Efficacy of T-cell assays for the diagnosis of primary defects in cytotoxic lymphocyte exocytosis

Hemophagocytic lymphohistiocytosis (HLH) is a hyperinflammatory disorder characterized by unremitting fever, splenomegaly, cytopenia, hypertriglyceridemia, hypofibrinogenemia, hemophagocytosis, as well as high serum ferritin and soluble CD25.^1-4 Early-onset, familial forms of HLH (FHL) are associated with autosomal recessive variants in PRF1, UNC13D, STXBP2, and STX11 encoding perforin, Munc13-4, Munc18-2, and syntaxin-11, respectively. Perforin expression is mainly restricted to nature killer (NK) cells as well as subsets of CD8⁺ T cells, where it is contained in cytotoxic granules.^5,6 In contrast, Munc13-4, Munc18-2, and syntaxin-11 are cytosolic proteins that mediate exocytosis by hematopoietic cells. The similarities in clinical presentation of patients with perforin-, Munc13-4–, Munc18-2–, and syntaxin-11–deficiency underscore their pivotal role in facilitating target cell killing by cytotoxic T lymphocytes (CTLs) and NK cells.^7-10 Individuals with autosomal recessive variants in RAB27A, LYST, and AP3B1 also display defective cytotoxic granule exocytosis and frequently develop HLH, with a majority manifesting characteristic hypopigmentation.^11-13 Impaired cytotoxic granule exocytosis has also been reported in patients with autosomal recessive AP3D1, MADD, ORAI1, RHOG, and STIM1.^14-18 Individuals with variants in SH2D1A, XIAP, CD27, or ITK may also manifest with HLH but lack a gross defect in lymphocyte cytotoxicity.^19-23 Individuals diagnosed with other inborn errors of immunity (IEI) may also develop HLH without displaying defective lymphocyte cytotoxicity.^24,25 HLH may also present in the context of infections or malignancy in individuals seemingly lacking a genetic predisposition.²⁶ 

Primary defects in lymphocyte cytotoxicity can be rapidly fatal unless promptly treated.²⁷ Although first-line treatment involves immunosuppression, the only currently available cure is allogeneic hematopoietic stem cell transplantation. Gene therapeutic approaches are being investigated.^1,28,29 Advances in DNA sequencing technology can facilitate rapid molecular diagnoses but may miss pathogenic noncoding genetic aberrations.^30-33 Moreover, variants of unknown significance require functional validation. Thus, functional analysis of lymphocyte cytotoxicity can provide timely and necessary answers that support vital treatment decisions.^4,8 

According to the HLH-2004 guidelines, defective NK-cell activity, as measured by the ⁵¹Cr-labeled K562 target cell lysis assay, represents 1 criterion for the diagnosis of HLH.¹ Using radioactive isotope, which is prohibited in many laboratories, this assay requires relatively large numbers of peripheral blood mononuclear cells (PBMCs). Results are also dependent on the frequency of NK cells among PBMCs, which may be influenced by inflammation. Hence, the ⁵¹Cr-labeled K562 target cell lysis assay does not unequivocally reflect the killing capacity of NK cells.^34,35 Laboratories have therefore adopted assays quantifying intracellular perforin expression as well as induced surface expression of CD107a (also known as lysosomal-associated membrane protein 1) in response to K562 cells for the diagnosis of patients with biallelic loss-of-function (LoF) variants in PRF1 or UNC13D, STX11, STXBP2, RAB27A, and LYST.^21,36-38 

Prf1 knockout mice have demonstrated that defective CTL activity is key to the pathogenesis of HLH, whereas defective NK-cell cytotoxicity may hamper control of adaptive immune responses and exacerbate disease.^39,40 Thus, measurements of CTL lytic activity may be of particular value for the diagnosis of primary patients with HLH. Additionally, stimulation can lead to functional recovery of NK- and T-cell exocytosis in patients with LoF variants in RAB27A, STX11, STXBP2, and LYST.^{8-10,21,41-43} Based on phenotypic characterization of peripheral blood CTL as CD8⁺CD57⁺ T cells, we previously published a protocol for sensitive evaluation of CTL exocytosis.⁶ This assay aided the diagnosis of patients with UNC13D variants, in which NK-cell assays yielded contradictory results.⁴⁴ Importantly, the efficacy of different NK-cell and CTL exocytosis assays for the overall diagnosis of patients with defects in lymphocyte cytotoxicity has not been systematically evaluated.

Here, we detail interindividual differences in cytotoxic lymphocyte exocytosis in a large cohort of healthy individuals and prospectively evaluate the efficacy of CTL and NK-cell exocytosis assays for patients with suspected primary HLH. Quantification of CTL exocytosis triggered by T-cell receptor (TCR) engagement or NK-cell exocytosis triggered by Fc receptor CD16 engagement provided the highest levels of accuracy for the diagnosis of primary defects in cytotoxic lymphocyte exocytosis, with K562 cell–triggered exocytosis assays being less sensitive. Combined evaluation of distinct cytotoxic lymphocyte subsets provided the most reliable and accurate diagnosis of patients with defects in cytotoxic lymphocyte exocytosis.

This study was approved by the Stockholm Regional Ethical Review Board. Patient samples were collected between November 2011 and May 2022. Healthy individual samples were obtained from the Karolinska University Hospital Blood Bank. For methods on sequencing, please refer to the supplemental Data, available on the Blood website.

Heparinized blood was received at room temperature, and PBMCs were extracted via density gradient centrifugation (Lymphoprep; Axis-Shield). From adults, 10 mL of blood was requested, whereas between 1 and 10 mL was obtained from infants and children. Live nucleated cells were counted (Muse; Cytek) and plated overnight unstimulated in complete medium (RPMI 1640 medium supplemented with 10% fetal bovine serum and 2 mM L-glutamate; all Hyclone). K562 cells (CCL243; American Type Culture Collection [ATCC]) and P815 cells (TIB64; ATCC) were maintained in complete medium. In total, 8 × 10⁵ PBMCs were generally used for individual analyses. A detailed protocol is specified in the supplemental Material.

Assessment of exocytosis was performed as previously described.⁶ Briefly, PBMCs were stimulated for 3 hours at with either K562 cells (CCL-243; ATCC), P815 cells (TIB-64; ATCC) coated with anti-CD16 antibody (3G8; BD Biosciences), or P815 cells coated with anti-CD3 (OKT3, BD Biosciences or S4.1; Thermo Fisher) antibody at 1:1 effector-to-target cell ratio. Cells were then surfaced stained with antibodies to CD3 (OKT3), CD8 (SK1), CD56 (HCD56), CD57 (HNK1), as well as an amine reactive viability dye (Thermo Fisher) and CD107a (H4A3). This test was also performed in combination with intracellular staining of signaling proteins in which surface staining was followed by intracellular staining with anti-FcεRγ (polyclonal rabbit; Upstate), anti-EAT-2 (polyclonal rabbit; Proteintech), and anti-SYK (clone 4D10.2; eBioscience) antibodies, as previously detailed.⁴⁵ Cells were acquired on a 4 laser BD Fortessa instrument. Flow cytometry data were analyzed with Flowjo (v9.9, Treestar).²¹ Gates for NK cells (Live CD3^–CD56⁺) and CTL (Live CD3⁺CD8⁺CD57⁺) were made based on the mock stimulated cells. These were then juxtaposed onto the various stimulated tubes, in which the delta value of CD107a expression was calculated by deducting the levels of background.

Statistics were generated with Prism (v7, GraphPad) as well as pROC R (v3.0.2) package for the generation of receiver operating characteristic curves.⁴⁶ Youden index and discriminant power (DP) were calculated as described.^47,48 Assay accuracy was defined as follows: (true positive + true negative)/(true positive + false positive + false negative + true negative). Unpaired t tests are reported as ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; and ∗∗∗∗P < .0001. Boxes and whisker plots indicate quartiles (25th, 50th, and 75th) and 2.5th and 97.5th percentiles for whiskers, and points were plotted individually for outliers. Bar graphs show lines representing standard deviation. Specific statistics used are reported in the figure legends.

This study was approved by the Stockholm Regional Ethical Review Board.

In our initial description of a sensitive T-cell exocytosis assay, analysis of PBMCs from 14 healthy adult volunteers revealed superior responses by CD8⁺CD57^dim and CD8⁺CD57^bright T cells after TCR engagement compared with other T-cell as well as NK-cell subsets after antibody-mediated engagement of CD16 or K562 cell stimulation.⁶ To validate these findings from a large number of healthy adults, we assessed cytotoxic lymphocyte exocytosis in 198 healthy adult volunteers (Figure 1A). Substantiating previous findings, assessment of CD107a surface expression in response to engagement of the TCR on CD8⁺CD57^dim and CD8⁺CD57^bright T cells or CD16 on NK cells using anti-CD3 or anti-CD16 monoclonal antibodies, respectively, demonstrated the strongest responses with a coefficient of variation of 0.22, 0.14, and 0.20, respectively. By comparison, CD3^–CD56^dim NK-cell responses to K562 cells and CD8⁺CD57^– T-cell responses to TCR engagement were weaker and displayed greater donor variability (coefficient of variation of 0.33 and 0.33, respectively). In summary, the exocytic response of cytotoxic CD8⁺CD57⁺ T cells upon TCR engagement or CD3^–CD56^dim NK cells in response to CD16 engagement was the most robust.

Interassay variability may represent a confounding factor in the diagnosis of HLH. Therefore, we repeatedly tested the same 2 donors over 3 months, examining exocytosis by CD8⁺CD57⁺ T cells in response to TCR engagement and CD3^–CD56^dim NK cells in response to CD16 engagement or K562 cell stimulation. PBMCs cryopreserved in multiple vials were thawed and run at different time points. Results were comparable throughout the period (Figure 1B-C). Although numerical values for exocytosis varied more for TCR engagement or CD16 engagement (Figure 1B-C; supplemental Table 1), the greatest relative variability was observed for CD3^–CD56^dim NK-cell responses to K562 cells (Figure 1D-F). Nonetheless, the overall high interassay consistency provided confidence when comparing data accumulated from independent experiments collected over several years.

With available data on sex, age, cytomegalovirus (CMV) serostatus, cell numbers, and differentiation of the 198 healthy adult volunteers for whom cytotoxic lymphocyte function was examined, additional factors influencing the magnitude of response in distinct cell subsets could be interrogated. When stratified by gender or age, no statistical difference was found among the 3 exocytosis assays (supplemental Figure 1).

Forming latent infection, CMV imprints human immune cell constitution and function, foremost promoting effector CD8⁺ T-cell differentiation, elevating serum interleukin-6 (IL-6) and IL-10, and influencing responses to these cytokines.^49,50 Notably, CD8⁺CD57⁺ T-cell exocytosis upon TCR engagement was somewhat higher in CMV seropositive individuals, whereas NK-cell responses to CD16 engagement or K562 cells were not correlated to CMV serostatus (Figure 2A).

CMV infection is associated with clonal expansions of long-lived, adaptive NK cells that lack expression of certain cytosolic signaling proteins in a variegated manner.^45,51-54 Such adaptive NK cells typically constitute a small subset of CD56^dim NK cells in most CMV seropositive individuals but may in some individuals dominate the NK-cell repertoire. Because adaptive NK cells display altered signaling and target cell recognition, we hypothesized that variability in NK-cell responses to K562 cells may be associated with adaptive NK-cell expansions. Indeed, NK cells lacking FcεRγ or SYK expression displayed less exocytosis in response to K562 cell stimulation, whereas responses to engagement of CD16 were maintained (Figure 2B). Among the 198 healthy individuals examined, expansions of NK cells lacking FcεRγ, EAT-2, or SYK expression did not affect CD3^–CD56^dim NK-cell responses to CD16 engagement (Figure 2C). However, highly reduced FcεRγ or SYK expression correlated with low CD3^–CD56^dim NK-cell exocytosis upon K562 cell stimulation (Figure 2D). A few healthy individuals with large adaptive CD3^–CD56^dimFcεRγ^– NK-cell subsets displayed K562 cell–induced exocytosis below the 10% threshold that has been deemed abnormal (Figure 2D).²¹ Lack of FcεRγ, EAT-2, and SYK expression is stochastically interlinked.⁴⁵ Boolean gating of adaptive NK-cell subsets demonstrated that lack of FcεRγ, but not EAT-2 or SYK, specifically correlated with reduced NK-cell exocytosis in response to K562 cell stimulation (Figure 2E). Thus, NK-cell responses to K562 cells were diminished in individuals with a high proportion of adaptive NK cells lacking expression of FcεRγ.

Finally, we compared TCR-triggered exocytosis in CD8⁺CD57⁺ T cells with CD16-triggered exocytosis in CD3^–CD56^dim NK cells. Notably, a few donors displayed low responses to both these stimuli and additionally displayed low CD3^–CD56^dim NK-cell exocytosis in response to K562 cells (Figure 2F-G) but were distinct from low responders with respect to K562 cell recognition, explained by epigenetic regulation of signaling protein expression (Figure 2H). Thus, combined assays may discriminate healthy adult individuals with overall mild impairments in cytotoxic lymphocyte exocytosis.

To gain insights to the efficacy of different cytotoxic lymphocyte exocytosis assays for the diagnosis of FHL, we prospectively evaluated cohorts of patients fulfilling HLH criteria, carrying HLH-associated LoF variants, or manifesting related hyperinflammatory disorders. Patients who exhibited reduced exocytosis according to established guidelines, or with at least 5 of 8 HLH criteria fulfilled, or with a family history of HLH were sequenced for HLH-associated genes.²¹ Furthermore, in patients with hypopigmentation, LYST, AP3B1, and RAB27A were sequenced.

Notably, a total of 92 patients analyzed harbored biallelic UNC13D, STX11, STXBP2, LYST, RAB27A, AP3B1, or RHOG LoF variants (termed EXO). Moreover, a total of 58 patients with biallelic or hemizygous PRF1, SH2D1A, XIAP, CD27, GATA2, MAGT1, ZNFX1, CYBA, or ITK LoF variants were analyzed (termed IEI). Thirty-three patients fulfilling HLH criteria but lacking pathogenic variants in genes associated with familial HLH or IEI and 30 patients diagnosed with macrophage-activation syndrome or systemic-onset juvenile idiopathic arthritis were also analyzed (termed HYPINF). These results were compared with those from 84 unrelated adult transport control samples or 198 healthy adult controls derived from the local blood bank. Three exocytosis stimulations were performed, namely, K562 cell–triggered NK-cell exocytosis, Fc receptor–triggered anti-CD16 antibody CD3^–CD56^dim NK-cell exocytosis, and TCR-triggered anti-CD3 antibody CTL exocytosis. Overall, we compared 5 sample groups and 3 stimulations (Table 1; Figure 3).

Patients in the EXO group displayed reduced exocytosis upon stimulation of CTL and NK cells compared with IEI or HYPINF patients, healthy adults, or transport controls (Figure 3A-B). Patients diagnosed with FHL3 (UNC13D), FHL4 (STX11), FHL5 (STXBP2), CHS (LYST), and GS2 (RAB27A) displayed defective exocytosis in all assays (Figure 3D), whereas patients diagnosed with FHL2 (PRF1), XLP1 (SH2D1A), XLP2 (XIAP), or other IEI generally displayed exocytosis almost at the level of healthy adults controls (Figure 3E). Notably, trend lines for exocytosis in IEI or HYPINF patients plotted vs age at analysis showed similar levels of exocytosis in pediatric patients to that of healthy adult control samples (supplemental Figure 2A).

Our exocytosis data were used to plot receiver operating characteristic curves (Figure 4A-D; supplemental Figure 2B-E). The area under the curve was larger for anti-CD3 CTL and anti-CD16 NK-cell exocytosis assays. The K562 cell exocytosis assay returned 93.1% sensitivity and 87.0% specificity at a 5.3% exocytosis cutoff. By comparison, the anti-CD16 stimulation of NK cells returned 96.6% sensitivity and 92.4% specificity at a 12.5% exocytosis, and the anti-CD3 stimulation of CTL was 94.8% sensitivity and 92.4% specificity at 23.5% exocytosis. The accuracy of the anti-CD3 and anti-CD16 assays was 93.3% and 98.7%, respectively, compared with 89.3% for the K562 cell assay (Table 2). The DP and Youden index showed a similar trend with K562 cell–induced NK-cell exocytosis having poorer results than the anti-CD3 CTL or anti-CD16 NK-cell exocytosis assay (Table 2).

Exocytosis values for the EXO, IEI, and HYPINF patient groups were plotted for the 3 different pair-wise comparisons (Figure 4E-G). Overall, the correlation between the exocytosis values obtained in the different assays was high, with a few exceptions of patients with low NK-cell exocytosis induced by K562 cells but with higher exocytosis induced by anti-CD16 NK-cell or anti-CD3 CTL assays.

In summary, the anti-CD3 CTL and anti-CD16 NK assays were more accurate than the current standard K562 cell–triggered NK-cell assay in predicting primary genetic defects affecting cytotoxic lymphocyte exocytosis.

Combining analysis of distinct cell types could potentially increase the accuracy of detecting primary defects in lymphocyte cytotoxicity. Therefore, we performed combined analyses of the sensitivity and specificity values of 2 assays, that is K562 cells with anti-CD16 NK-cell assays, anti-CD16 NK cell with anti-CD3 CTL assays, or K562 cell NK cell with anti-CD3 CTL assays. To have both true negative and true positive samples, the EXO and IEI groups were pooled. Taking cutoff values of 5.3% for K562 cell–triggered NK-cell exocytosis, 12.5% for anti-CD16 NK-cell assays, and 23.5% for anti-CD3 CTL assays, the results were evaluated with “AND” or “OR” Boolean functions. Using “AND” did not improve the accuracy, but anti-CD16 NK-cell OR anti-CD3 CTL assays returned 100% sensitivity and 98.3% specificity for a combined 99.3% accuracy (Table 3). Thus, combined analyses can enhance the diagnostic accuracy of functional diagnostic approaches.

Exocytosis testing is typically analyzed in larger experienced centers, necessitating sample transport. Some clinical laboratories have strict cutoff time points of <24 hours transport, but this is not always feasible from remote locations. Retrospectively, exocytosis assays from IEI patients were found impartial to shipment duration up to 48 hours (supplemental Figure 3A). Samples from healthy unrelated controls arriving <48 hours generally did not show any bias toward lower exocytosis results (supplemental Figure 3B). Familial controls (supplemental Figure 3D) show slightly poorer exocytosis at the >48-hour time point. When grouped by time of arrival (supplemental Figure 3C,E), a difference in function was observed when comparing samples arriving <24 hours with those arriving >48 hours. Notably, anti-CD16–triggered NK-cell exocytosis appeared the most time stable test. We further investigated the stability on blood samples from 10 healthy adult volunteers left for 1, 24, 48, and 72 hours before PBMC isolation. A significant reduction of exocytosis potential with all 3 stimulations was observed within 24 hours after venipuncture (supplemental Figure 3F). For K562 cell–triggered NK-cell exocytosis, the responses continuously dropped up to 48 hours, whereas anti-CD16–triggered assays remained stable during that time followed by a drop in function at >48 hours. The anti-CD3 CTL assay was the most robust with no statistically significant reduction in function even in samples stored >48 hours. The contrast in results between shipped and laboratory controls highlighted the difficulty in controlling for transportation stress because it is unique for each shipment and may include not just time but also variables such as temperature and agitation. In summary, because the anti-CD3 CTL exocytosis assay has inherently high signal-to-noise ratio and assay sturdiness, although not ideal, evaluation of cytotoxic lymphocyte function is possible with samples arriving after >48 hours.

We previously optimized an assay to evaluate exocytosis by freshly isolated human CTL, without prior stimulation or prolonged incubation.⁶ Here, we assessed the performance of different NK-cell and T-cell exocytosis assays in a large cohort of healthy individuals and prospectively evaluated patients with suspected primary HLH collected over 10 years. Our results demonstrate that assessment of CD8⁺CD57⁺ T-cell exocytosis in response to TCR engagement and CD3^–CD56^dim NK-cell exocytosis in response to CD16 engagement provided the highest diagnostic sensitivity. Importantly, combined analysis provided the highest diagnostic accuracy and safeguarded results in challenging patients when ambiguous results, long or suboptimal shipment, or low blood CTL or NK-cell frequencies might lead to assay failure.

Evaluation of ⁵¹Cr-labeled K562 target cell lysis has historically represented a gold standard for the evaluation of cytotoxic lymphocyte activity and diagnosis of primary HLH.^55,56 However, this test carries a number of drawbacks that may yield false positive results.¹ Hence, quantification of surface CD107a upregulation on NK cells after K562 cell stimulation was proposed as a diagnostic assay for patients with defective cytotoxic lymphocyte exocytosis.^21,57 Initially identified as an activation-induced surface protein on PBMC subsets, CD107a surface expression could play a role in self-protection against released cytolytic proteins and has been used to quantify cytotoxic T-cell, NK-cell, basophil, and mast cell exocytosis.^58-64 The K562 cell–triggered NK-cell exocytosis represents a safe, rapid, easily performed, and robust assay with good sensitivity and specificity,^21,37,65 but our efforts demonstrate that even higher diagnostic accuracy can be achieved with additional stimulations.

Primary HLH is commonly viewed as a T-cell–driven pathological response to infection.³⁹ The main aim of our study was to assess the efficacy of T-cell exocytosis assay for the diagnosis of congenital defects in lymphocyte exocytosis. The assessment of T-cell blast exocytosis was previously tested but may display a low signal-to-noise ratio especially with milder defects.^21,66 Confirming a pilot study, our data from 198 healthy adult donors revealed uniformly strong exocytic responses of CTL with relatively low interindividual variability.⁶ In comparison with assays evaluating NK- and T-cell exocytosis in response to engagement of CD16 or the TCR complex, respectively, the K562–NK-cell test displayed lower signal and higher interindividual variability. Notably, with our modification of the exocytosis assay examining 3 readouts of cytotoxic lymphocyte exocytosis, the cellular input is still lower than that typically used for ⁵¹Cr-release assays. Hence, flow cytometric exocytosis assays are more sensitive and require lower numbers of PBMCs than ⁵¹Cr-release assays, potentially facilitating measurements in prenatal infants or severely lymphopenic patients. CTL assays can facilitate assessment of patients with few NK cells, which we find represents an issue in ∼3% of pediatric and 6% of adolescent and adult patient samples.

Low NK-cell reactivity toward K562 cells among healthy individuals correlated with outliers harboring high proportions of CMV-associated adaptive NK cells lacking FcεRγ expression. Our results implicate FcεRγ-associated activating NKp30 and NKp46 receptors in recognition of K562 cells. Congruently, the NKp30-ligand B7-H6 is highly expressed on K562 cells and antibody-mediated NKp30 blockade inhibits NK-cell recognition of K562 cells.⁶⁷ Our results demonstrate that a high proportion of FcεRγ^– adaptive NK cells can negatively skew the overall NK-cell response to K562 cell stimulation. Although adaptive NK-cell expansions are associated with CMV seropositivity, in a European population, only 15% of healthy blood volunteers manifested sizable adaptive NK-cell expansions constituting >30% of the total CD56^dim NK-cell population.⁴⁵ CMV seropositivity is thus a poor indicator of adaptive NK-cell expansions. However, in a sub-Saharan population, adaptive NK-cell expansions were much more prevalent and constituted >50% of the total CD56^dim NK-cell population in a majority of participants.⁶⁸ Inflammation can promote adaptive NK-cell expansions.^69,70 Notably, patients with GATA2 variants occasionally present with HLH and can have extremely high frequencies of adaptive NK cells.^53,71,72 As such, diminished NK-cell exocytosis in response to K562 cell stimulation could reflect epigenetic cell differentiation processes rather than primary HLH and be especially prevalent in individuals with severe inflammation or underlying primary genetic defects. Quantification of NK or CTL exocytosis triggered by anti-CD16 or anti-CD3, respectively, can thus safeguard K562 cell–based analyses.

Prospectively, we evaluated 213 patients with suspected primary HLH. Patients with biallelic UNC13D, STX11, STXBP2, LYST, RAB27A, AP3B1, and RHOG variants consistently displayed strongly impaired CTL exocytosis compared with patients with PRF1, SH2D1A, XIAP, CD27, GATA2, MAGT1, ZNFX1, CYBA, or ITK variants, which have all been associated with HLH or low NK-cell numbers. Of note, healthy adult controls do not take into account several factors that might influence cellular analyses, including differences in age, time from sampling to analysis, medication/therapy, and blood inflammatory status. Thus, samples from patients diagnosed with other pathogenic HLH variants that do not impair exocytosis, secondary HLH, or other inflammatory syndromes represent better groups for comparison with patients with primary defects in cytotoxic lymphocyte exocytosis. Moreover, the interpretation of patient results is normally performed by juxtaposing a reference range obtained from a group of healthy adult donors. This study instead focuses on estimating NK and CTL exocytosis ranges of various genetically confirmed groups. Our patient data demonstrated superior sensitivity of the anti-CD3 CTL and anti-CD16 NK-cell exocytosis assays for the identification of patients with biallelic LoF variants in genes required for cytotoxic lymphocyte exocytosis compared with the established K562 cell stimulation assay, providing a high DP and Youden index, signifying the assays’ ability to discriminate between the 2 populations and avoid false positives and negatives, respectively.

Most cases of primary HLH are acute and require urgent diagnosis. Offering more confidence in findings, running the anti-CD3 CTL and anti-CD16 NK-cell exocytosis assays side by side increased accuracy and provided up to 100% sensitivity and 98% specificity. Infants aged <3 months normally have low numbers of CD57⁺ T cells, and it is not uncommon for patients to have too few NK cells for evaluation, suggesting a place for testing expanded T-cell exocytosis.^18,21,73,74 Independent assays also lessen the risk of human error leading to assay failure. Thus, running multiple assays increases the probability that at least 1 result is informative, avoiding delays and lowering the risk of misdiagnosis. Importantly, combining ≥2 assays also strengthens the overall diagnostic prediction. Generally, implementing changes in laboratory assays in a clinical setting can be costly in terms of efforts to fulfill national regulatory requirements, validate assays, and establish reference ranges. In terms of operating time and costs for implementing Fc receptor stimulation of NK cells and TCR stimulation of CTLs, use of multichannel pipettes and flow cytometers with plate readers should imply no drastic increase in operator time to run assays, whereas reagent costs are deemed to double due to an increase from 2 to 4 wells for each tested individual.

Although we report specific cutoff percentages for the 3 exocytosis assays, results will vary among laboratories. Moreover, our values were estimated by comparing controls and confirmed patient data, with the latter being quite rare and thus challenging for smaller laboratories to clinically validate. Determination of specific normal values are necessary for each laboratory and protocol. Results showing defective/absent exocytosis below the cutoff should be repeated to confirm, whereas those with reduced but not defective exocytosis should also be repeated and sequenced because data here suggest some patients with hypomorphic variants in genes required for exocytosis have lowered but not absent exocytosis. Here, the K562 cell exocytosis cutoff of 5.3% is in line with the previously stipulated 5%,²¹ even with protocol changes and extension from 2 to 3 hours of incubation. Importantly, the CTL exocytosis assay is also more robust than the K562 cell assay when dealing with suboptimal samples, due to long or harsh transportation.³¹ Large numbers of familial and healthy unrelated controls sent together with patient samples were minimally affected by transport times up to 48 hours, indicating that successful diagnosis of samples from farther locales or long shipping duration is feasible if well controlled.

We have demonstrated the accuracy of various exocytosis assays in diagnosing different primary genetic defects causative of defective exocytosis. We recommend laboratories run a combination of 2 exocytosis assays along with other panels to investigate the levels of perforin, SLAM-associating protein, X-linked inhibitor of apoptosis protein, CD27, NKT cells, and TCRα/β–double negative T cells to encompass a wider range of primary genetic defects linked to HLH, as illustrated in our suggested laboratory diagnosis algorithm (Figure 5).^22,23,75-83 We have previously reported how IL-2 stimulation can discriminate certain forms of FHL.^6,8,21,41-43 In this study, we focused analyses on freshly isolated samples in which functional assays can promptly identify patients with defective exocytosis and potentially direct treatment. With respect to molecularly categorizing patients, we advocate that genetic tests are a more reliable way to stratify patients with FHL and gain important genotype-functional correlations.

In conclusion, screening for anti-CD3 antibody–induced exocytosis of CTL and anti-CD16 antibody–induced exocytosis of NK cells is more accurate than the current standard K562 cell stimulation for the diagnosis of congenital defects in lymphocyte exocytosis. The data presented here thus suggest a change in the HLH-2004 criteria from low or absent NK-cell activity to low NK-cell or CTL exocytosis, except for FHL2/perforin deficiency. Performing 2 different exocytosis stimulations simultaneously gives high accuracy and acts as confirmatory assays. Such assays with increased sensitivity can aid the understanding of immunodeficiencies that may partially impair exocytosis.

The authors thank Felicia Er ZhiXin for technical assistance, MedH Flow Cytometry Facility, Uppsala Multidisciplinary Center for Advanced Computational Science, all patients and their families for contributing samples, as well as collaborative clinicians from around the world.

This work was supported by Deutsche Forschungsgemeinschaft SFB1160 (256073931; S.E.) TPA1; and by grants from the Swedish Research Council (2022-01178), Cancer Foundation (23 3090 Pj), Childhood Cancer Foundation, the Knut and Alice Wallenberg Foundation, and the Center for Innovative Medicine (all to Y.T.B.).

Contribution: S.C.C.C. designed the study, performed exocytosis assays, evaluated the data, and wrote the manuscript; B.T. and L.E.C. performed sequencing, evaluated the data, and performed receiver operating characteristic analyses; H.S. performed exocytosis assays with EAT-2 and and FcεRγ staining; J.T. partially performed statistical analysis; T.M.C., J.N.-Z., T.S., and S.W. performed exocytosis assays; M.M. performed sequencing and evaluated the data; K.M., W.A.-H., H.H.A., F.B.B., M.Y.C., O.D., T.A., M.I., I.M., M.S., E.U., S.U., W.J.I., K.K., K.C.G., and S.E. cared for patients and/or provided material; all pHLH collaborators cared for patients and provided clinical information; H.-G.L., M.N., A.H., and J.-I.H. revised the manuscript; Y.T.B. designed the study, evaluated the data, and wrote the manuscript; and all authors have critically read the manuscript before submission.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A complete list of all pHLH collaborators appears in “Appendix.”

Correspondence: Yenan T. Bryceson, Karolinska Institute, Department of Medicine, Center for Hematology and Regenerative Medicine, Hälsovägen 9, 14186 Stockholm, Sweden; email: yenan.bryceson@ki.se; and Samuel C. C. Chiang, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229; email: sam.chiang@cchmc.org.

The members of the pHLH collaboration are Miguel R Abboud, American University of Beirut, Beirut, Lebanon; Sevkiye Selin Aytac, Hacettepe Univerity Faculty of Medicine, Hacettepe, Turkey; Franziskus Johannes Bosse, Barne og Ungdomsklinikken, Haukeland Universitetssjukehus, Bergen, Norway; Sharon Choo, Royal Children's Hospital, Melbourne, Australia; Katarzyna Drabko, Klinika Onkologii, Hematologii i Transplantologii Pediatrycznej, Uniwersytet Medyczny w Poznaniu, Szpital Kliniczny im. Karola Jonschera, Instytut Pediatrii, Poznań, Poland; Reem Elfeky, Great Ormond Street Hospital, London, United Kingdom, and Children's hospital Ain Shams University, Cairo, Egypt; Dalia Helmy El-Ghoneimy, Children's Hospital, Ain Shams University, Cairo, Egypt; Zehra Fadoo, Aga Khan University, Karachi, Pakistan; Tatiana Greenwood, Karolinska Institutet, Karolinska University Hospital Solna, Stockholm, Sweden, and Astrid Lindgren Children’s Hospital, Karolinska University Hospital, Stockholm, Sweden; Britt Gustafsson, Karolinska Institutet, Stockholm, Sweden; Stefan Hagelberg, Astrid Lindgren's Children's Hospital, Karolinska University Hospital, Stockholm, Sweden; Henrik Hasle, Aarhus University Hospital, Aarhus, Denmark; Johanna Hästbacka, New Children’s Hospital, University of Helsinki and Helsinki University Hospital, Helsinki, Finland; Oleg Jadrešin, Children's Hospital Zagreb, Zagreb, Croatia; Martin Jädersten, Karolinska University Hospital, and, Center for Hematology and Regenerative Medicine, Karolinska Institutet, Stockholm, Sweden; Zuhre Kaya, Gazi University Faculty of Medicine, Ankara, Turkey; Ranon Lecumberri, Clínica Universidad de Navarra, Pamplona, Spain; Laura Marques, Centro Materno Infantil do Norte, CHUdSA, Porto, Portugal; Naureen Mushtaq, Aga Khan University Hospital, Karachi, Pakistan; Ahmed Naqvi, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; João Farela Neves, Hospital Dona Estefânia, CHULC-EPE, Comprehensive Health Research Centre and Chronic Diseases Research Center, NOVA Medical School, NOVA University of Lisbon, Lisbon, Portugal; Susana Nunes, Centro Hospitalar Universitário São João, Porto, Portugal; Martin Paucar, Karolinska University Hospital, Stockholm Sweden; Jeanette H. Payne, Sheffield Childrens Hospital, Sheffield, United Kingdom; Jelena Rascon, Vilnius University Hospital Santaros Klinikos, and Clinics of Children’s Diseases, Faculty of Medicine, Vilnius University, Vilnius, Lithuania; Terhi Susanna Ruuska, University of Oulu and Department of Pediatrics and Adolescent Medicine, Oulu University Hospital, Oulu, Finland; Ebru Tugrul Saribeyoglu, FÄ für Kinder- und Jugendmedizin, SP Pädiatrische Onkologie/Hämatologie, ZB Palliativemedizin, Charité-Universitätsmedizin Berlin, Berlin, Germany; Mikael C. Sundin, Karolinska University Hospital Huddinge and Clintec, Karolinska Institutet, Stockholm, Sweden; Jenny Svedenkrans, Karolinska University Hospital, Stockholm, Sweden; Natalia Świderska, Medical University of Gdańsk, Gdańsk, Poland; Ulf Tedgård, Skåne University Hospital, Lund, Sweden; Tor Henrik Tvedt, Oslo University Hospital, Oslo, Norway; Ayşegül Ünüvar, İstanbul School of Medicine, İstanbul University, Istanbul, Turkey; Jan A. M. Van Laar, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands; Sheila Weitzman, The Hospital for Sick Children, Toronto, Canada; Jacek Winiarski, Karolinska University Hospital Huddinge and Clintec, Karolinska Institutet, Stockholm, Sweden; Muhamma Zohaib Yaseen, Children's Hospital Karachi, Karachi, Pakistan; Mehmet Yildiz, Cerrahpasa Medical Faculty, Istanbul University, Istanbul, Turkey; Daniela Zantomio, Austin Health, Melbourne, Australia; Ingrid Øra, Lund University, Lund, Sweden; and Torstein Øverland, Oslo University Hospital, Oslo, Norway.