Monogenic Common Variable Immunodeficiency (Mo-CVID) Score for Optimizing the Genetic Diagnosis in Pediatric CVID Cohort

Abstract

Common variable immunodeficiency (CVID) represents an "umbrella" diagnosis due to its clinical and immunological heterogeneity. The primary objective of this study was to describe a cohort of CVID pediatric subjects from clinical, immunological, and genetic viewpoints. Secondary, we propose a model for prioritizing genetic investigations in these patients. Thirty-four patients with CVID followed at Meyer Children's Hospital, IRCSS, were enrolled. Whole exome sequencing was performed according to the latest International Union of Immunological Societies 2022 update. Genetic variants were identified in 16 patients (47%), including known variants in SLC39A7, PRKCD, STAT3, NFKB1, PIK3R1, PLCG2, RFXANK, PRKDC, TNFRSF13B, and novel variants in SPI1, NFKB1, NFKB2. Comparing the Gene+ and Gene- cohorts, we demonstrated that a monogenic cause is more likely to be found in cases of early disease onset, positive family history, autoimmunity, lymphoproliferation, and specific immunological alterations. Using these criteria, we developed a pediatric monogenic CVID (Mo-CVID) score to hypothesize when a CVID pediatric patient is more likely to carry a genetic mutation. A scoring system such as the Mo-CVID score could help physicians prioritize genetic testing. Genetic analysis in CVID patients can help stratify patients into different disease entities to predict complications and prognosis, ensure appropriate genetic counseling, and personalize treatment.

Keywords: children; common variable immunodeficiency; exome sequencing; primary immunodeficiency; prioritization.

FIGURE 1

Percentage of patients in whom at least one candidate genetic variant was identified, divided into single genes and their overall frequency in the cohort: Whole exome sequencing was performed according to the latest International Union of Immunological Societies 2022 update in 34 pediatric patients with diagnosis of Common variable immunodeficiency. A putative pathogenic variant was identified in 16 patients of the 34 (47%) patients under study. We identified variants in 11 genes (SLC39A7, SPI1, NFKB1, NFKB2, PRKCD, STAT3, PIK3R1, PLCG2, RFXANK, PRKDC, TNFRSF13B).

FIGURE 2

Bar chart giving an overview of the distribution of the eight significantly different (p < 0.05) features between Gene + and Gene− groups: Based on the results of the genetic analysis, patients were divided into two cohorts: gene‐positive (Gene +) with 16 patients and gene‐negative (Gene−) with 18 patients. Then, we compared the occurrence of demographic, clinical, and laboratory features in the two groups. We used Chi‐square or Fisher exact test for categorical variables, depending on the number of observations. Two‐sided p < 0.05 was considered statistically significant. The eight features are significantly different between Gene+ and Gene− groups were selected for the determination of the pediatric Mo‐CVID score: family history of IEIs, severe infections, infections with sequelae, pan hypogammaglobulinemia, absence of switched memory B cells, ≥3 autoimmune manifestations, ≥2 lymphoproliferative manifestations and clinical onset <4 years, or CVID diagnosis <7 years.

FIGURE 3

ROC curve and coordinates used for determination of relevant cut‐off scores to differentiate between likely and unlikely monogenic CVID cases: We identified 8 significantly different (p < 0.05) features between gene‐positive (Gene+, 16 patients) and gene‐negative (Gene−, 18 patients) groups and we calculated the sensitivity, specificity, positive predictive value, negative predictive value, accuracy and a weighted score for each characteristic. We determined cut‐off values of the total score to differentiate between likely and unlikely monogenic cases, based on a ROC curve. Cut‐off values of the scoring system were determined using a receiver operating characteristic (ROC) curve. Based on the sensitivity and the 1– specificity for each score, we determined a cut‐off for unlikely monogenic CVID at a score of 2 (sensitivity 100.0% [95% CI, 81.6−100.0]; specificity 80% [95% CI, 54.8−93.0]) and for likely monogenic CVID at a score of 7 (sensitivity 64.7% [95% CI, 41.3−82.7]; specificity 100.0% [95% CI, 79.6−100.0]).

FIGURE 4

Pediatric Mo‐CVID Scoring system and graphical distribution of our Gene + and Gene− patients according to the score: To design the scoring system, we selected 8 significantly different features between gene‐positive (Gene +, 16 patients) and gene‐negative (Gene−, 18 patients) groups. Then we calculated the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for each characteristic and we allocated a weighted score to each of them according to the significance of p‐value (p 0.05–0.03 = 1 point, p 0.029–0.002 = 2 points, p ≤ 0.001 = 3 points), with the maximum clinical score being 13 points (when all features are present). The total clinical score was calculated for each patient and we determined cut‐off values of the total score to differentiate between likely and unlikely monogenic cases, based on an ROC curve. According to the Mo‐CVID score at ≤ 2 points, the subject is unlikely to have a genetic mutation causing CVID, at 3–6 points, the presence of a genetic cause is possible, and at 7–13 points, the subject is very likely to have a genetic mutation causing CVID. In our cohort, 13 (72.2%) of the Gene− and none of the Gene + group had a score ≤ 2 points; 5 (27.8%) of the Gene− and 5 (31.3%) of the Gene+ had a 3–6 points score; none of the Gene− and 11 (68.8%) of the Gene+ had a 7–13 points score.