Hemophagocytic lymphohistiocytosis (HLH) is a rare disease and a life-threatening syndrome caused by pathological hyperactivation of the immune system. It can lead to hypercytokinemia and multi-organ failure [1
]. Adult HLH without genetic predisposition or family history is a rare clinical condition. Its underlying pathophysiological mechanisms have not been fully elucidated yet. HLH is difficult to diagnose because its clinical symptoms and signs are nonspecific, overlapping with those of other diseases such as sepsis, autoimmune disease, multi-organ failure, and progressed malignancies [2
]. Not all HLH patients exhibit hemophagocytosis. In addition, this can be observed in other diseases such as severe sepsis, post-transfusion or cytotoxic therapies, and hematologic malignancy [3
]. HLH-2004 guideline was developed for the diagnosis of pediatric HLH and commonly used to date for adult HLH diagnosis as well [4
]. This guideline includes six clinical and laboratory parameters (fever, splenomegaly, cytopenia, hyperferritinemia, hypertriglyceridemia/ hypofibrinogenemia, and hemophagocytosis in bone marrow) and three specialized tests (increased serum soluble interleukin-2 receptor (sIL-2r), decreased natural killer (NK) cell activity, and related gene test), of which 5 or more must be met for a diagnosis of HLH [4
The pathophysiological mechanism of HLH is mainly due to defective function of NK cells and cytotoxic T lymphocytes (CTLs). It results in uncontrolled activation of lymphocytes and macrophages which induce excessive production of cytokines [6
]. Therefore, NK cell function test is a hallmark and a valuable tool for HLH diagnosis. Human NK cells can be divided into various subsets by expression of CD16 and CD56 [7
]. CD16+CD56dim cells account for majority of NK cells and display cytotoxic activity while CD16-CD56bright cells secrete various cytokines, mainly interferon-gamma (IFNγ) [8
]. CD16-CD56bright cells also acquire potent cytotoxicity after activation. NK cells have direct cytotoxicity through granzyme and perforin release. NK cells can also control immune response by cytokine secretion. Therefore, measuring both NK-cytotoxicity and NK cell function for IFNγ release (NKA-IFNγ) might be valuable for the diagnosis of HLH. However, the diagnostic value of NK-cytotoxicity and NKA-IFNγ in HLH patients has not been properly evaluated yet.
To measure NK-cytotoxicity, 51
Cr-release assay has been known as the gold standard. However, it has many technical shortages such as spontaneous breakdown of 51
Cr from target cells that could affect analytical sensitivity and radioactive waste generation [9
]. Many publications have reported alternative flowcytometric methods. However, flowcytometric-based NK cytotoxic assays have not been standardized across laboratories. They are generally only available in specialized reference centers [10
NK Vue (ATGen, Seongnam-si, Korea) test is a newly developed in vitro diagnostic assay that can measure NKA-IFNγ using sandwich enzyme-linked immunosorbent assay (ELISA). The principle of this assay is stimulating whole blood with Promoca (engineered recombinant cytokines) that can specifically activate NK cells in whole blood. Released IFNγ level from activated NK cells is then measured [17
]. Previous studies reported that patients with malignancy show decreased NKA-IFNγ level compared to healthy controls [17
]. However, a study applying this assay in HLH diagnosis has not been reported yet. Thus, the objective of this study was to evaluate flowcytometry-based NK-cytotoxicity and NKA-IFNγ as diagnostic tools for adult HLH. Associations between NK cell function test and clinicopathological parameters including NK cell phenotype and cytokines were also investigated.
HLH is a life-threatening inflammatory condition. Early diagnosis and intervention are essential for patient survival [1
]. Among clinical and laboratory parameters in HLH-2004 criteria, three specialized tests (genetic test, serum sIL-2r test, and NK cell function test) are not easy to perform in clinical laboratory. Genetic tests are useful for detecting potential genetic predisposition to HLH in adult patients. However, genetic abnormalities are rarely detected. Pending results may delay clinical decision to treat HLH [22
]. Because the hallmark of HLH is deficiency of NK cell function, NK cell function test can be the most valuable parameter for HLH diagnosis [1
]. This is the first study to assess the diagnostic performance of NK cell function test by both NK-cytotoxicity and NKA-IFNγ in adult HLH patients. We confirmed that both flow cytometry-based NK-cytotoxicity and NKA-IFNγ levels were significantly decreased in HLH patients than those in non-HLH patients.
When we compared NK-cytotoxicity and NKA-IFNγ results, NK-cytotoxicity was not correlated with NKA-IFNγ level. Although the agreement between qualitative results from NK-cytotoxicity and NKA-IFNγ were high in HLH patients (88%), non-HLH patients showed lower agreement (58.0%) between the two NK cell function tests. It might be due to decreased NK cell function in malignancy and increased cytokine levels in infection. Several previous studies have also demonstrated decreased NK cell function in patients with malignancies (including colorectal cancer, prostate cancer, ovarian cancer and hematologic malignancies) compared to healthy controls [17
Current HLH diagnosis criteria specify ferritin level > 500 μg/L. However, a recent study reported that the specificity for predicting HLH was as low as 33% with this cut off ferritin level. They suggested that an optimal cutoff level of ferritin at 5775 μg/L for diagnosis of HLH had a sensitivity of 89.5% and a specificity of 63.2% [5
]. With cutoff value of 38.5% for NK-cytotoxicity and 250 pg/mL for NKA-IFNγ, the two tests showed sensitivities of 96.0% and 92.0% for diagnosing HLH, respectively. However, specificities of NK-cytotoxicity and NKA-IFNγ were low (36.2% and 17.4%, respectively). To improve the specificity of NK cell function test, we combined NK cell function results and ferritin levels. Combined NKA-IFNγ and ferritin levels (>10,000 µg/L) improved specificity to 94.2%. In addition, combined NK-cytotoxicity, NKA-IFNγ, and ferritin levels (>10,000 µg/L) showed specificity of 95.7%. In practice, there is a trade-off between sensitivity and specificity. Still, no single laboratory parameter is sufficient to diagnosis HLH and superiority of current diagnostic criteria is still investigated. Thus, further new findings could be incorporated in diagnostic and prognostic marker of adult HLH.
Conventional NK cell function assay requires time-consuming cell culturing process and radioactive reagents or specialized flow cytometric techniques [16
]. Although flow cytometry-based NK-cytotoxicity has been widely applied and constantly optimized, different laboratories have their own protocols and reference methods. Variations include different effector to target ratios, selection of fluorochromes to label target cells, NK cell purification process using magnetic beads, and the use of rIL-2 to enhance NK cell response [10
]. In addition, flow cytometry assays need fresh samples (drawn within 24 h of testing) to be valid. It is known that IFNγ released from NK cell and cytotoxic T cell can activate macrophage to induce secretion of inflammatory cytokines in HLH [30
]. As HLH is referred to as a cytokine storm, previous studies have demonstrated elevated levels of inflammatory cytokines including IFNγ and IFNγ-induced chemokines in HLH [31
]. Among several cytokines, only sIL-2r is included in HLH criteria. Unlike serum cytokine measurements, NKA-IFNγ assay is a functional assay that measures released IFNγ level from activated NK cells. In contrast to NK-cytotoxicity, NKA-IFNγ is a simple, verified test based on a simple commercially available kit. Supernatants containing released IFNγ can be frozen and stored until running the ELISA. Because adult secondary HLH is defined by pathologic immune activation and cytokine release leading to end organ damage, measurement of released IFNγ level from activated NK cells should be highlighted. Although NKA-IFNγ correlates with serum cytokine levels, evaluation of diagnostic performance of NKA-IFNγ is important for clinical implementation.
NK-cytotoxicity level, but not NKA-IFNγ level, correlated with NK cell percentage and NK cell count. Previous study reported NK-cytotoxicity results using isolated NK cells [33
]. However, NK-cytotoxicity using peripheral blood mononuclear cells (PBMC) can also provide natural circumstance, like in vivo environment as compared to the test using isolated NK cells. When encountering the target cells, the CD56dim cells secrete the perforin with the granzyme B and then induce cell cytotoxicity. The activated NK cells can also secret IFN-γ and the innate and adaptive immune system is reinforced by interaction of the activated NK cells and other immune cells [34
]. Therefore, using PBMC is more convenient and can provide more informative results than using NK cells in clinical laboratory setting [15
]. A previous study has revealed that NKA-IFNγ is not associated with NK cell counts in hematologic malignant patients [26
]. Based on this finding, we could conclude that decreased level of NKA-IFNγ reflects decreased ability of NK cell function itself, not numerical change of NK cell count. Gao et al. [30
] have found that secondary HLH patients have increased expression of inhibitory NKG2A receptor and decreased expression of activating NKG2D. Changes of NK cell receptor and its interaction with adaptors can impair the cytotoxic capacity of NK cells against pathogenic T cells. Therefore, hyperactive T cells continue to activate macrophage and lead to cytokine storm [30
]. In our data, patients with decreased NK-cytotoxicity showed tendency of increased expression of NKG2A but decreased expression of NKG2D than patients with normal NK-cytotoxicity, although these differences were not statistically significant. In addition, the expression of NKG2A or NKG2D was not associated with NKA-IFNγ results. NKA-IFNγ does not measure cytotoxicity capacity, but focuses on the regulatory capacity as IFNγ producers [35
]. Therefore, it is unlikely to be associated with NK cell receptors. In previous study, Mahapatra S. et al. [36
], measured both the percentage and median fluorescent intensity (MFI) of NK cell phenotypes in healthy adults and children. They reported that significantly different NK markers were not consistent between percentage and MFI and emphasized the importance of using both measures. While percentages measure the frequency of positively expressing NK cells at a population level, MFI reflect the molecules per cell for markers expressed by individual NK cells. Thus, further studies are needed to analyze the MFI of NK cell phenotypes in association with NK function in HLH patients. When we normalize NK-cytotoxicity data to CD56dim NK cells (%), HLH patients showed significantly decreased NK-cytotoxicity levels. However, normalized NKA-IFNγ results did not show significant differences between HLH and non-HLH patients. Further validation studies are needed for NKA-IFNγ tests in association with NK subpopulation.
This study has several limitations. First, this retrospective study was performed in a single institution. We did not simultaneously perform genetic test and sIL-2r tests for all our study population. Recently, serum sIL-2r has been reported as a good test with low cost for adult HLH [5
]. However, in the present study, sIL-2r assay only showed sensitivity of 69.6% in HLH patients who were available for the test. It might be due to different population and etiologic causes of HLH. In addition, we did not assess prognostic value of NK cell function test. Although our data originated from a retrospective study, we could assess NK cell functions in all patients with suspected HLH. HLH is a rare disease and hematologic malignancies are reported to be the main trigger of HLH [3
]. If clinicians do not have clinical suspicion of HLH and lack specific knowledge about HLH, the prevalence of HLH may be under- or overestimated, especially in adult patients [3
]. In the present study, all patients enrolled in this study had fever and cytopenia, therefore, the incidence of disease showed higher than expectation. We have evaluated the performance of newly developed NK function assay for diagnosis HLH. Diagnostic values of NKA-IFNγ should be validated by larger studies. In addition, further study is needed to compare NKA-IFNγ with NK-cytotoxicity using isolated NK cells with or without C107a expression. It might be worthwhile to conduct the intracellular cytokine staining to compare the predictability of NK function assay [37
In conclusion, our results suggest that both NK-cytotoxicity and NKA-IFNγ might be useful for HLH diagnosis. We expect that NKA-IFNγ assay is a convenient and supportive screening test for HLH. Compared to NK-cytotoxicity, NKA-IFNγ showed similar performance for predicting HLH. It also showed high level of agreement with NK-cytotoxicity. Further validation studies are needed for NK cell function tests in large population and the prognostic value of NKA-IFNγ.