Reply to “Leukocyte receptor expression as a biomarker for severe sepsis”

We thank Dr. Johansson and Dr. Sj€ oberg for their valuable comments and critical considerations about our work. In our observational study, we were interested in the early expression of leukocyte surface molecules in severe sepsis. In addition, we were interested in the effect of major surgery on early expression of these surface molecules. Patients undergoing major surgery commonly present diagnostic difficulties when infection is suspected. Our approach was a case–control study instead of a randomly selected group of patients. It is true that this design could lead to overestimation of the diagnostic accuracy of a test. However, in this pilot phase, we were more interested in whether the tested surface molecules are even feasible in clearly selected groups. The next phase would be to test these markers in an unselected intensive care unit (ICU) patient population. As suggested, our severe sepsis patients may well have been in the pro-inflammatory stage of the disease. In earlier studies, the timing of sampling has been variable or not clearly stated, which may explain the variation in study results, as we discussed in our paper. For example, in a previous study, CD11b was reported to be lower in non-survivors because of the anti-inflammatory stage. In our study, compared with surviving sepsis patients, expression of monocyte and neutrophil CD11b was actually higher in the six patients with severe sepsis who died within 7 days (P < 0.05, data not shown). This outcome may be consistent with an early pro-inflammatory response. We agree that in our study, the severe sepsis patients had higher ICU scores than the controls. As suggested, optimal control groups would be patients with similar disease severity. Pancreatitis patients have been reported to have elevated CD11b expression and even higher expression with more severe disease. The infection complications, however, are more common with severe cases; thus, it may be impossible to differentiate which part of CD11b expression is the result of infection and the underlying acute pancreatitis or burns per se. Our approach allowed comparison of septic and non-septic ICU patients. We think that our conclusions relate appropriately to our results. Future studies are needed with an unselected ICU population to demonstrate or refute whether leukocyte markers can been used as diagnostic tests for severe sepsis and could affect treatment choices and subsequent outcomes.