- Email: [email protected]
Nocturnal hypoxaemia and central-nervous-system events in sickle-cell disease
ARTICLES
Nocturnal hypoxaemia and central-nervous-system events in sickle-cell disease F J Kirkham, D K M Hewes, M Prengler, A Wade, R Lane, J P M Evans
Summary Background Central-nervous-system (CNS) events, including strokes, transient ischaemic attacks, and seizures are common in sickle-cell disease. Stroke can be predicted by high velocities in the internal-carotid or middle-cerebral arteries on transcranial doppler ultrasonography. We tested the hypothesis that nocturnal hypoxaemia can predict CNS events better than clinical or haematological features, or transcranial doppler sonography. Methods We screened 95 hospital-based patients with sickle-cell disease (median age 7·7 years [range 1·0–23·1]), but without previous stroke, with transcranial doppler and overnight pulse oximetry. Follow-up continued for a median of 6·01 (0·11–8·54) years. Findings 19 patients had CNS events (six ischaemic and one haemorrhagic stroke, eight transient ischaemic attacks, and four seizures). Mean overnight oxygen saturation ([SaO2] hazard ratio 0·82 per 1% increase [95% CI 0·71–0·93]; p=0·003) and higher internal-carotid or middle-cerebral artery velocity (1·02 for every increase of 1 cm/s [1·004–1·03]; p=0·009) were independently associated with time to CNS event. After accounting for mean SaO2, artery velocity, and haemoglobinopathy, high haemoglobin concentration was also associated with an increased risk of CNS event (1·7 per g/dL, [1·18–2·43]; p=0·004). Dips suggestive of obstructive sleep apnoea did not predict CNS events, and adenotonsillectomy seemed to have no effect, although the CI were wide and clinically important effects cannot be excluded. Interpretation Screening for, and appropriate management of, nocturnal hypoxaemia might be a safe and effective alternative to prophylactic blood transfusion for primary prevention of CNS events in sickle-cell disease. Lancet 2001; 357: 1656–59
Neurosciences Unit (F J Kirkham MB, D K M Hewes MB, M Prengler MD), Paediatric Epidemiology and Biostatistics Unit (A Wade PhD), Portex Unit of Anaesthesia, Intensive Care and Respiratory Medicine (R Lane PhD), and Department of Haematology (J P M Evans DM), Institute of Child Health, University College London, London, UK; and Great Ormond Street Hospital, London, UK (F J Kirkham, D K M Hewes, M Prengler, A Wade, R Lane, J P M Evans) Correspondence to: Dr F J Kirkham, the Wolfson Centre, Mecklenburgh Square, London WC1N 2AP, UK (e-mail: [email protected])
1656
Introduction Sickle-cell disease is a group of inherited haemoglobin disorders, including sickle-cell anaemia (homozygous SS), SC disease, and S-thalassaemia. Sickle-cell disease is the most common cause of childhood stroke, with an incidence of 0·28–0·61%, similar to normal elderly adults;1–3 25% and 10% of patients with SS and SC, respectively, have had a stroke by age 45 years.2 Stenosis or occlusion of the basal cerebral vessels, especially distal-internal-carotid or proximal-middle-cerebral arteries, is characteristic.4 Cerebrovascular disease is seen in young children5 with ischaemic stroke, and cerebral haemorrhage arises mainly in adults. Seizures are more common in patients with sickle-cell disease than in the rest of the population,6 can be difficult to distinguish from transient ischaemic attacks, and are associated with imaging abnormalities.7 Cerebral infarction (overt and covert) is associated with cognitive difficulties.8 Risk factors for ischaemic stroke include low fetal haemoglobin,9 recent chest crisis,3 and hypertension.3 Infants with dactylitis, haemoglobin values less than 7 g/dL, or leucocytosis are at risk of adverse outcomes, including stroke,2,10 and might be candidates for prophylactic therapy—for example, hydroxyurea to increase fetal haemoglobin concentrations. Transcranial doppler sonography predicts stroke risks of 40%, 7%, and 2% over the following 40 months in patients with velocities of more than 200 or more than 170, or less than 170 cm/s, respectively, in the internal-carotid or middle-cerebral arteries.11 Primary stroke prevention with long-term blood transfusion is possible in children with velocities greater than 200 cm/s.12 The risks of infection, alloimmunisation, and iron overload, needing parenteral chelating agents, make this approach punishing and potentially life threatening, however, which justifies the investigation of alternative preventative strategies. Snoring is a risk factor for stroke in adults.13 Both episodic and continuous nocturnal hypoxaemia are common in sickle-cell disease,14 possibly because of upper-airway obstruction secondary to adenotonsillar hypertrophy. The reported association between obstructive sleep apnoea and central-nervous-system (CNS) events in sickle-cell disease15,16 has not been studied prospectively. We tested the hypothesis that nocturnal hypoxaemia, documented by overnight pulse oximetry, offers better prediction of CNS events than clinical, haematological, or transcranial doppler sonography variables.
Methods Patients Permission for the study was granted by the Great Ormond Street Hospital Local Research Ethics committee, and written informed consent was received from all patients. From Jan 1, 1991, we invited all patients who had not had a stroke and were regularly attending the haemoglobinopathy clinic of Queen Elizabeth Hospital, London, UK, to participate. Follow-up was discontinued on April 30, 2000.
THE LANCET • Vol 357 • May 26, 2001
ARTICLES
Study protocol We used a TC-64B (Eden Medical Electronics, Uberlingen, Germany) or a Companion (Nicolet, Warwick, UK) transcranial doppler sonography machine to insonate the distal internal carotid, middle, anterior, posterior-cerebral, and basilar arteries.17 We recorded the highest average velocity measured on both sides of the internal-carotid and middle-cerebral arteries. We assessed patients at recruitment, at the time of the sleep study, and at follow-up in 1998. Individuals with high velocities were assessed more frequently than other patients. We used a pulse oximeter (Biox 3700, Datex-Ohmeda, Hatfield, Hertfordshire, UK) to record oxygen saturation (SaO2) during sleep. DKMH or RL, who were unaware of the clinical data, analysed the results. We measured the SaO2 of 63 patients at home. In these patients, parents turned on the oximeter after sleep onset. The rest of the study population visited the hospital sleep laboratory. Studies lasted 4·3–8·2 h (median 7·5). We excluded movement artefacts manually. We recorded the mean and minimum oxygen saturations and the proportion of sleep spent at saturations less than 90% and less than 80%, and examined the trace for dips (>4% from baseline) in oxygenation associated with acute pulse rate rises—a pattern suggestive of obstructive sleep apnoea. Patients with symptoms of obstructive sleep apnoea were referred for surgical management without reference to overnight pulse oximetry data. We obtained retrospective data on dactylitis in the first year of life, mean early haemoglobin concentrations and early white blood cell counts from year 2, baseline haemoglobin on the day of doppler sonography, and haemoglobin F from 1993 (except in those with SC disease). We recorded data on previous headache or transient ischaemic attack and new CNS events (seizures or focal deficits lasting about 24 h coded as stroke or transient ischaemic attack, respectively) at monthly joint haematology and neurology clinics. Magnetic resonance imaging and magnetic resonance angiography were done as soon as possible.18,19 We reviewed patients in 1998. Statistical analysis We compared patients who did and did not undergo overnight pulse oximetry. Exact tests and CI were calculated for categorical variables with StatXact (version 4·0·1). Mann-Whitney U tests were used to compare the medians of continuous variables. Associations between overnight pulse oximetry measurements and other variables were investigated with Spearman’s correlation coefficients, 2, Mann-Whitney U, and Kruskal-Wallis tests as appropriate. We used Cox’s proportional hazard regression models to identify predictors of subsequent CNS events, taking into account the varying length of follow-up. Time to CNS event from overnight pulse oximetry was investigated. The transcranial doppler sonography measurement made closest to the overnight data was used in the model. Adenotonsillectomy, which could have been done some time after the overnight pulse oximetry and transcranial doppler sonography studies, was entered as a time-varying covariate.20 Hazard ratios are presented with 95% CI. A Kaplan-Meier survival curve illustrates the relation between sleep study data and subsequent risk of stroke.
Results We enrolled 147 patients. Overnight pulse oximetry was done in 65% of these (table 1) at a median time of 218 days after recruitment. Before 1994, most had overnight
THE LANCET • Vol 357 • May 26, 2001
Overnight pulse oximetry
Sex (M/F) Type of sickle-cell disease Homozygous SC disease S-thalassaemia History Previous transient ischaemic attack Previous headaches Dactylitis before age 1 year Adenotonsillectomy Age at recruitment (years)
p
Yes (n=95)
No (n=52)
54 (57%)/41 (43%)
32 (62%)/20 (38%)
0·60 0·44
74 (78%) 14 (15%) 7 (7%)
37 (71%) 12 (23%) 3 (6%)
19 (20%)
3 (6%)
32 (34%) 23 (29%)* 27 (28%) 6·65 (4·71–9·50)
0·09
10 (19%) 6 (16%)† 4 (8%) 6·52 (3·68–10·44)
Haematological characteristics Haemoglobin concentration 8·1 (7·5–10·2)‡ 8·4 (7·7–9·9)§ at recruitment (g/dL) Mean haemoglobin between 8·9 (8·3–10·0)|| 8·7 (7·8–9·9)¶ 11 and 25 months (g/dL) Mean white cell count between 11·8 (8·3–15·0)** 10·8 (9·0–14·9)‡‡ 11 and 25 months (109/L) Higher (right or left) ICA or MCA 120·0 (100·0–139·0)§§ 109·0 (98·5–130·5) velocity at recruitment (cm/s)
0·10 0·18 0·02 0·35 0·57 0·56 0·67 0·28
Numbers and percentages are given for categorical variables, medians and IQR for continuous variables. ICA=internal-carotid artery; MCA=middle-cerebral artery. *n=79; †n=38; ‡n=93; §n=46; ||n=78; ¶n=30; **n=74; ‡‡n=29; §§n=92.
Table 1: Baseline characteristics of patients
pulse oximetry at home; thereafter patients were offered hospital sleep studies but often found it difficult to attend. 19 patients (17 with sickle-cell anaemia) had CNS events during follow-up (six ischaemic and one haemorrhagic stroke, eight transient ischaemic events, and four seizures). One patient with S-thalassaemia had a stroke. A child with SC disease and hydrocephalus died of cerebral oedema after developing severe headaches (normal shunt pressure), three vertebrobasilar transient ischaemic attacks, and a seizure. Two patients died from non-CNS causes (chest syndrome and sepsis) and were censored in the statistical analysis. Four patients with ischaemic stroke had only deep-white-matter lesions and one had a right parieto-occiptal cortical lesion; all had large vessel disease on magnetic resonance angiography. One patient developed bilateral frontal and parietal watershed lesions after seizures associated with facial infection, and another had a severe headache and mild left hemiparesis with three acute left-sided haemorrhagic lesions and developed right-sided deep-white-matter lesions 4 days later.19 These two children had normal magnetic resonance angiography and conventional arteriography. Five (including the SC patient) had vertebrobasilar transient ischaemic attacks with dizziness, diplopia, visual symptoms, ataxia, paraesthesiae, and sudden collapse. None of these five patients had infarction on magnetic resonance imaging, and only one had an abnormal magnetic resonance angiogram with bilateral turbulence in the terminal internal-carotid artery. Three individuals had transient hemiparesis consistent with anterior-circulation transient ischaemic attack, of whom two had small deep-white-matter lesions and bilateral middle-cerebral or internal-carotid artery turbulence on magnetic resonance angiography, and one had a small hemisphere contralaterally with normal magnetic resonance angiography. Four children had seizures: three had normal magnetic resonance imaging scans (one with normal magnetic resonance angiography and two with bilateral turbulence), and one had bilateral deep-whitematter lesions and magnetic resonance imaging evidence of left-sided internal-carotid-artery dissection. Previous transient ischaemic attack (95% CI for difference ⫺0·31 to 31·1) and headache (⫺2·3 to 32·0) 1657
ARTICLES
Hazard ratio (95% CI)
p
1·01 (0·89–1·15) 1·24 (0·49–3·15) 4·32 (0·57–32·57) 3·42 (0·21–55·37) 3·73 (1·46–9·55) 0·82 (0·29–2·28) 1·13 (0·42–3·04) 0·58 (0·18–1·91) 0·97 (0·76–1·24) 0·74 (0·50–1·10) 1·07 (0·96–1·18)
0·87 0·65 0·16 0·39 0·006 0·70 0·81 0·36 0·82 0·14 0·23
1·02 (1·01–1·03)
0·006
0·91 (0·81–1·02) 0·85 (0·77–0·95) 0·97 (0·65–1·46) 1·02 (1·00–1·04) 0·98 (0·94–1·02) 6·16 (1·64–23·17) 1·41 (0·45–4·38)
0·09 0·003 0·89 0·04 0·30 0·007 0·55
For all continuous variables hazard is change per unit increase in variable. ICA=internalcarotid artery; MCA=middle-cerebral artery.
Table 2: Univariate hazard ratios and 95% CIs
1·0 Proportion of patients CNS-event free
Age at sleep study (years) Men vs women SS vs SC disease S-thalassaemia vs SC Previous transient ischaemic attack Previous headaches Dactylitis less than 1 year Adenotonsillectomy Baseline haemoglobin (g/dL) Mean haemoglobin between 11 and 25 months (g/dL) Mean white cell count between 11 and 25 months, (109/L) Higher right or left ICA or MCA velocity at time of sleep study (cm/s) Baseline haemoglobin F Mean oxygen (%) Proportion sleep study <80% Proportion sleep study <90% Minimum SaO2 (%) Low baseline SaO2 vs normal Dips vs normal
Mean SaO2 ⭓96%
0·9 0·8 0·7 Mean SaO2 <96%
0·6 0·5 0·4 0·3 0·2 0·1 0 0
1
2 3 4 5 Time after sleep study (years)
6
7
Kaplan-Meier curves of difference in time to CNS event or death for patients with mean overnight SaO2 more and less than 96% n=50 >96% SaO2; n=45 <96% SaO2.
were not significantly more common in the pulse oximetry group than in the non-oximetry group (table 1), but adenotonsillectomy was (5·5–37·7), and this group contained all of the CNS events. Mean overnight SaO2 correlated positively with current haemoglobin (correlation coefficient 0·4, p<0·001). Mean SaO2 in SS (median value 95·5) was lower than in patients with SC and S-thalassaemia (97·9 and 97·8, p=0·002). Previous transient ischaemic attack was associated with lower mean SaO2 (p=0·041). Previous transient ischaemic attack, high velocity at the time of the sleep study, low mean SaO2, and a high proportion of the sleep study with SaO2 less than 90% were all significantly associated with a greater risk of future CNS event (table 2). Dips in SaO2 and adenotonsillectomy had no effect on risk of CNS events. However, CI were wide, and clinically meaningful associations cannot be excluded. Baseline haemoglobin F was of borderline significance. After accounting for mean SaO2, the other pulse oximetry measurements did not improve prediction of future CNS events. Mean SaO2 and velocity were independently associated with time to CNS event. After adjustment for these measurements, a history of transient-ischaemic attack became non-significant (p=0·07). High haemoglobin concentration was associated with future CNS events after mean SaO2 and velocity were taken into account (hazard ratio 1·33 [95% CI 0·99–1·79]; p=0·06). After accounting for haemoglobinopathy, the effect of haemoglobin increased in significance. Table 3 shows the fitted values and CI for a model containing haemoglobinopathy, mean SaO2, velocity, and haemoglobin. To identify a cut-off point for treatment, dichotomies for mean SaO2 at 90%, 92%, 94%, and 96% were investigated. The hazard ratio for patients with mean SaO2 less than 96% was greatest (5·6 [1·8–16·9]; p=0·0026, figure). For 94%, 92%, and 90%, the hazard
SS vs SC S-thalassaemia vs SC Baseline haemoglobin (g/dL) Higher (right or left) ICA or MCA velocity (cm/s) Mean oxygen (%)
Hazard ratio (95% CI)
p
9·77 (0·92–103·36) 10·13 (0·53–193·68) 1·7 (1·18–2·43) 1·02 (1·004–1·03) 0·82 (0·71–0·93)
0·06 0·12 0·004 0·01 0·003
ICA=internal-carotid artery; MCA=middle-cerebral artery.
Table 3: Adjusted hazard ratios and 95% CIs (n=91)
1658
ratios were 4·12, 4·65, and 2·48, respectively. Four children had internal-carotid and middle-cerebral artery velocity higher than 200 cm/s recorded during the study: two had strokes and two had no CNS events.
Discussion Our data suggest that nocturnal hypoxaemia predicts CNS events in patients with sickle-cell disease. The association between sleep-disordered breathing and stroke or transient ischaemic attack could involve either the generation of cerebrovascular disease or reduction of the threshold for infarction in the at-risk territory of a stenosed or occluded artery, or both. Hypoxia promotes polymerisation of sickle haemoglobin and the adhesion of red cells to endothelium via binding molecules such as VCAM-1,21 and can also increase platelet activation and endothelial adhesion.22,23 Erythropoietin production increases in hypoxia and could promote the release of reticulocytes, leucocytes, and platelets, which are more adherent to endothelium than older cells.24 Hypertension is an independent risk factor for stroke in sickle-cell disease3 and is more common in sleep-disordered breathing.22 There is evidence from a rat middle-cerebral artery occlusion model that ischaemic necrosis is exacerbated by hypoxia,25 and the reserve capacity of the cerebral circulation to vasodilate could be reduced under hypercapnic conditions, increasing the risk of infarction distal to a narrowed vessel. Seizures seem to be exacerbated by obstructive sleep apnoea;26 cognitive difficulties and cerebral oedema might also be a consequence of hypoxia.27,28 Chronic underventilation could allow continuing vascular and neuronal damage. In documenting sleep disordered breathing, many workers recommend inpatient monitoring, but this is expensive and the patient might not sleep normally. We therefore chose to monitor oxygen saturation in the home setting.29 People who participated in sleep studies generally had more symptoms, probably indicating the motivation of families and physicians to pursue this relatively time-consuming test. This happening could limit the generalisability of our results, but does not weaken the case for doing sleep studies in symptomatic patients. There is a good case for the further investigation of the role of hypoxaemia, however, as an abnormal sleep study is not necessarily predicted by clinical symptoms of
THE LANCET • Vol 357 • May 26, 2001
ARTICLES
obstructive sleep apnoea, and dips were not predictive of CNS events. Primary prevention of stroke in sickle-cell disease has been based on the early detection of cerebrovascular disease and prophylaxis with blood transfusion for patients at high risk. Screening for, and aggressive management of, nocturnal hypoxaemia might be a cost-effective alternative, with lower long-term morbidity. Ideally, prophylaxis should be associated with low morbidity and should have a high probability of preventing disease. Our data suggest that those with a mean SaO2 of less than 96% are at increased risk of CNS events, but rebound crises could occur after withdrawal of oxygen supplementation in sickle-cell disease.30 A controlled trial, funded by the Stroke Association, of primary prevention of CNS events in sickle-cell disease is underway in which the treatment group will receive overnight oxygen supplementation for nocturnal hypoxaemia. Contributors All investigators took part in the design, execution, and analysis of the study, and in writing the report. F J Kirkham and J P M Evans started and supervised the study and did the long-term follow-up. D K M Hewes, M Prengler, and R Lane obtained and interpreted the data. A Wade did the statistical analysis.
10
11
12
13 14
15
16
17
18
19
Acknowledgments We thank our colleagues, Parvis Habibi and Aiden Lafferty, for help with setting up overnight pulse oximetry, David Albert for helpful discussion, Darren Hargrave, and Anne O’Reilly for data entry, and Brian Neville for critical review of the manuscript. The study was supported by Action Research, FJK was supported by the Wellcome Trust, and MP by the Leo Baeck (London) B’nai B’rith and the Ian Karten Charitable Trust. The work was done by Great Ormond Street Hospital for Children NHS Trust, which received a proportion of its funding from the NHS Executive.
References 1 2
3
4
5
6 7
8
9
Earley CJ, Kittner SJ, Feeser BR, et al. Stroke in children and sicklecell disease. Neurology 1998; 51: 169–76. Balkaran B, Char G, Morris JS, Thomas PW, Serjeant BE, Serjeant GR. Stroke in a cohort of patients with homozygous sickle cell disease. J Pediatr 1992; 120: 360–66. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular accidents in sickle cell disease: rates and risk factors. Blood 1998; 91: 288–94. Rothman SM, Fulling KH, Nelson JS. Sickle cell anemia and central nervous system infarction: a neuropathological study. Ann Neurol 1986; 20: 684–90. Wang WC, Langston JW, Steen RG, et al. Abnormalities of the central nervous system in very young children with sickle cell anemia. J Pediatr 1998; 132: 994–98. Liu JE, Gzesh DJ, Ballas SK. The spectrum of epilepsy in sickle cell anemia. J Neurol Sci 1994; 123: 6–10. Kinney TR, Sleeper LA, Wang WC, et al. Silent cerebral infarcts in sickle cell anemia: a risk factor analysis—the Cooperative Study of Sickle Cell Disease. Pediatrics 1999; 103: 640–45. Armstrong FD, Thompson RJ Jr, Wang W, et al. Cognitive functioning and brain magnetic resonance imaging in children with sickle cell disease. Pediatrics 1996; 97: 864–70. Powars DR, Weiss JN, Chan LS, Schroeder WA. Is there a threshold
THE LANCET • Vol 357 • May 26, 2001
20 21
22
23
24
25 26
27
28 29
30
level of fetal hemoglobin that ameliorates morbidity in sickle cell anemia? Blood 1984; 63: 921–26. Miller ST, Sleeper LA, Pegelow CH, et al. Prediction of adverse outcomes in children with sickle cell disease. N Engl J Med 2000; 342: 83–89. Adams RJ, McKie VC, Carl EM, et al. Long-term stroke risk in children with sickle cell disease screened with transcranial Doppler. Ann Neurol 1997; 42: 699–704. Adams RJ, McKie VC, Hsu L, et al. Prevention of a first stroke by transfusions in children with sickle cell anemia and abnormal results on transcranial Doppler ultrasonography. N Engl J Med 1998; 339: 5–11. Partinen M, Palomaki H. Snoring and cerebral infarction. Lancet 1985; 2: 1325–26. Samuels MP, Stebbens VA, Davies SC, Picton-Jones E, Southall DP. Sleep related upper airway obstruction and hypoxaemia in sickle cell disease. Arch Dis Child 1992; 67: 925–29. Robertson PL, Aldrich MS, Hanash SM, Goldstein GW. Stroke associated with obstructive sleep apnea in a child with sickle cell anemia. Ann Neurol 1988; 23: 614–16. Davies SC, Stebbens VA, Samuels MP, Southall DP. Upper airways obstruction and cerebrovascular accident in children with sickle cell anaemia. Lancet 1989; 2: 283–84. Gillard JH, Kirkham FJ, Levin SD, Neville BG, Gosling RG. Anatomical validation of middle cerebral artery position as identified by transcranial pulsed Doppler ultrasound. J Neurol Neurosurg Psychiatry 1986; 49: 1025–29. Gadian DG, Calamante F, Kirkham FJ, et al. Diffusion and perfusion magnetic resonance imaging in childhood stroke. J Child Neurol 2000; 15: 279–83. Kirkham FJ, Calamante F, Bynevelt M, et al. Perfusion magnetic resonance abnormalities in patients with sickle cell disease. Ann Neurol 2001; 49: 477–85. Collett D. Modelling survival data in medical research. London: Chapman and Hall, 1994. Setty BN, Stuart MJ. Vascular adhesion molecule-1 is involved in mediating hypoxia-induced sickle red blood cell adherence to endothelium: potential role in sickle cell disease. Blood 1996; 88: 2311–20. Eisensehr I, Ehrenberg BL, Noachtar S, et al. Platelet activation, epinephrine, and blood pressure in obstructive sleep apnea syndrome. Neurology 1998; 51: 188–95. Inwald DP, Kirkham FJ, Peters MJ, et al. Platelet and leukocyte activation in childhood sickle cell disease: association with nocturnal hypoxaemia and disease severity. Br J Haematol 2000; 111: 474–81. Dale GL, Alberio L. Is there a correlation between raised erythropoietin and thrombotic events in sickle-cell anaemia? Lancet 1998; 352: 566–67. Miyamoto O, Auer RN. Hypoxia, hyperoxia, ischemia and brain necrosis. Neurology 2000; 54: 362–71. Britton TC, O’Donoghue M, Duncan JS, Hirsch NP. Exacerbation of epilepsy by obstructive sleep apnoea. J Neurol Neurosurg Psychiatry 1997; 63: 808. Steen RG, Xiong X, Mulhern RK, Langston JW, Wang WC. Subtle brain abnormalities in children with sickle cell disease: relationship to blood hematocrit. Ann Neurol 1999; 45: 279–86. Appenzeller O. Altitude and the nervous system. Arch Neurol 1998; 55: 1007–09. Rackoff WR, Kunkel N, Silber JH, Asakura T, Ohene-Fempong K. Pulse oximetry and factors associated with hemoglobin oxygen desaturation in children with sickle cell disease. Blood 1993; 81: 3422–27. Embury SH, Garcia JF, Mohandas N, Pennathur-Das R, Clark MR. Effects of oxygen inhalation on endogenous erythropoetin kinetics, erythropoesis and properties of blood cells in sickle-cell anemia. N Engl J Med 1984; 311: 291–95.
1659
Nocturnal hypoxaemia and central-nervous-system events in sickle-cell disease F J Kirkham, D K M Hewes, M Prengler, A Wade, R Lane, J P M Evans
Summary Background Central-nervous-system (CNS) events, including strokes, transient ischaemic attacks, and seizures are common in sickle-cell disease. Stroke can be predicted by high velocities in the internal-carotid or middle-cerebral arteries on transcranial doppler ultrasonography. We tested the hypothesis that nocturnal hypoxaemia can predict CNS events better than clinical or haematological features, or transcranial doppler sonography. Methods We screened 95 hospital-based patients with sickle-cell disease (median age 7·7 years [range 1·0–23·1]), but without previous stroke, with transcranial doppler and overnight pulse oximetry. Follow-up continued for a median of 6·01 (0·11–8·54) years. Findings 19 patients had CNS events (six ischaemic and one haemorrhagic stroke, eight transient ischaemic attacks, and four seizures). Mean overnight oxygen saturation ([SaO2] hazard ratio 0·82 per 1% increase [95% CI 0·71–0·93]; p=0·003) and higher internal-carotid or middle-cerebral artery velocity (1·02 for every increase of 1 cm/s [1·004–1·03]; p=0·009) were independently associated with time to CNS event. After accounting for mean SaO2, artery velocity, and haemoglobinopathy, high haemoglobin concentration was also associated with an increased risk of CNS event (1·7 per g/dL, [1·18–2·43]; p=0·004). Dips suggestive of obstructive sleep apnoea did not predict CNS events, and adenotonsillectomy seemed to have no effect, although the CI were wide and clinically important effects cannot be excluded. Interpretation Screening for, and appropriate management of, nocturnal hypoxaemia might be a safe and effective alternative to prophylactic blood transfusion for primary prevention of CNS events in sickle-cell disease. Lancet 2001; 357: 1656–59
Neurosciences Unit (F J Kirkham MB, D K M Hewes MB, M Prengler MD), Paediatric Epidemiology and Biostatistics Unit (A Wade PhD), Portex Unit of Anaesthesia, Intensive Care and Respiratory Medicine (R Lane PhD), and Department of Haematology (J P M Evans DM), Institute of Child Health, University College London, London, UK; and Great Ormond Street Hospital, London, UK (F J Kirkham, D K M Hewes, M Prengler, A Wade, R Lane, J P M Evans) Correspondence to: Dr F J Kirkham, the Wolfson Centre, Mecklenburgh Square, London WC1N 2AP, UK (e-mail: [email protected])
1656
Introduction Sickle-cell disease is a group of inherited haemoglobin disorders, including sickle-cell anaemia (homozygous SS), SC disease, and S-thalassaemia. Sickle-cell disease is the most common cause of childhood stroke, with an incidence of 0·28–0·61%, similar to normal elderly adults;1–3 25% and 10% of patients with SS and SC, respectively, have had a stroke by age 45 years.2 Stenosis or occlusion of the basal cerebral vessels, especially distal-internal-carotid or proximal-middle-cerebral arteries, is characteristic.4 Cerebrovascular disease is seen in young children5 with ischaemic stroke, and cerebral haemorrhage arises mainly in adults. Seizures are more common in patients with sickle-cell disease than in the rest of the population,6 can be difficult to distinguish from transient ischaemic attacks, and are associated with imaging abnormalities.7 Cerebral infarction (overt and covert) is associated with cognitive difficulties.8 Risk factors for ischaemic stroke include low fetal haemoglobin,9 recent chest crisis,3 and hypertension.3 Infants with dactylitis, haemoglobin values less than 7 g/dL, or leucocytosis are at risk of adverse outcomes, including stroke,2,10 and might be candidates for prophylactic therapy—for example, hydroxyurea to increase fetal haemoglobin concentrations. Transcranial doppler sonography predicts stroke risks of 40%, 7%, and 2% over the following 40 months in patients with velocities of more than 200 or more than 170, or less than 170 cm/s, respectively, in the internal-carotid or middle-cerebral arteries.11 Primary stroke prevention with long-term blood transfusion is possible in children with velocities greater than 200 cm/s.12 The risks of infection, alloimmunisation, and iron overload, needing parenteral chelating agents, make this approach punishing and potentially life threatening, however, which justifies the investigation of alternative preventative strategies. Snoring is a risk factor for stroke in adults.13 Both episodic and continuous nocturnal hypoxaemia are common in sickle-cell disease,14 possibly because of upper-airway obstruction secondary to adenotonsillar hypertrophy. The reported association between obstructive sleep apnoea and central-nervous-system (CNS) events in sickle-cell disease15,16 has not been studied prospectively. We tested the hypothesis that nocturnal hypoxaemia, documented by overnight pulse oximetry, offers better prediction of CNS events than clinical, haematological, or transcranial doppler sonography variables.
Methods Patients Permission for the study was granted by the Great Ormond Street Hospital Local Research Ethics committee, and written informed consent was received from all patients. From Jan 1, 1991, we invited all patients who had not had a stroke and were regularly attending the haemoglobinopathy clinic of Queen Elizabeth Hospital, London, UK, to participate. Follow-up was discontinued on April 30, 2000.
THE LANCET • Vol 357 • May 26, 2001
ARTICLES
Study protocol We used a TC-64B (Eden Medical Electronics, Uberlingen, Germany) or a Companion (Nicolet, Warwick, UK) transcranial doppler sonography machine to insonate the distal internal carotid, middle, anterior, posterior-cerebral, and basilar arteries.17 We recorded the highest average velocity measured on both sides of the internal-carotid and middle-cerebral arteries. We assessed patients at recruitment, at the time of the sleep study, and at follow-up in 1998. Individuals with high velocities were assessed more frequently than other patients. We used a pulse oximeter (Biox 3700, Datex-Ohmeda, Hatfield, Hertfordshire, UK) to record oxygen saturation (SaO2) during sleep. DKMH or RL, who were unaware of the clinical data, analysed the results. We measured the SaO2 of 63 patients at home. In these patients, parents turned on the oximeter after sleep onset. The rest of the study population visited the hospital sleep laboratory. Studies lasted 4·3–8·2 h (median 7·5). We excluded movement artefacts manually. We recorded the mean and minimum oxygen saturations and the proportion of sleep spent at saturations less than 90% and less than 80%, and examined the trace for dips (>4% from baseline) in oxygenation associated with acute pulse rate rises—a pattern suggestive of obstructive sleep apnoea. Patients with symptoms of obstructive sleep apnoea were referred for surgical management without reference to overnight pulse oximetry data. We obtained retrospective data on dactylitis in the first year of life, mean early haemoglobin concentrations and early white blood cell counts from year 2, baseline haemoglobin on the day of doppler sonography, and haemoglobin F from 1993 (except in those with SC disease). We recorded data on previous headache or transient ischaemic attack and new CNS events (seizures or focal deficits lasting about 24 h coded as stroke or transient ischaemic attack, respectively) at monthly joint haematology and neurology clinics. Magnetic resonance imaging and magnetic resonance angiography were done as soon as possible.18,19 We reviewed patients in 1998. Statistical analysis We compared patients who did and did not undergo overnight pulse oximetry. Exact tests and CI were calculated for categorical variables with StatXact (version 4·0·1). Mann-Whitney U tests were used to compare the medians of continuous variables. Associations between overnight pulse oximetry measurements and other variables were investigated with Spearman’s correlation coefficients, 2, Mann-Whitney U, and Kruskal-Wallis tests as appropriate. We used Cox’s proportional hazard regression models to identify predictors of subsequent CNS events, taking into account the varying length of follow-up. Time to CNS event from overnight pulse oximetry was investigated. The transcranial doppler sonography measurement made closest to the overnight data was used in the model. Adenotonsillectomy, which could have been done some time after the overnight pulse oximetry and transcranial doppler sonography studies, was entered as a time-varying covariate.20 Hazard ratios are presented with 95% CI. A Kaplan-Meier survival curve illustrates the relation between sleep study data and subsequent risk of stroke.
Results We enrolled 147 patients. Overnight pulse oximetry was done in 65% of these (table 1) at a median time of 218 days after recruitment. Before 1994, most had overnight
THE LANCET • Vol 357 • May 26, 2001
Overnight pulse oximetry
Sex (M/F) Type of sickle-cell disease Homozygous SC disease S-thalassaemia History Previous transient ischaemic attack Previous headaches Dactylitis before age 1 year Adenotonsillectomy Age at recruitment (years)
p
Yes (n=95)
No (n=52)
54 (57%)/41 (43%)
32 (62%)/20 (38%)
0·60 0·44
74 (78%) 14 (15%) 7 (7%)
37 (71%) 12 (23%) 3 (6%)
19 (20%)
3 (6%)
32 (34%) 23 (29%)* 27 (28%) 6·65 (4·71–9·50)
0·09
10 (19%) 6 (16%)† 4 (8%) 6·52 (3·68–10·44)
Haematological characteristics Haemoglobin concentration 8·1 (7·5–10·2)‡ 8·4 (7·7–9·9)§ at recruitment (g/dL) Mean haemoglobin between 8·9 (8·3–10·0)|| 8·7 (7·8–9·9)¶ 11 and 25 months (g/dL) Mean white cell count between 11·8 (8·3–15·0)** 10·8 (9·0–14·9)‡‡ 11 and 25 months (109/L) Higher (right or left) ICA or MCA 120·0 (100·0–139·0)§§ 109·0 (98·5–130·5) velocity at recruitment (cm/s)
0·10 0·18 0·02 0·35 0·57 0·56 0·67 0·28
Numbers and percentages are given for categorical variables, medians and IQR for continuous variables. ICA=internal-carotid artery; MCA=middle-cerebral artery. *n=79; †n=38; ‡n=93; §n=46; ||n=78; ¶n=30; **n=74; ‡‡n=29; §§n=92.
Table 1: Baseline characteristics of patients
pulse oximetry at home; thereafter patients were offered hospital sleep studies but often found it difficult to attend. 19 patients (17 with sickle-cell anaemia) had CNS events during follow-up (six ischaemic and one haemorrhagic stroke, eight transient ischaemic events, and four seizures). One patient with S-thalassaemia had a stroke. A child with SC disease and hydrocephalus died of cerebral oedema after developing severe headaches (normal shunt pressure), three vertebrobasilar transient ischaemic attacks, and a seizure. Two patients died from non-CNS causes (chest syndrome and sepsis) and were censored in the statistical analysis. Four patients with ischaemic stroke had only deep-white-matter lesions and one had a right parieto-occiptal cortical lesion; all had large vessel disease on magnetic resonance angiography. One patient developed bilateral frontal and parietal watershed lesions after seizures associated with facial infection, and another had a severe headache and mild left hemiparesis with three acute left-sided haemorrhagic lesions and developed right-sided deep-white-matter lesions 4 days later.19 These two children had normal magnetic resonance angiography and conventional arteriography. Five (including the SC patient) had vertebrobasilar transient ischaemic attacks with dizziness, diplopia, visual symptoms, ataxia, paraesthesiae, and sudden collapse. None of these five patients had infarction on magnetic resonance imaging, and only one had an abnormal magnetic resonance angiogram with bilateral turbulence in the terminal internal-carotid artery. Three individuals had transient hemiparesis consistent with anterior-circulation transient ischaemic attack, of whom two had small deep-white-matter lesions and bilateral middle-cerebral or internal-carotid artery turbulence on magnetic resonance angiography, and one had a small hemisphere contralaterally with normal magnetic resonance angiography. Four children had seizures: three had normal magnetic resonance imaging scans (one with normal magnetic resonance angiography and two with bilateral turbulence), and one had bilateral deep-whitematter lesions and magnetic resonance imaging evidence of left-sided internal-carotid-artery dissection. Previous transient ischaemic attack (95% CI for difference ⫺0·31 to 31·1) and headache (⫺2·3 to 32·0) 1657
ARTICLES
Hazard ratio (95% CI)
p
1·01 (0·89–1·15) 1·24 (0·49–3·15) 4·32 (0·57–32·57) 3·42 (0·21–55·37) 3·73 (1·46–9·55) 0·82 (0·29–2·28) 1·13 (0·42–3·04) 0·58 (0·18–1·91) 0·97 (0·76–1·24) 0·74 (0·50–1·10) 1·07 (0·96–1·18)
0·87 0·65 0·16 0·39 0·006 0·70 0·81 0·36 0·82 0·14 0·23
1·02 (1·01–1·03)
0·006
0·91 (0·81–1·02) 0·85 (0·77–0·95) 0·97 (0·65–1·46) 1·02 (1·00–1·04) 0·98 (0·94–1·02) 6·16 (1·64–23·17) 1·41 (0·45–4·38)
0·09 0·003 0·89 0·04 0·30 0·007 0·55
For all continuous variables hazard is change per unit increase in variable. ICA=internalcarotid artery; MCA=middle-cerebral artery.
Table 2: Univariate hazard ratios and 95% CIs
1·0 Proportion of patients CNS-event free
Age at sleep study (years) Men vs women SS vs SC disease S-thalassaemia vs SC Previous transient ischaemic attack Previous headaches Dactylitis less than 1 year Adenotonsillectomy Baseline haemoglobin (g/dL) Mean haemoglobin between 11 and 25 months (g/dL) Mean white cell count between 11 and 25 months, (109/L) Higher right or left ICA or MCA velocity at time of sleep study (cm/s) Baseline haemoglobin F Mean oxygen (%) Proportion sleep study <80% Proportion sleep study <90% Minimum SaO2 (%) Low baseline SaO2 vs normal Dips vs normal
Mean SaO2 ⭓96%
0·9 0·8 0·7 Mean SaO2 <96%
0·6 0·5 0·4 0·3 0·2 0·1 0 0
1
2 3 4 5 Time after sleep study (years)
6
7
Kaplan-Meier curves of difference in time to CNS event or death for patients with mean overnight SaO2 more and less than 96% n=50 >96% SaO2; n=45 <96% SaO2.
were not significantly more common in the pulse oximetry group than in the non-oximetry group (table 1), but adenotonsillectomy was (5·5–37·7), and this group contained all of the CNS events. Mean overnight SaO2 correlated positively with current haemoglobin (correlation coefficient 0·4, p<0·001). Mean SaO2 in SS (median value 95·5) was lower than in patients with SC and S-thalassaemia (97·9 and 97·8, p=0·002). Previous transient ischaemic attack was associated with lower mean SaO2 (p=0·041). Previous transient ischaemic attack, high velocity at the time of the sleep study, low mean SaO2, and a high proportion of the sleep study with SaO2 less than 90% were all significantly associated with a greater risk of future CNS event (table 2). Dips in SaO2 and adenotonsillectomy had no effect on risk of CNS events. However, CI were wide, and clinically meaningful associations cannot be excluded. Baseline haemoglobin F was of borderline significance. After accounting for mean SaO2, the other pulse oximetry measurements did not improve prediction of future CNS events. Mean SaO2 and velocity were independently associated with time to CNS event. After adjustment for these measurements, a history of transient-ischaemic attack became non-significant (p=0·07). High haemoglobin concentration was associated with future CNS events after mean SaO2 and velocity were taken into account (hazard ratio 1·33 [95% CI 0·99–1·79]; p=0·06). After accounting for haemoglobinopathy, the effect of haemoglobin increased in significance. Table 3 shows the fitted values and CI for a model containing haemoglobinopathy, mean SaO2, velocity, and haemoglobin. To identify a cut-off point for treatment, dichotomies for mean SaO2 at 90%, 92%, 94%, and 96% were investigated. The hazard ratio for patients with mean SaO2 less than 96% was greatest (5·6 [1·8–16·9]; p=0·0026, figure). For 94%, 92%, and 90%, the hazard
SS vs SC S-thalassaemia vs SC Baseline haemoglobin (g/dL) Higher (right or left) ICA or MCA velocity (cm/s) Mean oxygen (%)
Hazard ratio (95% CI)
p
9·77 (0·92–103·36) 10·13 (0·53–193·68) 1·7 (1·18–2·43) 1·02 (1·004–1·03) 0·82 (0·71–0·93)
0·06 0·12 0·004 0·01 0·003
ICA=internal-carotid artery; MCA=middle-cerebral artery.
Table 3: Adjusted hazard ratios and 95% CIs (n=91)
1658
ratios were 4·12, 4·65, and 2·48, respectively. Four children had internal-carotid and middle-cerebral artery velocity higher than 200 cm/s recorded during the study: two had strokes and two had no CNS events.
Discussion Our data suggest that nocturnal hypoxaemia predicts CNS events in patients with sickle-cell disease. The association between sleep-disordered breathing and stroke or transient ischaemic attack could involve either the generation of cerebrovascular disease or reduction of the threshold for infarction in the at-risk territory of a stenosed or occluded artery, or both. Hypoxia promotes polymerisation of sickle haemoglobin and the adhesion of red cells to endothelium via binding molecules such as VCAM-1,21 and can also increase platelet activation and endothelial adhesion.22,23 Erythropoietin production increases in hypoxia and could promote the release of reticulocytes, leucocytes, and platelets, which are more adherent to endothelium than older cells.24 Hypertension is an independent risk factor for stroke in sickle-cell disease3 and is more common in sleep-disordered breathing.22 There is evidence from a rat middle-cerebral artery occlusion model that ischaemic necrosis is exacerbated by hypoxia,25 and the reserve capacity of the cerebral circulation to vasodilate could be reduced under hypercapnic conditions, increasing the risk of infarction distal to a narrowed vessel. Seizures seem to be exacerbated by obstructive sleep apnoea;26 cognitive difficulties and cerebral oedema might also be a consequence of hypoxia.27,28 Chronic underventilation could allow continuing vascular and neuronal damage. In documenting sleep disordered breathing, many workers recommend inpatient monitoring, but this is expensive and the patient might not sleep normally. We therefore chose to monitor oxygen saturation in the home setting.29 People who participated in sleep studies generally had more symptoms, probably indicating the motivation of families and physicians to pursue this relatively time-consuming test. This happening could limit the generalisability of our results, but does not weaken the case for doing sleep studies in symptomatic patients. There is a good case for the further investigation of the role of hypoxaemia, however, as an abnormal sleep study is not necessarily predicted by clinical symptoms of
THE LANCET • Vol 357 • May 26, 2001
ARTICLES
obstructive sleep apnoea, and dips were not predictive of CNS events. Primary prevention of stroke in sickle-cell disease has been based on the early detection of cerebrovascular disease and prophylaxis with blood transfusion for patients at high risk. Screening for, and aggressive management of, nocturnal hypoxaemia might be a cost-effective alternative, with lower long-term morbidity. Ideally, prophylaxis should be associated with low morbidity and should have a high probability of preventing disease. Our data suggest that those with a mean SaO2 of less than 96% are at increased risk of CNS events, but rebound crises could occur after withdrawal of oxygen supplementation in sickle-cell disease.30 A controlled trial, funded by the Stroke Association, of primary prevention of CNS events in sickle-cell disease is underway in which the treatment group will receive overnight oxygen supplementation for nocturnal hypoxaemia. Contributors All investigators took part in the design, execution, and analysis of the study, and in writing the report. F J Kirkham and J P M Evans started and supervised the study and did the long-term follow-up. D K M Hewes, M Prengler, and R Lane obtained and interpreted the data. A Wade did the statistical analysis.
10
11
12
13 14
15
16
17
18
19
Acknowledgments We thank our colleagues, Parvis Habibi and Aiden Lafferty, for help with setting up overnight pulse oximetry, David Albert for helpful discussion, Darren Hargrave, and Anne O’Reilly for data entry, and Brian Neville for critical review of the manuscript. The study was supported by Action Research, FJK was supported by the Wellcome Trust, and MP by the Leo Baeck (London) B’nai B’rith and the Ian Karten Charitable Trust. The work was done by Great Ormond Street Hospital for Children NHS Trust, which received a proportion of its funding from the NHS Executive.
References 1 2
3
4
5
6 7
8
9
Earley CJ, Kittner SJ, Feeser BR, et al. Stroke in children and sicklecell disease. Neurology 1998; 51: 169–76. Balkaran B, Char G, Morris JS, Thomas PW, Serjeant BE, Serjeant GR. Stroke in a cohort of patients with homozygous sickle cell disease. J Pediatr 1992; 120: 360–66. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular accidents in sickle cell disease: rates and risk factors. Blood 1998; 91: 288–94. Rothman SM, Fulling KH, Nelson JS. Sickle cell anemia and central nervous system infarction: a neuropathological study. Ann Neurol 1986; 20: 684–90. Wang WC, Langston JW, Steen RG, et al. Abnormalities of the central nervous system in very young children with sickle cell anemia. J Pediatr 1998; 132: 994–98. Liu JE, Gzesh DJ, Ballas SK. The spectrum of epilepsy in sickle cell anemia. J Neurol Sci 1994; 123: 6–10. Kinney TR, Sleeper LA, Wang WC, et al. Silent cerebral infarcts in sickle cell anemia: a risk factor analysis—the Cooperative Study of Sickle Cell Disease. Pediatrics 1999; 103: 640–45. Armstrong FD, Thompson RJ Jr, Wang W, et al. Cognitive functioning and brain magnetic resonance imaging in children with sickle cell disease. Pediatrics 1996; 97: 864–70. Powars DR, Weiss JN, Chan LS, Schroeder WA. Is there a threshold
THE LANCET • Vol 357 • May 26, 2001
20 21
22
23
24
25 26
27
28 29
30
level of fetal hemoglobin that ameliorates morbidity in sickle cell anemia? Blood 1984; 63: 921–26. Miller ST, Sleeper LA, Pegelow CH, et al. Prediction of adverse outcomes in children with sickle cell disease. N Engl J Med 2000; 342: 83–89. Adams RJ, McKie VC, Carl EM, et al. Long-term stroke risk in children with sickle cell disease screened with transcranial Doppler. Ann Neurol 1997; 42: 699–704. Adams RJ, McKie VC, Hsu L, et al. Prevention of a first stroke by transfusions in children with sickle cell anemia and abnormal results on transcranial Doppler ultrasonography. N Engl J Med 1998; 339: 5–11. Partinen M, Palomaki H. Snoring and cerebral infarction. Lancet 1985; 2: 1325–26. Samuels MP, Stebbens VA, Davies SC, Picton-Jones E, Southall DP. Sleep related upper airway obstruction and hypoxaemia in sickle cell disease. Arch Dis Child 1992; 67: 925–29. Robertson PL, Aldrich MS, Hanash SM, Goldstein GW. Stroke associated with obstructive sleep apnea in a child with sickle cell anemia. Ann Neurol 1988; 23: 614–16. Davies SC, Stebbens VA, Samuels MP, Southall DP. Upper airways obstruction and cerebrovascular accident in children with sickle cell anaemia. Lancet 1989; 2: 283–84. Gillard JH, Kirkham FJ, Levin SD, Neville BG, Gosling RG. Anatomical validation of middle cerebral artery position as identified by transcranial pulsed Doppler ultrasound. J Neurol Neurosurg Psychiatry 1986; 49: 1025–29. Gadian DG, Calamante F, Kirkham FJ, et al. Diffusion and perfusion magnetic resonance imaging in childhood stroke. J Child Neurol 2000; 15: 279–83. Kirkham FJ, Calamante F, Bynevelt M, et al. Perfusion magnetic resonance abnormalities in patients with sickle cell disease. Ann Neurol 2001; 49: 477–85. Collett D. Modelling survival data in medical research. London: Chapman and Hall, 1994. Setty BN, Stuart MJ. Vascular adhesion molecule-1 is involved in mediating hypoxia-induced sickle red blood cell adherence to endothelium: potential role in sickle cell disease. Blood 1996; 88: 2311–20. Eisensehr I, Ehrenberg BL, Noachtar S, et al. Platelet activation, epinephrine, and blood pressure in obstructive sleep apnea syndrome. Neurology 1998; 51: 188–95. Inwald DP, Kirkham FJ, Peters MJ, et al. Platelet and leukocyte activation in childhood sickle cell disease: association with nocturnal hypoxaemia and disease severity. Br J Haematol 2000; 111: 474–81. Dale GL, Alberio L. Is there a correlation between raised erythropoietin and thrombotic events in sickle-cell anaemia? Lancet 1998; 352: 566–67. Miyamoto O, Auer RN. Hypoxia, hyperoxia, ischemia and brain necrosis. Neurology 2000; 54: 362–71. Britton TC, O’Donoghue M, Duncan JS, Hirsch NP. Exacerbation of epilepsy by obstructive sleep apnoea. J Neurol Neurosurg Psychiatry 1997; 63: 808. Steen RG, Xiong X, Mulhern RK, Langston JW, Wang WC. Subtle brain abnormalities in children with sickle cell disease: relationship to blood hematocrit. Ann Neurol 1999; 45: 279–86. Appenzeller O. Altitude and the nervous system. Arch Neurol 1998; 55: 1007–09. Rackoff WR, Kunkel N, Silber JH, Asakura T, Ohene-Fempong K. Pulse oximetry and factors associated with hemoglobin oxygen desaturation in children with sickle cell disease. Blood 1993; 81: 3422–27. Embury SH, Garcia JF, Mohandas N, Pennathur-Das R, Clark MR. Effects of oxygen inhalation on endogenous erythropoetin kinetics, erythropoesis and properties of blood cells in sickle-cell anemia. N Engl J Med 1984; 311: 291–95.
1659