Introduction

Rheumatoid arthritis (RA), with a prevalence of 0.5 to 1.0% [1], affects women three times more frequently than men [2,3]. While the reason for this unbalanced frequency between women and men is not fully understood, it is known as the result from multiple factors, including both biologically-driven sex concepts and socially constructed gender factors. Sex-based factors include observed differences in the immune response and hormonal levels between sexes [4–7]. Comparing adult men and women, antibody production and cell-mediated immunity are generally enhanced in women [7,8]. Particularly, T cell activation and proliferation, CD4/CD8 ratio, B cells, and immunoglobulins are increased in women, while men present higher levels of CD8+ T cells and regulatory T (Treg) cells [7]. Additionally, levels of estrogen (which fluctuate during the menstrual cycle) have a dose-, density-, and distribute-dependent role in the immune-response [7], with low levels enhancing it, and high levels (characteristic of pregnancy stage) having immune inhibitory effect [8]. Conversely, progesterone and testosterone have anti-inflammatory effects [7,8]. Furthermore, besides the sex-based factors, socially constructed gender bias and stereotypes may also play a role in the development of RA by influencing environmental or behavioral factors [4]. For example, smoking, associated to increased risk of incidence of seropositive-RA [9], is more common among men than women in Switzerland [10]. And while women may seek health consultation earlier, they tend to be prescribed lower dose [11] or less appropriate pain medication than men [12–15]. Likewise, a later referral from first physician encounter to the arthritis clinic was reported for women versus men [16].

In addition to the differences in RA prevalence in men and women, previous studies have found differences in the presentation and management of RA. Among RA patients, an increased disease severity and higher pain levels in women, compared to men, have been reported in several studies [17–22]. Moreover, observational studies showed a stronger clinical response and higher likelihood of achieving remission in men compared to women [23–28]. However, these findings were not always consistent depending on the study design. For example, Jawaheer et al. found that men, compared to women, had higher odds of achieving sustained remission in early RA but not in established RA [24,26]. Likewise, Couderc et al. found inconsistent results on regard to remission rates and response across men and women depending on the follow-up time-point [29]. Additionally, a recent meta-analysis of randomized clinical trials (RCTs) for biologics reported no significant differences in response rates between women and men [30].

Thus, investigating the influence of sex and gender on the course of RA remains of interest. In this study we characterized women and men with RA at the time of starting their first biologic or targeted synthetic disease-modifying anti-rheumatic drug (b/tsDMARD), and aimed to assess the impact of sex/gender in the subsequent clinical response.

Materials and methods

Results

The study included 3722 RA adult patients, of which 2839 (76.3%) were women and 883 (23.7%) were men. The flow chart for inclusion/exclusion criteria is depicted in Fig 1. Patient’s age at first symptoms and diagnosis, and baseline characteristics are presented in Table 1. Compared to women, men were older at first symptoms, diagnosis, and start of first b/tsDMARD. While both strata had similar time from first symptoms to diagnosis, men’s median time from diagnosis to the start of b/tsDMARD was shorter than the time in women (3.4 years versus 5.0 years). At baseline, men were more likely to be overweight (47.2% versus 27.0%) and a had higher frequency of ever smoking (49.2% versus 30.6%), compared to women. Conversely, men had a lower disease activity at baseline (DAS28-ESR 4.2 versus 4.4, p<0.001), and lower (better) median HAQ score (0.9 vs 1.0, p<0.001) and lower median TJC (4.0 vs. 5.0, p = 0.02), compared to women. When assessing patient responses to disease activity and pain level (on a scale from 1–10), we note that men had less reported pain (4.6 vs. 4.9, p = 0.05), and lower self-assessed activity of their rheumatic disease (4.8 vs. 5.1, p = 0.004) compared to women. Additionally, while men had higher frequency of hyperlipidemia, diabetes, and cardiovascular disease or cardiac events, women had significantly higher frequency of history of fibromyalgia, osteoporosis, and depression/anxiety.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of inclusion/exclusion criteria for the study population.

Abbreviations: RA rheumatoid arthritis; b/tsDMARD biologic or targeted synthetic disease-modifying anti-rheumatic drug; SCQM Swiss Clinical Quality Management in Rheumatic Diseases.

https://doi.org/10.1371/journal.pone.0275026.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient characteristics.

https://doi.org/10.1371/journal.pone.0275026.t001

Results from the main logistic regressions are presented in Fig 2. Compared to women, men had higher odds of achieving DAS28-remission within the first 12-months, with unadjusted odds ratio (OR) of 1.21 and 95% confidence interval (CI) 1.02–1.42. Adjusting for age and seropositivity led to similar results, with adjusted OR (ORadj) 1.24 (95% CI 1.05–1.46). Secondary outcomes DAS28-rem/LDA and RADAI-5 remission did not show a significant difference between men and women. Sensitivity analyses with 9- and 15-months follow-up showed similar results as those at 12-months for all three outcomes (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Logistic regression assessing the effect of sex/gender on the study outcomes, with 12-month maximum follow-up.

Abbreviations: n number; OR odds ratio; ORadj odds ratio adjusted for age and seropositivity; DAS28 Disease Activity Score 28; rem/LDA remission or low disease activity; RADAI-5 Rheumatoid Arthritis Disease Activity Index-5.

https://doi.org/10.1371/journal.pone.0275026.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Sensitivity analysis.

Logistic regression assessing the effect of sex/gender on the study outcomes, with maximum follow-up of 9- and 15-months.

https://doi.org/10.1371/journal.pone.0275026.t002

Sensitivity analysis excluding patients without follow-up information on the outcome (S2 Table), showed an even higher likelihood of men versus women on achievement of DAS28-remission within the first year (OR 1.36 [95% CI 1.12–1.64]). Likewise, this resulted in higher odds of men achieving DAS28-rem/LDA (OR 1.23 [95% CI 1.02–1.50]), whereas the findings for RADAI-5 remission remained non-significant.

Results from the mediation analyses are shown in Table 3, and S3 Fig. Among the investigated potential mediators, men were associated with lower odds of baseline DAS28≥3.2 and RA duration >2-years, and higher odds of having BMI≥25 in comparison to women. Subsequently, adding baseline DAS28 (threshold 3.2) to the main analysis removed the significance from the previously observed association between sex/gender and DAS28-remission. Similar observation was made for RA duration (threshold 2-years). Thus, baseline DAS28 and RA duration were identified as factors which may explain the association between sex/gender and DAS28-remisison at ≤12-months. Conversely, adding BMI to the main analysis did not attenuate, nor dismissed, the main findings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary findings from the mediation analyses (steps I and II), and findings from the main analysis for reference.

https://doi.org/10.1371/journal.pone.0275026.t003

Results from the post-hoc analysis assessing changes in individual clinical endpoints and composite disease activity scores are provided in S3 Table. We did not identify any significant differences, apart from CRP, which had substantial missingness.

Discussion

This study on 3722 RA patients new-users of b/tsDMARD had 76% women and 24% men. At baseline, men were older and presented higher frequency of overweight and ever smoking compared to women. Conversely, women had longer disease duration (5.0 versus 3.4 years in men), more active disease (DAS28, RADAI-5), and more frequent history of fibromyalgia, osteoporosis, and depression/anxiety. Men were more likely to achieve DAS28-remission within one year, in comparison to women. Our analyses suggested that this benefit in men may be explained by the differences in baseline disease activity (DAS28) and disease duration at start of first b/tsDMARD. Conversely, no differences were observed between men and women on their likelihood of achieving low disease activity (DAS28-rem/LDA) or RADAI-5 remission.

Results from our analyses showed 21% higher odds of DAS28-remission in men versus women, which was consistent with previous studies. For example, an observational cohort study with 2879 RA patients treated with tumor necrosis factor alpha inhibitor (TNFi) (etanercept/infliximab) in a British registry suggested that women were 40% significantly less likely to achieve DAS28-remission [27]. Similarly, another study in the same registry, seeking to identify predictors for infliximab response at 12-months, reported women to be significantly associated with lower DAS28 improvement compared to men [28]. Conversely, two studies of RA patients treated with rituximab and abatacept, respectively, showed no significant differences between men and women and their likelihood of achieving DAS28-remission [29,39]. However, these last-mentioned studies included in their main analyses adjustment for baseline disease activity and disease duration. While the authors treated these variables as confounders, we identified them as mediators, and adjustment for mediators in their analyses is expected to attenuate or remove the observed effect of the study exposure. Additionally, findings from the rituximab study showed higher remission rates in men compared to women [29], and the un-adjusted results from the abatacept study showed significantly reduced odds of achieving DAS28-remission in women compared to men at 6-, 12-, and 24-months, with an OR 0.56 (95% CI 0.37 to 0.87) in women at 12-months [39].

Similarly, a recent meta-analysis of RCTs of biologics did not find statistically significant difference in the American College of Rheumatology 20 (ACR20) rates between men and women. While both DAS28-remisson and ACR20 are commonly used clinical outcomes in RA, they are different in nature. The ACR20 is a relative score, defined by a 20% improvement in various disease attributes (e.g. SJC28, TJC28, pain, HAQ, CRP) [34]. Thus, ACR20 outcome is taking the baseline disease activity and HAQ into consideration. This, together with the above-mentioned role of the baseline disease activity as mediator in the association between sex/gender and clinical outcome, may explain why no differences were found between men and women when using ACR20 as outcome.

Our mediation analyses suggested baseline disease activity and disease duration as mediators of the effect of sex/gender on the achievement of DAS28-remisison. More concretely, the observed lower odds of having moderate or higher disease activity at baseline (DAS28≥3.2) and lower odds of a disease duration >2-years at start of b/tsDMARD explained the observed association between men and higher odds of achieving DAS28-remissison compared to women.

In our study, men were prescribed b/tsDMARD with a median 1.6 years shorter disease duration than women. This significant temporal gap between men and women escalating treatment was consistent with other observational studies on RA patients treated with biologics [29,39,40].

The study by Arkema et al. in the Swedish national biologics registry ARTIS suggested that, compared to men, women started TNFi therapy at a higher level of patient-reported outcomes (TJC28, patient’s global assessment, pain, HAQ) but at similar level of physician-reported outcomes (SJC28, physician’s global assessment) [40]. Our study reflected a similar trend, with women having worse symptoms or patient-reported outcomes (TJC28, RADAI-5), while similar physician-reported outcomes (SJC28, physician global disease activity), in comparison to men at baseline.

Thus, although overall decision on step-up to b/tsDMARD in the Swiss practice is done based on disease activity, we observed discrepancies on composite disease activity scores (DAS28, RADAI-5) and individual clinical features between the study groups. Hypotheses to explain these phenomena may include differential physician-driven and/or patient-driven aspects on treatment decision-making.

Although patient-reported outcomes, like pain or tenderness, are key features on RA disease management, sex-driven differences (e.g., hormonal level; expression of pain receptors) and socially-constructed gender stereotypes may affect their perception, expression, and interpretation across patients [41]. Studies on sex, gender, and pain reported that, while chronic pain syndromes are more frequent in women, and women generally report more and worse pain than men, there are numerous examples of women’s pain being discounted or underplayed in healthcare [12,13,42–46], and women often receive less or less appropriate pain medication [12–15]. While our findings show similar type of b/tsDMARD and similar frequency of comedication with csDMARD and steroids between men and women, the shorter disease duration and lower disease activity in men at b/tsDMARD start, compared to women, reflects a potential treatment gap between the study groups. Additionally, in the context of pain, we may discuss depression, as a common comorbidity with pain [14] and fibromyalgia. In our study, the female/male ratio for fibromyalgia was four to one (1.9% of women versus 0.5% of men), and frequency of depression/anxiety was 12.4% in women and 7.5% in men. While overall these frequencies are low, this along with the observed higher tender joints in women, suggest that women may have had higher pain at the start of their first b/tsDMARD, in comparison to men. And yet, it may be that a more frequent mention to pain by women may have led physicians to unconsciously “raise the bar” and undervalue the patient’s observations, a move that by itself may “increase the voice” of the patient, all in all turning into a vicious circle from which healthcare professionals should seek to escape.

In addition to the above-discussion on patient-reported outcomes, we may as well discuss the levels of inflammatory markers in both study groups. Conversely to the study by Arkema et al. [40], we observed higher ESR in women versus men at baseline, whereas in both theirs and our study, baseline CRP was higher in men. Both ESR and CRP are acute phase reactants commonly used in the clinic as biomarkers of inflammation. Both increase with age [47], may be differently affected by sex and BMI [48], and have different half-life [49]. It was suggested that ESR is more influenced by sex (elevated in females) and age, while the CRP levels may be affected by smoking, high blood pressure, and high BMI [50]. This could partially explain our higher baseline CRP in the male population, with higher BMI and prevalence of cardiovascular history and smoking. Likewise, it could partially explain the observed higher baseline ESR level in women. While evidenced interpretation of the controversy between the baseline findings on ESR and CRP remain unclear, our CRP results may be taken with caution due to the high missingness (>50% at baseline), and the elevated ESR levels (which agreed with the elevated tender joint counts and RADAI-5 score in women) depicted a more active disease in women versus men at start of treatment.

It remains of interest to investigate the impact of sex on the b/tsDMARD mode of action. RA clinical guidelines do not differentiate treatment regimen based on sex [41], and although further sex-specific pharmacological research remains of interest, this lays out of the scope and capacity of our study design. However, independently of sex-specific mechanisms, according to our study findings, a higher threshold for prescription of b/tsDMARD in women versus men may partially explain the observed gap on treatment. While this could be prescriber-driven, we also note that observations from our practice in Switzerland also describe women patients as more reluctant to b/tsDMARDs and prescription of injectable therapies compared to men.

Ultimately, the study findings suggest that earlier step-up treatment with b/tsDMARD in women could potentially increase their odds of remission and it is consistent with the suggestions that earlier treatment with biologics leads to higher likelihood of successful clinical outcome [51].

Results for our secondary outcome DAS28-rem/LDA showed no differences across men and women. While this could simply be due to the more approachable threshold of DAS28-rem/LDA compared to DAS28-remisison, exclusion of the patients without follow-up information on outcome (sensitivity analysis) tilted the odds towards a favorable result for men. This suggests a differential distribution of patients lacking follow-up information on outcome between sexes. Therefore, the study findings on DAS28-rem/LDA should be taken with caution. On regard to the outcome RADAI-5 remission, men and women had similar odds of achieving RADAI-5 remission in both the main and the sensitivity analysis. RADAI-5 is a mainly patient-driven score, and therefore, it reflects the patient’s perception of their own disease, unlike the DAS28, which mostly contains values measured by the physicians (3-item formula for DAS28, S1 Equations). Thus, these two outcomes measured two different clinical goals. A complete understanding on the RADAI-5 results remains of interest.

Conclusion

The study findings indicated that when starting their first b/tsDMARD, men had a shorter disease duration and lower disease activity (DAS28-ESR, RADAI) in comparison to women. Likewise, compared to women, men presented higher odds of achieving DAS28-remission within the first year. Finally, findings from mediation analyses indicate these discrepancies in the treatment decision making regarding start of first b/tsDMARD in men and women patients as potential explanation for the observed response gap between the sex groups. Thus, these findings suggest that step-up to b/tsDMARD treatment at an earlier disease stage in women, similar to the observed practice in men, may bring women the beneficial effect observed in the men group.

Supporting information

Acknowledgments

We thank all patients and rheumatologists contributing to the SCQM registry, as well as the entire SCQM staff. A list of rheumatology offices and hospitals which contribute to the SCQM registry can be found at http://www.scqm.ch/institutions. A. M. Burden acknowledges that her professorship is partly endowed by the Swiss National Pharmacy Association (PharmaSuisse) and the ETH Foundation.