Cigarette smoking associates with body weight and muscle mass of patients with rheumatoid arthritis: a cross-sectional, observational study

Rheumatoid arthritis (RA), the commonest inflammatory arthritis, is associated with altered metabolism [1]. Compared with healthy controls, RA patients exhibit elevated resting energy expenditure (REE) and enhanced muscle catabolism [2]. Such changes may lead to rheumatoid cachexia (that is, involuntary loss of fat-free mass [FFM] with a proportional increase of body fat [BF]) in the presence of stable body weight [3, 4]. Body composition changes, particularly BF increase, may remain largely undetected by traditional assessments such as the body mass index (BMI) [5]. Increased BF, together with reduced levels of physical activity due to joint inflammation and damage [3, 6], is associated with several comorbidities, including cardiovascular disease [7, 8] as well as increased mortality [3].

Cigarette smoking is an important risk factor for several diseases [9]. It is also known to decrease body weight in healthy individuals by reducing appetite and increasing REE [10]. In contrast, smoking cessation may associate with significant weight increase, which constitutes a major deterrent to smoking control [11].

We have recently demonstrated that smoking further increases REE in RA [12] and this could potentially augment rheumatoid cachexia in these patients. Given the RA-related alterations in body composition and the comorbidity associated with them, the examination of potential contributors to muscle wasting, such as smoking, is important. The aim of this cross-sectional study was to detect potential associations between smoking and body weight, body composition, and rheumatoid cachexia in RA patients.

Table 1 illustrates means ± standard deviations and the ANOVA results for all studied parameters. Current smokers had a significantly lower BMI than ex-smokers (mean difference: male -2.6, 95% CI: -3.5 to -1.7; female: -2.6, 95% CI: -4.8 to -0.5) and never-smokers (mean difference: male -1.8, 95% CI: -3 to -0.6; female: -1.4, 95% CI: -2.4 to -0.4). Current smokers also had a significantly lower BF compared with ex-smokers (mean difference: male: -4.3, 95% CI: -7.5 to -1.2; female: -3.4, 95% CI: -6.4 to -0.4) and never-smokers (mean difference: male: -3.3, 95% CI: -6.3 to -0.4; female: -2.1, 95% CI: -4 to -0.2). FFM did not differ between these groups (mean difference: current smokers versus ex-smokers, male: -4.6, 95% CI: -10.7 to 1.6; female: -1.2; 95% CI: -3.8 to 1.4; current smokers versus never-smokers, male: -2.7, 95% CI: -9.2 to 3.9; female: 0.1, 95% CI: -2.4 to 2.4). Current smokers had a significantly smaller waist circumference than ex-smokers (mean difference: male: -6.2, 95% CI: -10.4 to -1.9; female: -7.8, 95% CI: -13.5 to -2.1) but not never-smokers (mean difference: male: -2.9, 95% CI: -10.6 to 4.9; female: -3.9, 95% CI: -9.2 to 1.5). Also, ex-smokers had a larger waist circumference than never-smokers but the difference was significant for males only (mean difference: male: 3.3, 95% CI: 0.4 to 6.3; female: 3.9, 95% CI: -0.4 to 8.1).

In ANCOVA with gender and smoking as factors and age, DAS28, HAQ score, and disease duration as covariates, smoking was a significant and independent predictor for BMI (F_2,387 = 8; P < 0.001), BF (F_2,387 = 4.4; P < 0.05), and waist circumference (F_2,387 = 7.9; P < 0.001). Smoking also emerged as a significant predictor of FFM (F_2,387 = 5.1; P < 0.05), but inclusion of BMI as a covariate eliminated the effect of smoking on FFM (P > 0.05).

There was a significant negative correlation between pack-years and BF (r = -0.46; P < 0.001) in the current smoker and the ex-smoker groups. This remained significant after adjustment for gender, age, DAS28, HAQ score, and disease duration (F_1,389 = 4.8; P < 0.05). Following pack-year grouping into quartiles (pack-group), ANOVA did not reveal any differences for BMI or body composition among the current and ex-smoker pack-groups. However, an ANCOVA model with gender and pack-group as factors and age and weight as covariates (following stepwise elimination of ESR, CRP, DAS28, HAQ score, and disease duration) revealed a significant effect of pack-group on FFM (F_3,217 = 2.7; P < 0.05), with heavy smokers exhibiting the lowest values. Mean (95% CI) values of this variable in the pack-year subgroups appear in Figure 1.

Following BMI and BF grouping, chi-square analyses showed significant differences (P < 0.05) in the prevalence of overweight and obesity among smoking groups, with obesity being more prevalent in ex-smokers (50%) followed by never-smokers (39%) and current smokers (30%). Similarly, ex-smokers had a significantly (P < 0.05) higher prevalence of increased waist circumference (69%) compared with never-smokers (60%) and current smokers (49%). However, FFM did not differ between groups (P > 0.05) (Figure 2).

To our knowledge, this is the first study to identify significant associations between smoking, body weight, and body composition of RA patients: current smokers had a significantly lower BMI and BF compared with never-smokers. Both BMI and BF were significantly increased in ex-smokers, whereas very heavy smoking appeared to associate with reduced FFM. The study has several potential limitations. These are all cross-sectional associations, and although they can serve for hypothesis generation, they do not provide definitive evidence for causality or directionality: longitudinal studies are required for this. In addition, body composition was assessed by bioelectrical impedance. This method has been validated [19‐23] and is thought to be suitable for body composition studies in diverse populations [22‐25], correlates well with the 'gold standards' of dual-energy x-ray absorptiometry and hydrostatic weighing [23], and is widely used in RA research [5, 12, 24, 26, 27], but it has not actually been specifically validated in the RA population. Finally, although self-report of smoking, especially smoking history, is generally reliable, both under- and over-reporting can occur [28]. This is unlikely to have influenced the primary findings of this study (that is, the differences between current, ex-, and non-smokers), while any misreporting in pack-years may have been smoothed by the large number of participants. It is difficult to assess any other selection bias: the prevalence of current, ex-, and non-smokers among the participants of this study was similar to that reported for local general population subjects of similar age [29], although it was different from an RA cohort established more than 10 years ago [30].

Our observations for BMI are consistent with those in the general population. Both male and female smokers tend to have decreased BMI compared with their non-smoking counterparts [10, 31]. In contrast, significant BMI increases have been noted after smoking cessation [11]. Smokers have increased levels of leptin [32], which regulates food intake and fat deposition [33], and reduced hypothalamic neuropeptide Y [34], which regulates appetite [35]. Smoking-induced increases in the levels of epinephrine, norepinephrine, and thyroid hormones lead to increased energy expenditure at rest [36, 37] and during light physical activity [38‐40]. However, these effects are short-lived: after smoking cessation, leptin decreases to levels below those expected for non-smokers of similar weight [32] and resting energy expenditure (REE) returns to normal [41].

In patients with RA, smoking has been shown to elevate REE [12]; however, no data are available on other potential contributors to smoking-related weight loss or smoking cessation-related weight gain for this population. Although we did not assess energy intake and expenditure or related regulators (such as leptin), it is likely that the mechanisms behind the reduced body weight of current smokers and the increased body weight of ex-smokers with RA are similar to those described for the general population.

Interestingly, the lower BMI of current smokers in the present study seems to be due to decreased BF rather than FFM. A possible mechanism by which smoking may affect fat metabolism is through a reduction in neuropeptide Y. This molecule not only stimulates food intake, but also promotes white fat lipid storage and decreases brown fat thermogenesis [35], so its inhibition through smoking would be expected to have the opposite effects. Additionally, smoking results in decreased adipose tissue lipoprotein lipase activity [42], which diverts fat storage away from adipose tissue and toward utilization by muscle [43], possibly leading to the decreased BF of smokers [42, 44]. In the present study, the inverse association between smoking and BF appeared to be dose-dependent: increasing pack-years associated with reducing BF levels. Smoking cessation is thought to result in a reversal of the mechanisms described above, leading to increases in BF [42] and, most importantly, abdominal fat [45]. Indeed, among these RA patients, ex-smokers seemed to be the most 'unhealthy' group in terms of body weight and composition as they exhibited the highest BMI, BF, and waist circumference values.

In predominantly healthy people who are from the general population and who do not have wasting muscle disease, smoking of any intensity has been implicated in muscle wasting [10] by impairing the process of muscle protein synthesis [46]. In contrast, in the present study, only very heavy smoking appeared to associate with a reduction in FFM. It is possible that the effect of smoking on muscle is of less significance than the muscle loss associated with RA itself, as part of rheumatoid cachexia. This hypothesis is supported by the finding that increased duration of smoking (that is, pack-years) associated with lower FFM in both current and ex-smokers, which suggests the existence of a threshold below which smoking does not induce further muscle loss in RA patients. A longitudinal study of the impact of smoking intensity (and cessation) on the body composition of patients with RA may throw more light on the mechanistic basis of these observations.

Overall, this study suggests that, in RA, smoking associates with reduced body mass and fatness without inducing further muscle loss, except in very heavy smokers; in contrast, smoking cessation associates with increased body mass and fatness. This should not be interpreted as favouring what is a very unhealthy habit. Smoking cessation, even if it occurs in mid-life, reduces most of the later risk of death from tobacco [47]. However, smoking cessation is known to result in body weight increase, and this may affect some people's decision to stop smoking [11, 44, 45]. Therefore, any smoking cessation regime should be underpinned by more generalised lifestyle counselling, including advice on exercise and weight management. This is emphasized by the fact that, based on recently described RA-specific BMI [5], BF [16], and waist circumference thresholds [48], ex-smokers have the highest prevalence of obesity – both total and abdominal. FFM did not differ between groups and the prevalence of low FFM was comparable to that expected in age- and gender-matched healthy individuals [18].

Within the limitations of this study, it is concluded that RA smokers have a lower BMI and BF than RA non-smokers, while heavy smokers also have a reduced FFM. A history of smoking cessation appears to associate with increases in BMI, BF, and waist circumference. Nevertheless, given the numerous adverse effects of smoking on health, smokers with RA should be actively advised against it, but smoking cessation programs should include wider lifestyle counselling for weight control, also focusing on increased physical activity and a healthy diet.