Is the arteriovenous carboxyhemoglobin gradient really a useful marker in systemic inflammation?

We read with great interest the study in CHEST by Dr. Yasuda and colleagues (June 2004)1 elucidating the role of arteriovenous carboxyhemoglobin (a-vHb-CO) difference in patients with fevers of unknown origin. The authors reported that the presence of an a-vHb-CO difference may be a useful indicator to differentiate between pulmonary and extrapulmonary diseases. This hypothesis was generated from the idea that, in the setting of pulmonary inflammation, a-vHb-CO differences may be caused by increased carbon monoxide (CO) production secondary to the up-regulation of heme oxygenase-1.2 Conversely, we recently demonstrated that the a-vHb-CO gradient may just be a technical artifact that can be avoided by a special calibration (SAT100; Radiometer; Copenhagen, Denmark) eliminating the fetal COHb dependency on oxygen saturation.3 In addition, we evaluated the implication of a-vHb-CO differences in healthy and endotoxemic sheep, and found (1) that the a-vHb-CO difference, per se, does not reflect critical illness and (2) that measurements made without the special calibration (SAT100 calibration for ABL 625 blood gas analyzer; Radiometer; Copenhagen, Denmark) underestimate COHb concentrations measured with the blood gas analyzer (ABL 725 blood gas analyzer; Radiometer) in which this special calibration was routinely performed.4 In another in vitro experiment,5 we exposed venous blood samples to fixed CO doses at incremental oxygen concentrations and showed that the affinity of CO (200 and 400 ppm) to hemoglobin progressively increased with an inspiratory oxygen fraction from 0 to 15%, whereas at higher oxygen tensions this effect vanished. In summary, there is considerable evidence that the presence of an a-vHb-CO difference is influenced by the degree of tissue oxygenation. Therefore, a-vHb-CO difference is not inevitably a useful marker to define the site of inflammation. Due to current knowledge, the COHb levels measured by Yasuda et al1 are methodologically questionable, especially since a special calibration (SAT100) was not performed.