Preclinical trials in Alzheimer’s disease: Sample size and effect size for behavioural and neuropathological outcomes in 5xFAD mice

Abstract

5xFAD transgenic (TG) mice are used widely in AD preclinical trials; however, data on sample sizes are largely unaddressed. We therefore performed estimates of sample sizes and effect sizes for typical behavioural and neuropathological outcome measures in TG 5xFAD mice, based upon data from single-sex (female) groups. Group-size estimates to detect normalisation of TG body weight to WT littermate levels at 5.5m of age were N = 9–15 depending upon algorithm. However, by 1 year of age, group sizes were small (N = 1 –<6), likely reflecting the large difference between genotypes at this age. To detect normalisation of TG open-field hyperactivity to WT levels at 13-14m, group sizes were also small (N = 6–8). Cued learning in the Morris water maze (MWM) was normal in Young TG mice (5m of age). Mild deficits were noted during MWM spatial learning and memory. MWM reversal learning and memory revealed greater impairment, and groups of up to 22 TG mice were estimated to detect normalisation to WT performance. In contrast, Aged TG mice (tested between 13 and 14m) failed to complete the visual learning (non-spatial) phase of MWM learning, likely due to a failure to recognise the platform as an escape. Estimates of group size to detect normalisation of this severe impairment were small (N = 6–9, depending upon algorithm). Other cognitive tests including spontaneous and forced alternation and novel-object recognition either failed to reveal deficits in TG mice or deficits were negligible. For neuropathological outcomes, plaque load, astrocytosis and microgliosis in frontal cortex and hippocampus were quantified in TG mice aged 2m, 4m and 6m. Sample-size estimates were ≤9 to detect the equivalent of a reduction in plaque load to the level of 2m-old TG mice or the equivalent of normalisation of neuroinflammation outcomes. However, for a smaller effect size of 30%, larger groups of up to 21 mice were estimated. In light of published guidelines on preclinical trial design, these data may be used to provide provisional sample sizes and optimise preclinical trials in 5xFAD TG mice.

No Evidence of Sensory Neuropathy in a Traditional Mouse Model of Idiopathic Parkinson’s Disease

“…Moreover, sample size calculations were "uniformly absent" from the papers they reviewed. A lack of sample size calculations also occurs in preclinical Alzheimer's disease research [66] and in preclinical research on traumatic brain injury [67]; however, for example, we recently published sample sizes and expected power for many outcome measures for preclinical Alzheimer's disease research [27], and we present further calculations in this paper. Certainly, MPTP and 6-OHDA models will develop central dopaminergic cell death and associated motor disorders, but our data reiterate that this is not sufficient to model the disease.…”

“…Spontaneous activity in a novel environment, over a period of 1 h, was analysed as per [27] at baseline and at 5, 10, and 15 weeks post-surgery.…”

“…One section per mouse, approximately AP + 0.4, was stained for cresyl violet to confirm microinjection location, as per [27]. Photomicrographs were taken using a stereomicroscope (Zeiss Axio imager, Oberkochen, Germany) at 17× magnification to confirm appropriate needle placement.…”

“…For immunofluorescence, staining protocols were as per [27]. Control sections that were not exposed to primary antibodies were always run in parallel, and no specific staining was observed in these sections.…”

Effect Sizes of Cognitive and Locomotive Behavior Tests in the 5XFAD-J Mouse Model of Alzheimer’s Disease

“…Moreover, the sample size varies according to the sound-cue interval in the behavioral analysis of Contextual and Trace Fear Conditioning in the P301S tau mouse model [17]. A recent study suggesting sample and effect sizes for behavioral and pathological characteristics with 4-to 6-month-old 5XFAD-SJL AD model mice showed no alteration in cognitive impairment by the NOR and Y-maze tests [18]. In addition, the MWM behavior test showed no alteration in spatial learning between training and probe days, while cognitive function was decreased during the reverse learning task in the 5XFAD group [18].…”