Noninvasive PET Imaging of Pathologic VLA-4–Mediated Hyperadhesion in Sickle Cell Disease
Vanessa Shehu, Charles M Laymon, Leticia Candra, Xiyi Wang, Yi-Fan Chen, Sina Tavakoli, Scott E Snyder, Isabelle Tomassoli, Charles Jonassaint, Carolyn Anderson, Enrico M NovelliAbstract
Background
Vaso-occlusive episodes (VOE) pose a significant health burden on patients with sickle cell disease (SCD) and remain the leading cause of emergency department visits due to SCD (Jang et al., Journal of Translational Medicine, 2021). In the vascular pathophysiology underlying VOE, chronic hyper-adhesion of red and white blood cells to the endothelium acutely worsens in the microvasculature of bones and bone marrow, leading to onset of VOE. Hyperadhesion is driven by overexpression of adhesion molecules on inflamed endothelium. Among the adhesion molecules overexpressed in SCD VLA-4 plays an important role in reticulocyte and leukocyte binding and recruitment to sites of VOE. A major challenge in prediction, treatment, and prevention of VOE is the absence of objective, clinically measurable biomarkers for hyperadhesion, which hinders the ability to personalize treatment. VOE-associated pain is subjective and therefore an unreliable surrogate for VOE frequency and efficacy of anti-adhesion therapy, as pain perception varies among patients and may be difficult to distinguish from chronic, steady-state pain. To address this unmet need, we investigated the utility of the PET imaging tracer, [64Cu]LLP2A—that binds to VLA-4—as a noninvasive imaging biomarker of hyperadhesion during steady-state (baseline clinical status) SCD and VOE.
Methods
We obtained [64Cu]LLP2A PET/CT scans of eight patients with SCD (mean age 38.9 years; 62.5% males; 3 HbSS and 5 HbSC) in steady state, with two of the patients additionally scanned during VOE. We generated regions of interest (ROIs) on CT around humoral and femoral heads, bilaterally (Skellytour [https://github.com/cpwardell/Skellytour] and PMOD [https://www.pmod.com] imaging software) that are common sites of VOE and complications such as osteonecrosis (Mann et al., Radiology. Artificial intelligence, 2025). ROIs were applied to the registered PET to extract [64Cu]LLP2A standardized uptake values (SUVs). Pain was assessed via the visual analog score (VAS) and using Painimation, a digital pain assessment tool that allows patients to localize and describe pain through drawings and descriptive animations, respectively (O’Brien et al., The Journal of Pain, 2024). We performed preliminary correlation network analysis of baseline scans to investigate the association between pain measures and PET tracer uptake (mean SUV across the four ROIs). Spearman correlation coefficients are reported with 95% confidence intervals estimated using Fieller’s approach, which is robust for small sample sizes.
Results
Preliminary findings from correlation network analysis reveal a strong correlation between pain measures, with pain intensity highly correlated with body pain percentage (percentage of body areas with reported pain; r = 0.90, 95% CI: [0.42, 0.99]. Strong inter-region correlations were also observed among PET SUVs (r = 0.86-0.96; corresponding 95% CIs reported). The correlation between pain measures and PET SUVs was weaker (r = 0.33 to 0.45) but may yet have biological significance in the context of this small preliminary sample size.
Conclusions
Our preliminary findings suggest a weak-to-moderate correlation between patients’ pain measures and mean uptake of [64Cu]LLP2A within ROIs across humoral and femoral heads. Given our small preliminary sample size, this finding warrants further study of [64Cu]LLP2A as a biomarker of hyperadhesion in SCD.