Using biomarkers to predict the presence of FAS mutations in patients with features of the autoimmune lymphoproliferative syndrome

Abstract

To the Editor: The autoimmune lymphoproliferative syndrome (ALPS) is characterized by chronic lymphadenopathy, splenomegaly, autoimmune cytopenias, and expansion of T cell receptor (TCR) αβ+ CD3+CD4−CD8− (αβ-double-negative [DNT]) cells (see this article’s Table E1 in the Online Repository at www.jacionline.org). Approximately two thirds of the patients with ALPS symptoms are genetically characterized, and most have germline (ALPS Ia) or somatic (ALPS Ia-s) TNFRSF6 (FAS) mutations. A small number of patients have defects in genes encoding Fas ligand (ALPS Ib), caspase-10 (ALPS II), or neuroblatoma-RAS (NRAS) viral oncogene homolog (ALPS IV). In addition, a large group of patients with ALPS findings remain genetically uncharacterized (ALPS III), and yet another has an undefined ALPS-like syndrome (ALPS-phenotype; Table I).1,2 Given the clinical similarities among all these groups, we sought to develop a biomarkers-based algorithm to predict the presence or absence of FAS mutations in this setting. TABLE I Description of patients with ALPS and control groups included in the study To this end, we investigated 26 parameters including immunophenotyping, eosinophil and monocyte counts, serum or plasma vitamin B12 (B12), soluble FAS ligand (sFASL), immunoglobulins, and levels of 14 cytokines in 562 subjects classified into 6 categories (Tables I and E1). The number of measurements, medians, and first and third quartiles are presented in this article’s Tables E2 and E3 in the Online Repository at www.jacionline.org. A full description of the Methods can be found in the Online Repository at www.jacionline.org. Elevated αβ-DNT cells are a hallmark of ALPS, but their utility for predicting FAS mutations had not been previously evaluated. 3 Patients with ALPS Ia and Ia-s had a high percentage of αβ-DNT cells, with median values 5.1% and 7.7%, respectively, compared with 0.5% for control mutation-negative relatives (MNRs; P 4% found in 60% (90/152) of patients with type Ia and in the majority of patients with type Ia-s (7/9), but in only 13% (11/85) of patients with ALPS type III and ALPS-phenotype (Fig 1, A). This value was associated with a positive likelihood ratio (LR) of 5.0 and a posttest probability of 89.3% for harboring FAS mutations. Conversely, the presence of αβ-DNT cells in the 1% to 2% range decreased the posttest probability to 25%, with a LR of 0.19 (Fig 2, B and C; see this article’s Table E4 in the Online Repository at www.jacionline.org). FIG 1 Biomarkers in patients with ALPS and control groups. Dashed lines represent cut-off values used to calculate likelihood ratios. Bars denote median values. P values for the differences between groups were obtained by Mann-Whitney test and are shown above ... FIG 2 sFASL levels and combinations of biomarkers accurately predict FAS mutations. A, Scatter plot showing sFASL levels. Increasing (B) and decreasing (C) probabilities for having a FAS mutation according to the percentage of αβ-DNT cells, ... In line with previous reports, patients with ALPS, regardless of mutation status, had 16% made the diagnosis of ALPS very unlikely (LR = 0.17). Other described abnormalities including increased CD3+HLA-DR+ to CD3+CD25+ ratio and high number of B cells had no additional diagnostic utility.4 We also evaluated serum B12 levels in patients with ALPS and found very elevated median levels in ALPS Ia and Ia-s (2259 ng/L; 1653 ng/L) compared with control MNRs (474 ng/L; P 1500 ng/L was 4.0, with a posttest probability of 87%. In contrast, having B12 levels <1000 ng/L diminished the posttest probability to 35% (Fig 2, B and C; Table E4). Analysis of plasma cytokines revealed 2 additional biomarkers for ALPS: IL-18 and TNF-α. Median plasma IL-18 levels were elevated in patients with ALPS Ia and Ia-s compared with control MNRs (1041 pg/mL, 1526 pg/mL, and 208 pg/mL, respectively; P < .0001). Patients with ALPS III and ALPS-phenotype had median values of 521 pg/mL and 702 pg/mL, respectively (P < .001 compared with MNRs; Fig 1, D). Furthermore, IL-18 <500 pg/mL was rarely seen in patients with ALPS and FAS mutations (7/56), with an associated negative LR of 0.19. TNF-α levels were higher in all ALPS groups with median values of 5 pg/mL for ALPS Ia (P < .0001), 9 pg/mL for ALPS Ia-s (P < .05), 8 pg/mL for ALPS III (P < .0001), and 7 pg/mL for ALPS-phenotype (P < .0001) compared with 1.3 pg/mL for MNRs (Fig 1, E). As previously reported, IL-10 was markedly elevated in ALPS Ia and Ia-s compared with MNRs (P 40 pg/mL, contrasting with 26% (10/38) of patients with ALPS III and ALPS-phenotype. For levels of IL-10 >40 ng/mL, the positive LR was 3.8, with a posttest probability of 85% for having a FAS mutation. Notably, only 20% (29/141) of patients with ALPS Ia and no patients with ALPS Ia-s had IL-10 values <20 pg/mL, giving a negative LR of 0.31 and a posttest probability of 33% for FAS mutations (Fig 2, B and C; Table E4). A recent report documented high levels of sFASL in patients with ALPS.7We expanded these findings analyzing more than 200 patients and controls. Ninety-seven percent of patients with ALPS Ia (136/140) and all patients with ALPS Ia-s had plasma sFASL >200 pg/mL, with median values of 1114 pg/mL and 1329 pg/mL, respectively, compared with control MNR levels of 104 pg/mL (P <.0001 for both groups). Only modest elevations of sFASL were seen in patients with ALPS III and ALPS-phenotype, as well as healthy mutation-positive relatives, with median values of 208 pg/mL, 174 pg/mL, and 207 pg/mL, respectively (Fig 2, A). These findings make sFASL the most sensitive biomarker to rule out a FAS mutation, with values <200 pg/mL associated with a negative LR of 0.05 and a posttest probability of 7.7% (Fig 2, C; Table E4). Soluble FASL also showed a strong positive correlation with IL-10 (r = 0.8;P < .0001) and a moderate correlation with αβ-DNTcells (r = 0.6; P < .0001) and B12 levels (r = 0.69; P < .0001; see this article’s Fig E1, A, in the Online Repository at www.jacionline.org). The area under the ROC curve for sFASL, αβ-DNT cells, B12, and IL-10 levels were calculated to evaluate how well they discriminate patients with a FAS mutation from those without (Fig E1, B). The area under the curve for sFASL was 0.9 (defines an excellent test) and for αβ-DNT cells was 0.81. B12 and IL-10 exhibited areas significantly less than sFASL (P < .05), with values of 0.76 and 0.77. We next evaluated whether combinations of αβ-DNT cells, B12, IL-10, IL-18, and sFASL would have increased power to predict or exclude FAS mutations in patients suspected of ALPS (Fig 2, B; Table E4). The combination of αβ-DNT cells >4% with B12 >1500 ng/L or IL-10 >40 pg/mL or IL-18 >500 ng/mL or sFASL >300 pg/mL was associated with >95% probability of having a FAS mutation. Conversely, having αβ-DNT cells <2% in combination with IL-10 <20 pg/mL or B12 <1000 ng/L or IL-18 <500 ng/mL decreased the probability of a FAS mutation to less than 10% (Fig 2, C; Table E4). Finally, finding αβ-DNT cells <2% and sFASL <200 pg/mL resulted in <2% probability for a FAS mutation. In conclusion, the biomarkers described should aid in the selection of patients with findings of ALPS for further diagnostic workup. In addition, the presence of a combination of markers strongly suggestive of a FAS mutation in the setting of a negative genetic test should prompt a search for somatic mutations in sorted αβ-DNT cells.