Exacerbation history is recognised as a key predictor of future exacerbation risk, with results from other non-interventional studies such as ECLIPSE [18] helping to inform treatment recommendations [3]. The current analyses, using data collected at standard clinic visits from patients who were receiving triple therapy for at least six months prior to entry and who then either continued triple therapy or switched to a LABA/LAMA FDC prior to entry, support and extend these previous findings. In the overall population there was a clear separation between those individuals with a low exacerbation history (defined consistently with GOLD) and those with a high exacerbation history. Just under three quarters of patients with a low exacerbation history did not experience any exacerbations during the follow-up period, compared with approximately half of those with a high exacerbation history, and with significantly lower mean exacerbation rates in the low exacerbation history groups than the high exacerbation history groups when matched by baseline eosinophil category.
Although when analysed by treatment the low patient numbers resulted in high variability around the mean exacerbation rate, the same trend was seen with both treatments. It is of note that despite ICS treatment being discontinued prior to inclusion, the mean exacerbation rates over the follow-up period were all lower in those receiving a LABA/LAMA FDC than in those receiving triple therapy, with approximately 85% of patients receiving a LABA/LAMA FDC and who had a low exacerbation history not exacerbating during the follow-up period, suggesting that the treating physicians were able to select patients who could be ‘stepped down’ to a LABA/LAMA.
Baseline blood eosinophil count, however, did not correlate with subsequent exacerbations, either in the overall analysis or with either treatment. These results contrast somewhat with those of a previous analysis of pooled data, in which baseline eosinophil count correlated with subsequent exacerbation risk in those who had ICS withdrawn [10] (so the equivalent of those receiving a LABA/LAMA FDC in the current analyses, given these patients were previously receiving ICS + LABA + LAMA triple therapy). However, this previous analysis used data from a series of interventional clinical studies that recruited patients with a much higher exacerbation risk than those recruited into DACCORD, and who are much more likely to reflect patients with COPD in the ‘real world’ – 65% in the pooled analyses had a history of at least one moderate/severe exacerbation compared to 45% in the overall DACCORD population [17]. This difference clearly highlights that outcomes of clinical studies in highly selected patient populations cannot always be extrapolated to a real-world population, and emphasises the importance of conducting studies in patients who are as close to the ‘real life’ patient as possible.
The main limitations of these analyses are, of course, also associated with this key strength – the purely non-interventional nature of the study. The only data available are those collected from standard clinic visits, and so it would be unrealistic to recruit only patients who had a blood eosinophil count conducted just prior to study entry. We therefore selected a period of six months to collect these data – although 92.2% had values assessed within three months of entry [17]. Secondly, given the non-interventional nature of the study, the decision to switch the patient from triple to LABA/LAMA FDC therapy (or maintain triple therapy) had to be taken prior to inclusion into the study, with the protocol not impacting treatment choice, and thus patients could not be randomised to therapy, resulting in very low numbers of patients in some subgroups. In addition, all data were collected from the study centres’ own equipment and laboratories. However, these limitations are all consistent with standard care. Furthermore, we do not know why triple therapy was previously initiated in these patients. Finally, the non-interventional nature also meant that medication changes were not prohibited by the protocol, and so it was theoretically possible for all patients who switched to LABA/LAMA FDC to subsequently resume triple therapy. In the event, the majority of patients continued with the same treatment regimen for the duration of follow-up (87.7% and 89.3% of patients in the LABA/LAMA FDC and triple therapy groups respectively) [17].
In conclusion, although the majority of patients included in these analyses did not exacerbate during the follow-up period, whereas exacerbation history is a predictor of future exacerbations, blood eosinophil count is not. This suggests that although eosinophil count may help to guide initiation of ICS therapy (or indeed when not to initiate an ICS) [3], this is less of a consideration when ‘stepping down’ from triple therapy to a LABA/LAMA FDC. Indeed, as was shown in the main analyses, some patients may have a clinical benefit from this step-down [17].