Interferon-γ and skin test responses of schoolchildren in southeast England to purified protein derivatives from Mycobacterium tuberculosis and other species of mycobacteria

Authors


R. Weir, Immunology Unit, Department of Infectious and Tropical Diseases, London School of Hygiene & Tropical Medicine, Keppel Street, London WC1E 7HT, UK.
E-mail: rosemary.weir@lshtm.ac.uk

SUMMARY

The immune responses of schoolchildren in southeast England to Mycobacterium tuberculosis and other species of mycobacteria were studied prior to vaccination with bacille Calmette-Guérin (BCG). Data are presented for tuberculin (Heaf) skin test and interferon-γ (IFN-γ) responses to M. tuberculosis purified protein derivative (PPD), and IFN-γ responses to PPDs from eight other environmental mycobacteria, measured in 424 schoolchildren (13–15 years of age). Responses to M. tuberculosis PPD were detected in 27% of schoolchildren by in vitro IFN-γ response and in 20% by the Heaf test. IFN-γ responses were more prevalent to PPDs from species of mycobacteria other than M. tuberculosis, predominantly those of the MAIS complex and M. marinum (45–60% responders). Heaf test and IFN-γ responses were associated (P < 0·001) for M. tuberculosis, MAIS and M. marinum. These findings have implications for appropriate implementation of vaccination against tuberculosis.

INTRODUCTION

Bacille Calmette–Guérin (BCG) vaccine has been given routinely to tuberculin-negative teenagers in the UK since 1953. More recently it has been introduced as a neonatal vaccination in high-risk groups. Large-scale studies have shown a consistently high protective effect (60–70%) of BCG against pulmonary tuberculosis in the UK [1,2]. This contrasts with the failure of the same vaccine to protect young adults in other locations, such as Malawi in southern Africa, where a recent trial found no protection against pulmonary tuberculosis, although it did protect against leprosy [3]. Neonatal vaccination has been found to provide protection against pulmonary and other forms of tuberculosis in both the UK and the tropics [4,5].

The cellular mechanism underlying the protective immune response against Mycobacterium tuberculosis appears to involve a T helper 1 (Th1)-type response, including the production of interferon-γ (IFN-γ) by sensitized CD4+ [6] and CD8+ [7] T cells to induce macrophage activation. There is current interest in the use of in vitro assays, using IFN-γ production as an indicator of a protective response, to provide a more accurate correlate of natural and vaccine-induced protection [8] as well as a diagnostic tool for infection with M. tuberculosis and other species of mycobacteria [9,10]. Methods to measure IFN-γ production by cultured peripheral blood cells in response to mycobacterial antigens include reverse transcription–polymerase chain reaction (RT–PCR) of mRNA, fluorescence-activated cell sorter (FACS) analysis of stained intracellular cytokines [11], and enzyme-linked immunospot assay (ELISPOT) [12] or enzyme-linked immunosorbent assay (ELISA) of supernatants from undiluted whole-blood culture [13] to detect production after overnight stimulation. For the current study, we used a 6-day diluted whole-blood culture assay, as this fulfilled our requirement to test a large panel of antigens simultaneously using small blood samples from a large number of donors.

This article presents results from the pre-vaccination phase of a large study of immune responses to BCG vaccination in English schoolchildren [14]. We measured the in vitro IFN-γ response in whole-blood cultures to purified protein derivatives (PPDs) derived from M. tuberculosis (‘tuberculin’) and other species of mycobacteria (M. bovis, M. avium, M. intracellulare, M. scrofulaceum, M. marinum, M. fortuitum, M. kansasii and M. vaccae), as an indicator of the naturally acquired immune response to mycobacteria in these children before they received BCG vaccination. Prior sensitivity to mycobacteria is currently measured by the tuberculin skin test (performed as the multipuncture Heaf test in the UK): a Grade 0 or 1 test response is interpreted as ‘negative’ and BCG vaccination is required; a Grade 2 Heaf test response is interpreted as indicative of prior (protective) exposure to mycobacteria with no requirement for BCG vaccination; while a Grade 3 response warrants further investigation for tuberculosis. That our whole-blood assay system measures prior sensitivity was confirmed by a dramatic increase in IFN-γ production in response to M. tuberculosis antigens measured by this assay following BCG vaccination in these same subjects [14].

By measuring Heaf test responses and in vitro IFN-γ responses to PPDs from M. tuberculosis and other mycobacteria in these English schoolchildren, we addressed four questions:

  • • How responsive to mycobacterial antigens are English schoolchildren prior to BCG vaccination?
  • • How does the Heaf skin test response relate to the in vitro IFN-γ response to PPD from M. tuberculosis, and from other species, in this population?
  • • Are there differences between the different ethnic groups in their prior sensitivity to mycobacterial antigens?
  • • Does a history of travel to the tropics alter the prior sensitivity of a child to mycobacterial antigens?

Our findings are compared with analogous data from a parallel study of young people in Malawi [15,16].

MATERIALS AND METHODS

Recruitment of subjects

This work was carried out within the routine BCG programme carried out for schools in the Redbridge and Waltham Forest Health Authority (RWFHA) area in southeast England. Approval for the study was given by the RWFHA Local Research Ethics Committee and the Ethics Committee of the London School of Hygiene & Tropical Medicine. Seven schools were selected on the basis of their willingness to participate. Children were recruited via a letter distributed by the school nurse to parents/guardians, which explained the purpose of the study and contained a written consent form and questionnaire form for completion. The questionnaire recorded ethnic group, travel history (time spent outside the UK with details of location) and atopic status. All subjects gave informed consent. Sample collection was carried out between February 1999 and April 2000. Exclusion criteria were a record of BCG vaccination or clear evidence of a BCG scar found on the day of testing, or any serious infectious or immunomodulatory disease (these children were excluded in advance by the school nurse). Children were asked if they had taken any medication, either currently or within the last month.

Skin testing was carried out using the Heaf technique, with M. tuberculosis tuberculin PPD, BP (100 000 U/ml; Evans Medical Limited, Leatherhead, UK), following UK standard procedures [17].

A venous blood sample was collected immediately after the child had received a Heaf test in the other arm, and 5 ml of blood was transferred immediately into a sterile tube (Greiner Labortechnik Ltd, Stonehouse, Gloucestershire, UK) containing 50 U of preservative-free sodium heparin (Monoparin; CP Pharmaceuticals Ltd, Wrexham, UK) for whole-blood assay. The time of collection was recorded. Heparinized blood samples were stored at room temperature until required for use.

All children were followed-up after 7 days to have their skin test indurations graded. Children with a Heaf grade of ≥ 2 were ineligible for BCG vaccination. Children with a Heaf grade of ≥ 3 were referred for active investigation for tuberculosis, following standard procedures [17].

Whole-blood assay

All whole-blood assays and ELISAs were performed in the laboratory at the London School of Hygiene & Tropical Medicine. Heparinized whole blood (4·5 ml) was diluted 1 : 5 (total 22·5 ml) with serum-free medium [RPMI-1640 supplemented with 20 IU/ml penicillin and 20 µg/ml streptomycin plus 2 mm l-glutamine (Gibco BRL, Paisley, UK)] and plated in 96-well, round-bottomed tissue-culture plates (Nunc, Roskilde, Denmark) at 100 µl/well. Cells were stimulated in quadruplicate with antigen, mitogen or with serum-free medium in a volume of 100 µl/well, giving a final volume of 200 µl/well. Cell cultures were incubated at 37°C with 5% CO2. Supernatants were harvested on day 6 and stored at −70°C prior to ELISA.

Antigens

PPD, for in vitro use, from M. tuberculosis (batch RT48, lot 191), M. avium (batch RS10/2, lot 39), M. intracellulare (batch RS23, lot 28), M. scrofulaceum (batch RS95, lot 18), M. marinum (batch RS170, lot 11), M. kansasii (batch RS30, lot 19) and M. fortuitum (batch RS20, lot 17) were provided by Statens Seruminstitut (SSI), Copenhagen, Denmark, and used at a final concentration of 5 µg/ml. PPDs from M. bovis[prepared at the Central Veterinary Laboratory (CVL), Weybridge, Surrey UK; supplied by the National Institutes of Biological Standards and Control, Potters Bar (UK)] and M. vaccae[batch R877R; Dr J. Stanford, University College, London, UK (UCL)]; and PPDs for comparison experiments –M. avium (standard avian PPD; CVL), M. intracellulare (PPD-B; CDC, Atlanta, GA; batch IV-VI), M. kansasii (Dr J. Stanford, UCL) and M. bovis (SSI; batch RS7, lot 17) – were tested at a final concentration of 5 µg/ml. Details of preparation of these PPDs have been published previously [16]. Controls were phytohaemagglutinin (PHA) (Difco Laboratories/Becton-Dickinson, Oxford, UK; final concentration 5 µg/ml); a non-mycobacterial antigen, streptokinase–streptodornase (SK/SD, Varidase; Wyeth Laboratories, Maidenhead, Berks., UK; final concentration 250 U/ml); and serum-free medium alone as the negative control.

Measurement of cytokines

IFN-γ concentrations were measured in single, 100-µl aliquots of supernatant by quantitative ELISA in Immulon 4 ELISA plates (Dynex Technologies, Chantilly, VA) using commercially available antibody pairs (PharMingen, San Diego, CA). Recombinant IFN-γ (PharMingen) was used for the standard curve on each plate, over a range of 31–2000 pg/ml. The ELISA was developed using avidin peroxidase (Sigma Chemical Co., Poole, Dorset, UK) and orthophenylenediamine (OPD; Sigma) and stopped with 2-m H2SO4. The limit of detection of the assay was 31 pg/ml. All ELISA plates were read at 492 nm wavelength using an MRX1.1 plate reader and revelation software (Dynex); the cubic spline curve fit and extrapolation options were used. To measure inter-plate and intra-plate variation, a positive-control supernatant was tested in duplicate. The coefficient of variation between plates (n = 163) was 15% and the mean variability of duplicate measurements was 4%. Identical positive-control samples were tested in both the UK and Malawi laboratories and were found to be comparable: 278 ± 43 pg/ml IFN-γ (UK, n= 163) versus 246 ± 15 pg/ml IFN-γ (Malawi, n= 5).

Data analysis

Data text files were transferred from revelation into foxpro and analysed using stata 6.0. Negative control values were subtracted from all results. Resulting cytokine responses were grouped into the following categories: ≤ 31, 32–62, 63–125, 126–250, 251–500, 501–1000, 1001–2000 and > 2000 pg/ml. A ‘positive’ IFN-γ response was defined as being > 62 pg/ml, twice the limit of detection of the assay [14]. Age was grouped as 12–13 and 14–15 years. Ethnic group was categorized as Caucasian, Black (black African, black Caribbean, black other), Asian (Indian, Pakistani, Bangladeshi) and Other. The effects of age, gender, ethnic group and school on IFN-γ (M. tuberculosis, M. avium and M. marinum PPDs) and delayed-type hypersensitivity (DTH) (M. tuberculosis PPD) responses were analysed using logistic regression. Differences among ethnic groups were assessed further, restricting the analysis to children who had not travelled outside the UK. The effect of travel outside the UK on IFN-γ and DTH responses was also assessed, separately for each ethnic group. The Kruskal–Wallis test was used to compare median IFN-γ responses to tuberculin PPD, by category of skin-test response. Spearman rank correlations were used to quantify associations between pairs of antigens for the IFN-γ response.

RESULTS

Of 515 subjects initially willing and eligible to participate, 49 declined or were absent on the day, 41 were found to be BCG scar positive on the day, and phlebotomy failed for one individual, leaving a total of 424 children recruited into the study. The mean age was 13 years (range 12–15 years); 50% were male. The majority (321/424; 75%) of the study group were white Caucasian, with the remainder classified as Black (n = 37), Asian (n = 32) and Other (n = 34). The numbers of individuals taking medication that could affect the immune responses measured were small and exclusion of these individuals from the analysis did not affect the results.

The distributions of IFN-γ responses to the control stimuli are shown in Fig. 1(a), 1(b), 1(c). Three per cent of individuals produced > 62 pg/ml IFN-γ in unstimulated cultures, and > 99% produced > 62 pg/ml IFN-γ to the positive control, PHA. A range of responses was observed to SK/SD antigen, with 15% of individuals producing ≤ 62 pg/ml IFN-γ and 52% producing > 500 pg/ml.

Figure 1.

Frequency distribution of the interferon (IFN)-γ response to (a) a negative control (RPMI), (b) a positive control [phytohaemagglutinin (PHA)] and (c) a non-mycobacterial control antigen (SK/SD), of study subjects (n = 424), as measured in the whole-blood assay.

The distribution of IFN-γ responses to the mycobacterial antigens is shown in Fig. 2. Twenty-seven per cent of the study group gave a positive response (> 62 pg/ml) to M. tuberculosis PPD (Fig. 2a), with a median response in these individuals of 175 pg/ml. The highest prevalence of sensitivity, by IFN-γ response, was observed to PPDs from members of the MAIS complex (M. avium, M. intracellulare and M. scrofulaceum) and M. marinum, with 60% of the children producing an IFN-γ response of > 62 pg/ml to M. avium PPD (Fig. 2c).

Figure 2.

Frequency distribution of the interferon (IFN)-γ response of study subjects (n = 424) to purified protein derivative (PPD) of (a) Mycobacterium tuberculosis and (b) – (i) other species of mycobacteria, as measured in the whole-blood assay.

The proportions of the group making highest response to a particular PPD were as follows: M. avium, 28·1%; M. marinum, 18·9%; M. intracellulare, 9·2%; M. scrofulaceum, 7·3%; M. fortuitum, 3·5%; M. tuberculosis, 2·1%; M. kansasii, 1·9%; M. vaccae, 1·2%; and M. bovis, 0·7%. A total of 27·1% (n = 115) of the group made no response to any of these antigens.

Analysis of the pattern of responses showed a strong concordance in the IFN-γ response to the MAIS PPDs (Fig. 3). Cross-reactivity between responses to PPDs of all the species tested was also apparent from correlation analysis, with correlation coefficients ranging from 0·50 to 0·84 (0·78–0·84 between MAIS complex species) (Table 1). Species-specific patterns were still apparent: although M. marinum and M. tuberculosis are closely related [correlation coefficient (r) = 0·66 for responses to these two species], 55% of subjects produced a positive response to M. marinum (median 231 pg/ml), as compared with 27% responding to M. tuberculosis (median 175 pg/ml).

Figure 3.

Interferon (IFN)-γ responses to Mycobacterium avium, M. intracellulare and M. scrofulaceum (MAIS) purified protein derivatives (PPDs). The vertical axis shows the percentage of children responding to no antigen, or to one, two or all three antigens.

Table 1.  Spearman rank correlation coefficients between responses to different mycobacterial purified protein derivatives (PPDs)
PPDM. tuberculosisM. bovisM. aviumM. intracellulareM. scrofulaceumM. marinumM. kansasiiM. fortuitumM. vaccae
M. tuberculosis1·00·680·700·740·710·680·690·550·59
M. bovis    1·00·570·620·590·610·680·520·60
M. avium      1·00·800·840·780·700·500·58
M. intracellulare         1·00·790·760·730·560·58
M. scrofulaceum          1·00·760·720·510·51
M. marinum         1·00·710·520·56
M. kansasii          1·00·530·57
M. fortuitum           1·00·51
M. vaccae           1·0

Three pairs of PPDs, prepared from the same species of mycobacteria (M. avium, M. intracellulare and M. kansasii), but by different institutions, were compared. A strong correlation was observed for M. avium SSI versus M. avium CVL (r = 0·82) (although a greater number of subjects gave a positive response to the SSI PPD), for M. intracellulare SSI versus PPD-B (r = 0·78) and for M. kansasii SSI versus M. kansasii UCL (r = 0·67).

The distribution of Heaf test responses is shown in Fig. 4(a). There was a strong association between the median IFN-γ response to M. tuberculosis PPD and the DTH response to the Heaf skin test (P < 0·001) (Fig. 4b). However, discordant subjects were observed – of those children who gave a Heaf test response of Grade 2 or Grade 3, 14% (four of 29) did not produce a positive IFN-γ response, and of 342 children who gave a Heaf test response of Grade 0, 59 (17%) produced a positive IFN-γ response to M. tuberculosis PPD, 14 of whom gave an IFN-γ response of > 250 pg/ml. Median IFN-γ responses to PPDs from all the environmental species of mycobacteria were associated with the Heaf test response (P < 0·001; Fig. 4c). No association was observed between Heaf test response and IFN-γ response to the control antigen SK/SD.

Figure 4.

Heaf test responses to tuberculin in the study group (n = 424). (a) Frequency distribution of Heaf test responses. (b) Interferon (IFN)-γ responses to Mycobacterium tuberculosis purified protein derivative (PPD) as a function of the Heaf test response. (c) IFN-γ responses to M. tuberculosis PPD, environmental mycobacteria PPDs and SK/SD, as a function of the Heaf test response. The median IFN-γ responses associated with each level of skin test sensitivity are shown for: ▪, M. tuberculosis PPD; ○, M. avium PPD; ◊, M. intracellulare PPD; ▴, M. scrofulaceum PPD; □, M. marinum PPD; *, M. kansasii PPD; ×, M. fortuitum PPD; ▵, M. vaccae PPD; and •, SK/SD.

Analysis of the effects of age, gender, school and ethnic group, on the Heaf test response and on the IFN-γ response to M. tuberculosis and M. avium PPD, are shown in Table 2. There was no evidence for differences by age, gender or school attended, apart from one school (School 3, with a small number of pupils) that had relatively low responses.

Table 2.  Analysis of Heaf test response and interferon-γ (IFN-γ) response to Mycobacterium tuberculosis purified protein derivative (PPD) and M. avium PPD for all study individuals
Risk factorCategoryHeaf test gradeIFN-γ to M. tuberculosis PPDIFN-γ to M. avium PPD
+ ve/n*%Adjusted OR95% CIP-valuep/n%Adjusted OR95% CIP-valuep/n%Adjusted OR95% CIP-value
  • Threshold of response is defined as an IFN-γresponse of > 62 pg/ml, or Heaf grade > 0.

  • *

    + ve/n, number of group making positive Heaf test response/total number in group.

  • + ve/n, number of group making positive IFN-γresponse/total number in group.

  • CI, confidence interval; OR, odds ratio.

Age12–1343/23918   60/24125   138/24157   
14–1533/181180·9 (0·5–1·7)0·86149/183271·1(0·6–1·8)  0·829114/183621·2 (0·8–1·8)0·473
GenderFemale41/20920   61/21029   125/21060   
Male35/211170·9 (0·5–1·7)0·85248/214220·8(0·5–1·4)  0·461127/214590·9 (0·6–1·4)0·769
School 119/9819   23/9823    64/9865   
 2 9/61150·7 (0·3–1·7)0·40915/61251·0(0·5–2·3)  0·902 37/61610·8 (0·4–1·6)0·555
 3 2/23 90·4(0·07–1·7)0·197 4/23170·4(0·1–1·4)  0·146  7/23300·2(0·06–0·5)0·001
 414/60231·2 (0·5–2·8)0·62520/60331·8(0·9–4·0)  0·113 37/60620·9 (0·5–1·9)0·884
 513/80160·7 (0·3–1·5)0·34918/80220·8(0·4–1·6)  0·500 50/80620·8 (0·4–1·6)0·560
 6 7/60120·5 (0·2–1·3)0·16914/64220·7(0·3–1·6)  0·443 36/64560·6 (0·3–1·2)0·183
 712/38321·1 (0·4–3·0)0·914
0·399
15/38391·2(0·5–3·1)  0·723
  0·178
 21/38550·6 (0·2–1·4)0·206
0·035
Ethnic groupCaucasian51/32016   70/23122   187/32158   
Black 4/35110·9 (0·3–2·6)0·77717/37464·1(1·9–8·7)< 0·001 26/37702·5 (1·1–5·7)0·030
Asian13/32413·6 (1·5–8·8)0·00515/32473·1(1·3–7·4)  0·009 19/32591·2 (0·5–2·8)0·636
Other 8/33241·7 (0·7–4·1)0·219
0·033
7/34211·0(0·4–2·3)  0·912
< 0·001
 20/34591·2 (0·6–2·5)0·679 0
·154

Heaf test grades of ≥ 1 were more prevalent among Asian children (P = 0·005), and a higher percentage of Asian (P < 0·001) and Black children (P = 0·009) gave a positive IFN-γ response to M. tuberculosis compared with Caucasian children. These differences were also evident using a threshold of 250 pg/ml. The percentage of children giving a positive IFN-γ response to M. avium PPD was higher for Black children than for Caucasian children, but this difference was not evident using the higher threshold of 250 pg/ml. Ethnicity had no significant effect on the IFN-γ response to the control antigen, SK/SD (data not shown).

Sixty-four study recruits (15%) (20 Caucasian, 10 Black, 25 Asian and nine children of ‘Other’ ethnic origin) had visited at least one country in a non-temperate area. The countries visited were associated with ethnic group: none of the Caucasian or Black children, but 22/32 (69%) of the Asian children, had visited the Indian subcontinent; seven (19%) and two (5%) Black children had visited West Africa and the Caribbean, respectively, compared with none of the Caucasian and Asian children. Analysis of the effect of travel history on the Heaf test and IFN-γ response to M. tuberculosis PPD, stratified by ethnic group, suggested little effect for Caucasian children: 25% of children who had travelled to non-temperate areas gave a positive IFN-γ response compared with 22% among children who had not (P = 0·73). For Asian and Black children, the data suggested that travel to non-temperate areas might influence responsiveness to M. tuberculosis (52% of Asian children who had travelled, versus 29% of Asian children who had not, gave a positive IFN-γ response), but the analyses were limited by small sample size and the differences were not statistically significant. There was no evidence that a history of previous travel influenced the IFN-γ response to M. avium PPD – the percentage giving a positive IFN-γ response to M. avium among children who had or had not travelled to non-temperate areas was 70% versus 57% for Caucasian children, 70% versus 70% for Black children and 56% versus 71% for Asian children.

As travel history varied by ethnic group, and because the data suggested that travel to non-temperate areas could affect IFN-γ and Heaf test responses to M. tuberculosis PPD, analysis of differences among ethnic groups was then restricted to children who had never left the UK. Evidence that the IFN-γ response to M. tuberculosis PPD was higher in Black children than Caucasian children remained [48% (12/25) compared with 22% (65/297) gave a positive response]. The difference between Asian and Caucasian children was diminished: only 29% (two of seven) of Asian children who had never left the UK gave a positive IFN-γ response to M. tuberculosis PPD.

DISCUSSION

We have presented data comparing skin test and in vitro IFN-γ responses to M. tuberculosis and other mycobacterial PPDs in healthy non-BCG-vaccinated UK schoolchildren. Our observations have implications for BCG vaccination in this population. We compared these data with results from a parallel study of unvaccinated adolescents and young adults in Malawi [15,16].

Pre-BCG vaccination IFN-γ responses of UK subjects to the control stimuli (PHA and SK/SD), and IFN-γ responses of non-stimulated cultures, showed a similar profile to those observed in Malawi [15], which indicates that in both populations the whole-blood cultures were viable, the majority were not pre-activated to produce an IFN-γ response, and that responses to SK/SD (derived from Streptococcus) reflected the ubiquity of exposure to streptococci in both locations. A minority (27%) of the UK study group produced a positive IFN-γ response to M. tuberculosis PPD prior to BCG vaccination, in contrast to the majority (62%) of the Malawi group [15]. This lack of prior sensitivity to M. tuberculosis PPD in the UK was also evident in the skin-test responses, the majority of the UK group (80%) producing no response (Grade 0) to the Heaf test.

Sensitivity to PPDs from other species of mycobacteria suggest appreciable exposure to a variety of mycobacteria in this UK population, although most IFN-γ responses were of low magnitude. M. avium PPD was most commonly recognized (by 60% of the group producing a positive IFN-γ response) closely followed by other species of the MAIS complex and M. marinum. Prior sensitivity to mycobacteria was considerably lower, for all species of mycobacteria, in UK children than observed in Malawi (where 77–85% sensitivity to MAIS complex PPDs was observed [15]), with the exception of M. vaccae, which induced relatively low responses in children of both locations. Among responders, the highest median responses were to M. avium, M. intracellulare, M. scrofulaceum and M. marinum PPDs. There was a similar magnitude of response to each species among responders in each location, with the exception of M. tuberculosis PPD, which induced higher IFN-γ responses in Malawian subjects than in the English children (251 pg/ml versus 175 pg/ml; P < 0·001), M. bovis (210 pg/ml versus 180 pg/ml; P = 0·067) and M. scrofulaceum (309 pg/ml versus 256 pg/ml; P = 0·035). A high degree of cross-reactivity between antigens of the MAIS complex is to be anticipated, and strong concordance in IFN-γ response to these PPDs was observed in the English children, consistent with our previous findings in Malawi [15]. However, the relatively low recognition by English children of M. tuberculosis PPD (which shares many antigens with MAIS and M. marinum) indicates that these PPDs contain species-specific antigens and can discriminate, to some degree, between exposure to different species of mycobacteria. The maximum response of an individual to a PPD was most frequently seen to M. avium PPD. Among children who produced an IFN-γ response to M. tuberculosis PPD, for only 8% (n = 9) was this the highest response, with the majority (37%, n= 40) making the highest response to M. avium PPD.

It has been suggested that different potencies of PPDs prepared from different species could have an effect on their relative recognition in vitro or in vivo. We sourced our PPDs from the same laboratory as far as possible, to reduce the possibility of different production methods resulting in a different antigenic profile. Where we had the opportunity to test pairs of PPDs from the same species prepared by different laboratories, we did find a high degree of correlation between the two PPDs. This is consistent with our parallel studies in Malawi [16], and supports the relative prevalence of response to a PPD being an indicator  of  previous  exposure  to  antigens  found  in  that species rather than an artefact of the method of production of the PPD.

There is clinical and environmental evidence to corroborate the presence of MAIS and other species of mycobacteria in the UK environment. There have been few microbiological studies of environmental mycobacteria in the UK: a study of a farm in southwest England found more than 750 acid-fast isolates in soil and water samples, but the majority of these were difficult to characterize when compared with laboratory isolates, and the preparation methods may have selectively depleted certain species [18]. A recent study [19] found an increase in the reports of infections caused by opportunist mycobacteria in England and Wales between 1982 and 1994; most of these were caused by M. avium-intracellulare infections (which were associated with human immunodeficiency virus in < 40% of cases). Of the species included in our current study, MAI is the commonest cause of opportunist mycobacterial disease in the UK, followed by M. kansasii, M. marinum and, to a lesser extent, M. fortuitum, while M. scrofulaceum is quite rare. Although clinical disease is not a simple correlate of prevalence in the environment, these findings give an indication of what healthy teenagers in the UK may encounter. A close relationship between M. avium-intracellulare strains isolated from clinical samples and those isolated from the environment was found in a study in Africa [20]. The large Medical Research Council trial of BCG and vole bacillus vaccines in 1950 [21] showed that prior to vaccination, 32% (15 514/47 964) of English 13-year-olds at that time were positive (≥ 5 mm induration), by the Mantoux skin test, to 3 TU of tuberculin, indicating previous infection with M. tuberculosis. Of the remaining 68%, 19% (6153) were positive only to a higher dose of tuberculin (100 TU) – this group were probably sensitized to environmental mycobacteria. This earlier indication of the extent of pre-sensitization of English schoolchildren is confirmed by our current study, which may provide the first clear evidence of widespread sensitivity to particular species of environmental mycobacteria in children living in the UK, even in this suburban area of southeastern England. However, these responses are less prevalent, and of lower magnitude, than those measured in young adults living in Malawi [16,22].

The prevalence of IFN-γ responses to environmental mycobacteria was little affected by whether a child had travelled outside the UK, indicating that the observed responses are largely the result of exposure in southeast England. The responses to M. tuberculosis were higher among Black and Asian children who had travelled outside the UK compared to those who had not, but the analysis was limited by small sample size and the differences were not statistically significant. It has been shown that travellers to tuberculosis-endemic areas are exposed to a similar risk of contracting tuberculosis during their stay as the local population [23]. There was some evidence of IFN-γ response to M. tuberculosis PPD varying by ethnic group, even among those children who had never left the UK, although sample sizes were small. Within the UK, contact with visitors from areas of high tuberculosis endemicity and with tuberculosis patients probably varies by ethnic group. This study has provided suggestive evidence regarding the relative roles of ethnicity and travel history on pre-vaccination responses to mycobacterial antigens among a group of individuals. A study in which a high proportion of children are non-Caucasian is required to investigate this issue in greater detail.

The Heaf test result is used to guide vaccination strategy in the UK, although evidence shows that DTH is not a direct correlate of protection against M. tuberculosis[24]. This study provides an opportunity to understand more fully the implications of the Heaf skin test response to tuberculin, by comparing it with another manifestation of the host immune response, the in vitro IFN-γ response. We observed a strong association between the Heaf test grade and the in vitro IFN-γ response to tuberculin, although there was considerable variation in IFN-γ responses within Heaf test categories, and also considerable overlap in IFN-γ responses between Heaf categories (Fig. 4b), indicating a discordance in these two responses, similar to that previously observed in Malawi [15]. This implies that the two tests, although measuring related aspects of the immune response, do not measure exactly the same mechanism, consistent with findings from animal models [25]. To what extent the variation in, and discordance between, the tuberculin skin test and the in vitro IFN-γ response reflects different cellular pathways is unclear, but may point to important processes relevant to protection against mycobacterial infection or disease.

Tuberculin DTH induced by prior exposure to M. tuberculosis indicates infection and is associated with a high risk of tuberculous disease, whereas tuberculin DTH, induced by environmental mycobacteria, may be associated with protection against tuberculosis [24]. The students in this study have had 3–4 years of follow up since these pre-vaccination tests were carried out, and none of the five children with a Heaf grade of > 2 have developed tuberculosis. This suggests that their initial sensitivity was unlikely to be a result of infection with M. tuberculosis, but was provided by exposure to cross-reactive environmental mycobacteria.

The strong association observed between the Heaf test response and the IFN-γ response to the majority of mycobacterial PPDs tested indicates the role of environmental mycobacterial antigens, in particular those of the MAIS complex, in the induction of DTH to M. tuberculosis PPD in UK children. It has been shown in animal models that prior exposure to different environmental mycobacteria can affect the response to subsequent BCG vaccination, either by blocking proliferation of the BCG [26] or by ‘masking’ the protective effect of the vaccine by providing as much protection as the BCG [27]. Whether the protection induced by environmental mycobacterial exposure could be increased through some form of vaccine should become clear in the context of new tuberculosis vaccine trials. Studies such as this, which relate tuberculin DTH to in vitro responses to other mycobacterial antigens, provide important insights, with implications for effective tuberculosis prevention and the development of new tuberculosis vaccines.

ACKNOWLEDGEMENTS

This work was supported by LEPRA, with additional funds from the World Health Organization (WHO). We thank Ann Berry, Kathryn Brady, Sally Edwards, Anna Hadassi, Makki Hameed, Mary Hayde, Mary Heath, Barbara Holland, Freda Lock, Ann Marsden, Marie Murphy, Shakuntala Patel, Christine Sloczynska, Agnes Udom and Margaret Walsh in Redbridge and Waltham Forest Health Authority for help with the UK school study; the staff and students of Heathcote School, Highams Park School, Wanstead High School, Hainault Forest High School, Woodbridge High School, Trinity Catholic High School, and Woodford County High School for their co-operation and participation in this project; Sara Atkinson, Shweta Brahmbhatt, Heidi Robinson, Kevin Tetteh and Sally Stenson for laboratory assistance at the London School of Hygiene & Tropical Medicine; and Anna Randall for data entry. We thank Michael Brennan, Kaare Haslov, Glyn Hewinson, John Stanford and the Central Veterinary Laboratories, Weybridge, for providing the mycobacterial antigen preparations used in this study. This publication is dedicated to the memory of Dr Richard Aspinall.

Ancillary