Abstract
Introduction
Sex disparities in lung cancer risk assessment may lead to inequities in screening eligibility. Criteria based on smoking history and age can underrepresent women. Systemic inflammatory indexes derived from routine blood tests are potential biomarkers in cancer risk assessment. We investigated the association between systemic inflammatory indexes and lung cancer risk, stratified by sex, in a Danish cohort.
Methods
We conducted a population-based cohort study using data from the Lolland-Falster Health Study (LOFUS) linked to the Danish Pathology Data Bank (Patobank). Participants aged 40–74 years without a history of lung cancer were included. We calculated four systemic inflammatory indexes and assessed their association with lung cancer using logistic regression models adjusted for age, sex, and smoking status. Optimal cut-off points were determined using the Youden index. Bootstrapping was applied for internal validation.
Results
Among 10,887 participants with complete data, 58 (0.53%) were diagnosed with lung cancer during a maximum follow-up period of 6.6 years. High levels of all four inflammatory indexes were associated with increased lung cancer risk. Sex-stratified analyses revealed stronger associations between Neutrophil-to-Lymphocyte Ratio (NLR) and Systemic Immune Inflammatory Index (SII) and lung cancer in men, and between C-reactive protein to albumin ratio (CAR) and lung cancer in women. Model discrimination improved after adding inflammatory indexes.
Conclusion
Our findings support sex-specific lung cancer risk assessment and personalised screening strategies. Pre-diagnostic inflammatory indexes may serve as biomarkers to complement existing screening criteria. Further research is needed to validate their potential in improving lung cancer prediction models.
Keywords: Lung cancer screening, Sex differences, Systemic inflammatory indexes, Lolland-Falster Health Study, Biomarkers, Personalised medicine
Plain Language Summary
Lung cancer is the most common cause of cancer death worldwide. A way to find people at risk of developing lung cancer early is through special health checks. However, not everyone is invited to these checks. Today, the decision is mostly based on age and how much a person has smoked. These rules often leave women out, even though women can also get lung cancer with less smoking. One possible way to improve this process is to utilise information from simple blood tests that many people already undergo. These tests can measure changes in the blood linked to inflammation called “inflammatory indexes.” In our study, we followed more than 10,000 people in Denmark. We looked at their inflammatory indexes and checked who later developed lung cancer. We found that people with higher inflammation indexes were more likely to develop lung cancer. When we looked at women and men separately, some indexes worked better for men and others for women. This means that these indexes could help make early detection of lung cancer more accurate and more fair. They may help determine who should be invited for early checks, allowing both women and men at risk to be diagnosed and treated in time.
Introduction
Lung cancer is the leading cause of cancer-related death worldwide and has the highest cancer mortality rates among men and the second among women [1]. Randomised controlled trials (RCTs) on lung cancer screening with low-dose computed tomography (LDCT) in risk populations have shown that lung cancer screening reduces lung cancer mortality [2]. There is no universal consensus on the definition of the risk group that should be screened; the eligibility criteria vary among RCTs, as they use different inclusion criteria based on age, smoking history, and smoking status to identify the target group for lung cancer screening [3].
In our previous study, differences were observed in outcomes depending on the dichotomous criteria used in the respective RCTs; however, these differences were more pronounced among women than among men [4]. Similar sex differences have been found in other studies using dichotomous criteria and in risk prediction model as the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial model 2012 (PLCOm2012) [5]. Considering these differences, sex disparities require further investigation to ensure equity and to refine screening strategies accordingly [6].
There are sex-based immunological differences that change at different stages of life, varying the incidence of autoimmune, infectious diseases, and cancer between women and men [7]. The immune and inflammatory responses are essential in the tumour microenvironment and cancer development [8]. Circulating blood cells (CBCs), such as neutrophils, lymphocytes, and platelets, as well as levels of circulating proteins, including C-reactive protein (CRP), provide valuable information concerning the inflammatory response [9]. The inflammatory status can be assessed by so-called systemic inflammatory indexes, which are calculated from the CBCs and circulating proteins, such as the NLR, the SII, the Platelet-to-Lymphocyte Ratio (PLR), and the CAR [10]. Studies have shown that elevated inflammatory indexes are related to an increased risk of developing cancer [11, 12], including lung cancer [10, 13–15].
The study aims to contribute to the investigation of sex disparities in the identification of high-risk profiles for lung cancer. We explore the association between age, smoking status, systemic inflammatory indexes, and lung cancer among participants in a Danish population-based health survey and their differences among men and women.
Methods
Study Design and Data Sources
We conducted a population-based cohort study using data from the Lolland-Falster Health Study (LOFUS), linked with the Danish Pathology Data Bank. LOFUS is a population-based health survey conducted between 2016 and 2020 in the provincial municipalities of Lolland and Guldborgsund, Denmark. A random sample of people aged 18 and above, along with their entire household, was invited. Participants provided information on risk factors, diseases, lifestyle, and socioeconomic and family factors in self-administered questionnaires, underwent a simple health examination, and provided biological samples [16]. Information on incident lung cancer diagnoses was obtained through linkage with the Danish Pathology Data Bank, a national database containing data from pathological examinations performed by pathology departments in Denmark. Diagnoses are recorded using the Danish version of the Systematised Nomenclature of Medicine (SNOMED) [17].
Study Population and Variables
The study population consisted of participants in the LOFUS survey aged 40–74 at the time of participation, with complete data on the variables used in this study and without a lung cancer diagnosis before the date of invitation to the survey (Fig. 1). The study sample refers to the proportion of this group filtered according to different age range criteria. The primary analyses were conducted using the age range criterion of 50–74 years, which aligns with the NELSON study (Fig. 1).
Fig. 1.
Flow diagram of the construction of the study population.
From the LOFUS survey participants, we retrieved data on smoking status (categorised as current, former, or never smokers based on the question “Do you smoke?”), circulating blood cells (including neutrophils, lymphocytes, and platelets), and circulating blood proteins (specifically CRP and albumin). We calculated the following systemic inflammatory indexes: Neutrophil-to-Lymphocyte Ratio (NLR), calculated as neutrophil count divided by lymphocyte count; Systemic Immune Inflammatory Index (SII), calculated as neutrophil count multiplied by platelet count and divided by lymphocyte count; Platelet-to-Lymphocyte Ratio (PLR), calculated as platelet count divided by lymphocyte count; and C-reactive protein to albumin ratio (CAR), calculated as CRP divided by albumin. Incident lung cancer cases were identified using SNOMED codes indicative of lung cancer diagnoses recorded in the Danish Pathology Data Bank (Fig. 1; online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000551648) [17].
Statistical Analysis
We conducted analyses using prospective data from LOFUS and the Danish Pathology Data Bank to evaluate the association between systemic inflammatory indexes and lung cancer cases within both sexes and among women and men separately. We first used the Youden index to find the optimal cut-off point for each of the systemic inflammatory indexes. The Youden index measures the difference between the true-positive rate (sensitivity) and the false-positive rate (1-specificity) at various cut-off points on the receiver operating characteristic (ROC) curve [18]. We calculated the Youden index (sensitivity; 1-specificity) for each cut-off value, and the cut-off point with the highest value of this index was considered the optimal cut-off point [19, 20]. These cut-off values, presented in online supplementary Table 2, allowed us to dichotomise each index into “high” and “low” categories. Youden index was selected because no universally accepted cut-off values exist for these biomarkers in the context of lung cancer risk assessment in population-based screening settings. The Youden index is a well-established method for dichotomising continuous variables to maximise discriminative ability by balancing sensitivity and specificity, thereby limiting an excessive increase in false-positive classifications. Importantly, this data-driven approach allows cut-off values to be derived directly from the study population rather than relying on thresholds defined in other populations with different demographic, clinical, or epidemiological characteristics. The method is transparent, reproducible, and independent of outcome prevalence, making it particularly suitable for exploratory analyses in low incidence conditions.
To explore the relationship between systemic inflammatory indexes and smoking status, we conducted a tabulation analysis that examined the frequencies and distributions of different category combinations. Specifically, we analysed the number and proportion of lung cancer cases within subgroups defined by smoking status (current, former, and never smokers), sex, and inflammatory index levels (categorised as high vs. low) (online suppl. Table 4; Fig. 2). To evaluate whether cancers were concentrated in the high versus low category within each (smoking status × biomarker) combination and their statistical significance, we performed a chi-square goodness-of-fit test (online suppl. Table 8; Fig. 2).
Fig. 2.
Proportion of lung cancer cases by smoking status and inflammatory index. Study sample aged 50–74. a Both sexes (total). b Women. c Men. The statistical significance is expressed through p values (‡p ≤ 0.05). Exact p-values are provided in online suppl. Table 8.
Logistic regression analyses were used to assess the association between biomarker levels and the risk of lung cancer. We developed six generalised logistic models adjusted for age, sex, and smoking status. An all-index model includes all systemic inflammatory indexes. Additionally, we have a model for each index: the NLR model, the SII model, the PLR model, and the CAR model. We also developed a base model, which does not include indexes assessing the association between smoking status and lung cancer adjusted for age and sex. For the primary outcomes, we applied all models to the study sample aged 50–74. To explore potential differences in model performance between men and women, we further stratified these models by sex to analyse the model performance separately for males and females.
Internal validation was performed using non-parametric bootstrapping to assess the stability and reliability of the model estimates [21]. A total of 1,000 bootstrap samples of equal size to the original dataset were generated. For each resample, the logistic regression model was refitted, and the statistics of interest (e.g., area under the ROC curve [AUC] or odds ratio [OR] for a specific biomarker) were recalculated. This approach allowed us to estimate ORs and the 95% confidence intervals (CIs) for the dichotomised biomarker variables (e.g., comparing “high” vs. “low” groups; see Table 2) and to evaluate the discriminatory performance of each logistic model by calculating the AUC. For each model, we plotted the distribution of bootstrapped AUC values to visualise their variability and stability across resamples. These distributions (shown in Figure 3a, b) also facilitate comparisons between models.
Table 2.
Association between systemic inflammatory indexes and lung cancer risk, adjusted by sex, age, and smoking status
| Bootstrapping of the estimations of the logistic regression – combined model with all indexes and both sexes | ||
|---|---|---|
| | OR | 95% CI |
| (a) | ||
| Sex1 | ||
| Reference | 1 | − |
| Woman | 1.49 | 0.89; 2.59 |
| Age2 | ||
| Reference | 1 | − |
| 55–59 | 0.85 | 0.21; 4.29 |
| 60–64 | 2.15 | 0.81; 9.40 |
| 65–69 | 2.33 | 0.92; 12.02 |
| 70–74 | 2.96 | 1.16; 14.14 |
| Smoking status3 | ||
| Reference | 1 | − |
| Former | 0.54 | 0.28; 1.00 |
| Never | 0.13 | 0.04; 0.28 |
| Bootstrapping of the estimations of the logistic regression – model with each index within both sexes and woman and men separately | ||||||
|---|---|---|---|---|---|---|
| | both sexes | women | men | |||
| OR | 95% CI | OR | 95% CI | OR | 95% CI | |
| (b) | ||||||
| Index4 | ||||||
| NLR high | 2.05 | 1.16; 4.12 | 1.63 | 0.82; 3.59 | 3.69 | 1.33; 12.39 |
| SII high | 1.94 | 1.15; 3.55 | 1.53 | 0.74; 3.10 | 2.92 | 1.27; 9.52 |
| PLR high | 1.97 | 1.15; 3.87 | 2.03 | 0.96; 5.63 | 2.02 | 0.89; 6.5 |
| CAR high | 2.74 | 1.43; 6.89 | 3.79 | 1.49; 21.1 | 1.81 | 0.73; 7.3 |
Odds ratios (OR) and 95% confidence intervals (CI) for lung cancer risk by inflammatory index. For all the indexes, the risk is calculated above the cut-off point (high category). All the estimation are medians based on 1,000 bootstrap replicates.
1–3The estimates reflected are subtracted from the logistic regression model including age, smoking status, sex, and all indexes as independent variables. The reference is a man between 50 and 54 years old and current smoker.
4The estimates reflected are subtracted from the logistic regression model including age, smoking status, sex, and one of the indexes as independent variables.
Fig. 3.
Bootstrap distribution of AUC by model. Study sample aged 50–74. a Both sexes. b Women and men separately. The vertical black dotted line is the median bootstrap AUC. The shape of the distributions indicates the statistical variability of the AUC: symmetrical and centered distributions indicate stability.
We report the median of the bootstrapped estimates for ORs, as the median provides a robust measure of central tendency that is less influenced by skewed distributions or outliers. Sex-stratified bootstrap estimates were also computed to assess potential heterogeneity in biomarker model performance across sexes. All statistical analyses were performed in R (version 4.4.1), using the stats, pROC, and boot packages. This study was reported in accordance with the STROBE guidelines, and the completed STROBE checklist is provided in the online Supplementary Material.
Outcomes and Sensitivity Analyses
Primary analyses were conducted in participants aged 50–74 years, corresponding to the age inclusion criterion of the Nederlands-Leuvens Longkanker Screenings Onderzoek (NELSON) trial. This age range was selected based on findings from our previous work showing lower sensitivity of NELSON eligibility criteria compared with the National Lung Screening Trial (NLST), particularly among women.
Primary outcomes included the following: (1) the determination of the cut-off point for each systemic inflammatory index using the Youden method (online suppl. Table 2); (2) the frequencies and distributions of lung cancer in subgroups defined by smoking status, sex, and inflammatory index levels (online suppl. Table 4; Fig. 2); (3) the assessment of differences in the proportion of lung cancer cases between high and low index categories within each smoking status-biomarker combination using chi-square goodness-of-fit test (online suppl. Table 8; Fig. 2); (4) the association between high inflammatory index category and lung cancer risk, expressed as ORs with 95% CIs (Given the limited number of lung cancer cases in our sample, ORs are interpreted as risk estimates [22, 23]) (Table 2); and (5) the AUC distributions based on 1,000 bootstrap samples to assess the discriminatory performance of the models. The shape of these distributions provides information about statistical variability, with symmetrical and centred distributions indicating model stability (Fig. 3a, b).
Sensitivity analyses were performed to assess the robustness of the results to changes in the age eligibility criteria. The same analytical framework was applied to participants aged 40–74 (to explore ages below 50, which is the usual minimum age cut-off; this may be of interest in the case of women) and 55–74 years, which meets the NLST age inclusion criterion (online suppl. Tables 6, 7, Fig. 1, 2).
Follow-Up and Time to Event
LOFUS participants with a history of lung cancer prior to their visit to the survey were excluded. LOFUS visits began on January 8, 2016, and lung cancer diagnoses were tracked using the Danish Pathology Data Bank, data up to August 31, 2022. Follow-up is the period from the date of the participants’ visit to the LOFUS survey until August 31, 2022. Time to event refers to the number of days between the participants’ visit to the LOFUS survey and the lung cancer case. If there is no lung cancer case, the time to event is the number of days between the study visit and the end of follow-up.
Results
Baseline Characteristics
A total of 10,887 LOFUS participants aged 40–74 years were included in the study, of whom 58 (0.53%) were diagnosed with lung cancer. The maximum follow-up period in our study was 6.6 years (median: 4.2 years; range: 0.1–6.6 years). Women accounted for 53.9% of the study population and represented a higher proportion of lung cancer cases than men. Lung cancer cases occurred at a younger mean age in women than in men (65.6 vs. 70.3 years). Regarding smoking status, 19.5% of the total cohort were current smokers, 37.1% were former smokers, and 43.4% had never smoked. Lung cancer was most prevalent among former smokers (48.3%), followed by current smokers (38%), and was least common among never smokers (13.8%). This pattern was consistent across sexes. Across inflammatory indexes, individuals in the high category for NLR, SII, PLR, and CAR had a higher rate of lung cancer. Detailed baseline characteristics are presented in Table 1.
Table 1.
Baseline characteristics of the study population age 40–74
| Characteristics | Total | Women | Men | |||
|---|---|---|---|---|---|---|
| N (%) | lung cancer, n (%) | N (%) | lung cancer, n (%) | N (%) | lung cancer, n (%) | |
| Participants1 | 10,887 (100) | 58 (100) | 5,871 (53.9) | 35 (60.3) | 5,016 (46.1) | 23 (39.7) |
| Sex | | | | | | |
| Women | 5,871 (53.9) | 35 (60.3) | 5,871 (100) | 35 (100) | − | − |
| Men | 5,016 (46.1) | 23 (38.7) | − | − | 5,016 (100) | 23 (100) |
| Age | ||||||
| 40–44 | 958 (8.8) | <5 | 572 (9.7) | <5 | 386 (7.7) | 0 (0) |
| 45–49 | 1,227 (11.3) | <5 | 667 (11.4) | 0 (0) | 560 (11.2) | 0 (0) |
| 50–54 | 1,565 (14.4) | 5 (8.6) | 885 (15.1) | 5 (14.3) | 680 (13.6) | 0 (0) |
| 55–59 | 1,715 (15.8) | 5 (8.6) | 934 (15.9) | <5 | 781 (15.6) | <5 |
| 60–64 | 1,820 (16.7) | 14 (24.1) | 969 (16.5) | 9 (25.7) | 851 (17) | 5 (21.7) |
| 65–69 | 1,885 (17.3) | 16 (27.6) | 985 (16.8) | 9 (25.7) | 900 (17.9) | 7 (30.4) |
| 70–74 | 1,717 (15.8) | 17 (29.3) | 859 (14.6) | 8 (22.9) | 858 (17.1) | 9 (39.1) |
| Smoking status | ||||||
| Current | 2,123 (19.5) | 22 (38) | 1,112 (18.9) | 13 (37.1) | 1,011(20.2) | 9 (39.1) |
| Former | 4,036 (37.1) | 28 (48.3) | 2,053 (35) | 16 (45.7) | 1,983 (39.5) | 12 (52.2) |
| Never | 4,728 (43.4) | 8 (13.8) | 2,706 (46.1) | 6 (17.1) | 2,022 (40.3) | <5 |
| NLR | ||||||
| High | 6,088 (55.9) | 43 (74.1) | 3,095 (52.7) | 23 (65.7) | 2,993 (59.7) | 20 (87) |
| Low | 4,799 (44.1) | 15 (25.9) | 2,776 (47.3) | 12 (34.3) | 2,023 (40.3) | <5 |
| SII | ||||||
| High | 4,632 (42.5) | 36 (62.1) | 2,633 (44.8) | 20 (57.1) | 1,999 (39.8) | 16 (69.6) |
| Low | 6,255 (57.4) | 22 (37.9) | 3,238 (55.2) | 15 (42.9) | 3,017 (60.2) | 7 (30.4) |
| PLR | ||||||
| High | 6,312 (58) | 42 (72.4) | 3,609 (61.5) | 26 (74.3) | 2,703 (53.9) | 16 (69.6) |
| Low | 4,575 (42) | 16 (27.6) | 2,262 (38.5) | 9 (25.7) | 2,313 (46.1) | 7 (30.4) |
| CAR | ||||||
| High | 6,612 (60.7) | 48 (82.8) | 3,668 (62.5) | 30 (85.7) | 2,944 (58.7) | 18 (78.3) |
| Low | 4,275 (39.3) | 10 (17.2) | 2,203 (37.5) | 5 (14.3) | 2,072 (41.3) | 5 (21.7) |
Mean age at diagnosis both sexes: 67.5, among women: 65.6, among men: 70.3.
1Participants without missing data.
Primary Outcomes among the Study Sample Aged 50–74
Baseline characteristics of the study sample aged 50–74 are presented in online supplementary Table 3. Tabulation analysis by smoking status and inflammatory index showed that among current smokers, a higher number of lung cancer cases were observed in participants with elevated inflammatory markers across all indexes. The percentage of cancer cases with a high index was higher in women current smokers compared with men across all indexes (online suppl. Table 4). In women current smokers, all lung cancer cases occurred in those with high CAR, representing 100% of cancers among women who were current smokers and 38.2% of all cancers among women. In men, most cases clustered in the high NLR group (77.8% of cancers among men who were current smokers and 30.4% of all cancers among men) (online suppl. Table 4; Fig. 2). Among former smokers, 91.7% of the lung cancer cases in men occurred in those with high NLR, and the same high proportion showed elevated CAR. In women who were former smokers, the majority of lung cancer cases were observed in participants with high CAR and PLR. Among never smokers, lung cancer cases were 7 out of 57 (12.3%) with more cases observed in individuals with high inflammatory indexes across all indexes, being the highest proportion of lung cancer cases among women with high PLR (online suppl. Table 4; Fig. 2).
Figure 2 displays the distribution of lung cancer cases across high vs. low categories of each inflammatory index within smoking status strata, both sexes, and stratified by sex. Among current smokers (both sexes), high CAR, NLR, and SII were significantly associated with a higher proportion of cancers (p ≈ 6.5 × 10−4, 0.0105, and 0.0105, respectively). Among former smokers (both sexes), high CAR, NLR, and PLR were significant (p ≈ 2 × 10−4, 0.023, and 0.023). In never smokers (overall), only PLR reached significance (p = 0.008). Sex-stratified results showed significance for CAR and SII in women who were current smokers (p ≈ 3.1 × 10−4 and 0.0126), a trend for NLR (p = 0.052), and significance for CAR in women who were former smokers (p = 0.012), and within women who were never smokers, PLR had significance (p ≈ 0.025). In men who were former smokers, CAR and NLR were significant (both p ≈ 0.0039) (online suppl. Table 8).
The association between inflammatory indexes and lung cancer is reported in Table 2, using logistic regression models adjusted for age, sex, and smoking status. In the model with all indexes, being female was associated with a higher, though non-significant, risk of lung cancer compared to males. Older age groups generally showed increased risk, with the highest risk observed in the 70–74 age group. Compared to current smokers, former smokers had a reduced risk, and never smokers showed a substantially lower risk. In the models with each index and both sexes, higher values of all inflammatory indexes were significantly associated with increased lung cancer risk. In analyses including both sexes, CAR showed the strongest association (OR = 2.74; 95% CI: 1.43–6.9), followed by NLR (NLR: OR = 2.05; 95% CI: 1.16–4.12), SII (OR = 1.94; 95% CI: 1.15–3.55), and PLR (OR = 1.97; 95% CI: 1.15–3.87). Sex-stratified analyses revealed that men exhibited stronger associations for NLR (OR = 3.69; 95% CI: 1.33–12.39) vs. women (OR = 1.63; 95% CI: 0.82–3.59) and SII (OR = 2.92; 95% CI: 1.27–9.52) vs. women (OR = 1.53; 95% CI: 0.74–3.10), whereas CAR was most predictive among women (OR = 3.79; 95% CI: 1.49–21.06) vs. men (OR = 1.81; 95% CI: 0.73;7.3).
Regarding the discriminatory performance of the models, the addition of inflammatory indexes improved model discrimination compared with the base model including age, sex, and smoking status alone. In analyses including both sexes, the base model yielded an AUC of 0.752, which increased to 0.792 in the model including all inflammatory indexes (Fig. 3a).
Sex-stratified analyses showed higher overall discrimination in men, with particularly strong performance for NLR and SII model (AUC 0.818 for both). The base model had the lowest AUC (0.789), and the model with all indexes had the highest (0.841). In contrast, among women, CAR demonstrated the best performance (AUC = 0.776). The base model had an AUC of 0.742, while the model with all indexes had an AUC of 0.797 in women (Fig. 3b).
Overall, the addition of inflammatory indexes improved model performance across all groups, with variations observed between sexes. The predictive utility of NLR and SII appeared stronger and statistically significant in men, while CAR showed superior performance in women. Significant associations were observed for NLR and SII in men (p = 0.05 and p = 0.02, respectively) and for CAR in women (p = 0.02). Exploratory analyses in sex by biomarker interaction between high SII and lung cancer among the 70–74 year group yielded p = 0.05, although estimates were based on small numbers and should be interpreted with caution.
Secondary Outcomes
In our sensitivity analyses applying alternative age eligibility criteria, 40–74 (Table 1; online suppl. Table 5) and 55–74 years, the ORs for NLR and SII remain high and significant in men compared to women, and the OR for CAR remained high and significant in women compared to men, across both age criteria, supporting our primary outcomes (online suppl. Tables 6, 7; online suppl. Fig. 1, 2). These analyses highlight the differences between sexes and age regarding the risk of lung cancer, depending on the index. An extended narrative description of the results is provided in the online supplementary Results Appendix.
Discussion
Main Findings
In this population-based cohort study, we found that elevated systemic inflammatory indexes derived from pre-diagnostic blood tests were associated with an increased risk of lung cancer, beyond established risk factors such as age and smoking status. Importantly, these associations differed by sex, with stronger effects of NLR and SII observed in men, while CAR showed the strongest association and discriminatory performance in women. The inclusion of inflammatory indexes consistently improved the predictive performance of lung cancer risk models, both overall and in sex-stratified analyses. Together, these findings support the relevance of sex-specific inflammatory profiles in lung cancer risk assessment and suggest that systemic inflammatory markers may contribute to more personalised approaches to screening.
Lung Cancer Screening and Sex Disparities
Sex plays a role in the carcinogenesis, development, treatment, and prognosis of lung cancer [24]. Some studies have revealed differences between men and women in terms of lung cancer risk factors, screening practices, molecular mechanisms, treatment, and survival outcomes [6]. Our previous Danish study identified differences in the population likely to benefit from screening, with women being underrepresented. The eligibility criteria for the largest lung cancer screening studies, NLST and NELSON, have significantly lower sensitivity for women compared to men [4, 25]. Similar findings by Pasquinelli et al. [26] showed that women were less likely to meet the United States Preventative Services Task Force (USPSTF) lung screening criteria, and risk-based models, such as PLCOm2012, reduce but do not eliminate the sex gaps in screening sensitivity. Pinsky et al. [27] have shown that existing guidelines for lung cancer screening reveal some sex disparities. They found that even with expanded criteria, such as the 2021 USPSTF proposal, eligibility rates remain consistently higher among men compared to women. The same happened in our sensitivity analysis in which we tried two different age inclusion criteria: one identical to the NLST age criterion (55–74 years) and the other encompassing the entire study population (40–74 years), as illustrated in Figure 1. Comparing the age criteria of 40–74 with 55–74, the AUC improves for all indexes and models among women but remains lower compared with men (online suppl. Table 7; online suppl. Fig. 1, 2). Rivera et al. [28] conducted a literature review of Lung Cancer Screening (LCS), finding sex disparities in USPSTF 2013. They issued an Official Statement from the American Thoracic Society proposing strategies such as evaluating inclusion criteria to capture better high-risk individuals who do not meet the eligibility criteria, like women [28]. This statement aligns with the efforts in our study to revise eligibility criteria. Our results suggest that incorporating such changes may help address structural biases in current practices and improve outcomes for historically underserved groups.
These sex disparities may arise from not considering sex differences in lung cancer risk, like in smoking behaviour in women starting smoking at a later age and smoking less intensively than men, which do not fully explain increased lung cancer rates in young women and the trend of lung cancer being diagnosed in women at younger ages than in men [29, 30]. In our study, the mean age at diagnosis was 67.5 years for both sexes. Among women, it was 65.6 years, and among men, it was 70.3 years. Our tabulation analysis by smoking status and inflammatory index suggests that elevated inflammatory indexes are associated with higher lung cancer risk regardless of smoking status, showing a stronger association among current and former smokers.
Systemic Inflammatory Indexes and Lung Cancer
The observed sex differences in our findings suggest that inflammatory responses and systemic biomarkers may play distinct roles in the risk of lung cancer in men and women. Fan et al. [31] explored sex differences in immune cell infiltration in the lung tumour microenvironment, finding differences between male and female patients with lung adenocarcinoma. Among the tumour-infiltrating immune cells, they found that neutrophils accounted for a higher proportion in men compared with women in paracancerous and cancerous tissues [31]. Xie et al. [32] analysed the potential regulatory role of sex hormones in cancer-related inflammation. In their study of an all-male population, they found that albumin levels in the cancer group were significantly lower compared with the non-cancer group; neutrophil counts in the cancer group were higher compared with the non-cancer group; and the NLR was significantly elevated in the cancer group [32]. Our findings revealed a stronger association between NLR and lung cancer among men (Table 2).
Grieshober et al. [33] calculated the NLR index of systemic inflammation derived from DNA methylation data by dividing the estimated proportion of neutrophils by the estimated proportion of lymphocytes (obtaining methylation-derived NLR [mdNLR]), similar to the traditional NLR calculated from the CBC. They evaluated the mdNLR and the risk of lung cancer in individuals at high risk due to exposure to tobacco or asbestos. Their findings showed that mdNLR was higher in cases than in controls (2.06 vs. 1.86, p = 0.03), and the relationship between mdNLR and lung cancer showed an OR 1.21; 95% CI 1.01–1.45 [33]. This aligns with our results, where the NLR model showed that the association between NLR high and lung cancer had an OR of 2.05 (95% CI: 1.16–4.12). Grieshober et al. [33] highlighted the importance of a better understanding of the mechanism between NLR and cancer as a potentially useful marker for prevention and early detection in high-risk populations, which could be included in a screening program.
In our study, we found that high levels of NLR and PLR were significantly related to an increased risk of lung cancer in both sexes, and we also observed that there were differences between men and women, with the relationship between NLR and lung cancer being significantly higher in men than in women. This finding is consistent with the results reported by Budin et al. [34], who aimed to investigate the relationship between inflammation and carcinogenesis. They observed that NLR and PLR were significantly higher in patients with lung cancer and also found correlation differences between NLR values and the probability of developing lung cancer according to sex, being higher among men than among women. They also compared the discriminative power between models with and without the indexes, observing superior discriminative power between cancer and control patients, in the model with age, gender, NLR, and PLR indexes compared with a model without the indexes (AUC of 0.777 [0.772–0.782] vs. 0.700 [0.694–0.706]) [34]. In our study, the discriminative power also increased when we added each of the indexes as well as when we added all of them to the model, compared to the discriminative power of the model without indexes.
Regarding the SII index, Fest et al. [12] investigated the association between SII and incident cancers in a prospective cohort study. Their findings showed that the risk of developing a solid cancer (including lung cancer) after a high baseline SII was significantly higher within the first 6 months after baseline, with an HR of 2.00 (95% CI: 1.09–3.67). The risk persisted over time and increased for individuals with longer follow-up [12]. In our study, high SII values were significantly associated with an increased risk of lung cancer.
Zhang et al. [35] investigated the role of the CAR in patients with operable NSCLC, finding that CAR was significantly related to sex and smoking status. CRP is the numerator of the CAR index. Chaturvedi et al. [36] investigated the association of CRP and lung cancer risk. They found that elevated CRP levels were associated with an increased risk of lung cancer in current and former smokers, and this association was strongest in the period 2 to 5 years before lung cancer diagnosis [36]. Muller et al. [37] conducted an analysis investigating the association between circulating hsCRP concentration and risk of lung cancer by smoking status, finding a positive association stronger in the first 2 years of follow-up, particularly among current smokers. This underscores its potential in lung cancer prevention and screening.
Strengths and Limitations
This study has several strengths. It leverages data from a population-based cohort (LOFUS) with linkage to national pathology records, enabling accurate case identification and providing demographic coverage of one of the areas with the highest percentage of daily smokers in the Zealand region of Denmark [38]. The use of systemic inflammatory indexes derived from routine blood tests offers a low-cost and widely accessible strategy to enhance lung cancer risk stratification. Blood samples were collected at the time of participation in the LOFUS survey, and no participant included in our study had a diagnosis of lung cancer at that time, suggesting the potential of systemic inflammatory markers as pre-diagnostic measurements for future screening-oriented studies. Moreover, the sex-stratified approach enabled us to identify potential sex-specific differences in the predictive performance of these indexes, thereby potentially contributing to more equitable and personalised screening strategies. We also applied robust statistical methods, including logistic regression models adjusted for age, sex, and smoking status, along with internal validation through 1,000 bootstrap replicates. Sensitivity analyses across different age ranges further confirmed the stability of our findings.
Future studies should explore whether repeated or longitudinal measurements of systemic inflammatory indexes improve lung cancer risk stratification compared with single baseline assessments. In particular, characterising temporal changes in inflammatory profiles could help distinguish transient inflammatory elevations from persistent patterns and clarify how early such changes occur in relation to lung cancer diagnosis. This information may be especially relevant in screening-oriented settings, where identifying early risk signals is critical for timely intervention.
From a screening perspective, systemic inflammatory indexes derived from peripheral blood may represent an initial, minimally invasive step for population-level risk stratification. While the present study does not evaluate downstream diagnostic procedures, integrating such blood-based risk signals within a broader, stepwise screening pathway could help inform subsequent assessment and contribute to more personalised screening strategies.
Nonetheless, some limitations should be noted. The relatively small number of lung cancer cases (n = 58) may limit the statistical power of subgroup analysis. To mitigate this, we performed internal validation using non-parametric bootstrapping, complemented by sensitivity analyses across alternative age eligibility criteria. Although sex-stratified analyses yielded wider CIs due to smaller subgroup sample sizes, the associations remained statistically significant. Nevertheless, our findings should be interpreted cautiously, and external validation in larger and independent cohorts is required to confirm the stability, generalisability, and potential screening utility of the proposed approaches.
Additionally, systemic inflammatory indexes can be influenced by factors such as chronic diseases, infections, or medications. From an aetiological perspective, this may contribute to residual confounding and limits causal interpretation of the observed associations. However, our objective was to assess the potential of these indexes as a screening tool; we did not exclude participants with such conditions. These factors may reflect real-world risk and may be common in the population targeted for screening; adjusting for them could reduce the practical applicability of the model in clinical settings with screening purposes. There were limitations in the smoking intensity data, as this variable was not available in our dataset for former smokers; therefore, only smoking status was included in this study. Sensitivity analyses were performed in our population sample aged 50–74 (data not shown), in which pack-year data in current smokers were added to the model, increasing all ORs of all indexes in both sexes and between men and women separately, except for CAR in women and PLR in men, which was slightly lower. Another limitation is that there are no standardised values as a reference for the cut-off points of the indexes. The resulting cut-off values in our study were used for exploratory risk stratification and model comparison and should be interpreted cautiously and validated in independent external cohorts before any clinical application. Finally, while our findings are based on a Danish cohort, their generalisability to other populations may be limited and requires external validation.
Conclusion
To our knowledge, this is the first study to identify pre-diagnostic inflammatory indexes derived from a simple blood test that enhance the lung cancer prediction model in a female population. Across sexes, several indexes demonstrated a stronger relation with risk of lung cancer, depending on the index and the sex, underscoring a relevant sex-specific effect. Our sensitivity analysis underscores the importance of expanding age criteria, highlighting differences between sexes and age groups in the risk of lung cancer, which should be further investigated in larger cohorts. Given their simplicity, low cost, and availability in routine clinical practice, these biomarkers could serve as a complementary tool to refine risk stratification, especially within specific subgroups defined by age and sex.
Future multicentre and prospective studies are required to validate these findings, explore the underlying mechanisms driving these sex differences, evaluate the potential role of systemic inflammatory biomarkers in lung cancer early detection, define cut-off values, and establish their role in real-world settings. Our study could inspire new approaches and strategies for future lung cancer screening in Denmark, thereby striving towards ensuring equity between sexes.
Statement of Ethics
This study was registered with the Danish Data Protection Agency (p-2024-16362), and Region Zealand approved the use of patient record data from the Danish Pathology Data Bank (EMN-2022-05344). LOFUS was approved by Region Zealand’s Ethical Committee on Health Research (SJ-421) and registered with the Danish Data Protection Agency (p-2024-16360) and Clinical Trials (NCT02482896). The LOFUS participants provided written informed consent.
Conflict of Interest Statement
Zaigham Saghir has received payment for educational services from Chiesi Pharma. All other authors have no conflicts of interest to declare.
Funding Sources
This work was supported by Danish Cancer Society [R311-A18195]; Region Zealand Health Scientific Research Foundation [R22-A597], Helsefonden [22-B-0096], and Zealand University Hospital, Nykøbing F. [Grant No. N/A]. The funder had no role in the design, data collection, data analysis, and reporting of this study.
Author Contributions
María del Pilar Fernández Montejo: conceptualisation, investigation, data curation, methodology, formal analysis, writing ‑ original draft, writing ‑ review and editing, visualisation, and decision to publish. Uffe Bodtger and Zaigham Saghir: supervision, methodology, writing ‑ original draft, writing – review, and decision to publish. Randi Jepsen: supervision, resources, data curation, writing – review, and decision to publish. Søren Lophaven: methodology, formal analysis, data curation, writing – review, and decision to publish. Ioannis Fotopoulos and Oluf Dimitri Røe: methodology, writing – review, and decision to publish.
Funding Statement
This work was supported by Danish Cancer Society [R311-A18195]; Region Zealand Health Scientific Research Foundation [R22-A597], Helsefonden [22-B-0096], and Zealand University Hospital, Nykøbing F. [Grant No. N/A]. The funder had no role in the design, data collection, data analysis, and reporting of this study.
Data Availability Statement
Due to restrictions on the availability of the data, which were used under license for this study, access to and use of the data require approval from the respective data providers. Further enquiries can be directed to the corresponding author.
Supplementary Material.
Supplementary Material.
References
- 1. Cancer today [Internet]. [cited 2024 Dec 2]. Available from: https://gco.iarc.who.int/today/
- 2. Bonney A, Malouf R, Marchal C, Manners D, Fong KM, Marshall HM, et al. Impact of low-dose computed tomography (LDCT) screening on lung cancer-related mortality. Cochrane Database Syst Rev. 2022;8(8):CD013829. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Hardavella G, Frille A, Sreter KB, Atrafi F, Yousaf-Khan U, Beyaz F, et al. Lung cancer screening: where do we stand? Breathe. 2024;20(2):230190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Fernández Montejo MDP, Saghir Z, Bødtger U, Jepsen R, Lynge E, Lophaven S. Identifying the population to be targeted in a lung cancer screening programme in Denmark. BMJ Open Respir Res. 2024;11(1):e002499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Walter J, Kauffmann‐Guerrero D, Muley T, Reck M, Fuge J, Günther A, et al. Comparison of the sensitivity of different criteria to select lung cancer patients for screening in a cohort of German patients. Cancer Med. 2023;12(7):8880–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Baiu I, Titan AL, Martin LW, Wolf A, Backhus L. The role of gender in non-small cell lung cancer: a narrative review. J Thorac Dis. 2021;13(6):3816–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Klein SL, Flanagan KL. Sex differences in immune responses. Nat Rev Immunol. 2016;16(10):626–38. [DOI] [PubMed] [Google Scholar]
- 8. Faria SS, Fernandes PC, Silva MJB, Lima VC, Fontes W, Freitas-Junior R, et al. The neutrophil-to-lymphocyte ratio: a narrative review. Ecancermedicalscience. 2016;10:702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Sylman JL, Mitrugno A, Atallah M, Tormoen GW, Shatzel JJ, Tassi Yunga S, et al. The predictive value of inflammation-related peripheral blood measurements in cancer staging and prognosis. Front Oncol. 2018;8:78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Şahin F, Aslan AF. Relationship between inflammatory and biological markers and lung cancer. J Clin Med. 2018;7(7):160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tamas F, Tamas CI, Suciu BA, Manu DR, Cehan AR, Balasa AF. Comparative analysis of inflammatory indexes in lung cancer patients with and without brain metastases. Cureus. 2024;16(5):e60921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Fest J, Ruiter R, Mulder M, Groot Koerkamp B, Ikram MA, Stricker BH, et al. The systemic immune-inflammation index is associated with an increased risk of incident cancer-A population-based cohort study. Int J Cancer. 2020;146(3):692–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Mazzella A, Maiolino E, Maisonneuve P, Loi M, Alifano M. Systemic inflammation and lung cancer: is it a real paradigm? Prognostic value of inflammatory indexes in patients with resected non-small-cell lung cancer. Cancers. 2023;15(6):1854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Yu Y, Qian L, Cui J. Value of neutrophil-to-lymphocyte ratio for predicting lung cancer prognosis: a meta-analysis of 7,219 patients. Mol Clin Oncol. 2017;7(3):498–506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Goksel S, Ozcelik N, Telatar G, Ardic C. The role of hematological inflammatory biomarkers in the diagnosis of lung cancer and in predicting TNM stage. Cancer Invest. 2021;39(6–7):514–20. [DOI] [PubMed] [Google Scholar]
- 16. Jepsen R, Egholm CL, Brodersen J, Simonsen E, Grarup J, Cyron A, et al. Lolland-Falster Health Study: Study protocol for a household-based prospective cohort study. Scand J Public Health. 2020;48(4):382–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Erichsen . Existing data sources for clinical epidemiology: the Danish National Pathology Registry and Data Bank. CLEP; 2010. Vol. 51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. IBM SPSS statistics [Internet]. 2025. [cited 2025 June 13]. Available from: https://www.ibm.com/docs/en/spss-statistics/30.0.0?topic=schemes-area-under-curve
- 19. Smith LEF. Methodology for the selection of optimal cutoff point to dichotomize continuous covariates. Rev Cub Gen. 2013;7(3):36–42. [Google Scholar]
- 20. Martín-Conejero A. Validación de test diagnósticos (segunda parte). Test cuantitativos. Angiología. 2023;75(3):181–5. [Google Scholar]
- 21. Bootstrapping in R [Internet] . DataCamp. [cited 2025 June 3]. Available from: https://www.datacamp.com/doc/r/bootstrapping
- 22. Ranganathan P, Aggarwal R, Pramesh CS. Common pitfalls in statistical analysis: odds versus risk. Perspect Clin Res. 2015;6(4):222–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. 9.2.2.1 risk and odds. [Internet]. [cited 2024 Aug 29]. Available from: https://handbook-5-1.cochrane.org/chapter_9/9_2_2_1_risk_and_odds.htm
- 24. Gee K, Yendamuri S. Lung cancer in females: sex-based differences from males in epidemiology, biology, and outcomes – a narrative review. Transl Lung Cancer Res. 2024;13(1):163–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Livanou ME, Syrigos NK, Montgomery A, Moeckel C, Panagiotou E, Charpidou A, et al. Eligibility for screening with low-dose CT in a real-world cohort of patients with lung cancer in Greece: a brief report. Lung Cancer. 2023;186:107424. [DOI] [PubMed] [Google Scholar]
- 26. Pasquinelli MM, Tammemägi MC, Kovitz KL, Durham ML, Deliu Z, Guzman A, et al. Addressing sex disparities in lung cancer screening eligibility. Chest. 2022;161(1):248–56. [DOI] [PubMed] [Google Scholar]
- 27. Pinsky PF, Lau YK, Doubeni CA. Potential disparities by sex and race or ethnicity in lung cancer screening eligibility rates. Chest. 2021;160(1):341–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Rivera MP, Katki HA, Tanner NT, Triplette M, Sakoda LC, Wiener RS, et al. Addressing disparities in lung cancer screening eligibility and healthcare access. An official American thoracic Society statement. Am J Respir Crit Care Med. 2020;202(7):e95–112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Jemal A, Miller KD, Ma J, Siegel RL, Fedewa SA, Islami F, et al. Higher lung cancer incidence in young women than young men in the United States. N Engl J Med. 2018;378(21):1999–2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Senthil P, Kuhan S, Potter AL, Jeffrey Yang C-F. Update on Lung Cancer Screening Guideline. Thorac Surg Clin. 2023;33(4):323–31. [DOI] [PubMed] [Google Scholar]
- 31. Fan T, Li C, He J. Prognostic value of immune-related genes and comparative analysis of immune cell infiltration in lung adenocarcinoma: sex differences. Biol Sex Differ. 2021;12(1):64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Xie W, Zhang Z, Zhao D, Zhang Y, Feng C, Zhou Y. Cancer disrupts sex hormone-inflammation relationships: analysis of ALI in males from NHANES 2007–2018. PLoS One. 2025;20(6):e0325796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Grieshober L, Graw S, Barnett MJ, Thornquist MD, Goodman GE, Chen C, et al. Methylation-derived neutrophil-to-lymphocyte ratio and lung cancer risk in heavy smokers. Cancer Prev Res. 2018;11(11):727–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Budin CE, Cocuz IG, Enache LS, Rența IA, Cazacu C, Pătrîntaşu DE, et al. Neutrophil-to-Lymphocyte and platelet-to-lymphocyte ratio: side by side with molecular mutations in patients with non-small cell lung Cancer – the INOLUNG Study. Cancers. 2024;16(16):2903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Zhang F, Ying L, Jin J, Chen K, Zhang N, Wu J, et al. The C-reactive protein/albumin ratio predicts long-term outcomes of patients with operable non-small cell lung cancer. Oncotarget. 2016;8(5):8835–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Chaturvedi AK, Caporaso NE, Katki HA, Wong H-L, Chatterjee N, Pine SR, et al. C-Reactive protein and risk of lung cancer. J Clin Orthod. 2010;28(16):2719–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Muller DC, Larose TL, Hodge A, Guida F, Langhammer A, Grankvist K, et al. Circulating high sensitivity C reactive protein concentrations and risk of lung cancer: nested case-control study within Lung Cancer Cohort Consortium; 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.062333_resj_mat_sundhedsprofil_2021_tilnettetapril2022.pdf. [cited 2025 July 22]. Available from: https://app-rsjdxp-cms-prod-001.azurewebsites.net/media/ze1eei5l/062333_resj_mat_sundhedsprofil_2021_tilnettetapril2022.pdf
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Due to restrictions on the availability of the data, which were used under license for this study, access to and use of the data require approval from the respective data providers. Further enquiries can be directed to the corresponding author.
