Copeptin does not accurately predict disease severity in imported malaria

On admission, blood samples were taken for analysis of the red blood cell count, haematocrit, white blood cell count, platelet count, CRP, serum electrolytes, total bilirubin, serum creatinin, liver enzymes, and blood glucose. A separate blood tube was drawn on admission for determination of plasma lactate, which was immediately analysed after isolation of plasma. Malaria was diagnosed by QBC (Quantitative Buffy Coat) analysis, by a rapid diagnostic antigen test for malaria (Binax NOW^® Malaria Test, Binax Inc. Maine, USA) and by conventional microscopy of stained thick and thin blood smears to identify the causative Plasmodium species. In case of Plasmodium falciparum infections, parasite density was determined. When the parasitaemia was less than 0.5% infected erythrocytes, parasites were counted per 100 leucocytes in thick smears. When the parasitaemia was equal or higher than 0.5% infected erythrocytes, infected erythrocytes were counted in thin smear and expressed as a percentage of the total erythrocytes. The number of parasites per microliter was subsequently calculated from these data. Patients were classified as having severe malaria if they met one or more of the WHO criteria for severe malaria, as published [14].

Procalcitonin levels were determined in serum samples using a commercially available EIA test (VIDAS BRAHMS Procalcitonin, bioMérieux, Lyon, France). Normal serum values for procalcitonin are < 0.1 ng/ml. Copeptin was measured with a commercial sandwich immunoluminometric assay (Brahms Copeptin, Thermo Fisher Scientific, Hennigsdorf/Berlin, Germany). Normal values for serum copeptin are 1.70-11.25 pmol/L [2].

Differences between patients with severe malaria, non-severe P. falciparum malaria and non-severe non-falciparum malaria were analysed with the Kruskall-Wallis test followed by Dunn's post-hoc tests. Two group comparisons were done by Student's unpaired t-test, Student's unpaired t-test with Welch correction or with the non-parametric Mann Whitney test depending on the distribution of the data. The diagnostic performance of each biomarker was reported as sensitivity, specificity, positive and negative predictive value for severe malaria and their corresponding 95% confidence intervals. Youden's index J (J = sensitivity + specificity-1) was used to identify the most appropriate cut-off point for each biomarker. Of each test a Receiver Operating Characteristics (ROC) curve, a graphical plot of sensitivity (true positive rate) versus 1-specificity (false positive rate), was constructed as a summary statistic and the area under the ROC curve (AUROC) and its corresponding 95% confidence intervals were calculated. The AUROC of each of the biomarkers was compared to that of copeptin in a pair-wise comparison with the method of DeLong et al. [15].

In total, 204 travellers with imported malaria were included in this study; 63 patients were diagnosed with a non-falciparum infection (45 with Plasmodium vivax, 15 with Plasmodium ovale, three with Plasmodium malariae) and 141 patients were diagnosed with P. falciparum infection. Of the patients with P. falciparum infections, 25 (17.7%) patients had severe malaria. There were no cases of severe non-falciparum malaria. The infection was most frequently acquired in Africa (n = 156), followed by Asia (n = 27) and South America (n = 14). Thirty (14.7%) patients reported adequate use of malaria chemoprophylaxis, 24 (11.8%) used it inadequately and 143 (70.1%) patients did not use any chemoprophylaxis (data on prophylaxis use were not available in seven patients). The general characteristics of all patients are shown in Table 1.

Twenty-one of 25 patients (84.0%) with severe malaria acquired a P. falciparum infection in West-Africa. The others acquired severe malaria in East-Africa (n = 2), South-East Asia (n = 1) and South-America (n = 1). Twenty-four patients with severe malaria did not use any form of malaria chemoprophylaxis and one patient was using it inadequately. At admission to the intensive care unit (ICU), the 25 patients with severe malaria all fulfilled one or more of the severity criteria (Glasgow Coma Score ≤ 11, n = 2; haematocrit < 0.20 L/L n = 5; creatinine > 250 μmol/L n = 2; bilirubin > 50 μmol/L n = 12; lactate > 5.0 mmol/L n = 5; hyperparasitaemia > 5% n = 14; schizontaemia n = 14). The first arterial blood gas analysis on the ICU showed a median bicarbonate level of 21 mmol/L (range 9-25 mmol/L) and a median base excess of -2 mmol/L (range -18-3). The median Glasgow Coma Scale score was 15 (range 11-15). Two patients required mechanical ventilation. Two patients were referred to another hospital for renal replacement therapy because of persistent acute oliguric renal insufficiency. Thirteen patients received exchange transfusion as an adjunct therapy to parenteral anti-malarial treatment. One case fatality was observed. The laboratory results of travellers with severe malaria were further characterized by significantly lower platelet counts and haemoglobin levels and by significantly higher plasma lactate, bilirubin and CRP levels compared to patients with non-severe P. falciparum malaria (Table 1).

In order to evaluate the diagnostic accuracy of copeptin for severe malaria, the performance of copeptin was compared to that of sodium, CRP, lactate and procalcitonin. Assuming that AUROCs are a measure of diagnostic accuracy, the performance of copeptin for severe malaria was comparable to that of sodium, lactate and procalcitonin, as is shown in Figure 1. The AUROC of copeptin did not statistically differ from the AUROC of sodium, lactate and procalcitonin in a pair-wise comparison (Table 2). As shown in Figure 1 and Table 2, CRP had a significantly better performance than either copeptin, sodium and lactate for prediction of severe malaria but its usefulness in clinical practice is probably limited. For illustration, when a hypothetical decision rule for severe malaria was created solely based on a CRP cut-off level of ≥ 155 mg/L, 10 of 62 (or 16.1%) evaluable patients with non-falciparum disease and 26 of 116 (or 22.4%) evaluable patients with uncomplicated P. falciparum would have been falsely diagnosed as having severe malaria. In contrast, the diagnosis severe malaria would not have been considered in 5 of 24 (20.8%) evaluable patients with severe disease. Addition of either lactate, sodium or copeptin as a secondary selection criterion to the prediction rule did not result in a proper identification of all patients with severe disease on admission. Only after adding procalcitonin (at a cut-off of ≥ 0.9 ng/mL) as a secondary selection criterion to the prediction rule, all evaluable patients with severe malaria would have been correctly identified on admission (Additional file 1: Figure S1). However, still nine of 29 evaluable patients with non-severe P. falciparum malaria and 10 of 15 evaluable patients with non-severe, non-falciparum malaria would have been falsely diagnosed with severe malaria, respectively (Additional file 1: Figure S1).

When focusing on the 25 patients with severe malaria, 11 patients fulfilled a single criterion for severe malaria on admission, 11 patients fulfilled two malaria severity criteria, 2 patients fulfilled 3 malaria severity criteria and one patient fulfilled 4 malaria severity criteria on admission to the ICU, respectively. As a consequence of this distribution a proper stratification by numbers of malaria severity criteria present on admission was not feasible. When the severe malaria patients fulfilling a single malaria severity criterion on admission were compared to the patients with more than one criterion for severe malaria on admission, only for the biomarker plasma lactate a statistically significant difference was found (1.8 ± 0.6 versus 3.3 ± 1.6 mmol/L, p = 0.0042). See Additional file 2: Table S1 for further details.