End stage renal failure is associated with a massively increased risk of cardiovascular
disease, and evidence suggests that this increase in risk begins early in the development
of chronic kidney disease. This thesis considers the hypothesis that small reductions in
glomerular filtration rate (GFR), across the population range, are associated with a
clustering of cardiovascular risk factors and increased risk of cardiovascular disease and
death. Existing methods to accurately measure GFR are difficult to perform and
unsuitable for large studies, while GFR estimated from blood creatinine or cystatin c
concentration is inaccurate.
The performance of a new method of assessing GFR, in which total plasma iohexol
clearance is measured using dried capillary blood spots, was examined in a crosssectional study of 81 consecutive individuals undergoing routine measurement of GFR.
The new blood spot iohexol clearance (BSIC) method (using 3 blood spot samples)
assessed GFR accurately compared to traditional iohexol clearance using 3 timed plasma
samples (mean±standard deviation [BSIC - reference method]: 1.1±7.7 ml/min/1.73m");
prediction equations to estimate GFR from blood creatinine and cystatin c concentration
performed poorly. The results were similar when 2 blood spots (2 and 4 hours) were
used, but using only a single 4 hour blood spot resulted in some loss of accuracy.
The feasibility of using the BSIC method when the blood spot sampling is completed by
participants at home was assessed in a cross-sectional study of 111 individuals.
Following a short training, 100% of participants completed satisfactory baseline samples,
97% returned 2 timed samples through the post and 90% found the procedure acceptable.
However, 21% returned small or poor quality blood spots and there was statistical
evidence ofrounding-up of the recorded sampling time. Among 106 participants with
measurements of BSIC-GFR, GFR estimated from blood creatinine, and cystatin c
concentration, one or more measures of GFR were positively correlated with blood high
density lipoprotein cholesterol concentration, and were inversely correlated with blood
concentrations of triglyceride, C-reactive protein, fibrinogen, and homocysteine. BSICGFR was not more strongly related to cardiovascular risk factors than GFR estimated
from blood creatinine concentration (eGFR). GFR estimated from plasma cystatin c
concentration was strongly related to measures of body fat, but no relationship was seen
with the other GFR measures.
In a meta-analysis of cohort studies assessing the relationship between eGFR and risk of
death and cardiovascular events, which included 4 061 003 and 1 372 820 individuals
for each outcome respectively, a 30% lower eGFR was associated with a 20-30%
increase in risk of each outcome, depending on the type of study examined. However,
there was significant heterogeneity between the studies. These overall results may
underestimate the impact of lower eGFR among those with chronic kidney disease since,
in two large studies together contributing 85% of the deaths, lower eGFR was not
associated with increased risk of death until eGFR was below 60 ml/min/1.73m2. In one
large study contributing 218 000 deaths, the relative risk of death associated with lower
eGFR was larger at younger ages, but eGFR had greater absolute impact at older ages
and among those with prior vascular disease. Because eGFR is only weakly related to
measured GFR among healthy individuals, these results may underestimate the
relationship between 'true' GFR and risk of death and vascular disease.
This work demonstrates an association between measures of GFR and cardiovascular
risk factors, and increased risk of vascular disease, although the methods used to assess
GFR may have introduced bias. Large scale studies in which GFR is accurately
measured are needed. Using the new BSIC method for this purpose is potentially
feasible, but further work is required to ensure accuracy when the blood sampling is
completed by participants themselves.