A study examined the role of a humanized monoclonal antibody (mAb) that binds soluble forms of amyloid in treating Alzheimer disease. Patients were randomly assigned to a mAb injection or placebo group for 14 months. The outcomes included changes in cognitive performance assessed by different scores at week 72 compared to baseline. Higher scores indicate worse impairment. The following numbers were reported for patients with mild and moderate Alzheimer disease. Which of the following is the most accurate interpretation of these study results?

Question

A study examined the role of a humanized monoclonal antibody (mAb) that binds soluble forms of amyloid in treating Alzheimer disease.  Patients were randomly assigned to a mAb injection or placebo group for 14 months.  The outcomes included changes in cognitive performance assessed by different scores at week 72 compared to baseline.  Higher scores indicate worse impairment.  The following numbers were reported for patients with mild and moderate Alzheimer disease.  Which of the following is the most accurate interpretation of these study results?

Quizplus · Accepted Answer

Explanation: This trial compared cognitive performance scores with and without monoclonal antibody (mAb) therapy among patients in 2 disease categories (mild and moderate Alzheimer). These 2 categories were evaluated separately but presented in the same analysis, so a test for heterogeneity helps to exclude differences between them. Heterogeneity testing is particularly important when conducting a meta-analysis of several trials as it can provide insight about the combinability of multiple studies. A commonly used heterogeneity test is the Q statistic, which results in a corresponding P value (last column on the right in the table). In general, the null hypothesis (H₀) denotes that there is no difference between the groups studied. A small significant P value (<0.05) suggests that there is a difference and that H₀ should be rejected. However, large nonsignificant P value (>0.05) suggests that there is no difference between the groups studied so H₀ cannot be rejected. For the Q statistic in this example, H₀ represents no difference (no heterogeneity) in response to mAb therapy between patients with mild and moderate Alzheimer disease. The large P values obtained (0.67, 0.86, and 0.94) mean that H₀ cannot be rejected. In other words, there was no differential response to mAb therapy depending on disease severity. Another commonly used heterogeneity test is the I² index (derived from the Q statistic). An I² index = 0 represents no heterogeneity. I² <25% is traditionally considered to represent low heterogeneity (<25% of the variability seen is due to true heterogeneity); the cutoffs for moderate and high heterogeneity are I² = 50% and I² = 75%, respectively. (Choice A) The P values listed under "Mild Alzheimer disease" and "Moderate Alzheimer disease" relate to the differences in scores for patients treated with mAb versus those treated with placebo within each group. None of the P values are significant, so it is difficult to comment on the statistical significance of these score differences or on the effectiveness of mAb in slowing disease progression. (Choices B and C) Even though in most cases, the P values are smaller in the mild Alzheimer disease group than in the moderate Alzheimer disease group, this does not necessarily mean that the antibodies were more effective in the mild Alzheimer disease group, particularly because none of the P values are significant. (Choice D) Patients with mild Alzheimer disease did not have a sharper decline than patients with moderate Alzheimer disease for scores 2 (-1.6 in mild versus -1.9 in moderate Alzheimer disease) and 3 (-0.6 in mild versus -0.7 in moderate Alzheimer disease); only score 1 declined more (-1.5 in mild versus -0.8 in moderate Alzheimer disease). Furthermore, the confidence intervals all cross 0, so the results are not statistically significant. Educational objective: Tests for heterogeneity are useful when performing a meta analysis or comparing different trials. Common tests for heterogeneity include the Q statistic (no heterogeneity if P >0.05) and the I² index (25%, 50%, and 75% are traditionally considered cutoffs for low, moderate, and high heterogeneity, respectively). References: Quantifying, displaying and accounting for heterogeneity in the meta-analysis of RCTs using standard and generalised Q statistics (http://www.ncbi.nlm.nih.gov/pubmed/3102034) Assessing heterogeneity in meta-analysis: Q statistic or I2 index? (http://www.ncbi.nlm.nih.gov/pubmed/16784338)

A Study Examined the Role of a Humanized Monoclonal Antibody