Generic Plasmodium forward (5'-TCAGATACCGTCGTAATCTTA-3') and reverse (5'-TTCGCGCAAGCAGAAAGTT-3') primers were used to amplify a 180 bp region of P. vivax, P. ovale and P. malariae 18S rRNA gene by PCR as previously described 10. The amplified fragments were cloned into plasmid pCR2 in Escherichia coli INValphaF (Invitrogen, Carlsbad, California USA). Next, in vitro RNA was produced with the transcription kit SP6/T7 (Roche, Mannheim, Germany) and the cloned fragment was sequenced (Base Clear, Leiden, The Netherlands). Production of P. falciparum 18S rRNA based in vitro RNA was performed previously 10. Primers were selected on the basis of the DNA sequences of the three species (Figure 1). Homology of these sequences with published sequences and specificity of primers was analysed with the BLAST database for homology search 15. The NASBA forward primers were chosen on their specificity for the targeted species. P. falciparum: 5'GTCATCTTTCGACGTGACTT-3'; P. vivax: 5'-TTTCTCTTCGGAGTTTATTC-3'; P. ovale: 5'-CGACATTGTCATTCCATTTAC-3'; P. malariae: 5'-GAGTGTTTCTTTTAGATAGC-3'. A T7 promotor sequence was added to the generic Plasmodium NASBA reverse primer; 5'-AATTCTAATACGACTCACTATAGGGAGAAGGAACTTTCTCGCTTGCGCGAA-3'. The published pf18S molecular beacon 5'-6-carboxyfluorescein-CGATCGGAGAAATCAAAGTCTTTGGGCGATCG-dimethylaminoazosulfonic acid-3' was used as a detection probe 9.

The molecular work was performed under permit 02–080 granted on 22 February 2002 to KIT Biomedical Research by the Netherlands Ministry for Spatial Planning, Housing and the Environment.

Real-time QT-NASBA for 18S rRNA of P. falciparum, P. vivax, P. malariae and P. ovale was performed on an IQ5 Real-Time analyser (Bio-RAD). The reactions were performed with the Nuclisens Basic kit for amplification (BioMerieux) according to the manufacturers instructions with a KCl concentration of 80 mM for P. falciparum, P. vivax and P. malariae and 70 mM for P. ovale. The reaction mixture (5 μl) containing the primers (100 pmol/μl) molecular beacon (20 μM) and template RNA (2.5 μl) was incubated at 65°C for two minutes followed by two minutes at 41°C. Thereafter, 2.5 μl enzyme mixture from the basic kit was added to each reaction. Amplification was monitored for 60 minutes after which the results were analysed. A sample containing only water and reaction mixture was used as blank and served as control for background fluorescence. The signal produced by the blank samples is automatically subtracted from the analytical samples (Bio-RAD IQ5 software v. 1.0). In order to quantify the number of parasites in a clinical sample, a 10 fold serial dilution of 10⁹ to 10² molecules of in vitro RNA per amplification reaction of each respective Plasmodium species was run in triplicate in each test, wherein 10⁴ molecules corresponds to one Plasmodium parasite 10.

Blood samples for validation (n = 79) were obtained from patients diagnosed with malaria by detection and differentiation of Plasmodium parasites in Giemsa-stained slides with standard microscopy according to WHO recommendations 11 or PCR methods as established in the contributing medical centres. The samples were collected from returned travellers with clinical symptoms visiting the out patient clinic of the University Medical Centre Leiden (The Netherlands) and the Academic Medical Centre Amsterdam (The Netherlands) and from ongoing field studies in Vietnam, Turkey and Kenya (Table 1). The samples were tested blinded from the diagnostic results. Specificity of the assays was tested against in vitro RNA of the four Plasmodium species and RNA isolated from Plasmodium berghei isolates (12 samples). Furthermore 20 samples from Dutch healthy blood donors were tested as negative controls. Blood samples (50 μl) were mixed with 950 μl guanidium isothiocyanate lysis buffer and RNA was isolated as described previously 12.

Microscopy performed at initial diagnosis was considered as the golden standard for this purpose and all NASBA results were compared to these results. The agreement between microscopy and the real-time QT-NASBA assay was determined by calculating Kappa values with a 95% confidence interval (Altman, 1991). Kappa values express the agreement beyond chance and a kappa value of 0.21–0.60 is a moderate, a kappa value of 0.61–0.80 a good and kappa > 0.80 an almost perfect agreement beyond chance.

The lower detection limit of the developed assay as determined by analysing serial dilutions of in vitro RNA of P. vivax, P. ovale or P. malariae, respectively, was 100 – 1,000 molecules of in vitro RNA for each species, corresponding to 0.1 – 0.01 parasite per diagnostic sample (i.e. 50 μl of processed blood) (Figure 2, 3 and 4). No cross-hybridisation of the primers with the in vitro RNA of the other species and none with P. berghei RNA isolated from mice (n = 12) was observed. Blood samples form healthy donors (n = 20) showed no reaction in any of the assays.

The developed real-time QT-NASBA was further evaluated using 79 clinical samples (Table 2). Real-time QT-NASBA confirmed the initial diagnosis with microscopy of 69 out of the 79 clinical samples. The 37 P. vivax samples were found to be P. vivax with microscopy as well as with NASBA. However, one clinical sample gave an additional weak P. ovale signal with the real-time QT-NASBA. The four P. ovale samples were identified as such by real-time QT-NASBA. However, one sample gave a weak additional P. vivax signal. Two of the seven P. malariae mono-infections were found negative by NASBA. Furthermore, three of the seven P. malariae mono-infections were identified with real-time QT-NASBA as a P. falciparum/P. malariae mixed infection. Only 11 P. falciparum samples were analysed with real-time QT-NASBA in the present study, as a thorough analysis of this species has already been done 9. The microscopical diagnosis P. falciparum of the clinical samples tested in the present study was confirmed by NASBA. Of the 20 P. falciparum/P. malariae mixed infections, 19 were identified as a mixed infection with NASBA: 17 being a mixed infection of P. falciparum/P. malariae and two samples gave a weak P. vivax reaction instead of P. malariae. One sample was identified with real-time QT-NASBA as a P. falciparum mono-infection.

Of the 10 discordant results, seven samples were re-analysed by re-reading the blood slides by an expert microscopist who was blinded from the original microscopy and NASBA results. There were no back-up slides or PCR samples available from the other three discordant results. The microscopic re-analysis resolved five discordant results. In the three P. malariae mono-infections that were identified as a mixed infection by real-time QT-NASBA, a very low number of P. falciparum was found after re-reading the slides. In the two slides which were initially diagnosed by microscopy as being P. falciparum/P. malariae mixed infections, but with real-time QT-NASBA as being a P. falciparum/P. vivax mixed infection, the results of re-reading the blood slides were inconclusive for the presence of P. malariae and/or other Plasmodium species.

A high degree of agreement was observed between the real-time QT-NASBA assay and microscopy in the present study. A kappa value of 0.926 (95% CI 0.872–0.967) indicates a very good agreement beyond change.