Originally published as JCO Early Release 10.1200/JCO.2008.19.8440 on March 16 2009

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Allin, K. H. Articles by Nordestgaard, B. G. Search for Related Content PubMed PubMed Citation Articles by Allin, K. H. Articles by Nordestgaard, B. G. Social Bookmarking                            What's this?

Baseline C-Reactive Protein Is Associated With Incident Cancer and Survival in Patients With Cancer

From the Department of Clinical Biochemistry, Herlev Hospital, and The Copenhagen City Heart Study, Bispebjerg Hospital, Copenhagen University Hospital; and Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark.

Corresponding author: Børge G. Nordestgaard, MD, DMSc, Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Herlev Ringvej 75, DK-2730 Herlev, Denmark; e-mail: brno{at}heh.regionh.dk.

	ABSTRACT

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Purpose We tested the hypothesis that baseline plasma levels of C-reactive protein (CRP) are associated with risk of incident cancer in the general population and early death in patients with cancer.

Patients and Methods A total of 10,408 individuals from the Danish general population who had CRP measured at baseline were observed for up to 16 years; 1,624 developed cancer, and of these, 998 patients died during follow-up. Follow-up was 100% complete. We excluded individuals with a cancer diagnosis at baseline.

Results Baseline CRP levels more than 3 versus less than 1 mg/L were associated with multifactorially adjusted hazard ratios of 1.3 (95% CI, 1.0 to 1.6) for cancer of any type, 2.2 (95% CI, 1.0 to 4.6) for lung cancer, 1.9 (95% CI, 0.8 to 4.6) for colorectal cancer, and 0.7 (95% CI, 0.4 to 1.4) for breast cancer. Corresponding hazard ratios for the highest versus the lowest quintile of baseline CRP levels were 1.3 (95% CI, 1.0 to 1.6), 2.1 (95% CI, 1.2 to 3.8), 1.7 (95% CI, 0.8 to 3.2), and 0.9 (95% CI, 0.5 to 1.7), respectively. Multifactorially adjusted hazard ratios for early death in patients with cancer were 1.8 (95% CI, 1.2 to 2.7) for CRP more than 3 versus less than 1 mg/L and 1.4 (95% CI, 1.1 to 1.7) for the highest versus the lowest quintile. Elevated CRP levels were associated with early death in patients with cancer having localized disease, but not in those with metastases (interaction; P = .03).

Conclusion Elevated levels of CRP in cancer-free individuals are associated with increased risk of cancer of any type, of lung cancer, and possibly of colorectal cancer. Moreover, elevated levels of baseline CRP associate with early death after a diagnosis of any cancer, particularly in patients without metastases.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
C-reactive protein (CRP) is an acute-phase reactant that is elevated during bacterial infection, inflammatory disease, trauma, myocardial infarction, surgery, and cancer¹; CRP is produced in the liver in response to elevated cytokine levels after an inflammatory stimulus.² Two hypotheses have been proposed to explain the relationship between elevated CRP levels and cancer.³ The first hypothesis states that elevated CRP levels are a result of an underlying cancer or a premalignant state, whereas the second hypothesis states that chronic inflammation and elevated CRP might have a causal role in carcinogenesis.

Case-control studies have reported higher levels of CRP in patients with cancer compared with controls,³ whereas results from prospective studies are conflicting, with some studies suggesting that CRP is not merely a marker of prevalent cancer, but is also associated with incident cancer.^4–16 A limitation of the prospective studies is the use of a single CRP measurement at baseline, because lack of additional measurements precludes adjustment for regression dilution bias¹⁷ and thereby tends to underestimate any association. Elevated levels of plasma CRP have also been associated with poor survival in patients with cancer^7,18–22; however, in only two of these studies including few participants were CRP measured with a high-sensitivity assay,^7,19 allowing examination of associations in the high-normal interval just greater than 1 mg/L. Thus it is still unclear whether, and to which extent, CRP levels are associated with incident cancer as well as early death in patients with cancer.

We tested the hypothesis that baseline plasma levels of CRP in the general population are associated with risk of incident cancer, including the three most common cancers (ie, lung, colorectal, and breast cancer), and we corrected for regression dilution bias. In addition, we tested the hypothesis that baseline plasma levels of CRP are associated with early death in patients with cancer with any type of cancer.

	PATIENTS AND METHODS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Study Design
The Copenhagen City Heart Study is a prospective cohort study of the Danish general population, in which participants 20 years of age or older were drawn randomly from the Copenhagen Central Person Register.^23,24 It was initiated in 1976 to 1978, with follow-up examinations in 1981 to 1983, 1991 to 1994, and 2001 to 2003. Participants in the present study are from the 1991 to 1994 and/or 2001 to 2003 examination (additional information on patients and methods is described in the Appendix, online only).

For the analysis of plasma CRP levels and incident cancer, we included 10,520 individuals who had plasma CRP measured at the 1991 to 1994 and/or 2001 to 2003 examination. We excluded participants with liver cirrhosis diagnosed before or during the study period from analysis (n = 112) because their ability to produce CRP must be expected to be impaired, and 233 participants were excluded because of incomplete information about covariates. Furthermore, we excluded participants with a diagnosis of any cancer (n = 772), lung cancer (n = 54), colorectal cancer (n = 93), female breast cancer (n = 164), or other cancer (n = 581) before study entry, from the analysis of the respective cancer subtypes. Follow-up time for each participant began at entry into the study (1991 to 1994 or 2001 to 2003) and ended at occurrence of cancer, death, emigration, or July 2007, whichever came first. During the study period, we had 100% follow-up. The median and maximum follow-up periods were 13 and 16 years, respectively (Appendix Fig A1, online only).

Diagnoses of invasive cancer from 1947 until July 2007 were obtained from the Danish Cancer Registry,^25,26 which identifies 97.8% of all cancers in Denmark,²⁷ as well as from the national Danish Patient Registry (Appendix Table A1, online only).

Individuals with a diagnosis of cancer during the period of follow-up (n = 1624) were observed from their date of diagnosis until death, emigration, or July 2007, whichever came first. The median and maximum follow-up periods were 2 and 15 years, respectively. Information about mortality was obtained from the national Danish Civil Registration System, which is 100% complete.

Ethics
The ethical committee of Copenhagen and Frederiksberg, Denmark, approved the study (KF100.2039/91 and KF01-144/01). All participants gave written informed consent.

Measurement of CRP
Plasma levels of CRP were determined with a high-sensitivity assay on 8,778 participants from the 1991 to 1994 examination on plasma stored for 12 to 15 years and on 5,931 participants from the 2001 to 2003 examination on fresh plasma samples using turbidimetry or nephelometry (Dako, Glostrup, Denmark, or Dade Behring, Deerfield, IL).

Statistical Analysis
We analyzed data using the statistical software package STATA version 10.0 (StataCorp, College Station, TX). A two-sided P value less than .05 was considered statistically significant. A priori, we divided baseline CRP levels into quintiles as well as into three categories: low (< 1.0 mg/L), average (1.0 to 3.0 mg/L), or high (> 3.0 mg/L). The CRP groups were coded 1, 2, 3, 4, and 5 corresponding to the quintiles and 1, 2, and 3 corresponding to the categories for trend tests in log-rank and Cox statistics.

We plotted survival against follow-up time using the Kaplan-Meier method and tested differences between categories of CRP using log-rank trend tests. To estimate hazard ratios with 95% CIs for incident cancer, we used Cox proportional hazards regression. We used age as time scale, thus analyzing age at event using left truncation (ie, delayed entry). Hereby, age differences are automatically adjusted for. We used a multifactorial Cox regression model including sex and time-dependent covariates from the 1991 to 1994 and 2001 to 2003 examinations: smoking (never, former, current), smoking dose (cigarettes per day), alcohol consumption (women, 168 or > 168 g per week; men, 252 or > 252 g per week), body mass index (< 18.5, 18.5 to 24.9, 25 to 29.9, or 30.0 kg/m²), and, for women, also use of oral contraceptive therapy, postmenopausal status, and use of hormone replacement therapy.

To estimate hazard ratios with 95% CIs for early death of cancer, we used Cox proportional hazards regression analyzing time to event. We used a Cox regression model including age at diagnosis, sex, cancer type (lung, colorectal, breast, or other cancer), cancer stage (localized, regional metastases, or distant metastases), cancer histology (adenocarcinoma, squamous carcinoma, other carcinoma, or other histology), and time from blood sampling to diagnosis.

Because approximately 41% of the participants had CRP measured at both the 1991 to 1994 and 2001 to 2003 examination, we were able to correct hazard ratios for regression dilution bias with a nonparametric method.¹⁷ Spearman's rank correlation coefficient between the two CRP measurements was 0.5.

	RESULTS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Baseline characteristics of the participants by plasma levels of CRP at study entry are listed in Table 1. Levels of CRP were correlated with several of these characteristics. Median values of CRP were 1.7 mg/L among all participants, and 1.9, 2.3, 1.8, and 1.7 mg/L among individuals who developed any cancer, lung cancer, colorectal cancer, or breast cancer, respectively, during the period of follow-up.

View larger version (15K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 1. Hazard ratios for cancer incidence by plasma levels of C-reactive protein (P-CRP) in categories. Multifactorial adjustment was for age, sex, smoking, alcohol consumption, body mass index, and, for women, also oral contraceptive therapy, menopausal status, and hormone replacement therapy. P values for trend tests examine whether increasing levels of CRP are associated with increasing hazard ratios.

View larger version (22K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 2. Hazard ratios for cancer incidence by plasma levels of C-reactive protein (P-CRP) in quintiles. Multifactorial adjustment was for age, sex, smoking, alcohol consumption, body mass index, and, for women, also oral contraceptive therapy, menopausal status, and hormone replacement therapy. P values for trend tests examine whether increasing levels of CRP are associated with increasing hazard ratios.

Colorectal Cancer
Levels of CRP, by categories or quintiles, were not associated statistically significant with risk of incident colorectal cancer: P for trend = .10 and .33, respectively (Figs 1 and 2). Multifactorially adjusted hazard ratios were 1.9 (95% CI, 0.8 to 4.6) for CRP more than 3 versus less than 1 mg/L and 1.7 (95% CI, 0.8 to 3.2) for the highest versus the lowest quintile. For CRP more than 3 versus less than 1 mg/L and for the highest versus the lowest quintile, we had 90% statistical power to exclude a hazard ratio of 2.3 and 2.1, respectively.

Breast Cancer
Levels of CRP, by categories or quintiles, were not associated with risk of incident breast cancer: P for trend = .40 and .79, respectively (Figs 1 and 2). Multifactorially adjusted hazard ratios were 0.7 (95% CI, 0.4 to 1.4) for CRP more than 3 versus less than 1 mg/L and 0.9 (95% CI, 0.5 to 1.7) for the highest versus the lowest quintile. For CRP more than 3 versus less than 1 mg/L and for the highest versus the lowest quintile, we had 90% statistical power to exclude a hazard ratio of 1.9 and 1.9, respectively.

Sensitivity Analyses
After exclusion of individuals with CRP more than 10 mg/L, as an indication of overt inflammation at baseline, the results were similar to those in Figures 1 and 2. To eliminate an effect of occult cancer on CRP levels, we also conducted additional analyses excluding incident cases of cancer diagnosed within 2 years after blood sampling, and hazard ratios were attenuated (Figs 1 and 2). Mean time from blood sampling to a diagnosis of cancer was 6.3 years for all participants and 5.8 years for individuals with CRP more than 3 mg/L.

To examine whether CRP is associated with cancer of different stages at diagnosis, we stratified for localized cancer and metastases (Fig 3). Increasing levels of CRP, by categories, were associated with increasing risk of incident any, lung, and colorectal cancer with metastases: P for trend = .003, .04, and .02, respectively. Corresponding multifactorially adjusted hazard ratios were 2.3 (95% CI, 1.3 to 4.0), 4.4 (95% CI, 1.1 to 18), and 10.4 (95% CI, 0.9 to 118), respectively, for CRP more than 3 versus less than 1 mg/L.

View larger version (14K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 3. Hazard ratios for cancer incidence by plasma levels of C-reactive protein (P-CRP) stratified for cancer stage. Multifactorial adjustment was for age, sex, smoking, alcohol consumption, body mass index, and, for women, also oral contraceptive therapy, menopausal status and hormone replacement therapy. P values for trend tests examine whether increasing levels of CRP are associated with increasing hazard ratios.

Survival After a Diagnosis of Cancer
Survival after a diagnosis of cancer, unadjusted for potential confounders, decreased with increasing levels of plasma CRP at study entry: log-rank trend test P < .001 (Fig 4). Multifactorially adjusted hazard ratios were 1.8 (95% CI, 1.2 to 2.7) for CRP more than 3 versus less than 1 mg/L and 1.4 (95% CI, 1.1 to 1.7) for the highest versus the lowest quintile: P for trend = .002 and .002.

View larger version (14K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 4. Survival after a diagnosis of cancer by levels of C-reactive protein (CRP). P value for log-rank trend test examines whether increasing levels of CRP are associated with decreasing survival. Results are unadjusted. Multifactorially adjusted hazard ratios for early death in patients with cancer by levels of CRP are presented in Table 2. hs, high-sensitivity assay.

	DISCUSSION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

What is the biologic underlying mechanism of the association between levels of CRP and risk of cancer? Several findings support the hypothesis that elevated levels of CRP are a marker of occult cancer. First, tumor growth can cause tissue inflammation around the tumor and thus increase plasma levels of CRP.³ Second, tumor cells are known to produce various cytokines and chemokines that attract leukocytes, and some cancerous cells have been shown to express CRP and secrete interleukin-6 and interleukin-8, which stimulate CRP production in the liver.^3,28 Finally, it is possible that CRP is a part of a host immune response to the tumor.³ However, there is increasing evidence that chronic inflammation, of which CRP is a marker, is a causal factor in several malignancies.^28,29 It is evident that inflammatory cells act as tumor promoters, producing an attractive environment for tumor growth, inducing DNA damage, promoting angiogenesis, and favoring neoplastic spread and metastasis.²⁸ In the present association study, we cannot distinguish between elevated CRP being a marker of occult cancer or inflammation and elevated CRP being causative in carcinogenesis. However, mean time from blood sampling to diagnosis of cancer was 5.8 years in the CRP category more than 3 mg/L, suggesting that elevated levels of CRP might be more than just a marker of occult cancer. Conversely, when we excluded incident cancer cases diagnosed within 2 years of blood sampling, the association between CRP levels and incident any cancer was attenuated. Also in support of CRP being a marker of occult cancer, we found that elevated CRP levels were associated with any, lung, and colorectal cancer with metastases at diagnosis, but not with localized cancer only.

Thus far, seven prospective nested case-control studies and five prospective cohort studies have been published, but the results are inconsistent.^4–13,15,16 A recent meta-analysis found an odds ratio of 1.10 (95% CI, 1.02 to 1.18) for any cancer and of 1.32 (95% CI, 1.08 to 1.61) for lung cancer for a log unit increase in CRP level.⁶ Despite the considerable heterogeneity in the meta-analysis (I² > 70%), these findings are in support of our findings of baseline CRP levels being associated with risk of any cancer and lung cancer. In support of our finding of an increasing risk of lung cancer with increasing levels of CRP, studies have suggested an increased risk of lung cancer in individuals with asthma as well as in individuals with tuberculosis or other infectious inflammation of the lung.³⁰

An increased risk of colorectal cancer has been observed in patients with inflammatory bowel diseases,^15,28 suggesting that chronic inflammation and elevated CRP levels could be associated with risk of colorectal cancer. Although we did not find a statistically significant association between elevated levels of CRP and incident colorectal cancer, the hazard ratios for colorectal cancer were of a magnitude comparable to that for any cancer and lung cancer; the number of cases was just smaller for colorectal cancer. Furthermore, we found an association between elevated levels of CRP and risk of colorectal cancer with metastases. A recent meta-analysis found an odds ratio of 1.09 (95% CI, 0.98 to 1.21) for a log unit increase in CRP level, whereas another recent meta-analysis found that levels of CRP were weakly associated with an increased risk of colorectal cancer (odds ratio, 1.12; 95% CI, 1.01 to 1.25) for a log unit increase in CRP level.^6,14 In support of our finding of no association of CRP levels with risk of breast cancer, a recent meta-analysis found an odds ratio for breast cancer of 1.10 (95% CI, 0.97 to 1.26) for a log unit increase in CRP level.⁶ One could suspect that many of the breast cancer cases in the present study were screening detected (early stage) and that CRP could be associated with late stage at breast cancer diagnosis. However, we did not find an association between elevated levels of CRP and incident breast cancer with metastases.

Several studies have reported that CRP can be used as a prognostic marker in patients with cancer, but so far only two studies with few participants have measured CRP with a high-sensitivity assay.^7,19 Il'yasova et al⁷ found a hazard ratio of 1.4 (95% CI, 1.1 to 1.8) for early death of patients with cancer for a log unit increase in CRP level, and Heikkilä et al¹⁹ reported similar estimates. The results of our study, which included far more participants, support these findings and consequently the value of CRP as a prognostic marker in patients with cancer. Interestingly, our results do not seem to suggest any differential effect on the association between elevated levels of CRP and early death according to different cancer types or histology, or according to time from blood sampling to diagnosis. However, our results suggest that the prognostic value of CRP as a marker of early death is present primarily in patients with localized cancer without known metastases at time of diagnosis. Consequently, CRP does not merely seem to be a surrogate for known metastases at diagnosis, and CRP might provide prognostic information beyond that provided by cancer type, stage, and histology.

We studied the value of elevated CRP levels as a risk factor for cancer and as a prognostic marker in patients with cancer in a large prospective cohort study with up to 16 years of follow-up and had no losses to follow-up. In addition, we were able to adjust for regression dilution bias, because CRP was measured twice in approximately 41% of the participants. Potential limitations of our study include confounding and selection bias. However, we included several important potential confounders associated with CRP levels in the Cox regression model. Despite this, we naturally cannot exclude all possible confounders and the possibility that CRP may act as a surrogate for imperfectly measured risk factors for cancer as a result of not measuring exposure during the etiologically relevant time or misspecification of exposure, like body mass index as a surrogate for body fat. Selection bias of healthy individuals would draw the results in a direction toward the null hypothesis and therefore cannot explain our positive results. Despite adjustment for age at diagnosis, sex, cancer type, cancer stage, cancer histology, and time from blood sampling to diagnosis, the interpretation of the positive association found between elevated levels of CRP and early death after a cancer diagnosis is diminished by the heterogeneity of the cancer diagnoses and lack of information about treatment.

In conclusion, we have demonstrated that elevated levels of CRP in apparently cancer-free individuals are associated with increased risk of incident cancer of any type, lung cancer, and possibly colorectal cancer. Moreover, we have demonstrated that elevated levels of CRP at baseline are associated with early death after a diagnosis of cancer, particularly in patients without metastases.

	AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
The author(s) indicated no potential conflicts of interest.

	AUTHOR CONTRIBUTIONS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Conception and design: Kristine H. Allin, Stig E. Bojesen, Børge G. Nordestgaard

Financial support: Børge G. Nordestgaard

Provision of study materials or patients: Stig E. Bojesen, Børge G. Nordestgaard

Collection and assembly of data: Kristine H. Allin, Stig E. Bojesen, Børge G. Nordestgaard

Data analysis and interpretation: Kristine H. Allin, Stig E. Bojesen, Børge G. Nordestgaard

Manuscript writing: Kristine H. Allin, Stig E. Bojesen, Børge G. Nordestgaard

Final approval of manuscript: Kristine H. Allin, Stig E. Bojesen, Børge G. Nordestgaard

	Appendix

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Study Design
The Copenhagen City Heart Study is a prospective cohort study of the Danish general population, in which participants aged 20 years or older were drawn randomly from the Copenhagen Central Person Register.^23,24 It was initiated in 1976 to 1978, with follow-up examinations in 1981 to 1983, 1991 to 1994, and 2001 to 2003. Participants in the present study are from the 1991 to 1994 and/or 2001 to 2003 examination. In the 1991 to 1994 examination, 10,135 of 16,563 invited individuals participated (response rate, 61.2%) and in the 2001 to 2003 examination, 6,238 of 12,600 invited individuals participated (response rate, 49.5%).

For the analysis of plasma C-reactive protein (CRP) levels and incident cancer, we included 10,520 individuals who had plasma CRP measured at the 1991 to 1994 and/or 2001 to 2003 examination. We excluded participants with liver cirrhosis diagnosed before or during the study period from analysis (n = 112), because their ability to produce CRP must be expected to be impaired, and 233 participants were excluded because of incomplete information about covariates. Furthermore, we excluded participants with a diagnosis of any cancer (n = 772), lung cancer (n = 54), colorectal cancer (n = 93), female breast cancer (n = 164), or other cancer (n = 581) before study entry from the analysis of the respective cancer subtypes. Follow-up time for each participant began at entry into the study (1991 to 1994 or 2001 to 2003) and ended at occurrence of cancer, death, emigration, or July 2007, whichever came first. During the study period, we had 100% follow-up. The median and maximum follow-up periods were 13 and 16 years, respectively (Appendix Fig A1, online only).

Diagnoses of invasive cancer from 1947 until July 2007 were obtained from the Danish Cancer Registry,^25,26 which identifies 97.8% of all cancers in Denmark,²⁷ as well as from the national Danish Patient Registry (Appendix Table A1, online only). Diagnoses were classified according to the WHO International Classification of Diseases, seventh revision (ICD-7) and 10th revision (ICD-10).

Individuals with a diagnosis of cancer during the period of follow-up (n = 1624) were observed from their date of diagnosis until death, emigration, or July 2007, whichever came first. During the study period, we had 100% follow-up. The median and maximum follow-up periods were 2 and 15 years, respectively. Information about mortality was obtained from the national Danish Civil Registration System, which is 100% complete.

Ethics
The ethical committee of Copenhagen and Frederiksberg, Denmark, approved the study (KF100.2039/91 and KF01-144/01). All participants gave written informed consent.

Measurement of CRP
Plasma levels of CRP were determined with a high-sensitivity assay on 8,778 participants from the 1991 to 1994 examination on plasma stored for 12 to 15 years, and on 5,931 participants from the 2001 to 2003 examination on fresh plasma samples using turbidimetry or nephelometry (Dako, Glostrup, Denmark, or Dade Behring, Deerfield, IL). CRP measurements were daily assessed for precision (coefficient of variation ranged from 3% to 6% at a level of 1.7 mg/L) and monthly for accuracy with a Scandinavian quality control program. If the statistical analyses presented in this article in addition were adjusted for storage time from blood sampling to CRP measurement, the results were similar to those shown.

Statistical Analysis
We analyzed data using the statistical software package STATA version 10.0 (StataCorp, College Station, TX). A two-sided P value less than .05 was considered statistically significant. A priori, we divided baseline CRP levels into quintiles as well as into three categories: low (< 1.0 mg/L), average (1.0 to 3.0 mg/L), or high (> 3.0 mg/L), corresponding to approximate tertiles of CRP among more than 40,000 adults in more than 15 populations (Pearson TA, Mensah GA, Alexander RW, et al. Circulation 107:499-511, 2003); we used baseline CRP levels in time to event analysis, whereas updated CRP levels on the 41% of participants were used for calculation of regression dilution bias, as described below. If we used updated CRP concentrations in the statistical analyses, results were similar to those shown.

We compared individuals with higher levels of CRP with the reference group (individuals in the lowest quintile, respectively individuals with CRP < 1.0 mg/L). The CRP groups were coded 1, 2, 3, 4, and 5, corresponding to the quintiles, and 1, 2, and 3, corresponding to the categories for trend tests in log-rank and Cox statistics.

We plotted cumulative survival against follow-up time using the Kaplan-Meier method and tested differences between categories of CRP using log-rank trend tests. To estimate hazard ratios with 95% CIs for incident cancer, we used Cox proportional hazards regression. We used age as time scale, thus analyzing age at event using left truncation (ie, delayed entry). Thus age differences are automatically adjusted for. We used a multifactorial Cox regression model that included sex and time-dependent covariates from the 1991 to 1994 and 2001 to 2003 examinations: smoking (never, former, current), smoking dose (cigarettes per day), alcohol consumption (women, 168 or > 168 g per week; men, 252 or > 252 g per week), body mass index (< 18.5, 18.5 to 24.9, 25 to 29.9, or 30.0 kg/m²), and, for women, also use of oral contraceptive therapy, postmenopausal status, and use of hormone replacement therapy. To estimate hazard ratios with 95% CIs for early death of cancer, we used Cox proportional hazards regression analyzing time to event. We used a Cox regression model that included age at diagnosis, sex, cancer type (lung, colorectal, breast, or other cancer), cancer stage (localized, regional metastases, or distant metastases), cancer histology (adenocarcinoma, squamous carcinoma, other carcinoma, or other histology), and time from blood sampling to diagnosis. We tested the assumption of proportional hazards graphically by plotting log(cumulative hazard) as a function of age. Suspicion of nonparallel lines was tested by using Schoenfeld residuals. We detected no major violations of the proportional hazards assumption. Interaction between CRP and other risk factors in the models was tested for by using the likelihood ratio test of the model without the two-factor interaction term nested in the model with it. After correction for multiple comparisons using the Bonferroni method, none of the likelihood ratio tests remained significant. We cannot study the effects of age itself when age is used as the time scale in the Cox regression model. To test interaction of age with CRP levels, we therefore used years of follow-up as time scale analyzing time to event.

Because approximately 41% of the participants had CRP measured at both the 1991 to 1994 and 2001 to 2003 examination, we were able to correct hazard ratios for regression dilution bias with a nonparametric method.¹⁷ The regression dilution ratio provides an assumption-free estimate of the importance of regression dilution during the exposure period that is valid no matter what the sources of variation might have been over the particular time period.¹⁷ For this calculation, we included individuals with plasma CRP measured at both examinations without a diagnosis of cancer before the measurement of plasma CRP at the 2001 to 2003 examination (n = 3,769). A regression dilution bias ratio of 0.83 and 0.86 was computed when the population was divided into three categories of plasma CRP or into quintiles of plasma CRP, respectively. Spearman rank correlation coefficient between the two CRP measurements was 0.5. Age-adjustment in Table 1 was based on 10-year age strata or linear regression.

View larger version (10K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig A1. Study design. Participants with a diagnosis of any cancer (n = 772), lung cancer (n = 54), colorectal cancer (n = 93), female breast cancer (n = 164), or other cancer (n = 581) before study entry were excluded from analyses of levels of C-reactive protein (CRP) and risk of the respective cancer types.

	Acknowledgment

 
We thank Anja Jochumsen for technical assistance.

	NOTES

 
Supported by the Danish Heart Foundation, the Danish Medical Research Council, and the Research Council at Herlev Hospital, Copenhagen University Hospital.

The funding sources are public or nonprofit organizations and had no role in the design or conduct of the study, analyzing or interpreting the data, or approving the submitted manuscript.

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
1. Burtis CA, Ashwood ER, Bruns DE. Tietz Textbook of Clinical Chemistry and Molecular Diagnostics, ed 4 St Louis, MO: Elsevier Saunders, 2006.

2. Pepys MB, Hirschfield GM: C-reactive protein: A critical update. J Clin Invest 111:1805–1812, 2003.[CrossRef][Medline]

3. Heikkilä K, Ebrahim S, Lawlor DA: A systematic review of the association between circulating concentrations of C reactive protein and cancer. J Epidemiol Community Health 61:824–833, 2007.[Abstract/Free Full Text]

4. Erlinger TP, Platz EA, Rifai N, et al: C-reactive protein and the risk of incident colorectal cancer. JAMA 291:585–590, 2004.[Abstract/Free Full Text]

5. Gunter MJ, Stolzenberg-Solomon R, Cross AJ, et al: A prospective study of serum C-reactive protein and colorectal cancer risk in men. Cancer Res 66:2483–2487, 2006.[Abstract/Free Full Text]

6. Heikkilä K, Harris R, Lowe G, et al: Associations of circulating C-reactive protein and interleukin-6 with cancer risk: Findings from two prospective cohorts and a meta-analysis. Cancer Causes Control 20:15–26, 2009.[CrossRef][Medline]

7. Il'yasova D, Colbert LH, Harris TB, et al: Circulating levels of inflammatory markers and cancer risk in the health aging and body composition cohort. Cancer Epidemiol Biomarkers Prev 14:2413–2418, 2005.[Abstract/Free Full Text]

8. Ito Y, Suzuki K, Tamakoshi K, et al: Colorectal cancer and serum C-reactive protein levels: A case-control study nested in the JACC Study. J Epidemiol 15:S185–S189, 2005 (suppl 2.[CrossRef][Medline]

9. Otani T, Iwasaki M, Sasazuki S, et al: Plasma C-reactive protein and risk of colorectal cancer in a nested case-control study: Japan Public Health Center-based prospective study. Cancer Epidemiol Biomarkers Prev 15:690–695, 2006.[Abstract/Free Full Text]

10. Rifai N, Buring JE, Lee IM, et al: Is C-reactive protein specific for vascular disease in women? Ann Intern Med 136:529–533, 2002.[Abstract/Free Full Text]

11. Siemes C, Visser LE, Coebergh JW, et al: C-reactive protein levels, variation in the C-reactive protein gene, and cancer risk: The Rotterdam Study. J Clin Oncol 24:5216–5222, 2006.[Abstract/Free Full Text]

12. Suzuki K, Ito Y, Wakai K, et al: Serum heat shock protein 70 levels and lung cancer risk: A case-control study nested in a large cohort study. Cancer Epidemiol Biomarkers Prev 15:1733–1737, 2006.[Abstract/Free Full Text]

13. Trichopoulos D, Psaltopoulou T, Orfanos P, et al: Plasma C-reactive protein and risk of cancer: A prospective study from Greece. Cancer Epidemiol Biomarkers Prev 15:381–384, 2006.[Abstract/Free Full Text]

14. Tsilidis KK, Branchini C, Guallar E, et al: C-reactive protein and colorectal cancer risk: A systematic review of prospective studies. Int J Cancer 123:1133–1140, 2008.[CrossRef][Medline]

15. Zhang SM, Buring JE, Lee IM, et al: C-reactive protein levels are not associated with increased risk for colorectal cancer in women. Ann Intern Med 142:425–432, 2005.[Abstract/Free Full Text]

16. Zhang SM, Lin J, Cook NR, et al: C-reactive protein and risk of breast cancer. J Natl Cancer Inst 99:890–894, 2007.[Abstract/Free Full Text]

17. Clarke R, Shipley M, Lewington S, et al: Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol 150:341–353, 1999.[Abstract/Free Full Text]

18. Al Murri AM, Bartlett JM, Canney PA, et al: Evaluation of an inflammation-based prognostic score (GPS) in patients with metastatic breast cancer. Br J Cancer 94:227–230, 2006.[CrossRef][Medline]

19. Heikkilä K, Ebrahim S, Rumley A, et al: Associations of circulating C-reactive protein and interleukin-6 with survival in women with and without cancer: Findings from the British Women's Heart and Health Study. Cancer Epidemiol Biomarkers Prev 16:1155–1159, 2007.[Abstract/Free Full Text]

20. McMillan DC, Elahi MM, Sattar N, et al: Measurement of the systemic inflammatory response predicts cancer-specific and non-cancer survival in patients with cancer. Nutr Cancer 41:64–69, 2001.[CrossRef][Medline]

21. Nielsen HJ, Christensen IJ, Sorensen S, et al; The RANX05 Colorectal Cancer Study Group. Preoperative plasma plasminogen activator inhibitor type-1 and serum C-reactive protein levels in patients with colorectal cancer. Ann Surg Oncol 7:617–623, 2000.[Medline]

22. Scott HR, McMillan DC, Forrest LM, et al: The systemic inflammatory response, weight loss, performance status and survival in patients with inoperable non-small cell lung cancer. Br J Cancer 87:264–267, 2002.[CrossRef][Medline]

23. Nordestgaard BG, Benn M, Schnohr P, et al: Nonfasting triglycerides and risk of myocardial infarction, ischemic heart disease, and death in men and women. JAMA 298:299–308, 2007.[Abstract/Free Full Text]

24. Schnohr P, Jensen JS, Scharling H, et al: Coronary heart disease risk factors ranked by importance for the individual and community: A 21 year follow-up of 12000 men and women from The Copenhagen City Heart Study. Eur Heart J 23:620–626, 2002.[Abstract/Free Full Text]

25. Storm HH: The Danish Cancer Registry, a self-reporting national cancer registration system with elements of active data collection. IARC Sci Publ, 220–236, 1991.

26. Storm HH, Michelsen EV, Clemmensen IH, et al: The Danish Cancer Registry: History, content, quality and use. Dan Med Bull 44:535–539, 1997.[Medline]

27. Storm HH: Completeness of cancer registration in Denmark 1943-1966 and efficacy of record linkage procedures. Int J Epidemiol 17:44–49, 1988.[Abstract/Free Full Text]

28. Coussens LM, Werb Z: Inflammation and cancer. Nature 420:860–867, 2002.[CrossRef][Medline]

29. Balkwill F, Mantovani A: Inflammation and cancer: Back to Virchow? Lancet 357:539–545, 2001.[CrossRef][Medline]

30. Engels EA: Inflammation in the development of lung cancer: Epidemiological evidence. Expert Rev Anticancer Ther 8:605–615, 2008.[CrossRef][Medline]

Submitted August 26, 2008; accepted January 21, 2009.

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Allin, K. H. Articles by Nordestgaard, B. G. Search for Related Content PubMed PubMed Citation Articles by Allin, K. H. Articles by Nordestgaard, B. G. Social Bookmarking                            What's this?