Prevalence,Clinical Characteristics,and Management of Episodic Dyspnea in Advanced Lung Cancer Outpatients: A Multicenter Nationwide Study—The INSPIRA-DOS Study

Sample size

We assumed a dyspnea prevalence of 56% ¹³ for stage III-IV patients. Using these assumptions, with 95% confidence intervals (CIs), an accuracy of 6%, and an expected loss of 25%, we calculated that 351 patients would be needed to reliably estimate the prevalence of ED.

Statistical analyses

Categorical variables are shown as absolute frequency and percentages with 95% CIs. Continuous variables are presented as means with standard deviations or as medians and interquartile range (IQR). Categorical data were compared using either chi-square test or Fisher's exact test, as appropriate. Continuous variables were compared using the Student's t test or Mann-Whitney U test depending on the distribution (normal or non-normal). Normality was checked with Kolmogorov–Smirnov and Shapiro–Wilk test. To identify potential variables associated with the risk of ED, multivariate logistic regression models were fitted to test potential associations between patients reporting ED and principal confounders: sex, age, comorbidities, cancer-related variables, and current treatments. The model results are shown as odds ratios (ORs) with 95% CIs. For variables with missing data, we used multiple imputation from multivariate imputation by chained equations, ²¹ to include them in the model. Predictive mean matching was used for continuous variables, logistic regression for binary variables, and multinomial logistic regression for categorical variables. p-Values <0.05 were considered statistically significant. All statistical analyses were performed with the R statistical software v. 3.6.1. ²²

BD and ED

Of the 366 patients, 135 (35.5%) patients reported BD, with a median intensity of 4/10 (IQR: 3–5), 117 patients reported ED (90% of patients with BD and 31.9% overall). As shown in Table 2, performance status was significantly worse in patients who reported ED. Patients with ED were more affected by cachexia and received fewer disease-specific therapies. Intrathoracic comorbidities (COPD, pulmonary vascular disease, heart failure) were also significantly more common in patients with ED. Patients on PC had significantly more ED than patients receiving disease-specific treatments (p < 0.001).

Overall, patients reported a median of one ED episode per day, with a median intensity of 7/10 (≥5/10 in 75% of patients). Triggers were reported by 105 of the 117 patients with ED (89.9%); physical activity was the most common trigger (n = 101; 96.2%) followed by emotional triggers (n = 42; 40.8%). The characteristics and management of ED are shown in Table 3.

Treatment of ED

Table 4 shows the usual treatments administered for ED in the participants. A total of 154 patients (42.1%) took opioids on a regular basis, mainly morphine (n = 84; 22.9%) and transdermal fentanyl (n = 52; 14.2%). The use of opioids (p = 0.035), inhaled bronchodilators (p = 0.007), oral steroids (p = 0.001), and oxygen therapy (p < 0.001) was more frequently reported by patients experiencing ED (Table 3). ED was managed pharmacologically in 79 patients (67.5%), nonpharmacologically in 86 (84.6%), and with both strategies in 77 (65.8%). See Tables 2 and 4 for details.

Patients reported implementing pharmacological measures for ED later than nonpharmacological measures (median time: 5′vs2′; p = 0.013). The beneficial effects were faster with nonpharmacological measures (median time: 5′vs10′; p = 0.075). The median time to episode resolution was 15 minutes with drugs versus only 6.5 minutes with nonpharmacological strategies (p = 0.107).

Risk factors associated with ED

On the logistic regression model (Table 5), the presence of pulmonary vascular disease was associated with an increased risk of ED (OR: 3.50) although this did not reach statistical significance. Patients with ED had an increased risk of requiring continuous or intermittent oxygen therapy (OR: 9.89; p < 0.001 and OR: 3.61; p = 0.007, respectively).

ED frequency and clinical features

Previous studies have evaluated the prevalence of ED in mixed samples of patients with diverse advanced diseases^10,13,23 and cancer.^9,12,24,25 The aim of early studies on ED^10,12,13 was mainly to characterize this phenomenon with regard to the underlying condition. The only early study that included only patients with advanced cancer was performed by Reddy et al., ¹² but was underpowered. Even though the reported prevalence of ED in those studies ranged from 71% to 100%, nontriggered ED was reported in 22% to 49% of cases. However, others ²⁶ have found that nontriggered ED commonly has an underlying emotional driver. Mercadante et al. published three studies^9,24,25 on ED. The first of these was a survey ²⁵ of 921 patients (LC 22.6%) with different types and stages of cancer and different settings. ED rate in the previous 24 hours was 20.4% but 70.9% in those with BD. In the second study, ²⁴ the authors surveyed 347 patients (LC 23%) at home with an ED rate in the whole sample of 27.9% and 79.5% in the BD subgroup. The third study ⁹ included 439 cancer patients (LC 22.5%) admitted to an acute pain relief and supportive/PC unit. The overall ED rate was 6.1% and 67.5% in those with BD. In our study, the ED rate was much higher than previous studies (31.9% overall and 90% in the BD subset) probably due to patient population (advanced LC). Our data also demonstrate the relevance of BD in the presence of ED and the uncertain existence of ED without an identified trigger, confirming the observations made by Linde et al. ²⁶

In our study patients experienced a median of one episode of ED/day, although a substantial proportion (25%) reported ≥2 episodes/day, similar to the findings reported by Weingärtner et al. ¹³ in the subgroup of patients with LC. This indicates that ED in LC is rather common. Previous studies have shown that ED episodes are usually severe. ¹¹ In our cohort the median intensity of 7/10 was similar to the mean intensity values (7/10) for ED reported in other studies.^9,24,25 The most reported trigger of ED is exertion followed by emotional factors,^10,23,25 which is consistent with our findings. The duration of ED in our study was relatively short: 15 minutes after pharmacological treatment and 6.5 minutes after nonpharmacological measures. Other authors have also reported short ED durations (<10 minutes) in 84% ¹⁰ and 79.9% of cases of patients with LC. ¹³

Considering our sample of outpatients with relatively good performance status, we had expected to observe more daily episodes of ED. This finding can be explained by the wide variability among patients with LC in the perceived unpleasantness of dyspnea. In some cases, even though an external observer might consider the ED episode to be severe, some patients may consider the episode to be a normal fluctuation of their BD and, thus, do not report it as ED. ²⁷ Stowe and Wagland, ²⁸ in a sample of patients with advanced LC, reported that patients found it difficult to quantify the frequency, duration, and relative severity of ED.

ED and comorbidities

Comorbidities play a role in ED, not only limiting exertional capacity but also because exacerbations are triggers of ED.^5,6 We found that COPD, pulmonary vascular disease, heart failure, and cachexia were significantly more common in patients reporting ED. Exertion plays a prominent role as a trigger of ED (96.6% of patients), and the use of oxygen is nearly 10-times greater in patients with ED versus those without ED. Then, it is reasonable to speculate about the potential specific relevance of alveolar oxygen exchange failure in advanced LC patients who develop ED and how certain triggers (e.g., tachycardia with exertion and/or anxiety) may stimulate an unmet demand for more oxygen. Thus, alveolar oxygen-exchange failure could be considered the main differentiating etiopathogenic factor of ED in patients with LC versus other cancers with greater systemic involvement where muscle loss can play an important additional role. In fact, a recent study ²⁹ in a sample of 1027 cancer patients (LC = 26%) confirmed the association between weight and muscle loss with clinical worsening of dyspnea.

It is well known that COPD is an additional factor in exertional pulmonary hypertension and hypoxemia like pulmonary hypertension associated with pulmonary thromboembolism ³⁰ ; this can lead to increased right ventricular afterload and under-recruitment of pulmonary alveoli. This mechanism plays a key role in exercise triggered ED in this population, ³¹ as well as left heart failure ³² and cancer. ³³ Some reports have found that comorbidities, such as COPD ³⁴ or pulmonary embolism, ³⁵ are often underdiagnosed in patients with LC. Mercadante et al. ²⁵ found that COPD (OR: 3.2; p < 0.0001) increased the risk of BD but not ED. Those same authors ⁹ found that LC was independently associated with BD. Both findings suggest that ED may be superimposed on BD (mainly in patients with LC), a hypothesis that is supported by the high prevalence of ED in patients reporting BD but not in all LC patients, as observed in our study and previous epidemiological studies. The multicausal etiopathogenic origin of ED, including emotional factors, suggests that LC patients with BD would benefit from a comprehensive evaluation of ED (including cardiac ultrasound and lung circulation-ventilation scintigraphy) to identify the contributing factors and offer the most accurate etiopathogenic-directed interventions.

Emotion-triggered ED was highly prevalent (40.8%). In the literature, ²³ commonly panic, anger, and excitement ⁷ have been described as potential triggers of ED and associated with predictable and unpredictable episodes of dyspnea. ³⁶ Some authors ³⁷ have described a phenomenon called “dyspnea crisis” which is defined as a “sustained and severe resting breathing discomfort at rest that occurs in patients with advanced illness and overwhelms the patient and caregivers' ability to achieve symptom relief.” Dunger et al. ³⁶ described two different types of dyspnea crisis: “physical” and “psychological.”

Treatment of ED

Patients with ED used more background opioids, inhaled bronchodilators, and oral steroids compared to patients without ED. Other studies describe similar medications. ²⁵ For the specific treatment of ED, 44.4% of patients used opioids, most commonly nasal or transmucosal fentanyl and oral immediate release morphine. Due to the short duration of ED episodes (<10 minutes),^9,10 oral immediate release morphine might be inadequate for the management of most episodes ¹³ due to its slow onset of action (20 to 30 minutes). ³⁸ Transmucosal fentanyl formulations have a fast action (5 to 15 minutes), offering a potential role in ED management. ³⁹ Four double-blind randomized placebo-controlled trials^40–43 enrolling 106 mixed sample cancer patients (LC: 12%–36%) evaluated various transmucosal fentanyl formulations for the prophylactic treatment of induced exertional ED. All of those RCTs showed a potential benefit for fentanyl.

Opioids relieve both intensity and unpleasantness of dyspnea by acting in the limbic and cortical brain areas. ³⁹ Opioids may be underused in the treatment of BD and ED in the outpatient setting and could be a surrogate indicator for the high prevalence of COPD in LC patients. Anxiolytics were also used by 21.8% of patients in our sample.

We found that nonpharmacological and pharmacological interventions were used simultaneously. Nonpharmacological measures were initiated first, probably because exertion was the most common ED trigger, leading the patient to interrupt activity as the first nonpharmacological measure. Full therapeutic effect of nonpharmacological measures was also faster than drug interventions.

On the logistic regression model, oxygen use was significantly associated with ED (p < 0.001 and p = 0.007, respectively). This finding is consistent with a previous study showing that oxygen was widely used among patients to treat ED. ⁴⁴ Nevertheless, continuous or intermittent use of oxygen may be indicative for more advanced illness with limited pulmonary performance.

Limitations and strengths

Our study has several potential limitations. First, patients reported ED for the seven days before the visit, which could be influenced by recall bias, leading to underreporting of mild-moderate episodes of ED. Nevertheless, given that the study was carried out in an outpatient clinic, if we had assessed dyspnea for the day of the visit, this would likely have overestimated the frequency (due to increased previsit anxiety and exertion). Moreover, seven-day recall for symptoms is usual in outpatient clinical visits. Second, the possible exclusion of some patients that experienced ED would affect the reported prevalence of ED, but we believe that this effect should be minimal given the similar rates described in other studies. The high levels of emotional-triggered ED found may be affected by the negative feedback cycle of dyspnea and anxiety, which we tried to minimize. Third, the cross-sectional design limits the directionality of associations among the study variables. Another limitation of the study is that dyspnea was not explored with validated tools to assess either its impact in physical activities or in physical and emotional components of breathlessness. Finally, this is a model with quite a number of variables given the study sample. However, all the variables for which we have adjusted are clinically relevant and since the aim of the model is to estimate the individual effect of each variable on the response, we have met the criteria ⁴⁵ to avoid an overestimation.

The strengths of our study are the multicenter design and the large number of patients enrolled, making this the largest study to date to specifically assess ED in advanced LC. Another strength is the patients' diverse clinical status, who were recruited from both oncology and PC clinics.