Copeptin does not accurately predict disease severity in imported malaria

Abstract

Background: Copeptin has recently been identified to be a stable surrogate marker for the unstable hormone arginine vasopressin (AVP). Copeptin has been shown to correlate with disease severity in leptospirosis and bacterial sepsis. Hyponatraemia is common in severe imported malaria and dysregulation of AVP release has been hypothesized as an underlying pathophysiological mechanism. The aim of the present study was to evaluate the performance of copeptin as a predictor of disease severity in imported malaria.

Methods: Copeptin was measured in stored serum samples of 204 patients with imported malaria that were admitted to our Institute for Tropical Diseases in Rotterdam in the period 1999-2010. The occurrence of WHO defined severe malaria was the primary end-point. The diagnostic performance of copeptin was compared to that of previously evaluated biomarkers C-reactive protein, procalcitonin, lactate and sodium.

Results: Of the 204 patients (141 Plasmodium falciparum, 63 non-falciparum infection), 25 had severe malaria. The Area Under the ROC curve of copeptin for severe disease (0.66 [95% confidence interval 0.59-0.72]) was comparable to that of lactate, sodium and procalcitonin. C-reactive protein (0.84 [95% CI 0.79-0.89]) had a significantly better performance as a biomarker for severe malaria than the other biomarkers.

Conclusions: C-reactive protein but not copeptin was found to be an accurate predictor for disease severity in imported malaria. The applicability of copeptin as a marker for severe malaria in clinical practice is limited to exclusion of severe malaria.

Figure 1

Receiver Operating Curves (ROC) characteristics of the diagnostic performance of copeptin for severe P. falciparum malaria, compared to that of sodium, lactate, CRP and procalcitonin. The ROC curve is a graph of sensitivity (true positive fraction) plotted against 1-specificity (false positive fraction). The performance of a diagnostic variable can be quantified by calculating the area under the ROC curve (AUROC). The ideal test would have an AUROC of 1, whereas a random guess would have an AUROC of 0.5. The results of the pair-wise comparison of ROC curves are shown (with copeptin ROC curve as comparator). a. copeptin vs sodium (n = 204 pairs, p-value not significant); b. copeptin vs lactate (n = 124 pairs, p-value not significant); c. copeptin vs CRP (n = 202 pairs, p = 0.02); d. copeptin vs procalcitonin (n = 70 pairs, p-value not significant). The copeptin graphs are not all identical due to missing values in pair-wise comparisons.