More than half of our COPD patients who initiated IA treatment at StD initially presented intolerance, with this value increasing to 62% when including patients with intolerance to IA at LoD. The main symptoms of intolerance were dyspnoea and cough during the first 3 months of treatment, being most common after one month, with intolerance appearing later in 16% of patients. To the best of our knowledge, this is the largest study analysing the use of IA in patients with COPD and CBI by PA and is the first to specifically analyse intolerance to IA and the factors associated in these patients. It also evaluated the use of LoD of IA as an alternative treatment for patients unable to tolerate StD or considered at high risk of intolerance.

Respiratory symptoms such as dyspnoea and cough were the most common cause of treatment discontinuation in our study, as previously reported^13,14,15. In fact, the possibility of bronchospasm after IA is warned in the technical specifications of the commercial IA preparations. These symptoms could be related to direct irritation by the drug, as well as to the changes that these products can produce in the osmolarity of the respiratory tract¹⁸.

The efficacy and safety of IA have been assessed in patients with bronchiectasis in several studies; however, the few studies that have analyzed their impact on COPD basically focused on their effectiveness¹⁹. These studies described lower rates of intolerance compared to ours, although with a similar time of onset, being mainly during the first three months of treatment. Monton et al. reported that 21% of 53 patients with COPD and CBI by PA treated with nebulized colistin presented intolerance during the first three months and 9% after this period¹⁴. However, they excluded an unspecified number of patients with a decrease in FEV₁ of 200 mL and 12% right after the first dose of IA, while our protocol contemplates performing spirometry after 1–2 weeks. This may have increased our overall rate of intolerance as patients with a possible immediate decrease in FEV₁were not detected. In the study including the largest number of patients with COPD on IA, 25% of the patients presented similar adverse respiratory events in the first three months, after a median of 22.5 days¹⁵. However, it should be taken into account that this multicentre study included different IA and at various doses, and the IA protocols used were not described and very likely significantly differed among the various study centres. In another study, Bruguera et al.. did not include patients who received less than 3 months of treatment with colistin for CBI by PA, and thus initial intolerance, which was the most frequent in our study, was not reported¹³. Despite this limitation, 4 of 36 (11%) patients presented bronchospasm after 3 months of treatment.

In patients with bronchiectasis, a meta-analysis of RCT showed that treatment with IA was associated with wheezing and bronchospasm in up to 22% of cases^20,21. However, some studies were of short duration, and thus, longer-term effects were unknown, and in all of these studies the lung function of the patients was higher, with a mean FEV₁(%) > 50%. Likewise, a study on the efficacy and safety of dry powder IA in patients with bronchiectasis found that having comorbid COPD was an independent risk factor for intolerance with an OR of 2.3²². The present study showed that patients who did not tolerate IA had a lower baseline FEV₁(%). This relationship of worse tolerance in relation to greater severity of airway obstruction could explain the better tolerance to IA in patients with bronchiectasis. Interestingly, this same fact can make the management of patients with COPD and CBI by PA difficult, because both the frequency of CBI and the prevalence of PA increase as the severity of COPD increases^23,24. Our results suggest that close clinical and spirometric monitoring should be recommended in patients with COPD, especially during the first month of IA treatment, and in case of clinical worsening in patients receiving long-term IA, the possibility of late intolerance should also be considered.

Our study showed that in patients with COPD and CBI by PA treatment with various IA with a median duration of one year was associated with clinical improvement and decrease of exacerbations, as in previous studies^16,17,18. The decrease in hospital admissions was not statistically significant, probably due to the small sample size and the low number of admissions during follow-up.

A particularly novel aspect of our study was the use of IA at LoD in COPD, a practice that has already been used in other diseases, such as cystic fibrosis (CF) or ventilator-associated pneumonia^25,26,27. Although the numbers were small, almost half of the patients who did not tolerate StD were able to tolerate LoD, and in the latter cases, efficacy was demonstrated by the presentation of the same parameters as those obtained with StD. A possible concern when using LoD of antibiotics is the possible appearance of antimicrobial resistance, but this was not detected in any of our patients during treatment with LoD. This result is similar to that of some previous studies on colistin at LoD in CF, including a study using monthly on/off administration in which no antimicrobial resistance was detected after 6 months^25,26. Patients who did not tolerate LoD were again those with a lower FEV₁(%), but no worsening of FEV₁ was observed in those who tolerated LoD. These results are important since they describe a problem in clinical practice such as intolerance to IA at conventional doses, which, according to our results, occurs in more than half of the COPD patients with CBI by PA. In these cases, the administration of LoD could be tested with close monitoring. On the other hand, the option of starting treatment with LoD in patients with very severe airflow limitation may be considered and, depending on the evolution, the dose might be increased or maintained if proving effective, although more studies are needed. Interestingly, patients who did not tolerate LoD of an IA were also unable to tolerate another IA also at LoD. This suggests that while the type of IA could influence tolerance, the most relevant factor was impaired FEV₁ in our study using colistin and tobramycin.

Our study has several limitations: first, it was an observational, retrospective, single-centre study that did not include a control group. The small number of patients in the subgroups of analysis reduced statistical power and limited the number of variables included in the multivariate analysis. On the other hand, being a single centre study limits its external validity. Therefore, our findings have to be analyzed with caution and must be validated in future prospective and multicenter studies with a larger number of patients. Another limitation derives from the definition of intolerance based only on patients’ symptoms without objective spirometric data. However, the decision of IA discontinuation in routine clinical practice is mainly based on clinical symptoms. Finally, most patients received the IA via a nebulizer, and therefore, the results cannot be extrapolated to other delivery systems such as dry powder or other novel inhaled antibiotic formulations that are being developed²⁸.