The authors enrolled 106 consecutive adults presenting with cytopenias and hyperferritinemia who were undergoing bone marrow examination. The cohort was adjudicated against revised HLH-2004 criteria, yielding 50 patients classified as HLH and 56 as non-HLH after expert review. This was a single-center, observational cohort; the abstract reports those enrollment and classification details but does not list additional demographic breakdown or comorbidity distributions.
Physicians and the study team used an adapted strict Fardet-style HScore as the clinicobiochemical comparator. The adapted HScore served as the reference clinical tool against which peripheral blood immune-activation markers were measured, allowing direct comparison of a laboratory-based immunophenotype with a composite clinical scoring system.
Multiparametric flow cytometry quantified CD8+ T-cell subsets. The study measured single-positive CD8+CD38+ and CD8+HLA-DR+ populations and the dual-positive CD8+CD38+HLA-DR+ subset as a proportion of parent CD8+ T cells. The authors performed ROC analysis and derived Youden-index cut-offs to define optimal thresholds from their data.
The CD8+CD38+HLA-DR+ subset emerged as the best-performing flow-cytometric marker. The Youden-optimal threshold was reported as greater than 16.71% of parent CD8+ T cells, at which sensitivity reached 78% and specificity 81%. The team also reported an exploratory, higher threshold of 22.69% that produced reduced sensitivity (64.0%) but improved specificity (87.5%), suggesting a potential use-case for rule-in confirmation when a higher cut-off is applied.
Single-positive populations did not perform well diagnostically in this cohort. CD8+CD38+ alone had an AUC of 0.564 and CD8+HLA-DR+ alone had an AUC of 0.427, indicating poor discrimination between HLH and non-HLH patients by those single markers.
ROC analysis placed the adapted strict Fardet-style HScore substantially ahead of flow-cytometric markers in this dataset. The adapted HScore achieved an AUC of 0.970. At a threshold above 196, the adapted HScore demonstrated sensitivity of 94.0% and specificity of 87.5% for diagnosing adult HLH in this cohort.
In contrast, the dual-positive CD8+CD38+HLA-DR+ subset yielded an AUC of 0.824. While that represents reasonable discrimination for a single immunophenotypic marker, its sensitivity and specificity at the Youden-derived cut-off (78% and 81%, respectively) were inferior to the adapted HScore. The higher exploratory cut-off increased specificity to match the adapted HScore’s specificity (87.5%), but this was accompanied by a trade-off in sensitivity.
The study therefore positions the adapted HScore as the stronger clinicobiochemical framework for adult HLH diagnosis, while identifying CD8+CD38+HLA-DR+ as the most informative flow-cytometric adjunct available from the tested panels.
Data indicated that the adapted HScore should remain central to adult HLH evaluation because it performed best overall in sensitivity and specificity. The study team observed that CD8+CD38+HLA-DR+ offers added value in specific scenarios—primarily as a high-specificity rule-in adjunct when clinical scoring yields equivocal results. At a higher dual-positive threshold clinicians may gain confidence in diagnosing HLH in gray-zone cases where the HScore is non-diagnostic or borderline.
The authors recommended combining clinical scoring with immune-activation profiling: use the adapted HScore to provide a broad clinicobiochemical assessment, and employ CD8+CD38+HLA-DR+ quantification to strengthen diagnostic certainty in select patients. The abstract reports this integrated strategy as the practical conclusion derived from the cohort’s comparative performance metrics.
The report is limited by its single-center observational design; the abstract does not describe external validation or multicenter replication. The flow-cytometry approach is described as multiparametric, but the abstract does not provide detailed gating strategies, instrument platforms, or inter-laboratory standardization steps, so applicability across different laboratories remains unreported.
Additional unresolved questions remain in the abstract: whether the diagnostic thresholds will generalize to broader adult HLH populations with different prevalences, how the markers behave in competing hyperinflammatory syndromes such as sepsis or malignancy-associated inflammation in varied settings, and whether longitudinal sampling or treatment effects alter the performance of the CD8+CD38+HLA-DR+ subset. The investigators observed the marker’s potential as a rule-in adjunct but did not report external validation datasets within the abstract.
These limitations frame the next steps needed to move from a single-center observation to broader clinical adoption: prospective multicenter validation, standardized flow-cytometry protocols, and assessment of marker performance across differential diagnoses that commonly mimic adult HLH.