A QA Program for MRD Testing Demonstrates That Systematic Education Can Reduce Discordance Among Experienced Interpreters

Abstract

Background: Minimal residual disease (MRD) in B lymphoblastic leukemia (B-ALL) by flow cytometry is an established prognostic factor used to adjust treatment in most pediatric therapeutic protocols. MRD in B-ALL has been standardized by the Children's Oncology Group (COG) in North America, but not routine clinical labs. The Foundation for National Institutes of Health sought to harmonize MRD measurement among COG, oncology groups, academic, community and government, laboratories.

Methods: Listmode data from post-induction marrows were distributed from a reference lab to seven different clinical FCM labs with variable experience in B-ALL MRD. Labs were provided with the COG protocol. Files from 15 cases were distributed to the seven labs. Educational sessions were implemented, and 10 more listmode file cases analyzed.

Results: Among 105 initial challenges, the overall discordance rate was 26%. In the final round, performance improved considerably; out of 70 challenges, there were five false positives and one false negative (9% discordance), and no quantitative discordance. Four of six deviations occurred in a single lab. Three samples with hematogones were still misclassified as MRD.

Conclusions: Despite the provision of the COG standardized analysis protocol, even experienced laboratories require an educational component for B-ALL MRD analysis by FCM. Recognition of hematogones remains challenging for some labs when using the COG protocol. The results from this study suggest that dissemination of MRD testing to other North American laboratories as part of routine clinical management of B-ALL is possible but requires additional educational components to complement standardized methodology. © 2017 International Clinical Cytometry Society.

Keywords: minimal residual disease.

FIG. 1.

Initial Assessment with Four Laboratories. Results of first wet sample challenge. Two COG laboratories and two adult Cooperative Group laboratories participated in a wet sample challenge in which varying numbers of ALL blasts were spiked into remission bone marrow. For this initial assessment, the cut-off was set at 0.1% since this was an initial attempt at comparability of MRD among the laboratories. Samples 5 and 8 did not contain ALL blasts, while Sample 1 was just under the 0.1% level.

FIG. 2.

Results from five dry send outs comprised of listmode files with five cases with varying levels of residual disease or none. Black dots are the reference values and white circles are the reported results from the seven participating labs. Percent concordance was measured as a result, and was within a ½ log of the reference result. A Cases 1–5. B. Cases 6–10. C. Cases 11–15. D. Cases 16–20. E. Cases 21–25. Parts A, B, and C were from Phase 2, and D and E. were from Phase 3.

FIG. 3.

Dot plots from two cases from the study. A. Representative dot plots showing a pattern that is consistent with maturing B cell hematogones within the bone marrow and no residual disease present. B. Dot plots show residual disease with a distinct phenotype that is bright CD19, CD10 negative, CD20 negative, CD34 positive, and brighter CD58 than normal.

FIG. 4.

Dot plots of a case showing degradation/artefact within the analyzed sample. Although fresh specimens are ideal due to centralized testing shipped samples often experience some degeneration over time due to unavoidable pre-analytical issues. Recognizing this and using Boolean gating to remove this from the residual disease is necessary to accurately enumerate the disease present.