Combination of High Prevalence Sign/Symptom Pairs: An Approach to the Diagnosis of Cushing’s Syndrome

Abstract

Most of the clinical features of Cushing’s syndrome (CS) are nonspecific and could be present in obesity, particularly when this condition is associated with metabolic syndrome. Our objective was to evaluate the frequency of clinical manifestations and changes in general laboratory tests, in patients with confirmed endogenous CS, to identify diagnostic dyads. We evaluated in each patient the rate of coexistence of 2 elements either: symptoms, clinical signs, or laboratory alteration. The prevalence of a combined pair of clinical features or dyad in over 30% of the cases, was considered clinically significant. Fourteen dyads were identified as clinically relevant. Facies + buffalo hump; facies + eosinopenia; buffalo hump + supraclavicular fat pads and facies + supraclavicular fat pads, were present in over 50% of cases. Facies + muscular atrophy; centripetal fat distribution + muscular atrophy and facies + striae were present in 42–49%. Hirsutism/acne + eosinopenia; buffalo hump + eosinopenia; muscular atrophy+ eosinopenia; eosinopenia + accelerated weight gain; buffalo hump + muscular atrophy; hirsutism/acne + muscular atrophy and hirsutism/acne + supraclavicular fat pads, were observed in 33–38% of patients. Its application will facilitate the correct diagnosis of CS.

1. Introduction

Classical manifestations of Cushing’s syndrome (CS) are related to cortisol excess. Frequent physical signs are moon face and plethora, buffalo hump, supraclavicular fat pads, central obesity, width and dark red striae, thin limbs, thin skin, hirsutism, and acne. Frequent symptoms are accelerated weight gain, muscular weakness, amenorrhea, pathological fractures/osteoporosis. In general laboratory tests, alterations secondary to hypercortisolism can be found (examples: hyperglycemia, hypertriglyceridemia, hypokalemia, low eosinophils, and lymphocytes count.

Nevertheless, most of the clinical features of CS are nonspecific and could be found in obesity, particularly, in association with metabolic syndrome. For this reason, diagnosis is often overlooked, and treatment is delayed unless the evaluation is done by an expert endocrinologist.

A meta-analysis published in 2020 (which included 44 studies) showed that mean time delay to diagnosis CS was 34 months. This was shorter in ectopic CS (14 months) compared with adrenal CS (30 months) and pituitary Cushing’s (38 months), probably due to the more aggressive behavior of these neoplasms [1].

For this reason, it seems necessary to establish clinical features of suspicion of this disorder that can be used by non-specialists.

More than fifty years ago, it was reported that in about 50% of patients with suspected CS, the diagnosis could be confirmed or excluded using clinical or general laboratory tests with greater accuracy than steroid screening [2].

Nevertheless, with the increasing prevalence of obesity and its associated metabolic disorders, experts questioned the use of the classical clinical parameters and validated only specific signs related to hypercortisolism (osteopenia, thin skin, and proximal muscular weakness). When they are used to confirm or discard CS diagnosis, the probability is 95% [3, 4].

Over decades, different authors have proposed methods to approach clinical diagnosis of this syndrome (Table 1). In the sixties, Nugent et al. studied the incidence of different signs in patients with and without the syndrome. They reported that the more significant findings were osteoporosis, central or generalize obesity, weakness, plethora, leukocytosis (≥11,000/μL), acne, striae (red or purple), diastolic hypertension (≥105 mmHg), edema, hirsutism, ecchymoses, and low serum potassium (≤3.6 mEq/L). These signs were used to calculate the probability of CS. The results suggested that diagnosis could be confirmed or excluded with a high degree of confidence in at least half of patients with suspected CS [2].

Reference	n° patients	Discriminatory signs/symptoms	Parameter	Estimated value for diagnosis of CS
Nugent, 1964 [2]	38 CS/73 Non-CS	Weakness, ecchymoses, edema,; low serum potassium	Probability	0.9 to ≥ 0.99
Ross, 1982 [4]	70 CS/159 Non-CS	Bruising, myopathy, hypertension, plethora,; edema,; hirsutism, red striae	Discriminant index*	>2
Schneider, 2013 [5]	73 CS/369 Non-CS	Recurrent infections,; red striae, amenorrhea, abdominal fat distribution, plethora,; muscular weakness, hirsutism	Discriminant index**	≥ 2
León-Justel, 2016 [6]	389 at risk of CS	Muscular atrophy, osteoporosis, dorsocervical fat pad	ROC analysis score	4
Loriaux, 2017 [3]	Estimated prevalence 0.2%	Thin skin, osteopenia, ecchymoses	Probability***	95%

Table 1.

Diagnostic models of Cushing syndrome based on prevalent signs and symptoms.

*

Index calculated by dividing the prevalence of each feature in CS by its prevalence in reference [2] of 159 mostly obese patients in whom the diagnosis of Cushing’s syndrome was suspected but not biochemically substantiated.

**

Index calculated by dividing the prevalence in authors series with the prevalence reported in reference [4].

***

Obtained using a probability of 0.2% and a likelihood ratio of 116 for thin skin, 18 for osteopenia, and 4 for ecchymoses

In the early 80s, another paper showed that bruising, muscle weakness, and hypertension were the most discriminating features of CS (discriminant index over 4) [5].

Other authors evaluated the prevalence of signs and symptoms in a series of 73 cases of CS and calculated a discriminant index. The signs and symptoms, with a higher index (2 or greater) were recurrent infections, red striae, amenorrhea, abdominal fat distribution, plethora, muscular weakness, and hirsutism [6].

Another group developed a risk score to predict CS in a population of over 300 subjects at-risk. They propose two models, one based on the assessment of clinical symptoms and signs, and in the other, late-night salivary cortisol (LNSC) determination was added to the clinical features. The multivariate logistic regression analysis showed that muscular atrophy, osteoporosis, and dorsal cervical fat pad remained independent variables associated with CS. The ROC analysis showed that a score of 4 resulted in sensitivity and specificity of 96.2% and 82.9%, respectively. With this cut-off value, 83% of subjects without CS were correctly identified and only one of 26 CS was missed. However, they reported several false positives, mainly related with the LNSC level [7].

Applying technology, in a study of 20 patients with CS, a facial appearance classification software was tested to discriminate patients with Cushing from healthy controls. The software correctly classified 85% of patients and 95% of controls, with a total classification accuracy of 91.7%. Nevertheless, this study only evaluated women, which limited its utility [8].

On the other hand, within general laboratory tests, impaired blood glucose and abnormal lipid levels may be present in a high percentage of the obese population, however, low potassium levels in absence of diuretic use and low eosinophil count may be typical manifestations of hypercortisolism.

Eosinopenia is defined as a reduction of circulating eosinophils < 10/μL or < 0.1% of total leukocyte count. In physiological conditions, during acute stress, eosinopenia is mediated by adrenal glucocorticoids and epinephrine. In patients treated with corticosteroids, eosinopenia would result from an impairment in the release of these cells from the bone marrow and its sequestration of the blood pool [9]. Chronic hypercortisolemia in CS can explain this typical finding which should be taken into account when evaluating a suspected case.

Because laboratory tests performed to confirm the diagnosis of CS are not routinary done and are not widely available, the selection of cases that requires evaluation should be done by a specialist but given a large number of obese patients, this is impracticable. For this reason, combinations of signs, symptoms, and general laboratory tests could be a useful tool for clinicians when CS is suspected.

Our objective was to evaluate the frequency of clinical signs, symptoms, and biochemical alterations in patients treated in a single center with a confirmed endogenous CS and identify dyads of specific symptoms and/or signs.

2. Methodology

This was a retrospective review of clinical records, laboratory tests, radiological exams, and biopsies from patients treated in one university center in the period 1980–2017. We excluded patients under corticosteroid therapy, those whose physical examination was not performed by an endocrinologist and with incomplete records. At least two of the following tests confirmed the diagnosis of CS: free urinary cortisol (FUC), overnight 1 mg Dexamethasone suppression test (Nugent’s test), and LNSC. The etiology of CS was confirmed by the following: plasma ACTH, hormonal functional tests, and radiological exams. Operated patients had a confirmatory biopsy. This protocol was approved by the local scientific Ethics Committee.

We determined patient characteristics, including age, gender, reason for consultation, features at physical examination, and general laboratory tests, establishing the frequency for each parameter.

We evaluated the rate of coexistence of two features either: symptoms, clinical signs, or laboratory alterations in each patient. A prevalence of a combined pair or dyad in over 30% of the cases was considered clinically relevant. Finally, according to expert endocrinologist opinion and based on scientific literature on the subject, we selected diagnostic dyads with the more specific features of hypercortisolism. Statistical analysis was performed using STATA SE v 15.0 software (StataCorp LLC).

2.1 Definitions of some specific signs of hypercortisolism

• Accelerated weight gain: weight increase over 7–10% in less than 3 months.

• Changes in fat distribution: Increase of thoracoabdominal fat (centripetal fat depot) and decrease in buttocks and limbs.

• Buffalo hump: fat accumulation in dorsocervical region.

• Supraclavicular fat pads: fat accumulation in supraclavicular hollows.

• Striae: dark red or violet, ≥ 1 cm width, mainly in abdomen (paraumbilical and flanks), inner thighs, and arms.

• Thin skin/Ecchymoses: translucent appearance, frail / 3 or more ecchymoses.

• Muscular atrophy: decrease in quadriceps, buttocks, and biceps muscle mass on palpation.

• Hirsutism: characteristics and distribution of body hair according to Ferriman Gallwey score [10].

• Eosinopenia: eosinophils count < 10/μL or < 0.1% of total leukocyte count.

3. Results

In the analyzed period, 102 patients, 89 women (87.2%), 36.9 ± 13 years old (15–72 yo), had confirmed CS.

The etiology of CS was pituitary (68%), adrenal (23%), and ectopic ACTH (9%).

Clinical characteristics are shown in Table 2. The more frequent clinical symptoms were accelerated weight increase (57.3%), hirsutism and/or acne (23%), and muscular weakness (22%). At physical examination, the most frequent signs were moon face (90%), centripetal fat depot (78.4%), buffalo hump (65.7%), supraclavicular fat pads (53.9%), hirsutism and/or acne (mainly in women) (65%), systolic and/or diastolic hypertension (60%), muscular atrophy (52%) and striae (42%). At the laboratory tests, we found eosinophils count < 0.1% of total leukocytes in 84.2%.

N	102 (%)
Age (yr)	36.9 ± 13.0
Gender: Females; Males	89 (87.3%); 13 (12.7%)
Reason for consultation
Accelerated weight gain/changes in; fat distribution	59 (57.3%)
Muscular weakness	23 (22.1%)
Menstrual disorders	10 (9.7%)
Hirsutism/acne	24 (23.3%)
Hypertension	10 (9.7%)
Hyperglycemia/diabetes/dislipidemia	8 (7.8%)
Osteoporosis/fractures	5 (4.9%)
Skin alterations	8 (7.8%)
Edema	7 (6.8%)
Neuropsyquiatric	5 (4.9%)
Other	12 (11.7%)
Physical examination
Moon Face	91 (89.2%)
Hirsutism/acne	60 (58.8%)
Buffalo hump	67 (65.7%)
Supraclavicular fat pads	55 (53.9%)
Acanthosis	32 (31.4%)
Centripetal fat distribution	80 (78.4%)
Muscular atrophy	53 (52.0%)
Thin skin/ecchymoses	30 (29.4%)
Striae	43 (42.2%)
Hyperpigmentation	13 (12.7%)
Systolic blood pressure > 140 mm/Hg	62 (62.0%)
Diastolic blood pressure > 90 mm/Hg	57 (57%)
BMI (kg/m^2)
26−29	26 (26.8%)
30−35	27 (27.8%)
> 35	17 (17.5%)
General Laboratory
Eosinophil count <10/μL or <0.1% of; total leukocyte count	64 (84.2%)
Lymphocytes < 1000/μL	12 (18.5%)
Hyperglycemia	43 (50.0%)
Hypertriglyceridemia	22 (55.5%)
Hypokalemia	28 (31.1%)

Table 2.

Clinical features in patients with Cushing’s syndrome.

From 520 combinations of data pairs, we obtained 18 dyads present in over 30% of cases (Table 3). From these, we selected 14 dyads which feature combinations were considered more specific to hypercortisolism. Facies + buffalo hump; facies + eosinopenia; buffalo hump + supraclavicular fat pads and facies + supraclavicular fat pads, were present in over 50% of cases. Facies + muscular atrophy; centripetal fat distribution + muscular atrophy and facies + striae were present in 42–49%. Hirsutism/acne + eosinopenia; buffalo hump + eosinopenia; muscular atrophy+ eosinopenia; eosinopenia + accelerated weight gain; buffalo hump + muscular atrophy; hirsutism/acne + muscular atrophy and hirsutism/acne + supraclavicular fat pads, were observed in 33–38% of patients.

Facies + buffalo hump	60.7%
Facies + eosinopenia	54.9%
Buffalo hump + supraclavicular fat pads	52.9%
Facies + supraclavicular fat pads	50.9%
Facies + muscular atrophy	49.0%
Centripetal fat distribution+ muscular atrophy	43.1%
Facies + striae	42.1%
Hirsutism/acne + buffalo hump	38.2%
Hirsutism/acne + eosinopenia	38.2%
Buffalo hump + eosinopenia	36.2%
Systolic hypertension + eosinopenia	35.2%
Muscular atrophy+ eosinopenia	34.3%
Eosinopenia + accelerated weight gain	34.3%
Buffalo hump + muscular atrophy	33.3%
Diastolic hypertension + eosinopenia	33.3%
Muscular atrophy + systolic hypertension	32.3%
Hirsutism/acne + muscular atrophy	31.3%
Hirsutism/acne + supraclavicular fat pads	30.3%

Table 3.

Diagnostic dyads.

4. Discussion

Cushing’s Syndrome represents a clinical challenge even for specialists. As we mentioned before, many features related to hypercortisolism could be found in other clinical conditions. Some of them correspond to pseudo-Cushing’s states. This condition shares clinical signs of CS and equivocal evidence of hypercortisolism in laboratory tests [11]. One study disclosed that obesity (BMI>30 kg/m²) was significantly more frequent in pseudo-Cushing than in CS and only ecchymoses and osteoporosis were more frequent in CS [12]. Another study showed that obesity, moon face, and buffalo hump were present in over 50% of pseudo-Cushing’s patients [13].

Nonetheless, even in this condition, a meticulous clinical evaluation could improve diagnostic accuracy and the finding of combination of frequent signs, symptoms, and biochemical tests could be the best approach to continue with more specific evaluations.

Interestingly, classical signs of hypercortisolism were present in a high proportion of our cases, as is shown in Table 2. Accelerated weight gain and change in body fat distribution were frequent causes of consultation and feature at physical examination (58% and 78% respectively).

In vitro and in vivo studies have shown that glucocorticoids (GCs) increase hypothalamic endocannabinoids. GCs and cannabinoids have been shown to increase hypothalamic adenine monophosphate protein kinase-A (AMPK) activity, resulting in increased appetite. Moreover, GCs have been shown to upregulate gene expression of orexigenic peptides such as Y neuropeptide (NPY) and agouti-related peptides. This could explain cravings for fat and carbohydrates observed in endogenous GCs excess states as Cushing’s disease [14].

Some studies in human healthy controls or subjects with inflammatory conditions evaluated the effect of short-term use of oral GCs on energy intake, body weight, body composition, or appetite but, results have not been conclusive [15].

Still, more studies are necessary to establish the effects of long-term use of oral GCs and endogenous hypercortisolism, on appetite and dietary intake.

On the other hand, changes in fat redistribution are determined by chronic hypercortisolism, leading to an increase in centripetal adiposity and visceral abdominal fat. Regulation of adipose tissue mass is complex and involved mechanisms are multiple and still not completely understood. In general, the effects of GCs on adipose tissue depend on the duration of the exposure to GCs and on the type of adipose tissue considered (subcutaneous or visceral) [16]. Visceral fat would be differentially responsive to GCs than subcutaneous fat. In experimental models, GCs induce the differentiation of preadipocytes, specific to visceral fat, but not for subcutaneous fat. Glucocorticoid receptor expression is higher in visceral than in peripheral subcutaneous fat. Chronic corticoid exposure increases lipogenesis in visceral adipose compartment and increases lipolysis in subcutaneous adipose tissue [17]. Also, there is a higher cortisol production in visceral adipose tissue, due to an increased expression of GC receptors and 11B-hydroxysteroid dehydrogenase type 1 (11BHSD1) enzymatic activity. The 11BHSD1 is widely expressed throughout the body, including liver, visceral and subcutaneous fat, and promotes the conversion of inactive cortisone to cortisol. These mechanisms among others, explain the characteristic phenotype of CS [18].

Another relevant clinical feature is muscle involvement. In our case series, muscle atrophy was present in 52% of cases. In previous reports, the prevalence of muscle weakness or atrophy was 20–45% [6, 7, 12]. Chronic hypercortisolism generally has a more prominent effect on the proximal muscles. GC-induced muscle atrophy affects mainly fast-twitch or type II fibers with less or no impact on type I or slow-twitch fibers. This explains the characteristic findings at physical examination of thin limbs with loss of muscle mass.

GCs reduce skeletal muscle mass both by inhibiting protein synthesis and by increasing the rate of protein degradation. GCs stimulate myostatin, an inhibitory growth factor that downregulates protein synthesis and also, proliferation and differentiation of muscle satellite cells, precursors of skeletal muscle cells. GCs inhibit the transport of amino acids into the muscle and interfere with the stimulatory action of insulin and Insulin-like Growth Factor 1 (IGF1) on the protein synthesis pathways. Moreover, GCs inhibit the production of IGF1 in muscle, which contributes to muscle proteolysis and apoptosis [18]. On the other side, GCs can indirectly affect skeletal muscle by downregulating gonadal function and reducing the expression of the androgen receptor. This would explain the authors’ finding of high frequency of myopathy in males compared with females (65% vs 45%) [19].

Recently, some researchers compared muscle mass and muscle function in patients with CS and matched obese controls. The CS group showed a significant decrease in muscle function tests (chair rising time and hand grip strength) versus the obese non-CS group. Interestingly, both groups did not show differences in muscle mass, fat mass and waist-to-hip ratio, suggesting that CS could be associated with impaired muscle quality and functional alterations more than in trophic effects [20]. Functional tests could be other clinical tool to objectify the diagnosis of CS.

It is interesting that skin manifestations were not observed with high frequency in our cases (only 29%), as has been reported previously [21]. In that report, they used a caliper to measure the skinfold over the proximal phalanx of the middle finger of the nondominant hand. Our diagnosis was based on observation only. Also, we must consider that our patients have different ethnic backgrounds.

Another issue observed in CS patients is metabolic complications, as hyperglycemia, dyslipidemia and higher cardiovascular risk. These are due to several GCs related actions on the liver, skeletal muscle, pancreas, and adipose tissue [see Ref. 22]. In omental, but not in subcutaneous adipocytes, GCs induce insulin resistance by increasing free fatty acid, inducing deregulation in adipokines secretion, and increasing leptin and resistin levels. IL6 and TNFα receptor 1 are also increased in CS patients compared to BMI-matched controls. In some series, a prevalence of diabetes between 15–45% was reported [6, 7, 12, 19] and 46% of dyslipidemia [7]. In our series, 50% had hyperglycemia and hypertriglyceridemia. It should be mentioned that not all cases had a glucose tolerance test, so this probably underestimated the frequency of hyperglycemic disorders.

A remarkable finding observed, is a decrease in eosinophil count. Eosinopenia/aneosinophilia determined by excess of GCs. has been described for more than 6 decades. Patients with CS may present leukocytosis and neutrophilia and a low lymphocyte and eosinophil count or eosinopenia [9, 23, 24]. In the absence of exogenous corticosteroid use, eosinopenia would be the most specific laboratory finding. Patients with pseudo-Cushing, present various clinical signs and even altered laboratory tests such as FUC or Nugent’s test, but do not have eosinopenia [11]. We found an eosinophil count < 0.1% of total leukocyte count in over 80% of cases, unlike lymphopenia, observed only in 18% of patients. Then, we consider eosinopenia a significant diagnostic element.

The association of eosinopenia with signs such as hypertension or hirsutism, which are observed commonly in other pathologies with similar characteristics to CS, like polycystic ovary syndrome (PCOS), leads toward the diagnosis of Cushing’s. The combination of hirsutism/acne + eosinopenia and systolic hypertension + eosinopenia were prevalent in CS (38% and 35%).

Decreased bone mass density was frequent in other series [7, 12]. But, in our patients we found osteoporosis only in 5%. However, it should be noted that this was due to the fact that bone densitometry was performed in a minority of cases. Then, this sign was underestimated in our cases and always should be considered if it is present together with any of the diagnostic dyads.

5. Conclusion

We present a casuistry with a large number of patients with a confirmed Cushing’s syndrome which makes our findings very significant. The combinations of high-risk clinical pairs or dyads associated with determined disease have been used for the identification of patients at risk of other pathology such as acromegaly [25].

The presence of dyad combinations of clinical features may be a useful tool in a practical clinical setting, to assist physicians in identifying patients at risk of CS, without requiring complex or expensive tests.

With training, that allows recognizing the specific signs of hypercortisolism and general laboratory tests, the use of these sign/symptom dyads (shown in Table 3), will facilitate that the correct diagnosis of CS can be reached with high certainty.

An early diagnosis and treatment of Cushing syndrome, most probably will reduce the severe comorbidities associated with this condition and will improve the prognosis of these patients.

Acknowledgments

To the neurosurgery team of the University of Chile Clinical Hospital, who referred many patients for evaluation.

Conflict of interest

The authors declare no conflict of interest.