Association between C-reactive protein variability and cancer incidence: a longitudinal prospective cohort study

doi:10.21203/rs.3.rs-4894293/v1

Background

The association between inflammation and cancer has been widely studied, but research on the relationship between the magnitude of inflammatory fluctuations and cancer remains limited. Therefore, this study aims to investigate the association between C-reactive protein (CRP) variability and the occurrence of cancer.

Methods

A total of 42,514 participants were included, and their CRP levels were measured over a 4-year period. We used the coefficient of variation (CV) of CRP to quantify the variability in inflammation. Cox regression analysis was used to assess the association between CRP variability and cancer risk after adjusting for potential confounding factors.

Results

High CV of CRP significantly associated with reduced cancer risk (hazard ratio (HR) = 0.84; 95% CI: 0.75–0.94, P = 0.003). Participants with high CRP and low CV had a significantly increased risk of cancer (HR = 1.42; 95% CI: 1.18–1.70, P < 0.001). In the population with long-term stable CRP levels, there is a significant association between the CV and the risk of cancer (HR = 0.81; 95% CI: 0.72–0.92, P = 0.001). Furthermore, the association between the CV and lung cancer was most pronounced. Sensitivity analyses confirmed the stability of the association between CRP variability and cancer.

Conclusion

High CRP variability is significantly associated with reduced risk of cancer, particularly in the case of lung cancer. This provides a new perspective and evidence for exploring the relationship between inflammation and cancer.

Biological sciences/Cancer

Health sciences/Risk factors

C-reactive protein

cancer

variability

prospective

risk

Cancer stands as the second leading cause of global mortality(1), with its incidence showing a concerning annual increase(2). It is estimated that there will be 290,000 new cancer diagnoses and 50,000 cancer-related deaths in 2040, with over 16% of deaths occurring in low- and middle-income countries(3). Furthermore, cancer treatment brings tremendous physical and mental pain and heavy financial burden to patients at the same time(4, 5). This multifaceted issue transcends individual experiences, affecting families, careers, and the broader social fabric(6). As a result, many studies are actively exploring the risk factors associated with cancer, and a comprehensive understanding of the causes of cancer and effective interventions are urgent and necessary.

Inflammation, especially long-term chronic inflammation, is one of the factors contributing to tumor development(7, 8). This phenomenon is underpinned by various mechanisms, wherein inflammation impacts key processes involved in tumorigenesis—ranging from genomic instability, DNA damage, and oncogene activation to compromised tumor suppressor activity(9). Beyond fostering the initial stages of cancer formation, long-term chronic inflammation plays a pivotal role in cancer progression(10, 11); approximately 25% of all cancers are associated with chronic inflammation(12). Among the plethora of peripheral inflammatory markers, C reactive protein (CRP) has been widely studied because it has a long plasma half-life, minimal diurnal variation, and is not dependent on age or sex(13). Notably, a prospective cohort study showed a significant association between elevated CRP levels and cancer risk(14). Our previous investigations have further substantiated this connection, revealing an association between elevated CRP levels and unfavorable survival outcomes across various cancer types(15, 16). These studies demonstrated a strong association between CRP and cancer, but their limitation was that CRP levels were measured only once at baseline.

Nonetheless, cancer occurrence is often associated with the long-term effects of risk factors. CRP, as an acute phase response protein, is susceptible to fluctuations in response to both exogenous and exogenous stimuli, especially in the presence of inflammation, autoimmune disease or cancer(13). Relying solely on CRP levels measured at baseline as a predictor of cancer development imposes limitations. In contrast, employing multiple measurements of CRP over a period of time, encompassing both high and low CRP levels and their fluctuations, provides a comprehensive reflection of peripheral inflammation and the corresponding inflammatory response. Variability, denoting the extent of fluctuation in an indicator over time, serves as a crucial metric. Commonly utilized indicators of variability include the coefficient of variation (CV) and the variation independent of the mean (VIM). In this study, our approach involved scrutinizing changes in CRP over time, expressing variability through the CV, and incorporating VIM as a complementary indicator to explore the intricate association between CRP variability and the occurrence of cancer.

1. Study design and population

Data for this study were obtained from the Kailuan cohort, a prospective cohort located in northern China. In 2006, employees of the Kailuan Group, including retired employees, were invited to participate, and a total of 101,510 employees agreed to take part and provided signed consent forms. The participants underwent questionnaire interviews, health examinations, and laboratory assessments at Kailuan Hospital. Subsequent medical examinations and questionnaires were administered every 2 years. Participants who underwent fewer than three medical examinations between 2006 and 2010 were excluded. All personal information of the participants included in the final analysis was kept confidential. The study protocol adhered to the guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of Kailuan Medical Group, Kailuan Company.

2. Variable evaluation

All participants underwent fasting blood tests, including measurements of CRP, fasting blood glucose, and lipids. Participants with missing blood test results for any of the three medical examinations were excluded. To ensure stability and scientific rigor, participants with CRP levels >15 mg/L were also excluded. Blood measurements were performed using automated analyzers following standard operating procedures. Height and weight were measured by experienced nurses, and the body mass index (BMI) was calculated as weight (in kilograms) divided by height (in meters) squared.

The demographic information included sex, age, educational attainment, and income. Educational attainment above high school level was classified as “high,” and anything below that was considered “low.” Family monthly income exceeding 1000 yuan per person was classified as “high income,” whereas incomes below this threshold were considered “low income.” Lifestyle data, including smoking, alcohol consumption, and physical exercise, were obtained from the patients’ responses to the lifestyle questionnaire.

The primary outcome of this study was cancer incidence, and the secondary outcome was mortality. A prospective follow-up design was employed, tracking all participants from the baseline assessment conducted in 2010-2011 until the occurrence of cancer or death. Death data were obtained from the death certificates provided by the National Vital Statistics Bureau. The follow-up period spanned from the baseline assessment until 2021, during which all outcome events were carefully recorded.

3. Calculation of CRP variability

CRP was measured during three distinct periods: 2006-2007, 2008-2009, and 2010-2011. To assess the CRP variability, we used the CV, a widely accepted measure of variability. The CV was calculated by dividing the mean by the standard deviation (SD).

To strengthen the robustness and reliability of our findings, we also included the variability independent of the mean (VIM) as a supplementary measure. The VIM provides additional insights into the variability of CRP independent of its mean value. The VIM is a statistical transformation of the SD that is specifically designed to be uncorrelated with the mean levels. It was calculated by fitting a curve of the form y = kx^p to a scatter plot of the SD CRP (y-axis) against the mean CRP (x-axis) for all individuals in the cohort. The parameter p is estimated from the data, and k is a constant that can be chosen to ensure that the VIM is on the same scale as the SD. Specifically, if M represents the average value of the mean CRP in the cohort, then k is determined as M^p, and the VIM CRP value for any individual is calculated as VIM CRP = (k SD/x^p). This statistical approach allows the assessment of CRP variability independent of its mean, providing additional insights into the relationship between CRP fluctuations and cancer occurrence. The CV and VIM were categorized into three groups based on tertiles: low, moderate, and high.

VIM = k × SD(CRP)/Mean(CRP)^x

SD(CRP) = constant × Mean(CRP)

k = Mean (Mean(CRP))^x

4. Statistical analysis

Normally distributed continuous variables were summarized using the mean and SD and analyzed using t-tests. Non-normally distributed continuous variables were analyzed using the Wilcoxon rank-sum test. Categorical variables were analyzed using the chi-squared test. This approach ensured the appropriate analysis of different variable types.

We employed Cox regression analysis to investigate the relationship between CRP variability and the risk of cancer. Model A was unadjusted, whereas Model B was adjusted for sex, age, education, and income. In addition to the adjustments made in Model B, Model C was further adjusted for BMI, smoking, alcohol consumption, and physical exercise. We created Restricted Cubic Spline (RCS) plots to visually depict the intricate relationship between continuous CV of CRP and the risk of cancer. Furthermore, we fitted the model using the "TRAJ" program in SAS version 9.4 (SAS Institute) to group individuals with similar patterns of CRP change between 2006 and 2010. We then delved into examining the association between CRP variability and cancer occurrence, specifically within subgroups characterized by distinct patterns of change in CRP trajectories. We counted the incidence of different site-specific cancers during the follow-up period and explored the association between CV of CRP and site-specific cancers. We conducted stratified analyses to examine whether this association persists across populations with different characteristics. Several sensitivity analyses were conducted to ensure the robustness of the results and eliminate confounding factors. Owing to the potential lag in cancer incidence, participants who developed cancer within the initial 2-year follow-up period were excluded from the study. To address the impact of extreme values in CRP measurements on the results, participants with CRP levels exceeding 10 mg/L were also excluded. Acknowledging the potential impact of metabolic factors, including blood glucose and lipid levels, on cancer incidence, we included additional adjustments in our analysis. Furthermore, to emphasize that the identified association was specifically attributed to CRP variability during the follow-up period, we controlled for baseline CRP levels. Additionally, our analyses considered aspirin use as a relevant factor.

VIM as a complementary indicator of variability, we additionally used COX regression analyses and RCS plots to explore the association between VIM of CRP and cancer incidence. In addition, we used Cox regression analysis to explore the association between the CV and death, as well as cancer-specific mortality. The competing risk plots illustrated the relationships between different levels of CRP variability and the occurrence of cancer and death.

All statistical tests were two-sided and were considered significant at P < 0.05. Statistical analysis was performed using SAS (version 9.4) and R (version 4.2.3).

This study included 42,514 participants from Kailuan cohort. The detailed process of participant inclusion and exclusion is shown in Fig. 1 (Figure. 1). Among the included participants, 24% were women and 76% were men, with an average age of 53.24 (11.69) years and a median follow-up time of 10.98 years. Participants with higher CRP variability tended to be men, younger, with lower BMIs (Table. 1). In Supplementary Table 1, we categorized mean CRP as high ( ≧ 3 mg/L) or low (< 3 mg/L) based on clinical experience and stratified participants accordingly to present baseline information. The results showed that participants with higher CRP were more likely to be women, older, with higher BMIs (Table. S1).

We conducted Cox regression analysis to explore the relationship between CRP variability and cancer occurrence. The high CRP group had a 20% higher risk of cancer than the low CRP group (hazard ratio (HR) = 1.20, 95% CI: 1.07–1.33, P < 0.001). We used RCS to observe the association between continuous CV of CRP and cancer risk, and found that as CV increased, HR decreased (Figure. 2 (a)). Subsequently, we stratified CRP variability into three groups based on the tertile intervals of CV of CRP: low, medium, and high. Moderate CRP variability did not show a significant association with cancer risk compared to low variability (adjusted HR = 0.93, 95% CI: 0.84–1.04, P = 0.205), but high variability exhibited a notable protective effect against cancer (adjusted HR = 0.84, 95% CI: 0.75–0.94, P = 0.003). To further investigate the influence of CRP levels on the relationship between variability and cancer risk, we performed an additional Cox regression analysis by cross-classifying CRP levels and variability. The reference group consisted of individuals with low CRP levels and high variability, which had a significant protective effect against cancer. Compared to the reference group, the group with low CRP level and low variability had a 20% increased risk of cancer (adjusted HR = 1.20, 95% CI: 1.06–1.37, P = 0.05), and the group with low CRP level and moderate variability did not have a significantly different cancer risk (adjusted HR = 1.12, 95% CI: 0.99–1.28, P = 0.079). High CRP levels were associated with a higher risk of cancer occurrence regardless of variability, with the group exhibiting the lowest variability showing the highest risk (adjusted HR = 1.42, 95% CI: 1.18–1.70, P < 0.001) (Table. 2).

Next, we categorized participants into three distinct long-term CRP change patterns: stable, increasing, and decreasing (Figure. S1). We investigated the association between CRP variability and cancer risk in participants with each of these three different patterns. The results indicated that only in the stable group, CRP variability was significantly associated with the risk of cancer (High variability in stable-pattern: HR = 0.81, 95% CI = 0.72–0.92, P = 0.001) (Table. 3).

Furthermore, we explored the association between CRP variability and cancer at different specific sites. During the follow-up period, the incidence of cancer at various specific sites was presented in Supplementary Table 2 (Table. S2). The results indicated a significant association between CV of CRP and the incidence of lung cancer (HR = 0.77, 95% CI = 0.63–0.96, P = 0.018) (Table. 4).

To explore whether the association between CRP variability and cancer exists in populations with different characteristics, we performed subgroup analyses. Forest plots showed that significant associations between high variability and cancer were present in most subgroups compared to low variability. Moreover, high variability was more protective against cancer than moderate variability (Figure. 3).

We conducted sensitivity analyses to ensure the stability of the findings and to rule out other factors. First, to rule out potential reverse causality, we excluded participants who developed cancer within initial two years of follow-up. Second, we excluded participants with extremely large values of CRP. In addition, we corrected for blood glucose and lipids, CRP at baseline, and aspirin use separately from Model C. The results suggested that these conditions did not affect the association between CRP variability and cancer risk (Table. 5).

In addition, we used VIM as a complementary measure of variability. The RCS plot showed a decrease in HR for cancer risk with increasing VIM of CRP (Figure. 2 (b)). We used COX regression analysis to explore the association between VIM of CRP and cancer risk. The results showed that VIM demonstrated a similar association with cancer as CV. There was a significant association between higher VIM of CRP and reduced cancer risk (Figure. S2).

We further examined the association between CRP variability and the risk of mortality. The results showed no significant association between CRP variability and death and cancer-specific death after correction for confounders (Table. S3). Nonetheless, competing risk plots suggest that high CRP variability has a lower risk of cancer development and death than low variability (Figure. S3).

Our findings suggest a significant association between high variability in CRP and a decreased risk of cancer occurrence. Maintaining a low inflammatory state in the body holds crucial importance for the prevention of cancer. Participants with high levels of CRP accompanied by low variability had a significantly increased risk of developing cancer.

Previous studies have shown that CRP serves not only as a prognostic factor for cancer patients but is also associated with an increased risk of future cancer in seemingly healthy participants(17). Suthathar et al. suggested that baseline CRP was associated with a 17% increase in cancer risk, and elevated CRP over time was associated with an 8% increase in cancer risk(14). A study on the trajectory of CRP changes suggests that maintaining a long-term, stable, and low level of CRP is positively meaningful for cancer prevention(18, 19). Our results keep in line with this conclusion. Participants in a low inflammatory state (CRP < 3 mg/L, or with a trajectory of low-stable) exhibited a significantly lower risk of developing cancer compared to participants in a high inflammatory state (CRP ≥ 3 mg/L). Chronic hyperinflammatory states leads to cellular damage and DNA injury on the one hand. On the other hand, the activation of pro-inflammatory pathways, such as the NF-κB pathway, may excessively inhibit the immune system's surveillance and clearance of malignant cells(20). These factors could potentially contribute to the eventual presence and growth of cancer.

Previous studies exploring the association between CRP and cancer have focused on differences in the amount of CRP, neglecting the fluctuating changes in CRP. When the body is exposed to inflammation or infection, the immune system is stimulated, and the liver releases more CRP to respond to the inflammation or infection(21). Even in relatively healthy populations, CRP fluctuates to some extent. Additionally, unhealthy lifestyle factors such as smoking, obesity, chronic psychological stress, and chronic stress are likely to contribute to elevated CRP levels(22, 23). Evaluating the future risk of cancer based solely on the quantity of CRP is highly susceptible to interference from other factors. In this study, we focused on exploring the association between CRP variability and cancer occurrence. Our study revealed a particularly intriguing finding: the higher the variability of CRP, the lower the risk of cancer occurrence. We cross-combined high and low levels of CRP with the size of variability. It was found that the risk of developing cancer was 42% higher in those with high CRP level-low CRP variability than in those with low CRP level-high CRP variability. Among populations with different CRP trajectories, statistical results revealed that significant protective effects of high CRP variability are observed only in individuals with low-stable CRP patterns. In contrast, no significant protective effect was found in populations with CRP decreasing or increasing patterns, possibly due to sample size limitations, as the sample size for decreasing or increasing CRP patterns were much lower than that of populations with low stable CRP pattern. Furthermore, we observed a significant association between CRP variability and the occurrence of lung cancer. Compared to individuals with low CRP variability, those with high CRP variability had a 23% reduced risk of developing lung cancer. Prior research has demonstrated the relevance of CRP to the prognosis of non-small cell lung cancer patients(24, 25). Xu et al.'s study established that CRP plays a role in promoting lung cancer development in the tumor microenvironment(26). The Kailuan Cohort Study also previously found a significant association between high levels of baseline CRP and the occurrence of lung cancer(27). Our study further found a significant association between CRP variability and lung cancer.

In subgroup analyses, we observed an interesting phenomenon. The protective effect of high CRP variability against cancer incidence was more significant in older individuals compared to middle-aged individuals, in smokers than in non-smokers, and in obese than in non-obese. Both smoking and obesity put the body in a state of chronic inflammation, and CRP is not only a marker of chronic inflammation, but also serves as an internal exposure marker, reflecting the aging state of the organism(28). A prospective cohort study has shown that organismal aging is often accompanied by low-grade inflammation, and that aging of the immune system is accompanied by elevation of the pro-inflammatory marker CRP(29). CRP is also one of the most commonly used biomarkers of immune senescence(30). Cancer, as an age-related disease, is strongly associated with factors such as aging, smoking, and obesity. Our results indicate that, in these cancer risk groups, high CRP variability paradoxically exhibits a protective effect. This suggests that high CRP variability might imply a healthy immune mechanisms opposite to immune aging, hence playing a significant protective role in the mentioned high-risk populations. However, this is only our speculation, and the lack of indicators for the diagnosis of immune aging in our cohort precludes further exploration of the association between CRP variability and immune aging.

Through this study, we observed an association between CRP variability and the occurrence of cancer. Although our research cannot precisely explain the mechanistic aspects of this association, it still holds instructive significance for cancer prevention. Firstly, maintaining low levels of CRP is crucial for cancer prevention. Secondly, long-term monitoring of CRP fluctuations, coupled with early intervention and management, is advantageous in preventing the occurrence of cancer or other immune-related diseases. One of the strengths of our study is its pioneering exploration of the relationship between the amplitude of inflammation fluctuations and cancer. To the best of our knowledge, no similar study has been conducted to date. This provides new insights into understanding the longitudinal association between CRP changes and cancer risk. Additionally, we adjusted for many confounding factors, such as obesity and lifestyle, in the risk model, enhancing the reliability of the results.

Our study has certain limitations. First, there is a significant imbalance in sex distribution in our cohort, with a larger number of men than women. Despite conducting stratified analyses based on the sex, the results may still be prone to bias. Second, factors influencing cancer occurrence are multifaceted, including metabolic factors. Although we adjusted for TC and TG in sensitivity analyses, we couldn't completely eliminate the impact in this regard. Third, while we hypothesize an association between low CRP variability and immune aging, the lack of specific data for diagnosing immune aging in our cohort prevents us from confirming this hypothesis. In future research, it is essential to delve deeper into the specific mechanistic connections among CRP variability, immune aging, and cancer occurrence.

Through this study, we discovered the significant role of CRP variability in the occurrence of cancer, especially in lung cancer. High levels of CRP concentration coupled with low variability were strongly associated with a substantial increase in cancer risk. Furthermore, the protective mechanisms of high CRP variability in cancer need to be better elucidated in future experimental studies.

Conflict of Interest

The authors declare that there is no conflict of interest.

Ethics approval and consent to participate

This study followed the Helsinki declaration. All participants signed an informed consent form. Trial registration: Kailuan study, ChiCTR‐TNRC‐11001489. Registered 24 August, 2011‐Retrospectively registered, http://www.chictr.org.cn/showprojen.aspx?proj=8050.

Consent for Publication

Obtained

Availability of data and materials

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Funding

This work was supported by National Key Research and Development Program (grant number 2022YFC2010101); Laboratory for Clinical Medicine, Capital Medical University (grant number 2023-SYJCLC01); National Multidisciplinary Cooperative Diagnosis and Treatment Capacity Project for Major Diseases: Comprehensive Treatment and Management of Critically Ill Elderly Inpatients (grant number 2019.YLFW).

Authors’ contributions

CY wrote the manuscript. CY, WYM, ZX, LCA, LT, and LSQ analyzed and interpreted the patient data, DL, ZQS and SHP made substantial contributions to the conception, design, and intellectual content of the studies. All authors read and approved the final manuscript.

Acknowledgments

We would like to thank Editage (www.editage.cn) for English language editing. We are grateful to all the participants and staffs who have been part of the kailuan cohort which has enabled this research.

Declaration of generative AI in scientific writing

In the analysis and writing of this study, AI was used only for language polishing.

Ahmed, F., Samantasinghar, A., Soomro, A. M., Kim, S. & Choi, K. H. A systematic review of computational approaches to understand cancer biology for informed drug repurposing. J. Biomed. Inf. 142, 104373 (2023).
Chinnadurai, R. K. et al. Current research status of anti-cancer peptides: Mechanism of action, production, and clinical applications. Biomed. Pharmacother. 164, 114996 (2023).
Ranganathan, P. et al. The International Collaboration for Research methods Development in Oncology (CReDO) workshops: shaping the future of global oncology research. Lancet Oncol. 22 (8), e369–e76 (2021).
Wang, Y. et al. Overview and countermeasures of cancer burden in China. Sci. China Life Sci. (2023).
Xin, W., Yang, Y. & Shi, H-P. Hunger-free hospital initiative: fighting the skeleton in the hospital closet. Precision Nutr. 1 (3), e00018 (2022).
Zamudio-Sánchez, A. et al. Needs and Expectations of Long-Term Cancer Survivors: Multi-Centre Study Protocol. J. Pers. Med. ;14(1). (2024).
Romanos-Nanclares, A. et al. Inflammatory and insulinemic dietary patterns and risk of endometrial cancer among US women. J. Natl. Cancer Inst. 115 (3), 311–321 (2023).
Trabert, B. et al. Prediagnostic circulating inflammation markers and endometrial cancer risk in the prostate, lung, colorectal and ovarian cancer (PLCO) screening trial. Int. J. Cancer. 140 (3), 600–610 (2017).
Afshari, A. R. et al. Nanoparticle-based drug delivery systems in cancer: A focus on inflammatory pathways. Semin Cancer Biol. 86 (Pt 2), 860–872 (2022).
Sohrab, S. S. et al. Chronic Inflammation's Transformation to Cancer: A Nanotherapeutic Paradigm. Molecules ;28(11). (2023).
Ren, W. et al. Jak-STAT3 pathway triggers DICER1 for proteasomal degradation by ubiquitin ligase complex of CUL4A(DCAF1) to promote colon cancer development. Cancer Lett. 375 (2), 209–220 (2016).
Kawanishi, S., Ohnishi, S., Ma, N., Hiraku, Y. & Murata, M. Crosstalk between DNA Damage and Inflammation in the Multiple Steps of Carcinogenesis. Int. J. Mol. Sci. ;18(8). (2017).
Podeanu, M-A. et al. C-Reactive Protein as a Marker of Inflammation in Children and Adolescents with Metabolic Syndrome: A Systematic Review and Meta-Analysis. Biomedicines ;11(11). (2023).
Suthahar, N. et al. Association of Initial and Longitudinal Changes in C-reactive Protein With the Risk of Cardiovascular Disease, Cancer, and Mortality. Mayo Clin Proc. ;98(4):549 – 58. (2023).
Yang, M. et al. Association between C-reactive protein-albumin-lymphocyte (CALLY) index and overall survival in patients with colorectal cancer: From the investigation on nutrition status and clinical outcome of common cancers study. Front. Immunol. 14, 1131496 (2023).
Xie, H. et al. The inflammatory burden index is a superior systemic inflammation biomarker for the prognosis of non-small cell lung cancer. J. Cachexia Sarcopenia Muscle. 14 (2), 869–878 (2023).
Allin, K. H., Bojesen, S. E. & Nordestgaard, B. G. Baseline C-reactive protein is associated with incident cancer and survival in patients with cancer. J. Clin. Oncol. 27 (13), 2217–2224 (2009).
Liu, T. et al. C-reactive protein trajectories and the risk of all cancer types: A prospective cohort study. Int. J. Cancer. 151 (2), 297–307 (2022).
Cui, J. et al. Evidence-based guideline on immunonutrition in patients with cancer. Precision Nutr. 2 (1), e00031 (2023).
Fernandes, Q. et al. Chronic inflammation and cancer; the two sides of a coin. Life Sci. 338, 122390 (2024).
Olson, M. E. et al. A biofunctional review of C-reactive protein (CRP) as a mediator of inflammatory and immune responses: differentiating pentameric and modified CRP isoform effects. Front. Immunol. 14, 1264383 (2023).
Christofaro, D. G. D. et al. Are C-reactive protein concentrations affected by smoking status and physical activity levels? A longitudinal study. PLoS One. 18 (11), e0293453 (2023).
Staufenbiel, I. et al. Influence of Nutrition and Physical Activity on Local and Systemic Inflammatory Signs in Experimentally Induced Gingivitis. Nutrients ;15(15). (2023).
Nassar, Y. M. et al. C-Reactive Protein as an Early Predictor of Efficacy in Advanced Non-Small-Cell Lung Cancer Patients: A Tumor Dynamics-Biomarker Modeling Framework. Cancers (Basel) ;15(22). (2023).
Xu, B. P. & Shi, H. Precision nutrition: concept, evolution, and future vision. Precision Nutr. 1 (1). 10.1097/PN9.0000000000000002 (2022).
Hsu, W-L. et al. High-fat diet induces C-reactive protein secretion, promoting lung adenocarcinoma via immune microenvironment modulation. Dis. Model. Mech. ;16(11). (2023).
Yin, J. et al. Association Between Baseline C-Reactive Protein and the Risk of Lung Cancer: A Prospective Population-Based Cohort Study. Cancer Prev. Res. (Phila). 15 (11), 747–754 (2022).
Zhu, M. et al. C-reactive protein and cancer risk: a pan-cancer study of prospective cohort and Mendelian randomization analysis. BMC Med. 20 (1), 301 (2022).
van Sleen, Y. et al. Frailty is related to serum inflammageing markers: results from the VITAL study. Immun. Ageing. 20 (1), 68 (2023).
Van Tran, E. et al. Biomarkers of the ageing immune system and their association with frailty - A systematic review. Exp. Gerontol. 176, 112163 (2023).

Table. 1 Baseline characteristics of participants, stratified by the coefficient of variation of CRP (n=42514)

	CV of CRP
	Lowest tertile (<53.07)	Middle tertile (53.07~84.05)	Highest tertile (≧84.05)	*P-value*
N	14173	14172	14169
Sex, n (%)
Woman	3622 (25.6)	3504 (24.7)	3093 (21.8)	<0.001
Man	10551 (74.4)	10668 (75.3)	11076 (78.2)
Age	54.56 (11.82)	53.55 (11.77)	51.62 (11.47)	<0.001
Education, n (%)				<0.001
Low	10336 (72.9)	11040 (77.8)	10991 (77.6)
High	3837 (27.1)	3132 (22.1)	3178 (22.4)
Income, n (%)				<0.001
Low	12832 (90.5)	13207 (93.2)	13355 (94.3)
High	1341 (9.5)	965 (6.8)	814 (5.7)
Smoke, n (%)				<0.001
No	8880 (62.7)	9223 (65.1)	9088 (64.1)
Current and past	5293 (37.3)	4949 (34.9)	5081 (35.9)
Alcohol intake, n (%)				<0.001
No	7951 (56.1)	8606 (60.7)	8458 (59.7)
Yes	6222 (43.9)	5566 (39.3)	5711 (40.3)
Physical exercise, n (%)				<0.001
No	1204 (8.5)	1301 (9.2)	1465 (10.3)
Yes	12969 (91.5)	12871 (90.8)	12704 (89.7)
BMI	25.47 (3.39)	25.23 (3.38)	24.79 (3.29)	<0.001
FBG	5.75 (1.67)	5.68 (1.76)	5.58 (1.59)	<0.001
TG	1.36[0.93,2.06]	1.28 [0.88,1.91]	1.26[0.90,1.85]	<0.001
CRP (2010)	1.27[0.70,2.50]	1.21[0.68,2.43]	1.00[0.50,2.70]	<0.001
CRP (2008)	1.33[0.80,2.70]	1.60[0.81,3.03]	1.60[0.70,3.20]	<0.001
CRP (2006)	1.10[0.60,2.30]	0.70[0.29,1.80]	0.33[0.11,1.08]	<0.001
Mean CRP	1.30[0.75,2.52]	1.40[0.77,2.50]	1.57[0.95,2.60]	<0.001

Notes: Continuous variables were presented as mean ± standard deviation (SD). Continuous variables that were not normally distributed were expressed as the median (interquartile range). Categorical variables were presented as numbers and percentages. Differences in normally and non-normally distributed baseline characteristics were compared using the chi-square test or t-test and Wilcoxon rank sum test, respectively. CV, coefficient of variation; CRP, C reactive protein; BMI, body mass index; FBG, fasting blood-glucose; TG, triglyceride.

Table. 2 Association between serum C reactive protein levels and variability and cancer incidence

		Model A		Model B		Model C
	Case/N	HR (95%CI)	*P-value*	HR (95%CI)	*P-value*	HR (95%CI)	*P-value*
CRP levels
Low (<3mg/L)	1445/34118	ref.	ref.	ref.	ref.	ref.	ref.
High (≧3mg/L)	457/8396	1.33(1.19,1.47)	<0.001	1.22(1.09.1.35)	<0.001	1.20(1.07,1.33)	<0.001
CRP variability*
Low	689/14173	ref.	ref.	ref.	ref.	ref.	ref.
Moderate	645/14172	0.93(0.83,1.03)	0.181	0.93(0.83,1.04)	0.183	0.93(0.84,1.04)	0.205
High	568/14169	0.81(0.72,0.90)	<0.001	0.83(0.75,0.93)	0.002	0.84(0.75,0.94)	0.003
P for trend			<0.001		0.001		0.003
CRP levels and variability
Low CRP with high variability	428/11322	ref.	ref.	ref.	ref.	ref.	ref.
Low CRP with moderate variability	497/11495	1.16(1.02,1.32)	0.026	1.13(0.99,1.29)	0.062	1.12(0.99,1.28)	0.079
Low CRP with low variability	520/11301	1.24(1.09,1.41)	<0.001	1.22(1.07,1.38)	0.003	1.20(1.06,1.37)	0.005
High CRP with high variability	140/2847	1.33(1.1,1.61)	0.003	1.27(1.05,1.53)	0.015	1.25(1.03,1.51)	0.025
High CRP with moderate variability	148/2677	1.52(1.26,1.84)	<0.001	1.36(1.13,1.64)	<0.001	1.33(1.1,1.61)	0.003
High CRP with low variability	169/2872	1.65(1.38,1.97)	<0.001	1.45(1.21,1.73)	<0.001	1.42(1.18,1.70)	<0.001
P for trend			<0.001		<0.001		<0.001

Notes: Data are presented as hazard ratios (95% confidence intervals).

Model A: unadjusted; Model B: adjusted for sex, age, education, income; Model C: additionally adjusted for BMI, smoke, alcohol intake, physical exercise based on Model B.

*We categorized the variability into three groups based on the tertile intervals of CV of CRP: low, medium, and high.

CRP, C reactive protein; HR, hazard ratio; CI, confidence interval; P-value, probability.

Table. 3 Association between CRP variability and cancer incidence in different CRP trajectories (2006-2010)

	Case/N	HR (95%CI)	*p-value*
Continuous CV of CRP (per_SD)
Overall	1902/42514	0.95 (0.90,0.99)	0.013
Track 1 (stable)	1607/36579	0.95 (0.90,1.00)	0.032
Track 2 (increase)	138/2792	0.93 (0.79,1.08)	0.335
Track 3 (decrease)	157/3143	0.89 (0.76,1.03)	0.122
CV tertile interval of CRP
Track 1 (stable-pattern)
Lowest tertile	585/12394	ref.
Middle tertile	555/12255	0.95 (0.85,1.07)	0.426
Highest tertile	467/11930	0.81 (0.72,0.92)	0.001
Track 2 (increase-pattern)
Lowest tertile	45/701	ref.
Middle tertile	35/889	0.59 (0.38,0.93)	0.021
Highest tertile	58/1202	0.71 (0.48,1.05)	0.083
Track 3 (decrease-pattern)
Lowest tertile	59/1078	ref.
Middle tertile	55/1028	0.96(0.67,1.39)	0.848
Highest tertile	43/1037	0.74 (0.51,1.09)	0.127

Notes: COX regression model adjusted for sex, age, education, income, BMI, smoke, alcohol intake and physical exercise. Data are presented as hazard ratios (95% confidence intervals). CV, coefficient of variation; CRP, C reactive protein; HR, hazard ratio; CI, confidence interval; P-value, probability.

Table. 4 Association between CRP coefficient of variation and site-specific cancer incidence

	Lowest tertile	Middle tertile		Highest tertile
		HR (95%CI)	P-value	HR (95%CI)	P-value
Head and neck	ref.	1.68 (1.09,2.59)	0.018	1.32 (0.84,2.08)	0.236
Esophagus	ref.	1.98 (1.04,3.77)	0.037	1.30 (0.65,2.60)	0.463
Gastrointestinal	ref.	1.22 (0.82,1.82)	0.318	1.13 (0.75,1.71)	0.546
Colorectal	ref.	0.81 (0.61,1.09)	0.161	0.78 (0.58,1.05)	0.100
Liver and gallbladder	ref.	1.36 (0.93,1.96)	0.109	0.81 (0.53,1.24)	0.328
Pancreatic	ref.	1.63 (0.67,3.93)	0.280	1.36 (0.72,2.48)	0.245
Lung	ref.	0.74 (0.59,0.91)	0.005	0.77 (0.63,0.96)	0.018
Breast	ref.	0.90 (0.62,1.32)	0.587	0.71 (0.47,1.09)	0.117
Uterus and ovaries	ref.	1.08 (0.56,2.06)	0.825	1.18 (0.61,2.28)	0.626
Prostate	ref.	0.58 (0.32,1.03)	0.063	0.81 (0.47,1.40)	0.448
Kidney	ref.	1.14 (0.67,1.95)	0.631	0.78 (0.43,1.42)	0.422
Urinary bladder	ref.	0.97 (0.53,1.80)	0.935	0.79 (0.40,1.54)	0.488
Hematology*	ref.	0.81 (0.44,1.51)	0.511	0.63 (0.32,1.26)	0.191

Notes: Data are presented as hazard ratios (95% confidence intervals). Cox regression model adjusted for sex, age, education, income, BMI, smoke, alcohol intake, physical activity. HR, hazard ratio; CI, confidence interval; P-value, probability. Bolded text indicates statistical significance.

*Tumors of the hematology include lymphoma and leukemia.

Table. 5 Sensitivity Analyses

CV of CRP	HR(95%CI)	*P-value*
Exclude cancer occurrence within first 2 years (n=41948)
Lowest tertile	ref.
Middle tertile	0.89 (0.80-1.00)	0.059
Highest tertile	0.79 (0.70-0.90)	<0.001
Exclude CRP maxima (CRP>10 mg/L) (n=40412)
Lowest tertile	ref.
Middle tertile	0.94 (0.84-1.05)	0.283
Highest tertile	0.86 (0.77-0.97)	0.014
Additionally adjusted for blood glucose and lipid
Lowest tertile	ref.
Middle tertile	0.93 (0.84-1.04)	0.201
Highest tertile	0.84 (0.75-0.94)	0.003
Additionally adjusted for basic CRP
Lowest tertile	ref.
Middle tertile	0.93 (0.84-1.04)	0.209
Highest tertile	0.84 (0.75-0.94)	0.002
Additionally adjusted for taking aspirin
Lowest tertile	ref.
Middle tertile	0.93 (0.84-1.04)	0.212
Highest tertile	0.85 (0.76-0.95)	0.003

Notes: Data are presented as hazard ratios (95% confidence intervals). Cox regression model adjusted for sex, age, education, income, BMI, smoke, alcohol intake, physical activity. CV, coefficient of variation; HR, hazard ratio; CI, confidence interval; P-value, probability.

No competing interests reported.

Supplementary.docx

Association between C-reactive protein variability and cancer incidence: a longitudinal prospective cohort study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Methods

1. Study design and population

2. Variable evaluation

3. Calculation of CRP variability

4. Statistical analysis

Results

Discussion

Conclusion

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1