Overlooked, dismissed, and downplayed: reversion of Mycobacterium tuberculosis immunoreactivity

Abstract

Tuberculosis (TB) is caused by Mycobacterium tuberculosis (Mtb). Following infection, immune responses to Mtb antigens can be measured using the tuberculin skin test or an interferon-γ release assay. The gain of Mtb immunoreactivity, a change from a negative to a positive tuberculin skin test or interferon-γ release assay result, is called conversion and has long been used as a measure of Mtb exposure. However, the loss of immunoreactivity (reversion; a positive followed by a negative result) has often been overlooked. Instead, in clinical and epidemiological circles, Mtb immunoreactivity is commonly considered to persist lifelong and confer a lifetime of disease risk. We present a critical review, describing the evidence for reversion from cohort studies, ecological studies and studies of TB progression risk. We outline the inconsistent reasons why reversion has been dismissed from common understanding and present evidence demonstrating that, just as conversion predominantly indicates prior exposure to Mtb antigens, so its opposite, reversion, suggests the reduction or absence of exposure (endogenous or exogenous). Mtb immunoreactivity is dynamic in both individuals and populations and this is why it is useful for stratifying short-term TB progression risk. The neglect of reversion has shaped TB research and policy at all levels, influencing clinical management and skewing Mtb infection risk estimation and transmission modelling, leading to an underestimation of the contribution of re-exposure to the burden of TB, a serious oversight for an infectious disease. More than a century after it was first demonstrated, it is time to incorporate reversion into our understanding of the natural history of TB.

FIGURE 1

Observed population prevalence of immunoreactivity a) ≥10 mm to old tuberculin 0.1 mg in a clinic population of 7765 people in Williamson County, TN, USA, 1951–1955 by age and sex, reproduced from [61] with permission; b) ≥5 mm to 5 tuberculin units (TU) purified protein derivative (PPD) in 177 259 people in Ontario, Canada, 1958–1960, by population and age, reproduced from [37] with permission; c) ≥5 mm to 5 TU PPD in an Inuit population of Alaska in 1957, by sex and age, reproduced from [62] with permission; and d) to PPD prepared by Florence Siebert, 0.00002 and 0.005 mg per 0.1 cm³ from 1% tuberculin solution among 8420 Pima Indians of Arizona, South Dakota, south-eastern Alaska and North Dakota, by tribe (unstated induration and study years, but the study year was likely to be ∼1935), reproduced from [63] with permission.

FIGURE 2

Prospective cohort studies that have monitored tuberculosis (TB) progression in those with differing Mycobacterium tuberculosis (Mtb) immunoreactivity for longer than 4 years in relatively low-incidence settings (TB incidence <∼50 per 100 000). a) Residents of the USA with recent household contact with a person with TB, recruited from 1957 to 1960 [45]. b) Residents of Muskogee county, OK, USA, screened in 1950 during a large trial of bacille Calmette–Guérin vaccination [93, 94]. c) Residents of Norway assessed with the QuantiFERON-TB Gold In-Tube (QFT-GIT) for varying reasons, screened between 2009 and 2014 [95]. The arrows indicate that reactivation rates given in these studies beyond the previous year label were not right censored. d) Those determined to be at high risk of latent TB infection were recruited from 54 UK National Health Service centres and community settings between 2010 and 2015 in London, Birmingham and Leicester [96, 97]. Additional studies are presented in supplementary figure S2. TST: tuberculin skin test.

FIGURE 3

Prospective cohort studies that have monitored tuberculosis (TB) progression in those with differing immunoreactivity for longer than 4 years in higher incidence settings (TB incidence ≥∼50 per 100 000). a) French school children, recruited from 1949 to 1951 [98]. b) English school children, recruited from 1951 to 1952 [92]. c) Edinburgh school children, recruited from 1955 to 1969 [99]. d) Indigenous peoples in Alberta, Canada, recruited from 1972 to 1979 [100]. An additional study is presented in supplementary figure S3. TST: tuberculin skin test.

FIGURE 4

Three case studies from the Victorian Tuberculosis (TB) Program (1999–2022) illustrating the dynamic nature of tuberculin skin test results, even within the relatively small windows of testing in the context of an individual's life. All individuals had one or more known exposures to people with TB, and “Exp” marks the age at these known exposures; “BCG” indicates those with a history of bacillus Calmette–Guérin vaccination (which is likely to confer a degree of immunoreactivity for an uncertain period) [87] and “TB” indicates the age of TB notification. Dotted lines and arrows are to assist test sequence interpretation and should not be regarded as the true changes in immunoreactivity.

FIGURE 5

Illustration of a) a common tuberculosis (TB) dynamic transmission model structure with susceptible (S), latent (L), TB disease (I) and recovered (R) compartments/states and b) a model structure incorporating reversion, with additional states and flows shown with dotted lines. Those with prior exposure who have reverted are represented by the S_x compartment, the latent (L) compartment has been relabelled as immunoreactive (Imm) and additional flows represent the reversion probability (τ), and a low reactivation rate (ν_S) from those who have reverted. This reflects the fact that the reactivation rate in those without immunoreactivity is low in the absence of re-exposure, but not zero. Other parameter definitions are given in supplementary table S2.

FIGURE 6

Deterministic compartmental tuberculosis models using the structure in figure 5b and with identical parameters (supplementary table S2) except for the annual reversion probability from the “immunoreactive” to the “susceptible-prior exposure” (S_x) compartments, which varies from a) no reversion (Model A), to an annual reversion probability of b) 2% (Model B) and c) 4% (Model C). The proportion of disease due to fast progression (versus reactivation of latent Mycobacterium tuberculosis (Mtb) infection) in the baseline simulation is noted in each figure. TST: tuberculin skin test; IGRA: interferon-γ release assay.