Article Text

Download PDFPDF

Higher oxygen saturation with hydroxyurea in paediatric sickle cell disease
  1. Lisa van Geyzel1,
  2. Michele Arigliani2,
  3. Baba Inusa3,
  4. Bethany Singh1,
  5. Wanda Kozlowska1,4,
  6. Subarna Chakravorty5,
  7. Cara J Bossley1,
  8. Gary Ruiz1,
  9. David Rees5,
  10. Atul Gupta1
  1. 1 Department of Paediatric Respiratory Medicine, King's College Hospital NHS Foundation Trust, London, UK
  2. 2 Department of Medicine, Division of Pediatrics, University Hospital of Udine, Udine, Italy
  3. 3 Department of Paediatric Haematology, Guy's & St Thomas' NHS Foundation Trust, London, UK
  4. 4 Department of Paediatric Respiratory Medicine, Addenbrooke's Hospital, Cambridge, Cambridgeshire, UK
  5. 5 Department of Paediatric Haematology, King's College Hospital NHS Foundation Trust, London, UK
  1. Correspondence to Dr Atul Gupta, Respiratory Pediatrics, Kings College Hospital, London SE5, UK; atul.gupta{at}


Introduction Sickle cell disease (SCD) is one of the most common inherited diseases worldwide. It is associated with lifelong morbidity and reduced life expectancy. Hydroxyurea (HU) has been shown to reduce the frequency and severity of vaso-occlusive episodes in SCD. Hypoxaemia and intermittent nocturnal oxygen desaturations occur frequently in children with SCD and contribute to the associated morbidity, including risk of cerebrovascular disease.

Objective To evaluate the effect of HU on oxygen saturation (SpO2) overnight and on daytime SpO2 spot checks in children with SCD.

Methods A retrospective review of children with SCD and respiratory problems who attended two UK tertiary sickle respiratory clinics and were treated with HU. Longitudinal data were collected from 2 years prior and up to 3 years after the commencement of HU.

Results Forty-three children, 23 males (53%) with a median age of 9 (range 1.8–18) years were included. In the 21 children who had comparable sleep studies before and after starting HU, mean SpO2 was higher (95.2% from 93.5%, p=0.01) and nadir SpO2 was higher (87.2% from 84.3%, p=0.009) when taking HU. In 32 of the children, spot daytime oxygen saturations were also higher (96.3% from 93.5%, p=0.001).

Conclusion Children with SCD had higher oxygen saturation overnight and on daytime spot checks after starting HU. These data suggest HU may be helpful for treating persistent hypoxaemia in children with SCD pending more evidence from a randomised clinical trial.

  • haematology
  • sickle cell disease
  • hydroxyurea
  • hypoxaemia

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic?

  • Hypoxaemia and intermittent oxygen desaturation frequently occur in children with sickle cell disease (SCD) and contribute to the associated morbidity. Hydroxyurea reduces the incidence of painful crises and acute chest syndrome by increasing fetal haemoglobin percentage in the blood.

What this study adds?

  • Children with SCD had higher overnight oxygen saturation on sleep studies and higher daytime spot oxygen saturation checks after starting hydroxyurea. Hydroxyurea is a potential treatment for persistent hypoxaemia in SCD pending further evidence from randomised controlled trials.


Sickle cell disease (SCD) is one of the most prevalent inherited diseases, affecting approximately 300 000 newborns globally each year.1 Hypoxaemia, a frequent complication of SCD, increases the risk of painful crises,2 cognitive dysfunction,3 central nervous system events4 and acute chest syndrome (ACS).5 Children with SCD also have a high prevalence of obstructive sleep apnoea (OSA) that may contribute to nocturnal hypoxaemia.6–9 Hydroxyurea (HU) increases production of fetal haemoglobin (HbF).10 11 HbF percentage (HbF%) can rise from a mean of 5%–10% to 15%–20% during treatment with HU.12 Increasing HbF% lowers the concentration of HbS within red cells resulting in less polymerisation and increased total haemoglobin. This results in fewer episodes of pain, hospitalisation and ACS.13 The drug has a good safety profile in children with SCD.11 14 Among possible side effects, mild gastrointestinal symptoms,11 hyperpigmentation of the skin and darkening of the nails have been reported,15 beyond the most common excessive myelosuppression, which is transient and reversible.12 Chronic hypoxaemia is one the complications of SCD for which HU therapy is now recommended in the UK.16 However, evidence of a better oxygenation on HU has only been reported anecdotally by a case report,17 a small study using spot pulse oximetry18 and a cross-sectional study comparing children with SCD and suspected OSA who were on HU with a larger group of patients who were not.19 To our knowledge, there have been no longitudinal studies looking at the effect of HU on daytime and overnight oxygen saturations in SCD. As part of routine respiratory assessment in our centres, we have often conducted overnight oximetry/capnography in children with SCD who had respiratory complaints. Many who subsequently started HU therapy for various reasons had repeated sleep studies. We therefore had longitudinal data on which to test the hypothesis that oxygenation would improve in children with SCD on HU.


Our study group were paediatric SCD patients followed at a tertiary respiratory clinics (King’s College Hospital) or joint respiratory sickle cell clinics (Evelina London Children’s Hospital). As per standard practice, these children would have been assessed for respiratory comorbidities (most commonly including wheezing and asthma, sleep disordered breathing (SDB), chronic hypoxaemia, recurrent ACS, chronic cough and lung function abnormalities) and treated accordingly. We retrospectively collected longitudinal data on children with SCD who were seen in these clinics and who were commenced on HU between March 2006 and July 2014 (figure 1). We reviewed electronic and paper medical records for oxygen saturation as measured by pulse oximeter (SpO2) readings recorded routinely in clinic. A Nonin GO2 pulse oximeter (Nonin, Plymouth, Minnesota, USA) was used to measure SpO2, with the value recorded after at least 2 min of stable SpO2 readings and a clear pulsatile photoplethysmographic signal. Median values of these readings taken from 6 months prior to commencement of HU were compared with median values of readings taken up to 2 years after. Over the same time periods before and after starting HU, haemoglobin and fetal haemoglobin concentrations and spirometry data were also collected and averaged for comparison. Those who had oximetry/capnography sleep studies (TCM 4/40 monitoring system, Radiometer, software version 3.0, SpO2 averaging time 3 s) done from 2 years before and up to 3 years after starting HU were identified for comparison of sleep study parameters within the same patient. Sleep studies were ordered in those patients who reported symptoms of SDB (eg, loud snoring, witnessed apnoeas, restless sleep and mouth breathing) or had a previous history of OSA and were performed over a single night, during which parents or guardians kept a sleep diary. Artefacts due to poor perfusion, low signal identification and movement were manually excluded, as well as periods of wakefulness according to the sleep diary’s records. Studies with less than 4 hours of artefact-free data were excluded. Analysis software provided standard measures including overnight mean and nadir SpO2 and oxygen desaturation index (ODI), defined as the number of validated desaturations of at least 3% per hour of sleep.

All the assessments were performed when the patients were at steady state, outside acute SCD-related acute events (eg, vaso-occlusive crisis and ACS).

Statistical analysis

Descriptive statistics are reported as medians with interquartile ranges (IQR). Group comparisons were performed using Wilcoxon matched-pairs signed-rank test, χ2 test or Fisher’s exact test as appropriate. The relationship between night-time or daytime SpO2 and age at each data point was evaluated through Spearman’s rank correlation. A p value<0.05 was considered as statistically significant. Statistical analysis was performed using GraphPad, version 6 for Windows (GraphPad Software, La Jolla, California, USA).


Forty-three children, 23 (53%) male, were included. Three children were excluded due to lack of comparable data. Median age was 9 years (range 1.8–18). One child had HbS/β0 thalassaemia; all the others had HbSS. There were no smokers among the adolescent patients enrolled. Six (14%) children had a history of recurrent ACS (figure 1), and two had severe hypoxaemia. Fifteen of them (35%) had asthma, which had been diagnosed at least 1 year before starting HU. Moreover, 17 children (40%) had previously undergone adenotonsillectomy for OSA, three of them after they started HU.

Figure 1

Indications for starting hydroxyurea in 43 paediatric patients with sickle cell disease. Hb, haemoglobin; low steady state Hb: <70 g/L. *Microalbuminuria, retinopathy and recent severe pneumococcal sepsis.

Both at baseline and after starting HU, there were no statistically significant differences in median values of spot daytime SpO2 and overnight mean SpO2 between patients with or without asthma (data not shown).

The majority started HU due to frequent painful crises or very low steady-state haemoglobin (figure 1). The median dose of HU at 1–2 years after commencement was 22 mg/kg (IQR 20–26, range 15–30).

Overnight oxygenation

Comparable sleep studies before and after starting HU were available on 21 children. Parameters from sleep studies done a median of 9 months (IQR 3.5–15 months) before starting HU were compared with parameters from studies done a median of 9 months (IQR 5–16 months) after. Mean overnight oxygen saturations rose significantly from 93.5% to 95.2% on HU (p=0.01), while nadir overnight oxygen saturations were also significantly higher (84.3%–87.2%) on HU (p=0.009) (table 1). There was no significant difference in the ODI. None of these parameters was related to age at each data point. When removing from the analysis three patients who underwent adenotonsillectomy after starting HU, increases of mean and nadir overnight SpO2 from baseline to endpoint were still significant (p<0.05 for both outcomes; data not shown).

Table 1

Changes in oxygen saturation parameters and haematological indices with hydroxyurea

Among 18 children with SDB and comparable ODI results from nocturnal oximetry before and after starting HU, the frequency of ODI ≥3, a cut-off highly predictive of OSA in children with SDB,20 decreased from 8 (44%) pre-HU to 6 (33%) post-HU (p=0.7). Of these, five participants who had previously undergone adenotonsillectomy for OSA had an ODI persistently ≥3, both before and while taking HU.

Averaged spot daytime oxygen saturation and other measures

Comparable daytime spot checks of SpO2 before and after starting HU were available in 32 of the children. The median SpO2 rose by 2.8% (from 93.5% to 96.3%) after starting HU (table 1). Before the intervention, SpO2 had a moderate negative correlation with age (rs=−0.44, p=0.01), but this relationship lost significance after HU was introduced (rs=−0.18. p=0.2). As expected, Hb and HbF rose significantly on HU. There was a moderate positive correlation between changes in daytime SpO2 and changes in HbF level from baseline to endpoint (rs=0.47, p=0.02), as well as between changes in SpO2 and Hb concentration (rs=0.51, p=0.007). No significant changes in forced expiratory volume in 1 s and forced vital capacity % of predicted were found on HU, but longitudinal lung function data were available only in 10 children (data not shown).

Among SCD patients with asthma and with previous adenotonsillectomy, changes in daytime and nocturnal oxygen with HU therapy were similar to those of the whole sample, but only variations of spot daytime SpO2 resulted significant (online supplementary table S1), probably also for the limited number of observations in these subgroups.


Our principal finding was that overnight mean SpO2 and nadir SpO2 were significantly higher after the commencement of HU in children with SCD. Averaged spot daytime oxygen saturations were also higher in the children when they were on HU.

Hypoxaemia in SCD is reported to occur in 33%–44% of patients5 21 and is associated with increased risk of stroke,22 pain crises,2 increased tricuspid regurgitant jet velocity (TRJV)23 and left ventricular abnormalities.24 A low oxygen saturation in patients with SCD may depend on several causes, including elevated levels of carboxyhaemoglobin and methaemoglobin that are unable to carry oxygen,25 asleep hypoxaemia due to disordered breathing7 and the rightward shift of the oxygen–haemoglobin dissociation curve (ODC) when HbS polymerises.26–28 In addition, chronic anaemia, besides reducing oxygen carrying capacity, also leads to increased 2,3 DPG production and rightward shift of the ODC, resulting in a lower than expected oxyhaemoglobin saturation at a given arterial oxygen tension (PaO2).29

There is limited evidence for the effect of HU on oxygenation in children with SCD. In 2008, Singh et al 17 published a report of three cases of children with SCD, recurrent episodes of ACS and chronic hypoxaemia. Spot daytime oxygen saturations were measured before and after the commencements of HU. Oxygenation improved soon after the commencement of HU, and this effect was sustained up to the last follow-up, 20–24 months later.

Pashankar et al 18 conducted a small prospective study of 13 children with haemolobin SS or Sβ0 thalassemia who were prospectively treated with HU for 12 months for an elevated TRJV, which has been associated with oxygen desaturation. HU significantly increased spot oxygen saturation, and the improvement was sustained at 12 months post-treatment. Compared with these reports, our study included a larger number of patients and integrated data from both nocturnal and daytime SpO2 from several data points for each patient, in order to minimise possible bias due to variability in spot daytime SpO2 values at steady state.26

Narang et al 19 evaluated cross-sectionally overnight oxygen saturation in 37 children with SCD who were receiving HU compared with 104 who were not receiving HU. All the patients enrolled had been referred to a sleep laboratory, mainly for a history suggestive of OSA. Overnight sleep parameters for oxygenation and awake spot SpO2 values were significantly higher in the HU group. However, the cross-sectional design of this study precludes the possibility of inferring whether its findings depended on the effect of HU or on different baseline characteristics of the study and control group. Moreover, the fact that most of the patients included suffered from SDB hinders the generalisation of its conclusions to the entire SCD population. Our study confirmed the preliminary cross-sectional findings of Narang et al,19 showing that median asleep and waking SpO2 increased, respectively, by 2% (p=0.01) and 3% (p=0.001), on a longitudinal evaluation covering up to 3 years from the introduction of HU.

The precise mechanism by which HU increases HbF is not fully understood. Various theories have been postulated including a cytotoxic effect of HU on late erythroid precursors leading to recruitment of early erythroid precursors with increased capacity to produce HbF,30 modification of transcription factors that alters the ratio of haemoglobin A (HbA) to HbF31 and a nitric oxide-derived mechanism for HbF induction.32 New theories continue to emerge, but all culminate in an increased production of HbF, which decreases the rate of HbS polymerisation thereby reducing the downstream pathological events, including haemolysis, anaemia, inflammation and tissue infarction. Besides, HbF has a higher affinity for oxygen than HbS or HbA due to its lower affinity for 2,3 DPG, shifting the ODC to the left and increasing oxyhaemoglobin saturation for a given PaO2 in arterial blood. Interestingly, in the present study, though median increase of HbF after commencing HU was lower than expected (table 1), we could still appreciate a significant increase of asleep and waking SpO2, moderately correlated with changes in Hb and HbF.

Chronic hypoxaemia is an indication for children with SCD to be offered HU in the UK, although the recommendation is based on poor quality evidence (1C).16 There is a need for evidence of clinical benefit that this study attempts to provide. Only two of the children in this study started HU due to hypoxaemia, but a significant increase in oxygenation was still observed. We could only speculate whether a larger effect might have been seen if more of the children had had low baseline oxygen saturation.

In our study, the majority of patients were started on HU because of frequent episodes of pain. They were on moderate doses of HU with some leeway for making increases. It is possible that larger improvements in oxygenation might have been achieved on higher doses. Reaching the maximum tolerated dose may be an appropriate strategy if hypoxaemia were the indication for HU.

A strength of this study is that, at the best of our knowledge, it is the largest longitudinal evaluation of the effects of HU on oxygen saturation in patients with SCD. Furthermore, integrating both asleep and waking SpO2 data from multiple data points for each patient allowed to minimise possible biases due to intrasubject variability in SpO2 values, especially as regards to spot daytime SpO2.33

A major limitation is the absence of a control group of patients with SCD who did not start HU treatment. This does not allow to exclude that changes in SpO2 over the study period were due to factors other than HU. For example, changes in daytime and overnight oxygen saturation with HU in patients with asthma were similar to those of the general sample (online supplementary table S1), but since information on changes in severity of asthma symptoms and therapy over the study period was not available, we cannot exclude that the improved oxygen saturation in this subgroup was due to a better asthma control from baseline to endpoint. Moreover, since in participants with SDB only data from nocturnal oximetry but not from polysomnography were available, the exact prevalence and trends of OSA over the study period could not be established. However, considering that the frequency of ODI ≥3 in children with SDB did not change significantly from baseline to endpoint, it is unlikely that variations in the prevalence of OSA affected the results of nocturnal oximetry pre-HU and post-HU. Furthermore, when we performed the analysis excluding three patients who had adenotonsillectomy for OSA after starting HU (in these children improvements of SpO2 could depend on relief of upper airway obstruction rather than HU), differences in nocturnal SpO2 remained significant. Regarding other potential confounders, over the study period, the centres involved did not apply major changes in the standards of care for SCD patients (apart from increasing use of HU), and the SpO2 would have been expected to decrease, rather than increase, by time.5 21 34 Therefore, it is unlikely that variables not taken into account had a relevant influence on changes in oxygen saturation in patients on HU.

Another limitation of this study is related to the use of pulse oximetry as measure of steady state oxygen saturation in children with SCD, as pulse oximetry tends to overestimate arterial oxygen saturation compared with co-oximetry (gold standard).25 35

We were unable to measure adherence to HU therapy, but the rise in HbF provided some induction that it was being taken. The relatively small study’s retrospective design meant data gaps were inevitable. There were not enough lung function data available to draw any conclusions here. Comparable sleep study data were available on only half the children who were receiving HU, and even then the reports were not easily available, and some data had to be extracted from clinic letters. Selection bias could not be excluded as we only looked at children with SCD who were referred for respiratory symptoms. Indeed, it was not possible to control for confounding factors that may have affected SpO2 in addition to HU. To do so, a randomised placebo-controlled trial of HU therapy in children would be required. Ethical approval for such a trial now may not be straightforward. Confining a study to a particular subgroup of children with SCD, for example, those with respiratory complaints in our case, or a particular indication for HU, helps to reduce confounding factors but then begs the question of whether conclusions can be generalised to other children with SCD.


Granting all the limitations of this study, the data do indicate higher oxygenation, both on overnight and daytime measurements, with HU therapy. This was despite the fact that less than 5% of the children with SCD had started HU because of persistent hypoxaemia. In other words, even children without resting hypoxaemia appear to have higher SpO2 on HU. This study therefore provides important preliminary evidence on which to justify giving HU to treat persistent hypoxaemia in children with SCD until better evidence from a randomised controlled clinical trial is forthcoming.



  • LvG and MA contributed equally.

  • Contributors AG conceptualised and designed the study, led on the analysis and interpretation of the data and contributed to the manuscript. LvG contributed to the design of the study, carried out the data collection and drafted the initial manuscript. MA contributed to the analysis, contributed to the manuscript and drafted the subsequent revisions. BS carried out the data collection, analysis and reviewed the initial manuscript. CJB, SC, BI, WK, GR and DR contributed to the study design and critical revision of the initial manuscript as well as subsequent revisions. All authors approved the final manuscript as submitted.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available on reasonable request. Deidentified participant data are available on reasonable request, contacting Dr Atul Gupta, King’s College Hospital, Denmark Hill, London, UK, SE5 9RS (, 020 3299 4574.