Abstract
Background Smoking cessation is the only intervention known to affect disease progression in patients with COPD as measured by the rate of change in forced expiratory volume/1s (FEV1) over time. The need for new drugs to modify the progression of COPD is well recognised. We hypothesised that changes on CT in relation to smoking cessation may relate to changes in response to disease-modifying drugs, and therefore as a novel quantitative biomarker of drug efficacy. CT biomarkers of emphysema and airway wall thickness are increasingly used in research, but there has not been a systematic appraisal of the evidence to assess how these biomarkers evolve with a change in smoking exposure in COPD patients.
Methods We searched MEDLINE, Embase, the Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials (CENTRAL)), and Web of Science to 10th September 2019. We included longitudinal studies of smoking COPD patients who had CT scans before and after smoking cessation. Two review authors (DC, SA) independently screened studies, extracted outcome data and assessed the risk of bias, with a third reviewer (JRH) arbitrating conflicts.
Results Four studies were included in the final analysis. Three studies measured CT markers of lung density, which all, perhaps counter-intuitively, showed a significant decrease with smoking cessation. One study measured CT markers of airway wall thickness, which also significantly decreased with smoking cessation.
Authors’ conclusions Smoking cessation in COPD patients causes a fall in lung density, but the magnitude of the effect has not been rigorously assessed. One study has reported a decrease in airway wall thickness with smoking cessation. The number of studies is small, with some risk of bias. This question remains important for COPD researchers and requires further studies, in particular to assess whether changes with smoking cessation may model changes in response to novel pharmaceutical agents, and how to handle change in smoking status in relation to longitudinal observational imaging studies in COPD.
1 Introduction
Chronic obstructive pulmonary disease (COPD) is a condition characterised by airflow limitation and persistent respiratory symptoms. It is the fourth leading cause of death worldwide1 and its prevalence is increasing2. Disease progression has classically been assessed by measures such as the rate of change in forced expiratory volume over 1 second (FEV1). Despite the development of drugs to reduce symptoms and prevent exacerbations in COPD, smoking cessation is the only proven way to slow the progression of disease 3, as measured by rate of decline in FEV1, and remains the only disease-modifying intervention we can offer patients. In part, this likely reflects the heterogeneity in rate of FEV1 decline between patients with COPD. There is currently much interest in developing disease-modifying interventions in early COPD and one limitation is a current lack of quantitative biomarkers of disease progression.
Since its introduction more than four decades ago, computed tomography (CT) is now widespread in clinical practice and research. It has unique strengths in terms of image resolution and speed of acquisition which mean that it has utility in a variety of pulmonary conditions. In COPD, quantitative regional assessment of emphysema is already used to guide use of endobronchial valves, and has been used as a biomarker of response to alpha-1 antitrypsin augmentation, whilst quantitative assessments of airway wall geometry and functional small airway dysfunction have proved useful phenotyping tools in COPD research.
These CT biomarkers hold further promise as biomarkers of treatment response in COPD. We hypothesised that quantitative imaging biomarkers that change in relation to smoking cessation – the archetypal disease modifying intervention in COPD – may also be useful as markers to treatment response in relation to novel drugs.
We have therefore conducted a systematic review of studies examining the CT changes which occur as a result of smoking cessation in COPD patients. We aimed to synthesise the range of quantitative biomarkers measured, and the direction and size of effect with intervention. We provide a robust quality assessment of our included studies. Our results will be of interest to those developing imaging biomarkers of treatment response in COPD.
Our review question was: how does CT imaging in patients with COPD change when they stop smoking compared to if they continue smoking? Our outcome measures are quantitative changes in CT scan appearances.
2 Methods
2.1 Protocol and registration
A review protocol was published on the 4th September 2019 on PROSPERO [ref]. The protocol was structured according to the Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement 8. The protocol registration number is CRD42019144555.
2.2 Criteria for review
2.2.1 Types of study
We included original research studies which report the outcome (quantitative CT scan appearances) in our population (COPD patients) and which meet our definition of change in exposure (successful smoking cessation).
We included longitudinal studies, with no other restrictions on study design. Studies could be case controlled, cohort, or observational, prospective or retrospective. Since randomisation to continued smoking would be unethical, randomisation was not an inclusion criterion. Studies in any language were included.
We excluded studies which recorded outcome measures during acute hospitalization or during COPD exacerbations. We excluded review articles. We excluded longitudinal studies which reported outcome measures only before or only after intervention.
2.2.2 Types of patient
Our target patients were adults with a diagnosis of COPD with a significant smoking history. We did not make restrictions as to the diagnostic criteria the study authors chose for COPD diagnosis. We did not make restrictions as to the definition of ‘significant smoking’.
2.2.3 Types of intervention
Intervention was successful smoking cessation between two outcome time-points (CT scan). We did not place any restriction on the duration of smoking cessation. If a control group was used in a study, this would be taken as patients who continued to smoke until the second time point. A lack of a control group was not an exclusion criterion since some studies may elect to use extrapolation from longitudinal data as a comparison.
2.2.4 Types of outcome and prioritisation
The main outcomes were: (1) change in quantitative CT biomarkers of emphysema; and (2) change in quantitative CT biomarkers of airway wall thickness.
We also report qualitative change in CT appearances.
2.3 Search methods for identification of studies
2.3.1 Information sources
The bibliographic databases we searched included: OVID MEDLINE, EMBASE, The Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials (CENTRAL)), and Web of Science. Other sources included articles identified through discussion with experts, including experts on the review team.
2.3.2 Search strategy
The search strategy was formulated to include all articles which included the topics of chronic obstructive pulmonary disease, smoking cessation, and computed tomography. The technical differences in compiling this search were translated across the different bibliographic database. The Evidence Services Librarian at University College London was consulted to help build a robust search strategy during the protocol stage of the review. The search strategy was also appraised using the Peer Review of Electronic Search Strategies (PRESS) guideline 9. The full search strategy for each of the databases in included as appendix A. The search was carried out on 10th September 2019 for all three databases.
2.4 Data collection and analysis
2.4.1 Study selection
Titles of studies were retrieved using the search strategy from our information sources. These studies were combined with others found from other sources. Duplicates were removed. Two reviewers (DC and SA) independently screened the study titles for potential eligibility to meet the inclusion criteria. The abstracts for these studies were retrieved for a further round of screening. All studies which met inclusion screening by abstract had full text retrieved. The full texts of these articles were screened by two review members independently, and any discordance about eligibility were resolved by a third member of the review team (JRH).
2.4.2 Data extraction and management
Data was extracted using a standardised form (6Appendix B). Data was extracted independently by two study authors and any discrepancies resolved through discussion.
2.4.3 Assessment for risk of bias in individual studies
To assess for the risk of bias in individual studies, the formal risk of bias for internal validity tool ‘Risk Of Bias In Non-randomised Studies - of Interventions’ (ROBINS-I) was used 10. We first attempted to resolve disagreement between two reviewer judgements by consensus discussion; then by majority opinion with the third reviewer.
2.4.4 Data synthesis
A data extraction table was produced using the data items outlined above. We have provided a narrative synthesis of the findings of the included studies and reported the direction and magnitude of the effect size of summary statistics. We have also summarised the homogeneity and perceived efficacy of the smoking cessation intervention performed, and the different CT metrics reported by each study.
We have not proceeded to any formal meta-analysis. This is because of the small number of included studies, the heterogenous CT metrics used, the difference in time between stopping smoking, and the heterogenous clinical characteristics of the patient populations.
3 Results
3.1 Results of search
A total of 1467 studies were retrieved from the search for screening and combined with one study collected from discussion with experts. 194 duplicates were removed to give a total of 1273 studies to screen by title and abstract. At title and abstract screening, reviewer DC included 12 studies, whereas SA included 14. 6 of these were in conflict, with a total of 16 studies included by a minimum of 1 reviewer. These studies had full text retrieved for screening, of which 5 were included in the final analysis. Study selection is summarised in the PRISMA flowchart (Figure 1).
3.2 Excluded studies
1257 studies were excluded by screening by title and abstract. 11 studies were excluded by full text, and the reasons for this are detailed in table 5.4 ‘Characteristics of excluded studies‘. The most common reason was that there was no evidence of smoking cessation between the two time points, and studies examined CT differences between former smokers and current smokers (n=5). The next most common reason was that the population investigated were not COPD patients (n=4).
3.3 Included studies
The five remaining studies were published between 2011 and 2019. Details of the included studies are included in the table 5.3 ‘Characteristics of included studies‘. Four studies were published in English. One study was published in Chinese and was translated by reviewer DC for the review team.
3.4 Risk of bias
3.4.1 Risk of bias across studies
The formal risk of bias for internal validity tool ‘Risk Of Bias In Non-randomised Studies – of Interventions’ (ROBINS-I) was used to assess biases across all included studies. The confounders selected for consideration included: socioeconomic differences between intervention groups; and severity of disease and differences in previous history of smoking between intervention groups. Socioeconomic factors may influence co-variates such as environmental exposures which could influence the outcome. Patients with severe disease or a stronger smoking history may be less likely to successfully quit smoking and thus baseline lung pathology could influence the outcome.
Co-interventions such as nicotine replacement therapy or exercise were possibly different between intervention groups, but this was not thought likely to impact outcomes.
The ‘ideal’ randomised trial design was an individually randomised trial of COPD patients to either continue smoking or stop smoking immediately after the first outcome measure and remain abstinent until a repeat outcome measure at 1 year.
3.4.2 Risk of bias within included studies
The ROBINS-I tool was used to evaluate individual studies for potential bias due to confounding, participant selection, intervention classification, deviation from intervention, missing data, outcome measurement, and selective reporting. The authors’ risk of bias judgements for the individual studies are included in the supplementary table 5.3 ‘Characteristics of included studies‘.
Due to a critical risk of bias in the selective reporting, of the 5 included studies, one11 has not been included in the narrative synthesis, and the results presented hereafter are from the remaining four studies. The excluded study compared the change in %LAA-950 between the 5 lobes of the lung between two time points without evidence of multi-comparison correction for statistical significance.
The included studies detailed numbers of patients in each cohort who successfully stopped smoking according to their definition of smoking cessation. These definitions differed however, making direct comparison difficult. For example, the monitoring differed, with some authors replying on smoking diaries, and some on self-reported questionnaires. One study used carbon monoxide monitoring. None of the studies reported the time duration from smoking cessation to the outcome measures (CT scan).
3.5 Main outcomes
3.5.1 Quantitative change in CT markers of emphysema
Two of the included studies involving 81 patients who stopped smoking, reported changes in quantitative CT markers of lung density using the percentage or relative area of lung below an attenuation threshold 12,13. Both studies reported a significant increase in the %LAA/RA-950 or -910 HU, representing a fall in lung density. Both also reported a significant decrease in the 15th percentile of lung density (PD15), again representing a fall in lung density with smoking cessation.
Shaker and Hlaing demonstrated that up to a year after smoking cessation, there is a significant decrease in lung density (n=36, n=45 respectively), a difference that was demonstrated to be greater than a continued smoking control group from Shaker’s original study (Table 5.2).
Hlaing et al (2015) also demonstrated a significant reduction in the mean lung density (MLD, -7.7HU, SD=2.3, p<0.001).
3.5.2 Quantitative change in CT markers of airway wall thickness
Only one study involving 203 patients who stopped smoking reported changes in airway wall metrics on CT with smoking cessation 14. The authors reported a significant decrease in wall thickness of - 0.18mm (95% CI -0.23 to -0.13, p<0.001) on smoking cessation between two scans five years apart. The authors did not comment on duration of smoking cessation or time from smoking cessation to second scan but defined smoking cessation as self-reported abstinence from smoking after the first scan.
3.6 Other outcomes
3.6.1 Qualitative change in CT appearances
One study reported qualitative changes in appearances of emphysema on smoking cessation. This was performed blinded, with the radiologist comparing scans for emphysema, ground glass opacification or micronodules, documenting whether these features were more, less, or the same quantity as the baseline scan. The study reported no significant difference in the presence of emphysema visually with smoking cessation but did report a decrease in the presence of micronodules.
3.6.2 Change in FEV1 with smoking cessation
Three of the five studies reported a change in FEV1 with smoking cessation. Two papers reported no significant difference in this measure at one year, whilst Hlaing et al (2015) reported a significant decline in FEV1 after 1 year sustained smoking abstinence (-33ml, p<0.001). The authors noted that this change was less than patients who continued to smoke.
Dhariwal et al (2014) reported a transient improvement in FEV1 of mean 184ml at 6 weeks, but this decreased to 81ml at 12 weeks and not fully maintained at 1 year. Shaker et al (2011) reported no significant change in FEV1 within one year of smoking cessation (72ml, SD=47ml, p=0.14).
4 Discussion
4.1 Summary of main results
4.1.1 Measures of Lung Density
In summary, quantitative CT metrics of lung density fall with smoking cessation. The direction of effect of three markers across the four studies included was consistent. Relative area of low attenuation (RAA) - RAA-910 and RAA-950 are defined as the percentage of lung pixels, or the relative area, with lower than 910 and 950 Hounsfield units respectively (approximately the density of air). The advantage of these measures and the mean lung density is that they are objective, replicable and are quick to assess using automated software, however may be underestimated in the presence of transient consolidation. MLD suffers from the additional disadvantage in that it includes non-airspace densities even when the lungs are segmented meticulously. Density of the 15th percentile (PD15) is similarly an automated density measure, but one that is not affected by ‘outliers’ e.g. consolidation.
4.1.2 Measures of Airway Thickness
Only one study reported the change in airway wall thickness, suggesting that this decreases with smoking cessation, but because this result has not been replicated, and difficult to interpret with confidence. The Pi10 measure, the square root wall thickness for all airways with an internal diameter of 10mm, is a global measure of wall thickness using automated software. This measure is objective and replicable, but importantly is different to measures of lung density and therefore is theoretically prone to a different set of confounders. Charbonnier [**] demonstrated that the same cohort of quitters had both a decrease in Pi10 and RAA-910, suggesting that this effect is not due to a different patient cohort.
4.2 Limitations
Overall, a small number of studies were available for evidence synthesis, and they varied in quality and risk of bias. Of the five studies included after the final search, one had a critical risk of bias. The studies addressed prior smoking history and severity of COPD, with varying degrees of rigour.
At the outcome level, most of the studies used differing outcome measures which made direct comparison difficult and meta-analysis impossible. The outcomes chosen were also limited by the scans performed, with newer biomarkers such as parametric response mapping (PRM) not yet evaluated. The studies also varied in terms of length of follow up.
At the review level, limitations were the small number of studies which met our final inclusion criteria. Large longitudinal cohorts such as ECLIPSE likely contain the data required for this analysis, but researchers have not analysed the effect of smoking cessation 16 There likely exists a sizable dataset of imaging data for in lung cancer screening populations, and we were careful not to exclude these from our search, but we did not identify any which examined the subset of COPD patients. Even within the broader question of the longitudinal anatomical evolution of COPD is not well studied. As noted by Coxson et al, cohorts were limited by their lack of specific focus on COPD patients.
4.3 Conclusions
CT measures of COPD equate to lower lung density and thickened airways, although these studies demonstrate this more in the smoking cessation groups.
Pathologies that affect airway wall thickness and lung density can confound these CT biomarkers. Smoking may promote an inflammatory state which increases the density of lung parenchyma and thickens the airway walls15. The implication is that these biomarkers may be largely measuring transient inflammation, and that the hypothesised mechanism of lower lung density and thickened airways – which could be interpreted as COPD progression – requires further study. It is worth noting that no study scanned beyond 5 years.
The pulmonary CT imaging changes that occur with smoking cessation in patients with COPD needs further study, and it is not at present possible to test the hypothesis that smoking-cessation changes on CT in COPD may be useful as biomarkers of drug intervention.
5. Figures and tables
5.1 Figure 1: PRISMA diagram
5.3 Characteristics of included studies
5.3.1 Charbonnier et al 2019
5.3.2 Li et al 2018
This study was not included in the narrative synthesis due to critical risk of bias.
5.3.3 Hlaing et al 2015
5.3.4 Dhariwal et al 2014
5.3.5 Shaker et al 2011
5.4 Characteristics of excluded studies
Data Availability
The data referred to in the manuscript is included either in the manuscript or the supplementary data at the end. The raw data is also available on request from the study authors.
Appendix A Search Strategies
A.1 Ovid MEDLINE® (1946 to August Week 5 2019)
exp Pulmonary Disease, Chronic Obstructive/ (52490)
Lung Diseases, Obstructive/ (18135)
(obstruct$ adj3 (pulmonary or lung$ or airway$ or airflow$ or bronch$ or respirat$)).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (102899)
COPD.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (36917)
COAD.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (244)
emphysema*.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (32495)
(chronic* adj3 bronchiti*).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (10836)
1 or 2 or 3 or 4 or 5 or 6 or 7 (138142)
smoking cessation/ or smoking reduction/ or smoking/ or exp pipe smoking/ or exp tobacco smoking/ (150642)
((quit* or stop* or ceas* or cessation or abstain* or abstinen*) adj3 (smok* or tobacco or cigar*)).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (41160)
9 or 10 (157596)
exp Tomography, X-Ray Computed/ (410855)
CT scan.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (44507)
CAT scan.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (820)
CATSCAN.mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (3)
(comput* adj2 tomograph*).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] (300434)
12 or 13 or 14 or 15 or 16 (546904)
8 and 11 and 17 (769)
A.2 EMBASE (Embase Classic & Embase 1947 to 2019 September 09)
exp chronic obstructive lung disease/ (125489)
COPD.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (82778)
(obstruct$ adj3 (pulmonary or lung$ or airway$ or airflow$ or bronch$ or respirat$)).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (201771)
COAD.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (455)
(chronic$ adj3 bronchiti$).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (22805)
emphysema*.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (56796)
1 or 2 or 3 or 4 or 5 or 6 (271226)
exp smoking cessation/ (56555)
exp smoking reduction/ (149)
smoking cessation program/ (3217)
((smok* or tobacco or cigar*) adj3 (quit* or stop* or ceas* or cessation or abstain* or abstinen*)).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (71196)
8 or 9 or 10 or 11 (71244)
exp computer assisted tomography/ (1002372)
CT scan.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (98165)
CAT scan.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (1360)
CATSCAN.mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (17)
(comput* adj2 tomograph*).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword, floating subheading word, candidate term word] (992850)
13 or 14 or 15 or 16 or 17 (1100405)
7 and 12 and 18 (466)
A.3 The Cochrane Library
MeSH descriptor: [Pulmonary Disease, Chronic Obstructive] explode all trees-4858
MeSH descriptor: [Lung Diseases, Obstructive] this term only-2522
COPD OR (chronic obstructive pulmonary disease*)-18567
COAD OR (chronic obstructive airway* disease*)-8305
(obstruct* NEAR/2 (pulmonary OR lung* OR airway* OR airflow* OR bronch* OR respirat*))-18034
chronic bronchitis-2597
emphysema*-1646
{OR #1-#7}-25813
MeSH descriptor: [Smoking Cessation] explode all trees-3779
MeSH descriptor: [Tobacco Use Cessation] explode all trees-94
MeSH descriptor: [Smoking Reduction] explode all trees-9
stop* smok*-2339
(“tobacco control intervention”):kw-0
(“smoking cessation treatment”):kw-108
smok* cessation-10210
(quit* OR stop* OR cessation OR ceas*) NEAR (smok* OR tobacco OR cigar*)-11130 17.
{OR #9-#16}-11930
MeSH descriptor: [Tomography, X-Ray] explode all trees-4975
CT scan*-10513
comput* NEAR/2 tomograph*-19483
CAT scan*-226
CATSCAN*-17
A.4 Web of Science (All years 1900-2019)
TS=(COPD OR chronic obstructive pulmonary disease* OR COAD OR chronic obstructive airway* disease* OR chronic obstructive lung disease OR emphysema OR chronic bronchitis) [105,681]
TS=(CT OR comput* tomograph*OR CAT OR CATSCAN) [401,167]
TS=((smok* OR tobacco OR cigar*) AND (quit* OR stop* OR cessation OR ceas* OR giv* OR prevent* OR abstain* or abstinen*)) [97,885]
#3 AND #2 AND #1 [163]
Appendix B Data extraction form
Included as supplementary figure