Reye's syndrome and aspirin: a review

Reye's Syndrome and Aspirin: A Review596 Journal ofthe Royal Society of Medicine Volume 79 October 1986
Reye's syndrome and aspirin: a review
SusanM Hall MSc MFCM PHLS Communicable Disease Surveillance Centre, London NW9, and Department ofEpidemiology, Institute of Child Health, London
Keywords: Reye's syndrome, aspirin, encephalopathy
Reye's syndrome (RS) is a rare but severe neuro- logical disorder complicated by hepatic dysfunction and predominantly affecting children. It was first described as a clinicopathological entity by Reye, an Australian pathologist, in 19631. A brief review of the clinical and epidemiological
features is pertinent to an understanding of the issues ofstudying the risk factors for developing RS.
Clinical and epidemiological features ofRS In the typical case a previously healthy child acquires a common viral infection which is apparently taking a normal course when there is a sudden onset of profuse persistent effortless vomiting, accompanied or shortly followed by alteration in conscious level. Liver involvement is indicated by grossly elevated levels of serum aspartate and alanine transaminases and the blood ammonia level is also usually raised. There are histological and ultrastructural changes in the liver. These include diffuse panlobular micro- vesicular fatty infiltration and swollen, disrupted mitochondria. The Centers for Disease Control (CDC) have main-
tained national surveillance ofRS since 1974. Annual USA incidence figures have ranged from 0.3 to 0.88 per 100 000 population under 18 years2, although more intensive local studies have shown that the inci- dence may be as high as 6 per 100 0003. Other salient epidemiological features ofRS in the USA include an equal sex distribution, a median age of 8-9 years and an excess of cases living in rural areas and a consis- tent place-time association with influenza B and, to a lesser extent, influenza A. The next commonest viral prodrome is varicella4. The 1984 case fatality ratio was 20% and since 1980 there has been a marked reduction in overall annual incidence'. The British Reye's Syndrome Surveillance Scheme
(BRSSS) began in August 1981. Two hundred cases were reported in the first 3 years and the annual incidence in 1983/4 was 0.7 per 100 000 children under 16 years6. Although the epidemiology ofRS in Britain is similar to that in theUSA in two respects-an equal sex distribution and a significant excess of cases living in rural areas - there are important differ- ences: the median age, 14 months, is younger; the mortality at 50% is higher; and there is no clear, strong association with influenza but rather with a wide variety of viruses7. A significant problem in epidemiological studies of
RS is the nonspecificity of the case definition. The one used by CDC states, firstly, that there should be encephalopathy but no signs in the cerebrospinal fluid or in brain histology to indicate infection or inflammation. Secondly, for the fatty liver changes, either there should be typical histology or raised
serum transaminases or a raised blood ammonia. The third requirement is that there should be no other or more reasonable explanation for the condition5. This is a sensitive and practical case definition but it is also nonspecific because there is a wide differential diagnosis of children presenting with encephalo- pathy and abnormal liver function tests. This differ- ential includes not only certain inherited metabolic disorders which can even mimic RS histologically, but also salicylate toxicity8'9. The third requirement of the CDC case definition does not specify that all such altemative diagnoses must have been ruled out. Indeed, the facilities to do so and the diagnostic awareness is far from universal both in the USA and in Britain.
Aetiology ofRS The precise aetiology of RS is unknown but it is believed to be multifactorial, resulting from an abnormal reaction to a viral infection in a genetically susceptible host which is modified by an exogenous toxin4. Diverse toxins and chemicals have been impli- cated, including aflatoxins, insecticides and their emulsifiers, margosa oil and aspirin4.
Aspirin and RS: historicalperspective In the early 1960s and 1970s several authors observed that there were clinical and biochemical similarities between RS and salicylism and that a high proportion of cases had a history of aspirin ingestion9. In one later case-control study a dose-response relationship was noted between stage of coma and total dose of aspirin received'0. One study in the early 1980s found higher mean salicylate levels in RS patients on admission compared to controls" and another, a correlation between salicylate level at presentation and the maximum level ofcoma achieved'2. In 1983 a study demonstrating histopathological similarities between RS and salicylate toxicity was published'3. The pathogenesis ofRS is thought to depend on an
acute mitochondrial dysfunction, and experimental observations support the biological plausibility ofan RS-aspirin relationship. First, aspirin acts on iso- lated rat mitochondria: it uncouples oxidative phos- phorylation, inhibits mitochondrial dehydrogenases, causes an increase in liver membrane permeability and causes swelling and othermorphological changes consistent with those seen in RS". Second, in vitro, aspirin inhibits lymphocyte transformation and interferon production in response to viral infection' 5. IfRS is an abnormal response to a viral infection then an agentwhichmightcompromise theimmunesystem might, in a susceptible host, precipitate the atypical response. Third, animal models have shown that the syndrome can be induced by a combination of hyper-
Paper read to Section of Epidemiology & Community Medicine, 5 February 1986
0141-0768/86/ 010596-03/$02.00/0 C) 1986 The Royal Society of Medicine
Journal ofthe Royal Society ofMedicine Volume 79 October 1986 597
Table 1. Controlled studies of salicylate use during prodromal illnes8 ofReye's syndrome
Starko Waldman Halpin Hurwitz et al.1O et al.I8 et al.17 et al. 19
No. cases 7 37 97 30 No. controls 6 75 156 145 Case:control salicylate use (%) 100:50 97:57 97:71 93:46
ammonaemia, a viral infection and aspirin16. Fourth, one study of the pharmacokinetics of salicylates in RS survivors suggested that those patients may be slow metabolizers of aspirin9. The most controversial and widely publicized
studies linking RS and aspirin have, however, been the epidemiological case-control studies conducted in the USA between 1978 and 198410.17-19. Table 1 summarizes the findings of these studies, which all showed a significant excess risk of salicylate usage (specified as aspirin in all except that of Hurwitz et al. 19)byRS patients during theirprodromal illnesses. The findings of the first three studies led the
United States Surgeon General to issue a public and professional warning about possible RS-aspirin association in 198220. Parents and doctors were advised to avoid using aspirin in children with influenza and chickenpox. The American Academy of Pediatrics initially endorsed this warning but later reversed its decision, as did the then Secretary of State for Health, who decided at the end of1982 not to order the warning labels for all aspirin-containing products which had been considered earlier in the year. The reason for these reversals was that such controversy had arisen over the conduct of the epidemiological studies that it was recommended that a further study should be undertaken. The pilot phase of this, which was intended to
establish the appropriate methodology, took place in the first half of 1984. As a result of the pilot study findings19, legislation requiring product labelling was promulgated in late 1985, having been preceded by voluntary labelling and a public service education campaign. In the meantime the main case-control study is proceeding.
The RS-aspirin association: criticisms A detailed analysis of the apparent clinical, bio- chemical and histological similarities between RS and salicylate toxicity shows that there are in fact fundamental differences between the two entities in all three parameters, and the relevance of the early comparisons is now seriously questioned9. The 'dose-response' relationship reported by
Starko et al.10 was not observed in later case-control studies and the studies relating severity to serum salicylate levels have been criticized because of nonspecificity of the methodology; the results have not been repeatable using more sensitive assays21. In vitro and animal experiments can always be
criticized on the grounds of artificiality and unknown relationship to humans. Furthermore, although therapeutic doses may have been used in the experiments, it is not known how they relate to intracellular concentrations of aspirin in vivo in human subjects. It is also noteworthy that in one recent study an animal model of RS was developed
using influenza B and paracetamol22, and that in another there was experimental evidence supporting a hypothesis ofa direct viral effect without any toxin mediation23. Animal studies have also provided support for a role for insecticides or emulsifiers with viral infection as a cause ofRS related to an impaired
24interferon response The epidemiological evidence is also controversial.
It is hard to explain why an aspirin association would explain an excess occurrence ofRS in rural areas. The early reports of an RS-aspirin association are
difficult to interpret because they are uncontrolled. The case-control design is appropriate for studying risk factors for rare diseases but is inherently subject to several forms of biases, and it was the demon- stration of those in the first 3 studies, following release of the raw data, that led to the call for a new study and to the decision reversals outlined above. The criticisms have been reviewed in detail9'25'26 but can be summarized as follows: firstly, selection bias - did the cases really have RS? Was it certain, for example, that they did not include patients with aspirin toxicity? Were the controls equally at risk of developing RS in respect of all other parameters except aspirin usage (they were not matched for type, severity and viral cause ofthe prodrome).
Secondly, recall bias: memory is worse for controls than cases because the event is less significant and these interviews took place after a longer interval than those for cases.
Third, data collection bias: the interviewers were aware of the research hypothesis and of the designation oftheir subjects (case or control).
Fourth, categorization bias: product identification was more accurate for controls than for cases because the former were interviewed at home.
Fifth, protopathic bias: exposure to the risk factor should precede onset of disease. The problem is that the clinical marker for onset ofRS is unknown. It is normally taken to be the vomiting but it is always possible that it begins earlier. It was only possible to assess this in one part of one of the studies because of the methods ofdata collection. The methodology of the fourth (pilot phase) case-
control study attempted to overcome these and other criticisms, but it still retained the potential for serious biases27. The most important ofthese relate to the widespread knowledge of the association which means, firstly, that there could havebeen preferential diagnosis and/or reporting of patients with the diag- nostic criteria plus a history of aspirin ingestion. Secondly, parentsmay consciously (with litigation in mind) or unconsciously have reported aspirin use once they knew the diagnosis.
Conclusion If the criteria for judging causality are applied to the evidence outlined above for and againstan RS-aspirin association, an ambiguous picture emerges: in the case-control studies the temporal relationship ap- pears satisfactory at first but there are problems in defining the onset of RS. On the wider time scale, however, perhaps the most convincing piece of evi- dence in favour of the relationship is the natural 'experiment' illustrated by the long-term trends in the annual incidence ofRS in the USA. The steady fall since 1980, which accelerated in
19855, is temporally related to publicity about the association and has occurred in cases associated with
598 Journal of the Royal Society of Medicine Volume 79 October 1986
both varicella (which has a steady annual incidence) and influenza (whose type and activity fluctuates annually). Furthermore, there is evidence that this decline has been paralleled by a decline in aspirin usage among children28'29. The association has been consistently shown in
the American epidemiologial studies although the only other published study, from Japan, did not show such an association (but the methodology was not described)30. Further evidence of consistency of association was provided by finding an increased risk ofRS among children on long-term salicylate therapy for connective dissue disorders3 - 33. The association in the USA studies was strong but
those studies have been criticized for serious biases which, on reanalysis, reduced the strength; the association is not specific - there is experimental and epidemiological evidence linking other exogenous agents to RS. The association appears to be bio- logically plausible but the experimental evidence in support of this is conflicting.
In summary, a question mark has been hung over the safety of aspirin when given to children with common viral infections. The evidence to date is con- flicting and each piece, taken on its own, might not be convincing. Nevertheless, the accumulated weight of the evidence, together with the serious nature of RS and the availability ofalternative effective remedies, has led to the recent recommendations in Britain34 that aspirin should not be given to children under 12 years of age unless specifically indicated for childhood rheumatic conditions.
References 1 Reye RDK, Morgan G, Baral J. Encephalopathy and
fatty degeneration of the viscera, a disease entity in childhood. Lancet 1963;ii:749-52
2 Centers for Disease Control. Reye Syndrome - United States, 1984. CDC Morbidity and Mortality Weekly Report 1985;34:13-16
3 Lichtenstein PK, Hubi JE, Daugherty CC, et al. Grade One Reye's Syndrome. A frequent cause ofvomiting and liver dysfunction after varicella and upper respiratory tract infection. NEnglJMed 1983;809:133-9
4 Sullivan-Bolyai JZ, Corey L. Epidemiology of Reye's Syndrome. Epidemiol. Rev. 1981;3:1-26
5 Centers for Disease Control. Reye Syndrome - United States, 1985. CDC Morbidity and Mortality Weekly Report 1986;35:66-74
6 PHLS Communicable Disease Surveillance Center and British Paediatric Association. Reye's syndrome surveillance scheme: third annual summary report. Br Med J 1985;291:329-30
7 Hall SM, Bellman MH. Reye's Syndrome in the British Isles: the British Paediatric Association/PHLS Com- municable Disease Surveillance Centre joint surveil- lance scheme. In: Pollack JD, ed. Reye's syndrome IV. Bryan, Ohio: National Reye's Syndrome Foundation, 1985:32-46
8 Stanley CA, Coates PM. Inherited defects of fatty acid oxidation which resemble Reye's Syndrome. In: Pollack JD, ed. Reye's Syndrome IV. Bryan, Ohio: National Reye's Syndrome Foundation, 1985:190-200
9 Rodgers GC. Analgesics and Reye's syndrome: fact or fiction? In: Pollack JD, ed. Reye's syndrome IV. Bryan, Ohio: National Reye's Syndrome Foundation, 1985: 117-34
10 Starko KM, Ray CG, Dominguez LB, Stromberg WL, Woodall DF. Reye's syndrome and salicylate use. Pediatrics 1980;66:859-64
-11 Partin JS, Partin JC, Schubert WK, et al. Serum salicy- late concentrations in Reye's Syndrome. Lancet 1982;i: 191-4
12 Tonsgard JH, Huttenlocker PR. Salicylates and Reye's syndrome. Pediatrics 1981:68:747-9
13 Starko KM, Mullick FG. Hepatic and cerebral patho- logy findings in children with fatal salicylate intoxi- cation: further evidence for a cause or relationship between salicylate and Reye's Syndrome. Lancet 1983;i: 326-9
14 Kwan-Sa-You. Salicylate and mitochondrial injury in Reye's syndrome. Science 1983;221:163-4
15 Glezen WP. Aspirin and Reye's syndrome. Am J Dis Child 1982;163:971-2
16 Deshmukh DR, Maassab HF, Mason M. Interactions of Aspirin and other potential aetiologic factors in an animal model of a Reye's syndrome. Proc Natl Acad Sci USA 1982;79:7557-60
17 Hallpin TJ, Holtzhauer FJ, Campbell RJ, et al. Reye's syndrome and medication use. JAMA 1982;248:687-91
18 Waldman RJ, Hall WN, McGee H, VanAmburg G. Aspirin as a risk factor in Reye's syndrome. JAMA 1982;247:3089-94
19 Hurwitz ES, Barrett MJ, Bregman D, et al. Public Health Service Study on Reye's Syndrome and Medications: Report of the pilot phase. N Engl J Med 1985;313:849-57
20 Centers for Disease Control. Surgeon General's Advisory on the use of salicylates and Reye's syndrome. CDC Morbidity, Mortality Weekly Report 1982;31: 289-90
21 Clark JH, Nagamori K, Fitzgerald JF. Confirmation of serum salicylate levels in Reye's syndrome: a compari- son between the Natelson colorimetric method and high performance liquid chromatography. Clin Chim Acta 1985;145:243-7
22 Macdonald MG, McGrath PP, McMartin DN, Washington GC, Huda KG. Potentiation of the toxic effects ofactaminophen in-mice by concurrent infection with influenza B virus: a possible mechanism forhuman Reye's syndrome? Pediatr Res 1984;18:181-7
23 Ogburn PL, Shorop H, Lloyd-Still JD et al. Abnormal polyunsaturated fatty acid patterns of Serum Lipids in Reye's Syndrome. Proc Natl Acad Sci USA 1982; 79:908-11
24 Crocker JFS, Fung D, Hudson R, Faith SH. Examin- ation of the role of surfactants in Reye's syndrome. In: Pollack JD, ed. Reye's syndrome IV. Bryan, Ohio: National Reye's Syndrome Foundation, 1985:135-40
25 Daniels SR, Greenberg RS, Ibrahim MA. Scientific uncertainties in the studies of salicylate use and Reye's syndrome. JAMA 1983;249:1311-16
26 RS Working Group. Reye's syndrome and salicylates: a spurious association. Pediatrics 1982;70:158-60
27 White JM. Reye's syndrome and salicylates. N Engl J Med 1986;314:920
28 Remington PL, Rowley D, McGee H, Hall NN, Monto AS. Decreasing trends in Reye Syndrome and Aspirin Use in Michigan 1979 to 1984. Pediatrics 1986;77:93-8
29 Barrett MJ, Hurwitz ES, Schonberger LB, Rogers MF. Changing epidemiologyofReyeSyndrome in the United States. Pediatrics 1986;77:598-602
30 Yamashita F, Eiichiro 0, Kimura A, Yoshida I. Reye's syndrome in Asian countries. In: Pollack JD, ed. Reye's syndrome IV. Bryan, Ohio: National Reye's Syndrome Foundation, 1985:47-60
31 Remington PL, Shabino CL, McGee H, et al. Reye Syndrome and Juvenile Rheumatoid Arthritis in Michigan. Am JDis Child 1985;139:870-2
32 Hansen JR, McCray PB, Bale JF, Corbett AJ, Flanders D)J. Reye syndrome associated with aspirin therapy for systemic lupus erythematosus. Pediatrics 1985;76:202-5
33 Rennebohm RM, Heubi JE, Dougherty CC, Daniels SR. Reye syndrome in children receiving Salicylate therapy for corrective tissue disease. JPediatr 1985;107:877-80
34 Committee on Safety ofMedicines. CSM Update: Reye's syndrome and aspirin. Br Med J 1986;292:1590
(Accepted 16June 1986)