SPECIAL ARTICLE
Affair With Triphasic WavesdTheir Striking Presence, Mysterious
Significance, and Cryptic Origins: What are They?
Peter W. Kaplan* and Raoul Sutter†‡
Summary: Triphasic waves, which have been recorded in the EEG of
encephalopathy for more than 50 years, remain clearly identifiable but
historically purportedly of uncertain significance. Initially described with
liver failure and high serum ammonias, they came to be reported in an ever-
expanding list of metabolic, toxic, and structural conditions. Often a dynamic
finding (in which the occurrence of triphasic waves might increase or
decrease with stimulation or arousal of the patient during EEG), there has
been increasing insight into their correlation with multiple concurrent
conditions, including subcortical white-matter disease, infections and meta-
bolic disturbances, and their prognostic significance. There are sparse data,
but there is active controversy into their confusion for, or occurrence in,
nonconvulsive status epilepticus. This review and commentary discuss our
current understanding of triphasic waves and the newer areas of contention
surrounding this mysterious EEG morphology.
Key Words: EEG, Encephalopathy, Subcortical, White matter disease,
Nonconvulsive status epilepticus, Delirium, Confusion, Organ failure,
Hyperammonemia, Prognosis, Neurocritical care.
(J Clin Neurophysiol 2015;32: 401–405)
T riphasic waves (TWs) are visually striking complexes that appear
in EEG in some patients with encephalopathy (Sutter et al.,
2013a). In this waveform, there are three principal phases: the main
deflection being downward, representing a surface positive change.
This dominant phase is usually preceded by a low-amplitude (often
rounded or even absent) negative deflection and followed by a long,
slow, broad slow-rising deflection, giving the entire complex
a triphasic contour. Ever since, TWs were described by Foley et al.,
1950 and by Bickford and Butt, 1955, they have fascinated and
puzzled electroencephalographers and neurologists regarding their
exact nature and prognostic value. Were these eye-catching
morphologies epileptic discharges that had been blunted by toxic
or metabolic encephalopathy? Or, did they represent an EEG pattern
pathognomonic of specific types of encephalopathy (hepatic, renal,
or other) (MacGillivray and Kennedy, 1970)? To confuse the issue
further, the question was raised whether TWs could be a rare pattern
in nonconvulsive status epilepticus?
Early reports on single cases and case series enlarged the
clinical settings in which TWs appeared, listing as presumed
causes, hepatic, renal, hypoxic, structural, toxic, and metabolic
insults. This expansion (albeit with relatively few cases of each)
From the *Department of Neurology, Johns Hopkins Bayview Medical Center,
Baltimore, Maryland, U.S.A.; †Clinic for Intensive Care Medicine, University
Hospital Base, Basel, Switzerland; and ‡Department of Neurology, Division of
Clinical Neurophysiology, University Hospital Basel, Basel, Switzerland.
Address correspondence and reprint requests to Peter W. Kaplan, MBBS, FRCP,
Department of Neurology, Johns Hopkins Bayview Medical Center, 4940
Eastern Avenue, Baltimore, MD 21224, U.S.A.; e-mail: pkaplan@jhmi.edu.
Copyright Ó 2015 by the American Clinical Neurophysiology Society
ISSN: 0736-0258/15/3205-0401
Journal of Clinical Neurophysiology  Volume 32, Number 5, October 2015
“watered down” the specificity and hence the “diagnostic value” of
finding TWs, hence “relegating” the morphologies to nonspecific
status. Were TWs then merely a morphological curiosity without
diagnostic value? Were they now to be viewed as a finding that
would indiscriminately encourage the clinician to cast wide the
diagnostic net for myriad possible causes that might range from
Anemia to Zollinger–Ellison syndrome?
DISTINGUISHING TRIPHASIC WAVES FROM
ICTAL PATTERNS
In their infancy, TWs were looked on as a curiosity. They
were not found in animal models or in young humans, indeed they
were rarely reported in patients younger than 20 years. Making
their first literary appearance in 1950, they were featured as TWs
and described as high-amplitude slow waves with an initial sharp
positive transient, followed by a negative phase, with a complex
frequency of 1.5 to 3 Hz. Early on and then subsequently, the
discharges were seen to have a fronto-occipital delay or “time-
lag,” and to appear singly or in runs for a few seconds at a time, in
the setting of slow EEG background activity (Sheridan and Sato,
1956). In 1955, Bickford and Butt enlarged on this description,
referring to them as “blunted sharp-slow waves” and they were
paraded largely as evidence of hepatic insufficiency or failure.
With the appearance of larger case series, some of these
challenges were met by exploring the morphological character-
istics of TWs, evaluating their clinical associations and settings
and attempting to differentiate them from other encephalopathic
patterns (Bahamon-Dussan et al., 1989; Fisch and Klass, 1988).
The resemblance of some forms of TWs to some types of non-
convulsive status epilepticus EEG patterns had been noted in passing
by Granner and Lee, 1994 and Kaplan, 1996 and more formally by
Boulanger et al., 2006 with studies aimed at determining how to
distinguish an encephalopathic type from an ictal variant. One
drawback was the lack of objectivity in diagnosing, determining, or
classifying what was a TW, leading to reports of atypical and typical
forms. Panels of EEGers might “decide” which forms were true TWs
and which forms were epileptic transients (nonconvulsive status
epilepticus), producing a somewhat circular argument; there were no
proven objective criteria that established an episodic transient as
a TW, as opposed to being, for example, a distorted epileptic
discharge. Some clarity was shed on the subject when a single patient
was reported who simultaneously had both TWs (from hepatic failure)
and genetic generalized frontocentral spike waves, enabling the
patient to be his “own control,” and thus generating potentially
objective criteria that might be used to distinguish TWs from genetic
generalized spike-wave morphologies at least (Kaplan and Schlatt-
man, 2012). The distinctions between the TW and the epileptic
discharge included (1) more acute narrow-angled, briefer, and
frontopolar emphasis for generalized spike waves and (2) conversely,
401
P. W. Kaplan and R. Sutter Journal of Clinical Neurophysiology  Volume 32, Number 5, October 2015

longer duration, frontocentral, stimulus-altered blunter TWs
(Figs. 1A and 1B).
SOURCE LOCALIZATION
Using a dipole source and distributed source models, source
localization of TWs of various causes revealed a single dipole with
radial orientation to the frontal pole. Current density in TWs was
directed to the medial frontal regions (Kwon et al., 2007). Other
studies on generalized spike wave-complexes (GSWCs) in idiopathic
generalized epilepsy using source analysis revealed dipoles localized to
bilateral frontal, parietal, and temporal lobes, prominent in the
frontomedial and frontoorbital regions with a current source in the
midsagittal region (Santiago-Rodriguez et al., 2002). Our patient with
both TWs and GSWCs had TWs maximally in the posterior frontal
regions or frontocentral regions (F3-C3; F4-C4), with GSWCs in the
anterior frontal or frontopolar regions (Fp1-F3; Fp2-F4), similar to
patients with juvenile myoclonic epilepsy (Fig. 1A).

FIG. 1. A, Reveals TWs in an individual with both GSWCs. B, Montage of an individual’s GSWCs. The TWs subtended a larger angle
than GSWCs and are “blunted.” Complex duration for TWs is 0.32 6 0.02 seconds, while GSWCs averages 0.12 6 0.03 seconds. The
typical TW location is posterofrontal at F3-C3 and F4-C4, whereas GSWCs favor frontopolar (anterior frontal) regions: Fp1-F3; Fp2-F4.
Amplitude comparisons show amplitude changes between phases II and III, and peak phase III are similar. The amplitude difference
for TWs is 110 mV (614 mV); amplitude difference for GSWCs was 87 mV (619 mV). Adapted with permission from Kaplan and
Schlattmann (2012). FIRDA, frontal intermittent rhythmic delta activity; GSWCs, generalized epileptic spike waves; TWs, triphasic
waves.

402 Copyright Ó 2015 by the American Clinical Neurophysiology Society

Journal of Clinical Neurophysiology  Volume 32, Number 5, October 2015 Affair With Triphasic Waves

CLINICAL AND RADIOLOGICAL ASSOCIATIONS OF
TRIPHASIC WAVES
With added scrutiny and the publication of larger series, there
was an exposée of the loose and promiscuous associations of TWs.
They were indiscriminately coupled to various diseases, such as
hypertensive encephalopathy, hypernatremia and hyponatremia, hypo-
glycemia, hypercalcemia, brain abscesses, septic shock and sepsis-
related encephalopathy, postictal states, lithium and baclofen toxicity,
and strokes. Sundaram and Blume (1987) and Fisch and Klass (1988)
further unmasked the TW mystique when they reported that no
particular characteristic of the waves could be linked to a particular
etiology. Blatt and Brenner (1996) and Aguglia et al. (1990) further
provided TWs with increased press, linking them to dementias, brain
tumors, subcortical encephalopathy, and psychiatric conditions (Amo-
dio et al., 2006; Blatt and Brenner, 1996). In the 50 patients reported
by Karnaze and Bickford, 1984, their reputation now in free-fall, TWs
were seen with renal failure, hyperosmolarity, anoxia, and hypogly-
cemia, although half the cases were still linked to the alcoholic
liver, hepatic encephalopathy, and hyperammonemia.
These were tough times for the specificity of TWs. Patients
with TWs had mortalities ranging from 30% to 100% depending on
whether the cause was renal, hepatic, or anoxic encephalopathy
(Karnaze and Bickford, 1984). Bahamon-Dussan et al. (1989) noted
that whatever the cause, there was a 50% 1-month mortality. At
almost 2 years, it rose to 77%. Even today, the morbidity and
mortality remain highly elevated despite heightened awareness and
earlier detection of TWs, better approaches to reversing metabolic
disorders including liver transplantation, and rapid correction of
hyperammonemic states (Fig. 2) (Sutter et al., 2013a).
More recently, there has been a concerted effort to quantify and
qualify the gradations of changes that occur with disease progression,
notably with hepatic disease associated with hyperammonemia. In a study
of 154 encephalopathic patients, TWs were associated with liver or
multiorgan failure, and the odds for TWs increased significantly with
every additional unit of elevated serum ammonia levels (Sutter et al.,
2013a). A number of classifications and spectral analyses, which have
included ANNES (the artificial neuronal network expert system) have
examined correlations between the severity of liver disease, psychiatric
features, and various classification systems, along with biological
markers of liver disease and prognosis (Amodio et al., 2006), revealing
its sensitivity in states of minimal hepatic encephalopathy. Noted were
increases in theta power over the posterior head regions and decreases in
mean dominant frequencies with increase in delta power (Amodio and
Gatta, 2005). Such correlations have enabled the assessment of prognosis
and response to treatment after liver transplantation.
A case-control study on a larger cohort of encephalopathic
patients spanning 9 years has shed some light on the clinical
significance of TWs, the clinical and radiologic correlations, and their
prognostic import (Sutter and Kaplan, 2014). Adult encephalopathic
patients with TWs (95 cases) were matched 1:1 with encephalopathic

FIG. 2. The presence of clinical, biochemical, and neuroanatomic abnormalities in encephalopathic patients with different EEG
patterns. Metabolic problems were renal and/or liver insufficiency. Structural abnormalities included white matter lesions, brain
atrophy, cerebral infarcts, intracranial hemorrhage, brain tumors, encephalitis, posterior reversible encephalopathy, and traumatic
brain injury. FIRDA, frontal intermittent rhythmic delta activity; TWs, triphasic waves. Adapted with permission from Sutter and
Kaplan (2013).

Copyright Ó 2015 by the American Clinical Neurophysiology Society 403

P. W. Kaplan and R. Sutter Journal of Clinical Neurophysiology  Volume 32, Number 5, October 2015
patients without TWs (95 controls) by Glasgow Coma Scale and the
frequency range of EEG background activity. Alcohol abuse, liver
insufficiency, infections, and subcortical brain atrophy were indepen-
dently associated with TWs in patients matched for clinical and EEG
features of encephalopathy. These associations strengthen the hypoth-
esis that TWs evolve from an interplay of pathological neurostructural,
metabolic, and toxic conditions.
BEHAVIOR OF TRIPHASIC WAVES
Many authors have noted that TWs may increase or decrease
with stimulation. This reactivity is not universal, and the effect of
stimulation or arousal on the prevalence of TWs remains poorly studied.
Bickford and Butt (1955) found no effect of arousal, whereas Fisch and
Klass (1988) reported that stimulation increased TWs in 2 of 53 patients
and decreased it in 9. Our unpublished data on a cohort of 109 TW
patients revealed that TWs increased in 32 and decreased in 46 and
revealed no change in 19; 4 were not tested. In 8 patients with
continuous TWs, there was no change. Most (up to 85%) TWs in
encephalopathy are reactive (Sutter et al., 2013b). In a recent study of
a different cohort of 105 encephalopathic patients with TWs and
different underlying clinical pathologic conditions (Fig. 3), the univari-
able analysis revealed that the frequency of TWs tended to increase more
frequently in nonsurvivors (13/21 nonsurvivors vs. 40/84 survivors), and
the proportion of patients who decreased their TW occurrence with
stimulation was greater in the survivors (32/84 survivors vs. 4/21
nonsurvivors) (Sutter et al., 2013b). However, the only predictor of death
that was independent from measured confounders was the lack of EEG
background activity. In contrast, GSWCs in genetic epilepsies show little
reactivity with arousal or stimulation.
ELECTROPHYSIOLOGICAL BASIS
Perhaps, the most difficult aspect regarding TWs is what they
actually are and how they are produced on an electrophysiological
basis. The neuropathic and biochemical substrates of TWs and
FIG. 3. Proportional distribution of infections, metabolic
derangements, and structural brain abnormalities in 105
encephalopathic patients with triphasic waves. Adapted with
permission from Sutter et al. (2013b).
404
hepatic encephalopathy are still being explored. A review by
Shawcross and Jalan (2005) of the ammonia hypothesis noted the
increase in astrocytic pH with hyperammonemia, which induces
calcium-dependent release of glutamate, N-methyl-D-aspartate acti-
vation, and excito-toxicity. The effects of hyperammonemia are
believed to decrease chronic astrocytic glutamate supplies, inactivate
glutamate transportation, and hence decrease the postsynaptic
glutamate receptor numbers on astrocytes and neurons. In this
fashion, there is an increase in cerebral inhibition and increased
gamma–aminobutyric acid tone. The increased gamma–aminobutyric
acid–A receptor activation increases postsynaptic neuron inhibition
(Ahboucha and Butterworth, 2004). These acute changes are
associated with astrocytic and cytotoxic edema (Shawcross and
Jalan, 2005), which may affect the subcortical cerebral white matter.
The EEG slowing has been thought to reflect the level of
hyperammonemia and to correlate with the degree of liver disease. In
a series of studies, Amodio et al demonstrated worsening of the EEG
profile with slowing of the alpha frequency, theta activity that appeared
in the frontal and temporal regions, and the appearance of diffuse theta
activity, which further degraded into diffuse theta and then delta activity
to the point of isoelectricity (Amodio et al., 2006).
But we must return to the TWs as a hallmark, the apparent
biomarker that is reflected on the EEG. What are TWs and how are
they produced? Even from the beginning, there had been an effort at
describing TWs as being produced at the scalp’s surface by distant
“projected rhythms” somehow related to thalamocortical sleep or
epileptic relay systems or was this just so much hand-waving? The
mechanism by which TWs are generated was postulated by early
investigators to involve a traveling wave of positivity that swept
across the cortex from the frontal to the occipital regions at
a calculated rate of 1.5 m/s. The authors postulated that the primary
disturbance occurred at a thalamic (subcortical) level with changes in
the potential field observed on the cortical surface reflecting the
primary disturbance thalamic disturbance through thalamocortical
relays (Bickford and Butt, 1955).
There seemed to be no cogent explanation in clinical or
electrophysiologic texts, they just were. In fact, not only could they
be produced seemingly by almost any cerebral dysregulation, they
might even indicate some forms of subclinical or nonconvulsive
status epilepticus (Boulanger et al., 2006). How could they be so
striking and yet so elusive; unexplainable and nonspecific? Were
they the mysterious figure wrapped in mystique that appeared at
gatherings of confused states?
Much work in animal models has been done to help elucidate
the interplay of thalamocortical slow-firing systems from which the
production of TWs might be inferred, since TWs do not occur in an
animal model. To summarize, TWs may represent the sequential
activation (usually in an anteroposterior direction) of midline
thalamic reticular nuclei (with a “slow” firing rate) and oscillatory
system involved with spindle waves (Kim et al., 1995). This
activation may be facilitated by acquired subcortical (white matter)
disease that impairs the reverberating or reentrant corticothalamic
inputs that in turn suppress thalamic activity. Arousal through the
ascending reticular activating system increases thalamocortical
activation (and hence might increase the generation of TWs at the
cortex, added by Kaplan and Sutter) from the thalamic reticular
“slow-firing” activation (Llinas and Paré, 1991).
SUMMARY
The voluptuous TW has long been a mysterious presence at
the “Encephalopathy Ball”. With no appearance earlier than in
Copyright Ó 2015 by the American Clinical Neurophysiology Society
Journal of Clinical Neurophysiology  Volume 32, Number 5, October 2015
adulthood, no presence in other species, and with the press reports of
indiscriminate and suspect associations with any number of impaired
conditions, TWs have remained a striking but puzzling bystander in
impaired brains. More recent data certainly suggest (as reported early
on) that there are indeed classical or typical and atypical forms.
Atypical TWs may largely represent “TW look-alikes” such as with
drug toxicities (e.g., lithium, baclofen, and cefepime) or in forms of
nonconvulsive status epilepticus. These atypical TWs could be seen
with particular clinical features, associations, spatial distribution, and
morphology that might differ from the original and contemporary
descriptions of TWs as being symmetric, often maximally fronto-
central in distribution, often reactive to arousal level, and occurring
with high ammonia or more recently with white matter disease.
Atypical forms not related to metabolic problems or white matter
disease might be asymmetric, have a more sharply contoured phase I
and steeply declining phase II, be more diffusely and sporadically
distributed, and less affected by arousal.
To have such a sentinel biomarker remain unresolved, is to
potentially lose a valuable electrophysiological insight into the
etiology, specificity, correlations, and prognostic value of this
pattern. Recent work from single centers has clearly established that
TWs are significantly associated with white-matter disease, infec-
tions, and some metabolic derangements, but it would seem that the
door is still open to a more systematic multicentered investigation
using contemporary imaging and modern laboratory tests to
demystify this occult figure.
All interested parties are welcome to the Ball!
REFERENCES
Aguglia U, Gambardella A, Oliveri RI, et al. Nonmetabolic causes of triphasic
waves: a reappraisal. Clin Electroencephalogr 1990;21;120–125.
Ahboucha S, Butterworth RF. Pathophysiology of hepatic encephalopathy: a new look
at GABA from the molecular standpoint. Metab Brain Dis 2004;19:331–343.
Amodio P, Gatta A. Neurophysiological investigations of hepatic encephalopathy.
Metab Brain Dis 2005;20:369–379.
Amodio P, Pellegrini A, Ubiali E, et al. The EEG assessment of low grade hepatic
encephalopathy: comparison of an artificial neural network-expert system
(ANNES) based evaluation with visual EEG readings and EEG spectral
analysis. Clin Neurophysiol 2006;117:2243–2251.
Bahamon-Dussan JE, Celesia GG, Grigg-Damberger MM. Prognostic significance
of EEG triphasic waves in patients with altered states of consciousness. J Clin
Neurophysiol 1989;6:313–319.
Copyright Ó 2015 by the American Clinical Neurophysiology Society
Affair With Triphasic Waves
Bickford RG, Butt HR. Hepatic coma: the electroencephalographic pattern. J Clin
Invest 1955;34:790–799.
Blatt I, Brenner RP. Triphasic waves in a psychiatric population: a retrospective
study. J Clin Neurophysiol 1996;13:324–329.
Boulanger JM, Deacon C, Lécuyer D, et al. Triphasic waves versus non-
convulsive status epilepticus: EEG distinction. Can J Neurol Sci
2006;33:175–180.
Fisch BJ, Klass DW. The diagnostic specificity of triphasic wave patterns.
Electroencephalogr Clin Neurophysiol 1988;70:1–8.
Foley JM, Watson CW, Adams RD. significance of the electroencephalographic
changes in hepatic coma. Trans Am Neurol Assoc 1950;51:161–165.
Granner MA, Lee SI. Nonconvulsive status epilepticus EEG analysis in a large
series. Epilepsia 1994;35:42–47.
Kaplan PW Nonconvulsive status epilepticus in the emergency room. Epilepsia
1996;37:643–650.
Kaplan PW, Schlattman D. Comparison of triphasic waves and epileptic
discharges in one patient with genetic epilepsy. J Clin Neurophysiol
2012;29:458–461.
Karnaze DS, Bickford RG. Triphasic waves: a reassessment of their significance.
Electroencephalogr Clin Neurophysiol 1984;57:193–198.
Kim U, Bal T, McCormick DA. Spindle waves are propagating synchronized
oscillations in the ferret LGNd in vitro. J Neurophysiol 1995;74:
1301–1333.
Kwon OY, Jung KY, Park KJ, et al. Source localization of triphasic waves:
implications for the pathophysiological mechanism. Clin EEG Neurosci
2007;38:161–167.
Llinas RR, Paré D. Of dreaming and wakefulness. Neuroscience
1991;44:531–535.
MacGillivray B, Kennedy JK. The triphasic waves of hepatic encephalopathy.
Electroencephr Clin Neurophysiol 1970;28:428.
Santiago-Rodriguez E, Harmony T, Fernández-Bouzas A, et al. Source analysis of
polyspike and wave complexes in juvenile myoclonic epilepsy. Seizure
2002;11:320–324.
Shawcross D, Jalan R. The pathophysiologic basis of hepatic encephalopathy:
central role for ammonia and inflammation. Cell Mol Life Sci
2005;62:2295–2304.
Sheridan PH, Sato S. Triphasic waves of metabolic encephalopathy versus spike-
wave stupor. J Neurol Neurosurg Psychiatry 1956;49:108–109.
Sundaram MB, Blume WT. Triphasic waves: clinical correlates and morphology.
Can J Neurol Sci 1987;14:136–140.
Sutter R, Kaplan PW. Clinical and electroencephalographic correlates of acute
encephalopathy. J Clin Neurophysiol 2013;30:443–453.
Sutter R, Kaplan PW. Uncovering clinical and radiological associations of
triphasic waves in acute encephalopathy: a case-control study. Eur J Neurol
2014;4:660–666.
Sutter R, Stevens RD, Kaplan PW. Clinical and imaging correlates of EEG
patterns in hospitalized patients with encephalopathy. J Neurol
2013a;260:1087–1098.
Sutter R, Stevens RD, Kaplan PW. Significance of triphasic waves in patients with
acute encephalopathy: a nine-year cohort study. Clin Neurophysiol
2013b;124:1952–1958.
405