Pulmonary Pitfalls

Using and Interpreting Carbon Monoxide Diffusing Capacity (Dlco) Correctly

Lam-Phuong Nguyen, DO, Richart W. Harper, MD, and Samuel Louie, MD

Lam-Phuong Nguyen, DO; Richart W. Harper, MD; Samuel Louie, MD
UC Davis Medical Center, Sacramento, California.

Nguyen LP, Harper RW, Louie S. Using and interpreting carbon monoxide diffusing capacity (Dlco) correctly. Consultant. 2016;56(5):440-445.


Carbon monoxide diffusing capacity (Dlco) probably is the least understood pulmonary function test (PFT) in clinical practice worldwide, even among experienced pulmonologists. Every clinician knows that Dlco measures the quantity of carbon monoxide (CO) transferred per minute from alveolar gas to red blood cells (specifically hemoglobin) in pulmonary capillaries, and that this value, expressed as mL/min/mm Hg, represents mL of CO transferred per minute for each mm Hg of pressure difference across the total available functioning lung gas exchange surface.1 But has anyone stopped to ask why Dlco measurement is ordered, how it is determined, and what it means when it is reduced or not?

What Is Dlco?

Dlco is a calculated, derived value that indirectly assesses the ability of the lungs to “transfer” oxygen to blood through the use of a test gas (namely, CO) that has a greater affinity for blood hemoglobin. When Dlco is below the predicted reference range (75% to 140% of predicted) it becomes a clue to the presence of a physiologic problem that ultimately may impair exercise, and even affect long-term survival from common lung diseases and disorders.

The specificity and sensitivity of Dlco for specific lung diseases has not been studied extensively until recently, particularly for pulmonary arterial hypertension (PAH) and systemic sclerosis with or without interstitial lung disease (ILD).2 Both PAH and ILD can reduce Dlco, the former by reducing capillary blood volume and the latter by causing fibrosis of the delicate interface necessary for gas diffusion between alveolar air and capillary blood. Alone, Dlco is not enough to confirm the presence of or differentiate between the 2 lung conditions. Additionally, Dlco may predict mortality in a variety of lung diseases (including cancer), various ILDs (including idiopathic pulmonary fibrosis), and severe PAH.

Dyspnea is the most common reason for ordering a Dlco test, but there are many situations and presentations in which a higher than predicted or lower than predicted Dlco suggests the possible presence of lung or heart disease (Table 1). Respiratory tract symptoms and abnormalities on chest radiographs and/or chest computed tomography (CT) scans are essential to properly interpret any PFT, including Dlco.


Table 1


Overlooking a reduced Dlco can delay early diagnosis and treatment of a disease. For example, Dlco is low in chronic obstructive pulmonary disease (COPD) with emphysema, or amiodarone lung toxicity, and it is even lower in ILD with PAH. While Dlco serves as a surrogate marker of the available lung surface area and its properties that enable diffusion to take place, blood in the capillaries—or more accurately, unbound hemoglobin—is the essential driver in the diffusion of CO from the alveolar air across the alveolar-capillary membrane barrier into hemoglobin in red blood cells. Remember, blood in the airways also can bind CO, hence Dlco can rise with hemoptysis and pulmonary hemorrhage.

Inhaled CO is used because of its very high affinity for hemoglobin. CO has a 200 to 250 times greater affinity for hemoglobin than does oxygen. Because anemia can lower Dlco, all calculations of Dlco are adjusted for hemoglobin concentration to standardize measurements and interpretation.1 In the PFT laboratory, a very small amount of CO (0.3% of the total test and room air gases) is inhaled by the patient during the test, and the level is not dangerous—CO poisoning with tissue hypoxemia does not occur with the Dlco measurement. But a cornucopia of lung disorders that disturb oxygen uptake by hemoglobin in the lungs (and increase the work of breathing, perceived as dyspnea) can be detected by a reduction in Dlco.

How the reduction in Dlco is interpreted can influence clinical decisions in patients with unexplained dyspnea or dyspnea that fails to improve with initial treatments such as bronchodilators. Variability in how Dlco is reported is a concern. The unfortunate adoption of certain nomenclature, primarily Dlco/Va (where Va is alveolar volume) can cause confusion on how Dlco assessment is best applied in clinical practice.

How Is Dlco Determined?

Simply put, Dlco is the product of 2 primary measurements, the surface area of the lung available for gas exchange (Va) and the rate of alveolar capillary blood CO uptake (Kco).1,3 An understanding of how these 2 variables are determined provides important insight into the clinical implications of Dlco.

Several techniques are available to measure Dlco, but the single breath-hold technique is most often employed in PFT laboratories. The patient breathes through a mouthpiece with nose clips in place to acclimate to the equipment, followed by unforced exhalation to residual volume (RV). From RV, the patient rapidly inhales test gases (typically 0.3% CO combined with either helium or methane, mixed in remaining portions of room air) to total lung capacity (TLC) and holds his or her breath for 10 seconds. The patient then is asked to perform an unforced, complete exhalation in less than 4 seconds.

After elimination of estimated dead-space exhaled breath, a volume of exhaled breath is sampled to measure test gas concentrations (Figure). Spirometry is performed simultaneously with measurement of test gas concentrations in order to calculate Va and Kco to derive Dlco, which then is adjusted for hemoglobin concentration.

figure 1

Figure. Single breath-holding technique: The patient takes a deep breath from residual volume to total lung capacity, inhaling a known concentration of helium and carbon monoxide, and holds this for 10 seconds. At the end of the breath hold, a sample of gas is collected for measurement of exhaled helium and carbon monoxide after discarding approximated dead-space ventilation. Abbreviations: COe, exhaled CO concentration; COi, inhaled CO concentration; COo, carbon monoxide concentration after redistribution during single breath hold; Hee, exhaled helium concentration; Hei, inhaled helium concentration; Heo, helium concentration after redistribution during single breath hold. Legend: Black line, spirometric measurement during single breath hold; #, discarded gas (approximated dead space); *, collected gas for He and CO measurements.


Va is calculated by a change in the concentration of an inhaled inert gas (such as helium or methane) after that gas has had an opportunity to mix throughout the lungs. Because an inert gas is used, it is reasonably assumed that a change in exhaled concentration from the inhaled concentration is purely due to redistribution (“dilution”) of the gas into a larger volume. This is because there is no loss of the gas through uptake by pulmonary tissues (as with oxygen) or into the capillary bed. Using helium as the inert gas, the concentration of the inhaled helium (Hei) would be known, and because the inhaled volume (Vi) is measured, measuring the concentration of exhaled helium (Hee) will give the volume of lungs exposed to helium, or Va, as follows:

Vi × Hei = Va × Hee

Vi is the volume of inhaled gas minus the estimated dead space (since dead space will not contain any helium). Rearranging this equation gives us a way to determine Va from carefully measured values of Vi, Hei, and Hee:

Va = Vi × Hei/Hee

Unlike TLC, Va is calculated from a single breath. Despite this, Va typically approximates TLC within a few percentage points (Va/TLC > 95%) in the normal lung. The Va/TLC ratio does not depend on age, sex, height, or weight but decreases when there is intrapulmonary airflow obstruction and/or uneven distribution of ventilation. A reduction in Va will reduce Dlco unless the rate of CO uptake or Kco increases.

As Marie Krogh first modeled in 1915, CO leaves the alveolar space at an exponential rate related to the gradient of CO between the alveolar compartment and the pulmonary capillary compartment. Therefore, the rate of CO uptake is calculated from the difference between the initial and final alveolar CO concentrations over the period of a single breath-hold (10 seconds). This rate, kco, which has units of seconds-1, is calculated as follows:

kco = loge(COo/COe)/t

COo is the initial alveolar concentration, COe is the alveolar concentration at the end of the breath hold, and t is the breath-hold time in seconds. COo cannot be directly measured, since we only know the inhaled CO concentration (COi) and the exhaled CO concentration (COe). However, CO on a single breath-hold will dilute proportionately with helium (Figure), so that immediately at the end of inhalation:

COo = COi(Hee/Hei)

Combining equations 3 and 4, we can determine kco by measuring inhaled and exhaled concentrations of helium (or methane) and CO.

Note that Dlco is not equivalent to Kco! Dlco is the product of Va and Kco, the rate of diffusion across a membrane that is dependent upon the partial pressure of the gas on each side of the alveolar membrane. Because CO in the pulmonary capillary compartment is usually close to zero, the partial pressure gradient of CO across the alveolar-capillary integrated interface, or “membrane,” is estimated to be partial pressure of CO in the alveolar compartment alone (or atmospheric pressure–water vapor pressure at 37°C). Therefore, Dlco is defined as follows:

Dlco = Va × kco/Pb, or 

Dlco = Va × Kco

Pb is atmospheric pressure–water vapor pressure at 37°C, and Kco is kco/Pb.

As shown above, Dlco is the product of a volume (determined by the dilution of helium) and a “decay” rate of CO over a specific breath-hold time for a given atmospheric pressure, all of which are derived from measured values of exhaled CO and helium (or other inert gas).

Confusion arises in how PFT laboratories, by convention, report Dlco and the related measurements Va and Dlco/Va. This has had the unintended consequence of many clinicians considering Dlco/Va to be the Dlco “corrected” for the Va, when it is actually Kco—a rate constant for CO uptake in the lung. Kco is not the lung CO diffusing capacity.

The term Dlco/Va is best avoided because Kco (the preferred term) is not derived from measurement of either Dlco or Va! Furthermore, Kco is not a surrogate measurement for Dlco. Whenever Dlco is reduced, the predominant reason for this reduction (eg, whether it is predominantly a reduced Va, or reduced Kco, or both) has critical diagnostic and pathophysiologic implications. The key questions that should be asked include: Is the reduction in Dlco due to a reduction in Va, Kco, or both?

A common pitfall when considering Dlco measurements is not appreciating the relationship between Va and Kco. As is made obvious in equation 5, reductions in either Va or Kco (aka, Dlco/Va) will result in a reduction in Dlco.

Dlco can be falsely reduced in patients with COPD or severe restrictive diseases in which the patient is unable to take in an adequate breath. The inspired CO under these circumstances may not completely reach all the functioning alveolar-capillary units.

Another common but underappreciated fact is that as lung volume falls from TLC to RV, Dlco does not fall as much as would be predicted based on the change in Va. The reason is that as the lung volume falls, Kco actually rises. This ensures that Dlco remains relatively constant at various volumes from tidal breathing to TLC.

The reason Kco increases with lower lung volumes in certain situations can best be understood by the diffusion law for gases. For a given gas, the rate of diffusion for this gas, Dl, is dependent upon the thickness of the diffusing membrane (DM, the alveolar-capillary membrane), the rate of uptake of a gas by red blood cells, Θ, and the pulmonary capillary blood volume, Vc. Specifically for CO, the rate of diffusion is as follows:

1/Dlco = 1/DMco + 1/Θco∙Vc

The values for DMco and Θco remain relatively constant in the normal lung at various inspired volumes, which indicates that a change in Vc is the predominant reason why Dlco does not fall directly in proportion to Va. At lower lung volumes, Kco increases, because more capillary blood volume is accessible to absorb CO.

Va/Kco Relationships in Disease States

Understanding the anatomic and pathologic processes that affect Va and Kco enables the clinician to properly interpret the significance and underlying mechanisms leading to a low Dlco.

For example, group 1 PAH, early pulmonary vasculitis, and pulmonary arteriovenous malformations may produce a lower than predicted Dlco primarily due to a reduction in Kco or due to reduced Vc, while Va remains relatively preserved (see equation 6). Conversely, obesity, kyphoscoliosis, and neuromuscular disease will reduce Va, but Kco, due to relatively increased Vc for a given Va, will be increased, resulting in a normal range or slightly decreased Dlco.

Similarly, disease states that result in loss of alveolar units, such as pneumonectomy, lobectomy, or lobar collapse as reflected by a low Va can reduce Dlco. Blood flow of lost alveolar units can be diverted to the remaining units, resulting in a slight increase in Kco, and as a result, Dlco falls relatively less than expected given the reduction in Va.

Emphysema or ILD can feature a loss of both Vc and Va, which can result in a more profound reduction in Dlco.

This understanding is particularly useful in clinical situations in which the expected values do not correlate clinically or with other PFTs such as TLC.

As mentioned, neuromuscular disease may demonstrate a Dlco in the “normal” range with a reduced Va and an elevated Kco (Dlco/Va) because of increased CO transfer to higher than normal perfused lung units (eg, the Va may be 69% predicted with a Kco of 140% predicted). However, in this same patient, if the Kco were 80% predicted (still in the normal range as an isolated value), the Dlco may become abnormally low due to a combination of low Va and “normal” Kco. In this situation, it would be incorrect to state that the Dlco “corrects” for Va, because the Kco should be much higher. Hence, seeing a low Kco would be a clue that the patient with neuromuscular disease has a concomitant disease or disorder that impairs gas exchange (ie, pulmonary fibrosis or pulmonary vascular disease) on top of the lower alveolar volume. This demonstrates that Dlco could be lowered by 2 different mechanisms in the same patient.

The American Thoracic Society/European Respiratory Society statement on PFT interpretation advocates the use of a Dlco percent predicted of 80% as the normal cutoff. It also indicates that 79% to 60% of predicted is a mild reduction, 59% to 40% is a moderate reduction, and that Dlco values less than 40% of predicted are severely reduced.1

The Fick law of diffusion can explain factors that influence the diffusion of gas across the alveolar-capillary barrier:

V = A × D × (P1 − P2)/T

V is volume of gas diffusing, A is surface area, D is the diffusion coefficient of gas, T is the thickness of the barrier, and P1− P2 is the partial pressure difference of gas across the alveolar-capillary barrier.

As one might postulate, a proportional decrease in Dlco would be expected if there were a reduction in lung volume and hence alveolar surface area, as seen in patients after pneumonectomy. However as noted, blood flow of lost alveolar units is diverted to the remaining units, resulting in a slight increase in Kco; as a result, Dlco falls relatively less than Va and not always proportionately.

In summary, a reduced Dlco is sensitive but not specific for:

  • Decreased volume of pulmonary capillary blood or hemoglobin volume
  • Decreased surface area integrated between capillaries and alveoli
  • Ventilation/perfusion mismatching or intrapulmonary shunting from atelectasis


Important Elements of Standardization

At the UC Davis Medical Center’s Pulmonary Services Laboratory, the Dlco measurement begins with a patient being asked to inhale from RV to TLC a test gas composed of 0.3% methane, 0.3% CO, 21% oxygen, and the remaining proportion nitrogen. Other institutions may use 10% helium as the tracer gas instead of methane.

The patient needs to hold his or her breath for 10 seconds, then exhale quickly and completely back to RV. The exhaled breath from alveolar lung volume is collected after the washout volume (representing anatomic dead space) and is discarded as described in the Figure.

A vital capacity (VC) of at least 1.5 L is required to perform the Dlco measurement with sufficient accuracy, because 0.75 to 1.0 L needs to be discarded as washout volume from dead space, and a Va sample of at least 500 mL must be available for calculating Dlco. If the patient’s VC is less than 2.0 L, it is recommended that the washout volume be reduced to 0.5 L. The averages of the 2 Dlco measurements must be within 10% of each other. It is recommended that no more than 5 tests be performed at a sitting. The uptake of CO can be calculated from the Va and inspired and expired CO concentrations.

In obstructive lung diseases. It is a common pitfall to “correct” Dlco for Va and thus misinterpret Dlco/Va that appears in the normal range in patients with obstructive lung diseases such as COPD and asthma-COPD overlap syndrome (ACOS), which can produce spuriously normal results, leading to errors in interpretation and decision-making. Dlco “correction” by Va cannot reliably rule out the presence of underlying emphysema or parenchymal lung disease.4

Dlco usually is decreased in COPD when emphysema is present; it typically is normal in chronic bronchitis alone or in asthma, where it even could be increased during acute attacks.5

An isolated low Dlco can suggest emphysema is present in the context of normal spirometry and lung volumes, but a normal Dlco cannot rule out emphysema, whereas a CT scan will.

In restrictive lung diseases and disorders. Dlco can be normal or slightly decreased in extrinsic restrictive disorders (underlying lung physiology is normal except for atelectasis) such as Guillain-Barré syndrome, myasthenia gravis, amyotrophic lateral sclerosis, and corticosteroid-induced myopathy, given a decrease in Va but a normal to elevated Kco (Dlco/Va). Chest wall disease, such as morbid obesity, pleural effusions, and kyphoscoliosis, can display a normal Dlco or a slightly decreased Dlco, but the Dlco/Va remains normal.

Intrinsic restrictive lung diseases such as ILD (specifically pulmonary fibrosis from collagen vascular disorders and sarcoidosis) commonly have a reduced Dlco. A reduced Dlco (primarily from reduction in Kco) is a useful tool for detecting early ILD before lung volumes become decreased, for detecting pulmonary vascular diseases from venous thromboembolism or PAH, and for monitoring response to therapy and disease progression. Although it is nonspecific, a reduced Dlco requires an adequate explanation in every case.

Dlco is not very helpful in differentiating among the causes of ILD, but it can be helpful in suggesting the diagnosis and other conditions (eg, emphysema, PAH) in patients with unexplained dyspnea, in assessing disease severity, and in predicting prognosis (eg, a severely decreased Dlco in nonspecific interstitial pneumonitis and idiopathic pulmonary fibrosis augurs a very poor prognosis).

In drug-induced lung diseases. Dlco is helpful in detecting drug-induced lung disease. For example, chronic interstitial pneumonitis is the most common form of amiodarone-induced lung disease and usually is recognized after 2 or more months of therapy where the daily dose exceeds 400 mg. The diagnosis should be suspected in a patient taking amiodarone with nonproductive cough, dyspnea, and weight loss accompanied by an abnormal chest radiographs demonstrating chronic interstitial lung changes. The prevalence is approximately 5%, and the condition may improve when amiodarone is stopped, with or without adding systemic corticosteroids.

The presence of the following suggests the diagnosis of amiodarone-induced lung disease: new or worsening symptoms or signs; new abnormalities on chest radiographs; and a decline in TLC of 15% or more, or a decline in Dlco of more than 20%.

Other drugs that can cause lung diseases include amphotericin, methotrexate, cyclophosphamide, nitrofurantoin, cocaine, bleomycin, tetracycline, and many of the newer biologics.

Another striking example of where Dlco is helpful are cases of difficult-to-control young adult asthmatic women with normal spirometry and lung function who subsequently are diagnosed with PAH secondary to dieting pills or methamphetamines. The diagnosis often is made after an unexpectedly reduced Dlco prompts a search for the reasons.

Similarly, it is important to recognize the conditions that most frequently are associated with an elevated or high Dlco (ie, greater than 140% predicted)—namely asthma, obesity, or both and, uncommonly, polycythemia and left-to-right shunts.6 Any condition that typically reduces Dlco, such as emphysema, pulmonary vascular disease, or cancer, can deceptively bring supranormal Dlco into the normal range.

A checklist can be helpful in establishing a regular routine for interpreting Dlco, Va and Kco (Tables 2 and 3).

Table 2

Table 3



Additional take-home messages

  • Anemia, COPD with emphysema, ILD, and pulmonary vascular diseases can decrease Dlco below the normal range.
  • Asthma, obesity, and less commonly polycythemia, congestive heart failure, pregnancy, atrial septal defect, and hemoptysis or pulmonary hemorrhage can increase Dlco above the normal range.
  • A reduced Dlco also can accompany drug-induced lung diseases.
  • A decreasing Dlco is superior to following changes in slow vital capacity (SVC) or TLC in ILDs.
  • Dlco is a specific but insensitive predictor of abnormal gas exchange during exercise. Low Dlco less than or equal to 50% predicted can predict hypoxemia with exercise. A normal Dlco does not rule out oxygen desaturation with exercise.
  • A decrease in Dlco in persons with HIV independently predicts the development of opportunistic pneumonia or pneumocystis pneumonia and is due to loss of capillary blood volume with regional air-trapping or early emphysema.7
  • Reduced Dlco in the context of normal spirometry, lung volumes, and chest radiographs suggests underlying lung disease such as ILD, emphysema, or PAH.
  • In the setting of a normal chest radiograph, early ILD or pulmonary vascular disease or both can be present. This observation underscores the need for chest CT for confirming the diagnosis of ILD.
  • A Dlco below 30% predicted is required by Social Security for total disability.
  • Routine reporting of Dlco “corrected to normal with Va” without fully understanding the implications is misleading and can cause clinicians to lose their clinical index of suspicion and underdiagnose diseases when in fact Dlco still is abnormal.


The bottom line is that a reduced Dlco is not normal, requires explanation, and may indicate the presence of clinically significant lung disease or pulmonary vascular disease. A Dlco within the normal range (75% to 140% predicted) cannot completely rule out lung disease when the patient is persistently and genuinely dyspneic.


Pitfalls to avoid


Lam-Phuong Nguyen, DO, is chief fellow in the Division of Pulmonary, Critical Care, and Sleep Medicine in the Department of Internal Medicine at UC Davis Medical Center in Sacramento, California.

Richart W. Harper, MD, is a professor of medicine in the Division of Pulmonary, Critical Care, and Sleep Medicine at UC Davis Medical Center.

Samuel Louie, MD, is a professor of medicine in the Division of Pulmonary, Critical Care, and Sleep Medicine at UC Davis Medical Center.



  1. MacIntyre N, Crapo RO, Viegi G, et al. Standardization of the single-breath determination of carbon monoxide uptake in the lung. Eur Respir J. 2005;26(4):720-735.
  2. Sivova N, Launay D, Wémeau-Stervinou L, et al. Relevance of partitioning DLCO to detect pulmonary hypertension in systemic sclerosis. PLoS ONE. 2013;8(10):78001.
  3. Hughes JMB, Pride NB. Examination of the carbon monoxide diffusing capacity (DlCO) in relation to its Kco and Va components. Am J Respir Crit Care Med. 2012;186(2):132-139.
  4. Uvieghara AO, Lanza J, Vasudevan VP, Arjomand F. Volume correction for diffusion capacity: use of total lung capacity by either nitrogen washout or body plethymography instead of alveolar volume by single breath methane dilution. Poster presented at: American Thoracic Society 2010 International Conference; May 14-19, 2010; New Orleans, LA. Accessed April 11, 2016.
  5. Salzman SH. Which pulmonary function tests best differentiate between COPD phenotypes? Respir Care. 2012;57(1):50-57.
  6. Saydain G, Beck KC, Decker PA, Cowl CT, Scanlon PD. Clinical significance of elevated diffusing capacity. Chest. 2004;125(2):446-452.
  7. Diaz PT, King MA, Pacht, ER et al. The pathophysiology of pulmonary diffusion impairment in human immunodeficiency virus infection. Am J Respir Crit Care Med. 1999;160(1):272-277.