Risk of prostate cancer is not associated with levels of C-reactive protein and other commonly used markers of inflammation†
Role of funding organizations: The funding organizations had no influence in the design and conduct of the study, collection, management, analysis and interpretation of data, and preparation, review or approval of the manuscript. The corresponding author had full access to all of the data and the final responsibility to submit for publication
Most population-based studies studied the association between inflammation and prostate cancer (PCa) by assessing C-reactive protein (CRP). As these findings have shown inconsistent results, we aimed to also study different markers that have been commonly taken as indications of inflammation. A cohort based on four groups of men (n = 34,891), according to age at cohort entry (45, 55, 65 and 75 years), with measurements of glucose, triglycerides, total cholesterol, haptoglobin, albumin, hemoglobin and leukocytes were selected from the Apolipoprotein Mortality Risk database. A total of 17,937 men had measurements of non-high-sensitive CRP. Multivariate Cox proportional hazard models were used to analyze associations between inflammatory markers and PCa. A total of 49 of 12,063 men developed PCa in the age 45 group, whereas 207 of 9,940, 472 of 8,266 and 276 of 3,618 were diagnosed in the age 55, 65 and 75 groups, respectively. Mean follow-up time was 7.5 years (SD: 3.9). No markers showed an association with PCa risk, nor was there a trend by quartiles or an indication for different PCa risks by strata of hypercholesterolemia, hyperglycemia and hypertriglyceridemia status. The studied markers were not found to be associated with PCa risk. These null findings might be due to methodological issues; however, it is unlikely that strong and long-lasting associations between inflammation and PCa risk were missed as this was a large database with long follow-up. This indicates need for international consensus on appropriate inflammatory markers in the context of cancer that may be practically applied in large studies.
Inflammation is known to be one of the mechanisms that play a role in the progression and severity of many cancers. In the context of prostate cancer (PCa), De Marzo et al. proposed that environmental factors such as infectious agents and dietary carcinogens, and hormonal imbalances may lead to injury of the prostate and to the development of chronic inflammation and proliferative inflammatory atrophy.1 The California Men's Health Study, including 68,675 men of whom 1,658 developed PCa, showed that men having a history of prostatitis had an increased risk of PCa compared to those with no history (Hazard ratio (HR): 1.30; 95% CI: 1.10–1.54). They did not find an association between sexually transmitted diseases (STDs) and overall PCa risk, but when stratifying by ethnicity Latinos were found to have an increased risk of PCa when reporting any STDs (HR: 1.43; 95% CI: 1.07–1.91).2 Another example of epidemiological evidence for a link between inflammation and PCa is a recent study in the Prostate Cancer Prevention Trial that showed an association between high intake of dietary polyunsaturated fat and risk of high-grade PCa (HR: 2.41 (95% CI: 1.33–4.38), comparing the fourth with the first quartile).3 The n-6 fatty acids, which constitute the majority of dietary polyunsaturated fats, are proinflammatory, suggesting a role for inflammation in PCa development3; however no strong evidence is available from epidemiological studies.4 Inflammation is also a commonly suggested mechanism linking the metabolic syndrome and PCa. Obesity, one of the components of the metabolic syndrome, is associated with a state of low-grade chronic inflammation, with infiltrating macrophages within adipose tissue and elevated concentrations of proinflammatory molecules as tumor necrosis factor (TNF)-α, interleukin-6 (IL-6), haptoglobin and C-reactive protein (CRP).5, 6 Adipose tissue is thus an active organ that secretes a large number of proteins, including proinflammatory factors such as cytokines, suggesting a possible involvement in PCa biology.6
Most observational studies investigating inflammation and risk of PCa have focused on the positive acute-phase protein CRP. In a recent meta-analysis of five studies including 628 PCa cases, CRP was not associated with PCa risk (random pooled risk estimate: 1.00; 95% CI: 0.88–1.13).7 The authors of this meta-analysis also published the first study that found a positive association of CRP and PCa, although the small sample size (36 PCa cases) entailed a limited statistical precision.8 However, the largest study to date, a prospective case–control study nested within the Physicians' Health Study, did not find a trend for quartiles of CRP and risk of PCa (552 PCa cases).9
Many of the inflammatory markers such as circulating proinflammatory cytokines and adhesion are not yet available in routine laboratory practice. Apart from non-high-sensitive(hs)CRP, our study therefore also evaluated the association between inflammation and risk of PCa by assessing four other markers that are commonly measured in clinical practice and that may be indicative for inflammation (albumin, haptoglobin, hemoglobin and leukocytes).10 Moreover, most population-based studies today studied the association between proinflammatory markers and PCa by assessing CRP in databases including less than 300 men with PCa.11–14 As findings have shown both positive and null associations, we aimed to also study different markers that have been commonly taken as indications of inflammation and are widely used in medical practice. These were studied in the context of PCa risk in a prospective database (n = 34,891) that also provided information on metabolic markers such as triglycerides (TG), total cholesterol (TC) and glucose.
Material and Methods
The Central Automation Laboratory (CALAB) database (1985–1996) includes data obtained from 351,487 men and 338,101 women, mainly from the greater Stockholm area (Sweden). All individuals were either healthy individuals referred for clinical laboratory testing as part of a general health checkup or outpatients referred for laboratory testing. No individuals were inpatients at the time their blood samples were taken, and none was excluded for disease symptoms or because of treatment. Apart from the information on blood testing, no clinical data were included in the CALAB database.15 This database was linked to several Swedish national registries such as the National Cancer Register, the Hospital Discharge Register, the Cause of Death Register, the consecutive Swedish Censuses during 1970–1990 and the National Register of Emigration by using the Swedish ten-digit personal identity number to provide information on socioeconomic status (SES), vital status, cancer diagnosis and emigration. This linkage of national registers to the CALAB database is called the Apolipoprotein MOrtality RISk (AMORIS) study, and it has been described in detail elsewhere.15–21 Our study complied with the Declaration of Helsinki, and the ethics review board of the Karolinska Institute approved the study.
As the risk of PCa is strongly age dependent, we created a subcohort of the AMORIS study consisting of four age groups (45, 55, 65 and 75 years old) of men, with information on TG, TC, glucose, albumin, haptoglobin, hemoglobin and leukocytes available from the same health examination (n = 34,891). Non-hsCRP was measured for a subgroup of 17,937 men. Our study design has been described in detail elsewhere.21 It allows for a more unbiased comparison by age of the association between inflammation and PCa. Briefly, men were allocated to age group 45 if they had their blood sample taken within 10 years before age 45 and were not diagnosed with PCa before the age of 45, nor were they diagnosed with PCa or did they die within 3 months after they turned 45. The same reasoning applies to age groups 55, 65 and 75. Follow-up time was defined for each men as the time from age group entry (age 45, 55, 65 or 75) until the date of PCa diagnosis, death or study closing date (December 31, 2002), whichever occurred first.21
The following information was retrieved from the CALAB database: serum TG (mmol/L), serum TC (mmol/L), serum glucose (mmol/L), serum albumin (g/L), serum haptoglobin (g/L), serum hemoglobin (g/L), leukocytes (109/L), CRP (mg/L) and age. All other information was taken from the aforementioned national registries. SES is based on occupational groups and classifies gainfully employed subjects into manual workers and nonmanual employees, below designated blue-collar and white-collar workers.22 TC and TG were measured enzymatically as described previously.16, 23 Glucose was measured enzymatically with a glucose oxidase/peroxidase method. Albumin was measured with a bromcresol green method (coefficient of variation < 1.8%), hemoglobin and leukocytes with routinely used hematology analyzers. Haptoglobin and CRP were measured by an immunoturbidimetric method (reagents from Orion Diagnostics, Finland). Total imprecision calculated by the coefficient of variation was 5.6% at haptoglobin level 1.1 g/L and 12% at CRP level 40 mg/L. hsCRP assessment was not available yet during the whole period of blood sampling collection and analyses (1985–1996).10 All methods were fully automated with automatic calibration and accredited laboratory facilities.16
As already mentioned, before 1996, immunochemical assay methods for plasma proteins had limited sensitivity so that CRP concentrations <10 mg/L could not be measured precisely (i.e., non-hsCRP), and this cutoff of 10 mg/L was widely accepted as the upper limit of the health-associated reference range.24 To our knowledge, no study has investigated the effect of using hsCRP instead of non-hsCRP in the context of inflammation and cancer risk, but it is likely that low-grade inflammation is not captured by using this cutoff of 10 mg/L, resulting in an underestimation of the association between CRP and PCa. However, the use of this cutoff value is thought to be satisfactory for the purpose of medical events such as ischemic necrosis24 and has been used in several other studies looking into the association between CRP and cancer diagnosis and prognosis.25, 26
Associations between continuous values and quartiles of inflammatory markers and risk of PCa were analyzed by using multivariate Cox proportional hazards models in an age-stratified analysis, which allowed for different baseline hazards in different age groups. All models were adjusted for glucose (continuous), TG (continuous), TC (continuous), fasting status, SES and time between measurement and age at group entry. A test for trend was conducted by using assignment to quartiles as an ordinal scale. All analyses were repeated for each age group separately. CRP was investigated by using five categories of CRP (<10, 10–15, 15–25, 25–50 and >50 mg/L). All analyses were also conducted for categorical values of each inflammatory marker based on their clinical cutoffs, with the normal range being the reference level. Levels of glucose, TG and TC were dichotomized for stratified analyses, based on the National Cholesterol Education Program guidelines (cutoffs: 6.11, 1.71 and 6.50 mmol/L for glucose, TG and TC, respectively).27 To study the effect of reverse causation, a sensitivity analysis was conducted in which all men who had their measurements taken within 1.5 years before PCa diagnosis were deleted.28 All analyses were conducted with Statistical Analysis Systems release 9.1.3 (SAS Institute, Cary, NC).
A total of 49 of 12,063 men developed PCa in the age 45 group, whereas 207 of 9,940, 472 of 8,266 and 276 of 3,618 were diagnosed in the age 55, 65 and 75 groups, respectively. All participant characteristics by PCa status are shown in Table 1. About 90% of the population was gainfully employed because the measurements were taken as part of health examinations done at company health checkups. The majority of PCas were detected after 1997 (66.73%), and the mean time between measurement and cohort entry was comparable for men with and without PCa (3.40 and 4.40 years, respectively). The distribution of hypertriglyceridemia (TG ≥ 1.71 mmol/L), hypercholesterolemia (TC ≥ 6.50 mmol/L) and hyperglycemia (glucose ≥ 6.11 mmol/L) was similar for men with and without PCa. Little difference was observed for the distribution of the inflammatory markers by cancer status. The participant characteristics in the subgroup with measurements of CRP were comparable to those of the total cohort studied in Table 1 (results not shown).
Table 1. Descriptive statistics of the study population by prostate cancer status
The correlation coefficients between each marker of inflammation are shown in Table 2. All markers were statistically significant correlated with each other. The strongest correlation was seen between haptoglobin and leukocytes (r = 0.33), haptoglobin and CRP (r = 0.24) and hemoglobin and albumin (r = 0.23).
Table 2. Correlation matrix for markers currently taken as an indication of inflammation
Table 3 shows age-stratified HRs of PCa risk for continuous values and quartiles of each marker as well as HRs for each association by age group. None of the markers showed an association with PCa risk, nor was there a trend by quartiles. Also, for CRP, no association or trend was found. The association between inflammatory markers and PCa risk was also analyzed by strata of hypercholesterolemia, hyperglycemia and hypertriglyceridemia status, but there was no indication for different PCa risks by these strata (results not shown).
Table 3. Hazard ratios of prostate cancer in quartiles of inflammatory markers, adjusted for glucose (continuous), triglycerides (continuous), total cholesterol (continuous), fasting status, SES and time between measurement and age at group entry
When using clinical cutoffs for each marker, there was also no association observed between the markers and risk of PCa (Table 4). The age-stratified analysis for high levels of hemoglobin (>175 g/L) showed a statistically significant increased risk of PCa (HR: 4.15, 95% CI: 1.04–16.62), but this association disappeared when analyzing it by different age groups (Table 4).
Table 4. Hazard ratios of prostate cancer for clinical cutoffs of inflammatory markers, adjusted for glucose (continuous), triglycerides (continuous), total cholesterol (continuous), fasting status, SES and time between measurement and age at group entry
Only 26 men had their measurements of inflammatory markers taken within 1.5 years before their PCa diagnosis, resulting in no changes in the previous findings when excluding these men (results not shown). When extending the sensitivity period to 3.0 years, 56 men were excluded, but this also did not change the previous findings.
This is the first large study assessing an association between PCa risk and widely available clinical markers of inflammation other than CRP. Even though there is experimental evidence to support such an association, we did not find any indication for a link between the inflammatory markers studied and PCa risk.
Despite the growing experimental evidence for a link between inflammatory components (e.g., TNF-α, IL-6 or IL-8) and PCa risk, none of them are standard diagnostic tools in the current medical practice.29, 30 Moreover, few population-based studies have investigated the association between proinflammatory markers and PCa, and results are contradictory.8, 13, 14, 31–33 Most studies investigating inflammation and risk of PCa have focused on CRP.8 Some studies also looked into the association between risk of PCa and other proinflammatory molecules such as COX2 and IL-6.1, 34 To our knowledge, no observational study has yet looked into a possible association between albumin, haptoglobin, hemoglobin, leukocytes, CRP and PCa risk.
Leukocytes and albumin, an acute-phase protein, have often been studied as markers of systematic inflammation in the context of cancer survival: low levels of albumin and high levels of leukocytes are associated with worse cancer prognosis.35 In a prospective cohort of 4,831 subjects, these markers were also studied to assess the relation between inflammation and lung cancer. It was found that participants with leukocyte counts in the upper tertile were 2.81 times (95% CI: 1.58–5.01) more likely to develop lung cancer as those with counts in the lowest tertile. Serum albumin was not found to be related to lung cancer risk.36 The full scope of the biological function of haptoglobin, a positive acute-phase protein, is not yet defined, but experimental studies have identified increased intact haptoglobin in sera from, for instance, ovarian cancer patients and have hypothesized that haptoglobin polymorphisms might contribute to increased oxidative stress and low-grade chronic inflammation.37, 38 Haptoglobin has also been used as a marker of inflammation in the context of obesity.39, 40 In a cross-sectional study of 562 persons, it was found that haptoglobin was positively correlated with body mass index (BMI) and waist–hip ratio (r = 0.35 and 0.22, respectively, with p-value < 0.05).39 Hemoglobin is also thought to be associated with inflammation, as high CRP levels have been shown to be predictive for less stable hemoglobin levels, which is related to poor survival.41 Moreover, it is thought that inflammation (e.g., cytokines released by activated leukocytes) contributes to the reduction in hemoglobins via different mechanisms such as the inhibition of erythropoietin release from the kidney by TNF-α, resulting in anemia of chronic disease.42
Our study showed no association between CRP and PCa risk. By using other commonly used markers of inflammation and stratification by metabolic statuses, we aimed to assess the link between inflammation and PCa in more detail. However, our results did not show a link between any of the markers studied and risk of PCa. If anything, there was an emerging pattern for hemoglobin, but despite its statistically significance the pattern was difficult to interpret.
The major strength of this analysis is the large number of men with PCa endpoints and prospective measurements of inflammatory markers in the AMORIS database, all measured at the same clinical laboratory. This database provided complete follow-up for each person as well as linkage to other registers allowing for detailed information on cancer diagnosis and time of death. The population in the AMORIS study was selected by analyzing blood samples from health checkups in nonhospitalized persons. However, any healthy cohort effect would not affect the internal validity of our study. We have no indication that these inflammatory markers were measured due to disease symptoms. A limitation of our study is that it did not have access to other commonly measured markers for inflammation such as hsCRP or IL-6, nor did we have repeated measurements to verify the time line between changes in inflammatory markers and risk of PCa. In the AMORIS study, it was not possible to study hsCRP because at the time of blood sampling and analysis (1985–1996) only a non-hsCRP assay was available. To our knowledge, no study has yet addressed the difference between using non-hsCRP and hsCRP in the context of cancer risk. We have used the cutoff that has previously been suggested as medically relevant when using non-hsCRP.24 However, it is a limitation that we cannot specify the CRP values <10 mg/L. Furthermore, we did not have information on other possible confounders such as obesity. Obesity is associated with a state of low-grade chronic inflammation, characterized by infiltrating macrophages within adipose tissue and elevated concentrations of proinflammatory molecules.5, 6 To date, it is unclear whether inflammation is an intermediate on the pathway between obesity and PCa or whether obesity is confounding the association between inflammation and PCa. As our study focused on the latter, we believe that residual confounding due to lack of information on BMI is minor. Our study did, however, provide stratified analyses by metabolic abnormalities (measurements of TG, TC and glucose), which are strongly correlated with obesity.27 Nevertheless, the results must be interpreted cautiously without additional adjustment for BMI. From Table 1, it can be seen that the majority of PCa was detected after 1997, suggesting that some cancers might have been detected with PSA screening. However, we had no individual information on how PCa was detected so that it is not possible to estimate whether screening, and as a consequence PCa stage, had an effect on the association between inflammation and PCa. Finally, no information was available on tumor severity.
Current empirical evidence from many sources supports a link between inflammation and PCa.1 However, in our study, markers of inflammation commonly used in clinic were not found to be associated with PCa risk. These null findings might be due to possible methodological issues; however, it is unlikely that strong and long-lasting associations between inflammation and PCa risk were missed as this was a large database with long follow-up. Therefore, there is the need for an international consensus on appropriate markers of inflammation in the context of cancer risk that may be practically applied in large studies.43