Exertional Desaturation in Patients with Chronic Obstructive Pulmonary Disease

ABSTRACT

Although the Centers for Medicare and Medicaid Services oxygen prescription guidelines utilize a threshold arterial oxygen tension ≤55 mmHg or an oxygen saturation ≤88%, a range of oxygen levels and relative declines have been used in investigations of exertional desaturation in patients with chronic obstructive pulmonary disease (COPD). There is no uniform definition of exertional hypoxemia or standardized exercise protocol to elicit decreases in oxygen levels in individuals with COPD. The causes for exertional desaturation in patients with COPD are multifactorial with ventilation-perfusion mismatching, diffusion-type limitation, shunting and reduced oxygen content of mixed venous blood all contributing to some degree. Neither resting oxygen saturation nor pulmonary function studies can reliably predict which patients with COPD will develop exertional desaturation. However, preserved pulmonary function, especially diffusing capacity, reliably predicts which patients with COPD will sustain oxygenation during exercise. Although exertional desaturation in patients with COPD appears to portend a poor prognosis, there is no evidence that maintenance of normoxemia during exercise improves the survival of these patients. Studies of the effect of supplemental oxygen on exercise performance in individuals with COPD who desaturate with exertion have yielded conflicting results. The use of short-term or “burst” oxygen either prior to or after exertion may not have significant clinical benefit. Differences in the definition of desaturation, mode of exercise, and characteristics of the patient population make it difficult to compare studies of exertional desaturation and its treatment and to determine their applicability to clinical practice.

Keywords: :

• Chronic obstructive pulmonary disease
• Hypoxemia
• Exertional desaturation
• Oxygen

CLINICAL DEFINITION OF EXERTIONAL DESATURATION

The Centers for Medicare and Medicaid Services (CMS) defines the most widely used clinical threshold for exertional hypoxemia in individuals with chronic stable COPD as a PaO₂ ≤ 55 mmHg or an arterial oxygen saturation ≤88% measured during exercise (1). In contrast, various investigators have used different absolute values or relative decrements to define a significant decrease in oxygenation during exertion. Absolute limits for adequate oxygen tension include PaO₂ ≤ 55 mmHg or ≤8–8.5 kPa (60.0–63.8 mmHg). Threshold values for SaO₂ are usually ≤ 88–90% but significant relative declines in SaO₂ vary from 2 to 5%. Most studies consider any measurement beyond these absolute thresholds or relative declines significant. The length of desaturation is infrequently considered but some studies require a desaturation below a threshold value be maintained for a specified duration (usually between 0.5 and 5 minutes) before it is deemed significant.

Last, the type of exertion varies from activities of daily living, 6-minute walk, hall walk, step test exercise, or treadmill walk to incremental maximal or steady state cycle ergometry. Some studies are performed with patients breathing supplemental oxygen whereas, in other investigations, the subjects are breathing room air. There is no uniform definition of exertional desaturation or standardized exercise protocol to elicit decreases in oxygen levels in individuals with COPD.

The use of supplemental oxygen during exercise to ameliorate exertional desaturation varies from fixed dose supplemental oxygen, titrating oxygen flow rates to maintain oxygen saturation above a defined level, relieve symptoms, or improve objective measures. Although oxygen is prescribed frequently for use with exertion in individuals with exertional desaturation, verification of maintenance of normoxemia throughout everyday activities is rarely performed and compliance is rarely documented. Other practices such as short burst oxygen advocate the use of oxygen as needed by the patient to relieve breathlessness.

The clinical characteristics of patients enrolled in studies of exertional desaturation in individuals with COPD vary greatly; the definition of desaturation and type of exertion also differ. Although CMS oxygen prescription guidelines utilize a threshold PaO₂ ≤ 55 mmHg or an arterial oxygen saturation ≤ 88%, a range of oxygen limits and relative declines have been used in investigations of exertional desaturation in patients with COPD. Therefore, it is very difficult to compare studies and determine their applicability to clinical practice. In this review, we will state explicitly the definition of desaturation, mode of exertion, and characteristics of the patient population when discussing specific studies.

WHAT ARE THE CAUSES OF EXERTIONAL DESATURATION OR HYPOXEMIA?

The causes of hypoxemia with exertion are similar to the causes of hypoxemia at rest in patients with lung disease. These 6 causes are listed in Table 1 and all except reduced ambient oxygen tension are discussed in the following sections.

Table 1.  Causes of Hypoxemia

Causes of Hypoxemia

• Ventilation to perfusion mismatching or inequality: V/Q mismatching

• Shunting of venous blood to the arterial side without contact with the gas exchange surfaces

• Diffusion-type limitation: primarily loss of pulmonary capillary surface area

• Alveolar hypoventilation with elevation of carbon dioxide tension or content in blood

• Reduced content of oxygen in mixed venous blood returning to the lungs

• Reduced oxygen tension in the inspired air: at altitude or in oxygen-depleted environments

Ventilation to perfusion mismatching (V/Q mismatching)

In order for adequate gas exchange to occur allowing oxygen to enter from the lungs into the pulmonary circulation and carbon dioxide to be eliminated from the lungs into the atmosphere, appropriate matching between the distribution of inhaled air and the blood flow through the pulmonary vascular bed must occur. At the extremes of the relationship between ventilation and perfusion are shunt (blood flow but no ventilation to an area of the lung) and dead space ventilation (ventilation to an area of lung where there is no blood flow). In general, the lung is a composite of multiple regions or lung units where there are different ratios of ventilation to blood flow or perfusion (V/Q) from 0 (shunt) to infinity (dead space ventilation). The more uniform the distribution of those ratios and the closer the mean of that distribution is to 1.0, the better the overall gas exchange is for the lung as a whole (2, 3).

Under conditions of lung disease such as exists in COPD, there are alterations in both the distribution of ventilation and the distribution of blood flow to areas of the lung. As a result, the ratios of ventilation to perfusion have a much broader distribution with decreases in oxygen content of blood leaving the lung (hypoxemia) (4). It is also thought that such imbalance of ventilation to perfusion ratios can contribute to carbon dioxide retention (5). This mismatching of ventilation to perfusion is thought to be the most common cause of hypoxemia in lung disease.

Shunting of venous blood to the arterial circulation

There are circumstances in certain lung and cardiac conditions where the mixed venous blood returning the right side of the heart through the vena cavas and into the right atrium bypasses the gas exchange units of the lung and is added to oxygenated blood that enters the systemic arterial circulation. These shunts can occur at the level of the heart (inter-atrial or inter-ventricular right to left shunts) or within the lungs themselves (pulmonary arteriovenous malformations or atelectasis or fluid filled alveolar spaces).

As a result, blood that is less oxygenated (mixed venous blood at rest is usually considered to be 75% saturated with oxygen and having an oxygen tension of 40 mmHg) is mixed with arterial blood that is 95–98% saturated with oxygen with a tension of 80–90 mmHg. The degree of hypoxemia depends on the percent of blood that is shunted past the gas exchange surfaces of the lung. Even in a normal individual without lung or heart disease, there is a 4% shunt of mixed venous blood that reaches the systemic arterial circulation.

Diffusion-type limitation

Oxygen from the inspired air and carbon dioxide from the metabolic processes in the body must diffuse across the blood-gas barrier in the lungs in order for oxygen to be delivered to the systemic circulation and for carbon dioxide to be eliminated in the expired air. The ability of the lungs to take up oxygen can be estimated by a diffusing capacity measurement that uses a small concentration of carbon monoxide. A decrease in the diffusing capacity measurement in most instances does not refer to a limitation at the pulmonary blood-gas barrier because of increased thickness of the diffusion barrier (epithelium, interstitium and endothelium), but instead is due to a loss of the pulmonary vascular surface area available for gas diffusion to take place (loss of vascular bed in emphysema or obstruction of the pulmonary vascular bed in disease such as pulmonary emboli).

Usually, this diffusion-type limitation does not result in hypoxemia at rest but can be a major contributor to hypoxemia in exertion. Under resting conditions, there is adequate time for the blood moving through the pulmonary vascular bed to become equilibrated with the oxygen tension in the alveoli and thus saturated with oxygen. With exertion, the cardiac output is increased as there is greater demand for oxygen in the peripheral exercising muscles. As a result, there is a decreased residence time for the blood to be in the pulmonary circulation. In a normal individual, without diffusion-type limitation, there is still enough time for the saturation of the hemoglobin to take place by equilibration with alveolar air. However, when there is diffusion-type limitation with loss of pulmonary vascular surface area and under conditions of exertion with increased cardiac output and reduced residence time or transit time in the lungs, the exiting blood is no longer able to be fully equilibrated with alveolar air and hypoxemia occurs.

Alveolar hypoventilation

In blood, there is a balance between the partial pressure of oxygen and the partial pressure of carbon dioxide. The partial pressure of carbon dioxide in turn is dependent upon the alveolar ventilation and the production of carbon dioxide in the body.

PaCO₂ = VCO₂/V_A × K, where PaCO₂ is the systemic arterial carbon dioxide tension in mmHg, VCO₂ is the production of carbon dioxide in ml/min, and V_A is the alveolar ventilation in L/min or the ventilation that occurs at the level of the alveolus and is involved in gas exchange.

When the alveolar ventilation is reduced, the carbon dioxide tension or PaCO₂ increases. When that occurs, the PaO₂ will decrease. PaO₂ is determined first by knowing the PaO₂ (alveolar oxygen tension) according to the Alveolar Air Equation (in simplified form). As can be seen, an increase in PaCO₂ or hypercarbia as found in some lung conditions will result in a reduction in PaO₂. There is usually a small difference between the PaO₂ (alveolar oxygen tension) and the PaO₂ (arterial oxygen tension) which is known as the A-a oxygen difference (at rest normally 5–15 mmHg and increasing with age). Hence, the increase in PaCO₂ will result in a decrease in PaO₂ or hypoxemia.

Reduced oxygen content of the returning mixed venous blood to the lungs

As oxygen is delivered to the periphery, it is used for the generation of energy (ATP) by the citric acid cycle and oxidative phosphorylation. The extraction of oxygen from the arterial blood in the capillary beds varies from organ to organ and from tissue to tissue. The mixed venous blood that is then returned to the heart and lungs will have a reduced oxygen content and tension compared with arterial blood. As noted above, values for mixed venous blood are usually 75% of the hemoglobin saturated with oxygen and an oxygen tension of 40 mmHg.

Under conditions of exertion, the exercising muscles will extract a greater amount of oxygen for every ml of blood so that the mixed venous blood will have even lower values of oxygen content and tension. Exercise in individuals who are elite athletes or in patients with cardiovascular disease can result in mixed venous oxygen tensions as low as 18–20 mmHg with corresponding oxygen saturations as low as 15–20% (6). Any of the other causes of hypoxemia (Table 1) will be accentuated or worsened when the mixed venous oxygen content is reduced by increased extraction in the periphery.

CAUSES OF EXERTIONAL DESATURATION/HYPOXEMIA IN COPD

Although the primary cause of hypoxemia in individuals with COPD at rest is considered to be ventilation/perfusion mismatching, other causes (Table 1) may play more of role in hypoxemia or desaturation during exercise. Worsening ventilation-perfusion mismatching during exercise was originally considered the major pathophysiological mechanism for the development of exertional hypoxemia (7, 8). However, subsequent studies demonstrated a significant role for reduced mixed venous oxygen tension (PvO₂). Dantzker and D'Alonzo (9) studied pulmonary gas exchange in 7 patients with COPD during recumbent exercise on a cycle ergometer using the multiple inert gas technique. The mean FEV₁ was 0.56 l and the PaO₂ decreased from 76 mm Hg at rest to 63 mm Hg with maximal exercise. The PaCO₂ rose from 56 to 62 mm Hg and the PvO₂ decreased from 38 to 32 mm Hg. Because there was no significant change in VA/Q, the decline in arterial oxygenation was due to the decline in the PvO2 and the limited ventilatory response to exercise. Wagner and coworkers (4) demonstrated similar findings and concluded that exercise-induced desaturation in individuals with COPD is due to the decline in PvO₂ and not increasing VA/Q mismatching or diffusion-type limitation.

OXYGEN SATURATION VARIES WITH DIFFERENT FORMS OF EXERTION

The form of exertion may affect the detection of exercise-induced desaturation in individuals with COPD. Poulain and colleagues (10) measured oxygen saturation during 6 minute walk testing compared with maximal cardiopulmonary exercise testing (CPET) in 80 individuals with COPD (mean FEV₁ 62.4% of predicted). Desaturation (> 4% decrease in SpO₂ for the last 3 min of exercise) occurred in 21% during both forms of exercise but in 29% only during the 6-minute walk and not during CPET. Other studies have shown more profound reductions in PaO₂ during walking than during cycling in individuals with COPD (11, 12). In contrast, Mathur and coworkers (13) did not find a significant difference in peak oxygen consumption during cycle and treadmill exercise in patients with severe COPD (mean FEV₁ 0.69 l). Thus, the form of exercise may affect the ability to detect exertional desaturation in individuals with COPD.

FREQUENCY OF EXERTIONAL DESATURATION IN INDIVIDUALS WITH COPD

The prevalence of exertional desaturation in patients with COPD is unknown. In a group of 30 patients with moderate to severe COPD (mean FEV₁ 37% of predicted and mean PaO₂ 68 mm Hg), Schenkel and colleagues (14) measured oxygen saturation continuously. Oxygen desaturation was defined as a ≥4% decrease in the SaO₂ followed by an increase of at least 4%. The number of desaturations was greater during walking, 13.1 desaturations per hour, washing, 12.6 desaturations per hour, and eating, 9.2 desaturations per hour than at rest, 5.3 desaturations per hour. The mean SaO₂ at night and during the day were similar, 88% and 89% respectively, but the number of desaturations was greater during the day, 8.6 desaturations per hour, than at night, 6.8 desaturations per hour. In a study of 88 patients with stable COPD (mean FEV₁/FVC 38 ± 13%) and mild hypoxemia (mean PaO₂ 63.7 ± 2.8 mm Hg) who underwent 24-hour ambulatory monitoring, 33 patients (38%) were desaturators (SaO₂ < 90% for ≥ 30% of the recorded time) (15). The nocturnal SaO₂ was < 90% for > 30% of the night time in all the desaturators, whereas only 59% of the desaturators' SaO₂ was < 90% for > 30% of the daytime. Desaturators were significantly more hypoxemic and hypercarbic than non-desaturators.

Pilling and Cutaia (16) measured oxygen saturation continuously during normal activities of daily living in a group of 27 patients with COPD (mean FEV₁ 0.93 l) who were receiving supplemental oxygen. Desaturations were defined as a SpO₂ < 90% for ≥ 30 seconds. The mean SpO₂ was 92.0 ± 0.5% and the average minimal SpO₂ was 73.9 ± 1.2%. These patients experienced a mean of 0.9 ± 0.1 desaturations per hour and the SpO2 was < 90% for 24.6 ± 3.8% of the recorded time. Similar results were demonstrated in a group of 34 patients with COPD, resting PaO₂ ≤55 mmHg, who were treated with supplemental oxygen to raise the PaO₂ to > 65 mmHg at rest (17). Each patient experienced an average of 10 desaturations (SaO₂ < 90% for at least 5 minutes during which the minimal SaO2 was ≤85%) during 24 hours of measurement. The SaO₂ was ≤90% for 30 ± 27% of the daytime recording during which time the patients used their supplemental oxygen for 8.7 ± 3.2 hours. The mean SaO₂ was 92 ± 3.2% and the minimal SaO2 was 59 ± 13.8%. Thus, even during treatment with supplemental oxygen, exertional desaturation occurs frequently in individuals with COPD.

PREDICTING EXERTIONAL DESATURATION IN PATIENTS WITH COPD

Although numerous studies have attempted to correlate various clinical tests with exertional desaturation in patients with COPD, no test has been shown to have significant positive predictive value (Table 2).

Table 2.  Predicting Exertional Desaturation

Study	Correlation with Exertional Desaturation in Patients with COPD
Histopathology
1. Emphysema Severity Score	1. Significant positive correlation
2. Bronchiolar Score	2. No correlation
Chest CT Imaging
Emphysema severity	Mixed results
Pulmonary Function Tests
1. FEV1	1. Mixed results
2. FVC	2. No correlation
3. TLC	3. No correlation
4. DLCO	4. Negative predictive value
Resting oxygen saturation	Mixed Results: high normal value has negative predictive value

*Negative Predictive Value: when the value is normal, it is less likely that exertional desaturation will occur.

Histopathology and imaging

In a study correlating lung histopathology and exercise gas exchange in patients with COPD, Barbera and colleagues (18) performed progressive exercise testing and pulmonary function testing in 17 patients undergoing surgical resection of a lung cancer. The mean FEV₁ was 2.4 l and the PaO₂ increased from 81 mmHg at rest to 86 mm Hg during exercise. The airways and lung parenchyma of the resected lung specimens were quantitatively analyzed. The emphysema severity score correlated with a lower PaO₂ at rest and with exertion whereas the bronchiolar score was only associated with the level of oxygenation at rest.

Although one study has not demonstrated a significant correlation between exertional desaturation and the degree of emphysema measured by quantitative chest computed tomography (CT) (19), Taguchi and coworkers (20) showed a significant correlation between the severity of emphysema measured by chest CT and the minimal oxygen saturation and the change in oxygen saturation in a series of 32 patients who underwent 6-minute walk testing.

Pulmonary function tests

Numerous studies have attempted to correlate exertional desaturation with pulmonary function testing in patients with COPD and have not found a consistent relationship. Arterial blood gases were measured during exercise testing on a cycle ergometer using a gradually increasing work load in a group of 48 patients with COPD and resting oxygen tension > 55 mmHg (21). Exercise testing was stopped due to dyspnea or desaturation to < 75%. All saturation measurements were normalized for the change in oxygen consumption during exercise and desaturation was defined as a decrease of > 3%/ml/min.

The FEV₁ and DLCO (0.89 l and 7.1 ml/min/mmHg, respectively) were significantly lower in the 19 patients with desaturation compared with the 29 individuals who did not desaturate (1.44 l and 15.3 ml/min/mmHg, respectively) whereas FVC, FRC, TLC, and RV were not significantly different in the two groups. Desaturation did not develop in any patient with a DLCO > 55% of predicted but approximately 75% of patients with a DLCO or FEV₁ < 35% of predicted desaturated during graded exercise. Nordenfelt and Svensson (22) suggested that a transfer factor < 50% of predicted correlated with desaturation (PaO₂ < 8–8.5 kPa (60.0–63.8 mm Hg)) during maximal exercise in a group of 30 patients with COPD, restrictive lung disease, and mixed disorders. However, only 8 patients had COPD and only one of these had a transfer factor < 50%.

Other studies suggest that diffusion-type limitation may also contribute significantly to hypoxemia in patients with COPD during exercise (21–26). Patients who desaturate with exertion typically have a low diffusing capacity whereas those with a normal diffusing capacity usually do not desaturate with exercise. In a study of 40 patients with COPD, Ries and colleagues (27) were unable to predict which patients desaturated with exertion based upon their pulmonary function studies but determined that FEV₁/FVC > 0.5 and DLCO > 20 ml/min/mm Hg excluded those patients with exertional desaturation (decrease in PaO₂ > 3 mmHg during incremental cycle ergometry).

Similarly, Owens and colleagues (21) demonstrated that a diffusing capacity > 55% of predicted was 100% predictive in excluding desaturation during exertion. Mohsenifar and colleagues (28) determined the relationship between diffusing capacity and resting and exertional oxygenation in 1071 patients undergoing evaluation for the National Emphysema Treatment Trial. The mean (± SD) FEV₁ was 0.76 + 0.2 L (26.6 ± 7.2% of predicted) and the DLCO was 8.0 ± 3.1 mL/min/mm Hg (28.4 ± 9.8% of predicted). Quintiles of DLCO correlated with the resting PaO₂ after adjusting for age and gender and with the need for supplemental oxygen to maintain the SaO₂ ≥ 90% for at least 3 minutes while walking on a level treadmill at 1 mile per hour for 6 minutes. Only 38% of those with a DLCO > 35% required supplemental oxygen during this treadmill walk compared with 84% of patients with a DLCO ≤20%. These studies suggest that a diffusing capacity > 50–55% of predicted prognosticates maintenance of normoxemia during exercise in individuals with COPD but that a reduced DLCO is not a specific predictor of exertional desaturation.

In a large study of approximately 8,000 patients with various lung diseases undergoing pulmonary function studies and submaximal step test exercise with pulse oximetry, nearly 75% of the patients had obstructive lung disease (defined as both FEV₁/FVC and FEV₁ below the 95% confidence interval limit of normal) (25). Desaturation was defined as a ≥ 4% decrease in SpO2. In the group of patients with obstruction and a decreased DLCO (below the 95% confidence interval limit of normal), 5.4% desaturated with exercise whereas 12.8% of patients with obstruction and a normal DLCO desaturated. The sensitivity and specificity for predicting exercise-induced desaturation were approximately 75% when the DLCO threshold was set at 59.7% of predicted.

Resting oxygen saturation

Knower and colleagues (29) measured oxygen saturation continuously in 81 patients with COPD and FEV₁/FVC ≤70% during a 6-minute walk. Approximately half of patients (19/37) with a resting SaO₂ ≤ 95% experienced desaturation (defined as a decrease in SaO2 of ≥ 4% or a minimal SaO₂ of ≤88%) whereas only 16% (7/44) with a resting SaO₂ > 95% desaturated with exertion. All patients whose SaO₂ declined 4% or more also reached a minimal SaO₂ ≤88%. Receiver operating characteristic curve analysis determined that a resting SaO₂ of 95% produced the optimal sensitivity (73%) and negative predictive value (84%) for exertional desaturation. In the subgroup of 70 patients with DLCO measurements, no patients with DCLO > 36% and resting SaO₂ > 95% desaturated during exercise.

Fussell and coworkers (30) compared oxygenation at rest, during a 6 minute walk, and during continuous ambulatory monitoring in a group of 20 patients with stable COPD and mild hypoxemia (mean FEV₁1.16 ± 0.7 l and mean SpO₂ 94.9 ± 2.8%). Sixteen patients desaturated to ≤88% during the 6 minute walk and the mean minimal SpO₂ was 83.4 ± 7.0%. However, during ambulatory monitoring, the SpO₂ was < 88% for > 10% of the time in only 3 patients. There was no significant correlation between resting oxygen saturation and either the mean ambulatory saturation or the time with SpO₂ < 88%; further, the minimal oxygen saturation during the 6 minute walk did not correlate significantly with either the mean ambulatory saturation or the time with SpO₂ < 88%. Resting oxygen saturation is therefore a poor predictor of minimal saturation during ambulatory monitoring or 6-minute walk in patients with COPD and mild hypoxemia.

In a study of 70 patients with asthma and COPD undergoing 6 minute walk testing, Mak and coworkers (31) showed that the mean oxygen saturation correlated best with the diffusing capacity, FEV₁/FVC, and peak expiratory flow but not with the distance walked, perceived exertion, or perceived breathlessness.

Thus, pulmonary function tests especially diffusing capacity, predict which patients with COPD will maintain PaO₂ during exercise but have poor positive predictive value in determining exertional desaturation. Neither resting oxygen saturation nor pulmonary function studies can reliably predict which patients with COPD and resting mild hypoxemia will develop exertional desaturation.

WHAT ARE THE CONSEQUENCES OF EXERTIONAL DESATURATION?

Desaturation with exertion appears to portend a poorer prognosis for patients with COPD (32–35). During a 4-year period after step exercise testing, 19 of 50 patients with COPD died (32). The mean resting PaO₂ was greater in survivors than in those who died, 75.5 ± 3.7 and 63.6 ± 5.8 mmHg, respectively, and the PaO₂ declined to 56.9 ± 7.2 in the nonsurvivors but did not change in the survivors. In a nearly 10 year study of 58 individuals with COPD (mean FEV₁ 1.29 l) with resting normoxemia who underwent symptom limited ramp exercise testing on a cycle ergometer, Tojo and colleagues (33) showed that the survival rate for those patients with PaO₂max < 60 mmHg was significantly less than that of those with PaO₂max ≥ 60 mmHg.

In a retrospective study of 144 patients undergoing pulmonary rehabilitation who were followed for a mean duration of 8.4 years, oxygen desaturation > 6% during 6 minute walk testing was a significant predictor of mortality (34). In a 3-year longitudinal study of 576 individuals with COPD (mean FEV₁ 1.17 L), those who desaturated during 6 minute walk testing had a higher mortality, relative risk 2.63 (35). However, exertional desaturation was a less powerful predictor of mortality than resting PaO₂.

The PaO₂ slope (delta PaO₂/delta VO₂) during incremental cardiopulmonary exercise testing predicts prognosis in patient with COPD (33, 36, 37). Age and PaO₂ slope were the two most significant independent prognostic factors associated with survival time in a group of 120 patients with COPD. In a cohort of 64 patients with hypercapnea followed for up to 15 years, the decline in SaO₂ during exercise was significantly greater in those who died (38). Although exertional desaturation in patients with COPD appears to predict a poor prognosis, there is no evidence that maintenance of normoxemia during exercise improves the survival of these patients.

Hemodynamic studies in patients with COPD who are receiving supplemental oxygen and have their oxygen temporarily discontinued provide potential insight into the pathophysiological consequences of desaturation. Hemodynamic monitoring was performed before and after temporary oxygen discontinuation at rest and with exertion in twenty patients with COPD, mean FEV₁ 60 ± 2 % and PaO₂ 51 ± 2 mmHg, who were receiving 1–2 lpm supplemental oxygen (39). The mean PaO₂ decreased from 74 ± 4 to 55 ± 2 mm Hg with discontinuation of supplemental oxygen. The pulmonary vascular resistance index increased by approximately one third at rest and with exercise. The pulmonary artery pressure significantly increased at rest.

With exercise, the mean pulmonary artery pressure nearly doubled both with and without supplemental oxygen. Discontinuation of oxygen reduced arterial and mixed venous oxygen levels, oxygen delivery, and stroke volume index but did not affect systemic arterial pressure or vascular resistance at rest or with exercise. The pulmonary vascular changes reached steady state levels approximately 120–180 minutes after the discontinuation of oxygen. Thus, desaturation due to interruption of oxygen therapy in patients with COPD causes elevations in pulmonary artery pressures and pulmonary vascular resistance that occur over 2–3 hours but has no effect on systemic blood pressure or resistance.

In patients with resting mild hypoxemia and exercise-induced desaturation, the pulmonary artery pressure increases as oxygenation declines (40). In 17 patients with COPD (mean FEV₁ 1.0 ± 0.3 l, resting PaO₂ 10.6 ± 1.1 kPa (79.5 ± 8.3 mmHg)), the mean pulmonary artery pressure increased from 19.9 ± 4.5 at rest to 35.0 ± 2.2 mmHg with exercise equivalent to activities of daily living and the mean PaO₂ declined to 9.7 ± 0.4 kPa (72.8 ± 3.0 mmHg). Two thirds of the patients developed significant pulmonary hypertension, Ppa > 30 mmHg. However, the development of significant pulmonary hypertension was not always accompanied by hypoxemia and hypoxemia was not consistently associated with elevated pulmonary artery pressures. Although these findings suggest that desaturation during exercise may not be a significant contributing factor to the development of pulmonary hypertension in patients with COPD and mild resting hypoxemia, the effect of chronic intermittent exertional desaturation on pulmonary vascular hemodynamics is not as well studied as nocturnal desaturation.

WHAT ARE THE BENEFITS OF SUPPLEMENTAL OXYGEN TREATMENT OF EXERTIONAL DESATURATION IN COPD?

In a study of survival of 7,700 patients with COPD who were prescribed supplemental oxygen, 1,425 (18.5%) had a resting PaO₂ > 8 kPa (60 mmHg) and were presumed to have received oxygen for either nocturnal or exercise-related desaturation (41). The survival of this group of patients was reduced compared with a gender and age matched general population but similar to patients with COPD and resting hypoxemia, PaO₂ between 6.7–8 kPa (50–60 mmHg) who were also treated with supplemental oxygen.

In a randomized study of survival in patients with COPD and mild hypoxemia, Gorecka and colleagues (42) found no difference between patients treated with supplemental oxygen and untreated control patients who breathed room air. Drummond and colleagues (43) analyzed the 8-year survival of 471 participants in the National Emphysema Treatment Trial with resting normoxemia and exertional desaturation (SpO₂ < 90% during a 7-minute treadmill walk) who received medical management and found similar survival rates in those individuals treated with continuous, intermittent, or no oxygen.

Supplemental oxygen may enhance exercise performance in individuals with COPD who are normoxemic at rest but who desaturate with exertion (44, 45). Jolly and coworkers measured the 6-minute walk distance and level of dyspnea in 11 patients with COPD (mean FEV₁ 0.9 l) with a resting SaO2 of 94.7% and a mean minimal exertional SaO2 of 84.8% (45). The level of dyspnea decreased and the distance walked increased with supplemental oxygen compared with room air whereas neither measurement changed while breathing compressed air. In a double-blind trial comparing supplemental oxygen with compressed air during 7 weeks of high intensity cycle ergometer exercise in 29 patients with COPD and resting normoxemia, Emtner and colleagues (46) showed that the training work rate increased more rapidly in the group treated with oxygen. This group also achieved a higher maximal work load and greater endurance.

In contrast, other studies have not demonstrated significant improvements in the exercise capacity of patients with exertional desaturation treated with supplemental oxygen. A single blind trial comparing oxygen with compressed air in 20 patients undergoing 8 weeks of treadmill exercise training did not demonstrate significant differences in the improvement in the distance walked in 6 minutes between the two groups (47).

These investigators concluded that “training of patients with COPD and exercise-induced hypoxaemia can be conducted without supplemental oxygen.” However, the joint ACCP/AACVPR evidence-based clinical practice guidelines for pulmonary rehabilitation state there is a strong rationale to utilize supplemental oxygen during exercise training in individuals with exertional desaturation from a safety perspective (48). Thus, although oxygen does improve maximal exercise performance during acute testing, its augmentation of the exercise training effect and requirement during training are less certain (48).

Rooyackers and colleagues (49) compared supplemental oxygen with room air during 10 weeks of inpatient pulmonary rehabilitation in 24 patients with COPD and exertional desaturation to < 90%. Both groups improved exercise performance and quality of life but oxygen did not add to the benefit of rehabilitation while breathing room air. Throughout this study the work rate was adjusted to maintain the SaO2 > 90%. Another placebo controlled trial compared supplemental oxygen with compressed air during pulmonary rehabilitation in a group of 25 patients with nonhypoxemic COPD (mean FEV₁ 0.76 l), who desaturated with exertion (50).

Dyspnea was reduced in the group treated with oxygen but exercise tolerance, health status, mood state, and performance of daily activities were not different. Supplemental oxygen in concentrations up to 50% reduces respiratory rate and dynamic hyperinflation during exercise in patients with COPD and mild hypoxemia (51). A meta-analysis of oxygen supplementation during exercise training in patients with COPD concluded that oxygen did not augment the benefits of exercise training and that, because of the benefits of pulmonary rehabilitation, any exercise program, with or without oxygen supplementation was recommended (52).

Health-related quality of life is improved in patients with COPD and exertional desaturation treated with supplemental oxygen (53). In a group of 50 patients with COPD (mean FEV₁ % predicted 25.9 +/− 8.0, resting PaO₂9.2 +/− 1.0 kPa (69.0 ± 7.5 mm Hg), and SaO2 after 6-minute walk, 82 +/− 5.4%), Eaton and coworkers (53) evaluated the effect of supplemental oxygen on HRQL in a 12-week double-blind, randomized, crossover trial. Twenty-eight patients (68%) increased their distance in the 6-minute walk (≥54 m) or decreased their Borg dyspnea scale (≥1). Oxygen significantly improved all measures in the Chronic Respiratory Questionnaire, levels of anxiety and depression determined by the Hospital anxiety and depression scale, and general health, physical and emotional roles measured by the short-form 36 health survey questionnaire. The effect of oxygen on the 6 minute walk distance and the Borg dyspnea scale did not correlate with the survey responses, however. Interestingly, despite the significant measured benefit, 41% of the responders did not want to continue supplemental oxygen after the 12 week trial.

Jensen and colleagues (54) measured cerebral oxygenation using near-infrared spectrophotometry during incremental exercise on a cycle ergometer in 13 patients with terminal lung disease (7 had COPD or emphysema and 3 had alpha 1 anti-trypsin deficiency). The mean SaO₂ decreased from 91.6 +/− 1.5% to 87.3 +/− 1.7% at peak exercise while breathing room air but were significantly elevated with supplemental oxygen and did not change with exercise, 95.8 +/− 0.7% and 96.0 +/− 1.0, respectively. During exercise while breathing room air, cerebral oxygenation declined due to an increase in deoxyhemoglobin. In contrast, while breathing supplemental oxygen during exercise, deoxyhemoglobin declined and cerebral oxygenation improved. Thus, supplemental oxygen may prevent exercise-induced reduction in cerebral oxygenation and sustain cerebral function.

AS NEEDED OR “SHORT BURST OXYGEN THERAPY”

Intermittent use of oxygen for short periods before or immediately after exertion to relieve breathlessness (short burst oxygen) is frequently used in the United Kingdom (55–58). Several early studies suggested that the administration of oxygen for short periods immediately before or just after exertion reduced dyspnea and may increase 6 minute walk distance (59–61). Subsequent studies have concluded short duration treatment with oxygen prior to or immediately after exertion does not reduce breathlessness (62–64).

Short-burst oxygen supplementation either before or after a 6 minute walk does not improve the distance walked or the Borg dyspnea scale in individuals with COPD and normoxemia at rest and desaturation with exertion (65). In a randomized, single blind, crossover study, Stevenson and Calverley (66) compared the effect of oxygen (FiO2 0.4) with air for 15 minutes during recovery after incremental cycle ergometer exercise in a group of 18 patients with COPD (mean FEV₁% predicted 40 ± 16 %).

The resting SaO₂ was 95.9 ± 1.66% and six patients desaturated during exercise to a minimal SaO₂ of 88%. Breathing supplemental oxygen after the completion of exercise decreased the time for recovery of dynamic hyperinflation compared with air by 6.61 ± 1.65 minutes. Oxygen did not effect the time to return to baseline breathlessness or maximal perception of dyspnea during recovery. A meta-analysis of short-burst oxygen therapy in chronic obstructive pulmonary disease (67) found that short-burst oxygen therapy does not reduce breathlessness and that its effect on other outcome measures such as exercise capacity, oxygen saturation, and other ventilatory parameters was not consistent. Differences in patient characteristics, study protocols, amount, duration, and method of oxygen delivery make assessment of studies of short burst oxygen difficult and clinical applicability of these findings is uncertain.

WHAT ARE THE UNTOWARD EFFECTS OF SUPPLEMENTAL OXYGEN?

Levels of 8-isoprostane and IL-6 within the exhaled breath condensate of 18 patients with COPD increased after breathing 28% oxygen for 1 hour suggesting that supplemental oxygen may cause oxidative stress and inflammation within the lower airway (68). The mean FEV₁ % predicted was 54.1 +/− 3.1 and the mean PaO₂ was 7.2 +/− 0.4 kPa (54.0 ± 3.0 mmHg). Seven patients were treated with supplemental oxygen for 2–5 hours daily chronically.

Fires caused by the ignition of oxygen can cause severe facial burns. These fires are usually ignited by lit cigarettes but stoves and even cellular phones may be the cause (69–72). In a retrospective review of 3673 adults admitted to a burn center over 10 years, 27 patients had burns directly attributable to supplemental oxygen (73). Between 2000 and 2007, four states (Maine, Massachusetts, New Hampshire, and Oklahoma) recorded 38 deaths related to fires associated with supplemental oxygen (74). The overall fatality rate was 3.8 deaths per 10 million population per year. Thirty-four (89%) of the decedents were using supplemental oxygen and smoking at the time of the fire.

CONCLUSION

Exertional desaturation occurs frequently in patients with COPD and in patients with COPD who are treated with supplemental oxygen. Desaturation with exertion does appear to predict an increased risk of mortality in patients with COPD. Although neither pulmonary function studies nor resting SaO₂ predict exertional desaturation with high sensitivity and specificity, preservation of the diffusing capacity appears to correlate with maintenance of normoxemia during exertion.

Supplemental oxygen does not reduce mortality in patients with COPD and mild resting hypoxemia and it is not known if oxygen treatment improves the longevity of patients with exertional desaturation. Use of oxygen has variable and inconsistent effects on exercise capacity, sensation of breathlessness, and quality of life in individuals with COPD and exertional desaturation. Differences in the definition of desaturation, mode of exercise, and characteristics of the patient population make it difficult to compare these studies and apply them to clinical practice. There is a need for the development of a uniform definition of exertional desaturation and standardized exercise protocol to elicit decreases in oxygen levels in individuals with COPD.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.