Variation in C - reactive protein response according to host and mycobacterial characteristics in active tuberculosis

James Brown, Kristina Clark, Colette Smith, Jennifer Hopwood, Oliver Lynard, Michael Toolan, Dean Creer, Jack Barker, Ronan Breen, Tim Brown, Ian Cropley, Marc Lipman

The C - reactive protein (CRP) response is often measured in patients with active tuberculosis (TB) yet little is known about its relationship to clinical features in TB, or whether responses differ between ethnic groups or with different Mycobacterium tuberculosis (M.tb) strain types. We report the relationship between baseline serum CRP prior to treatment and disease characteristics in a metropolitan population with TB resident in a low TB incidence region. People treated for TB at four London, UK sites between 2003 and 2014 were assessed and data collected on the following characteristics: baseline CRP level; demographics (ethnicity, gender and age); HIV status; site of TB disease; sputum smear (in pulmonary cases) and culture results. The effect of TB strain-type was also assessed in culture-positive pulmonary cases using VNTR typing data. Three thousands two hundred twenty-two patients were included in the analysis of which 72 % had a baseline CRP at or within 4 weeks prior to starting TB treatment. CRP results were significantly higher in culture positive cases compared to culture negative cases: median 49 mg/L (16-103 mg/L) vs 19 mg/L (IQR 5-72 mg/L), p = <0.001. In those with pulmonary disease, smear positive cases had a higher CRP than smear negative cases: 67 mg/L (31-122 mg/L) vs 24 mg/L (7-72 mg/L), p < 0.001. HIV positive cases had higher baseline CRPs than HIV negative cases: 75 mg/L (26-136 mg/L) vs 37 mg/L (10-88 mg/L), p <0.001. Differing sites of disease were associated with differences in baseline CRP: locations that might be expected to have a high mycobacterial load (e.g. pulmonary disease and disseminated disease) had a significantly higher CRP than those such as skin, lymph node or CNS disease, where the mycobacterial load is typically low in HIV negative subjects. In a multivariable log-scale linear regression model adjusting for host characteristics and M.tb strain type, infection with the East African Indian strain was associated with significantly lower baseline-CRP (fold-change in CRP 0.51 (0.34-0.77), p < 0.01). Host and mycobacterial factors are strongly associated with baseline CRP response in tuberculosis. This analysis suggests that there are important differences in innate immune response according to ethnicity, Mtb strain type and site of disease. This may reflect differing mycobacterial loads or host immune responses.