Association between Dietary Anthocyanidins and Risk of Lung Cancer

Background: Anthocyanidins are a kind of water-soluble flavonoids widely found in flowers and fruits of many plants. Although the beneficial effect of anthocyanidins in cancer prevention has been discussed, the value of anthocyanidins in lung cancer prevention requires further investigation. In this study, we aimed to explore the role of dietary anthocyanidins in the prevention of lung cancer in population-based prospective studies. Methods: Data of participants in this study were collected from the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated in Cox proportional hazards regression for the association of dietary anthocyanidins and lung cancer risk. The dose-response relationship was explored between total anthocyanidins and the incidence of lung cancer. Results: A total of 97,993 participants were included in this study. The calculated HRs showed a trend that a higher quartile of total anthocyanidins indicated lower risk of lung cancer after adjusting for covariates (HRQ4vsQ1: 0.63; 95% CI: 0.55,0.73; p for trend < 0.001). A non-linear association between total anthocyanidins and lung cancer risk was found in the restricted cubic spline model. Conclusion: A protective association between dietary anthocyanidins and risk of lung cancer in Americans was investigated.


Introduction
The number of newly diagnosed lung cancer cases has risen for years. Causing an estimated 1.8 million deaths each year, lung cancer has been regarded as the leading cause of cancer deaths worldwide; thus, the prevention of lung cancer is an important research area [1,2]. In recent decades, the association between diet and the prevention of lung cancer has been discussed. Some healthy eating habits have been proven to be closely related to reduced lung cancer risk [3][4][5]. Anthocyanins are a kind of water-soluble flavonoids [6], the basic structures of which are anthocyanidins. In other words, anthocyanins are in the form of glycoside while anthocyanidins are in the form of aglycone [7]. The six main members of anthocyanidins are cyanidin, delphinidin, malvidin, peonidin, petunidin, and pelargonidin [7,8]. Commonly, the dietary sources of anthocyanidins include plants, especially flowers, fruits, and tubers containing a large amount of natural pigments [7]. Research shows that anthocyanidins participate in many health-promoting activities and have antioxidant, anti-inflammatory, anti-diabetic, anti-adipogenesis, and anti-cancer effects [9][10][11]. In particular, the beneficial effects of anthocyanidins in specific cancer prevention have been extensively discussed. Reviews have clearly demonstrated remarkable anti-cancer activity of anthocyanidins [12][13][14][15]. Anthocyanidins reduced the cell proliferation of tumors

Lung Cancer Ascertainment
In this study, the outcome was the incidence of lung cancer. In the PLCO trial, the confirmation of the diagnosis of lung cancer was based on reports abstracted from the annual study update forms, and then ICD-O codes were used for extracting relevant medical records using standardized forms. Of note, carcinoid lung cancer was not considered as a target of lung cancer screening in the PLCO trial; thus, it was not confirmed as lung cancer in this study.

Statistical Analysis
Baseline characteristics of participants are presented as the quartile of total anthocyanidins (quartile 1 to quartile 4). Continuous variables are presented as mean (standard deviation), and categorical variables are presented as numbers (percentage). The Kruskal-Wallis test and chi-square test were used to compare continuous and categorical variables across the groups of participants, respectively. To test whether a trend across quartiles of anthocyanidins existed for the outcome, tests for trends across anthocyanidin quartiles were conducted by assigning each quartile the median value and treating the variable as a linear term in the regression models. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated with Cox proportional hazards regression for the association of total anthocyanidins, cyanidin, delphinidin, malvidin, pelargonidin, peonidin, and petunidin and lung cancer incidence in all included participants and sex-specific groups, respectively. Sub-analyses were further performed to evaluate associations with different histological types, including adenocarcinoma, squamous cell carcinoma, large cell carcinoma, and small cell carcinoma. Covariates included in the multivariate regression models were based on the literature review and clinical judgement. In detail, age, gender, BMI, total energy intake, family history of lung cancer, marital status, race/ethnicity, cigarette pack-years, and alcohol intake were adjusted as covariates. The dose-response relationship was explored between total anthocyanidins and the incidence of lung cancer. We excluded the highest 5% of the value of total anthocyanidins in order to avoid the impact of outliers on the trend in the dose-response analysis. A restricted cubic spline model with three knots at the 10th, 50th, and 90th percentiles was employed [24]. We chose the median value of total anthocyanidins as the reference level [24]. Prespecified subgroup analyses were performed to evaluate whether the observed association of intake of total anthocyanidins with lung cancer incidence was modified by age (>65 vs. ≤65 years old), BMI (>25 vs. ≤25 kg/m 2 ), family history of lung cancer (yes vs. no or possible), race/ethnicity (white vs. non-white) or smoking status (never vs. 0-20 pack-years vs. >20 pack-years). Effect modification by variables was examined by adding the cross-product of each effect modifier with total anthocyanidin quartiles in the multivariable-adjusted model. In addition, we conducted the sensitivity analysis to test the robustness of the results by excluding participants (1) with extreme energy intake (<800/>4000 kcal/day for males and <500/>3500 kcal/day for females) [25], (2) with the highest 1% intake of anthocyanidins (including total anthocyanidins, cyanidin, delphinidin, malvidin, peonidin, petunidin, and pelargonidin), or (3) with a follow-up less than 2 years. A two-tailed p value less than 0.05 was considered significant. The statistical analyses were conducted using STATA 15.1, SPSS 25.0, and R 3.6.1 software.

Baseline Characteristics
Data of a total of 97,993 participants were extracted after excluding the participants according to exclusion criteria. The detailed flow chart is presented in Figure 1. We divided participants into quartiles of total anthocyanidins intake (24,533 in Q1; 24,467 in Q2; 24,516 in Q3; 24,477 in Q4). A total of 1631 lung cancer cases were obtained. There were 50,218 (51.25%) participants enrolled in the intervention group and 47,775 (48.75%) participants recruited to the control group. Significant differences were obtained in total energy intake, age, gender, BMI, race/ethnicity, marital status, cigarette pack-years, alcohol intake, family history of lung cancer, and family history of any cancer (all p < 0.05). In the highest quartile group of total anthocyanidins, participants had higher daily energy, were older, had lower cigarette pack-years, and had higher rates of clear family history of lung cancer. More detailed information is shown in Table 1.
at the 10th, 50th, and 90th percentiles was employed [24]. We chose the median value of total anthocyanidins as the reference level [24]. Prespecified subgroup analyses were performed to evaluate whether the observed association of intake of total anthocyanidins with lung cancer incidence was modified by age (>65 vs. ≤65 years old), BMI (>25 vs. ≤25 kg/m 2 ), family history of lung cancer (yes vs. no or possible), race/ethnicity (white vs. nonwhite) or smoking status (never vs. 0-20 pack-years vs. >20 pack-years). Effect modification by variables was examined by adding the cross-product of each effect modifier with total anthocyanidin quartiles in the multivariable-adjusted model. In addition, we conducted the sensitivity analysis to test the robustness of the results by excluding participants (1) with extreme energy intake (<800/>4000 kcal/day for males and <500/>3500 kcal/day for females) [25], (2) with the highest 1% intake of anthocyanidins (including total anthocyanidins, cyanidin, delphinidin, malvidin, peonidin, petunidin, and pelargonidin), or (3) with a follow-up less than 2 years. A two-tailed p value less than 0.05 was considered significant. The statistical analyses were conducted using STATA 15.1, SPSS 25.0, and R 3.6.1 software.

Baseline Characteristics
Data of a total of 97,993 participants were extracted after excluding the participants according to exclusion criteria. The detailed flow chart is presented in Figure 1. We divided participants into quartiles of total anthocyanidins intake (24,533 in Q1; 24,467 in Q2; 24,516 in Q3; 24,477 in Q4). A total of 1631 lung cancer cases were obtained. There were 50,218 (51.25%) participants enrolled in the intervention group and 47,775 (48.75%) participants recruited to the control group. Significant differences were obtained in total energy intake, age, gender, BMI, race/ethnicity, marital status, cigarette pack-years, alcohol intake, family history of lung cancer, and family history of any cancer (all p < 0.05). In the highest quartile group of total anthocyanidins, participants had higher daily energy, were older, had lower cigarette pack-years, and had higher rates of clear family history of lung cancer. More detailed information is shown in Table 1.

Association between Anthocyanidins and Lung Cancer Risk
In the unadjusted analyses, the intakes of total anthocyanidins, cyanidin, delphinidin, malvidin, pelargonidin, peonidin, and petunidin were statistically significantly associated with at least a 40% reduction in the risk of lung cancer for comparison of the highest vs. lowest quartiles [(HR Q4vsQ1 for total anthocyanidins: 0.50; 95% CI: 0.44,0.58; p for trend < 0.001); (HR Q4vsQ1 for cyanidin: 0.55; 95% CI: 0.48,0.63; p for trend < 0.001); (HR Q4vsQ1 for delphinidin: 0.60; 95% CI: 0.53,0.69; p for trend <0.001); (HR Q4vsQ1 for malvidin: 0.54; 95% CI: 0.47,0.62; p for trend < 0.001); (HR Q4vsQ1 for pelargonidin: 0.52; 95% CI: 0.46,0.60; p for trend < 0.001); (HR Q4vsQ1 for peonidin: 0.53; 95% CI: 0.47,0.61; p for trend < 0.001); (HR Q4vsQ1 for petunidin: 0.60; 95% CI: 0.52,0.69; p for trend <0.001)]. In the multivariate adjusted regression model, the calculated adjusted HRs showed a trend that a higher quartile of total anthocyanidins, cyanidin, delphinidin, malvidin, pelargonidin, peonidin, and petunidin indicated lower risk of lung cancer [(HR Q4vsQ1 for total anthocyanidins: 0.63; 95% CI: 0.55,0.73; p for trend < 0.001); (HR Q4vsQ1 for cyanidin: 0.73; 95% CI: 0.63,0.84; p for trend < 0.001); (HR Q4vsQ1 for delphinidin: 0.70; 95% CI: 0.60,0.80; p for trend < 0.001); (HR Q4vsQ1 for malvidin: 0.65; 95% CI: 0.56,0.75; p for trend < 0.001); (HR Q4vsQ1 for pelargonidin: 0.75; 95% CI: 0.65,0.87; p for trend = 0.001); (HR Q4vsQ1 for peonidin: 0.63; 95% CI: 0.55,0.73; p for trend < 0.001); (HR Q4vsQ1 for petunidin: 0.70; 95% CI: 0.60,0.81; p for trend = 0.001)]. Overall, differences in the results by sex were not statistically significant for total anthocyanidins and the subclasses (all p-interaction due to sex > 0.05). However, there was a tendentious suggestion that the results differed between females and males for intakes of delphinidin, pelargonidin, and petunidin for comparison across quartiles. In men, higher intake of delphinidin (p for trend <0.001), pelargonidin (p for trend = 0.003), and petunidin (p for Nutrients 2022, 14, 2643 6 of 13 trend <0.001) was associated with a reduced risk of lung cancer, and a significant difference was not found in women (p for trend > 0.05) ( Table 2). We also examined the associations by lung cancer histologic type. In the unadjusted regression model, a significant inverse association between delphinidin and adenocarcinoma was not obtained (p for trend = 0.054). A significant inverse association between cyanidin and large cell carcinoma was also not obtained (p for trend = 0.072). However, other significant inverse associations between higher intakes of anthocyanidins and the risk of lung cancer cell types were observed. In the multivariate analyses, each of the associations was greatly attenuated when adjusted for potential covariates, for comparison of the highest vs. lowest quartiles. An inverse association between the intake of cyanidin, malvidin, and peonidin and the risk of adenocarcinoma was found, but not for total or other subclasses of anthocyanidins. An inverse association between the intake of total and subclasses of anthocyanidins and the risk of squamous cell carcinoma was observed. For large cell carcinoma, only the intake of malvidin was found to be associated with reduced cancer incidence. For small cell carcinoma, an inverse association was not found between Nutrients 2022, 14, 2643 7 of 13 the intake of delphinidin and the risk of small cell carcinoma. However, an association was found between total anthocyanidins and other subclasses of anthocyanidins and the risk of small cell carcinoma. Although a significant trend was not observed, the highest quartile of total anthocyanidins and pelargonidin was significantly associated with reduced risk of adenocarcinoma compared with the lowest quartile. The highest quartile of delphinidin was significantly associated with reduced risk of large cell carcinoma and small cell carcinoma compared with the lowest quartile. The highest quartile of peonidin was significantly associated with reduced risk of large cell carcinoma compared with the lowest quartile. Detailed information is shown in Table 3.

Additional Analyses
In order to explore the trend that the probability of lung cancer changed with the intake of total anthocyanidins, we conducted the dose-response analysis. A non-linear association between the intake of total anthocyanidins (reference value = 11.62 as median value) and lung cancer risk was found in the restricted cubic spline model (p for non-linear = 0.001) ( Figure 2). The stratified analyses showed that the association between total anthocyanidins and lung cancer risk could be modified by smoking status (p-interaction = 0.004). The highest quartile of total anthocyanidins intake was associated with an elevated risk of lung cancer compared with the lowest quartile in non-smokers (HR Q4vsQ1 : 2.18; 95% CI: 1.25,3.78) but associated with a reduced risk of lung cancer in former/current smokers with >20 cigarette pack-years (HR Q4vsQ1 : 0.55; 95% CI: 0.47,0.65). The association between total anthocyanidins and risk of lung cancer differed in two age groups (p-interaction = 0.035). However, the association between total anthocyanidins intake and lung cancer risk was not found to be modified by BMI (>25 vs. ≤25 kg/m2), family history of lung cancer (yes vs. no or possible), or race/ethnicity (white vs. non-white) ( Table 4). In sensitivity analyses, the HRs of total anthocyanidins did not change significantly by excluding participants with extreme energy intake, with the highest 1% intake of total anthocyanidins, or with a follow-up less than 2 years, indicating the robustness of the association between total and subclasses of anthocyanidins intake and the incidence of lung cancer (Table 5).

Additional Analyses
In order to explore the trend that the probability of lung cancer changed with the intake of total anthocyanidins, we conducted the dose-response analysis. A non-linear association between the intake of total anthocyanidins (reference value = 11.62 as median value) and lung cancer risk was found in the restricted cubic spline model (p for non-linear = 0.001) (Figure 2). The stratified analyses showed that the association between total anthocyanidins and lung cancer risk could be modified by smoking status (p-interaction = 0.004). The highest quartile of total anthocyanidins intake was associated with an elevated risk of lung cancer compared with the lowest quartile in non-smokers (HRQ4vsQ1: 2.18; 95% CI: 1.25,3.78) but associated with a reduced risk of lung cancer in former/current smokers with >20 cigarette pack-years (HRQ4vsQ1: 0.55; 95% CI: 0.47,0.65). The association between total anthocyanidins and risk of lung cancer differed in two age groups (p-interaction = 0.035). However, the association between total anthocyanidins intake and lung cancer risk was not found to be modified by BMI (>25 vs. ≤25 kg/m2), family history of lung cancer (yes vs. no or possible), or race/ethnicity (white vs. non-white) ( Table 4). In sensitivity analyses, the HRs of total anthocyanidins did not change significantly by excluding participants with extreme energy intake, with the highest 1% intake of total anthocyanidins, or with a follow-up less than 2 years, indicating the robustness of the association between total and subclasses of anthocyanidins intake and the incidence of lung cancer (Table 5).

Figure 2.
Dose-response relationship between anthocyanidins and risk of lung cancer adjusted for age, gender, BMI, total energy intake, family history of lung cancer, marital status, race/ethnicity, cigarette pack-years, and alcohol intake (p for non-linear = 0.001).

Figure 2.
Dose-response relationship between anthocyanidins and risk of lung cancer adjusted for age, gender, BMI, total energy intake, family history of lung cancer, marital status, race/ethnicity, cigarette pack-years, and alcohol intake (p for non-linear = 0.001).

Discussion
This prospective large-scale cohort study suggests a reverse association between anthocyanidins and lung cancer risk in the American population. Dietary intake of total anthocyanidins and subclasses including cyanidin, malvidin, pelargonidin, peonidin, and petunidin are suggested to be related to a reduced risk of lung cancer after adjusting for potential confounding factors.
Differences in the results by sex were not statistically significant for anthocyanidins intake, though the tendentious association between certain subclasses of anthocyanidins and the risk of lung cancer was not significantly observed in the female group. In the analyses for different histologic types of lung cancer, we observed that the inverse association between total anthocyanidins and the risk of adenocarcinoma, squamous cell carcinoma, and small cell carcinoma was still significant in the highest quartile compared with the lowest quartile of total anthocyanidins. However, the association was not observed in large cell carcinoma. Whether the phenomenon is attributed to certain mechanism or the small number of cases needs further investigation. The dose-response analysis showed a non-linear relationship between total anthocyanidins and the risk of lung cancer. The lung cancer risk changed with the increased intake of total anthocyanidins in a non-linear manner. Increasing the intake of anthocyanidins may lead to a waning increase in the preventive effects of anthocyanidins on lung cancer. The waning increase in the preventive effects of anthocyanidins on lung cancer with the increasing intake of anthocyanidins indicates that the efficient intake of anthocyanidins against lung cancer may be no more than 20 mg/day according to the dose-response analysis. The stratified analyses suggest that the inverse association between total anthocyanidins and the risk of lung cancer could be modified by age and smoking status. In participants under 65 years old, the inverse association between total anthocyanidins and lung cancer risk was more clearly observed, compared with individuals older than 65 years. Moreover, high intake of total anthocyanidins may decrease the risk of lung cancer in heavy smokers of more than 20 cigarette pack-years. Previous evidence has shown that smokers have a high inflammatory response in the body, which is an important risk factor for lung cancer [26,27]. The evidence that anthocyanidins have anti-inflammatory effects [28] may explain why heavy smokers have opposite responses to anthocyanidins than non-smokers. Surprisingly, high intake of total anthocyanidins may increase the risk of lung cancer in non-smokers, whether anthocyanidins only have protective effect on lung cancer in people with high levels of inflammation merits further investigation. Sensitivity analyses indicated the robustness of the association between total anthocyanidins intake and the incidence of lung cancer.
Overall, this study shows differences from the previous two human studies mentioned above [18,19]. We notice that the daily intake of anthocyanidins in one previous study [18] was too small to compare with this study. And the other study did not describe the intake of anthocyanidins of participants in detail [19]. The concentration of anthocyanidins declines sharply in human blood through digestion. The attenuation of the concentration may induce inconspicuous anticancer effects to exert chemopreventive effects inhibiting the growth of malignant cells, inducing apoptosis and regulating carcinogenic signal transduction in human body [29]. Moreover, the biological activities of anthocyanidins were often influenced by intestinal absorption and mediated by microbial catabolites through habitual dietary intake, which decreases the effect of anthocyanidins in the human body [13]. Thus, low dietary intake of anthocyanidins may be not effectively biologically active in the human body [13]. Despite much lower intake of anthocyanidins than in this study, Cutler et al. suggested that there was a trend that a higher level of dietary anthocyanidins was associated with reduced risk of lung cancer in female ever-smokers (HR: 0.83; 95% CI: 0.67,1.01) compared with a lower intake of anthocyanidins [18]. It is also worth noting that the number of lung cancer cases in the previous two studies was relatively small and resulted in wider confidence intervals.
In summary, this study showed strengths and offers value for lung cancer prevention. This is the first large-scale prospective study exploring the association between the intake of anthocyanidins including total and subclasses and lung cancer risk in both females and males in the American population. We also evaluated the different associations for different genders and for different histologic types of lung cancer. Related potential confounding factors were adjusted to obtain more accurate results. Subgroup analyses evaluated factors that could modify the results, and consequently, we found that age and smoking status could significantly modify the relationship. In addition, the sensitivity analyses ascertained the robustness of the outcomes. Furthermore, the dose-response analysis showed the real relationship between the intake of total anthocyanidins and lung cancer risk in a visual way.
Flaws also exist in this study. Firstly, as dietary habits may change during the long-term follow-up, using only baseline diet to evaluate the dietary intake generally yielded weaker associations with the incidence of disease than using the cumulative dietary intake [30]. Secondly, confidence limits existed in the individual's self-analysis regarding diet patterns, although the questionnaires were validated. Thirdly, the potential co-linearity between anthocyanidins and main food such as vitamins and fiber that may be responsible for the associations observed should be considered in future investigation.

Conclusions
In this study, we observed a protective association between dietary anthocyanidins and the risk of lung cancer in the American population. The lung cancer risk changed with the increase in intake of total anthocyanidins in a non-linear manner. Increasing intake of anthocyanidins (less than 20 mg/day) may lead to the waning increase in the preventive effects of anthocyanidins on lung cancer. Further studies should be conducted to confirm the association.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of the United States NCI.

Informed Consent Statement:
Written informed consent has been obtained from participants included in the PLCO study.

Data Availability Statement:
Data that support the findings of this study have been deposited in the PLCO trial (https://biometry.nci.nih.gov/cdas/plco/ accessed on 2 March 2022) upon reasonable request.