OUP user menu

A meta‐analysis for headache in systemic lupus erythematosus: the evidence and the myth

Dimos D. Mitsikostas, Petros P. Sfikakis, Peter J. Goadsby
DOI: http://dx.doi.org/10.1093/brain/awh146 1200-1209 First published online: 26 March 2004


Controversies in the occurrence and implications of headache in patients suffering from systemic lupus erythematosus (SLE) triggered us to conduct an extensive literature search in order to answer five clinical questions. (i) Is headache prevalence higher in SLE patients than in the general population? (ii) Is ‘lupus headache’ a separate entity? (iii) Is there a distinct pathogenetic mechanism of headache in SLE? (iv) Is headache related to CNS involvement or general SLE activity? (v) Is headache related to anxiety‐ and depression‐like symptoms in SLE? All published articles reporting data from >30 SLE patients were classified into four classes (I, IIa, IIb and III) by the quality of their evidence. We found no prospective controlled study (class I), but we identified seven controlled (class II) and 28 uncontrolled studies (class III) that retrospectively investigated the occurrence of headache in SLE patients. Eight out of 35 studies applied the International Headache Society (IHS) criteria for headache classification, whereas only four uncontrolled studies investigated paediatric SLE populations (class III). Pooled data from eight studies (controlled and uncontrolled) that used the IHS criteria show that 57.1% of SLE patients reported any type of headache (migraine 31.7% and tension‐type headache 23.5%). Pooled data from seven controlled studies showed that the prevalence of all headache types, including migraine, was not different from controls. Insufficient evidence was found for the concept of ‘lupus headache’. No particular pathogenetic mechanism of headache in adult SLE patients has been identified, nor an association between headache and the disease status, including CNS involvement. There is no good evidence that headache is associated with anxiety and depression in SLE. Insufficient data (class III) do not allow safe conclusions for headache among paediatric SLE patients. These findings suggest that the occurrence of headache in adult SLE patients does not itself require further investigation and that headache in those patients should be classified according to IHS criteria and managed as primary headache if there is no specific indication of a role for SLE in the patient. These recommendations should be verified by a properly controlled and prospective study in both adult and paediatric populations.

  • headache; migraine; systemic lupus erythematosus
  • aCL = anticardiolipin antibody; IHS = International Headache Society; SLE = systemic lupus erythematosus; TTH = tension‐type headache


Clinical observations since the 1970s have suggested an association between primary headaches and systemic lupus erythematosus (SLE). In the first study that appeared in the literature, brain involvement in SLE was not associated with headache (Atkinson and Appenzeller, 1975). Three years later, another report suggested that migraine may be associated with vascular dysfunction in SLE (Brandt and Lessell, 1978). Since then, all subsequent studies have agreed that headaches are common in SLE, occurring in a large proportion of patients, ranging from 30 to 65% (Brandt and Lessell, 1978; Isenberg et al., 1982; Levine et al., 1987; Anzola et al., 1988; Hogan et al., 1988; Vazquez‐Cruz et al., 1990; Markus and Hopkinson, 1992; Montalban et al., 1992; Hietaharju et al., 1993; Omdal et al., 1988, 2001; Sfikakis et al., 1998; Amit et al., 1999; Fernandez‐Nebro et al., 1999; Menon et al., 1999; Ainiala et al., 2001a; Glanz et al., 2001). Even if secondary headaches are excluded, e.g. headaches associated with renal dysfunction (Isenberg et al., 1982), non‐steroidal anti‐inflammatory drug use (Omdal et al., 1988), infection (Appenzeller, 1998) or cerebral venous thrombosis (Vidailhet et al., 1990; Laversuch et al., 1995; Haji Muhammad Ismail Hussain et al., 1999; Yuen et al., 2001), the incidence is still high (Yuen et al., 2001). Moreover, in some studies, the headache occurrence has been related to SLE exacerbation (Brandt and Lessell, 1978; Markus and Hopkinson, 1992; Hietaharju et al., 1993) and Raynaud’s phenomenon (Hogan et al., 1988). Others showed that there is a link between migraine and anticardiolipin antibodies (aCLs) (Cronin et al., 1988; Hogan et al., 1988), but not all investigators have agreed (Levine et al., 1987; Markus and Hopkinson, 1992; Montalban et al., 1992). Thus, although a high prevalence of various headache types has been suggested by several studies, the specific relationship of headache to SLE features still remains unclear. It is important to note that rheumatologists believe that headache is a part of the clinical picture in SLE, using terms, not included in the International Headache Society (IHS) classification of headaches, such as ‘intractable headache’, ‘vascular headache’ and ‘tension headache’. There are reports suggesting that MRI should be performed in all SLE patients that report any type of headache (Bovim et al., 1990; Rozell et al., 1998).

The American College of Rheumatology recently published the nomenclature and case definitions for neuropsychiatric SLE (American College of Rheumatology, 1999), where headaches, although non‐specific, were considered as a part of the clinical spectrum of SLE. Three previous studies have been published indicating that migraine is indeed associated with SLE (Ainiala et al., 2001a; Glanz et al., 2001; Brey et al., 2002), whereas in another no association was found (Omdal et al., 2001). These recently published studies with contradictory findings triggered us to review critically available published data to attempt to clarify whether primary headaches occur more frequently in SLE patients than in the general population and, if they do, which particular type of headache is the most frequent. We further aimed to search for scientific evidence indicating a particular pathogenetic mechanism for headaches seen in SLE patients. If so, then the term ‘lupus headache’ should clearly be established and used in clinical practice, otherwise it may have no use. We also searched for possible association of headache with either CNS involvement in SLE (neuro‐SLE) or with SLE disease activity. If there is such an association, then headaches could serve as either an early clinical sign of neuro‐SLE, as a clinical indication for laboratory CNS investigation, or both. Apart from these major objectives, we also addressed a possible relationship of headache in SLE to Raynaud’s phenomenon, or aCLs in the context of SLE. Thus, the aims of the review were to answer the following questions. (i) Is headache prevalence higher in SLE patients than in the general population? (ii) Is ‘lupus headache’ a separate entity? (iii) Is there evidence for a distinct pathogenetic mechanism of headache in SLE? (iv) Is headache related to CNS involvement or general SLE activity? (v) Are headaches related to anxiety‐ and depression‐like symptoms in SLE?


We conducted an extensive Medline/PubMed and EMBASE search, using as keywords ‘headache’ and ‘lupus’, or ‘migraine’ and ‘lupus’ (by December 2002). All reports published in any language were reviewed and classified according to criteria presented in Table 1. From all scanned studies, no controlled prospective study was found (class I), but seven retrospective controlled studies were identified (class II) (Isenberg et al., 1982; Markus and Hopkinson, 1992; Montalban et al., 1992; Goh et al., 1997; Sfikakis et al., 1998; Fernandez‐Nebro et al., 1999; Ainiala et al., 2001b). Several uncontrolled studies were filtered further by selecting those reporting specific findings on headache frequency and type, obtained from >30 SLE patients. We excluded studies with <30 patients since these were either case reports, or focused on a specific, clinical aspect of SLE other than headaches. Using these criteria, we had 28 uncontrolled studies (class III) (Atkinson and Appenzeller, 1975; Anzola et al., 1988; Cronin et al., 1988; Omdal et al., 1988; Alarcon‐Segovia et al., 1989; Vazquez‐Cruz et al., 1990; Vitali et al., 1992; Hietaharju et al., 1993; Wysenbeek et al., 1993; Parikh et al., 1995; Steinlin et al., 1995; Donders et al., 1998; Fanopoulos et al., 1998; Rozell et al., 1998; Amit et al., 1999; Banecka‐Majkutewicz, 1999; Camilleri and Mallia, 1999; Menon et al., 1999; Rood et al., 1999; Galeazzi et al., 2000; Quintero‐Del‐Rio and Van Miller, 2000; Ainiala et al., 2001a; Glanz et al., 2001; Mok et al., 2001; Omdal et al., 2001; Swaak et al., 2001; Brey et al., 2002; Sibbitt et al., 2002). From all selected studies, only eight used the IHS diagnostic criteria for headache classification (Fig. 1) (Goh et al., 1997; Rozell et al., 1998; Sfikakis et al., 1998; Fernandez‐Nebro et al., 1999; Ainiala et al., 2001b; Glanz et al., 2001; Omdal et al., 2001; Brey et al., 2002). When populations were homogeneous, we pooled data and χ2 tests were applied to detect significant differences between groups. These analyses were restricted to class IIa studies. In studies that did not use IHS criteria, it was not clear if all SLE patients with headache and CNS involvement should be considered to have primary or secondary headaches. We decided to consider those headaches under the term ‘all headache types’. Data from one Finish survey were published twice, once as a case‐controlled study (Ainiala et al., 2001b), and once as an uncontrolled study (Ainiala et al., 2001a). When we pooled data from studies that used the IHS criteria, we used the case‐controlled report (Ainiala et al., 2001b). Since children had severe CNS involvement and there were relatively fewer studies in paediatric populations, we excluded these studies from the calculations. A summary of the above studies is presented in Tables 234.

Fig. 1 The headache prevalence (%) in SLE patients as has been reported by IHS‐based studies. The averages of pooled data from all studies are presented as the vertical axis in each headache type. Filled triangles correspond to class IIa studies (controlled, IHS‐based) and open triangles to class III (uncontrolled studies). The asterisk indicates that headache prevalence was significantly different between SLE patients and controls. No further analysis for migraine and TTH was provided in this study.

View this table:
Table 1

Classification of evidence relating headache and SLE

IEvidence provided by more then one well‐designed, prospective*, IHS‐based, controlled clinical trial
IIaEvidence provided by more then one well‐designed, retrospective, IHS‐based, study with concurrent controls (e.g. case–control or cohort studies)
IIbEvidence provided by more then one well‐designed, retrospective, non‐IHS‐based, study with concurrent controls (e.g. case–control or cohort studies)
IIIEvidence provided by expert opinion, case series, case reports and studies with historical controls

*Prospective design requires headache evaluation based on headache diaries.

View this table:
Table 2

Class II studies that searched for assocations between headaches and SLE

Isenberg et al. (1982)Montalban et al. (1992)Markus and Hopkinson (1992)Goh et al. (1997)Sfikakis et al. (1998)Fernandez‐Nebro et al. (1999)Ainiala et al. (2001b)
Headache criteriaNon‐IHSNRNon‐IHSIHSIHS***IHSIHS
No. of SLE patients301039050887146
No. of controls30589050 and 208**897146
Controls wereObstetric out‐patientsNon‐SLE migraineursHospital staffNon‐SLE in‐patients and GPGPNon‐lupus out‐patientsGP
Participant’s approachInterviewInterviewInterviewInterviewInterview and examinationInterviewInterview and examination
Migraine in SLE17 (56.6%)32 (31%)31 (34%)8 (16%)7 (7.9%)16 (22.5%)18 (39.1%)
Migraine in controls9 (30%)58 (100)15 (17%)10 (20%) and 30 (14.4%)NR13 (18.3)9 (19.5%)
Migraine and SLEAssociatedNRAssociated No associationNo association No associationNR
TTH in SLENR21 (20.4%)24 (27%)*15 (30%)22 (25%)17 (23.9%)7 (15.2%)
TTH in controlsNRNR16 (18%)*16 (32%) and 74 (35.6%)NR17 (23.9%)2 (4.3%)
TTH and SLENRNRNo associationNo associationNo associationNo associationNR
All headache types in SLENR54 (52.4%)55 (61.1%)39 (78%)29 (32.9%)33 (46.5%)25 (54%)
All headache types in controlsNR58 (100%)31 (34.4%)41 (82%) and 165 (79.3%)26 (30%)31 (43.7%)11 (24%)
All headache types and SLENRNRNRNo associationNo associationNo associationAssociated
Other significant associationsNRNRAge of migraine onsetAnxietyAnxiety/depression?Age of migraine onsetNR
SLE features and migraineRaynaud: no assocationNRSLE activityNRNo associationNo associationNR
aCL and migraineNRNo associationNo associationNRNo associationNo associationNR

NR = not reported; GP = general population; *reported as non‐migrainous headaches; **two control groups were used; ***more than one attack per 2 weeks.

View this table:
Table 3

Class III studies that searched for associations between headaches and SLE

ReferenceNo. of SLE patientsHeadache criteriaAll headache types (%)Migraine (%)TTH (%)Neuro‐SLE (%)Neuro‐SLE and headachesaCL and headachesRaynaud and headachesSLE activity and headachesOther SLE feature related to headaches
[Ainiala et al. (2001a)]$46IHS25 (54)18 (39.1)7 (15.2)42 (91)NRNRNRNRSteroid use
Alarcon‐Segovia et al. (1989)500Non‐IHSNR66 (13.2)NR54 (10.8)&NRNo associationNRNRNR
Amit et al. (1999)148Non‐IHS52 (35.1)NRNRNRAssociatedNo associationNRAssociatedMyoskeletal features
Anzola et al. (1988)82Non‐IHS34 (41.4)25 (30.4)9 (10.9)NRNRNRNRNRNR
Atkinson and Appenzeller     (1975)61Non‐IHS27 (44.6)15 (21.3)9 (14.7)28 (46)No associationNRNRNRNR
Banecka‐Majkutewicz (1999)44Non‐IHS18 (40.9)3 (6.8)9 (20.4)28 (63.6)AssociatedNRNRNo associationNR
Brey et al. (2002)128IHS73 (57)51 (38.8)21 (29)(80)NRNRNRNRNR
Camilleri and Mallia (1999)58Non‐IHS17 (29.3)NRNRNRNRNRNRNRNR
Cronin et al. (1988)64Non‐IHSNR14 (22.9)NRNRNRAssociatedNRNRNR
Donders et al. (1998)175Non‐IHSNR18 (10.3)NRNRNRNo associationNRNRNR
Fanopoulos et al. (1998)48§NRNR3 (6.2)NR13 (27)NRNRNRNRNR
Galeazzi et al. (2000)448Non‐IHSNR75 (16.7)NR188 (41.9)NRNRNRNRaGM1‐IgG
Glanz et al. (20010186IHS115 (62)72 (39)43 (23.1)NRNRNo associationNo associationNRNR
Hietaharju et al. (1993)42NRNR10 (3.8)NR12 (28.5)NRNRNRNRNR
Menon et al. (1999)45NR6 (12.5)5 (10.4)NRNRNRNRNRNRNR
Mok et al. (2001)518Non‐IHS5/96 (4)*3/96 (3.1)*2/96 (1.9)*96 (19)NRNRNRNRNR
Omdal et al. (2001)58IHS38 (66)22 (38)21 (36.2)NRNRNo associationNRNo associationNR
Omdal et al. (1988)30Non‐IHS18 (70)12 (40)NR25 (83)NRNRNRNRNR
Parikh et al. (1995)108**Non‐IHS16 (14.8)NRNR25 (23.1)NRNRNRNRNR
Quintero del Rio (2000)86**Non‐IHS6/25 (24)*NRNR25 (29)NRNRNRNRNR
Rood et al. (1999)191Non‐IHS7 (14.3)*4 (8.1)*NR49 (25.6)NRNRNRNRNR
Rozell et al. (1998)40IHS29 (72.5)18 (45)11 (27.5)NRNo associationNRNRNo associationDuration of SLE
Sibbit et al. (2002)75**Non‐IHS(72)(36)NR(95)NRNRNRNo associationHospital admission
Steinlin et al. (1995)91**non‐IHS9 (22.5)*NRNR40 (43.9)NRNRNRNRNR
Swaak et al. (2001)187non‐IHSNRNR (15–20)NRNRNRNRNRNRNR
Vaquez‐Cruz et al. (1990)76Non‐IHS52 (68.4)24 (31.5)25 (32.9)15 (19.7)AssociatedNRNo associationNo associationThrombocytopenia
Vitali et al. (1992)704Non‐IHSNRNR (17.2)+NRNR (23)NRNRNRNRNR
Wysenbeek et al. (1992)83NR49 (59)&NRNRNRNRNRNRNRFatigue
All studies reported percentage of all headache types
1039528 (50.8)

aGM1‐IgG = anti‐ganglioside antibodies; $not included in the calculations because the paper by Ainiala et al. 2001b is included in class II studies; *percentage in neuro‐SLE patientes only, not in all SLE patients; **chlildren; non‐migraine; &calculated, not reported; +reported as headache/migraine; 48% among headache sufferers and 22.9% among SLE patients without headache; paediatric populations not included; §controlled study focused on anti‐β2‐glycosylphosphatidylinositol and aCLs; criteria for migraine were very close to those suggested by IHS. NR = not reported.

View this table:
Table 4

Migraine with and without aura in IHS‐based case control studies

Goh et al. (1997)Nebro et al. (1999)Ainiala et al. (2001b)All studies togetherχ2
Migraine (all cases)8/5010/5016/7113/7118/469/4642/16732/167NS
 Migraine without auraNRNR13/7112/716/464/4619/11718/117NS
 Migraine with auraNRNR3/711/7112/465/4615/1176/117P < 0.05
All headaches39/5041/5033/7131/7125/4611/4697/16791/167NS

NR = not reported; NS = not significant.


Is headache prevalence higher in SLE patients than in the general population?

We considered only class I or II studies could answer this question. Among class II studies, one did not address this issue, aiming to search for a possible relationship between migraine and aCLs (Montalban et al., 1992). The remaining six studies had somewhat contradictory results (Isenberg et al., 1982; Goh et al., 1997; Sfikakis et al., 1998; Fernandez‐Nebro et al., 1999; Ainiala et al., 2001b). With one exception, all class IIa studies agreed that the prevalence of all primary headache syndromes in SLE patients is not significantly different from that found in the control group, i.e. the general population (Goh et al., 1997; Sfikakis et al., 1998) or non‐lupus patients (Goh et al., 1997; Fernandez‐Nebro et al., 1999). However, there is one class IIa study that showed a significant association between all headache types and SLE (Ainiala et al., 2001b). This was a cross‐sectional, general population‐based survey covering an area with 440 000 inhabitants in Finland. The odds ratio of this association was estimated at 3.0 (95% confidence interval 1.3–7.1). Interestingly, despite this finding, the authors suggest that headaches do not contribute to SLE, because of the high prevalence seen in their control group (Ainiala et al., 2001b). Among class IIb studies, one (Markus and Hopkinson, 1992) reported that 55 out of 90 SLE patients and 31 out of 90 hospital staff they used as the control group had various headaches, indicating that the difference may be significant, although they did not report an analysis. A further study (Isenberg et al., 1982) only reported data for migraine. Thus, the overall impression from class II studies is that no evidence is provided for a higher prevalence of primary headache in SLE patients when compared with control groups. In class III studies, the headache prevalence in patients with SLE varies from 12.5% (Menon et al., 1999) to 72.5% (Rozell et al., 1998). When we pooled the data from all studies that reported prevalence for all headache types seen in SLE patients (no paediatric populations included), the prevalence was calculated as 50.2% (versus 57.1% in IHS‐based studies). Although uncontrolled studies provide no clear conclusion, these data do not look different from those in the general population (Rasmussen, 2001).

Tension‐type headache prevalence in SLE

Tension‐type headache (TTH) is the most frequent primary headache syndrome in the general population (Rasmussen, 2001). Three class IIa studies reported data on TTH (Goh et al., 1997; Fernandez‐Nebro et al., 1999; Ainiala et al., 2001b) and none reported significant association with SLE. Pooled data from these studies show that 39 out of 167 SLE patients, versus 35 out of 167 controls, reported TTH headaches (χ2, not significant). Thus, based on available class IIa studies, there is no evidence for higher prevalence of TTH in SLE. Among class III studies, the TTH prevalence varies from 1.9 (Mok et al., 2001) to 32.9% (Vazquez‐Cruz et al., 1990). Considering data from all IHS‐based studies, TTH prevalence is 23.5% (Fig. 1). There is some evidence that TTH may be less frequent in SLE patients than in the general population, or that TTH may occur less often than migraine in these patients, in contrast to the general population. The average prevalence of TTH is 23.5% versus 31.7% for migraine, in IHS‐based studies (Fig. 1). In only one class IIa, study the TTH prevalence was higher than that of migraine, but it was also higher than in the control group (Ainiala et al., 2001b). The sample size was limited in the latest study to 46 participants from the general population. Apart from this, and considering the overall prevalence of TTH, both class II and class III studies do not provide evidence for a particular change in the TTH prevalence among SLE patients.

Migraine prevalence in SLE

The prevalence of migraine in SLE patients seems more complicated. Two class II studies either did not report migraine data (Sfikakis et al., 1998) or the control group consisted of migraineurs but without SLE (Montalban et al., 1992), thus offering no insight into this issue. Among class IIa studies that showed data on migraine (Goh et al., 1997; Fernandez‐Nebro et al., 1999; Ainiala et al., 2001b), one reported a significant association between migraine and SLE (Ainiala et al., 2001b), but that was not confirmed by the other two studies (Goh et al., 1997; Fernandez‐Nebro et al., 1999). When we pooled data from these studies, migraine prevalence was 42 cases out of 167 SLE patients versus 36 cases out of 167 controls (χ2, not significant). In class IIb studies, a significant association between migraine and SLE was found (Isenberg et al., 1982; Markus and Hopkinson, 1992), the migraine prevalence in SLE patients being almost twice that of controls. Since a clear definition of migraine is crucial in such studies, we have assigned the results of class IIb studies much less weight. In class III studies, migraine prevalence varies from 3.1 to 45% (Rozell et al., 1998; Mok et al., 2001). Migraine with or without aura does not seem to differ between SLE patients and controls in these studies. One class IIa study showed that migraine with aura in SLE was higher (26%) than in the general population (10.8%) (Ainiala et al., 2001b). The authors did not provide an analysis of this finding. In addition, pooled data from IHS‐based studies revealed a significant increase of migraine with aura prevalence in SLE patients (15 cases versus six in controls out of 115 observations; Table 4). The small number of cases limits the value of this finding. In other class III studies, migraine with aura was very frequent, reaching 42% (Brey et al., 2002) and 44% (Glanz et al., 2001) of SLE patients. Although selection bias may have influenced the uncontrolled studies, these observations suggest important issues with possible implications.


The prevalence of all headache types, particularly of TTH and of migraine, does not differ between SLE patients and controls (class IIa evidence). However, migraine with aura may be more frequent among SLE patients. Further investigation on this matter is needed.

Is ‘lupus headache’ a separate entity?

In four class III reports, the term ‘lupus headache’ was used (Vazquez‐Cruz et al., 1990; Steinlin et al., 1995; Menon et al., 1999; Omdal et al., 2001), but without a clear definition. Among 48 adult SLE patients, one reported severe lupus headache in the first screening, whereas in the second assessment 12 or 18 months later, four patients reported such headaches (Menon et al., 1999). In another uncontrolled study that used the IHS criteria, the participants did not report ‘lupus headache’; one headache case was unclassifiable. The authors mentioned that ‘lupus headache’ contributes to a ‘severe, disabling, persistent and not responsive to narcotic analgesics headache’. No such phenotype was found among the 58 patients they investigated specifically for headaches (Omdal et al., 2001), nor did any class IIa study identify any headache case with these characteristics. In a study in children, nine ‘lupus headaches’ out of 91 patients (22%) were reported (Steinlin et al., 1995). The authors describe these headaches as intractable and severe, not responding to treatment, but without any other detailed description (Steinlin et al., 1995). All these children had CNS manifestations of SLE; thus these headaches may not have been primary.


Not enough evidence (only class III) is available in the literature regarding the term ‘lupus headache’, and therefore no safe conclusion could be made.

Is there a distinct pathogenetic mechanism of headache in SLE?

Since there are no data from class IIa studies indicating a different prevalence of headaches in SLE patients from that in the general population, this question is moot. The issue has been addressed in the past because it was considered that headache was more common in patients with SLE. In a retrospective class III study (Atkinson and Appenzeller, 1975), the files of 61 SLE patients were reviewed; 27 (45%) suffered from headaches classified by older criteria with 13 of vascular and nine of muscle contraction type. The primary aim of the investigators was to explore the hypothesis that small vessel disease may contribute to headache pathogenesis, so they looked for a possible association between CNS involvement and headache frequency. They did not find such a relationship and concluded that headache in SLE patients without neurological symptoms or signs does not indicate CNS involvement, and that small vessel disease within the brain without inflammation cannot count as a cause of headache (Atkinson and Appenzeller, 1975). In a study not included in the analysis because of the small sample size, 11 SLE patients with headache and visual hallucinations typical of those found in migraine were reported (Brandt and Lessell, 1978). The authors concluded that migraine‐like phenomena, such as visual hallucinations, might arise as a result of vascular dysfunction in SLE. These symptoms were typical of visual aura (fortification spectra) and responded to therapy with both corticosteroids and ergotamine or aspirin (Brandt and Lessell, 1978). In a subsequent brief report, it was found that 34 out of 83 SLE patients complained of various types of headache, 25 (30%) of whom were classified as migraineurs (non‐IHS criteria). They suggested that in a minority of cases (4% of headache sufferers), SLE triggered migraine that was indistinguishable from idiopathic migraine, while in 13 SLE patients (17%) headache was caused directly by SLE. The criterion used to obtain this different influence of SLE in migraine was the family history for headaches. If it was present, they speculated that SLE triggered migraine, if not, then SLE caused de novo migraine (Anzola et al., 1988). While such a genetic basis is attractive in theory, no clear evidence for this approach was provided. On the other hand, it has been emphasized that when migraine is assessed by proband report, it is not satisfactory for diagnosing migraine in relatives and that direct clinical interview of the affected relatives is necessary in family studies of migraine (Russell et al., 1996). Beyond these class III studies, no other reports have attempted to clarify this point.


There is no scientific evidence to support a particular mechanism for headache pathogenesis in SLE patients (class III evidence).

Is headache related to CNS involvement or general SLE activity?

To answer this question, we first looked for epidemiological data. Studies in adults reported headache prevalence that varied from 4 to 14.3% of neuro‐SLE cases. Only six studies, however, looked specifically for a possible association between headaches and neuro‐SLE features. In the only class IIa study, no significant association was identified between neuro‐SLE clinical features and any headache type (Sfikakis et al., 1998). Among the class III studies, three documented a significant relationship (Vazquez‐Cruz et al., 1990; Amit et al., 1999; Banecka‐Majkutewicz, 1999) and two did not (Atkinson and Appenzeller, 1975; Rozell et al., 1998). In another class IIb study (Weiner et al., 2000) not included in our analysis because of the limited number of patients, 28 SLE patients were evaluated with brain MRI and PET and compared with 10 healthy volunteers. Headache was considered as a sign of mild CNS involvement, together with depression, anxiety, cognitive dysfunction and anxiety disorders. All patients with either severe or mild neuro‐SLE showed hypometabolism in PET scans, as compared with two out of five patients without neuro‐SLE (P < 0.0025). Limitations in the sample size and the heterogeneity of the CNS symptoms in SLE patient groups do not allow safe conclusions regarding headache or migraine. Brain H‐MRS scans in patients with neuro‐SLE revealed abnormalities in N‐acetylaspartate, as compared with controls, but again the study was not focused on headache or migraine (Sibbitt. et al., 1997). Neuroimaging studies in migraineurs have also shown abnormalities, in migraine with aura in particular (Ferbert et al., 1991; Igarashi et al., 1991; Osborn et al., 1991; Fazekas et al., 1992; De Benedittis et al., 1995), as well as in SLE patients with CNS involvement (Bosma et al., 2000), although not all studies found abnormalities (Gonzalez‐Crespo et al., 1995). It is noteworthy that children with SLE seem different. When we pooled data from studies in paediatric populations, we found that headaches (not classified) were reported by 34% of patients with neuro‐SLE (Steinlin et al., 1995; Quintero‐Del‐Rio and Van Miller, 2000), whereas when we pooled data from adult SLE patients, the headache prevalence (all headaches) was calculated at 8.3% (Rood et al., 1999; Mok et al., 2001). This was a significant difference between children and adults.

Among the studies identified, only two showed that headache was related to general SLE activity. One was a class IIb study (Markus and Hopkinson, 1992) and the other class III (Amit et al., 1999). All other studies (Vazquez‐Cruz et al., 1990; Sfikakis et al., 1998; Banecka‐Majkutewicz, 1999; Fernandez‐Nebro et al., 1999), controlled and uncontrolled, agreed that there was no such association. Regarding other SLE features, most studies were focused on the presence of aCLs, and all agree that there is no such association (Markus and Hopkinson, 1992; Montalban et al., 1992; Tsakiris et al., 1993; Sfikakis et al., 1998; Fernandez‐Nebro et al., 1999), with one exception (Cronin et al., 1988). However, in another class IIb study, aCL was found to be elevated in the CSF of SLE patients with ‘lupus headache’. In the same study, aCL within the CSF was associated further with acute psychosis, cognitive impairment, high cortical dysfunction and altered consciousness. The authors suggested that intrathecal diffusion rather than the diffusion of immunoglobulin G aCL from serum to CSF occurred in those SLE patients with CNS involvement (Lai and Lan, 2000). These findings deserve further evaluation. Other manifestations were also investigated, such as Raynaud’s phenomenon and anti‐ganglioside antibodies. For Raynaud’s phenomenon, no association was proven (class IIb) (Vazquez‐Cruz et al., 1990; Glanz et al., 2001), whereas immunoglobulin G anti‐ganglioside antibody was related to migraine in a large multicentre cohort study (class IIb) (Galeazzi et al., 2000). Genetic analysis of a Dutch family with autosomal dominant vascular retinopathy, migraine and Raynaud’s phenomenon revealed significant evidence of linkage to chromosome 3p21 (Terwindt et al., 1998; Ophoff et al., 2001). The significance of these findings remains unclear. The onset of headaches may be related to the onset of SLE, although there are contradictory results on this issue: in one controlled study, there is indeed an association (Fernandez‐Nebro et al., 1999), but this was not confirmed by two other class IIa studies (Goh et al., 1997; Sfikakis et al., 1998). It is appropriate to mention that apart from SLE, other chronic disease could evoke headaches, but no such association has been shown for the case of SLE (Goh et al., 1997). All studies examining this question used a retrospective methodology, e.g. the patients ‘felt’ that they had more or fewer headaches, thus limiting their value. It seems, therefore, that there is no established association between the onset of SLE and the frequency, and even more the onset, of headache. In a recent paper not included in our analysis (Brunner et al., 2003), headaches reported by children with SLE were unlikely to be associated with disease activity.


In adults, there is no established association between neuro‐SLE clinical features and any headache type (class IIa evidence). In contrast, for children, there is class III evidence supporting a possible association between CNS involvement and headache. Further investigation is needed. Headache is not related to SLE activity and serum aCL, or to Raynaud’s phenomenon (class II evidence). There is class IIb evidence that headache is related to the presence of aCL in the CSF, or with anti‐ganglioside antibody in serum. These findings deserve further investigation.

Is headache related to anxiety‐ and depression‐like symptoms in SLE?

It is speculated that headaches in SLE may be associated with anxiety‐ and depression‐like symptoms (Goh et al., 1997; Sfikakis et al., 1998). In one study, 59% of SLE headache patients reported two or more symptoms of stress (unspecified) compared with 18.2% of the SLE‐non‐headache patients (χ2 = 5.71, P = 0.0017) (Goh et al., 1997). Applying the Hamilton rating scales for anxiety and depression in all SLE headache patients, it was reported that most of them complained of anxiety‐ and depression‐like symptoms (73% had a score >14 in the scale for anxiety and 60% for depression) (Sfikakis et al., 1998), but the authors did not apply these scales to a suitable control group. Looking specifically at the psychological profile of those SLE patients that report headache, it was noted that migraine was associated with a tendency to social isolation and anxiety, while TTH was associated with psychological distress, such as somatic complaints, reduced energy, mental tension, social discomfort and withdrawal, as well as depressive mood (Minnesota Multiphasic Personality Inventory and Beck Depression Inventory). Although the study was not controlled, the authors suggested that these are the characteristics usually seen in migraine and TTH outside SLE, and thus migraine and TTH may not be considered as part of the neuropsychiatric disease spectrum (Omdal et al., 2001). Similarly, two recent class II or III studies showed that SLE patients very often suffer from mood (44 and 48.8%) and anxiety (16 and 24%) disorders (Ainiala et al., 2001b; Brey et al., 2002).


There is only class III evidence that headache is associated with anxiety and depression.


By critically reviewing all available published information, we found no difference in the prevalence of headaches between SLE patients and controls. Moreover, no distinct headache syndrome or pathogenetic mechanism for headaches has been demonstrated in SLE patients. Thus, the occurrence of headache in SLE patients does not itself require further investigation. Headache in SLE patients should be classified according to IHS criteria and managed as primary headache in the absence of any complicating factors. If headaches are frequent, prophylactic treatment may be required. Although the presence of ‘lupus headache’ is considered important in activity indices for SLE patients (Bombardier et al., 1992), we found neither enough, nor good evidence for this particular headache type. Therefore, this term should be avoided in the absence of further studies. Evaluation for mood disorders may be helpful in making decisions related to the management of those patients for reasons of co‐morbidity not causality. Finally, there is need for a new, large‐scale class I study with neuroimaging, particularly in the paediatric population, to verify the findings of this review.


View Abstract