Intelligence

Analysis

Myeloma MRD Across Laboratories: Why 0.90 and 0.63 Tell Different Stories

An Italian multi-centre pilot found that re-reading shared flow files was more concordant than processing distributed marrow samples. What the ICCs, detection limits and separate NGS workstream really mean.

Technical Review of a peer-reviewed multicentre pilot in Cancer Medicine (17 March 2026), DOI 10.1002/cam4.71678. Original CFCM interpretation, not independent validation.

Why identical sensitivity targets do not guarantee comparable myeloma MRD

Two laboratories can state the same minimum analytical sensitivity and still return different cell percentages. For next-generation flow cytometry (NGF) measurable residual disease in multiple myeloma, comparability depends on acquisition depth, marrow quality, sample transport, assay harmonisation and analysis. The Italian MM-MRD network studied exactly these operational sources of disagreement, but its pilot should not be mistaken for prospective proof of clinical interchangeability.

Stage 1: reinterpreting six common FCS files

Seven laboratories first compared the evaluation of six shared multiple-myeloma files. Inter-operator agreement for quantitative detection of monoclonal plasma cells was high (intraclass correlation coefficient, ICC 0.90; 95% confidence interval 0.72 to 0.98). This is primarily a test of interpretation of already-acquired material, not the entire specimen-to-result process.

Stage 2: the same protocol meets real specimen variability

The next step distributed 12 freshly obtained bone-marrow samples for local processing within 48 hours; two were excluded owing to excessive cell death. Quantitative agreement across laboratories was lower (ICC 0.63; 95% CI 0.38 to 0.87). The authors discuss differences in sample cellularity, transportation, viability and processing, plus software-dependent handling of acquired events. A binary MRD-positive or MRD-negative decision can agree while the reported percentage and its analytical limits differ. The cohort is too small to treat any one observed ICC as a universal performance estimate.

LOD and LOQ are not decorative numbers

Across NGF measurements, median limit of detection was 8 × 10^-6 and median limit of quantification was 2 × 10^-5, with wider lab-to-lab ranges. LOD describes evidence of detecting an abnormal population; LOQ indicates the support for quantifying it. The parent population, sufficient acquired cells, events defining the clone, the fraction of viable cells and marrow hemodilution must be documented before a negative or low-level result is compared across sites.

Original CFCM four-stage conceptual diagram showing the separate NGF shared-file review, NGF distributed marrow processing, LOD/LOQ dependency and independent NGS mock-sample QC; educational, not clinical validation.
Original CFCM conceptual diagram. Educational synthesis; not clinical validation.

Do not turn the NGS QC programme into an NGF-versus-NGS trial

The separate molecular workgroup comprised four laboratories using targeted IGH and IGK sequencing and mock-sample quality-control rounds. It reported full concordance for clonotype identification in its first QC round and 81% and 91% overall concordance in two later MRD mock-sample rounds. These findings concern a distinct experimental workflow, not paired clinical NGF and NGS results establishing equivalence or superiority.

What to include in a transferable laboratory verification

A reproducibility study should predefine the same reference gates and viable event denominator, acquisition and merging rules, number of evaluable cases, borderline samples, control materials, specimen delay, reportable range, laboratory-to-laboratory analysis methods and discrepancy review. Report negative calls together with sample-specific sensitivity. This pilot provides a practical failure-mode checklist, not validated thresholds for changing myeloma treatment.

Daily Learning Pill

Agreement on whether MRD is present is not the same as agreement on how much MRD is present. Ask whether a comparison holds the FCS file constant or deliberately includes transport, staining and acquisition variability.

Explore CFCM Resources and CFCM Troubleshooting for cytometry verification concepts. Study the full primary article before adopting any method claim.