Figure 3.
Binomial logistic regression analysis predicting the association between class II HED scores and IAA phenotypes. (A) Forest plots showing the results of the binomial logistic regression analysis predicting the likelihood for class II and locus-specific HED scores of being associated with an IAA “immune-enriched” phenotype. All (n = 263); IAA responders (n = 141); IAA >20 years (n = 216); IAA with PNH clones (n = 135); IAA >20 years with PNH clone and responders (n = 59). (B) Multivariable logistic regression analysis testing the independent effect of HED and risk alleles on idiopathic BMF phenotype. The length of the bars indicates the OR; the error bars show the 95% CI, the numbers on the bars depict the P values resulting from the likelihood ratio test. Two distinct models are built for DRB1 and DQB1 locus. (C) Violin plots representing the mean class II HED distribution across healthy controls and IAA patients not carrying DRB1*15:01. Wilcoxon signed-rank test was used to calculate the P value. (D) Violin plots representing the DRB1 HED distribution across HCs and IAA patients not carrying DRB1*15:01. Wilcoxon signed-rank test was used to calculate the P value. (E) Violin plots representing the mean class II HED distribution across HCs and IAA patients not carrying DQB1*06:02. Wilcoxon signed-rank test was used to calculate the P value. (F) Violin plots representing the DQB1 HED distribution across HCs and IAA patients not carrying DQB1*06:02. Wilcoxon signed-rank test was used to calculate the P value. Comment on Figure 3C-F: Here, our intention was to clarify whether the pattern of lower class II HED seen in IAA/PNH patients was related only to the enrichment in certain risk alleles (such as DRB1*15:01 or DQB1*06:02) or could instead uncover a broader immunogenetic aspect encompassing risk allele profiles. We therefore analyzed HED class II configurations in patients vs controls not carrying those risk alleles and showed that their scores were lower than controls, as a result of higher structural similarity between 2 DRB1 and DQB1 alleles in patients, regardless of the presence of risk genotypes.

Binomial logistic regression analysis predicting the association between class II HED scores and IAA phenotypes. (A) Forest plots showing the results of the binomial logistic regression analysis predicting the likelihood for class II and locus-specific HED scores of being associated with an IAA “immune-enriched” phenotype. All (n = 263); IAA responders (n = 141); IAA >20 years (n = 216); IAA with PNH clones (n = 135); IAA >20 years with PNH clone and responders (n = 59). (B) Multivariable logistic regression analysis testing the independent effect of HED and risk alleles on idiopathic BMF phenotype. The length of the bars indicates the OR; the error bars show the 95% CI, the numbers on the bars depict the P values resulting from the likelihood ratio test. Two distinct models are built for DRB1 and DQB1 locus. (C) Violin plots representing the mean class II HED distribution across healthy controls and IAA patients not carrying DRB1*15:01. Wilcoxon signed-rank test was used to calculate the P value. (D) Violin plots representing the DRB1 HED distribution across HCs and IAA patients not carrying DRB1*15:01. Wilcoxon signed-rank test was used to calculate the P value. (E) Violin plots representing the mean class II HED distribution across HCs and IAA patients not carrying DQB1*06:02. Wilcoxon signed-rank test was used to calculate the P value. (F) Violin plots representing the DQB1 HED distribution across HCs and IAA patients not carrying DQB1*06:02. Wilcoxon signed-rank test was used to calculate the P value. Comment on Figure 3C-F: Here, our intention was to clarify whether the pattern of lower class II HED seen in IAA/PNH patients was related only to the enrichment in certain risk alleles (such as DRB1*15:01 or DQB1*06:02) or could instead uncover a broader immunogenetic aspect encompassing risk allele profiles. We therefore analyzed HED class II configurations in patients vs controls not carrying those risk alleles and showed that their scores were lower than controls, as a result of higher structural similarity between 2 DRB1 and DQB1 alleles in patients, regardless of the presence of risk genotypes.

Close Modal
This Feature Is Available To Subscribers Only

or Create an Account

Close Modal
Close Modal