Research Group
- Assist. Prof. Howard Slater, Principal Investigator
- Prof David Power, Co-Investigator
- Assist. Prof. Frank Ierino, Co-Investigator
- Dr Bruno Damien, Co-Investigator
- Ms Devika Ganesamoorthy, Res. Associate
- Assist. Prof. Nigel Toussaint, Collaborator
- Assist. Prof. John Kanellis, Collaborator
Location
- Murdoch Childrens Research Institute, Parkville, Australia
Title
- A Novel, Simple, Universal Blood Test Based on Copy Number Variants for Organ Transplant Monitoring
Aim 1: To develop a non-invasive, sensitive blood test consisting of copy number variation (CNV)-deletion assays to quantify 'donor-derived' DNA released into the recipient's plasma from the transplant organ.We have developed a panel consisting of 10 copy number deletion (CND) assays that allow estimation of graft-derived cell-free DNA (gdcfDNA) using real-time quantitative PCR. Validation studies performed in triplicate revealed a cycle threshold coefficient of variation (CtCV%) of 0.1% to 1.7% (median 0.6%) for cfDNA target concentrations between 16-16,000 GE/mL (Figure 1). The CtCV% (median) for 4 GE triplicate reactions was 4.4 fold greater (2.6%) than the median value for16-16,000 GE reactions, suggesting a lower limit of quantification of 16 GE/mL. We have published and presented the concept and assay details (1, 2). The concept has also been patented (3).We also modelled and observed the likelihood that each CND would be homozygous-null in the recipient but one- or two-copy in the graft (an "informative marker"). We predicted that a panel of 10 CNDs would result in three or more informative markers in 38% of recipients and no informative markers in 13% (Figure 2). The observed distribution matched this modelling, however initial expectations that one or two informative markers would be sufficient to confidently calculate the gdcfDNA resulted in limited data quality. It is now clear that confident, accurate determination of gdcfDNA will benefit from at least three informative markers in all patients. This facilitates calculation of error bars, allowance for two or more CNV in the graft and redundancy in the event of aberrant results or failed assays. To address this, we are expanding our CND panel to 30 assays which should yield at least 3 informative CNDs in 99% of unrelated donor-recipient pairs and at least 5 informative CNDs in 89% (Figure 2).The experience gained from this work has highlighted the inherent methodological weakness of real-time quantitative PCR (rtPCR). This technique relies on reference standard curves for quantification. The result is relative quantification rather than absolute quantification which is dependent on theoretical exponential amplification performance of PCR and limits performance accuracy and precision. Additionally, the technology is reliant on arbitrary specification of a number of technical variables which significantly impact results (such as determining the cycle threshold). All of these issues introduce error into the methodology. With better availability of emerging technologies such as digital PCR and increasing concern within the literature regarding real-time PCR, we feel a move to a digital PCR platform is an important step in preparing the panel for commercial application.Digital droplet PCR (ddPCR) is a relatively new method which importantly provides absolute quantification of nucleic acid that is less dependent on PCR reaction efficiency and is accurate at low target concentrations. Each PCR reaction is distributed into approximately 15,000x1 microlitre water-in-oil emulsion droplets. The distribution of target DNA sequence into droplets is dependent on concentration. After the PCR reaction is run with a hydrolysis probe, a droplet reader quantifies fluorescence in each individual droplet. The number of positive droplets divided by the total number of generated droplets is used to calculate the target concentration.An additional benefit following on from the introduction of a 30 CND assay panel and use of digital PCR will be streamlined recipient genotyping. Our prior methodology required melt-curve analysis to genotype recipients for homozygous-null CNDs which could be subsequently assayed for informative cfDNA quantification. Using our new approach, genotyping will occur concurrent with data generation on the first sample for each patient, simplifying the workflow and reducing overall costs. Subsequent samples will be targeted to informative markers only, saving further on reaction costs.Aim 2: To determine inter-individual variation in baseline levels of 'donor-derived' DNA in the plasma of 'stable' transplant recipients.The level of gdcfDNA in 69 stable patients is 30-75 GE/ml (30GE/ml is very near the limit of qPCR quantification). To make sound correlations with clinical events and biopsy histopathology, intra-patient variation levels over time will need to be determined.Aim 3: To measure 'donor-derived' DNA levels in the plasma of patients with graft damage.We have recruited 140 patients into the study and have collected >400 plasma samples. Analysis at one year, indicated in Figure 3, shows distinctly higher gdcfDNA levels in patients with antibody-mediated rejection (AMR) but no clear distinction in 'stable' and cell-mediated rejection (CMR) patients (4). Due to the significant lymphocyte infiltration and tubulointerstitial distribution of inflammation in CMR, we feel that consideration of tcfDNA and donor fraction in the diagnostic approach as well as quantification of these markers in urine may improve sensitivity of this test for this condition. We have applied our methodology to urine samples and have been able to successfully quantify gdcfDNA in these samples (Figure 4).Using a serial sampling strategy, we have observed high levels of gdcfDNA during the immediate post-transplant period which rapidly drop to baseline levels in stable individuals without rejection on biopsy (Figure 5). This observation is consistent with other groups and likely reflects gdcfDNA release as a result of ischemia-reperfusion injury(5). Additionally, consistent with other groups, we have also noted that gdcfDNA appears to correlate with successful treatment of rejection and treatment failure (Figure 6).
Figure 1. The cycle threshold coefficient of variation (CtCV%) for each CND assay is shown across the linear dynamic range. Between 16-16,000 GE, the CtCV% ranged from 0.1% to 1.7% (median 0.6%). The CtCV% (median) for 4 GE triplicate reactions was 4.4 fold greater (2.6%) than the median value for 16-16,000 GE reactions. This is consistent with a lower limit of quantification of 16 GE for these CND assays. Notably, reaction failures were recorded only for 1 GE reactions (typically 1-2 out of 3 replicates failed) with no differences between CND qPCR assays.
Figure 2. Observed and simulated projections of the number of informative markers obtained using panels of 10 (existing panel), 20 and 30 CNDs in unrelated donor-recipient pairs. The cumulative proportion for 0, at least 3 and at least 5 informative markers if given in the inset table.
Figure 3. Linear mixed effects model on log2 transformed cfDNA level data (accounting for repeated measures arising from different subsets of markers assayed for each individual i.e. random effects for individuals and for markers) showing completely separate levels in the 8 patients with biopsy-determined AMR (fixed effect estimate 8.39 with 95% CI [7.57,9.20]) compared to 'stable' patients (n=69) and 4 patients with biopsy-determined CMR (fixed effect estimate 4.6 with 95% CI [3.76,5.47]).
Figure 4. Relatively higher-levels of gdcfDNA measured in the urine of 2 patients with biopsy-defined CMR, 1 patient with biopsy-defined AMR and 1 'stable' patient. The matched plasma gdcfDNA measurements appear in grey. The significance of this relationship requires further study.
Figure 5. Graft-derived cfDNA measured in two individuals plotted over time since kidney transplant (day 0). In both cases, gdcfDNA levels are high for the first two weeks post-transplantation, after which a stable baseline below 50 GE/mL is established. Indication biopsy at 105 days in case 1, a sensitised recipient treated with peri-operative plasma exchange and intravenous immunoglobulin (IVIg), revealed no rejection consistent with stable gdcfDNA levels.
Figure 6. Graft-derived cfDNA measured over time in two individuals treated for biopsy-proven AMR. In case 3, successful treatment is confirmed with establishment of baseline gdcfDNA levels which correspond with absence of AMR on follow-up biopsy. In case 4, treatment failure is evidenced by persisting elevated gdcfDNA levels which correspond with persisting AMR on follow-up biopsy.
References
- D. L. Bruno et al., Use of copy number deletion polymorphisms to assess DNA chimerism. Clin Chem 60, 1105 (Aug, 2014).
- D. Ganesamoorthy, D. L. Bruno, F. L. Ierino, D. A. Power, H. R. Slater, paper presented at the Combined Meetings of Australasian Mutation Detection & Molecular Genetics Society of Australasia, Port Douglas, Queensland, Australia, 17-21 September 2012.
- D. L. Bruno, H. R. Slater. (2013), vol. WO/2013/049892.
- D. L. Bruno et al., paper presented at the Australian and New Zealand Society of Nephrology Annual Scientific Meeting, Melbourne, Victoria, Australia, 25-27 August 2014.
- I. De Vlaminck et al., Circulating cell-free DNA enables noninvasive diagnosis of heart transplant rejection. Science translational medicine 6, 241ra77 (Jun 18, 2014).
