Grading the evidence for CPET variables in predicting perioperative morbidity and long-term mortality in lung cancer surgery: a systematic review and meta-analysis

Abstract

Objective

Current cardiopulmonary exercise testing (CPET) guidelines for lung cancer surgery rely on expert consensus and scientific statements, lacking robust, graded evidence. This study aimed to provide an evidence-based evaluation of CPET variables for assessing perioperative morbidity and long-term mortality risks.

Methods

Electronic databases were searched through December 20, 2025, for cohort studies on CPET variables and lung cancer surgery. Quality was assessed using the Newcastle-Ottawa Scale. Heterogeneity was evaluated by the Q-test and I² statistic, with fixed or random effects models. Publication bias was assessed via funnel plots and Egger’s test, with adjustment using the trim-and-fill method when necessary. Effect sizes were pooled, and evidence was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.

Results

Of 3080 studies, 47 studies involving 11,356 surgeries and 13 CPET variables were included. Meta-analysis and GRADE assessment showed lower peak oxygen uptake (peak $\dot{V}$O₂) increased the risk of perioperative complications by 2.04–3.02 times (P < 0.001), using a cutoff point of 15 mL/kg/min or peak $\dot{V}$O₂ expressed as %pred. A higher minute ventilation to carbon dioxide production ($\dot{V}$E/$\dot{V}$CO₂) slope was associated with a 2.44-fold increased risk (95% CI [1.82–3.27], P < 0.001), with a cutoff point of 35 (high-level). Limited studies suggest peak $\dot{V}$O₂ (L/min) and $\dot{V}$E/$\dot{V}$CO₂ slope relate to cancer-specific and all-cause mortality. Evidence for variables such as $\dot{V}$O₂ at anaerobic threshold and peak ventilation remains low.

Conclusion

Among 13 CPET variables, only peak $\dot{V}$O₂ (15 mL/kg/min or %pred values) and $\dot{V}$E/$\dot{V}$CO₂ slope (35) are supported by high-certainty evidence for perioperative risk. Further studies are needed to establish CPET’s role in predicting cancer-specific mortality in lung cancer surgery.

Systematic review registration

PROSPERO CRD42023405083.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13643-026-03258-y.

Highlights

• Current guidelines show significant variability in recommending CPET variables and cutoff points for assessing perioperative morbidity risk in lung cancer surgery, highlighting the need for standardization.

• This study is the first to apply the GRADE classification, offering an evidence-based framework for selecting CPET variables to assess perioperative risk in patients undergoing lung cancer surgery.

• Among 13 commonly used CPET variables, peak O₂ (cutoff: 15 mL/kg/min or %pred values) and $\dot{V}$E/$\dot{V}$CO₂ slope (cutoff: 35) were supported by high-certainty evidence for perioperative risk assessment, while the remaining 11 variables show low- or very low-certainty evidence.

• Preliminary evidence suggests that peak O₂ (L/min) and E/CO₂ slope may correlate with cancer-specific mortality, highlighting potential areas for further research and clinical application.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13643-026-03258-y.

Introduction

Lung cancer remains a leading cause of cancer-related deaths worldwide [1], accounting for an estimated 1.8 million deaths annually. Surgery remains the most effective treatment for early-stage lung cancer. However, the significant risk of complications and the variability in long-term outcomes heavily influence surgical decision-making and treatment efficacy. Therefore, identifying high-risk patients through non-invasive preoperative testing is crucial for individualizing management strategies, reducing surgical risks, and improving prognosis.

Cardiopulmonary exercise testing (CPET) has emerged as a pivotal method for risk stratification in this context [2]. By simulating surgical stress through maximal exercise, CPET measures cardiorespiratory functional reserve and screens for asymptomatic cardiovascular conditions. This aids clinicians in predicting an individual’s capacity to withstand surgical stress and the risk of postoperative complications. CPET variables can be categorized into traditional cardiovascular stress test variables (e.g., exercise ECG, heart rate, blood pressure, and symptoms) [3] and ventilatory/gas exchange variables.

While ECG stress test variables have been relatively well-studied, recent research has increasingly focused on the relationship between gas exchange variables and perioperative complication risks as well as long-term outcomes in lung cancer patients [4, 5]. However, heterogeneous results have been observed in these studies, possibly arising from variations in study design, patient populations, and methodology. Additionally, current guidelines on applying CPET in lung cancer surgery are primarily based on expert consensus and position statements rather than rigorously graded evidence (Table 1) [2, 7, 10]. These gaps in high-certainty evidence undermine the utility and interpretation of CPET in lung cancer surgery.

Table 1.

Varied guideline recommendations on CPET gas exchange variables and cutoff points for perioperative complication assessment

Recommending body	Type	Recommended variable	Recommended cutoff point	Patients representativeness of referenced study a
JACS, 2025 [6]	Scientific reports	peak $\dot{V}$O2	10 mL/kg/min, 35%pred	Yes
POETTS, 2018 [7]	Expert consensus	peak $\dot{V}$O2, $\dot{V}$ O2@AT	No	Yes
EACPR/AHA, 2016 [2]	Scientific statement	peak $\dot{V}$O2	10 mL/kg/min, 16 mL/kg/min, 20 mL/kg/min	No (heart failure)
		$\dot{V}$E/$\dot{V}$CO2 slope	30, 36, 45
		$\dot{V}$O2@AT	11 mL/kg/min
ACC/AHA, 2014 [8]	Expert consensus	$\dot{V}$O2@AT	10 mL/kg/min	NA
ACCP, 2013 [9]	Scientific statement	peak $\dot{V}$O2	10 mL/kg/min, 20 mL/kg/min, 35%pred, 75%pred	Yes
EACPR/AHA, 2012 [10]	Scientific statement	peak $\dot{V}$O2	10 mL/kg/min, 16 mL/kg/min, 20 mL/kg/min	Yes
		$\dot{V}$E/$\dot{V}$CO2 slope	30, 36, 45
		PETCO2	33 mmHg
BTS/SCTS, 2010 [11]	Expert consensus	peak $\dot{V}$O2	10 mL/kg/min, 20 mL/kg/min, 40%pred, 75%pred	Yes
ERS/ESTS, 2009 [12]	Expert consensus	peak $\dot{V}$O2	10 mL/kg/min, 20 mL/kg/min, 35%pred, 75%pred	Yes
ACCP, 2007 [13]	Scientific reports	peak $\dot{V}$O2	10 mL/kg/min, 15 mL/kg/min, 20 mL/kg/min	Yes
ATS/ACCP, 2003 [14]	Scientific reports	peak $\dot{V}$O2	50%	Yes
BTS, 2001 [15]	Expert consensus	peak $\dot{V}$O2	15 mL/kg/min	Yes

ATS American Thoracic Society, ACCP American College of Chest Physicians, ACC American College of Cardiology, AHA American Heart Association, BTS British Thoracic Society, EACPR European Association for Cardiovascular Prevention & Rehabilitation Representative, ERS European Respiratory Society, ESTS European Society of Thoracic Surgeons, JACS The Japanese Association for Chest Surgery, SCTS Society for Cardiothoracic Surgery in Great Britain and Ireland, POETTS Perioperative Exercise Testing and Training Society, CPET cardiopulmonary exercise testing, peak $\dot{V}$ O₂ peak oxygen uptake, $\dot{V}$ E/ $\dot{V}$ CO₂ slope minute ventilation to carbon dioxide production slope, AT anaerobic threshold, PETCO₂, partial pressure of end-tidal carbon dioxide, NA not available

^aYes indicates that the participants were patients who either underwent or were undergoing lung surgery, ensuring a representative sample of this patient population. No indicates that the participants were not patients who underwent or were undergoing lung surgery

Accordingly, this systematic review and meta-analysis aimed to perform a comprehensive, evidence-based evaluation of CPET variables and cutoff points. We will use the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to evaluate the level of the evidence [16], thereby ensuring that our recommendations for the clinical application of CPET gas exchange variables in lung cancer surgery patients are based on the highest standards of evidence-based medicine.

Methods

Due to limited randomized controlled trials on the association between CPET and perioperative morbidity as well as long-term mortality risks in lung cancer surgery, we included cohort studies for this systematic review and meta-analysis. The study was designed to meet the high methodological standards of Assessing the Methodological Quality of Systematic Reviews 2 (AMSTAR2), was registered with PROSPERO (CRD42023405083), and was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [17, 18] and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines [19] (checklists in Tables S1-S3). Ethical approval and informed consent were not required because this study involved no clinical interventions.

Inclusion and exclusion criteria

Inclusion criteria

Studies were included if they (1) were prospective or retrospective cohort studies; (2) enrolled patients undergoing lung cancer surgery; (3) analyzed the association between pre-surgery CPET and perioperative complications or all-cause mortality, the CPETs performed immediately before surgery in all studies; and (4) reported odds ratios (OR) or hazard ratios (HR) or provided sufficient data to calculate these metrics. Perioperative complications included pulmonary or cardiovascular issues such as atelectasis, pneumonia, mechanical ventilation or tracheostomy requirement, myocardial infarction, heart failure, arrhythmia, stroke, empyema, wound infection, bronchial stump leakage, massive hemothorax, or death within 30 days post-surgery.

Exclusion criteria

Studies were excluded if they (1) failed to report CPET completion quality or reasons for non-completion; (2) had a follow-up period of less than one year for long-term all-cause mortality outcomes; (3) reported on overlapping participants across multiple studies by the same author, with preference given to the more recent or larger sample size study; (4) had low evidence quality, e.g., conference abstracts without relevant data, editorials, expert opinions, studies with fewer than 30 participants, reviews, or animal experiments; or (5) were not published in Chinese or English.

Search strategy

We searched six databases (PubMed, Web of Science, Embase, CNKI, Wanfang Data, and the Chinese Biomedical Literature Service System) from inception to December 20, 2025. Search terms followed were constructed using the PICOS (Population, Exposure/Intervention, Comparison, Outcome, and Study design) framework.

Population: “Lung Neoplasms,” “Carcinoma,” “Non-Small-Cell Lung,” “Pulmonary surgical procedures,” “Thoracic surgical procedures,” “Pneumonectomy.”

Intervention and Comparison: “Exercise Test,” “Exercise Tolerance.”

Outcome: “Morbidity,” “Mortality,” “Postoperative Complications,” “Perioperative Complications,” “Recurrence,” and prognostic terms from the McMaster University Health Information Research Unit. Study design: cohort study.

Additional searches were performed in relevant clinical trial registries. Full search strategies are detailed in Supplementary Table S4.

Literature screening, data extraction, and quality assessment

Studies were screened by title, abstract, and full text against the predefined criteria, including those from reference lists of relevant articles. Data extraction included author, publication year, nationality, study type, participant characteristics, sample size, study variables, complication types, OR or HR values, follow-up duration, and outcomes. Quality assessment was performed using the Newcastle–Ottawa Scale (NOS) [20], which evaluates selection, comparability, and outcome assessment. Three researchers independently carried out these procedures, resolving disagreements through discussion or, when necessary, by consulting a fourth researcher. Literature screening was performed using EndNote X9.

Statistical analysis

Outcomes with insufficient data or high heterogeneity were systematically reviewed without meta-analysis. For outcomes with adequate statistical power and lower heterogeneity, meta-analyses were performed using ORs derived from logistic regression and HRs from Cox proportional hazards models. When effect sizes were not reported but at least five events occurred in both the exposed and non-exposed groups, ORs and 95% CIs were calculated from available raw data. All ORs were computed as high-risk group/low-risk group; results using the low-risk group/high-risk group comparison were adjusted accordingly.

Heterogeneity was assessed using the Q-test (α = 0.1) and the I² statistic. Fixed-effect models were used for low heterogeneity (I² < 50%), while random-effects models were applied for high heterogeneity (I²> 50%). Sensitivity analyses identified studies affecting stability or heterogeneity, while subgroup analyses explored heterogeneity sources. For outcomes with ≥ 6 studies, funnel plots and Egger’s test were used to evaluate publication bias. If publication bias was detected, the trim-and-fill method estimated the number of missing studies and adjusted results. Stability was assumed if changes in the effect size were negligible after adjustments [21].

The quality of evidence for CPET gas exchange variables was appraised using the GRADE Evidence table of GRADEpro Guideline Development Tool (GDT) [22], considering study numbers, study design, risk of bias, inconsistency, indirectness, imprecision, and other considerations. To support the GRADE assessment, potential biases were specifically evaluated using the Risk of Bias in Non-randomized Studies – of Exposures (ROBINS-E) tool [23]. Statistical analyses were conducted using Review Manager 5.3, Stata 15.1, and R software, with the significance defined as two-sided 0.05 (two-sided).

Results

Literature search results

We identified 3080 relevant articles, with 47 studies (involving 11,356 lung cancer surgery patients across 14 countries) meeting the inclusion criteria. The selection process is illustrated in Fig. 1. The broad geographic distribution of included studies may enhance robustness and generalizability.

Fig. 1.

PRISMA flow diagram of the study selection process. The flowchart illustrates the identification, screening, and inclusion of studies from the initial search across multiple databases

Quality assessment

Among the 43 studies on perioperative complications [5, 24–65], 40 were rated as high-quality, according to the NOS (7–9 points), while 3 studies scored 6 points (Table S5). All 8 studies assessing long-term cancer-specific or all-cause mortality were rated as high quality (8–9 points) (Table S6) [5, 25, 33, 44, 66–69].

CPET variables and perioperative complications

Forty-three studies explored the relationship between CPET gas exchange variables and perioperative complications in lung cancer surgery [5, 24–65], comprising 19 prospective and 24 retrospective studies. These studies involved 10,138 patients from 14 countries, with a mean age range of 51 to 79 years. The proportion of males ranged from 41.1% to 100%, reflecting varied gender distributions across studies. The largest study included 1684 participants, while the smallest had 30. Four studies employed a treadmill for CPET, 33 used a cycle ergometer, and 6 did not specify the exercise modality. However, cutoff points measured on both the cycle and treadmill did not differ significantly, both included peak $\dot{V}$O₂ at 15 mL/kg/min, 20 mL/kg/min, and peak $\dot{V}$O₂ (%pred). All studies used symptom-limited exercise protocols, analyzing 13 CPET gas exchange variables: absolute peak $\dot{V}$O₂ in L/min, body weight-relative peak $\dot{V}$O₂ in mL/kg/min, percentage of predicted maximal values in %pred, minute ventilation/carbon dioxide production ($\dot{V}$E/$\dot{V}$CO₂) slope, oxygen uptake at anaerobic threshold ($\dot{V}$O₂@AT), minute ventilation/carbon dioxide production at anaerobic threshold($\dot{V}$E/$\dot{V}$CO₂@AT), peak oxygen pulse (peak O₂pulse), metabolic equivalents (METs), peak ventilation during exercise (peak $\dot{V}$E), respiratory rate (RR), partial pressure of end-tidal carbon dioxide production at anaerobic threshold (PETCO₂@AT), peak $\dot{V}$O₂ combined with $\dot{V}$E/$\dot{V}$CO₂ slope, and breathing reserve (BR). Thirty studies monitored complications within 30 days post-surgery, one study had a 21-day follow-up, two studies had a 7-day follow-up, one study had a 12-month follow-up, another had a 3-month follow-up, and 8 did not specify the follow-up duration. Detailed results are provided in Supplementary (Table S7).

Oxygen uptake ($\dot{V}$O₂)

Thirty two studies [24–28, 30–35, 37, 38, 40–45, 47, 49–51, 54, 56–63] analyzed the relationship between $\dot{V}$O₂ variables and perioperative complications in lung cancer surgery, including absolute peak $\dot{V}$O₂ in L/min, body weight-relative peak $\dot{V}$O₂ in mL/kg/min, percentage of predicted maximal values in %pred, $\dot{V}$ O₂@AT, peak O₂ pulse, and METs.

Our meta-analysis of 14 studies [24, 27, 30, 31, 34, 40, 42–45, 47, 57–59] indicated that patients with peak $\dot{V}$O₂ below 15 mL/kg/min had a 2.61 times higher risk of perioperative complications compared to those with peak $\dot{V}$O₂ greater than 15 mL/kg/min (95% CI [1.76–3.86], P < 0.001). Excluding 2 heterogeneous studies [27, 44], the risk increased to 3.02 times (95% CI [2.26–4.04], P < 0.001) (Fig. 2A). Funnel plot and Egger’s test indicated no publication bias (Egger’s test P = 0.228) (Figures S1, S2).

Fig. 2.

Forest plot of peak $\dot{V}$O₂ as a predictor of perioperative complications with different cutoff points. A ≥ 15 mL/kg/min vs. < 15 mL/kg/min. B  ≥ 20 mL/kg/min vs. < 20 mL/kg/min. C High peak $\dot{V}$O₂ (%pred) vs. Low $\dot{V}$O₂ (%pred). peak $\dot{V}$O₂, peak oxygen uptake

Our meta-analysis of 7 studies [28, 31, 33, 41, 42, 57, 62] found that patients with peak $\dot{V}$O₂ below 20 mL/kg/min had a 2.04 times higher risk of perioperative complications compared to those with peak $\dot{V}$O₂ greater than 20 mL/kg/min (95% CI [1.57–2.64], P < 0.001) (Fig. 2B). However, Egger’s test suggested publication bias (P = 0.008), and after correction by the trim-and-fill method, the risk was 1.86 times higher (95% CI [1.45–2.38], P < 0.001) (Figure S3).

Our meta-analysis of 9 studies [32, 37, 42, 43, 47, 56, 58, 61, 62] showed that patients with low peak $\dot{V}$O₂ (%pred) had significantly higher perioperative complication rates compared to those with high peak $\dot{V}$O₂ (%pred) (OR [95% CI], 2.53 [1.58–4.05], P < 0.05). After excluding one study [32], the result was 2.86 (95% CI [2.07–3.95], P < 0.001) (Fig. 2C), with no publication bias as indicated by funnel plot and Egger’s test (P = 0.244) (Figure S4, S5). The variation in cutoff values for peak $\dot{V}$O₂ (%pred) across studies—or the absence of cutoff values with reliance on continuous variable analysis—hinders the exploration of optimal thresholds and their subsequent clinical application.

The results for the relationship between changes in peak $\dot{V}$O₂ per unit variable and perioperative cardiopulmonary complications [25, 26, 31, 32, 35, 37, 41, 50, 51, 54, 60, 63] are presented in Table S8; the results for peak $\dot{V}$O₂@AT [35, 58], peak O₂ pulse [35, 49, 51, 58], and MET [40] are shown in Table S9.

$\dot{\mathit{V}}$E/CO₂ slope

Our meta-analysis of 11 studies [5, 25, 26, 29, 36, 39, 44, 48, 52, 53, 62] showed that patients with high $\dot{V}$E/$\dot{V}$CO₂ slope had a 2.32 times higher risk of complications compared to those with low $\dot{V}$E/$\dot{V}$CO₂ slope (95% CI [1.54–3.52], P < 0.001). After excluding 1 study [53], the result was 2.44 (95% CI [1.82–3.27], P < 0.001) (Fig. 3), with no publication bias as indicated by funnel plot and Egger’s test (P = 0.068). Funnel plots and sensitivity analysis plots are shown in Supplementary (Figure S6, S7). The commonly used cutoff point is 35 [25, 36, 52]. Additionally, 8 cohort studies [31, 46, 50, 54, 55, 60, 64, 65] found that $\dot{V}$E/$\dot{V}$CO₂-derived indicators were also significantly associated with perioperative complications risk (Table S10).

Fig. 3.

Forest plot of the association between $\dot{V}$E/$\dot{V}$CO₂ slope and perioperative complications. $\dot{V}$E/$\dot{V}$CO₂ slope, minute ventilation to carbon dioxide production slope

Other CPET variables

Several studies reported the associations between RR [58] and PETCO₂@AT [26] with perioperative complications, while BR [51] and peak $\dot{V}$O₂ combined with $\dot{V}$E/$\dot{V}$CO₂ slope [38] were not associated. The association between peak $\dot{V}$E [40, 50] and perioperative complications remains controversial. Specific results are shown in Table S11.

CPET variables and long-term cancer-specific and all-cause mortality

Eight studies explored the relationship between CPET gas exchange variables and long-term cancer-specific or all-cause mortality in lung cancer surgery patients [5, 25, 33, 44, 66–69], including 5 prospective and 3 retrospective studies. These involved a total of 3051 patients from 7 countries, with a mean age range from 57 to 69 years. The proportion of males ranged from 55 to 83%. Sample sizes ranged from 110 to 1169. CPET gas exchange variables studied included peak $\dot{V}$O₂ mL/kg/min, L/min, peak $\dot{V}$O₂ (%pred) values, and $\dot{V}$E/$\dot{V}$CO₂ slope. All studies employed symptom-limited exercise protocols, with 7 using a cycle ergometer and 1 not reporting the exercise mode. Further details are summarized in Table 2. Follow-up duration ranged from a median of 21.0 months to 69.2 months. The relationships between peak $\dot{V}$O₂ and $\dot{V}$E/$\dot{V}$CO₂ slope with long-term all-cause mortality risk in lung cancer surgery patients remain inconsistent across studies, and the studies showed large heterogeneity (Figure S8). Some studies identified a significant association between lower peak $\dot{V}$O₂ or higher $\dot{V}$E/$\dot{V}$CO₂ slope and increased all-cause mortality, while other studies have not observed this relationship (Table 2). Only two studies have specifically examined the association between CPET variables and cancer-specific mortality in this patient population. Lindenmann et al. reported that peak $\dot{V}$O₂ was not statistically associated with cancer-specific mortality [68]. In contrast, Dun et al. found that patients with a $\dot{V}$E/$\dot{V}$CO₂ slope ≥ 31 had a 1.38-fold increased risk of cancer recurrence or death compared to those with a $\dot{V}$E/$\dot{V}$CO₂ slope < 31 [5]. Details of included studies are provided in Table S12.

Table 2.

Association between CPET gas exchange variables and long-term cancer-specific and all-cause mortality in lung cancer surgery patients

Author	Year	Country	Sample size	Number of events	Research variables	Reference value	All-cause mortality HR (95% CI)	Cancer-specific mortality HR (95% CI)
Dun [33]	2023	China	895	156	peak $\dot{V}$O2 (mL/kg/min)	< 20.0	1.98 (1.31–2.98)*	NA
Jones [67]	2010	America	335	237	peak $\dot{V}$O2 (mL/kg/min)	< 13.9 vs > 17.3	1.16 (0.83–1.64)	NA
						< 13.9 vs 14.0–17.3	1.14 (0.83–1.56)	NA
Lindenmann [68]	2020	Australia	342	245	peak $\dot{V}$O2 (mL/kg/min)	NA	1.04 (1.01–1.08)*	1.03 (0.99–1.06)
Jones [67]	2010	America	335	237	peak $\dot{V}$O2 (L/min)	< 0.96 vs 0.96–1.29	1.52 (1.05–2.17)*	NA
						< 0.96 vs > 1.29	1.69 (1.12–2.50)*	NA
Win [69]	2008	Britain	110	59	peak $\dot{V}$O2 (L/min)	NA	1.08 (1.01–1.15)*	NA
Lindenmann [68]	2020	Australia	342	245	peak $\dot{V}$O2 (%pred)	NA	1.02 (1.01–1.03)*	1.01 (1.00–1.02)
Win [69]	2008	Britain	110	59	peak $\dot{V}$O2 (%pred)	NA	1.00 (0.99–1.01)	NA
Brunnelli [66]	2014	Italy	157	113	peak $\dot{V}$O2 (%pred)	< 60%	2.40 (95% CI, NA)*	NA
Bédat [25]	2022	Switzerland	145	19	$\dot{V}$E/$\dot{V}$CO2 slope	> 35	2.00 (0.50–7.90)	NA
						> 40	3.00 (0.80–10.80)
Dun [5]	2023	China	895	142	$\dot{V}$E/$\dot{V}$CO2 slope	> 31	1.69 (1.15–2.48)*	1.38(1.02–1.88)
Miyazaki [44]	2018	Britain	172	51	$\dot{V}$E/$\dot{V}$CO2 slope	> 40	1.05 (1.01–1.10)*	NA

CPET cardiopulmonary exercise testing, peak $\dot{V}$O₂ peak oxygen uptake, $\dot{V}$E/ $\dot{V}$CO₂ slope minute ventilation to carbon dioxide production slope. The asterisk “*” indicates P < 0.05

Evidence levels for CPET variables

Using the GRADE methodology, we assessed the evidence for 13 commonly used gas exchange variables in studies on perioperative complications and long-term all-cause mortality in lung cancer patients (Tables S13, S14). The evidence supporting peak $\dot{V}$O₂ (cutoff, 15 mL/kg/min or %pred values) and the $\dot{V}$E/$\dot{V}$CO₂ slope (cutoff, 35) for perioperative risk assessment is high. The evidence for peak $\dot{V}$ O₂ with a cutoff of 20 mL/kg/min is moderate. However, other commonly used variables such as peak O₂ pulse, $\dot{V}$O₂ @AT, PETCO₂@AT, BR, $\dot{V}$E/$\dot{V}$CO₂@AT, peak $\dot{V}$E, and peak RR are of very low quality and cannot be substantiated. The associations between these variables and long-term outcomes remain inconclusive and warrant further research. Additional details are provided in Fig. 4.

Fig. 4.

CPET gas exchange variables for assessing perioperative morbidity risk and long-term prognosis in lung cancer patients undergoing surgery. The figure categorizes CPET variables into high, moderate, and very low grades of evidence based on their predictive value for perioperative complications and long-term mortality. CPET, cardiopulmonary exercise testing; peak $\dot{V}$O₂, peak oxygen uptake; $\dot{V}$E/$\dot{V}$CO₂ slope, minute ventilation to carbon dioxide production slope; peak O₂ pulse, peak oxygen uptake divided by peak heart rate; METs, metabolic equivalents; AT, anaerobic threshold; peak $\dot{V}$E, peak ventilation during exercise; PETCO₂, partial pressure of end-tidal carbon dioxide. *The differences in the equations used to predict peak $\dot{V}$O₂ across various racial groups have made it challenging to establish optimal cutoff points

Discussion

This systematic review and meta-analysis, conducted using the GRADE approach, evaluated the current evidence on CPET gas exchange variables association with perioperative morbidity and long-term mortality risks in lung cancer surgery. Our findings indicate that, among 13 commonly used CPET variables, only peak $\dot{V}$O₂ in mL/kg/min, peak $\dot{V}$O₂(%pred) values, and $\dot{V}$E/$\dot{V}$CO₂ slope are supported by high-certainty evidence, respectively, for perioperative risk assessment. Although specific cutoff values varied across studies, a threshold of 15 mL/kg/min for peak $\dot{V}$O₂ and 35 for the $\dot{V}$E/$\dot{V}$ CO₂ slope emerged as clinically relevant. In contrast, the remaining 11 variables demonstrated low or very low evidence levels.

Peak $\dot{V}$O₂ is a widely recognized key measure of cardiorespiratory fitness and is recommended as a vital sign by the American Heart Association [70]. In patients undergoing lung cancer surgery, a higher peak $\dot{V}$O₂ indicates better physiological reserve, a greater capacity to withstand surgical stress, and a lower likelihood of asymptomatic cardiovascular diseases [2], correlating with a lower risk of perioperative complications. Although several studies have proposed thresholds of 16 mL/kg/min [26], 17 mL/kg/min [41], and 19 mL/kg/min [37], or distinct unit-based cutoffs [25, 31, 35, 50, 51, 60, 63], the most commonly recommended thresholds in clinical guidelines are 10, 15, and 20 mL/kg/min, or peak $\dot{V}$O₂ (%pred) [28, 31, 33, 41, 42, 57, 62]. Derived indices such as peak O₂ pulse, $\dot{V}$O₂@AT, and METs have also been associated with perioperative complications in lung cancer surgery [35, 40, 49, 51, 58]. Our meta-analysis, supported by GRADE assessment, confirms a strong association between peak $\dot{V}$O₂ in mL/kg/min and perioperative morbidity risk.

Despite strong evidence supporting the association between the percentage of predicted maximal $\dot{V}$O₂ with preoperative risk, variations in maximal $\dot{V}$O₂ and prediction equations across different racial groups [71] have prevented the identification of universally applicable cutoff points. Future multinational studies with prospective cohort designs are warranted to harmonize prediction formulas and to define appropriate thresholds for diverse populations.

A steeper $\dot{V}$E/$\dot{V}$CO₂ slope indicates poorer ventilatory efficiency, often associated with increased dead space ventilation, ventilation-perfusion mismatch, and diminished pulmonary reserve [72]. Existing studies have consistently demonstrated the association between the $\dot{V}$E/$\dot{V}$CO₂ slope and perioperative morbidity in lung cancer surgery [5]. However, there is a lack of robust evidence to define precise thresholds for clinical use. This study recommends a cutoff of 35 for the $\dot{V}$E/$\dot{V}$CO₂ slope, based on its frequency of use in previous research [25, 36].

Although the association between peak $\dot{V}$O₂ and long-term prognosis has been firmly established in various populations [73], and the $\dot{V}$E/$\dot{V}$CO₂ slope has been validated as a prognostic marker in patients with heart failure [72], limited studies have reported the prognostic value of these variables in lung cancer surgery patients. Among the few available studies, two studies with small sample sizes have linked peak $\dot{V}$O₂ in L/min to all-cause mortality [67, 69]. For instance, Dun et al. [5] conducted a prospective observational study and reported that a $\dot{V}$E/$\dot{V}$CO₂ slope ≥ 31 was associated with a 1.38-fold higher risk of reduced relapse-free survival compared to a lower $\dot{V}$E/$\dot{V}$CO₂ slope. These findings suggest that peak $\dot{V}$O₂ and $\dot{V}$E/$\dot{V}$CO₂ slope may be associated with long-term outcomes in postoperative lung cancer patients. However, the current evidence is insufficient to draw definitive conclusions. More high-quality prospective studies are urgently needed to determine if these variables can aid in assessing long-term prognosis, such as cancer-specific mortality in lung cancer surgery patients.

The timing and application of CPET in routine clinical practice for lung cancer surgery patients remain areas of interest. Currently, CPET is not a standard preoperative assessment for lung cancer patients but is used as a supplementary tool when abnormalities are found in static lung function tests, ECGs, or other surgical risk assessments. Future research should focus on integrating CPET metrics with other clinical test results to enhance the accuracy of assessments related to surgical complications and long-term prognoses. The development of machine learning and multimodal prediction models offer a methodological foundation for combining CPET with additional clinical tests. Incorporating CPET into preoperative assessment protocols may enhance surgical outcomes by facilitating personalized treatment plans. High-risk patients identified via CPET could benefit from prehabilitation (exercise training to improve cardiorespiratory fitness preoperatively), optimized perioperative care, and vigilant postoperative monitoring, potentially reducing complication rates and improving long-term survival.

While a recent systematic review by Arbee-Kalidas et al. established a general association between CPET variables and postoperative outcomes in lung cancer surgery [74], our study advances the field by bridging the gap between statistical association and clinical application. First, by updating the literature search to December 2025, we synthesized data from 47 studies, providing the most current evidence base available. Second, unlike prior works, we stratified risk using specific, clinically relevant cutoff values (specifically 15 mL/kg/min and 20 mL/kg/min for peak V̇O₂ and specific thresholds for V̇E/V̇CO₂ slope). This granular analysis offers surgeons precise actionable benchmarks rather than broad associations. Finally, we are the first to apply GRADE classifications to this specific body of evidence, offering a rigorous framework to support future clinical guidelines regarding perioperative risk stratification and long-term cancer-specific survival.

This study has several limitations. First, heterogeneity was observed in the meta-analysis due to differences in the definitions of exposure and outcome measures across studies. Although sensitivity and subgroup analyses, along with trim-and-fill methods were employed to address some biases, residual errors may persist. Second, differences in exercise modalities may have contributed to further heterogeneity, given that treadmill exercise engages additional muscle groups that are not typically recruited during cycle ergometry. Third, limited studies on certain variables prevented meta-analysis, highlighting the need for large-sample, high-quality prospective research to assess their prognostic value. Finally, excluding non-Chinese and non-English studies may introduce bias, though the most relevant cohort studies were published in English.

Conclusions

This study supports the use of peak $\dot{V}$O₂ with risk stratification thresholds set at 15 mL/kg/min, %pred and $\dot{V}$E/$\dot{V}$CO₂ slope with a cutoff of 35 as primary variables for perioperative risk assessment in lung cancer surgery. Other commonly used clinical CPET variables currently lack sufficient supporting evidence. Future research is needed to determine the association between CPET variables and cancer-specific as well as all-cause mortality in this patient population.

Authors’ contributions

Yaoshan Dun: conceptualization, methodology, writing—original draft, writing—review and editing, project administration, funding acquisition. Jialin Yu: investigation, data curation, formal analysis, writing—original draft, visualization. Yang Gao: conceptualization, writing—review and editing. Randal J. Thomas: conceptualization, investigation, writing—review and editing. Ray W. Squires: investigation, writing—review and editing. Thomas P. Olson: investigation, writing—review and editing. Wenliang Zhang: investigation, writing—review and editing. Shaoping Wu: investigation, data curation, formal analysis, writing—original draft. Yang Du: investigation, writing—review and editing, funding acquisition. Suixin Liu: conceptualization, writing—review and editing, supervision, funding acquisition. All authors contributed to the review and revision and have seen and approved the final version of the manuscript.

Funding

This work was funded by the National Natural Science Foundation of China (82372576 to Suixin Liu, 82272613 to Yaoshan Dun), and the Natural Science Foundation of Hunan Province (2022JJ40777 to Yang Du). GanPo Talents Program-Leading Talents in Innovation (gpyc20250228 to Yaoshan Dun); Health Research Project of Hunan Provincial Health Commission (20257661 to Yaoshan Dun). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Declarations

Ethics approval and consent to participate

Ethical approval is not required because this study did not involve patients and clinical interventions.

Consent for publication

All authors have read and approved the final version of the manuscript and consented to its publication in Systematic Reviews.

Competing interests

The authors declare that they have no competing interests.