Iron, inflammation, and early death in adults with sickle cell disease

Abstract

Rationale: Patients with sickle cell disease (SCD) have markers of chronic inflammation, but the mechanism of inflammation and its relevance to patient survival are unknown.

Objective: To assess the relationship between iron, inflammation, and early death in SCD.

Methods and results: Using peripheral blood mononuclear cell transcriptome profile hierarchical clustering, we classified 24 patients and 10 controls in clusters with significantly different expression of genes known to be regulated by iron. Subsequent gene set enrichment analysis showed that many genes associated with the high iron cluster were involved in the toll-like receptor system (TLR4, TLR7, and TLR8) and inflammasome complex pathway (NLRP3, NLRC4, and CASP1). Quantitative PCR confirmed this classification and showed that ferritin light chain, TLR4, and interleukin-6 expression were >100-fold higher in patients than in controls (P<0.001). Further linking intracellular iron and inflammation, 14 SCD patients with a ferroportin Q248H variant that causes intracellular iron accumulation had significantly higher levels of interleukin-6 and C-reactive protein compared with 14 matched SCD patients with the wild-type allele (P<0.05). Finally, in a cohort of 412 patients followed for a median period of 47 months (interquartile range, 24-82), C-reactive protein was strongly and independently associated with early death (hazard ratio, 3.0; 95% confidence interval, 1.7-5.2; P<0.001).

Conclusions: Gene expression markers of high intracellular iron in patients with SCD are associated with markers of inflammation and mortality. The results support a model in which intracellular iron promotes inflammatory pathways, such as the TLR system and the inflammasome, identifying important new pathways for additional investigation.

Keywords: gene expression; inflammation; iron; mortality; sickle cell disease.

Figure 1. A subgroup of SCD patients cluster together in a group with very high expression of iron regulated genes

PBMC transcriptome profiling in 24 SCD patients and 10 healthy controls identified 3 clusters of subjects with high, low and intermediate expression of iron regulated genes. patients (ClinicalTrials.gov identifiers: NCT00072826). The Heat map shows unsupervised two-way hierarchical clustering of the pre-defined set of iron regulated genes on the y-axis (for a tabular version of this list, see supplemental data) and the clustering of study participants on the x-axis. None of the controls was found to be in the high iron cluster. “S” means SCA patient; “C” means control.

Figure 2. QPCR results confirm the relation between iron regulated genes and genes involved in inflammatory and anti-inflammatory pathways

To validate the microaray results we performed qantitative polymerase chain reaction (QPCR) normalized to the housekeeping gene RNA18S5 on the same samples. (ClinicalTrials.gov identifiers: NCT00072826) (A) This diagram shows a good correlation (R=0.72, P=0.023) between the fold change as found by microarray and QPCR. (B) Bars show the fold change in expression of the group of patients with highly expressed iron regulated genes over the other groups. (C) The number of PCR cycles to reach treshold compared to the housekeeping gene RNA18S5 (ΔCT) of FTL was highly correlated to TLR4 in both patients (R=0.930, P<0.001) as controls (R=0.943, P=0.005) although mean expression of both genes was much higher in patients than in controls (P<0.001). TLR4 was also highly correlated to ALDH1A1, GAPDH, SAT2 and HMOX expression (data not shown). To get an better impression of actual gene expression X-and Y-axis are reversed and originate 20 ΔCT. (D) Bars show the fold change in expression of selected genes in SCD patients compared to healthy controls. ■ Black bars represent genes that were included in the pre-defined set of iron regulated genes (see Supplemental table I). □ White bars represent genes that were identified as signifcantly differentially expressed between the three iron clusters or were included to validate the microarray data. GAPDH: Glyceraldehyde-3-phosphate dehydrogenase, TLR4: Toll-like receptor 4, ALDH1A1: Aldehyde Dehydrogenase 1A1, FTL1: ferritin light chain 1, HMOX: Heme oxygenase-1, SAT2: spermidine/spermine N1-acetyltransferase family member 2, IL6: interleukin-6, ECE1: endothelin converting enzyme 1, ST13: suppression of tumorigenicity 13.

Figure 3. Patients with the ferroportin Q248H mutation which is associated with high retention of intracellular iron have higher levels of CRP and interleukin-6

To assess relationship between intracellular iron and inflammation we used available ferroportin Q248H mutation status. Patients were analyzed previously for ferroportin mutation status.(A,B) In the gene-expression part of the study (ClinicalTrials.gov identifier: NCT00072826) the five patients with a ferroportin Q248H mutation (GT and/or TT genotype) had a significant (p<0.05) higher plasma level of CRP and log 10 IL-6 trended (P=0.08) to be higher compared to 19 patients with wild type ferroportin (GG genotype).(C,D) As a sensitivity analysis we repeated this in a case-control study in selection of patients of the NIH SCD cohort (ClinicalTrials.gov identifier: NCT00011648). Fourteen patients with Q248H mutation were matched on genotype and ferritin level. We found that the 14 patients with the mutation had a significant higher CRP and log10 il6 than the matched controls. Because the ferroportin Q248H is also associated with less sensitivity for hepcidin, and hepcidin levels are strongly associated with ferritin levels in SCD we decided to perform this last analysis only in patients with a ferritin level smaller than 1000 mg/dL. * P<0.05

Figure 4. Kaplan Meier curve showing a significant difference in survival between patients with low and high CRP

SCD Patients (n=412) of the NIH pulmonary hypertension screening cohort (ClinicalTrials.gov identifier: NCT00011648) were divided in a high CRP group and a low CRP group at the 75^th percentile level of CRP (0.8 mg/dL, (upper limit of normal)). Median follow up time was 47 months, (IQR 24-82, range 2-132). After 100 months patients were censored. Log rank test showed that patients with CRP above 0.8 mg/dL had significantly lower (p=0.0035) survival rate by time from enrollment. Five year mortality percentages for the low, intermediate and high CRP groups were respectively 12.4%, 20.5 and 25.8%. In a multivariate analysis, CRP was an independent predictor of mortality (see Table 2).