534

CEREBRAL ANAEROBIC GLYCOLYSIS ANDREDUCED CEREBRAL OXYGEN TRANSPORT

IN HUMAN CEREBRAL MALARIA

D. A. WARRELL1,2N. VEALL3

P. CHANTHAVANICH1JUNTRA KARBWANG1

N. J. WHITE1,2S. LOOAREESUWAN1R. E. PHILLIPS1,2

PRANEET PONGPAEW1

Faculty of Tropical Medicine, Mahidol University, Bangkok 10400,Thailand;1 Nuffield Department of Clinical Medicine, University of

Oxford,2 and Radioisotopes Division, MRC Clinical ResearchCentre, Harrow, Middlesex3

Summary In 12 patients comatose with cerebral

malaria, cerebral blood flow was 52·2 (SE4·0) ml/100 g per min, within the reported range for healthycontrols, but cerebral vascular resistance was raised at 1·66

(0·19) mm Hg/ml per 100 g per min. Cerebral oxygenconsumption (1·90 [0·23] ml/100 g per min), and cerebralarteriovenous oxygen content difference (3·5 [0·43] ml/dl)were subnormal, while cerebral venous pO2 (5·7 [0·2] kpA)was raised. After recovery of consciousness there were

significant decreases in arterial lactate concentration (2·44[0·45] to 1·19 [0·45] &mgr;mol/l) and cerebral lactate production(17·4 [7·9] to 5·6 [1·1] mmol/100 g per minute). Theseresults provide evidence of cerebral anaerobic glycolysisassociated with inadequate oxygen delivery to the brainconsistent with either inhibition of cerebral oxidativemetabolism or the microcirculatory obstruction envisagedin the "mechanical" hypothesis for cerebral malaria.

Introduction

CEREBRAL malaria is the most important severe

manifestation of Plasmodium falciparum infection and inThailand carries a mortality of about 20% despite the good

facilities in provincial hospitals. 1 Ignorance of the

underlying mechanism has been a major obstacle to

improving management.The "mechanical" hypothesis is currently the most

compelling explanation for cerebral malaria. Blood flowthrough the brain is thought to be impeded by the sticking("cytoadherence") of parasitised erythrocytes (PE) to

venular endothelium by a specific interaction between amalarial adhesion present on knob-like protuberances on thesurface of PE and an endothelial Uganda possibly associatedwith thrombospondin.3 Intererythrocytic adhesion anddecreased deformability of PE may also contribute to

mechanical obstruction in the cerebral microcirculation.4 Infatal cases of cerebral malaria, choking of cerebral venulesand capillaries by erythrocytes containing mature parasiteshas been recognised since the nineteenth century. Duringlife, a critical reduction in cerebral blood flow and hence"stagnant anoxaemia" might be the cause of coma. Thesequestration and tight packing of PE in cerebral vesselspredicted by the mechanical hypothesis has been found inpatients who died with cerebral malaria.6 Other predictableconsequences of cytoadherence are reduction in delivery ofoxygen to the brain and the development of cerebral hypoxiawith consequent anaerobic glycolysis. We have been able todemonstrate these changes in patients with cerebral malaria.

Patients and Methods

Cerebral malaria was defined as unrousable coma with non-

localising or absent motor response to painful stimulation in

patients with asexual forms of Plasmodium falciparum in theirperipheral blood. Other causes of encephalopathy were excluded.’Patients were admitted to the intensive care unit of Pra PokklaoProvincial Hospital, Chantaburi, eastern Thailand. Relatives gaveinformed consent to the study, which had been approved by theethical committee of the Faculty of Tropical Medicine, MahidolUniversity. History was obtained from accompanying relatives, and

TABLE I-CLINICAL DETAILS OF 12 MEN WITH CEREBRAL MALARIA

*Fatal case complicated by hypotension, hypoglycaemia, lactic acidosis, gram-negative septicaemia. Died of intractable hypotension 5 h after admission.t0=rousable to full consciousness; I = impaired consciousness but purposeful response to stimuli; II = unrousable, motor response non-localising;III = unresponsive, tendon reflexes intact; IV = unresponsive, areflexic.

535

Fig 1-Values for cerebral blood flow (CBF) in 12 patients(numbered 1-12 as in table i) with cerebral malaria while theywere comatose breathing air or 95% oxygen + 5% CO2 and afterthey became rousable.

Values are means of two or three measurements made during a period of2-3 h while the patients’ clinical condition was stable. (Thick horizontal barsare mean values.) Case 3 was fatal. There is no significant difference in CBFvalues in the three clinical states.

a full physical examination and routine laboratory investigationswere conducted as described previously.1 Quinine treatment wasstarted by intravenous infusion as soon as the diagnosis wasestablished. Intensive care procedures included monitoring ofbrachial arterial and internal jugular venous pressures (throughindwelling catheters) and arterial blood gas tensions. The jugularvenous catheter was inserted by a percutaneous Seldingertechnique.’ Cerebral blood flow and metabolism were measuredwhile the patients, initially comatose and unsedated, lay supine inbed. Cerebral venous blood was sampled by advancing the internaljugular venous catheter until its tip was in the jugular bulb.

Measurement of Cerebral Blood Flow

The Kety-Schmidt technique8 was modified by replacingcontinuous administration of the tracer substance by a single initialintravenous pulse injection,9 and frequent sampling of arterial andcerebral venous blood by continuous sampling.10 The theoreticalbasis for the former modification has been explored by Veall andGibbs," and by Chonacky and Thomas.12 The modified techniquewas validated in piglets by comparison with indicator fractionation(radioactive microsphere technique)13,14 and in patients underintensive care it was compared with the original nitrous oxidemethod.15 Approximately 700 KBq of tritiated water (AmershamInternational code TRS8) was injected intravenously. A doublesyringe pump (Braun Perfusor VI) withdrew integrated arterial andcerebral venous samples simultaneously into identical heparinisedglass syringes through identical arterial and internal jugular venouscatheters and connecting tubing at about 2 ml/min for an exfusiontime of 3-7 min depending on the expected flow rate, the longerperiod being required at low flow rates to achieve equilibrium. Atthe end of the sampling period venous and arterial spot sampleswere withdrawn immediately to check equilibration of isotope andfor measurement of blood gases and pH. If previous measurementshad been made a preinjection venous sample was withdrawn toserve as a blank.

An estimate of the cerebral blood flow (ml/ 100 g per min) isobtained from the formula:

CBF, ml/100 g per min =100(VT-VB)&lgr;&lgr;_ _

(A - V)t(A, V, VB, and VT are radioactivities of the integrated arterial.integrated venous, preinjection, and terminal venous samples,respectively, in arbitrary units. t is the integrating time in minutesand ^ the tissue/blood partition coefficient.) 3H20 &lgr; is 0-95 at flowrates less than 30 ml/100 g per min. At higher flow rates the singlepassage extraction fraction, E for 3H20=1.11-0.0037 CBF. 16CBF values in this range are corrected by assuming that the bloodvolume is about 45% of the brain weight 17 and calculating the"effective" value of , thus: &lgr;EFF = 0.95E + 0-045 (1-E).

ß-emission from pure water samples prepared by vacuumdistillation18 was measured by means of an LKB-Wallac (Type81 000) liquid scintillation spectrometer. To minimise statisticalerrors, 40 000 counts were recorded or a counting period of 20min was used; and the mean value of four or more suchmeasurements was taken. Cerebral vascular resistance (CVR) wascalculated as the ratio of cerebral perfusion pressure (meanarterial pressure-mean jugular venous pressure) to cerebralblood flow.

Metabolic Measurements

Blood for measurement of arterial (brachial) and cerebral venous(jugular bulb) p02, pC02, pH oxygen content, and haematocritwas drawn anaerobically into syringes lubricated with heparin(1000 units per ml) which were capped, mixed by rotation, andstored in iced water until analysed. Blood gas tensions and pHwere measured in triplicate at 37°C by use of Radiometerelectrodes (BME33/BGA 3 system). Gas partial pressures andpH were corrected to the patient’s core (rectal) temperature."Blood total oxygen content was measured in triplicate by thedirect galvanic cell method (Lex O2 Con-K, LexingtonInstruments Corporation).20 Oxygen extraction ratio (OER) andcerebral oxygen transport (COT) were calculated as follows:

OER= CMRO2

arterial O2 content x CBFCerebral arteriovenous O2 content difference

arterial O2 contentCOT = CBF x arterial O2 content ml/ 100 g per min.

Glucose (hexokinase method, ’Glucose Rapid Test’, Roche) andlactate (fully enzymatic UV method, Boehringer Mannheim)concentrations were measured in parts of the integrated arterial andcerebral venous blood samples drawn for cerebral blood flowmeasurement and taken into fluoride oxalate and perchloric acid,respectively. Measurements were repeated within 2-3 h while thepatients were breathing oxygen or 95 % oxygen + 5 % CO2, As soonas they became rousable, measurements were repeated while theybreathed air. Intravascular catheters were then removed. PairedStudent’s t-test (2 tailed) was used for statistical comparisons..

Results

Clinical (table I)

The 12 patients in this study resembled the others in ourseries of cases of cerebral malaria. 9 of them had had a

generalised convulsion 7-57 (mean 17-5) h before the firstmeasurement of CBF. Only 1 patient died. The other 11recovered fully without neurological sequelae and weredischarged from hospital after 7-28 (mean 12-4) days. Inno case did the breathing of oxygen + 5 % CO2 improvethe level of consciousness.

Cerebral Blood Flows (fig 1, table II)The difference between CBFs repeated within 2-3 h

while the patient’s clinical state was stable, on twelveoccasions, was 50 (SD 35) ml/100 g brain per min. CBFand cerebral erythrocyte flow (CBF x haematocrit) values

536

TABLE II-CEREBRAL BLOOD FLOW AND METABOLISM IN PATIENTS WITH CEREBRAL MALARIA

*p=0.011; tp = 0.036;‡p = 0.015. Values are mean (SE); no of cases.

while the patients were comatose breathing air or 95%oxygen+5% CO2 and those measured when the patientsbecame rousable were not significantly different. CBFvalues were within the normal range for healthy youngadults, (table II)8,9,17,21,22 but cerebral erythrocyte flowswere strikingly reduced in all cases.CBF was responsive to changes in PaC02 over the normal

range (fig 2) correlation r = 0-75, p = 0 02, df= 9.CVR values were higher than in healthy normal subjects

and fell slightly, but not significantly, when patientsbreathed 5 % CO2 + oxygen and when they recoveredconsciousness.

Arterial pCO2, haematocrit, plasma fibrinogenconcentration, and rectal temperature showed no

correlation with CBF.

Fig 2-Effect of changes in arterial pC02 on cerebral blood flow(CBF) in 9 patients with cerebral malaria.

In some patients CBF was measured first while they breathed air and laterafter they had been switched to 95% oxygen+5% CO2, The others werestarted on oxygen + CO2 (for clinical reasons) and later switched to air. Theresults indicate responsiveness of CBF to changes in pC02 (r = 0-75, p = 0-02,df = 9, slope of regression line = 99).

Cerebral arteriovenous 02content differences were lowerthan in normal controls9 and did not change significantlywhen the patients became rousable. Cerebral venous pO 2was generally high (table II).There was a small but significant increase in COT when

patients breathed 95% oxygen+5% CO2-namely, 6-3(SE 0-8) to 7-0 (0-6) ml/100 g per min (p= 0-01)—butthere was no significant change in either COT or OERwhen patients became rousable. OER and CMR02remained low throughout.

Cerebral Metabolism (table II)Coefficients of variation for duplicate measurements of

blood oxygen content, glucose, and lactate were 2-0, 43,and 11-9%, respectively. Mean differences in oxygencontents and CMRO2 values repeated in the same patientwithin 2-3 h were 0-56 (SD 0-76) ml/100 ml (16 pairs) and0-44 (0-65) ml/100 g per min (12 pairs), respectively.Compared with reported values from healthy controls

breathing air and hypoxic mixtures, CMR02 was lowduring both comatose and rousable states and CMR

glucose was increased in comatose patients breathing air.CMR02and CMR glucose did not increase significantly

either when the patients became rousable or when theywere switched from breathing air to oxygen plus 5% CO2while they were comatose.

Cerebral lactate production and arterial lactateconcentration decreased significantly when patientsbecame rousable-17-4 (SD 7-9) compared with 56 (11)µmol/100 g per min (p=0-03) and 244 (045) comparedwith 1 19 (0-45) mmol/1 1 (p = 0-015), respectively.

Discussion

With few exceptions, coma is associated with a reductionin cerebral metabolic rate and cerebral blood flow. This istrue of diabetic, hepatic, uraemic, and traumatic comasand of anaesthesia induced by barbiturates and otherdrugs,21,23 In these conditions there is a failure ofutilisation rather than supply of energy. In our comatosepatients with malaria, cerebral oxygen consumption waswell below the normal range, despite their fever and, in 9, ahistory of recent convulsions. However, recovery of

537

Fig 3-Cerebral blood flow (CBF) values in 12 patients with cerebralmalaria plotted against arterial oxygen content.

Superimposed is the regression line and range of values (shaded area)obtained by Brown et al 12 in a group of 54 normal subjects and patients withvarious haematological conditions causing a range of haemoglobinconcentrations from 5-7 to 19 1 g/dl. (By permission of the authors andpublisher of Brain.)

consciousness was not associated with a significantincrease in CMRO2. Although it is possible that cerebralmalaria is also primarily a metabolic coma, and that thelow oxygen transport to the brain simply reflects reducedrequirements, the increased cerebral lactate productionfound in this study argues in favour of inadequate energysupply. Barbiturate coma is associated with subnormalcerebral lactate content and, unlike cerebral malaria, withreduced glycolysis.z4 Excessive lactate production in ourpatients was due to ischaemia of various tissues, butanother possible source was the metabolism of malariaparasites in erythrocytes sequestered in microvascular

beds, notably in the brain. Mature parasites have anextremely active metabolism and convert 90% of ingestedglucose into lactic acid. From biochemical studies of Pfalciparum cultured in vitro, some workers estimate that athigh parasitaemias (approximately 30% of erythrocytesparasitised) the parasite biomass could produce about 1-2moles of lactate per day-a doubling of the adult basal rateof lactate production.zs In our comatose patients, cerebralblood flows were generally within the range of publishedvalues from healthy controls2l and did not increase

significantly with recovery of consciousness. However,these patients showed alterations in a number of the

physiological determinants of CBF. They were

hypocapnic (mean PaC02 34 mm Hg) and most werehyperfibrinogenaemic (range 300-680 mg/dl)—factorstending to reduce CBF. However, reductions in CAAR02and arterial oxygen content to the levels found in cerebralmalaria override the effect of hypocapnia 26 and

hyperproteinaemia.2’ Unlike some patients withcerebrovascular diseasesZ1 and viral encephalitis,28,29 thecomatose patients with cerebral malaria showed a normalresponse of CBF to changes in PaC02 over the normalrange.Z6 The patients were anaemic (mean haematocrit28%) and febrile (rectal temperatures up to 41.1°C)-

abnormalities that would be expected to increase CBF by15-30%.30,31With so many confounding variables it is helpful to

consider CBF values in relation to transport of oxygen tothe brain (COT) and extraction of oxygen by the brain(OER). In a group of patients with haemoglobinconcentrations varying from 5-7 to 19-1 g/dl, Brown andMarshall found a close inverse relation between arterialoxygen content and CBF.27,32 The patients with cerebralmalaria had low CBFs in relation to arterial oxygencontent while comatose and rousable (fig 3); despite lowarterial oxygen contents CBF was not increased to

maintain COT. The low values of the OER and cerebralarteriovenous oxygen content differences and the highvalues of cerebral venous p02 all suggest that the brain waseither unable to take up or unable to utilise the oxygendelivered. This could be explained in several ways:

sequestered erythrocytes may interfere with oxygen

transport across the capillary wall; there may be areas oflow flow with adjacent regions of the microcirculation inwhich flow is reflexly high ("luxury perfusion");33 or thebrain may be effectively anaesthetised by local metabolicchanges. The concept of luxury perfusion or uneven

distribution of microcirculatory blood flow would explainthe increased cerebral lactate production without a majorreduction in total flow. However, if arterial oxygencontent is the principal determinant of cerebral bloodflow, the values for CBF and oxygen delivery to the brainin our patients with cerebral malaria are unequivocallylow, irrespective of the brain’s ability to utilise oxygen.The explanation for low CBF and relatively high CVRcould be mechanical obstruction of the cerebral circulation

by cytoadherent parasitised erythrocytes as envisaged inthe mechanical hypothesis for cerebral malaria.

We are grateful to the Director (Dr Chaisit Dharakul) and staff at PraPokklao Hospital, Chantaburi, for their enthusiastic cooperation, to ProfTranakchit Harinasuta and Prof Danai Bunnag for their advice,encouragement and support, to Dr M. J. Warrell for help with

organisation, to Mrs Kamolrat Silamut and Mrs Vanapom Wuthiekanunfor excellent technical help, to Mrs Ann Watson for help with calculations,and to Mrs Patchari Prakongpan, Miss Nucharee Cholvilai, and MissEunice Berry for typing the manuscript. This study was part of theWellcome-Mahidol University, Oxford Tropical Medicine Research

Programme, funded by the Wellcome Trust.

Correspondence should be addressed to D. A. W., Nuffield Department ofClinical Medicine, John Radcliffe Hospital, Headington, Oxford OX39DU.

REFERENCES

1. Warrell DA, Looareesuwan S, Warrell MJ, et al. Dexamethasone proves deleterious incerebral malaria. A double-blind trial m 100 comatose patients. N Engl J Med 1982;306: 313-19.

2. Leech JH, Aley SB, Miller IH, Howard RJ. Plasmodium falciparum malaria:cytoadherence of infected erythrocytes to endothelial cells and associated changes inthe erythrocyte membrane. Progr Clin Biol Res 1984; 155: 63-77.

3 Roberts DD, Sherwood JA, Spitalnik SL, et al Thrombospondin binds falciparummalana parasitised erthrocytes and may mediate cytoadherence. Nature 1985; 318:64-66

4. Miller LH, Usami S, Chien S. Alteration in the rheological properties of Plasmodiumknowlesi-infected red cells. A possible mechanism for capillary obstruction. J ClinInvest 1971; 50: 1451-55.

5. Marchiafava E, Bignami A. On summer-autumn malarial fevers (translated from thefirst Italian edition by J. H Thompson). London: New Sydenham Society, 1894.

6. MacPherson GC, Warrell MJ, White NJ, Looareesuwan S, Warrell DA. Humancerebral malana: a quantitative ultra-structural analysis of parasitised erythrocytesequestration. Am J Pathol 1985, 119: 385-101

7. Branthwaite MA, Bradley RD. Measurement of cardiac output by thermal dilution in man. J Appl Physiol 1968, 24: 434-38.

8. Kety SS, Schmidt CF The nitrous oxide method for the quantitative determination ofcerebral blood flow in man: theory, procedure and normal values. J Clin Invest1948; 27: 476-83.

9. Scheinberg P, Stead EA. The cerebral blood flow in male subjects as measured by thenitrous oxide technique. J Clin Invest 1949; 28: 1163-71.

538

10. Lassen NA, Høedt-Rasmussen K. Human cerebral blood flow measured by two inertgas techniques. Circ Res 1966; 19: 681-88.

11. Veall N, Gibbs GP. The accurate determination of tracer clearance rates andequilibrium distribution volumes from single injection plasma measurements usingnumerical analysis. In: Joekes AM, Constable AR, Brown NJG, eds.

Radionuclides in nephrology. New York. Grune and Stratton, 1982: 125-230.12. Chonacky NJ, Thomas FD. Evaluation of plasma clearance by synthesis of continuous

infusion: principles and use J Nucl Med 1986; 27: 1611-22.13. Wootton R, Flecknell PA, John M. Accurate measurements of cerebral metabolism in

the conscious, unrestrained neonatal piglet. I Blood flow. Biol Neonate 1982; 41:209-20.

14. Flecknell PA, Wootton R, John M. Accurate measurement of cerebral metabolism inthe conscious, unrestrained neonatal piglet. II Glucose and oxygen utilization. BiolNeonate 1982; 41: 221-26.

15. Veall N, Bell JA, Warrell DA, Reeve J. Observations on the use of a single injection125I-iodo-antipyrine for studies of cerebral blood flow and metabolism in comatosepatients. Br J Radiol 1975; 48: 505.

16. Eichling JO, Raichle ME, Grubb RL. Evidence of the limitations of water as a freelydiffusible tracer m brain of the rhesus monkey. Circ Res 1974; 35: 358-63.

17. Frackowiak RSJ, Wise RJS, Gibbs JM, Jones T. Positron emission tomographicstudies in aging and cerebrovascular disease at Hammersmith Hospital. Ann Neurol1984, suppl. S112-S118.

18. Hesp R, Smith T, Veall N The use of saliva samples to determine total body water bytritium dilution. Clin Phys Physiol Meas 1982; 3: 155-57

19. Kelman GR, Nunn JF. Computer produced physiological tables for calculationsinvolving the relationships between blood oxygen tension and content. London:Butterworth, 1968.

20. Valeri CR, Zaroulis CG, Narchionni L, Patti KJ. A simple method for measuringoxygen content in blood. J Lab Clin Med 1972; 79: 1035-40.

21. Siesjo BK Brain energy metabolism. Chichester: Wiley, 1978

22. Cohen PJ, Alexander SC, Smith TC, Reivich M, Wollaman H. Effects of hypoxia andnormocarbia on cerebral blood flow and metabolism in conscious man J ApplPhysiol 1967; 23: 183-89.

23 Sokoloff L. Neurophysiology and neurochemistry of coma Exp Biol Med 1971, 4:15-33.

24. Nilsson L, Siesjo BK. Influence of anaesthetics on the balance between production and utilisation of energy in the brain.. N Neurochem 1974; 23: 29-36

25 Zolg JW, Macleod AJ, Scaife JG, Beaudoin RL. The accumulation of lactic acid andits influence on the growth of Plasmodium falciparum in synchronized cultures InVitro 1984; 20: 205-15.

26. Kety SS, Schmidt CF. The effects of altered arterial tensions of carbon dioxide andoxygen on cerebral blood flow and cerebral oxygen consumption of normal youngmen. J Clin Invest 1948; 27: 484-91.

27. Brown MM, Marshall J. Regulation of cerebral blood flow in response to changes inblood viscosity. Lancet 1985; i: 604-09.

28. Shapiro W, Eisenberg S. Pathophysiology of epidemic St Louis encephalitis III

Cerebral blood flow and metabolism. Ann Intern Med 1969; 75: 691-70229. Paulson OB, Brodersaon P, Hansen EL, Kristensen HS. Regional cerebral blood flow,

cerebral metabolic rate of oxygen, and cerebrospinal fluid acid-base variables inpatients with acute meningitis and with acute encephalitis Acta Med Scand 1974.196: 191-98.

30. Paulson OB, Parving HH, Olesen J, Skinhoj E. Influence of carbon monoxide and ofhaemodilution on cerebral blood flow and blood gases in man. J Appl Physiol 1973.35: 111-16

31. Heyman A, Patterson JL, Nicholas FT. The effects of induced fever on cerebralfunctions in neurosyphilis. J Clin Invest 1950; 29: 1335-41

32. Brown MM, Wade JPH, Marshall J. Fundamental importance of arterial oxygencontent and in the regulation of cerebral blood flow in man. Brain 1985; 108: 81-93

33. Lassen NA. The luxury-perfusion syndrome and its possible relation to acutemetabolic acidosis localised within the brain. Lancet 1966; ii: 1113-15

DETECTION OF HTV1 DNA IN INFANTS ANDCHILDREN BY MEANS OF THE POLYMERASE

CHAIN REACTION

F. LAURE1C. ROUZIOUX2

F. VEBER3C. JACOMET2

V. COURGNAUD1S. BLANCHE3M. BURGARD2C. GRISCELLI3

C. BRECHOT1,4

INSERM U75,1 Laboratoire de Microbiologie,2 and Unitéd’Immunologie et Hematologie, INSERM U132,3 Centre

Hospitalier Universitaire Necker; Unité d’Hépatologie, HôpitalLaennec;4 and Laboratoire Hybridotest, Institut Pasteur,5 Paris,

France

Summary The polymerase chain reaction (PCR)assay was used to investigate the possibility

of HIV1 DNA detection in uncultured peripheral bloodmononuclear cells from newborn infants and children ofHIV-infected mothers. HIV1 DNA sequences were

detected in mononuclear cells of six of fourteen symptom-free newborn infants of seropositive mothers. Only one ofthese infants had detectable HIV antigenaemia. In addition,HIV1 DNA was identified in the mononuclear cells of fiveof ten children (2-5 years old) of infected mothers who hadbecome seronegative 12-15 months after birth; amongthese, four children had only mild clinical features related toHIV infection, while the other had none. HIV1 DNA wasshown in all of eight seropositive children with HIVinfection and none of fifteen normal seronegative controls.The PCR assay thus provides an early and directidentification of HIV infection in newborn infants andseronegative children born to infected mothers.

Introduction

PERINATAL transmission of human immunodeficiencyvirus (HIV) from infected mothers is associated with a highrisk of infection and of mortality in the children.l-6 Rates ofHIV infection in newborn infants of 20-60% have been

reported.1-6 However, these studies are of limited valuebecause seropositivity in the newborn infant is not sufficient

evidence for HIV infection; IgG antibodies against HIVcross the placenta and may persist for up to 15 months .3

HIV antigenaemia is generally not detected and IgManti-HIV antibodies have not been reliable markers in thesecircumstances.7,8 In addition, children born to infectedmothers can become seronegative despite the developmentof clinical features suggesting HIV infection? Earlydemonstration of viral infection can be reliably based onlyon the isolation of the virus through cell culture or on thedetection of the viral genome. The sensitivity of standardhybridisation procedures, including in-situ hybridisation, istoo low to detect HIV DNA sequences in uncultured cells,owing to the limited number of infected blood cells.9The polymerase chain reaction assay has been developed

to amplify specifically short HIV1 DNA sequences to a leveldetectable by hybridisation, allowing their identification inuncultured peripheral blood mononuclear cells. 10.11 We haveused this approach to analyse the perinatal transmission ofHIV from infected mothers to their children.

Patients and Methods

Four groups of patients were studied with the informed consentof the parents (see table). Blood samples (3-5 ml) were obtained 2 or3 days after birth from fourteen symptom-free infants bom toHIV-infected mothers; for two infants cord blood samples wereanalysed. The second group consisted of ten children (2-5 years old)born of HIV1-infected mothers who had become seronegative forHIV 10-15 months after birth. Five of them had no symptoms andfive had mild clinical features related to HIV infection: adenopathy,hypergammaglobulinaemia and, in one patient, low numbers ofCD4 lymphocytes. The third group consisted of eight seropositivechildren (2-12 years old) with overt clinical signs of HIVI infection,acquired through blood transfusion in one (12 years old) and frominfected mothers in seven. The control group consisted of fifteen

HIV-seronegative blood donors.Anti-HIV antibodies were tested by enzyme immunoassay and

western blot (Diagnostics Pasteur and Dupont de Nemours). HIVp24 antigenaemia was determined by enzyme immunoassays(Abbott and Diagnostics Pasteur).

Peripheral blood mononuclear cells from patients and healthydonors were separatedd by means of ’Ficoll-hypaque’. DNA wasextracted after treatment of cells in a lysis buffer containing 10mmol/1 "tris"-HCl pH 8, 10 mmol/1 ethylene diamine tetra-acetate