Frailty and Pre-Frailty in Patients With Lung Cancer and Its Association With Long-Term MACCE: A Longitudinal Cohort Study.

1
Introduction
Lung and bronchus (hereinafter lung) cancer represents a significant global disease burden, with an estimated 1.8 million deaths in 2020 [1, 2]. According to a population‐based registry, the prevalence of cardiovascular co‐morbidity among lung cancer (LC) patients is about twice as high as in the general population [3]. With improved cancer treatments and a rising survivorship population, which is estimated to be 26 million by 2040, [4] it is critical to manage cardiovascular comorbidities in patients with cancer, especially in the elderly individuals [5].
Frailty, an emerging public health concern worldwide paralleled with population aging, is characterized by a decline in functioning across multiple physiological systems, with a resultant increased susceptibility to stressors [6, 7]. Frailty is an important shared risk factor for both cancer and cardiovascular disease [8]. The prevalence of frailty in LC was45% [9]. In patients with LC, frailty was associated with a 3‐fold increased risk for mortality [9]. While the relationship between frailty and mortality in patients with cancer is well documented, the long‐term cardiovascular consequences, specifically the risk of major adverse cardiac and cerebrovascular events (MACCE), remain poorly understood in LC patients. Pre‐frailty is a condition characterized by the presence of some but not all features of frailty, representing an intermediate stage in the continuum from health to frailty [10]. Pre‐frailty is considered a critical stage for early intervention, as it is associated with an increased risk of progression to frailty and mortality, especially for older patients with cancer [11]. Moreover, the role of pre‐frailty as a precursor to frailty and its potential impact on cardiovascular risk in this population has yet to be explored in depth.
In this study, we utilized the UK Biobank, a large nationwide cohort study, to further investigate the association between frailty and pre‐frailty with long‐term MACCE in LC patients. By identifying frailty as a risk factor for MACCE in LC patients, this study has the potential to inform clinical practice, leading to improved risk stratification and better long‐term outcomes for LC patients at risk for cardiovascular events.

2
Methods
2.1
Study Group
This is a multicenter, long‐time‐span cohort study utilizing the UK Biobank database. The UK Biobank recruited over 500,000 participants, with an average age at recruitment of 56.5 years (interquartile range (IQR): 50.0 to 63.0) [12]. Participants were recruited from various regions across the United Kingdom, encompassing diverse ethnic backgrounds. The UK Biobank assessed baseline characteristics and tracked participants' primary care records, hospital admissions, cancer records, and mortality records. We obtained data from the UK Biobank online platform in September 2024, with baseline assessment between 2006 and 2010 and follow‐up until October 31, 2022. Out of the 502,173 participants, we excluded those marked as lost to follow‐up (n = 1297) and those who had LC prior to recruitment (n = 346). Overall, this study included 6095 who were diagnosed with LC over the follow‐up period (Figure 1A).

2.2
Ascertainment of LC and MACCE
The diagnosis of LC was confirmed using ICD‐9 (International Classification of Diseases, 9th Revision) and ICD‐10 (International Classification of Diseases, 10th Revision) codes [13] (Table S3).
We defined MACCE as the outcome, which was a composite outcome encompassing any of the following events: all‐cause mortality, HF, respiratory failure (RF), stroke, shock, and coronary heart disease (CHD). HF, RF, stroke, shock, and CHD were classified using ICD‐10 codes (Table S8). All‐cause mortality was obtained from national death registries. The cut‐off date for updates to the outcomes was October 31, 2022.

2.3
Ascertainment of Frailty and Pre‐Frailty
The original conceptualization of the frailty phenotype was articulated and implemented in the Cardiovascular Health Study by Fried and colleagues [14]. This framework has since been adapted for use in the UK Biobank [15, 16]. In this adaptation, weight loss was determined through self‐reporting via the question: “Compared with one year ago, has your weight changed?” with responses coded as “yes, lost weight” = 1 and all other responses = 0. Exhaustion was similarly self‐reported using the question: “Over the past two weeks, how often have you felt tired or had little energy?” Responses indicating “more than half the days or nearly every day” were coded as 1, with all other responses coded as 0. Physical activity was quantified using total metabolic equivalent task (MET) minutes per week, encompassing the sum of walking, moderate, and vigorous activities.
Participants were categorized into quintiles based on sex‐ and age‐specific total MET minutes per week, with the lowest quintile classified as indicative of “low physical activity.”
Gait speed was assessed through self‐report with the question: “How would you describe your usual walking pace?” where ‘slow’ was coded as 1, and all other responses as 0. Hand grip strength was measured using a Jamar J00105 hydraulic hand dynamometer, with results expressed in kilograms adjusted for sex and BMI. Cutoff points for grip strength were aligned with those established by Fried and colleagues [14].
Participants were categorized as frail if they met three or more criteria (weight loss, exhaustion, low physical activity, slow gait speed, weak grip strength), pre‐frail if they met one or two criteria, and non‐frail if they met none.

2.4
Ascertainment of Other Covariates
We chose sex, age, ethnic background, family cancer history, smoking status, Townsend deprivation index, obesity, diabetes, hypertension, chronic obstructive pulmonary disease (COPD), and histology classification as they are potential confounders in the frailty and MACCE association or prognostic factors for MACCE among LC participants [17]. Ethnic background is a self‐reported variable, including White, Black and Black mixed, Asian and Asian mixed, and other backgrounds. The study population predominantly consists of individuals with a White ethnic background, while other ethnic backgrounds are more complex and account for a very small proportion of the total population (< 5%). Thus, in Cox regression analysis, we categorized “Black and Black Mixed” and “Asian and Asian Mixed” as “Other”. Body mass index (BMI) was constructed based on height and weight, which were measured during the initial assessment center visit. Some participants' BMI was measured multiple times, and we used the data from the latest visit to the assessment centers. BMI greater than or equal to 30 was defined as obesity. The Townsend deprivation index was used to measure the degree of material deprivation in the population [18]. This index was calculated prior to the recruitment, and each participant is assigned a score corresponding to the output area in which their postcode is located. A larger Townsend deprivation index means a higher degree of material deprivation. The indicator of participants' educational attainment is based on the self‐reported level of academic education and professional skills. If participants provided multiple data points, the most recent data were used. For specific classification methods, please refer to the section on educational attainment classification in Table S1.
Lifestyle‐related indicators include physical activity, smoking status, and alcohol consumption frequency. Regarding physical activity, we used the Summed MET (Metabolic Equivalent Task) minutes, which measure participants' weekly physical activity, covering walking, moderate, and vigorous activities. An average was used if multiple values during follow‐up were provided. Smoking status and alcohol consumption frequency were based on the most recent self‐reported data during participants' visits to the assessment centers.
Tumor histology information was obtained from the cancer registry using the ICD‐O‐3 (International Classification of Diseases for Oncology, 3rd Edition) histology code. For specific tumor classification methods, please refer to the section on tumor histology classification in Table S2. The UK Biobank does not directly provide the stage of LC at initial diagnosis. Tumor stages were assigned based on pathological results, lymph node metastasis, and treatment methods at the time of the initial diagnosis of LC, following the consensus recommendations of the American Cancer Society [19].

2.5
Statistical Analysis
Categorical variables were summarized as frequencies and percentages, while continuous variables were expressed as mean ± standard deviation. The normality of continuous data was evaluated using quantile‐quantile plots. In cases where the data deviated from a normal distribution, the interquartile range was reported. To compare means of continuous variables between two groups, t‐tests or Wilcoxon rank‐sum tests were employed. For categorical variables, chi‐square tests or Fisher's exact tests were utilized. Comparisons of continuous variables across multiple groups were conducted using ANOVA or Kruskal–Wallis tests, while chi‐square tests or Fisher's exact tests were applied for categorical variables. Cox regression analysis was performed to estimate the hazard ratio of frailty on MACCE, adjusting for sex, age, obesity, family history of cancer, smoking status, Townsend deprivation index, diabetes, hypertension, COPD, tumor type, physical activity, and alcohol consumption. Nearest neighbor propensity score matching (PSM) was performed within each of the study groups. Variables that each cohort was matched included sex, age, family cancer history, smoking status, Townsend deprivation index, obesity, diabetes, hypertension, pathological subtype, and COPD. Continuous variables were rounded to two decimal places, and p‐values were rounded to three decimal places. Two‐sided p‐values < 0.05 were considered statistically significant. All statistical analyses were conducted using R Project Software (version 4.2.2, The R Foundation).

3
Results
3.1
Baseline Characteristics
The study included a total of 6095 participants, stratified into three groups based on frailty status: non‐frail (n = 2669, 43.79%), pre‐frail (n = 2938, 48.20%), and frail (n = 488, 8.01%). Table 1 summarizes the baseline characteristics of the study population. The mean age at recruitment was similar across the three groups, with no statistically significant differences observed (non‐frail: 61.50; pre‐frail: 61.46; frail: 61.38, p = 0.912). In contrast, the mean age at diagnosis of LC showed significant differences between certain groups. While there was no statistical difference between the non‐frail and pre‐frail groups (69.61 vs. 69.28, p = 0.081), the frail group was diagnosed at a significantly younger age compared to the non‐frail group (68.75 vs. 69.61, p = 0.007). The frail group had a higher proportion of females compared to the non‐frail group (54.30% vs. 46.87%, p = 0.001). White ethnicity predominated across all groups, but the frail group exhibited a slightly higher proportion of participants from Black and Asian ethnicities. Frail participants also had higher rates of obesity (44.88%, p < 0.001) compared to the non‐frail and pre‐frail groups. Frail participants were more socioeconomically disadvantaged, as indicated by a higher Townsend deprivation index (p < 0.001), and had lower levels of physical activity (p < 0.001) compared to pre‐frail and non‐frail participants.
Comorbidities were more prevalent in frail individuals compared to the other groups. Diabetes was present in 26.84% of frail participants, compared to 17.63% in the pre‐frail group and 9.78% in the non‐frail group (p < 0.001). Hypertension affected 58.61% of frail individuals, compared to 47.96% in the pre‐frail group and 42.04% in the non‐frail group (p < 0.001). COPD was also notably more common in the frail group (45.49%, p < 0.001) compared to the pre‐frail and non‐frail groups.

3.2
Outcomes
Table 2 outlines the outcomes and cancer‐related treatments among the study population. Frail participants were less likely to undergo surgery (21.93%, p = 0.005) and chemotherapy (28.69%, p < 0.001) compared to their non‐frail and pre‐frail counterparts. However, the rates of radiotherapy were similar across all groups (approximately 12%). Furthermore, we conducted a series of sensitivity analyses to test the robustness of our findings, including stratified analyses by sex, smoking status, pathological subtype, age, obesity, and excluding newly diagnosed lung cancer patients in the first year (Tables 3, 4, 5, 6, 7, 8).
Frail participants had the highest incidence of MACCE, with 81.35% experiencing at least one event, compared to 74.47% in the pre‐frail and 71.56% in the non‐frail groups (p < 0.001). Frail individuals also demonstrated significantly higher all‐cause mortality (77.25%) compared to pre‐frail (69.91%) and non‐frail participants (67.48%) (Table 9).

3.3
Kaplan–Meier Survival Analysis
Figure 1B displays Kaplan–Meier survival curves for mortality and MACCE stratified by frailty status. Frail participants had significantly worse survival outcomes compared to the non‐frail group, with a reduction in survival rates over the 10‐year follow‐up period. For mortality, frail individuals demonstrated significantly lower survival (p < 0.001), while pre‐frail individuals also exhibited a trend toward poorer survival, although the difference was not statistically significant (p = 0.128). Similarly, frail participants had significantly lower survival from MACCE (p < 0.001), while pre‐frail individuals also showed reduced survival compared to non‐frail participants (p = 0.019). Furthermore, we compared long‐term MACCE after PSM, and the results showed the frail participants had significantly lower survival from MACCE (p = 0.001) after balancing the baseline characteristics across frailty groups (Figure 2).

3.4
Predictors of MACCE
Cox regression analysis (Figure 3) identified frailty as an independent predictor of MACCE. Being frail was associated with a 21% higher hazard for MACCE (HR = 1.21, 95% CI: 1.07–1.38, p = 0.002), while pre‐frailty conferred a 10% increased hazard (HR = 1.10, 95% CI: 1.02–1.18, p = 0.010). Smoking status also emerged as a significant predictor, with current smokers (HR = 1.47, 95% CI: 1.31–1.65, p < 0.001) and ex‐smokers (HR = 1.24, 95% CI: 1.11–1.38, p < 0.001) having higher risks compared to non‐smokers. Female (HR = 0.76, 95% CI: 0.71–0.81, p < 0.001) and a diagnosis of NSCLC (HR = 0.67, 95% CI: 0.61–0.75, p < 0.001) were associated with lower risks of MACCE.

3.5
Frailty Components and Outcomes
Figure 4 presents the associations between individual frailty components and various clinical outcomes. Among these components, low physical activity and slow gait speed were consistently associated with increased risks of adverse outcomes. Low physical activity was associated with a 16% higher risk of MACCE (HR = 1.16, 95% CI: 1.05–1.27, p = 0.002), a 17% higher risk of mortality (HR = 1.17, 95% CI: 1.06–1.29, p = 0.001), and a 61% higher risk of stroke (HR = 1.61, 95% CI: 1.06–2.44, p = 0.025). Similarly, slow gait speed was associated with a 68% higher risk of HF (HR = 1.68, 95% CI: 1.12–2.50, p = 0.012) and a nearly 5‐fold higher risk of shock (HR = 4.99, 95% CI: 1.32–18.89, p = 0.018).
Other frailty components, such as weight loss, exhaustion, and low grip strength, demonstrated weaker or statistically insignificant associations with most outcomes, indicating variability in the prognostic value of these components. Notably, slow gait speed and low physical activity were more strongly associated with cardiovascular, cerebrovascular, and respiratory outcomes compared to other components.

4
Discussion
Based on nationwide cohort studies of half a million UK Biobank participants, 6095 participants were diagnosed with LC. The prevalence of pre‐frailty and frailty, based on a widely used Fried frailty phenotype, was 48.20% and 8.01%, respectively. We found that frailty and pre‐frailty were independently associated with an increased risk of long‐term MACCE in LC patients. Specifically, frail patients exhibited a significantly higher risk of MACCE compared to non‐frail participants, while pre‐frail patients also showed an elevated but somewhat lesser risk. Additionally, specific components of frailty, notably low physical activity and slow gait speed, were robust predictors of adverse outcomes, underscoring the multifaceted nature of frailty in this population.
Previous studies have established the association between frailty and poor outcomes in cancer patients [11] particularly in terms of treatment (chemotherapy, radiotherapy, and surgery) tolerance [20, 21, 22] and survival [9, 23, 24, 25] However, the relationship between frailty and long‐term cardiovascular outcomes in LC patients has been less well explored. Our findings demonstrate that both frailty and pre‐frailty are independently associated with an increased risk of MACCE in LC patients. Specifically, frail individuals exhibited a 21% higher hazard (HR = 1.21, 95% CI: 1.07–1.38, p = 0.002) and pre‐frail individuals a 10% higher hazard (HR = 1.10, 95% CI: 1.02–1.18, p = 0.010) compared to non‐frail counterparts. This association persists even after adjusting for potential confounders, including demographic factors, comorbidities, and lifestyle indicators. The elevated risk of MACCE in frail and pre‐frail patients may be attributable to the cumulative physiological decline and the presence of multiple comorbid conditions, such as diabetes, hypertension, and COPD, which were more prevalent in these groups.
The high prevalence of pre‐frailty (48.20%) observed herein corroborates findings from other cohorts, emphasizing the importance of early identification in at risk populations [26]. The high prevalence of pre‐frailty is particularly noteworthy, as it represents a critical intermediate state that precedes full‐fledged frailty. Pre‐frailty, characterized by the presence of one or two frailty criteria, indicates early physiological decline and increased vulnerability, making it an essential target for early intervention strategies. The identification of nearly half of the LC patient cohort as pre‐frail underscores the imperative need for routine screening and proactive management to prevent progression to frailty, which is associated with more severe adverse outcomes in LC patients.
Among the frailty components assessed, low physical activity and slow gait speed emerged as significant predictors of both MACCE and all‐cause mortality, whereas low grip strength did not demonstrate a significant association. Low physical activity was associated with a 16% higher risk of MACCE (HR = 1.16, 95% CI: 1.05–1.27, p = 0.002) and a 17% higher risk of mortality (HR = 1.17, 95% CI: 1.06–1.29, p = 0.001). Similarly, slow gait speed was linked to a 68% higher risk of heart failure (HR = 1.68, 95% CI: 1.12–2.50, p = 0.012) and a nearly fivefold increase in the risk of shock (HR = 4.99, 95% CI: 1.32–18.89, p = 0.018). These findings suggest that certain aspects of physical function are more closely related to cardiovascular and overall health outcomes in LC patients. Low physical activity may contribute to cardiovascular deconditioning and exacerbate underlying heart disease, while slow gait speed may reflect broader systemic frailty and reduced physiological reserve [27, 28, 29]. The lack of significant association with low grip strength may indicate that muscle strength alone is insufficient to capture the multifaceted nature of frailty in the context of LC, where other factors such as endurance and mobility play more critical roles in determining patient outcomes.
4.1
Strengths and Limitations
This study has several strengths. Firstly, as the first nationwide, large‐sample‐sized cohort focusing on the long‐term prognosis of LC patients without frailty, with pre‐frailty, and with frailty, the results and conclusions of this study could represent a broader population with better generalizability instead of being limited to clinical settings from retrospective, single‐center, or small‐sample‐sized studies from specialized medical centers. While the large sample size of UK Biobank enhances statistical power, the healthy volunteer bias and limited diversity of its participants may constrain the external validity, and thus the generalizability of our findings should be interpreted with caution. Secondly, owing to the long time span follow‐up, we could evaluate the long‐term prognosis of LC patients and provide important evidence for long‐term cardiovascular outcomes among LC patients without frailty, with pre‐frailty, and with frailty. Thirdly, we utilized baseline characteristics, socioeconomic factors, lifestyle factors, disease history, cancer histologic subtypes, and cancer‐related treatments, aiming to provide a comprehensive and longitudinal perspective on the impact of pre‐frailty and frailty on LC patients by comparing LC participants without frailty, with pre‐frailty, and with frailty.
This study has several limitations. Firstly, due to the long time span of this cohort and rapid development in clinical practice, there will inevitably be heterogeneity in diagnostic and therapeutic approaches for LC and frailty. Secondly, information on socioeconomic level and lifestyle factors (including physical activity, smoking status, and frequency of alcohol intake) was mainly self‐reported. Most of the indicators were measured only once, but the reality is that the values of these indicators may change over time. Thirdly, although the physical activity component was operationalized using self‐reported questionnaires in UK Biobank, which allows large‐scale assessment of habitual activity, it may be less objective than the original Fried criteria that relied on standardized activity questionnaires and direct estimates of energy expenditure. Finally, although we controlled for key individual characteristics and comorbidities, residual confounding was still possible.

5
Conclusion
To conclude, LC participants had a high prevalence of pre‐frailty and frailty. The presence of frailty and pre‐frailty significantly increased the risk of MACCE in long‐term LC survivors. Notably, slow gait speed and low physical activity were strongly associated with MACCE compared to other frailty components. The results emphasize the essential need for routine frailty assessments in LC patients to enable effective risk stratification, thereby enhancing both cardiovascular and overall outcomes in this vulnerable population and improving survivorship in lung cancer care.

Author Contributions

Fang Zhu: funding acquisition, writing – original draft, visualization, software, data curation, formal analysis, methodology, validation, investigation, conceptualization. Qian Zhang: funding acquisition, writing – original draft, visualization, writing – review and editing, formal analysis, project administration. Hang Hao: visualization, data curation, investigation, writing – review and editing. Wenjie Cai: formal analysis, writing – review and editing. Qingquan Luo: supervision, resources, project administration, writing – review and editing.

Funding
This study was funded by the National Natural Science Foundation of China (82400483, 82100409).

Ethics Statement
The UK Biobank has obtained ethics approval from the North West Multi‐Centre Research Ethics Committee (MREC), which covers the entire project. Participants in the UK Biobank provided informed consent at the time of enrollment.

Conflicts of Interest
The authors declare no conflicts of interest.

Supporting information

Table S1: Education category.

Table S2: Neoplasm category.

Table S3: Lung and/or bronchus malignant neoplasm.

Table S4: Diabetes.

Table S5: Hypertension.

Table S6: Chronic obstructive pulmonary disease.

Table S7: Metastasis.

Table S8: Outcome.