Abstract
Transfer coefficient of the lung for carbon monoxide (KCO) and alveolar volume (VA) increase the yield of clinical information obtained from transfer factor of the lung for carbon monoxide (TLCO) measurements in clinical practice http://ow.ly/AVgu30na1vu
Case history
A 67-year-old morbidly obese female (body mass index: 46.3 kg·m−2) with a history of long-term cigarette smoking (>30 pack-years) was referred from Cardiology to Respirology due to progressive dyspnoea and recent findings of extensive mosaic attenuation of the lungs on a high-resolution computed tomography (HRCT) scan (figure 1). She had been followed by Cardiology on the grounds of multivalvular disease (severe aortic stenosis and moderate mitral regurgitation), ischaemic heart disease, hypertension and hypercholesterolaemia.
A non-contrasted chest HRCT scan on inspiration showing diffuse areas of mosaic attenuation of the lungs with enlarged central pulmonary vessels and pulmonary artery trunk (3.6 cm).
The patient reported exertional dyspnoea in the past 5 years, which had worsened markedly in the past year. In fact, her dyspnoea has progressed from a previous grade of 2 according to the 1–5 Medical Research Council scale to a current grade of 4. Apart from occasional, non-productive cough she denied any other respiratory symptom. As part of her pre-operative assessment for potential transcatheter aortic valve implantation, a chest radiograph showed ill-defined, bilateral haziness which was deemed inconsistent with her current haemodynamic status. As mentioned, a subsequent chest HRCT scan (figure 1) prompted referral to the respiratory service.
Question
How can the pulmonary function testing (PFT) laboratory be helpful in guiding further investigations to uncover the cause(s) of the abnormalities observed on chest HRCT?
Answer
The mosaic pattern of lung attenuation seen in figure 1 is characterised by a patchwork of regions of differing attenuation with well-defined borders corresponding to the secondary pulmonary lobules [1]. As expected by the secondary lobules' anatomy, the mosaic pattern may be due to pulmonary vascular, small airway or infiltrative lung disease [2]. Infiltrative lung disease and airway disease can be reliably differentiated by an experienced radiologist. However, pulmonary vascular disease is often misinterpreted as an airway disease [1]. Comparing inspiratory and expiratory CT scans can be helpful in indicating the presence of gas trapping secondary to small airway disease although the support for this approach comes mainly from studies involving patients with high likelihood of gas trapping, e.g. asthma, bronchiolitis obliterans and cystic fibrosis [3–5]. Thus, clinical interpretation of PFTs in the current scenario should be focused on answering the following question: is the patient's mosaic pattern on chest HRCT more likely related to an airway- or a pulmonary vascular-centred disease?
Results of routine PFTs are shown in table 1. Spirometry was essentially within normal limits with a trend towards mild and proportional decrease in FEV1 and FVC either pre- or post-bronchodilator [6]. Body plethysmography also suggested a trend towards restriction with a borderline decrease in TLC, a high IC and a low ERV. There was no evidence of gas trapping either absolute (high RV) or relative (high RV/TLC); in fact, FRC and RV were low. The VA, as determined by methane dilution in the single-breath TLCO measurement [7], was normal-to-low in tandem with TLC, i.e. the VA/TLC ratio was within normal limits (>0.8) [8–10]. Thus, there was no significant maldistribution of ventilation [9, 10] Of note, the haemoglobin-corrected TLCO was severely reduced as well as the KCO (TLCO/VA ratio) [11]. Inhaled bronchodilator had no significant effect on the recorded variables (table 1).
Standard pulmonary function tests pre- and post-inhaled bronchodilator
Collectively, these results indicate: an incipient restrictive ventilatory defect coupled with other findings (increased IC, decreased FRC, decreased ERV and decreased RV) which could be largely ascribed to patient's body habitus (morbid obesity) [12]; associated with an out-of-proportion decrease in TLCO and a severe impairment in gas exchange efficiency (low KCO).
Thus, a pulmonary vascular aetiology was considered the most likely reason behind the observed mosaic pattern on the patient's HRCT.
Clinical follow-up
Based on the PFTs data suggesting pulmonary vascular disease and the mosaic pattern on HRCT with an enlarged pulmonary artery trunk, an urgent transthoracic echocardiogram confirmed pulmonary hypertension (PH). A ventilation/perfusion (V′/Q′) scan was requested with a specific concern of thromboembolic disease [13]; in fact, it did show multiple segmental mismatched V′/Q′ deficits. A CT pulmonary angiogram confirmed chronic thromboembolic material within dilated central pulmonary arteries. Right heart catheterisation results were consistent with pre-capillary PH (mean pulmonary arterial pressure: 39 mmHg, pulmonary arterial wedge pressure: 10 mmHg). In this context, the patient was diagnosed with chronic thromboembolic pulmonary hypertension (CTEPH) [13] and referred to the institutional PH clinic.
Discussion
The transfer of any inspired gas with the ability of binding to haemoglobin (e.g. O2) in the red blood cells (RBCs) crossing the pulmonary capillaries depends on [14]:
the pre-haemoglobin resistance to gas flow (literally from the mouth to the interior of the RBCs), collectively called the “membrane” component; and
the concentration and binding properties of haemoglobin.
When carbon monoxide (CO) is used instead of O2 to assess the integrity of this complex process (haemoglobin's affinity for CO is 230 times that of O2), it has been shown that ∼80% of the TLCO signal comes from the blood phase, i.e. the number of RBCs and/or the number of open capillary vessels (capillary volume, VC) [15]. It is therefore understandable that TLCO is considered a “window to the pulmonary microcirculation” [16]. It may also partially explain why a low TLCO is associated with increased areas of high ventilation–perfusion relationship [17]. It follows that patients with chronic obstructive pulmonary disease presenting with a low TLCO need to ventilate in excess to a given metabolic load (i.e. poor exertional ventilatory efficiency) to overcome an enlarged physiological dead space [18]. Patients with low TLCO (<50% predicted) are also at higher risk of presenting with another source of increased ventilatory stimuli: hypoxaemia [19]. Thus, a low TLCO may be associated with a high respiratory neural drive on exertion, a key correlate of exertional dyspnoea [20]. It is not surprising, therefore, that a low TLCO is an independent predictor of decreased physical activity in this patient population [21].
The clinical interpretation of TLCO, however, might be confounded by the fact that it is critically dependent on:
1) the lung volume at which it is measured and, importantly,
2) “how many” gas exchanging unities at a given lung volume have received the inhaled gas mixture [22, 23].
Regarding the first point, TLCO increases as the alveoli are distended because the surface area for gas exchange increases and because the alveolar–capillary membrane may become thinner. Conversely, as the lung deflates TLCO decreases because the surface area decreases while it may become progressively thicker [24]. However, alveolar distension also compresses the lung capillaries, whereas alveolar emptying gives more space to capillary filling [25] (although this is partially compensated by a squeezing effect on the total intra-thoracic blood volume at very low lung volumes) [26]. It follows that [24–26]: from mid-lung volumes upwards, TLCO increases less than it would be expected from the increase in VA because the latter effect is partially counterbalanced by a (small) decrease in VC; and from mid-lung volumes downwards, TLCO decreases less than it would be expected from the decrease in VA because VC decreases less than the VA [23]. The final result is a curvilinear, exponential-like increase in the VC/VA ratio as the VA diminishes. Translating those concepts to TLCO measurements, they imply an essentially linear decrease in TLCO but an exponential increase in KCO (TLCO/VA) as VA decreases (figure 2) [23]. These considerations carry important messages:
they might lead to a “pseudo-normal” KCO in a restrictive defect that decreases both TLCO and VA [27]; and
1 “unit” change in VA does not lead to 1 “unit” change in TLCO across all possible VA values, i.e. it is highly misleading to state that “the TLCO/VA ratio represents TLCO corrected by lung volume”.
The linear decrease in TLCO as the accessible VA decreases is associated with an exponential increase in KCO. For instance, TLCO decreases ∼25% while KCO increases by ∼90% from TLC to FRC. It follows that at a given VA below TLC, KCO (% predicted) will always be a higher value than TLCO (% predicted). See the text for further elaboration on the clinical interpretation of those measurements. VA/VA TLC%: accessible VA as a percentage of the VA found at TLC. Modified from [23] with permission from the publisher.
Concerning the second point, it should be remembered that during the TLCO manoeuvre we measure only the fraction of TLC which can be accessed by the inhaled mixture (which is provided by VA), a volume that depends on how well the peripheral units connect with the large airways [10, 28]. In most healthy subjects, the VA/TLC ratio is ≥0.85 [8, 29] or ≥0.80 [9] provided that the subject performed a maximal inspiratory manoeuvre. However, in the presence of airway disease and extensive maldistribution of ventilation, VA will be a poor representation of the lungs as a whole (TLC) and VA/TLC ratio is <0.80–0.85 [28]. This might lead to a “pseudo-normal” KCO in an obstructive defect that decreases VA [27].
How to deal with these potential pitfalls in VA and KCO interpretation? We suggest the following simplified interpretation algorithm when TLCO is reduced (figure 3) [23].
A simplified algorithm for the differential diagnosis of a low TLCO taking into consideration the potential pitfalls involved in interpretation of the accessible VA and KCO. Symbols ↑, ↓ and ↔ represent values above, below and within the normal range, respectively. See the text for further elaboration.
When VA is normal there is a high negative predictive value for restriction (because VA is always a fraction of TLC; thus, if VA is normal so is TLC in the great majority of subjects) [30]. If obstruction with maldistribution of ventilation were present, it would decrease VA secondary to a low VA/TLC [31]. It follows that a normal VA, when associated with a low KCO (provided there is no anaemia or recent smoking), suggests pulmonary vascular disease (e.g. PH or vasculitis), intrapulmonary right-to-left shunting or early emphysema, i.e. mild enough to not (yet) be associated with a low VA/TLC.
When VA is low the next step is to check the VA/TLC ratio.
If the VA/TLC ratio is low (<0.8) indicating obstruction with ventilation distribution abnormalities [10, 28], the KCO might turn “normal”. In this scenario, no further valid inferences can be made regarding KCO; however, if KCO is low despite those caveats (which, as mentioned, would otherwise tend to increase the TLCO/VA ratio), there is extensive impairment in pulmonary gas exchange efficiency, e.g. severe emphysema [32].
If the VA/TLC ratio is normal (≥0.8), the only explanation for a low VA is a low TLC, i.e. restriction [30]. Thus:
1) A high KCO indicates a predominance of VC over VA due to: a) incomplete alveolar expansion but preserved gas exchange unities frequently leading to KCO >120–140% or even higher, i.e. extra-parenchymal restriction (e.g. pleural, chest wall or neuromuscular disease) [11]; b) an increase in pulmonary blood flow from areas of diffuse (pneumonectomy) or localised (local destructive lesions/atelectasis) loss of gas exchange units to areas with preserved parenchyma frequently leading to more modest increases in KCO (although a high KCO can also be seen with normal or near-normal VA when there is increased pulmonary blood flow or redistribution such as left-to-right shunt and asthma); and c) extra-vascular haemoglobin, i.e. alveolar haemorrhage.
2) A low KCO points towards intra-parenchymal restriction with impaired gas exchange efficiency as in some interstitial lung diseases (ILD) [33].
3) A normal KCO is consistent with intra-parenchymal restriction with preserved KCO. This is the most common finding in patients with HRCT abnormalities showing a pattern consistent with idiopathic interstitial pneumonia [33]. Thus, a normal KCO should not be misinterpreted as indicative of “no ILD” [34, 35]. In this context, a normal KCO may either indicate that the main mechanism underlying the low TLCO is loss of lung volume (e.g. fibrotic ILD) or the low VA led to a normalisation of KCO due to the curvilinear increase in KCO as VA decreases (figure 2).
Owing to the trend for KCO to be higher than TLCO regardless the underlying reason(s), the severity of functional impairment should always be gradated based on TLCO based on the European Respiratory Society recommendations: mild (less than lower limit of normal, but above 60% predicted according to the recently published Global Lung Initiative equations [36]), moderate (40–60% predicted) and severe (>40% predicted) [6].
In the present case, a low VA but normal VA/TLC (coupled with the absence of obstruction on spirometry, gas trapping or high specific airway resistance) is not consistent with maldistribution of ventilation due to small airway disease. Thus, VA decreased in proportion to a low TLC, i.e. secondary to the patient's morbid obesity and/or to the known trend for CTEPH patients to present with mild restriction due to microatelectasis and fibrotic changes [37]. As shown in figure 2, KCO should increase as VA decreases if there was no impairment in gas exchange efficiency; in fact, this is one of the justifications as to why obesity is associated with a high KCO [38]. It follows that the only plausible explanation for a low KCO in the present case is a low VC, i.e. a pulmonary vascular disease. Some key interpretative issues regarding TLCO and its derived measurements are provided in table 2.
Key recommendations to avoid frequent pitfalls in the interpretation of TLCO, KCO and VA in clinical practice
Key points
Clinical interpretation of the TLCO frequently benefits from associated analysis of the accessible VA, VA/TLC ratio and the diffusion coefficient (KCO).
The TLCO/VA ratio (KCO) should not be erroneously interpreted as “TLCO corrected by lung volume” because a one unit change in VA does not lead to one unit change in TLCO across all possible VA values.
Preserved VA and a low KCO (provided the absence of anaemia or recent smoking) suggests pulmonary vascular disease, intrapulmonary right-to-left shunting or early emphysema.
A low VA in the presence of a normal VA/TLC ratio suggests restriction: a low KCO points towards intra-parenchymal restriction with impaired gas exchange efficiency whereas a high KCO is consistent with extra-parenchymal restriction.
A low VA in the presence of a low VA/TLC ratio indicates obstruction with ventilation distribution abnormalities: whereas a preserved KCO is not amenable to interpretation, a low KCO indicates extensive impairment in pulmonary gas exchange efficiency, e.g. severe emphysema.
A normal KCO should not be used to rule out major pulmonary pathology, particularly when VA is reduced.
Self-evaluation questions
In each of the following patients, please indicate which is the most likely underlying diagnosis which could explain the observed abnormalities (indicated by a # symbol).
TLC % pred TLCO % pred VA % pred VA/TLC KCO % pred 132# 52# 64# 0.56# 87 a) Idiopathic PH
b) Neuromuscular disease
c) Emphysema
d) Asthma
e) ILD
TLC % pred TLCO % pred VA % pred VA/TLC KCO % pred 50# 68# 44# 0.97 145# a) Idiopathic PH
b) Neuromuscular disease
c) Emphysema
d) Asthma
e) ILD
TLC % pred TLCO % pred VA % pred VA/TLC KCO % pred 91 54# 82 0.86 64# a) Idiopathic PH
b) Neuromuscular disease
c) Emphysema
d) Asthma
e) ILD
TLC % pred TLCO % pred VA % pred VA/TLC KCO % pred 69# 44# 67# 0.93 58# a) Idiopathic PH
b) Neuromuscular disease
c) Emphysema
d) Asthma
e) ILD
TLC % pred TLCO % pred VA % pred VA/TLC KCO % pred 107 112 99 0.82 129# a) Idiopathic PH
b) Neuromuscular disease
c) Emphysema
d) Asthma
e) ILD
Suggested answers
c. Moderately reduced TLCO. Low VA is associated with a low VA/TLC and a high TLC indicating extensive ventilation distribution abnormalities in the context of an obstructive airway disease. Of note, despite low VA/TLC, KCO is reduced indicating extensive impairment in gas exchange efficiency. Collectively, emphysema is the most likely explanation for these findings.
b. Mildly reduced TLCO. Markedly low VA is associated with an equally low TLC and preserved VA/TLC. Thus, KCO is supra-normal. Collectively, these results are in line with extra-parenchymal restriction. Given the options available, neuromuscular disease is the most likely explanation for these findings (differentials being pleural and chest wall disease).
a. Moderately reduced TLCO. VA, TLC and VA/TLC are within normal limits. Thus, KCO is reduced. Collectively, the results are in line with impaired intrapulmonary gas exchange efficiency in the absence of restriction or significant airway disease. Given the options available, idiopathic PH is the most likely explanation for these findings (differentials being other pulmonary vascular diseases and right-to-left shunt).
e. Moderate-to-severe impairment in TLCO. Low VA is associated with an equally low TLC and preserved VA/TLC. KCO is also reduced. Collectively, the results are in line with intra-parenchymal restriction. Given the options available, ILD (with impaired gas exchange efficiency) is the most likely explanation for these findings.
d. Preserved TLCO but supra-normal KCO in the context of normal TLC. Given the options available, asthma is the most likely explanation for these findings (differential being left-to-right shunt).
Acknowledgements
This article is dedicated to Dr. J. Michael B. Hughes and Dr. Neil B. Pride (National Heart and Lung Institute, Imperial College, Hammersmith campus, London, UK) whose work has been instrumental to advance the clinical interpretation of TLCO and its derived measurements.
Footnotes
Conflict of interest: J.A. Neder has nothing to disclose.
Conflict of interest: M. Marillier has nothing to disclose.
Conflict of interest: A-C. Bernard has nothing to disclose.
Conflict of interest: D.E. O'Donnell has nothing to disclose.
- Copyright ©ERS 2019
Breathe articles are open access and distributed under the terms of the Creative Commons Attribution Non-Commercial Licence 4.0.