NGS-Based Minimal Residual Disease (MRD) after Stem Cell Transplantation (SCT) Is More Specific for Relapse Prediction Than qPCR and Suggests the Possibility of False-Positive qPCR Results

Background

MRD detection using qPCR for IG/TR rearrangements has been the gold standard approach for relapse prediction and for guiding post-transplant therapeutic interventions for already two decades. Despite its thorough standardization within the EuroMRD laboratories and very strict rules for defining MRD positivity (van der Velden, Leukemia, 2007) this method carries the risk of false-positive MRD results, mainly in post-transplant time-points with massive B-cell regeneration (Fronkova, Bone Marrow Transplantation, 2008). This may lead to application of unneeded therapeutic interventions and to the misinterpretations in the post-SCT survival analyses.

Our previous study employing amplicon NGS-based MRD detection in frontline paediatric ALL patients (Kotrova, Blood, 2015) showed, that the sensitivity of NGS and qPCR is comparable, whereas NGS was proven to have better specifity for relapse prediction. We hypothesized that amplicon NGS of IG/TR rearrangements, providing the sequences of entire CDR3 regions, is a perfect tool for distinction between false and real post-transplant positivities.

Aims

To detect MRD by amplicon NGS in:

1) qPCR low-positive (non-quantifiable) samples from paediatric ALL patients after allogeneic SCT, who remained in long-therm complete remission

2) qPCR low-positive post-transplant samples from a control cohort of paediatric ALL patients who subsequently relapsed

Methods

Sequencing libraries were prepared from 500ng of bone marrow/peripheral blood DNA employing two-round PCR. In the first PCR virtually all IGH or TRG rearrangements present in the investigated sample are amplified using universal V(D)- and J-regions primers. In the second round, sequencing adaptors for Ion Torrent sequencers and sample-specific barcodes are added.

Libraries were sequenced on Ion Torrent PGM/Ion Proton sequencers using the Hi-Q chemistry, according to the manufacturer's instructions. For the detection of clonal leukemia-specific IG/TR rearrangements we developed own in-house bioinformatics algorithm. NGS-MRD positivity was defined as the presence of leukemia-specific sequence(s) as identified at diagnosis in the follow-up samples.

Results

We sequenced 27 qPCR-positive (not quantifiable) post-transplant samples from 17 patients (14 patients carrying IGH and 3 patients TRG targets) who remained in complete remission after SCT with a median follow-up of 10.4 years (range 2.2-15.8). Only 1 of 27 (3.7%) follow-up samples was found to be MRD-positive by NGS. This sample belonged to the patient with the shortest follow-up and unclear outcome so far (2.2 years). Also, parallel testing in 6 patients with available fusion gene (BCR/ABL1, ETV6/RUNX1, KMT2A/MLLT3, KMT2A/MLLT10, KMT2A/MLLT1) revealed negative results. No treatment interventions were applied in this group of patients, except for two of them who received 4xDLI/Glivec.

The control cohort consisted of 13 patients who subsequently relapsed. We re-evaluated 15 low-positive qPCR post-transplant samples from these patients by NGS. Ten of 15 samples were also positive by NGS, which significantly differed from the non-relapse group (1 of 27 vs. 10 of 15, p=0.002).

To find the cause of false positive qPCR results we searched for matches of the clone-specific qPCR primers sequences from samples which were qPCR-positive and NGS-negative in all resulting fastq files. For all but one primers the stretches of nucleotides matching the primer sequences with maximum 2bp mismatch were present in fastq files from different patients, thus likely explaining the possibility of non-specific amplification due to random annealing of primers to regions with a high sequence similarity.

Conclusions

Our results indicate that NGS is more specific in relapse prediction also in regenerating post-transplant samples and that in this specific situation of massive regeneration after the transplant qPCR can detect false positive results, despite high degree of standardization and strict rules for defining positivity. Therefore low (positive, not quantifiable) qPCR positivities after SCT should be interpreted with great caution. If possible we recommend to perform repeated sampling, to check the length of PCR product and also check the complementary MRD targets.

Support: AZV 16-32568A, NPU I LO1604, GACR P302/12/G101, GAUK 394214, 00064203 (Ministry of Health, Czech Republic), AZV 15-31540A.