| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The most commonly used genotyping method involves amplifying family-specific alleles of the polymorphic genes merozoite surface protein 1 (msp1) and merozoite surface protein 2 (msp2) using nested polymerase chain reaction (PCR) and gel electrophoresis.2 Interpretation of agarose gels has been shown to vary significantly even when performed repeatedly by the same individual.3 When pretreatment and recurrent parasitemia samples are compared with each other, genotypes of two unrelated strains may appear the same by chance alone, leading to the misclassification of a new infection as a recrudescence. Such misclassification is dependent on the number of molecular markers used in genotyping, the population diversity of the molecular markers, and the number of strains present in an individual infection. Misclassification may also occur when minority strains are not detected or when false-positive alleles are detected. Compared with low transmission areas such as Southeast Asia, where genotyping techniques were first developed, and patients are typically infected with one or a few strains, misclassification is more likely to occur in areas of high transmission intensity where infections with multiple strains are common.1 Misclassification of genotyping outcomes will lead to inaccurate estimates of drug efficacy, compromise the ability to compare results from different studies, and limit studies aimed at identifying risk factors for drug resistance.
The accuracy of genotyping may be improved by increasing the number of molecular markers assessed. Other polymorphic genes with large repeats that have been evaluated for this purpose, such as circumsporozoite protein and glutamine-rich protein, can be characterized with gel electrophoresis, but show less diversity then msp1 and msp2. Microsatellites, or simple sequence repeats, have been used for studies of population structure4 and are good candidates for additional markers. Hundreds have been described,5 they are generally not under immune selection, and the sizes of alleles fall at predictable, discrete lengths that may enable easy comparison across multiple samples and laboratories. By measuring the size of microsatellites with capillary electrophoresis, which has a resolution of one nucleotide and is highly reproducible, the full diversity of length polymorphisms present in a population can be used. The use of microsatellite markers has been explored in two drug efficacy trials,6,7 but has not, to our knowledge, been validated using polyclonal controls.
We designed a new method for genotyping P. falciparum microsatellites, including an automated algorithm for interpreting results, and optimized the technique for conditions likely to be encountered in clinical samples from highly endemic areas. We validated this method on genomic DNA samples of P. falciparum laboratory clones prepared from dried blood spots on filter paper and compared the results to a commonly used genotyping method using msp1 and msp2. Finally, we applied this method to a small set of field samples and showed that adding microsatellite genotyping to msp1 and msp2 alone strengthened the association between a known molecular marker of drug resistance and recrudescence.
|
MATERIALS AND METHODS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Genotyping of msp1 and msp2. The surface antigen loci msp1 and msp2 were amplified using previously described primers.11 Briefly, 2 µL of template DNA was amplified using nested polymerase chain reaction (PCR), with second round primers specific to allelic families: K1, MAD20, and RO33 for msp1 and IC3D7 and FC27 for msp2. PCR products were separated on a 2.5% agarose gel (UltraPure Agarose; Invitrogen, Carlsbad, CA) and stained with ethidium bromide. A technician blinded to the composition of the samples used GelCompar II software (Applied Maths, Sint-Martens-Latem, Belgium) to select alleles and estimate the size of PCR products using a standardized approach.12
Genotyping of microsatellites.
Twelve trinucleotide repeat microsatellite loci have been previously described for use in genotyping filter-paper blood samples.13 Of these, the six loci located on separate chromosomes with the largest heterozygosity were chosen. Primers were designed for single-round PCR using Primer3 software14 (Table 1
). Forward primers were labeled at the 5' end with either a HEX or 6-FAM fluorophore, and a 5' GTGTCTT "tail" was added to the reverse primers to promote addition of an extra adenosine base for more uniform PCR product size.15 Reaction and cycling conditions were optimized to minimize formation of stutter peaks and maximize detection of minor alleles. Annealing temperatures were varied from 55°C to 70°C, extension temperatures were varied from 62°C to 72°C, the lengths of the denaturing, annealing, and extension steps were varied, the number of cycles was varied, different concentrations of Mg2+ and dNTP were tested, and three different polymerases were evaluated. Final reactions contained 1 µL of template DNA, 200 nmol/L of each primer, 40 µmol/L of each dNTP, 1.5 mmol/L Mg2+, 50 mmol/L KCl, 10 mmol/L Tris-HCl, and 0.4 U of AmpliTaq Gold DNA Polymerase (Applied Biosystems, Foster City, CA) in a total volume of 10 µL. PCR was performed in a Dyad thermal cycler (MJ Research, Waltham, MA) under the following conditions: 5 minutes at 94°C, 10 cycles of 20 seconds at 94°C, 30 seconds at 60°C, decreasing 0.5°/cycle, 25 seconds at 68°C, 30 cycles of 20 seconds at 94°C, 30 seconds at 55°C, 25 seconds at 68°C, and 10 minutes at 68°C.
|
In addition to the primary peak for each allele, distinct patterns of "stutter" peaks were observed for each microsatellite marker. Stutter peak formation is common during amplification of microsatellites and is primarily caused by DNA slippage during PCR at intervals equal to the size of the repeat.16 To create an objective algorithm for filtering out stutter peaks, the height ratio and relative location of observed stutter peaks to true peaks was calculated for each microsatellite. Thresholds were determined to minimize both background noise and stutter peaks while maintaining sensitivity (Appendix 1). These thresholds were automatically applied to data exported from GeneMapper using Microsoft Access (Microsoft Corp., Redmond, WA); peaks with heights below the thresholds were removed, and the remaining peaks were considered alleles. Because thresholds were set relative to the height of other peaks present in the sample, the filtering algorithm does not depend on knowledge of the sample composition and was applied independently of known sample composition.
Interpretation of alleles. Allele sizes for single-clone samples were used to guide the interpretation of alleles in polyclonal samples. For msp1 and msp2, size ranges were used for each clone, taking into account variation within and between gels. A range of 1 base was used for microsatellite alleles because sizes varied by < 0.5 base across all samples. A laboratory clone present in a sample was classified as detected if an allele of the correct size (and correct allelic family for msp1 and msp2) was recorded for that sample. Any alleles recorded for a sample that did not correspond to clones present were designated false-positive alleles.
Genotyping of field samples. To evaluate these methods on clinical isolates, we genotyped filter-paper blood samples from a previously completed clinical trial in Burkina Faso.17 Paired samples from patients with recurrent parasitemia within 28 days after therapy with sulphadoxine-pyrimethamine were extracted and genotyped using the methods described above for msp1, msp2, and microsatellites. For msp1 and msp2, alleles from original and recrudescent samples were considered a match if they were within 10 bp in size. For microsatellites, alleles were considered a match if they were within 1 base in size. A recurrent parasitemia sample was considered a recrudescence if at least one allele matched at every locus tested, and a new infection if no allele matched in at least one locus tested. Outcomes were classified using msp1 and msp2 alone and also with the addition of microsatellites. In addition, all pretreatment samples were evaluated for the presence of the dhfr 59R mutation using previously published methods.18 The association between dhfr 59R and recrudescence was measured using Fisher exact test.
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
was often unable to detect more than two alleles present in a sample; TA87 frequently produced false-positive alleles when multiple clones were present (data not shown). These two markers were therefore discarded, and the data are presented are for msp1, msp2, and the remaining four microsatellite markers (TA40, TA60, TA81, and PfPK2).
Ability to detect clones in controls.
For each DNA sample, observed alleles were compared with the known composition of laboratory clones using two methods: 1) nested amplification of msp1 and msp2 followed by gel electrophoresis of amplified fragments and 2) amplification of four microsatellites followed by capillary electrophoresis (Table 2; more detailed results are available in Appendix 2).
|
). Samples containing the same ratios of clones but different parasite densities generally gave similar results.
|
). A similar number of alleles were missed by the microsatellite markers and by msp1 and msp2 genotyping. The factors contributing to missed alleles were similar for the microsatellites as for msp1 and msp2: a low proportion of parasites and larger allele size were associated with missing an allele (Figure 1). Detection of false-positive alleles in controls. Observed alleles not expected based on the known composition of laboratory clones were designated false-positive alleles.
msp1 and msp2.
False-positive alleles were frequently detected using msp1 and msp2 genotyping (Table 3; Figure 2; more detailed results are in Appendix 3). In single-clone samples, msp1 genotyping detected no false-positive alleles, and msp2 genotyping detected zero to two false-positive alleles. When multiple clones were present, msp1 genotypes contained up to one false-positive allele, and msp2 genotypes always contained one to three false-positive alleles. No pattern was detected that could reliably predict the occurrence of false-positive alleles.
|
|
). Field samples. In Burkina Faso, dhfr 59R is the primary mutation associated with resistance to sulphadoxine-pyrimethamine. Patients without this mutation would not be expected to fail therapy. Classifying outcomes based on the results of msp1 and msp2 alone, four of seven patients presenting with parasites containing the dhfr 59R mutation experienced recrudescence versus two of eight patients presenting without the mutation (57% versus 25%, P = 0.31). When outcomes were reclassified adding microsatellite data, four of seven patients with dhfr 59R were considered recrudescent versus zero of eight patients without the mutation (57% versus 0%, P = 0.03).
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and TA87) were inaccurate for polyclonal samples, showing the importance of validating new genotyping techniques on these types of samples. The other four microsatellite markers performed at least as well as msp1 and msp2 in correctly identifying alleles present in filter-paper blood samples containing multiple strains of P. falciparum and produced fewer false-positive alleles. Finally, we showed that the addition of these four microsatellites to msp1 and msp2 genotyping in a small number of field samples strengthened the association between a known molecular marker of drug resistance and recrudescence. Because our control samples were artificial mixtures of laboratory clones, we observed trends but cannot precisely estimate the probability of failing to detect alleles for field isolates collected from a clinical trial. Missed alleles occurred at a similar frequency for the four microsatellite markers as for msp1 and msp2, suggesting that these microsatellite methods are as sensitive for detecting minority alleles in polyclonal samples. As seen previously,3,19 relative densities of clones strongly influenced detection at all six loci, and larger alleles were more likely to be missed.20 The greater size differences between msp1 and msp2 alleles than those of microsatellites may have counterbalanced the sensitivity advantage of family-specific PCR,21 resulting in similar rates of detection overall. Although not observed in this study, it is also possible that our microsatellite genotyping method may miss a strain by inappropriately filtering out a minority allele that is at the same size as a stutter peak.
False-positive alleles were detected at msp1, msp2, and two of the four microsatellites (TA40 and PfPK2), with msp2 producing more false-positive alleles than other loci. For msp1 and msp2, false-positive alleles may have been caused by primer or product carry-over from the first round of PCR, heteroduplex formations detected as additional bands,22 recombination during PCR resulting in formation of chimeric products,23 DNA strand slippage during PCR, or PCR contamination. False-positive alleles at TA40 and PfPK2 could have been created by the same factors except for carry-over, because our microsatellite PCR was not nested, and heteroduplex formation, because capillary electrophoresis was performed under denaturing conditions. Some or all of these factors are likely to influence the accuracy of any PCR-based genotyping system.
In choosing the most appropriate methods for genotyping a clinical drug efficacy trial, characteristics of both the genotyping techniques and the study site need to be considered. The probability of two independent strains having the same genotype depends on the diversity of the genetic loci assessed and the ability of the genotyping technique to measure that diversity. When unrelated strains appear the same, new infections may be misclassified as recrudescence. When more strains are simultaneously present, such as in sub-Saharan Africa, the probability of this misclassification increases. One way we have attempted to adjust for this problem in the past has been to classify sample pairs containing both matched and unmatched alleles as new infections.12 However, the persistence of even one of a number of strains originally present may suggest drug failure and would ideally define recrudescence.2 When using multiple loci, a recrudescence should be defined as the presence of at least one matched allele at every locus; if at least one locus shows only unmatched alleles, the outcome should be classified as a new infection. By increasing both the resolution of the genotyping method and the number of loci assessed, discriminatory power may be increased enough to enable use of these definitions. Increased resolution enables the detection of small differences between strains at a given locus, and adding additional loci further increases the probability of detecting genetic differences between strains because they may have the same genotype at one locus but have detectable differences at another.
Using these definitions, missing alleles in genotypes from drug efficacy trials can lead to the misclassification of a recrudescence as a new infection. For example, a minority allele present in a pretreatment sample may be missed, especially if it is larger in size than the dominant allele(s). If the corresponding parasite strain is the only one able to survive drug therapy and causes recurrent parasitemia, the recurrence will be classified as a new infection, even if the allele is only missed at one locus. For this reason, the increase in discrimination afforded by adding additional loci needs to be balanced with the increased risk of misclassifying a recrudescence as a new infection if alleles are missed at those loci. False-positive alleles, on the other hand, can lead to misclassification of a new infection as a recrudescence at a particular locus. Adding an additional locus, however, will not increase the chance of misclassifying a new infection as a recrudescence, even if false-positive alleles are present at that locus.
By adding four microsatellite loci to msp1 and msp2 genotyping in a small set of field samples, we reclassified two outcomes from recrudescence to new infection. Both patients with reclassified outcomes presented with wild-type alleles at the dhfr 59 locus and would not have been expected to fail therapy with sulphadoxine-pyrimethamine. The increase in strength of association between the mutant allele dhfr 59R and recrudescence after adding the microsatellite loci (P = 0.31 versus P = 0.03) shows that a change in even a small number of genotyping outcomes may improve the ability to identify risk factors for drug resistance.
Our microsatellite genotyping method offers a number of advantages over genotyping using surface antigen loci and agarose gel electrophoresis. Paired clinical samples genotyped using agarose gels need to be run in adjacent lanes on the same gel to make meaningful comparisons, given the large variability in size measurements across gels. The precise sizing of capillary electrophoresis and discrete allele sizes of microsatellites make comparisons between multiple samples evaluated at different times and across different laboratories possible. The high resolution of capillary electrophoresis also enables measurement of the full diversity present at these loci; in contrast, with agarose gels, a resolution of 6 bp7 limits the ability to measure allelic diversity. Analysis of agarose gel results is inherently subjective3; our method, relying on a simple algorithm for determining true alleles, is more objective and minimizes investigator bias. Finally, our method differs from most surface antigen and microsatellite genotyping techniques in that it does not depend on nested PCR. This decreases the probability of contamination, eliminates the problem of artifact from first-round primer or product carryover, and saves on time and reagent costs. Capillary electrophoresis has the disadvantage of not being available in many resource-poor settings. However, it is available in most research centers that perform automated DNA sequencing, and at our institution, the cost of analyzing four microsatellite loci is similar to that of genotyping msp1 and msp2 when all reagent costs are considered. The additional cost we incurred in genotyping four microsatellites for our field samples was $1.51/sample.
Choosing the optimal method of genotyping samples from a P. falciparum clinical drug efficacy trial requires knowledge of the performance and limitations of genotyping techniques when applied to a particular setting. In areas where polyclonal infections are common, additional loci may need to be added to commonly used genotyping techniques to accurately distinguish recrudescence from new infection. We have proposed an efficient and objective method for genotyping four independent loci that compares favorably with existing techniques and may aid in the interpretation of results from drug efficacy trials, especially in regions of high endemicity where rates of both polyclonal infections and new infections during follow up are high. We are currently in the process of applying this method to a large number of field samples from different sites to determine the optimal combination of markers, including microsatellites, msp1, and msp2, for use in different epidemiologic settings.
Note: Supplemental material appears online at www.ajtmh.org.
Received May 12, 2006. Accepted for publication July 12, 2006.
Acknowledgments: The authors thank Jiri Gut, Jennifer Legac, and Sarah Baxter for culturing parasites and performing flow cytometry.
Financial support: Support was provided from the National Institutes of Allergy and Infectious Disease (AI052142), Fogarty International Center/National Institutes of Health (TW00007), and the Doris Duke Charitable Foundation. P.J.R. is a Doris Duke Charitable Foundation Distinguished Clinical Scientist.
* Address correspondence to Bryan Greenhouse, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143. E-mail: bgreenhouse{at}medsfgh.ucsf.edu 
Authors’ addresses: Bryan Greenhouse, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143, Telephone: 415-206-8844, Fax: 415-648-8425, E-mail: bgreenhouse{at}medsfgh.ucsf.edu. Alissa Myrick, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143, Telephone: 415-206-3352, Fax: 415-648-8425, E-mail: amyrick{at}medsfgh.ucsf.edu. Christian Dokomajilar, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143, Telephone: 415-206-3349, Fax: 415-648-8425, E-mail: cdokomajilar{at}medsfgh.ucsf.edu. Jonathan Woo, Institute for Human Genetics, University of California, San Francisco, 401 Parnassus Avenue, Bldg. LPPI, Rm. LP70, Campus Box 0984, San Francisco, CA 94143, Telephone: 415-502-1576, Fax: 415-476-7389, E-mail: jonathan.woo{at}ucsf.edu. Elaine Carlson, Institute for Human Genetics, University of California, San Francisco, 401 Parnassus Avenue, Bldg. LPPI, Rm. LP70, Campus Box 0984, San Francisco, CA 94143, Telephone: 415-514-2289, Fax: 415-476-7389, E-mail: elaine.carlson{at}ucsf.edu. Philip J. Rosenthal, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143, Telephone: 415-206-8845, Fax: 415-648-8425, E-mail: prosenthal{at}medsfgh.ucsf.edu. Grant Dorsey, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143, Telephone: 415-206-4680, Fax: 415-648-8425, E-mail: gdorsey{at}medsfgh.ucsf.edu.
Reprint requests: Bryan Greenhouse, Department of Medicine, University of California, San Francisco, Box 0811, San Francisco, CA 94143.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Liu, J. Mu, H. Jiang, and X.-z. Su Effects of Plasmodium falciparum Mixed Infections on In Vitro Antimalarial Drug Tests and Genotyping Am J Trop Med Hyg, August 1, 2008; 79(2): 178 - 184. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. K. Laufer, A. A. Djimde, and C. V. Plowe Monitoring and Deterring Drug-Resistant Malaria in the Era of Combination Therapy Am J Trop Med Hyg, December 1, 2007; 77(6_Suppl): 160 - 169. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Greenhouse, C. Dokomajilar, A. Hubbard, P. J. Rosenthal, and G. Dorsey Impact of Transmission Intensity on the Accuracy of Genotyping To Distinguish Recrudescence from New Infection in Antimalarial Clinical Trials Antimicrob. Agents Chemother., September 1, 2007; 51(9): 3096 - 3103. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Dorsey, S. Staedke, T. D. Clark, D. Njama-Meya, B. Nzarubara, C. Maiteki-Sebuguzi, C. Dokomajilar, M. R. Kamya, and P. J. Rosenthal Combination Therapy for Uncomplicated Falciparum Malaria in Ugandan Children: A Randomized Trial JAMA, May 23, 2007; 297(20): 2210 - 2219. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
