Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Jun 1;129(22):2993-2999.
doi: 10.1182/blood-2016-12-753830. Epub 2017 Mar 7.

Perforin and CD107a Testing Is Superior to NK Cell Function Testing for Screening Patients for Genetic HLH

Affiliations
Free PMC article

Perforin and CD107a Testing Is Superior to NK Cell Function Testing for Screening Patients for Genetic HLH

Tamar S Rubin et al. Blood. .
Free PMC article

Abstract

Primary hemophagocytic lymphohistiocytosis (HLH) can be caused by biallelic mutations in PRF1, encoding perforin, or UNC13D, STXBP2, STX11, RAB27A, LYST, and AP3B1, encoding proteins involved in cytotoxic lymphocyte degranulation. Natural killer (NK)-cell cytotoxicity assays can quickly screen for all of these genetic diseases, facilitating treatment, but combining NK-cell perforin expression and CD107a upregulation tests can as well. To determine the relative diagnostic accuracies for each approach, we retrospectively reviewed screening test performance in 1614 patients referred for HLH evaluation. For each test, we generated a receiver operating characteristic (ROC) curve, and calculated area under the curve (AUC) and diagnostic parameters at optimal threshold. We generated an AUC for combining perforin and CD107a tests by creating a logistic regression model and applying model-generated coefficients to patient values. Sensitivities of NK-cell function, perforin mean channel fluorescence (MCF), and CD107a MCF to detect biallelic mutations were 59.5%, 96.6%, and 93.8%, with specificities of 72.0%, 99.5%, and 73%. AUCs for NK-cell cytotoxicity, perforin MCF, CD107a MCF, and combined perforin and CD107a MCFs were 0.690, 0.971, 0.860, and 0.838. Perforin and CD107a tests are more sensitive and no less specific compared with NK cytotoxicity testing for screening for genetic HLH and should be considered for addition to current HLH criteria.

Figures

Figure 1.
Figure 1.
Patient samples included and excluded from analyses.
Figure 2.
Figure 2.
Diagnostic accuracy of NK cytotoxicity for detecting biallelic HLH gene mutations. (A) Sensitivity/specificity/PPV/NPV of low NK lytic units to distinguish patients with any biallelic HLH gene mutations (PRF1, UNC13D, STX11, STXBP2, RAB27A, LYST, or AP3B1) from patients with all other sequencing results. (B) Sensitivity/specificity/PPV/NPV of low NK lytic units to distinguish patients with any biallelic HLH mutation from those with normal sequencing results. (C) Distribution of NK lytic unit results among patients with no mutations, 1 or more VUCS, carriers (with or without additional VUCS), and biallelic mutations. (D) ROC curve showing diagnostic accuracy of NK lytic units to distinguish patients with biallelic mutations from patients with all other sequencing results; sensitivity and specificity are shown at the optimal diagnostic cutoff of LU ≤ 0.1.
Figure 3.
Figure 3.
Diagnostic accuracy of perforin expression for detecting biallelic PRF1 mutations. (A) Sensitivity/specificity/PPV/NPV of low perforin MCF to distinguish biallelic PRF1 mutations from all other sequencing results. (B) Sensitivity/specificity/PPV/NPV of low perforin MCF to distinguish biallelic PRF1 mutation from patients with normal sequencing results. (C) Distribution of perforin MCF results among patients with no mutations, 1 or more VUCS (including A91V mutations), patients with A91V VUCS, carriers (with or without additional VUCS), and patients with biallelic mutations. (D) ROC curve showing diagnostic accuracy of perforin MCF to distinguish patients with biallelic PRF1 mutations compared with patients with all other sequencing results; sensitivity and specificity are shown at the optimal diagnostic cutoff of perforin MCF ≤ 38.
Figure 4.
Figure 4.
Diagnostic accuracy of CD107a expression for detecting biallelic degranulation gene mutations. (A) Sensitivity/specificity/PPV/NPV of low CD107a MCF to distinguish any biallelic degranulation gene mutation (UNC13D, STX11, STXBP2, RAB27A, LYST, or AP3B1) from patients with all other genetic-sequencing results. (B) Sensitivity/specificity/PPV/NPV of low CD107a MCF to distinguish any biallelic degranulation gene mutation from patients with normal genetic results. (C) Distribution of CD107a MCF results among patients with no mutations, 1 or more VUCS, carriers (with or without additional VUCS), and patients with biallelic mutations. (D) ROC curve showing diagnostic accuracy of CD107a MCF to distinguish patients with biallelic degranulation gene mutations compared with patients with all other sequencing results; sensitivity and specificity are shown at the optimal diagnostic cutoff of CD107a MCF ≤ 143.
Figure 5.
Figure 5.
Logistic regression model predicting probability of biallelic HLH mutations when combining perforin and CD107a expression tests, with or without NK cytotoxicity. ROC display of logistic regression model, showing diagnostic performance of testing with perforin MCF and CD107a MCF together in the model (A) and perforin MCF, CD107a MCF, and NK lytic units together in the model (B) to distinguish patients with biallelic HLH-associated mutations compared with patients with all other genetic results. est., coefficient of the logistic regression model; lr.eta, optimal cut point for the fitted values obtained from the logistic regression model; s.e., standard error for the coefficient; Sens, sensitivity; Spec, specificity.

Comment in

Similar articles

See all similar articles

Cited by 17 articles

See all "Cited by" articles

Publication types

MeSH terms

Feedback