In an earlier post I noted that pathology experts believe next generation sequencing will be required to bring minimal residual disease (MRD) into focus as a more valuable endpoint for clinical trials of leukemia treatments. I’ve reviewed the technique, which starts with a bone marrow sample from which the laboratory isolates the DNA, increases the total amount through PCR amplification, and then sequences the genes using fluorescent markers. The different genes are counted by type, and those associated with tumor cells divided by the total number of cells in the sample to produce a measurement of MRD.

The agency’s proposed target for sponsors to use MRD as a disease-free endpoint is a measured amount of tumor cells that is <0.01% of total cell number. This is a pretty ambitious target, since it means being able to detect less than 1 in 10,000 tumor cells in a sample that likely contains a wide variety of cell types as well as cellular debris following treatment.

The biggest problem I see with next generation sequencing is the lack of a true sample enrichment or matrix removal process. Bone marrow samples for MRD are often low in overall numbers of cancer cells, and preparations of these samples for sequencing will thus have low numbers of gene fragments specific for those cells mixed in with gene fragments from non-cancerous cells. Amplification by PCR is performed, which will reduce the amount of non-genetic material. However, this process only increases the number of all gene fragments by straightforward copying and does not do anything to increase the signal (the number of cancer gene fragments) relative to the noise (the number of non-cancer gene fragments). 

A good reference for the technique, including a complete discussion of validation requirements and challenges, may be found in Guidelines for Validation of Next-Generation Sequencing-Based Oncology Panels. The authors of this article propose that board-certified pathologists provide the final say on selection of samples and sub-samples for testing to ensure a high proportion of tumor cells in the sample. Since we already have a shortage of board-certified pathologists, this pre-selection step seems counter-productive in a method meant to deliver rapid and confident data for clinical decision making. Perhaps it would be wiser to develop some tools to simplify or automate sample/sub-sample selection as part of validating the method for a particular end point. If the validated limit of detection (LOD) is less than 0.01%, the method may prove useful in assessing MRD.

Text Copyright © 2018 Katrina Rogers