Table 4

 Summary of prospective studies of the association between circulating CRP and cancer

Study [author, year]Study designNumber of cancer casesAgeNo. (%) femaleMain aim of the studyMain results relating to CRPComments
Colorectal cancer
CLUE II, US [Erlinger, 2004]Case-control, nested in CLUE II cohortColorectal cancer: 172 Controls: 342Mean (SD): Cases 63.6 (11) Controls 63.4 (11)Cases: 95 (55.2) Controls: 189 (55.3)Aetiology To assess the association between CRP and incident colorectal cancer.OR for colorectal cancer across quartiles of CRP (mg/l) [95% CI]: 1st (<0.92): 1.00 2nd (0.92, 1.93): 1.16 [0.65, 2.09] 3rd (1.94, 3.69): 1.42 [0.82, 2.46] 4th (>3.69): 2.00 [1.16, 3.46] p for trend = 0.008Up to 2 controls, matched for age, sex, race and date of blood sample, were selected from a community-based cohort for each case. Further adjustment for age, sex, smoking status, BMI and use of hormones or NSAIDs.
Women’s Health Study, US [Zhang, 2005]Cohort, nested in an RCTColorectal cancer: 16945+ at baseline27 913 (100)Aetiology To evaluate whether plasma CRP concentrations predict colorectal cancer.HR for colorectal cancer across categories of CRP, mg/l, [95% CI]: Crude <1 mg/l: 1.00 [ref.] 1–3 mg/l: 0.89[0.62, 1.29] >3 mg/l: 0.80 [0.55, 1.15] p = 0.24 Age-adjusted: <1 mg/l: 1.00 [ref.]Participants recruited from a community-based RCT. Adjusted for age, BMI, family history of colorectal cancer, physical activity, smoking, alcohol intake, menopausal status, use of aspirin, vitamin E, multivitamin, oral contraceptive or postmenopausal hormones.
1–3 mg/l: 0.77 [0.53, 1.11]
>3 mg/l: 0.67 [0.46, 0.97]
p = 0.05
Multivariate adjusted:
<1 mg/l: 1.00 [ref.]
1–3 mg/l: 0.79 [0.53, 1.17]
>3 mg/l: 0.66 [0.43, 1.03]
p = 0.09
Various locations, Japan [Ito, 2005]Case-control nested in prospective cohortColorectal cancer: 141 Controls: 327Range: 40–79Cases: 78 (55.3) Controls: 179 (54.7)Aetiology To investigate whether serum CRP is associated with colorectal cancer in the Japanese.Crude OR [95% CI] for incident colorectal cancer, by tertiles of CRP: 1st: 1.0 (ref.) 2nd: 0.93 [0.52, 1.65] 3rd: 0.91 [0.50, 1.66] p for trend: 0.77Study nested in a large community-based multi-centre cohort. Controls matched for age, sex and site. Adjusted for BMI, smoking and alcohol consumption.
Multivariable-adjusted OR [95% CI] for incident colorectal cancer, by tertiles of CRP:
1st: 1.00 (ref.)
2nd: 1.05 [0.57, 1.94]
3rd: 0.97 [0.51, 1.83]
p for trend: 0.98
Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study, Southwest Finland [Gunter, 2006]Prospective case-control nested in an RCTColorectal cancer: 130 Controls: 260Median (IQR): Colorectal cancer: 56 (53–61) Controls: 57 (53–59)0 (0)Aetiology To investigate the relationship between serum CRP and incident colorectal cancer and to examine whether any association differs by body size or cancer site.Age-adjusted OR [95% CI] for incident colorectal cancer, by quartiles of CRP, mg/l: 1st: 1.0 (ref.) 2nd: 1.5 [0.8, 2.7] 3rd: 1.0 [0.5, 1.9] 4th: 2.0 [1.1, 3.7] p = 0.02 Multivariable-adjusted OR [95% CI] for incident colorectal cancer, by quartiles of CRP, mg/l:Study nested in an RCT where the participants were male smokers. Controls were matched on age, date of the baseline blood draw and intervention group (α-tocopherol, β-carotene, both or placebo). Further adjustment for age, BMI, aspirin use, smoking duration (years) and usual no. of cigarettes smoked per day.
1st: 1.0 (ref.)
2nd: 1.9 [1.0, 3.8]
3rd: 1.2 [0.6, 2.6]
4th: 2.9 [1.4, 6.0]
p = 0.006
Japan Public Health Center-based Prospective Study (JPHC), various locations, Japan [Otani et al., 2006]Prospective case-control nested in a cohortColorectal cancer: 375 Controls: 750Mean: Colorectal cancer: 56.7 Controls: 56.6Colorectal cancer: 179 (47.7) Controls: 358 (47.7)Aetiology To assess the association of circulating CRP and colorectal cancer risk in relation to tumour location and invasion.Multivariable-adjusted OR [95% CI] for incident colorectal cancer, by quartiles of CRP, mg/l: 1st: 1.0 (ref.) 2nd: 1.5 [0.99, 2.2] 3rd: 1.3 [0.85, 2.0] 4th: 1.6 [1.1, 2.5] p for trend  = 0.053 By cancer type, highest vs. lowest quartile of CRP:All participants selected from people with available blood samples, within a large community based cohort. Controls matched for age, sex, date of blood draw, time since last meal and study location. Further adjustment for smoking (pack-years), alcohol consumption, exercise, and family history of colorectal cancer.
Colon (n = 244): 1.6 [0.99, 2.7]
p for trend = 0.041
Rectal (n = 111): 1.4 [0.63, 3.3]
p for trend  = 0.82
Colon, intramucosal (n = 94): 2.6 [1.1, 6.2]
p for trend  = 0.017
Colon, invasive (n = 146): 1.2 [0.64, 2.4]
p for trend  = 0.55
Prostate cancer
CLUE II, US [Platz, 2004]Case-control study nested in CLUE II cohortProstate cancer: 264 Controls: 26418+0 (0)Aetiology To examine the association of CRP with prostate cancer.Geometric mean (SD) CRP (mg/l): Prostate cancer: 1.24 (2.94) Controls: 1.41 (2.97) p = 0.16 Adjusted OR for prostate cancer across quartiles of CRP [95% CI]: 1st: 1 2nd: 1.29 [0.80, 2.08] 3rd: 0.98 [0.61, 1.58]Controls and cases selected from male participants of the CLUE II cohort who were Washington County residents. Controls matched on age, date of blood draw, race and time since last meal. Further adjustment for BMI, age at diagnosis and smoking history did not change the effect estimates.
4th: 0.95 [0.57, 1.58]
p for trend  = 0.66
Any cancer
The Women’s Health Study, US [Rifai, 2002]Case-control, nested in an RCTAny cancer: 513 Controls: 513Mean: 56.71026 (100)Aetiology To examine predictive value of CRP for cancer and coronary heart disease in women.Crude RR for incident cancer across quartiles of baseline CRP (mg/l) [95% CI]: 1st (<1.0): 1.0 2nd [1.0, 2.3): 1.1 [0.8, 1.5] 3rd (2.5–5.6): 1.0 [0.7, 1.4] 4th (⩾5.7): 1.2 [0.9, 1.6] p for trend >0.2Community controls, matched on age and smoking status. Further adjustment for BMI, hypertension, diabetes, hyperlipidemia, exercise, parental history of heart disease and random assignment to vitamin E, aspirin or both. High baseline CRP in cancer patients may be a result of an already begun cancerous process.
Baseline CRP was not associated with breast, ovarian or uterine, colon, lung, hematopoietic, thyroid, bladder, brain or pancreatic cancers, melanoma or other types of cancer (data not shown).
Health Aging and Body Composition study, Memphis, TN and Pittsburgh, PA, US [Il’yasova, 2005]CohortTotal participants: 2438 Any incident cancers: 296Median (IQR): Cases: 74 (71–76) Non-cases: 73 (71–76)Cases: 55% Non-cases: 51% Numbers not citedAetiology To analyse the association between circulating inflammatory markers and incident cancer in elderly people.HR [95% CI] for incident cancer events/log CRP (mg/l): Crude: 1.17 [1.02, 1.33] Adjusted: 1.25 [1.09, 1.43] HR [95% CI] for cancer events by type/log CRP (mg/l): Colorectal: 1.44 [1.03, 2.02] Lung: 1.64 [1.20, 2.24] Breast: 1.32 [0.91, 1.93] Prostate: 0.94 [0.70, 1.28] Estimates adjusted for age, gender, race and site.Participants recruited from Medicare beneficiaries. Exclusion criteria were very poor physical condition, life-threatening disease or intent to leave the area in the subsequent 3 years. Adjustment for age, gender, race and site. Further adjustment for BMI, pack-years of cigarettes smoked, physical activity, education, baseline medical conditions and medication used did not change the effect estimates (data not shown).
Colorectal and lung cancer analyses were based on subgroups of 41 and 42 events, respectively.
European Prospective Investigation into Cancer and Nutrition cohort, Greece [Trichopoulos, 2006]Case-control, nested in EPIC cohortAny cancer: 496 Controls: 996No. (%) participants: <55 years: Cancer: 153 930.8) Controls: 289 (29.0) ⩾55 years: Cancer: 215 (43.3) Controls: 707 (71.0)Cancer cases: 253 (51.0) Controls: 512 (51.4)Diagnosis To assess the association between CRP and subsequent risk of cancer and to determine whether CRP could be used to identify individuals at high risk at a sub-clinical stage of the disease.Adjusted OR [95% CI] for incident cancer per 1 SD of log CRP: All cancers: 1.20 [1.10, 1.32] Cancers by site: Stomach: 1.10 [0.82, 1.47] Colon-rectum: 1.17 [0.93, 1.46] Liver cancer: 1.51 [1.20, 1.90] Pancreas: 1.29 [0.89, 1.87] Lung: 1.31 [1.11, 1.53] Skin: 1.24 [0.95, 1.62] Kidney: 1.48 [1.11, 1.53] Bladder: 1.21 [0.91, 1.61] Brain: 1.00 [0.54, 1.85] Leukemia/lymphoma: 1.26 [1.05, 1.51]Community controls matched for age, sex and date of cohort enrolment. Cases were participants free of cancer at baseline. Adjustment for age, sex, BMI, smoking, alcohol and NSAID use and duration of storage of plasma samples. Some site specific OR estimates were based on small subgroups. Effect estimates strengthened somewhat with removal of cases occurring in the first year of follow-up.
Breast (women): 1.16 [0.95, 1.41]
Cervix uteri: 1.31 [0.72, 2.35]
Corpus uteri: 1.34 [1.03, 1.74]
Ovary: 1.00 [0.67, 1.48]
Prostate: 0.74 [0.37, 1.47]
All cancers, stratified by smoking:
Never-smokers: 1.17 [0.96, 1.42]
Ever-smokers: 1.50 [1.22, 1.84]