..DiscussionOur whole-blood expression profiling data indicate that H1N1 influenza A pneumonia has an immune response detectable at a gene-expression level. This immune response persists beyond the first 24 hours of admission to the ICU and was present throughout 5 days of follow-up. In addition, the H1N1 influenza A signature was highly consistent, as it remained detectable in a subset of patients with concurrent bacterial infection. Furthermore, this signature is highly specific to viral pneumonia (influenza A), because it is distinctively different from the gene-expression profile of bacterial pneumonia, or any conditions that may mimic an inflammatory host response. Our data therefore provide proof-of-concept evidence that gene-expression profiling may identify the etiology of acute pulmonary infection in critically ill patients, allowing more-specific patient care.Our study addresses an important issue in the current diagnosis of influenza infection. The current diagnosis in critically ill patients is difficult because of a lack of correlation between influenza virus antigen test and clinical status. For example, many influenza-infected individuals test negative for influenza virus [25]. Our study showed that the key to diagnosis is the presence of an abnormal immune response associated with influenza virus. This makes biological sense because it is the abnormal immune response that determines the progression to a more-severe illness, or sometimes, death. The 29-gene signature reflects the virus-specific host immune response. This allows influenza infection to be diagnosed correctly, independent of the result of the viral antigen test. Furthermore, the persistence of the 29-gene signature over time makes it possible to diagnose viral pneumonia for at least 5 days after ICU admission. Many critically ill patients present late, often with impending respiratory failure. By this stage, the viral shedding is minimal, and the pick-up rate for viral antigen testing is low. Another useful application of our gene-expression signature will be to assist the diagnosis in patients with bacterial coinfection. During the influenza season, many patients with bacterial pneumonia also test positive for the influenza virus, making it difficult to ascertain the etiology of the infection. The 29-gene viral gene signature has the potential to resolve diagnostic uncertainty in this situation by directly demonstrating the presence of a virus-specific immune response. These results warrant further exploration in a future diagnostic study in which the gene-expression signature can be validated in a large independent patient cohort.