Plain English summary

Patients with lung cancer undergoing chemotherapy often face a diminished Health-related Quality of Life (HRQoL) and a substantial symptom burden. While prior research predominantly concentrated on individual symptoms and overall quality of life, symptoms tend to cluster, thereby amplifying their impact on HRQoL. Given the multifaceted nature of HRQoL, it is imperative to conduct a comprehensive analysis of each domain, as an aggregate score may fail to capture the complete picture. Consequently, this study aimed to identify symptom clusters in lung cancer patients undergoing chemotherapy and explore their correlation with HRQoL. Three symptom clusters were identified: Somato-psychological (encompassing pain, fatigue, sleep disturbance, distress, drowsiness, and sadness), Respiratory (involving shortness of breath, coughing, expectoration and chest tightness), and Gastrointestinal (including nausea, poor appetite, vomiting, constipation, and weight loss). The Somato-psychological SC and Gastrointestinal SC negatively impact patients' HRQoL. Health care providers should prioritize these SCs to identify high-risk patients early and implement targeted preventive and intervention measures for each SC.

Introduction

Lung cancer is a significant contributor to global cancer incidence and cancer-related mortality, with 2,206,771 new patients with lung cancer and 1,796,9144 deaths in 2020 [1]. It has the highest incidence and death rate among men and shows a persistent increasing trend [2]. Owing to its heigh incidence rates and substantial number of patients, lung cancer has become a prominent public health challenge [3]. Lung cancer patients lack distinct features in their early stages, making them easily overlooked. Therefore, it is often diagnosed at advanced stages, and is accompanied by severe symptom burden [4]. In addition to the severity of disease characteristics, treatment also plays a crucial role. While immunotherapy and targeted therapy are emerging, chemotherapy remains the mainstream treatment [5]. Chemotherapy, a nonspecific chemotherapeutic drug, not only inhibits the growth of tumor cells but also affects normal cells, causing a range of side effects, such as pain, fatigue, vomiting, and numbness [6].

When Miaskowski and Dodd introduced the concept of symptom clusters (SCs), medical professionals gradually recognized that symptoms frequently did not manifest in isolation but rather presented as clusters, multiplying their impact on patients [7, 8]. Previous studies have identified several SCs. In patients with lung cancer undergoing chemotherapy, the number of SCs ranges from three to five. Common SCs include the somatic SC (mainly including pain, fatigue, and sleep disturbance), the psychological SC (primarily consisting of sadness, distress, and feeling nervous), and the respiratory SC (also known as the lung cancer-specific SC, comprising coughing, expectoration, chest tightness, hemoptysis, and shortness of breath) [9,10,11,12,13,14,15]. In our previous study of patients receiving chemotherapy, we also identified respiratory SC, gastrointestinal SC, and somatic SC [9, 15]. In patients receiving immunotherapy, Yang [16] identified five SCs: lung cancer-related, emotional-related, physical, skin, and neural SCs. In patients after CT-guided microwave ablation (MWA), Liu et al. [17] measured symptoms at 24, 48, and 72 h post-MWA. They identified four SCs: MWA-related, somatic, respiratory, and gastrointestinal SCs. These SCs persisted across the three time points, although the specific symptoms within each cluster varied slightly. These uncomfortable SCs directly affect patients' HRQoL [18, 19]. Different treatment methods for patients can lead to variations in the identified SCs. A gold standard for identifying SCs has not yet been established, and larger sample studies are needed to confirm the classification of SCs in lung cancer patients undergoing chemotherapy.

Health-related quality of life (HRQoL) encompasses a comprehensive concept, including physical health, psychological state, level of independence, social relationships, and relationships with essential elements in the environment [20]. It has the potential to predict disease prognoses and patient mortality [21]. Therefore, understanding the HRQoL of patients is crucial for assessing the overall impact of treatments and care interventions. The HRQoL of patients with lung cancer is generally lower than that of patients with other malignancies [22, 23]. Some studies have explored the relationships between symptom clusters and HRQoL in lung cancer patients, and reported that symptom clusters negatively impact HRQoL during chemotherapy, surgery, immunotherapy, and targeted therapy [11, 16, 24,25,26,27]. Although the above studies revealed that SCs might reduce HRQoL, they all used exploratory factor analysis to identify SCs, which does not account for the interrelationships between symptoms. Moreover, the sample sizes for these studies were relatively small. Therefore, this study aims to (1) use network analysis to identify SCs in lung cancer patients undergoing chemotherapy and (2) analyze the impact of each SC on HRQoL and each of its functions.

Methods

Study design, patients, and procedures

The study was a cross-sectional study. We recruited lung cancer patients undergoing chemotherapy through convenience sampling at a tertiary hospital in Guangdong Province, China, from March to December 2023. The inclusion criteria for patients with lung cancer undergoing chemotherapy were as follows: (1) had a pathological diagnosis of lung cancer; (2) had a treatment regimen that included chemotherapy, and at least one chemotherapy session was completed; (3) were age ≥ 18 years; and (4) were able to communicate independently and participate voluntarily in the study. The exclusion criteria for patients were as follows: (1) a history of severe cognitive impairment or mental illness; and (2) the presence of severe physical diseases or other tumors. The minimum sample size for estimating cross-sectional network models typically needs to exceed 20 times the number of nodes [28]. With 19 symptom nodes in this study, the required minimum sample size was 380 patients. The larger the sample size is, the more accurate the network analysis. The researchers screened patients the day before their admission and contacted them by phone to confirm their admission time as well as willingness to participate in the study. If they agreed to participate, the researchers met them in person on the day of admission. After obtaining informed consent, the researchers distributed paper questionnaires, which the patients completed in a quiet room. Upon completion, the researchers promptly checked for any missing responses and asked the patients to fill in any omitted items.

Measurements

Sociodemographic and clinical characteristics questionnaire

Sociodemographic and clinical characteristics included age, BMI, sex, residential area, educational level, employment status, monthly income, economic burden, personality traits, course of illness and cancer stage. The demographic characteristics were recorded by the patient, and in cases where patients were not aware of their clinical characteristics, researchers retrieved and completed them from the hospital's electronic medical record system.

The M.D. Anderson Symptom Inventory and Lung Cancer Module

The M.D. Anderson Symptom Inventory (MDASI) was originally developed by The University of Texas M. D. Anderson Cancer Center [29] and subsequently translated into Chinese by Wang et al. [30]. The MDASI consists of 13 core symptom items that are rated based on their presence and severity (pain, fatigue, nausea, disturbed sleep, distressed, shortness of breath, remembering things, lack of appetite, drowsy, dry mouth, sad, vomiting and numbness or tingling) and 6 symptom interference items that are rated based on the level of symptom interference with life (general activity, mood, work, relations, walking and enjoyment of life). Severity scores for each item range from 0 (asymptomatic) to 10 (as severe as one could imagine). A higher score indicates a greater burden of these symptoms and interference levels, whereas a lower score suggests the opposite. The Cronbach’s α for the MDASI in this study was 0.847. The Lung Cancer Module of the M.D. Anderson Symptom Inventory (MDASI-LC) was developed by our group in a previous study [31]. The MDASI-LC includes 6 symptoms (coughing, expectoration, hemoptysis, chest tightness, constipation, and loss of weight) and uses the same scoring method as the MDASI. This scale is widely used to measure the degree of symptoms in Chinese patients with lung cancer, and the Cronbach’s α for the MDASI-LC in this study was 0.903.

Quality of life questionnaire-core 30

The quality of life of lung cancer patients was measured using the Quality of Life Questionnaire-core 30 (QLQ-C30). The QLQ-C30 was developed by the European Organization for Research and Treatment of Cancer (EORTC) [32] and translated into Chinese by Zhao [33]. The QLQ-C30 includes six single items (dyspnea, insomnia, appetite loss, constipation, diarrhea and financial difficulties), five functional scales (physical, role, emotional, cognitive, and social functional scales), three symptom scales (fatigue, pain, and nausea and vomiting symptom scales) and one global health status/QoL scale, totaling 30 items. With the exception of global health status, which was assessed on a scale of 1–7, the remaining items were scored on a four-point response rating scale ranging from 1 to 4 and represented by “not at all, a little, quite a bit, and very much”. The higher the functional scale score and the global health status score are, the healthier the functional domain and the overall quality of life. All scale measures range from 0 to 100. When calculating the score of functional scales, the first step is to estimate the average of these scales, referred to as the raw score. The raw scores subsequently undergo standardization through linear transformations to ensure comparability across dimensions. The Cronbach’s α for the QLQ-C30 in this study was 0.950.

Statistical analysis

Data analysis was conducted using R (R language: A Language and Environment for Statistical Computing, v 4.2.3, R Core Team) and SPSS (Statistical Package for the Social Sciences, v 25.0, IBM, USA). Count data are presented frequencies and percentages. Measurement data following a normal distribution were described via means and standard deviations, whereas nonnormally distributed measurement data were expressed via medians and interquartile intervals. T tests and one-way analysis of variance were used to assess inter group differences in demographic and clinical characteristics related to global health status. Post hoc tests were conducted using the Bonferroni method, with the P value adjusted on the basis of the number of comparisons: adjusted P value = 0.05/number of comparisons.

Network analysis is a novel method for identifying SCs that is capable of not only comprehensively evaluating relationships within SCs, but also visualizing the relationships between symptoms [34, 35]. In this study, we employed the Walktrap method in network analysis to identify SCs. The Walktrap method is based on the premise that nodes within communities are often connected by shorter random walks. It identifies communities (SCs) through iterative mergers, minimizing the overall distance between nodes (symptoms) measured by random walks. The maximum path length of these random walks can be adjusted to create either more tightly knit or more diverse communities. The Walktrap method is favored for selecting community structures or for exploring different cutoff points within communities [36]. We also calculated centrality indices and determined the central symptoms within each SC on the basis of strength centrality [37, 38]. The score of each SC is the average score of all symptoms included in the SC. The Pearson correlation method was used to investigate the correlation between each function of HRQoL and SCs.

To explore the influencing factors of each function of HRQoL, multiple linear regression was conducted with each HRQoL function as the dependent variable and the patient's demographic characteristics, along with scores of SCs, as the independent variables. Dummy variables were introduced for categorical variables with more than two categories, with the first category designated the reference variable. All dummy variables were included using the enter method, whereas other variables were selected via the stepwise method. The entry criterion for variables in the regression model was set at 0.05, the exclusion criterion was set at 0.10, and a significance level of P < 0.05 was considered statistically significant. The results were visualized using the qgraph package. The nodes represent the influencing factors for each function of HRQoL, and the edge color indicates the direction of impact, with blue for positive and red for negative. The thickness of the edge represents the slope, indicating the change in the dependent variable caused by a one-unit change in the independent variable.

Results

Patient characteristics

A total of 623 lung cancer patients, comprising 64.53% males and 35.47% females, participated in this study. The mean age of the patients was 56.45 years (SD = 11.06), and the majority of patients (43.50%) resided in urban areas. In terms of education, 65.81% had completed junior high school or had lower educational attainment. Additionally, 87.80% of the patients were in the advanced cancer stage (Table 1).

Table 1 Demographic and clinical characteristics of lung cancer patients (n = 623)
Full size table

Prevalence and severity of symptoms

The prevalence of symptoms ranged from 5.62% to 89.73%. The highest prevalence of symptoms was observed for sleep disturbance (89.73%), followed by fatigue (86.52%) and distress (80.42%). Conversely, hemoptysis had the lowest prevalence, at 5.62%. The most severe symptom was fatigue (4.49 ± 3.05), followed by sleep disturbance (3.66 ± 2.53) and distress (2.65 ± 2.32). Regarding the impact on patients' lives, work (including housework) had the greatest effect (3.46 ± 2.67), whereas relationships with others had the least impact (1.62 ± 1.98). Furthermore, the global health status score was 59.71 ± 21.09, as detailed in Table 2.

Table 2 Prevalence, severity and interference of symptoms (n = 623)
Full size table

Symptom clusters of lung cancer patients

Three SCs were identified using the walktrap method. The first was Somato-psychological SC, which included pain, fatigue, sleep disturbance, distress, drowsiness, and sadness. The second was the Respiratory SC, which included shortness of breath, coughing, expectoration, and chest tightness. The third was the Gastrointestinal SC, which comprised nausea, poor appetite, vomiting, constipation, and weight loss. Strength centrality emerged as the most dependable centrality index; the higher the strength centrality is, the greater the likelihood that the symptom is the central symptom. Fatigue (r strength = 1.339) was the most central symptom of the Somato-psychological SC, coughing (r strength = 0.533) was the most central symptom of the Respiratory SC, and vomiting (r strength = 1.712) was the most central symptom of the Gastrointestinal SC (Fig. 1).

Fig. 1
figure 1

Symptom network of lung cancer patients. The symptom network layout was generated using the Fruchterman-Reingold algorithm, which clusters more strongly connected nodes together to estimate the optimal layout. In the symptom network diagram, circular nodes represent symptoms, edges between nodes indicate correlations between two symptoms, and the thickness of the edges reflects the strength of the correlation. The colors of the spheres indicate different symptom clusters: blue for the Somato-psychological SC, yellow for the Respiratory SC, and pink for the Gastrointestinal SC

Full size image

Correlations between SCs and HRQoL

The Pearson correlation results indicated significant negative correlations between the Somato-psychological SC, Respiratory SC, along with Gastrointestinal SC and physical functioning, role functioning, emotional functioning, cognitive functioning, social functioning, and global health status. The P values for all correlation coefficients were less than 0.05. The strongest correlation was observed between the Somato-psychological SC and emotional functioning (r = − 0.57, P < 0.01), followed by the correlation between the Somato-psychological SC and global health status (r = − 0.55, P < 0.01). In contrast, the weakest correlation was identified between Gastrointestinal SC and cognitive functioning (r = − 0.20, P < 0.01). For detailed results, please refer to Fig. 2 and Supplementary Table S1.

Fig. 2
figure 2

Heatmap of the correlations between SCs and HRQoL. In this heatmap, red indicates a negative correlation, whereas blue indicates a positive correlation. The size of the sphere and the intensity of the color correspond to the absolute value of the correlation coefficient, with larger and darker representations indicating stronger correlations. The lower-left section of the figures reveals negative correlations between the three symptom clusters and the five functional dimensions and the global health status of HRQoL. SC1 Respiratory symptom cluster, SC2 Gastrointestinal symptom cluster, SC3 Somato-psychological symptom cluster, F1 Physical functioning, F2 Role functioning, F3 Emotional functioning, F4 Cognitive functioning, F5 Social functioning, F6 Global health status

Full size image

Multiple linear regression analysis of HRQoL

Age, BMI, sex, residential area, education level, employment status, monthly income, economic burden, personality traits, course of disease, cancer stage, and the degree of Somato-psychological SC, Respiratory SC, and Gastrointestinal SC were utilized as independent variables. Simultaneously, physical functioning, role functioning, emotional functioning, cognitive functioning, social functioning, and global health status served as dependent variables in the multiple linear regression analysis. All variance inflation factors were less than 10, indicating the absence of multicollinearity. The significant variables influencing physical functioning were monthly income, age, the Somato-psychological SC, and the Gastrointestinal SC (R2 = 0.260, F = 9.141, P < 0.001). Role functioning was significantly associated with educational level, employment status, monthly income, cancer stage, age, the Somato-psychological SC, and the Gastrointestinal SC (R2 = 0.303, F = 11.314, P < 0.001). The factors contributing to emotional functioning included employment status, monthly income, age, sex, the Somato-psychological SC, and the Gastrointestinal SC (R2 = 0.419, F = 17.934, P < 0.001). Factors related to cognitive functioning were employment status, monthly income, age, the Somato-psychological SC, and the Gastrointestinal SC (R2 = 0.250, F = 8.655, P < 0.001). Social functioning was significantly associated with educational level, monthly income, and the Somato-psychological SC (R2 = 0.286, F = 11.430, P < 0.001). The significant variables influencing global health status were economic burden, the Somato-psychological SC, and the Gastrointestinal SC (R2 = 0.351, F = 14.702, P < 0.001). The results of the multiple linear regression analysis are presented in Fig. 3, with detailed information available in Supplementary Table S2.

Fig. 3
figure 3

Network of influencing factors of HRQoL. In this diagram, plots a-f illustrate the influencing factors of the functional dimensions and global health status of HRQoL. The large circle at the center symbolizes the functional dimensions and global health status of HRQoL (dependent variables), whereas the smaller circle beside it represents the influencing factors (independent variables). The thickness of the edges signifies the slope of these variables, with thicker edges indicating a greater change in the dimensions of HRQoL resulting from a change in the influencing factor. The blue edges denote a positive influence, whereas the red edges indicate a negative influence. SC symptom cluster, Somato SC Somato-psychological symptom cluster

Full size image

Discussion

This study identified three SCs: the Somato-psychological SC (including pain, fatigue, sleep disturbance, distress, drowsiness, and sadness), the Respiratory SC (including shortness of breath, coughing, expectoration, and chest tightness), and the Gastrointestinal SC (including nausea, poor appetite, vomiting, constipation, and weight loss). These results were consistent with those of the SCs previously identified by our research team [9, 15]. The first SC was the Somato-psychological SC. Some studies have divided this SC into two separate SCs, the Somato SC and the Psychological SC, but the primary symptoms of these SCs mostly include fatigue, sleep disturbance, and sadness [10, 11, 13, 14]. This result suggests that physiological discomfort and psychological distress may share common biological causes or have a synergistic relationship. The study results also indicated that the central symptom of the Somato-psychological SC was fatigue. A systematic review and a study of 1330 patients with 7 types of cancer also identified fatigue as the most common and central symptom [12, 39]. This situation highlights the need for health care providers to pay particular attention to fatigue symptom [40].

The second SC was the Respiratory SC. Some studies have referred to the SC as the lung cancer-specific SC [10, 13, 14, 16]. The components of this SC are generally consistent, mostly comprising respiratory-related symptoms such as shortness of breath, coughing, expectoration, and chest tightness. Yang [16] identified a lung cancer-related SC in patients receiving immunotherapy, which included weight loss, shortness of breath, cough, lack of appetite, and fatigue; the differences may be due to variations in treatment protocols. Li [11] included hemoptysis in the Respiratory SC, but in this study, the prevalence of hemoptysis was low and not closely related to other symptoms; thus, it was not included in the SC. The central symptom identified in the Respiratory SC was coughing. A qualitative analysis also suggested that coughing plays a central role in the Respiratory SC [41]. The result may be because coughing consumes more oxygen, increasing the respiratory rate and depth, thereby affecting chest tightness and shortness of breath [42].

The final SC was the Gastrointestinal SC. The symptoms associated with this SC are relatively consistent and primarily include nausea, poor appetite, vomiting, and constipation. Russell [13] and Wong [14] used the MSAS questionnaire to survey lung cancer patients undergoing chemotherapy and referred to this SC as the nutritional SC, which included decreased appetite and weight loss. This study used the MDASI for the survey, and the results were similar. The central symptom identified in the Gastrointestinal SC was vomiting. However, research specifically focusing on central symptoms is limited, and the central symptoms for lung cancer patients undergoing chemotherapy require further validation. In summary, the SCs identified in this study are consistent with those identified in most previous studies. While there may be some differences due to the treatment methods, the majority of symptoms are the same.

The results of this study indicated that the global health status score was 59.71 ± 21.09, which is lower than the normative scores reported by Nolte (66.1 ± 21.7) [43], Young (62.3 ± 23.7) [44], and Pilz (68.2 ± 20.1) [45] for the general population. This decrease in HRQoL was attributed to the study population being composed of lung cancer patients, whose HRQoL was diminished due to the disease. Compared with that of other cancer patients, the HRQoL of the lung cancer patients in this study was lower, similar to the findings of Wan (56.9 ± 24.6) [46] and Machingura (60 ± 22) [47]. This disparity may be due to the more severe side effects of lung cancer treatment than other types of cancer, resulting in a lower HRQoL [48]. However, compared with other studies, our results indicated a higher HRQoL [49, 50]. This improvement is likely due to better medical conditions and continuous advancements in medical technology and treatment methods, leading to an increased HRQoL for lung cancer patients compared with that reported in the past [51, 52].

The results of this study indicated that the Somato-psychological SC negatively affected the global health status and all five functional domains of HRQoL, which is consistent with previous findings [11, 16, 27]. Additionally, our study revealed that the Somato-psychological SC was the primary predictor of patients' HRQoL. The Somato-psychological SC included various physical symptoms (such as fatigue and pain) and psychological symptoms (such as distress and sadness). Discomfort, such as fatigue, pain, and sleep disturbances, directly reduces patients' physical functioning and participation in life activities [53, 54]. Negative emotions such as distress and sadness can affect patients' social activities, work ability, and overall life satisfaction [48, 55]. The multidimensional impact of these symptoms significantly influences patients' HRQoL, making the Somato-psychological SC the most impactful for their HRQoL.

The Gastrointestinal SC negatively impacted global health status, physical functioning, role functioning, emotional functioning, and cognitive functioning, which is consistent with previous research findings [11, 56]. Gastrointestinal symptoms such as nausea and vomiting could lead to poor appetite and malnutrition, limiting patients' activity levels and affecting their ability to work, engage in hobbies, or participate in leisure activities [57, 58]. Consequently, these factors reduce patients' HRQoL. Li et al. [11] reported that the Gastrointestinal SC negatively affects patients' social functioning. However, in this study, the Gastrointestinal SC did not significantly impact patients' social functioning. This discrepancy could be due to the strong familial support influenced by traditional Chinese filial piety, which provides stable social support for most cancer patients [59, 60]. Additionally, the widespread use of advanced online communication tools allows patients to communicate seamlessly across time and distance, mitigating the impact of the Gastrointestinal SC on social functioning [61, 62].

In this study, the Respiratory SC did not impact patients' HRQoL, whereas previous research indicated a negative impact [11, 16, 26]. The inconsistency could be attributed to the fact that most patients receive treatment, thereby reducing the severity of respiratory symptoms compared with other SCs and influencing factors. This reduced impact may not significantly affect patients' HRQoL. Future research could explore the specific effects of the Respiratory SC on HRQoL through subgroup analysis or long-term follow-up studies involving different patients.

In addition, the findings indicated that patients with higher incomes had higher scores in all five functional domains. Patients who were still employed presented higher scores in role functioning, emotional functioning, and cognitive functioning compared to those who were not employed. Age negatively affected physical, role, and cognitive functioning, but positively affected emotional functioning. Although the above demographic characteristics could influence patients' HRQoL, the impact of SCs was the most significant. Therefore, health care providers should be encouraged to perform personalized assessments and health guidance for SCs. Interventions should be implemented for the Somato-psychological SC and the Gastrointestinal SC, such as exercise and psychosocial interventions for the Somato-psychological SC, and dietary care along with nutritional management for the Gastrointestinal SC, to maximize improvements in patients' HRQoL.

Limitations

The current study was a cross-sectional correlational investigation. While a significant reduction in symptoms may lead to improvements in certain functional domains, deterioration in function could also result in more frequent or severe symptoms. However, the cross-sectional design of this study limited our ability to establish causality. Therefore, future research should consider longitudinal prospective studies to validate whether symptom improvement indeed predicts better HRQoL.

Conclusions

Lung cancer patients who are undergoing chemotherapy experience various symptoms, which form SCs that significantly affect their HRQoL. This study identified three SCs. The Somato-psychological SC had negative impacts on global health status as well as all five functional domains, and had the greatest impact. The Gastrointestinal SC negatively affected global health status and four functional domains, excluding social functioning. The Respiratory SC did not have a significant effect on patients' HRQoL. In addition, the study results indicated that fatigue and vomiting were the central symptoms of the Somato-psychological SC and Gastrointestinal SC, respectively. These findings highlight the importance of health care providers prioritizing the management of fatigue and vomiting. Health care providers should pay close attention to patients' Somato-psychological SC and Gastrointestinal SC, promptly assess and identify high-risk individuals, and develop targeted interventions for these SCs to alleviate symptom burden and improve HRQoL.