Antibodies to blood stage antigens of Plasmodium falciparum in rural Gambians and their relation to protection against infection

TRANSACTIONS OF THE ROYAL SOCIE~

Antibodies Gambians

OF TROPICAL MEDICINE AND HYGIENE (1989) 83, 29%303

to blood stage antigens of Plasmodium fakiparum and their relation to protection against infection

293

in rural

K. Marsh’*, L. Otoo’, R. J. Hayes’, D. C. Carson2 and B. M. Greenwood’ Laboratories, Fajara, The Gambia, West Africa; ‘Tropical Epidemiology Tropical Medicine, Keppel Street, London, WCIE 7HT, UK

Abstract Cross-sectional and longitudinal studies were performed in a rural population living in The Gambia to examine the relationship between several in vitro assaysof the host immune responseto asexual stages of Plasmodiumfakipwn and protection from malaria in vivo. Assays included an enzyme-linked immunosorbent assayfor antibodies to schizont antigens; an indirect immunofluorescence assay for total antiblood-stage antibodies; an immtmofluorescence assay on glutaraldehyde-fixed parasites to detect antibodies to antigen Ff 155; an assayfor serum inhibition of red blood cell invasion; a micro-agglutination assay to detect antibodies to neo-antigens on the surface of infected red blood cells; and an assayusing polymorphonuclear leucocytes to detect antibodies capable of opsonizing schizont infected red blood cells. There were marked differences in the age-related pattern of response for different assays performed on sera obtained at a cross-sectionalsurvey of 280 individuals. Examination of the correlation between the various immune responses and malariometric indices at the population level and at the individual level provided no evidence that any of the in virro assayswere related to protective immunity. The relationship between in vitro measurements of the anti-malarial immune response and protection from clinical episodes of malaria was examined in a group of 134 children aged 11 years and under who were monitored weekly throughout an entire malaria transmission season. The only immune factor to show a consistent protective effect against clinical malaria was the titre of antibodies to neo-antigens on the infected erythrocyte surface (P=O*Ol). The same longitudinal techniques were used to examine the effect of two non-immunological factors, sickle cell trait and mosquito net usage, both of which showed significant protection against clinical episodes and malaria. Introduction

Plasmodium falcipatwn infections in naive subjects of any age have high attendant morbidity and mortality. In areassubject to stable malaria transmission the ill effects of malaria fall on young children, older subjects develop a state of -functional but non-sterilizine immunitv. The characteristics of this immunity b&e been described in detail in many endemic areas (MCGREGOR, 1986) and although the exact timing of events varies with differing degreesof transmission the general features are constant.

‘Medical Research Council Unit, London School of Hygiene and

The last 20 years have seena remarkable expansion in our knowledge of malaria parasites, particularly at the molecular level. Unfortunately this has not been matched by progress in understanding the mechanisms involved in the development and maintenance of naturally acquired immunity. There are two important practical reasonsfor investigating these mechanisms, apart from the fundamental questions of hostparasite biology. Firstly, an understanding of the mechanismsof natural immunity may be useful in the development of anti-parasite vaccines. Secondly, naturally acquired immunity in combination with innate factors remains the main protection against malaria in endemic areas and the monitoring of changes in this vital defence should be an important part of any mass anti-malarial intervention. Two problems confront studies which aim to identify host factors protective against malaria. Firstly, there are conceptual and methodological problems in distinguishing the state of malarial parasitization, which may be almost universal in endemic areas,from the clinical disease ‘malaria’. Secondly, the host’s immune response to parasitaemia is complex and available measurementsusing crude antigen preparations tend to reflect exposure to, rather than protection from, malaria. In the present study we have tried to overcome the first problem by using a functional criterion of ‘malaria resistant status’ based on the actual clinical experience of children exposed to malaria, in addition to the use of classical malariometric indices. To approach the second problem we have examined a number of delined antibody responses which are thought to play an important role in inducing natural immunity to P. falciparum. In a previous paper (MARSH et al., 1988) we have described the antibody response to a sporozoite antigen and its relationship to clinical protection in a rural Gambian community. In this paper we report the pattern of antibody responsesto a number of blood stage antigens in the same population and their relationship to the development of clinical immunity. To examine whether the methods that we have used to detect clinical immunity are appropriate we have also analysed our epidemiological findings in relation to possessionof the haemoglobin genotype AS and the use of bed nets, two factors which have been shown in previous studies to have a protective effect against malaria. Materials and Methods Studv area and PoPulation

*Author for correspondence. Presentaddress: Institute of Molecular Medicine, John FtsdcliffeHospital, Headington,

Oxford, OX3 9DU, UK.

The study was carried out in a rural area near the town of Farafenni on the north bank of the Gambia river about 100 km inland. The geographical and

294 demographic features of this areahave been described previously (GREENWOOD et al., 1987). Briefly, it is an area of flat Sudan Savannahwith some mangrove and rice swamps near the river. The climate is dry from mid-October to June. In the 1984rainy seasona total of 606 mm of rain fell in Farafenni. Clinical casesof malaria are similarly seasonal, the majority of cases occurring in the 3 month period between September and November each year, though other malariometric indices show much less seasonalvariation. Our study involved the central village of Kataba and 5 outlvine hamlets, about 10 km tothe east of Farafenni. Thi habitations lie about 2 km from the river and are surrounded by cultivated arable land. Before the study there were no health services available in Kataba; the nearest dispensary manned by a dresserdispenser was about lb km away. At a-compound enumeration in November 1983, the total population was 656 distributed amongst 50 compounds. The population consisted of roughly equal numbers of subjects of Wollof and Mandinka ethnic origin, with a few Fula compounds.

naire was completed and the axillary temperature taken. A thick blood film was nrenared from subiects with a temperature above 3i*5’C. An episode of malaria was delined as a febrile illness with an asexual P. fulciparum parasitaemia of >lOOO parasites per microlitre. Blood films

Thick blood films prepared in the field were air-dried and stained with Giemsa. 100 high power fields were examined and the number of malaria parasites of each species and stage recorded. The number of parasites per microlitre of blood was determined with reference to the number of white cells counted in the same 100 fields (TRAPE, 1985). Huemolgobin genotype

Haemoglobin genotype was determined by electrophoresis on acetatesheetsusing haemoglobin from clotted blood samples which had been stored at -20°C.

Antibodies to asexual stages of P. falciparum Field methods

From April 1984 two Medical Research Council (MRC) fieldworkers went to live in the community and carried out updating of demographic information, mosouito net survevs (BRADLEY et al.. 1986). and sens&zation of the population to the aim; and procedures of the study. From mid-May morbidity monitoring of children under the age of 11 years (end of year age)was instituted on a fortnightly basis and a clinic held in Kataba on alternate weeks. A cross-sectional survey was carried out in the lirst week of Auaust 1984 at which all children then resident between the ages of 1 and 11 years and a randomly chosen 1 in 3 sample of older subjects were seen. At this time two Fula compounds with a total population of 42 declined to participate in the study. Subjects were examined for splenomegaly (children standing, older subjects lying down) and a blood sample taken. Children were bled (24 ml) from the dorsum of the hand using a 25 or 23 gauge ‘butterfly’ and adults from the antecubital fossa. Duplicate thick films were prepared in the field. A small aliquot of blood was saved in ethylene diaminotetraacetic acid and the remainder allowed to clot in glass tubes. Blood fihns of children renorted to be unwell were read that day and a curativedose of chloroquine given if asexual malaria parasites were present. Following the-cross-sectional survey 141 children aged between 1 and 11 years were followed closely with morbidity monitoring on a weekly basis (see below). Further episodesof illness were detected at a fortnightly clinic run by one of us (K.M.). It was decided in advance that monitoring would be terminated after 2 consecutive weeks without cases of malaria being detected. This occurred in the last 2 weeks of December and a second cross-sectional survey of the cohort children and a sample of older subjects was then carried out. Morbidity

monitoring

We detected clinical episodes of malaria in cohort children using techniques previously developed for field use in The Gambia. Each child was visited weekly by a field worker, when a morbidity question-

Two assayswere used. In the indirect fluorescence method 12 spot multiwell slides were coated with blood containing asexual parasites from an in vitro culture (predominantly mature parasites), acetonefixed and dried. For these studies a Gambian isolate (GAM 83/l) cultured by the method of TRAGER& JENSEN(1976) was used. The slides were incubated with dilutions of test sera for 30 min at room temperature, washed 3 times with phosphate-buffered saline (PBS) and incubated for 30 min with sheep anti-human immunoglobulin (Ig) G labelled with fluorescein isothiocyanate (FITC). After washing and counterstaining with Evans blue, slides were examined by fluorescence microscopy. The end point was taken as the last dilution showing clear parasitespecific fluorescence. An enzyme-linked immunosorbent assay (ELISA) for asexual antigens was performed as previously described (0~00 et al., 1988) using sonicated schixonts, from a heavily infected placenta, coated on 96-well plates. Test sera were diluted 1:200, and 100 nl were applied to the wells for 120 min. After washing 3 times in PBS with 1% Tween 20, a secondary antihuman IgG (Dako) coupled to horseradish peroxidase was applied. Results were expressed as optical density (OD) at 492 mn. Antibodies to Pf 155

We used the method of PERLMANNet al. (1984). Briefly, monolayers of ring-infected red blood cells on multispot slides were fixed with 1% glutaraldehyde. Slides were incubated sequentially with test sera, biotinylated goat anti-human IgG (Vector) and finally with an avidin-FITC conjugate (Vector). Slides were counterstained with ethidium bromide before viewing by fluorescencemicroscopy. Positive results consisted of a bright rim of fluorescence restricted to the membrane of erythrocytes containing parasites.

Parasite inhibition by serum

All serawere dialysed 3 times against PBS and once against RPM1 1640 and then filtered through 0.22 pm membranes (Millipore). We wished to distinguish inhibition of the intra-erythrocytic develop-

295 ment of the parasite from inhibition occurring around the time of schixont rupture and invasion. The methods and results of assaysused to detect intraerythrocytic retardation have been reported by MARSHet al. (1987). To detect inhibition occurring around the time of schixont rupture, highly synchronixed schixont-infected cells at a haematocrit of 1.5% and a parasitaemia of 0.3% were suspended in test serum at a dilution of 1:s in RPM1 1640 medium supplemented with 7.5% non-immune serum, 20 mM Hepes, 20 mu bicarbonate and 2 g/litre glucose. Cultures were set up in quintuplicate, 200 ul per well, in 96-well microtitre plates incubated in a candle jar. After 12 h, 150 ul of medium were removed and replaced with complete medium containing 1 uCi/ well of tritiated hypoxanthine. After a further 24 h parasitesfrom triplicate wells were harvested on filters which were dried and counted in scintillant fluid on a beta-counter. Thin films were made from the remaining 2 wells and stained with Giemsa. To ensure comparability of results all sera were tested in two batches, with sera from all age groups mixed on a single plate. Each plate contained positive and negative control sera to allow for interplate variation.

plastic tubes at 200 g for 5 min, the preparations were incubated at 37°C for 30 min. Cvtocentrifuaed smearswere stained with May-Grunwald and Giemsa and the number of infected cells phagocytosedby 100 polymorphonuclear cells determined by light microscopy. Results were analysed as log,0 (n+l) because the distribution was positively skewed. Results Malaria in the study population

150children aged between 1 and 11 years were seen at the August 1984 survey. Nine of these children were excluded from the subsequent longitudinal study because of inadequate documentation. Seven further exclusions were made from analysis of longitudinal data: one child died of unknown causeswhilst away from the village and in 6 casesa finger prick was not obtained at the time of a febrile episode. Parasite rates and densities at the 2 cross-sectional surveys undertaken at the beginning and at the end of the malaria transmission seasonare shown in Figs 1 and 2. With the exception of the youngest children, who generally had higher rates in January, the differences beetween the 2 surveys were not pro-

Parasite-infected cell agglutination assay

Antibodies directed against determinants on the unfixed surface of red cells infected with mature parasites were detected using methods previously described (MARSHet al., 1986). Briefly, a 5 ul pellet containing 5% late trophozoites/early schizonts of a parasite isolate expressing erythrocyte neo-antigens was suspendedat a haematocrit of 10% in dilutions of test serum containing 40 @ml ethidium bromide and rotated for 1 h at 37°C. 25 ul of the suspension were placed under a coverslip. Agglutinates of parasitized cells were clearly visible under ultraviolet illumination. Results are expressed as log* (titre/ lO)+l, with reference to the last dilution causing micro-agglutination of parasite-infected cells. None of our established in vitro parasite lines expressed parasite-dependentred cell neo-antigens as judged by immunofluorescence or micro-agglutination with pooled immune sera (MARSH et al., 1986). We therefore used a single wild isolate, obtained from an adult with clinical malaria, for all micro-agglutination assays. The same isolate was used for opsonixation assays(see below). The blood was cryopreserved in O-5 ml aliquots in liquid nitrogen. Aliquots were thawed and cultured as previously described (MARSH & HOWARD, 1986). Parasites thus treated grew normally to mature schixonts and were capable of several cycles of growth in vitro; however, all assays were performed using parasites in the first cycle of growth. Opsonization of schizont infected cells

Antibodies capable of opsonixing red cells containing mature parasiteswere assayedby the technique of CELADAet al. (1983). Briefly, 2.5~ lo5 polymorphonuclear cells from a non-immune donor (same donor for all experiments) were mixed with 5~ lo6 red cells containing early schixonts in 100 ul of a 1:5 dilution of test serum in RPM1 1640medium. All assayswere performed using aliquots of a single wild isolate (see above)?infected cells being purified to >80% parasitaerma on Plasmagel. After centrifugation in 3 ml

2

50-

w 4: 3

40: 30-

p

20-

+ -

AUG84 JAN85

10 0

0

10

20

30

40

50

AGE (years) Fig. 1. Parasite rates (percentage of each age group with any asexual P. fakipanrm parasites detected by examining 100 high power fields ofa thick blood film) in August 1984and January 1985(+ 1 standard error).

g

38

c8

36

j, g

34 32

P

3.0

5

28

i

2.6

i

2.4

;

2.2

8

2.0

0.

1.8

9 -.-

0

10

AUG84 JAN65

20

30

40

AGE (years)

Fig. 2. Positive densities (log n+ 1 of number of P. fakrparm asexual parasitesper microlitre of blood io subjectswith a positive blood film) in August 1984 and January 1985 (+l standard error).

Table

1. Episodes

of malaria in 134 children

aged l-11 years followed

Number of chilren with 2 or more No episode 1 episode episodes

Age (Yf=S> l-2 Ei 7-8 9-11

from May 19&&January

1985

Percentage with more than 1 episode

Number of episodes per child

10 15”

15 11 9

6 4

2:

0.55 A:E

::

12 6

:

z 36

0.48 0.45

‘Age at end of 1984.

nounced. 92 episodes of clinical malaria (as defined above) were detected in 134 children between May and December 1984 (Table 1 and Fig. 3); 63 cases were detected by morbidity monitoring and a further 29 were detected at clinics. (8 caseswhich occurred before or during the August 1984 survey were included in the analyses of the relationship between non-acquired factors and malaria but these caseswere excluded when considering the role of acquired factors measured in August.) Parasite density did not differ appreciably between casesdetected by morbidity monitoring (mean log10 parasite density k standard deviation (SD)=4*43f0*53) and casespresenting at clinic (4*41+0.49). Parasite densities in clinical cases did not vary with age group.

60-

J

FMAMJJASONDJ

F morbidity monitoring

t

t

Fig. 3. Febrile episodes and episodes of malaria (febrile illness with asexual P. fakipamm density of ~1000 parasites per tnicrolitre) in 134 children between May 1984 and January 1985. Table 2. The effect of mosquito net usage oo parasite rates and parasite densities io children aged 1-11 years

Mosquito net usage No net Net used used Parasite rate August 1984 January 1985

10/17(59%) 6/l 5(40%)

Parasite densityb August 1984 January 1985

2.05f2.01 1.03f1.36

Signifmnce of difference

50/130(38%) 2.28’,ns 60/l 12(549/o) 0.76’,ns 1.14k1.57 1.58f1.62

2.41b,I’=0.02 l.3Sb,ns

‘Mantel-Haensel x2 allowing for age (grouped as 1-2, U, 5-6,7-S, 9-11 years); ns=not significant. bAmalysedas log,, (n+ 1). Tables shows means f standard deviations for all children. ‘t-test calculated by 2-way ANOVA allowing for age (groups as in note a); ns=not significant.

Table 3. The effect of mosquito

Mosquito net use Net used Net not used

Episodes per child O-67 1.08

net usage on clinical

Mosquito net usage and protection against malaria

Mosquito net usageis common in this area of West Africa and only 10% of the children aged l-l 1 years did not use them. Parasite rates and densities were lower in subjects using nets at the time of the August 1984 survey, while there was no such difference in January 1985 (Table 2). Children sleeping under nets experienced significantly fewer episodes of clinical malaria than children who did not (Table 3). Haemoglobin genotype and protection against malaria

The overall prevalence of sickle-cell trait in the population was 20% (1.3% of the population had haemoglobin AC; only one subject homoxygous for haemoglobin S was detected, a female of 14 years who was apparently well). Mean parasite densities were significantly lower in children with the haemoglobin genotype AS in August 1984, but not in January 1985 (Table 4), and children with sickle cell trait experienced many fewer episodes of malaria than children with a normal genotype (Table 5). The protective effect of the AS genotype against clinical episodes remained significant after allowing for mosquito net usage and vice versa (data not shown).

episodes of malaria in children

aged 1-11 years

Number with no episode

Number with 1 episode

Number with 2 episodes


55 4

48 4

13 5

3.67, P=O.O5 4~00,P<0~05”

“Corrected for age (groups as in Table 2).

297 Table 4. The effect of haemoglobin years

genotype on parasite rates and parasite densities in children aged l-11

Haemoglobin genotype AA

AS

Significance of difference

Parasite rate August 1984 January 1985

52/l 19 (44%) 51/103 (50%)

7128 (25%) 14/123(61%)

3.W, P=O*O8 0.67a, ns

Parasite density August 1984 January 1985

1.35k1.70 1.48+1-63

0.71k1.32 1.70f1.55

Table 5. The effect of haemoglobin

genotype on clinical episodes of malaria in children aged l-11 years of age

2~01',P=0~05 0*63', ns “Mantel-Haensel x2 allowing for age (grouped as l-2, 3-4~ 5-6, 7-8, 9-11 years); ns=not significant. bAnaiysed as log,, (n+l). Table shows + standard devianons for all children. ‘t-test calculated by 2-way ANOVA allowing for age (groups as above); ns=not significant.

Haemoglobin genotype it

Episodes per child

Number with no episode

Number with 1 episode

Number with 2 episodes

o-35 O-80

41 17

49 4

16 2

6.97, 5.78,P=O-02 P
aCorrected for age (groups as in Table 2).

Fig. 4. Anti-malarial antibody profdes by age measured in August 1984. PAR. RATE=parasite rate; B.S. IFA=blood stage antigen by indirect imm~ofluorescence (log, (titre/lO+ 1); B.S. ELISA=schizont antigen (optical density=492); lSS.IFA=indirect immtmofluorescence assay for Pfl55 (log? (titre/lO)+ 1); PICA=parasite-infected cell micro-agglutination (log, (titre/lO)+ 1); SCHIZ OPS=schizont infected cell opsonization (log,,(n+ 1); INHIB=serum inhibition of parasite growth in vitro (l/log,, cpm).

Antibodies to blood stage malaria parasites and protection against malaria

Relationships between the immune responses which we measured and malaria were examined in 3 ways: firstly, by determining the temporal relationship between responsesand malariometric indices at a population level; secondly by em ’ ‘ng the correlation between such factors and malariometric indices at the individual level; and thirdly by seeking evidence of protection from clinical attacks of malaria in individuals followed longitudinally.

The age-pattern of the various immune responses studied are presented in Fig. 4, which shows that different aspectsof the immune response have different kinetics. Age-corrected correlation coefficients for the relationship between different immune factors and 2 commonly used malariometric indices and for the relationship between the factors themselves are shown in Table 6 (corrected by calculating partial correlation coefficients with allowance for age in the following groups 1-2, 3+ 5-6, 7-8, 9-11, 12-14, 15-19, 20-29, 30-44 and 245 years). A number of

298 Table 6. Age-corrected correlation coefficients for malariometric indices and in vitro assays of antimalarial immunity in subjects of 11 years or less, and 12 years or more, in August 1984 children aged 11 years or less Subjects aged 12 years or more

Parasitaemia (log n+l)

Parasitaemia (log n+l)

Packed cell volume

Anti-blood stage (ELBA)

Anti-blood stage (IF.9

Anti-Pf 155 (IFA)

-0.35’ (131)

0.38’ (133)

0.45’ (133)

0.36’ (136)

-0.25’

-0.23’ (124)

Packed cell volume

0.10 (109)

-0.17

Anti-blood stage (ELBA)

$2;

-0.08 (100)

q

Anti-blood stage (IFA)

0.19 ( 79)

-0.38’ ( 53)

0.59’ ( 74)

q

Anti-Pf 155 (IFA)

0.06 (11%

-0.03 ( 86)

0.41’ W)

0.68’ ( 76)

Serum inhibition

0.16 ( 31)

@16 ( 24)

-0.08 ( 29)

0.26 ( 16)

0.02 ( 33)

0.26 ( 25)

-0.31 ( 31)

-0.08 ( 18)

-0.05

agglutiuation Scbizont opsonizing antibodies

0.05 ( 26)

0.03

-0.25

@12

-0.37

(

(126)

:

(121)

Serum inhibition 0.03

(102) 0.26 ( w

0.00

Infected cell agglutination

Scbizont opsonizing antibodies

0.02 (135)

(107)

-049

0.02 -0.18

(126)

(102)

0.59’

050’

(125)

(128)

ww

0.36’ (129)

0.13 (107)

0.69’ (132)

0.05 ( 9)

0.26’

(128)

0.20 ww

z

0.03 ( 98)

0.27’ (131)

0.18 W)

-0.01 ( 19)

1

0

0.18 ( 87)

Infected Cdl

‘PiOOl;

21)

(

26)

(

(

1%

(

-0.22 ( 28)

21)

-0.11

18)

(

22)

1

0.38’ (105)

0.37 ( 23)

q

numbers in parentheses=degrees of freedom.

Table 7. The relationship between experience of clinical malaria

immunological

factors

measured

in August

1984 and subsequent

El

Definition of clinical episodea boo

108

-0.16 (0.95)

-0.22 (0.83)

-0.74 (0.46)

110

- 1.35 (0.18)

- 1.30 (0.20)

- 1.87 (046)

110

-0.88 (0.38)

-0.79 (0.43)

-0.68 (0.5)

84

-0.69 (0.49)

-0.65 (0.52)

-1.46 (0.15)

Infected cell agglutination

-2.78 (0*006)*

-2.55 (O-01)*

-2.27 (O-02)’

Infected cell opsonization

-1.19 (0.24)

-0.94 (0.35)

-0.86 (0.39)

Df?Eitiof Total anti-blood stage antibodies (IFA) Total anti-schizont antibodies (ELISA) Anti-Pf 155 (IFA) Serum inhibition

bOO

- 1.57 (0.12) 111 -164 (0.10) -1.19 (0.24) AI&(NANP)~~ “The table shows z-test values for trend in the parameter among children with 0, 1 or 22 episodes obtained by analysis of covariance with allowance for age, mosquito net usage,genotype and housing cluster. Clinical episodes defined as febrile illness with any P. fukipurum asexual parasites(El), >lOOO parasitesper ul (Elm) and >5000 parasites per nl (Esm). Values of probability (P) shown in parentheses. bAntibodies to the conserved sequenceof the P. fulcipatwmcircumsporozoite protein. These are included in the analysis for completeness (see MARSH et al., 1988). ‘Signifkantly different values.

299 points must be considered concerning the use of such information. Firstly, such tables can be used only to detect relationships of possible interest which must then be subjected to further analysis; low levels of significance must be treated with extreme caution as some such results could occur by chance when screening a large number of correlations. Secondly, becauseof limiiations of the serum volumes required, or oarasite availabilitv in the caseof wild isolates, only a hmited number of assays for serum inhibition, parasitized cell agglutination and schixont infected cell opsonization was performed in subjects over the ageof 12 to provide a population profile. Therefore it would be inappropriate to over-interpret data relating to the older age group. In general the assayswere positively correlated with parasite density in younger subjects at the cross-sectional survey in August 1984. As expected, most of the assays were negatively correlated wnh packed cell volume (PCV); however, children with higher levels of serum parasite inhibitory activity did have significantly higher PCVs (r= +0.26; P~0.01). To assesspossible relationships between immunological factors measured at the beginning of the transmission season and a subject’s subsequent experience of clinical malaria we examined the trend for each factor in children who subsequently experienced 0, 1 or 22 episodes. We used analysis of covariance with allowance for age (l-2, 3-4, 5-6, 7-8, 9-11 years), mosquito net usage (yes/no>, haemoglobin genotype (AA/AS), and housing cluster (l-6). Results are shown in Table 7. Because the level of 1000 asexual parasites per microlitre in association with a febrile illness could, on the basis of published data, be considered too low (TRAPEet al., 1985) or too high (GREENWOOD et al., 1987), we repeated the analysis using cut off points of any parasitaemia and of 5000 per ~1. The only immunological factor to show a consistent correlation with protection was the titre of antibodies to neo-antigens on the infected erythrocyte surface detected by micro-agglutination. In view of the steep increase of this response with age (Fig. 4), we repeated the analysis in single year age groups. The protective effect remained significant at all levels of parasitaemia. As an alternative we repeated the analysis using logistic regression and the outcome variable y=O (no episodes), or y=l (1 or more episodes) and again allowing for age, mosquito net usage, genotype and housing cluster. Antibodies detected by infected cell agglutination were again seen to have a significantly protective effect (x2=7.76, P=O*OOS).None of the other immunological factors that we assayed appeared to be protective. Discussion We wished to determine which, if any, of a range of assaysof the host response to malaria provide useful information concerning the ‘immune status’ of individuals in an endemic area. Longitudinal monitoring of individual subjects offers a possible way of determining this. A method of longitudinal monitoring of children was originally developed in The Gambia to measure malarial morbidity in a large rural community (GREENWOODet al., 1987). It has proved sensitive in detecting protective effects of impregnated mosquito nets (SNOW et al., 1988), the effects of iron supple-

mentation (A. Smith, personal communication) and the effects of chemoprophylaxis with Maloprim ~GREENWOOD et al.. 1988). In an endemic environment, where even ‘adults may have a cumulative parasite rate approaching 100% (BRUCE-CHWA’M’, 1%3), detetmining whether an illness episode is due to co-existent parasitaemia is extremely difficult. Parasite density alone is not necessarily a good indicator. Although TRAPEet al. (1985) found that the level of 5000 per microlitre gave good discrimination in the Congo, we have found previously in this areaof The Gambia that even very low parasitaemiasmay be causally related to symptoms (GREENWOOD et al:, 1987). By contrast, apparently asymptomatic individuals with high parasitaemiasare often encountered on cross-sectionalsurveys. We based our operational deiInition of malaria as a febrile illness with >lOOO asexual parasitesper microlitre on a combination of a review of 1200documented casesof malaria presenting at the MRC hospital in Fajara and on local clinical experience. We accept that we will have included somefalse positive and false negative episodes. In this regard it is probably reasonabie to regard the detection of the nrotective effect of sickle cell trait and mosquito neis against clinical episodesas a test of the validity of our criteria, asindependent evidence shows that sickle cell trait (WILLCOX et al., 1983) and probably mosquito nets protect against malarial morbidity (BRADLEY et al., 1986). It should be noted that the protective effects of both sickle cell trait and mosquito net usage are less clear cut with respect to malariometric indices measured cross-sectionally. There are several possible explanations for this observation but the implication is that, although cross-sectional malariometric indices may reflect the development of antimalarial immunity at a population level, they may not adequately reflect clinical resistance or immunity at the individual level. The importance of fully allowing for the confounding effect of age in any analysis of the natural history of malaria is well illustrated by our data. Clinical episodesof malaria were most common in the first 3 years of life and thereafter declined. This was mirrored to some extent by the decline in positive parasite densities and in our previously published finding (GREENWOOD et al., 1987) that deaths from malaria in this population ceasearound the age of 5 years. However, spleen rates and parasite rates continued to rise despite this evidence of increasing clinical resistanceand did not peak until between the ages of 7 and 10 years. One possible explanation of this pattern is that younger children were receiving more curative chemotherapy; we think that this is unlikelv to have been true in Auaust 1984 as chloro&ine availability and usage w&e extremely limited. Whatever is the case it again illustrates the point that ‘immunity’ to malaria, as reflected by changes in malariometric indices at the population level, is not synonymous with ‘immunity’ to clinical disease. The immunological responsesthat we measured were all strongly age-dependent and it is important to avoid over-interpretation of apparent relationships between putative protective responses and indices of malarial parasitixation or illness. In our analysis of both cross-sectional and longitudinal data several apparently significant relationships disappeared after fully correcting for the effect of age.

300 The use of methods employing crude blood-stage antigens to measure the humoral anti-malarial responsedoes not allow the differentiation of protective responsesfrom those that merely reflect cumulative exposure. At the August 1984 survey there was a strong correlation at the individual level between parasite density in children under 12 years, and total anti-malarial antibody levels measured by ELISA or immunofluorescent assay.At the population level the responseflattened between 10 and 15years of age, and in older subjects there was not a significant correlation with parasitaemia. It is generally assumed that if humoral mechanisms are important in protecting against malaria, they must be hidden in the mass response detected by the use of crude antigen preparations. We therefore examined a number of specific aspectsof the antimalarial immune response for which in vitro assays are available. A novel antigenic modification of recently invaded red blood cells was independently described by PERLMANNet al. (1984) and by COPPELer al. (1984). Naturally occurring anti-Pf 155 antibodies inhibit parasite growth in vitro and several lines of evidence have suggestedthat this responsemay be involved in naturally acquired immunity to malaria (WAHLIN et al., 1984). In the population that we studied there was no evidence from either cross-sectionalor longitudinal data to suggest that the anti-Pf 155 response as measured by modified immunofluorescence was protective against malaria. We emphasizethat our results do not lead to the conclusion that the anti-Pf 155 response itself is unimportant, because the method used to measure the response is relatively crude and protective responses to the Pf 155 molecule in non-human primates is epitope-specific (COLLINSet al., 1986). Further studies will be required to dissect the naturally occurring response to Pf 155. Although the relative importance of cellular and humoral immune responses to malaria parasites remains to be clarified it has long been evident that resistanceto several malarial speciescan be produced by the passive transfer of immune serum (COHENet al., 1961). Many workers have shown inhibitory effects of serum or gamma globulin from immune monkeys or humans on P. falciparum growth in z&o, but more recent evidence castsdoubt on the relevance of such in vitro assays: in vitro inhibition did not correlate at all with immunity to P. falciparum induced in the saimiri monkey, or with the ability of serum to transfer immunity passively (FANDEURet al., 1984). Moreover, purified IgG from functionally immune adults living in Papua New Guinea was often not inhibitory to parasite growth and in 25% of cases actually promoted increased growth (BROWNet al., 1982). The situation is further complicated by the fact that serum inhibition may show isolate specificity (WILSON& PHILLIPS, 1976; PATARAPOTIKULet al., 1983). It will be clear from the foregoing that there are considerable difficulties in interpreting the significance of in vitro assays of parasite inhibition by serum. We argued that, despite such problems, if inhibition in vitro reflects to any degree processes which are involved in maintaining naturally developed immunity, then this should at least be reflected in the age-related inhibitory abilities of serum from subjects who belong to a single defined population when tested against a parasite which they are likely to encounter in viva.

The development of serum inhibition showed a completely different pattern to that of the other immune responses which we measured (Fig. 4). Inhibitory responsesdeveloped rapidly with age but then declined. Although a few ‘immune’ adults did have markedly inhibitory sera, the growth inhibitory properties of sera from Gambian adults as a group were not signiticantly different from those of nonimmune Europeans. There are a number of (not mutually exclusive) possible explanations for the unusual population picture that we observed. Firstly, it may be that the parasite isolate used in our assays was one that had been common for a limited period in the community and that many older subjects had not been exposed to it. Secondly, it may be that serum inhibition assaysmeasure a rather primitive response which is subsequently superceded by more effective immune responses. -Finally, it may be that the inhibitors effects of sera from older subiects are masked in vitro by the presence of anti-idiotypic antibodies or blocking antibodies. Some support for the last possibility is given by the observation that B cells from Gambian subjects whose serum is not inhibitory can sometimesproduce inhibitory antibody when transformed by Epstein-Barr virus in vitro (unpublished observation). There was no clear evidence that the inhibitory responsesthat we detected were protective against malaria in vivo; if isolatespecific inhibitory responsesare important it will be necessaryto study responsesto multiple isolates in a closely monitored population. Sporozoites and merozoites are the only parasite stagesexposed directly to the host immune system, and this, combined with their relative fragility and their obligatory requirement to enter to host cells, has led to a concentration on them as potential targets for anti-parasite immunity. However, a third potential target is the membrane of the infected host erythrocyte, which undergoes major morphological, functional and antigenic changes on parasitization. Parasite-dependent red blood cell neo-antigens (PDNs) are exposed to the host’s immune system for about half of each erythrocytic cvcle (HOWARD,1988). The length of exposure, and the fact that in a number of species PDNs exhibit clonal antiaenic variation (BROWN & BROWN, 1965; MCLEAN et al., 1986; HANDUNE?TIet al., 1987), are strong arguments for their potential role asinducers of, and targets for, host anti-parasite responses. Study of the human response to PDNs of P. falciparum is made difficult by the fact that culture in vitro is often associated with loss of expression of these antigens (MARSHet al., 1986). We therefore used a single wild isolate of P. falciparum in our assays. The wide diversity of PDNs in wild parasites (MARSH& HOWARD,1986) presents problems similar to those discussed with regard to serum inhibition of parasite multiplication; again., we have argued that if a response is important it will be reflected in the age pattern of responsesin a single population when tested against a parasite which it is likely to encounter in vivo. The development of the humoral response to PDNs showed a quite distinct pattern; the development of seropositivity was delayed in the first few years of life compared with the anti-Pf 155 response or measurements of total antimalarial antibody. However, between the ages of about 3 and 8 years the development of antibodies is rapid and the response levelled several years earlier

301 than the total asexual response, and 20 years earlier than the anti-sporozoite response (MARSH et al., 1988). There was no evidence from the August 1984 cross-sectional survey that antibodies were protective against parasitaemia per se, but initial analysis of 1Ongirudinal data indicated a strong negative relationship with clinical episodes. As expected, age is an important confounding variable. Rather surprisingly, haemoglobin genotype was also a confounding variable and mean log titres of anti-PDN antibody were significantly higher in sickle trait carriers (2.74) than in normal subjects (1.80, PcO.05); this interesting finding deserves further study. When analysis of covariance was performed allowing for factors identified as possibly confounding (age, genotvpe, mosthe broiective quito net usage; and housi&&t&] effect of anti-PDN antibodies was still evident. The question arises whether we were measuring responses specific to the wild isolate used in the assay or responses to epitopes shared by different isolates (MARSH& HOWARD,1986). In the absenceof clear characterization of PDNs at a molecular level this remains a moot point. However, the fact that high titre sera were similarly reactive with other parasite isolates (data not shown) indicates that our results are unlikely to represent idiosyncratic responses to a single isolate. Antibodies to PDNs could promote a number of possible anti-parasite effector mechanisms including complement-mediated lysis (GABRIEL & BERZINS, 1983), antibody-dependent cytotoxicity (BROWN& SMALLEY,1980), opsonixation (CELADAet al., 1982, 1983) and prevention of infected cell sequestration (UDEINYA et al.. 1983). Levels of anti-PDN antibodies measured’bv micro-agglutination show a correlation with the ability ofsera to interfere with infected red blood cell (IRBC) bindine in the C32 melanoma model of parasitizedcell sequestration (K. Marsh, unpublished) but we have not been able to resolve technical difficulties in scaling this assaydown to allow us to examine large numbers of sera. We initially attempted to assaythe opsonizing activitv of sera using peripheral blood monocytes (C~LDAetb., 1983). but found that attachment of (IRBCsl to mon&cytes was high even in the absence of serum. This result is explained by the presenceon monocytes of the platelet giycoprotein IV?hrombospondin cbmdex. which functions as a receotor for IRBC bindina io a’ range of cells (BARNWELLet al., 1986). Wi therefore subsequently used peripheral blood polymorphs as phagocytic cells in the opsonixing assay. The overall age-relatedresponsewas similar to that of the micro-agglutination assay and there was a significant correlation between the 2 assays.There was no evidence of protection against clinical illness. However, we cannot confidently exclude this possibility for the following reasons: firstly it was evident that we detected opsonixing factors other than IgG, as a number of children’s sera showed opsonizing activity after absorption with protein A (data not shown). IgM antibodies reactive to the IRBC surface can be detected onlv rarelv (MARSH et al.. 19861.but it is possible that other ‘factors such ‘as acute phase proteins are capable of opsonixing IRBCs. Secondly, serawith hiaher levels of anti-PDN antibodies caused considerabl~agglutination of target cells, reducing the effective target cell ratio and probably leading to

artificially low results in this group. There is a need to develop better assays for the assessmentin vizro of infected cell opsonixation. Malarial parasitixation in endemic areas is an exceedingly complex phenomenon. We wish to stress that the analysis of inter-correlated data generated by Studies of this phenomenon requires great caution. Results, both positive and negative, are best regarded as indicating directions for further investigation rather than providing definitive answers. The relationship with ageof practically all important variables concerned with malaria provides an example of problems which may be encountered, failure to allow adequately for which may result in spurious apparent relationships between otherwise unrelated variables. Against this background of caution we draw a number of conclusions from this sNdv . Firstlv . the differences between analysing malarhl parasiti&ion in terms of cross-sectional population indices and aCNai clinical illness are well illustrated. Our casedetection system is by no means perfect and there is scope for further refinement of both field techniques and parasitological criteria. Nevertheless, the ability of the technique to detect the protective effects of a number of factors in this and other studies leads us to suggestthat it should be applied in other endemic areas. It is particularly important that future intervention studies, such as chemoprophylaxis or vaccine trials, include an assessment of protection for morbidity. Secondly, this study provides some reasonably clear-cut negative answers in the search for in vitro assaysof mahuial immune status. As expected, crude tests of the total humoral anti-malarial response give no useful information on individual immune status. Previously described assays to detect anti-Pf 155 antibody and schixont-infected cell opsonixing antibodies were uninformative, though we stress that these conclusions apply to the assaysthemselves and not necessarily to the underlying factors beiig measured. We have previously reported that non-antibody crisis form factor did not play a role in this community (htARSH et al., 1987), but conclusions with regard to other serum inhibitory factors are less clear-cut. Our view that the humoral anti-sporoxoite response was not an important factor in clinical immunity in this population has also been reported previously (MARSH et al., 1988). Finally, we have identified the response to neoantigens on the infected red cells as warranting further investigation. The kinetics of the responseat a population level and evidence of protection at the individual level indicate the potential importance of this host defencemechanism. To our knowledge, this is the Srst defined aspectof the human host’s immune response to malaria which has been shown to be associatedwith protection from clinical malaria. The previously described antigenic diversity of PDNs, and the possibility that they may undergo antigenic variation, combine to suggest that much better characterization of these antigens is necessary to facilitate the extension of this work in further field studies. This study has a number of deficiencies, including the relatively small study woulation in relation to the complexity of the problems -andthe fact that we have examined only a limited range of potential protective factors. In particular, we have not included any

302

studies on cell mediated immunity. This is not because we feel that other aspects of immunity are unimportant but rather reflects the lack of suitable techniques for application in field studies using relatively small samplesof blood. We hope that such methods will be developed and that the type of study described here will be viewed as a prototype for larger and more comprehensive future investigations. The need for such studies is apparent when it is realized that we may soon be in a position of having to test anti-malarial vaccines without being able to monitor their effect on naturally acquired immunity to the disease.

fakipamm in the Saimiri monkey. 3oumal of Immutwlogv,

Lance& ii, 204207.

132, 432437. Gabriel, J. & Berzins, K. (1983). Specificlysis of Plasntodium voelii infected mouse ervthrocvtes with antibodv and complement. Clinical and &per&ma1 Zmmunolo&, 52, 129-134. Greenwood, B. M., Bradley, A. K., Greenwood, A. M., Byass, P., Jammeh, K., Marsh, K., Tulloch, S., Oldfield, F. S. & Hayes, R. (1987). Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene, 81, 478-486. Greenwood, B. M., Greenwood, A. M., Bradley, A. K., Snow, R. W., Byass, P., Hayes, R. J. & NiJie, A. B. H. (1988). A comparison of two drug strategies for the control of malaria within a primary health care programme in The Gambia, West Africa. Lancet? ii, 1121-1127. Handunnetti, S. M., Mendis, K. N. & David P. H. (1987). Antigenic variation of cloned Plasmodium fragile in its natural host Macaca sinica. Journal of Experimental Medicine, 165, 1269-1283. Howard, R. J. (1988). Malarial proteins at the membrane of Ptizsmadium fakiparum infected erythrocytes and their involvement in cytoadherence to endothelial cells. In: Malaria Immunology, Perhnann, P. 81 Wizzell, L. H. (editors). Progress in Allergv, pp. 98-147. Marsh, K. & Howard, R. J. (1986). Antigens induced on erythrocytes by natural Plasmodium fakiparum infections: expression of antigenically diverse and conserved determinants. Science, 231, 150-153. Marsh, K., Sherwood,J. & Howard, R. J. (1986).Parasite infected cell agglutination and indirect immunofluorescent assavs for detection of human serum antibodies bound to* antigens on Plasmodium falcipamm infected erythrocytes. Journal of Zmmunological Methods, 917,

al and Experbnental Immunology, 41, 423-429.

Marsh, K., Otoo, L. N. & Greenwood? B. M. (1987). Absence of crisis form factor in subjects immune to Plasnwdium fakipamm in The Gambia, West Africa. Transactions of the Royal Socieryof Tropical Medicine and Hygiene, 81, 514515.

Acknowledgements We thank Dr A. K. Bradlev. .< Mr B. N’deve. -, ~: Mr P. Camara,Dr R Snow, Mr I. SambouandMr H. Robinsonfor their invaluable assistance.The study would not have been possible without the active co-owration of the residents of

Ihe Kataba villages. This investigationreceivedfinancial support from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases.

References Barnwell, J. W., Ockenhouse, C. F. & Knowles, D. M. (1985). Monoclonal antibody OKMS inhibits the in-&o binding of Plasmodium falcipanmr-infected erythrocytes to monocytes, endothelial, and C32 melanoma cells. 7ouma1 of Immurwloev. 135. 3484-3497. Bra&ey, A. .K., Greenw&d, g. M., Greenwood, A. M.,

Marsh, K., Byasq,P., Tulloch, S. & Hayes,R. (1986). Bed nets (mosqmto nets) and morbidity from malaria.

Brown, K. N. & Brown. I. N. (1965). Immunitv fo malaria: antigenic variation h chro& in&ions of~l?&nwdium kwzulesi. Nature, 208, 1286-1288. Brown, J. & Smalley, M. E. (1980). Specific antibodydependent cellular cytotoxicity in human malaria. ClinicBrown, G. V., Anders, R. F., Mitchell, G. F. & Heywood, P. F. (1982). Target antigens of purified human immunoglobulins which inhibit growth of Plasmodium falciparum in vitro. Nature, 297, 591-593.

Bruce-Chwatt,L. J. (1963).A longitudinalsurveyof natural malaria in a group of West African adults. Part I. West

African Medical 3ouma1, 12, 141-173.

Celada, A., Cruchaud, A. & Perrin, L. H. (1982). Opsonic activity of humanimmune serum on in vitro phagocytosis of Plasmodium falcipancm infected red blood cells by F30Gpes.

Clinical and Experimental Immunology, 47,

Celada?R., Cruchaud, A. & Perrin, L. H. (1983). Phagocytosls of Plasmodiumfakiparum parasitized red blood cells by human polymorphonucleti leucocytes. 3oumal of Parasitology, 69, 49-53. Cohen, S., McGregor, I. A. & Carrington, S. P. (1961). Gamma globulin and acquired immunity to human malaria. Nature, 192, 733-737. Collins, W. E., Anders, R. F., Pappaioanou, M., Campbell, G. H., Brown, G. V., Kemp, D. J., Coppel, R. L., Skinner, J. C., Andrysiak, P. M., Favaloro, J. M., Corcoran, L. M., Broderson, J. R., Mitchell, G. F. & Campbell, C. C. (1986). Immunization of Aotus monkeys with recombinant proteins of an erythrocyte surface antigen of Plasmodium fakiparum. Nature, 323,259-262. Coppel, R. L., Cowman, A. F., Anders, R. F., Bianco, A. E., Saint, R. B., Lmgelbach, K. R., Kemp, D. J. & Brown, G. V. (1984). Immune sera recognise on erythrocytes a Plasmodiumfakiparum antigen composed of repeated amino acid sequences.Nature, 310,78%791. Fandeur, T., Dubois, P., Gysin, J., Dedet, J. P. & Pereira da Silva, L. (1984). In vitro and in vivo Stud& on protective atid inhibitory antibodies against Plasmodium

1n7-115. __ ___

Marsh,K., Hayes,R. J., Carson,D. C., Otoo, L., Shenton, F., Byass, P,, Zavala, F. & Greenwood, B. M. (1988). Anti-sporozoae antibodies and immunity to malaria in a rural Gambian population. Transactions of the Royal Society of Tropical Medicine and Hygiene, 82, 532-537. McGregor, I. A. (1986). The development and maintenance of immunity fo malaria in highly endemic areas.Clinics in Tropical Medicine and Communicable Diseases, 1, 29-53. McLean, S. A., Pearson, C. D. & Phillips, R. S. (1986). Antigenic variation in Plasmodium chubaudi: analysis of parent and variant populations by cloning. Parasite Itiamunology, 8, 415-424. Otoo, L. N., Snow, R. W., Menon, A., Byass, P. & Greenwood, B. M. (1988). Immunitv to malaria in vouna Gambian children Bfter ‘a two ye& period of chemoprophylaxis. Transactions of the Royal Society of Tropical Medicine and Hygiene, 82, S-65.

Patarapotikul, J., Tharavanij, S. & Poonthong, C. (1983). Multiple strains of Plasmodium falciparum are necessary for the growth inhibition assay.South-East AsianJoumal of Tropical Medicine and Public Health, 14, 149-153. Perlmann, H., Berzins, K., Wahlgren, M., Carlsson, J., Bjorkman, A., Patarroyo, M. E. & Perlmann, P. (1984). Antibodies in malarial sera to parasite antigens in the membrane of erythrocytes infected with early asexual stagesof Plasmodium falciparum. Journal of Experimental Medicine, 159, 16861704. Snow, R. W., Lindsay, S. W., Hayes, R. J., Greenwood, B. M. (1988). Permethrin-treated bed nefs (mosquito nets) prevent malaria in Gambian children. Transactionsof the Royal Society of Tropical Medicine and Hygiene, 82, 838-842. Trager, W. & Jensen, J. (1976). Human malaria parasites in continuous culture. Science, 193, 673-675.

303 Trape, J. F. (1985).Rapid evaluationof malaria parasite density and standardizationof thick smearexamination for epidemiological+zsti@ons. Trunsu+ms of the T8yl Soctetyof Tropwal Medtctne and Hypne, 79,181TraPe,J. F., Perlmsan,P. 81Morault-Peelman,B. (1985). Criteria for diagnosm8clinical malariaamon8a semiimmune population exposed to intense and perennial transmission.Transactionsof the Royal Societyof Tropical Medicine and Hygiene, 79, 435-442.

Udeinya,I. J., Miller, L. H., McGregor,I. A. & Jensen,J. B. (1983).Plosmodium fakiparm strain specific antibody blocks binding of infected erythrocytesto amelanotic melanomacells. Nature, 303, 42w31. Wahlin, B., Wahlgren, M., Perlmann,H., Ben&s, K., Bjorlunan,A., Patsrroyo,M. E. & Perlmann,P. (1984).

TRANSACTIONS OF THE ROYAL SOCIETY OF TROPKAL

1 BookReview

Humanantibodies to a M, 155,000F%snwdium falciparurn antigen efficiently inhibit merozoiteinvasion. Proceedingsof the National Academy of Sciences,USA, 81, 7912-7916. Willcox, M., Bjorkman,A., Brohalt, J., Pehrson,P. O., Rombo, L. & Bengtsson, E. (1983). A case-controlstudy in northern Liberia of Plasmodiumfakiparum malaria in haemoglobinS and @halassaemiatraits. Amtuls of Tropical Medicine and Parasiwlogy, 77, 239-246.

Wilson,,R. J. M. & Phillips, R. S. (1976).Method to test inhibitory antibodiesin humanserato wild Populations of Plasmodiumfalciparum. Nature, 263, 132-134. Received 26 September 1988; acceptedfor publication 27 October 1988

MEDICINE AND HYGIENE (1989) 83, 000-000

1

Medicine and Health in Africa: a bibliography witb critical abstracts 1984. London: Bureau of Hvaiene

and Tropical Diseasesand CAB International,-f988: xii + 527~~. + index. Price: E20. A vast amount of information has by now been carefully documented concerning the medical problems which face the varied populations of the great continent of Africa; however, a major difficulty facing the primary health care worker, practising doctor, research scientist and even the administrator or interested observer, exists in the tracing of relevant facts from these accumulated data. With generous support (unspecified) from UNICEF, C. A. Brown of the- Bureau- of Hygiene and Tropical Diseases (BHTD) and 1. P. StanIield of the African Medicine and Research-Foundation, Nairobi, have brought together nealy 2000 abstracts relevant to medicin;in Africa which were nublished in the Trohcal Diseases Bulletin during 198’4(most of these repiesent papers which originated in 1983); some are straight factual accounts of published articles, while others are detailed critical assessments.The maior criterion for selection was that the original article had some bearing (direct or indirect) on African health and/or disease investigations did’not necessarily have to be carried out locally. The entries are broadly classified under subject or diseaseentity; there are also subject and author indexes (both of which are comprehensive), and a limited geographical one (which covers

the more important field studies); it is therefore relatively easy to trace abstracts relating to required topics. Attention is drawn in the introduction to the fact that the last attempt to bring together nubhshed work from sub-Saharan-Africa was probably in 1972 with ‘Medicine in a Tronical Environment which emanated from Makerere’ University, Uganda, during the ‘golden age’. The present book, which has been cheaply produced in an impressive paper-backed format, is therefore to be warmly welcomed. But why 1984? This happens to be the year in which the BHTD started storing data, including abstracts, by electronic means, and an exercise such as this became for the first ume relatively straightforward. It is absolutely essential, of course, that the editors keep it going; this volume represents a mere ‘one-year’s worth’ of published work. Will the benefactor be able to finance an annual follow-up? It is certainly to be hoped so, for a single isolated volume, such as this, is of only limited value. It is certainly the editors’ expressed intention to launch a second collection covering 1985-1986. And thev would annarentlv value sutraestions about making future vol&es even more u&&l. As clearly statedin the introduction, however, success of the venture depends upon the overall readership response; much work has gone into this first volume and it certainly deservesenthusiastic support not only from libraries, but also from those involved in many different disciplines and who are either concerned wivifiaor merely interested in, the health problems of G. C. Cook