Article Text

Download PDFPDF

Recurrence of spondylarthropathy among first-degree relatives of patients: a systematic cross-sectional study
  1. E Dernis1,2,
  2. R Said-Nahal3,4,
  3. M-A D’Agostino3,4,
  4. P Aegerter4,5,
  5. M Dougados1,2,
  6. M Breban3,4,6,7,8
  1. 1
    Rheumatology Division, Cochin Hospital, Assistance Publique des Hôpitaux de Paris, Paris, France
  2. 2
    Université Paris-Descartes, Paris, France
  3. 3
    Rheumatology Division, Ambroise Paré Hospital, Assistance Publique des Hôpitaux de Paris, Boulogne-Billancourt, France
  4. 4
    Université Versailles-Saint-Quentin en Yvelines, Versailles, Paris, France
  5. 5
    Epidemiology and Statistics, Ambroise Paré Hospital, Assistance Publique des Hôpitaux de Paris, Boulogne-Billancourt, France
  6. 6
    INSERM U567, Paris, France
  7. 7
    CNRS UMR8104, Paris, France
  8. 8
    Institut Cochin, Paris, France
  1. M Breban, Rheumatology Division, Hôpital Ambroise Paré, 9 Avenue Charles de Gaulle, 92100, Boulogne, France; maxime.breban{at}


Objective: To examine the recurrence of manifestations belonging to the spectrum of spondylarthropathy (SpA) in first-degree relatives of patients with SpA, and to estimate the recurrence risk ratio.

Methods: Parents and siblings of consecutive SpA probands have been thoroughly investigated, including clinical data collection, pelvic x ray and human leukocyte antigen (HLA)-B27 status determination. The diagnosis of SpA was made according to European Spondylarthropathy Study Group and/or the Amor criteria. The recurrence risk ratio λ1, which gives an estimate of the weight of genetic factors, was calculated as the ratio of the recurrence risk of SpA in first-degree relatives compared with the population prevalence of SpA. The λnon-HLA was obtained by similar calculations restricted to HLA-B27+ individuals.

Results: Most manifestations of SpA were more frequent among the 157 HLA-B27+ relatives of 83 probands than among their 111 HLA-B27- relatives. A diagnosis of SpA was made in 50 relatives of 31 (37%) probands. Recurrence was very similar between parents and siblings, without gender difference, resulting in overall recurrence risk of 12% in first-degree relatives and of 22.7% in HLA-B27+ relatives. The λ1 value was 40 and the λnon-HLA value was 6.5, very close to the λHLA value of 6.25 estimated from linkage study in SpA.

Conclusions: A similar recurrence risk of SpA was observed between parents and siblings, consistent with a model of inheritance with no dominance variance and without sex influence. The weight of the non-HLA genetic component was equivalent to that estimated for the HLA locus, and fitted a model of multiplicative interaction between HLA and non-HLA genetic components.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The spondylarthropathies (SpA) are a group of inflammatory diseases that have been brought together on the grounds of shared skeletal manifestations and frequent association with overlapping extra-articular features.1 Diagnostic entities that make up SpA are ankylosing spondylitis (AS), which is characterised by predominant axial skeletal involvement and advanced radiographic sacroiliitis; reactive arthritis, which follows a triggering infection; a subset of psoriatic arthritis; arthritis associated with idiopathic inflammatory bowel diseases (IBD); and undifferentiated SpA. The validity of a unifying concept of SpA is supported by anatomical studies, which have identified enthesitis as a characteristic feature of SpA,2 and also by the observation that distinct entities of SpA often occur in different individuals within the same families.3 Part of the latter observation is explainable by the role of human leukocyte antigen (HLA)-B27, a genetic predisposing factor shared by the different varieties of SpA.4

By studying families with multiple cases of SpA, we have shown that distinct subtypes of SpA appeared quite evenly distributed within families.5 Most importantly, all articular and extra-articular manifestations belonging to the spectrum of SpA appeared linked together and to HLA-B27.6 These results supported a model in which different forms of SpA were determined by shared predisposing factors, whereas secondary factors were responsible for phenotypic variations. We inferred that SpA subtypes should be considered as phenotypic variations of a unique disease rather than truly different conditions, and should be studied together in genetic studies aimed at identifying genes determining susceptibility to SpA.79 The validity of our approach was further supported by the whole-genome screen that we performed in SpA, which allowed us to identify a significant linkage with a locus on 9q31–34,10 overlapping with one of the regions independently reported as being putatively linked with AS.11

As a complex disorder, SpA is thought to result from a combination of genetic predisposition and environmental exposure. Genetic epidemiology studies evaluating the familial aggregation of a disease are important to give a sense of its heritability. This type of investigation is routinely performed by studying the frequency of a disease among relatives of randomly selected patients, which can then be compared to its population prevalence. In SpA, such studies have largely been focused on AS, a strictly defined form of SpA, since its diagnosis relies on advanced radiographic sacroiliitis.12 However the radiographic criterion is only met after a variable time-course of disease, and a number of patients lacking this criterion can be misclassified as unaffected, leading to truncated data and to erroneous conclusions.13 14 In the present study, we sought to establish the actual recurrence of disease among parents and siblings of patients with SpA by considering all patients with SpA as affected. We then calculated the recurrence risk ratio of SpA for first-degree relatives, as compared to the general population.


Subjects and study design

Consecutive outpatients with a definite diagnosis of SpA according to European Spondylarthropathy Study Group (ESSG) criteria15 and/or the criteria of Amor et al,16 and having consulted at Cochin Hospital tertiary care rheumatology centre in a predefined period of time were proposed to participate to the study. Only patients from the surrounding area (Ile-de-France) having at least one first-degree relative living in France were included, if they signed an informed consent. They are referred to as probands.

All available parents and siblings of each proband were screened by a questionnaire previously validated for the detection of SpA.6 Physical examination by a rheumatologist was performed for all individuals with positive answer by questionnaire. The diagnosis of psoriasis required the presence of typical lesions or a diagnosis previously made by a dermatologist; a diagnosis of anterior uveitis was retained, if established by an ophthalmologist; the diagnosis of IBD required typical gut endoscopic or histological findings of Crohn disease or ulcerative colitis. Standard anteroposterior radiographs of the pelvis were performed on all symptomatic individuals, except for those for whom a recent x ray (<6 months old) or a radiograph showing evidence of at least bilateral grade II or unilateral grade III sacroiliitis was available. Additional skeletal radiographs, tomodensitometry and magnetic resonance imaging were obtained when necessary to help with the diagnosis. All radiographs were examined by a single qualified examiner (RSN), using an established grading system.12 A grade of II bilaterally or of III unilaterally was required for a definite diagnosis of sacroiliitis.

As for probands, the diagnosis of SpA was made according to ESSG criteria15 or/and to the criteria of Amor et al.16

HLA-B typing

HLA-B typing was routinely performed by a polymerase chain reaction-based sequence-specific method on DNA extracted from peripheral venous blood leukocytes.17 For individuals already typed as positive for HLA-B27, retyping was not systematically performed.

Recurrence risk and recurrence risk ratio

The recurrence risk of SpA in parents was the number of observed cases divided by the number of studied parents.

To calculate the recurrence risk of SpA in siblings, we accounted for the following ascertainment bias due to the proband-based study design: the chance of sampling a sibship is proportional to the number of SpA cases in that sibship. Thus, we applied a correction by dividing the number of sibships with given SpA cases, by the number of SpA cases in that sibship, including the proband.18

The recurrence risk ratio in first-degree relatives, denoted as λ1, is the ratio of the recurrence risk in those relatives and the population prevalence of the disease. We set the population prevalence of SpA in the French adult population to 0.3% (95% CI 0.17 to 0.46).19 The λnon-HLA reflects the contribution of non-HLA loci to the genetic predisposition. Since HLA-B27 accounts for most of the HLA-borne genetic predisposition,9 this value is approximately the ratio between the recurrence risk of SpA in HLA-B27+ relatives and its prevalence in HLA-B27+ individuals of the population. The latter value was set to 3.5%, considering a prevalence of HLA-B27 in the French population of 7.2%,20 and that 80% to 90% of patients with SpA are HLA-B27+.16 21 22 The λHLA for SpA was estimated in a linkage study to 6.25.10

Statistical analysis

Differences between groups were tested by unpaired t test (numerical data) or by the Fisher exact test (categorical data). A p value<0.05 was considered as significant. When calculable, the relative rate (RR) and its 95% CI were also determined, for comparison of categorical data between groups.


SpA features in probands and siblings

In all, 83 probands (52 males, 31 females) and their relatives consisting of 147 siblings (62 brothers, 85 sisters) and 141 parents (65 fathers, 76 mothers) were included in the study. Two probands were siblings of each other and appeared twice in the results (as proband and as sibling). Pelvic radiograph and HLA-B27 status were obtained in 59% and 93% of the relatives, respectively. Table 1 shows the frequency of SpA-related manifestations in relatives with known HLA-B27 status. All skeletal manifestations and uveitis were more frequent in HLA-B27 positive than HLA-B27 negative relatives. Psoriasis was also more frequent in HLA-B27 positive than HLA-B27 negative siblings. The frequency of IBD was independent of the HLA-B27 status of relatives (table 1).

Table 1 Prevalence of spondylarthropathy (SpA)-related manifestations in relatives of SpA probands, in reference to human leukocyte antigen (HLA)-B27 status

A total of 81 (98%) probands fulfilled the Amor and ESSG criteria, respectively. A diagnosis of SpA was made in 50 relatives of 31 (37%) probands, consisting of 35 siblings and 15 parents fulfilling the Amor and ESSG criteria (3 previously undiagnosed). Characteristics of patients are shown in table 2. HLA-B27 was positive in 93% of the probands and 98% of their SpA relatives. Male to female ratio was 1.68 among probands and 0.72 among SpA relatives, the difference being significant (p = 0.03). The latter value was similar to that observed among all relatives (0.79), or among HLA-B27 positive relatives (0.85). Age and disease duration were much higher in parents, than in probands or siblings. However, age at onset was significantly lower in probands. About two-thirds of the probands and parents fulfilled AS criteria, whereas only 37% of the SpA siblings did so (table 2). Manifestations such as peripheral enthesitis, peripheral arthritis and psoriasis were significantly more prevalent in probands than in SpA relatives.

Table 2 Demographic, clinical and radiographic features in spondylarthropathy (SpA) probands and their SpA relatives

There was no significant difference for any characteristic of the probands between those with and those without affected relatives (data not shown). There was no concordance between gender or SpA manifestations of probands and the same presentation features observed in their relatives, as shown in table 3 for all relatives. This result applied to parents as well as siblings (data not shown).

Table 3 Concordance of characteristics between probands and spondylarthropathy relatives

Recurrence risk of SpA in relatives

The size of the sibships ascertained varied from 2 to 10 and the number of SpA cases from 1 to 4, including the proband. Correction applied to take in account the ascertainment bias is shown in table 4.

Table 4 Distribution of spondylarthropathy (SpA) cases among siblings, according to the size of sibship

Recurrence rate of SpA in relatives was similar between genders, and close for parents (10.6%) and siblings (13.9%) (table 5). One single SpA case was ascertained among HLA-B27 negative relatives, in a sister of an HLA-B27 negative proband. The recurrence risk was 46-fold greater among HLA-B27 positive, than HLA-B27 negative relatives. The numbers of HLA-B27 negative and HLA-B27 positive parents were relatively well balanced. In the latter group, the recurrence was 21.4% (table 5). In contrast, there were 1.5-fold more HLA-B27 positive than HLA-B27 negative siblings. The recurrence rate was 24.4% in HLA-B27 positive siblings.

Table 5 Recurrence of spondylarthropathy in first-degree relatives

We tested whether the sex of the proband influenced the recurrence rate in relatives (table 6). There was a general trend towards a greater recurrence rate in parents (relative risk: 1.4) and siblings (relative risk: 2) of male, than female probands, but the difference was not significant (table 6).

Table 6 Recurrence of spondylarthropathy in first-degree relatives of probands, according to gender

Recurrence risk ratio for SpA

The λ1 was very close for parents and siblings (35 and 45, respectively) and overall λ1 was 40. Calculations restricted to HLA-B27+ individuals yielded a λnon-HLA value for first-degree relatives of 6.5. Assuming multiplicative interaction between non-HLA and HLA components, the λHLA value would be 6.15, very close to the 6.25 value, which was calculated from our linkage study in French SpA families.10


Several studies aimed at identifying genetic factors of predisposition to AS and SpA, have recently been undertaken, including genomewide linkages.7 10 23 Such investigations relied on the assumption that the genetic basis of those disorders could not be entirely accounted for by the HLA region.24 However, data supporting such statement have remained relatively scarce to date.

The recurrence risk ratio, denoted as λ1 for first-degree relatives and λs for siblings, allowed us to estimate the weight of genetic factors underlying predisposition to a disease.25 It is the ratio between the recurrence risk in the kind of relatives studied and the population prevalence of the disease. Thus, proper ascertainment of the recurrence rate of a disease in first-degree relatives of patients is an important step towards prediction of its degree of inheritance. For studies of relatively infrequent disease such as SpA, a proband-based design is the most realistic one, but a number of biases need to be avoided.2628 Probands should be representative of the patient population. The status of relatives should be ascertained and not just classified as affected or not on the basis of the proband claim. Comprehensive ascertainment of relatives should be obtained inasmuch as possible, to limit the natural tendency to over-report affected individuals. Calculations should be adapted to the type of relatives, and calculations of the recurrence risk among siblings should account for the increased chance of randomly sampling sibship in proportion to the number of cases that it contains.18 In the case of AS and SpA, none of the studies published to date have eliminated all these possible biases. Separate calculations of the relative risk for different kinds of relatives was only rarely performed, and even in those studies, no correction was applied to the sibling recurrence risk calculation.2931

To date, most studies of the recurrence risk among relatives have been focused on the AS phenotype.24 Several reports in the past have emphasised the high frequency of non-AS cases belonging to SpA among relatives of patients with AS,3236 but only one study has examined the corresponding patients on a systematic basis,37 and none has embraced the full spectrum of SpA among probands and relatives. Since the radiographic criterion is sometimes met only after a very long period of time,38 failing to ascertain SpA cases without advanced radiographic sacroiliitis results in truncated data that may lead to erroneous conclusions.13 14 Such an issue concerns particularly female patients, who tend to be misclassified as unaffected, because of their weaker propensity to develop radiographic sacroiliitis than males.39

The present study was the first to systematically examine the recurrence risk of SpA among first-degree relatives of a large number of consecutive SpA cases. All manifestations of SpA, including extra-articular features to the exception of IBD, were increased in HLA-B27 relatives, supporting the theory that HLA-B27 might predispose to all of them.6 40 More than 50% of the relatives classified as affected did not meet AS criteria. However, 98% of them were typed positive for HLA-B27, a finding that supports the validity of the diagnostic procedure. There was no concordance of SpA features between probands and their affected relatives, in agreement with the conclusion drawn in the familial context that distinct SpA phenotypes depend on ubiquitous genetic or environmental influences, rather than any major specific factors.6 The recurrence risk of SpA was 12% among first-degree relatives, much higher than the 4.9% recurrence risk derived from an AS-restricted study previously used to build inheritance models.24 30 By contrast, the 21.4% recurrence risk that we calculated in HLA-B27+ parents was nearly identical to the 21% recurrence of AS in HLA-B27+ first-degree relatives aged ⩾45 years, previously reported by van der Linden et al.41 The high similarity of our calculation with that of the latter study might not be surprising, considering that (i) the risk of missing affected patients by using AS criteria, such as in the study by van der Linden et al, was minimised in the oldest relatives; and (ii) a greater proportion of cases was likely identified in our study by using SpA criteria, but was counterbalanced by the lack of correction applied to the sibling recurrence risk in the study by van der Linden et al.

We observed a very similar recurrence risk between parents and siblings, which is consistent with a model of inheritance for SpA with no dominance variance as previously proposed for AS.24 Assuming a multiplicative model of interaction between HLA and non-HLA genetic components, the respective weight of these components, as estimated by their λ values (ie, 6.15 and 6.5, respectively) were almost equivalent to each other and fully consistent with the 6.25 λHLA value resulting from our linkage study in SpA.10 Those estimates of λ140 and λHLA are at variance with those formerly used for AS, which assumed a λ1 of 80 and a λHLA of 3.6. Nevertheless, the predicted number of non-HLA genes would be similar (ie, about 4, supposing that each of them has a λ value in the range of 1.5–1.75 and interacts multiplicatively with HLA-B27, independently of each other). Interestingly, such a model fits our prediction regarding the first non-major histocompatibility complex (MHC) SpA locus mapped by linkage to chromosome 9q31-34.10

As a possible bias inherent to the present study performed in a tertiary care centre, it is likely that the probands were more severely affected than are patients with SpA in general. First, the proportion of HLA-B27+ probands was higher than usual in SpA,16 21 22 and HLA-B27+ cases tend to be the most serious ones.42 There was a male predominance among probands, contrary to SpA relatives, as previously reported.37 Probands had a younger age at onset and more frequent peripheral arthritis, enthesitis and psoriasis than relatives. These differences were formerly found to distinguish between two phenotypes of familial SpA, which might correspond to distinct severity patterns.38 39 As another difference, affected siblings had lower frequency of advanced radiographic sacroiliitis than probands and parents. This could be explained by the influence of gender and disease duration on radiographic sacroiliitis.38 39 Thus, it is likely that siblings had less radiographic sacroiliitis than probands because of their female predominance and shorter disease duration, whereas parents who were also more frequently females than probands had a similarly high proportion of sacroiliitis because of the much longer disease duration.

There was a sex imbalance in probands but not in relatives. Such an equal genetic risk for SpA between genders might have previously been overlooked because of truncated data and/or bias ascertainment, as discussed above.29 Our result is consistent with an equivalent estimate of SpA prevalence between genders in the French population.19 We did not observe an increased risk of SpA in siblings of female probands, as previously described for AS,29 but a non-significant trend in the opposite direction. The risk of SpA was greatest in relatives of males, and especially in their siblings, which could result from ascertainment bias. Hence, unaffected relatives of males could have been more reluctant to enrol in this study than the affected ones, in contrast to an easier recruitment of all relatives of females. Such bias could also account for an excess of HLA-B27+ siblings as compared to the theoretical expectation, albeit not more pronounced among relatives of males than females (data not shown). Alternatively, males of families with several cases could be more prone to consult in tertiary care centre than females, an interpretation fully consistent with the predominance of males in probands, but not in relatives.

In conclusion, this study of recurrence in first-degree relatives of patients allowed us for the first time to estimate the inheritability of SpA. The genetic risk appeared to be lower than previously found in AS studies, identical in males and females and similar between parents and siblings, fitting a model of inheritance with no dominance variance or sex influence. The predicted weight of the non-HLA component was close to that of HLA.


We gratefully acknowledge the contributions of Dr Selam El Hassani to subject recruitment and of Alexandre Santenero to HLA-B27 typing.



  • Competing interests: None declared.

  • Funding: This work was supported by grants from Schering-Plough (2001) and Programme Cohortes et Collections Biologiques from Institut National de la Santé et de la Recherche Médicale (2002).

  • Ethics approval: Ethics approval was obtained.