Cyclic alternating pattern

Liborio Parrino MD (Dr. Parrino of the University of Parma has no relevant financial relationships to disclose.)
Antonio Culebras MD, editor. (Dr. Culebras of SUNY Upstate Medical University has no relevant financial relationships to disclose.)
Originally released February 9, 2014; last updated November 1, 2016; expires November 1, 2019


Cyclic alternating pattern is an EEG marker of sleep instability that modulates the flexibility of sleep in both physiological conditions and in sleep disorders. Together with sleep duration, sleep intensity, and sleep continuity, cyclic alternating pattern represents a topical pillar of sleep quality.

Key points


• Expressed by the physiological phasic events of NREM sleep, including arousals, cyclic alternating pattern represents the most complex arrangement of sleep microstructure.


• Cyclic alternating pattern is organized in sequences organized in successive cycles. Each cyclic alternating pattern cycle is composed of a phase A (phasic events) and a phase B (background rhythm).


• Cyclic alternating pattern involves not only cerebral activities during sleep, but paces, and is reciprocally influenced by ongoing autonomic and motor functions, with activation during phase A and deactivation during phase B.


• The interaction between cyclic alternating pattern, neurovegetative fluctuations, and motor events determines the pathophysiology of several sleep disorders and the effect of medication and continuous positive airway pressure (CPAP) treatment.


• Recent studies indicate the possibility of scoring cyclic alternating pattern automatically, opening new perspectives for a wider exploitation of this fundamental mechanism of the sleeping brain.

Historical note and terminology

Sleep staging: the conventional scoring rules. The standardized Rechtschaffen and Kales criteria for sleep staging have constituted the internationally accepted system of sleep scoring for 45 years (Rechtschaffen and Kales 1968). An enormous body of data has been produced and published in the scientific literature using this system. The American Academy of Sleep Medicine (AASM) manual introduced in 2007 revised Rechtschaffen and Kales sleep staging to address digital methodology as well as the scoring of arousals, respiratory events, sleep-related movement disorders, and cardiac abnormalities, with consideration of pediatric and geriatric age groups (Iber et al 2007). In particular, the new visual scoring of sleep expands the number of channels indicated in the Rechtschaffen and Kales rules (single central derivation), allowing a more extensive coverage of scalp areas where significant sleep-wake rhythms and waveforms are better represented.

In the AASM system, the distinction between wakefulness (W), non-REM sleep (N), and REM sleep (R) is maintained, but the non-REM stages are reduced to 3: N1 (previous stage 1), N2 (previous stage 2), and N3 (previous stages 3 + 4). Unfortunately, the lumping of stages 3 and 4 in a single stage N3 downsizes the nocturnal development of the process “S,” characterized by decreasing peaks across the night, according to the decreasing power of the EEG delta band (Borbely 1982). Instead of providing additional information on the power of the EEG signal, the adopted solution is a reductive simplification of the previous sleep stages. Moreover, obliterating the exponential profile of process “S,” the restorative function of slow-wave sleep, and its direct relationship with waking activities are shadowed.

In the Rechtschaffen and Kales rules, EEG arousals were excluded from conventional staging procedures. In 1992, the American Sleep Disorders Association (ASDA; later named the AASM) defined arousals as markers of sleep disruption representing a detrimental and harmful feature for sleep (American Sleep Disorders Association 1992). In the following years, however, a number of studies clarified that spontaneous arousals are natural guests of sleep and undergo a linear increase along the lifespan following the profile of maturation and aging (Boselli et al 1998). Moreover, the spectral composition of arousals and their ultradian distribution throughout the sleep cycles reveal that arousals are endowed within the texture of physiologic sleep under the biological control of REM-on and REM-off mechanisms (Terzano et al 2005). Arousal scoring is now considered a fundamental process in staging classification as well as spindles and K-complexes. In the Rechtschaffen and Kales system, a K-complex was unmistakably a marker of sleep stage 2. According to the AASM manual, if a K-complex is associated with an arousal the epoch is scored as N1. In other words, the Rechtschaffen and Kales approach privileged the role of EEG synchrony in the scoring process, whereas the AASM manual enhances the impact of EEG desynchrony (Parrino et al 2009a).

In spite of the differences regarding sleep stages, arousals, and K-complexes, the 2 scoring systems share the same cultural frameworks, ie, rigid epochs of 30 seconds, neglecting evidence that sleep is a continuous function that cannot be restricted to a static sequence of artificial segments. In effect, each 30-sec epoch encompasses several short-time events that carry important information that does not show up in the classical sleep staging reports. These features have been grouped under the general label of sleep phasic events and detailed as K-complexes, sleep spindles, delta bursts, arousals, and K-alpha. The most comprehensive method for their analysis and report is the so-called cyclic alternating pattern (Terzano et al 1985). Because cyclic alternating pattern spans across long periods of NREM sleep, it overcomes the boundaries of rigid epochs and offers a dynamic contribution to the static framework of conventional scoring.

Scoring rules and significance of cyclic alternating pattern. Cyclic alternating pattern is a well-defined marker of the physiological cerebral activity occurring under conditions of reduced vigilance (sleep, coma), translating a state of arousal instability and involving muscle, behavioral, and autonomic functions (Terzano et al 2002b; Parrino et al 2006; Parrino et al 2012a).

During NREM sleep, cyclic alternating pattern is organized in sequences. A cyclic alternating pattern sequence is composed of a succession of cyclic alternating pattern cycles.

Image: Cyclic alternating pattern cycle
The cyclic alternating pattern cycle is composed of a phase A (lumps of sleep phasic events) followed by a phase B (return to EEG background). All cyclic alternating pattern sequences begin with a phase A and end with a phase B. Each phase of cyclic alternating pattern is 2 to 60 s in duration. This cut-off relies on the consideration that the great majority (about 90%) of A phases occurring during sleep are separated by an interval of less than 60 s (Terzano and Parrino 1991).

The absence of cyclic alternating pattern for more than 60 s is scored as non-cyclic alternating pattern and coincides with a condition of sustained physiological stability (Terzano et al 1986). An isolated phase A, ie, a phase A separated from another phase A by more than 60 s, is classified as non-cyclic alternating pattern.

Image: Non-cyclic alternating pattern

The reactivity of cyclic alternating pattern. Cyclic alternating pattern and non-cyclic alternating pattern can be consistently manipulated by sensorial inputs.

Applying separately the same arousing stimulus during the 2 EEG components of cyclic alternating pattern, phase B is the one that immediately assumes the morphology of the other component, whereas the inverse transformation never occurs when the stimulus is delivered during phase A. This stereotyped reactivity persists throughout the successive phases of cyclic alternating pattern with lack of habituation. In contrast, when the same stimulus is presented during non-cyclic alternating pattern, the EEG responses are generally brief, hypersynchronized (slow-waves), and proceed toward progressive habituation (Terzano et al 1990; Terzano and Parrino 1991).

However, a robust or sustained stimulus delivered during non-cyclic alternating pattern induces the immediate appearance of repetitive cyclic alternating pattern cycles that display the same morphology and reactive behavior of spontaneous cyclic alternating pattern sequences. The evoked cyclic alternating pattern sequence may herald a lightening of sleep depth or continue as a damping oscillation before the complete recovery of non-cyclic alternating pattern.

General rule. Cyclic alternating pattern cannot be measured without having scored sleep stages in advance; however, it is not limited to epoch fragmentation and spans over long periods of NREM sleep.

An A phase is scored within a cyclic alternating pattern sequence only if it precedes and/or follows another phase A in the 2 to 60 s temporal range.

At least 2 consecutive cyclic alternating pattern cycles are required to define a cyclic alternating pattern sequence.

Image: Sequence of cyclic alternating pattern cycles)
Consequently, 3 or more consecutive A phases must be identified with each of the first 2 A phases followed by a phase B (interval <60 s), and the third phase A followed by a non-cyclic alternating pattern interval of more than 60 s.

Cyclic alternating pattern sequence onset must be preceded by non-cyclic alternating pattern (a continuous non-REM sleep EEG pattern for >60 s), with the following 3 exceptions. There is no temporal limitation: (1) before the first cyclic alternating pattern sequence arising in non-REM sleep; (2) after a wake to sleep transition; and (3) after a REM to non-REM sleep transition (Terzano et al 2002b).

Cyclic alternating pattern sequences have no upper limits for duration and number of cyclic alternating pattern cycles. In normal young adults, 2.5 min is the approximate mean duration of a cyclic alternating pattern sequence, containing an average of 6 cyclic alternating pattern cycles (Smerieri et al 2007).

Stage shifts. Within non-REM sleep, a cyclic alternating pattern sequence is not interrupted by a sleep stage shift if cyclic alternating pattern scoring requirements are satisfied. Consequently, because cyclic alternating pattern sequences can extend across adjacent sleep stages, a cyclic alternating pattern sequence can contain a variety of different phase A and phase B activities.

Cyclic alternating pattern in REM sleep. Cyclic alternating pattern sequences commonly precede the transition from non-REM to REM sleep and end just before REM sleep onset. REM sleep is characterized by the lack of EEG synchronization; thus, phase A features in REM sleep consist mainly of desynchronized patterns (fast low-amplitude rhythms), which are separated by a mean interval of 3 to 4 min (Schieber et al 1971). Consequently, under normal circumstances, cyclic alternating pattern does not occur in REM sleep. However, pathologic conditions characterized by repetitive A phases recurring at intervals less than 60 s (eg, periodic REM-related sleep apnea events), can produce cyclic alternating pattern sequences in REM sleep (Terzano et al 1996).

Recording techniques and montages. Cyclic alternating pattern is a global EEG phenomenon involving extensive cortical areas. Therefore, A phases should be visible on all or most EEG leads. Bipolar derivations such as Fp1-F3, F3-C3, C3-P3, P3-O1 or Fp2-F4, F4-C4, C4-P4, and P4-O2 guarantee a favorable detection of the phenomenon. A calibration of 50 mV/7 mm with a time constant of 0.1 s and a high-frequency filter in the 30 Hz range is recommended for EEG channels. Monopolar EEG derivations (C3-A2 or C4-A1 and O1-A2 or O2-A1), eye movement channels, and submentalis EMG, currently used for the conventional sleep staging and arousal scoring, are also essential for scoring cyclic alternating pattern.

Amplitude limits. Changes in EEG amplitude are crucial for scoring cyclic alternating pattern. Phasic activities initiating a phase A must be a third higher than the background voltage (calculated during the 2 s before onset and 2 s after offset of a phase A). However, in some cases, a phase A can present ambiguous limits due to inconsistent voltage changes. Onset and termination of a phase A are established on the basis of an amplitude/frequency concordance in the majority of EEG leads. The monopolar derivation is mostly indicated when scoring is carried out on a single derivation. All EEG events that do not clearly meet the phase A characteristics cannot be scored as part of phase A.

Time limits. The minimal duration of a phase A or a phase B is 2 s. If 2 consecutive A phases are separated by an interval of less than 2 s, they are combined as a single phase A. If they are separated by an interval of 2 s or more, they are scored as independent events.

The A phases of cyclic alternating pattern.

Subtype classification. Phase A activities can be classified into 3 subtypes.

Image: Phase A subtypes
Subtype classification is based on the reciprocal proportion of high-voltage slow waves (EEG synchrony) and low-amplitude fast rhythms (EEG desynchrony) throughout the entire phase A duration. The 3 phase A subtypes are described as follows (Terzano et al 2002b).

Subtype A1. EEG synchrony (high-amplitude slow waves) is the predominant activity. If present, EEG desynchrony (low-amplitude fast waves) occupies less than 20% of the entire phase A duration. Subtype A1 specimens include delta bursts, K-complex sequences, vertex sharp transients, and polyphasic bursts with less than 20% of EEG desynchrony.

Subtype A2. The EEG activity is a mixture of slow and fast rhythms with 20% to 50% of phase A occupied by EEG desynchrony. Subtype A2 specimens include EEG arousals and polyphasic bursts with more than 20%, but less than 50%, of EEG desynchrony.

Subtype A3. The EEG activity is dominated by rapid low-voltage rhythms with more than 50% of phase A occupied by EEG desynchrony. Subtype A3 specimens include K-alpha, EEG arousals, and polyphasic bursts with more than 50% of EEG desynchrony. A movement artifact within a cyclic alternating pattern sequence is also classified as subtype A3.

Cyclic alternating pattern sequences include different phase A subtypes. The majority of EEG arousals occurring in non-REM sleep (87%) is inserted within the cyclic alternating pattern sequences and basically coincides with a phase A2 or A3. In particular, 95% of subtype A3 and 62% of subtype A2 meet the AASM criteria for arousals (Parrino et al 2001; Terzano et al 2002a). The broad overlap between arousals and subtypes A2 and A3 is further supported by their similar evolution in relation to age and to their positive correlation with the amount of light NREM sleep and negative correlation with the amount of deep NREM sleep.

Spectral composition. Power spectral analysis of cyclic alternating pattern components shows that the different phase A subtypes in non-REM sleep are variants of a continuous 2-fold process: an initial high-voltage slow-wave component, which predisposes the cerebral cortex to a greater readiness and opens the way to the more rapid activity, correlated with strong activating effects (De Carli et al 2004; Ferri et al 2005a). What distinguishes the single event is the build-up and reciprocal distribution of the EEG components. In the A1 phases of cyclic alternating pattern, which exclusively host K-complexes and equivalent slow-wave activities (vertex potentials and delta bursts), the starting delta power increase is maintained and prevails throughout the entire activation process. A balanced representation of slow and fast EEG frequency bands is the main characteristic of A2 phases, whereas rapid EEG activities are the dominant feature of A3 subtypes and arousals. This does not mean that all activating complexes exert equivalent effects on sleep structure and on autonomic functions. A hierarchical activation from the slower EEG patterns (moderate autonomic activation without sleep disruption) to the faster EEG patterns (robust autonomic activation associated with visible sleep fragmentation) has been described in different studies (Guilleminault and Stoohs 1995; Ferri et al 2000; Sforza et al 2000; Halasz et al 2004; Togo et al 2006). If AASM arousal is a sign of transient sleep discontinuity, the finding of phasic EEG delta activities during enhancement of autonomic functions indicates the possibility of physiological activation without sleep disruption (Halasz 1993). In effect, non-visible sleep fragmentation induced by acoustic tones has been associated with increased daytime sleepiness, indicating that the processes of sleep consolidation may be impaired (in this case by sensorial stimulation) without evidence of sleep discontinuity (Martin et al 1997).

The content you are trying to view is available only to logged in, current MedLink Neurology subscribers.

If you are a subscriber, please log in.

If you are a former subscriber or have registered before, please log in first and then click select a Service Plan or contact Subscriber Services. Site license users, click the Site License Acces link on the Homepage at an authorized computer.

If you have never registered before, click Learn More about MedLink Neurology  or view available Service Plans.