In patients with cancer, the cause of AI can be primary (adrenal), secondary (pituitary), or tertiary (hypothalamus). Unfortunately, Kwon et al.1 did not differentiate subtypes of AI. Instead, the authors reported that corticosteroid and megestrol acetate use were risk factors for AI.1 Exogenous use of corticosteroids caused tertiary AI, since chronic high-dose corticosteroids decreased the synthesis of proopiomelanocortin and secretion of ACTH. Han et al.2 also revealed that at least three short courses of high-dose dexamethasone every two to four weeks as antiemetics led to AI within three to six months in 15% of patients with cancer. However, in Kwon et al.’s study,1 corticosteroid use of any duration, regardless of dosage, was a risk factor for AI, which needed to be reevaluated. Moreover, the proportion of high-dose corticosteroid users was only 36% in Kwon et al.’s study.1 The other purpose of low or medium-dose users should be specified. The duration of corticosteroid use was calculated as the cumulative duration. Nevertheless, the effect of intermittent or daily use would be quite different. Moreover, the withdrawal period might affect the GI-AI. It would be more informative if the authors suggested the duration or dosage of drugs and the withdrawal period.

As another risk factor, Kwon et al.1 identified the use of megestrol acetate, which has glucocorticoid activity via binding to the glucocorticoid receptor in the pituitary gland.3 Thus, megestrol acetate causes secondary AI. Han et al.2 also found that patients who also received megestrol acetate had a threefold higher risk of AI.

Notably, in Kwon et al.’s study, eosinophilia was a better indicator for AI rather than hyponatremia.1 Therefore, this finding suggests that cancer patients with eosinophilia might have AI instead of drug hypersensitivity reaction or other diseases. However, relative lymphocytosis and hypercalcemia, which were also recognized as a manifestation of AI, needed to be evaluated.4

If there are simple tools for diagnosing AI in patients with cancer, it would be helpful to circumvent the stimulation test for AI. Kwon et al.1 suggested that morning fasting cortisol < 6.2 µg/dL showed a high specificity of AI and < 12.85 µg/dL excluded AI with a probability of 98.9%. In other studies, early morning serum cortisol < 3.6 µg/dL was identified for diagnosing AI, which was lower than Kwon et al.’s study.5 At this point, it remains unexplained, but it might reflect basal stress status in patients with cancer.

Overall, Kwon et al.1 showed that discriminatory features such as eosinophilia and significant risk factors such as the history of glucocorticoid or megestrol acetate increased the probability of AI in cancer patients with non-specific symptoms. In addition, they gave us useful tips that morning basal cortisol measurement might be enough to diagnose AI in some patients. Future studies should seek other discriminatory features and validate the pre-specified cutoff values of morning basal cortisol to diagnose AI.