ALL
Precision analysis, including variation from instrument set, operator, processing day, processing run, and reagent lot, is reported as %CV for each tested MRD frequency at each DNA input. The analysis was done separately for MM and ALL and is summarized in Tables 1 and 2. Precision ranged from 29.6% to 70.0% CV for MM and 25.9% to 76.9% CV for ALL.
In these tables, MRD frequency range refers to the central 95% range of MRD estimates that were observed across all of the patient samples tested at each DNA input and frequency condition. These data were used to define the 95% confidence intervals that are used in patient reports.
Table 1: Precision of the clonoSEQ Assay in MM Samples
Note: Some contrived samples included a subset of patient samples.
Table 2: Precision of the clonoSEQ Assay in ALL Samples
Precision of Malignant Cells Detected
The precision of malignant cells detected was evaluated across the range of tested malignant cells (2.14 – 612.56). For this analysis, the results from all of the DNA inputs across MM and ALL were pooled into a single analysis that is summarized in Table 3. As expected, the precision was primarily influenced by cell numbers being evaluated. Precision ranged from 72% CV at 2.14 cells to 19% CV at 612.56 cells. The majority of the observed variation is due to residual variability; the tested factors (Operator, Instrument Sets, Reagent Lots, Day, and Run) minimally contributed to variability with attributable %CV ranging from 0% to 3% (Table 3).
Table 3: Summary of the clonoSEQ Assay Precision
* These values were aggregated across diseases (ALL
and MM) and
total DNA input levels
The precision for each sample at each tested condition across all DNA inputs is summarized in a Sadler’s precision profile (Figure 3). The Sadler’s precision profile visualizes the relationship between the number of sampled malignant cells and precision as measured by %CV. This analysis demonstrates that the precision of the clonoSEQ Assay is largely dependent on the number of malignant cells that are being evaluated by the assay.
Figure 3: Sadler’s Precision Profile (Coefficient of Variation) of the clonoSEQ Assay as a Function of Input Cancer Cells
2.5 Analytical Sensitivity
2.5.1 Limit of Blank
The LoB was determined by measuring the specificity of a patient’s trackable immunoglobulin (Ig) sequences. These sequences were identified from 66 samples from patients diagnosed with a lymphoid malignancy (23 MM, 21 ALL, and 22 other malignancy). The LoB was determined by searching for the presence and abundance of these trackable sequences in healthy bone marrow samples. The 95th percentile of sample MRD frequencies for these trackable sequences was zero at 500 ng and 20 μg gDNA input. Therefore, the LoB was zero, demonstrating that trackable Ig sequences are highly patient-specific.
2.5.2 Limit of Detection/Limit of Quantitation
The LoD and LoQ were determined by blending gDNA extracted from 66 specimens from patients with lymphoid malignancies (23 MM, 21 ALL, and 22 with other malignancy) into 500 ng and 20 μg of gDNA from bone marrow. A dilution series of 22.97, 10.72, 4.59, 2.14 and 0.94 malignant cell equivalents was made for each patient at each DNA input level. Each sample was tested in duplicate for each of four reagent lots resulting in eight results for each of the 66 samples at each dilution condition. A probit approach was used to determine the LoD to be 1.903 malignant cells (95% CI; 1.75 – 2.07) based on the combined data from both DNA inputs (Table 4). The LoQ was defined as the lowest absolute number of malignant cells whose frequency can be quantitatively determined with an accuracy of 70% relative total error. The LoQ was found to be 2.390 malignant cells (95% CI; 1.90 – 9.14) (Table 4).
Table 4: LoD/LoQ by MRD Cell Counts and by MRD Frequency
*Calculated from samples with 500 ng and 20 μg of DNA input.
The clonoSEQ Assay can use a range of DNA inputs from 500 ng to 20 µg of DNA. The LoD/LoQ by MRD frequency will vary based on the DNA input and the total nucleated cells that are evaluated by the assay. The estimated LoD/LoQ at 500 ng and 20 µg of DNA input are shown in Table 4.
To confirm the LoD and LoQ, 1.903 and 2.390 malignant cell equivalents were spiked into 200 ng, 500 ng, 1 μg, 2 μg, 5 μg, 10 μg, 20 μg, and 40 μg of gDNA extracted from bone marrow of healthy subjects. The results showed that the LoD and LoQ of malignant cells detected remained consistent across all DNA input levels.
2.6 Analytical Specificity
2.6.1 Interfering Substances
Testing was performed to characterize the effects of five endogenous (Table 5) and three exogenous (Table 6) substances on the clonoSEQ Assay to identify potential interfering substances.
Table 5: Endogenous Interfering Substances Tested
* Chloroform was used assolvent to resuspend bilirubin (unconjugated) and cholesterol.
Table 6: Exogenous Interfering Substances Tested
* The BMA samples were shipped to Adaptive containing
EDTA (“Low” concentration) for anti-coagulation purposes. Additional EDTA was spiked in to
achieve the High
† Chloroform inhibition was tested at a single spiked-in volume (2.5µl).
The potential exogenous and endogenous substances were spiked separately into 250 μl aliquots of bone marrow from four different donors. Each condition was replicated for a total of eight times (four donors with two replicates each) and all conditions passed the pre-specified MRD frequency equivalence margin of ± 30%. This study concluded that MRD results were not substantially influenced by the presence of the tested interfering substances.
2.6.2 Cross-Contamination/Sample Carryover
The assessment of cross-contamination included two studies: one study to measure contamination of ID samples during automated DNA extraction of BMA and BMMC and one study to measure contamination of gDNA from MRD samples during PCR, library pooling, and sequencing with the clonoSEQ Assay.
Cross-contamination of ID samples during automated DNA extraction was assessed using a panel of three ALL cancer cell lines and three MM cell lines each spiked to 10% of total cells in a BMA pool of two healthy subjects or a bone marrow mononuclear cell (BMMC) pool of four healthy subjects. PBS (blank) samples were included in this study. Samples were evaluated as to whether they correctly calibrated. There were no false calibrations for run-to-run with 0/44 BMA and 0/44 BMMC false calibrations. There was one false calibration for the well-to-well study with 1/44 BMA and 0/44 BMMC samples falsely calibrating. The falsely calibrated sequence was found in a PBS sample with 83 total templates and the sequence was not associated with any of the six cell lines. The PBS sample provided a sensitive test for contamination since there was no background DNA and a contamination of 83 templates would not be expected to cause false calibration of a clinical specimen.
Cross contamination of incorrectly calling samples MRD positive was assessed using gDNA from peripheral blood from healthy subjects as MRD-negative specimens and blends of cell line gDNA and gDNA from peripheral blood of healthy subjects spiked to a concentration of 5%. The 5% level was used to simulate a patient with clinical relapse. This study evaluated for the presence of a clonal sequence and molecular barcode simultaneously. There were no run-to-run contamination events observed in 0/36 tests. Well-to-well cross contamination was observed in 8/712 comparisons; this was likely caused by contamination of a primer barcode plate sourced from a vendor. All contamination events were below 4x10-6. This low level of contamination is unimpactful because tracked clonotype sequences are highly specific to each patient, so contamination between samples from different patients would not affect the reported MRD result. Cross contamination between samples from the same patient is prevented by process controls that disallow co-processing of samples from the same patient.
Linearity of the clonoSEQ Assay using three ALL cell lines (SUP-B15, GM14952, and GM20930) and three MM cell lines (IM9, U266, and L363) was evaluated by blending cell line gDNA with gDNA from healthy subjects using DNA inputs of 200 ng, 2 µg and 20 µg gDNA and tested at zero and across 11 MRD frequencies at each DNA input. This study was performed to measure the linearity of the clonoSEQ Assay at depths beyond the sensitivity of conventional tools. The frequency range of 6.0x10-5 to 1.0 was tested at the 200 ng DNA input. The frequency range of 6.5x10-6 to 1.0 was tested at the 2 µg DNA input. The frequency range of 6.6x10-7 to 0.1 was tested at 20 µg DNA input. The linear range of the assay was determined by finding the input range where the maximum deviation from linearity (based on a quadratic or cubic fit to the data) was less than 5%. Linearity was established for each sample input and disease type tested across the entire tested range (Table 7), with data shown in Figure 4. This study demonstrated linearity of MRD frequencies across several orders of magnitude for each condition tested.
Figure 4: Linearity of the clonoSEQ Assay. Expected (x-axis) and Observed (y-axis) MRD Frequency of Six Cell Lines.
Table 7: Linearity of the clonoSEQ Assay using Cell Lines
Linearity using Clinical Specimens
Linearity was confirmed using clinical samples from the precision study (section 2.4), which evaluated blended gDNA extracted from 23 MM and 21 ALL specimens at three DNA inputs and six MRD frequencies per DNA input. These data were re-analyzed to confirm linearity at the lower frequency range of the assay. The linear range of the assay was determined by finding the input range where the maximum deviation from linearity (based on a quadratic or cubic fit to the data) was less than 5%. Results are summarized in Table 8. The slopes and intercepts are reported as the average and range of values across all clinical specimens that were tested at each DNA input by disease indication. Results from three representative specimens for each ALL and MM are shown in Figure 5. This study demonstrated linearity across a wide range of MRD frequencies for each condition tested using clinical specimens.
Table 8: Linearity of clonoSEQ Assay using Clinical Specimens
Figure 5: Linearity of clonoSEQ Assay. The Expected (x-axis) and Observed (y-axis) MRD Frequency of Six Clinical Samples
2.8 Reagent Stability
2.8.1 In-Use Reagent Stability
An in-use stability study was executed to determine stability needs of the clonoSEQ Assay for reaction mixes and intermediate steps. The following critical steps were evaluated: pre-amp and PCR primer mix stability, master mix stability, complete reaction stability, and process pause stability. gDNA was tested using seven replicates for all conditions tested. Acceptance criteria were based on sequencing results meeting all QC metrics; all of the conditions tested met the pre-specified acceptance criteria and the clonoSEQ Assay in-use stability needs.
2.8.2 Real Time Stability of Pre-Amp and PCR Mixes
The real-time reagent stability studies used the primer QC processes to assess primer performance and determine primer stability. The primer QC process uses a set of synthetic double-stranded molecules representing rearrangements of the targeted exons to determine whether each manufactured lot of pre-amp PCR primers and PCR primers are performing within specification. The priming sites on synthetic molecules are identical to biologic priming sites on targeted exons. Data from these molecules were analyzed and assessed for the ability of the primers to amplify each identified exon at acceptable levels and the presence of primer sequences. These data were used to confirm that the performance of the pre-amp and PCR primers was adequate and consistent with previous primer lots. The performance of the amplification of the synthetic molecules met the pre-specified acceptance criteria.
This real-time reagent stability study established a 12-month shelf life of pre-amp and PCR primer mixes when stored at -20 ± 5° C. These data were confirmed by assessing the equivalence of MRD frequency in 40 clinical samples amplified with primer lots of different ages, and by tracking the stability of MRD measurements of synthetic molecules over time. The conditions tested in the real time stability study met the pre-specified acceptance criteria of a pairwise equivalence test of clinical specimens to be within ± 30% MRD frequency.
2.9.1 Quantitation Accuracy
The analytical quantitation accuracy of the clonoSEQ Assay for MRD testing was assessed relative to multiparametric flow cytometry (mpFC):
- Assessment of the clonoSEQ Assay accuracy in cell mixtures compared to mpFC
- Concordance of the clonoSEQ Assay and mpFC in two clinical validation studies: ALL and MM
- Assessment of the clonoSEQ Assay accuracy in 44 clinical samples
2.9.2 Assessment of clonoSEQ Assay Accuracy in Cell Mixtures Comparing to
Multiparametric Flow Cytometry (mpFC)
Accuracy was assessed using cell line blends. Measured MRD frequencies were compared against known frequencies based on diluting cell lines into background mononuclear cells at specific MRD levels. The mpFC lab screened a panel of cancer cell lines and selected 2 MM and 2 ALL cancer cell lines that performed well with mpFC.
This study evaluated two MM cell lines (U266B1 and NCI-H929) and two ALL cell lines (SUP-B15 and GM20390); the mpFC laboratory screened a panel of cancer cell lines and selected these four cell lines based on performance of the mpFC assay. Each cell line was tested at five dilutions from 5x10-7 to 1x10-2. Two replicates of each sample were assessed by the clonoSEQ assay and mpFC. A pairwise comparison of MRD frequency measurements is shown in Figure 6. This study demonstrated similar quantitative accuracy comparing clonoSEQ with mpFC at frequencies above 1x10-4.
Figure 6: Pairwise Comparison of MRD Frequency Measurements from mpFC (x-axis) and the clonoSEQ Assay (y-axis)
2.9.3 Concordance with mpFC in Clinical Samples
Two concordance studies between mpFC and the clonoSEQ Assay were performed using clinical samples. For both studies concordance was assessed two ways: concordance of MRD positive or negative calls and concordance of reported MRD frequency. One study used 273 ALL samples from the Children’s Oncology Group (COG) AALL0331 (standard risk) and AALL0232 (high risk) regimens and compared the clonoSEQ Assay to a validated mpFC assay. The other study performed a similar comparison using 91 MM samples from the Dana Farber Cancer Institute (DFCI) Study 10-106 that were measured by both the clonoSEQ Assay and mpFC. MRD negativity was defined as < 1x10-4 for mpFC in ALL (a commonly used threshold in that patient population) and <1x10-5 for mpFC in MM. For the clonoSEQ Assay, MRD calls were assessed at the LOD in both studies. The positive percent agreement (PPA) between the clonoSEQ Assay and mpFC was 93.5% for ALL and 97.9% for MM. Negative percent agreement (NPA) reflects the higher sensitivity of the clonoSEQ Assay with 117 ALL and 23 MM cases reported as positive for clonoSEQ and negative for mpFC (Table 9).
Table 9: Summary of mpFC vs. the clonoSEQ Assay Concordance Data for ALL and MM
Concordance of MRD frequency was visualized by plotting reported MRD frequency of mpFC against the clonoSEQ Assay for both MM and ALL (Figure 7). Concordance of MRD call is indicated by color; blue circles indicate samples had concordant MRD positive calls, while orange triangles and red squares denote discordant calls, with orange triangles indicating that clonoSEQ identified the sample as MRD positive and red squares indicating that mpFC identified the sample as MRD positive. To simplify the plot, samples with concordant MRD negative calls were not plotted. To quantify the similarity of reported MRD frequencies, correlations were calculated for samples with either concordant MRD calls or mpFC positive calls; MRD frequencies were highly concordant (ALL, concordance correlation coefficient = 92.8%; MM, concordance correlation coefficient = 91.9%). These data demonstrate that at high disease burdens mpFC and clonoSEQ report similar MRD levels, while clonoSEQ continues to detect MRD at lower frequencies.
Figure 7: Measurements of the clonoSEQ Assay Compared to Flow Cytometry Measurements from ALL (left) and MM (right) Clinical Studies.
2.9.4 Analysis of Quantitation Bias on Clinical Specimens
The precision study described evaluated blended gDNA extracted from 23 MM and 21 ALL specimens at three DNA inputs and six MRD frequencies per DNA input. These data were reanalyzed to evaluate if there was a quantitation bias for clonoSEQ. Sample MRD frequencies measured with the assay were compared to the expected MRD value, as calculated using flow cytometry on the original clinical sample and applying the appropriate dilution factor.
For ALL and MM, the quantitation accuracy of the clonoSEQ Assay was within ±25% across all tested diseased cell inputs (Figure 8). The assay tended to have a modest upward bias in MRD estimation at lower MRD frequencies and a modest downward bias at higher MRD frequencies (Figure 8). These biases were deemed to be minimal and acceptable.
Figure 8: Bias Estimates of the clonoSEQ Assay by MRD Frequency using Clinical Specimens
2.9.5 Repeatability of Nucleotide Base Calls
The repeatability of sequences generated by the clonoSEQ Assay was assessed using a two-step process. First, ID samples from 72 lymphoid malignancy samples and nine cell lines were processed to determine the baseline calibrating clonotype nucleotide sequences. Next, 20 replicates of the samples were run at disease inputs of ~2 to 600 malignant cells across four DNA inputs (10 ng, 500 ng, 2 μg, 20 μg). The replicates were tested using three operators, two instrument sets, and four reagent lots. These data were used to assess the observed rate of agreement between the nucleotide sequences chosen for tracking and the nucleotide sequences observed in contrived samples from the same biological specimens.
For each calibrating clonotype sequence in an ID sample, all sequences in the corresponding MRD samples within N bp were included for assessment of overall percent agreement (OPA), where N is defined for each sequence as the number of allowable mutations determined during specimen characterization by our calibration algorithm. N is chosen to capture somatic variation among B cells from the same clonal lineage without incorrectly grouping sequences from different clonal lineages. Once this population was established, the OPA between the original calibrating clonotype sequence and the sequences identified in the MRD assessment was calculated.
Table 10 reports the number of nucleotides assessed, the OPA, the lower and upper 95% confidence limits, and the OPA restated in the same terms as a Phred quality score (i.e., -10 x log10 disagreement rate). This test assessed approximately 442.5 million nucleotides for sequence agreement, with an overall disagreement rate of approximately 3.5 parts per 100,000 (corresponding to a Phred score of about 44.5; in typical NGS applications a Phred score of 30 or higher constitutes a high-quality base call).
Table 10: Summary of Sequence Agreement Metrics
The observed sequence error rates are extremely low. Sequence error does not represent a reasonable risk for generating false negative results even when a single copy of the relevant MRD sequence is observed.
2.10 Amplification Bias by Clonotype
Two types of studies were executed to assess amplification bias. One study used a comprehensive panel of synthetic double-stranded molecules representing rearrangements of the targeted exons, while the other used clinical samples. Data from amplification of the synthetic templates demonstrate that the clonoSEQ Assay amplifies the targeted exon segments efficiently and consistently with nominal bias. These conclusions were supported by data from clinical samples which show that patients who carry certain exons in their malignant clonotypes do not have biased precision profiles.
3. CLINICAL STUDIES
Clinical validation was demonstrated using an analysis of samples obtained from two previously conducted clinical studies in ALL, one ongoing study in MM and an analysis of a completed study in MM. Samples for the analysis of the clonoSEQ Assay performance in MM were obtained from an ongoing randomized, open label, Phase III Study of a lenalidomide and bortezomib in a combination therapy regimen (DFCI Study 10-106). Multiple timepoints were assessed in this two-arm analysis and not all patients have the same number of MRD assessments (see data in section 3.2). Patients on Arm A (blinded to Adaptive Biotechnologies) had assessments after eight cycles of RVD, and then after lenalidomide maintenance. Patients on Arm B (blinded to Adaptive Biotechnologies) were assessed following three cycles of RVD, following auto transplant and again after two more cycles of RVD consolidation, and then following lenalidomide maintenance.
3.1 Clinical Validation of the clonoSEQ Assay for Acute Lymphoblastic
Children’s Oncology Group (COG) Studies AALL0232 and AALL0331
The primary objective of the study was to establish the ability of the clonoSEQ Assay to predict event-free survival (EFS) at the MRD threshold of 10-4 using available bone marrow samples from patients who were enrolled in previously conducted COG studies AALL0232 and AALL0331. The study was also designed to evaluate the clinical utility of the clonoSEQ Assay using alternative MRD thresholds and continuous MRD measures.
COG study AALL0331 is a Phase III randomized study of different combination chemotherapy regimens in pediatric patients with newly diagnosed standard risk B-precursor acute lymphoblastic leukemia. COG study AALL0232 is a Phase III randomized study of dexamethasone versus prednisone during induction and high-dose methotrexate with leucovorin rescue versus escalating-dose methotrexate without leucovorin rescue during Interim Maintenance I in patients with newly diagnosed high-risk acute lymphoblastic leukemia. Within these studies, bone marrow was collected at six separate time points to assess response to treatment; however, only the post induction marrow was used for MRD determination.
Clinical samples (pre-treatment BMA and day 29 post-induction BMA) were collected from 619 individuals, with samples from 315 patients who were enrolled as part of the “high risk” COG protocol AALL0232 and samples from 304 patients enrolled as part of the “standard risk” COG protocol AALL0331. Available specimens from these trials were tested with the clonoSEQ assay and results from both studies were pooled into a single analysis. Specimens were selected based on having a sufficient quantity of gDNA, available MRD flow cytometry results and patients with study related endpoints for EFS and overall survival.
A subset of 283 of the 619 patients originally enrolled in COG studies AALL0232 and AALL0331 had leftover samples of sufficient amount that could be tested with the clonoSEQ Assay. The population characteristics between these 283 patients were compared against the remaining 336 that were not tested and there were no significant differences in any characteristic that was evaluated, including age, gender, presence of specific genetic fusions, trisomy, and progression free survival. The 283 bone marrow specimens were tested to evaluate the clinical performance of the clonoSEQ Assay and to demonstrate concordance in MRD measurements between the clonoSEQ Assay and results of original testing with a previous version of the clonoSEQ Assay and mpFC. Ten specimens did not pass QC, leaving results from 273 specimens available for the final analysis.
The clonoSEQ Assay MRD negativity at ≤ 10-4 was found to predict improved EFS irrespective of age (P=0.0034; Figure 9). Results demonstrate a 2.74-fold higher event risk in MRD positive patients (MRD > 10-4) compared to MRD negative patients (95% CI: 1.330-5.656). Similar findings were published in a broader COG analysis of the relationship between EFS and MRD negativity by an earlier version of the clonoSEQ Assay in pediatric ALL (Wood et al. 2018).
Figure 9: Kaplan-Meier Survival Curve for EFS using the clonoSEQ Assay at an MRD Cutoff of 10-4 in ALL
Cox regression analysis of MRD and EFS using continuous MRD values demonstrates that the clonoSEQ Assay is significantly associated with EFS after adjusting for age (P=0.0057) and that each 10-fold increase in MRD level is associated with a 1.499-fold increase in event risk (95% CI: 1.139-1.974). These data further demonstrate that the MRD level remains a significant predictor of EFS even after accounting for age, gender, and genetic abnormalities, which demonstrates the utility of MRD measurement in ALL.
Qualitative assessments of MRD were also evaluated with MRD negativity defined as ≤ 10-5 (Figure 10) is significantly associated with EFS (P=8.4x10-4).
Figure 10: Kaplan-Meier Survival Curve for EFS using the clonoSEQ Assay at an MRD Cutoff of 10-5 in ALL
The clonoSEQ Assay was used to assess MRD at various disease burden thresholds to determine the correlation of MRD level with EFS. Patients who are clonoSEQ MRD negative (≤10-5) have longer EFS, followed by patients with MRD between 10-5 and 10-4 and patients with MRD ≥10-4 (P=6.5x10-4; Figure 11). These data demonstrate that patients with the lowest levels of MRD have better outcomes than patients with higher disease burden regardless of risk stratification.
Figure 11: Kaplan-Meier Survival Curve for EFS using the clonoSEQ Assay with Three MRD Bins in ALL: ≤10-5, 10-5 – 10-4, ≥10-4.
These analyses demonstrated that MRD estimation by the clonoSEQ assay is associated with patient outcomes for B-cell precursor ALL.
3.2 Clinical Validation of the clonoSEQ Assay for Multiple
Two separate studies were analyzed to support that MRD as estimated with the clonoSEQ Assay is associated with patient outcomes in MM, including the DFCI Study 10-106 and the ALCYONE study.
The objective of this study was to establish that the clonoSEQ Assay is predictive of progression-free survival (PFS) and disease-free survival (DFS) in MM. Patient samples were accrued under DFCI Study 10-106, “A Randomized Phase III Study Comparing Conventional Dose Treatment Using a Combination of Lenalidomide, Bortezomib, and Dexamethasone (RVD) to High-Dose Treatment with Peripheral Stem Cell Transplant in the Initial Management of Myeloma in Patients up to 65 Years of Age.”
A subset of 365 of the 720 patients originally enrolled in DFCI Study 10-106 had leftover samples of sufficient amount to be tested with the clonoSEQ Assay. The population characteristics between these 365 patients were compared against the remaining 355 patients that were not tested and there were no significant differences in any characteristic that was evaluated, including age, gender, ISS staging, cytogenetic status and progression free survival. Samples from 365 patients were tested and results from 323 patients were evaluable and passed QC. Seventy-five of the samples were available from patients in complete response (CR) at the time of first MRD assessment. This study aimed to demonstrate the association of the first MRD measurement with DFS in patients who achieved CR and with PFS in all evaluable patients.
Samples from 75 patients who had achieved CR were evaluable for analysis. Continuous clonoSEQ MRD levels were modestly associated with DFS in patients who have achieved CR (P=0.064) such that patient with lower MRD levels were less likely to progress.
The ability of the clonoSEQ Assay MRD measurements to predict PFS in all 323 evaluable patients was also assessed. clonoSEQ measurements demonstrated that MRD status at a threshold of 10-5 significantly predicts PFS in all patients (P= 0.027, Figure 12).
Figure 12: Kaplan-Meier Survival Curve for PFS using the clonoSEQ Assay at an MRD cutoff of 10-5 in MM
Cox regression analysis using a continuous measure of MRD was also associated with disease progression (P=1.9x10-7). For every 10-fold increase in continuous clonoSEQ MRD measurement, the likelihood of an event is 1.69 times higher (95% CI:1.071-2.67).
The ALCYONE Trial was a multicenter, randomized, open-label, active-controlled phase 3 trial that evaluated daratumumab plus bortezomib, melphalan and prednisone (D-VMP) versus bortezomib, melphalan and prednisone (VMP) in 706 patients with newly diagnosed multiple myeloma who were ineligible for stem-cell transplantation. The result of this study was reported in (Mateos et al. 2018).
Within this trial, MRD was assessed by the clonoSEQ Assay using of BMA collected at screening, at the time of confirmation of complete response or stringent complete response, and at 12, 18, 24, and 30 months after the first dose in patients having a complete response or stringent complete response. Patients who did not achieve a CR were considered to be MRD positive. An MRD threshold of 10-5 was used for analysis.
Regardless of treatment group, patients who were MRD negative by the clonoSEQ Assay at ≤10-5 had longer PFS compared to MRD positive patients (Figure 13). In patients with persistent MRD, PFS was longer in the daratumumab group than in the control group.
Figure 13: Analysis of MRD with Progression-Free Survival. Patients who were MRD negative by the clonoSEQ Assay had longer PFS compared to MRD positive patients.
In summary, MRD negativity as measured by the clonoSEQ Assay was associated with improved patient outcomes in studies of ALL and MM. These data support the use of the clonoSEQ Assay to measure MRD and to be used for monitoring MRD in patients diagnosed with ALL and MM, to monitor changes in burden of disease during and after treatment.
In summary, the MRD negativity as measured by the clonoSEQ assay was associated with improved patient outcomes in studies of ALL and MM. These data support the use of clonoSEQ to assess MRD and monitor residual disease in patients diagnosed with ALL and MM, in order to track changes in burden of disease during and after treatment. In samples from patients diagnosed with ALL and MM, the clonoSEQ Assay provides robust quantitative measurements of MRD frequencies from 10-6 - 10-4, with sensitivity increasing in proportion to the amount of input material. Clinical outcomes are strongly associated with MRD levels measured by the assay for the purposes of detecting and monitoring residual disease in patients diagnosed with ALL and MM, in accordance with clinical guidelines.
1. Borowitz M.J., et.al., Prognostic significance of minimal residual disease in high risk B-ALL: a report from Children's Oncology Group study AALL0232. Blood. 2015 Aug 20;126(8):964-71.
2. Mateos M.V., et.al., for the ALCYONE Trial Investigators, Daratumumab plus Bortezomib, Melphalan, and Prednisone for Untreated Myeloma. New England Journal of Medicine. 2018 Feb 8;378(6):518-528.
3. NCCN Clinical Practice Guidelines in Oncology: Acute Lymphoblastic Leukemia. Version 1. 2018.
4. NCCN Clinical Practice Guidelines in Oncology: Multiple Myeloma. Version 1.2018.
5. Randomized Trial of Lenalidomide, Bortezomib, Dexamethasone vs High-Dose Treatment with SCT in MM Patients up to Age 65 (DFCI Study 10-106), NCT01208662
6. Wood B, et.al., Measurable residual disease detection by high-throughput sequencing improves risk stratification for pediatric B-ALL. Blood. 2018 Mar 22;131(12):1350-1359.
6. SPECIMEN AND SHIPPING INSTRUCTIONS
Please refer to the clonoSEQ web site for detailed requirements: