Abstract
The carbon monoxide transfer factor (TL,CO) is the product of the two primary measurements during breath-holding, the CO transfer coefficient (KCO) and the alveolar volume (VA). KCO is essentially the rate constant for alveolar CO uptake (Krogh's kCO), and in healthy subjects, increases when VA is reduced by submaximal inflation, or when pulmonary blood flow increases. Recently, new reference values were proposed for clinical use which included the observed VA at full inflation; this was claimed to “eliminate the need for KCO”.
In this commentary, some mechanisms e.g. respiratory muscle weakness, lung resection, diffuse alveolar damage and airflow obstruction, which decrease or increase total lung capacity (TLC) are reviewed.
Even when alveolar structure and function are normal, the change in KCO at a given VA varies according to the underlying pathophysiological mechanism. The advantages and disadvantages of normalizing KCO and TL,CO to predisease predicted TLC or to the patient's actual VA (using lack of expansion or loss of alveolar units models) are considered.
Examination of carbon monoxide transfer coefficient and alveolar volume separately provides information on disease pathophysiology which cannot be obtained from their product, the carbon monoxide transfer factor.
A few years ago, a paper in the European Respiratory Journal 1 concluded that: “… the use of TL/VA (the carbon monoxide (CO) transfer coefficient) cannot be justified on scientific grounds”. Apart from one letter of disagreement 2, this view that TL/VA (or KCO) is a redundant and misleading measurement has not been challenged. This is surprising because measurements of TL/VA have continued to be published in respiratory journals.
The single breath method for measuring CO uptake by the lung, which is used world-wide, was introduced by Krogh 3 in 1915; this measurement was termed diffusion constant. Subsequently the diffusion constant for CO was renamed the diffusing capacity (DL,CO) or the transfer factor (TL,CO), with the uptake being measured at total lung capacity (TLC). Krogh 3 pointed out that TL,CO was the product of two separate measurements, which potentially varied widely (and independently), the rate constant for CO removal from alveolar gas (called the permeability factor (kCO)) and the alveolar volume (VA).
kCO is measured as the exponential decay in fractional
concentration of CO over a period of breath-holding (BHT):where
CO0 and COt are the alveolar CO concentrations at the
start and finish of the breath-holding period. The units of kCO are s−1 or min−1.
The total CO transfer of the lung is calculated as:where PB
and PH2O are the barometric pressure and the
water vapour pressure (at 37°C) which standardize for the driving
pressure for CO uptake, i.e.the pressure of CO in the alveoli (PA,CO). VA is the alveolar volume
measured at standard temperature and pressure, dry (STPD).
In modern usage, M. Krogh's kCO is rarely employed; instead, the carbon monoxide transfer coefficient is substituted, whose units of mmol·min−1·kPa−1·L−1 (at body temperature and ambient pressure, and saturated with water vapour (BTPS)) give the appearance of being a ratio, an impression enhanced by its terminology (TL/VA or DL/VA). In fact, kCO converts to the carbon monoxide transfer coefficient by dividing by the STPD to BTPS conversion (1.2), by a L to mmol change (1,000/22.4) (if in SI units), and by the barometric pressure term (PB−PH2O). In SI units, kCO (min−1) converts to KCO (TL/VA) by dividing by 2.56.
The objection of Chinn et al. 1 to the use of TL/VA is that “VA was the largest single contributor to the variance in TL/VA”; unfortunately, this gives the misleading impression that TL/VA is derived from TL,CO by dividing TL by VA, whereas TL/VA and VA are the two primary measurements used to obtain TL,CO. An unambiguous way to rephrase this objection would be to say that the rate constant for CO uptake varies with VA, as shown (within an individual) by Krogh 3 in 1915, and confirmed by all subsequent authors.
The variation in KCO with VA in normal subjects has been investigated extensively since 1959 4; in 1994, Stam et al. 5 suggested that in restrictive lung disease values of TL,CO and KCO should be compared with reference values both at the patient's predicted total lung capacity (TLC) and at the lung volume equal to the patient's actual TLC; this suggestion has been endorsed subsequently 1, 6, 7. The novelty in the approach of Chinn et al. 1 rests on the development of reference values for TL,CO and KCO, which include a term for VA (at TLC) as well as a height term, i.e. they take into account variation in TLC at a standard height. Extrapolating from this, they suggest that their reference equations may be used to interpret TL,CO when VA is reduced or increased in disease, “… and eliminate the need for the carbon monoxide transfer coefficient”. On the contrary, the present authors argue that both primary measurements (KCO and VA) should always be examined, especially in disease.
Determinants of carbon monoxide transfer coefficient in normal subjects
Within individuals
In a healthy subject, the degree of lung inflation and the pulmonary capillary volume are probably the major determinants of TL,CO and KCO. Figure 1⇓ shows that KCO and TL,CO are functions of alveolar expansion 5, cardiac output 8, and haemoglobin concentration 9. The extensive studies of Stam et al. 5 have emphasized that with a reduction in alveolar expansion down to 50% TLC the rise of KCO is linear, although earlier studies 4, 10 found a steeper rise at VA <50% VA,max. Despite this increase in KCO, the product KCO×VA (i.e. TL,CO) falls as VA declines. On exercise, KCO (and TL,CO) increases from its value at rest (cardiac output 5 L·min−1) by ∼20% per 5 L min−1 increase in blood flow.
In a) the transfer factor (TL,CO) and b) carbon monoxide transfer coefficient (KCO) are plotted against alveolar volume (VA) as per cent of the VA value at total lung capacity (TLC), at different levels of alveolar expansion (indicated by arrow). There is a systematic change with increasing age. (Data replotted from 5.) In c) and d) KCO (normalized as indicated, and measured at TLC) is plotted against c) pulmonary blood flow at rest ( - - - -) and on exercise, data from 8; and d) against haemoglobin concentration (data from 9). TL,CO would behave similarly.
The physiological explanation for these changes is given in the Roughton-Forster
equation 11, corrected for VA:where Dm is the membrane diffusing capacity (mmol·
min−1·kPa−1), θ is the
reaction rate of CO with haemoglobin adjusted to a standard haemoglobin (Hb)
concentration (mmol·min−1·kPa−1·L−1) and Qc is the
pulmonary capillary volume (L); the units of all three terms are
mmol−1·min·kPa·L−1.
As the expansion of the lung diminishes, Dm in absolute terms falls, but Qc does not change in any systematic way 12, 13. Therefore, the fall in TL,CO (fig. 1a⇑) is dominated by the fall in Dm. KCO, on the other hand, is dependent on the ratios, Dm/VA and Qc/VA. In the sitting position 12, 13, the fall in Dm is almost proportional to the fall in VA, so the rise in KCO as VA falls (fig. 1b⇑) is dominated by the rise in Qc/VA.
Several other physiological factors influence KCO at a given VA. As cardiac output rises on exercise (fig. 1c⇑), Qc/VA increases by capillary distension and recruitment; Dm/VA also increases slightly because vascular distension expands the alveolar surface available for gas exchange. In contrast, anaemia, by reducing θ/VA will reduce KCO (fig. 1d⇑) and TL,CO similarly. A low alveolar oxygen tension (Pa,O2), as occurs at altitude, will increase KCOθ by increasing VA 14, and any accompanying polycythaemia will enhance this.
Technical factors can influence the value of KCO such as the speed of the initial inspiration (it should be rapid) and the method used to measure the BHT. As shown in fig. 1a and b⇑, inadequate inflation of the lungs to VA,max in the single breath test, will result in a low TL,CO and a high KCO. For clinical purposes, the recommendation 15–17 is that the preceding inspired volume from residual volume (RV) should be at least 90% of the subject's vital capacity (VC) so that, with normal gas mixing, the TL,CO and KCO≥ measurements are made at 90% of actual TLC 18. Because gas mixing is not quite complete in the 10 s BHT, VA,max in normal subjects is on average 94±7% of TLC, or 0.1–0.6 L less in absolute terms 18.
Between individuals
After standardization for age, height and sex, Chinn et al. 1 found a very similar relation between TL,CO and VA measured at full inflation in their population study (i.e. an inverse relationship between KCO and VA) to that found with submaximal inflation in an individual. Therefore, they propose an additional VA term to improve the relatively inaccurate predictions of reference values of TL,CO and KCO. They support their own population study by reviewing the mean values of predicted TL,CO and VA from nine published studies of reference values and, at least in males, find these share a similar relation of TL,CO to VA. Unfortunately, the ratio VA/TLC was not available in any of these studies, but using TLC predicted (TLCpred) from mean age and height, eight studies had VA/TLCpred≥0.90 while the remaining study 19, which has a disproportionate influence on the slope, had VA/ TLCpred of only 0.77. Therefore, further studies, which include individual measurement of VA/TLC, are needed to establish the presence and size of any effect of differences in TLC at a given height on values of KCO and TL,CO in a healthy population.
Effects of altered alveolar volume on transfer factor for carbon monoxide and transfer coefficient in respiratory disease
Reduction in alveolar volume and total lung capacity
As discussed above, Stam et al. 5 suggested that when TLC is reduced by disease, TL,CO values should be compared with reference values based on the observed VA, but they cautioned that “this assumes that the effect of decreasing lung volume by disease has the same effect on TL/VA as the voluntary reduction in lung volume in healthy volunteers”. Some of the different mechanisms of reductions in VA at TLC are outlined in table 1⇓, and will be reviewed to emphasize the weaknesses of this assumption.
Different mechanisms reducing singlebreath alveolar volume (VA) in respiratory disease
Respiratory muscle weakness
The most obvious simulation of voluntary reduction in VA,max (table 1⇑“”, lack of lung expansion mechanism), occurs when acute inspiratory muscle weakness prevents the achievement of a normal TLC; in this case, the lack of inflation of the lung can be expected to be relatively uniform and associated with a reduced lung elastic recoil pressure (PL) at VA,max, and preservation of a similar distribution of cardiac output and pulmonary capillary volume as in normals. Thus TL,CO should fall and KCO should rise from the conventional TLC reference values as predicted in fig. 1a and b⇑. In six patients with severe isolated diaphragm weakness 20, the mean TL,CO was 65% pred (range 44–78) and the mean KCO was 128% pred (range 101–167) at 60% of predicted maximum VA; a TL,CO of 80–85% and a KCO of 130–140% would have been predicted on a reduced VA expansion model (fig. 1⇑). A possible explanation for the lower TL,CO and KCO than expected is secondary atelectasis; the remaining aerated lung units would then be more expanded than indicated by the actual level of VA, and have a lower KCO.
Loss of alveolar units
The physiological situation with a reduction in VA and TLC (table 1⇑, loss of units mechanism) from lung resection, e.g. pneumonectomy, is completely different. First, Pl and the dimensions of the remaining airspaces are normal or even increased 21 at full inflation. Secondly, total pulmonary blood flow probably remains at preresection levels so that, depending on the flow-partitioning preoperatively, flow to the remaining lung per unit volume will increase up to two-fold (as if cardiac output had doubled from 5 to 10 L min−1), a situation analogous to the KCO versus cardiac output plot in fig. 1c⇑. This relationship between KCO and pulmonary blood flow can be transposed into a plot of KCO against loss of alveolar units (as VA/VA,max %), where 50% VA,max is equivalent to the KCO· for the whole lung at double the resting cardiac output (10 L min−1) and 33% VA, max is equivalent to a three-fold increase of blood flow per unit volume (fig. 2b⇓). The TL,CO which results from these opposing changes of KCO andVA is also shown (fig. 2a⇓).
Predicted changes in a) the transfer factor (TL,CO) and b) carbon monoxide transfer coefficient (KCO) at full inflation when total lung capacity (TLC)
is reduced by disease. TL,CO and KCO
are plotted against alveolar volume (VA) as
a fraction of VA at predisease TLC for two different causes
of VA reduction, incomplete alveolar expansion (),
and loss of alveolar units. Incomplete expansion follows figure 1a
and 1b⇑. The loss of units plot (
)
is derived from figure 1c⇑
by transposing the KCO (TL,CO
is similar) at twice pulmonary blood flow at rest to KCO at 50% VA,max, and the KCO at 1.5 times blood flow increase to 66% VA,max, etc. The dashed line indicates different benchmarks for a VA’ of 60% of the predisease value against which
a patients KCO or TL,CO at that VA could be compared to the standard reference point (shown
by an arrow). See text for explanation.
Preservation of Qc is the reason why, for a given VA, TL,CO and KCO in figure 2⇑ are higher in the incomplete alveolar expansion situation than for loss of alveolar units. At 50% VA, max, for loss of units, the data of Hsia et al. 8, expressing the values for one lung at twice resting pulmonary blood flow as per cent of both lungs at resting flow, would predict a Dm of 58% and a Qc of 67%. On the other hand, with voluntary reduction to 50% VA,max, Dm (as % Dm at VA,max) would also be 58% but Qc would be 100% 12. Both models presuppose that the alveolar units of the VA have normal function; deviations from the expected values will occur when this is not the case.
From the data in 28 patients Corris et al. 22 established an empirical relationship
for the increase in KCO postpneumonectomy:where x is the percentage
flow to the resected lung preoperatively, based on a radioisotope lung perfusion
scan. For equal flow to both lungs before pneumonectomy (x=50%),
they found that postpneumonectomy KCO was 110–131%
for a mean KCO preoperatively of 98%. Since VA,max after pneumonectomy averaged 50% of pred TLC 22, the loss of units model (fig. 2b⇑) implies a doubling of pulmonary blood
flow per unit volume with a KCO in the range 117–125%
pred, which is similar to the results of Corris et al. 22. The reduced alveolar expansion model,
conversely, would predict a much higher KCO of 145–155% (fig. 2b⇑).
Diffuse alveolar damage
In the preceding two examples, the structure and expansion of the lung remains uniform, whereas in chronic interstitial lung disease (table 1⇑, diffuse alveolar damage mechanism) the structural and functional changes are characteristically nonuniform. In the most abnormal ventilated alveolar units, volume, Dm and Qc are reduced and KCO is low. On the other hand, there may be some redistribution of blood flow to the least abnormal alveolar units whose KCO may be increased (fig. 2b⇑, loss of units). Depending on the overall weighting, the whole lung KCO (using standard reference values) may be low or even normal. In fibrosing alveolitis, for example, a KCO of 100% pred at a low VA implies from fig. 2b⇑ some degree of diffuse alveolar damage.
Stam et al. 7 have recently studied the pre- and postdisease dependence of KCO and TL,CO on VA in a group of young males without previous pulmonary disease, some of whom developed changes in the lungs, accompanied by modest reductions in TL,CO and KCO, when treated with bleomycin for a germ cell tumour. In these males (and in one 11-yr-old female with interstitial lung disease 23), the absolute change in TL,CO and KCO with change in VA (L) was similar before and after disease developed, supporting their contention that the extent of disease was assessed more correctly, and appeared greater, if values of TL,CO and KCO were compared to reference values for the actual TLC rather than to values for the predicted (predisease) TLC. While this may be justified in the unique circumstances of their study, usually predisease TLC is unknown.
Airflow obstruction
TLC is normal or increased in most patients and the low single-breath VA (VA,SB) in airflow obstruction (table 1⇑, VA<TLC due to incomplete mixing mechanism) is caused by incomplete mixing, within the BHT, between the inspired He-CO gas mixture and the RV in the lungs. Without airflow obstruction, the VA at full inflation should be ∼90–95% of the TLC 18, but, with airflow obstruction, VA,SB/TLC is often <80%. In the derivation of TL,CO, the volume (VA) term could either be the true TLC (minus the anatomic dead space), as originally proposed by Ogilvie et al. 24 (this would give a maximum or upper-bound value for TL,CO) or VA,SB (which would give a minimum or lower bound TL,CO). The European Respiratory Society guidelines recommend the use of TLC, but most pulmonary function laboratories prefer to use VA,SB because no extra measurement is required. The higher bound value for TL,CO (equivalent to KCO×TLC) includes the poorly ventilated units by assigning them a KCO equal to that of the well ventilated units (equivalent to measured KCO). The lower bound value for TL,CO (KCO×VA,SB) excludes the poorly ventilated units (equivalent to TLC−VA difference). Nevertheless, asthma apart, it is probable that the poorly ventilated units will be more affected by the disease process, so that the true gas-exchanging potential will lie closer to the lower bound TL,CO value.
The use of the carbon monoxide transfer coefficient in clinical practice
KCO is an index of alveolar gas exchange efficiency in terms of available surface area (Dm/VA) and vascular density (Qc/VA). Disease processes, which reduce alveolar surface and capillary density (emphysema, fibrosis), or which, more selectively, lead to loss of the microvasculature (vasculitis, intrapulmonary shunting, heart failure) reduce the KCO (table 2⇓) and TL,CO, often severely. In practice, by using the standard reference values for TL,CO at the predicted TLC, the upper or lower-bound values of TL,CO and KCO (% pred) have shown good correlations in emphysema with anatomical measurements of airspace surface area per unit lung volume on subsequently resected lobes 30–32. In addition, the KCO correlates with X-ray computed tomography (CT) scan hypodensity in vivo 32. In the assessment of patients with bullous emphysema for lung surgery, the KCO is a guide to the physiological status of the nonbullous lung, and complements the CT scan.
Some of the most common causes of a carbon monoxide transfer coefficient (KCO) which is lower or higher than the reference value.
The causes of a high KCO are less familiar. Discrete loss of alveolar units and lack of alveolar expansion have already been discussed (table 2⇑, fig. 2⇑). Alveolar haemorrhage 27, redistribution of pulmonary blood flow in asthma 31 and a high cardiac output state e.g. atrial septal defect (ASD) 28 all increase KCO. Alternatively, when the KCO is high, TL,CO may be reduced by lack of expansion or loss of units, normal (as in asthma) or even increased (alveolar haemorrhage or ASD).
Patients with a TL,CO of 60% pred, for example, have a similar reduction in their gas exchange capacity at rest. Nevertheless, this defect may result from a variety of changes in KCO or VA, as shown in table 3⇓; examining these patterns provides information on the underlying pathophysiology which will be overlooked if attention is focused solely on the TL,CO. Further examples of these patterns are discussed in more detail elsewhere 34.
Hypothetical combinations of carbon monoxide transfer coefficient (KCO) and single-breath alveolar volume (VA,SB) giving rise to a carbon monoxide transfer factor (TL,CO) of 60% pred at full inflation
Normalizing the KCO for a low VA
The consequences of three different ways of normalizing the KCO in disease for a current VA of 60% of the VA at predisease TLC are shown in columns 4, 5 and 6 of table 3⇑. The conventional way (column 4) is to compare the observed value with the value predicted at the predisease TLC. An alternative (column 6), proposed by Stam et al. 5 and Frans et al. 6, is to compare the observed value with the KCO at the patient's actual VA from studies of voluntary restriction of expansion in normal subjects (fig. 1b⇑). A third normalization procedure (column 5) compares the observed value with the expected KCO at a VA reduced by loss of lung tissue in which pulmonary blood flow per unit lung volume is high and increases the expected KCO (fig. 1c⇑), but to a lesser extent than with the lack of alveolar expansion model. The same arguments apply to normalizing the TL,CO (fig. 2a⇑).
The importance of choosing an appropriate model for reference values is shown in table 3⇑. If the diagnosis is acute neuromuscular disease (first example), the appropriate model is “lack of alveolar expansion” and the observed value is 105% pred. But, if the same values of KCO and VA were due to transient alveolar haemorrhage, the appropriate reference is “loss of units” (VA loss due to alveolar units filled with blood) and the observed value is increased at 134% pred. In lung resection (second example), “loss of alveolar units” is again the appropriate model (100% pred), whereas the “lack of expansion model” falsely suggests a degree of alveolar damage (81% pred). The appropriate models for diffuse alveolar damage and micro vascular damage are (third, fourth and fifth examples) not obvious. Referencing the measured KCO to the expected KCO at predisease TLC results in an overestimated (or upper bound) value compared to predictions of KCO at the actual VA. Indeed, in diffuse alveolar damage, the KCO expressed in the conventional way may be ≥100% pred (fourth example), and familiarity with the relationships between TL,CO and KCO shown in figure 2⇑ would be needed if a correct clinical interpretation is to be made.
Conclusions
The KCO is a measurement of the rate constant for alveolar uptake of CO during breath-holding in the single breath measurement of TL,CO at full inflation. The TL,CO is derived as the product of the KCO and the single breath alveolar volume (VA) divided by PB-PH2O.
In respiratory disease, at least four different pathophysiological mechanisms are responsible for the reduction in single-breath VA, with only acute inspiratory muscle weakness simulating the effects of voluntary submaximal inflation of the normal lung.
With normal alveolar structure and function, the increase in KCO at a given low VA with incomplete alveolar expansion is greater than the corresponding increase due to lung resection.
The advantages and disadvantages of normalizing KCO (and TL,CO) to predisease predicted TLC (the conventional method) or to the actual VA using lack of expansion or loss of alveolar units models, are discussed.
As originally pointed out by Krogh 3, different combinations of alveolar volume and carbon monoxide transfer coefficient may occur in disease for a given value of carbon monoxide transfer factor, each pattern providing different pathophysiological information which would be overlooked if attention was focused solely on the carbon monoxide transfer factor.
- Received February 17, 2000.
- Accepted June 26, 2000.
- © ERS Journals Ltd