The most appropriate primary outcomes to design clinical trials on Huntington's disease: meta-analyses of cohort studies and randomized placebo-controlled trials.