Extensive genetic diversity of Plasmodium falciparum isolates collected from patients with severe malaria in Dakar, Senegal

TRANSACTIONS OFTHEROYALSOCIETY OFTROPICALMEDICINEANDHYGIENE(1996) 90,704-711

Extensive genetic diversity of Pfasmodium fakiparum patients with severe malaria in Dakar, Senegal

isolates

collected

from

F. Robertl, F. Ntoumil’, G. Ange12, D. Candito2, C. Rogier3, T. Fandeur4, J.-L. Sarthou3* and 0. Mercereau-Puijalon 1 ‘Unite’ d’lmmunologie Molhdaire des Parasites, Institut Pasteur, 2.5 rue du Dr Roux, 7501.5 Paris, France; 2HGpital Principal, Dakar, S&&al; 31nstitut Pasteur, Dakar, StWgal; 41nstitut Pasteur, Cayenne, French Guiana Abstract While some genetic host factors are known to protect against severe Plasmodium falciparum malaria, little is known about parasite virulence factors. We have compared the genetic characteristics of I? falciparum isolates collected from 56 severe malaria patients and from 30 mild malaria patients recruited in HGpital Principal, Dakar, Senegal. All isolates were typed using polymerase chain reaction amplification of polymorphic genetic loci (MSP-1, MSP-2, HRPI, GLURC,CSP, RESA, and the multigene family Pf60). The complexity of infections was lower in severe than in mild malaria and the parasite genetic diversity in both groups was very large. No specific genetic make-up was associated with severity; there were, however, marked differences in allele frequencies in both groups, with a prevalence up to 60% of MSP-2 alleles specifically observed in the severe malaria isolates. In addition, the presence of MSP-I/R033 alleles was significantly associated with a higher plasma level of tumour necrosis factor CIreceptor 1 (P
malaria, Plasmodiumfalciparum, genetic diversity, Senegal

Introduction The severity of a Plasmodium falciparum infection depends on the complex interplay of the patient’s immune status and genetic background, and possibly on parasite virulence factors. Several genetic traits have been identified as protecting humans against severe malaria, such as the sickle cell trait, thalassaemia, glucose-6-phosphatedehydrogenase deficiency (reviewed by MILLER, 1994), and the carriage of human leucocyte antigen (HLA) BW53 or DRI 1302 alleles in some ethnic groups (HILL et al., 1991). On the other hand, some genetic factors such as carriage of the TNF2 allele, responsible for higher levels of tumour necrosis factor (TNF) a production, predispose to severe malaria (MCGUIRE~~al., 1994). The hypothesis that some strains of P. falciparum mieht be more virulent than others has been debated for a long time and has recently received renewed attention (GUPTAet al., 1994). Experimental infections in humans have indeed indicated that some strains consistently induced more severe infections than others (JAMESet a&, 1932). The identification of specific virulence factors 1s complicated by the fact that clinical malaria presents a broad range of symptoms. Cerebral malaria, severe anaemia or respiratory distress are the most frequent clinical presentations of severe malaria. These symptoms can be isolated or associated with each other, or can coexist with other clinical manifestations (MOLYNEUX et al., 1989; BREWSTERet al., 1990; WARRELL et al., 1990; MARSHet al., 1995). Moreover, mild malaria presents a wide spectrum of clinical manifestations, most of which are not specific. Little is known to date about the parasite factors that could be responsible for increased virulence and/or for specific clinical manifestations. ALLAN et al. (1993) have shown that P. falciparum strains isolated from cerebral malaria patients induced higher production of TNFcx by monocytes than parasites collected from mild malaria cases. However, this finding was not confirmed in a subsequent study (ALLAN et al., 1995). Similarly, the capacity of some strains to form rosettes was associated with cerebral malaria (CARLSONet al., 1990; RINGWALDet al., 1993; ROWEet al., 1995), but these results are still controversial (HO et al., 1991; ALYAMANet al., 1995).It is usually believed that the pathophysiology of cerebral malaria is a consequence of the massive sequestration of infected red blood cells (IRBC) *Present addresses: F. Ntoumi, CIRMF, Franceville, Gabon; J.-L. Sarthou, Institut Pasteur, Cayenne, French Guiana. Address for correspondence: Odile Mercereau-Puijalon, Unit6 d’Immunologie Mokulaire des Parasites, Institut Pasteur, 25 rue du Dr Roux, 75015 Paris, France.

in brain microvessels, involving expression of specific ligands on the IRBC membrane and nrobablv uuregula-

Gon of specific receptors on the bra& endotheiial;ells (HOMMEL, 1993; TURNERetal., 1994),resulting from the elevated cytokine levels induced by the parasite (GRAUet al., 1989). An additional factor contributing to the outcome of an infection is the parasite burden itself (MOLYNEUXet al., 1989; ROGIERet al., 1996), suggesting that parasite growth rate might be crucial (WHITE & KRISHNA,1989). The work reported here was undertaken as the first step of an analysis of the genetic characteristics of parasites responsible for severe forms of malaria. In particular, we wanted to investigate whether P. falciparum clinical isolates obtained from patients with severe malaria were genetically more closely related than strains collected from mild malaria cases living in the same region, a characteristic predicted from the hypothesis that some strains might be more virulent than others. Experimental evidence in favour of this has been obtained with Toxoalasma pondii lsee SIBLEY& BOOTHROYD. 1992)and numerous bicteriai genera. We have therefore comiared the isolates collected from severe or mild malaria patients living in Dakar, a hypoendemic area of Senegal. The strategy of our study was based on a molecular analysis using polymerase chain reaction (PCR) amplification of several polymorphic genetic loci, used here as typing markers to assess genetic relatedness of strains. The markers chosen were derived from the multigene familv Pf60. recentlv characterized in our laboratorv

(CAREYet al., 1994, i995), as well as from single cop; genes, such as MSP-1 (TANABEet al., 1987) and MSP-2 (SMYTHEet al., 1991), encoding merozoite surface antigens, HRPI coding for the knob-associated histidine rich protein (POLOGE& RAVETCH,1988),GLURP, encoding a protein of the parasitophorous vacuole (BORREet al.. 19911,RESA, encoding a urotein of the dense granules (F~VAL~ROet al., 1986),-a&l CSP, encoding thi sporozoite CS surface protein (DAME et al., 1984). The results obtained for both groups of patients were compared, in order to determine the genetic diversity of the isolates, both in terms of complexity (number of distinct alleles of single loci genes per patient) and allele distribution. Materials and Methods Patients and samples Patients were recruited at the Hapital Principal de Dakar, Senegal. As Dakar is in a hypoendemic region, malaria episodes, including severe ones, occur at any age. With the view of studying the actual epidemiological picture in Dakar, we included in the study all pa-

705

GENETICDIVERSITYOF PLASMODIUMFALCIPARUM

tients presenting at the hospital for malaria?regardlessof their age, sex or ethnic group. Two clinically distinct arouns of uatients were investigated. The first contained 36 Senegaiesepatients from the Service de Reanimation, with severe l? falciparum malaria: 27 children (< 15 years old) and 29 adults. The mean Glasgow coma score was calculated as described by TEASDALE & JENNETT (1974). All patients in this group presented one or more of the svmntoms comorisina the World Health Oraanization definition for ‘sever: malaria (WARRELL et al., 1990), i.e., altered consciousness, convulsions, hvpoglycaemia, acidosis, respiratory distress, severe anaemia, and imnaired visceral functions. Fifteen uatients (27%) died: 4 children aged 6, 8,9 and 14 years,and 11 adults, aged 16, 18 (2 cases),19, 22, 30, 38, 40, 50 (2 cases)and 53 years. A wide range of investigations-biochemical (arterial pH, glycaemia, HCO, creatinine and urea levels), haematological (red blood cell count, haemoglobin level), and immunological (levels of serum cytokines: interleukin (IL) 6, IL-8, IL-lo, TNF-a, TNF-B, interferon y, and of their soluble, circulating receptors: rsIL-6, Rl TNF, R2 TNF)-was conducted. These data will be reported in detail elsewhere. Parasitaemia was calculated after examination of thick blood films bv lieht microscopy. The second group under study consistgd of 30 patients (10 Senegaleseadults, 7 Senegalesechildren and 13 French military personnel), who presented at the hospital dispensary with a mild P. falciparum malaria attack which did not require admission to hospital and who were releasedafter appropriate treatment. Red blood cell pellets were prepared from venous blood samples as described by CONTAMIN et al. (1995) and frozen in liquid nitrogen. All samples were subsequently stored at -80°C. DNA extraction

Deoxyribonucleic acid (DNA) was extracted according to the procedure described by CONTAMIN et al. (1995). Briefly, after lysis of the red blood cells, the parasites were treated with proteinase K to remove haemoglobin. DNA was purified by 3 phenol-chloroform-isoamyl alcohol (25:24:1 vol./vol.) extractions,

Table 1. Oligonucleotide sequences and corresponding falciparum genetic loci examined

Annealing temperature Target MSP-I a

Primer A B :: Ml M2

Mad20

:; 57

MSP-2a : 2 3

HRPI-A HRPI-D GLURP-E GLURP-F

Pf60 CSP RESA

68 62

R033

GLURP

55 62

Kl

HRPl

(“Cl

;:. CSP-A CSP-B RESA-A RESA-F

%onserved region.

57 65 55 61 62 57

precipitated with ethanol, and resuspended in sterile water in a volume corresponding to the initial blood sample volume. PCR typing HRPI, GLURP, the multigene family

Pf60, CSP, and were typed using a single PCR amplification. The various alleles amplified were characterized on size criteria. MSP-I block 2 was typed using 2 strategies: (i) a single PCR using conserved primers followed by hybridization using family-specific probes, and (ii) nested PCR using family-specific primers to drive the second reaction. Amplification was carried out using external primers (A+B) derived from conserved regions flanking block 2. The fragments generated were assigned to an allelic family by hybridization of the PCR products with family-specific DNA probes (CONTAMIN et al., 1995). In order to identify alleles potentially present in low abundance, nested PCRs were carried out, in which the different MSP-1 allelic families (Kl, MAD20 and R033) were specifically amplified from the first PCR products generated using the conserved primers A+B. Nine independent secondary PCRs were carried out, driven either by homologous combinations of family-specific primers (Kl/K2, Ml/M2, RljR2) or by hybrid combinations (i.e., combining a 5’ primer specific for one family with a 3’ primer specific for another family). This strategy allowed enumeration of the various alleles present-in a samule and identification of notential hvbrids. For the MSh-2 gene, 2 consecutive amplification reactions were performed using 2 sets of primers (the first with 1+4, ihe second with-2+3), derived from conserved regions of the gene. The nroducts of the secondarv PCR. renresent& the ceniral polymorphic domain-of the-M&-Z gene, were assigned to one-or the other allelic family (3D7 or FC27) bv hvbridization usina familv-suecific

RESA

probes (CONTAkINet k., 1995; NTOLJMIet al., 19’953.

All amplifications were carried out with 1 yL of DNA, i.e., the amount of DNA corresponding to approximately 1 uL of blood. The outimum hvbridization temnerature of the oligonucleotides was determined for each pair of primers. The sequencesof the various primers and the annealing temperature

used for PCR amplification

of the P.

Oligonucleotide sequences 5’-AAG CTT TAG AAG ATG CAG TAT TGAC 5’-ATT CAT TAA TTT CTT CAT ATC CATC 5’-GAA ATT ACT ACA AAA GGT GCA AGTG 5’-AGA TGA AGT ATT TGA ACG AGG TAA AGTG 5’-GAA CAA GTC GAA CAG CTG TTA 5’-TGA ATT ATC TGA AGG ATT TGT ACG TCT TGA 5’-GCA AAT ACT CAA GTT GTT GCA AAGC 5’-AGG ATT TGC AGC ACC TGG AGA TCT 5’-ATG AAG GTA ATT AAA ACA TTG TCT ATT ATA 5’-ATA TGG CAA AAG ATA AAA CAA GTG TTG CTG 5’-AAC GAA TTC ATA AAC AAT GCT TAT AAT ATG AGT 5’-GAT GAA TTC TAG AAC CAT GCA TAT GTC CAT GTT 5’-CCG GAT CCC ACC CCA TGG TGC AGGC Y-AGA ATT CCA TTG TCC TTT ATT TGT TGC GGC 5’-ATG AAT TTG AAG ATG TTC ACA CTG AAC 5’-AAA TAT TAC TAT ATC CTT TGC TAT TCC 5’-TGG TAC TAG AAC CTA GTG GTA AC 5’-GGA TAA TTA TAT TCT TCT CCAC 5’-AGA AAA TTA GCT ATT TTA TCT GTT TCT TCC 5’-TTG ACC TAT TTA CGA CAT TAA ACA CAC TGG 5’-GAG ACC TTT TCA TGC ATA TAG TTG GA 5’-AGT GAA TTC AAC TCA CTT ATA TGA GGA ATG GC

F.ROBERTETAL.

706

corresponding annealing temperature used are listed in Table 1. For all reactions, parasite clones or lines were used as positive controls (FANDEUR et al., in press). The PCR products were separated on 1.5% agarosegels, except for the Pf60 amplification products, for which 3.5% agarose gels were used. The DNA was stained with ethidium bromide, and visualized under ultraviolet (u.v.) light. Hybridization For MSP-1 and MSP-2 allelic typing, agarosegels of PCR products obtained with primers A+B (MSP-1) and after nested I’CR (MSP-2) were blotted on to Nylon N@

membranes (Amersham, Les Ulis, France) in 2 x SSC buffer (SSC is 0.15 M NaCl and 0.015 M Na3 citrate, pH7.0), and DNA was fixed under U.V.light. [32P]d-ATP labelled DNA probes were prepared by nick translation (Boehringer Mannheim kit) from PCR products generated from reference parasite lines or clones using allelespecific primers. Hybridization was carried out overnight as described elsewhere (CONTAMIN et al., 1995). After several washing steps under different stringency conditions (2 x SSC; 0.5 x SSC, and finally 0.1 x SSC at 65”C), membranes were exposed at -20°C to Amersham photographic film.

Interpretation of results MSP-I alleles hybridizing to one family specific probe

and amplified by nested PCR were used in the analysis. Within one allelic family, different alleles were identified according to their -size and were given arbitrary codes as indicated in Table 3. All fragments hvbridizina to a particular family-specific probe were grouped toy gether and considered as alleles assigned to that specific family. Every ethidium bromide stained band amplified in the primary PCR reaction, that neither hybridized nor produced fragments by nested PCR, was classified as an untyped allele. On the contrary, nested PCR products that failed to hybridize to any probe under non-stringent conditions were not considered as bona fide alleles and were excluded from the analysis. For MSP-2, allelic typing was based on hybridization results obtained using the product of the secondary PCR. and alleles were aroused in the 2 maior allelic families? 3D7 and FC27.-An LMSP-2 fragment implified with primers 2+3 that could not be subsequently typed was considered as an untyped allele. For the other loci (HRPl, GLURP, CSP, RESA and Pf60), the analysis was based on the presenceor absence of an amplified fragment and on the size of the fragments. Results Patients

Among the 56 Senegalesepatients with severemalaria, the mean parasite density was 54 312 IRBC/pL (95% confidence interval [CI] 25 436-83 187). In that group of patients, the following manifestations of severe malaria Table 2. Amplification laria cases

characteristics

were noted: repeated generalized convulsions (21.5% of patients), hypoglycaemia (41%), acidosis (7%), respiratory distress (19.5%), and severe anaemia (12.5%). The mean Glasgow coma score was 8.2 (95% CI 7.1-9.4). The mortality rate was 26.8%. The mean parasite density in the group of mild malaria caseswas 8916 IRBC/yL. Only mild symptoms were observed in these patients. The Glasgow coma score was normal, and no fatal outcome was observed. No clinical or biological difference was observedbetween the Senegaleseand French sub-groups of the mild malaria cohort. Typing efficiency

All DNAs were amplified using conserved primers derived from the HRpl, RESA, -CSP, GLU~P, MSP-I, MSP-2 loci and from the multinene familv Pf60. Although all samplesanalysed were iollected from symptomatic patients with detectable peripheral parasitaemia (i.e., 21-484 800 IRBC/yL), some reactions did not yield an amplification product. Two samples (NP17 and NP42) did not yield a product in any of the 7 typing reactions. As indicated in Table 2, the typing efficiency ranged from 57% to lOO%,depending on the locus investigated. In the samples in which no fragment was amplified in the HRPl and/or RESA reactions, the hvvothesis that failure of amplification might result’from a deletion of the gene was investigated using a PCR driven by an internal primer paired with a primer consisting of telomer-specific repeats (FANDEUR et al., 1993). Not a single‘positive reaction was observed using this alternative typing strategy, while controls such-as the FCR3 DNA. deleted for HRPI POLOGE & RAVETCH, 1988). and the FUP/SP DNA, deleted for RESA (FANDE~R et ai; 1993), yielded the predicted fragment (data not shown): These results indicated that failure to detect an amvlification product was most probably not due to deletion of these genes.For all loci investigated, efficient amplification was observed with the samples with low parasite counts, indicating that sensitivity of the reactions was not a matter of cancer+ at least in the range of parasite density observed in the isolates under study. Complexity of infection Apart from the Pf60 reaction, in which primers annealed to several distinct members of the Pf60 multigene

family and hence amplified numerous fragments per genome, the other typing reactions used here investigated single copy loci and generated one single fragment per genome. Amplification of several fragments from a blood sample indicated, therefore, that the isolate contained several distinct alleles. Evidence for multivle infections was obtained for all reactions investigating single copy loci, except for RESA, where all reactions generated a 720 basepair (bp) fragment. A substantial proportion of isolates were typed as containing more than one MSP-1 or MSP-2 allele (Fig. l), while samples with multiple HRPl, GLURF or CSP alleles were less common. The proportion of isolates pre-

of each locus examined

in l? fulciparum

isolates from severe and mild ma-

Mild cases(n = 30) Severecases(n=56) No. of No. of No. of fragments No. with No. of No. of No. of fragments fragments multiple per multiple positive fragments multiple per multiple per isolate infections infection (mean)a PCR per isolate infections infection (mean)a Locus 1 l-2 HRPI l-2 ;“; [gq 2 l-3 i3 2,l l-2 ;;;RP ; 2 24 (80;) l-2 2 l-2 46 (82;) 1 0 28 (93%) 1 i RESA 2-2 l-7 l-3 21 17 (57%) 9 3,l MSP-I ;‘f ,‘;;y 214 30 (100%) l-4 MSP-2 l-4 2,3 iGb 3-9 Pf60 42 (75:) 2-9 20 (66%) $1, 5,2 527 aMeannumberof bandsobtainedwith isolatesfrom multiple infections. bNotapplicable. No. with positive PCR 44 (78%) ;; i;;;]

PLASMODIUMFALCIPARUM

GENETICDIVERSITYOF

HRPI

GLURP

CSP

MSP-1

707

MSP-2

Fig. 1. Prevalence of isolates of I? falciparwnpresenting more than one allele for single copy loci. The percentage of isolates containing more than one allele is shown for each polymorphic genetic locus investigated. The number of samples harbouring more than one GLURP or MSP-2 allele was significantly higher in mild (hatched bars) than in severe malaria cases(grey bars) (P
Table 3. Distribution of alleles in isolates from patients with severe and mild falciparum malaria

Locus RESA

Allele size SevereMild (bp) case? casesa Locus 720 46 28 GLURP

HRPI Lit !i

580 640 680 700

:;

1 37 1 6

i3

E

i

E G7

CSP ::

1050 0 1100 5

:; C6

1150 1200 1250 1300

C3 -.

MSP-1 Kl-a Kl-b Kl-c Kl-d Kl-e Kl-f Kl-g Kl-h it:;

130 150 160 180

14 10 9 2

0 3 1 0

190

0

210

5

235 250 270 300

4 5 2 0

Mad20-a 180 Mad20-b 190 Mad20-c 220 MadZO-d250 Mad20-e 300 R033-a 150 R033-b 180 R033-c 220 R033-d 280 Untyped

0 2 9 0 1 34 0 1 2 0

:; :, 12 i 0 1 Fi 3 : 0 : 1 i 2 :, 11 2 :, 5

GlO Gil G12 G13

MSPZ 3D7-a 3D7-b 3D7-c 3D7-d 3D7-e 3D7-f 3D7-g 3D7-h 3D7-i 3D7-j 3D7-k 3D7-1 3D7-m 3D7-n 3D7-0 FC27-a FC27-b FC27-c FC27-d FC27-e FC27-f FC27-g FC27-h FC27-i Untyped

Allele size Severe Mild (bp) casesacasesa 1000 1050 1100 1150 1200 1250

1 1 4 5 5 6

0 0 3 6” 5

::z::” 5 1400 4 1430 1 1500 1

i 2

1550 1 1600 1

:, 0

470 500 540 560 580 600 640 680 700 720 760 780 840 880 900 540

1 3 0 3 12 0 8 8 3 3 1 0 1 0 0 2

560 580 620 650

1 8 0

: 0 3

700 680

39

ii

780 740

1 7

z 6

0” :, 0 ; 0 5 : 7 1 : 1

aFor each locus, the number of samples in which each individual allele was detected is indicated, for each clinical group.

senting multiple GLURP or MSP-2 alleles was significantly lower in the severe cases than in the mild malaria patients (Student’s t test, PcO.05 and P
Polymorphism Table 3 summarizes the typing data and indicates the size of the various alleles identified and their relative abundance. As already indicated above, the RESA amplification reaction yielded a single size fragment (720 bp). Four different HRPI alleles (Hl-H4) were identified. Allele H2 (640 bp) was by far the most predominant one, representing 82% and 92% of the alleles detected in the severe and mild cases, respectively. Six different CSP alleles (Cl-C6) were amplified. Allele C3 was detected in 35% and 44.4% of severe and mild cases, respectively. The overall distribution of the various CSP alleles differed significantly between the 2 clinical groups (x2 test, 3 degrees of freedom [dfj, P~0.01). Amplification of the GLURP marker yielded 13 distinct alleles (Gl-G13). Polymorphism of this locus was illustrated by the wide distribution among the various patients and by the relatively low frequency of each individual allele. Seven alleles (G3-G9) accounted for 87% and 97% of all the alleles in severe cases and mild cases, respectively. Allelic polymorphism of MSP-I was very large, as 19 distinct alleles were enumerated in a total of 105 PCR fragments (Table 3). In the group of severe malaria samples, 20/69 MSP-1 fragments (29%) were typed as belonging to the Kl family, and 12/69 (17%) and 37/69 (54%) were assigned to the MAD20 and R033 families, respectively. Six distinct Kl alleles, 3 MAD20 and 3 R033 alleles were observed in this group. The 150 bp R033 fragment largely - __predominated (49% of the MSP-1 alleles), while the other 11 alleles were present at a low freauencv. Thirtv-six fraaments were amnhfied from 17 mifd malaria samples: 3f%, 14% and 39% typed as Kl, MAD20 and R033, respectrvely, while 14% could not be assigned to a specific family (untyped alleles). Seven distinct Kl alleles. 3 MAD20 and 3 R033 were identified in this group. fiere also, the 150 br, R033 fragment was the most predominant (allele frequency 30%).‘kllele distribution did not differ sienificantlv between the 2 groups except for a trend forhigher frequency of R033 alleles in the severe malaria group (x2 test, 2 df, P=O.O9). MSP-2 polymorphism was also very extensive. Twenty-four alleles were observed in a total of 134 PCR fragments. Fifteen distinct 3D7 and 9 distinct FC27 alleles were identified. The overall frequency of each family was similar in both groups, with a slight predominance of 3D7 alleles which represented 57% and 53% of the MSP-2 alleles in severe and mild malaria cases, resuectivelv. However, there were differences in the distribution of individual alleles amongst the 2 groups of patients. The most striking one concerned alleles 3D7-e and 3D7-h, which represented 20/43, i.e., 46.5%, of 3D7type alleles in the severe malaria samples but were not detected in the mild malaria cases (x2 test, 5 df, P
F.ROBERTETAL. the individual genotype of each parasite clone present in mixed infections obviously could not be deduced from the analytical approach used here, the multiple alleles detected in some samples have been simply listed in Table 4. This undoubtedly showed an extremely large genetic diversity of the isolates studied. No specific genotype association could be detected in any group, confirming that indeed both groups contained a large number of distinct strains.

severe

mild

MSP-2

severe MSP-1

mild

severe

mild

CSP

Fig. 2. Percentage of MSP-1, MSP-2, and CSP alleles which were specifically detected in one clinical group and not the other (severe and mild falciparum malaria cases). the FC27 alleles in severe malaria and was undetected in the samples from mild malaria. However, the overall FC27 allelic distribution was similar in the 2 clinical groups (x2 test not significant with 1,2 or 3 df). The percentage of MSP-1, MSP-2 and CSP alleles which were group-specific, i.e., which were exclusively identified in one or other of the clinical groups, is illustrated in Fig. 2. Amplification of members of the Pf60 multigene family using primers I’1 and P2 produced complex profiles. Typical examples are shown in Fig. 3. Each isolate pro@p)

M 1 2 3 4 5 6 7 8 9 10111213141516

Fig. 3. PCR amplification of the P. falciparum multigene family using the primers PlIP2 as described by CARCY et al. (1995). Ethidium bromide staining of amplification products generated from severe malaria isolates NP43 to NP57 (lanes 1 to 15, respectively); lane 16: negative control (no DNA added). M indicates the molecular weight marker (Boehringer Mannheim marker VI, consisting of a mixture of pBR328 BGlI and Hinf I digests); bp=base pairs.

Pf60,

duced a specific pattern, regarding both the size and the number and relative intensity of bands. As indicated in Table 2, the number of bands amplified varied from 2 to 9. There was no correlation between the number of fragments amplified and the complexity of the isolate, as demonstrated by MSP-I/MSP-2 typing. No group-specific pattern emerged from the Pf60 results and no common band could be consistently observed in any group. Genotypes The genetic make-up of each isolate is represented in Table 4, where the individual alleles of each locus have been allocated the specific code indicated in Table 3. As

Discussion The pathogenesis process leading from inoculation of sporozoites to more or less severe clinical manifestations of malaria is still poorly understood. The purpose of this study was to analyse and compare the genetic make-up of isolates collected from either severe or mild malaria cases. The cohorts studied reflected the heterogeneity of patients presenting with malaria in Dakar, where the entomological inoculation rate is low (TRAPE et al., 1992) and malaria is hypoendemic, and as a consequence antimalaria immunity is low. Such an epidemiological context is an advantage in the search for potential parasitederived elements contributing to virulence? as acquired immunity rapidly reduces the clinical severity of malaria infections. One factor that might potentially compromise a comparison of genotypic characters of isolates collected from patients experiencing a clinical episode is uncontrolled drug intake before presentation to the hospital. It was difficult to ascertain precisely what proportion in each group had used some sort -of medicat-ion. While such treatments evidentlv moved inefficient. thev could nevertheless have cleared some parasites and, as a consequence, the blood sample collected at the time of hospital admission may not have adequately reflected the parasite infection experienced by the patient. Some uncertainty existed about the actual location where the various patients received their infective bite, but this was true-for both clinical groups. All severe malaria patients were Seneaalese, while 13 of 30 mild malaria oatients were French expatriates and 17 were Senegal&e. We analysed the subset of isolates collected from French patients separately and did not find any difference in complexity or allelic types present compared to the Senegalese patients of the same clinical groups. We have previously shown that, for single copy genes, amplification efficiency varies with the locus investigated and with the primer combination used (CONTAMIN et al.. 1995). In the work renorted here. 54 of 56 DNA preparations from severe malaria, and’all of 30 from mild malaria, generated a PCR product in at least one reaction, indicating efficient DNA extraction. However, variable amplification yields were noted, in particular for the MSP-I reaction, complicating interpretation of some results. Whether this reflected the presence of partially degraded parasite DNA, or resulted from a mutation in the sequence targeted by the primers, remains to be investigated. Despite these reservations, some clear conclusions emerge from this work. The first concerns the incidence of mixed infections in both groups. Thirty-two of 54 severe malaria isolates, and 23 of 30 mild malaria isolates, harboured more than one allele for at least one locus. A high prevalence of multiple infections is consistent with the reports by CREASEYet al. (1990) and PRESCOTTet al. (1994), who observed 83% and 40%, respectively, of mixed infections in Zimbabwe and in the Solomon Islands. CONWAY et al. (1991) also observed mixed infections in Gambian malaria patients, with an average of 2 distinct clones per isolate. Contrasting with the conclusions of CONWAY et al. (1991), however, we found that the complexity of infection was lower in patients with severe malaria, as the mean number of MSP-2 and GLURP alleles detected per patient was significantly lower in severe isolates than in mild cases. The other typing reactions also suggested less complexity in severe malaria cases than in those with mild malaria. This might simply reflect more fre-

GENETICDIVERSITY OF PLASMODIUM

709

FALCIPARUM

Table 4. Genotypes of the P. fulciparum

isolates from patients with severe and mild malaria

Patient no. Isolate genotype Severe malaria NPl H2 G7 G9 Sl C5 Kg Ra Da De Fc Ff NP2 H2 G13 Sl Ra Ff NP3 H2 G9 Sl Ra Ra Da NP4 H2G6SlC4KgFy NP5 H2 G4 Sl C2 C4 Ra Ra* Fi NP6 u2 NE’7 H2 G8 Sl C3 Kf De Ff NP8 H2 G4 Sl C3 MC Dh NP9 H4 G6 Sl C5 Kg De Di Fd NPlO H2 G6 G12 Sl C3 Ra Ra* Dg NPll H4G9SlC3KfDd NP12 H2 GlO Sl C3 Kg Dd Ff NP13 Hl G8 Sl C3 Ki De Fd NP14 Q NP15 NP16 H2 :lhDh NP17 NP18 H2 G9 SlRa Db De NP19 H4 G7 Sl C3 Ra Ra* De NP20 G3 Sl NP21 Ra De NP22 H2G7SlC5KhRaFf NP23 H4 G7 Sl C5 Ki Ra Di Nl’24 RaU2aU2b ’ NP25 Ra NP26 H2 G3 Sl C6 MC U2 NP27 H2 H3 G7 Sl C5 Ra Db Dg Fg NP28 H2G8SlC4Fd NP29 H2 G3 Sl C4 Di NP30 H2 G8 Sl C5 Ra Ra* Dg Dm Fg NP31 H2G5SlC4RaRa*U2 NP32 H2G7SlC4KhMcDi NP33 H2 Gil Sl C4 Kh Ra Ff NP34 H2 G7 Sl C4 MC Me Ra Ra* Db Dg Fb NP35 H2 G5 Sl C4 Dh NP36 H2 G7 Sl C6 Kh Dj NE’37 Sl MC Ra Fg NP38 Sl C3 Ra Dk NP40 H2 G8 Sl C3 Kh Ra Ra* De Dh Fh NP41 C5 De NP42 NP43 H2 G6 Sl C3 Kb Ra Dg Fd NP44 H2G3SlC2KfMcRdFd NP45 H2 G5 Sl C5 Ra Ra* Dg NP46 H4 G2 Sl C2 Kf Mc Rd Fa NP47 H2 G5 Sl Ra Dh U2 NP48 H2 MC Ff NP49 H2 G8 Sl C3 Mb Dh NE’50 H2 G6 Sl C3 Ff U2 NP51 H4 Gl G5 Sl C2 C4 Kb MC Dh Fa NP52 H2 G7 Sl C5 De NE’53 H2 G4 Sl C2 Kb Mb Rc De NP54 H2 G8 Sl Dg Fd NP55 H2 G4 G7 Sl Kb De Fd NP56 H2 G6 Sl C3 Ra Ra* Dd NP57 H2 G4 Sl C3 Kf Fd

Patient no. Mild malaria MM1 MM2 MM3 MM4 MM5 MM6 MM7 MM8 MM9 MM10 MM11 MM12 MM13 MM14 MM15 MM16 MM17 MM18 MM19 MM20 MM21 MM22 MM23 MM24 MM25 MM26 MM27 MM28 MM29 MM30

Isolate genotype H2 G7 Sl C3 Kf Fe Ff H2 G8 Sl C4 Ma Ra DC Dg Fb H2 Sl Ul Da Ff H2 G8 Sl C4Dl G5 Sl C3 Ka Ul Fb H2 G5 Sl C3 Ul Ff H4 G3 G7 G8 Sl Cl C3 MC Ra Ra* Dg Dl Ff H2G8GllSlC3DgFe H2 G7 Sl C3 Da Dn H2 G5 G8 Sl Cl C3 Mu Ul Fa Fe G3 G7 Sl C3 Kf Fb H2 G8 Sl C3 Kd Dg Dl U2 H2 Sl Dl U2 H2 G7 G9 Sl C3 C4 Ra Ra* Rb Ul Df Fb H2 G8 Sl C3 Ki Di U2 Sl Dk H2G5 Sl C4DgDl G5 Sl C4 Da Dl H2 G6 Sl FyU2 H2 Di H2 G6 Sl C2 Rc Do Fc H2 H4 GlO Sl C2 Dg U2 H2 G3 G9 Sl C2 Kd Kh Df Di Fc H2G6SlC2KeDkFf H2 G6 GlO Sl C4 Ki Ra Ra* Di Fc H2Sl C2Ff ’ H2 G5 G9 Sl Cl Kd Ki Kj Ma Md Ra Ra* Di Do Fc U2 G9 Sl C2 Dl Dm Ff H2 G6 Sl C3 Ra Ra* Rb Di Ff

aThe genetic make-up, including the presence of multiple alleles at the same loci, is presented. For each individual allele of the HRPl, GLURP, CSP and RESA loci, the code used is that indicated in Table 3. For the MSP-1 and MSP-2 loci, the alleles were coded as follows: a capital letter corresponding to the allelic family (K for Kl, M for MAD20, R for R033, D for 3D7 and F for FC27), and a small letter corresponding to allele size as indicated in Table 3. Ra and Ra* refer to 2 distinct 150 bp R033 fragments, which could be differentiated by their amplification profile using 3’ allele-specific primers (primers R2 and M2, respectively). The code Ul refers to MSP-I alleles which could not be assigned to any of the 3 MSP-I allelic families (untyped MSP-1 alleles) and U2 corresponds to MSP-2 alleles which could not be assigned to any of the 2 MSP-2 allelic families (untyped MSP-2 alleles); U2a and LJ2b indicate 2 unassigned alleles of distinct size in the same sample. Code Sl refers to the 720 bp RESAband. quent drug intake before hospital admission by the patients with severe malaria compared to atients experiencina a mild attack. However, in sub-Saharan Africa severe malaria symptoms can occur quite suddenly (MOLYNEUX et al., 1989; WARRELL et al., 1990), leaving little time for medication before hospital admission. In both severe and mild malaria, the number of parasite

types harboured was similar in adults and children and the trend was for a reverse relationship between parasite densitv and comnlexitv: severe malaria isolates had higher parasite density ihan mild malaria samples, and yet their complexity was lower. This contrasted with the observation made with isolates collected from asymptomatic carriers living in a holoendemic area, where com-

F.ROBERT ETAL.

710 plexity was markedly reduced in adults and was positively correlated with parasite density (NTOUMI et al., 1995).These observations are compatible with the interpretation that immunity reduces parasite burden and number of types hosted. In this scenario, severemalaria would be associatedwith high parasite density, resulting from overwhelming multiplication of a limited number of clones, while mild malaria would be causedby a large number of clones reaching a lower density. The second major observation concerns the large extent of polymorphism among the isolates studied. All strains isolated from severe caseswere genetically distinct. This conclusion was derived from the genotyping patterns basedon single loci polymorphism, as indicated in Table 4, as well as from the amplification patterns observed for the multigene family Pf60 which appeared to be polymorphic in both the number of bands amplified and their size. These data certainly indicate that there is no single ‘virulent strain’. The situation strikingly contrasts with T. go&ii, where the genetic make-up of virulent parasites is remarkably homogeneous (SIBLEY & BOOTHROYD,~~~~). Analysis of allele prevalence revealed interesting trends. We observed a higher prevalence of 3D7 than FC27 alleles of the MSP-2 locus in the samplesanalysed, consistent with the data reported by CREASEYet al. (1990) concerning clinical malaria samples collected in Brazil and Zimbabwe. This contrasts with the observation by ENGELBRECHT et al. (1995) that, in Papua New Guinea, clinical malaria was associated with MSP-2 FC27 alleles. MARSHALLet ~2. (1994) also noted a predominance of FC27 allelic type in patients from Irian Jaya. The results we obtained with clinical samples collected in Dakar contrasted with our observation of a higher prevalence of FC27 types in asymptomatic carriers in Dielmo, a holoendemic village located about 280 km south-east of Dakar (NTOUMI et al., 1995). Whether this reflects irrelevant geographical differences in parasite population or a genuine tendency for some association of 3D7 alleles with clinical malaria in Senegal requires further investigation of a larger number of isolates originating from this region. Interestingly, the frequency of alleles of MSP-I, MSP2 and CSP differed substantially in both clinical groups. Due to the uncertainty about the location where the various patients studied had received their infective bites, inherent in any analysis of urban malaria, we cannot rule out the possibility that the observed differences in allele frequency between both clinical groups reflected fortuitous local differences in parasite populations. As no specific focus of more severe malaria has been identified so far in this region, and as the patients studied here were recruited sequentially over several months, this hypothesis is unlikely. We prefer to interpret such an imbalance in the distribution of genetic characters as an indication that there are distinct subsets of strains in those quite different clinical groups. We do not know whether the unbalanced distribution of some alleles-it is striking to note that about 60% of the MSP2 alleles detected in the severe malaria caseswere specific for that group-reflected their implication in some step of the pathogenesis process or whether some factor contributing to virulence was genetically linked to one or more of the loci investigated. Further research is needed, in particular to unravel the exact biological role of most of the proteins encoded by the loci investigated here, which is still obscure. Another potentially interesting trend emerging from this study was the association of R033 alleles with severe malaria. The higher incidence of R033 alleles in the severe malaria g&p compared to the mild malaria group did not reach statistical significance (P=O.O9),but the limited number of mild malaria isolates that could be successfully typed for the MS&l locus precludes any definite conclusion. We think that the association of R033 alleles with severe malaria should be further ex-

plored, as the only statistically significant association between any particular allele and the various immunopathological features investigated in the severe malaria group was indeed the association between the presence of R033 MSP-1 alleles and increased plasma levels of TNFa receptor 1 (P
References Allan, R., Rowe, A. & Kwiatkowski, D. (1993).Plasmodium falciparum varies in its ability to induce turnour necrosis factor. Infection and Immunity, 61,4772-4776. Allan, R., Beattie, I’., Bate, C., Boele van Hensbroek, M., Morris-Jones, S., Greenwood, B. & Kwiatkowski, D. (1995). Strain variation in tumour necrosis factor induction by parasites from children with acute falciparum malaria. Infection and Immunity, 63,1173-l 175. Al-Yaman, F., Genton, B., Mokela, D., Raiko, A., Kati, S., Rogerson, S., Reeder?J. & Alpers, M. (1995). Human cerebral malaria: lack of significant association between erythrocyte rosetting and diseaseseverity. Transactions of the Royal Society of Tropical Medicine and Hygiene, 89,55-X

Borre, M., Dziegiel, M., Hogh, B., Petersen, E., Rieneck, K., Riley, E., Meis, J., Aikawa, M., Nakamura, K., Harada, M., Wind, A., Jakobsen, f., Cowland, J., Jepsen, S., Axelsen, N. & Vuust, J. (1991). Primary structure and localization of a conserved immunogenic Plasmodium.faZciparum glutamate rich protein (GLURP).expressedin both the pre-erythrocytic and ervthrocvtic staaes of the vertebrate life cvcle. Molecular and Biochemkal Par&tology,

49,119-132.

.

Brewster, D., Kwiatkowski, D. & White, N. (1990). Neurological sequelae of cerebral malaria in children. Lancet, 336, 1039-1043. Carey, B., Bonnefoy, S., Guillotte, M., Le Scanf, C., Grellier, P., Schrevel, J., Fandeur, T. & Mercereau-Puijalon, 0. (1994). A large multigene family expressed during the erythrocytic schizogony of Plasmodiumfalciparum. Molecular and Biochemical Parasitology,68,221-233. Carey, B., Bonnefoy, S., Schrevel, J. & Mercereau-Puijalon, 0. (1995). Plasmodium falciparum: typing of malaria parasites based on polymorphism of a novel multigene family. Molecular and Biochemical Parasitology, 80,463-472.

Carlson, J., Helmby, H., Hill, A., Brewster, D., Greenwood, B. & Wahlgren, M. (1990). Human cerebral malaria: association with erythrocyte rosetting and lack of anti-rosetting antibodies. Lance&336,1475-1460. Contamin, H., Fandeur, T., Bonnefoy, S., Skouri, F., Ntoumi, F. & Mercereau-Puijalon, 0. (1995). PCR typing of field isolates of Plasmodiumfalciparum. Journal of Clinical Microbiology, 33,944-95 1. Conway, D. J., Greenwood, B. M. & McBride, J. S. (1991). The epidemiology of multiple-clone Plasmodiumfalcipanrm infections in Gambian patients. Parasitology, 103, l-6. Creasey, A., Fenton, B., Walmker, A., Thaithong, S., Oliveira, S.. Mutambu. S. & Walliker. D. 11990).Genetic diversitv of Plasmodium f&iaarum shows aeograohical variation. AkericanJournal 03T&al Medicinekd’H&?eie, 42,403-413. Dame, J. B., Williams, J. L., McCutchan, T. F., Weber, J. L., Wirtz, R. A., Hockmeyer, W. T., Maloy, W. L., Haynes, J. D., Schneider, I., Roberts, D., Sanders, G. S., Reddy, E. P., Diggs, C. L. & Miller, L. H. (1984). Structure of the gene en-

GENETICDIVERSITYOF

PLASMODIUMFALCIPARUM

coding the immunodominant surface antigen on the sporozoite of the human malaria parasite Plasmodium fukiparum. Science, 225,593-599. Engelbrecht, F., Felger, I., Genton, B., Alpers, M. & Beck, H.P. (1995). Plasmodium fulciparum: malaria morbidity is associated with specific merozoite surface antigen 2 genotypes. Experimental Parasitology, 81,90-96. Fandeur, T., Vazeux, G. & Mercereau-Puijalon, 0. (1993). The virulent Saimiri-adapted Palo Alto strain of Plasmodium fulcipurum does not express the ring-infected erythrocyte surface antigen. Molecular and Biochemical Parasitology, 60,241-248. Fandeur. T.. Mercereau-Puiialon. 0. & Bonnemains. B. (in press).’ Plasmodium falcipahm: ‘genetic diversity of several strains infectious for the squirrel monkey (Saimiri sciureus). Experimental Parasitology. Favaloro, J. M., Coppel, R. L., Corcoran, L. M., Foote, S. J., Brown, G. V., Anders, R. F. & Kemp, D. J. (1986). Structure of the RESAgene of Plasmodium falciparam. Nucleic Acids Research, 14,8265-8277. Grau, G. E., Taylor, T. E., Molyneux, M. E., Wirima, J. J., Vassali. I’., Hommel, M. & Lambert, I’. H. (1989). Tumour necrosis factor and hisease severity in children with falciparum malaria. New EnglandJournal of Medicine, 320, 1586-1591. Gupta, S., Hill, A., Kwiatkowski, D., Greenwood, A., Greenwood, B. & Day, K. (1994). Parasite virulence and disease patterns in Plasmodium fulcipurum malaria. Proceedings of the National Academy ofSciences of the USA, 91,3715-3719. Hill, A. V., Allsopp, C. E. M., Kwiatkowski, D., Anstey, N. M., Twumasi, I’., Rowe, I’. A., Bennett, S., Brewster, D., McMichael, A. J. & Greenwood, B. M. (1991). Common West African HLA antigens are associated with protection from severe malaria. Nature, 352,595-600. Ho, M., Davis, T,, Silamut, K., Bunnag, D. & White, N. (1991). Rosette formatton of Plasmodium fulcipamm-infected erythrocytes from patients with acute malaria. Infection and Zmmunity, 59,2135-2139. Hommel, M. (1993). Amplification of cytoadherence in cerebral malaria: towards a more rational explanation of disease pathophysiology. Annals of Tropical Medicine and Parasitology, 8?,627-635. James, S. I’., Nicol, W. D. & Shute, P. G. (1932). A study of induced malignant tertian malaria. Proceedings of the Royal Societv ofMedicine. 25. 1153-1181. Kern-l’., Hemmer,‘C. J., Gallati, H., Neifer, S., Kremsner, P,., Dietrich, M. & Porzsolt, F. (1992). Soluble tumour necrosis factor receptors correlate with parasitaemia and disease severity in human malaria. Journal of Infectious Diseases, 166, 930-934. Marsh, K., Forster, D., Waruiru, C., Mwangi, I., Winstanley, M., Marsh, V., Newton, C., Winstanley, I’., Warn, I’., Peshu, N., Pasvol, G. & Snow, R. (1995) Indicators of life-threatening malaria in African children. New England Journal of Medicine, 332, 1399-1404. Marshall, V. M., Anthony, R. L., Bangs, M. J., Purnomo, Anders, R. F. & Coppel, R. L. (1994). Allelic variants of the Plasmodium falciparum merozoite surface antigen 2 (MSA-2) in a geographically restricted area of Irian Jaya. Molecular and Biochemical Purusitologv, 63,13-21. McGuire, W., Hill, A. V., Allsopq, C. E. M., Greenwood, B. M. & Kwiatkowski, D. (1994). Vartation in the TNF-a promoter region associated with susceptibility to cerebral malaria. Nature.371. 50X-5 11. Miller, L. H. (1994). Impact of malaria on genetic polymorphism and genetic diseases in Africans and African Americans. Proceedings of the National Academy of Sciences of the USA, 91,2415-2419. Molyneux, M. E., Taylor, T. E., Wirima, J. J. & Borgstein, A. (1989). Clinical features of prognostic indicators in paediatric cerebral malaria: a study of 131 comatose Malawian children.

711

Quarterly Journal of Medicine, 71,441-459. Molyneux, M. E., Engelmann, H., Taylor, T. E., Wirima, J. J., Ardeka, D., Wallach, D. & Grau, G. G. (1993). Circulating plasma receptors for tumour necrosis factor in Malawian children with severe falciparum malaria. Cytokine, 5,

_“. _“_.

hO4-609

Ntoumi, F., Contamin, H., Rogier, C., Bonnefoy, S., Trape, J. F. & Mercereau-Puijalon, 0. (1995). Age-dependent carriage of multiple Plasmodium fulcipurum merozoite surface antigen2 alleles in asymptomatic malaria infections. AmericanJournal of Tropical Medicine and Hygiene, 52,81-88. Pologe, L. & Ravetch, J. (1988). Large deletions result from breakage and healing of Plasmodium fukparum chromosomes. Cell, 55,869-874. Prescott, N., Stowers, A. W., Cheng, Q., Bobogare, A., Rzepczyk, C. M. & Saul, A. (1994). Plasmodium fulcipurum genetic diversity can be characterised using the polymorphic merozoite surface antigen 2 (MSA-2) gene as a single locus marker. Molecular and Biochemical Parasitology, 63,203-212. Ringwald, P., Peyron, F., Lepers, J. I’., Rabarison, P., Rakotomalala, C., Razanamparany, M,., Rabodonirina, M., Roux, J. & Le Bras, J. (1993). Parasite virulence factors during falciparum malaria: rosetting, cytoadherence by cytokines. Infection and Immunity, 61,5198-5204. Rogier, C., Commenges, D. & Trape, J.-F. (1996). Evidence for an age-dependent pyrogenic threshold of Plasmodium fulciparum parasitemia in highly endemic populations. American Journal of Tropical Medicine and Hygiene, 54,613-619. Rowe, A., Obeiro, I., Newbold, C. & Marsh, K. (1995). Plusmod&n falciparum iosetting is ‘associated with malaria severity in Kenya. Infection and Immunity, 63,2323-2326. Sibley, D. & Boothroyd, J. (1992). Virulent strains of Toxoplasma gondii comprise a single clonal lineage. Nature, 359382-85. Smythe, J., Peterson, M., Coppel, R., Day, K., Martin, R., Oduola, A., Kemp, D. & Anders, R. (1991). Structural diversity in the Plasmodium fulcipurum merozoite surface antigen 2. Proceedings of the National Academy of Sciences of the USA, 88, 1751-1755. Tanabe, K., Mackay, M., Goman, M. & Scaife, J. (1987). Allelic dimorphism in a surface antigen gene of the malaria parasite Plasmodium fulciparum. Journal of Molecular Biology, 195, 273-287. Teasdale, G. & Jennett, B. (1974). Assessment of coma and impaired consciousness, A practical scale. Lancet, ii, 81-84. Trape, J.-F., Lefebvre-Zante, E., Legros, F., Ndiaye, G., Bouganali, H., Druilhe, P. & Salem, G. (1992). Vector density gradients and the epidemiology of urban malaria in Dakar, Senegal. American Journal of Tropical Medicine and Hygiene, 47.181-189 ..I ---

--_.

Turner, G., Morrison, H., Jones, M., Davis, T., Looareesuwan, S., Bulev, I., Gatter. K.. Newbold. C.. Pukritavakamee. S.. Nagashihta, B., White, N. & Berendt, A. (1994). An immunehistochemical study of the pathology of fatal malaria. Evidence for widespread endothelial activation and a potential role for intercellular adhesion molecule-l in cerebral sequestration. AmericanJournal OfPathology, 145, 1057-1069. Warrell, D. A., Molyneux, M. E. & Beales, P. F. (editors) (1990). Severe and complicated malaria, 2nd edition. Trunsactions of the Royal Society of Tropical Medicine and Hygiene, 84, supplement 2. White, N. J. & Krishna, S. (1989). Treatment of malaria: some considerations and limitations of the current methods of assessment. Transactions of the Royal Society of Tropical Medicine and Hygiene, 83,767-777. Received 4 March 1996; revised 2 July publication 2Ju2y 1996

1996; accepted for