Why is COPD phenotyping like sorting diamonds?

ACOS is a global problem

Despite the previously limited acceptance of ACOS, studies now clearly show that this is a global problem. Hardin et al. [4] confirm ACOS in North American Caucasians and demonstrate ACOS in African Americans. ACOS has also been reported in Europeans [5, 6], in Australians and New Zealanders [3, 7–9], in Latin American populations [10], and in Asian groups [11]. ACOS is therefore a consistently recognisable subtype of COPD that is seen around the world, the prevalence of which is also high, with one population study reporting an overlapping phenotype in greater than 50% of people of aged 50 years [3].

Does this matter? It seems so, as patients with ACOS experience a greater disease burden than those with asthma or COPD alone. Hardin et al. [4] demonstrate this clearly in the COPDGene study. This study represents the largest sample of people with ACOS who have been clinically characterised. It confirms the work of others in showing significantly greater quality of life impairment in ACOS. In addition, we see that the BODE (body mass index (BMI), airflow obstruction, dyspnoea and exercise capacity) score was higher in ACOS. A BODE score of 3, as seen in ACOS, is associated with a 5-year mortality of 30% compared with a 5-year mortality of 20% when the BODE score is less than 3, as in the COPD-alone group. Overlap subjects also had a greater exacerbation burden. They reported an average of 1.7 times more lung attacks per year in those with ACOS, with one in three ACOS patients requiring hospitalisation or emergency department attendance for a lung attack.

This increased disease burden from ACOS seen by Hardin et al. [4] confirms the work of others, such as the PLATINO study [10]. Given the established association between mortality and both BODE score and exacerbations, this mandates that we place a greater emphasis on the identification and management of ACOS.

What can explain the increased disease burden and exacerbation risk in ACOS?

In this study, Hardin et al. [4] used quantitative chest computed tomography analysis to characterise the patients radiologically and they showed that there was more airway disease in patients with ACOS, with increased bronchial wall thickening and less emphysema, confirming a similar report by Simpson et al. [12]. This is related to the extent of gas trapping and potentially symptom impairment, but it did not explain the exacerbation risk. This was also not explained by the higher BMI in ACOS. Importantly, treatment profiles were not reported, and this would be a useful future investigation to examine any changes in inhaled corticosteroid, long-acting β₂-agonist and long-acting muscarinic antagonist use. It is possible that the treatment intensity was different in ACOS and this may modify clinical outcomes. More detailed mechanistic studies are warranted to identify the reason for the exacerbation risk in ACOS.

The genetics of ACOS

A clear goal of COPDGene study was to identify specific genetic markers for ACOS. None of the markers identified reached statistical significance, which confirms work of others showing that the genetics of asthma and COPD appear similar. This was identified previously when Meyers et al. [13] reported IL13 (interleukin-13) to be a common susceptibility gene for asthma and COPD. Recently, Kaneko et al. [14] integrated the published asthma and COPD genetic studies, and found considerable overlap in genetic markers of susceptibility for asthma and COPD. A network analysis was undertaken and over 200 overlapping molecules were identified that formed a single complex network. This would suggest that there are unlikely to be specific genetic markers that will differentiate ACOS from asthma or from COPD alone.

Hardin et al. [4] chose a different interpretation and considered the genetics underpowered, with the familiar implication that all we need is more (genome-wide association study (GWAS)) data. Certainly, there were some markers that looked interesting, even though they failed to reach statistical significance. But before rushing in with more and more GWAS data, we should pause and consider study design. With modern genetics, we obtain incredible precision where the genotype is reported to the level of individual nucleotides. Contrast this to the clinical characterisation, where ACOS is determined by answers to questions like “do you wheeze?” and “when did it begin?”. Perhaps genetic epidemiologists should demand much better clinical precision, approaching that offered by genetics. Resolving this precision mismatch just might clean up this whole area. A quote from 2004 seems relevant here; it asks us to reflect on what have we done in clinical phenotyping over the last 10 years: “Now that we have sophisticated molecular level genotyping, we need to employee [sic] equally sophisticated phenotyping. Only when both are employed with equal precision, will progress be made toward linking specific genes to specific phenotypes” [15].

COPDGene does this with detailed imaging of the lung. The authors were persuaded by some of the single-nucleotide polymorphisms identified, which can be related to ACOS with biological plausibility. These relate to concepts of the origin of ACOS as a phenotype that is believed to start with asthma, which then develops into secondary COPD from associated smoking, longstanding severe asthma or other unidentified exposures. In this context, the association of ACOS with SOX5 is interesting, as this gene is associated with both COPD and abnormal lung development [16]. There is mounting evidence showing lung function defects in asthma during the period of lung growth and development in utero and in childhood [17] that persist into adult life and could lead into COPD. GPR65 was also linked to ACOS. It is associated with eosinophil activation, a typical feature of eosinophilic asthma.

Another approach might be to identify biomarkers that are typical of ACOS. Iwamoto et al. [18] reported elevated sputum lipocalin, a neutrophil marker, as differentiating COPD and ACOS. In contrast, Bafadhel et al. [19] found that the mediator profile reflected the pattern of granulocytic inflammation, which is not specific for a diagnosis of asthma or COPD. This makes more biological sense when one recognises the heterogeneity of the inflammatory responses seen in airway diseases. Induced sputum eosinophilia predicts response to inhaled corticosteroid in ACOS, just as in asthma and COPD [20].

ACOS exists as a recognisable phenotype of COPD and severe asthma. It is important because the ACOS phenotype of COPD is associated with an increased illness burden that may be modifiable. Additionally, ACOS or its elements may precede the development of severe COPD, and so there is the possibility that ACOS recognition and management offers a path to COPD prevention.

What needs to happen now?

We should work to improve the definition and recognition of ACOS. Consensus meetings may do this, as the successful Ciba symposia showed in the last century. Studies like that of Hardin et al. [4] offer the possibility of specific testing but require replication. ACOS now needs to be included in guidelines and randomised controlled trials of COPD treatment. Just as a diamond requires a multidimensional assessment that includes colour, cut, clarity and carat to determine its value and ultimate use, so in COPD, multidimensional assessment may prove a useful path to revealing the true value of what we can offer our patients [21]!