Independent Associations Between Blood Lipid Proles and Lung Cancer Risk

Introduction: Dyslipidemia is a common intermediate aggravating factor of various cancers, but the relationship between blood lipid proles and lung cancer remains unclear. This study was performed to assess the non-linear and linear relationships between them. Material and methods: We enrolled 1593 newly diagnosed lung cancer patients and 1593 age- and sex-matched healthy controls between 2017 and 2019. Biochemical indicators, including lipid proles and tumor markers, were collected. Odds ratios and 95% con ﬁ dence intervals were calculated using conditional logistic regression analysis. The restricted cubic spline analysis and multiple linear regression were used to explore the non-linear and linear associations. Results: Lung cancer patients had lower values for total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). After multivariate adjustment, hyper-TC and low-HDL-C were associated with lung cancer risk. In addition, non-linear relationships were identied between lung cancer risk and TC/HDL-C. Multiple linear regression analyses showed that triglycerides were signi ﬁ cantly associated with squamous cell carcinoma antigen (SCC). As well, HDL-C was signi ﬁ cantly associated with neuron-specic enolase (NSE), cytokeratin 19 fragments (CYFRA21-1), SCC, and tumor-associated antigen 125 (CA-125). Signi ﬁ cant linear trends were observed between increasing triglyceride quartiles and hyper-NSE/hyper-CYFRA21-1, and between increasing HDL-C quartiles and hyper-NSE/hyper-CYFRA21-1/hyper-CA-125. Conclusions: Lipid levels were signicantly lower in lung cancer patients, and there were negative non-linear associations of TC and HDL-C with lung cancer risk. In addition, there were negative linear trends across TG and HDL-C quartiles for the risk of abnormal tumor markers among lung cancer patients. were negative non-linear associations between both TC and HDL-C and lung cancer risk. We also found negative linear trends among lung cancer patients for the risk of abnormal tumor markers across TG and HDL-C quartiles. These results suggest that it is important to consider lipid levels in populations at high-risk of lung cancer, and that dyslipidemia may be a potential modiable factor for lung cancer prevention. Therefore, good lipid control may be a new and promising therapeutic strategy for the personalized treatment of lung cancer.


Introduction
Lung cancer is the most common tumor in the world, accounting for 11.6% of total cancer cases, and it is also the leading cause of cancer-related death, accounting for 18.4% of total cancer mortality [1].
Although many epidemiologic studies have suggested that smoking is the major risk factor, the increasing incidence of lung cancer in nonsmokers indicates that other risk factors may exist [2]. As the tumorigenesis in the body is a systemic process, abnormal metabolism, including dyslipidemia, may play a role in the occurrence and deterioration of tumors [3].
Dyslipidemia is a common intermediate aggravating factor in many chronic diseases, including various types of cancer. Elevated triglyceride (TG) levels have been associated with prostate cancer [4], and elevated total cholesterol (TC) levels could increase the risk of prostate and colorectal cancers [4,5]. Decreased high-density lipoprotein cholesterol (HDL-C) levels were associated with breast cancer and lymphoma [6,7], and low-density lipoprotein cholesterol (LDL-C) levels were correlated with poor prognosis in pancreatic carcinoma [8]. However, the relationship between blood lipids pro le and lung cancer has thus far resulted in contradictory conclusions and remains unclear. Lin et al. [9] demonstrated Page 3/15 that TC and HDL-C were negatively associated with lung cancer risk, while TG was positively associated with lung cancer risk. Hao et al. [10] reported that blood lipid pro les were not associated with lung cancer risk, with the exception of HDL-C. Lyu et al. [11] reported a non-linear relationship between lung cancer risk and both TG and TC, respectively, while low LDL-C was associated with increased lung cancer risk; and there was no signi cant relationship between HDL-C and lung cancer. Therefore, we performed this study to investigate the linear and non-linear relationships between blood lipid pro les and lung cancer risk.
In addition, much concern has been aroused in recent years regarding the value of molecular-based techniques in cancers. Cytokeratin 19 fragments (CYFRA21-1), squamous cell carcinoma antigen (SCC), serum carcinoembryonic antigen (CEA), neuron-speci c enolase (NSE), and tumor-associated antigen 125 (CA125) have been the most commonly used in lung cancer because of their excellent sensitivity [12]. To comprehensively evaluate the linear relationship between blood lipid pro les and lung cancer risk, the association between abnormal tumor markers and lipid pro les among lung cancer patients was also explored.

Study population
We enrolled 2390 newly diagnosed lung cancer patients admitted to Qilu Hospital of Shandong University between 2017 and 2019. The exclusion criteria were as follows: 1) history of treatment for lung cancer; 2) use of lipid-lowering drugs; 3) severe chronic diseases, such as chronic kidney disease or hyperparathyroidism; 4) other malignancy; 5) < 18 years of age; 6) undergoing systemic steroid treatment; and 7) missing clinical data. Ultimately, 1593 eligible patients were included in the study. In addition, 1593 age-and sex-matched healthy controls were collected from the Health Examination Center, Qilu Hospital of Shandong University.

Biochemical Indicators
Fasting blood samples were taken in the morning from each participant's anterior cubital vein.
Determination of serum lipid pro les, including TC, TG, HDL-C, and LDL-C, and tumor marker pro les, including NSE, CYFRA21-1, SCC, and CA125, were performed using routine procedures in the hospital laboratory. Dyslipidemia was de ned as hyper-TC ≥ 5.17 mmol/L, hyper-TG ≥ 1.7 mmol/L, low HDL < 1.03 mmol/L, and high LDL ≥ 3. Statistical analysis All data were tested for normality prior to statistical analysis. Data are expressed as medians (interquartile range) for continuous variables with skewed distribution and percentages for categorical variables. Differences in skewed continuous variables were examined using Mann-Whitney U test. Differences in categorical variables were examined using Chi-squared tests. Correlations were estimated using Spearman's correlation analysis. Odds ratios (ORs) and 95% con dence intervals (95% CIs) for lung cancer were calculated for binary or quaternary lipid pro les using conditional logistic regression adjusted for age, sex, smoking status, alcohol consumption, and other lipid pro le components. The restricted cubic spline (RCS) analysis was used to explore the non-linear association between blood lipid pro les and lung cancer, and knots for blood lipid pro les were identi ed at the 5th, 25th, 50th, 75th, and 95th percentiles. Supplementary analyses, including multiple linear regression and logistic regression, were performed among lung cancer patients to estimate the linear association between the blood lipids pro le and lung cancer. All statistical analyses were performed using SPSS 25.0 software (SPSS Inc., Chicago, IL, USA) and R software package 3.6.2 (http://www.r-project.org/). Two-tailed P-values < 0.05 were considered signi cant.

Basic characteristics of the study participants
In total, 3186 participants, including 1593 patients and 1593 controls, were included in this study. The basic characteristics of the two groups are shown in Table 1. Lung cancer patients presented with higher percentages of smoking and drinking, and all tumor markers, including NSE, CYFRA21-1, SCC, and CA-125 were signi cantly higher in this group. Analysis of lipid profiles revealed that the lung cancer patients had lower TC, HDL-C, and LDL-C values, whereas there were no signi cant differences in the TG values. Similarly, the percentages of individuals with hyper-TC, low HDL-C, and high LDL-C were signi cantly higher in lung cancer patients.
Relationship Between lipid profilesand lung cancer risk As shown in Table 2, hyper-TC, low HDL-C and high LDL-C were all signi cantly associated with lung cancer risk. However, after adjusting for confounding factors, including age, sex, smoking status, alcohol consumption, and other lipid pro le components, only hyper-TC and low HDL-C remained signi cantly associated with lung cancer risk. In particular, hyper-TC was associated with decreased lung cancer risk, with an OR (95% CI) of 0.632 (0.527, 0.759), while low-HDL-C was associated with increased lung cancer risk, with an OR (95% CI) of 1.992 (1.599, 2.482).
To further explore the non-linear relationship between plasma lipid levels and the risk of lung cancer, RCS was performed with adjustments for age, sex, smoking status, and alcohol consumption. The results showed that there was a non-linear relationship between lung cancer risk and both TC and HDL-C levels (both P-overall<0.001, P-non-linear<0.001, Figure 1). However, the non-linear relationship was not signi cant for TG and LDL-C (P-non-linear>0.05, Figure 1).

Relationship between lipid profiles and tumor markers among lung cancer patients
Supplementary analyses, including multiple linear regression and logistic regression, were performed to estimate the linear relationship between blood lipid pro les and tumor markers among lung cancer patients. Multiple linear regression revealed that TG was significantly associated with SCC (β=-0.241, P=0.017), while HDL-C was significantly associated with NSE, CYFRA21-1, SCC, and CA-125 (P<0.05), following adjustment for age, sex, smoking status, and alcohol consumption (Table 3). We then performed logistic regression between the ORs (95% CI) for abnormal tumor markers and increasing TG/HDL-C quartiles. Significant negative linear trends were observed for hyper-NSE and hyper-CYFRA21-1 with increasing TG quartiles, with the ORs (95% CI) for hyper-NSE being 1 (reference), 1.104 (0.758, respectively (all P for linear trend < 0.01) ( Table 4).

Discussion
The present study demonstrated that the levels of TC, HDL-C, and LDL-C were signi cantly lower in lung cancer patients. Notably, signi cant non-linear and negative associations between TC and HDL-C and lung cancer risk were found after multivariate adjustment. Furthermore, we identi ed negative linear trends for the risk of abnormal tumor markers across TG and HDL-C quartiles among lung cancer patients when all four lipid indicators were considered jointly.
Several previous studies have demonstrated a relationship between lipid pro les and risk for various types of cancer, including prostate, breast, and colorectal cancers [4][5][6][7][8]; however, the relationship between blood lipid pro les and lung cancer remains unclear. Studies aiming to evaluate the association between TC and non-small cell lung cancer have drawn inconsistent conclusions. Lin et al. [9] and Lyu et al. [11] demonstrated that low TC levels were associated with lung cancer risk. In contrast, Chandler et al. [15] found that high TC levels were associated with increased incidence of lung cancer, although the signi cance disappeared after multivariable adjustment. In the present study, we demonstrated that hyper-TC was associated with decreased lung cancer risk after adjusting for age, sex, smoking status, alcohol consumption, and other lipid pro le components, consistent with the ndings of Lin et al. and Lyu et al. However, it remains to be determined whether the observed relationship is causal or due to the effect of pre-diagnosed cancer on serum cholesterol levels. There are several underlying mechanisms that could be involved in this relationship. Plasma polyunsaturated fatty acids, which can decrease TC levels, were found to be higher in lung cancer patients [16]. As well, low cell cholesterol has been associated with increased NF-κB activity, upregulated mevalonate pathway activity, and suppressed immunity [17,18]. Finally, due to the rapid growth and high rate of division, malignant cells require far more cholesterol, which could lead to lower levels of blood cholesterol, including TC levels. We additionally found that there was a negative non-linear relationship between TG/TC and lung cancer risk. The identi cation of this non-linear pattern revealed the complexity of the relationship arising from interactions between multiple risk factors and suggests that lung cancer risk cannot be reduced simply by lowering blood lipids. This was demonstrated by the U-shaped association, which indicated that adjusting TC levels as low as possible did not reduce lung cancer risk.
HDL-C plays an important role in the reverse transport process of cholesterol, which facilitates the removal of excess cholesterol from peripheral tissues, and it is widely recognized as a protective factor in cardiovascular disease [19]. However, the role of HDL-C in the occurrence and progression of cancer, especially lung cancer, has not been determined. A meta-analysis of randomized controlled trials concluded that there was a signi cant inverse association between HDL-C and the risk of cancer [20], although this conclusion contradicts those of other studies [21,22]. Some researchers have demonstrated that HDL-C was negatively associated with a risk of lung cancer [10], while Lyu et al. [11] found that no signi cant relationship existed. Our present results indicate that low HDL-C was associated with increased lung cancer risk, and there exists a signi cant non-linear and negative association between HDL-C and lung cancer risk after multivariate adjustment. There are numerous underlying mechanisms that may account for these ndings. First, HDL-C could confer anti-in ammatory effects and organ protection through leukocyte adhesion and cytokine production; thus, the decrease in HDL-C level may lead to in ammation, which plays a role in the development of neoplasms [23,24]. Second, lower HDL may have induced a reduction in antioxidant activity, which is associated with lung cancer [25]. Third, a decrease in the ability of HDL-C to inhibit apoptosis, resulting from decreased HDL-C levels, may contribute to the development of cancer [26,27].
Research to evaluate the relationship between TG and lung cancer risk has drawn inconsistent conclusions as well. Some researchers demonstrated that TG was positively associated with increased lung cancer risk [9,28], while Siemianowicz et al. [29] found that TG levels were lower in lung cancer patients. In our study, there was no signi cant difference in TG levels between lung cancer patients and healthy controls, and hyper-TG was not associated with lung cancer risk after multivariable adjustment. LDL-C, another component of cholesterol, was found to be associated with various cancers [30][31][32][33], while the relationship between LDL-C levels and lung cancer risk has not been clari ed [10,11]. The present study indicated that LDL-C levels were lower in lung cancer patients, and that high LDL-C was associated with decreased lung cancer risk. Of note, the association disappeared after multivariable adjustment. We speculate that interactions between the confounding factors may exist, and that this relationship requires further investigation. In addition, large-scale prospective studies are needed to clarify these relationships.
The association between biomarkers and tumors has aroused much attention [34,35]. In order to comprehensively evaluate the linear relationship between blood lipid pro les and lung cancer risk, we explored the association between abnormal tumor markers and lipid pro les among lung cancer patients. Tumor markers, including CYFRA21-1, SCC, CEA, NSE, and CA125 have been commonly used to predict lung cancer because of their excellent sensitivity [12,36]. We found that TG was associated with a decreased risk of hyper-NSE and hyper-CYFRA21-1, and that HDL-C was associated with a decreased risk of hyper-NSE, hyper-CYFRA21-1, and hyper CA-125, after multivariable adjustment. To our best knowledge, there is no research to explore the relationship between blood lipids and tumor markers of lung cancer, and the involved mechanisms were not clari ed. One possible mechanism relating TG with abnormal tumor markers was that hyper TG was associated with the production of reactive oxygen species, which could affect normal cell proliferation [37]. However, hyper TG was not associated with lung cancer risk in aforementioned study. As to HDL-C, the involved mechanism may be related to in ammation, reduced antioxidant activity and disability to inhibit apoptosis [23][24][25][26][27]. In order to clarify the biological mechanism, large-scale prospective cohort studies taking tumor markers as primary target need to be performed.
Although our investigations were performed on a large sample that adjusted for multiple potential confounding factors, there were still several limitations. First, this observational study had a case-control rather than a prospective design; thus, the cause-consequence relationship cannot be evaluated. Second, although many common confounding factors were adjusted for, other potential confounding factors such as dietary habits, obesity, and physical activity were not considered. Therefore, further research is warranted to clarify the causal relationship and to enhance credibility.

Conclusion
The present study investigated the non-linear and linear relationships between lipid pro les and lung cancer risk after multivariate adjustment. We found that lipid levels were signi cantly lower in lung cancer patients, and that hyper-TC and low HDL-C were associated with lung cancer risk. In addition, there were negative non-linear associations between both TC and HDL-C and lung cancer risk. We also found negative linear trends among lung cancer patients for the risk of abnormal tumor markers across TG and HDL-C quartiles. These results suggest that it is important to consider lipid levels in populations at highrisk of lung cancer, and that dyslipidemia may be a potential modi able factor for lung cancer prevention. Therefore, good lipid control may be a new and promising therapeutic strategy for the personalized treatment of lung cancer.     Figure 1 Restricted cubic spline regression of the non-linear relationship between lipid pro les [triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C)] and lung cancer risk. There was a non-linear relationship between lung cancer risk and both TC and HDL-C levels (both P-overall<0.001, P-non-linear<0.001). However, the non-linear relationship was not signi cant for TG and LDL-C (P-non-linear>0.05).