A blood protein and its fragments split bacterial pneumonia from viral
Telling bacterial pneumonia from viral pneumonia at the bedside is still mostly guesswork, and the guess decides whether a patient gets antibiotics they may not need. A study published this week in the Journal of Translational Medicine ↗ points at a familiar blood protein, and the small peptides it sheds, as a way to make that call with numbers instead of instinct.
The protein is alpha-1 antitrypsin, best known for the inherited deficiency that damages lungs and liver. It is an acute-phase protein, meaning its level in blood climbs during inflammation. The researchers measured it in 81 adults hospitalized with community-acquired pneumonia, 36 with a bacterial cause and 45 with a viral one, sampling on the day of admission and again on day three.
Alpha-1 antitrypsin ran higher in the bacterial group, and the gap was not a fluke. It reached statistical significance on admission (p equals 0.006) and grew stronger by day three (p below 0.001). What makes it interesting is that the signal held up independently of C-reactive protein, the inflammation marker doctors already order. A model combining the protein with age and white blood cell count separated bacterial from viral cases with an area under the curve of 0.803, where 1.0 is perfect and 0.5 is a coin flip.
The more novel part sits in the protein's breakdown products. Alpha-1 antitrypsin sheds short peptides from its C-terminal end, and the team could detect four of nine such fragments in patient blood. These behaved less like a single dial and more like separate readouts. One fragment, labeled C37, ran higher in bacterial infection (p equals 0.010). Another, C36, fell between admission and day three (p equals 0.006), tracking recovery. A third, C40, moved the opposite way, rising in viral cases (p equals 0.017), and the ratio of C40 to the parent protein was higher in viral infection too (p equals 0.014).
That divergence is the point. The whole protein leans bacterial, one fragment leans harder in the same direction, and another leans the other way toward viral. A panel that reads several of these at once could carry more information than any single marker, including the ratio between a fragment and the protein it came from, which cancels out some of the noise from how sick a patient is overall.
The limits are real and worth stating. Eighty-one patients is small, the split into bacterial and viral rests on the usual imperfect clinical workup, and no fragment here is ready to change a prescription. Peptide measurement by mass spectrometry is also not a bedside test yet. But the appeal is that this reuses a protein clinical labs already understand, and it suggests the fragments a body makes while fighting an infection carry a readable fingerprint of what kind of infection it is. The next step is a larger cohort to see whether the fragment signals hold when the patients are messier than a study population.