Are gender-specific approaches needed in diagnosing early axial spondyloarthritis? Data from the SPondyloArthritis Caught Early cohort

Spondyloarthritis (SpA) refers to a group of chronic inflammatory diseases that, according to the Assessment of SpondyloArthritis international Society (ASAS) classification criteria, may be divided by the predominant site of involvement into axial SpA (axSpA) and peripheral SpA [1, 2]. In addition, axSpA can be further split into radiographic axSpA (r-axSpA, also known as ankylosing spondylitis), characterised by typical radiographic changes in sacroiliac joints (SIJ) [3], and non-radiographic axSpA, a form of the disease in which structural damage in the SIJ is not (yet) present.

Although axSpA affects both men and women, r-axSpA is more common in males, with a reported male to female ratio of 2–3:1 [4–6]. The higher incidence of r-axSpA in males compared to females has stimulated research on gender differences in this disease. In patients with longstanding r-axSpA, males have more severe radiographic damage and often more inflammation [7–11]. On the other hand, some authors have reported that female r-axSpA patients more often have cervical spine involvement, peripheral pain, and worse patient-reported outcomes (PROs) [7, 10, 12, 13]. Moreover, a longer time frame between symptom onset (based on patient history) and diagnosis has been repeatedly noticed in female r-axSpA patients [9, 14]. This finding poses the question of whether there is an earlier presentation of the disease in males or an earlier detection of the disease after its onset. If there were phenotypical gender differences at the start or diagnosis of the disease, with less typical findings in women, this could result in a delayed diagnosis. The importance of gender differences in the entire axSpA spectrum has only recently been appreciated, especially since an earlier diagnosis of axSpA has become more feasible. Few studies have been performed in cohorts with early disease such as the German Spondyloarthritis Inception Cohort (GESPIC) and the DEvenir des Spondylarthropathies Indifférenciées Récentes (DESIR) cohort [15, 16], with only one report specifically focusing on possible gender differences [16]. In the GESPIC cohort, composed of patients with a certain diagnosis of axSpA and a mean disease duration approaching 4 years, male gender was independently associated with more radiographic damage [15]. In the DESIR cohort, which includes patients with a high likelihood of SpA (inflammatory back pain highly suspected of axSpA), in the subgroup of patients who fulfilled the ASAS criteria, males were younger at diagnosis, had higher C-reactive protein (CRP) levels, and more often had radiographic sacroiliitis and/or signs of inflammation of SIJ and the spine on magnetic resonance imaging (MRI) [16]. Both studies include only patients with an established (or at least very likely) diagnosis and do not have subjects without axSpA for comparison.

Given the clear gender difference in established r-axSpA, it would be important to know whether differences between male and female axSpA patients are already present early in the disease. If so, different diagnostic strategies may be needed for males and females, considering that diagnostic delay is reported to be higher in females [17, 18]. Therefore, we set out to analyse gender differences at the moment of diagnosis of axSpA in a large inception cohort of chronic back pain (CBP) patients of unknown origin. Our aim was to assess differences in disease presentation between males and females in early axSpA, and to determine whether different SpA features are important for the diagnosis of axSpA in the two genders, possibly demanding separate diagnostic strategies.

Out of the 719 CBP patients, 301 patients (42%) were diagnosed with axSpA by the rheumatologist with a level of confidence (LoC) ≥ 7 (axSpA patients). An additional 82 patients (35% males) out of 719 received a diagnosis of axSpA with a level of confidence in diagnosis < 7 (here included in the no-axSpA group).

Of the 275 male CBP patients, 146 (53%) received a certain diagnosis of axSpA, and of the 444 female CBP patients, 155 (35%) received a certain diagnosis of axSpA, showing that male patients with CBP were significantly more likely to be diagnosed with axSpA (OR 2.1, 95% CI 1.5–2.9). When the patients with a low certainty diagnosis of axSpA (LoC < 7) were removed from the no-SpA group and added to the axSpA group, the gender difference remained the same (OR 2.0, 95% CI 1.5–2.7).

Among axSpA patients, most SpA features were similarly distributed between the two genders. However, males were significantly more often HLA-B27⁺ (80% vs 60%; p < 0.001) and more often had imaging abnormalities (any positive imaging 78% vs 64%; p = 0.007) than axSpA female patients. The latter was mainly due to the fact that more males than females had both a positive MRI-SI and a positive X-SI (30% vs 17%; p = 0.006). Moreover, although syndesmophytes were uncommon in these patients with early disease, more males than females had syndesmophytes present (8% vs 1%; p = 0.02). In contrast, psoriasis appeared to be more common in females (24% vs 15%), although this difference was not statistically significant (p = 0.051). Interestingly, the total number of SpA features without HLA-B27 and imaging was not different between males and females (3.5 (1.7) vs 3.5(1.6); p = 0.99). Females were less frequently smokers than males (10% vs 22%; p = 0.02).

In both genders, all SpA features were significantly more common in axSpA patients as compared to no-axSpA patients except for IBD and family history. The latter was actually more common in male axSpA patients compared to male no-axSpA patients, but such a difference was not observed in female patients. Interestingly, HLA-B27 and positive imaging were still more prevalent in female axSpA patients than in no-axSpA patients, both females (HLA-B27⁺ 23% and positive imaging 7%) and males (HLAB27⁺ 34% and positive imaging 11%) (all p < 0.01).

Male gender was also positively associated with MRI-SIJ⁺ alone (OR 1.6, 95% CI 1.0–2.8) and with X-SIJ⁺ alone (OR 1.9, 95% CI 1.1–3.3) in the multivariable models having MRI-SIJ⁺ and X-SIJ⁺ separately as outcomes (data not shown).

Patient-reported outcomes, metrological indices, and disease activity indices of axSpA patients were recorded in the cohort and are presented in Additional file 1: Table S1. Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) scores as well as the percentage of patients presenting with BASDAI > 4 were higher in females (4.5 (2.2) vs 3.4 (1.9) and 57% vs 34%; both p < 0.001). The Bath Ankylosing Spondylitis Functional Index (BASFI) and Short Form (SF)-36 outcomes were not significantly different between genders. Disease activity based on the Ankylosing Spondylitis Disease Activity Score (ASDAS) was similar in terms of absolute value, but females more frequently had a high or very high disease activity (ASDAS ≥ 2.1, 65% vs 46%; p = 0.004). Similarly, total MASES was similar between males and females, but the percentage of patients with current entheseal involvement (MASES > 0) was higher in females (70% vs 41%; p < 0.001). Finally, more female than male axSpA patients (43% vs 20%; p = 0.002) had an extreme BASDAI score (score ≥ 7 on a 0–10 scale in at least three questions of the BASDAI).

The analysis of gender differences in this large CBP cohort with short duration of complaints allowed us to conclude that, while females are the majority of patients in the original CBP population, males were twice as likely to be diagnosed with axSpA. At the moment of diagnosis of axSpA, males were younger and more often presented with HLA-B27 positivity and imaging abnormalities, including a higher prevalence of syndesmophytes, even though symptom duration was similar to female patients. However, the prevalence of most other SpA features was equally distributed between genders.

Although we noticed differences in disease presentation, the results of our study suggest that similar elements are relevant for the diagnosis of male and female axSpA patients. In fact, despite HLA-B27 positivity and imaging abnormalities being more common in male axSpA patients, they were significantly associated with diagnosing axSpA in both genders. We believe this is this first study to examine the possible relevance of different SpA features in making a diagnosis stratified for gender. This is the first report showing in a CBP cohort that males have a higher probability to be diagnosed with axSpA. Since all patients, suspected of axSpA or not, in our cohort underwent the same protocol (including extensive imaging), the gender difference cannot be attributed to differences in the diagnostic process.

Nevertheless, it is obvious that early axSpA does not have such a clear male gender skewing as r-axSpA, since female patients still represent nearly half of the axSpA population, as previously noted [15, 16]. This suggests that specific gender disparities (e.g. radiographic evidence of sacroiliitis) may become more apparent later in the disease course.

Our finding of male patients being younger at disease onset is in line with findings in the DESIR cohort [16]; considering the strict age span allowed by the inclusion criteria and given that males and females in our cohort had similarly short symptom duration, this likely reflects a true difference in the disease onset, rather than a better recognition of the disease in males. Previous reports on age of onset in male and female axSpA patients mainly concern longstanding disease, and some of them reported that males seemed to have an earlier onset [12, 23, 24]. The mechanism behind this is unclear and deserves further investigation.

The results of this study also confirm that imaging abnormalities are more frequently found in male axSpA patients, already at the beginning of the disease [15, 16, 24]. Moreover, we show that male sex is independently associated with positive imaging, both in terms of MRI inflammation and radiographic sacroiliitis. Remarkably, syndesmophytes were more common in males at baseline, showing that differences in radiographic damage between genders, which have been repeatedly observed in r-axSpA [7, 10, 11], can already be found in very early disease. Also, radiographic or MRI sacroiliitis was the most important item for diagnosis in women, and was significantly more common in both male and female axSpA patients compared to no-axSpA patients.

HLA-B27 was also more common in male axSpA patients, and the overall prevalence of HLA-B27 in axSpA patients in our study was similar to other early axSpA cohorts [25]. However, it has to be noted that in other axSpA populations, such as those of the clinical trials, where axSpA patients are included only if they fulfil ASAS criteria and not based solely on the rheumatologist diagnosis, often display higher percentages of HLA-B27⁺ patients [26, 27].

In studying the factors associated with sacroiliitis in patients with axSpA, a positive family history was protective of positive imaging even in univariate analysis. At present we have no clear explanation for this observation. Little is known about differences in axial spondyloarthritis between patients with and without a familial history. One study reported that familial ankylosing spondylitis was significantly milder than sporadic disease [28], while another found no difference in disease phenotype [29].

Although PROs were not the main outcome of our study, it is interesting to observe how, in line with data from the literature, female patients in our cohorts presented with a higher subjective disease burden [7, 10, 12, 13]. Unfortunately, our cohort lacks assessment of fibromyalgia, but the observation of female axSpA patients presenting with more extreme PROs than males suggests that fibromyalgia could have been a frequent co-morbidity for our axSpA patients, possibly influencing the degree of symptom complaint in females. Indeed, it has been described that about a third of axSpA patients could have a neuropathic pain component [30].

A clear limitation in our analysis is that our diagnostic models for axSpA do not truly reflect a normal diagnostic work-up in clinical practice, in which looking for—and excluding—other causes of CBP is pivotal. However, the validity of our observations derives from the fact that all of the information included in the model was available to the rheumatologist and was used at the moment of making the diagnosis (axSpA/no axSpA). On the other hand, rheumatologists certainly performed a diagnosis according to their background knowledge and beliefs, and therefore there could have been a tendency towards a specific interpretation of the results (e.g. women could have been under-diagnosed if the rheumatologists interpreted axSpA as a predominantly male disease). Finally, we cannot exclude that possible referral bias may have interfered with the proportion of males and females in the study population or with other results such as the nearly significant prevalence of psoriasis in females. Indeed, the reasons why GPs or medical specialists referred CBP patients to the rheumatologists is unknown, but they could have been more prone to refer a female CBP patient if SpA features (besides HLA-B27 and imaging) were already present or because they had worse symptoms, thus a channelling mechanism cannot be excluded. However, other cohorts have reported a higher frequency of CBP in females [31]; moreover, the male to female ratio in our axSpA population is in line with previous findings [15, 16].

In summary, in patients with CBP of unknown origin, males are more likely to be diagnosed with axSpA. In addition, significant differences between genders, including age at onset and imaging abnormalities, already exist at a very early disease stage. Nevertheless, HLA-B27 and imaging are still key elements for a diagnosis of axSpA in both genders, and therefore our data do not support the need for gender-specific diagnostic strategies for early axSpA patients.