Introduction

Treatment of obesity-related comorbidities is a major factor in the treatment of obesity, such that in the UK, National Institute for Health and Clinical Excellence (NICE) guidelines suggest a BMI of ≥ 35 kg/m2 threshold for patients with at least one comorbidity for access to bariatric surgery [1].

Obesity is associated with both increased severity of asthma and more frequent exacerbations [2,3,4], while higher BMI has been linked to greater prevalence of asthma with a clear dose-response relationship [4, 5]. Indeed, the Global Initiative for Asthma (GINA) management strategy cites obesity as a modifiable risk factor in asthma management [6].

The pathophysiology of asthma is complex, and several phenotypes have been identified by various authors [7,8,9,10,11]; the ‘obesity phenotype’ characterised by late-onset, absence of atopy and moderate airway hyperresponsiveness. Bariatric surgery has been demonstrated to result in greater weight loss, more physical activity, and 20% lower mean energy intake compared to nonsurgical control subjects over a 10-year period. Furthermore, 2- and 10-year incidence rates of hypertriglyceridemia and diabetes are more favourable [12].

An analysis of the Michigan Bariatric Surgery Collaborative (MBSC) clinical registry of 13,057 patients [13] found 18.6% of all patients undergoing bariatric surgery to have asthma with the use of at least one medication—a significant disease burden. Reportedoutcomes of bariatric surgery in asthma patients mainly consists of observational studies, some with large numbers; however, no study, including RCTs have directly compared bariatric surgery to nonsurgical weight loss interventions for asthma control, likely due to the logistical and methodological difficulties this would entail.

This systematic review aims to identify and compare the current evidence in support of surgical and nonsurgical weight loss interventions for the purpose of improving asthma outcomes.

Methods

Two reviewers conducted a systematic review according to PRISMA guidelines using the PubMed (MEDLINE) and Google Scholar databases from inception to October 2020. The following search terms were used: ‘weight loss’, ‘bariatric surgery’ and ‘asthma’. All language articles were searched. Review articles reference lists were scanned for further studies to include.

Inclusion criteria included the following: adult age >18 years, patients with obesity and asthma confirmed before intervention and human studies only. All types of bariatric surgery were included such as laparoscopic adjustable gastric banding (LAGB), sleeve gastrectomy (SG), Roux-en-Y gastric bypass (RYGB) and biliopancreatic diversion duodenal switch (BPD-DS). All nonsurgical weight loss programmes were evaluated including hypocaloric diet regimes with or without exercise programmes. Case series and cohort studies were included where data for an asthma subgroup could be identified.

Population exclusions include mice models, obstructive sleep apnoea and COPD. Comparison exclusions were outcome in patients without asthma or obesity. Outcome-based exclusion criteria were use of CPAP or resolution of OSA (Fig. 1).

Fig. 1
figure 1

PRISMA flow chart to demonstrate identification and inclusion of studies used in this systematic review

Full size image

Results

Thirty-nine original studies were identified between 1993 and 2019, including a total of 5185 patients. Major outcome measures are summarised in Table 1.

Table 1 Major outcome measures described in RCTs, cohort studies and case series
Full size table

Studies of Bariatric Surgery

Twenty-six studies including 4359 patients were identified (Table 5). No RCTs were found. Follow-up ranged from 3 months to 5 years.

Weight loss when reported was significant in all bariatric surgery studies (Table 2) with Baltieri et al. [18] demonstrating the greatest average weight loss of 40.52 kg or 40.5% total weight loss (%TWL), p < 0.0001.

Table 2 Summary of changes in weight or BMI
Full size table

Asthma Medication Use

Eighteen studies [9, 13, 16, 17, 20, 21, 25,26,27,28,29,30, 32,33,34,35,36,37,38] demonstrated cessation or reduction in at least one medication. Seven studies [21, 25, 27, 30, 32, 35, 38] reported cessation of all asthma medication in more than 40% of their respective cohorts.

Guerron et al., the largest study (n = 751), reported a reduction in mean number of medications from 1.4 ± 0.6 to 0.8 ± 0.8 at 1 year postoperatively (p < 0.0001). Sikka (n = 40) [29] found mean respiratory prescriptions reduced from 7.0 to 3.8 at one postoperative year (p = 0.002), while Reddy [13] (n = 257) found 101 patients became asthma medication free following bariatric surgery (p < 0.0001).

Reddy et al. [13] reported significantly lower inhaled bronchodilator and ICS use (p < 0.0001); Dixon et al. [9] demonstrated a significant reduction in SABA usage (p = 0.001). Lombardi et al. [28] demonstrated significant reduction in ICS use (p < 0.03), while van Huisstede et al. [20] reported a small reduction in ICS use in their treatment group (BS+A).

Symptom scores and Quality of Life

Ten studies [9, 15, 17,18,19,20, 26, 27, 31, 34] considered symptoms, with all but Forno et al. [17], describing significant improvement. Major scoring systems including asthma control questionnaire (ACQ) [50], asthma control test (ACT) [51], AQLQ [52, 53] and mini-AQLQ [54] (Table 3).

Table 3 Summary of scoring systems used for asthma symptoms and quality of life
Full size table

Asthma Control Questionnaire

Boulet et al. [27] and Dixon et al. [9] reported statistically and clinically significant ACQ improvement after surgery (p = 0.03, p < 0.001). van Huisstede et al. [20] demonstrated improvement from baseline (p < 0.05) not only in the intervention group (BS+A) but also in the control group (NBS+A) (p < 0.05).

Asthma Control Test

Baltieri et al. [18] described improvement from 18 to 25 (p < 0.0001) and Maniscalco et al. [31] from 18.7 to 22.2 (p < 0.001), with the control group unchanged. Maniscalco et al. [19] reported significant improvement at 1 year after LAGB (p < 0.0001), persisting at 5 years. Forno et al. [17], in a large post hoc analysis, found only modest changes from 21.0 at baseline to 20.8 at 60 months; however, it is worth noting that baseline ACT was already considered ‘adequate’ in this study.

Other Symptom Scores

Santos et al. [15] reported CARAT [56] score improved by 3.9 (p = 0.017) in upper airways and 4.2 points (p = 0.027) in lower airways, and GINA [6] treatment step decreased by 1.8 (p = 0.017). Asthma severity score (ASS) was reduced from 44.5 to 14.3 (p < 0.001) in one study [34].

Quality of Life Scores

Both bariatric studies describing AQLQ showed significant improvement [9, 20]. Improvement in mini-AQLQ [19] (p = <0.001) and ALQ [15] (p = 0.017) was also reported.

Lung Function

FEV1 and FVC increased in almost all bariatric studies which considered lung function [9, 15, 19, 20, 23, 27, 31], with the exception of Baltieri et al. [18], while Al-Alwan et al. [24] reported significantly increased FEV1 but not FVC. Hewitt et al. reported significant improvement overall in FEV1 and FVC, though this was not associated with asthma status [25].

FEV1/FVC ratio, a marker of airway obstruction, was not found to be changed significantly [9, 18, 20, 23], except by Maniscalco et al. (2017) [19] (p ≤ 0.001) and only after 5 years follow-up. Chapman et al. and Baltieri et al. found no change in PEFR [18, 23].

van Huisstede et al. and Boulet et al. reported improvement in FRC (p = 0.02 and p < 0.001) [20, 27] while van Huisstede et al. and Santos et al. demonstrated improvements in TLC (p = 0.018, p = 0.036) [15, 20]. Boulet et al. [27] also showed vital capacity VC (p < 0.001), and ERV (p = 0.006) to significantly increase, with no change in the control group.

Airway Hyperresponsiveness

Airway hyperresponsiveness (AHR) is a pathological hallmark of asthma; it is measured using a stimulus, usually methacholine which acts on muscarinic (M3) receptors [57], known as a methacholine challenge test (MCT). AHR recorded by, for example, PD20 relates to the dose required to achieve a ≥20% fall in baseline FEV1. Six studies reported this outcome [9, 20, 23, 24, 27, 28].

van Huisstede et al. [20], Al-Alwan et al. [24], Boulet et al. [27] and Dixon et al. [9] all demonstrated significant improvement in PD20 or PC20 following bariatric surgery (p = 0.001, p < 0.001, p < 0.001 and p = 0.03). Dixon et al. reported significant PC20 increase in those with normal IgE levels (p = 0.001) but not those with elevated IgE levels (p = 0.89), with similar results by Chapman et al. [23] Conversely, van Huisstede et al. reported PD20 increase only in their IgE-high subgroup (p = 0.003). Boulet et al. found improvement of PC20 to be independent of atopic status.

Markers or Mediators of Inflammation

Markers or mediators of inflammation are reported by seven bariatric studies (Table 4). Leptin was significantly decreased [18, 20] or unchanged [9], while adiponectin was significantly increased [9, 18, 20]. IL-6 was significantly increased and IL-8 decreased in two studies [9, 18], though van Huisstede et al. [20] found no significant change in IL-6 or IL-8. CRP [18] and hs-CRP [20, 27] were significantly reduced; IgE [9, 20] was unchanged and exhaled nitric oxide (FeNO) was either unchanged [20, 31] or decreased [28]. The effect on TNF-α was variable [9, 18, 20].

Table 4 Summary of changes in markers or mediators of inflammation in Bariatric surgery and nonsurgical weight loss studies
Full size table

Healthcare Encounters/Exacerbations

Hasegawa et al. [22] (n = 2261) demonstrated ED attendance or hospitalisation decreased significantly from 22.0 to 10.9% at 12 months (OR 0.42, 95% CI 0.35–0.50, p < 0.001). Other studies [34, 36] reported asthma-related hospitalisation in 27.2% and 21.2% of patients 12 months preceding surgery, and none 12 months postoperatively (p values not given). Lombardi et al. (n = 14) [28] stated no patients experienced asthma-related hospitalisations at 12 months postoperatively (Table 5).

Table 5 Summary of adult studies on effect of bariatric surgery in asthma * = Primary outcome measure (if given)
Full size table

Macgregor et al. [38] found 12.5% (n = 5) reported no asthma attacks, 18.5% (n = 7) fewer attacks and 12.5% (n = 5) only seasonal/allergy-related attacks postoperatively with significant correlation between weight loss and improvement in asthma symptoms (p = 0.0093).

Studies of Nonsurgical Weight Loss

Thirteen studies [14, 39,40,41,42,43,44,45,46,47,48,49, 58] including eight RCTs [14, 39,40,41,42,43,44,45] describe nonsurgical weight loss interventions in 826 patients, with follow up time ranging from 40 days to 24 months (Table 6).

Table 6 Summary of adult studies on effect of non-surgical intervention in asthma
Full size table

Of the RCTs, only Stenius-Aarniala et al. [45] and Hernandez-Romero et al. [44] demonstrated a greater than 10% total body weight loss, while Ma et al. [40] demonstrated the most modest weight loss of 4.0%. Of other studies, Pakhale et al. [47] and Hakala et al. [49] reported a greater than 10% weight loss (p < 0.001) (Table 2).

Asthma Medication Use

Stenius-Aarniala et al. [45] reported a decrease in rescue medication of 0.5 doses in their treatment group (p = 0.002) (vs. 0 in the control group), fewer exacerbations (p = 0.001), but no significant decrease in oral steroid courses (p = 0.07).

Hernandez-Romero et al. [44] demonstrated use of rescue medication to be 20–30% less for salbutamol, theophyllin and ICS in diet A (1200–1500 kcal/ day of powdered feed) but was unchanged in diet B (1200–1500 kcal/day personalised meal regime) except for a 10% reduction in salbutamol, despite both diet regimes having the same calorie content.

Ma et al. [40] found no statistically significant difference in medication, including the number of asthma exacerbations requiring systemic corticosteroids.

Turner et al. [46] (n = 48) demonstrated 23 patients (48%) requiring reduced dosages (no p value given); Pakhale et al. [47] and Hakala et al. [49] did not demonstrate any significant change in medication use.

Symptom Scores and Quality of Life

Asthma Control Questionnaire

RCTs demonstrated some conflicting results: Scott et al. [43] found statistically and clinically significant (>0.5 points) improvement in ‘dietary‘ and ‘combined’ treatment arms (p < 0.001 and p ≤ 0.05), but not in the ‘exercise’ arm, while Freitas et al. [39] recorded improvement in combined weight loss and exercise (WL+E) (p < 0.001) but no change in the weight loss-only (WL+S) group.

Dias-Junior et al. [42] showed within- and between-group improvement (p < 0.001); however, Ma et al. [40] reported no changes in ACQ within or between groups (p = 0.92) but describe clinically significant improvement in ACQ (>0.5 points) with a weight loss of 5–10% or ≥10%. Other studies by Pakhale et al. [47] and Johnson et al. [48] describe significant improvement (p < 0.001, p < 0.0015).

Asthma Control Test

Dias-Junior et al. [42] and Özbey et al. [14] reported a mean improvement of 5.17 points (p < 0.001) and 2 points (p < 0.001) respectively with significant between-group improvement in ACT. Ma et al. [40] found no significant change.

St George’s Respiratory Questionnaire [59]

Stenius-Aarniala et al. [45] and Dias-Junior et al. [42] found significant between-group improvements in total St George’s Respiratory Questionnaire (SGRQ) score (p = 0.02, p = 0.011)). Dias-Junior et al. [42] found significant improvement within the intervention group (p < 0.001) in all domains, and between-groups improvement for activity (p = 0.005) and impact (p = 0.028) domains.

Visual Analogue Scale and ASUI

Using Visual Analogue Scale (VAS), Stenius-Aarniala et al. [45] demonstrated improvement in dyspnoea (p = 0.03) but not cough (p = 0.67) compared with controls. Hakala et al. [49] found a similar improvement in ‘dyspnoea’ (p < 0.001) but not ‘cough’. Johnson et al. [48] reported asthma symptom utility index (ASUI) [60] improved by 0.25 (p < 0.002).

Quality of Life

Pakhale et al. [47] and Özbey et al. [14] showed that AQLQ improved overall (p = 0.003, p < 0.001) with no change in their controls. Scott et al. [43] reported significant within-group improvement in dietary (p < 0.01) exercise (p ≤ 0.05) and combined (p < 0.01) groups.

Freitas et al. [39] demonstrated significant AQLQ improvement in all domains from baseline for the WL+E group, but only in the ‘environmental stimuli’ domain for the weight loss only (WL+S) group; between-group improvement was only found in ‘activity limitation’ (p = 0.007).

Johnson et al. [48] demonstrated an improvement in mini-AQLQ [54] (p < 0.004), while Ma et al. [40] found no change.

Lung Function

FEV1 and FVC showed significant between-group increase in RCTs by Özbey et al., Stenius-Aarniala et al. [14, 45], Freitas et al. (which includes an exercise intervention) [39], and two other studies [47, 49]. Another RCT, Dias-Junior et al. [42], reported FVC increased compared to a control group (p = 0.006) but found no change in FEV1. RCTs by Scott et al. [43] and Ma et al. [55] showed no significant improvement in FEV1 and FVC, while Johnson et al. [48] found no significant change in FEV1.

Improvement in FEV1/FVC was only reported by Özbey et al. [14]; all other studies showed no improvement [42, 47, 49, 55].

Scott et al. [43] showed no increase in FRC between groups; however, TLC increased in the exercise and combined intervention groups compared to dietary intervention (p = 0.037); they also found >10% weight loss resulted in significantly increased FRC (p = 0.018) and ERV (p = 0.008). Within their dietary group, ERV improved significantly (p < 0.05). Freitas et al. [39] found between group improvement in WL+E compared to WL+S (p = 0.038) for ERV; however, TLC remained unchanged while Dias Junior showed no change in ERV.

Airway Hyperresponsiveness

Decrease in AHR was mostly not significant within the nonsurgical studies, including studies which demonstrated significant symptom score improvement [42].

Scott et al. [43] found AHR decreased significantly only within the combined exercise and dietary intervention group, from 100 to 66.7% (p = 0.027). Dias-Junior et al. [42] demonstrated a nonsignificant increase in PD20 of 1.20 mg in their treatment group, while Pakhale et al. [47] found PC20 improvement in their intervention group approaching significance (p = 0.051). Hakala et al. [49] and Aaron et al. [58] found no significant improvement in AHR.

Markers or Mediators of Inflammation

These are reported by five studies (Table 4). Leptin was either decreased [39, 43] or unchanged [42, 48], while adiponectin was either increased [39] or unchanged [43]. IL-6 was significantly decreased [39, 43, 44], as was IL-8 [44] and TNF-α [39, 44, 48]. CRP was unchanged [39, 42, 43, 48], while IgE was unchanged [39, 42] or decreased [44]. FeNO was unchanged [42] or decreased [39].

Healthcare Encounters/Exacerbations

Only Dias-Junior et al. [42] report a statistically significant reduction in ED visits (p = 0.0095) in the treatment group. Ma et al. [40] (n = 330) found no significant change in asthma-related hospitalisations or ED visits (p = 0.40, p = 0.26, respectively).

Stenius-Aarniala et al. [45] report a statistically significant reduction in exacerbations in their treatment group, median 1 (range 0–4) compared to 4 (range 0–7) in the controls (p = 0.001).

Discussion

This review demonstrates significant improvement in asthma outcomes with weight loss, while bariatric surgery seems to offer more consistent clinical improvement in medication use, symptom scores, exacerbations and hospital attendance and AHR compared to nonsurgical programmes, as well as greater total body weight loss (22–36% vs 4.1–14.2%).

Clinical Implications

Arguably the most important clinical finding is the reduction in asthma medication use following bariatric surgery. Guerron et al. [16] found that preoperative BMI has a significant effect on asthma medication use and showed significant improvement with surgery, as did Reddy et al. [13] (n = 257). Bariatric surgery also dramatically reduced the risk of hospital admission or ED attendance for asthma by between 50% [22] and 100% within 12 months [28, 34, 36]. This may reduce the economic burden of asthma both on health systems and individuals. In clinical practice, it may be worthwhile to include respiratory or primary care clinicians in the perioperative process, analogous to the way diabetes specialists are currently involved, on the understanding that asthma medication doses will be titrated down or stopped altogether.

That symptom scores and quality of life were significantly improved in all but one bariatric study is important; these outcomes can be discussed preoperatively with patients as a possible benefit of surgery, especially for those whose main comorbidity is asthma. Routine preoperative symptom scoring for these patients may be worthwhile. Nonsurgical studies showed some improvement with weight loss, although not uniformly and many included an exercise component which may have improved symptom scores and respiratory function independently of weight loss.

A target weight loss may be relevant, with larger weight losses being achievable through surgery. Indeed, the largest RCT of nonsurgical weight loss [40] described clinically significant improvement in ACQ scores only with weight loss >5% [40], while studies reporting >10% weight loss [45, 47, 49] showed improvement in symptom scores. Interestingly, Forno et al. describe patients who underwent RYGB had significantly greater improvement on ACT scores than those who underwent other types of bariatric surgery (p = 0.002); this may influence the type of bariatric surgery offered to patients after multidisciplinary team (MDT) discussion.

Lung Function and Physiology

Obesity significantly reduces ERV, and FRC, the resting lung volume, due to lung compression which leads to narrowing of airways and alveolar de-recruitment, while FEV1 and FVC are also slightly reduced; FEV1/FVC ratio and VC normally remains the same [61]. This review has largely shown these changes in FEV1, FVC and ERV to be reversed with weight loss, although more consistently with bariatric surgery. Improvement in these mechanical effects on lung function has also been demonstrated in nonasthmatics [20].

That weight loss has been demonstrated to improve asthma outcomes can be explained further in two main ways. Firstly, in the ‘obesity phenotype’ of asthma [7,8,9,10,11, 62,63,64], individuals experience abnormal collapse and increased sensitivity to closure of peripheral airways at a higher body mass; thought to be related to decreased airway wall thickness [65] and increased elastance of the peripheral airway [24, 66]. In this phenotype, characterised by late-onset, reduced IgE and CD4+ T helper 2 cytokines (TH2-low), weight loss has a greater effect than in early-onset atopic asthma, characterised by elevated IgE and CD4+ T helper 2 cytokines (TH2-high).

Dixon et al. found that AHR improved significantly in ‘low IgE’ patients (p = 0.001) but not those with normal IgE levels (p = 0.89) [9], with similar results found by Chapman et al. They postulate that obesity reduces tethering forces of the lung parenchyma on the small airways and predisposes them to closure [23].

Secondly, increased adiposity causes increased airway inflammation due to low-grade systemic inflammation from adipose tissue breakdown resulting in macrophage activation and generation of proinflammatory cytokines, which impact the lung [67]. Inflammatory mediators such as CRP [68, 69], IL-6 [70] and tumour necrosis factor alpha (TNF-α) [71] have been demonstrated in higher levels in obese patients. Furthermore, adipose tissue gives rise to adipokines which are implicated in the inflammatory response: leptin being proinflammatory and increased in obesity, and adiponectin anti-inflammatory and decreased in obesity [72]. Mice models have shown increased AHR when exogenous leptin is administered [73], while exogenous adiponectin reduced AHR and airway inflammation [74]. Airway epithelial cells express both leptin and adiponectin receptors [75,76,77]; it has been suggested that these offer an alternative route of pathogenesis of airway reactivity separate from systemic inflammation, by promoting alteration of alveolar macrophage function and airway remodelling in late-onset, TH2-low, asthma [78].

Significant reductions in leptin [18, 20, 39, 43] and increases in adiponectin [18, 20, 39] in both bariatric and some nonsurgical interventions were demonstrated while CRP was significantly reduced in mostly bariatric studies [18, 20, 27, 48]. Other inflammatory markers showed variable responses which may suggest their lack of specificity to asthma and its various phenotypes [11].

Other routes of pathogenesis of airway reactivity exist which have not been described in studies in this review, such as the advanced glycation end products (AGE) pathway mediated by arginine dysregulation in obesity and metabolic syndrome and may be modulated by incretins such as glucagon-like peptide 1 (GLP-1) [79]. These represent areas of further study in obesity-related asthma.

Quality of Studies

The quality of the evidence available for bariatric surgery includes cohort studies and case series. There were no RCTs; therefore, all studies in this group may have confounding factors which are not accounted for. For example, those with severely uncontrolled asthma may not pass anaesthetic assessment for surgery or may undergo pre-optimisation. Many bariatric series included subgroups of asthmatic patients from which outcome data was extracted, though these were usually not primary outcome measures. Nonsurgical studies were fewer in number, but higher quality evidence; most being RCTs with clear selection criteria and outcome measures.

Limitations of This Review

There is much heterogeneity between studies. They differ in populations, specific outcome measures, reporting (e.g. between-group and within-group/from baseline changes) and units used (e.g. litres vs. %pred); therefore, only a narrative review, rather than a quantitative synthesis, was possible.

Conclusion and Future Direction

The relationship between asthma and obesity is complex and likely mediated by a combination of susceptibility to weight-related peripheral airways collapse, systemic inflammation and adipokine imbalance resulting in increased airway hyperresponsiveness. Notwithstanding the limitations of the review and the quality of the studies, it appears that bariatric surgery is more effective in treating asthma. We do, however, require good-quality studies which focus on the effect of bariatric surgery on mild and severe asthma. Should patients with BMI >35 kg/m2 with mild asthma as their only comorbidity be offered bariatric surgery? Similarly, should patients with severe life-threatening asthma be considered for bariatric surgery in spite of anaesthetic risks? These questions may need to be answered with future research.