serious adverse events associated with yellow fever...

For personal use. Only reproduce with permission from The Lancet Publishing Group.

ARTICLES

THE LANCET • Vol 358 • July 14, 2001 91

Summary

Background The yellow fever vaccine is regarded as one of thesafest attenuated virus vaccines, with few side-effects oradverse events. We report the occurrence of two fatal cases ofhaemorrhagic fever associated with yellow fever 17DDsubstrain vaccine in Brazil.

Methods We obtained epidemiological, serological, virological,pathological, immunocytochemical, and molecular biologicaldata on the two cases to determine the cause of the illnesses.

Findings The first case, in a 5-year-old white girl, wascharacterised by sudden onset of fever accompanied byheadache, malaise, and vomiting 3 days after receiving yellowfever and measles-mumps-rubella vaccines. Afterwards shedecompensated with icterus and haemorrhagic signs and diedafter a 5-day illness. The second patient—a 22-year-old blackwoman—developed a sore throat and fever accompanied byheadache, myalgia, nausea, and vomiting 4 days after yellowfever vaccination. She then developed icterus, renal failure,and haemorrhagic diathesis, and died after 6 days of illness.Yellow fever virus was recovered in suckling mice and C6/36cells from blood in both cases, as well as from fragments ofliver, spleen, skin, and heart from the first case and from theseand other viscera fragments in case 2. RNA of yellow fever viruswas identical to that previously described for 17D genomicsequences. IgM ELISA tests for yellow fever virus were negative

*Members listed at end of paper

Instituto Evandro Chagas/Fundação Nacional de Saúde (FUNASA),Belém, Brazil (P F C Vasconcelos MD, V L R S Barros MD,S G Rodigues PharmBioc, E S T Rosa PharmBioc, A C R Cruz MSc);Instituto Adolfo Luz, São Paulo, Brazil (T L Coimbra PharmBioc,C L S Santos PhD, I M Rocco PharmBioc, V A F Alves MD,R T M Santos PharmBioc); Universidade de Campinas, Campinas,Brazil (L J Silva MD, C M O Papaiordanou MD, L D Andrade MD);Fundação Oswaldo Cruz, Rio de Janeiro, Brazil (R Galler PhD);CENEPI/FUNASA, Brasília, Brazil (E Lacava MD, Z G Costa MD);Secretaria Municipal de Saúde, Goiânia, Brazil (C Laval MD);Secretaria de Saúde do Estado de São Paulo (H K Sato MD,M F G Vilela MD); Acambis Inc, Cambridge, MA, USA(T P Monath MD); Secretaria Municipal de Saúde, Americana, Brazil(E Lacava MD, L M R Almeida MD); Secretaria Municipal de Saúde,Santa Bárbara, Brazil (G B Froguas MD); and Pan American HealthOrganization, Washington, DC, USA (O F P Oliva MD)

Correspondence to: Dr Pedro F da Costa Vasconcelos, WHOCollaborating Center for Reference and Research on Arbovirus,Department of Arbovirus of the Instituto Evandro Chagas/FUNASA,Av Almirante Barroso 492, 66090-000 Belém, PA, Brazil(e-mail: [email protected])

in case 1 and positive in case 2; similar tests for dengue,hantaviruses, arenaviruses, Leptospira, and hepatitis virusesA–D were negative. Tissue injuries from both patients weretypical of wild-type yellow fever.

Interpretation These serious and hitherto unknowncomplications of yellow fever vaccination are extremely rare,but the safety of yellow fever 17DD vaccine needs to bereviewed. Host factors, probably idiosyncratic reactions, mighthave had a substantial contributed to the unexpected outcome.

Lancet 2001; 358: 91–97See Commentary page 84

IntroductionBrazil, like several other countries in South America, has aregion of extensive Amazonian forest, in which jungleyellow fever is endemic, and a coastal zone, in which yellowfever does not occur.1,2 The boundary between these zonescan be affected by periodic expansions in epizootic activity.The densely populated coastal zone has become reinfestedwith the urban vector of yellow fever virus, Aedes aegypti,and is thus receptive to the introduction and spread ofyellow fever from the endemic area.2

The endemic area includes 12 states in the western two-thirds of the country, which is inhabited by 29·3 millionpeople. Beginning in the first quarter of 1998 andcontinuing to 2000, one of the largest epizootics in historyoccurred, leading to the occurrence of 192 registeredhuman cases of jungle yellow fever, 88 (46%) of which hada fatal outcome.3 The epizootic is believed to be linked toexcessive rainfall caused by the recent El Niño/southernoscillation event.4 The cases of yellow fever occurred over awide area from Roraima along the Venezuelan border in thewest, to Pará state in the north, Minas Gerais and Bahia inthe east, and São Paulo state in the south.3,4 The remainingreceptive, non-endemic coastal zone includes 15 states andis inhabited by 126·3 million people.

Due to the upsurge in yellow fever activity and theincreasing mobility of the human population, cases of jungleyellow fever acquired in the endemic zone have beenexported to cities of the coastal zone including Rio deJaneiro, São Paulo, and Campinas. The risk of urban yellowfever is increased by the expanding distribution and densityof A aegypti, which now includes all 27 states of thecountry.5,6

In response to the increase in epizootic activity and thethreat of urban yellow fever, use of yellow fever vaccine hasincreased strikingly in Brazil. Additionally, a new policy wasestablished in 1998 to include the vaccine in the nationalprogramme of childhood immunisation. Between 1990 and

Serious adverse events associated with yellow fever 17DD vaccinein Brazil: a report of two cases

Pedro F C Vasconcelos, Expedito J Luna, Ricardo Galler, Luiz J Silva, Terezinha L Coimbra, Vera L R S Barros, Thomas P Monath, Sueli G Rodigues, Cristina Laval, Zouraide G Costa, Maria F G Vilela, Cecília L S Santos, Cristina M O Papaiordanou, Venancio A F Alves, Liliana D Andrade, Helena K Sato, Elisabeth S T Rosa, Gustavo B Froguas,Ethel Lacava, Leda M R Almeida, Ana C R Cruz, Iray M Rocco, Raimunda T M Santos, Otávio F P Oliva, and the Brazilian YellowFever Vaccine Evaluation Group*

Articles


2000, about 85 million doses of yellow fever vaccine weregiven across the whole country. Vaccine coverage in 1999was estimated to exceed 90% in most states in the endemiczone. However, theoretical coverage estimated by thenumber of doses as a function of total population does notprovide an accurate picture, because revaccination iscommonplace and populations with easy access to fixedcentres are immunised, whereas pockets of the populationin more rural, remote, and inaccessible areas escapeimmunisation. For this reason, the use of mobilevaccinating teams has recently been reinstated.

The live, attenuated yellow fever 17D virus is one of themost successful vaccines developed to date.7 Its productionin embryonated chicken eggs is well defined and efficient, aseed-lot system is used to control passage level, and thevaccine is subject to strict quality control including monkeyneurovirulence testing of seed viruses. It induces long-lasting immunity in almost 100% of people vaccinated witha single dose, and it is also inexpensive. Its use has beenestimated at more than 400 million doses, and it has anexcellent record of safety. Only 21 cases of post-vaccinalencephalitis have been recorded since the implementationof the seed-lot system in 1945; the rate of encephalitis invery young infants is 0·5–4·0 per 1000 in those youngerthan 9 months, and 1 per 8 million in those 9 months orolder.7,8

Yellow fever vaccines approved by WHO aremanufactured in Brazil, the USA, England, France, Russia,and Senegal. Brazil uses the 17DD substrain, whereas othermanufacturers use 17D. The two substrains representindependent passage lineages from the original 17D virusdeveloped by Theiler and Smith.9 Vaccination with 17DDvirus has been associated with three reports of adversereactions in Brazil, all of which were neurological symptomssimilar to those previously reported after vaccination withother the 17D substrains.7,8 There was one case ofencephalitis and two of paralysis, giving a prevalence of 0·09per million (unpublished data), which is not different fromthe historical prevalence of post-vaccinal encephalitis—anexpected, rare complication of yellow fever vaccination.

Recently, however, two fatal adverse events with anunexpected clinical syndrome occurred during a period ofintensified yellow fever vaccination in Brazil. We report theclinical details of these two cases associated with use of theyellow fever 17DD vaccine virus.

Patients and methodsPatientsCase 1 was a 5-year-old white girl from Goiânia, Goiásstate, who was given simultaneous but separate injections ofyellow fever 17DD (lot 98UFB088Z) and measles-mumps-rubella vaccines on Oct 8, 1999. 3 days later, she developedfever and diarrhoea, and the next day anorexia, vomiting,and vulval pruritis and exudate, at which point she was seenin the emergency room. On examination, she was febrile(temperature 39·6ºC), tachypnoeic, had a hyperaemicpharynx without exudate, and no visceromegaly. She wasanaemic (haematocrit 34%) and had raised serumaminotransferase concentrations (aspartate aminotrans-ferase 114 IU/L, alanine aminotransferase 160 IU/L). Thebilirubin concentration was slightly raised (18·8 �mol/L).The diagnoses of possible hepatitis A and urinary-tractinfection were made.

On the fourth day of illness, she decompensated, withprostration, respiratory distress, increased vomiting anddiarrhoea, dehydration, and scleral icterus. Abdominalexamination revealed an enlarged, tender liver. Possiblemeningismus was noted, and pneumonia and meningitiswere added to the differential diagnosis. The cerebrospinal

fluid examination was normal. Chest radiography revealeddiffuse interstitial infiltrate in the left lung. The haematocritfell to 30% despite dehydration, and the total leucocytecount was 12 000/L (4% bands, 88% neutrophils, 8%lymphocytes). She was hydrated and treated withantibiotics. The next day, she returned to the hospital inextremis and was admitted to intensive care. On admission,she was hypotensive, not perfusing, cyanotic, and in acuterespiratory distress. She had a cardiorespiratory arrest, fromwhich she was resuscitated, but died within the next hour.

The patient had not travelled before the event. A reviewof the past medical history of the patient revealed that shewas a low-birthweight baby, who had had repeated episodesof diarrhoea and bronchitis during childhood. 3 monthsbefore the current illness, she had been admitted to hospitalwith aseptic meningitis. Her parents were in good healthand underwent testing for HIV, with negative results.

Case 2 was a 22-year-old black woman from Americana,São Paulo state, who was vaccinated against yellow fever(lot 995F029Z) on Feb 17, 2000. 4 days later, sheexperienced pain in the left arm (the site of vaccination) andfever. She was seen in the emergency room and analgesicswere prescribed. 3 days after illness onset, she soughtmedical treatment at a clinic in the same city for fever,headache, generalised myalgia, and sore throat. She wasgiven symptomatic treatment. That evening she againsought medical attention at the emergency room;examination revealed intense pharyngitis. She was givenpenicillin and released. The next morning (4 days afteronset) she returned to the clinic complaining of epigastricpain; she was given cimetidine and released. That night, shereturned to the emergency room with complaints of feelingincreasingly ill, fever, and myalgia. She was noted to bejaundiced and was admitted with an initial diagnosis ofhepatitis. On admission, her haematocrit was 43%, white-blood-cell count 9600/L (7% bands, 77% neutrophils, 8%lymphocytes), platelet count 54 000/L, creatinine concen-tration 247·5 �mol/L, aspartate aminotransferase con-centration 430 IU/L, and alanine aminotransferaseconcentration 190 IU/L. The urinalysis showed 1+ protein,2+ urobilinogen, 103 000 white blood cells per L, and 36 000 red blood cells per L. She was treated withantibiotics and released.

The next day (5 days after onset), she again soughtmedical attention and was referred to a regional hospital,where she was admitted into intensive care with a tentativediagnosis of leptospirosis, dengue, or an adverse reaction toyellow fever vaccine. She had noted diminishing urineoutput and oedema of the arms and legs during the 24 hbefore admission. On admission she was acutely ill, alert,icteric, and had oedema of the arms and legs, andhepatomegaly. Her blood pressure was 110/60 mm Hg. Shewas treated with intravenous penicillin, ranitidine,furosamide, mannitol, and dopamine. Her white cell countwas 27 700/L with a striking left-shift, haematocrit 37·7%,platelet count 38 000/L, total bilirubin concentration194·9 �mol/L, aspartate aminotransferase concentration511 IU/L, alanine aminotransferase concentration 91 IU/L,� glutamyl transpeptidase concentration 62 U/L, alkalinephosphatase concentration 530 IU/L, creatine kinaseconcentration 155 IU/L, and creatinine concentration3447·6 �mol/L. She was treated with fresh frozen plasma.

On Feb 27 (6 days after onset), she was transferred to atertiary-care hospital in Campinas city because of continueddeterioration. On admission she was in critical condition, inrespiratory distress, icteric, cyanotic, and oliguric with ablood pressure of 110/50 mm Hg. Physical examinationrevealed conjunctival injection and a right periorbitalhaematoma. While still in the emergency room she

ARTICLES

92 THE LANCET • Vol 358 • July 14, 2001


underwent cardiopulmonary arrest and was resuscitated,intubated, and mechanically ventilated. She was transferredto the intensive-care unit and treated with antibiotics (forpossible sepsis), fresh frozen plasma, furosamide,dopamine, and norepinephrine. A central line was placed.The patient had a large haemothorax, which was drained bychest tube. She had a significant decrease in her haematocritto 17·9%; other laboratory values were not substantiallydifferent from those noted above. She had another cardiacarrest and died. A complete necropsy was done.

The patient had been in good health and had no recenttravel history. She had received no other concomitantvaccines. There was a history of hepatitis A and nephritis as

a child. There was a history of rats in the house, but aninspection revealed none. The patient lived in ametropolitan urban area in which no cases of yellow feverhad been reported.

ProceduresBlood and tissue samples were collected after obtainingconsent from the parents of the patients. Specimens forvirus isolation and serology were stored at �70ºC untilused; those for histological studies were preserved at roomtemperature in 10% buffered formalin.

Serum samples were initially tested by haemagglutinationinhibition against yellow fever antigen (strain BeH111).

ARTICLES


Figure 1: Histopathological examination of case 1A: Liver showing intense mid-zonal necrosis and hepatocytes preserved near centrilobular area (haematoxylin-eosin stain, original magnification �200). B: Higher magnification (�400) of liver showing vesicular fatty degeneration and bodies of Councilman. C: Liver showing yellow fever virus antigens byimmunohistochemistry (�400). D=Spleen showing intense congestion and focal haemorrhage in red pulp area (haematoxylin-eosin, �400). E: Spleenshowing yellow fever virus antigens by immunohistochemistry, especially in white area (�400). F: Heart showing intense interstitial oedema and focalhaemorrhage (haematoxylin-eosin, �400).


Tests were done as described by Clarke and Casals10 withacetone-extracted serum samples and a microtitre assay.11

Serum samples were also tested for IgM antibodies with theMAC-ELISA procedure described by Kuno andcolleagues.12

For virus isolation, tissue fragments were homogenised inphosphate-buffered saline (pH 7·4) and clarified bycentrifugation (2000 g, 10 min, 4ºC). 0·1 mL of a 10%(w/v) suspension was inoculated into Vero cells and C6/36A albopictus cells. The methods used for virus isolation havebeen previously published.13 Samples from case 1 werecultured at Instituto Evandro Chagas, Belém, and those ofcase 2 at Instituto Adolfo Lutz, São Paulo. Culturesupernatants were harvested 6 days after inoculation.Simultaneously, 3-day-old suckling mice (outbred, Swissstrain) were inoculated by the intracerebral route with 0·02 mL suspension. Mice were observed for signs ofillness.

Samples of frozen tissues or cell-culture supernatantswere homogenised in Trizol (Life Technologies, Rockville,MD, USA) for RNA extraction. Total RNA (1 �g) wasused for cDNA synthesis and amplification witholigonucleotides specific for yellow fever virus DNA.14,15

PCR products were gel purified (Qiaquick gel extraction kit,Qiagen, Valencia, CA, USA) and sequenced with an ABIPRISM dye terminator kit and ABI 373 or 377 instruments(Perkin Elmer, Shelton, CT, USA).

All virus strains isolated were identified by immuno-fluorescence staining of cells by use of polyclonal andmonoclonal antibodies16,17 and by complement-fixation testson mouse-brain suspension.13

Sections of liver tissue were stained with haematoxylinand eosin and examined by light microscopy. Paraffin-embedded sections of liver and other viscera were used forthe detection of specific viral antigens by immuno-histochemistry.18

ResultsCase 1Serum collected on the fifth day after vaccination (secondday after onset of illness) was negative by haemagglutinationinhibition tests for antibodies to yellow fever virus, otherflaviviruses (St Louis encephalitis, Ilhéus, Rocio,Bussuquara, dengue 3, and dengue 4 viruses), alphaviruses(Eastern, Western, and Venezuelan equine encephalitisviruses, and Mayaro virus), and bunyaviruses (Oropouche,Catu, Caraparu, Guaroa, Maguari, and Tacaiuma viruses).ELISA for IgM antibodies against yellow fever virus,dengue 1 and 2 viruses, hepatitis B core protein, hepatitis Avirus, hepatitis B surface antigen, hantaviruses (sin nombrevirus), arenavirus (Sabia virus), and Leptospira werenegative. Virus isolation in suckling mice and C6/36 cellswere negative.

Post-mortem biopsy specimens of liver, heart, spleen, andskin were examined, as well as blood obtained by fromcardiac puncture. Virus was isolated from blood, heart,liver, spleen, and skin in suckling mice and C6/36 cells.Reisolation of yellow fever virus was obtained from bloodand viscera fragments in Vero and C6/36 cells. The virusesrecovered from blood and tissues were identified byimmunofluorescence with polyclonal and monoclonalantisera. Virus in heart tissue suspension contained thehighest titre in suckling mice (median lethal dose log 6·7 /mL) and Vero cells (log 4·7 plaque-formingunits/mL).

Histopathological examination revealed changesconsistent with those seen in wild-type yellow fever. Theliver showed midzonal necrosis, microvesicular fatty change,and eosinophilic degeneration of hepatocytes (Councilman

bodies). The spleen showed congestion, haemorrhage, andhypoplasia of the white pulp. The heart showed congestionand focal haemorrhage. Immunocytochemical staining withpolyclonal antibody against yellow fever virus showed largeamounts of antigen in hepatocytes, Councilman bodies, andKupffer cells (figure 1). Yellow fever antigen was alsodetected in spleen, but not in the heart tissue.

Reverse transcription PCR on RNA extracted directlyfrom liver, spleen, and heart fragments were positive foryellow fever virus (figure 2). Sequence analysis of theamplicons corresponding to the 3� untranslated region(figure 3) revealed that the virus recovered from case 1 wasthe 17D vaccine virus and not wild-type virus (neither of thetwo genotypes present in South America).17 Thisobservation is supported by the fact that nucleotides 28–93are absent in the 3� untranslated region of South Americanwild-type viruses, but present in the vaccine viruses andAfrican genotype I virus from which the 17D virus wasderived.17 Sequence analysis also excluded African genotypeII from being the causative agent, given the absence ofnucleotides 32–79, the presence of nucleotides 114–116,and the overall genetic difference. The cytosine (C) residueat position 66 (figure 3) instead of a thymidine (T) suggeststhat the virus from case 1 corresponded to 17D virus.Further sequencing of the envelope protein gene revealed itto be 17DD virus.

Case 2Serological tests on serum obtained 10 days aftervaccination (6 days after onset) were positive by IgMELISA for yellow fever virus, and negative for dengue 1 and2 viruses. Haemagglutination inhibition tests for antibodiesto yellow fever, St Louis encephalitis, Ilhéus, Rocio, andEastern equine encephalitis viruses were negative. IgMELISA and agglutination tests for Leptospira, IgM ELISAfor hantavirus (sin nombre virus), and IgM for hepatitis Avirus, hepatitis B core protein, and hepatitis B surfaceantigen were negative.

Blood cultures for bacteria were negative. Virus isolationin suckling mice and C6/36 cells was attempted from heartblood, as well as from brain, spinal cord, cerebellum,kidney, liver, spleen, lung, and heart fragments. All tissuesyielded virus in one or both systems. The liver isolate waspassed in suckling mice, which became ill on day 5. Thevirus was identified by indirect immunofluorescence (inC6/36) or complement fixation (mouse brain) as aflavivirus, most probably yellow fever virus. The finalidentification was made by sequencing a PCR amplicon ofthe 3� untranslated region and the E gene. The virus wasidentical to the isolate from case 1 (figure 3).

ARTICLES


Figure 2: Electrophoretic analysis of cDNAs synthesised fromviral RNA in tissuesLanes 4 and 11: molecular-weight markers. Lanes 1–3: amplicons ofstructural region of viral genome with 1064, 848, and 1024 nucleotides,respectively, from liver suspension of case 1. Lanes 5–10: amplificationwith primers at 3� untranslated region. Lanes 5–6: spleen samplesnegative and positive, respectively (case 1) for yellow fever virus. Lane 7:heart sample negative for yellow fever virus. Lane 8 amplicon derivedfrom C6/36 cell culture infected with heart suspension (case 1). Lanes9–10: amplification from negative and positive liver samples (case 1).


Histopathological examination of liver showed changeswere consistent with those seen in cases of wild-type yellowfever and were similar to those of case 1, althoughsomewhat milder. Immunocytochemistry was positive foryellow fever antigen in hepatocytes, Councilman bodies,and Kupffer cells.

Epidemiological investigationsAn extensive epidemiological investigation was done in theresidential neighbourhood of the affected families both inGoiânia and Americana. Case 1’s family resided in a lowersocioeconomic suburb of Goiânia, near an area of residualforest. A serosurvey of 1791 residents showed no evidenceof recent yellow fever by immunofluorescence assay andIgM ELISA. An entomological survey revealed a lowdensity of A aegypti and Haemagogus spp; virus isolationattempts on mosquito pools were negative. There was

judged to be no evidence of transmission of wild-type yellowfever virus.

Case 2’s family lived in a middle-class district ofAmericana. No cases of yellow fever have been reported inSão Paulo state since 1952, and there have been none in thisregion of the state for more than 100 years. No cases ofjungle yellow fever have occurred within 500 km of thisarea.

DiscussionWild-type yellow fever virus strains have the potential tocause severe damage to liver lobules in humans andmonkeys. This so-called viscerotropism is characterised bydiffuse disorganisation of the parenchyma, mid-zonalnecrosis, and degenerative lesions including fatty infiltrationand hyaline intracellular degeneration (Councilmanbodies). By contrast, yellow fever 17D viruses do not have

ARTICLES


Asibi TGAAACACCATCTAATAGGAATAACCGGGATACAAACCACGGGTGGAGAACCGGACTCCCCACyf17dd ...............................................................Goi nia ...............................................................yf17d213 ...............C...............................................yfb35 ...G.....G.T--CC.A........A------------------------------------yfb96 ...G.....G.T--CC.A........A------------------------------------yfper95 ...G..G....T.T.C...G.C....A------------------------------------CAR77/900 ...G..CAGCCA.TG-T.A.T...........-------------------------------

70 80 90 100 110 120Asibi AACTTGAAACCGGGATATAAACCACGGCTGGAGAACCGGACTCCGCACTT---AAAATGAyf17dd ...C..............................................---.......Goi nia ...C..............................................---.......yf17d213 ...C..............................................---.......yfb35 -------------------------------.....T....CTT.T..A.---.G...A.yfb96 -------------------------------.....TA....TT.T....---.A.....yfper95 -------------------------------.....AT...ATT.T.TA.---.G....GCAR77/900 ----------------.A....T....G................C.....CAA.GC..C.

130 140 150 160 170 180Asibi AACAGAAACCGGGAT-AAAAACTACGGATGGAGAACCGGACTCCACACATTGAGACAGAAyf17dd ...............-............................................Goi nia ...............-............................................yf17d213 ...............-............................................yfb35 ...T...........-...................................A.A......yfb96 ...T...........-..................................CA.A...C..yfper95 G.TT...........-...................................A.A.T.A..CAR77/900 .CG............A......C.........................T.AA.AGGCT.-

190 200Asibi GAAGTTGTCAGCCCAGAyf17dd .................Goi nia .................yf17d213 .................yfb35 ..G.A............yfb96 ..G.A............yfper95 ..G.A............CAR77/900 -TT.AC..........G

Figure 3: Alignment of sequences of 3� untranslated region from yellow fever wild-type and vaccine virusesAsibi virus=genotype I (West Africa); Yf17dd=yellow fever vaccine strain used for immunisation in Brazil; Goiânia=sequence in case 1 virus;Yf17d213=yellow fever vaccine strain obtained by Theiler and Smith (1937);9 Yfb35 and Yfb96=genotype I of South America isolated in Brazil in 1935 and1996, respectively; Yfper95=genotype II of South America isolated in Peru in 1995; CAR77/900=genotype II found in Central and East Africa. Sequencefrom the case 2 (Americana) virus is identical to case 1 (Goiânia) virus sequence. Sequences were downloaded from GenBank and aligned using Clustal.


these invasive and destructive capabilities.19 Until now,neither preclinical safety tests in monkeys nor clinicalexperience have indicated that 17D vaccine has residualviscerotropism. A meeting of yellow fever experts wasconvened by the Pan American Health Organization inBrasilia in May, 2000, to consider the clinical, laboratory,and epidemiological features of the two cases. The panelunanimously concluded that the vaccine virus was theprobable cause of fatal infections, which closely resembledwild-type yellow fever. A viscerotropic infection resemblingwild-type yellow fever represents a hitherto unknowncomplication of vaccination with yellow fever 17D viruses,and has never been reported in animals, including monkeys,which are more susceptible hosts than humans to yellowfever.7

This conclusion regarding causality was based on thefollowing points: (1) onset of illness in close temporalproximity (3–4 days) to immunisation; (2) nearly identicalclinical course—an initial non-specific febrile illnessfollowed by rapid progression with jaundice, renal failure,hypotension, shock, and death on the fifth to sixth day afteronset; (3) clinical laboratory findings with raised serumaminotransferase concentrations (aspartate amino-transferase predominating as in wild-type yellow fever),bilirubin concentrations, creatinine concentrations, andthrombocytopenia; (4) histopathological changes in the liver(and heart and spleen in one case examined) pathogno-monic for yellow fever, with yellow fever antigen present byimmunohistochemistry; (5) yellow fever 17DD virusisolated from multiple tissues; (6) no epidemiologicalevidence of jungle yellow fever transmission in the citieswhere the two patients lived, no history of trips to endemicareas, nor history of exposition to potentially risky situationsfor jungle yellow fever transmission; and (7) absence of anyother clear explanation for the illnesses.

Certain clinical features of these cases remaincontroversial and poorly understood. The aminotransferaseconcentrations were lower than those typically seen in wild-type yellow fever. This observation suggests that tissueinjury is less severe than in the wild-type disease, butnevertheless deaths occurred rapidly—perhaps more rapidlythan in typical wild-type virus infections in which deathoccurs 5–10 days after illness onset. The rapid evolutioncould explain the relatively low conentrations of serumaminotransferase enzymes. We could not exclude thepossibility that concomitant infection, possibly bacterialsepsis, might have contributed to illness and death. Bothpatients had possible sources for bacterial infection (urinarytract or lung), raised white-cell counts, and a left shift(especially case 2). Although these alterations could becaused by bacterial sepsis, marked leucocytosis is also afrequent late event in wild-type yellow fever. Moreover,histological examination of tissues did not reveal evidence ofbacteria.

There were no changes in manufacturing methods of17DD vaccine in Brazil that could account for the adverseevents. The secondary seed lot 102/84 SC 285 produced in1984 has been used for production of all vaccine lots since1985, and 370 million doses have been produced. Thissecondary seed was tested in the monkey safety(neurovirulence) test in 1984, and has been usedsubsequently as the reference material in five other monkeyneurovirulence tests between 1987 and 1999. In each case,the safety profile (viraemia as indicative for viscerotropism,clinical scores, and neuropathological scores) were similarand within WHO specifications.20 The lyophilised vaccinecontains the same stabiliser that has been used for manyyears. The potency of virus in the final filled containers hasnot changed, and is log 4·4 (SD 0·2) plaque-forming units.

The lots associated with the two fatal cases were released attitres of log 4·56 and log 4·63 plaque-forming units perdose. No significant variation is seen in potency of the finalbulks or the dilution factor used for final formulation of thevaccine.

One explanation is the occurrence of genetic variationduring replication of the vaccine virus in the host, or theselection of a genetic variant already present in the vaccinevirus population (quasispecies). Nucleotide sequenceanalysis of the PCR products in the E gene and the 3� enduntranslated region revealed no changes in the virusgenome, as was seen with 17D virus isolated from a fatalcase of encephalitis.21 We have now concluded moreextensive sequence analyses on the secondary seed lot virus,the viruses from both vaccine lots, and the viruses from bothcases. These analyses proved that the viruses are identicaland that no major alteration took place in the host. Furtherevaluation of virulence of these viruses in monkeys androdents has also suggested that the viruses are similar sinceno alterations on pathology were seen (unpublished data).

Despite the likely vaccine association, the possiblecontribution of host factors to the severity of the casesshould be discussed. The two cases occurred as isolatedevents in an estimated population of 45 million who wereimmunised between January, 1998, and April, 2000, andamong about 2 million people who received the samevaccine lots, suggesting that idiosyncratic host factors wereresponsible. Indeed, case 1 was a low-birthweight baby, anda sickly child who had a history of recurrent diarrhoea andbronchitis during childhood, raising the possibility that shewas immunocompromised. We could not determinewhether or not she responded with antibodies to thevaccine, since the only serum sample was taken too earlyafter vaccination. Case 2 had an IgM response to thevaccine by day 10—a clear sign that her humoral immunesystem was responding to infections.

Although immunosuppression has long been regarded asa theoretical contraindication to yellow fever vaccination,there are no data to suggest that immune deficiencyproduces untoward events. There are no previous reports ofadverse events associated with yellow fever vaccine inimmunosuppressed patients. Small numbers of patientswith HIV infection have received the 17D vaccine, andadverse events were not reported, although a highproportion failed to respond immunologically.22 Given thelow frequency of serious adverse events after yellow fevervaccination, factors that could predict or identify risk in anindividual patient are unlikely to be found.

The recognition of a new clinical syndrome of severemultisystem adverse events requires that surveillance forsuch events be intensified. Definition of the prevalence ofthese events as accurately and quantitatively as possible iscrucial for estimating the risk-benefit of vaccination. Thetwo fatal events require a change in the perception of yellowfever vaccine as being completely safe. The risk of seriousadverse events is still extremely low but needs to beestablished, whereas the risk of acquiring wild-type yellowfever, which has a case-fatality rate of about 50%, is muchhigher in the endemic zone. For this reason, we recommendno change in the public health policy for immunisation forresidents of the endemic zone and areas threatened byyellow fever by virtue of virus emergence—eg, the epizooticzone. Moreover, since travellers to the endemic zone are athigh risk and can also become vectors for introduction ofyellow fever on return to the receptive areas of the country,vaccination is mandatory.

The apparent rarity of serious adverse events requires thatvery large surveillance studies be employed; however, thenumber of participants in a study of this kind is probably

ARTICLES



prohibitively large. Increased passive surveillance forsyndromes meeting one or more case definitions (eg, febrileillness with jaundice) might be an alternative. Surveillancecould include hospital admissions and death registries.Education of health-care providers with respect to reportingcases and obtaining appropriate specimens for diagnosis isrequired, especially if other fatal cases were to be reported.Collection of specimens for investigation must be a priority.

On the basis of the new vaccine safety concerns, auniversal vaccination policy in Brazil or other countries inthe South American continent where yellow-fever risk isclearly partitioned between endemic and non-endemiczones is no longer advisable. In areas at no risk of jungleyellow fever, such as the coastal zones, vaccination shouldbe substituted by an effective policy of surveillance foryellow fever and readiness for emergency interventionincluding A aegypti control.

ContributorsPedro Vasconcelos was responsible for preparation of the paper, dataorganisation and interpretation, virus isolation and identification from case 1,and epidemiological studies; Expedito Luna contributed to preparation ofthe paper, and data organisation and interpretation; Ricardo Galler wasinvolved with paper preparation, data organisation and interpretation, PCRdetection and sequence determination of the virus from case 1, andcomparative analysis of sequences from both cases; Luiz Silva contributed tothe preparation of the paper, data organisation and interpretation, isolationand identification of virus from case 2, and epidemiological studies;Terezinha Coimbra isolated virus from case 2 in mice, and did cell culture;Vera Barros was responsible for the histopathological diagnosis of yellowfever in case 1, and for overseeing the histological diagnosis in both cases;Thomas Monath organised and interpreted data, and helped prepare thepaper; Sueli Rodrigues contributed to isolation of virus from case 1,diagnosis by PCR, and nucleotide sequence determination; Cristina Lavalwas responsible for treatment of and collection of clinical data from case 1;Maria Vilela did epidemiological studies; Cecília Santos identified the virusfrom case 2 by PCR, and determined the nucleotide sequence; CristinaPapaiordanou did the necropsy and histological examination on case 2;Venancio Alves did the histochemical assays for case 2 virus, and providedan overview of histological diagnosis for both cases; Liliana Andrade wasresponsible for clinical characterisation of case 2; Helena Sato did clinicaland epidemiological studies, diagnosis, and virus identification for case 2;Elisabeth Rosa contributed to case 1 virus isolation and characterisation byserological assays; Gustavo Froguas was case 2’s clinician, and provided earlydescription and data collection; Ethel Lacava and Leda Almeida didepidemiological studies for case 2; Ana Cruz helped with case 1 virusisolation, diagnosis by PCR, and nucleotide sequence determination;Iray Rocco contributed to case 2 virus isolation and characterisation byserological assays; Raimunda Santos analysed serological data for differentialdiagnosis; and Otávio Oliva contributed to data organisation andinterpretation, and to the preparation of the paper.

Brazilian Yellow Fever Vaccine Working GroupMaria da Luz Leal (FIOCRUZ, Brazil); Luisa T M de Souza (InstitutoAdolfo Lutz, Brazil); José Cássio de Moraes, Giselda Katz, Gabriel Oselka(Secretaria de Saúde de São Paulo, Brazil); Marty Cetron (Centers forDisease Control and Prevention, GA, USA); Jarbas Barbosa da Silva Jr,Roseli C Oliveira, Emanuel C Maritns, Maria de Lourdes S Maia,Ademir L Silva (CENEPI/FUNASA, Brazil); Francisco P Pinheiro,Bernardus Ganter, Angel Valencia (Pan American Health Organization,Brazil); Mário P Moraes (University of Brasília, Brazil).

AcknowledgmentsWe thank Renato Marchevsky (FIOCRUZ, Rio de Janeiro, Brazil), GilbertaBensabath (Instituto Evandro Chagas, Belém, Brazil), Akemi Suzuki(Instituto Adolfo Lutz, São Paulo, Brazil), André Schenka and Marcelo deCarvalho Ramos (University of Campinas, Campinas, Brazil), and Clelia SS Aranda (Secretaria de Saúde do Estado de São Paulo), for their help in theinvestigation of the cases, and laboratory and epidemiological expertise

during preparation of the material for discussion in May, 2000, in Brasília.The work had financial support from the Pan American HealthOrganization, Division of Vaccines and Immunization, Washington, DC,USA; CENEPI Fundação Nacional de Saúde (FUNASA), Ministry ofHealth, Brasília, Brazil; and CNPq (process 521294/97-5).

References1 Robertson SE, Hull BP, Tomori O, et al. Yellow fever: a decade of

reemergence. J Am Med Assoc 1996; 276: 1157–62.2 Vasconcelos PFC, Rodrigues SG, Dégallier N, et al. An epidemic of

sylvatic yellow fever in the southeast region of Maranhao State, Brazil,1993-1994: epidemiologic and entomologic findings. Am J Trop MedHyg 1997; 57: 132–37.

3 Vasconcelos PFC, Travassos da Rosa APA, Rodrigues SG, et al. Yellowfever in Pará State, Amazon Region of Brazil, 1998–1999: entomologicand epidemiologic findings. Emerg Infect Dis (in press).

4 Vasconcelos PFC, Costa ZG, Travassos da Rosa ES, et al. An epidemicof jungle yellow fever in Brazil, 2000: implications of climatic alterationsin the disease spread. J Med Virol (in press).

5 Vasconcelos PFC, Travassos da Rosa APA, Pinheiro FP, et al. Aedesaegypti, dengue and re-urbanization of yellow fever in Brazil and otherSouth American Countries: past and present situation and futureperspectives. WHO Dengue Bull (New Delhi) 1999; 23: 55–66.

6 Travassos da Rosa APA, Vasconcelos PFC, Travassos da Rosa ES, et al.Dengue epidemic in Belém, Pará, Brazil, 1996–97. Emerg Infect Dis2000; 6: 298–301.

7 Monath TP. Yellow Fever. In: Plotkin S, Orenstein W, eds. Vaccines.New York: W B Saunders, 1999: 815–998.

8 Robertson SE. The immunological basis for immunization, 8: yellowfever. Document WHO/EPI/GEN/93.18. Geneva: WHO, 1993.

9 Theiler M, Smith HH. Use of yellow fever modified by in vitrocultivation for human immunization. J Exper Med 1937; 65: 787–800.

10 Clarke DH, Casals J. Techniques for hemagglutination andhemagglutination-inhibition with arthropod-borne viruses. Am J TropMed Hyg 1958; 7: 561–73.

11 Shope RE. The use of a microhemagglutination-inhibition test to followantibody response after arthropod-borne virus infection in a communityof forest animals. Anais Microbiol (Rio de Janeiro) 1963; 11: 167–71.

12 Kuno G, Gomez I, Gubler DJ. Detecting artificial antidengue IgMimmune complexes using an enzyme-linked immunosorbent assay. Am J Trop Med Hyg 1987; 36: 153–59.

13 Beaty BJ, Calisher CH, Shope RE. Arboviruses. In: Schmidt NJ,Emmons E, eds. Diagnostic procedures for viral, rickettsial andchlamydial infections, 6th edn. Washington: American Public HealthAssociation, 1989: 797–855.

14 Duarte dos Santos CN, Post PR, Carvalho R, et al. Complete nucleotidesequence of the genome from yellow fever virus 17DD and 17D-213vaccine strains. Virus Res 1995; 35: 35–41.

15 Wang E, Weaver SC, Shope RE, et al. Genetic variation in yellow fevervirus: duplication in the 3� noncoding regions of strains from Africa.Virology 1996; 225: 274–81.

16 Tesh RB. A method for the isolation and identification of dengueviruses, using mosquito cell cultures. Am J Trop Med Hyg 1979; 28:1053–59.

17 Barrett ADT, Pryde A, Medlen AR, et al. Examination of the envelopeglycoprotein of yellow fever vaccine viruses with monoclonal antibodies.Vaccine 1989; 7: 333—36.

18 Hall WC, Crowell TP, Watts DM, et al. Demonstration of yellow feverand dengue antigens in formalin-fixed paraffin-embedded human liverby immunohistochemical analysis. Am J Trop Med Hyg 1991; 45:408–17.

19 Theiler M. The virus. In: Strode GK, ed. Yellow fever. New York:McGraw Hill, 1951: 46–136.

20 WHO Expert Committee on Biological Standardization. Requirementsfor yellow fever vaccine. WHO Tech Rep Ser 1995; 1: 1–16.

21 Jennings AD, Gibson CA, Miller BR, et al. Analysis of a yellow fevervirus isolated from a fatal case of vaccine-associated human encephalitis.J Infect Dis 1994; 169: 512–18.

22 Sibailly TS, Wiktor SZ, Tsai TF, et al. Poor antibody response to yellowfever vaccination in children infected with human immunodeficiencyvirus type 1. Pediatr Infect Dis J 1997; 16: 1177–79.

ARTICLES


serious adverse events associated with yellow fever...

Documents