Letter to the Editor: “Dynamic Pituitary-Adrenal Interactions in the Critically ill After Cardiac Surgery”

In your modeling analysis of cortisol appearance and elimination rates in critically ill (CI) patients and controls (1), your finding that cortisol secretion rates (CSR) are increased in CI is consistent with most (2, 3) but not all (4) previous studies. By contrast, your finding that cortisol half-lives are similar in CI and controls differs from previous studies (2, 3), as does your conclusion that maximal cortisol secretion rates (CSR_max) are decreased in CI (3). In order to better understand whether these discrepancies are related to assumptions made in your modeling analysis or unique aspects of the CI study group, we seek clarification regarding your modeling methods. First, it appears from the convolution integral (Methods) that you are using a single-compartment model based on total cortisol concentrations to estimate (total) cortisol appearance and elimination rates (1). The decision to use a single rather than 3-compartment model is curious, since your authorship has been instrumental in the development (5) and validation (6) of the 3-compartment model, which accounts for dynamic equilibrium of cortisol between free and protein-bound compartments. Since increased free cortisol concentrations (as a percent of total) are consistently observed in CI (3, 4), what was the rationale for using a model that may produce error by not accounting for individual or group differences in concentrations of free cortisol and cortisol binding proteins?

Second, your model for (total) cortisol elimination used a bi-exponential function with rapid and slow half-lives. Evidently the same value (2.41 minutes) for rapid cortisol half-life is assigned in both groups, and similar values for slow cortisol half-life are estimated: 53 vs 56 minutes for CI vs controls. How is the finding of similar cortisol half-lives in CI and controls reconciled with the consistent body of literature (3, 4) indicating longer cortisol half-life in CI? We are concerned that low corticosteroid binding globulin concentrations in the CI group would result in underestimation of cortisol half-life using the single compartment model (7). In addition, please clarify if there is an experimental basis for assignment of the same rapid half-life values to both groups. Also, please clarify what fractions of total cortisol you assigned to rapid vs slow half-life in your analysis.

Third, you report that CSR_max (defined as the cortisol secretion that obtains when ACTH concentrations at or above the threshold for maximal CSR; termed “efficacy” in Gibbison et al) is decreased in CI. By contrast, CSR_max was found to be normal or increased in CI relative to controls when was assessed directly during cosyntropin stimulation (3). The ACTH concentration at which CSR is maximal exceeds the upper limit of normal, and in healthy controls has been variably estimated to be 80 to 200 pg/mL (8). Although ACTH concentrations were not reported in your study, your finding of similar ACTH secretion and elimination rates in CI and controls is consistent with previous literature indicating normal or decreased ACTH concentrations in CI (2, 4). Therefore it seems to us that you may have too few measurements to derive the CSR-ACTH curve at ACTH concentrations above the inflection point of the logistic function. If you obtain CSR_max by extrapolation of existing data, please explain the extrapolation method and clarify whether you use symmetric or asymmetric logistic function (1, 5, 8) to extend the nonlinear ACTH-CSR curve.

Abbreviations

Abbreviations
 
• ACTH

  adrenocorticotropic hormone

 
• CI

  critically ill

 
• CSR

  cortisol secretion rate

 
• CSR_max

  maximal cortisol secretion rate

Acknowledgments

Financial Support: This work was supported with resources and the use of facilities at the New Mexico VA Healthcare System.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

References