1st, 2nd, and 3rdauthors deserve equal credit.

Red cell mass is tightly controlled by hypoxia, a process regulated by hypoxia-inducible transcription factors (HIFs). Obstructive sleep apnea (OSA) involves chronic intermittent hypoxia during sleep and has multiple organismal detrimental effects including hypertension, diabetes, cognitive impairment, heart attacks, and thrombotic complications. Polycythemia is not recognized as a feature of OSA, but we see occasional patients with polycythemia and OSA whose hematocrits are normalized by successful therapy with continuous positive airway pressure (CPAP).

We set out to analyze the effect of OSA on erythropoiesis in 18 non-smoking OSA patients without any chronic pulmonary disease and with normal hematocrits. Their blood was drawn before and after an overnight polysomnography (PSG) and, to account for possible circadian oscillation changes, at the same time of day. Patients ranged in OSA severity based on apnea-hypopnea index from mild to severe. An average of 98.5 minutes was spent with oxygen saturations of 89% or lower, with a 4% oxygen desaturation index of 22.9/hr. The average lowest oxygen saturation value was 78%, 6/18 subjects had hypertension and 4/18 diabetes. Three of these patients achieved remission of OSA over 3 months of successful therapy with CPAP, and were reanalyzed.

We considered the possibility that daily rapid changes from hypoxia to normoxia can lead to overcorrection of augmented erythropoiesis by preferential destruction of erythrocytes made in hypoxia; i.e. neocytolysis. We attempted to replicate data using our neocytolysis mouse model in OSA patients wherein neocytolysis is mediated by excessive generation of ROS from increased reticulocyte mitochondrial mass due to HIF-regulated reduction of Bnip3L levels upon return to normoxia. The reticulocyte-derived ROS then destroy hypoxia-produced young RBCs having reduced catalase (CAT) downregulated by increased miR-21 (see J. Song at this ASH). We also analyzed HIFs levels by QT-PCR of selected HIF-regulated gene transcripts (GLUT1, HK1, VEGFA, and PDK1) and the principal HIF inhibitor, PHD2, in reticulocytes. HK1 transcripts were higher in OSA patients than in controls. The expression of other HIF-targeted genes in OSA patients showed the same trend with HK1 expression but did not reach statistical significance.

In uncontrolled OSA (before and after overnight PSG), there were significant changes in CAT and BNIP3L transcript levels in reticulocytes compared to CPAP-corrected OSA. We found that CAT and BNIP3L transcript levels increased after CPAP treatment, indicating that in uncontrolled OSA, catalase and BNIP3L are low, causing generation of excessive ROS from mitochondria and, in keeping with this observation, we observed a significant decrease in miR21 leading to increased CAT. Compared to after correction of OSA, these BNIP3L and CAT changes would account for the absence of polycythemia during OSA. Increased mitochondrial mass generates additional ROS, thus replicating mouse neocytolysis in human OSA. Pre-OSA HIF activity is not markedly different from post-OSA HIF activity, perhaps reflecting the limitations of analyzing terminally differentiated cells. It is likely that miRNA changes in reticulocytes do not reflect acute changes that are present in erythroid progenitors with transcriptional activity. This limitation will be corrected by repeated analyses performed a few days after initiation of CPAP therapy.

We also analyzed erythropoiesis and observed a decreasing trend in erythropoietin level and erythroferrone transcripts after correction of OSA, findings compatible with augmented erythropoiesis in uncontrolled OSA.

Correlation of these and future data with those few OSA subjects who develop erythrocytosis is expected to uncover individual variations in hypoxic response leading to changes in HIFs response. Elucidation of erythroid, inflammatory (granulocyte analyses) and thrombotic (platelet analyses) OSA complications offers the possibility of targeted interventions by HIF inhibitors (i.e., digoxin) and/or HIF augmentation (PHD2 inhibitors).


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.