Response: The in vivo half-life of human neutrophils

We have read the comments raised by Tofts and colleagues and Li and colleagues with great interest. We share their surprise about our finding that—under normal homeostatic conditions—peripheral blood neutrophils have an average life span of 5.4 days, and we welcome their constructive replies to this important topic.

Tofts et al correctly show that an alternative model in which neutrophils have an infinitely short life span in the peripheral blood provides an equally good fit to our deuterium data. However, their model is compatible with our data only if neutrophils spend approximately 11 days in the bone marrow after having completed their last division in the marrow. If neutrophils were to leave the bone marrow sooner after their last division, the whole curve presented by Tofts et al would shift to the left and would no longer provide a good fit to our data. We neglected the possibility presented by Tofts et al, because in vivo data show that the delay in the postmitotic pool is much shorter than 11 days. In one individual included in our study, whose first measurement was taken at day 7, deuterium-labeled neutrophils could already be found in the peripheral blood. Also our best fits to the data presented in Table 1 of our original article suggested that the delay after the last division in the bone marrow is at most 6 days. These estimates are completely in line with other in vivo labeling studies of granulocytes in humans under homeostatic conditions with ³H-thymidine¹  (salvage pathway) or ²H-glucose²  (de novo pathway), which both yielded a delay of granulocytes in the postmitotic pool of 6.5-7 days. Thus, although the total residence time (mitotic and postmitotic pool) in the bone marrow may be as long as—or even longer than—11.2 days,³  the delay in Tofts' alternative model (parameter b) reflects the delay after the last division in the bone marrow, which should be much shorter.^1,2 

The authors also question the “inexplicably high” level of enrichment that we measured in the deoxyribonucleic acid of blood neutrophils. It is important to realize, however, that our method analyzes deuterium enrichment in deoxyribose (dR) derivatives that harbor multiple hydrogen atoms independent of the particular hydrogen location that is enriched. Because the enrichment in body water is low, double-labeling within one dR hardly occurs, and the fraction of single-labeled dR derivatives is thus a nearly linear sum of the enrichment at each of the dR hydrogen locations. This allows the fraction of dR derivatives with one hydrogen atom enriched to exceed the level of enrichment found in the body water, as we and others have previously described.^4,5 

Taken together, we think the issues raised by Tofts et al do not affect the interpretation of our results.

Li et al, however, raise an alternative possibility that may have large implications for the interpretation of our data. They propose that the slow kinetics of labeled cells observed in the blood may in fact be because of the slow production of neutrophils in the bone marrow, rather than a long half-life in the blood. Indeed, if neutrophil production in the bone marrow were to be the rate limiting kinetic step (ie, if the average time between subsequent divisions of neutrophil precursors in the marrow would be of the order of days), the resulting labeling kinetics of human neutrophils in the blood could be much slower than their actual dynamics in the blood. Unfortunately, in the absence of labeling data from the bone marrow of healthy humans, we cannot exclude this alternative possibility. Our concern remains, however, that previous studies based on ex vivo labeling, on labeling with DFP³² (which might affect neutrophil activation⁶ ) or on non–steady state or toxic conditions, have likely caused aberrant (homing) behavior of neutrophils⁷  and may thereby have underestimated the circulatory half-life of blood neutrophils. The only way to distinguish between the 2 alternative possibilities proposed by Li et al is to investigate the turnover of bone marrow neutrophils in healthy humans using in vivo labeling techniques such as deuterium labeling, to ensure that neutrophil kinetics are studied under normal physiologic circumstances. Given the importance of the issue, we very much support such future in vivo labeling studies of neutrophils in the bone marrow.