PDF
Citation
Print
INTRODUCTION
Natural killer (NK) cells are innate immune cells displaying potent cytolytic effector functions [1, 2]. They play a crucial role in defending against pathogens, participating in immune surveillance, and actively eliminating cancer cells. Numerous diseases, and in particular hemophagocytic lymphohistiocytosis (HLH) [3, 4], have been linked to impaired NK cell activity (NKA). As low to no NKA can serve as a diagnostic criterion of HLH, accurate measurement of NKA is essential. Reduced NKA has also been observed in various other cancer types [5, 6]. Consequently, there is a growing demand for reliable and standardized methods for assessing NKA levels that ensure high sensitivity and precision.
Several approaches to measuring NKA exist. The radioactive chromium (51Cr) release assay is considered the gold standard for measuring NKA. However, it has several limitations, including radioactivity, inter-laboratory variability, and low sensitivity for apoptotic cell death [7-11]. As an alternative, the flow cytometry-based NK cytotoxicity assay (FC-NKA) has been widely applied and continuously optimized for use in clinical laboratories. However, the FC-NKA has not been standardized across laboratories, and a consensus on the reference interval for healthy individuals has not been reached [7, 9, 12-14].
Several previous studies have suggested a possible reference interval for interpreting FC-NKA results, but the number of healthy controls was insufficient to establish a statistically reliable reference interval [15-17]. Furthermore, in addition to the inherent challenges of standardizing functional immunoassays [18], the factors influencing FC-NKA performance have not been fully elucidated.
The present study aims to propose a reference interval for the FC-NKA in healthy controls and provide an optimal cutoff value for identifying those with low or no NKA. Furthermore, potential factors contributing to assay performance are also investigated.
MATERIALS AND METHODS
1. Study population
In the single-center retrospective study, the internal quality control data was examined. The data of 143 healthy controls subjected to the FC-NKA test from January 2019 until March 2023 at Seoul St. Mary’s Hospital were included. The healthy control samples included leftover samples from transplant donor pretransplant cross-matching tests or freshly collected samples from volunteered healthcare workers. Available demographic information for each control includes age, sex, and complete blood count (CBC) parameters. To validate the reference interval and determine the optimal cutoff of the assay, we compared the FC-NKA results of 16 HLH patients with those of the healthy controls. The study was approved by the Institutional Review Board of Seoul St. Mary’s Hospital (KC23EISI0221). Because the current study retrospectively analyzed quality control data obtained from routine laboratory protocols, informed consent was waived by the board.
2. Flow cytometry-based NK activity (FC-NKA) assay
FC-NKA measurements were performed according to the standard procedure in our laboratory [13]. In brief, CML-derived K562 cell lines (ATCC No. CCL-243™) were cultured with Dulbecco’s modified Eagle’s medium (DMEM, Welgene, Daegu, Korea) supplemented with 100 U/mL penicillin (Gibco, BRL, NY, USA), 100 µg/mL streptomycin (Gibco), and 10% fetal bovine serum. Fresh peripheral blood mononuclear cells (PBMC) were isolated from 10 mL of heparinized blood using Uni-sep lymphocyte separation tubes (Novamed, Jerusalem, Israel). K562 cells were stained with carboxyfluorescein diacetate succinimidyl ester (CFSE) (ThermoFisher Scientific, Carlsbad, CA, USA) to label the target cells. Effector cells (i.e., PBMC) were incubated at 37°C with CFSE-labeled K562 cells at an effector-to-target ratio of 32:1 for 4 hours. For the negative control, K562 and effector cells were individually subjected to the test. After incubation at 37°C under 5% CO2 for 4 hours, 7-AAD (Beckman Coulter, Marseille, France) was added, followed by analysis using a Navios flow cytometer (Beckman Coulter, San Diego, CA, USA). At least 1,000 to 2,000 target cells were examined. NK cell activity is determined by calculating the ratio of 7-AAD-stained cytolyzed target cells to the total number of CFSE-positive cells.
3. Statistical analysis
Statistical analyses were performed using Prism version 9.4.1 for Windows (GraphPad, San Diego, CA, USA) and MedCalc Statistical Software version 20.114 (MedCalc Software Ltd., Ostend, Belgium). Baseline characteristics were compared using either the chi-square test or t-test. The correlation between each CBC para-meter and the assay result was evaluated using Pearson’s correlation coefficient. Cytolyzed target cell percentages were compared among subgroups based on age and sex using a two-way analysis of variance (ANOVA). Following the C28-A3 CLSI guideline, we used the 2.5th to 97.5th percentiles to determine the reference interval [15]; moreover, the 5th and 10th percentiles were used to determine additional lower limits as it is crucial to elucidate reduced NK cytotoxicity for medical decision-making. The differences between the healthy controls and patients diagnosed with HLH were determined using the Mann-Whitney test. Receiver operating characteristic (ROC) analysis was performed using data from 16 patients diagnosed with HLH and 143 healthy controls collected within a similar timeframe, and the optimal cutoff was suggested based on Youden’s index. The variability between lots and interpreters was assessed using the Kruskal-Wallis test. All values were tested for normality before selecting either the parametric or non-parametric statistical method. Two-sided P-values of <0.05 were considered statistically significant.
RESULTS
1. Influence of age and sex
The study analyzed data from 143 healthy individuals using a total of 183 collected blood samples. Among these, 15 individuals were repeatedly tested as healthy controls, ranging from two to 11 times, with eight individuals participating more than three times. Only the first assay result from multiple sampling events was considered for determining the reference interval. Table 1 shows the distribution of individuals across different age groups (<30, 30–39, 40–49, and ≥50 years old) and sexes, along with a summary of the CBC results from each group. The analysis revealed that the only significant difference based on sex was related to hemoglobin levels. Nevertheless, no correlation was found between hemoglobin levels and the FC-NKA assay result (data not shown). Notably, no significant differences are identified in the FC-NKA results among the age- and sex-based subgroups, as depicted in Fig. 1.
2. Determination of reference interval
Given that no significant differences were observed within the age and sex subgroups, we provided a single reference interval for FC-NKA without stratification by age or sex. Our data showed that the 2.5–97.5 percentile reference interval for FC-NKA ranged from 13.8% to 82.9%. Additional lower limits of the 5th and 10th percentiles resulted in lower reference values of 17.9% and 26.8%, respectively. We used data from 16 patients diagnosed with HLH for comparison. The 16 patients had a median age of 23 years (range from 1 month to 65 years old) and consisted of nine males and seven females (Table 1). Their median age was significantly lower than that of the healthy controls (P=0.0012), and they revealed lower hemoglobin levels (P<0.0001) and platelet counts (P=0.0026). The assay results showed a significant decrease in K562 cytolysis (P<0.0001) in patients diagnosed with HLH compared to those in healthy controls (Fig. 2A). Additionally, the ROC curve indicated an impressive area under the curve (AUC) of 96.8% (95% confidence interval: 92.8% to 99.0%) based on Youden’s index (sum of 93.8% sensitivity and 93.0% specificity–100%). The optimal threshold value for defining decreased NK cell function while maintaining satisfactory diagnostic accuracy was determined to be 20.0% (Fig. 2B).
The healthy individuals with NKA levels below the recommended threshold were significantly younger (median age, 31.5 years old) than those with NKA levels above 20% (median age, 42.0 years old). No differences were observable in terms of sex proportion or CBC parameters.
3. Potential factors contributing to assay performance
We estimated the baseline intra-individual biological variation of the assay by analyzing samples obtained from a single individual undergoing three or more blood draws (N=8). The mean intra-individual coefficient of variation (CV) was 13.4%, with values ranging from 4.6% to 26.9%. The blood samples of the individual with an intra-individual CV of 26.9% were drawn at an interval of approximately six months. No discernible differences were observed for lot-to-lot or inter-interpreter variability (Fig. 3).
DISCUSSION
Because NK cells play a crucial role in the disease mechanism of HLH and other cancers, accurate and reproducible measurement of their activity is imperative [19-22]. Despite its widespread use, consensus regarding the reference intervals for the FC-NKA is lacking. Here, we propose a feasible FC-NKA assay that can be implemented in clinical laboratories, along with a reference interval based on CLSI C28-A3 and an optimal cutoff value for the diagnosis of HLH.
As no significant differences among the age and sex subgroups were observed, we concluded that stratification based on age and sex was not necessary to establish the reference interval for the FC-NKA. Therefore, for the 143 healthy individuals, we proposed a single reference interval of 13.8–82.9% for the 2.5–97.5th percentiles, respectively. Additionally, to establish a lower limit for HLH diagnosis, the 5th and 10th percentiles were used, providing lower limits of 17.9% and 26.8%, respectively. There is no consensus on which percentile to use as the optimal diagnostic threshold, and in practice, the three cutoffs listed above have all been implemented previously [12, 14, 16]. We further validated the cutoff value for the detection of HLH by comparing the FC-NKA results of healthy individuals with those of 16 HLH patients. Finally, we identified a threshold of 20.0%, which exhibited a sensitivity and specificity of 93.8% and 93.0%, respectively. These results suggest that the lower limits of the 5th and 10th percentiles can serve as diagnostic thresholds depending on the purpose of the assay, that is, whether it prioritizes sensitivity or specificity.
Reported differences among FC-NKA results based on age and/or sex are conflicting across studies. Some studies have reported that NKA is higher in males than females [12, 17], while another study reported no difference [16]. Our data did not show significant differences in the FC-NKA results based on variables such as age, sex, and CBC parameters. This may be due to the sample size and heterogeneity of the participants. Further studies with a larger sample size that include both HLH patients and healthy individuals with well-defined health data are needed. Particular attention should be given to the younger population, as our data showed that healthy individuals with NKA <20% were significantly youn-ger. This finding may suggest the need for a separate reference interval for younger individuals. Considering familial HLH, where children are at high risk [23], our data also revealed significantly younger patient epidemiology, and the identification of decreased NKA in this age group is of utmost importance.
The NKA assay holds particular significance in individuals with immunomodulated conditions that are susceptible to lymphopenia, thus necessitating reliable isolation of the effector cells to ensure accurate results. However, due to limited capacities for obtaining PBMCs, it is necessary to optimize the effector-to-target ratio. The 51Cr assay, esteemed as the gold standard for estimating NKA, has been previously suggested to have reference intervals of 31.9–51.2% [24] and 29.4–63.2% [25]. The introduction of FC-based assays has led to the utilization of various effector-to-target ratios, but a consensus on the ideal ratio has not been reached [9, 16, 22]. Reaching a consensus on the effector-to-target ratio closest to the results obtained from the 51Cr assay will yield valuable and readily interpretable data.
Although FC-NKA is an effective method for assessing the functional aspect of NK cell activity, it is unsuitable for use as a quantitative assay, as is the case with many other functional immunoassays. Additionally, the intra-individual CV of FC-NKA ranged from 4.6% to 26.9%, which may be higher than that of conventional immunoassays. However, this range is generally consistent with previous findings regarding functional immunoassays [26]. Considering the high biological variation of NKA [27], it is suggested that the analytical imprecision can be well controlled, allowing for a high degree of harmonization. Laboratories performing FC-NKA should implement internal and external quality controls or interlaboratory comparisons. Furthermore, a consensus would lead to the standardization of laboratory diagnostic strategies. The limitations of the study are as follows. The data in our study lack results from the pediatric population as healthy controls. Separate references for subgroups according to demographics could not be addressed in this study. Additionally, the number of included HLH patients was limited, and their median age differed from that of the healthy controls. Additionally, we only analyzed results tested using an effector-to-target ratio of 32:1; future studies are needed to confirm whether this ratio is optimal for the evaluation of NK cytotoxicity.
However, a notable advantage of our study lies in its utilization of a well-controlled routine laboratory protocol, which ensures consistency and reliability in establishing a reference interval for FC-NKA assays in a larger healthy control sample size than analyzed in previous studies. In conclusion, this study proposed a single reference interval for the FC-NKA and validated an optimal cutoff value for HLH diagnosis.
XML Download