Article Text

Download PDFPDF

Concise report
Female patients with ankylosing spondylitis: analysis of the impact of gender across treatment studies
  1. Irene E van der Horst-Bruinsma1,
  2. Debra Jeske Zack2,
  3. Annette Szumski3,
  4. Andrew S Koenig3
  1. 1Department of Rheumatology, VU University Medical Centre, Amsterdam, The Netherlands
  2. 2Inflammation Global Clinical Development, Amgen Inc., Thousand Oaks, California, USA
  3. 3Inflammation and Immunology, Department of Specialty Care, Pfizer Inc., Collegeville, Pennsylvania, USA
  1. Correspondence to Dr Irene E van der Horst-Bruinsma, Department of Rheumatology, VU University Medical Centre, 3A-47, PO Box 7057, Amsterdam1007 MB, The Netherlands; IE.vanderHorst{at}


Objectives To examine the impact (if any) of gender on the clinical, functional and patient-reported outcomes of treatment using data pooled from four controlled clinical trials.

Methods Study data were pooled from four clinical control trials in which 1283 adult patients with active ankylosing spondylitis (AS) were treated with etanercept, sulfasalazine or placebo. Patients were stratified by gender and analysed for differences/similarities in baseline demographics, disease characteristics, and efficacy in AS outcome measures and safety and discontinuation rates after 12 weeks of therapy.

Results Significant baseline differences were observed between 326 female patients compared with 957 male patients. Female patients had an older mean age of disease onset (35.0 vs 31.2 years; p<0.001), shorter mean time of disease duration (7.4 vs 9.5 years; p<0.001) and lower mean baseline C-reactive protein (13.1 vs 20.9 mg/l; p<0.001); a lower proportion was HLA-B27 positive (76.3% vs 85.2%; p<0.001) compared with male patients. Women had significantly (p<0.001) smaller differences in all week 12 efficacy assessments including AS disease activity score (0.87 vs −1.08), Bath AS disease activity index (−19.22 vs −23.41) and Bath AS functional index (−13.89 vs −16.88) relative to men. A similar relationship was observed between women and men in the adjusted mean difference of nocturnal back pain (4.04, 95% CI 0.77 to 7.32; p<0.05), total back pain (3.80, 95% CI 0.77 to 7.32; p<0.05) and patient global assessment (4.79, 95% CI 1.51 to 8.08; p<0.01).

Conclusions Women had a higher burden of disease and less improvement in AS outcome measures compared with men. This was observed despite women having a later disease onset of shorter duration; the mechanism behind this observation is unclear. Additional research is necessary to better understand female patients with AS and the burden of disease in this population.

  • Ankylosing Spondylitis
  • Anti-TNF
  • DMARDs (biologic)
  • Treatment
View Full Text

Statistics from


Ankylosing spondylitis (AS) is a chronic rheumatic disorder characterised by inflammatory back pain, enthesitis and arthritis of the peripheral joints that can result in substantial quality-of-life impairment and disability.1–3 Historically, AS has been considered a disorder that occurs overwhelmingly in men, with early studies reporting a ratio of 9–10:1 for prevalence in men to women.4 ,5 However, from more recent estimates, based on less severe cases and with new detection techniques such as MRI, the prevalence appears higher in men by only a 2–3:1 ratio.6–12

The persistent belief that AS primarily affects men may have contributed to the underdiagnosis or late recognition of this disease in women.13 Previously published data have been limited because a substantially lower percentage of women were included, and therefore the natural history of the disease was not studied uniformly across genders. Enrolment of female patients into clinical AS treatment trials remains a challenge. Clinicians’ perception of AS as predominantly a male disease may introduce bias. Studies suggest both similarities and differences between the genders in terms of clinical parameters, x-ray results and function.13 ,14 Little is known about potential differences in treatment outcomes. Most studies are not powered to test similarities/differences across genders because of enrolment imbalance; therefore, it is a rare opportunity to pool data to address some of these questions concerning gender disparity in AS.

The objective of this analysis is to examine the impact (if any) of gender on the clinical, functional and patient-reported outcomes of treatment using data pooled from four blinded controlled global clinical trials.15–18


Study data were pooled from four published clinical control trials in which adult patients (aged ≥18 years) with active AS were treated with etanercept, sulfasalazine or placebo.15–19 Randomised patients who received ≥1 dose were stratified by gender and analysed for baseline demographics and disease characteristics. Efficacy analyses between genders and treatments included data from patients who received ≥1 treatment dose collected at baseline and after 12 weeks of treatment (modified intent to treat population, observed case). Week 12 efficacy and safety assessments included the Bath AS disease activity index (BASDAI), Bath AS functional index (BASFI), patient global assessment, AS disease activity score (ASDAS), total and nocturnal back pain, incidence of adverse events (AEs) and study discontinuation rates.

Statistical analyses

For analyses of gender differences in baseline variables, continuous variables were analysed using a one-way analysis of variance and categorical variables were analysed using a Cochran–Mantel–Haenszel test of general association. For analyses of gender differences in efficacy variables, continuous responses were analysed using an analysis of covariance model of week 12 change in response with baseline of response, gender, study, treatment and geographic region as predictors. Adjusted means of ASDAS (absolute and change) for each gender were obtained from this model and adjusted for the other model predictors. Adjusted mean gender difference (with 95% CI) and corresponding p value were also obtained from this model. Dichotomous responses were presented as the proportion of patients with response for each gender, and the difference in proportions (with 95% CI) between genders was analysed using a logistic model of response at 12 weeks of treatment with baseline of response, gender, study, treatment and geographic region as predictors. The OR, adjusted for other model predictors, was obtained for women versus men, with corresponding χ2 p value. Patients in the etanercept 25 mg twice weekly treatment group were combined with patients in the etanercept 50 mg once weekly treatment group for analysis.

Adverse events of clinical interest were summarised by gender. No statistical tests were conducted because of the small number of events. For each treatment group, gender differences in the proportion of patients who discontinued were analysed by log-rank test, adjusted for the study. A HR of study discontinuation for women versus men was calculated from a Cox proportional hazards model, adjusted for study and treatment.


Baseline assessments

Baseline analysis includes 1283 patients (table 1). Significant baseline differences between genders were observed, with female patients having an older mean age of disease onset (35.0 vs 31.2 years; p<0.001) and a shorter mean time of disease duration (7.4 vs 9.5 years; p<0.001) compared with male patients. Interestingly, women had lower mean baseline C-reactive protein (CRP) levels (13.1 vs 20.9 mg/l; p<0.001), and a lower proportion was positive for the HLA-B27 allele (76.3% vs 85.2%; p<0.001) compared with men. Female patients exhibited a significantly more severe disease state at baseline compared with male patients as measured by BASDAI, patient global assessment, and total and nocturnal back pain (p≤0.05). Mean baseline BASFI and ASDAS scores were not significantly different between genders.

Table 1

Baseline characteristics

Efficacy assessments

Efficacy assessments for 1203 patients (297 (24.7%) women and 906 (75.3%) men) with baseline and week 12 data were carried out. Although both gender groups showed improvement in AS signs and symptoms, women had significantly smaller changes in all end points examined at week 12 relative to men. Female and male patients had an adjusted mean change in ASDAS scores of −0.87 and −1.08, respectively, for a significant mean difference of 0.21 (95% CI 0.09 to 0.33; p<0.001). Women also experienced a lower improvement in disease activity (BASDAI), with a mean change of −19.22 compared with −23.41 for male patients, for a significant difference of 4.19 (95% CI 1.43 to 6.95; p<0.01). Improvement in physical functioning was lower in women compared with men, with a BASFI mean change of −13.89 versus −16.88, respectively, for an adjusted mean change of 2.99 (95% CI 0.22 to 5.77; p<0.05). A similar relationship was observed between women and men in the adjusted mean difference of nocturnal back pain (4.04, 95% CI 0.77 to 7.32; p<0.05), total back pain (3.80, 95% CI 0.77 to 7.32; p<0.05) and patient global assessment (4.79, 95% CI 1.51 to 8.08; p<0.01).

After 12 weeks of treatment, female patients had less improvement in AS than male patients based on the proportional differences in ASDAS cut-off scores for low (<1.3), moderate (<2.1) and high (<3.5) disease activity (table 2). Logistic modelling indicated female gender was a significant predictor of poorer ASDAS improvement compared with male gender. The adjusted OR (95% CI) for ASDAS improvement ≥1.1 in week 12 female-versus-male outcomes was 0.55 (95% CI 0.40 to 0.75; p=0.0002), and for ASDAS improvement ≥2 was 0.60 (95% CI 0.42 to 0.87; p=0.0068).

Table 2

Improvements in ankylosing spondylitis disease activity score (ASDAS) at week 12 of treatment

Safety and discontinuations

Selected AEs of clinical interest were found in similar patterns across treatments and were frequently flares of pre-existing conditions (table 3). However, lower exposure-adjusted event rates were observed for infections and uveitis in patients receiving etanercept than in those receiving placebo. Exposure-adjusted event rates were lower among etanercept-treated women versus etanercept-treated men for most selected AEs (except malignancies, which did not occur in any patient).

Table 3

Selected adverse events and disease flares by gender, total events (exposure-adjusted event rate per 100 patient-years of exposure)

The proportion of patients who discontinued from studies at 12 weeks of treatment was primarily similar across the treatment arms. However, higher proportions of women receiving etanercept 50 mg once weekly and sulfasalazine 3 g once weekly were discontinued than were men (figure 1; p<0.05). Overall, women had a 50% higher rate of study discontinuation relative to men (HR=1.486; p=0.0077).

Figure 1

Study discontinuations at week 12 by gender and treatment. p Value from log-rank test, adjusted study. BIW, twice weekly; ETN, etanercept; PBO, placebo; SSZ, sulfasalazine; QW, once weekly.


Historically, AS has been known to be much more prevalent in men relative to women; however, improvements in diagnostic techniques reveal more women with AS than was previously thought. Women with AS may represent a largely under-diagnosed and under-studied population. To understand and effectively treat this under-recognised population, gender comparisons are warranted.

Lee et al13 summarised AS gender differences over time from published literature and noted the obvious delay in diagnosis in women compared with men as recently as the early 2000s, highlighting the perception by clinicians of AS being primarily a male-specific disease. Although time to diagnosis is equalising today, women remain under-diagnosed. Missed diagnoses may be attributable to differing presentation in women versus men. Women are more likely to have peripheral arthritis/cervical pain, and less thoracic/lumbar radiographic severity/damage compared with men.13 ,14 Currently no differences in functional or permanent work disability have been clearly established; however, data in this area are limited because AS gender differences/similarities have been only recently explored.14 ,20–22

In this post hoc analysis, female patients with AS had a more severe burden of disease at baseline and less improvement in most treatment outcomes compared with male patients. At baseline, women had significantly shorter disease duration and therefore were diagnosed earlier compared with men. Interestingly, the mean age of symptom onset appears to be lower in men (21.7 years) compared with women (27.6 years). The mechanism behind women having a later disease onset of shorter duration is unclear. The higher age at diagnosis in women may contribute to the more severe burden of disease at baseline, possibly leading to more resistant disease and a reduced response to treatment. The smaller improvement in disease outcomes shown by women compared with men supports the observation that women do have more refractory disease, whether from more severity at baseline or from a longer latency from first symptom to first diagnosis. The higher number of ASDAS improvers in men can be partly explained by the higher CRP levels in the male participants at baseline, which are part of the ASDAS calculation. In addition, the higher discontinuation rates observed among women in this analysis may relate to this higher burden of disease and reduced response to treatment.

The number of those diagnosed with fibromyalgia during the trial period was low (4 men, 10 women). As all investigators and clinical trial sites were highly qualified AS centres, it is unlikely that the reduced response to treatment in women is due to an undiagnosed, concurrent condition such as fibromyalgia. However, through these observations we do acknowledge the need for formal fibromyalgia assessments in AS clinical trials.

A limitation to this research is that the gender-by-study-interaction analyses showed significant differences among the four studies regarding gender impact on dichotomous ASDAS end points but not on continuous end points. These studies were not designed to evaluate gender differences, nor were they uniformly distributed geographically and cannot assess the channelling bias introduced by some healthcare professionals regarding who does or does not have AS based on gender. This is a post hoc subgroup analysis and may not generalise to the entire population of AS patients. These patients represent the more severe disease state, and results of this analysis may differ in an analysis of the earlier or mild cases. These studies were based on the modified New York criteria and may not be the same for patients diagnosed based on the new Assessment in Spondyloarthritis International Society (ASAS) non-axial spondyloarthritis criteria.23 ,24 Data are limited with this pooled analysis, and additional research is needed on gender differences and on AS work impact/work productivity and symptom duration.

Lastly, the assessments, like most of rheumatology, are a mix of objective (MRI/CRP), semi-objective (ASDAS) and subjective tools (pain scales). As a consequence, it may be challenging to discriminate other conditions that may impact pain and function that are not related to AS. For example, it is difficult to detect fibromyalgia and non-AS musculoskeletal back pain by objective measures although some differentiation may occur based on semi-objective and subjective tools. More research is needed to discern related and unrelated symptoms as well as improved baseline screening tools in order to refine populations, document baseline co-morbid diseases and manage treatments.

These data highlighting a heightened disease severity and diminished treatment response in female patients should bring new attention to this often overlooked segment of the AS patient population. This was observed despite women having a later disease onset of shorter duration with an unclear aetiology. Additional research is needed to explore the impact of gender on disease burden and treatment response in AS and examine strategies to improve current AS management in clinical practice.


View Abstract


  • Contributors IvdH-B, DJZ and ASK participated in the study design and data interpretation, development, review, and approval of the manuscript. AS participated in the study design, data analysis, critical review and approval of the final version of the manuscript.

  • Funding The clinical studies were sponsored by Wyeth, which was acquired by Pfizer Inc. in October 2009. The study sponsor was involved in the study design, collection, analysis and interpretation of the data, writing of the report, and in the decision to submit the paper for publication.

  • Competing interests Medical writing support was provided by Stephanie Eide of UBC Scientific Solutions and funded by Pfizer Inc. IvdH-B is an investigator for clinical trials sponsored by Pfizer Inc. DZ is a former employee of Amgen. AS is an employee of Inventiv Health Inc who is a paid contractor to Pfizer Inc. in the development of this manuscript. AK is an employee of Pfizer Inc.

  • Ethics approval Multi-regional.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.