Non-cirrhotic hyperammonaemia: are we missing the diagnosis?

Abstract

Hepatic encephalopathy secondary to hyperammonaemia is a known complication of chronic liver disease. In contrast, non-cirrhotic hyperammonaemia is a lesser-known entity that should be considered in a patient with acute encephalopathy as part of the diagnostic workup as prompt identification can help to avoid complications such as seizures and cerebral oedema. We present a case of a middle-aged woman who presented electively for a total pancreatectomy–duodenectomy with splenectomy, hepatico-jejunostomy, gastro-jejunostomy and developed encephalopathy on postoperative day 10 due to non-cirrhotic hyperammonaemia.

Background

Non-cirrhotic hyperammonaemia is an unusual cause of encephalopathy. The presentation of this condition is very variable and ranges from acute confusion to unconsciousness. Due to the heterogeneity of presentation, it may be missed by unsuspecting clinicians, thereby delaying prompt treatment and contributing to poor patient outcomes. Non-cirrhotic hyperammonaemia has largely been associated with enzyme deficiencies in the urea cycle¹ and drugs such as sodium valproate and 5-fluorouracil.² The main surgery described in conjunction with this condition is Roux-en-Y gastric bypass surgery³ performed for morbid obesity, and no cases have been reported following pancreatico-duodenectomy surgery. Here, we present an interesting case of non-cirrhotic hyperammonaemia accounting for acute encephalopathy post pancreatico-duodenonectomy surgery and review the literature with regards to this condition.

Case presentation

A 67-year-old Malay woman, with known type 2 diabetes mellitus, hypertension and dyslipidaemia, was diagnosed to have biopsy-proven large pancreatic serous cystadenoma measuring 17.5×14.0 cm with a superior mesenteric vein, inferior vena cava and proximal main portal vein compression. There was no biliary dilatation and the superior mesenteric artery was patent. She was electively admitted for angioembolisation prior to total pancreatectomy–duodenectomy with splenectomy, hepatico-jejunostomy, gastro-jejunostomy on 16 November 2018. This was complicated by thrombosis of the superior mesenteric vein and portal vein with mesenteric congestion and bowel oedema. On postoperative day 3, she developed worsening lactic acidosis, suspicious for mesenteric ischaemia in view of the venous thromboses, and underwent an exploratory laparotomy and washout, which was negative. As the bowels appeared healthy, the decision was made not for open thrombectomy, and heparin anticoagulation was started instead. Thereafter, the patient improved clinically, and enteral feeds were escalated.

Unfortunately, on postoperative day 10, she was noted to have an acute drop in Glasgow Coma Scale (GCS) from 15 to 6 (E3V2M1). Bilateral pupils remained brisk and reactive to light, gaze was central and motor reflexes were normal. She was emergently intubated for airway protection and transferred to the intensive care unit (ICU).

CT of the brain did not reveal any acute intracranial event (figure 1). MRI scan of the brain, however, showed bilateral pontine and thalamic lesions, likely metabolic (figure 2). The CT scan of abdomen/pelvis was negative for rim-enhancing collections but showed worsening of left portal vein thrombosis and stable thromboses of the main portal vein and superior mesenteric veins. Her initial laboratory results are shown in table 1. Her liver function test remained normal with downtrending of the Gamma-Glutamyl Transferase (GGT) to 104 U/L on discharge from the ICU. The highest ammonia levels were on admission (257 μmol/L) and this improved to 47 μmol/L with lactulose and opening of bowels (figure 3). She was also treated with intravenous piperacillin–tazobactam for pseudomonas wound infection.

Figure 1.

CT brain on day 1 of intensive care unit admission—unremarkable.

Figure 2.

MRI brain diffusion-weighted imaging (DWI) sequence showing bilateral thalamic and pontine lesions.

Table 1.

Investigations

Investigation (units)	Value (normal ranges)	Investigation (units)	Value (normal ranges)
Haemoglobin (g/L)	92 (120–160)	pH	7.44 (7.35–7.45)
WBC count (×109/L)	22.4 (4.0–10.0)	pCO2 (mm Hg)	25 (35–45)
Platelets (×109/L)	448 (140–440)	pO2 (mm Hg)	95 (75–100)
Sodium (mmol/L)	127 (135–145)	HCO3 (mmol/L)	18 (21–27)
Potassium (mmol/L)	4.8 (3.5–4.5)	BE (mmol/L)	−7 (−2 to 2)
Urea (mmol/L)	16.1 (2.0–6.5)	Lactate (mmol/L)	2.2 (0.5–2.2)
Creatinine (mmol/L)	86 (40–75)	Albumin (g/L)	34 (38–48)
aCa (mmol/L)	2.29 (2.15–2.50)	Ammonia (umol/L)	257 (16–53)
Mg (mmol/L)	1.2 (0.7–1.0)	TFT	Normal
Phosphate (mmol/L)	0.7 (0.8–1.4)	LFT (U/L)	Normal except for GGT 126 (10–80)

aCa, adjusted calcium to albumin; BE, base excess; GGT, Gamma-Glutamyl Transferase; LFT, liver function test; TFT, thyroid function test.

Figure 3.

Trend of serum ammonia levels over the patient’s stay. ICU, intensive care unit.

Differential diagnosis

The initial concern when the patient presented with an acute drop in GCS was to rule out an acute intracranial event. This was evaluated with radiological imaging of the brain which was fortunately negative. We also wanted to ensure that she did not have septic encephalopathy from the wound infection and hence performed blood investigations and cultures to assist with the diagnosis. Lastly, metabolic causes would also need to be excluded such as electrolyte abnormalities and ammonia levels, of which the latter was significantly elevated.

Treatment

The possible aetiologies for the elevated ammonia levels were investigated. Her drug history was reviewed, but no inciting agents such as diuretics, isoniazid, antiepileptics or chemotherapeutic medications had been taken. Anaesthetic agents causing hyperammonaemia were unlikely as her last surgery was a week from ICU presentation. While she had been receiving total parenteral nutrition since postoperative day 3, this was at a constant protein concentration of 1.3 g/kg and was continued throughout her ICU stay, hence making this a less likely cause. Another possibility would be that of sepsis-induced hyperammonaemia as she had concomitant pseudomonas wound infection and an elevated total white cell count. (This differential was subsequently determined to be of low possibility as there was no evidence of sepsis during her readmission to ICU.)

Three days after admission to the ICU, her GCS score improved to E4VtM6, after having six episodes of large bowel movements. A retrospective review of her notes showed that she had not emptied her bowels for 5 days before ICU admission.

Outcome and follow-up

She was extubated on her fourth day of ICU stay and discharged on day 6. Unfortunately, 4 months later, she was readmitted to the ICU for encephalopathy and an elevated ammonia level of 173 umol/L; her symptoms again resolved with bowel clearance, and she was discharged well on lactulose of 10 mL once daily.

Discussion

This case illustrates an uncommon metabolic cause of encephalopathy which should be considered as a differential in patients following upper gastrointestinal/hepatic surgeries, particularly those complicated by portal vein thrombosis, as this can lead to impaired metabolism of ammonia, hyperammonaemia and confusion.

Endogenous ammonia production is primarily in the gastrointestinal tract, where protein is broken down by urease amino acid oxidase to ammonia and from protein catabolism. Less significant sources of ammonia are from urea-splitting Gram-negative bacteraemia, or the breakdown products of blood in the gastrointestinal tract. Ammonia is absorbed into the systemic circulation and normally enters the liver to be metabolised via the Krebs cycle to form urea, which is then excreted by the kidneys. A small amount of ammonia is taken up and broken down by skeletal muscles. In cases where a portacaval shunt exists (eg, portal vein thrombosis, chronic liver disease), the ammonia in the systemic circulation is shunted away from the liver and leads to an elevated ammonia level in the blood. The high ammonia in the brain contributes to astrocyte swelling, cerebral oedema and confusion. Clinically, symptoms may manifest as vomiting, lethargy, respiratory alkalosis which can evolve to metabolic acidosis when haemodynamic instability occurs, seizures and encephalopathy.

The causes of hyperammonaemia can be grossly divided into gastrointestinal/renal related dysfunction or extraneous sources. In the former, these include congenital causes such as urea cycle defects, gastrointestinal haemorrhage, renal disease, uretero-sigmoidostomy, porto-systemic shunting, urease-producing organism resulting in urinary tract infection as well as systemic mycoplasma hominis/ureaplasma infections. Massive gastrointestinal haemorrhage has been described as a cause of high ammonia levels as the significant amount of ammonia produced bypasses the liver and enters into the systemic circulation.⁴ Extraneous sources can be from shock, severe physical exertion, parenteral nutrition, drugs (valproate, barbiturates, narcotics, diuretics), cigarette smoking and salicylate intoxication.

There is limited consensus in the literature with regards to the radiological findings in encephalopathy due to hyperammonaemia. In a small case series of four patients studied by a group in Toronto,⁵ they found that bilateral symmetrical restricted diffusion was seen in the insular cortex and cingulate gyruses on MRI FLuid Attenuation Inversion Recovery (FLAIR) and T2-weighted imaging sequences. The parietal, frontal, temporal and occipital cortices were also affected in all patients, but this was more variable and asymmetrical. All MRIs were performed 1–4 days postsymptom onset. Cortical changes appear to reflect early radiological changes and are reversible if early and prompt treatment is administered.^{6 7} There was a high mortality of 50% in this case series.

In a patient with hyperammonaemic encephalopathy, the first and most important step in the management involves determining whether the patient has chronic liver disease. For patients with chronic liver disease, he should be evaluated for hepatic encephalopathy and the precipitating factors for it. Thereafter, ammonia-lowering treatment should be promptly commenced. In comparison, for patients without any known history of chronic liver disease, acute liver disease must be ruled out through a liver function test. If this has been excluded, other sources of high ammonia (as discussed above) should be sought. The presence of porto-systemic shunts should also be worked up via ultrasound Doppler of the liver. Ammonia-lowering agents are also started and any inciting drugs, if present, are stopped. Ammonia-lowering agents such as lactulose act by their osmotic activity, thereby reducing the absorption of endogenous nitrogen from the intestinal lumen. Endogenous nitrogen breakdown can be suppressed by a high carbohydrate intake, but this is not commonly performed. Other measures to care for these patients are mainly supportive: ventilatory support for reduced conscious states while awaiting for neurological recovery, treatment of seizures and measures to reduce intracranial pressures in patients with cerebral oedema. For patients with urea cycle enzyme deficiencies, sodium benzoate may be useful.

There is a paucity of literature on whether a low-protein diet should be instituted in patients with non-cirrhotic hyperammonaemia and much of the work has been mainly conducted in cirrhotic patients. In a randomised controlled trial performed by Cordoba in 2004,⁸ he found that protein restriction did not affect the outcomes of patients with hepatic encephalopathy, and in fact, the group assigned to low- protein diet showed a higher protein breakdown. Elevated resting energy expenditure is also seen in cirrhotic patients,⁹ and the increase in rates of gluconeogenesis to maintain splanchnic glucose levels results in increased proteolysis.¹⁰ The European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines, therefore, recommend a daily intake of 1.2–1.5 g/kg body weight in this group of patients.¹¹ However, whether this can be extrapolated to non-cirrhotic patients remains a question. As most of the patients with non-cirrhotic hyperammonaemia do not have increased baseline metabolism rates, they may still benefit from a short-term dietary protein restriction to reduce the amount of endogenous dietary protein.

Conclusion

In conclusion, encephalopathy secondary to non-cirrhotic hyperammonaemia remains a diagnostic challenge to intensivists. Porto-systemic shunting can lead to a reduction in blood flow through the liver, thereby affecting ammonia metabolism and resulting in an elevated ammonia level. A high index of suspicion is required to reach this diagnosis, while prompt and aggressive management is needed to treat this potentially reversible condition.

Learning points.

• Non-cirrhotic hyperammonaemia remains a diagnostic challenge to medical practitioners

• In particular, patients who have had recent upper abdominal or hepatic surgery with possible porto-systemic shunting should be evaluated for this diagnosis as the shunting of blood away from the liver will result in a reduction in ammonia metabolism and the ensuing features of hyperammonaemia

• The management of patients with high ammonia levels is conservative with the use of ammonia clearing agents, removal of inciting agents, bowel clearance and symptomatic control.