Clinical Chemistry 62:12

1565–1569 (2016) Clinical Case Study

Critically High Plasma Ammonia in an

Adolescent Girl
Dennis J. Orton,1,2 Jessica L. Gifford,1,2 Isolde Seiden-Long,1,2 Aneal Khan,3 and Lawrence de Koning1,2*

Downloaded from https://academic.oup.com/clinchem/article/62/12/1565/5612056 by University of Toronto user on 28 September 2023

CASE DESCRIPTION
QUESTIONS TO CONSIDER
A 12-year-old female patient presented with nonspecific
1. What is plasma ammonia and why is it important?
symptoms including fatigue, fever, and headache persist-
ing for approximately 4 months. Plasma ammonia was 2. What are the physiologic causes of hyperammonemia?
230 ␮g/dL (135 ␮mol/L) (critical value ⬎187 ␮g/dL or
3. What is the appropriate treatment for hyperammonemia?
110 ␮mol/L). The patient was referred to a pediatric
emergency center for urgent evaluation of hyperam- 4. What are the preanalytical causes of hyperammonemia?
monemia. Plasma and urine amino acid testing was per-
formed to investigate the possibility of a urea cycle disor-
der; however, all results were normal. Over the following Discussion
2 weeks, the patient was treated with intravenous sodium
benzoate/sodium phenylacetate and L-arginine to reduce Ammonia is a metabolic waste product produced mainly
the blood ammonia concentration. However, the pa- in the gut, kidney, and skeletal muscle during exercise
tient’s ammonia concentration never dropped below (1 ). It is generated from microbial metabolism, deami-
nation of amino acids, nucleic acid degradation, and pro-
94 ␮g/dL (55 ␮mol/L) (upper limit of reference interval,
tein catabolism. Ammonia is normally removed from cir-
80 ␮g/dL or 47 ␮mol/L), with concentrations as high as
culation by incorporation into the urea cycle after
296 ␮g/dL (174 ␮mol/L) with inconsistent associations
condensation with bicarbonate and phosphorylation to
between ammonia concentrations and headaches. Dur- form carbamoyl phosphate in the liver. The urea cycle
ing the course of treatment, the patient suffered substan- includes 5 enzymes and 2 transporters that remove am-
tial side effects of intravenous therapy and underwent monia by generating urea and fumarate. Fumarate is in-
numerous costly clinical and laboratory investigations. corporated into the Krebs cycle, whereas urea is excreted
Amino acid investigations were repeated but results re- in urine. At physiological pH, ammonia exists mainly as
mained normal. Given the large fluctuations in blood ammonium ions (NH4⫹), with 1%–2% as ammonia gas
ammonia concentrations with inconsistent response to (NH3). Ammonia readily crosses the blood– brain barrier
intravenous medications, the metabolic service contacted where it combines with H⫹ to form ammonium, which
the laboratory to inquire about a possible preanalytical or competes with potassium for transport across neuronal
analytical cause. One week before this query, a commu- membranes (2 ). Symptoms of hyperammonemia include
nity physician also contacted the laboratory regarding lethargy, irritability, vomiting, and poor feeding in chil-
hyperammonemia in several patients. However, no obvi- dren, as well as hyperventilation, seizures, coma, and
ous preanalytical or analytical errors were discovered eventually death.
despite numerous audits of preanalytical conditions and Common causes of hyperammonemia include liver
review of quality control data. failure (hepatocyte destruction and reduced urea cycle
enzymes), drug reactions (e.g., inhibition of the urea
cycle by valproic acid), hemolytic disease (release of am-
monia from red blood cells), gastrointestinal bleeds (in-
creased ammonia generation due to microbial catabolism
1
Calgary Laboratory Services, Calgary, Alberta, Canada, and Departments of 2 Pathology of hemoglobin), or urea cycle disorders (defective or de-
and Laboratory Medicine and 3 Medical Genetics and Paediatrics, University of Calgary,
Calgary, Alberta, Canada.
creased urea cycle enzymes) (1 ). Urea cycle disorders are
* Address correspondence to this author at: Alberta Children’s Hospital, Rm. B3–724, inborn errors of metabolism that normally present with
2888 Shaganappi Trail Northwest, Calgary, AB T3B 6A8, Canada. Fax 403-955-2321; severe hyperammonemia in the first few weeks of life,
e-mail [email protected].
This study was previously presented at the Joint Conference of the Canadian Society of
although less severe forms can manifest at any age (3 ).
Clinical Chemists (CSCC) and Canadian College of Medical Geneticists (CCMG) in Edmon- Early diagnosis and treatment is important to avoid de-
ton, Alberta, Canada, on June 20, 2016. lays in mental development.
Received April 18, 2016; accepted June 10, 2016.
DOI: 10.1373/clinchem.2016.259473 Ammonia is frequently measured in the clinical lab-
© 2016 American Association for Clinical Chemistry oratory with a one-step enzymatic assay using glutamate

1565
Clinical Case Study

mia in several patients; however, no obvious preanalytical

or analytical causes were discovered at that time. In these
patients, ammonia results were either disregarded or pa-
tients received limited follow-up. On day 7, the meta-
bolics service contacted the laboratory regarding their
patient, prompting a second investigation.
The second investigation by laboratory staff found

Downloaded from https://academic.oup.com/clinchem/article/62/12/1565/5612056 by University of Toronto user on 28 September 2023

that the centrifuge (Eppendorf 5810 R with swinging
bucket, rotor radius ⫽ 173 mm) used to separate samples
Fig. 1. Patient ammonia results over time. for ammonia testing was set at 1200 rpm instead of
Dashed line: critical value for pediatric patients (187 μg/dL or 110
1200g, resulting in a relative centrifugal force (RCF) of
μmol/L). Dotted line: upper limit of reference interval (80 μg/dL
only 280g (Eq. 1). This centrifuge’s display distinguished
or 47 μmol/L). Correction of the centrifuge speed occurred at day
between rpm and g with an asterisk (i.e., “1200” ⫽ 1200
7, indicated by the arrow. To convert ammonia results in μmol/L to
rpm; “1200*” ⫽ 1200g), which was missed during the
μg/dL, multiply by 1.703.
initial investigation. The centrifuge speed was immedi-
ately corrected and the patient ammonia results quickly
normalized (Fig. 1, day 7):
dehydrogenase (GLDH),4 ␣-ketoglutarate, and NADPH.
When ammonia is added to this mixture, glutamate and
NADP⫹ are generated by GLDH activity, with a decrease
RCF ⫽ 1.12 ⫻ Radius(mm) ⫻ 冉 冊
rpm 2
1000
. (1)

in NADPH absorbance at 340 nm proportional to ammo- To confirm that incorrect centrifugation speed was the
nia concentration. In this case, the ammonia assay (Randox cause, 2 prechilled 3-mL tubes of K2-EDTA anticoagu-
GLDH) also included a background subtraction at 700 nm. lated whole blood were collected from 10 healthy volun-
Ammonia concentrations can be falsely increased by teers. Tubes were stored on ice before separation in a
numerous preanalytical conditions. Delays in blood cen- refrigerated (4 °C) centrifuge. One tube from each vol-
trifugation and room temperature storage can lead to unteer was spun at low speed (1200 rpm, 280g), and the
increased ammonia due to in vitro generation by other at normal speed (1200g), each for 10 min. Plasma
␥-glutamyltransferase (GGT) activity (4, 5 ). Traumatic was aliquoted on ice and tested for ammonia, GGT,
blood collection and hemolysis can increase ammonia, as complete blood count, and Roche serum indices (hemo-
red blood cells contain a high concentration of ammonia lysis, icterus, lipemia). Differences were assessed using
compared to plasma (6 ). Excess muscle activity (e.g., the Wilcoxon signed rank test.
seizures) or muscle activity under ischemia (e.g., fist Plasma from samples centrifuged at low speed were
clenching with tourniquet) increases ammonia genera- visibly more turbid and had significantly (P ⬍ 0.01)
tion via AMP-deaminase activity (7 ). To limit the influ- higher ammonia and platelets than those spun at normal
ence of preanalytical error, blood collection is frequently speed (Fig. 2, A and B). There were no detectable red or
conducted without a tourniquet using prechilled tubes white blood cells in any of the plasma samples and no
transported on ice for separation in a refrigerated centri- significant difference in hemolysis or icterus indices (data
fuge. Dipotassium EDTA is the anticoagulant of choice not shown). Consistent with the visual assessment of tur-
as heparin interferes with several ammonia assays (8 ). bidity, the lipemia index was significantly increased (data
In this case, the patient presented with vague symp- not shown; P ⬍0.01) in samples centrifuged at low
toms and a critical (⬎187 ␮g/dL or ⬎110 ␮mol/L) speed. There was no significant difference in GGT (data
plasma ammonia which raised suspicion for a urea cycle not shown; P ⫽ 0.56). Linear regression was used to
disorder (Fig. 1, day ⫺13). Ammonia results over the determine the relationship between platelet count and
next few days remained high despite the use of intrave- ammonia, yielding a coefficient of determination (R2) of
nous ammonia scavengers, ranging from 94 to 296 0.97, a slope of 0.19 ␮mol/L/109 platelets/L, and an
␮g/dL (55–174 ␮mol/L) (Fig. 1, days ⫺13 to ⫺9). intercept of 20.9 ␮mol/L ammonia (Fig. 2C). Differ-
There were also significant fluctuations in ammonia con- ences were less pronounced when the study was repeated
centrations taken hours apart with no apparent change in using 1.0-mL microcollection containers, likely due to
clinical symptoms. On day 0 (Fig. 1), a community phy- the smaller number of platelets available to concentrate in
sician contacted the laboratory regarding hyperammone- an incompletely separated sample (data not shown).
As these results suggested that the patient’s high am-
monia concentrations were from residual platelets, a sec-
4
Nonstandard abbreviations: GLDH, glutamate dehydrogenase; GGT, ␥-glutamyltrans- ond experiment was performed with pooled K2-EDTA
ferase; RCF, relative centrifugal force. whole blood. Blood was centrifuged at 1200 rpm, 1200g,

1566 Clinical Chemistry 62:12 (2016)

 Clinical Case Study

Downloaded from https://academic.oup.com/clinchem/article/62/12/1565/5612056 by University of Toronto user on 28 September 2023

Fig. 2. Results from volunteers.
Light gray: samples spun at low speed. Dark gray: samples spun at normal speed. There was significantly (P <0.05) higher ammonia (A) and
platelets (B) in plasma spun at low vs normal speed. Ammonia versus platelet count (C). Dotted lines: ammonia upper limit of normal (80
μg/dL or 47μmol/L). To convert ammonia results in μmol/L to μg/dL, multiply by 1.703.

and 2000g, and plasma from each condition was split centrifuges to prevent accidental use when speeds were
into 2 aliquots— one of which was additionally centri- changed. In this case, the speed of the centrifuge in ques-
fuged at 2500g for 15 min to completely remove residual tion was changed to 1200 rpm to process a bone marrow
platelets. Ammonia and platelets were significantly de- sample; however, it was not changed back to 1200g. Fi-
creased with increasing centrifugation speed and the ad- nally, (c) samples collected for ammonia testing would be
ditional centrifugation at 2500g removed nearly all resid- considered STAT priority to reduce processing and anal-
ual platelets, consistently yielding the lowest ammonia ysis time.
concentration (data not shown). This experiment con- While these changes combined with staff education
firmed that increased ammonia in samples spun at low will reduce preanalytical errors in ammonia testing, they
speed was due to residual platelets in plasma. Interest- cannot eliminate them. Communication with phleboto-
ingly, this phenomenon was first described by Cowley et mists, clinician teams, and biochemical genetics labora-
al. over 30 years ago (9 ); however, to our knowledge this tories is essential to determine whether an increased am-
is the first report of a significant patient impact resulting monia result is truly real. This is because review of QC
from a falsely high ammonia result owing to incomplete
results (i.e., Levey–Jennings charts) cannot uncover er-
specimen separation.
rors that occur in the preanalytical phase of testing. Fur-
Other causes of increased ammonia were investi-
thermore, patient ␦ checks appear to be seldom used in
gated; however, these were quickly ruled out. For exam-
ammonia testing. Tracking mean patient ammonia re-
ple, while samples spun at low speed were visibly turbid,
the degree of turbidity assessed by the Roche lipemia sults over time could help detect large-scale preanalytical
index (dichromatic absorbance at 660 and 700 nm) was errors; however, this would require a large ammonia test-
too low to cause a significant interference, which inter- ing workload.
estingly would have been a negative interference (false Regardless of the test, preanalytical quality must be
low) for this assay. Interestingly, GGT concentrations maintained for physicians to make appropriate clinical
were within the reference interval and did not signifi- decisions. In this case, an incorrectly set centrifuge re-
cantly differ according to centrifugation speed. sulted in costly additional treatment that turned out to be
As there was a clear relationship between centrifuga- unnecessary, repeated testing, and significant physical
tion speed, residual platelets, and ammonia concentra- and psychological harm to a patient. This caused several
tion, several engineering controls and operational quality improvements to be made regarding the control
changes were made to reduce the chances of centrifuges and monitoring of centrifuge use. Because of their central
being set incorrectly in the future: (a) centrifuge speeds importance in clinical laboratories, centrifuges and their
would be required to be verified at the start of each shift, operating procedures should be carefully evaluated to
and (b) a brightly colored stop sign would be attached to minimize preanalytical errors (10 ).

Clinical Chemistry 62:12 (2016) 1567

Clinical Case Study

POINTS TO REMEMBER
Author Contributions: All authors confirmed they have contributed to
• Ammonia is a toxic nitrogenous waste product that is the intellectual content of this paper and have met the following 3 require-
normally processed by the urea cycle into urea. ments: (a) significant contributions to the conception and design, acquisi-
tion of data, or analysis and interpretation of data; (b) drafting or revising
• Urea cycle disorders can be responsible for hyperam- the article for intellectual content; and (c) final approval of the published

Downloaded from https://academic.oup.com/clinchem/article/62/12/1565/5612056 by University of Toronto user on 28 September 2023

monemia and usually appear in childhood. article.

• Hyperammonemia is a serious condition that must be Authors’ Disclosures or Potential Conflicts of Interest: Upon man-
uscript submission, all authors completed the author disclosure form. Dis-
urgently treated.
closures and/or potential conflicts of interest:
• Ammonia testing is confounded by numerous preana- Employment or Leadership: L. de Koning, Calgary Laboratory
lytical conditions. Residual platelets in plasma can Services.
cause large false increases of ammonia. Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
• Control of preanalytical error is paramount to prevent Honoraria: None declared.
treatment based on an incorrect laboratory result, as Research Funding: None declared.
this can cause patient harm. Expert Testimony: None declared.
Patents: None declared.

References
1. Adeva MM, Souto G, Blanco N, Donapetry C. Ammo- increased plasma gamma-glutamyltransferase is a 1987;30:661– 8.
nium metabolism in humans. Metabolism 2012;61: source of rapid ammonia formation in blood and 8. Wannasilp N, Sribhen K, Pussara N, Hwanpuch T, Wang-
1495–511. plasma specimens. Clin Chem 1990;36:1479 – 82. chaijaroenkit S, Opartkiattikul N. Heparin is an unsuit-
2. Thrane VR, Thrane, AS, Wang F, Cotrina ML, Smith NA, 5. Nikolac N, Omazic J, Simundic AM. The evidence based able anticoagulant for the detection of plasma ammo-
Chen M, et al. Ammonia triggers neuronal disinhibition practice for optimal sample quality for ammonia mea- nia. Clin Chim Acta 2006;371:196 –7.
and seizures by impairing astrocyte potassium buffer- surement. Clin Biochem 2014;47:991–5. 9. Cowley DM, Nagle BA, Chalmers AH, Sinton TJ. Effects of
ing. Nat Med 2013;19:1643– 8. 6. Apushkin M, Das A, Joseph C, Leung EK, Yeo KT, Baron platelets on collection of specimens for assay of ammo-
3. Summar ML, Dobbelaere D, Brusilow S, Lee B. Diagno- JM, Baron BW. Reducing the risk of hyperammonemia nia in plasma. Clin Chem 1985;31:332–3.
sis, symptoms, frequency and mortality of 260 patients from transfusion of stored red blood cells. Transfus 10. Pyle-Eilola, A. Centrifuges: labs unsung high performance
with urea cycle disorders from a 21-year, multicentre Apher Sci 2013;49:459 – 62. instruments. Clin Lab News; 2015. https://www.aacc.org/
study of acute hyperammonaemic episodes. Acta Pae- 7. Valen PA, Nakayama DA, Veum J, Sulaiman AR, Wort- publications/cln/articles/2015/july/centrifuges-labs-
diatr 2008;97:1420 –5. mann RL. Myoadenylate deaminase deficiency and unsung-high-performance-instruments (Accessed April
4. da Fonseca-Wollheim F, Deamidation of glutamine by forearm ischemic exercise testing. Arthritis Rheum 2016).

Commentary
D. Robert Dufour, MD*

The case described by Orton et al. is of a young girl who nosis less likely. Plasma ammonia concentrations are the
presented with nonspecific complaints of headache and most common screening test for this condition, as well as
fatigue. One possible explanation for such symptoms is for the similar hepatic encephalopathy that occurs in
an inborn error in urea cycle enzymes, a group of rare adults (2 ).
disorders that more commonly presents acutely and in An experienced clinician often recognizes that labo-
early childhood (1 ). Less commonly, chronic presenta- ratory results are inconsistent with the clinical picture, as
tions with nonspecific symptoms can occur later in life happened in this case. Despite another clinician contact-
(sometimes even in adults) and can be difficult to diag- ing the laboratory about apparently incorrect ammonia
nose. Most cases also have various neuropsychiatric concentrations, it took a second inquiry from the clini-
symptoms in addition to headache and fatigue, and the cians caring for this child for the laboratory to complete a
lack of these being described in this child make this diag- full investigation, leading to further complications of
treatment in the patient and, presumably, large medical
bills.
George Washington University Medical Center, Washington, DC.
A good working relationship between clinicians and
* Address correspondence to the author at: 3825 N. Kenmore Ave. #1S, Chicago, IL 60613. laboratorians is critical to optimal laboratory practice.
Fax 240-432-7760; e-mail [email protected]. The National Academies report on Improving Diagnosis
Received September 7, 2016; accepted September 14, 2016.
DOI: 10.1373/clinchem.2016.263566 in Health Care recommends such improved communica-
© 2016 American Association for Clinical Chemistry tion between clinicians and laboratorians (3 ). Clinicians

1568 Clinical Chemistry 62:12 (2016)