Progression of type 1 diabetes is associated with high levels of soluble PD-1 in islet autoantibody-positive children

Aims/hypothesis Type 1 diabetes is an autoimmune disorder that is characterised by destruction of pancreatic beta cells by autoreactive T lymphocytes. Although islet autoantibodies (AAb) are an indicator of disease progression, specific immune biomarkers that can be used as target molecules to halt development of type 1 diabetes have not been discovered. Soluble immune checkpoint molecules (sICM) play a pivotal role in counteracting excessive lymphocyte responses, but their role in type 1 diabetes is unexplored. In this longitudinal study, we measured sICM levels in AAb-positive (AAb+) children to identify molecules related to type 1 diabetes progression. Methods We measured the levels of 14 sICM in the sera of AAb+ children (n=57) compared to those with recent-onset type 1 diabetes (n=79) and healthy children (n=44), obtained from two cohorts. AAb+ children were followed up and divided based on their progression to type 1 diabetes (AAbP) or not (AAbNP) (if they lost islet autoimmunity and did not develop disease in subsequent years). sICM were also measured in the sample taken at the visit closest to disease onset in AAbP children. Results We found that AAb+ children had a distinct sICM profile compared with healthy children and those with recent-onset type 1 diabetes. In addition, AAb+ children who progressed to type 1 diabetes (AAbP) had higher sICM concentrations than non-progressors (AAbNP). Further, sICM levels decreased in AAbP children close to disease onset. Application of Cox regression models highlighted that high concentrations of soluble programmed cell death protein 1 (sPD-1) are associated with type 1 diabetes progression (HR 1.71; 95% CI 1.16, 2.51; p=0.007). Conclusions/interpretation This study reveals an sICM profile that is dysregulated during the preclinical stage of type 1 diabetes, and identifies sPD-1 as a pathophysiologically-relevant molecule that is associated with disease progression, offering a potential target for early interventions in autoimmune diabetes. Graphical Abstract Supplementary Information The online version contains peer-reviewed but unedited supplementary material available at 10.1007/s00125-023-06075-3.


ESM Table 1.
Characteristics of children enroled in the study at the time of sICM measurement.Data are reported using median (range) in the case of numerical variables and absolute frequencies and percentages for categorical factors.Kruskall-Wallis ANOVA and Chi -square test were used as omnibus tests.

Characteristics
Healthy children

ESM Table 2.
Characteristics of AAb + children at the time of sICM measurement divided in children who developed type 1 diabetes in subsequent years (AAb P ) and those who did not (AAb NP ).Data are reported using mean ± standard deviation (SD) with range or median with range in case of numerical variables and absolute frequencies and percentages for categorical factors.Accordingly, differences between groups were assessed through t-test for unpaired samples or Chi-square test.a Heterozygosity for the two riskassociated haplotypes DR4-DQ8 and DR3-DQ2 (DR3/4).b Homozygosity for the two risk-associated haplotypes DR4-DQ8 and DR3-DQ2 (DR4/4 and DR3/3) or risk-associated haplotype DR4-DQ8 combined with a neutral haplotype (DR4/X).c Risk-associated haplotype DR3-DQ2 combined with a neutral haplotype DR15-DQ6 (DR3/X).d Homozygosity for neutral haplotypes (DRX/X) or riskassociated haplotypes DR4-DQ8 and DR3-DQ2 combined with the protective haplotype DR15-DQ6 (DR4/15 and DR3/15).e Neutral haplotype combined with protective haplotype (DRX/15).NA, not applicable.

ESM Table 3.
Levels of sICM measured in healthy, AAb + and recent-onset type 1 diabetes (T1D) children subdivided for the ABIS and preT1D-UNINA cohorts, respectively.Data are reported using median [Interquartile range] (min to max).

Healthy children
AAb + children T1D children (a)  Correlation between sPD1 concentrations (pg/ml) and age at time of sICM measurement in healthy children (HC, grey line and dots), islet-autoantibody positive (AAb + , orange line and dots) and type 1 diabetes (T1D, blue line and dots) children.R=-0.322(p=0.102)  for HC, R=-0.656 (p<0.001) for AAb + , R=-0.359 (p=0.001) for T1D by Pearson correlation.p<0.05 is considered statistically significant.(b-c) Box plots showing the distribution of sPD-1 concentrations (pg/ml) in healthy children (HC, grey circles), AAb + (orange triangles) and type 1 diabetes (T1D, blue squares) under (b) or above (c) 10 years of age.Data are shown as the median (horizontal line in the box) and Q1 and Q3 (borders of the box).Whiskers show the lowest and highest values that are not outliers (i.e., points below Q1 − 1.5 × IQR or above Q3 + 1.5 × IQR).Dots outside the whiskers represent outlier values.***p<0.001by Mann-Whitney U test.(d) Box plots showing the distribution of sPD-1 levels in AAb + children, under 10 years of age, who developed (AAb P , violet circles) or did not develop (AAb NP , green triangles) type 1 diabetes in the following years.Data are shown as the median (horizontal line in the box) and Q1 and Q3 (borders of the box).Whiskers show the lowest and highest values that are not outliers (i.e., data points below Q1 − 1.5 × IQR or above Q3 + 1.5 × IQR).Dots outside the whiskers represent outlier values.**p<0.01 by Mann-Whitney U test.

Table 5 .
Left, levels of sICM measured in AAb + children who developed type 1 diabetes (AAb P ) at the time of