Introduction

Rapid diagnosis is required for the effective treatment of malaria. Currently, microscopic examination of thin or thick blood smears is widely used, but this method is time-consuming and requires an experienced microscopist for success. According to Moody [1], the mean prevalence of parasitemia among patients returning from endemic areas is between 0.1% and 0.001%. During the last decade, rapid diagnostic tests based on the detection of histidine-rich protein II (HRP-2), a soluble Plasmodium antigen, have been marketed [2, 3]. They are easy to use and self-contained, and the results are visible to the naked eye. New tests, such as NOW ICT MALARIA P.f./P.v. (Binax, Portland, ME, USA), which combines the detection of HRP-2 and aldolase [4], and OptiMAL IT (Diamed, Cressier, Switzerland), which is based on the detection of parasite-specific lactate dehydrogenase [5], have recently been developed. However, cross-reactions with rheumatoid factor [6] and other false-positive reactions of various origins, in addition to lack of sensitivity and cost considerations, have precluded widespread use of these rapid tests [3, 5]. In order to compare the diagnostic performance of these two rapid tests in a non-endemic setting, we conducted a prospective study of non-selected patients symptomatic for malaria seen in three centres in the Lyon area of France.

Patients and methods

The study was conducted from March to August 2002, and all subjects with symptoms of malaria were included. Information was collected on travel to malaria-endemic areas, chemoprophylaxis, recent episodes of fever, presenting signs and symptoms and treatment. In keeping with French Ministry of Health guidelines, outpatients were controlled on day 3 and day 7 after the onset of treatment. Standard microscopic examination of air-dried, methanol-fixed thin blood films stained with 10% buffered Giemsa was the reference method. Slides were examined with an oil-immersion lens at 1,000× magnification, for at least 20 min, by two expert microscopists in each centre. Parasite density was determined per 104 erythrocytes.

The NOW ICT test detects HRP-2, which is produced by Plasmodium falciparum trophozoites and young gametocytes, and aldolase, an enzyme of the glycolytic pathway of the four Plasmodium species, which is used as a panmalaria antigen. The appearance of both lines in the test points to infection by P. falciparum, either alone or in association with another species.

The Optimal IT test detects lactate dehydrogenase, an enzyme produced by sexual and asexual forms of Plasmodium species. The test is a new configuration of OptiMAL presented in a sealed cassette format as an individual test [5]. According to the manufacturer, the test can detect a parasite density as low as 100–200 per microliter of blood.

The two rapid tests were performed in a blinded manner and according to the manufacturer’s instructions by separate technicians at each centre after a short training period. If the results were borderline (i.e. faint or ambiguous lines), the technician was allowed to consult a third party for a consensual decision. Venous blood was collected in EDTA-coated tubes (Becton Dickinson, Franklin Lakes, NJ, USA), and the tests, which had been stored at room temperature, were performed either immediately or within 8 h, with samples being stored at 4°C.

Real-time PCR using LightCycler technology (Roche Molecular Biochemicals, Mannheim, Germany) was applied to specimens for which the two tests disagreed. The sequences used to identify the four species of Plasmodium that affect humans have been reported elsewhere [7]. Plasma from samples giving discordant results were assayed using high-pressure liquid chromatography (HPLC) for pyrimethamine, sulfadoxine, chloroquine, quinine, halofantrine and doxycycline.

All data were recorded using Excel 2000 software (Microsoft Corp., Redmond, WA, USA). Since the kinetics of target antigenemia differ between the two tests [1], the sensitivity, specificity, positive and negative predictive values of each test, and the likelihood ratios for positive and negative results were calculated using VassartStats software (http://faculty.vassar.edu/lowry/clin1.html) on patient data collected at the time of initial sampling. The MacNemar test was used to compare the sensitivity and specificity of the two rapid detection tests.

Results and discussion

A total of 556 patients were enrolled in the study. Since positive cases were followed up and some negative patients were rechecked for malaria, a total of 685 assays were performed with both tests. The mean age of the patients was 34 years (range, 1 month to 90 years), and 57% of the patients were female. Three-quarters of the patients had visited Africa (mainly sub-Saharan Africa), 10% Asia and 6% South America. The rest of the patients had either never left Europe (7%) or provided no information on travel to malaria-endemic areas (2%). One hundred nine (19.6%) patients were diagnosed with malaria.

Table 1 shows the different species identified by thin film microscopy and the results obtained by the two commercial assays. All three methods were used on the first blood sample obtained from each patient. Table 2 shows the comparative performance of the two commercial assays. NOW ICT had a sensitivity of 96.3%, whereas OptiMAL IT had a significantly lower sensitivity of 79.8% (chi-square, 13.7; P-value, 0.0001). NOW ICT initially gave negative results for two patients infected with P. falciparum, but in one case the test became positive the following day (parasitemia <0.01%) thus ruling out the possibility of a mutation abrogating HRP-2 production [1]. Secondarily positive results have been reported previously for rapid detection tests based on HRP-2 detection [8]. With OptiMAL IT, ten false-negative results were recorded and nine of these patients had P. falciparum parasitemia above 0.01%. All ten patients reported they had not taken antimalarial drugs, and drug assays were negative for eight. OptiMAL IT was less efficient than NOW ICT for detecting sexual forms of P. falciparum and mixed infections (Table 1).

Table 1 Comparison of thin film microscopy with two rapid commercial assays for detecting 109 documented cases of malaria. PCR was performed only on samples with false-negative results in one of the two rapid tests
Full size table
Table 2 Diagnostic performance of two rapid detection assays for testing 556 subjects with suspected malaria
Full size table

Five false-positive results were produced by NOW ICT. The corresponding patients had all returned from endemic areas, had a history of fever during the 3 weeks before admission, and had received antimalarial drugs. In three of these false-positive cases, PCR resulted positive for P. falciparum. In all cases antimalarial drug assays were negative. It is probable that these false-positive cases were due to the persistence of HRP-2, as has been reported previously [1, 3]. In contrast, OptiMAL IT gave seven false-positive results. In one case, PCR demonstrated the presence of P. vivax DNA. Five days before admission, the patient had returned from Indonesia and had started self-administered quinine therapy. In the other six cases, PCR and antimalarial assays were negative. Three of the patients had hepatic or infectious diseases. PCR negativity suggests that these false-positive results were due to interfering factors such as heterophilic antibodies [5].

Thirteen patients with P. falciparum infection were followed for at least 7 days after the onset of treatment. In 12 cases OptiMAL IT turned negative before NOW ICT, i.e. after an average of 3 days. NOW ICT remained positive longer than the reference method and was still positive on day 7 in ten patients. As parasitemia declined, the panmalaria antigen band faded and turned negative before the HRP-2 band. Our data support the view that HRP-2 leaves a trail of circulating antigens [9], while lactate dehydrogenase reflects the presence of viable parasites [10] and can be used to monitor efficacy of treatment [4, 11]. Both tests had a positive likelihood ratio greater than ten (Table 2). Thus, in a non-endemic area where the pretest probability of contracting malaria is generally low, a positive rapid test result implies a high probability of having the disease. A negative NOW ICT test result (negative likelihood ratio, 0.03) tends to rule out malaria, whereas a negative OptiMAL IT test result (negative likelihood ratio, 0.2) offers a lower degree of confidence. With regard to P. falciparum infections, our data are in keeping with previously reported results from studies conducted using the former versions of the two tests [11, 12]. For the diagnosis of P. vivax, our data support the results previously reported for NOW ICT [4] and OptiMAL IT [13]. In Australia, Playford and Walker [12] found that the former version of NOW ICT (ICT MALARIA P.f./P.v.) had a sensitivity of only 44% and did not detect cases of P. vivax parasitemia above 0.1%. One possible explanation could be a genotypic difference between P. vivax strains, since ten of our patients were infected in French Guyana.

In conclusion, the two rapid tests we evaluated should not be used as first-line methods for the diagnosis of malaria in returning travellers when microscopic examination of thick or thin blood films is an available alternative. However, they could be useful in settings where microscopists lack experience in the diagnosis of malaria.