Author's personal copy

Original Article

Insomnia cycling with a 42-day infradian period: Evidence for two uncoupledcircadian oscillators?

Luca Vignatelli a,b,*, Simone Masetti c, Mario Amore d, Caterina Laterza d, Katia Mattarozzi e,Giulia Pierangeli b, Pietro Cortelli b, Stefano Vandi b, Roberto Vetrugno b, Giuseppe Plazzi b,Pasquale Montagna b

a Ambulatorio di Neurologia, Polo Sanitario ‘‘Barberini” di Crevalcore, AUSL di Bologna, Italyb Dipartimento di Scienze Neurologiche, Università di Bologna, Italyc Dipartimento di Fisica, Università di Bologna, Italyd Divisione di Psichiatria, Dipartimento di Neuroscienze, Università di Parma, Italye Dipartimento di Psicologia, Università di Bologna, Italy

a r t i c l e i n f o

Article history:Received 10 June 2009Received in revised form 8 September 2009Accepted 14 September 2009Available online 4 February 2010

Keywords:Sleep initiation and maintenance disorders,Bipolar disorderChronobiology disordersCircadian rhythmSuprachiasmatic nucleusTheoretical models

a b s t r a c t

Objective: To describe the unique case of a middle-aged woman with severe insomnia recurring with aregular infradian period without any other significant clinical condition. To infer the existence of a circa-dian dysfunction modeled according to the physical phenomenon of the ‘‘beats.”Patient/Methods: A two-year prospective observation by means of a sleep log was performed during thepatient’s normal life. She underwent one month of motor activity recording and also polysomnography,circadian rhythm of body core temperature and psychiatric evaluation during periods with and withoutinsomnia.Results: Visual inspection of the 293-day plot of the sleep log disclosed a regular 42-day rhythm ofinsomnia recurrence confirmed by a Discrete Fourier Transform. During the periods of insomnia, lasting5–7 days, only moderate mood symptoms (depressive overlapping hypomaniac symptoms) were present.Treatment with sodium valproate was effective in curtailing insomnia.Conclusion: The wax and wane infradian modulation of the sleep length suggested the presence of a basicmechanism similar to the physical phenomenon of the ‘‘beats,” i.e., a long period modulation of theamplitude of an oscillating system due to the interference of two uncoupled oscillators with a slightlydifferent oscillation frequency. Hypothesizing a dysfunction of the circadian component of sleep, namelytwo uncoupled circadian cycles, a simple mathematical model estimated the difference of their periods ofoscillation |34 ± 2 min| and reproduced the sleep-log data of the drug-free period of observation.

� 2010 Elsevier B.V. All rights reserved.

1. Introduction

Complex behaviours may manifest with regular infradian peri-odicity both in healthy people and in patients with psychiatric dis-orders [1–4]. Menstrual migraine has, by definition, a monthlyrepetition, and cluster headache may recur with infradian regular-ity [5]. Some cyclic disorders have been described in the field ofhematology, and cyclic neutropenia, the most studied, has a regu-lar periodicity of 21 days [6].

Insomnia, the commonest sleep complaint and symptom of sev-eral neurological and psychiatric disorders, may display an infra-dian recurrence if driven by mood symptoms in a rapid-cyclingbipolar disorder [7]. Insomnia cycling with regular infradian recur-rence, however, has never been reported as a dominant complaint.

We describe the case of a middle-aged woman with a severecyclic insomnia whose original points were (1) the regular infra-dian period of recurrence demonstrated upon extensive prospec-tive clinical observation and neurophysiological monitoring, (2)the presence of moderate mood symptoms following the insomnia,(3) a good overall clinical response to treatment with sodium val-proate. In this patient we hypothesized a dysfunction of the circa-dian component of sleep, which we modeled according to thephysical phenomenon of the ‘‘beats,” i.e., a long period modulationof the amplitude of an oscillating system as the result of twouncoupled oscillators cycling with slightly different frequency.

2. Case description

2.1. Subject

In June 2001, a 43-year-old woman with no history of neurolog-ical or psychiatric disorders came to the Sleep Center of the Depart-

1389-9457/$ - see front matter � 2010 Elsevier B.V. All rights reserved.doi:10.1016/j.sleep.2009.09.006

* Corresponding author. Address: Dipartimento di Scienze Neurologiche, Univer-sità di Bologna, 40123 Bologna, Italy. Tel.: +39 0512092950; fax: +39 0512092963.

E-mail address: vigna@interfree.it (L. Vignatelli).

Sleep Medicine 11 (2010) 343–350

Contents lists available at ScienceDirect

Sleep Medicine

journal homepage: www.elsevier .com/locate /s leep

Author's personal copy

ment of Neurological Sciences, University of Bologna, Italy, com-plaining of ‘‘recurrent insomnia.” She lives in a cottage in a smalltown of about 1500 inhabitants (429 meters of altitude, latitude43.75, longitude 12.31) in a mountainous region of Central Italy.She had been in an orphanage from 2 to 14 years of age becauseof her family’s economic difficulties, had a regular menstrual cyclesince age 13, married at 17, had one daughter at 18 and held a mid-dle school diploma. She worked as a waitress. She ate regularly, didnot smoke or abuse alcohol or drugs. She defined her sleep (exceptfor the ‘‘insomnia period;” see below) as ‘‘light but satisfactory andrest-giving;” she took afternoon naps of about 60 min two or threetimes a week.

2.2. Clinical features

The sleep disorder began when the patient was 36 years old,coinciding in the onset epoch with her husband’s night shifts. Fromthat time up to our observation – an interval of nine years – sheexperienced the following monthly features (we will call it the‘‘insomnia period”). On a given night, she had difficulty fallingasleep; after finally falling asleep, sleep was fragmented with asubjective restriction of total sleep time. In the following nightsthe disorder worsened with increasing sleep latency and a furtherrestriction of total subjective sleep time. When awake in the nightshe lay in bed quietly and sometimes got up and to do her house-work. This insomnia waxed to a maximum in 3–4 days, with 1 or 2nights of total or nearly total subjective absence of sleep. A gradualrecovery of subjective sleep occurred in the following 2–4 days.During the daytime in this ‘‘insomnia period” she usually felt a lit-tle depressed, anxious and tired, with a reduced ability to concen-trate for the first two days; in the following 3–5 days she wouldreport a slight elevation of mood, irritability, pressure of ideasand speech, distractibility, increased involvement in social andworking activities, and moderate somatic symptoms (headache,musculoskeletal and abdominal pain, itch). Nevertheless, she wasable to function normally in her family, social and working activi-ties. Daytime naps seldom occurred during this ‘‘insomnia period,”and there was no recovery of sleep during the day. Her relatives,directly interviewed, confirmed her clinical report. Neurologicalexamination, EEG and brain MRI scan were normal.

3. Methods

3.1. Protocol of observation and intervention

A long prospective clinical observation (sleep log and psychiat-ric evaluation) of the patient during her normal life was devised.Neurophysiological observations (motor activity recording, poly-somnography, circadian rhythm of body core temperature) werealso planned to be performed during periods with and withoutinsomnia of one single infradian cycle by means of brief recoveriesin the sleep laboratory. The observation was organized in a firstperiod without any pharmacological intervention (drug-free peri-od) and a second period of prospective sleep log observation duringthe therapy with sodium valproate (therapy period).

3.2. Sleep log

The patient recorded the hours slept in a sleep log (30 min res-olution) each morning on rising. Beginning at 12 a.m., sleep log re-sults were plotted in 24-hour segments, each 24-hour segmentdisplayed beneath the preceding one. By analogy with the proce-dure by Kim and Young [8] to verify if a homeostatic ‘‘sleep debt”was produced by the insomnia period, we arbitrarily compared thenumber of hours subjectively slept seven days before the nadir of

insomnia (= night with fewest hour slept) with those slept sevendays after the nadir. To reveal periodicities and the relativestrengths of any periodic components, we performed a DiscreteFourier Transform of the daily recordings (hours slept in a day)for the period of observation when the patient was therapy free.This power spectrum gives a plot of the portion of the signal’spower falling within given frequency bins.

3.3. Psychiatric evaluation

A standardized interview was carried out by a psychiatrist toyield demographic, socioeconomic and historical information.The patient’s history was reviewed for psychotic episodes, suicideattempts, substance abuse, psychiatric or neurological illnessesand family psychiatric history. The Structured Clinical Interviewfor DSM-IV was used for evaluation. Personality disorders wereevaluated using the Structured Interview for DSM-IV Personality.Psychopathological status was assessed using the Brief PsychiatricRating Scale (BPRS) [9], the Young Mania Rating Scale (YMRS) [10],the Hamilton Depression Rating Scale (HAMDS) [11], and the Clin-ical Global Impression for Bipolar Disorder (CGI-BP) [12].

3.4. Motor activity recording

Spontaneous motor activity was measured by a wrist actigraph(Actigraph Mini Motionlogger, Ambulatory Monitoring Inc. Ards-ley, NY, USA) worn on the non-dominant wrist. Data were collectedin 60-second epochs in 24-hour periods. The collected data wereanalyzed by a computer program (Action 3, Ambulatory Monitor-ing Inc.). The mean (with standard deviation, SD) and median val-ues of motor activity (movement counts per minute) werecalculated. Beginning at 12 a.m., histograms of data were plottedin 24-hour segments and each 24-hour segment was displayed be-neath the preceding one.

3.5. Polysomnography

The sleep–wake cycle was monitored by a portable dynamicpolygraphic recorder (VITAPORT�) recording the electroencephalo-gram (EEG: C3-A2, CZ-A2, O2-A1), right and left electro-oculogram(EOG), electrocardiogram (ECG) and electromyogram (EMG) of themylohyoideus and the right and left tibialis anterior muscles. Theall night summary recording variables were derived from the vi-sual scoring of recordings in 30-second epochs by an expert sleeptechnician blinded with respect to the sleep log data. Polygraphicdata were analyzed according to standard criteria [13].

3.6. Circadian rhythm of body core temperature (BcT�)

BcT� was evaluated by monitoring rectal temperature every2 min by means of a Mini–logger™ portable device. The analysisof rythmicity was performed evaluating the time series for BcT�according to the single cosinor method [14] using a computerizedprocedure [15]. This procedure determines whether or not there isa rhythm within a 24-hour period and evaluates the followingparameters (and 95% confidence interval): (1) mesor (Midline Esti-mating Statistic of Rhythm): rhythm adjusted 24-hour average; (2)amplitude: the difference between the maximum value measuredat the acrophase and the mesor in the cosine curve; (3) acrophase:lag between reference time (12 p.m.) and the time of highest valueof the cosine function used to approximate the rhythm. The 24-hour pattern of BcT� was evaluated calculating the mean BcT� ofeach hour.

344 L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350

Author's personal copy

4. Results

4.1. Drug-free period

4.1.1. Sleep logTwo hundred ninety-three days of continuous observation were

obtained in the drug-free period (from June 26, 2001 to April 13,

2002). Seven periods of insomnia cycling with a long regular infra-dian period were observed (Fig. 1A). Each period lasted from 5 to7 days. No correlation was found with menstrual cycles, periodsof work or vacation, seasonality or moon cycles. In the seven daysfollowing the nadir of insomnia the patient slept on average onehour and 36 min more per day than the seven days preceding thenadir. The Discrete Fourier Transform Power Spectrum revealed a

Fig. 1. Plot of the sleep log. Horizontal bold bars denote the hours slept. Beginning at 12 am, sleep log results are plotted in 24-hour segments (x axis, hours per day). Each 24-hour segment is displayed beneath the preceding one (y axis, calendar date). (A) The period free from therapy (293 days, from June 26, 2001 to April 13, 2002). Seven periodsof insomnia (indicated by the circled letter ‘‘I”) cycling with a long regular infradian period were observed. See the corresponding period with actigraphic recording (32 days,from January 28, 2002 to March 1, 2002 – marked ‘‘ACT” on the right of the plot) in Fig. 3. (B) The period with sodium valproate therapy (262 days, from April 14, 2002 to May18, 2003 – marked ‘‘THERAPY” on the left of the plot). Only 3 insomnia periods (indicated by the circled letter ‘‘I”) were observed at the beginning of the therapy period. In thelast 92 days no insomnia periods were recorded.

L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350 345

Author's personal copy

principal cycle at frequency f0 = 0.0240 cycles/days (±0.0017), cor-responding to a period T0 = 42 days (±3). The second and third har-monics, at frequencies 2f0 and 3f0, were also observed (Fig. 2).

4.1.2. Psychiatric evaluationThe HAMDS, YMRS and CGI-BP scored 18–20 (HAMDS screened

level of depression: 10–13 mild; 13–17 mild to moderate; >17moderate to severe), 24–27 (normal YMRS score: 620), and 4(moderately ill at CGI-BP severity scale; range: from 1 = normalto 7 = among the most extremely ill patients) points respectivelyon the insomnia days, with complete remission on the intercriticaldays.

4.1.3. Motor activity recordingThirty-two consecutive days of actigraphic recording were

made from the 217th day of the sleep log (January 28, 2002) tothe 248th day (March 1, 2002). One period of insomnia was cap-tured (Fig. 3A). Mean motor activity was 106.7 movements/min(SD 21.50) in the 7 days of the insomnia period (from 8.30 p.m.to 8.29 a.m.) and 64.2 movements/min (SD 21.0) in the remaining25 days.

4.1.4. Polysomnography and circadian rhythm of BcT�Along with the actigraphic recording, during the intercritical

period, one night of sleep (218th day of sleep log) and 72 consec-utive hours of BcT� (days 218th, 219th, 220th) were recorded(Fig. 3B). During the insomnia period 72 consecutive hours of pol-ysomnography (days 231st, 232nd, 233rd) and 48 consecutivehours of BcT� (days 231st, 232nd) were also obtained (Fig. 3C).

Sleep data are compared in Table 1. In the intercritical period,total sleep time (TST) was 380 min, with normal sleep latencyand 22% of stages 3–4 NREM sleep. In the insomnia period onenight of nearly total agrypnia (232nd day) was captured (sleep effi-ciency of 11%, TST of 37 min, 1 min of stage 2 NREM sleep, absenceof 3–4 NREM and REM sleep). Another two nights of strongly de-creased sleep efficiency (33% and 37%) preceded and followed.Throughout these three nights the patient lay in bed quietly. A cir-cadian rhythm of the BcT� was confirmed on all days of recording,with a physiological nocturnal fall of BcT� comparable for theinsomnia and intercritical periods. Mesor and rhythm amplitudewere normal. Acrophase ranged from 14.08 to 14.44 h.m. During

the insomnia period Mesor was +0.4 �C higher than during theintercritical period.

4.1.5. Other laboratory testsDuring the insomnia and intercritical periods blood and urine

parameters (including toxicological testing) were normal, exceptfor a slight increase in CPK (380 U/L, normal value range 0–180)and total and direct bilirubin (1.04, normal value range 0.20–1.00; 0.36 mg/dl, normal value range 0–0.25) in the insomniaperiod.

4.2. Therapy period of observation

Suspecting a condition similar to a rapid-cycling bipolar disor-der, the patient was given sodium valproate therapy (500 mg bid,after ten weeks of tapering; plasma level of 68 mg/L at 107 daysfrom the start) on April 14, 2002. The log was filled in for 170 days(from April 14, 2002 to September 29, 2002) and for another92 days (from February 15, 2003 to May 18, 2003) (Fig. 1B). Duringthe first period of sleep log, 3 insomnia periods were observed ofthe 4 expected. Then, during a further 138 days without sleep logcompilation, no periods of insomnia of the 3 expected were re-ported. Finally, in the last 92 days of sleep log observation noinsomnia periods were observed of the 2 expected. Stable euthy-mia was also reported throughout. Thus, treatment with sodiumvalproate was effective in curtailing insomnia.

5. Discussion

We report the clinical and neurophysiological features of a pa-tient with insomnia cycling with a regular 42-day infradian period,in the absence of any structural neurological disorder. Moderatemood symptoms, present only in the insomnia period (depressiveoverlapping hypomaniac symptoms), were apparently triggeredby the insomnia. Mood symptoms were never experienced as ‘‘ill-ness” by the patient or her relatives, nor did they produce substan-tial working or family difficulties. Another feature was the goodresponse of the whole picture to sodium valproate therapy.

From the nosographical point of view the observed picture canbe labeled as either a primary insomnia or a rapid-cycling bipolardisorder, in both cases with peculiar features. Our patient sharessome aspects of primary insomnia, such as difficulty falling asleepand maintaining sleep together with behavioural and thermichyperarousal state in the absence of a clear etiologic factor [16].The moderate psychiatric symptoms could be considered second-ary to sleep deprivation by analogy with the hypothesis that‘‘changes in the timing of the sleep–wake cycle relative to the tim-ing of other circadian rhythms may trigger affective episodes,” pro-posed by Wehr and collegues [17,18]. But a regular infradianrecurrence has never been reported as a feature of primary insom-nia. On the other hand, we may classify our patient as affected by aclinical phenotype within the spectrum of rapid-cycling bipolardisorders with prevailing sleep symptoms. Among the various bio-logical rhythms implicated in mood disorders, the sleep–wake cy-cle is believed to play a major role [19,20]. Clinical observationssuggest a close association between sleep loss and onset of manicepisodes in patients with manic-depressive disorders (see thecomprehensive review by Harvey [19]) and specifically in thosewith rapid cycling illness with 48-hour cycles [1,18,21–27]. Otherinvestigators have reported that treatment of patients with bipolardisorder with extended sleep may be helpful in preventing manicepisodes and rapid cycling [23]. Moreover, not only sleep durationbut also shifts in the timing of sleep can induce subsequent moodchanges in patients with manic-depressive illness [28–30]. Lastly,some subsyndromal sleep alterations seem to persist during the

Fig. 2. Discrete Fourier Transform Power Spectrum of the sleep log during theperiod of observation when the patient was free from therapy. A principal cycle atfrequency f0 = 0.0240 cycles/days ±0.0017, corresponding to a period T0 = 42 days±3, was disclosed. Second and third harmonics at about 21 and 14 days were alsoobserved.

346 L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350

Author's personal copy

euthymic phase [19,20] reflecting enduring trait or vulnerabilitymarkers, although findings are not always consistent and longitu-dinal studies are lacking. Thus, even if in our case the whole picturedoes not fulfill the DSM-IV criteria for any bipolar disorder, it may

include one of the four alternative definitions of rapid-cyclingbipolar disorder proposed by Maj et al. [31], considering recurrentinsomnia without any accompanying affective component as oneend of the spectrum of bipolar affective disorder.

Fig. 3. (A) Thirty-two consecutive days of actigraphic activity (from January 28, 2002 to March 1, 2002 – see also the corresponding period in Fig. 1) during the period freefrom therapy. Beginning at 12 a.m., histograms (black denotes movements) are plotted in 24-hour segments, each 24-hour segment is shown beneath the preceding one. (B) A24-hour hypnogram and 3 consecutive days of BcT� during the intercritical period with normal sleep (corresponding to the days 218th, 219th, 220th of the sleep log; see alsothe Table 1). (C) Three consecutive 24-hour hypnograms and 2 consecutive days of BcT� during the insomnia period (corresponding to the days 231st, 232nd, 233rd of thesleep log; see also the Table 1).

L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350 347

Author's personal copy

The feature peculiar to our patient was a cycle of 42 days of pre-vailing insomnia. To our knowledge, only three reports share withour case a similar periodicity, even if the symptoms are clearly dif-ferent. Wehr et al. [32] observed a patient with a rapid-cyclingbipolar disorder who showed an oscillation of sleep symptomsconcurrently with a cycle of 6–7 weeks between depression andmania–hypomania (see Figs. 1 and 4 of the original paper). Theauthors stated: ‘‘When depressed, he was inactive, went to bedlate, slept 10–12 h, and arose late in the morning or afternoon.When hypomaniac, he was hyperactive, went to bed earlier, slept3–4 h, and arose before dawn. When he switched from depressionto hypomania he often went without sleep for one or more alter-nate night, with a pattern of 48-hour sleep–wake cycles.” In an

experimental setting of temporal isolation without regular sche-dule this patient had a free-run pattern of the sleep–wake cycle(see Fig. 1 of the original paper). The patient did not respond totherapy with valproate and partially responded to a forced sche-dule of sleep–wake cycle. Bryson and Martin [33] observed a 39-year-old male with regular mood-swings between mania (18 dayson average), normality (4 days) and depression (16 days), togetherwith a mirror cyclicity of urinary 17-ketosteroids and eosinophilblood cell count. The best fitting period of one whole cycle was40.96 in the untreated period of observation [34]. Unfortunately,no information was reported about the patient’s sleep characteris-tics. De Silva et al. [35] reported the case of a man who, instead ofinsomnia, manifested a stupor (EEG features of stage 1 sleep)

Table 1Subjective and neurophysiologic data on sleep, during 3 days of the intercritical period (days 218th, 219th, 220th) and of the insomnia period (days 231st, 232nd, 233rd).

Intercritical period Insomnia period

218th day 219th day 220th day 231st day 232nd day 233rd day

Sleep logEstimated sleep time (min) 390 450 510 60 <30 150

Actigraphic recordinga

Mean motor activity (movement/min) 46.6 63.7 50.0 89.6 127.0 118.3Standard deviation (movement/min) 68.6 81.3 78.0 106.6 105.9 122.4Median motor activity (movement/min) 14 22 12 24 128 46

Sleep recordingsNight

Total sleep period (min) 493 416 346 497Total sleep time (min) 380 137 37 186Stage 1 NREM sleep (min – %) 16–4.2 40–29.2 36–97.3 14–7.5Stage 2 NREM sleep (min – %) 191–50.3 78–56.9 1–2.7 98–52.7Stages 3–4 NREM sleep (min – %) 83–21.8 19–13.9 0–0 62–33.3REM sleep (min – %) 90–23.7 0–0 0–0 12–6.4Sleep efficiency (%) 77 33 11 37Sleep latency (min) 17 124 20 8REM sleep latency (min) 6 – – 57PLM index (movement/h) 4.5 – – –

Daylight napTotal sleep time (min) 123 – – 17REM sleep latency (min) 72 – – –

Rhythmometric analysis of body core temperatureMesor (�C) 36.97 36.97 37.37 37.37Amplitude (�C) 0.30 0.40 0.36 0.45Acrophase (h.m) 14.44 14.08 14.20 14.44

PLM = periodic limb movement.a Actigraphic data are reported from 20.30 h.m to 8.29 h.m.

Fig. 4. Comparison between the experimental sleep-log data and the sleep behaviour predicted by the model calculation. The computed sleeping time is the duration of thesleep-phase of the model. The model predicts insomnia periods appearing with an infradian recurrence of 42 days.

348 L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350

Author's personal copy

recurring every 4–6 weeks. These episodes lasted 4–5 h and wouldbe repeated over 4–5 days before resolution. He could be arousedonly with vigorous stimuli, and on arousal he had ‘‘impaired vigi-lance, dysarthria, limb dysmetria and gait ataxia.” Olanzapineand lithium were successful in controlling the symptoms.

Whatever nosographical collocation is considered, from a path-ophysiological point of view we tried to explain the 42-day rhythmof insomnia referring to the two-process model of sleep regulationbased on the existence of a circadian process and a homeostaticprocess (so called ‘‘Process C” and ‘‘Process S”) [36,37]. Accordingto this model sleep onset is triggered when a ‘‘sleep-regulating var-iable” (Process S) approaches an upper threshold and awakeningoccurs when it reaches a lower threshold. The thresholds show acircadian rhythm controlled by a circadian pacemaker (ProcessC). We observed that the 42-day modulation of the amplitude ofsleep duration could be conceptually explained by the so-calledphysical phenomenon of the ‘‘beats,” that is, the long period mod-ulation of the amplitude of an oscillating system due to the inter-ference of two uncoupled oscillators with a slightly differentoscillation frequency. According to this hypothesis, we modeledthe observed data assuming that the circadian component of sleepin our patient could be affected by a dysfunction consisting in thecoexistence of the ‘‘normal” circadian oscillator, entrained to theexternal stimulus of the 24-hour light–darkness cycle, and a ‘‘path-ological” oscillator dominated by an endogenous stimulus with aperiod slightly differing from the 24 h, insensitive to the externalstimulus. Assuming also that the resulting circadian rhythm isthe superposition of the normal cycle and a pathological one,‘‘beats” are generated: the total oscillation will have a principalperiod of 24 h, but will present an amplitude modulation thatexhibits a maximum every 42 days; in the nearest of this maxi-mum the sleep onset threshold cannot be reached. This corre-sponds to the emergence of the infradian episodes of insomnia.In this conceptual framework the mathematical modeling of thesleep log data, together with the previous simple assumptions nec-essary to reproduce the clinical manifestation, predicted theappearance of the insomnia period (Fig. 4) and estimated the abso-lute value of the difference between the periods of the normal andpathological oscillators in 34 ± 2 min (analysis not shown).

But, there are some limitations to our hypothesis. First, thepresence of other high peaks in the spectral analysis of the sleeplog corresponding to 21-day and 14-day rhythms could testify toa somewhat more complex oscillatory phenomenon than whatwe modeled. Nevertheless, as those harmonics only determinethe shape of the dominant 42-day cycle, this finding does not inval-idate the basic principle of interference of the two basic oscillators.Second, our explanation does not provide for a role of the homeo-static component of sleep, namely a pathological interaction be-tween the circadian and homeostatic processes. Indeed, there is agrowing amount of data suggesting that these two processes, evenif independently generated, may interact in determining the tim-ing, duration and quality of both sleep and wakefulness. For in-stance, clock genes regulating the circadian system in thesuprachiasmatic nucleus (SCN), the master of circadian rhythmsin mammals located in the anterior hypothalamus [38], seem toplay a role in sleep homeostasis when expressed in brain areas out-side the SCN. Moreover, certain ‘‘circadian” phenotypes (morningand evening types), in some individuals, seem to result from theaccelerated build up of sleep pressure [39]. A ‘‘circadian hypothe-sis,” however, is supported by the following elements. Therapywith valproate resolved the insomnia. This molecule has a provenaction on circadian locomotor behaviour of the drosophila [40], theability to decrease the sensitivity of melatonin to light and tolengthen the circadian period in humans [41], and may alter theexpression of circadian genes in the amygdala of mice [42]. More-over, a circadian alteration has been proposed to account for the

pathogenesis of bipolar disorders [2,18,21,30,43,44] and has alsobeen claimed as the pathophysiologic background of primaryinsomnia [45]. In the context of bipolar disorders, the model ofthe beats of two interfering circadian cycles was heralded by Hal-berg [34] and reconsidered by others [18,32,44,46,47]. Recently,studies in human populations have begun to identify polymor-phisms in circadian genes associated with bipolar disorder [48–50] and with the ‘‘insomniac” sleep pattern in bipolar patients[51]. Furthermore, an animal model of mania was obtained creat-ing a mouse with CLOCK gene mutation [52]. Finally, changes inthe phase relationship between two circadian oscillators (whenin antiphase to each other) have been shown to account for infra-dian recurrent complex behaviour in animals, such as recurrentmigratory restlessness (‘‘Zugunrhue”) in migratory songbirds,making some of them switch (seasonally) from being exclusivelydiurnal to either exclusively nocturnal or active both day and night[53]. Even if the neural substrate of this hypothetic circadian alter-ation remains enigmatic, the regulation of circadian rhythms dis-plays some features that may be the source of uncoupledoscillators: circadian rhythms seem to depend upon a web whoseeffectors, both inside and outside the brain, are hierarchically coor-dinated and synchronized by the SCN [54]. The SCN itself has neu-rons with heterogenic frequency signals [55], and the SCN of onebrain hemisphere may be forced to cycle in antiphase with respectto the nucleus on the other side [56].

In conclusion, this unique case of a regular 42-day cyclinginsomnia suggests that chronic sleep disorders could be studiedfrom the novel point of view of infradian recurrence.

Disclosures

The authors have nothing to disclosure.

Acknowledgements

Our warm thanks go to the patient and her family for their col-laboration. Giorgio Barletta and Laura Solieri gave technical sup-port. Anne Collins and Alessandra Laffi edited the English text.

References

[1] Bunney WE, Hartmann EL, Mason JW. Study of a patient with 48-hour manicdepressive cycles. Part II strong positive correlation between endocrine factorsand manic defense patterns. Arch Gen Psychiatry 1965;12:619–25.

[2] Eastwood MR, Whitton JL, Kramer PM, Peter AM. Infradian rhythms. Acomparison of affective disorders and normal persons. Arch Gen Psychiatry1985;42:295–9.

[3] Gjessing R, Gjessing L. Some main trends in the clinical aspects of periodiccatatonia. Acta Psychiatr Scand 1961;37:1–13.

[4] Hersey RB. Emotional cycles in man. J Mental Sci 1931;77:151–69.[5] Zee PC, Grujic ZM. Neurological disorders associated with disturbed sleep and

circadian rhythms. In: Turek FW, Zee PC, editors. Regulation of sleep andcircadian rhythms. New York, Basel: Marcel Dekker Inc.; 1999. p. 557.

[6] Haurie C, Dale DC, Mackey MC. Cyclical neutropenia and other periodichematological disorders: a review of mechanisms and mathematical models.Blood 1998;8:2629–40.

[7] American Psychiatric Association. Diagnostic and statistical manual of mentaldisorders. Washington, DC: American Psychiatric Association; 1994.

[8] Kim H, Young T. Subjective daytime sleepiness: dimensions and correlates inthe general population. Sleep 2005;28:625–34.

[9] Ventura J, Green M, Shaner A, Liberman R. Training and quality assurance withthe Brief Psychiatric Rating Scale: ‘‘the drift busters”. Int J Methods PsychiatrRes 1993;3:221–7.

[10] Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability,validity and sensitivity. Br J Psychiatry 1978;133:429–35.

[11] Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry1960;23:56–62.

[12] Spearing MK, Post RM, Leverich GS, Brandt D, Nolen W. Modification of theClinical Global Impressions (CGI) scale for use in bipolar illness (BP): the CGI-BP. Psychiatry Res 1997;73:159–71.

[13] Rechtschaffen A, Kales A. A manual of standardized terminology: techniquesand scoring system for sleep stages of human subjects. Los Angeles: BrainInformation Service/Brain Research Institute, UCLA; 1968.

L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350 349

Author's personal copy

[14] Halberg F, Reinberg A. Rythmes circadiens et rhythmes de basses fréquencesen physiologie humaine. J Physiol (Paris) 1967;59:S117–200.

[15] Mojon A, Fernandez JR, Hermida RC. ChronoLab: an interactive software forchronobiologic time series analysis written for Macintosh computer. Vigo(Spain): Bioengineering and Chronobiology Laboratories, E.T.S.I.Telecommunication, University of Vigo; 1994.

[16] Roth T, Roehrs T, Pies R. Insomnia: pathophysiology and implications fortreatment. Sleep Med Rev 2007;11:71–9.

[17] Wehr TA, Sack D, Rosenthal N, Duncan W, Gillin JC. Circadian rhythmdisturbances in manic-depressive illness. Fed Proc 1983;42:2809–14.

[18] Wehr TA, Sack D, Rosenthal N. Sleep reduction as a final common pathway inthe genesis of mania. Arch Gen Psychiatry 1987;144:201–4.

[19] Harvey AG. Sleep and circadian rhythms in bipolar disorder: seekingsynchrony, harmony, and regulation. Am J Psychiatry 2008;165:820–9.

[20] Plante DT, Winkelman JW. Sleep disturbance in bipolar disorder: therapeuticimplications. Am J Psychiatry 2008;165:830–43.

[21] Leibenluft E, Albert PS, Rosenthal N, Wehr TA. Relationship between sleep andmood in patients with rapid-cycling bipolar disorder. Psychiatry Res1996;63:161–8.

[22] Wehr TA, Goodwin FK, Wirz-Justice A, Breitmaier J, Craig C. 48-Hour sleep–wake cycle in manic depressive illness. Naturalistic observation and sleepdeprivation experiments. Arch Gen Psychiatry 1982;39:559–65.

[23] Wirz-Justice A, Quinto C, Cajochen C, Werth E, Hock C. A rapid-cycling bipolarpatient treated with long nights, bedrest, and light. Biol Psychiatry1999;45:1075–7.

[24] Hudson JI, Lipinski JF, Frankenburg FR, Grochocinski VJ, Kupfer DJ.Electroencephalographic sleep in mania. Arch Gen Psychiatry1988;45:267–73.

[25] Jenner FA, Gjessing LR, Cox JR, Davies-Jones A, Hullin RP, Hanna SM. A manicdepressive psychotic with a persistent forty-eight hour cycle. Br J Psychiatry1967;113:895–910.

[26] Schreiner R, Mirisch S, Vesely Z, Wiegand MH. Sleep and sleep–wake cycle inan 81-year-old patient with de novo ultra-rapid cycling bipolar disorder. EurArch Psychiatry Clin Neurosci 2001;251:29–31.

[27] Voderholzer U, Weske G, Ecker S, Riemann D, Gann H, Berger M.Neurobiological findings before and during successful lithium therapy of apatient with 48-hour rapid-cycling bipolar disorder. Neuropsychobiology2002;45:S13–9.

[28] Riemann D, Hohagen F, König A, Schwarz B, Gomille J, Voderholzer U, et al.Advanced vs. normal sleep timing: effects on depressed mood after responseto sleep deprivation in patients with a major depressive disorder. J AffectDisord 1996;37:121–8.

[29] Sack DA, Nurnberger J, Rosenthal NE, Ashburn E, Wehr TA. Potentiation ofantidepressant medications by phase advance of the sleep–wake cycle. Am JPsychiatry 1985;142:606–8.

[30] Wehr TA, Wirz-Justice A, Goodwin FK, Duncan W, Gillin JC. Phase advance ofthe circadian sleep–wake cycle as an antidepressant. Science 1979;206:710–3.

[31] Maj M, Magliano L, Pirozzi R, Formicola AMR, Tortorella A. Reliability andvalidity of four alternative definitions of rapid-cycling bipolar disorder. Am JPsychiatry 1999;156:1421–4.

[32] Wehr TA, Turner EH, Shimada JM, Love CH, Barker C, Leibenluft E. Treatment ofa rapidly cycling bipolar patient by using extended bed rest and darkness tostabilize the timing and duration of sleep. Biol Psychiatry 1998;43:822–8.

[33] Bryson RW, Martin DF. 17-Ketosteroid excretion in a case of manic-depressivepsychosis. Lancet 1954;2:365–7.

[34] Halberg F. Physiologic considerations underlying rhythmometry, with specialreference to emotional illness. In: de Ajuriaguerra J, editor. Cycles Biologiqueset Psychiatrie. Geneva: Masson et Cie; 1968. p. 73.

[35] De Silva DA, Ling LL, Yeong NB, Puvanendran K. An unusual case of episodicstupor. Sleep Med 2006;7:380–1.

[36] Daan S, Beersma DGM, Borbély AA. Timing of human sleep: recovery processgated by a circadian pacemaker. Am J Physiol 1984;246:R161–78.

[37] Borbély AA, Achermann P. Concepts and models of sleep regulation: anoverview. J Sleep Res 1992;1:63–79.

[38] Mistlberger RE. Circadian regulation of sleep in mammals: role of thesuprachiasmatic nucleus. Brain Res Rev 2005;49:429–54.

[39] Franken P, Dijk DJ. Circadian clock genes and sleep homeostasis. Eur J Neurosci2009;29:1820–9.

[40] Dokucu ME, Yu L, Taghert PH. Lithium- and valproate-induced alterations incircadian locomotor behaviour in drosophila. Neuropsychopharmacology2005;30:2216–24.

[41] Hallam KT, Olver JS, Norman TR. Effect of sodium valproate on nocturnalmelatonin sensitivity to light in healthy volunteers.Neuropsychopharmacology 2005;30:1400–4.

[42] Ogden CA, Rich ME, Schork NJ, Paulus MP, Geyer MA, Lohr JB, et al. Candidategenes, pathways and mechanisms for bipolar (manic-depressive) and relateddisorders: an expanded convergent functional genomics approach. MolPsychiatry 2004;9(11):1007–29.

[43] Gottschalk A, Bauer MS, Whybrow PC. Evidence of chaotic mood variation inbipolar disorder. Arch Gen Psychiatry 1995;52:947–59.

[44] Mansour HA, Monk TH, Nimgaonkar VL. Circadian genes and bipolar disorder.Ann Med 2005;37:196–205.

[45] Pigeon WR, Perlis ML. Sleep homeostasis in primary insomnia. Sleep Med Rev2006;10:247–54.

[46] Atkinson M, Kripke DF, Wolf SR. Autorhythmometry in manic-depressives.Chronobiologia 1975;2:325–35.

[47] Kronauer RE, Czeisler CA, Pilato SF, Moore-Ede MC, Weitzman ED.Mathematical model of the human circadian system with two interactingoscillators. Am J Physiol 1982;242:R3–R17.

[48] Nievergelt CM, Kripke DF, Barrett TB, Burg E, Remick RA, Sadovnick AD, et al.Suggestive evidence for association of the circadian genes PERIOD3 and ARNTLwith bipolar disorder. Am J Med Genet B Neuropsychiatr Genet2006;141B:234–41.

[49] Mansour HA, Wood J, Logue T, Chowdari KV, Dayal M, Kupfer DJ, et al.Association study of eight circadian genes with bipolar I disorder,schizoaffective disorder and schizophrenia. Genes Brain Behav 2006;5:150–7.

[50] Kripke DF, Nievergelt CM, Joo E, Shekhtman T, Kelsoe JR. Circadianpolymorphisms associated with affective disorders. J Circadian Rhythms2009;7:2.

[51] Serretti A, Benedetti F, Mandelli L, Lorenzi C, Pirovano A, Colombo C, et al.Genetic dissection of psychopathological symptoms: insomnia in mooddisorders and CLOCK gene polymorphism. Am J Med Genet B NeuropsychiatrGenet 2003;121B:35–8.

[52] Roybal K, Theobold D, Graham A, DiNieri JA, Russo SJ, Krishnan V, et al. Mania-like behavior induced by disruption of CLOCK. Proc Natl Acad Sci USA2007;104:6406–11.

[53] Bartel PA, Gwinner E. A separate circadian oscillator controls nocturnalmigratory restlessness in the songbird Sylvia borin. J Biol Rhythms2005;20:538–49.

[54] Fuller PM, Gooley JJ, Saper CB. Neurobiology of the sleep–wake cycle: sleeparchitecture, circadian regulation, and regulatory feedback. J Biol Rhythms2006;21:482–93.

[55] Herzog ED, Tosini G. The mammalian circadian clock shop. Semin Cell Dev Biol2001;12:295–303.

[56] de la Iglesia HO, Meyer J, Carpino Jr A, Schwartz WJ. Antiphase oscillation ofthe left and right suprachiasmatic nuclei. Science 2000;290:799–801.

350 L. Vignatelli et al. / Sleep Medicine 11 (2010) 343–350