Cerebral Blood Flow and Pain Perception in Sickle Cell Disease: A Mediation Analysis

Background:

Sickle Cell Disease (SCD) is a genetic disorder characterized by the presence of sickle-shaped red blood cells, leading to various complications, including chronic pain and cerebrovascular issues. Understanding the mechanisms underlying pain perception in SCD patients is crucial for developing targeted treatments. Cerebral blood flow (CBF) has been implicated in the modulation of pain experiences, but the pathways through which CBF influences pain perception remain unclear. Cognitive functions, hemoglobin levels, and heart rate are potential mediators in this process, given their roles in both neurological and physiological regulation. This study aims to investigate the causal pathways through which CBF impacts pain experience in adult SCD patients, considering the influence of various mediators and the potential confounding effect of SCD genotype variability.

Method:

The study analyzed data from nine SCD patients. Two patients with brain damage were excluded to focus on those without brain injury. The final sample included seven patients with various SCD genotypes to account for differences in disease severity. Pearson correlation examined the linear relationship between each mediator and the exposure, each mediator and each outcome, and the exposure and each outcome. ANOVA compared reported pain experience differences across genotypes. Mediation analysis investigated the pathways through which cerebral blood flow (CBF) impacts pain experience, defined as the total score on the PainDETECT survey or the brief pain inventory (BPI). The analysis considered cognitive functions (cognitive flexibility, executive function, and processing speed), hemoglobin levels, and heart rate as mediators, with CBF as the exposure, and genotype as a confounder. PainDETECT and BPI were considered as outcomes.

Results:

The Pearson correlation analysis showed no significant linear correlations between PainDETECT and any of the exposure or mediator variables. Similarly, the BPI did not exhibit any significant correlations with the exposure or mediator variables. ANOVA results showed that genotype had no significant effect on PainDETECT (p=0.1006), whereas it did affect BPI significantly (p=0.0110), with the SC genotype having the highest mean and Sβ⁰ genotype showing the lowest mean. Mediation analysis revealed that when PainDETECT is considered as an outcome, no relationship is observed between mediators, exposure and outcome. However, mediation analysis showed that CBF can impact BPI through hemoglobin with a total effect of 0.02 (p=0.76) when hemoglobin is considered as a mediator. This includes a direct individual impact of 0.08 (p=0.0010) and an indirect impact through hemoglobin of -0.08 (p=0.0490). When processing speed is taken into consideration regardless of any other mediators, mediation analysis revealed that CBF can impact BPI with a total effect of 0.02 (p-value=0.76). The direct and indirect effects of this impact are as follows: NDE (-0.07, p=0.09) and NIE (0.07, p=0.07). Other mediators did not reach mediation significance.

Conclusion:

Our findings indicate that neither PainDETECT nor BPI alone showed a direct linear correlation with any exposure or mediator variables. However, previous studies have shown that higher cerebral blood flow (CBF) correlates with lower hemoglobin (Hb) levels and processing speed. When we introduced CBF as an exposure variable and Hb or processing speed as mediators, we observed a positive natural direct effect (NDE) on BPI, suggesting that increased CBF directly increases BPI. Conversely, the natural indirect effect (NIE) was negative, indicating that increased CBF leads to decreased Hb levels, thereby reducing BPI. This suggests partial mediation of the effect of CBF on BPI through Hb levels. In contrast, when processing speed is considered as a mediator, increased CBF was found to directly decrease processing speed and possibly increase BPI consequently. These results suggest that neither CBF nor Hb and processing speed individually affect BPI, but when considered together in the model, they demonstrate a significant effect. This underscores the importance of considering these factors collectively rather than as independent predictors when predicting BPI.

Smith:Pfizer: Consultancy; Vertex: Honoraria.