Background Adults with sickle cell disease (SCD) and pulmonary hypertension have high mortality but death in SCD children with pulmonary hypertension is rare. The authors hypothesised that pulmonary hypertension in SCD children may be secondary to anaemia-induced high cardiac output rather than pulmonary vascular disease.
Methods Two independent, validated techniques were used to estimate pulmonary vascular resistance (PVR) in 50 SCD children and 50 matched controls. Tricuspid regurgitant jet velocity (TRV) and right ventricular outflow tract velocity time integral were measured using Doppler echocardiography; PVR was calculated from their ratio. Acetylene rebreathing technique using respiratory mass spectrometry was also performed to calculate pulmonary blood flow and stroke index, an estimate of PVR.
Results TRV was higher in SCD children compared with controls (2.28 vs 2.14 m/s, p=0.02). Fifteen of 34 (44%) children with haemoglobin of the SS genotype (HbSS) versus 1/16 (6%) children with haemoglobin of the SC genotype (HbSC) had pulmonary hypertension (TRV≥2.5 m/s) (p=0.009). Right ventricular stroke volume was higher (p<0.05) and Doppler PVR lower (1.20 (0.19) vs 1.31 (0.20) Wood units, p=0.04) in SCD children with pulmonary hypertension compared with controls. Qpeff and stroke index were higher in SCD children compared with controls (p<0.001 for both) and correlated with anaemia (p<0.001) and TRV (p=0.03). There was no correlation between TRV and history of asthma or acute chest syndrome.
Conclusions Pulmonary hypertension due to raised cardiac output is common in HbSS SCD children and is associated with normal PVR. PVR should be measured before therapy with agents such as sildenafil or bosentan is contemplated.
Statistics from Altmetric.com
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.
Pulmonary hypertension is associated with a 10-fold greater risk of death1,–,5 in adults with sickle cell disease (SCD) in whom Doppler measurement of peak regurgitant flow across the tricuspid valve regurgitation velocity (TRV) during systole correlates well with invasive measurement of pulmonary vascular resistance (PVR).1 5 Pulmonary hypertension in SCD is defined as a peak TRV≥2.5 m/s, corresponding to an estimated systolic pulmonary artery pressure (PAP) of at least 30 mm Hg.1 Although the elevation in PAP tends to be mild, pulmonary hypertension is the leading diagnosis in young adults dying with SCD.2 5 6 It is unclear whether pulmonary hypertension is the cause of death or a marker for a more generalised severe vasculopathy or end-organ damage.7 8 Although the reported prevalence of pulmonary hypertension in adults and children with SCD is similar, at approximately 32%,1 5 9,–,18 children with pulmonary hypertension do not have the high mortality of adults.14 19 20 We hypothesised that the better prognosis in children with pulmonary hypertension may be because pulmonary hypertension in children is a flow-mediated phenomenon caused by anaemia-associated, raised cardiac output rather than increased PVR associated with pulmonary vascular disease (PVD).
What is already known on this topic
▶ Pulmonary hypertension is a common and serious complication in adults with sickle cell disease (SCD) with an estimated prevalence of 32% and 40% mortality within 40 months of diagnosis.
▶ Pulmonary hypertension diagnosed using echocardiography is equally common in children with SCD, but death is rare and the precise aetiology and age of onset are unclear.
What this study adds
▶ Pulmonary vascular resistance (PVR) assessed non-invasively is lower in SCD children with pulmonary hypertension compared with controls and is associated with anaemia-related increased pulmonary blood flow.
▶ Accurate, possibly invasive measurement of PVR is essential in SCD children with elevated tricuspid regurgitant jet velocity before therapy with sildenafil or bosentan is initiated.
Methods and materials
We performed a single-centre, prospective study of the prevalence of pulmonary hypertension and estimated PVR by using two validated, independent, non-invasive methods in 50 (26 female) children with SCD and 50 control subjects. Consecutive patients from the haematology outpatient clinic at St George's Hospital with haemoglobin of the SS genotype (HbSS) or haemoglobin of the SC genotype (HbSC) disease were invited to participate. Eligibility criteria were age above 10 years, genotype HbSS or HbSC and Afro-Caribbean ethnicity in at least one parent. Exclusion criteria were HIV infection, a history of stroke and congenital heart disease. All children were in a baseline state of health, defined as being free from acute illness for a minimum of 14 days prior to the study. Fifty (25 female) healthy age-, sex- and race-matched control subjects were recruited from friends and siblings of patients. A full clinical history and examination including Tanner pubertal staging was performed in all subjects.21 Pulmonary symptoms including number of hospital admissions with vaso-occlusive crisis (VOC) in the previous 2 years, total number of admissions with acute chest syndrome (ACS), history and severity of doctor-diagnosed asthma (according to British Thoracic Society guidelines),22 atopic disease and prescribed medications were recorded in all subjects.
Doppler echocardiography to assess the prevalence of pulmonary hypertension by measuring TRV was performed in all SCD patients and controls (n=100). In a subgroup of patients (n=29; HbSS=20, HbSC=9) and controls (n=28), Doppler echocardiography was performed to estimate PVR. On a separate occasion, acetylene rebreathing using respiratory mass spectrometry (RMS) was performed in patients and controls (n=96) to measure pulmonary blood flow (Qpeff) and calculate stroke index, an accurate predictor of PVR.23 Haemoglobin (Hb), reticulocyte count, percentage of fetal Hb and Hb electrophoresis were determined in all subjects and compared with baseline oxygen saturation, history of VOC and ACS, and doctor-diagnosed asthma to determine relevant associated factors.
Measurement of TRV and right ventricular stroke volume
Echocardiography was performed by either a senior echocardiographer (CH) or cardiology research fellow (TK), both blinded to subject status. A Vivid-7 Ultrasound machine (GE Healthcare, Chalfont St Giles, UK) was used to obtain continuous 2D colour Doppler measurements after a minimum 10 min rest period. A single good quality velocity/time wave across the tricuspid valve was selected for measurement of peak TRV (m/s) in each subject. TRV is routinely used to calculate peak right ventricular systolic pressure (RVSP; mm Hg) which in the absence of pulmonary outflow tract obstruction equates to peak systolic PAP. Any subject who had TRV≥2.5 m/s was asked to attend for repeat measurement. The median interval between visits was 20 months (range 16–27.5). Reproducibility and repeatability studies were performed in 10 subjects.
We used the simplified Bernoulli equation,1 PAP (mm Hg)=4(TRV2)+right atrial pressure (RAP) to determine PAP. RAP was assumed to be 5 mm Hg in all cases.10 Right ventricular stroke volume (RVSV; ml) was calculated by multiplying the right ventricular outflow tract velocity time integral (RVOT VTI) by pulmonary valve cross-sectional area, as previously described.24
RVOT VTI is a pulsed Doppler value measured in centimetres, obtained by placing a 1–2 mm sample volume in the RV outflow tract adjacent to the pulmonary valve and parallel to the blood flow direction in the parasternal short axis view. After manually tracing the obtained Doppler spectrum, the RVOT VTI values are measured by the built-in calculation package of the ultrasound unit and are equal to the area enclosed by the baseline and Doppler spectrum.24
Derivation of PVR
PVR can be reliably derived from Doppler studies using the following equation: Doppler PVR (Wood units)=10×(TRV (m/s)/RVOT VTI (cm))+0.16.25,–,27 The derived PVR value is equivalent to Wood units and is therefore applicable to clinical practice.27
Measurement of stroke index using RMS
RMS was performed using commercially available equipment (AMIS 2000; Innovision, Odense, Denmark) on a different day, after Doppler echocardiography, as part of a study to be reported separately on exercise performance in SCD patients. All subjects were at a baseline state of health and free of acute illness for at least 14 days before testing. No subject had been admitted to hospital, or had had ACS or blood transfusion during the study. The mean of three measurements of baseline oxygen saturation was calculated in each participant prior to commencing RMS. Acetylene (C2H2), a highly soluble gas, and the insoluble gas sulphurhexafluoride (SF6) were used as part of a rebreathing protocol to measure effective Qpeff (l/min),28,–,30 namely that part of the Qpeff which perfuses gas exchanging alveoli. Qpeff is approximately 95% of cardiac output in health, in the absence of significant right to left shunting (excluded on echocardiography) and its measurement by RMS is reliable in patients with pulmonary hypertension.31 Stroke index (ml/m2), calculated by dividing Qpeff by pulse rate, shows excellent inverse correlation with direct measures of PAP at cardiac catheterisation in patients with pulmonary hypertension and in normal control subjects.23 Details of the RMS protocol have been validated by others and reported previously.29 31 In brief, the subject wore a nose clip and re-breathed the test gas mixture at a rate of 40 breaths a minute for 20 s. This manoeuvre was repeated five times at 3 min intervals. The mean of the final three measurements was used for analysis. During re-breathing, C2H2 concentration decreases as a result of dilution by mixing with native gas already within the lungs and being dissolved in blood perfusing the gas-exchanging alveoli. The extent of dilution is corrected for from the simultaneous changes in the concentration of SF6. Cardiac output is calculated from the disappearance curve of acetylene.
Echocardiography reproducibility and repeatability studies
TK and CH each re-analysed TRV data from 10 of their patients and controls blinded to their original findings, to determine intra-observer reproducibility; they then exchanged the studies and reviewed them to measure inter-observer reproducibility for CH. To test repeatability, five patients and five controls were randomly selected and asked to attend for repeat scans; there was an average delay of 9.8 months. As previously, all subjects were in a steady state of health and free of any VOC for at least 2 weeks. Each operator (MM and CH) made three sequential measurements of TRV within 20 min of each other.
The Stata 9.2 (Stata Corp Lp, Texas, USA) package was used for analysis. Fisher's Exact Test (two-sided) was used for categorical data. Haematological, echocardiographic and RMS data, presented as mean and SD, were analysed using independent t tests. Mass spectrometry data was analysed using SPSS 14.0 (SPSS, Chicago, Illinois, USA). Linear regression analysis was used to determine clinical association. Remaining data including oxygen saturation are presented as median and IQR and analysed using Mann–Whitney tests. The Bland and Altman method was used for repeatability and reproducibility of echocardiographic data.32
Clinical characteristics of SCD patients and control subjects
Patient and control demographics are shown in table 1. Although SCD patients and controls were matched for age and gender, 82% of patients compared with 100% of controls were pubertal (p=0.003). Accordingly, patients were significantly shorter than controls although there was no difference in body mass index (BMI) (table 1). The prevalence and severity of doctor-diagnosed asthma were the same in SCD patients as in controls (table 1). Four children (HbSS) were on chronic transfusion therapy and two (HbSS) were being treated with hydroxyurea. Hb was lower in HbSS compared with HbSC patients: 8.8 (1.7) versus 11.2 (0.7) g/dl (p<0.0001). Reticulocyte count, a marker of haemolysis, was higher in patients than in controls: 227 (94) versus 88 (37)×109/l (p<0.001) and correlated with the degree of anaemia: r=−0.73 (p<0.001). Eleven SCD patients (three female, eight male) experienced ACS prior to study enrolment. The three female patients had 1, 2 and 5 ACS episodes and the eight male patients had 1, 1, 1, 1, 2, 2, 3 and 5 ACS episodes. No SCD patient experienced ACS during the study period.
TRV and estimated pulmonary artery systolic pressure
A satisfactory TRV envelope was obtained in all 100 subjects. TRV was higher in SCD patients compared with controls: 2.28 (0.60) versus 2.14 (0.27) m/s (p=0.02), and higher in HbSS compared with HbSC patients, although this was not statistically significant: 2.33 (0.37) versus 2.16 (0.21) m/s (p=0.08). Sixteen of 50 (32%) SCD patients and 4/50 (8%) controls had a TRV≥2.5 m/s, suggestive of pulmonary hypertension (figure 1). Pulmonary hypertension was more common in HbSS than in HbSC patients: 15/34 (44%) versus 1/16 (6%) (p=0.009) (table 2). Estimated PAP was higher in SCD patients compared with controls: 26.2 (6.2) versus 23.6 (4.5) mm Hg (p=0.018), and higher in HbSS compared with HbSC patients: 27.3 (6.8) versus 23.8 (3.7) mm Hg (p=0.058). There was weak inverse correlation between estimated PAP and plasma Hb levels: r=−0.27 (p=0.007). Linear regression analysis showed that in SCD patients, there was no correlation between pulmonary hypertension and age, pubertal status, sex, number of ACS, number of VOC or asthma. Only one control had TRV>2.5 m/s. Her initial measurement was 2.9 m/s, and 2.5 m/s when repeated. Three controls had TRV=2.5 m/s. Repeat echocardiography performed on a different day after exercise testing had been completed, showed that two had TRV=2.4 m/s, while the remaining control had TRV=2.5 m/s. Of the 14/16 SCD patients with pulmonary hypertension, only one had a repeat TRV<2.5 m/s, and mean repeat TRV was unchanged: 2.65 (0.18) versus 2.63 (0.21) m/s (p=0.74). No SCD patient was admitted to hospital during the study.
Right ventricular stroke volume
SCD patients with pulmonary hypertension had significantly higher RVSV compared with SCD patients without pulmonary hypertension and controls with normal TRV values: 95.1 (28.0), 84.4 (25.0) and 72.2 (26.0) ml, respectively (p<0.05). RVSV was higher in HbSS compared with HbSC patients: 93.3 (26.0) versus 77.0 (24.7) ml, but this was not significant (p=0.1). Three controls had elevated RVSV (>80 ml)33 and previously diagnosed anaemia, with Hb between 9.0 and 11.0 g/dl.
Pulmonary vascular resistance
Doppler PVR was lower in SCD patients than in controls: 1.20 (0.19) versus 1.31 (0.2) Wood units (p=0.04), but there was no difference in HbSS compared with HbSC patients: 1.18 (0.2) versus 1.25 (0.18) (p=0.4). Lower PVR tended to be associated with lower Hb, but this was not significant (p=0.08).
Echocardiography reproducibility studies
Intra-observer results for CH were: mean difference 0.004, 95% CI −0.309 to 0.317. Intra-observer results for TK were: mean difference 0.004, 95% CI −0.183 to 0.193. Inter-observer results for CH versus TK were: mean difference −0.003, 95% CI −0.244 to 0.239.
Echocardiography repeatability studies
As with reproducibility studies, there were good levels of agreement in the repeatability studies. Intra-observer results for CH were: mean difference −0.037, 95% CI −0.930 to 0.856. Intra-observer results for MM were: mean difference 0.069, 95% CI −0.474 to 0.612. Inter-observer results for CH versus MM were: mean difference 0.106, 95% CI −0.0422 to 0.634.
Respiratory mass spectrometry derived pulmonary blood flow and stroke index
Of 96 subjects who attended for RMS, three SCD patients had baseline oxygen saturation ≤92% in air and were not permitted to proceed with rebreathing tests. Qpeff was higher in SCD patients than in controls: 3.89 (0.98) versus 2.97 (0.72) l/min/m2 (p=0.02), and higher in HbSS than in HbSC patients: 4.17 (0.93) versus 3.29 (0.82) l/min/m2 (p=0.004). Qpeff correlated inversely with Hb: r=−0.54 (p<0.001) (figure 2). Stroke index was higher in SCD patients than in controls: 49.6 (12.8) versus 40.0 (11.59) ml/m2 (p<0.001), and tended to be higher in HbSS than in HbSC patients: 50.0 (12.6) versus 42.1 (12.3) (p=0.058). Stroke index correlated with TRV (r=0.24; p=0.03) (figure 3) and inversely with Hb levels (r=−0.41; p<0.001) (figure 4).
There was no difference in median (interquartile range, IQR) baseline oxygen saturation between SCD patients (n=43) and controls (n=45): 99% (97%–100%) versus 99% (98%–100%) (p=0.98). There was also no difference according to genotype: HbSS 98.3% (97%–100%) versus HbSC 99.3% (98%–100%) (p=0.31). Baseline oxygen saturation was inversely correlated with TRV in SCD patients but not in controls: r=−0.39 (p=0.03) and r=−0.2 (p=0.3), respectively.
This study is the first to prospectively evaluate pulmonary hypertension in SCD children in a baseline state of health using two independent, validated techniques to estimate PVR. Overall pulmonary hypertension prevalence was 32% in SCD children but was significantly higher in those with HbSS genotype compared with HbSC (44% vs 6%, respectively). Pulmonary hypertension in SCD children was associated with high RVSV and higher systolic PAP but lower Doppler PVR, suggesting that pulmonary hypertension was secondary to increased Qpeff. The lower Hb levels observed in children with HbSS may explain their higher incidence of pulmonary hypertension in comparison with HbSC children. Our prevalence rate for pulmonary hypertension is similar to other reported rates in SCD children in a baseline state of health10 11 13 18 34 but is higher than the 11% rate observed in a recent prospective study where pulmonary hypertension was defined as TRV≥2.6 m/s.35 If we were to use the same definition in our study, then 24% of SCD children would have pulmonary hypertension.
Increased cardiac output is a recognised feature of chronic anaemia, and generally, the higher the cardiac output, the lower the peripheral vascular resistance.36 Others have shown that in children with HbSS, the increase in cardiac output results from an increase in left ventricular stroke volume and is usually not accompanied by significant pulmonary hypertension.37 The only other prospective study of pulmonary hypertension and echocardiographically derived PVR index in SCD children found no difference in PVR between SCD and controls, although SCD children with pulmonary hypertension did have higher PVR than those without pulmonary hypertension.38 High cardiac output with normal PVR has also been reported in SCD adults.5 10 We confirmed increased Qpeff in SCD children in our study using acetylene rebreathing and measured resting stroke index which is an accurate predictor of PVR.23 The correlation between stroke index and PVR is inverse, due to the excessive after-load imposed by PVD restricting right ventricular output. Stroke index was higher in SCD children compared with controls in this study and correlated with TRV, indicating that pulmonary hypertension was likely to be due to high Qpeff rather than raised PVR. However, one of the weaknesses in our study design is that we do not have data obtained from right heart catheterisation which would have enabled us to make direct measurements of PVR and compare them with non-invasively derived values. A second weakness is that this is the first time acetylene rebreathing has been used to determine stroke index in patients with anaemia, although the method has been validated by comparison with PVR and cardiac output data obtained during right heart catheterisation in normal adults and adults with primary pulmonary hypertension.23 We thought right heart catheterisation to be unethical in the context of a study involving healthy or (in the case of SCD) relatively symptom-free children. Other weaknesses include the use of data from SCD adults for the definition of pulmonary hypertension,1 but the advantage of these data is that they have been validated against data measured invasively during cardiac catheterisation, whereas there are no such data for children. We assumed RAP to be 5 mm Hg in everyone and this may have introduced error. Acetylene rebreathing was not performed on the same day as echocardiography as this was not logistically possible, but the consistency in repeat TRV values over time suggests that this did not introduce serious error. Strengths of the study include the use of two independent, validated methods to determine PVR prospectively, the inclusion of matched controls and the assessment of the reproducibility and repeatability of echocardiographic variables.
In contrast to previous prospective studies in adult SCD patients,1 9 and a retrospective study in paediatric SCD patients,12 we were unable to demonstrate a correlation between pulmonary hypertension and age, sex, pubertal status, ACS, number of VOC admissions in the previous 2 years or asthma. The lower incidence of asthma in SCD children was surprising and may be an underestimate since in one study, 53% of SCD children with normal lung function had received a doctor-diagnosis of asthma39 and bronchial hyper-reactivity is common in SCD, occurring in up to 77.5% of children.40 Conversely, respiratory symptoms in SCD children may have been falsely interpreted as due to asthma in other studies. SCD children and controls in this study underwent screening for asthma and bronchial hyper-reactivity and these results will be reported separately. Nocturnal oxygen desaturation in SCD children is associated with increased markers of cellular activation and has been implicated in the development of VOC.41 The desaturation tends to be modest (90%) and is associated with lower daytime oxygen saturation (91%).41 Only three SCD patients in our cohort had low daytime oxygen saturation (<92%) but were not included in the current study. There is no evidence to date to show a causal link between nocturnal oxygen desaturation and the development of PVD in SCD, although this is a possibility. Correlation between TRV and markers of haemolysis has been demonstrated in children10 11 13 18 and adults,1 9 and although reticulocyte count was higher in our SCD patients compared with controls, there was no correlation with TRV. Lactate dehydrogenase is a useful indicator of haemolysis in SCD,42 but studies demonstrating this had not been published when we designed our protocol. Four (8%) of 50 controls had elevated TRV, being 2.5 m/s in three subjects and 2.9 m/s in the fourth. There are no published studies on the reproducibility and repeatability of echocardiographic variables in healthy black children. However, screening studies in adults have indicated possible racial variation in PAP43 and ethnic-specific normal echocardiography data are necessary therefore to properly define pulmonary hypertension.
We have shown that pulmonary hypertension in SCD children in our study is accompanied by increased Qpeff and that this is probably related to anaemia. Children with SCD had lower PVR and higher stroke index than controls, indicating that the pulmonary hypertension is likely to be due to high blood flow. We suggest that careful assessment of PVR in SCD children, possibly invasively, is essential before beginning treatment with agents such as sildenafil or bosentan. Our findings provide a possible explanation for the different mortality observed in children and adults with SCD who have an echocardiographic diagnosis of pulmonary hypertension based on measurement of TRV.
Funding St George's Healthcare NHS Trust Charitable Trustees and the Sobell Institute (UK) provided funding for this study.
Competing interests None.
Ethics approval Ethics approval was granted by Wandsworth Local Ethics Committee, ref no. 04/Q0803.
Provenance and peer review Not commissioned; externally peer reviewed.