- Email: [email protected]
Long-term outcomes after AKI—a major unmet clinical need
commentary
Long-term outcomes after AKI—a major unmet clinical need Nicholas M. Selby1,2 and Maarten W. Taal1,2 In this issue of Kidney International, a comprehensive systematic review and meta-analysis provides an up-to-date picture of the long-term risks of death, chronic kidney disease (CKD), and end-stage kidney disease (ESKD) that follow an episode of acute kidney injury (AKI). Results confirm the significant event rate of these adverse outcomes following AKI and demonstrate that AKI severity and the clinical setting in which AKI occurs are important determinants of risk. In this commentary we discuss the implications of this study and how the results signal some key priorities for future research in an area of substantial clinical need. Kidney International (2019) 95, 21–23; https://doi.org/10.1016/j.kint.2018.09.005 Copyright ª 2018, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.
see clinical investigation on page 160
A
n early focus on the immediate complications and poor survival associated with AKI has been followed by a growing awareness of adverse long-term outcomes, which include increased risks CKD, cardiovascular events, heart failure, hospital readmission and fractures, along with higher mortality.1,2 AKI may be a mechanistic driver of these increased risks— for example, the pathways elucidated in experimental models by which AKI leads to CKD.3 Alternatively, the occurrence of AKI may be a signpost of more vulnerable, often multimorbid, patient groups. Although awareness has increased, we currently have no proven therapeutic strategies to reduce postAKI sequelae, nor do we have robust 1 Centre for Kidney Research and Innovation, School of Medicine, University of Nottingham, Derby, UK; and 2Department of Renal Medicine, Derby Teaching Hospitals NHS Foundation Trust, Derby, UK
Correspondence: Nicholas M. Selby, Centre for Kidney Research and Innovation, Division of Medical Sciences and Graduate Entry Medicine, University of Nottingham, Royal Derby Hospital Campus, Uttoxeter Road, Derby, DE22 3DT, UK. E-mail: [email protected] Kidney International (2019) 95, 19–30
evidence to inform strategies for the provision of health care. In this issue of Kidney International, See and colleagues4 report the results of a systematic review and meta-analysis of the long-term outcomes after AKI. Although previous meta-analyses have been performed,1,5 this well-conducted and more comprehensive study is timely. The authors used a sensitive search strategy and included only studies that used consensus definitions of AKI. This approach differs from that used in previous reviews in which previous definitions of AKI may have led to a bias toward participants receiving renal replacement therapy. In addition, studies were required to have a nonAKI comparator arm and a minimum of 1 year follow-up. Outcomes of interest were death, the development of new or progressive CKD, and end-stage kidney disease (ESKD). A total of 82 eligible studies incorporating more than 2 million patients were included. It comes as no surprise that patients who sustained AKI had higher odds of all of these outcomes compared with patients who did not sustain AKI. In addition, the severity of AKI was strongly and
consistently associated with increased risk across all outcomes. So what’s new? As well as confirming the adverse long-term consequences of AKI and reporting event rates relevant to current clinical practice, this systematic review shines a light on characteristics of the current evidence base, which in turn suggests some key priorities for future research. First, it is immediately obvious that most current data come from retrospective studies (77% of included studies, comprising 96.5% of patients; Figure 1). The authors reported significant heterogeneity across studies, although without any qualitative effects on the overall pattern of results, and although most studies were assessed as being good quality, few studies were at low risk of bias across all domains (including the usual caveats with retrospective studies relating to residual confounding and ascertainment bias). Additionally, none of the current studies incorporate the development of albuminuria in their definition of CKD. This is a major limitation given published data indicating that a substantial proportion of patients experience albuminuria after AKI and a large body of evidence identifying albuminuria as an independent risk factor for adverse outcomes.6 Perhaps of greatest importance, particularly because the link between AKI and long-term outcomes is now so clearly established, is that retrospective studies have limited ability to further our understanding of the mechanisms by which AKI is associated with adverse long-term outcomes. At present we do not know if long-term AKI sequelae are modifiable. If they are modifiable, we also need to determine which outcomes to target and with what (biologically plausible) interventions. Perhaps lessons can be taken from CKD, for which—in contrast to the AKI literature—more than 30 prospective cohort studies have been performed that together include almost three-quarters of a million patients.7 Such studies have generated novel insights into drivers of CKD progression and cardiovascular disease. Clinical risk prediction tools 21
commentary
-
a a a a a a a a a a a
Categories with at least one prospective study.
Figure 1 | Representation of studies included in the meta-analysis by See et al. by study type (prospective or retrospective) and by clinical setting. AAA, abdominal aortic aneurysm; AKI, acute kidney injury; ICU, intensive care unit; TAVI, transcatheter aortic valve implantation.
have been developed and validated and subsequently have been translated into clinical practice with endorsement of their use in international guidelines.7 For AKI, therefore, it appears essential that more prospective studies be performed with deep patient phenotyping, including assessment of albuminuria, long-term follow-up, and the capability of investigating mechanistic pathways. Second, See and colleagues4 highlight the importance of the clinical setting in which AKI occurs. A strength of the current study is that this information was reported transparently and the effects on outcomes were explored. Historically this aspect of AKI has been somewhat overlooked, but because AKI is a heterogeneous syndrome with multiple potential causes, it is logical to hypothesize that outcomes may differ across different AKI etiologies and patient groups. The authors4 reported that clinical setting did have an impact on outcomes, particularly with respect to mortality. Relative risks were particularly high in patients who sustained AKI after angiography compared with other settings. The hazard ratio for mortality was similar across cardiac surgery, general hospital, and intensive care unit populations, although this finding may be explained if differences in illness severity were reflected equally in both AKI and comparator groups. Again, this 22
aspect of the analysis provides direction for future research, because the limitations of the current evidence base did not allow more granular assessment, suggesting a need for improved reporting of AKI etiology and more detailed study of how this factor affects rates and mechanisms of long-term outcomes. Results also demonstrate the degree to which different clinical settings have been studied, with cardiac surgery being by far the most common and “general” hospital settings (where the majority of the caseload lies) relatively underrepresented (Figure 1). Third, it is important to move beyond relative risks associated with AKI versus non-AKI comparator groups and consider the differences in absolute event rates between the different longterm outcomes. In patients with AKI, pooled event rates were highest for new or progressive CKD (17.8 cases per 100 person years), followed by mortality (13.2 deaths per 100 person-years). Post-AKI CKD was driven mainly by the development of earlier stages of CKD, with much lower event rates for development of CKD stage G4 (1.3 cases per 100 person-years) and lowest of all for ESKD (0.5 cases per 100 person-years). One caveat is that the median duration of studies was only 2.9 years, which is likely to be insufficient to adequately study the development of ESKD. In addition, the vast number of
patients worldwide who sustain AKI annually means that even lower event rates can translate into a large absolute number of patients sustaining ESKD because of AKI. However, despite the fact that the event rates were derived from different studies and are not directly comparable, a clear pattern exists: earlier stages of CKD and mortality are common post-AKI sequelae, whereas advanced CKD and ESKD are less so. This balance of competing risks therefore poses a challenge; any strategies aimed at reducing the AKI to CKD transition must contend with the high rate of mortality, which has the potential to reduce the chances of successful clinical trials. Again, the solution would appear to be more data: can specific patient groups be identified in which competing risks differ and one particular outcome outweighs another? Obvious examples include stratification by age, in that younger patients (including pediatric cohorts) are likely to have a higher lifetime risk of ESKD (as has been reported in CKD populations8) and potentially would have the greatest benefit from interventions to reduce the AKI to CKD transition. Alternatively, identifying patients at higher risk of fluid overload/ heart failure may lead to targeting the avoidance of decompensation and hospital readmission. Further data to explore pathways by which post-AKI CKD links with cardiovascular disease also would be Kidney International (2019) 95, 19–30
commentary
valuable, particularly because the development of post-AKI CKD appears to modulate the risk of mortality.9 In addition to prospective research studies, the establishment and formal evaluation of the impact of post-AKI follow-up services is another strategy that may provide answers to some of these questions. In summary, the study by See et al.4 once again highlights the pressing clinical need posed by AKI by definitively describing the higher risks of long-term adverse outcomes and identifying groups at particular risk. At the same time, the study emphasizes critical areas in which more research is needed to advance our understanding to a point where interventions aimed at reducing the risk of long-term adverse outcomes ultimately can be developed and undergo trial. At present, the long-term future after AKI is anything but rosy; we must continue to work to change that situation. DISCLOSURE
All the authors declared no competing interests. REFERENCES 1. Coca SG, Yusuf B, Shlipak MG, et al. Long-term risk of mortality and other adverse outcomes after acute kidney injury: a systematic review and meta-analysis. Am J Kidney Dis. 2009;53(6): 961–973. 2. Wang WJ, Chao CT, Huang YC, et al. The impact of acute kidney injury with temporary dialysis on the risk of fracture. J Bone Miner Res. 2014;29(3):676–684. 3. Venkatachalam MA, Griffin KA, Lan R, et al. Acute kidney injury: a springboard for progression in chronic kidney disease. Am J Physiol Renal Physiol. 2010;298(5):F1078– F1094. 4. See EJ, Jayasinghe K, Glassford N, et al. Longterm risk of adverse outcomes after acute kidney injury: a systematic review and metaanalysis of cohort studies using consensus definitions of exposure. Kidney Int. 2019;95: 160–172. 5. Coca SG, Singanamala S, Parikh CR. Chronic kidney disease after acute kidney injury: a systematic review and meta-analysis. Kidney Int. 2012;81(5):442–448. 6. Horne KL, Packington R, Monaghan J, et al. Three-year outcomes after acute kidney injury: results of a prospective parallel group cohort study. BMJ Open. 2017;7(3): e015316. 7. Tangri N, Grams ME, Levey AS, et al. Multinational assessment of accuracy of equations for predicting risk of kidney failure: a meta-analysis. JAMA. 2016;315(2): 164–174. Kidney International (2019) 95, 19–30
8. van den Brand J, Pippias M, Stel VS, et al. Lifetime risk of renal replacement therapy in Europe: a population-based study using data from the ERA-EDTA Registry. Nephrol Dial Transplant. 2017;32(2):348–355.
9. Bucaloiu ID, Kirchner HL, Norfolk ER, et al. Increased risk of death and de novo chronic kidney disease following reversible acute kidney injury. Kidney Int. 2012;81(5): 477–485.
Stimulation of erythropoietin release by hypoxia and hypoxemia: similar but different Chang-Joon Lee1,2, David W. Smith1, Bruce S. Gardiner1,2 and Roger G. Evans3 Erythropoietin is released from the kidney in response to tissue hypoxia. Montero and Lundby found that increases in plasma erythropoietin induced by reducing arterial oxygen content in healthy humans were independent of arterial oxygen tension. Their observations accord with the established physiology of kidney oxygenation and can be predicted by a computational model of renal oxygen transport. However, model simulations indicate that the interpretation implicit in the title of their paper may be an oversimplification. Kidney International (2019) 95, 23–25; https://doi.org/10.1016/j.kint.2018.09.025 Copyright ª 2018, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.
see clinical investigation on page 173
B
lood levels of erythropoietin (Epo) increase in humans when they ascend to altitude and so are exposed to a lower partial pressure of oxygen (PO2) than experienced at sea level (a.k.a. hypoxia).1 Blood Epo concentration is also increased in human subjects2 and experimental animals3 subjected to acute or chronic anemia (a.k.a. normoxic hypoxemia). In all these cases, the dominant stimulus for release
1 Faculty of Engineering and Mathematical Sciences, The University of Western Australia, Perth, Australia; 2School of Engineering and Information Technology, Murdoch University, Perth, Australia; and 3Cardiovascular Disease Program, Biomedicine Discovery Institute and Department of Physiology, Monash University, Melbourne, Australia
Correspondence: Roger Evans, Department of Physiology, P.O. Box 13F, Monash University, Victoria 3800, Australia. E-mail: [email protected]
of Epo is renal tissue hypoxia, which in turn results in reduced breakdown of hypoxia-inducible factor (HIF)-2a in peritubular interstitial fibroblasts in the inner cortex and outer medulla, the major source of Epo.4 HIF-2a then forms a dimer with HIF-b, which induces transcription of the EPO gene.4 In the current issue of Kidney International, Montero and Lundby5 (2018) report the results of a carefully controlled blinded crossover study in healthy male volunteers, comparing the effects on the plasma concentration of Epo, of hypoxia (11% inspired oxygen) and normoxic hypoxemia (inspired carbon monoxide titrated to achieve w18% circulating carboxyhemoglobin). The use of carbon monoxide breathing allowed them to reduce the oxygen-carrying capacity of arterial blood without the need for hemodilution. The 2 interventions resulted 23
Long-term outcomes after AKI—a major unmet clinical need Nicholas M. Selby1,2 and Maarten W. Taal1,2 In this issue of Kidney International, a comprehensive systematic review and meta-analysis provides an up-to-date picture of the long-term risks of death, chronic kidney disease (CKD), and end-stage kidney disease (ESKD) that follow an episode of acute kidney injury (AKI). Results confirm the significant event rate of these adverse outcomes following AKI and demonstrate that AKI severity and the clinical setting in which AKI occurs are important determinants of risk. In this commentary we discuss the implications of this study and how the results signal some key priorities for future research in an area of substantial clinical need. Kidney International (2019) 95, 21–23; https://doi.org/10.1016/j.kint.2018.09.005 Copyright ª 2018, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.
see clinical investigation on page 160
A
n early focus on the immediate complications and poor survival associated with AKI has been followed by a growing awareness of adverse long-term outcomes, which include increased risks CKD, cardiovascular events, heart failure, hospital readmission and fractures, along with higher mortality.1,2 AKI may be a mechanistic driver of these increased risks— for example, the pathways elucidated in experimental models by which AKI leads to CKD.3 Alternatively, the occurrence of AKI may be a signpost of more vulnerable, often multimorbid, patient groups. Although awareness has increased, we currently have no proven therapeutic strategies to reduce postAKI sequelae, nor do we have robust 1 Centre for Kidney Research and Innovation, School of Medicine, University of Nottingham, Derby, UK; and 2Department of Renal Medicine, Derby Teaching Hospitals NHS Foundation Trust, Derby, UK
Correspondence: Nicholas M. Selby, Centre for Kidney Research and Innovation, Division of Medical Sciences and Graduate Entry Medicine, University of Nottingham, Royal Derby Hospital Campus, Uttoxeter Road, Derby, DE22 3DT, UK. E-mail: [email protected] Kidney International (2019) 95, 19–30
evidence to inform strategies for the provision of health care. In this issue of Kidney International, See and colleagues4 report the results of a systematic review and meta-analysis of the long-term outcomes after AKI. Although previous meta-analyses have been performed,1,5 this well-conducted and more comprehensive study is timely. The authors used a sensitive search strategy and included only studies that used consensus definitions of AKI. This approach differs from that used in previous reviews in which previous definitions of AKI may have led to a bias toward participants receiving renal replacement therapy. In addition, studies were required to have a nonAKI comparator arm and a minimum of 1 year follow-up. Outcomes of interest were death, the development of new or progressive CKD, and end-stage kidney disease (ESKD). A total of 82 eligible studies incorporating more than 2 million patients were included. It comes as no surprise that patients who sustained AKI had higher odds of all of these outcomes compared with patients who did not sustain AKI. In addition, the severity of AKI was strongly and
consistently associated with increased risk across all outcomes. So what’s new? As well as confirming the adverse long-term consequences of AKI and reporting event rates relevant to current clinical practice, this systematic review shines a light on characteristics of the current evidence base, which in turn suggests some key priorities for future research. First, it is immediately obvious that most current data come from retrospective studies (77% of included studies, comprising 96.5% of patients; Figure 1). The authors reported significant heterogeneity across studies, although without any qualitative effects on the overall pattern of results, and although most studies were assessed as being good quality, few studies were at low risk of bias across all domains (including the usual caveats with retrospective studies relating to residual confounding and ascertainment bias). Additionally, none of the current studies incorporate the development of albuminuria in their definition of CKD. This is a major limitation given published data indicating that a substantial proportion of patients experience albuminuria after AKI and a large body of evidence identifying albuminuria as an independent risk factor for adverse outcomes.6 Perhaps of greatest importance, particularly because the link between AKI and long-term outcomes is now so clearly established, is that retrospective studies have limited ability to further our understanding of the mechanisms by which AKI is associated with adverse long-term outcomes. At present we do not know if long-term AKI sequelae are modifiable. If they are modifiable, we also need to determine which outcomes to target and with what (biologically plausible) interventions. Perhaps lessons can be taken from CKD, for which—in contrast to the AKI literature—more than 30 prospective cohort studies have been performed that together include almost three-quarters of a million patients.7 Such studies have generated novel insights into drivers of CKD progression and cardiovascular disease. Clinical risk prediction tools 21
commentary
-
a a a a a a a a a a a
Categories with at least one prospective study.
Figure 1 | Representation of studies included in the meta-analysis by See et al. by study type (prospective or retrospective) and by clinical setting. AAA, abdominal aortic aneurysm; AKI, acute kidney injury; ICU, intensive care unit; TAVI, transcatheter aortic valve implantation.
have been developed and validated and subsequently have been translated into clinical practice with endorsement of their use in international guidelines.7 For AKI, therefore, it appears essential that more prospective studies be performed with deep patient phenotyping, including assessment of albuminuria, long-term follow-up, and the capability of investigating mechanistic pathways. Second, See and colleagues4 highlight the importance of the clinical setting in which AKI occurs. A strength of the current study is that this information was reported transparently and the effects on outcomes were explored. Historically this aspect of AKI has been somewhat overlooked, but because AKI is a heterogeneous syndrome with multiple potential causes, it is logical to hypothesize that outcomes may differ across different AKI etiologies and patient groups. The authors4 reported that clinical setting did have an impact on outcomes, particularly with respect to mortality. Relative risks were particularly high in patients who sustained AKI after angiography compared with other settings. The hazard ratio for mortality was similar across cardiac surgery, general hospital, and intensive care unit populations, although this finding may be explained if differences in illness severity were reflected equally in both AKI and comparator groups. Again, this 22
aspect of the analysis provides direction for future research, because the limitations of the current evidence base did not allow more granular assessment, suggesting a need for improved reporting of AKI etiology and more detailed study of how this factor affects rates and mechanisms of long-term outcomes. Results also demonstrate the degree to which different clinical settings have been studied, with cardiac surgery being by far the most common and “general” hospital settings (where the majority of the caseload lies) relatively underrepresented (Figure 1). Third, it is important to move beyond relative risks associated with AKI versus non-AKI comparator groups and consider the differences in absolute event rates between the different longterm outcomes. In patients with AKI, pooled event rates were highest for new or progressive CKD (17.8 cases per 100 person years), followed by mortality (13.2 deaths per 100 person-years). Post-AKI CKD was driven mainly by the development of earlier stages of CKD, with much lower event rates for development of CKD stage G4 (1.3 cases per 100 person-years) and lowest of all for ESKD (0.5 cases per 100 person-years). One caveat is that the median duration of studies was only 2.9 years, which is likely to be insufficient to adequately study the development of ESKD. In addition, the vast number of
patients worldwide who sustain AKI annually means that even lower event rates can translate into a large absolute number of patients sustaining ESKD because of AKI. However, despite the fact that the event rates were derived from different studies and are not directly comparable, a clear pattern exists: earlier stages of CKD and mortality are common post-AKI sequelae, whereas advanced CKD and ESKD are less so. This balance of competing risks therefore poses a challenge; any strategies aimed at reducing the AKI to CKD transition must contend with the high rate of mortality, which has the potential to reduce the chances of successful clinical trials. Again, the solution would appear to be more data: can specific patient groups be identified in which competing risks differ and one particular outcome outweighs another? Obvious examples include stratification by age, in that younger patients (including pediatric cohorts) are likely to have a higher lifetime risk of ESKD (as has been reported in CKD populations8) and potentially would have the greatest benefit from interventions to reduce the AKI to CKD transition. Alternatively, identifying patients at higher risk of fluid overload/ heart failure may lead to targeting the avoidance of decompensation and hospital readmission. Further data to explore pathways by which post-AKI CKD links with cardiovascular disease also would be Kidney International (2019) 95, 19–30
commentary
valuable, particularly because the development of post-AKI CKD appears to modulate the risk of mortality.9 In addition to prospective research studies, the establishment and formal evaluation of the impact of post-AKI follow-up services is another strategy that may provide answers to some of these questions. In summary, the study by See et al.4 once again highlights the pressing clinical need posed by AKI by definitively describing the higher risks of long-term adverse outcomes and identifying groups at particular risk. At the same time, the study emphasizes critical areas in which more research is needed to advance our understanding to a point where interventions aimed at reducing the risk of long-term adverse outcomes ultimately can be developed and undergo trial. At present, the long-term future after AKI is anything but rosy; we must continue to work to change that situation. DISCLOSURE
All the authors declared no competing interests. REFERENCES 1. Coca SG, Yusuf B, Shlipak MG, et al. Long-term risk of mortality and other adverse outcomes after acute kidney injury: a systematic review and meta-analysis. Am J Kidney Dis. 2009;53(6): 961–973. 2. Wang WJ, Chao CT, Huang YC, et al. The impact of acute kidney injury with temporary dialysis on the risk of fracture. J Bone Miner Res. 2014;29(3):676–684. 3. Venkatachalam MA, Griffin KA, Lan R, et al. Acute kidney injury: a springboard for progression in chronic kidney disease. Am J Physiol Renal Physiol. 2010;298(5):F1078– F1094. 4. See EJ, Jayasinghe K, Glassford N, et al. Longterm risk of adverse outcomes after acute kidney injury: a systematic review and metaanalysis of cohort studies using consensus definitions of exposure. Kidney Int. 2019;95: 160–172. 5. Coca SG, Singanamala S, Parikh CR. Chronic kidney disease after acute kidney injury: a systematic review and meta-analysis. Kidney Int. 2012;81(5):442–448. 6. Horne KL, Packington R, Monaghan J, et al. Three-year outcomes after acute kidney injury: results of a prospective parallel group cohort study. BMJ Open. 2017;7(3): e015316. 7. Tangri N, Grams ME, Levey AS, et al. Multinational assessment of accuracy of equations for predicting risk of kidney failure: a meta-analysis. JAMA. 2016;315(2): 164–174. Kidney International (2019) 95, 19–30
8. van den Brand J, Pippias M, Stel VS, et al. Lifetime risk of renal replacement therapy in Europe: a population-based study using data from the ERA-EDTA Registry. Nephrol Dial Transplant. 2017;32(2):348–355.
9. Bucaloiu ID, Kirchner HL, Norfolk ER, et al. Increased risk of death and de novo chronic kidney disease following reversible acute kidney injury. Kidney Int. 2012;81(5): 477–485.
Stimulation of erythropoietin release by hypoxia and hypoxemia: similar but different Chang-Joon Lee1,2, David W. Smith1, Bruce S. Gardiner1,2 and Roger G. Evans3 Erythropoietin is released from the kidney in response to tissue hypoxia. Montero and Lundby found that increases in plasma erythropoietin induced by reducing arterial oxygen content in healthy humans were independent of arterial oxygen tension. Their observations accord with the established physiology of kidney oxygenation and can be predicted by a computational model of renal oxygen transport. However, model simulations indicate that the interpretation implicit in the title of their paper may be an oversimplification. Kidney International (2019) 95, 23–25; https://doi.org/10.1016/j.kint.2018.09.025 Copyright ª 2018, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.
see clinical investigation on page 173
B
lood levels of erythropoietin (Epo) increase in humans when they ascend to altitude and so are exposed to a lower partial pressure of oxygen (PO2) than experienced at sea level (a.k.a. hypoxia).1 Blood Epo concentration is also increased in human subjects2 and experimental animals3 subjected to acute or chronic anemia (a.k.a. normoxic hypoxemia). In all these cases, the dominant stimulus for release
1 Faculty of Engineering and Mathematical Sciences, The University of Western Australia, Perth, Australia; 2School of Engineering and Information Technology, Murdoch University, Perth, Australia; and 3Cardiovascular Disease Program, Biomedicine Discovery Institute and Department of Physiology, Monash University, Melbourne, Australia
Correspondence: Roger Evans, Department of Physiology, P.O. Box 13F, Monash University, Victoria 3800, Australia. E-mail: [email protected]
of Epo is renal tissue hypoxia, which in turn results in reduced breakdown of hypoxia-inducible factor (HIF)-2a in peritubular interstitial fibroblasts in the inner cortex and outer medulla, the major source of Epo.4 HIF-2a then forms a dimer with HIF-b, which induces transcription of the EPO gene.4 In the current issue of Kidney International, Montero and Lundby5 (2018) report the results of a carefully controlled blinded crossover study in healthy male volunteers, comparing the effects on the plasma concentration of Epo, of hypoxia (11% inspired oxygen) and normoxic hypoxemia (inspired carbon monoxide titrated to achieve w18% circulating carboxyhemoglobin). The use of carbon monoxide breathing allowed them to reduce the oxygen-carrying capacity of arterial blood without the need for hemodilution. The 2 interventions resulted 23