Evaluating Heterogeneity of Hippocampal Avoidant Whole Brain Radiotherapy (HA-WBRT) Treatment Effect on Neurocognitive Function (NCF): Secondary Analysis of NRG-CC001

Abstract

<h3>Purpose/Objective(s)</h3> NRG-CC001, a trial evaluating the neuroprotective effect of HA-WBRT versus standard WBRT in patients with brain metastases (BM), demonstrated a significantly lower risk of NCF decline. Given the heterogeneous BM patient population, we sought to systematically identify subgroups of patients through secondary analysis of trial data that could derive differential benefit from HA. <h3>Materials/Methods</h3> We performed a post hoc analysis of NRG-CC001, a phase III clinical trial in which 518 patients with BM were randomly assigned to WBRT plus memantine or HA-WBRT plus memantine. In this secondary analysis, cumulative incidence function (CIF) estimated NCF failure between arms and subgroups was tested using Gray's test. Covariate and subgroup association with differential treatment response was calculated using Cox proportional hazards modeling; these factors were then tested for interaction with treatment arm. <h3>Results</h3> Age (>61), education (<high school) irrespective of treatment arm were predictive of higher risk of NCF failure (p=0.005 and p=0.007, respectively). Patients living <4 months derived no benefit from HA-WBRT (Gray's test p=0.793), but for those living >=4 months, there was a significant benefit favoring HA-WBRT (p=0.03). We found evidence of heterogeneity of treatment effect based on primary tumor histology (interaction test p=0.01): primary lung histology patients (HR 0.58, 95% CI: 0.43-0.77, p=0.0007) compared to those with non-lung histology primary tumors (HR 1.15, 95% CI: 0.78-1.71, p=0.48). Baseline MD Anderson Symptom Inventory for brain tumor (MDASI-BT) cognitive factor scores were also found to be highly associated with treatment response with significance on test for interaction (p=0.03); those with MDASI-BT baseline scores in quartiles 1-3 had significantly greater differential treatment response (HR=0.64, 95% CI: 0.48-0.85, p=0.002) compared to those with the highest MDASI-BT baseline scores (HR=1.24, 95% CI: 0.76-2.04, p=0.39). There was no significant heterogeneity of treatment effect based on age, education level, and baseline KPS. <h3>Conclusion</h3> This post-hoc subgroup analysis of NRG-CC001 evaluating heterogeneity of treatment effect confirmed that the neurocognitive protective benefit of HA-WBRT was most substantial for patients living > 4 months; for those surviving <4 months, statistically significant benefit was not identified, possibly because cognitive decline post-WBRT requires a window of time to manifest. We also found that treatment response was significantly associated with several patient and tumor characteristics. Patients with lung histology as well as those with lower baseline MDASI-BT cognitive factor scores, denoting less severe cognitive impairment at enrollment, had significantly greater benefit from HA-WBRT. These data may better help clinicians decide which subgroups of patients benefit most from HA-WBRT.