Cell mediated immunity after measles in Guinea-Bissau: historical cohort study

Subjects and methods

The study area was on the outskirts of Bissau, the capital of Guinea-Bissau, and comprised two semirural districts, Bandim 1 and Bandim 2, and an urban district, Belem. The houses are multifamily dwellings of mud brick, and polygamy is common. The Papel is the largest of the ethnic groups living in the study area.

Measles has been studied in Bissau since 1979. Epidemics have occurred every few years, and cases of measles have been ascertained by clinical examination or interviews with mothers during the epidemics. In Guinea-Bissau mothers can recognise measles infection reliably.13 14 Measles vaccination has also been documented. Most vaccinated children have received the standard Schwarz vaccine, but a few have received medium or high doses of Edmonston-Zagreb vaccine.

A major measles epidemic occurred between October 1990 and June 1991. Measles cases were ascertained by daily contact with the health centre and paediatric department of the hospital, by contact tracing whenever a new house with measles cases was detected, by ongoing weekly morbidity surveys in some houses in the study area, by demographic survey of all houses every three months, and by four specific measles surveys during and after the peak of the epidemic. The timing of the rash of multiple cases in each house was documented to determine whether children were primary or secondary cases. Cases who developed a rash 0–6 days after the primary case in the same house were considered to be coprimary cases. Cases who developed a rash more than six days after the primary case were classified as secondary cases.5

Children were eligible for this study if they were aged 0–9 years during the 1991 epidemic and were still living in the study area in 1994. Controls were selected from children who had been vaccinated before the 1991 epidemic and had no documentation of measles. Cases of measles were selected from the 1991 epidemic. After exclusion of cases infected outside Bissau or at the hospital, a total of 707 cases in children aged 0–9 years were ascertained during the epidemic; 51 of these children died in the acute phase (within one month) and 37 died over the next three years. Thus 619 cases of measles were, to our knowledge, still alive. However, many of these children had moved before 1994 or were temporarily living outside Bissau for the duration of the study. In all, 131 of the 304 primary cases (43%) and 139 of the 315 secondary cases (44%) were identified and included in the study. Overall, 65% (175/270) of cases of measles had been seen by a doctor or nurse in the epidemic. The remaining cases had been diagnosed on the basis of a history from mothers or guardians. Each week, after selecting secondary cases, a similar number of primary cases and controls was chosen from the same broad age bands (0–2, 3–5, 6–9 years in 1991) and from the same house as the secondary cases or from neighbouring houses. Mothers or guardians were asked for permission to include their child in the study, and none refused. Cases included clinical vaccine failures—that is, children who had measles despite having been vaccinated 21 days or more before a measles rash developed. For children exposed to measles infection at home, the clinical vaccine efficacy was 87% (range 77–93%) for children under 3 years of age. However, measles vaccination coverage was high in the study area and therefore a substantial proportion of the cases in this study had received vaccine before contracting measles.

Children were assessed at home between May and September 1994. Height, head circumference, and mid-upper arm circumference were measured to the nearest millimetre with a portable stadiometer and steel tape measure. Weight was measured with electronic scales (Seca, Birmingham) to the nearest 100 g. Children were examined for splenomegaly and the presence of a BCG scar. Thick and thin blood films were taken from fingerprick samples to count malaria parasites from a subset of children during the rainy season, which begins in June.

Delayed type hypersensitivity was assessed with the Multitest CMI (Merieux, Lyons). This instrument allows seven recall antigens and a glycerin control to be administered simultaneously in a standardised fashion and gives reproducible results.15 It has been used in epidemiological studies of children in developed and developing countries.16 17 18 19 20 The antigens include proteus, trichophyton, candida, tetanus, diphtheria, streptococcus, and tuberculin. The kits were stored in a refrigerator before use.

The instrument was applied to the paravertebral skin of the upper back,17 and reactions were assessed blind to the history of measles 48 (range 45–50) hours later by only one of us (CBH) throughout the study. The presence and limits of induration were determined using the ballpoint technique.21 The perpendicular diameters of induration were measured to the nearest millimetre. A positive reaction was defined as an induration with a mean diameter of >/=2 mm (Merieux). Subjects were considered to be anergic if the reactions to all antigens were negative. A sample of 28 children (20 cases and eight controls) who were seen in May were retested at the end of the study.

At the end of the assessment a questionnaire was administered to the mother or guardian. This asked about ethnic group; the child's health, including current respiratory symptoms; illness in the previous two weeks; chloroquine treatment in the previous month; past illnesses; duration of breast feeding; birth order; and mother's age and schooling. Fourteen children who had been selected as controls had a history of measles according to the mother when interviewed in 1994. Some cases of measles might not have been documented previously if children had measles infection while temporarily out of the study area. Therefore these 14 children were excluded from the analyses.

Background data available for subsets of children included arm circumference measured under 3 years of age before the epidemic; birth weight if born in hospital in Bissau; information about the house and living conditions from a recent survey during 1993–4; and documentation of diphtheria, tetanus, and pertussis vaccination.

STATISTICAL METHODS

Descriptive analyses and linear regressions were performed using the SPSS/PC+ package. Exact confidence intervals for differences in proportions were obtained using the CIA package.22 Arm circumferences before the epidemic were adjusted for age and sex within the population studied, and z scores were calculated using multiple linear regression. Cell mediated immunity was analysed by examining anergy using multiple logistic regression and the number of positive reactions using Poisson regression (EGRET). The repeatability of the number of positive reactions in the same individuals was assessed with a paired t test. In addition, a score of reactivity was calculated by adding the size of all positive reactions (Merieux). Scores were compared between groups by the Mann-Whitney U test.

BIAS AND CONFOUNDING

Two thirds of measles cases were confirmed by a doctor or nurse. The remainder were ascertained from the history given by the mother or guardian. Serological studies in Guinea-Bissau have shown that, although some infections may be missed by mothers, a positive history is nearly always correct.13 14 We therefore believe that the degree of misclassification of measles infection in this study will have been small. We cannot exclude the possibility that the association between measles and anergy was biased by losses from the 1991 population through migration or death. However, if this were so measles infection would have to be associated with a lower prevalence of anergy in children who had migrated or died, which seems unlikely.

Anergy was more common in younger children and in girls, which is consistent with results from studies in developed countries.16 17 The increase in anergy in children who had recently taken chloroquine may reflect the immunosuppressive effects of the drug.23 Anergy was less common in children who had been breast fed for more than a year, and there is some evidence from the United States that cell mediated immunity to tuberculin, assessed in vitro, is enhanced in breast fed infants.24 25 The relation between measles and anergy remained after adjustment for age, sex, chloroquine, and breast feeding.

Two other potentially confounding factors were nutritional state and vaccination with antigens that might contribute to delayed type hypersensitivity. In other populations prenatal growth retardation, as defined by low birth weight,26 27 and postnatal growth retardation, as measured by low weight for age28 and low weight for height,29 have been linked to impaired delayed type hypersensitivity. However, we found no difference between anergic and non-anergic children in birth weight, arm circumference before the 1991 epidemic, or current anthropometric results after adjusting for age. Deficiency of micronutrients such as vitamin A might lead to depression of cell mediated immunity. Vitamin A deficiency is associated with impaired delayed type hypersensitivity,30 and clinical deficiency in children is associated with abnormalities in T cells.31 However, a previous survey of children in the same community found no evidence of clinical deficiencies in vitamins.32 There were no significant differences in BCG vaccination or number of doses of diphtheria, tetanus, and pertussis vaccine between anergic and non-anergic children.

Infection with HIV is associated with cutaneous anergy as measured by the Multitest.33 However, this is unlikely to be an important cause of anergy in the age group that we studied. Seroprevalence surveys in 1987 and 1989 in Bissau found that HIV-2 infection but not HIV-1 infection was present in adults, but the prevalence of HIV-2 infection in children aged 4–14 years was less than 1% and there was no evidence for vertical transmission.34 35 36

IMMUNE SUPPRESSION OR STIMULATION

In a retrospective study of this kind we cannot discount the possibility that anergic children were more likely to develop measles, but this does not seem biologically plausible. Measles virus is highly infectious, and subclinical infection is rare. Pre-existing immunological abnormalities are therefore unlikely to modify the risk of clinical infection.

There are therefore two possible explanations for our findings. Firstly, measles caused suppression of general cell mediated immunity that lasted for three years. If this were so we would have expected such suppression to be more profound in secondary cases than in primary cases as secondary cases have a higher delayed mortality.5 6 The difference in the prevalence of anergy between primary and secondary cases, though not significant, is consistent with this hypothesis, although a smaller study of an earlier epidemic in Bissau found no evidence for persistent suppression of T cell subsets after measles.37

Secondly, general cell mediated immunity was enhanced long term in children who had received standard measles vaccine. Studies from Guinea-Bissau have shown that vaccination in early infancy is associated with improved long term survival beyond that expected from protection against measles alone.11 12 This effect was specific for measles vaccine and not confounded by other vaccines,12 which is in keeping with our findings for cell mediated immunity.

DOSE MAY BE IMPORTANT

The long term effects of measles vaccine and natural infection on cell mediated immunity may depend on the dose of measles virus. In this respect high dose live attenuated vaccine may be analogous to natural infection. Reductions in CD4:CD8 ratios in Guinea-Bissau38 and poorer responses to Multitest antigens in Peru20 have been found two to three years after vaccination in girls who have received high dose vaccine rather than standard low dose vaccine. In Guinea-Bissau high dose vaccine is also associated with delayed mortality over a similar period.39 The higher prevalence of anergy in secondary cases compared with primary cases is in keeping with the hypothesis that secondary cases are infected with a higher dose of virus than are primary cases.40 In Senegal secondary cases continue to have significantly higher mortality than primary cases up to three years after acute measles.6

SEASONAL EFFECT

A striking finding was the noticeable increase in the prevalence of anergy during the rainy season both in the population and in a subset of children who had repeat measurements. A population based study in Bangladesh has also noted seasonal variation in the prevalence of anergy in children19 and another study in Guinea-Bissau found that children had lower CD4 counts and CD4:CD8 ratios in the rainy season.41 We do not know what factors might be responsible for the increase in anergy during the rains or why this increase was predominantly in cases of measles. Malaria and respiratory infections are more common in the rains. Data from other west African countries are conflicting as to whether acute malaria leads to impaired delayed type hypersensitivity.42 43 A previous study in Bissau found a reduction in the percentage of CD4 T lymphocytes in asymptomatic children with malaria parasitaemia,44 although we found no relation between parasitaemia and anergy in a subgroup of children. Respiratory infections can cause anergy,45 but the seasonal effect was independent of reported recent illness and cough.

CONCLUSIONS

We found a significant reduction in cell mediated immunity three years after infection in children who had had measles compared with vaccinated controls. These findings offer an explanation for the divergent survival of these two groups of children in Guinea-Bissau, which has been reported previously2 and been confirmed in our cohort (unpublished data). If this explanation is correct we would expect cases who have died to have had even more reduced cell mediated immunity, resulting in an increased susceptibility to infections. Prospective studies are needed to test this hypothesis further and to determine what infections are responsible for the long term difference in mortality between children who have had measles and children who have been vaccinated. Studies from other developing countries have shown that anergy is a significant predictor of diarrhoeal morbidity in children.18 19 29

Differences in functioning of the immune system between these two groups of children may persist for longer than three years. In a study of young adults in Bissau we found differences in atopy 15 years after measles.46 Testing for delayed type hypersensitivity is ideal for assessing the integrity of cell mediated immunity in its entirety in the field and may provide more information than quantitative T cell studies alone.33 However, more detailed immunological studies are now needed to understand which components of cell mediated immunity were abnormal and responsible for the anergy that we observed.

The death rate from acute diseases, particularly diarrhoea,47 increases twofold during the rainy season among children in Bissau.11 If depression of cell mediated immunity contributes to the increase in the morbidity and mortality of children during the rainy season in west Africa then identification of the seasonal factors responsible may be of profound importance for public health.

We thank Paulo Frederico Gomes for invaluable help with the fieldwork. We acknowledge the contributions of the late Henning Andersen and the late Anja Volmer, who organised data collection during the 1991 epidemic. Graham Wield assisted with the computing.