Exercise tolerance, lung function abnormalities, anemia, and cardiothoracic ratio in sickle cell patients


  • Conflict of interest: Nothing to report


Many patients with sickle cell disease (SCD) have a reduced exercise capacity and abnormal lung function. Cardiopulmonary exercise testing (CPET) can identify causes of exercise limitation. Forty-four consecutive SCD patients (27 HbSS, 11 HbSC, and 6 HbS-beta thalassemia) with a median age (interquartile range) of 26 (21–41) years underwent pulmonary function tests, CPET, chest x-ray, and echocardiography to further characterize exercise limitation in SCD. Peak oxygen uptake (V′O2-peak), expressing maximum exercise capacity, was decreased in 83% of the studied patients. V′O2-peak correlated with hemoglobin levels (R = 0.440, P = 0.005), forced vital capacity (FVC) (R = 0.717, P < 0.0001). Cardiothoracic ratio on chest x-ray inversely correlated with FVC (R = −0.637, P < 0.001). According to criteria for exercise limitation, the patients were limited in exercise capacity due to anemia (n = 17), cardiovascular dysfunction (n = 2), musculoskeletal function (n = 10), pulmonary ventilatory abnormalities (n = 1), pulmonary vascular exercise limitation (n = 1), and poor effort (n = 3). In the present study we demonstrate that anemia is the most important determinant of reduced exercise tolerance observed in SCD patients without signs of pulmonary hypertension. We found a strong correlation between various parameters of lung volume and cardiothoracic ratio and we hypothesize that cardiomegaly and relative small chest size may be important causes of the impairment in pulmonary function, that is, reduced long volumes and diffusion capacity, in SCD. Taking into account anthropomorphic differences between SCD patients and controls could help to interpret lung function studies in SCD better. Am. J. Hematol. 89:819–824, 2014. © 2014 Wiley Periodicals, Inc.


Sickle cell disease (SCD) is an inherited disorder in which a single DNA point mutation results in the production of abnormal hemoglobin. Signs and symptoms are found in homozygous (HbSS) patients, as well as in heterozygous patients if the mutation is accompanied by another heterozygous hemoglobin disorder, such as hemoglobin C (HbSC) or β-thalassemia (HbSβ0 or HbSβ+). SCD is characterized by chronic hemolytic anemia and (micro)vascular occlusion, resulting in accumulating organ damage, a decreased quality of life and reduced life expectancy. Pulmonary hypertension (PH) may develop in about 6% of adult SCD patients during the course of their disease; and, restrictive lung disease, cardiomegaly, and a reduced exercise capacity can be observed in most patients [1, 2]. Various studies examined the mechanisms responsible for the reduced exercise capacity in SCD [3-8]. Proposed mechanisms are various, and include pulmonary parenchymal damage [9], cardiac limitation [4, 5], lowered general physical fitness [6], restrictive and obstructive pulmonary disease [1, 10], pulmonary vascular disease [7], peripheral vascular limitation [7], lowered oxygen affinity, inferior rheology of sickle blood [8], and a reduced oxygen carrying capacity.

Cardiopulmonary exercise testing (CPET) can identify causes of exercise limitation. In patients with PH the results are related to survival [11]. Since only few studies were performed before in highly selected groups of SCD patients, CPET using current ATS/ACCP standards could be very informative in patients with SCD [7, 12-14]. In a case–control study including SCD patients with PH, Anti et al. performed CPET and concluded that reduced exercise capacity was mainly related to PH and anemia [13]. Callahan et al. described CPET findings in a group of 17 women with SCD and concluded that a pulmonary vascular limitation was the main cause of the decreased exercise tolerance [7]. No other observational studies in unselected adult SCD patients have been performed.

The aim of the present study was to determine the cause of cardiopulmonary exercise limitation in an unselected group of adult SCD patients. To further interpret the results we additionally performed chest x-ray, pulmonary function tests, and echocardiography in all patients.



Consecutive adult sickle cell patients (HbSS, HbSC, HbSβ+-thalassemia, or HbSβ0-thalassemia, confirmed by high performance liquid chromatography), who visited the outpatient SCD clinic of the Department of Hematology of the Academic Medical Centre (AMC) for routine scheduled evaluation were considered eligible for this study. The AMC SCD cohort is a highly characterized group of relative young SCD patients. Results on radiologic signs of pulmonary fibrosis, lung perfusion defects, coagulation abnormalities, vascular autoregulation, and nitric oxide (NO)-metabolism were published previously [15-21]. Inclusion criteria were: age 18 years or older and the physical capacity for CPET. Exclusion criteria were: pregnancy, a recent vaso-occlusive crisis (<2 weeks before) or acute chest syndrome (<4 weeks before).

Study design

The cross-sectional study design included two visits. Laboratory testing (complete blood count, Hb-electrophoresis, creatinine, urea, LDH, bilirubin, NT-pro BNP), pulmonary function testing and transthoracic echocardiography (TTE) was routinely performed during scheduled evaluation. During a second visit, within 2 weeks after the first, a symptom limited CPET was performed.

The research protocol for this study was approved by the local Institutional Review Board and the study was conducted in accordance with the principles of the Declaration of Helsinki. All patients gave written informed consent before enrolment in the study.


Symptom limited CPET was performed and assessed according to the guidelines of the American Thoracic Society [12]. Briefly, patients were placed on a cycle ergometer in the upright position and continuous measurements were made of minute ventilation (V′E), oxygen consumption (V′O2), carbon dioxide production (V′CO2), heart rate (HR), blood pressure, and electrocardiography. Workload was increased by 5–15 Watt, depending on the predicted maximum exercise capacity and in such a way that maximal effort was attained within 10–15 minutes. Peak oxygen consumption (V′O2-peak) was defined as the highest attained value averaged over 8 breaths. Oxygen pulse (O2-pulse) was calculated as V′O2 divided by HR. Anaerobic threshold was determined using the V-slope method. Instead of Oxygen saturation HbO2 was used which is independent of carbon monoxide and methemoglobin concentration.

We used standardized CPET criteria to attribute various patterns of physiological changes during exercise to specific causes [22-24]. Signs of abnormal gas exchange are a peak exercise HbO2 below 88%, and or presence of dyspnea [7, 11].

Pulmonary function testing

Spirometry was performed according to ERS/ATS standards (Jaeger MS diffusion, Wuerzburg, Germany) and included forced vital capacity (FVC), forced expiratory volume in one second (FEV1) and inspiratory slow vital capacity (IVC). Total lung capacity (TLC) was determined by body plethysmography (Jaeger MS Body, Wuerzburg, Germany). Carbon monoxide diffusion capacity (DLCO) and carbon monoxide diffusion capacity corrected for the alveolar volume (KCO) were measured by the single breath technique [25]. Pulmonary function test variables were expressed as percentage of the predicted value for Caucasians [26]. DLCO was corrected for hemoglobin concentration according to ATS/ERS criteria. Patients with SCD have higher HbCO concentrations than normal controls. Although, in general a 1% increase in HbCO reduces DLCO by 1% we choose to apply no correction. All pulmonary function tests were performed according to the ATS/ERS guidelines [11, 16, 25, 26, 28].

Cardiothoracic ratio

Standard anterior–posterior chest x-rays were performed in the erect position. The cardio-thoracic ratio (CTR) was calculated by dividing the transverse cardiac diameter by transverse diameter of the thorax on a standing postero-anterior chest radiograph, taken at a standard 2 m patient-film distance and at deep inspiration. The transverse cardiac diameter was the sum of the maximum extensions of the heart to the right and left of the midline. The thoracic diameter was measured as the horizontal distance between the inner borders of the ribs at the level of the right diaphragm. A cardiothoracic ratio of 0.54 was defined as abnormal [27].


TTE was performed as described before, using conventional clinical echocardiographic equipment with 2.5 or 3.5 MHz transducers [15, 29]. According to the generally used SCD-specific criteria, an elevated tricuspid regurgitation jet flow velocity (TRV) was defined as a peak TRV of at least 2.5 m sec−1, corresponding to a cut-off value two standard deviations above the mean TRV in the general population [30, 31]. Patients with trace or no tricuspid regurgitant jet were considered to have normal pulmonary pressures and were designated to have a TRV of 1.3 m sec−1 [31].

Statistical analysis

Results are expressed as medians with interquartile range (IQR). All analyses were performed with the SPSS 17.0 statistical software (SPSS; Chicago, IL). Pearson's correlation coefficient was calculated for correlation studies, and was tested for two-sided significance. Multivariate linear regression analysis of all individual parameters that correlated significantly with either vital capacity or exercise capacity was performed to calculate their predictive value. A value of P < 0.05 was considered statistically significant.


Forty four consecutive adult SCD patients from the outpatient hematology clinic of the Academical Medical Centre (AMC), Amsterdam the Netherlands participated in this study. Patients included in this study had homozygous (HbSS, n = 27) SCD or compound heterozygous SCD (11 HbSC, 3 HbSβ0, and 3 HbSβ+). Median (IQR) hemoglobin concentration was 9.5 (8.3–11.0) mg dL−1 [5.9 (5.2–6.9) mmol L−1]. Other baseline characteristics are presented in Table 1. None of the patients was on chronic transfusion and six patients were on hydroxyurea.

Table 1. Patient Characteristics and Baseline Laboratory Parameters (n = 44)
  1. Values are numbers or medians (inter quartile range).

Age (yrs)26 (21–41)
Gender (male/female)12/32
Hemoglobin (mg dL−1)9.5 (8.3–11.0)
Fetal hemoglobin (%)4.2 (1.6–13.8)
WBC (×109/mL)8.6 (6.9–10.5)
Bilirubin (umol L−1)40 (24–67)
LDH (U/L)346 (225–463)


The CPET variables are presented in Table 2. None of the patients had signs or symptoms of cardiac ischemia during exercise. Thirty four of the studied patients (83%) had by definition a reduced exercise capacity with a decreased peak oxygen uptake (V′O2-peak), that is, below 84% of the predicted value (Fig. 1). On average, VO2-peak, W-max, HR-max, and V′E-max were decreased while mean RER (SD) at top was 1.13 (0.11), indicating maximal effort. Thirty two out of 41 patients stopped exercise due to leg discomfort; 3 complained of shortness of breath and 5 patients stopped because of a combination of dyspnea and leg discomfort. In none of the patients during exercise a HbO2 below 88% was observed; median (IQR) difference in HbO2 between peak exercise and rest was 1% (0–2) (Table 3). Only one patient had a gas exchange pattern that was consistent with pulmonary vascular limitation to exercise. Echocardiography showed an elevated TRV of 2.8 m sec−1 in this patient. Based on CPET criteria, the other patients were limited in exercise capacity due to anemia (n = 17), cardiovascular dysfunction (n = 2), musculoskeletal function (n = 10), pulmonary function abnormalities (n = 1), and poor effort (n = 3). Percentage of predicted V′O2-peak (as well as V′O2-peak itself) was strongly correlated to the hemoglobin level (R = 0.435, P = 0.005) and FVC (R = 0.581, P < 0.0001) (Figs. 2 and 3). The observed correlations did not change after exclusion of patients with SC disease. A p(A-a)O2 >30 mmHg and/or a p(a-ET)CO2 >0 can be considered indicators of ventilation/ perfusion mismatch. Of the 21 patients that had blood gas results available non had a p(a-ET)CO2 >0 and only two patients (both HbSC) had a p(A-a)O2 difference slightly higher than 30 mmHg.

Table 2. Cardiopulmonary Exercise Test Results (n = 41)
 MeasuredPercentage of predicted
  1. V′O2 peak is peak oxygen uptake at maximum exercise; V′O2 @AT is the oxygen uptake at anaerobic threshold; V′E peak is peak ventilation at maximum exercise; O2-pulse is the oxygen uptake divided by heart rate; EqCO @AT is ventilatory equivalent for carbon dioxide at anaerobic threshold; HR-max is maximum heart rate/min; Delta HbO2(%) is the difference in oxygen saturation between rest and at maximum exercise.

V′O2 peak1.47 (1.23–1.73)68 (61–78)
V′O2 @ AT1.15 (0.98–1.27)54 (41–62)
V′E peak58.2 (51.4–76)63 (51–70)
O2-pulse8.66 (7.66–11.13)n.a.
EqCO @AT28.1 (25.7–29.8)n.a.
HR-max164 (153–180)86 (82–90)
Delta HbO2(%)−0.65 (−2.0 to 0.2)n.a.
Figure 1.

V′O2-peak compared with predicted V′O2-peak. The patients in our cohort had a relative low peak amount of oxygen uptake during exercise (VO2-peak) with a median (interquartile range) of only 1.45 (1.23–1.68) L min−1, which was only 68 (59–78) percentage of expected. The dotted line represents the 84% of predicted V′O2-peak. Dots under this line represent cases which have a significantly reduced exercise capacity.

Table 3. Arterial Blood Gas Results
 BaselinePeak exercise
  1. Percentage (interquartile range) of oxyhemoglobin (HbO2), carboxyhemoglobin (HbCO), and methemoglobin (Methb). Corrected oxyhemoglobin is the percentage oxyhemoglobin of total hemoglobin minus Methb and HbCO.

HbO2 (%)94 (92–95)93 (90–94)
HbCO (%)3 (2–4)3 (2–4)
Methb (%)1 (1–2)1 (1–2)
Corrected HbO2 (%)97 (96–99)96 (94–98)
Figure 2.

Univariate correlations with exercise capacity. The percentage of predicted (A,B) as well as the maximum amount (C,D) of oxygen uptake during exercise (V′O2-peak) were significantly correlated with hemoglobin concentration and vital capacity.

Figure 3.

Univariate correlation of FVC with DLCO(TL,co). FVC was strongly correlated with DLCO in our patients (R = 0.677, P < 0.001). TLC and Va were also strongly correlated with DLCO. The correlations did not differ using the actual measured values or percentage of predicted values. This illustrates that reduced lung volume is a major contributor to a reduced DLCO in the SCD patients in our cohort.

Table 4. Results Lung Function Tests (n = 44)
 MeasuredPercentage of predicted
  1. Results on forced vital capacity (FVC), forced expiratory volume in one second (FEV1) total lung capacity (TLC), and alveolar volume (VA). Transfer factor for carbon monoxide (TL,CO or DLCO) was determined using the single breath technique [29]. DL,CO was corrected for hemoglobin concentration according to ATS/ERS criteria.

FVC (L)3.1 (2.8–3.6)85 (71–93)
FEV1 (L)2.6 (2.3–3.0)79 (71–89)
FEV1/FVC0.83 (0.79–0.86)100 (96–104)
TLC (L)4.2 (3.8–4.8)79 (72–87)
DL,CO (mmol kPa−1 min−1)7.3 (6.1–7.7)77 (66–84)
DL,CO/VA (mmol kPa−1 min−1 L−1)1.8 (1.6–2.1)105 (91–117)
VA (L)3.9 (3.5–4.5)75 (66–84)

Pulmonary function and cardiothoracic ratio

Results of pulmonary function testing are shown in Table 4. On average, FVC, FEV1, and TLC were reduced in the studied patients. Average FEV1/FVC was normal. Average DLCO was decreased, in contrast to the KCO which was normal to high normal in our patients. The median (IQR) cardio-thoracic ratio on chest x-ray was 0.53 (0.50–0.57) and based on a cut-off level of 0.54, 16 (48.5%) patients of 33 patients with available chest x-rays had an elevated CTR. The CTR correlated with FVC (R = −0.637, P < 0.0001), TLC (R = −0.601; P < 0.0001), VA (R = −0.650; P < 0.0001), and FEV1 (R = −0.660; P < 0.0001) (Fig. 4A–D). This didn't change significantly after exclusion of patients with SC disease. After correcting for body height in a multiple linear regression model CTR remained significantly correlated to the lung volume.

Figure 4.

Univariate correlations with cardiothoracic ratio. Cardiothoracic ratio's (CTR) as measured on chest x-ray gives an estimation of chest size in relation to cardiac size. Small chest size is associated with low lung volume in SCD [29]. Cardiomegaly is associated with decrease in lung volume in patients with heart failure [45]. Here we show a strong correlation of pulmonary function parameters and CTR in our patients (A–D). We did not find correlations with CTR and markers that are clearly lung volume independent: DLCO/Va and FEV1/FVC ratio (data not shown).


In our cohort of 44 unselected patients 43 had baseline TTE, 9 patients (21%) had an elevated TRV at rest (2.5–2.8 m sec−1). Nine patients had an undetectable TRV and were designated to have a TRV of 1.3 m sec−1 [31]. Median (IQR) TRV was 2.2 (1.9–2.5) m sec−1. No significant correlation was found between any echocardiographic parameter at rest and V′O2-peak, or any of the other exercise parameters.


Here, we present the results of CPET in 44 consecutive adult patients with SCD. We show that in this group of adult patients with SCD the median exercise capacity was only 68 (59–78) percent of predicted and, by definition, in 81% of the patients the exercise capacity was decreased. We show that as a group these patients had slightly reduced lung volumes with an associated increase in the cardiothoracic ratio. The decrease in lung volumes could explain the previously reported low DLCO in this patient group [1].

V′O2 peak was associated with hemoglobin level and DLCO in our patients. Both anemia and low DLCO could be a potential cause of exercise limitation. A low hemoglobin level leads to a decrease in total oxygen content, but a decrease in DLCO may in theory also lead to the same by a lower oxygen uptake in the lungs. In states of high cardiac output even a relatively minor reduction in pulmonary diffusing capacity may have impact on oxygen uptake capacity of the lung because of the associated increased red cell transit time than in states of normal cardiac output [32, 33]. It is well known that SCD patients have a reduced diffusion capacity [29], and with a median DLCO of 77 (66–84) percent of predicted this was the case in our patient group as well. As SCD patients in general have higher cardiac output compared with controls a small reduction in pulmonary diffusing capacity could contribute to a decrease in exercise capacity in this patient group. However, mean arterial HbO2 did not decrease during exercise in our patients. Therefore, we argue that the observed slightly lower DLCO is of no clinical relevance to exercise capacity in this patient group.

Low oxygen carrying capacity seems obviously, the most important cause of decreased exercise capacity in these anemic patients [34]. This is also suggested by the previously reported negative association of the distance walked in the 6 minute walk test with hemoglobin level in these patients [35, 36]. The inverse relation between hemoglobin level and exercise capacity in our cohort may, however, not necessarily be defined by a reduced oxygen carrying capacity alone. Continuous intravascular hemolysis results in release of cell free hemoglobin which readily reacts with NO at the near diffusion limit, thereby reducing NO bioavailability. Even small amounts are sufficient to inhibit all NO signaling and produce vasoconstriction [37]. By releasing its iron into the plasma membrane of endothelial cells, cell-free heme catalyses nonenzymatic production of reactive oxygen species, resulting in endothelial damage and further reduction in NO bioavailability [38]. Hypothetically, the resulting hemodynamic changes could reduce exercise capacity. However, as limited exercise capacity was associated with pulmonary vascular abnormality in only one patient, NO associated hemodynamic changes seem to be of limited importance in the reduced exercise capacity in SCD.

To further address potential causes of exercise limitation we used pre-defined and generally accepted criteria for interpretation of clinical exercise testing [22, 39-41]. The individual interpretation of all exercise tests in our study showed that the majority of patients had a reduced exercise capacity because of anemia or musculoskeletal dysfunction which includes deconditioning. One out of 41 (∼2%) patients in our cohort had a pulmonary vascular limitation that was associated with an increased TRV at rest and therefore is likely to suffer from PH [22-24, 29]. The recent study of Parent et al. showed that PH was present in 24 out of 398 (∼6%) screened SCD patients [2]. This lines up quite well with our results. Our findings, however, are in contrast to another study, showing a pulmonary vascular limitation in 11 out of 17 selected female SCD patients [7]. Interestingly, these results found by Callahan et al. resemble the results of the SCD patients with right heart catherization proven PAH in the case–control study by Anthi et al [13]. Callahan and co-workers did not report TRV values nor the prevalence of PH in their study group, but we speculate that the reported pulmonary vascular limitations are likely to be PH-related as well and therefore the included patients represent a highly selected subgroup of SCD patients [7].

Taken together, our results suggest that anemia appears the most important factor determining exercise capacity in mildly affected SCD patients. This is in contrast with more severely affected patients with SCD-related PH in whom pulmonary vascular limitations are likely to be more important [13].

Reduced pulmonary diffusing capacity

The carbon monoxide (CO) uptake from the lung is measured as a concentration fall in alveolar CO per unit time [29]. When multiplied by the volume of the lung containing CO [alveolar volume (VA)], the total uptake of CO by the lung (DLCO) is obtained [29]. Therefore, by definition the DLCO is correlated to the lung volume. Both DLCO and measures of lung volume are repeatedly reported to be low in SCD [15, 29]. The KCO (DLCO corrected for VA), however, is reported to be relatively high, as compared with DLCO and measures of lung volume such as FVC, TLC, and VA in SCD [15, 26, 34, 38]. Also, in our study the KCO levels [median 105 (91–117)] were normal to slightly elevated. Therefore, the reduced DLCO observed in SCD is mainly caused by a reduction in lung volume.

The cause of the lung volume reduction in SCD

It is well known that SCD patients have reduced lung volumes as compared with healthy controls [42, 43]. As both heart and lung are trapped in the narrow space of the chest, an increase in heart size or a decrease in chest size may both cause a decrease in lung volume [44, 45]. Because the cardio-thoracic ratio incorporates an estimate of heart size as well as an estimate of chest size we added the CTR as parameter to this study, which appeared to be elevated in most patients and was correlated to pulmonary volumes. SCD patients are known to have cardiomegaly but echocardiographic estimates of heart size and heart weight in autopsy series are arguably not extensive enough to significantly influence lung volumes [46-49]. It is less well noticed that SCD patients have relative small chest size compared with body size compared with healthy controls [5]. Studies by Serjeant et al. suggested that differences in chest size partly explained the observed differences in lung function between SCD patients and healthy controls [44, 50]. Therefore, expressing lung volumes as a percentage of the predicted values based on body height will cause an overestimation of the prevalence of restrictive lung disease in these patients [44]. Thus, chest size must be taken into account when assessing lung volume in SCD. Combined, we suggest that a reduction in chest size might be an often overlooked, but clinically important determinant of reduced lung volumes in SCD.

Additional comments and study limitations

A few methodological aspects of our study need additional comment. Firstly, although we are not aware of a report on CPET in SCD that included more patients, our study is still relatively small. This study is therefore underpowered to find small effects or differences. Any statements on absence of correlation or difference need to be taken into account carefully. Secondly, we did not perform right heart catheterization because none of the patients had signs or symptoms of PH and/or a TRV ≥ 3.0 m sec−1. Therefore, we cannot fully exclude that one or even more patients with PH might have been included in this patient group. However, given the 8%–11% prevalence of catheterization-proven PH in adults with SCD [2], and the absence of a very high TRV in the vast majority of our SCD patients, we feel the results can be extrapolated to all patients with SCD with comparable low to moderately elevated TRV. Thirdly, it can be speculated from our data that in SCD patients with low TRV and without any other sign or symptom suggestive for PH, the decrease in lung volume and DLCO as well as the reduced exercise capacity is merely the consequence of the reduced chest size and cardiomegaly and not caused by intrapulmonary pathophysiologic factors. However, it does not imply that the observed changes do not contribute to the course and the severity of the disease. We want to emphasize that patients with SCD do have less pulmonary reserve capacity as compared with healthy controls. This is underlined by the fact that even mild PH or a slightly elevated TRV are strong risk markers for early death in this patient group [2, 31].


In conclusion, in the present study we demonstrate that anemia is the most important determinant of reduced exercise tolerance observed in SCD patients without signs of PH. We found a strong correlation between various parameters of lung volume and CTR and we hypothesize that cardiomegaly and relative small chest size may be important causes of the impairment in pulmonary function, that is, reduced long volumes and diffusion capacity, in SCD. Taking into account anthropomorphic differences between SCD patients and controls could help to interpret lung function studies in SCD better.


We are extremely grateful to all patients who took part in this study. The authors would like to thank P.J. Sterk for his help and advice. (Department of Pulmonology, Academic Medical Centre, University of Amsterdam, The Netherlands).