www.elsevier.com/locate/sleep

Sleep Medicine 9 (2008) 425–433

Original article

Clinical, behavioural and polysomnographic correlatesof cataplexy in patients with narcolepsy/cataplexy

Katia Mattarozzi a, Claudia Bellucci a, Claudio Campi a, Carlo Cipolli a, Raffaele Ferri c,Christian Franceschini b, Michela Mazzetti a, Paolo Maria Russo a, Stefano Vandi b,

Luca Vignatelli b, Giuseppe Plazzi b,*

a Department of Psychology, University of Bologna, Bologna, Italyb Department of Neurological Sciences, University of Bologna, Bologna, Italy

c Department of Neurology I.C., Oasi Institute (IRCSS), Troina, Italy

Received 12 December 2006; received in revised form 17 May 2007; accepted 17 May 2007Available online 2 August 2007

Abstract

Background: Cataplexy is the main motor symptom of narcolepsy/cataplexy and is considered a form of rapid eye movement(REM) sleep motor dyscontrol appearing during wakefulness and elicited by emotions. This study examined the relationshipbetween the frequency of cataplectic attacks in patients with narcolepsy/cataplexy and (a) the clinical and behavioural characteris-tics of cataplectic attacks, including the emotional tone of trigger events, and (b) the polysomnographic characteristics of daytimesleepiness, nocturnal sleep structure and indices of motor disorders during sleep.Methods: A consecutive series of 44 first-diagnosed drug-naive patients with narcolepsy/cataplexy, fulfilling the International Classi-fication of Sleep Disorders, 2nd edition (ICSD-2) clinical and polysomnographic diagnostic criteria, were interviewed to estimate thefrequency and clinical characteristics of cataplectic attacks and the occurrence of REM sleep behaviour disorder (RBD). All patientsalso underwent a video-polysomnographic recording to assess their sleep parameters and indices of altered motor control during sleep.Results: Patients were divided into two groups on the basis of the frequency of cataplectic attacks, namely high-frequency (n = 30)or low-frequency (n = 14) depending on whether they estimated they had more or less than one attack per month. High-frequencypatients (with a larger proportion of men) reported attacks more often affecting mainly the head, jaw and shoulder muscles andexperienced more events among those listed as possible triggers of attacks. Sixty-one percent of patients reported RBD and 43%had an RBD episode at video-polysomnography regardless of the frequency of cataplectic attacks or gender. Lastly, the frequencyof periodic leg movements (PLM) per hour was higher in men than women and increased with age.Conclusions: Patients with more than one cataplectic attack per month had more frequent involvement of head, jaw and shouldermuscles and were mainly men. The proportions of patients with clinically assessed RBD and an RBD episode documented by video-polysomnography, as well as conspicuous values of PLM per hour, are fairly consistent with those reported in recent small-groupstudies. Therefore, it seems legitimate to argue that RBD and PLM are nocturnal manifestations intrinsic to narcolepsy/cataplexyand that the gender-related differences in the frequency of attacks and the value of PLM per hour may be indicative of a larger dif-ference in the clinical and polysomnographic characteristics of narcolepsy/cataplexy than hitherto suspected.� 2007 Elsevier B.V. All rights reserved.

Keywords: Narcolepsy; Cataplexy; Emotions; Trigger events; Motor disorders during sleep; NREM; REM sleep

1389-9457/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.sleep.2007.05.006

* Corresponding author. Tel.: +39 51 2092926; fax: +39 51 2092958.E-mail address: giuseppe.plazzi@unibo.it (G. Plazzi).

1. Introduction

Cataplexy is the main motor symptom of narcolepsy.When associated with excessive daytime sleepiness, cat-

426 K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433

aplexy is considered pathognomonic for narcolepsy/cat-aplexy according to International Classification of SleepDisorders, 2nd edition (ICSD-2) criteria [1]. Clinically,cataplexy ranges from mild manifestations, such as sag-ging jaw and leg or arm weakness, to more severe man-ifestations, such as buckling of the knees, falling down,head dropping, sudden facial hypotonia and difficultiesin speech articulation [2]. Cataplectic attacks last froma few seconds up to 10 min (usually less than one min-ute), have a highly variable frequency and are mainlytriggered by emotionally charged events. The emotionalvalence (or tone) of trigger events is more often ‘‘posi-tive’’ (e.g., joking, laughter and elation) than ‘‘negative’’(e.g., anger) or ‘‘undefined’’ (i.e., without a clear-cutdefined valence, such as surprise) [3].

On one hand, the close association between narco-lepsy/cataplexy and hypocretin deficiency [4] suggeststhat the pathological features of narcolepsy may providefurther information on the factors making patients vul-nerable to cataplexy. On the other hand, a more refinedassessment of the disease phenotype, including the fre-quency and the characteristics of cataplectic attacksand the phenomena of altered motor control duringsleep, may facilitate a comprehensive account of theclinical evolution of narcolepsy/cataplexy.

Several aspects of cataplectic attacks have yet to beinvestigated, namely whether their frequency is relatedto their clinical–pathological and behavioural features(including the emotional tone of the events triggeringattacks) and to the indices of motor disorders duringsleep (i.e., rapid eye movement [REM] sleep behaviourdisorders [RBD] and periodic leg movements [PLM]).

This study investigated a consecutive series of 44 first-diagnosed patients with narcolepsy/cataplexy to estab-lish (a) whether the frequency of attacks is associatedwith the specific clinical features of cataplectic attacks,(b) the types of emotionally charged events experiencedby patients as triggers and (c) the indices of daytimesleepiness, nocturnal sleep organization and alteredmotor control during sleep. We examined the standardindicators of diurnal sleepiness and nocturnal sleeporganization and also calculated the occurrence ofRBD episodes, the proportion of REM sleep withoutatonia (or RWA) and the number of PLM per hour dur-ing REM and non-rapid eye movement (NREM) sleepon polysomnographic (PSG) recordings.

2. Subjects and methods

2.1. Participants

Forty-five first-diagnosed, untreated patients withnarcolepsy/cataplexy were included in the study. Theywere examined from January 2004 to December 2005at the Sleep Disorders Clinic of the University of Bolo-gna. The diagnosis of narcolepsy/cataplexy fulfilled the

ICSD-2 clinical and PSG criteria. Inclusion criteria forthe enrollment of patients were (a) excessive daytimesleepiness demonstrated by the Epworth SleepinessScale (ESS) [5] and by a mean sleep latency less thanor equal to eight minutes at the multiple sleep latencytest (MSLT) [6], (b) at least two sleep-onset REM(SOREM) episodes at MSLT, (c) a clinical history ofunequivocal cataplexy and (d) human leukocyte antigen(HLA) DQB1*0602 positivity, albeit not pathogno-monic for narcolepsy [7,8]. Patients with more than fiveobstructive apnoeas per hour of sleep and with otherneurological or psychiatric disorders, including moder-ate to severe depression (i.e., a score higher than 18 onthe Beck Depression Inventory [BDIl; 9]) were excluded.All patients gave informed consent to take part in thestudy, which was approved by the Local EthicsCommittee.

2.2. Clinical interview

Before entering the study for clinical data to be col-lected (including a full sleep–wake history, also using asemistructured interview to detect RBD; [10]) for narco-lepsy/cataplexy diagnosis, patients were interviewed bya neurologist who was an expert in sleep disorders.Patients were also interviewed by a psychologist toscreen for any psychiatric history or current psychiatricsymptoms, which would have been an exclusion crite-rion, and then underwent (a) assessment of cognitivefunctioning and affective-emotional status (usingWeschler Adult Intelligence Scale [WAIS; 11]) andBDI, respectively), (b) evaluation of the frequency andclinical features of cataplectic attacks using items 77–79, 81 and 85 of the Stanford Scale for Narcolepsy(SSN) [3] and (c) detection of which of 18 possible eventslisted in the SSN had actually been experienced as trig-gers for cataplectic attacks (items 54–71).

2.3. Frequency of cataplectic attacks

The frequency of cataplectic attacks experienced byeach patient was calculated on the basis of the estimatescollected independently by the neurologist and psychol-ogist in the interviews. One case of incongruous esti-mates was excluded from statistical analyses whichwere carried out on 44 patients.

All patients taken for the present study fit at leastthree of the four criteria indicated by Mignot et al. [7]for definite (i.e., clear-cut) cataplexy, namely, (a) a lossof muscle tone with visible effect or the involvement ofother muscle groups in addition to legs, (b) durationof attacks shorter than 10 min, (c) preservation of con-sciousness, whereas (d) a frequency of attacks higherthan one per month was ascertained in 30 of the 44patients. Patients were thus divided on the basis ofattack frequency into two groups of 30 (high-frequency

K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433 427

group [HFG]) and 14 patients (low-frequency group[LFG]), respectively.

2.4. Video-polysomnography and MSLT

After a night for adaptation to the laboratory setting,a complete night was subjected to constant and time-synchronized video-monitoring and PSG (video-PSG,from 10 p.m. to 7 a.m.). Electrodes were placed accord-ing to the international system for scoring sleep stagesusing the standard criteria [12]. The montage of elec-trodes was set up record the following signals: electroen-cephalogram (EEG; four channels, two centrals, onevertex, one occipital, according to the international10–20 system), electro-oculogram (EOG, two channels),electrocardiogram (ECG, one channel) and electromyo-gram (EMG; three channels: one for the submentalismuscle, one for the right and one for the left tibialisanterior muscles). The respiratory pattern was moni-tored using an oral–nasal airflow thermistor and a tho-racic gauge. Sleep stages were visually scored followingstandard criteria [12] on 30-s epochs.

Daytime sleepiness was evaluated the next day by theMSLT carried out every two hours from 9 a.m. to 5 p.m.

2.5. Data analysis

Alpha value was fixed at 0.05 for all the statisticalanalyses reported below carried out using the SPSS11.0 package [13].

(1) Demographic and clinical indicators. One-wayanalyses of variance (ANOVAs) were used to comparethe two groups of patients for age, years of education,disease duration (estimated on the time elapsed betweenthe onset of daytime sleepiness or cataplexy and thediagnosis) and duration of cataplexy (estimated on thetime elapsed between the first recalled cataplectic attackand the time of diagnosis).

The v2 test was used to compare categorical variables,such as gender and clinical features of cataplecticattacks of the two groups (see Table 1).

(2) Trigger events and emotional tone. According toMignot et al.’s [7] distinction, the 18 possible triggerevents were classified as ‘‘Positive’’, ‘‘Negative’’ and‘‘Undefined’’ (see Table 2). The individual proportionsof events for these three categories were calculated foreach patient as ratios between the sum of positive/nega-tive/undefined events actually experienced and all possi-ble events of each category. A two-way multivariateanalysis of variance (MANOVA) was carried out onthese proportions, taking ‘‘group’’ (LFG/HFG) as thetwo-level between-subjects factor and ‘‘emotional tone’’as the three-level within-subjects factor. A multivariateanalysis was preferred to a univariate one, in order toavoid the infraction to homogeneity assumption forthe three-level factor (i.e., ‘‘emotional tone’’) [14].

The correlation between the rank order of the pro-portions of HFG and LFG patients who experiencedspecific types of events as triggers was calculated usingSpearman’s correlation coefficient. A logistic regressionanalysis (using Wald’s forward method) was carriedout to establish the predictive power of possible factorsfor the distinction between HFG and LFG.

(3) Indicators of daytime sleepiness, sleep organization

and altered motor control during sleep. (a) One-wayMANOVAs were used to compare the scores of daytimesleepiness (ESS and MSLT scores, i.e., sleep latency andnumber of SOREM) and the values of nocturnal sleepparameters (scored using the standard criteria [12]), tak-ing ‘‘group’’ (HFG/ LFG) as between-subjects factor(see Table 3). The sleep parameters considered for thestudy were total sleep time, sleep efficiency, sleeplatency, REM latency, proportions of REM sleep andNREM sleep stages.

(b) The presence of RBD was established on the basisof the two main diagnostic criteria listed in the ICSD-2[1]: (a) sleep-related injurious or disruptive behaviour byhistory or documented during video-PSG and (b) pres-ence of REM sleep without atonia (or RWA sleep) onPSG. Thus, RBD was assessed on the basis of a clear-cut clinical history, while the occurrence of one or moreepisodes of RBD during PSG was established by tworaters, blind to the design of the study and expert inexamining video-PSG files for detection of sleep disor-ders. They were instructed to score as RBD episodesthose periods of RWA sleep where there were also oneor more indices of motor behaviour involving legs orarms or vocal systems [1], for at least 15 s. By using thistemporal constraint we attempted to weaken the possi-bility of erroneous detection of an RBD episode as aconsequence of the lack of dream report, the collectionof which was excluded to avoid further sleepfragmentation.

Each 30-s epoch of REM sleep was classified as withor without atonia (hereinafter indicated as REM sleepand RWA sleep, respectively), according to the scoringcriteria originally suggested by Lapierre and Montplaisir[15] and then adapted by Consens et al. [16]. Therefore,we scored each epoch as RWA if tonic chin EMG activ-ity was present for more than 50% of 30 s or phasic chinEMG activity was present in at least 6 of 10 three-sec-ond miniepochs. Finally, RWA was expressed as thepercentage of total REM sleep.

The occurrences of clinical RBD and of RBD epi-sodes on video-PSG files were considered categoricalvariables and thus analyzed by the v2 test comparingthe proportions of patients with/without RBD in theLFG and HFG groups. In addition, a one-way MANO-VA was carried out on age, disease duration and cata-plexy duration of patients with/without video-PSGRBD. Finally, 2 two-way analyses of covariance(ANCOVAs) were carried out on the proportion of

Table 1Demographic and clinical data (data are given as means ± SD)*

Low-frequency group n = 14 High-frequency group n = 30 p Value

Demographic data

Gender (women), % (No.) 57.1 (8/14) 26.7 (8/30) 0.051Age, years 36.4 ± 14.4 40.7 ± 15.0 NSEducation level, years 11.6 ± 4.2 10.9 ± 4.2 NS

Clinical data

Cataplexy duration (years from first cataplectic attack) 17.0 ± 13.0 15.4 ± 13.7 NSDisease duration (years from first narcolepsy symptom) 21.7 ± 16.5 18.9 ± 15.4 NS

Muscle group involved in attacks

Legs and knees, % 100 100 NSArms and hands, % 71.4 78.6 NSHead, jaw, face and shoulders, % 69.2 96.7 < 0.05

Behavioural features of attacks

Normal state of consciousness, % 100 100 NSSlurred speech, % 90.0 96.0 NS

* Statistical tests applied: v2 test; ANOVA.

428 K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433

epochs of RWA sleep, taking ‘‘age’’ as covariate in bothanalyses, ‘‘group’’ (HFG/LFG) and ‘‘gender’’ asbetween-subjects factors for the first analysis and the

Table 2Proportions of patients who experienced specific events as triggers

Low-frequencygroup n = 14

High-frequencygroup n = 30

Positive tone

Laughter 0.786 0.900Excitement 0.385 0.483Remembering a happy moment 0.000 0.138Own verbal response in a

funny context0.308 0.800

Sexual intercourse 0.154 0.103Elation 0.538 0.690Joke 0.462 0.821

Negative tone

Anger 0.615 0.667Embarrassment 0.770 0.310Stress 0.154 0.586Tension 0.154 0.393

Undefined tone

Surprise 0.385 0.567Being startled 0.231 0.414During sport activities 0.770 0.379After sport activities 0.770 0.100Remembering an

emotional event0.770 0.138

Romantic thought or moment 0.770 0.690Being emotionally moved 0.385 0.400

Events are classified with respect to emotional tone using Mignotet al.’s typology [7]*.

* Statistical analyses (MANOVA and post hoc pairwise compari-sons) were carried out on the individual proportions of positive/neg-ative/undefined events experienced as triggers of cataplectic attacks.Individual proportions were calculated on the sum of events experi-enced out of all possible events for each cluster.

‘‘presence/absence of video-PSG RBD’’ and ‘‘gender’’as between-subjects factors for the second one.

(c) PLM were visually detected following the Ameri-can Sleep Disorders Association (ASDA) scoring crite-ria [17]. Leg movements were included when the EMGincreased to 8 lV or higher above the resting baseline;the ending point was when the EMG decreased to lessthan 2 lV above the resting level and remained belowthat value for 0.5 s. The duration of a leg movementwas at least 0.5 s and no longer than 10 s. PLM indexin NREM and REM sleep was calculated as the numberof leg movements included in a series of four or morePLMs, separated by more than five seconds and lessthan 90 s per hour of NREM and REM sleep,respectively.

The proportions of patients with a PLM index higherthan 15 per hour in NREM and REM sleep (followingthe ICSD-2 criteria [1]) were compared in women andmen. Moreover, a three-way ANCOVA was carried outon the PLM index, taking ‘‘age’’ as covariate and ‘‘group’’and ‘‘gender’’ as between-subjects factors and ‘‘type ofsleep’’ (NREM/REM) as within-subjects factor.

3. Results

(1) Demographic and clinical data are listed inTable 1.

The two groups of patients did not significantly differwith regard to age, education level, duration of diseaseor duration of cataplexy, whereas the higher numberof women in the LFG and men in the HFG groupsalmost reached significance (v2 = 3.831, df = 1,p = 0.051). In addition, the involvement of head, jawand shoulder muscles proved to be more frequent in cat-aplectic attacks in HFG than LFG patients (v2 = 6.644,df = 1, p < 0.05).

Table 3Indicators of daytime sleepiness and of organization of nocturnal sleep (data are given as means ± SD)

Low-frequency group n = 14 High-frequency group n = 30 p Value

Daytime sleepiness

Sleep latency at MSLT, min 2.8 ± 2.0 3.7 ± 2.5 NSSOREMs at MSLT, Nr 3.8 ± 1.6 3.9 ± 1.2 NSEpworth Sleepiness Scale, total score 16.0 ± 4.3 15.1 ± 4.8 NS

Nocturnal sleep parameters

Sleep latency, min 7.7 ± 8.8 7.2 ± 7.4 NSREM latency, min 15.2 ± 22.9 17.4 ± 31.3 NSTotal sleep time, min 439.5 ± 95.7 435.1 ± 70.2 NSSleep efficiency, % 84.5 ± 8.0 83.8 ± 8.4 NSStage 1, % 20.1 ± 16.3 15.6 ± 7.0 NSStage 2, % 36.1 ± 15.2 39.7 ± 6.7 NSStage 3 + Stage 4, % 17.6 ± 8.3 17.4 ± 8.1 NSREM, % 25.6 ± 10.3 27.6 ± 8.7 NS

MSLT, multiple sleep latency test; SOREM , sleep-onset REM; ANOVA (F).

K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433 429

(2) The proportions of patients who experienced spe-cific events as triggers of cataplectic attacks are reportedin Table 2.

The two-way MANOVA on the individual ratios ofpositive/negative/undefined events experienced as trig-gers showed a significant effect for ‘‘group’’(F1,41 = 6.313; p < 0.05), the proportions of types ofevents experienced as actual triggers being higher inHFG (mean 0.43 ± 0.17) than LFG (mean0.26 ± 0.22), and ‘‘emotional tone’’ (F2,40 = 11.064;p < 0.001). Pairwise comparisons (using Bonferroni’scorrection, with significance at 0.05 level) showed thatthe proportion of positive events (0.49 ± 0.22) was sig-nificantly higher than that of undefined events(0.26 ± 0.23) and the proportion of negative eventswas almost significantly higher than that of undefinedevents (0.41 ± 0.37). The interaction between ‘‘group’’and ‘‘emotional tone’’ was not statistically significant(F2,40 = 0.727; n.s.).

The rank order of the proportions of patients whoexperienced specific types of events as triggers was verysimilar in the two groups (Spearman’s r = 0.85,p < 0.01).

A logistic regression conducted taking all variablessignificantly different in the two groups as predictors(gender, involvement of head, jaw and shoulder musclegroup, proportions of positive, negative and undefinedtriggers) showed a significant contribution of two vari-ables to distinguish between HFG and LFG. These vari-ables were the involvement of head, jaw and shouldermuscles in attacks (odds ratio [O.R.] = 14.02; b = 2.64,standard error [SE] = 1.27; Wald = 4.35, p < .05) andgender (O.R. = 0.186; b = �1.68, SE = 0.81;Wald = 4.31, p < .05) with a Nagelkerke r2 equal to0.34 (Omnibus tests of model coefficients v2 = 11.25;df = 2; p < .05; �2log likelihood = 39.01). Noneof the other variables significantly contributed toprediction.

(3a) One-way MANOVAs did not show significantdifferences between the two groups (see Table 3) in thevalues of diurnal sleepiness (ESS and MSLT scores)(F3,40 = 1.014, n.s.) and parameters of organization ofnocturnal sleep (F9,34 = 0.720, n.s.).

(3b) The proportions of patients whose history wassuggestive of RBD did not significantly differ in LFGand HFG, being 57.1% and 63.3%, respectively(v2 = 0.748, df = 1, n.s.). Nor did the proportion ofpatients with RBD episodes at video-PSG differ inLFG and HFG, being 35.7% and 46.7%, respectively(v2 = 0.534, df = 1, n.s.). While the proportion ofwomen did not differ from that of men (56.25% vs64.28%, respectively) for clinical RBD, it was less thanhalf (25% vs 53.57%) that of men for video-PSG RBD(v2 = 3.366, df = 1, p = 0.07).

The numbers of patients with clinical RBD and withvideo-PSG documented RBD are reported in the upperpart of Table 4. Nineteen patients (4 women and 15men) with RBD at video-PSG had all reported RBDat clinical interview; 3 patients (2 men) had two RBDepisodes during the night. The occurrence during thefirst two periods of REM sleep did not significantly dif-fer with respect to that in the following ones (7 vs 15,respectively: v2 = 2.909, df = 1, n.s.). A one-wayMANOVA on age, disease duration and cataplexy dura-tion showed a trend toward a significant differencebetween patients with and without video-PSG RBD(Wilks’ k = 0.833, F3,40 = 2.465, p = 0.072). Subse-quently, a separate univariate ANOVA showed that dis-ease duration was longer in patients with video-PSGRBD (25.3 ± 19.2 years) than in those without(15.3 ± 10.6 years) (F1,42 = 4.515, p < 0.05).

A two-way ANCOVA on the proportion of RWAsleep out of total REM sleep time of all patients, consid-ering ‘‘age’’ as covariate, did not show any significant dif-ference with respect to ‘‘group’’ (F1,39 = 0.288, n.s.),‘‘gender’’ (F1,39 = 4.261, n.s.) or the ‘‘group’’ · ‘‘gender’’

Tab

le4

Ind

ices

of

alte

red

mo

tor

con

tro

ld

uri

ng

slee

p(d

ata

are

give

nas

mea

ns

±S

D)*

Lo

w-f

req

uen

cygr

ou

pn

=14

Hig

h-f

req

uen

cygr

ou

pn

=30

To

tal

sam

ple

n=

44

W,

n=

8M

,n

=6

W,

n=

8M

,n

=22

Pts

wit

hR

BD

Cli

nic

alR

BD

,%

(No

.o

fp

ts)

62.5

(5/8

)50

.0(3

/6)

50.0

(4/8

)68

.2(1

5/22

)61

.4(2

7/44

)vi

deo

-PS

GR

BD

,%

(No

.o

fp

ts)

37.5

(3/8

)33

.3(2

/6)

12.5

(1/8

)59

.1(1

3/22

)43

.2(1

9/44

)

%R

WA

slee

pep

och

so

ut

of

RE

Mep

och

sA

llp

atie

nts

,%

15.0

±13

.09.

3±

3.8

11.7

±5.

516

.4±

8.0

14.3

±8.

3P

tsw

ith

vid

eo-P

SG

RB

D,

%22

.1±

17.1

9.0

±1.

113

.2±

0.0

17.7

±8.

8417

.3±

9.9

Pts

wit

hou

tvi

deo

-PS

GR

BD

,%

10.8

±6.

99.

6±

4.3

11.5

±6.

014

.2±

6.6

12.0

±6.

1

PL

Min

dex

PL

M-N

RE

M>

15p

erh

ou

r,N

o.

of

pts

34

618

31P

LM

s-N

RE

Mal

lp

ts,

No

./h

13.6

±18

.840

.2±

24.5

23.5

±15

.828

.7±

18.7

26.6

±19

.7P

LM

-RE

M>

15p

erh

ou

r,N

o.

of

pts

34

416

27P

LM

s-R

EM

all

pts

,N

o./

h16

.6±

22.5

24.6

±16

.017

.7±

15.9

32.3

±22

.626

.0±

21.3

*T

he

stat

isti

cal

test

sap

pli

ed:v2

for

cate

gori

cal

vari

able

s,A

NO

VA

and

MA

NO

VA

for

par

amet

ric

mea

sure

s.

430 K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433

interaction (F1,39 = 3.144, n.s.). The covariate ‘‘age’’ wasalso not significant (F1,39 = 0.282, n.s.). A supplementarytwo-way ANCOVA was then carried out on the samemeasure, having divided patients on the basis of the pres-ence/absence of video-PSG documented RBD. This anal-ysis showed a significantly higher value for patients withthan in those without RBD episodes at video-PSG(17.3 ± 9.9 vs 12.0 ± 6.1, respectively: F1,39 = 5.643,p < .05), whereas it did not significantly differ with respectto ‘‘gender’’ (F1,39 = 0.021, n.s) or the ‘‘gender’’ · ‘‘pres-ence–absence of video-PSG RBD’’ interaction(F1,39 = 0.594, n.s.). The covariate ‘‘age’’ was also not sig-nificant (F1,39 = 0.236, n.s.) (see mid-part of Table 4).

(c) It was preliminarily ascertained that in the wholesample the proportion of women with a PLM indexhigher than 15 was almost significantly lower than thatof men in REM sleep (v2 = 3.290, df = 1, p = 0.068),whereas it was farther from significance in NREM sleep(v2 = 2.437, df = 1, p = 0.112).

Then a three-way ANCOVA carried out on the PLMindex of all patients, taking ‘‘age’’ as covariate, showeda significant difference with respect to ‘‘gender’’ (the aver-age value being lower in women than men: F1,39 = 4.911,p < 0.05), but not with respect to ‘‘group’’ (F1,39 = 0.002,n.s.) or ‘‘type of sleep’’ (F1,39 = 0.130, n.s); all interactionswere also considered not to be significant. ‘‘Age’’ was sta-tistically significant (F1,39 = 6.387, p < 0.02): a subse-quent analysis of correlation showed that age waspositively related to PLM index in NREM sleep (Pear-son’s p = 0.431, p < 0.01) but not in REM sleep(q = 0.217, n.s.) (see lower part of Table 4).

4. Discussion

The present study addressed the clinical, behaviouraland PSG correlates of the frequency of cataplecticattacks in patients with narcolepsy/cataplexy. The datacollected on 44 consecutive patients showed that the fre-quency (higher/lower than one attack per month: HFGand LFG) was closely associated with some clinical andbehavioural features (including the emotional tone oftrigger events) of attacks but not with PSG indices ofdaytime sleepiness and nocturnal sleep organization.In addition, several relationships were found betweengender and PSG indices of motor disorders during sleep.

In interpreting these findings we did not raise thequestion of whether the relations observed were indica-tive of clinical subtypes of narcolepsy/cataplexy or ofdistinct stages in the evolution of the disease, given thesmall size of the sample of patients examined. With thiscaveat in mind, our findings yield the followinginferences:

(a) The high-frequency of cataplectic attacks wasassociated almost significantly (p < 0.051) with genderand significantly with the involvement of head, jawand shoulder muscle districts. The difference in the pro-

K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433 431

portions of women (higher in LFG) and men (higher inHFG) in the two groups is a new finding. Indeed, a sig-nificant gender-related difference with respect to diseaseseverity, with which the frequency of cataplectic attackshas been associated [18], has never been reported,whereas a lower prevalence of narcolepsy in womenthan men (ranging from 1: 1.6–1.8 [19,20]) has beenestablished. If confirmed in larger samples, this gender-related difference in some aspects of cataplexy may shedlight on the evolution of cataplexy severity and narco-lepsy/cataplexy disease in general.

The more frequent involvement of head, jaw andshoulder muscles in the attacks experienced by HFGthan LFG patients is consistent with the indications pro-vided by investigations on the H-reflex in patients withnarcolepsy–cataplexy. These studies showed a linearcorrelation between the attenuation of the H-reflex dur-ing laughter and cataplexy severity [21]. The presentfinding could also be of clinical interest if confirmed inlarger samples of differentiated clinical phenotypes ofnarcolepsy/cataplexy.

(b) The proportion of events capable of triggeringcataplectic attacks was significantly higher in HFGpatients. Moreover, the proportion of undefined triggerevents was significantly lower than the proportion ofevents with a positive emotional tone (or valence) andalmost significantly lower than the proportion of eventswith a negative tone. These findings suggest a relation-ship between the number of trigger events and the expe-rienced frequency of cataplectic attacks and supportprevious observations that cataplexy is triggered overallby events with positive valence [3].

It seems legitimate to argue that the variations inthe frequency of attacks are related to the numberof trigger events experienced rather than to the occur-rence of specific events for two reasons. First, the pro-portions shown here were comparable for almost allevents to those reported in studies on larger samples[3,22]. Second, the proportions of patients experienc-ing specific trigger events were very similar in thetwo groups, as indicated by the extremely high corre-lation between the ranks of the events theyexperienced.

This hypothesis could be further corroborated by col-lecting patients’ evaluations of both the emotionalvalence and level of arousal of potential and actual trig-ger events, estimated according to bidimensional modelsof emotions [23]. This type of information would also beuseful to establish whether subjective descriptions of theemotional features of trigger events are accurate enoughto be taken into account in clinical settings, namely fordiagnostic purposes.

The possibility of pre-existing individual predisposi-tions capable of increasing the vulnerability to specifictrigger events remains open and should be tested in fur-ther studies using more detailed interviews and diary

formats for collection of information on the evolutionof narcolepsy and cataplectic attacks.

In general terms, the reliability of the emotional char-acteristics of trigger events is entwined with the accuracyof recalling the frequency and clinical characteristics ofcataplectic attacks. The problem cannot be resolved sim-ply because the clinical interview is made by the samewell-trained psychologist but requires concomitant vali-dation through pertinent indicators. Indeed, the possi-bility of a recall bias in estimating the frequency ofattacks cannot be ruled out, given that cataplexy maylast for up to 20–40 years before diagnosis [24,25], asalso shown in our sample (see Table 1). Hence, the esti-mates of the frequency and characteristic features maybe imprecise at the time of the diagnosis, based onlyon retrospective evaluation by the patient. This suggestsfuture studies should measure the influence of these pos-sible confounding factors by collecting not only retro-spective but also prospective estimates of both thefrequency and clinical characteristics of cataplecticattacks and the emotional features of trigger events.Measurement could be made by means of a cataplexydiary (including scales for assessing emotional toneand level of arousal of the events) to be compiled forfairly long periods after the diagnosis of narcolepsy/cat-aplexy. A cataplexy diary could be also useful to moni-tor disease evolution in narcolepsy/cataplexy patientswith or without medication [26].

(c) No indicators of daytime sleepiness, nocturnalsleep organization or disorders of motor control duringsleep were significantly related to the frequency of cata-plectic attacks, whereas some indices of motor dyscon-trol during sleep were significantly related to gender.The indications from these findings are important giventhe increasing interest in the role of motor disorders dur-ing sleep in the evolution of narcolepsy/cataplexy [27–29], and the reliability of the values of the indicatorsof disorders of motor control disorders which were com-parable in our study. In principle, the indications for therelationship between gender and indices of motor dys-control during sleep can be extended to the populationof patients with narcolepsy/cataplexy.

The proportion of patients with RBD ascertained atclinical interview was very similar in LFG and HFG(respectively, 57.1% and 63.3%) and comparable tothose reported in recent studies on small samples of nar-colepsy/cataplexy patients [27,29].

Moreover, RBD episodes were also documented onvideo-PSG in a substantial proportion of patients,which was again fairly similar in LFG and HFG (respec-tively, 35.7% and 46.7%) and comparable to thatobserved by Nightingale et al. on video-PSG of 13patients (38.4%). Likewise, the proportion of RWAsleep was significantly higher in patients with an RBDepisode documented by video-PSG (17.03%) than inthe others (12.01%).

432 K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433

It is noteworthy that while the proportion of womendid not differ from that of men (respectively, 56.25% vs64.28%) for clinical RBD, it was less than half (25% vs53.57%) that of men for video-PSG RBD. This picturesuggests that the screening of RBD based on video-PSG is less sensitive for women than men. Conse-quently, since the proportion of women in PSG investi-gations on RBD linked to other aetiology has beenestimated to be much lower than that of men (for exam-ple, 13% in Olson et al.’s study [30]), the identification ofa clinical RBD in a woman should be followed by a dif-ferential diagnosis for narcolepsy [27].

Besides corroborating the indications from recentstudies, the above three findings also enhance the reli-ability of clinical RBD evaluations obtained by meansof a semistructured interview [10]. Indeed, RBD epi-sodes were documented on a substantial proportion ofvideo-PSG of both LFG and HFG patients. It is appar-ent that the difference between the estimates drawn fromvideo-PSG (43.2%) and clinical history (61.3%) simplydepends on the fact that RBD is not an every-nightoccurrence in patients with narcolepsy/cataplexy, unlikeother types of patients (e.g., those with multiple systematrophy who nearly always displayed RBD at video-PSG [90%]) [31].

The proportion of patients with a PLM index higherthan 15 (as indicative of a PLM disease, according toICSD-2 [1]) was comparable to those reported inpatients with narcolepsy/cataplexy [28,32,33]. In addi-tion, a PLM index higher than 15 per hour was foundin an almost significantly larger proportion of men dur-ing REM sleep. Moreover, the mean value of PLMindex per hour, comparable to that reported in a recentstudy where it proved to be higher than in normal sub-jects [28], was significantly lower in women regardless ofthe frequency of cataplectic attacks (LFG vs HFG) andthe type (REM/NREM) of sleep. In addition, an age-related increase in the value of PLM index only duringNREM sleep was also in keeping with literature data[20,28].

In general terms, the consistency of clinical indicatorswith PSG indices confirms that RBD and PLM are noc-turnal manifestations intrinsic to narcolepsy/cataplexy.This finding seems to deserve further investigation toshed more light on the clinical evolution of the disease(as suggested by the higher proportion of epochs ofRWA sleep in patients with a video-PSG documentedRBD), pharmacological treatment (e.g., tricyclic antide-pressants, venlafaxine, SSRIs sodium oxybate, clonaze-pam or melatonin) and the behavioural consequences(namely, the risk of sleep-related injury and violence).

As the indicators of daytime sleepiness and sleepstructure did not prove to be distinctive for patients withdifferent frequency of cataplectic attacks or for womenwith respect to men, EMG parameters during sleepshould be investigated more extensively. In particular,

it should be ascertained whether the frequency of cata-plectic attacks (taken as a possible index of cataplexyseverity) is related to the level of altered motor controlduring sleep rather than the degree of daytime sleepi-ness, in that the indices of motor control are modulatedby age (as in the general population: [28]) and gender (asshown here).

Acknowledgements

This study was supported in part by grants awardedby MIUR (PRIN Grant No. 2005065029) to G. Plazziand by Fondazione Cassa Risparmio of Bologna toC. Cipolli. The authors are indebted to an anonymousreviewer for numerous thorough and helpfulsuggestions.

References

[1] American Academy of Sleep Medicine. The international classi-fication of sleep disorders diagnostic and coding manual. 2nd ed.Westchester, IL: American Academy of Sleep Medicine; 2005.

[2] Guilleminault C, Gelb M. Clinical aspects and features ofcataplexy. Adv Neurol 1995;67:65–77.

[3] Anic-Labat S, Guilleminault C, Kraemer C, Meehan J, ArrigoniJ, Mignot E. Validation of a cataplexy questionnaire in 983 sleep-disorders patients. Sleep 1999;22:77–87.

[4] Nishino S, Ripley B, Overeem S, Lammers GJ, Mignot E.Hypocretin (orexin) deficiency in human narcolepsy. Lancet2000;355:39–40.

[5] Johns MW. A new method for measuring daytime sleepiness: theEpworth Sleepiness Scale. Sleep 1991;14:540–5.

[6] Thorpy MJ, Westbrook P, Ferber R, et al. The clinical use of theMultiple Sleep Latency Test. Sleep 1992;15:268–76.

[7] Mignot E, Hayduk R, Black J, Grumet FC, Guilleminault C.HLA DQB1*0602 is associated with cataplexy in 509 narcolepticpatients. Sleep 1997;20:1012–20.

[8] Sturzenegger C, Bassetti L. The clinical spectrum of narcolepsywith cataplexy: a reappraisal. J Sleep Res 2004;13:395–406.

[9] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. Aninventory for measuring depression. Arch Gen Psych1961;4:561–71.

[10] Scaglione C, Vignatelli L, Plazzi G, Marchese R, Negretti A,Rizzo G, et al. REM sleep behaviour disorder in Parkinson’sdisease: a questionnaire-based study. Neurol Sci 2005;25:316–21.

[11] Wechsler D. Wechsler memory scale-revised manual. NewYork: The Psychological Corporation; 1987.

[12] Rechtschaffen A, Kales A. A manual of standardized terminologytechniques and scoring system for sleep stages of humansubjects. Los Angeles, CA: Brain Information Service; 1968.

[13] Norusis MJ. SPSS/PC +1.10: statistical data analysis. UpperSaddle River, NJ: Prentice Hall; 2002.

[14] Vasey MW, Fhayer JF. The continuing problem of false positivesin repeated measures ANOVA in psychophysiology: a multivar-iate solution. Psychophysiology 1987;24:479–86.

[15] Lapierre O, Montplasir MD. Polysomnographic features of REMsleep behavior disorder: development of a scoring method.Neurology 1992;42:1371–4.

[16] Consens FB, Chervin RD, Koeppe RA, Little R, Liu S, Junck L,et al. Validation of polysomnographic score for REM sleepbehaviour disorder. Sleep 2005;28:993–7.

K. Mattarozzi et al. / Sleep Medicine 9 (2008) 425–433 433

[17] Bonet Mand the ASDA Atlas Task Force. Recording and scoringleg movements. Sleep 1993;16:748–59.

[18] American Sleep Disorders Association [ASDA]. The internationalclassification of sleep disorders (ICSD): diagnostic and codingmanual. Rochester, MN: ASDA; 1997.

[19] Dauvilliers Y, Arnulf I, Mignot E. Narcolepsy with cataplexy.Lancet 2007;369:499–511.

[20] Longstreth Jr WT, Koepsell TD, Ton TG, Hendrickson AF, vanBelle G. The epidemiology of narcolepsy. Sleep 2007;30:13–26.

[21] Lammers GJ, Overeem S, Tijssen MA, van Dijk JG. Effects ofstartle and laughter in cataplectic subjects: a neurophysiologicalstudy between attacks. Clin Neurophysiol 2000;111:1276–81.

[22] Krahn EL, Lymp JF, Moore WR, Slocumb N, Silber MH.Characterizing the emotions that trigger cataplexy. J Neuropsy-chiatry Clin Neurosci 2005;17:45–50.

[23] Lang PJ, Greenwald MK, Bradley M, Hamm AO. Looking atpictures: affective, facial, visceral and behavioral reactions.Psychophysiology 1993;30:261–73.

[24] Bassetti CL, Aldrich M. Narcolepsy. Neurol Clin 1996;14:545–71.[25] Overeem S, Mignot M, van Dijk GJ, Lammers GL, Dijk JG.

Narcolepsy: clinical features, new pathophysiologic insights andfeature perspectives. J Clin Neurophysiol 2001;18:78–105.

[26] Vignatelli L, D’Alessandro R, Mosconi P, Ferini-Strambi L,Guidolin L, De Vincentiis A, et al. Health-related quality of life

in Italian patients with narcolepsy: the SF-36 health survey. SleepMed 2004;5:467–75.

[27] Nightngale S, Orgill JC, Ebrahim IO, de Lacy SF, Agrawal S,Williams AJ. The association between narcolepsy and REMbehaviour disorder (RBD). Sleep Med 2005;6:253–8.

[28] Ferri R, Zucconi M, Manconi M, Bruni O, Ferini-Strambi L,Vandi S, et al. Different periodicity and time structure of legmovements during sleep in narcolepsy/cataplexy and restless legssyndrome. Sleep 2006;29:1587–94.

[29] Marelli S, Fantini M, Busek P, Castronovo V, Manconi M, FeriniStrambi L. Association between REM sleep behavior disorderand narcolepsy with and without cataplexy. Sleep2006;29:A229–30. abstract supplement.

[30] Olson EJ, Boeve BF, Silber MH. Rapid eye movement sleepbehaviour disorder: demographic, clinical and laboratory findingsin 93 cases. Brain 2000;123:331–9.

[31] Plazzi G, Corsini R, Provini F, Pierangeli G, Martinelli P,Montagna P, et al. REM sleep behavior disorders in multiplesystem atrophy. Neurology 1997;48:1094–7.

[32] Baker TL, Guilleminault C, Nino-Murcia G, Dement W. Com-parative polysomnographic study of narcolepsy and idiopathiccentral nervous system hypersomnia. Sleep 1986;9:232–42.

[33] Harsh J, Peszka J, Hartwig G, Mitler M. Night-time sleep anddaytime sleepiness in narcolepsy. J Sleep Res 2000;9:309–16.