Relation between BCG vaccine scar and an interferon-gamma release assay in immigrant children with “positive” tuberculin skin test (≥10 mm)

Immigrants from countries with high incidence of tuberculosis (TB) are offered medical examination and screening for TB and other infectious diseases when they arrive to Sweden. The screening for TB consisted until recently of physical examination and a tuberculin skin test (TST). If TST was ≥ 10 mm a chest X ray was performed. If there were no signs of active TB, preventive chemotherapy was recommended to children and adolescents (in the following called children) if compliance could be expected to be good. From 2007 children coming to the Gothenburg area with TST ≥10 mm, irrespective if a BCG scar was visible or not, were referred to clinics specialising in TB where Quantiferon was assayed. Preventive chemotherapy was offered if Quantiferon was positive. The TST limit of 10 mm follows the recommendations of the National Board of Health and Welfare.

TST is not ideal for screening for latent TB. It requires two visits and it crossreacts with antigens in the Bacillus Calmette-Guérin (BCG) vaccine and with many nontuberculous mycobacteria (NTM) [1‐8]. Repeated testing can induce false positive results. Furthermore, the definition of a positive test varies from 5 to 15 mm) [1‐8]. The US Centers for Disease Control and Prevention adjust for underlying diseases, history of vaccination and origin of the immigrant (http://​www.​cdc.​gov/​tb/​publications/​factsheets/​testing/​skintesting.​htm). In the UK, 5 mm is used as cut-off in children with BCG scar (National Institure for Health and Care Excellence, https://​www.​nice.​org.​uk/​guidance/​ng33/​chapter/​recommendations#diagnosing-latent-tb-in-children-and-young-people).

The interferon gamma release assays (IGRAs) solve many, but not all, problems in the evaluation of latent TB. They require only one visit, do not cross-react with BCG vaccines and cross-react with fewer NTM than TST. Several studies have shown a rather good correlation between TST and IGRAs except in very young children [9‐13] but a recent study in 126 adults showed a relatively poor correlation between TST and IGRAs and also between two IGRAs (QuantiFERON -TB Gold In-Tube® (QFT) and T-SPOT.TB®) [12]. The use of IGRAs is therefore controversial and has not replaced TST for diagnosis of latent TB in many countries [14, 15].

It is usually difficult to obtain a reliable vaccination history from the accompanying adult family member and impossible if the child arrives alone to Sweden. The only way to obtain information about BCG vaccination is to look for typical BCG scars even though BCG vaccination does not always result in a scar and scars can wane with time [1, 16‐18].

A total of 762 children fulfilled all inclusion criteria. BCG scars were found in 492 of them (65 %) (Table 2). Scars were found in significantly lower proportions in children from Somalia and the rest of Sub-Saharan Africa (56 %) than in all other children combined (77 %) (p < 0.0001).

QFT was positive in 163/492 (33 %) children with a BCG scar and in 205/270 (76 %) without a scar (p < 0.0001). Significant differences were found in children from Somalia, the Middle East + North Africa, Sub-Saharan Africa, Southeast Asia, Europe and China. Children from India and South America were too few for separate calculations (Table 3). Somalian children with a BCG scar were significantly more often QFT positive (55 %) than all other groups combined (22 %) (p < 0.0001). Significant differences of the same magnitude were found in boys and girls.

Among the 215 unaccompanied children aged 11 – 18 years 131/215 (61 %) had positive Quantiferon compared to 179/383 (47 %) children aged 11–18 years who came with their families (p = 0.0009). Among those who came with their families and were aged between 11 and 18 years, the proportion with BCG scars was higher 266/383 (69 %) than in the unaccompanied children, 113/215 (53 %) (p < 0.0001).

The median TST diameters were significantly higher in QFT positive than in QFT negative children (p <0.0001) both in children with and without scar (Table 4). There was, however, a considerable overlap (Fig. 1). Among the children with negative QFT, 279 had TST 10–14 mm and 115 had TST ≥ 15 mm.

Chest X ray changes were seen in 25/368 QFT positive and in 2/394 QFT negative children. The changes consisted of enlarged hilar or tracheal lymph nodes or mild to moderate inflammatory infiltrates. One QFT negative child (6 year old boy from Somalia) had a TST just at the cut-off value of 10 mm. No mycobacteria could be isolated from bronchoalveolar lavage. He had elevated erythrocyte sedimentation rate (ESR) and C- Reactive Protein Test (CRP). His changes disappeared after 6 months of antituberculosis treatment for active TB with initially four drugs. This QFT negative child became QFT positive after 1.5 years, when he had finished treatment for active TB. The other QFT negative child (15 years old boy from Somalia) had a small basal inflammatory infiltrate. He had both ESR < 15 mm/h and CRP < 1.0 mg/L on repeated testing. He became QFT positive after 7 weeks. No mycobacteria were isolated from bronchoalveolar lavage. He was also treated for active TB with 4 drugs initially and chest X ray normalised.

The most important difficulty in comparing TST and QFT is due to the absence of a “golden standard” for the diagnosis of latent TB. Based on theoretical assumptions a positive QFT is, however, more specific than a large TST which has no generally approved definition of “positive” and no positive and negative controls. This must be kept in mind when the present and a very large number of comparative studies are evaluated.

The study showed that there is a highly significant correlation between presence of BCG scar and positive QFT. The reason for this could not be answered by the results. The finding that a higher proportion of children with scar had negative QFT than children without scar (67 % verus 24 %) speaks in favour of the possibility that BCG vaccination resulting in a scar influences TST while the finding of the same TST diameter in QFT negative children with and without scar speaks against an effect of BCG scar on TST (12 mm in both groups).

Assuming that a negative QFT makes latent TB unlikely something else may have induced the large TST of ≥10 mm in many QFT negative children. This could be infections with NTM, which is unlikely because NTM would hardly give any difference between children with and without scar. Another possibility is previous TST testing with booster effect (of which we could get no information), which is also unlikely because in low-income countries with ongoing civil wars, latent TB is hardly looked for. The difference between children with and without a BCG scar makes BCG vaccination resulting in a scar the most likely possibility, even though the causes might be multifactorial. This result was unrelated to country of origin, gender and to age ≥10 years and 5–9 years. The great majority of children with a BCG scar can be assumed to have been vaccinated at birth or soon afterwards, since the World Health Organisation recommends BCG vaccination as early as possible. Since most of the children in the present study were adolescents the effect of BCG on TST seems to last for several years. This finding is in contrast with a study of school children in India stating that TST can be used for survey of TB irrespective of BCG vaccination [19] and a study from Spain indicating that BCG can give a false positive TST result in children <3 years but not in older children [20]. The most likely explanation of our finding is that BCG vaccination can induce an increase in the TST reaction, while QFT is unaffected by BCG vaccination [1‐8, 20, 21].

The reasons why some children with no scar have TST ≥10 mm and negative QFT could be that they had received a vaccination without getting a scar [1, 16‐18] or that the scar had waned with time which according to a study from India may occur in 30 % after a five year period [22]. Besides, different BCG vaccines have different ability to induce a scar [23, 24]. Their TST might still react with antigens in BCG. Another possibility could be that the children had been infected with NTM, many of which can induce an increase in TST without cross-reacting with the antigens in QFT [2‐8]. Besides in patients with active TB negative QFT is found in about 20 % [25, 26]. In the absence of a reference diagnostic method for latent TB, it is not possible to state how common falsely negative QFT is in patients with latent TB.

The median TST diameter was significantly larger in QFT positive than in QFT negative children but the overlap was great, so it is not possible to predict the result of QFT based on the size of the TST diameter in the individual case, which agrees with two previous studies of Turkish adults and of immigrant children in Norway [27, 28].

A British study of adults, most of them from Asia, showed that chest X-ray at the port of entry contributed little or nothing to finding TB cases [12]. The present study in children agreed only partly with this. Although chest X ray findings were significantly more common in QFT positive than in QFT negative children, chest X ray in only QFT positive children would have led to missed or delayed diagnosis in two children, who had late QFT conversions. Besides there were very few children <5 years. In this age group immunologic tests like IGRA may not work well.

The study shows that TST with a cut-off value between positive and negative of 10 mm should not be used alone for the diagnosis of latent TB. This would have led to an overdiagnosis of more than two-fold in children with a BCG scar and of one-third in children without a BCG scar, in the present study. Considering the costs of prophylactic medication, side effects and controls of adherence this seems unacceptable. Confirmation with the more specific IGRA’s seems necessary both in vaccinated and nonvaccinated children. A Canadian study performed among North American Indians also concluded that TST alone is unsatisfactory for diagnosing latent TB in school children [8]. A two-step procedure with screening using the less expensive TST followed by an IGRA may be justifiable if compliance of return for reading the TST is high. A study from the United Kingdom advocates IGRA without preceding TST [12]. The U.K. study differs in two respects from the present. The immigrants in the U.K. study were mainly young adults coming from the Indian subcontinent and other Asian countries without civil wars while the patients in the present study were children and adolescents and > 50 % of them came as refugees from countries with devastating civil wars where the exposure may be expected to be high both in the home country and during the difficult journey to Europe. QFT alone is, however, not perfect for diagnosing latent or early active TB because of the late seroconversion in some patients, as shown in two cases in the present sttudy. If resources allow and if compliance for return can be expected to be good, we recommend that both TST and QFT are performed and that patients with positive TST and negative QFT are followed with repeated QFT for at least two years.

The present study can not be used for evaluation of the efficacy of BCG vaccination in neonates or early infancy but it seems obvious that any vaccine efficacy must have waned considerably. This was particularly obvious in Somalians and in unaccompanied minors. Though the children in this study were not systematically questionned about their travelling to Sweden a large number of Somalian and/or unaccompanied minors told spontaneously about several months of being hidden by smugglers in crowded vans, small apartments or tents where the risk of exposure must have been considerable. Many had also spent several months in crowded Libyan, Ethiopian, Turkish and Greek refugee camps. Children from other countries and children who came with adult relatives usually had spent less time and lived under better conditions between leaving the home countries and arrival to Sweden. Another reason for the higher proportion of Quantiferon positivity found in children coming alone or Somalian children could be that these groups had a lower frequency of BCG scars, so it is possible that the vaccination protected them during the first years of life.

The study showed a significant relation between presence of BCG scar with negative QFT. It did not give a definite answer whether a BCG scar was related to size of TST. The diagnosis of latent TB in health examinations with no known exposure should not be based on a large TST alone but preferably on a combination of TST and QFT. A confirmatory IGRA should be performed in patients with large TST. QFT negative children with large TST should be repeatedly tested by IGRA. The TST diameter was larger in QFT positive than in QFT negative children but could not be used to predict the QFT result in the individual patient. Chest X ray contributes very little to the diagnosis of TB in asymptomatic QFT negative children but should not be omitted because of the possibility of falsely negative QFT or late conversion of QFT in some patients.