Predictors of renal and patient outcomes in chronic kidney disease.
MD thesis, University of Glasgow.
Full text available as:
Chronic kidney disease (CKD) is associated with an increased risk of cardiovascular disease and end stage renal failure. Accurate identification of those with a reduced glomerular filtration rate and significant proteinuria facilitates early diagnosis and risk stratification.
This thesis explores the optimal measure of proteinuria, to accurately quantify proteinuria and as a predictor of renal and patient outcomes. We examine the prevalence of CKD in a general population cohort and assess the impact of different estimated glomerular filtration rate (eGFR) formulae. We explore the prognostic role of reduced eGFR and proteinuria in patients with hypertension and present the baseline characteristics of a community cohort study of patients with predominantly early CKD. They will be followed for ten years to identify predictors of cardiovascular and renal outcome.
Urine total protein:creatinine ratio (TPCR) and albumin:creatinine ratio (ACR) have largely replaced 24-hour urine collections for proteinuria quantification. The performance of these spot measures to identify significant proteinuria is compared in a cohort of 6842 patients attending a general nephrology clinic. Both tests perform well overall but TPCR is statistically significantly superior as a predictor of 24-hour total proteinuria than ACR (as measured by the area under the receiver operator characteristic (ROC) curve to predict 1g/day total proteinuria). On sub-group analysis the performance of the spot samples is poorer in women and the elderly, likely as a result of low muscle mass and low urine creatinine (the denominator in TPCR/ACR).
The performance of TPCR and ACR were then compared as predictors of outcome in a similar cohort of 5586 CKD patients using a hierarchical Cox survival model. TPCR and ACR both performed well as independent predictors of death, commencement of renal replacement therapy (RRT) and doubling of serum creatinine. Notably TPCR performed well at low levels where albuminuria has been considered superior. These findings are novel. The spot samples performed as well as 24-hour collections in the sub-group with timed urine collections.
The National Institute for Health and Clinical Excellence in England recommend ACR to monitor all patients with CKD; the Scottish Intercollegiate Guidelines Network recommend TPCR for non-diabetic renal disease. Therefore, we investigated the implications of these recommendations using survival modelling. The same cohort was divided into 5 groups: no proteinuria, low proteinuria (using TPCR and ACR), high proteinuria (TPCR and ACR) and two groups where TPCR and ACR were discordant (i.e. TPCR above the diagnostic threshold but ACR below it and vice versa) using the recommended thresholds (ACR 30mg/mmol/TPCR 50mg/mmol to predict 0.5g/day total proteinuria and ACR 70mg/mmol/TPCR 100mg/mmol to predict 1g/day total proteinuria). Using univariate survival analysis the discordant group had significantly poorer outcomes (using the same outcomes as previously) than those with significant proteinuria as measured by both tests. The discordant group was older with poorer renal function and some of the excess risk was abolished on multivariate analysis, however the risk did not return to the level of those without detectable proteinuria. TPCR, but not ACR, measures non-albumin proteins and these may have pathophysiological roles in progression. This requires further study. However this analysis confirmed that TPCR identifies patients at high risk of adverse outcomes.
TPCR and ACR may vary as a result of muscle mass. We adjusted TPCR and ACR for estimated creatinine excretion (ECE) (calculated using the Cockcroft and Gault formula) and performed cross-sectional and longitudinal analyses. Adjusting TPCR and ACR for ECE improves prediction of significant proteinuria in sub-groups with poor baseline test performance (such as women and the elderly) using ROC curve analysis. However when adjusted and unadjusted values were compared as predictors of outcome (using a net reclassification index analysis) adjusted values were significantly inferior. Urine creatinine is an independent predictor of mortality and hence may be directly contributing to the predictive value of TPCR and ACR rather than simply correcting for urine flow rate. As such, adjusting for ECE may act to remove the effect of a second independent predictor, leading to inferior test performance. Therefore the decision to adjust TPCR and ACR for ECE depends on the test application: to predict significant proteinuria adjustment of TPCR and ACR is of benefit, but adjustment leads to inferior performance as a prognostic test.
The prevalence of CKD stages 3-5 was assessed using a general population laboratory database. Overall population prevalence was 5.63% using the modification of diet in renal disease (MDRD) formula and fell to 4.94% when the CKD-Epidemiology group (CKD-EPI) formulae were applied. Those reclassified to an earlier stage of CKD were predominantly middle aged women. Prevalence over a five year period was found to be stable using the CKD-EPI formulae but rose slightly according to MDRD.
Proteinuria and eGFR were assessed as predictors of outcome in a large specialist hypertension clinic cohort. On multivariate survival analysis both baseline dipstick proteinuria and an eGFR<60ml/min/1.73m2 remained strong independent predictors of cardiovascular and all-cause mortality, despite intensive specialist intervention to control blood pressure. These simple tests should be advocated for risk stratification in these patients.
Lastly the baseline characteristics of a community CKD cohort are presented. We recruited 411 participants from seven general practices around Ayrshire and a detailed baseline clinical and biochemical assessment was performed. Patients were invited to participate if they were included in the primary care register of CKD stages 3-5. Over a quarter had an eGFR>60ml/min/1.73m2 on the meat-fasted study sample. Proteinuria was of notably low prevalence and the cohort had a large burden of cardiovascular disease. Complications of renal disease were uncommon. The characteristics of the cohort differ from those under hospital follow-up. Their long term outcomes should contribute to refining risk stratification in this population.
Proteinuria and eGFR are key aspects of diagnosis and monitoring in CKD. Identification of the optimal measures of both is essential and findings presented here contribute to that. There is a need to refine risk stratification in CKD, to identify those who require intensive intervention, and to reassure the rest. The findings of this thesis also contribute to that. Further study is required to refine the core aspects of diagnosis and investigation of CKD.
Actions (login required)