NK Cells Contribute to the Immune Risk Profile in Kidney Transplant Candidates

Abstract

Background: A previously proposed immune risk profile (IRP), based on T cell phenotype and CMV serotype, is associated with mortality in the elderly and increased infections post-kidney transplant. To evaluate if NK cells contribute to the IRP and if the IRP can be predicted by a clinical T cell functional assays, we conducted a cross sectional study in renal transplant candidates to determine the incidence of IRP and its association with specific NK cell characteristics and ImmuKnow® value. Material and Methods: Sixty five subjects were enrolled in 5 cohorts designated by age and dialysis status. We determined T and NK cell phenotypes by flow cytometry and analyzed multiple factors contributing to IRP. Results: We identified 14 IRP+ [CMV seropositivity and CD4/CD8 ratio < 1 or being in the highest quintile of CD8+ senescent (28CD-/CD57+) T cells] individuals equally divided amongst the cohorts. Multivariable linear regression revealed a distinct IRP+ group. Age and dialysis status did not predict immune senescence in kidney transplant candidates. NK cell features alone could discriminate IRP- and IRP+ patients, suggesting that NK cells significantly contribute to the overall immune status in kidney transplant candidates and that a combined T and NK cell phenotyping can provide a more detailed IRP definition. ImmuKnow® value was negatively correlated to age and significantly lower in IRP+ patients and predicts IRP when used alone or in combination with NK cell features. Conclusion: NK cells contribute to overall immune senescence in kidney transplant candidates.

Keywords: BK virus; NK cells; dialysis; end stage renal disease; immune risk profile; immune senescence; kidney transplant.

Figure 1

Complete separation of IRP+ and IRP– groups indicate contributions from both clinical and immune factors aside from those used to define IRP. PLS-DA plot of IRP+ and IRP– patients using the collected immune data except those that were used to determine IRP (CMV status, CD4 and CD8, and CD8+ senescent (28CD–/CD57+) cells. Each dot on the plot represents a subject, where blue represents IRP+ samples and orange represents IRP– samples. Confidence ellipses for each group were plotted to highlight the strength of the discrimination (confidence level set to 95%). Groups centroids were represented in the center of each group with circles.

Figure 2

Multivariate analysis using partial least-square discriminant analysis (PLS-DA) shows NK activation markers and ImmuKnow® variable segregate IRP– and IRP+ subjects. (A–C) Receiver operating characteristic (ROC) curves for NK markers alone (A), ImmuKnow® only (and NK markers + ImmuKnow® (C). Receiver operating characteristic (ROC) curve of PLS-DA model predicting IRP outcome. Receiver operating characteristic (ROC) curve showing the accuracy of the PLS-DA model in predicting IRP+ vs. IRP– outcomes using NK markers alone (A), ImmuKnow® alone (B), or NK markers with ImmuKnow® combined (C). The true positive rate (sensitivity) is plotted in function of the false positive rate (100-specificity). The area under the ROC curve (AUC) measures how well the model distinguishes IRP+ from IRP– subjects for each analysis. (D) Individual PLS-DA plot of IRP+ and IRP– patients using NK markers and ImmuKnow® as predictors and IRP status as outcome. Each dot on the plot represents a subject, where blue represents IRP+ samples and orange represents IRP– samples. Confidence ellipses for each group are plotted to highlight the strength of the discrimination (confidence level set to 95%). Groups centroids were represented in green circles. (E,F) Loading weights of each feature selected on the first the second component of the multivariate model. Loading plots represent the top features selected on the first (E) and the second (F) component of the PLS-DA model using tune sPLS-DA with color indicating the class with a maximal mean expression value for each feature. (G) Features correlation circle plots. Correlation circle plots display the correlation between measured features and latent components. Each variable coordinate is defined as the Pearson correlation between the original data and latent components 1 and 2. The contribution of each feature to each component is represented by features proximity to the large circle of radius 1. This plot shows also the correlation between features (clusters of features). The cosine angle between any two points represent the correlation (negative, positive, or null) between two features. A threshold of 0.5 is set to remove weaker correlations and to plot only features with major importance. Strongly correlated variables are projected in the same direction from the origin. The distance from the origin is correlated to the strength of the association.