Four candidate standards were included in this study. Sample AA was a freeze-dried preparation of blood, obtained from a patient infected with P. falciparum following an exchange transfusion. Local Research Ethical Committee approval was obtained at the London Hospital for Tropical Diseases for the collection and subsequent use of this sample. The blood used to prepare sample AA, was determined to have an overall parasitaemia of approximately 9.8% by light microscopy. Sample AA was stored at -70°C until lyophilization. The conditions for lyophilization were as previously described 1314. Lyophilized vials of AA have been stored at -20°C with constant temperature monitoring, at the National Institute for Biological Standards and Control (NIBSC). Sample BB was a liquid preparation, of in vitro cultured P. falciparum (3D7 strain) prepared in leucodepleted blood, with a parasitaemia of approximately 10%. Sample CC was a liquid preparation of approximately 6.9% parasitaemia of P. falciparum infected blood, obtained from the same patient as sample AA. Sample DD was a 1:1000 dilution of sample CC prepared in leucodepleted blood. Samples BB, CC and DD were dispensed in 0.5 ml volumes and stored as liquid/frozen preparations at -70°C. All samples tested negative by PCR for the following viral markers: HBV DNA, HCV RNA, HAV RNA, HIV-1 RNA and parvovirus B19 DNA.

Fourteen laboratories participated in the collaborative study. Laboratories were sent four vials of each of the candidate standards. The participants were requested to store samples at or below -70°C until analysis. The lyophilized preparation (AA) was reconstituted using 0.5 ml of nuclease-free deionized water immediately prior to analysis, with gentle agitation for 20 minutes to fully dissolve the contents. The liquid preparations (BB, CC and DD) were to be thawed quickly before use. Participants were requested to test the panel of candidate standards in four independent assays for P. falciparum DNA, using fresh vials of the four candidates for each assay run. In the case of qualitative assays, serial dilutions of the samples were analysed in four independent assays. In the first qualitative assay, ten-fold dilutions were performed to determine the end point for the detection of P. falciparum DNA. In each of the subsequent three assays, a minimum of two half-log₁₀ (i.e. 1:3.2) dilutions either side of the pre-determined end-point were assayed, and results reported as positive or negative. Participants were requested to prepare dilutions in the sample diluent normally used in their assay system. In the case of quantitative assays for P. falciparum DNA, results were used directly as returned by laboratories, after correction for any dilutions made.

The qualitative end-point assays were analyzed using the Poisson model, as previously described for collaborative studies for other NAT standards 1315. All estimates were expressed as log₁₀ PCR detectable units/ml. Overall means were calculated as arithmetic means of the log₁₀ estimates.

The lyophilized candidate standard (AA) was analysed for its stability over time in accelerated thermal degradation studies. Samples were incubated in temperature controlled environments, withdrawn at specified times and analysed for P. falciparum DNA content as described below. Following reconstitution, 200 μl volumes of sample for AA were extracted using the MagNA Pure LC instrument with software version 4.0 (Roche Applied Science, Mannheim, Germany). Samples were extracted using the DNA Isolation Kit I (Roche Applied Science, Mannheim, Germany) according to the manufacturer's instructions. Elution was performed with 100 μl of elution buffer. Real-time PCRs were performed on the LightCycler 2.0 instrument (Roche Applied Science, Mannheim, Germany). An in-house TaqMan assay was performed using primers selected from the most conserved regions of the 18S ribosomal RNA (rRNA) gene. The primer sequences were as follows: forward primer, 5' CAG ATG TCA GAG GTC AAA TTC TAA GAT T 3'; reverse primer, 5' TCC CTT AAC TTT CGT TCT TGA TTA ATG 3'. The sequence of the fluorogenic hydrolysis probe was as follows: 5' (FAM) CTG GAG ACG GAC TAC TGC GAA AGC ATT TG (TAMRA) 3'. Amplification reactions were performed using the LightCycler FastStart DNA Master Hybprobe kit (Roche Applied Science, Mannheim, Germany). The amplification conditions were as follows: 95°C for 10 min, then 45 cycles of the following sequential steps: 95 °C for 15s, 60°C for 1 min. Fluorescence data was collected during the combined annealing/extension step and detected at 530 nm. A standard curve was generated using serial ten-fold dilutions of a sample with a concentration of 180,000 parasites per μl, as determined by Giemsa-stained thin film microscopy by a proficient operator, with independent confirmation.

Fourteen laboratories from 10 countries, returned results from four assays for each of the four separate preparations. The participants included parasitology laboratories from tropical medicine institutes, universities, hospitals, quality control laboratories, manufacturers of diagnostic kits and other laboratories. The majority of assays used by the participating laboratories were ones developed in-house, although participants returned results from commercial assays. Throughout this study, a code number has been allocated at random for each laboratory and does not necessarily represent the order of the participants described in Table 1. The types of assays used by participants are listed in Table 2.

Participating laboratories were requested to test the samples in qualitative or quantitative format in order to determine the P. falciparum DNA content for each of the four candidate standards. The majority of laboratories provided data from qualitative end-point dilution assays. These laboratories reported the number of samples positive out of the number tested at various dilutions. These were treated as a dilution series and used to provide a single estimate of "PCR-detectable units/ml" in the undiluted sample using the method of maximum likelihood for "dilution assays" as described for the analysis of other NAT standards 1315. The model assumes that the probability of a positive result at a given dilution follows a Poisson distribution (with the mean given by the expected number of "copies" in the sample tested, and that a single copy will lead to a positive result). Calculations were carried out using the statistical package GLIM 1315.

Where a laboratory marked a result as uncertain or +/-, it was treated as negative. Where a laboratory performed more than one assay method, the results were analyzed separately, treated as if from separate laboratories, and coded as for example as laboratories 1A and 1B. Laboratory 1B did not continue the dilution series for sample AA until negative responses were observed. It was therefore not possible to calculate an estimate of units/ml for this sample (see below). For assays 2–4, Laboratory 9 only tested single log dilutions to the endpoint rather than half-log dilutions at the supposed end point. It was still possible to determine estimates of detectable units/ml, but they may be less precise than for other laboratories. Laboratory 10 (10A & 10B) had some positives at very high dilutions. Because the Poisson model could not be fitted with these results, they were excluded as potential false positives. Laboratory 11 (11A & 11B) reported "dilutions not well executed" for assay 2, which was therefore excluded. It was not possible to obtain an estimate of detectable units/ml for sample CC for 11A, as all results were positive. Some of the remaining results from 11B were also inconsistent. Laboratory 14B had some inconsistent negatives at 10^-1 and 10^-2 for sample DD in assay 2, which were excluded.

Two laboratories provided quantitative results. Laboratory 3B provided quantitative results using a commercial kit (Artus RealArt Malaria LC PCR kit, Qiagen, Hamburg, Germany) for a range of dilutions for each sample. The results were quoted as copies/μl. Laboratory 5, provided results from an in-house quantitative assay, giving results as a mean log₁₀ Par/ml based on two sets of triplicate determinations for each sample.

The estimated log₁₀ PCR-detectable units/ml for each of the four candidate standards from the qualitative end-point assays are shown in histogram form in Figure 1 and in Table 3. Each box represents the estimate from one laboratory, and is labeled with the laboratory code number. The overall means, range and between laboratory standard deviations are shown in Table 4. The agreement between laboratories is similar to that observed in the collaborative study to establish the 1^st International Standard for HCV RNA for NAT-based assays 13. The quantitative results from laboratories 3B and 5 are shown in Table 5. For laboratory 3B, the results are based on the neat and 1 in 10 dilutions only, as the results did not appear linear for dilutions beyond this. Results were expressed as "copies/ul". For laboratory 5, results were expressed as "Par/ml". The estimates from laboratory 3B are much higher than the results from the qualitative assays, particularly for samples AA, BB and CC. For laboratory 5, the estimates for samples AA, CC and DD are much lower than those from the qualitative assays. The estimate for sample BB is of the same order as the qualitative assays, but 4 log₁₀ higher than the equivalent estimates of AA and CC. This may be due to the fact that this particular assay is amplifying RNA and not DNA, compared to the other assays. Samples AA, CC and DD were subject to multiple rounds of freeze-thawing during their preparation, and due to cell lysis, the labile nature of RNA and its susceptibility to degradation by nucleases have resulted in loss of potency of these preparations when RNA levels are examined. Sample BB only underwent a single freeze-thaw cycle before dispatch to study participants and hence has a higher potency in this particular assay for Laboratory 5. These results emphasize that this candidate standard is not suitable for RNA assays.

Figure 2 shows the performance of the four candidate standard preparations in a real-time TaqMan PCR assay (as described in Stability Studies section of the Materials and Methods), which further demonstrates mean estimated PCR detectable units/ml values shown in Table 4. Near identical crossing cycle (Ct) values are found for samples AA, BB and CC, with sample DD generating a much later Ct values reflecting the dilution of this material when it was originally prepared. It should be noted that sample AA was derived from blood from an exchange transfusion that was determined to have an overall parasitaemia of approximately 9.8% which had been determined by light microscopy. Sample BB was a liquid preparation of in vitro cultured P. falciparum (3D7 strain) with a parasitaemia of approximately 10% was also evaluated in this study. The mean estimates of these two preparations AA and BB are 8.51 and 8.45 log₁₀ PCR detectable units/ml respectively based upon the results of the end point dilution assays, demonstrating that the patient derived sample AA gives results that are commutable to the in vitro propagated sample derived from the 3D7 laboratory strain of P. falciparum.

The stability of sample AA was examined over time in accelerated degradation studies. Samples were incubated in temperature controlled environments, withdrawn at specified times and tested for P. falciparum DNA content. A baseline titre was determined by comparison to vials of sample AA stored at -20°C (i.e. the long term storage temperature for lyophilized standards). The overall geometric means (log₁₀) for the estimated parasites/ul were determined for the sample AA at each of the different incubation temperatures. This is displayed as the log₁₀ drop (Table 6). At the higher temperatures there is clearly some degradation, with a drop of around 0.3 log₁₀ after a year at +20°C, and around 0.6 log₁₀ at higher temperatures. There was little observed loss at +4°C however, with an apparent increase at 8 months, and a loss of 0.06 log₁₀ at 12 months. This data is based on a limited number of vials however, and these differences are within the expected assay variability. If the Arrhenius model for accelerated degradation 2425 is applied to this data, a predicted loss per year at -20°C of around 20%, or 0.1 log₁₀ is obtained. However this is not a precise estimate. This prediction does not appear to be consistent with the observed data for the sample stored at +4°C, which has not shown degradation of the order of that predicted for the -20°C sample. It is not clear whether the Arrhenius model, which is based on a simple first order degradation process, is appropriate for material of this type. Although there is degradation at higher temperatures, there is little observed degradation after 12 months at +4°C and this observed stability is in line with that observed for the WHO International Standards for the blood borne viruses 131517. Studies of real-time stability are of sample AA are ongoing.