Remember me
ORIGINAL ARTICLE
Year : 2023 | Volume
: 12
| Issue : 1 | Page : 43-48
Epidemiology of the koch phenomenon of infants after bacillus calmette–Guerin vaccination by interferon-γ release assay status, Japan, 2013–2019
Masaki Ota1, Yoshiyuki Furuichi2, Susumu Hirao1
1 Division of Technical Assistance to National Tuberculosis Programmes, Research Institute of Tuberculosis, Tokyo, Japan
2 Department of Paediatrics, Yamanashi Red Cross Hospital, Yamanashi, Japan
Correspondence Address:
Masaki Ota
No. 3-1-24 Matsuyama, Kiyose 204-8533, Tokyo
Japan
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/ijmy.ijmy_203_22
Background: The epidemiology of infants who exhibited the Koch-like phenomenon after Bacillus Calmette–Guérin vaccination and who subsequently tested positive in interferon-γ release assay (IGRA) was compared to that of those who tested negative. The reports of pediatricians on the phenomenon to the health authorities of Japan were retrieved and analyzed. Methods: In 2013–2019, 790 infants with such events were reported with IGRA test results available, of whom 81 (10.3%) tested positive and 709 (89.7%) negative. Results: The infants who were IGRA positive did not show an increasing trend (P = 0.06, P = 0.60), whereas those who were IGRA negative showed a significantly increasing trend (P = 0.42, P = 0.0002). The infants who were IGRA positive did not exhibit seasonality, whereas those who were IGRA negative had a higher number of cases in winter than in summer. The rates of infants who were IGRA positive per 10 million live births showed a significant correlation with the tuberculosis (TB) notification rates by prefecture (P = 0.41, P = 0.004), whereas those who were IGRA negative did not (P = 0.04, P = 0.78). Conclusion: The IGRA-positive infants were distributed quite differently from those who were IGRA negative and appeared more likely to be infected with TB. Reports of pediatricians on the Koch-like phenomenon should continuously be collected as the reports reflect a risk of TB infection including TB outbreaks among infants in Japan. The reports should include IGRA test results as IGRA is more specific than tuberculin skin testing. Infants with IGRA-positive results should be followed up for 2–3 years to determine their final outcomes.
Keywords: Bacillus Calmette–Guerin, epidemiology, interferon-γ release assay, Japan, tuberculosis
Japan has successfully reduced the burden of tuberculosis (TB) in the past seven decades from 590,684 cases (698/100,000 population) in 1951 to 12,739 cases (10.1/100,000 population) in 2020.[1],[2] However, almost 5000 cases of sputum smear-positive pulmonary TB are still reported annually,[2] and these potentially infectious TB cases pose a public health threat, including in the form of outbreaks,[3],[4],[5],[6],[7],[8] to the community. Young children, particularly those under 5 years of age, are known to be more vulnerable to TB, if infected.[9] Although only about 30 cases were reported annually among those aged <5 years in the past decade,[2] Bacillus Calmette–Guérin (BCG) vaccination is still administered to all infants,[10] to prevent them from developing severe forms of TB.[11]
BCG vaccination started in 1943 in Japan, initially for selected adolescents, and then for children under 7 and a half years old.[12] Before 2004, TB screening using tuberculin skin testing (TST) was conducted for children to determine whether they had already been infected with TB. However, in 2005, the procedure was changed and TST is no longer conducted before BCG vaccination. In addition, the target infant population was narrowed to infants <7 months old[13] in line with the World Health Organization's recommendation.[14] In 2013, the target infant population was raised to <1 year old because of fear of BCG osteitis/osteomyelitis.[15]
However, one of the shortcomings related to administering BCG vaccines without TB screening using TST is early skin reactions within a few days of the vaccination, known as the Koch phenomenon,[16] or to be more precise the “Koch-like” phenomenon as the reactions are milder than the original meaning (necrosis on the skin), caused possibly by prior infection by Mycobacterium tuberculosis or more probably by other nontuberculous mycobacteria (NTM), such as Mycobacterium marinum and Mycobacterium avium.[17],[18] An early study reported that about 200 such incidences were reported each year from 2005 to 2008 to the Ministry of Health, Labour and Welfare (MHLW) of Japan as adverse events.[19] However, the cases were diagnosed using only TSTs in that study and the accuracy of diagnosis of TB infection was quite ambiguous.
Since 2006 when the interferon-γ release assay (IGRA) was introduced in Japan, it has become the common method for diagnosing TB infection, as IGRA is more specific and can avoid interference caused by BCG vaccination compared to the conventional TST.[20],[21] Initially, it was speculated that the sensitivity of IGRA was inferior to TST, particularly for young children, but later, it was reported that the sensitivity was superior, or at least equivalent, even in children younger than 2 years old.[22] However, there has never been a large-scale study conducted using IGRA for infants who exhibited the Koch-like phenomenon after BCG vaccination. This study aims to characterize the infants who exhibited the Koch-like phenomenon, including those with either IGRA-negative or IGRA-positive results, in terms of time, place, and person, after BCG vaccination from 2013 to 2019.
MethodsThis was an observational study on infants who exhibited the Koch-like phenomenon and were reported by physicians after BCG vaccination from April 2013 through March 2019. The information on the infants with the Koch-like phenomenon in the study period was retrieved from MHLW.[23] The information included age (in months), sex, the month and the year of the report, the prefecture where each infant lived, the IGRA test results, and the diameter of erythema (not induration) of TST.[24] It should be noted that there was no information of whether the infants developed active TB or not in the data from MHLW. The identification information of the infants was not included in the release of the MHLW.
The number of active TB cases among children under 1 year old from 2013 through 2018 and the TB notification rates by prefecture from 2013 through 2018 were retrieved from the National TB Surveillance Data System. The number of live births and the number of the population by prefecture were retrieved from the National Statistics Centre of Japan. The number of infants who were vaccinated with BCG was retrieved from MHLW.[25]
The data were analyzed in terms of time of onset (long-term trend and seasonality) and the number of days between the BCG vaccination and when the Koch-like phenomenon was noticed by their guardians, the places the infants lived, the age and sex distributions, and the IGRA-positivity in relation to the size of the erythema of TSTs.
Statistical tests, including calculations of 95% confidence intervals (CIs), were conducted using R (Ver x64 4.0.2. The R Foundation for Statistical Computing, Vienna, Austria). Spearman's correlation analysis was employed for the trend analysis. Tukey's honestly significant difference test was used for multiple comparisons. Cochran–Armitage tests were conducted for trend analysis for proportions. P < 0.05 was considered statistically significant.
ResultsFrom April 2013 through March 2019, there were 2684 infants who exhibited the Koch-like phenomenon after a BCG vaccination, of whom 790 (29.4%) underwent an IGRA test. There were 1894 (70.6%) who were not tested or whose status was unknown. Of the 790, 709 (89.7%) tested negative for IGRA, whereas 81 (10.3%) tested positive. The numbers of infants who were vaccinated with BCG increased from 877,419 in 2013, to 996,844 in 2014, and 1,003,475 in 2015, then decreased to 988,723 in 2016, 946,852 in 2017, and 898,837 in 2018 respectively.
The number of infants with the Koch-like phenomenon as a whole and those who were IGRA negative showed significantly increasing trends (P = 0.24, P = 0.04 and P = 0.42, P = 0.0002, respectively), whereas those who were IGRA positive did not show either an increasing or decreasing trend (P = 0.06, P = 0.60) [Figure 1]. For the record, the numbers of infants with the Koch-like phenomenon as a whole per 100,000 BCG shots from 2013 to 2018 were increasing (42.4, 42.4, 44.3, 53.9, 37.5, 59.5, P < 0.0001), so are those who were IGRA negative (8.9, 9.4, 12.1, 13.9, 11.6, 18.1, P < 0.0001). However, those who were IGRA positive per 100,000 BCG shots did not show any trend (1.7, 1.2, 0.9, 1.6, 1.2, 2.0, P = 0.51).
Figure 1: Long-term trends of infants who exhibited the Koch-like phenomenon (Upper left), those who were IGRA negative (Upper right), and those who were IGRA positive (Lower left), after Bacillus Calmette–Guérin vaccination, Japan, 2013–2019. (Upper left) n = 2663, P =0.24, P = 0.04, (Upper right) n = 703, P =0.42, P = 0.0002, (Lower left) n = 80, P =0.06, P = 0.60. IGRA: Interferon-γ release assayWith regard to seasonality, the average number of cases of the Koch-like phenomenon was higher in January, October, and December than in April through August. The situation is quite similar for those with IGRA negative, except that the average number of cases reported in January is not as high as those reported in June or July. The average number of cases of the Koch-like phenomenon with IGRA positive did not show any difference by month [Figure 2].
Figure 2: Distribution of the average number of Koch-like phenomena (Left), those who were IGRA-negative (Middle), and those who were IGRA-positive (Right) by month, after Bacillus Calmette–Guérin vaccination, Japan, 2013–2019. (Left) Differences in the number of reports between the months that were statistically significant: January versus April, January versus May, January versus June, January versus July, January versus August, February versus August, April versus October, April versus December, May versus October, May versus December, June versus October, June versus December, July versus October, July versus December, August versus October, August versus November, and August versus December n = 2663. (Middle) Differences in the number of reports between the months that were statistically significant: January versus April, January versus May, January versus August, April versus October, April versus November, April versus December, May versus October, May versus December, June versus December, July versus October, July versus December, August versus October, and August versus December n = 703. (Right) There was no difference in the number of reports among the months. n = 80. IGRA: Interferon-γ release assayThe Koch-like phenomenon was noticed by the guardians of the infants on day 0 (the day of the BCG vaccination) in 228 cases (8.5%), on day 1 in 1663 cases (62.0%), on day 2 in 530 cases (19.7%), on day 3 in 119 cases (4.4%), and on day 4 or later in 144 cases (5.4%). The findings were similar for both IGRA-negative and IGRA-positive cases. Of the IGRA-negative cases, 63 (8.9%) were noticed on day 0, 406 (57.3%) on day 1, 171 (24.1%) on day 2, 29 (4.1%) on day 3, and 40 (5.6%) on day 4 or later. Of the IGRA-positive cases, 2 (2.5%) were noticed on day 0, 49 (60.5%) on day 1, 18 (22.2%) on day 2, 8 (9.9%) on day 3, and 4 (4.9%) on day 4 or later.
The geographic distributions of rates of infants with the Koch-like phenomenon per 1 million population, including both IGRA-negative and IGRA-positive cases, are shown in [Figure 3]. By prefecture, the rates of infants with the Koch-like phenomenon as a whole per 10 million live births and those who were IGRA negative did not show correlations with the TB notification rates (P = −0.09, P = 0.54 and P = 0.04, P = 0.78, respectively), whereas those with IGRA-positive results had a statistically significant correlation with the TB notification rates (P = 0.41, P = 0.0037).
Figure 3: Geographic distribution of Koch-like phenomenon (Left), those who were IGRA-negative (Middle), and those who were IGRA-positive (Right), after Bacillus Calmette–Guérin vaccination, Japan, 2013–2019. IGRA: Interferon-γ release assayFor the infants with the Koch-like phenomenon, the median age was 6 months (range: 2–12 months, n = 2207), for those who were IGRA negative, it was 6 months (range: 3–12 months, n = 609), and for those infants with IGRA-positive results, it was 6 months (range: 5–11 months, n = 69) [Figure 4].
Figure 4: Age (in months) distribution of infants who exhibited the Koch-like phenomenon (Left), those who were IGRA-negative (Middle), and those who were IGRA-positive (Right), after Bacillus Calmette–Guérin vaccination, Japan, 2013–2019. (Left) n = 2207, Median: 6 months, (Middle) n = 609, Median: 6 months, (Right) n = 81, Median: 6 months. IGRA: Interferon-γ release assayRegarding the sex distribution of the infants with the Koch-like phenomenon, the number of males was 1106 (48.2%, 95% CI: 46.1%–50.2%) of the 2296 cases for which the sexes were available. Of the 610 who were IGRA negative, 290 were male (47.5%, 95% CI: 43.5%–51.6%) as were 33 of the 67 who were IGRA positive (49.2%, 95% CI: 36.8%–61.8%).
For 726 cases of the infants with the Koch-like phenomenon, the results of both TST and IGRA were available. The number of IGRA positives was 0 of 211 (0.0%, 95% CI: 0.0%–1.7%) for the erythema diameters of 0–4 mm, 5 of 127 (3.9%, 95% CI: 1.3%–8.9%) for those of 5–9 mm, 46 of 219 (21.0%, 95% CI: 15.8%–27.0%) for 10–14 mm, 20 of 108 (18.5%, 95% CI: 11.7%–27.1%) for 15–19 mm, 9 of 37 (24.3%, 95% CI: 11.8%–41.2%) for 20–24 mm, 0 of 14 (0.0%, 95% CI: 0.0%–23.2%) for 25–29 mm, and 0 of 9 (0.0%, 95% CI: 0%–33.6%) for 30 mm or over, and the trend was statistically significant (P < 0.0001).
DiscussionIn this study, the authors characterized infants with the Koch-like phenomenon with IGRA-negative or IGRA-positive results after BCG vaccination. From 2013 through early 2019, almost 2700 cases of infants with the Koch-like phenomenon were reported. IGRA test results were available for about 800 cases and approximately 10% of those were IGRA positive. The infants with the Koch-like phenomenon who were IGRA negative showed an increasing trend, whereas those who were IGRA positive did not. Seasonal variations were found with a higher number of infants with the Koch-like phenomenon who were IGRA negative in winter than in summer; however, those who were IGRA positive did not show a significant difference. There was no correlation by prefecture of the rates of infants with the Koch-like phenomenon who were IGRA negative with the TB notification rate, whereas there was a statistically significant correlation of the infants who were IGRA positive with the TB notification rates. These results showed that the infants with the Koch-like phenomenon who were IGRA negative were distributed quite differently from those who were IGRA positive in terms of the long-term trend, seasonality, and TB notification rates by prefecture, and they appear unlikely to have had true TB infection, whereas the IGRA-positive cases were caused by true TB infection. Considering that there were 61 active TB cases among those <1 year old from 2013 through 2018 (an average of 12.2 cases/year), the number of infants with the Koch-like phenomenon who were IGRA positive (which was 81) does not seem to be overestimated.
The reason why almost 90% of the infants with the Koch-like phenomenon with IGRA tests results available were IGRA negative is that they may have been infected with NTM, such as M. marinum[17] and M. avium,[18] as some authors previously reported, rather than M. tuberculosis. The reason why the number of Koch-like phenomenon reports increased in winter is that skin reactivity in general increases in dry and cold weather conditions in winter in Japan;[26],[27],[28],[29] however, other mechanisms may need to be explored.
This study was conducted using the reports of physicians on the Koch-like phenomenon to MHLW required by the Immunization Act of 1948 of Japan,[30] and the findings in this study are considered robust, highly reproducible, and representative. In addition, this is the first study on infants with the Koch-like phenomenon with IGRA-positive test results compared to those with IGRA-negative results after BCG vaccination, since a previous study did not report about IGRA positivity among the infants with the Koch-like phenomenon.[19]
The previous study reported a finding similar to ours: two-thirds of the infants with the Koch-like phenomenon were noted on day 1 after the vaccination.[19] However, there are dissimilar findings as well. The mean age of the cases was 4 months because, before 2013, BCG was normally administered at the age of 3–4 months.[19] The previous study reported about half the number of infants who exhibited the Koch-like phenomenon compared to our study.[19] The reason is that the infants vaccinated with BCG were younger in those years and had less opportunity for getting infected with NTM before the vaccination. After the target infant population was extended to those <1 year old in 2013, more infants over 6 months old who had already been infected with NTM got BCG vaccination and showed the Koch-like phenomenon. After 6 months of age, infants in general start being curious about the world around them, often grabbing things, and putting them in the mouth[31] and are more likely to be infected with NTM than those who are <6 months old. The start of weaning food of infants is between 5 and 6 months old in Japan,[32] and this may have contributed to more infants who had already been exposed to NTM when they were vaccinated with BCG after 2013. In addition, the previous study reported that the number of cases of the Koch-like phenomenon decreased from 269 cases in 2005 to 159 in 2008,[19] whereas in our study, the number of cases of the Koch-like phenomenon as a whole and those who were IGRA negative was significantly increasing. The reasons for this contradiction are unclear.
There are some limitations to this study. This study relied on the data collected, collated, and released by MHLW, and there may be some errors, incompleteness, or misclassifications in the data. One of the examples is the unavailability of information, particularly about two-thirds of the reports lacking IGRA test results, and this decreases the representativeness of IGRA positivity among the infants with the Koch-like phenomenon. This is, though, an inevitable limitation using routinely collected data. Another limitation would be that in some of our analysis, we compared the absolute number of infants with the Koch-like phenomenon because we did not have proper populations for calculating proportions or rates (e.g. analysis of seasonality and the age in months of the infants of the onset of the Koch-like phenomenon). However, in terms of the seasonality, since we do not have very long (over 10 days) holiday seasons in Japan, we assumed that the number of BCG vaccinations is more or less evenly distributed among the months. Third, a lack of the standard case definition for reporting an infant who exhibited a Koch-like phenomenon to MHLW may have led to a potential selection bias for the study. However, the authors do not believe an existence of any potential bias preclude in this study. Physicians in general tend to report an adverse event of a vaccination if the reaction is severe, long-lasting, or both, whereas they may not report if the reaction is mild, short-lasting, or both regardless of case definition. They may also report the infants with Koch-like phenomenon who tested positive for IGRA because the infants are likely infected with TB regardless of severity. Thus, we do not believe most infants who exhibited the Koch-like phenomenon who tested positive for IGRA are underestimated, rather those who tested negative for IGRA may have been underreported because the physicians may have thought the infants were unlikely to be infected with TB. Fourth, our data on TST of the infants who exhibited the Koch-like phenomenon were expressed by diameter of erythema, rather than induration that is universally used in this field. However, in Japan, erythema is commonly used to report the TST results from the outset and the physicians do not report the induration. In addition, it was reported that diameters of erythema and induration are proportional[24] and the authors still believe that there are some values reporting the TST results for the infants with the Koch-like phenomenon.
There are some implications from the findings of our study. First, MHLW should continue collecting reports on the adverse events in relation to BCG vaccination, including reports on the Koch-like phenomenon, since these reports reflect a risk of possible TB infection among infants in Japan and provide a small window through which we can estimate that risk and can possibly discover TB outbreaks. Second, MHLW should also instruct municipalities that implement BCG vaccination for infants that reports on the Koch-like phenomenon should include IGRA test results, particularly for those with erythema of TST equal to or over 5 mm in diameter, since IGRA is more specific than TST and provides more accurate risk assessment of TB infection among infants.[20],[21] Third, the infants with IGRA-positive results should be followed up for 2–3 years and their mid-term outcomes should also be studied to determine whether they develop TB disease or not. This also gives us a clue in terms of more accurate risk assessment of TB disease and infection in Japan. Those with IGRA-negative result may need to be followed up for 6 months or possibly 1 year with occasional chest X-ray examination (s) in addition to some instructions to their guardians to see their attending pediatricians if cough and/or fever persists for long, since the sensitivity of IGRA for bacteriologically positive TB is 80%[22] and probably lower for latent tuberculosis infection.
Ethical statement
This investigation did not need an ethical review because it used the infants' data on the Koch-like phenomenon after BCG vaccination that had already been released by MHLW and did not involve patients' confidential information.
Financial support and sponsorship
This study was partially supported by the Japan Agency for Medical Research and Development (Grant #: JP21fk0108127).
Conflicts of interest
There are no conflicts of interest.
References
Comments (0)