Exercise across the Lung Cancer Care Continuum: An Overview of Systematic Reviews

Background: Growing evidence supports exercise for people with lung cancer. This overview aimed to summarise exercise intervention efficacy and safety across the care continuum. Methods: Eight databases (including Cochrane and Medline) were searched (inception—February 2022) for systematic reviews of RCTs/quasi-RCTs. Eligibility: population—adults with lung cancer; intervention: exercise (e.g., aerobic, resistance) +/− non-exercise (e.g., nutrition); comparator: usual care/non-exercise; primary outcomes: exercise capacity, physical function, health-related quality of life (HRQoL) and post-operative complications. Duplicate, independent title/abstract and full-text screening, data extraction and quality ratings (AMSTAR-2) were completed. Results: Thirty systematic reviews involving between 157 and 2109 participants (n = 6440 total) were included. Most reviews (n = 28) involved surgical participants. Twenty-five reviews performed meta-analyses. The review quality was commonly rated critically low (n = 22) or low (n = 7). Reviews commonly included combinations of aerobic, resistance and/or respiratory exercise interventions. Pre-operative meta-analyses demonstrated that exercise reduces post-operative complications (n = 4/7) and improves exercise capacity (n = 6/6), whilst HRQoL findings were non-significant (n = 3/3). Post-operative meta-analyses reported significant improvements in exercise capacity (n = 2/3) and muscle strength (n = 1/1) and non-significant HRQoL changes (n = 8/10). Interventions delivered to mixed surgical and non-surgical populations improved exercise capacity (n = 3/4), muscle strength (n = 2/2) and HRQoL (n = 3). Meta-analyses of interventions in non-surgical populations demonstrated inconsistent findings. Adverse event rates were low, however, few reviews reported on safety. Conclusions: A large body of evidence supports lung cancer exercise interventions to reduce complications and improve exercise capacity in pre- and post-operative populations. Additional higher-quality research is needed, particularly in the non-surgical population, including subgroup analyses of exercise type and setting.


Introduction
Globally, lung cancer accounted for over 2.2 million (11.4%) incident cancers in 2020 [1]. The majority of diagnoses, 54% in the United States, occur once the disease has metastasized and is considered incurable [2]. In Australia, although lung cancer incidence is reducing, 90,000 new cases are predicted between 2040-2044 due to the ageing population. This incidence is the third and fourth highest of all tumour types for males and females, respectively [3]. Significantly, the number of people dying from lung cancer is predicted to decline in both sexes between 2020 and 2040-2044 [3]. These factors together will mean a growing number of people with lung cancer.
As an area of research, the evaluation of the effects of exercise on people with lung cancer has developed more slowly than in other areas. Study interventions commonly involve combinations of aerobic, resistance and respiratory training (some include inspiratory muscle training). Interventions aim to reduce the risk of post-operative pulmonary complications and aid recovery following surgery. In inoperable populations, the focus of interventions is on reducing symptom burden, commonly fatigue, pain and dyspnoea, and preventing a decline in physical function and health-related quality of life [4]. The first systematic review was published by Granger and colleagues in 2011 and included two randomised controlled trials (RCTs), nine case series and two cohort studies [5]. In contrast, a search of the International prospective register of systematic reviews (PROSPERO) in August 2022 using the terms 'lung cancer' AND ('exercise' OR 'rehabilitation') retrieved 145 registered systematic reviews, highlighting the rapid development of research in this area. The increasing volume and strength of evidence, media attention and an increasing number of lung cancer survivors place increasing demands on limited-exercise oncology services [6]. Evidence-based strategies to direct services to patients who are likely to benefit the most are essential. Overviews of reviews to synthesise the growing evidence base are now required. In people having surgery for lung cancer, Zhou and colleagues report an overview of systematic reviews, published prior to October 2019, of lung cancer exercise interventions delivered during the perioperative period [7]. Findings included low-quality evidence that pre-operative programs reduce post-operative pulmonary complications and hospital length of stay and increase exercise capacity and pulmonary function. Moderate to high-quality evidence supported post-operative programs in increasing exercise capacity and muscle strength. There was very low to low-quality evidence that programs improved health-related quality of life (HRQoL) and reduced dyspnoea [7]. Exercise adherence varies widely and further research to identify enablers of participation is required; an adherence of between 9-125% is reported by pre-and rehabilitation lung cancer studies which include a home-based component [8].
This overview of systematic reviews aimed to synthesise findings from and evaluate the quality of the current systematic review evidence on the efficacy and safety of exercise for people with both operable and inoperable lung cancer, delivered across the care continuum. The secondary aims were to report subgroup meta-analysis findings according to the type of exercise intervention/s and delivery settings, where possible.

Materials and Methods
This overview of reviews was guided by the Cochrane Handbook of Systematic Reviews of Interventions [9] and is reported according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA2020) guidelines [10]. The protocol was registered prospectively on the PROSPERO database (CRD42015001068 https: //www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021257938 (registered on 6 July 2021)).

Eligibility Criteria
The inclusion criteria for systematic reviews were as follows: Population-Patients (≥18 yo) diagnosed with lung cancer (non-small cell or small cell); Intervention-Any supervised or unsupervised exercise interventions delivered alone or in combination with any non-exercise interventions (e.g., nutrition, symptom management, psychological support); Comparator-Usual care or a non-exercise intervention; Context-Any setting (hospital, community or home); Outcomes-At least one health-related outcome. The primary outcomes of interest were exercise capacity, physical function, HRQoL and postoperative pulmonary complications (PPCs). Table S1 in the Supplementary Materials provides further details. Systematic reviews, with or without a meta-analysis, including only RCTs or quasirandomised controlled trials (qRCTs), were included. Systematic reviews including findings from other study designs were included if RCT or qRCT findings were reported separately. Abstract-only citations (e.g., conference proceedings) and narrative or non-systematic reviews were excluded. Additionally, systematic reviews where the study population included mixed cancer groups with <50% lung cancer, or mixed cancer groups with lung cancer findings not reported separately were excluded. Only systematic reviews available in English were included.

Literature Search
A comprehensive literature search of eight databases was performed; the Cochrane Systematic Review Database, the Database of Abstracts of Reviews of Effectiveness (DARE), Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library), Ovid SP MEDLINE, Ovid SP EMBASE, SPORTDiscus and CINAHL via EBSCO host and PEDro from inception until the 18 May 2021 and updated on 21 February 2022. The search string was developed in consultation with content specialists and a research librarian from the University of Melbourne, using the medical subject headings (MeSH) dictionary in MEDLINE to identify key terms and was adapted for use in CINAHL, SPORTDiscus, PEDro, CENTRAL and EMBASE. The full search strategy for each database is provided in the Supplementary Materials (Table S2).
Two researchers (LE and AB) independently screened the titles and abstracts of all articles retrieved and conducted a full-text review of all articles considered potentially relevant. Consensus between the two researchers was used to resolve any disagreements and a third researcher (LD) was available if a consensus could not be reached. Reference management software Covidence was used for managing all retrieved records [11]. Additional potentially relevant articles were identified by screening the reference lists of all included full-text articles.

Data Extraction and Quality Assessment
Two researchers (AB and NB) independently extracted data using a standardised form developed prior to database searching. A third researcher (LE) resolved any discrepancies in the data extraction. Data were collected for (1) review characteristics, including search dates, language restrictions, synthesis and quality appraisal methods; (2) primary study designs; (3) review population, intervention/s, comparator/s and outcomes; and (4) metaanalysis results for the overview primary outcomes. The authors of the included systematic reviews were contacted by email where discrepancies were unable to be resolved by the researchers or where data were not reported. Grades of Recommendations, Assessment, Development and Evaluation (GRADE) ratings, reported in the included systematic reviews, were extracted. Where GRADE was not reported for an overview primary outcome (e.g., physical function in the 'During and post-treatment (non-surgical)' included systematic reviews), two researchers (LE and SA) independently performed GRADE evidence certainty assessments, with any disagreements resolved by a third researcher (LD) [12].
The overall quality of reporting and methodological rigour of included systematic reviews was critically appraised by two researchers (LE and AB) independently, using the PRISMA-2020 item and abstracts [13] checklists and the Assessment of Multiple Systematic Reviews (AMSTAR-2) [14]. PRISMA-2020 is a 27-item (some with sub-items) checklist covering 7 areas (title, abstract, introduction, methods, results, discussion and other information (e.g., protocol registration, funding support). In case of disagreement, a consensus was reached by discussion and a third researcher (LD) was available if consensus could not be reached. The AMSTAR-2 contains 16 items, 7 of which are regarded 'critical' (protocol registration, appropriate literature search, excluded study justification, individual study risk of bias, methodological appropriateness of meta-analyses, risk of bias considered in result interpretation and publication bias likelihood/impact). In line with AMSTAR-2 scoring guidelines, confidence in the findings of the systematic reviews was rated 'critically low' if there was more than one critical flaw, 'low' for systematic reviews with one critical flaw, 'moderate' for those with no critical flaws but more than one non-critical weakness and 'high' for systematic reviews with one non-critical weakness, or with none [14].

Synthesis of Results
Findings from meta-analysed data included in the overview were synthesised narratively at the review level for each overview primary outcome. Included systematic review characteristics, findings, risk of bias and GRADE assessments were reported in subgroups according to the stage on the cancer treatment continuum that the interventions were delivered: (1) pre-treatment only (surgical); (2) post-treatment only (surgical); (3) preand post-treatment (surgical); (4) pre, during and/or post-treatments (mixed surgical and non-surgical) and; (5) during and post-treatments (non-surgical only).
The primary study overlap between the included systematic reviews was reviewed by authors (LE and AB) and visually mapped using a citation matrix. In the event of updated systematic reviews or systematic reviews that included identical studies and addressed the same research question, the more recent review was included. A citation matrix was created, and the corrected covered area (CCA) was calculated to assess the degree of primary study overlap. A CCA within the range of 0-5% indicates slight overlap, 6-10% indicates a moderate overlap, 11-15% indicates a high overlap and >15% indicates a very high amount of overlap [15].

Protocol Deviations
Two deviations from the prospectively registered protocol were made: the addition of a language restriction to English only, and the inclusion of electrical stimulation as an additional exercise intervention within the eligibility criteria.

Search Results
The search retrieved 5068 articles, and 4114 were screened for eligibility following the removal of duplicates (refer to Figure 1 PRISMA flow diagram for further details). Ninety-seven articles were retrieved for full-text review. Details of excluded articles at full-text screening and reasons for exclusion are provided in the Supplementary Materials (Table S3). Thirty systematic reviews (SRs) involving 6440 participants were included in the overview, of which 25 included synthesised findings in at least one meta-analysis.

Methodological Rigour and Quality of Reporting
AMSTAR-2 ratings were 'critically low' in 22 (73.3%), 'low' in 7 (23.3%) and 'moderate' in 1 (3.3%) of the included SRs. All SRs used an appropriate method for assessing the risk of bias of individual studies. Nine SRs (30%) provided a list with justification for excluded studies and ten (33.3%) accounted for individual study risk of bias in interpreting findings. Twelve SRs (40%) provided justification for their study design inclusion criteria. Further details are provided in the Supplementary Materials (Tables S4 and S5).
All SRs met PRISMA 2020 guideline items relating to reporting a rationale for the review and flow diagram of included studies; outlining included study characteristics; and interpreting their findings. Twenty (67%) SRs did not provide citations and justification for the exclusion of studies and nineteen (63%) did not assess certainty of the evidence. The Supplementary Materials provides further details (Tables S6 and S7). Table 1 provides details of included SRs, including participants, interventions, comparators, outcomes, synthesis and quality appraisal methods. A narrative summary according to stage on the treatment continuum is provided below.

Methodological Rigour and Quality of Reporting
AMSTAR-2 ratings were 'critically low' in 22 (73.3%), 'low' in 7 (23.3%) and 'moderate' in 1 (3.3%) of the included SRs. All SRs used an appropriate method for assessing the risk of bias of individual studies. Nine SRs (30%) provided a list with justification for excluded studies and ten (33.3%) accounted for individual study risk of bias in interpreting findings. Twelve SRs (40%) provided justification for their study design inclusion criteria. Further details are provided in the Supplementary Materials (Tables S4 and S5).
All SRs met PRISMA 2020 guideline items relating to reporting a rationale for the review and flow diagram of included studies; outlining included study characteristics; and interpreting their findings. Twenty (67%) SRs did not provide citations and justification for the exclusion of studies and nineteen (63%) did not assess certainty of the evidence. The Supplementary Materials provides further details (Tables S6 and S7).  Table 1 provides details of included SRs, including participants, interventions, comparators, outcomes, synthesis and quality appraisal methods. A narrative summary according to stage on the treatment continuum is provided below.

Pre-Treatment Only (Surgical)
Eight SRs assessed the effects of exercise training delivered pre-operatively [4,[16][17][18][19][20][21][22]. Sample sizes ranged from 203 to 791 per SR. Exercise training was commonly multimodal combinations of aerobic and resistance training. Two SRs focused on the effects of respiratory exercises [20] and inspiratory muscle training (IMT) [22]. Three SRs included exercise-only interventions whilst those remaining included RCTs where exercise was delivered as part of pulmonary rehabilitation or with nutrition, stress management and relaxation, psychological education or pharmacotherapy interventions. Program duration ranged from one to eight weeks. The reported primary outcome of interest was PPCs in four of the SRs, unspecified in two SRs, and exercise capacity and HRQoL in one SR, respectively.

Post-Treatment Only (Surgical)
Four SRs focused on exercise interventions delivered in the post-operative period [23][24][25][26]. Sample sizes ranged from 262 to 4381 participants per SR. One focused specifically on respiratory physiotherapy interventions delivered alone or combined with aerobic and resistance training [26]. Interventions were heterogeneous in the other three (combinations of aerobic, resistance, IMT, breathing exercises, balance and whole body vibration). Program duration ranged from 5 days [26] supervised in the inpatient setting to 20 weeks [24] supervised in the inpatient or outpatient setting combined with unsupervised home exercise. The primary outcome was exercise capacity in two SRs [24,25], HRQoL in one SR [23] and unspecified in the fourth SR.

Pre and Post-Treatment (Surgical)
Six SRs reported on the effects of exercise interventions pre-and post-surgery [27][28][29][30][31][32]. The sample sizes of included reviews ranged from 157 to 2068 participants per SR. Exercise interventions in all SRs included combinations of aerobic, resistance and respiratory training. The four most recent SRs included high-intensity interval training (HIIT) [29][30][31][32] and three SRs also included balance training [30][31][32]. One SR focused on the effect of respiratory exercise [29]. Program duration ranged from 5 days to 8 weeks pre-treatment and 5 days to 20 weeks in the post-treatment period. Outcomes of interest included exercise capacity, lung function, complications (including pulmonary), hospital length of stay (LOS), fatigue, HRQoL and mortality. The primary outcome was not specified by three SRs and was pulmonary function, HRQoL and PPCs/hospital LOS in the other three, respectively.
Pre, during and/or Post-Treatment (Surgical and Non-Surgical) Ten SRs [33][34][35][36][37][38][39][40][41][42] synthesised the effects of exercise for people with lung cancer pre, during or following both surgical and non-surgical treatment. Sample sizes ranged from 187 to 2109 participants per SR. One SR question specifically focused on the effects of respiratory exercises [33] and another on HIIT [41]. Interventions included within SRs were typically combined aerobic, resistance and respiratory training and six SRs also included Tai Chi [34,36,[38][39][40]42]. The program duration ranged from 1 to 20 weeks. Several SRs reported on the effects of exercise on single outcomes, namely depressive symptoms [33], sleep quality [34], HRQoL [37] and cancer-related fatigue [40,42]. In the remaining SRs, the primary outcome/s of interest was not reported by four SRs and was dyspnoea and exercise capacity in one SR.
During and/or Post-Treatment (Non-Surgical) Two SRs evaluated the effects of exercise during and following treatment only and included only non-surgical participants [43,44]. One SR included only participants receiving curative or palliative chemotherapy [44]. Sample sizes were 221 and 244 participants per SR. Both SRs included aerobic, resistance and respiratory exercise interventions of 6 to 12 weeks duration. Exercise capacity was reported as the primary outcome of interest in one SR [43] and was not reported by the other SR [44].

Intervention Effects
A GRADE evidence summary map is shown in Figure 2. Figures 3-5 and Supplementary Figures S2 and S3 provide a summary of meta-analyses and GRADE evidence certainty for the primary outcomes of interest of this overview of SRs and the SR quality ratings (AMSTAR-2). Supplementary Table S8 summarises details of the exercise interventions of these meta-analyses. Further details are provided narratively below.  [18][19][20][21][22] and three of these also analysed CPET VO 2 peak [18][19][20]. The sample size in meta-analyses ranged from 81 to 523. All meta-analyses reported significant positive effects of exercise training on exercise capacity, with mean differences (95% CI) ranging from 18.23 m (8.5, 27.96) [4] to 71.25 m (39.68, 102.82) [19]. The GRADE evidence certainty was 'low'-'high'. Gravier

Health-Related Quality of Life
Health-related quality of life was synthesised in the meta-analyses of three SRs [18,20,21], with sample sizes ranging from 161 to 266 participants. Non-significant between-group differences were reported in all meta-analyses. The GRADE evidence certainty was 'very low'-'moderate'.

Post-Treatment Only (Surgical)-Figure 4 Exercise Capacity
The positive effects of exercise training on exercise capacity (VO 2 peak or 6 MWT) were reported by two of three SRs immediately post-program [24,25]. This was not maintained at longer-term follow up (SMD 0.09 (−0.44 to 0.61, n = 56) [24]. Only one SR reported a subgroup analysis of rehabilitation timing, reporting no significant effects of exercise training on exercise capacity with early initiated rehabilitation, in contrast to interventions initiated later (SMD 0.58 (0.07, 1.09)) [24]. The GRADE evidence certainty was 'low'-'high'.

Health-Related Quality of Life
Three SRs [23][24][25] completed meta-analyses of HRQoL globally or by domain. The findings were non-significant aside from two meta-analyses of positive findings for HRQoL physical components immediately post-program (SMD 0.50 (0.19, 0.82) and MD 5.02 (2.30, 7.73)) which were not maintained in the one SR that assessed at 1-year follow-up (SMD 0.27 (−0.78, 0.25)) [24]. The GRADE evidence certainty was 'very low'-'low'.

Post-Operative Pulmonary Complications
Two SRs reported on post-operative complications with neither reporting significant effects of post-operative exercise, the GRADE evidence certainty was 'moderate' [23,26].

Health-Related Quality of Life
One SR meta-analysed HRQoL and reported significant effects favouring exercise interventions for the physical HRQoL domain (SMD 0.68 (0.47, 0.89), the GRADE evidence certainty was 'high') [31]. The findings for mental and emotional HRQoL domains and global HRQoL were not significantly different between groups.
3.4.4. Pre, during and/or Post-Treatment (Surgical and Non-Surgical)- Figure S2 Exercise Capacity

Health-Related Quality of Life
One SR meta-analysed HRQoL and reported significant effects favouring exercise interventions for the physical HRQoL domain (SMD 0.68 (0.47, 0.89), the GRADE evidence certainty was 'high') [31]. The findings for mental and emotional HRQoL domains and global HRQoL were not significantly different between groups.

Pre, during and/or Post-Treatment (Surgical and Non-Surgical)-Figure S2 Exercise Capacity
Four meta-analyses synthesised the effects of exercise capacity [35,36,41,42]. Three of four reported significant increases in exercise capacity favouring the intervention (MD between 20.4 and 37.7 m for the 6 MWT) [35,36,41], the GRADE evidence certainty was 'very low'-'moderate'. In subgroup analyses, Singh et al. reported significant effects from metaanalyses involving aerobic-only and combined interventions and programs of <12 weeks or ≥12 weeks; both resulted in significant increases in exercise capacity compared to usual care. Subgroup analyses of supervised interventions were more effective than unsupervised (SMD 0.54 (0.32, 0.76) versus SMD 0.95 (−0.25, 2.16)) [36]. Both breathing exercises only and breathing exercises combined with aerobic and resistance significantly increased exercise capacity [35].

Health-Related Quality of Life
Two meta-analyses demonstrated the positive effects of exercise interventions for global HRQoL with 'low' GRADE evidence certainty [36,42] and these were maintained in subgroup analyses of exercise type (aerobic only or combined) and program duration (<12 weeks or ≥12 weeks). Positive effects remained for supervised programs (SMD 0. and the other nonsignificant effects (SMD 0.38 (−0.42, 1.18)), the GRADE evidence certainty was 'very low'-'low' [43,44].

Physical Function
Upper limb strength was reported to be significantly improved following exercise in one meta-analysis (SMD 1.39 (0.80, 1.98), the GRADE evidence certainty was 'low' [44].

Health-Related Quality of Life
Peddle-McIntyre et al. reported non-significant meta-analysis findings for physical HRQoL and positive findings favouring the intervention group for general HRQoL, GRADE evidence certainty was 'low' [43]. Lee reported significant benefits favouring the intervention group for physical, social, functional and general well-being, the GRADE evidence certainty was 'moderate' [44].

Safety
One SR evaluated the safety of exercise interventions in the post-operative population. Four of eight of the included RCTs reported on adverse events, with only a single serious adverse event occurring [25]. In the one meta-analysis performed pre, during and/or post treatment, there were no significant differences in adverse events (grade 3-5 CTCAE severity ratings) between intervention and usual care participants (32 RCTs, n = 2109, 64 events (intervention) versus 61 events (usual care), risk difference −0.01 (−0.02, 0.01), I 2 = 17%). The differences remained non-significant for subgroup analyses of exercise type (aerobic only, resistance only, combined or other), supervised/unsupervised and program duration (<12 weeks or ≥12 weeks) [36]. Data regarding the safety of exercise during and/or following treatment in people with lung cancer managed non-surgically was limited. One SR reported on adverse events and supported the safety of exercise training in advanced lung cancer with no serious adverse events (e.g., mortality, fractures) and limited minor adverse events (musculoskeletal injuries) occurring [43].

Overlap of Included Systematic Reviews
The corrected covered area calculation was 4.9% (see Figure S1 in the Supplementary Materials). This represents a slight overlap of primary studies (RCTs) included within this overview.

Discussion
This overview of reviews synthesised findings from 30 systematic reviews of over 6000 participants and investigated the efficacy and safety of exercise for people with surgical and non-surgical lung cancer across the care continuum. AMSTAR-2 ratings of the included systematic review quality were predominantly 'very low' to 'low', highlighting areas for improvement in future systematic review conduct and reporting. Clear efficacy exists for exercise interventions in lung cancer surgical populations, the minority of those diagnosed, particularly with respect to the prehabilitation period, for the outcomes of exercise capacity and post-operative pulmonary complications. It should be noted, however, that the GRADE certainty of evidence for these outcomes ranged from 'low' to 'high'. In non-surgical lung cancer populations, additional higher-quality evidence is required to support the efficacy of exercise interventions. Only three systematic reviews, across both operable and inoperable populations, synthesised safety (adverse event) findings and all three reported few adverse events associated with exercise across the lung cancer care continuum. The need for improved transparency and consistency of reporting within studies is evident, with safety often not reported by the included RCTs.
Adding to the weight of evidence supporting lung cancer prehabilitation synthesised in this overview of reviews, the recently updated Cochrane systematic review, including 10 RCTs of over 600 participants, found high certainty evidence of a large reduction in post-operative pulmonary complication risk (RR (95% CI) 0.45 (0.33 to 0.61)) and moderate certainty evidence of an increase in exercise capacity (VO 2 peak MD 3.36 mL/kg/min (2.70 to 4.02)) associated with pre-operative exercise [45]. In the lung cancer surgical population, our attention needs to now focus on cost-effectiveness studies and the implementation of research to identify effective strategies for implementing exercise interventions into usual care. An excellent example of this is the UK-based 'Prehab4Cancer' lung cancer program which commenced in 2019 and services the greater Manchester area. Developed through a multi-disciplinary collaboration between clinical groups, a regional cancer alliance and community leisure centers, this community-based service includes exercise with nutrition and psychology also provided, dependent on screening criteria. In the 11 months prior to COVID-19, 377 people with lung cancer from 11 hospitals were referred. Seventy-four percent completed a baseline assessment and 48% completed the prehabilitation phase. The median program attendance was six sessions. Significant and clinically meaningful post-program improvements in objective (a 43 m increase in 6 MWT distance) and patientreported (physical activity and HRQoL) outcomes were reported, and there were no adverse events recorded [46].
Within the systematic reviews included in this overview, there was significant heterogeneity in terms of the exercise modalities included in the interventions and elements of exercise prescription. This high degree of heterogeneity of interventions limited the ability of included systematic reviews to perform subgroup analyses according to inter-vention characteristics and delivery settings, a secondary aim of this overview. Further research, including network meta-analyses, is needed to establish the optimal exercise intervention features (including modality, intensity and duration) and identifying those most likely to benefit. Lu et al. will investigate in the lung cancer surgical population the effects of different types of exercise training on HRQoL, exercise capacity, lung function, adverse events and mortality in a Bayesian network meta-analysis [47]. A recent RCT posttreatment randomised 90 people with stage I-III lung cancer, and cardiorespiratory fitness lower than normative values, to 1 of 4 training groups (stretching attention control, aerobic, resistance or combined aerobic and resistance) [48]. The trial reported high intervention attendance (median 90%) and minimal loss to follow-up (10%). Post-program findings included improvements in exercise capacity (VO 2 peak ) in the aerobic and combined training groups compared to the attention control group. Muscle strength was also improved in the resistance and combined groups compared to the aerobic or attention control groups. It must be noted that only 56% (90/160) of the required sample size were recruited to this trial, resulting in a higher likelihood of statistical error, and findings should be interpreted with caution. The relative dose intensity, defined as the ratio of total 'completed' to total 'planned' exercise was higher in the aerobic training group, indicating that aerobic training may be more tolerable for survivors [48]. In line with previous findings in other cancer types [49], the meta-analyses of systematic reviews included in this overview reported exercise intervention effectiveness for supervised rather than unsupervised interventions [36]. Advances in the fields of real-time monitoring and reporting of exercise programs need to continue to support the fidelity of performance. This will facilitate patient-centred care for people with lung cancer, where preference is often for home or community-based programs [50], whilst ensuring programs are supervised to enhance effectiveness.
The overview protocol was developed and registered a priori and guided by a robust methodology which included a duplicate performance of all overview stages. Only the included evidence from study designs at lower risk of bias was included; RCTs and qRCTs. The decision to include all eligible systematic reviews was aligned with the overview aim of summarising the body of evidence but resulted in an overlap of primary studies included in the overview and potential double counting of outcome data. However, robust methods were used to assess and document the degree of primary study overlap of the included systematic reviews.

Conclusions
This overview has synthesised a large body of evidence and provides a clear understanding of the gaps in the current evidence base regarding exercise for people with lung cancer and directions for future research. The evidence synthesised in this overview supports lung cancer exercise interventions to reduce complications and improve exercise capacity in pre-and post-operative populations, and research should now focus on implementation. Additional higher-quality research is needed, particularly in non-surgical populations, including subgroup analyses to determine optimal exercise types and settings.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm12051871/s1, Table S1. Overview PICO research question criteria, Table S2. Database search strategies, Table S3. Excluded full-text references and reasons for exclusion, Table S4. AMSTAR-2 items and questions, Table S5. AMSTAR-2 methodological quality ratings, Table S6. PRISMA reporting items, Table S7. PRISMA reporting ratings, Table S8. Exercise interventions from RCTs included in meta-analyses for overview primary outcomes, Figure  S1. Correlation correction matrix, Figure S2. Pre, during and/or post treatment (surgical and nonsurgical)-meta-analysis findings for overview primary outcomes, Figure S3. During and/or post treatment (non-surgical)-meta-analysis findings for overview primary outcomes.

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article.