Dissertations & Theses (Open Access)

APPLICATION OF GROUP-BASED TRAJECTORY MODELING TO INVESTIGATE BIOMARKERS CHANGE WITH TIME IN INTRACEREBRAL HEMORRHAGE PATIENTS

NGOC QUYNH HOANG NGUYEN, UTHealth School of Public Health

Abstract

Intracerebral hemorrhage (ICH) is one of the world deadly diseases that accounted for a high number of deaths. It occurs as tissues in the brain start to bleed which then lead to strokes and other complications. This poses as a threat in public health and many researches are still struggling to find an effective care and treatment. This study aimed to evaluate hematoma volume and edema volume collected from baseline to day 70 since stroke onset into distinct profiles to understand their relationship with the functional outcomes. A type of finite mixture modeling called group-based trajectory modeling (GBTM) was applied to classify patients who shared similar patterns of hematoma and edema volumes into distinct trajectories. Patient diagnosed with ICH were recruited in a prospective, randomized, blinded controlled experiment given dose-escalation of pioglitazone. Their hematoma volume and edema volume were assessed through magnetic resonance imaging at pre-specified time points up to day 70 since stroke onset. Functional outcomes measured include modified Rankin Scale (mRS), Barthel Index, Stroke Impact Scale-16, and EuroQol evaluated at 90 and 180 days. Three trajectory groups were identified for both biomarkers classified as low volume group, moderate volume group, and high volume group for 78 patients. For hematoma volume analysis, history of diabetes (p = .011), and baseline characteristics such as initial systolic blood pressure (p = .043), Glasgow Coma Scale (GCS) (p <.001), National Institutes of Health Stroke Scale (NIHSS) (p = .004), hematoma volume (p <.001) as well as hematoma region (p = .021) were statistically related to the identified trajectories. After adjusting for potential confounders, we found that the high volume group was associated with worse functional outcomes comparing to the low volume group. For example, comparing to low volume group, the high volume group was associated with lower Stroke Impact Scale-16 (adjusted OR=0.05, 95% CI= 0.00, 0.60) and lower EuroQol (adjusted OR=0.04, 95% CI=0.00, 0.51) at 180 days. Similar results were obtained for edema volume analysis. Baseline variables such as GCS (p = .001), NIHSS (p = .016), hematoma volume (p <.001), and edema volume (p <0.001) were associated with the three identified trajectory groups. The high edema volume group was associated with lower Stroke Impact Scale-16 (adjusted OR=0.03, 95% CI=0.00, 0.98), and lower EuroQol (adjusted OR= 0.01, 95% CI= 0.00, 0.66) at day 180 comparing to the low edema volume group after controlling potential confounding effects. The identified trajectories projected the patterns of the two imaging biomarkers for which they were important in predicting clinical outcomes. Using GBTM, we have shown that the hematoma volume and edema volume could serve as potential biomarkers for predicting long-term functional outcomes and can be added to the knowledge of ICH to help with developing new treatment and intervention.