Clinical Features and Risk Factors of Adrenal Insufficiency in Patients With Cancer Admitted to the Hospitalist-Managed Medical Unit

Min Kwan Kwon; Junhwan Kim; Jonghwa Ahn; Chang-Yun Woo; Hyeonjeong Kim; Hye-Seon Oh; Mingee Lee; Seungha Hwang; Keun Hoi Park; Young Hak Lee; Jakyung Yu; Sujeung Kang; Hyo-Ju Son

doi:10.3346/jkms.2022.37.e222

Journal List > J Korean Med Sci > v.37(28) > 1516078950

Go to TopGo to Top Go to BottomGo to Bottom

TOOLS

Kwon, Kim, Ahn, Woo, Kim, Oh, Lee, Hwang, Park, Lee, Yu, Kang, and Son: Clinical Features and Risk Factors of Adrenal Insufficiency in Patients With Cancer Admitted to the Hospitalist-Managed Medical Unit

Original Article

Oncology & Hematology

Journal of Korean Medical Science 2022; 37(28): e222.

Published online: 6 July 2022

DOI: https://doi.org/10.3346/jkms.2022.37.e222

Clinical Features and Risk Factors of Adrenal Insufficiency in Patients With Cancer Admitted to the Hospitalist-Managed Medical Unit

Min Kwan Kwon^1,²

, Junhwan Kim^1,²

, Jonghwa Ahn^1,^2,³

, Chang-Yun Woo^1,²

, Hyeonjeong Kim^1,²

, Hye-Seon Oh^1,²

, Mingee Lee^1,²

, Seungha Hwang^1,²

, Keun Hoi Park^1,²

, Young Hak Lee^1,²

, Jakyung Yu²

, Sujeung Kang²

, Hyo-Ju Son^1,^2,⁴

¹Department of Internal Medicine, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea.

²Medical Hospitalist Unit, Asan Medical Center, Seoul, Korea.

³Division of Endocrinology and Metabolism, Department of Internal Medicine, Korea University College of Medicine, Seoul, Korea.

⁴Department of Infectious Diseases, Uijeongbu Eulji Medical Center, University of Eulji College of Medicine, Uijeongbi, Korea.

Address for Correspondence: Hyo-Ju Son, MD. Department of Infectious Diseases, Uijeongbu Eulji Medical Center, University of Eulji College of Medicine, 712 Dongil-ro, Uijeongbu 11759, Republic of Korea. hjson923@eulji.ac.kr

Received 25 February 2022 Accepted 20 June 2022

The Korean Academy of Medical Sciences

https://creativecommons.org/licenses/by-nc/4.0/

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background

The symptoms of adrenal insufficiency (AI) overlap with the common effects of advanced cancer and chemotherapy. Considering that AI may negatively affect the overall prognosis of cancer patients if not diagnosed in a timely manner, we analyzed the incidence, risk factors, and predictive methods of AI in cancer patients.

Methods

We retrospectively analyzed the medical records of 184 adult patients with malignancy who underwent a rapid adrenocorticotrophic hormone stimulation test in the medical hospitalist units of a tertiary hospital. Their baseline characteristics and clinical features were evaluated, and the risk factors for AI were identified using logistic regression analysis.

Results

Of the study patients, 65 (35%) were diagnosed with AI, in whom general weakness (63%) was the most common symptom. Multivariate logistic regression showed that eosinophilia (adjusted odds ratio [aOR], 4.28; 95% confidence interval [CI], 1.10–16.63; P = 0.036), history of steroid use (aOR, 2.37; 95% CI, 1.10–5.15; P = 0.028), and history of megestrol acetate use (aOR, 2.71; 95% CI, 1.38–5.33; P = 0.004) were associated with AI. Baseline cortisol levels of 6.2 μg/dL and 12.85 μg/dL showed a specificity of 95.0% and 95.4% for AI diagnosis, respectively.

Conclusion

AI was found in about one-third of patients with cancer who showed general symptoms that may be easily masked by cancer or chemotherapy, suggesting that clinical suspicion of AI is important while treating cancer patients. History of corticosteroids or megestrol acetate were risk factors for AI and eosinophilia was a pre-test predictor of AI. Baseline cortisol level appears to be a useful adjunct marker for AI.

Graphical Abstract

Keywords: Adrenal Insufficiency, Neoplasms, Eosinophilia, Megestrol Acetate, Hospitalists

INTRODUCTION

Adrenal insufficiency (AI) is a life-threatening condition that increases the risk of readmission and mortality.1 2 AI is caused by an insufficient production of steroid hormones in the adrenal glands, among which glucocorticoids are related to the regulation of blood glucose metabolism, calcium metabolism, growth and development, immune system, cardiovascular system, and central nervous system.3 Patients with malignancy are at a particularly high risk of AI because many cancer patients have a history of steroid use, which is a known risk factor for AI.4 5 6 The development of AI is also associated with metastatic lesions in the adrenal glands, resection of the adrenal glands, and the use of immune checkpoint inhibitors.7 8 9 10 Supplement of glucocorticoids could alleviate AI-related symptoms, but can also lead to adverse effects such as osteoporosis, muscle atrophy, hyperglycemia, dyslipidemia, and heart failure.11 Therefore, it is important to appropriately diagnose and treat AI in patients with malignancy.

AI is diagnosed by performing a rapid adrenocorticotrophic hormone (ACTH) stimulation test, which involves the measurement of blood cortisol levels in serial blood samples.12 However, it is difficult to measure serial blood cortisol levels in the outpatient setting. Moreover, the common presenting symptoms of AI—general weakness, fatigue, weight loss, loss of appetite, nausea, vomiting, myalgia, hypoglycemia, hypotension, and fever13—are also often caused by advanced cancer or chemotherapy, thereby hindering the discerning of the cause of these symptoms and making a timely diagnosis of AI in cancer patients.

Hospitalist units were introduced to Korea in the late 2010s in order to reduce the workload of clinicians and improve the safety and in-hospital management of patients.14 Importantly, hospitalist units may have a unique advantage in examining the prevalence and clinical features of AI in cancer patients because patients hospitalized in these units have ample time for rapid ACTH tests and their symptoms are closely and continuously observed by dedicated hospitalists. Therefore, in this study, we aimed to analyze the prevalence and risk factors of AI in cancer patients being treated at a medical hospitalist unit, and tried to determine the values of a single cortisol level that can predict the occurrence of AI in cancer patients.

METHODS

Patients and study design

This retrospective study was conducted in patients with malignancy who were suspected of AI and underwent a rapid ACTH stimulation test during admission at the medical hospitalist unit at Asan Medical Center (Seoul, Korea) between January 2020 and June 2021. The medical hospitalist team at our institution is operated in two wards and is composed of 11 internal medicine specialists (hospitalists) who provide medical care 24 hours a day, 365 days a year.

Rapid ACTH stimulation tests were performed when the hospitalists decided to determine if symptoms such as general weakness, weight loss, loss of appetite, nausea, vomiting, and abdominal pain were associated with AI and not advanced cancer or chemotherapy, or if the reason for the signs such as hyponatremia, fever, and hypotension were not clear and AI should be ruled out. Baseline characteristics, laboratory findings, and clinical outcomes of the study patients were reviewed through electronic medical records.

ACTH stimulation test

Rapid ACTH stimulation tests with 250 μg of synthetic ACTH (Tetracosactrin, Synacthen^® DALIM BIOTECH, Wonju, Korea) were used for the diagnosis of AI. The tests were started in the morning at 8 AM; after collecting baseline serum cortisol and plasma ACTH by blood sampling, synthetic ACTH was administered intravenously. Additional serum cortisol levels were collected twice at 30 and 60 minutes after synthetic ACTH injection. AI was defined when the peak cortisol level after synthetic ACTH administration was less than 18 μg/dL (500 nmol/L).12

Definitions

Eosinophilia was defined as an eosinophil count of more than 500 cells/μL.15 Chronic kidney disease was defined as a glomerular filtration rate of less than 30 or the requirement of renal replacement therapy such as dialysis. Corticosteroid and megestrol acetate usage were classified into using any duration, more than 7 days, and more than 30 days according to the cumulative period. Corticosteroid doses were classified according to the prednisone equivalent dose as low dose (≤ 7.5 mg/day), medium dose (7.5–30 mg/day), and high dose (> 30 mg/day). Immune checkpoint inhibitors included PD-1 inhibitors (i.e., pembrolizumab, nivolumab) and PD-L1 inhibitors (i.e., atezolizumab, avelumab).16 17 18 The primary outcome was independent risk factors for AI, and the secondary outcome was the ideal baseline cortisol cutoff for predicting AI.

Statistical analysis

Categorical variables were compared using the χ² test or Fisher’s exact test, as appropriate, and continuous variables were compared using Student’s t-test or Mann-Whitney U test, as appropriate. All tests of significance were two-tailed, and P values < 0.05 were considered statistically significant. Risk factors for the occurrence of AI in patients with malignancy were analyzed using logistic regression analysis. We fitted multiple logistic regression model with variables that were included by backward elimination with P value threshold < 0.2. Variables with clinical importance, such as sex and immune checkpoint inhibitors, were included additionally. The diagnostic accuracy of the baseline cortisol level was assessed using receiver operating characteristic (ROC) curves, and the Youden index was used to select the optimum cutoff points on the ROC curves (optimal balance between sensitivity and specificity). Statistical analyses were performed using SPSS for Windows software package, version 24 (SPSS Inc., Chicago, IL, USA), and ROC curves were drawn using the R Studio (version 1.4.1717; R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org).

Ethics statement

This observational study was approved by the Institutional Review Board (IRB) of Asan Medical Center (IRB No. 2021-1185). To protect personal privacy, identifying information in the electronic database was encrypted. Informed consent was waived by the IRB because no intervention was involved and no patient-identifying information was included.

RESULTS

Clinical features of hospitalized patients with malignancy

During the study period, 3,201 patients with malignancy were hospitalized, of whom 184 (5.7%) patients underwent a rapid ACTH stimulation test. Of them, 65 (35%) patients were diagnosed with AI. Table 1 shows the baseline characteristics of patients with AI (n = 65) and those without (n = 119). The median age of the AI group and the non-AI group was 69 years and 66 years, respectively (P = 0.130), and the proportion of male sex was not significantly different between the two groups (60% vs. 58%; P = 0.791). Except for chronic kidney disease (11% vs. 3%; P = 0.043), the distribution of underlying diseases was not significantly different between the two groups. General weakness (63% vs. 52%) and loss of appetite (28% vs. 33%) were the most common symptoms in both groups. In the AI group and the non-AI group, 15 (23.1%) and 39 (32.8%) patients did not have overt symptoms, respectively. The total sum of symptomatic cases was 126% when considering overlapping symptoms.

Table 1

Baseline characteristics of hospitalized patients with malignancy

Characteristics			Total (N = 184)	AI (n = 65)	Non-AI (n =119)	P value
Age at diagnosis, yr			66.5 (60–74)	69 (58–73)	66 (62–74)	0.130
Male sex			108 (58.7)	39 (60)	69 (58)	0.791
Underlying disease
	Diabetes		57 (31)	24 (36.9)	33 (27.7)	0.197
	Hypertension		50 (27.2)	19 (29.2)	31 (26.1)	0.643
	Liver cirrhosis		17 (9.2)	4 (6.2)	13 (10.9)	0.285
	COPD		3 (1.6)	1 (1.5)	2 (1.9)	0.942
	Tuberculosis		9 (4.9)	4 (3.2)	5 (5.8)	0.557
	Chronic kidney disease		11 (6)	7 (10.8)	4 (3.4)	0.043^*
Symptoms
	General weakness		103 (56.0)	41 (63.1)	62 (52.1)	0.152
	Loss of appetite		57 (31.0)	18 (27.7)	39 (32.8)	0.476
	Weight loss		5 (2.7)	2 (3.1)	3 (2.5)	0.825
	Nausea		20 (10.9)	5 (7.7)	15 (12.6)	0.306
	Vomiting		12 (6.5)	3 (4.6)	9 (7.6)	0.439
	Depression		6 (3.3)	3 (4.6)	3 (2.5)	0.445
	Other symtoms^a		20 (20.0)	10 (15.4)	10 (12.9)	0.146
	No symtoms		54 (29.3)	15 (23.1)	39 (32.8)	0.167
Laboratory findings
	ACTH, pg/mL		51.0 ± 77.3	47.5 ± 100.5	52.8 ± 61.5	0.657
	Cortisol, μg/dL
		0 min, baseline	13.5 ± 12.3	5.8 ± 4.0	17.7 ± 13.3	< 0.001^***
		30 min	21.3 ± 14.5	11.1 ± 4.0	27.0 ± 15.0	< 0.001^***
		60 min	23.7 ± 14.3	12.6 ± 4.2	29.8 ± 14.2	< 0.001^***
	WBC, /μL		9,778 ± 8,398	9,419 ± 7,250	9,943 ± 8,987	0.122
	Hb, g/dL		9.4 ± 1.7	9.2 ± 1.5	9.6 ± 1.8	0.121
	Eosinophil count, /mm³		177.3 ± 15.5	165.7 ± 3.9	183.6 ± 3.9	0.752
		Eosinophilia, > 500/μL	12 (6.5)	8 (12.3)	4 (3.4)	0.019^*
	Sodium, mmol/L		133.0 ± 6.7	133.3 ± 6.3	132.8 ± 6.9	0.441
		Hyponatremia, < 135 mmol/L	105 (57.1)	38 (58.5)	67 (56.3)	0.777
		Severe hyponatremia, < 130 mmol/L	54 (29.3)	20 (30.8)	34 (28.6)	0.754
	Potassium, mmol/L		4.0 ± 0.8	3.9 ± 0.7	4.1 ± 0.8	0.051
		Hyperkalemia, > 5.5 mmol/L	5 (2.7)	1 (1.5)	4 (3.4)	0.467
	Albumin, g/dL		2.5 ± 1.2	2.4 ± 0.5	2.5 ± 1.4	0.441
		Severe hypoalbuminemia, < 2.5 g/dL	103 (56.0)	36 (55.4)	67 (56.3)	0.905
	Fever, > 38°C		17 (9.2)	3 (4.6)	14 (11.8)	0.109
	qSOFA score ≥ 2		12 (7.0)	4 (6.2)	9 (7.3)	0.209
Radiologic findings in adrenal glands
	Metastases in both adrenal glands		0 (0)	1 (0.8)	1 (0.5)	0.999
	Adrenalectomy or metastasis in one adrenal gland		3 (4.6)	3 (2.5)	6 (3.3)	0.667
Medication history
	Corticosteroids		100 (54.3)	40 (61.5)	60 (50.4)	0.148
		More than 7 days	70 (38.0)	32 (49.2)	38 (31.9)	0.021^*
		More than 30 days	42 (22.8)	22 (33.8)	20 (16.8)	0.008^**
		Low-dose^b	15 (15.0)	9 (22.5)	6 (10.0)	0.153
		Medium-dose^c	49 (49.0)	18 (45.0)	31 (51.7)	0.653
		High-dose^d	36 (36.0)	13 (32.5)	23 (36.3)	0.702
	Megestrol acetate		78 (42.4)	39 (60)	39 (32.8)	< 0.001^***
		More than 7 days	18 (9.8)	8 (12.3)	10 (8.4)	0.394
		More than 30 days	43 (23.4)	25 (38.5)	18 (15.1)	< 0.001^***
	Immune checkpoint inhibitors		21 (11.4)	9 (13.8)	12 (10.1)	0.443
		PD-1 (Pembrolizumab, Nivolumab)	16 (8.7)	6 (9.2)	10 (8.4)	0.849
		PD-L1 (Atezolizumab, Avelumab)	5 (2.7)	3 (4.6)	2 (1.7)	0.242
Re-admission			55 (29.9)	22 (33.8)	33 (27.7)	0.386
In-hospital mortality			69 (37.5)	19 (29.2)	50 (42)	0.087

Data are presented as median (interquartile range), number (%), or mean ± standard deviation.

AI = adrenal insufficiency, COPD = chronic obstructive pulmonary disease, ACTH = adrenocorticotropic hormone, WBC = white blood cell, Hb = hemoglobin, qSOFA = quick Sepsis Related Organ Failure Assessment, PD-1 = programmed cell death protein 1, PD-L1 = programmed cell death-ligand 1.

^aArthralgia, myalgia, dizziness, confusion, coma.

^b≤ 7.5 mg/day of prednisone equivalent.

^c> 7.5 mg/day and ≤ 30 mg/day of prednisone equivalent.

^d> 30 mg/day of prednisone equivalent.

^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

In terms of laboratory findings, the AI group and the non-AI group did not show significant differences except for cortisol levels (all P < 0.001) and the proportion of those with eosinophilia (12% vs. 3%; P = 0.019). In the AI group, the mean levels of ACTH and cortisol at 60 minutes were 47.5 ± 100.5 pg/mL and 12.6 ± 4.2 μg/dL, respectively.

In terms of medication history, the AI group had higher proportions of patients who used corticosteroids for more than 7 days (49% vs. 32%; P = 0.021) or 30 days (34% vs. 17%; P = 0.008) than the non-AI group. Megestrol acetate use of any duration was significantly more common in the AI group (60% vs. 33%; P < 0.001), but the use of immune checkpoint inhibitors was not significantly different between the two groups (14% vs. 10%; P = 0.443).

In the AI group, the rates of readmission and in-hospital mortality were 34% and 29%, which were not significantly different from those of the non-AI group (28%, P = 0.386; 42%, P = 0.087, respectively).

The distribution of the types of malignancy is shown in Table 2. The underlying malignancy in the AI group and the non-AI group was solid cancer in 80% and 85% (P = 0.523), respectively. Of the 8 patients with multiple myeloma who were suspected of having AI, 7 (87.5%) were diagnosed with AI (P = 0.003). No statistically significant difference between the AI group and the non-AI group was observed in other malignancies.

Table 2

Types of malignancy in hospitalized patients

Characteristic of patients		Total (N = 184)	AI (n = 65)	Non-AI (n = 119)	P value
Underlying malignancy
Solid cancer		153 (83.2)	52 (80.0)	101 (84.9)	0.523
	Lung cancer	28 (15.2)	9 (13.8)	19 (16.0)	0.866
	Breast cancer	8 (4.3)	0 (0)	8 (6.7)	0.052
	Esophageal cancer	8 (4.3)	2 (5.0)	6 (3.1)	0.714
	Stomach cancer	15 (8.2)	4 (6.2)	11 (9.2)	0.580
	Cholangiocarcinoma	26 (14.1)	14 (21.5)	12 (10.1)	0.056
	Hepatocellular carcinoma	10 (5.4)	4 (6.2)	6 (5.0)	0.744
	Pancreatic cancer	18 (9.8)	4 (6.2)	14 (11.8)	0.302
	Renal cell carcinoma	5 (2.7)	3 (4.6)	2 (1.7)	0.348
	Colorectal cancer	17 (9.2)	4 (6.2)	13 (10.9)	0.423
	Others^a	18 (9.8)	8 (12.4)	10 (8.4)	0.554
Hematologic malignancy		31 (16.8)	13 (20.0)	18 (15.1)	0.523
	AML	4 (2.2)	1 (1.5)	3 (2.5)	0.999
	CML	1 (0.5)	0 (0)	1 (0.8)	0.999
	ALL	1 (0.5)	0 (0)	1 (0.8)	0.999
	CLL	2 (1.1)	0 (0)	2 (1.7)	0.541
	Lymphoma	12 (6.5)	5 (7.7)	7 (5.9)	0.871
	Multiple myeloma	8 (4.3)	7 (10.8)	1 (0.8)	0.003^**
	Others^b	3 (1.6)	0 (0)	3 (2.5)	0.553

Data are presented as number (%).

AI = adrenal insufficiency, AML = acute myeloid leukemia, CML = chronic myeloid leukemia, ALL = acute lymphoblastic leukemia, CLL = chronic lymphocytic leukemia.

^aFive gallbladder cancer, three prostate cancer, three bladder cancer, two tongue cancer, one duodenal cancer, one ovarian cancer, one small bowel gastrointestinal stromal tumor, one neuroendocrine tumor, and one angiosarcoma were included.

^bThree myelodysplastic syndrome were included.

^**P < 0.01.

Predictors and risk factors for AI in hospitalized patients with malignancy

Multivariate analysis showed that eosinophilia (> 500/μL) (adjusted odds ratio [aOR], 4.28; 95% confidence interval [CI], 1.10–16.63; P = 0.036), steroid use for more than 30 days (aOR, 2.37; 95% CI, 1.10–5.15; P = 0.028), and megestrol acetate use of any duration (aOR, 2.71; 95% CI, 1.38–5.33; P = 0.004) were significantly associated with the occurrence of AI in hospitalized patients with malignancy (Table 3).

Table 3

Univariate and multivariate analysis of predictors and risk factors for adrenal insufficiency in hospitalized patients with malignancy

Risk factors	Univariate analysis		Multivariate analysis
Risk factors	OR (95% CI)	P value	aOR (95% CI)	P value
Age	1.02 (0.99–1.05)	0.130	1.03 (0.99–1.06)	0.130
Sex	0.92 (0.50–1.70)	0.791	0.97 (0.50–1.90)	0.929
Diabetes	1.52 (0.80–2.90)	0.199
General weakness	1.57 (0.85–2.92)	0.153
Fever	0.36 (0.10–1.31)	0.122
Eosinophilia, > 500 /μL	4.04 (1.17–13.97)	0.028^*	4.28 (1.10–16.63)	0.036^*
Hyponatremia, < 130 mmol/L	1.19 (0.98–1.45)	0.081
Hyperkalemia, > 5.5 mmol/L	0.76 (0.50–1.15)	0.197
Steroid use, > 30 days	2.73 (1.27–5.85)	0.010^**	2.37 (1.10–5.15)	0.028^*
Megestrol acetate use	3.08 (1.64–5.76)	< 0.001^***	2.71 (1.38–5.33)	0.004^**
Immune checkpoint inhibitor use	1.43 (0.57–3.61)	0.445	1.14 (0.41–3.19)	0.809
Chronic kidney disease	3.47 (0.98–12.34)	0.055	2.98 (0.74–12.01)	0.125

The model fitted the data well in terms of discrimination (C-statistic = 0.74) and calibration (Hosmer-Lemeshow goodness-of-fit statistic =14.26, P = 0.075).

OR = odds ratio, CI = confidence interval, aOR = adjusted odds ratio.

^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Predicting and excluding values of baseline cortisol level for AI in hospitalized patients with malignancy

Fig. 1. shows the ROC curve of baseline cortisol for predicting AI. The area under ROC was 0.92 (95% CI, 0.88–0.96). The optimal cutoff value as determined by the Youden index was 10.65 μg/dL, at which the sensitivity and specificity were 87.7% and 80.7%, respectively. At 6.2 μg/dL and 12.85 μg/dL, the sensitivities of baseline cortisol were 53.9% and 65.6%, respectively, and the specificities were 95.0% and 95.4%, respectively (Table 4).

Fig. 1

Receiver operating characteristic curve of baseline cortisol for predicting adrenal insufficiency in hospitalized patients with malignancy.

AI = adrenal insufficiency, AUROC = area under receiver operating characteristic, CI = confidence interval.

Table 4

Baseline cortisol levels for predicting and excluding adrenal insufficiency in hospitalized patients with malignancy

Baseline cortisol	Cortisol, μg/dL	Sensitivity % (n/N, 95% CI)	Specificity % (n/N, 95% CI)	Positive predictive value (95% CI)	Negative predictive value (95% CI)	Positive likelihood ratio (95% CI)	Negative likelihood ratio (95% CI)
Lower level for diagnosing AI	< 3.05	32.3 (21/65, 21–45)	100.0 (119/119, 97–100)	100.0	73.0 (70–76)		0.7 (0.6–0.8)
	< 6.20	53.9 (35/65, 41–66)	95.0 (113/119, 89–98)	85.4 (72–93)	79.0 (74.3–83)	10.7 (4.7–24.0)	0.5 (0.4–0.6)
	< 8.55	76.9 (50/65, 65–86)	90.8 (108/119, 84–95)	82.0 (72–89)	87.8 (82.2–92)	8.3 (4.7–14.8)	0.3 (0.2–0.4)
	< 10.65	87.7 (57/65, 77–95)	80.7 (96/119, 72–87)	71.3 (63–78)	92.3 (86–96)	4.5 (3.1–6.6)	0.2 (0.1–0.3)
Upper level for excluding AI	> 11.30	75.6 (90/119, 67–83)	90.8 (59/65, 81–97)	93.8 (87–97)	67.0 (59–74)	8.2 (3.8–17.7)	0.3 (0.2–0.4)
	> 12.85	65.6 (78/119, 56–74)	95.4 (62/65, 87–99)	96.3 (90–99)	60.2 (54–66)	14.2 (4.7–43.2)	0.4 (0.3–0.5)
	> 15.50	50.4 (60/119, 41–60)	100.0 (65/65, 94–100)	100.0	52.4 (48–57)		0.5 (0.4–0.6)

AI = adrenal insufficiency, CI = confidence interval.

DISCUSSION

Patients with malignancy have multiple risk factors for the development of AI, and steroid replacement offers significant benefits in cancer patients with AI; however, studies on the risk factors of AI in cancer patients are lacking. To our knowledge, this is the first study to evaluate clinical features of cancer patients with AI, identify the risk factors thereof, and determine the optimal baseline cortisol levels for predicting AI in such patients. By analyzing a cohort of cancer patients admitted to our hospitalist unit, we found that AI was quite common (35%) among cancer patients who showed relevant symptoms and signs such as general weakness and loss of appetite, which are prevalent among cancer patients receiving chemotherapy. Corticosteroid use and megestrol acetate use were independent risk factors for AI, and eosinophilia was an independent pre-test predictor of AI. Importantly, a baseline cortisol level of 10.65 μg/dL was determined as the optimal cutoff for predicting the development of AI (sensitivity, 87.7%; specificity, 80.7%), which may be useful in situations such as outpatient clinics in which serial testing for cortisol levels cannot be performed.

Of the 184 patients with malignancy who underwent a rapid ACTH stimulation test, 65 (35.3%) were confirmed with AI, suggesting that as much as one-third of patients with ongoing advanced cancer or chemotherapy may have AI if they complain of symptoms such as general weakness, loss of appetite, and nausea. Therefore, clinicians and hospitalists should be aware of the possibility of accompanying AI when patients with malignancy persistently complain of the abovementioned symptoms despite symptomatic treatment. Interestingly, although the overall incidence of multiple myeloma was small (8/184 [4.3%]), we found that the proportion of patients with multiple myeloma was significantly higher in the AI group than in the non-AI group (11% vs. 1%; P = 0.003). We assume that this was because patients with multiple myeloma tend to receive repeated treatments with salvage regimens including high-dose glucocorticoids.19 20

In our results, eosinophilia was a significant pre-test predictor for the occurrence of AI. It is not clear why eosinophilia was associated with AI, but the modulation of the expression of adhesion and migration factors by corticosteroids may result in a higher transition rate of eosinophils out of the bloodstream to tissues and confer an eosinopenic effect by stimulating eosinophil apoptosis.21 22 23 Therefore, it is plausible that eosinophilia may be a direct effect of the corticosteroid deficiency caused by AI, and eosinophilia can be utilized as a marker of the possible existence of AI. Other laboratory findings that were known to be related to AI, such as hypoalbuminemia or hyponatremia, had no statistical significance in the prediction of AI; this is presumably because cancer patients often have decreased oral intake due to chemotherapy and cachexia.

We also found that history of steroid and megestrol acetate use were significant risk factors for increasing the incidence of AI. The mechanism by which steroids cause AI is relatively well-known: when corticosteroids are administered, cortisol regulation and releasing system functions are suppressed due to the hormone feedback system.24 25 26 Accordingly, previous studies have reported that the use of dexamethasone for anti-emesis in cancer patients receiving chemotherapy could cause AI.4 25 26 Our finding is consistent with these previous findings. In terms of megestrol acetate, only a few case reports27 28 29 and one recent retrospective study20 reported the history of megestrol acetate use as a risk factor for AI. Through multivariable analysis that adjusted for confounding variables, our study provides further evidence that megestrol acetate could cause AI in patients with malignancy.

A total of 21 (11.4%) patients received immune checkpoint inhibitors, of whom 9 (42.9%) were diagnosed with AI. The development of AI-related to immune checkpoint inhibitors is a critically important issue, as the use of immune checkpoint inhibitors is increasing in cancer patients and their effects are irreversible.16 17 18 Although more than 40% of the suspected patients treated with immune checkpoint inhibitors were diagnosed with AI, the proportion of patients treated with immune checkpoint inhibitors did not significantly differ between the AI group and the non-AI group. Further studies with larger cohorts are needed to clarify this issue.

Our study provides valuable information on predicting AI based on baseline cortisol level in patients with malignancy. A previous study also evaluated the predictive value of baseline cortisol on the development of AI,30 but none has focused on patients with malignancy. We therefore determined the diagnostic performance and usefulness of baseline cortisol levels in a specific population with malignancy. Our results showed that AI can be predicted with a 95% probability if the baseline cortisol level is less than 6.2 μg/dL, and AI can be excluded with a 95% probability if the baseline cortisol level is over 12.85 μg/dL. These cortisol level values were similar to those reported in other studies; one study suggested that AI could be excluded with a 95% probability when the baseline cortisol level was higher than 12.7 μg/dL,30 and suggested that the optimal baseline cortisol level for predicting AI with a 95% probability was 4.5 μg/dL.30 Our findings on the value of baseline cortisol would be useful for clinicians seeking to predict the risk of outpatients with malignancy or those with poor vessel conditions, in whom measuring serial blood cortisol levels after injection of cortisol stimulating hormone is difficult.

Hospital medicine aims to improve the quality and efficiency of generalized care for hospitalized patients.31 Because medical hospitalist units in Korea are mainly operated in tertiary hospitals, the proportion of cancer patients in those units is high. In this study, the presence of AI was not significantly associated with the reduction of mortality rate and readmission rate. Yet, the prediction of AI in cancer patients will be of great help in improving the quality of life in patients and the quality of inpatient care in hospitalist units.

This study had a few limitations. First, selection bias might be present because it is a retrospective study conducted at a single center and ACTH stimulation tests were only conducted on patients who were clinically suspected of AI. This selection bias hinders the generalization of the high incidence rate of AI (up to 35%) in patients with malignancy. Therefore, further studies are needed by securing participation from multiple centers, and well-designed studies are also needed to define the criteria for performing ACTH stimulation tests in cancer patients. Second, AI might have been over-diagnosed because of the inherent limitations of rapid ACTH stimulation tests. The rapid ACTH stimulation test measures cortisol levels in the blood, which are mostly bound with proteins such as corticosteroid-binding globulin or albumin. This could be affected by the fact that patients with malignancy often have hypoalbuminemia due to chronic inflammation, decreased systemic protein synthesis, decreased liver function, malnutrition, and changes in the normal distribution of albumin.32 However, within our limited data, albumin levels were not significantly associated with the occurrence of AI. Further studies are needed to evaluate the occurrence of AI in patients with hypoalbuminemia. Despite these limitations, our study is meaningful in that it examined the incidences and risk factors of AI in patients with malignancy.

In conclusion, AI was diagnosed in more than one-third of cancer patients at a medical hospitalist unit who complained of common symptoms such as general weakness and loss of appetite. Patients with eosinophilia and history of corticosteroids or megestrol acetate use should be closely monitored for the development of AI. Baseline cortisol level may be a useful adjunct marker for AI.

Notes

Disclosure: The authors have no potential conflicts of interest to disclose.

Author Contributions:

Conceptualization: Son HJ, Kwon MK, Kim J, Ahn J, Woo CY, Kim H, Oh HS, Lee M, Hwang S, Park KH, Lee YH, Yu J, Kang S.
Data curation: Kwon MK, Son HJ.
Formal analysis: Kwon MK, Son HJ.
Investigation: Son HJ, Kwon MK.
Methodology: Son HJ, Kwon MK.
Software: Son HJ, Kwon MK.
Validation: Son HJ, Kwon MK.
Visualization: Son HJ, Kwon MK.
Writing - original draft: Kwon MK, Son HJ.
Writing - review & editing: Kwon MK, Son HJ, Kim J, Ahn J, Woo CY, Kim H, Oh HS, Lee M, Hwang S, Park KH, Lee YH, Yu J, Kang S.

References

1. Bergthorsdottir R, Leonsson-Zachrisson M, Odén A, Johannsson G. Premature mortality in patients with Addison’s disease: a population-based study. J Clin Endocrinol Metab. 2006; 91(12):4849–4853. PMID: 16968806.

2. Bensing S, Brandt L, Tabaroj F, Sjöberg O, Nilsson B, Ekbom A, et al. Increased death risk and altered cancer incidence pattern in patients with isolated or combined autoimmune primary adrenocortical insufficiency. Clin Endocrinol (Oxf). 2008; 69(5):697–704. PMID: 18727712.

3. Biddie SC, Conway-Campbell BL, Lightman SL. Dynamic regulation of glucocorticoid signalling in health and disease. Rheumatology (Oxford). 2012; 51(3):403–412. PMID: 21891790.

4. Laugesen K, Broersen LH, Hansen SB, Dekkers OM, Sørensen HT, Jorgensen JO. Management of endocrine disease: glucocorticoid-induced adrenal insufficiency: replace while we wait for evidence? Eur J Endocrinol. 2021; 184(4):R111–R122. PMID: 33449912.

5. Almeida MQ, Mendonca BB. Adrenal insufficiency and glucocorticoid use during the COVID-19 pandemic. Clinics (Sao Paulo). 2020; 75:e2022. PMID: 32555949.

6. Karangizi AH, Al-Shaghana M, Logan S, Criseno S, Webster R, Boelaert K, et al. Glucocorticoid induced adrenal insufficiency is common in steroid treated glomerular diseases - proposed strategy for screening and management. BMC Nephrol. 2019; 20(1):154. PMID: 31060510.

7. Redman BG, Pazdur R, Zingas AP, Loredo R. Prospective evaluation of adrenal insufficiency in patients with adrenal metastasis. Cancer. 1987; 60(1):103–107. PMID: 3581024.

8. Mitchell J, Barbosa G, Tsinberg M, Milas M, Siperstein A, Berber E. Unrecognized adrenal insufficiency in patients undergoing laparoscopic adrenalectomy. Surg Endosc. 2009; 23(2):248–254. PMID: 19037697.

9. Hirai H, Kuwana K, Kusano Y. Late onset adrenal insufficiency after adrenalectomy due to latent nonclassical 21-hydroxylase deficiency: a case report. Medicine (Baltimore). 2018; 97(33):e11888. PMID: 30113485.

10. Percik R, Shlomai G, Tirosh A, Tirosh A, Leibowitz-Amit R, Eshet Y, et al. Isolated autoimmune adrenocorticotropic hormone deficiency: from a rare disease to the dominant cause of adrenal insufficiency related to check point inhibitors. Autoimmun Rev. 2020; 19(2):102454. PMID: 31838158.

11. Oray M, Abu Samra K, Ebrahimiadib N, Meese H, Foster CS. Long-term side effects of glucocorticoids. Expert Opin Drug Saf. 2016; 15(4):457–465. PMID: 26789102.

12. Munro V, Elnenaei M, Doucette S, Clarke DB, Imran SA. The effect of time of day testing and utility of 30 and 60 minute cortisol values in the 250 mcg ACTH stimulation test. Clin Biochem. 2018; 54:37–41. PMID: 29458002.

13. Reddy P. Clinical approach to adrenal insufficiency in hospitalised patients. Int J Clin Pract. 2011; 65(10):1059–1066. PMID: 21762316.

14. Kim J. The impact of hospitalist care in Korea. J Korean Med Sci. 2019; 34(25):e177. PMID: 31243936.

15. Kuang FL. Approach to patients with eosinophilia. Med Clin North Am. 2020; 104(1):1–14. PMID: 31757229.

16. Deng R, Bumbaca D, Pastuskovas CV, Boswell CA, West D, Cowan KJ, et al. Preclinical pharmacokinetics, pharmacodynamics, tissue distribution, and tumor penetration of anti-PD-L1 monoclonal antibody, an immune checkpoint inhibitor. MAbs. 2016; 8(3):593–603. PMID: 26918260.

17. Sunshine J, Taube JM. PD-1/PD-L1 inhibitors. Curr Opin Pharmacol. 2015; 23:32–38. PMID: 26047524.

18. Li B, Chan HL, Chen P. Immune checkpoint inhibitors: basics and challenges. Curr Med Chem. 2019; 26(17):3009–3025. PMID: 28782469.

19. Ahn SY, Kim HK, Kang HC, Kim M, Song GY, Jung SH, et al. Adrenal insufficiency in hospitalized patients with multiple myeloma. Leuk Lymphoma. 2021; 62(2):501–503. PMID: 33135942.

20. Yoon JH, Ahn SY, Jung SH, Lee JJ, Choi W, Park JY, et al. Prevalence and risk factors for adrenal insufficiency in patients with multiple myeloma receiving long-term chemotherapy including corticosteroids: a retrospective cohort study. Biomed Res Int. 2021; 2021:2330417. PMID: 34938804.

21. Mouloudi E, Katsanoulas K, Aslanidis TH, Lampiri CL, Papageorgiou CH, Tholioti TH, et al. Eosinophilia: an early marker of adrenal insufficiency in critically ill patients with septic shock? Greek e J Perioper Med. 2018; 17(a):61–70.

22. Angelis M, Yu M, Takanishi D, Hasaniya NW, Brown MR. Eosinophilia as a marker of adrenal insufficiency in the surgical intensive care unit. J Am Coll Surg. 1996; 183(6):589–596. PMID: 8957461.

23. Beishuizen A, Vermes I, Hylkema BS, Haanen C. Relative eosinophilia and functional adrenal insufficiency in critically ill patients. Lancet. 1999; 353(9165):1675–1676. PMID: 10335792.

24. Miller WL. The hypothalamic-pituitary-adrenal axis: a brief history. Horm Res Paediatr. 2018; 89(4):212–223. PMID: 29719288.

25. Paragliola RM, Papi G, Pontecorvi A, Corsello SM. Treatment with synthetic glucocorticoids and the hypothalamus-pituitary-adrenal axis. Int J Mol Sci. 2017; 18(10):E2201. PMID: 29053578.

26. Gjerstad JK, Lightman SL, Spiga F. Role of glucocorticoid negative feedback in the regulation of HPA axis pulsatility. Stress. 2018; 21(5):403–416. PMID: 29764284.

27. García-Castells AM, Argente-Pla M, García-Malpartida K, Querol-Ripoll R, Merino-Torres JF. Adrenal insufficiency induced by megestrol acetate: report of two cases. Endocrinol Nutr. 2015; 62(10):515–516. PMID: 26250842.

28. González Villarroel P, Fernández Pérez I, Páramo C, Gentil González M, Carnero López B, Vázquez Tuñas ML, et al. Megestrol acetate-induced adrenal insufficiency. Clin Transl Oncol. 2008; 10(4):235–237. PMID: 18411198.

29. Bulchandani D, Nachnani J, Amin A, May J. Megestrol acetate-associated adrenal insufficiency. Am J Geriatr Pharmacother. 2008; 6(3):167–172. PMID: 18775392.

30. Manosroi W, Phimphilai M, Khorana J, Atthakomol P. Diagnostic performance of basal cortisol level at 0900-1300h in adrenal insufficiency. PLoS One. 2019; 14(11):e0225255. PMID: 31738804.

31. Wachter RM. An introduction to the hospitalist model. Ann Intern Med. 1999; 130(4 Pt 2):338–342. PMID: 10068402.

32. Ñamendys-Silva SA, González-Herrera MO, Texcocano-Becerra J, Herrera-Gómez A. Hypoalbuminemia in critically ill patients with cancer: incidence and mortality. Am J Hosp Palliat Care. 2011; 28(4):253–257. PMID: 21057142.

TOOLS

Similar articles