All patients will be extensively phenotyped using identical, comprehensive protocols. These protocols comprise an extensive MRI protocol providing information on the structure and function of heart, aorta, cerebropetal arteries, cerebral microvasculature, and brain, a recently developed and validated neuropsychological protocol, and a comprehensive set of relevant biomarkers. The acquired data will be stored and analyzed centrally and algorithms will be defined for identifying hemodynamic parameters, other biomarkers, and combinations thereof that are associated with cognitive impairment and that predict development of cognitive decline.
The primary aims in each of the three clinical populations are identical: 1) to assess the relationship between cardiovascular function/structure and cerebral perfusion and 2) to determine the relation between cerebral perfusion and cognition. Therefore, in each of the three populations we will assess heart function, vascular structure and function, cerebral perfusion, brain MRI abnormalities, and cognition in a standardized fashion. Addressing the role of co-morbidities and medication use will be a key target in both the clinical work-up and the analyses. Uniform diagnostic criteria for cognitive impairment, HF, and COD will be used in all three populations to be able to compare data across these populations. The presence of cognitive impairment will be based on performance on a comprehensive neuropsychological test battery.
For the diagnosis of HF, standard MRI of the heart will be used in addition to assessment of signs and symptoms to objectively determine structure and function of both the left and the right ventricle; N-terminal prohormone brain natriuretic peptide (NT-proBNP) levels will be used as suggested by the guidelines ; heart rate and blood pressure will be assessed; and a standard 12-lead electrocardiogram will be obtained in all patients. These records will be assessed by an expert committee of HF cardiologists, who will decide whether a patient has HF.
The presence of COD will be based on an MR-angiography study. With a sample size of 175 patients in each patient cohort we will be able to detect associations in which the determinant explains 5% or more of the variance in the dependent variable (i.e., the equivalent of a correlation coefficient of 0.2 or more with alpha 0.05, power 90%), taking up to 10 to 20 relevant co-variates (such as confounders, medication, and co-morbidities) into account. Assuming that the dropout rate will not exceed 25% over two years, we will have 80% power to detect associations of the same strength at follow-up. We argue that variables that explain less than 5% in our primary dependent variables (i.e., cerebral perfusion and cognition) should not be regarded as key determinants and that therefore the selected sample size is sufficient to detect clinically significant associations.