Alcazar, Roberto

Introduction: Functional outcome scales are commonly used in neurological clinical trials. Prior studies suggested that these scales may reflect symptomatic hemorrhage (SH), but their performance in detecting treatment effects has not been assessed in clinical trials of cerebral cavernous malformations (CCM). And they have not been correlated with CCM bleeding as reflected by lesional iron content assessed by quantitative susceptibility mapping (QSM) MRI. Hypothesis: We hypothesized that functional outcome measures would detect differences between treatment arms in a randomized clinical trial of hemorrhagic CCM, and reflect new SH or a ≥6% increase in mean lesional QSM reflective of CCM bleeding. Methods: Eighty adults with CCM-related SH in the prior 12 months of enrollment were randomized 1:1 to atorvastatin 80 mg/day or placebo for 2 years. Functional outcomes, including mRS, EQ-5D index, EQ-VAS, and PROMIS-29 domains, were collected at baseline, Year 1, and Year 2. Analyses were conducted in 63 participants (32 atorvastatin, 31 placebo) in the modified intention-to-treat trial cohort who had contributed at least one paired annual clinical and imaging assessment. Results: No significant differences were found between atorvastatin and placebo groups in any functional outcome measure across timepoints. Rates of functional decline and improvement were comparable, except for PROMIS Fatigue which improved more in the atorvastatin than placebo patients from baseline to Year 1 (p=0.019). Pooled analyses showed that several functional domains had strong specificity for identifying SH. PROMIS Sleep Disturbance decline from baseline to Year 2 was associated with SH (p=0.0045; specificity 92.3%, sensitivity 57.1%). A ≥10-point drop in EQ-VAS from baseline to Year 1 identified all SH cases during that interval (p=0.007; sensitivity 100%, specificity 83.6%). Functional decline had low sensitivity in detecting CCM bleeding on QSM, though specificity was high (>75%) across domains. Conclusions: Atorvastatin did not influence functional change over 2 years by any measure. PROMIS Sleep Disturbance and EQ-VAS demonstrated high specificity, and in the case of EQ-VAS also a high sensitivity for identifying SH. Functional outcomes across domains also had high specificity for occult bleeding in CCMs as measured by QSM. Poor sensitivity and consistent improvements over time demonstrate the limitations of functional outcome measures as endpoints of drug effects on CCM bleeding.

Background: It is not clear if changes in lesion volume might represent a clinically relevant and statistically advantageous outcome parameter in early phase trials of novel drugs aimed at preventing bleeding in cerebral cavernous malformations (CCM). Methods: We analyzed the modified intention to treat cohort (33 patients receiving atorvastatin and 31 placebo) who contributed at least one paired imaging biomarker in the recently completed double-blinded randomized controlled AT CASH EPOC trial (clinicaltrials.gov NCT02603328). We examined prospective changes in CCM lesion volume and correlated these with symptomatic hemorrhage (SH) events. We contemporaneously assessed changes in lesional iron content by quantitative susceptibility mapping (QSM) on MRI, and objective (mRS) and subjective (EQ-VAS) functional status. We also simulated sample sizes needed to detect potential effects of a drug on lesion volume. Results: There was excellent correlation between independently assessed maximum lesion diameter and CCM lesion volume (R 2 = 0.68, P<0.0001). The mean (SD) percentage change in lesion volume per year was +6 (65) and +18 (94) in atorvastatin and placebo patients, respectively (P= 0.43). All cases with SH had increases in lesion volume as well as increases in mean lesional iron content by QSM, and most had declines in functional status on mRS and EQ-VAS. But there were many cases with increases in lesion volume with improved functional status or decreased mean lesional QSM, and other cases with worsening in function or increased lesional QSM and decreased lesion volume. There were no significant correlations of changes in lesion volume with concomitant changes in mean lesional QSM, mRS or EQ-VAS (all R/R 2 <0.5 and Bonferroni corrected P>0.05), including in brainstem lesions analyzed separately. Sample size simulations indicate larger numbers of cases needed to demonstrate a drug effect on changes in lesion volume than those needed to detect an effect on QSM change with a=0.05 and b=0.8, and even larger samples to detect effects on the proportion of cases with increased or decreased lesion volume. Conclusion: Changes in lesion volume occur frequently in CCM lesions after recent SH, but do not correlate with measures of bleeding in the lesions, nor appear to reflect functional changes. And they are not statistically advantageous in comparison to QSM measures for detecting potential drug effects.