Detection of malaria parasites by PCR: a reply

TRANSACTIONS OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE

J Correspondence J Detection of malaria parasites by PCR One of the most common problems with the polymer­ ase chain reaction (PCR) is contamination. This problem is brought to the fore in studies comparing the detection of a given organism using PCR as compared to the cur­ rent 'gold standard'. Are the samples which are positive by PCR due only to the greater sensitivity of this method or are they due to contamination of the PCR with pre­ viously amplified products? Great care must be taken in the design of PCR studies so that this dilemma can be ad­ dressed. G. Snounou et al. (1993: Transactions, 87, 649653) report a far greater sensitivity of the PCR over microscopy, yet 30% of 41 samples reported as negative by the first PCR were positive on subsequent PCR. These samples were subsequently reclassified as being positive. Their explanation for this was that there was very low parasitaemia in these samples, which was at the limit of detection by PCR. An alternative explanation is that these inconsistent results were due to an underlying rate of contamination. The authors refute this by stating that none of their negative controls was ever positive. Yet only one negative control was included per PCR run. At least as many negative controls as samples assayed should have been performed. With only one negative control it is impossible to say that there was no contamination; SO% of the reactions could be contaminated in one run with the negative control having an even chance of still being negative. If the rate of contamination was as low as 10%, then on average only one of the 10 negative controls would have been positive. In fact 10% is the rate of con­ tamination routinely observed in our own laboratory des­ pite taking all the precautions described in detail in the report by Snounou et al. The fact that only Plasmodium falciparum products were detected in further assays does not discount contamination. P. falciparum product was most often amplified from the samples so it is not unrea­ sonable to assume that this was the most abundant ampli­ fication product contaminating the environment and the most likely to find its way into new PCRs. Of course, PCR is intrinsically more sensitive than microscopy and this study probably reflects this. But before drawing such far-reaching conclusions from PCR studies, it is im­ portant properly to control for contamination so that there is no room for argument and its occurrence can be irrefutably discounted. Stuart M. Wilson Department of Clinical Sciences London School of Hygiene and Tropical Medicine Keppel Street 10 December 1993 London, WC1E 7HT, UK

Detection of malaria parasites by PCR: a reply As a result of the extreme sensitivity of the polymerase chain reaction (PCR), contamination with extraneous PCR product will always remain an Achilles' heel of this technique. The risk of contamination is further exacer­ bated by the use of the so-called 'nested' PCR (i.e., reac­ tions performed in 2 steps). Thus it is understandable that this point was raised by Dr S. M. Wilson in his let­ ter (above), in view of the discrepancy between results from microscopical examination and those obtained by PCR (Snounou et al., 1993: Transactions, 87, 649-653) when diagnosing infections with Plasmodium falciparum using the assay described by SNOUNOU et al. (1993:

(1994) 88,

CORRESPONDENCE

363

Molecular and Biochemical Parasitology, 61, 315-320). In­ consistency in PCR diagnosis was found in 12 of the 74 cases (16%) that were assayed in duplicate. Nevertheless, we are confident that contamination was not a factor which significantly affected our results. An underlying level of contamination approaching 10% is out of the question in our laboratory for the fol­ lowing reasons. (i) Although only one negative control was used in each PCR run, the controls remained nega­ tive in all the runs performed. To complete the analysis of all the samples, 17 separate first amplification runs were required, the product of which was amplified 4 times, once for each of the 4 human malaria species. Contamination in our laboratory, detected by a negative PCR control giving a positive reaction, was detected in only very few (5 or so) of the numerous PCR runs per­ formed(> 700). Once contamination was detected, all re­ agents, as well as the data obtained, were immediately discarded. Experimental work was resumed only when the work areas, equipment and reagents were shown to be 'clean' by a test PCR run in which no amplification was observed in the negative controls. (ii) A similar nested PCR assay is commonly used in our laboratory to amplify 6 polymorphic genetic markers of P. falciparum. The pattern of PCR products, comprising 1 to 6 discrete products, is different between the vast majority of samples, and is always reproduced in duplicate experi­ ments, in spite of varying the sample order. This demon­ strates that physical 'carry-over' of the PCR product in nested PCR, or indeed between DNA samples, is very rare in our laboratory. (iii) The suggestion by Dr Wilson (lac. cit.) that as many negative controls as the number of samples be assayed would indeed help to confirm the ac­ curacy of the assay, but would prove a costly exercise when large numbers of samples have to be analysed. We have now adopted the regular performance of a PCR run in which 10 or more positive samples are interspersed with an equal number of negative samples, as a means of detecting contamination resulting from the environment, or indeed from careless handling. In the first 2 runs; in which 20 positive/negative samples were assayed, no con­ tamination was detected. Since our article was primarily concerned with epi­ demiology, a description of all the extensive precautions, required when any method involving PCR is used, was not included. However, we would be very happy to pro­ vide full details of our procedure to any interested col­ league. It is difficult to envisage a situation in which con­ tamination can be irrefutably discounted; however, precautions aimed at physically separating the PCR reac­ tion from the PCR product should result in adequate control of contamination (WILSON, S. M., 1993: Trans­ actions, 87, 609-611). Had we experienced the levels of contamination suggested by Dr Wilson, we certainly would not have considered our data suitable for publica­ tion. On the contrary, we are confident that the level of contamination in our laboratory leaves little room for ar­ gument with the results presented. Georges Snounou & K. Neil Brown National Institute forMedical Research The Ridgeway, Mill Hill London, NW7 1AA, UK Virgilio E. do Rosario Instituto de Higiene eMedicina Tropicais Centra deMalaria e Outras Doenfas Tropicais RuaJunqueira 96, 1300 Lisbon, Portugal 21January 1994