Mori, M., Shirado, A.N., Morita, T., Okamoto, K., Matsuda, Y., Matsumoto, Y., . . . Iwase, S. (2017). Predictors of response to corticosteroids for dyspnea in advanced cancer patients: A preliminary multicenter prospective observational study. Supportive Care in Cancer, 25, 1169–1181.

To explore potential factors predicting the response to corticosteroids for dyspnea in patients with advanced cancer.

Measurement variables were recorded at two time points as a part of routine practice for patients who received corticosteroids: baseline (day 1) and in the evening on day 2 after administration of corticosteroids. Patients were followed until one month after administration of corticosteroids and dates of discontinuation or death were recorded. Recommended doses of corticosteroids were betamethasone 2-8 mg per day; dexamethasone 2-8 mg per day, prednisolone 15-60 mg per day, methylprednisolone 10-40 mg per day, and hydrocortisone 50-200 mg per day given orally, IV, or subcutaneously.

• N = 74 patients: 82 enrolled, 6 patients died before day 3, and 2 patients had missing values of numerical rating scale (NRS) of dyspnea on day 3 for unknown reasons.  
• AGE: Mean age = 68 years
• MALES: 53%  
• FEMALES: 47%
• CURRENT TREATMENT: Unknown
• KEY DISEASE CHARACTERISTICS: Patients were included if they had metastatic or locally advanced cancer, were receiving specialized palliative care services, and had a dyspnea intensity of greater than 3 on NRS, worst during the last 24 hours. Patients were excluded if they were unable to communicate verbally due to delirium, dementia, or organic brain disorders; had contraindications to corticosteroids; had dyspnea due to acute exacerbation of underlying non-malignant comorbidities; or were planned or did undergo a thoracentesis during the study period.  
• OTHER KEY SAMPLE CHARACTERISTICS: The following patient characteristics were noted: primary lung tumor (39%), primary tumor of the stomach, colon, or rectum (18%), primary tumor of the uterus and ovary (14%). Other patient characteristics include: metastasis to lung (58%), metastasis to pleura (54%), metastasis to peritoneum (31%), metastasis to liver (30%), and metastasis to brain (8.1%). The majority of patients (84%) had an Eastern Cooperative Oncology Group (ECOG) performance status 3 or 4; 76% had palliative prognostic index (PPI) greater than 6; 64% used supplemental oxygen. Median oxygen flow rate was 2 L per minute and median survival was 26 days.

• SITE: Multi-site   
• SETTING TYPE: Not specified    
• LOCATION: 17 sites in Japan

• PHASE OF CARE: End-of-life care
• APPLICATIONS: Palliative care

Prospective, observational study

Primary end-point of worst dyspnea by NRS the last 24 hours was on the evening of day 3. NRS format for dyspnea from the Japanese version of the MD Anderson Symptom Inventory (MDSAI) was used. A response to corticosteroids was defined as a priori greater than or equal to one-point reduction in NRS of dyspnea. Secondary endpoints included support team assessment schedule, Japanese version (STAS-J), patient-perceived changes in dyspnea, confusion assessment method, short version (CAM), and memorial delirium assessment scale. Potential predictors of response to corticosteroids were recorded prior to administration of corticosteroids by the treating palliative care physicians. Potential predictors included patient demographics, indicators of general conditions, palliative prognostic score (PaP), laboratory findings and oxygen variables, etiologies of dyspnea and clinical manifestations, physician-predicted response of a six-point Likert-type scale, baseline dyspnea severity, and dose of corticosteroid.  
Survival times were calculated using Kaplan-Meier methods. Paired student t tests were used to compare NRS values of dyspnea before and after the administration of corticosteroids. Patients with missing dyspnea NRS values on day 3 due to severe dyspnea or delirium caused by corticosteroids were classified into a non-responder group and Last Observation Carried Forward method was applied. Frequencies and 95% confidence intervals of the proportion of patients with positive CAM tests and those with MDAS item 9 scores greater than or equal to 1 were calculated. Cohen’s kappa was calculated to explore the agreement between CAM-positive and MDAS item great than or equal to 1. Treatment responses between patients with and without each potential predictor was compared using chi-square tests. A logistic multivariate regression analysis was used to identify independent factors predicting greater than or equal to one-point reduction in dyspnea NRS. An alpha (two sided) and power = 0.8, 28 patients per group was needed to determine differences.

Patients had a 1.9 reduction of mean dyspnea NRS worst after administration of corticosteroids (p < 0.001). 50 patients showed a greater than one-point reduction in NRS worst, and 40 patients showed a greater than two-point reduction. 47 patients perceived their condition to be better. Predictor factors that were associated with greater than or equal to one-point reduction in dyspnea were age 70 years or older (p = 0.008), absence of liver mets (p = 0.001), presence of pleuritis carcinomatosa with small collection of pleural effusions (p = 0.011), and presence of audible wheezes (p = 0.002).  Major airway obstruction (p = 0.088), non-purulent serous secretions (p = 0.088), and absence of liver mets (p = 0.055) were associated with a two-point reduction in NRS. Multivariate analysis showed that independent factors predicting response to corticosteroids were PPI greater than 6 (p = 0.021), baseline NRS of dyspnea greater than or equal to 7 (p = 0.036), and absence of liver mets (p = 0.029).

Corticosteroids improved the majority of patients mean dyspnea NRS score. Patients also perceived that the corticosteroids improved their dyspnea. Caution should be taken to monitor for the development of delirium with starting corticosteroids in this patient population.

• Small sample (< 100)
• Risk of bias (no control group)
• Risk of bias (no blinding)
• Risk of bias (no random assignment)
• Risk of bias (no appropriate attentional control condition)
• Selective outcomes reporting
• Other limitations/explanation: Patients received co-interventions such as palliative treatment; therefore, effects cannot be exclusively attributed to corticosteroids. There was no set protocol about the dose and administration methods of corticosteroids therapy. The validity and inter-rater reliability of detecting etiologies and clinical manifestations were not formally tested. Some predictors were based solely on clinical impression. The sample size was not large enough to identify some predictors, such as major airway obstruction. It is unclear how steroids affect patients who have delirium or the inability to interpret symptoms. The benefits of steroids was measured after only 3 days of treatment. Some patients had treatable causes for their dyspnea, which may have contributed to a decreased dyspnea rating.

Nurses may consider using corticosteroids management of dyspnea in the palliative setting based on certain predictors such as bronchial constriction and no evidence liver mets. Even if patients are cognitively impaired, they may still experience symptoms of dyspnea and should be considered a candidate with alternative means of assessing dyspnea.