Causality on longitudinal data: Stable specification search in constrained structural equation modeling.

A typical problem in causal modeling is the instability of model structure learning, i.e., small changes in finite data can result in completely different optimal models. The present work introduces a novel causal modeling algorithm for longitudinal data, that is robust for finite samples based on recent advances in stability selection using subsampling and selection algorithms. Our approach uses exploratory search but allows incorporation of prior knowledge, e.g., the absence of a particular causal relationship between two specific variables. We represent causal relationships using structural equation models. Models are scored along two objectives: the model fit and the model complexity. Since both objectives are often conflicting, we apply a multi-objective evolutionary algorithm to search for Pareto optimal models. To handle the instability of small finite data samples, we repeatedly subsample the data and select those substructures (from the optimal models) that are both stable and parsimonious. These substructures can be visualized through a causal graph. Our more exploratory approach achieves at least comparable performance as, but often a significant improvement over state-of-the-art alternative approaches on a simulated data set with a known ground truth. We also present the results of our method on three real-world longitudinal data sets on chronic fatigue syndrome, Alzheimer disease, and chronic kidney disease. The findings obtained with our approach are generally in line with results from more hypothesis-driven analyses in earlier studies and suggest some novel relationships that deserve further research.

Abdominal aortic calcification in patients with CKD.

BACKGROUND: Abdominal aortic calcification (AAC) is independently associated with cardiovascular events in dialysis patients and in the general population. However, data in non-dialysis chronic kidney disease (CKD) patients are limited. We analyzed determinants and prognostic value of AAC in non-dialysis CKD patients. METHODS: We included patients with CKD not receiving renal replacement therapy from the MASTERPLAN study, a randomized controlled trial that started in 2004. In the period 2008-2009, an X-ray to evaluate AAC was performed in a subgroup of patients. We studied AAC using a semi-quantitative scoring system by lateral lumbar X-ray. We used baseline and 2-year data to find determinants of AAC. We used a composite cardiovascular endpoint and propensity score matching to evaluate the prognostic value of AAC. RESULTS: In 280 patients an X-ray was performed. In 79 patients (28 %) the X-ray showed no calcification, in 62 patients (22 %) calcification was minor (<4), while 139 patients (50 %) had moderate or heavy calcification (≥4). Older age, prior cardiovascular disease, higher triglyceride levels, and higher phosphate levels were independent determinants of a calcification score ≥4. AAC score ≥4 was independently associated with cardiovascular events, with a hazard ratio of 5.5 (95 % confidence interval 1.2-24.8). CONCLUSIONS: Assessment of AAC can identify CKD patients at higher cardiovascular risk, and may provide important information for personalized treatment. Whether this approach will ultimately translate into better outcomes remains to be answered.

Kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin as prognostic markers in idiopathic membranous nephropathy.

BACKGROUND: Urinary excretion of alpha-1-microglobulin and beta-2-microglobulin reflects tubular damage and predicts outcome in patients with idiopathic membranous nephropathy with reasonable accuracy. Urinary kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin are novel biomarkers of tubular damage. We investigated if these markers could improve prediction of outcome in idiopathic membranous nephropathy. METHODS: We measured kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin in urine samples from patients with idiopathic membranous nephropathy, who had nephrotic proteinuria and normal renal function. Excretion of alpha-1-microglobulin and beta-2-microglobulin had been measured previously. Progression was defined as a serum creatinine rise >30%, a rise in serum creatinine to an absolute value of ≥135 µmol/L, or a clinical decision to start immunosuppressive therapy. Remission was defined as proteinuria <3.5 g/day and >50% reduction from baseline. RESULTS: Sixty-nine patients were included. Median follow-up was 35 months (interquartile range 18-63 months). Progression occurred in 30 patients (44%), and spontaneous remission in 36 (52%). Kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin excretion rates were significantly correlated with each other, and with alpha-1-microglobulin and beta-2-microglobulin. The areas under the receiver operating characteristic curves for progression were 0.75 (0.62-0.87) for kidney injury molecule-1 and 0.74 (0.62-0.87) for neutrophil gelatinase-associated lipocalin. In multivariate analysis with either alpha-1-microglobulin and beta-2-microglobulin, kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin did not independently predict outcome. CONCLUSION: Kidney injury molecule-1 and neutrophil gelatinase-associated lipocalin excretion rates correlated with excretion rates of other tubular damage markers and predicted outcome in patients with idiopathic membranous nephropathy. They did not add prognostic value compared to measurement of either alpha-1-microglobulin or beta-2-microglobulin.