Refractory hypokalaemia and hypertension with metabolic alkalosis: an acute presentation of Cushing’s disease secondary to a pituitary macroadenoma

Abstract

A 68-year-old woman presented with right arm cellulitis, not responsive to oral antibiotics. Intravenous antibiotics were commenced, and an ultrasound scan confirmed a collection that was surgically drained. She developed refractory hypokalaemia with normal magnesium, no gastrointestinal losses and no iatrogenic cause. She was hypertensive, hyperglycaemic, alkalotic, clinically obese with proximal myopathy and skin bruising. These clinical findings and refractory hypokalaemic hypertension with metabolic alkalosis raised a suspicion of Cushing’s syndrome (CS). 24-hour urinary free cortisol (24 hours) was grossly raised on two occasions. The adrenocorticotropic hormone (ACTH) was significantly raised at 154 ng/L, confirming ACTH-dependant CS. A CT scan of the thorax, abdomen and pelvis excluded an ectopic source of hypercortisolaemia. MRI pituitary revealed an invasive macroadenoma. Treatment with endoscopic debulking resulted in the resolution of hypokalaemia and metabolic alkalosis with significant improvement in hyperglycaemia and hypertension.

Background

Endogenous Cushing’s syndrome (CS) is a rare disorder with an incidence of ~2.4 cases per million population year and is more common in women. Suboptimal management increases the risk of mortality by five-fold, and therefore, early recognition and treatment are of paramount importance.^{1 2} CS results from chronic exposure to excessive glucocorticoids and is most commonly (~70%) caused by Cushing’s disease (CD), which occurs due to overproduction of adrenocorticotropic hormone (ACTH) by a pituitary corticotroph adenoma.^{1 3} The vast majority of ACTH-secreting pituitary adenomas are ≤10 mm in diameter (microadenomas). However, ~10% of them are ≥10 mm in diameter (macroadenomas).⁴ Hypokalaemia, defined as a serum potassium level of <3.5 mmol/L, is more common with ectopic CS than CD.⁵ A pituitary macroadenoma and hypokalaemia are uncommon with CD; however, the authors describe both of these atypical findings in this case.

Case presentation

A 68-year-old woman with a background of primary hypothyroidism, gout, hypertension, type 2 diabetes mellitus of recent onset (diet-controlled), gout and cholecystectomy was referred to the plastic surgery team with a 2-week history of right hand and forearm cellulitis, which was not responding to oral flucloxacillin. She was taking levothyroxine 100 µg once a day (OD), atenolol 50 mg OD, ramipril 10 mg OD, allopurinol 300 mg OD, lacidipine 4 mg OD and simvastatin 20 mg OD. She was a non-smoker and did not consume alcohol. On review, there was evidence of a tender, erythematous, warm and swollen right hand and digits with proximally extending erythema and swelling involving the forearm. There was no history of trauma or penetrating injury. There was no history of fever, nausea or sweating. There was evidence of a fluctuant swelling on the dorsum of the right hand and forearm with no evidence of crepitus on clinical examination. Her baseline blood tests (summarised in table 1) showed hypokalaemia, raised inflammatory markers, metabolic alkalosis and grossly elevated haemoglobin A1c (HbA1c), suggesting poor glycaemic control.

Table 1.

Baseline blood test results

Blood test	Result	Normal range
Sodium	137 mmol/L	135–144
Potassium	2.8 mmol/L	3.5–5.3
Urea	3.6 mmol/L	3.0–7.6
Creatinine	42 µmol/L	55–87
CRP	84 mg/L	0–8
Haemoglobin	128 g/L	120–160
WCC	13.6×109/L	4.0–11.0
Platelet count	437×109/L	150–400
MCV	90.6 fL	80–100
PT	11.6 s	10.1–13.3
APTT	21.6 s	28.0–36.0
HbA1c	158 mmol/mol	<48
Adjusted calcium	2.26 mmol/L	2.20–2.60
Phosphate	0.66 mmol/L	0.8–1.5
Magnesium	0.82 mmol/L	0.7–1.0
Serum osmolality	293 mOsm/kg	275–295
pH	7.50	7.35–7.45
Bicarbonate	33 mmol/L	22–26
Base excess	11 mmol/L	+2 to –2
Lactate	1.3 mmol/L	0–1.3

* Significant abnormal baseline blood test results are highlighted in bold text. (1) A raised pH with an elevated bicarbonate level and base excess suggests metabolic alkalosis which is associated with hypokalaemia. (2) Raised inflammatory markers are consistent with infection. (3) In the clinical context, an elevated HbA1c level suggests presence of uncontrolled type 2 diabetes mellitus.

APTT, activated partial thromboplastin time; CRP, C reactive protein; HbA1c, haemoglobin A1c; MCV, mean corpuscular volume; PT, prothrombin time; WCC, white cell count.

An ultrasound scan of the right arm confirmed a subcutaneous loculated collection in the region of swelling. Blood cultures grew Staphylococcus aureas which was sensitive to flucloxacillin. The patient was commenced on intravenous flucloxacillin, oral potassium replacement, metformin and the collection was surgically drained under a general anaesthetic. Postoperative progress was satisfactory apart from generalised weakness and intermittent brief episodes of delirium, which responded well to parenteral haloperidol. She then developed persistent hypokalaemia in the range of 1.9–2.8 mmol/L (3.5–5.3) despite adequate intravenous potassium replacement. Her magnesium level was normal, there were no gastrointestinal losses, and no iatrogenic causes could be isolated. Random urinary potassium was 70 mmol/L (~25 mmol/L), which suggested urinary potassium loss.

The capillary blood glucose (CBG) was between 15 and 23 mmol/L. To control the CBG, in addition to metformin, insulin humulin I 40 units daily was commenced, and an Endocrinology review was sought. She was reviewed by the Endocrinology team who, in addition to refractory hypokalaemia and hyperglycaemia, noted persistent metabolic alkalosis on venous blood gas analysis and hypertension with the systolic blood pressure (BP) in the range of 180–210 mm Hg. Clinically, there was evidence of central obesity with multiple non-specific abdominal striae, bilateral pedal oedema, proximal myopathy and skin bruising (in the absence of antiplatelets and anticoagulants). Visual field examination was normal. There was no history of headache, visual disturbances or nausea. There was no history of fractures.

Because of refractory hypokalaemic hypertension with metabolic alkalosis, the differential diagnosis was between Cushing’s, Conn’s and Liddle syndrome. Plasma aldosterone was suppressed at <55 pmol/L (100–850), and plasma renin activity was borderline low-normal at 0.5 nmol/L/h (0.5–3.5), and thus Conn’s syndrome was excluded. Acute presentation, patient age group, absence of family history and clinical features suggestive of hypercortisolaemia went against a diagnosis of Liddle syndrome. Two 24 hours—urinary free cortisol (UFC) collections returned grossly abnormal at 5381 nmol/24 hours (11–110) and 4626 nmol/24 hours (11–110) with a urine volume of 2955 mL/24 hours and 2210 mL/24 hours, respectively. Random serum cortisol was 1229 nmol/L, 1314 nmol/L and 1441 nmol/L (200–700) on three separate occasions. This confirmed hypercortisolaemia. She was not on any short-term or long-term exogenous steroids or taking any estrogen-containing preparations. An ACTH level was done to assess the source of excess cortisol, which came back elevated at 154 ng/L (0–46), confirming ACTH-dependant CS. The next step was to assess whether the hypercortisolaemia was from an ectopic source in the body or due to over-secretion from a pituitary adenoma. A CT thorax, abdomen and pelvis (TAP) excluded an ectopic source of cortisol excess. MRI scan of the pituitary gland revealed evidence of a 16 mm anteroposterior (AP) × 12 mm transverse (TR) × 16 mm craniocaudal (CC), lobulated, poorly enhancing pituitary adenoma invading the right cavernous sinus and protruding into the right shenoid sinus (figure 1). At this stage, usually, an inferior petrosal sinus sampling (IPSS) is done to differentiate CD from ectopic CS. Because of clear clinical and biochemical evidence of hypercortisolaemia in the absence of exogenous steroids, no ectopic source found on CT TAP and a clear invasive pituitary macroadenoma isolated on MRI pituitary gland, a diagnosis of CD was made, and IPSS was not offered. The pituitary multidisciplinary team, which included endocrinologists, neurosurgeons and radiologists, felt that due to the reasons mentioned above, IPSS would not provide any further information and the risk of the procedure outweighed the benefits.

Figure 1.

MRI pituitary gland T2 weighted image in the coronal view showing an invasive pituitary macroadenoma (arrow).

To control hypercortisolism, the patient was commenced on metyrapone, which is an aromatic ketone that inhibits glucocorticoid synthesis in the adrenal cortex. By competitively inhibiting the enzyme 11-beta-hydroxylase, metyrapone blocks the conversion of 11-deoxycortisol to cortisol and deoxycorticosterone to corticosterone, respectively.⁶ The introduction of metyrapone stabilised the potassium levels between 3.3 and 3.7 mmol/L (3.5–5.3), 24 hours–UFC fell to 107 nmol/24 hours (11–110) and insulin requirements fell to humulin I 20 units daily.

Because of an acute presentation of CD with refractory hypokalaemia, hyperglycaemia, hypertension and severe infection, the patient underwent endoscopic trans-nasal trans-sphenoidal (TSS) debulking of the tumour on an urgent inpatient basis (preoperative pituitary function tests summarised in table 2).

Table 2.

Baseline preoperative pituitary function tests

Blood test	Result	Normal range
Insulin-like growth factor 1	9.3 nmol/L	4–23
Prolactin	270 mU/L	70–550
ACTH	154 ng/L	0–46
Cortisol (random)	1229 nmol/L	200–700
Thyroid-stimulating hormone (TSH)	0.29 mU/L	0.35–4.7
Free thyroxine	13 pmol/L	7.8–21
Follicle-stimulating hormone	0.50 iU/L	3–9
Luteinising hormone	<0.10 iU/L	2–11

Significance: (1) in the absence of exogenous steroid use, raised serum adrenocorticotropic hormone (ACTH) and cortisol levels, when interpreted in the context of grossly elevated 24 hours—urinary free cortisol levels, suggest ACTH dependent Cushing’s syndrome (CS) (serum cortisol is not recommended in screening for CS). (2) Normal prolactin level is reassuring as this suggests that the tumour is not a prolactinoma (which should be treated medically with dopamine agonist therapy initially). (3) Low TSH may be due to acute illness, transient loss in thyrotrophic activity due to a pituitary mass (which would also explain the low gonadotropin levels) and in this scenario, a low TSH can also be explained by cortisol excess itself.

On the third postoperative day, serum cortisol was 206 nmol/L, the systolic BP improved to 125–145 mm Hg (ramipril and lacidipine were stopped and atenolol was switched to bisoprolol 1.25 mg OD), serum ACTH 18 ng/L (0–46), potassium stabilised between 3.7 and 4.9 mmol/L (3.5–5.3), bicarbonate improved to between 26 and 28 mmol/L (22–26) and the insulin requirements continued to fall. The patient was switched to Humulin M3 10 units (morning) and 8 units (evening), maintaining the CBG between 6 and 12 mmol/L. The metyrapone was stopped, and hydrocortisone 20/10/10 mg was commenced. The histology of the pituitary mass revealed an adenoma with ACTH and growth hormone expression with a Ki-67 index of 15%, suggesting a potential risk of tumour recurrence and hence a need for close follow-up.

Differential diagnosis

In view of refractory hypokalaemic hypertension with metabolic alkalosis, the differential diagnosis included Cushing’s, Conn’s and Liddle syndrome. Based on the clinical findings of central obesity, proximal myopathy, skin bruising, hypertension and severe infection, CS was considered more likely. Conn’s syndrome was excluded with a low aldosterone/renin ratio, while the possibility of Liddle syndrome was discounted based on patient age group, negative family history and acute nature of the presentation. Grossly elevated 24 hours—UFC, in the absence of exogenous steroids, confirmed hypercortisolaemia, while a raised ACTH level suggested ACTH-dependent CS. CT TAP excluded an ectopic source of ACTH secretion. MRI pituitary confirmed an invasive pituitary macroadenoma as the likely source of ACTH-dependent CS, and therefore an IPSS was not done (figure 2).

Figure 2.

A diagrammatic illustration of the Endocrinology team’s real-time stepwise thought process, which led to the diagnosis of Cushing’s disease in this case. ACTH, adrenocorticotropic hormone; UFC, urinary free cortisol.

Outcome and follow-up

Four months postoperatively, a short Synacthen test (SST) revealed a baseline cortisol T⁰ 30 nmol/L and peak cortisol T³⁰ 174 nmol /L suggesting a need to continue hydrocortisone replacement therapy. (The SST was done after the last dose of hydrocortisone was administered on the previous evening at 1600 hours and omitted on the morning of the test. After Synacthen administration, the cortisol should rise by 190 nmol/L from the baseline level to a peak level of at least 540 nmol/L.) Hydrocortisone was continued but tapered down to 10/10/5 mg. The HbA1c improved from 158 mmol/L (<48) to a range between 52 and 55 mmol/mol (<48) on metformin and insulin humulin M3. The 24 hours—UFC was 43 nmol/24hours, and an MRI pituitary gland showed no significant residual tumour (figure 3). A graphical representation of serum cortisol and 24 hours—UFC, serum potassium level and serum bicarbonate level is shown in figure 4, depicting a change in these parameters following endoscopic TSS. The patient remains under Endocrinology follow-up.

Figure 3.

MRI pituitary T2-weighted image in the coronal view showing the postoperative appearance of the pituitary fossa. No significant residual tumour seen (arrow).

Figure 4.

(A) graphical representation of serum cortisol and 24 hours—urinary free cortisol (UFC) levels. The arrow depicts the point when the endoscopic TSS was done and shows a drop in serum cortisol and 24 hours—urinary free cortisol (UFC) levels leading to the resolution of hypercortisolaemia. (B) graphical representation of serial serum potassium levels. The arrow depicts the point when the endoscopic TSS was done and shows the potassium levels rising thereafter leading to resolution of hypokalaemia. (C) graphical representation of serial serum bicarbonate levels. The arrow depicts the point when the endoscopic TSS was done and shows the bicarbonate levels falling thereafter leading to the resolution of metabolic alkalosis.

Discussion

Hypokalaemia is a commonly encountered electrolyte abnormality in clinical practice and is defined as a serum potassium level of <3.5 mmol/L.⁷ In an outpatient population undergoing blood tests for various causes, 14% will have mild hypokalaemia, whereas clinically significant hypokalaemia is seen in approximately 4% of the hospitalised patients.^{8 9} Inadequate dietary potassium intake, excessive urinary or gastrointestinal losses and intracellular shift are among the main mechanisms leading to hypokalaemia.¹⁰ In the majority of the cases, hypokalaemia results from urinary loss secondary to medication, for example, diuretics and gastrointestinal loss due to diarrhoea and vomiting.^{10 11} Hypokalaemia is usually asymptomatic; however, it can become symptomatic when potassium is <3.0 mmol/L, and the severity depends on the duration and rate at which this occurs. Mild hypokalaemia can result in weakness, whereas severe hypokalaemia can result in fatal cardiac arrhythmias.^{9 12} Endocrine causes of hypokalaemia (table 3) result from excessive urinary potassium loss.⁹

Table 3.

Some important endocrine causes of hypokalaemia

Pathology	Blood pressure	Aldosterone: Renin ratio	Acid/base disturbance
Primary hyperaldosteronism			Metabolic alkalosis
Cushing’s syndrome			Metabolic alkalosis
Liddle syndrome			Metabolic alkalosis
Apparent mineralocorticoid excess			Metabolic alkalosis
Glucocorticoid resistance			Metabolic alkalosis
11-beta hydroxylase deficiency			Metabolic alkalosis
17-alpha hydroxylase deficiency			Metabolic alkalosis
Bartter syndrome			Metabolic alkalosis
Gitelman syndrome			Metabolic alkalosis

Key: high ; low ; normal.

Hypokalaemia is much more common with ectopic CS compared with CD.^{5 13} In a retrospective study of 195 patients with CD, Fan et al studied the association of hypokalaemia with cortisol and ACTH levels and found that cortisol levels in CD were negatively correlated with the levels of plasma potassium while there was no significant correlation between plasma ACTH and potassium levels.¹³

Cortisol and aldosterone have an equal affinity to the mineralocorticoid receptor (MR); however, circulating plasma cortisol levels are 100 times higher than aldosterone. 11β-hydroxysteroid dehydrogenase (11β-HSD2) converts the active cortisol to receptor-inactive cortisone and therefore protects the MR from cortisol mediated activation. In CD, a surge in plasma cortisol may overwhelm the ability of 11β-HSD2 to convert it to the inactive cortisone and therefore, cortisol mediated activation of MR results in excessive urinary potassium loss leading to hypokalaemia and sodium and water retention leading to hypertension. MR activation also leads to renal tubular excretion of hydrogen ions (bicarbonate retained), leading to metabolic alkalosis. There is ample literature suggesting that 11β-HSD2 is inhibited by ACTH and in cases of ACTH-dependent CS, excess cortisol can activate the MR receptor and exert the above described effects.^{11 13 14}

CS presents a diagnostic challenge to Endocrinologists across the globe as its presentation varies widely; correct implementation and interpretation of diagnostic tests are required to establish the diagnosis while being careful of the many pitfalls of the investigations used.

There are numerous overlapping signs, and symptoms of CS observed in the general population with other health conditions, for example obesity. However, the presence of facial plethora, proximal myopathy, easy bruising, reddish-purple >1 cm striae and weight gain with reduced growth velocity in children is highly suggestive of underlying CS.¹⁵ In one series of 423 patients with CS, truncal obesity was the most common feature, present in 97% of the cases, whereas moon facies, hypertension, skin bruising and glucose intolerance were seen in ≥70% of the cases.¹⁶ Once there is clinical suspicion of CS, clinicians should exclude history of oral, topical or inhaled steroid use. The use of certain medication can interfere with the evaluation of diagnostic tests for CS, which include CYP3A4 inducers (false positive), CYP3A4 inhibitors (false negative) and drugs increasing cortisol binding globulin, which may falsely elevate cortisol. The first step is to establish hypercortisolaemia and the Endocrine Society Guidelines recommend one of the following tests: at least two measurements of 24 hours—UFC or late-night salivary cortisol, 1 mg overnight dexamethasone suppression test (1 mg ODST), or the low-dose DST. In case of an abnormal result, another recommended test should be done to confirm hypercortisolaemia and clinicians should exclude physiological and other conditions as a cause of hypercortisolaemia in the absence of CS. These include depression, alcohol excess, malnutrition, anorexia nervosa, pregnancy, chronic intense exercise and uncontrolled diabetes, among others.^{15 16} In addition to 24 hours—UFC, random cortisol was incidentally completed in our case which was grossly elevated but it is not recommended in the evaluation of CS as it can be normal and is influenced by many factors, including acute illness (infection). A further confirmatory test to establish hypercortisolaemia was not done in our case as there was overwhelming evidence of the presence of cortisol excess (in the absence of exogenous steroids/oestrogen use) in the form of clinical features, hypokalaemic hypertension with metabolic alkalosis, grossly raised 24 hours—UFC levels, persistently raised serum cortisol and grossly raised ACTH levels. Furthermore, there was an urgent clinical need (persistent hypokalaemia/intermittent delirium) to control the hypercortisolaemia, therefore metyrapone therapy was commenced. Any confirmatory testing for hypercortisolaemia while on metyrapone would have yielded inaccurate results.

Once hypercortisolaemia is established, serum ACTH can differentiate between ACTH-dependant and independent CS, where the former can result from pituitary or ectopic source and the latter may be from unilateral or bilateral adrenal pathology. Adrenal imaging can identify an adrenal source of cortisol excess. If an ACTH-dependent CS is being considered, as in our case, a CT TAP is useful to investigate a potential ectopic source, and if negative, MRI pituitary gland can isolate a central cause. Often, lesions are so discreet that imaging is unreliable in locating them and hence, IPSS is useful to differentiate a central from an ectopic cause of ACTH-dependent CS.1 3 15 16 Furthermore, a high dose dexamethasone suppression test or a peripheral corticotropin-releasing hormone stimulation test may also help distinguish central (ie, Cushing disease) from an ectopic cause of ACTH-dependent CS.¹⁶

However, as described, this was not required in our case as hypercortisolaemia was robustly established (absence of interfering medication, physiological or clinical conditions), negative CT TAP and a clear-cut pituitary macroadenoma on MRI.

This case illustrates the importance of a stepwise approach in dissecting the cause of refractory hypokalaemia in clinical practice as the underlying cause may be rare, and if missed, can have catastrophic clinical consequences for the patient. CS is a rare cause of hypokalaemia but should be screened for in cases of refractory hypokalaemia.

Patient’s perspective.

We asked the patient about her experience. This was a rapidly evolving and concerning set of circumstances for the patient. She felt unwell and the list of diagnoses seemed to be piling up while the doctors searched for a cause of all the problems. She was relieved once the diagnosis of Cushing’s disease was made but there were apprehensions with regards to the impending surgery and postoperative recovery. She had an extended inpatient stay and as she was being looked after in a hospital away from her hometown, this took its toll. After the operation and in the follow up period she felt very well in herself and noticed a clear improvement in her health.

Learning points.

• In cases of refractory hypokalaemia, authors suggest screening for hypercortisolaemia.

• While screening for hypercortisolaemia, ensure patients are not on exogenous steroids and estrogen-containing preparations as this will skew the results.

• Although hypokalaemia is more common with ectopic Cushing’s syndrome while Cushing’s disease (CD) is usually caused by a pituitary microadenoma, this case highlights that CD can occur secondary to an invasive pituitary macroadenoma and result in persistent hypokalaemia.

• As described in this case, adrenolytic therapy (Metyrapone) effectively controls hypercortisolaemia leading to improvement in various biochemical disturbances and where surgery is required urgently, it is useful as bridging therapy. However, it is vital to establish the diagnosis reliably before commencing adrenolytics as premature therapy will interfere with diagnostic evaluation by blocking adrenal steroidogenesis.

Ethics statements

Patient consent for publication

Obtained.