Implementation of biomarkers in the management of Community-Acquired Pneumonia

Abstract

Chapter 1: We developed a real-time PCR test to detect pneumococci. The test was accurate both in vitro and oropharyngeal patient samples. It showed good reliability and has a detection threshold of 1-10 DNA copies per reaction, with a very low chance of false positive results. Chapter 2: We assessed a prediction score for identifying Legionella pneumonia at hospital admission. The score uses six variables: high fever, dry cough, low sodium, high LDH, high CRP, and low platelet count. The score was quite accurate, with an AUC of 0.89. A score of 4 or more showed a sensitivity of 58.8% and specificity of 93.1%. This score could help initiating Legionella specific antibiotic treatment Chapter 3: We investigated how levels of cytokines could discriminate between viral, pneumococcal, and mixed pneumonia. We found that specific cytokine patterns could predict viral pneumonia with high accuracy (AUC=0.911). This method shows promising results, but more prospective studies are needed. Chapter 4: We tested whether using a procalcitonin (PCT) and CRP algorithm could reduce the days on antibiotic treatment in patients hospitalized with community-acquired pneumonia (CAP). Patients guided by CRP or PCT had fewer days on antibiotics (4 and 5.5 days) than those receiving standard care (7 days). This approach can reduce antibiotic use by 30%. Chapter 5: We checked if MR-proADM levels could predict short-term problems like treatment failure or re-admission in CAP patients. MR-proADM levels did not show clear differences related to these outcomes, making it not very useful for predicting short-term risks.