Association of Albuminuria and Cancer Mortality

Albuminuria is an important and independent risk marker for cardiovascular disease in the general population and, more specifically, in patients with hypertension and diabetes mellitus (1-3). Whether albuminuria is a contributory factor in atherothrombosis or merely a marker remains unclear. However, evidence suggests that common pathophysiologic processes such as endothelial dysfunction, chronic low-grade inflammation, and transvascular leakage of macromolecules may underlie the association between microalbuminuria and cardiovascular disease (4). Microalbuminuria is also considered an early sign of intrarenal vascular dysfunction and a potent marker for the subsequent development of progressive kidney disease (5-7).

Albuminuria is associated with increases in all-cause and cardiovascular mortality in uncomplicated hypertensives and type 2 diabetics (8, 9). When combined as indicators of renal dysfunction, decreased glomerular filtration rate and albuminuria predict both all-cause and cardiovascular mortality (10). In the Prevention of Renal and Vascular End Stage Disease study conducted in the Netherlands, a 2-fold increase of albuminuria was associated with a relative risk (RR) of 1.29 and 1.12 for cardiovascular and noncardiovascular mortality, respectively (1).

The Prevention of Renal and Vascular End Stage Disease study proposed that albuminuria-associated noncardiovascular deaths might represent greater fatalities due to malignant neoplasms, although specific etiologies were not analyzed. Little is known regarding the association of albuminuria with cancer mortality. Because cancer is currently second only to cardiovascular disease as a leading cause of death in the United States, the hypothesis was formulated that albumin in the urine would be associated with an increased risk of cancer mortality. The objective of this study was to test the hypothesis by evaluating the relationship between albuminuria and cancer mortality using the large-scale nationally representative sample from the Third National Health and Nutrition Examination Survey (NHANES III, 1988-1994) and its follow up in the NHANES III–linked mortality data set (through December 31, 2006; refs. 11, 12).

NHANES III was conducted by the National Center for Health Statistics from 1988 to 1994 using a stratified, multistage, and cluster sampling design to obtain a representative sample of the non-institutionalized civilian U.S. population. Detailed survey operations manuals, consent documents, and brochures of the NHANES are available on the NHANES web site (11). The NHANES III was reviewed and approved by the National Center for Health Statistics Institutional Review Board. Of 7,593 adults ≥50 years of age that completed the NHANES III interview and examination between 1988 and 1994, the current analysis was limited to non-Hispanic Whites, non-Hispanic Blacks, and Mexican Americans by excluding 266 participants identified as other race. Participants with previous history of malignancy (n = 547), missing data for follow-up information (n = 19), and other covariates (e.g., urinary albumin measurement, n = 664) were also excluded from the cohort. The remaining 6,112 subjects (2,996 men and 3,116 women) in the analyses represent an estimated U.S. non-institutional population of 49,914,220 persons ages 50 and older after adjusting for sample weights.

In NHANES III, data were collected at all study sites by trained personnel according to standardized procedures. Information on age, sex, and smoking status was self-reported. Subjects with serum cotinine values >14 ng/mL were classified as current smokers, otherwise as nonsmokers (13). Subjects were classified as diabetic based on a self-reported diagnosis, prescribed medication for diabetes (insulin or oral agents), or a fasting plasma glucose of ≥126 mg/dL (14, 15). A diagnosis of hypertension was assigned if the subjects reported taking prescribed medication for hypertension, a physician diagnosis of hypertension, or if the systolic blood pressure was ≥140 mm Hg and/or diastolic blood pressure was ≥90 mm Hg. Three to four blood pressure (BP) determinations were made using a mercury sphygmomanometer from the right arm by a physician. Averaged systolic and diastolic blood pressure determinations were obtained. Body mass index (BMI) was defined as body weight (kg) divided by the square of the body height (m²). Blood was sampled, processed locally, stored, and then shipped to central laboratories for analysis (12).

A spot urine sample was collected to determine urinary albumin using a solid phase fluorescent immunoassay, capable of measuring values below the limit of detection by the dipstick method (16). Urine creatinine was analyzed using the Jaffé reaction. The urinary albumin-to-creatinine ratio (UACR in mg/mmol) was calculated by dividing the urinary albumin value by the urinary creatinine concentration for each subject. The presence of albuminuria was defined as UACR >2.5 mg/mmol in men and >3.5 mg/mmol in women (17).

Serum creatinine was measured by means of the modified kinetic Jaffé reaction, and standardized serum creatinine values were calibrated by using an enzymatic creatinine assay traceable to a gold standard reference method (18). The glomerular filtration rate was calculated from the modification of diet in renal disease (MDRD) Study equation: MDRD estimated glomerular filtration rate (eGFR; mL/min/1.73 m²) = 175 × (serum creatinine, mg/dL)^−1.154 × (age)^−0.203 × (0.742 if female) × (1.212 if African American; ref. 19). The presence of chronic kidney disease was defined as MDRD eGFR <60 mL/min/1.73 m² (20).

Several of the most common cancers were selected for analysis. Lung, colorectal and prostate were examined in the current study based on cause of death codes from the International Classification of Diseases, 10th Revision (ICD-10). Cancer mortality was analyzed for all-cause cancer (ICD codes: C00-C95), lung cancer (ICD codes: C33-34), colorectal cancer (ICD codes: C18-21), and prostate cancer (ICD codes: C61). Cancers other than lung cancer, colorectal cancer, or prostate cancer (in men) were defined as other types of cancer. Follow-up for each participant was calculated as the difference between the NHANES III examination date and the end of follow-up (date of death or December 31, 2006, whichever occurred first). Those found alive were right-censored at the last date known alive or at the end of the follow-up. For cancer-specific analysis, those who died from other cancers were also right-censored at the time of death.

Continuous variables were reported as median (interquartile range) and categorical data in percentages. For outcome measurements, the Cox proportional hazard model with the Efron method was used to calculate the RR, also known as hazard ratio (HR). RRs were determined for mortality risk associated with albuminuria for all-cause and cause-specific cancer (lung, prostate, colorectal, and other types of cancer). Possible confounding factors, such as age, race/ethnicity, smoking, BMI, hypertension, diabetes, and chronic kidney disease were added to adjust the association of albuminuria with cancer mortality. Except for age and BMI, all variables were fit as categorical terms, which were coded as indicator (dummy) variables. For instance, two indicator variables were created to code for three race/ethnicity categories: non-Hispanic Whites, non-Hispanic Blacks, and Mexican Americans. Statistical analyses were conducted using SUDAAN 10.01 (Research Triangle Institute, 2009) with the Taylor series linearization and were adjusted for sampling weights to obtain unbiased national estimates in all analyses. Two-tailed P < 0.05 values were accepted as statistically significant.

The baseline demographics of study subjects (2,996 men and 3,116 women) are presented in Table 1. Except smoking (both men and women) and BMI (men only), all demographic and clinical variables were statistically different between those with and those without albuminuria (P < 0.05) for both sexes. In general, albuminuria tended to be more prevalent among the elderly and subjects with chronic kidney disease, hypertension, or diabetes. Albuminuria was also associated with race/ethnicity such that albuminuria was more common in non-Hispanic blacks. Although statistically different and suggestive of a marginal reduction in renal function, it is unclear whether the higher serum creatinine and lower MDRD eGFR are biologically significant differences.

Between 1988 and 2006, a total of 613 subjects (8.8%) died of cancer (all-cause) in an average follow-up of 149 months (range, 1-217 mo). This included 184 lung cancers (123 men and 61 women), 65 colorectal cancers (48 men and 17 women), 55 prostate cancers, and 309 other cancer deaths (152 men and 157 women). This corresponds to age-adjusted mortality rates (year 2000 U.S. standard population) of 3.2 for lung, 0.94 for colorectal, 1.1 for prostate, and 3.7 for other cancers for men and 1.7 for lung, 0.44 for colorectal, and 4.4 for other cancers for women per 1,000 person-years, respectively (data not shown). As shown in both Kaplan-Meier survival curves (Figs. 1 and 2 for men and women, respectively) and univariate Cox proportional hazard regression (Table 2), subjects with albuminuria are at increased mortality risk from all all-cause cancer, prostate cancer (men only), and the collected cancers other than lung, prostate, and colorectal cancers, as compared with those without albuminuria for both sexes. For instance, the RR of all-cause cancer deaths were 83% higher among men with albuminuria than those without it [RR, 1.83; 95% confidence interval (CI), 1.36-2.47; P < 0.001]. Despite a numerically increased mortality risk for both lung and colorectal cancer deaths in men (RR, 1.39 and 1.99, respectively), these results failed to reach statistical significance. In comparison, there is not even a suggestion of an association of albuminuria with lung and colorectal cancer mortality in women.

Similar results were obtained by applying the multivariate model which included adjustments for age, race/ethnicity, smoking, BMI, and hypertension, diabetes, and chronic kidney disease status. However, the resulting RRs were generally reduced by approximately 20% to 40%, and none of the associations reached statistical significance. For instance, the risk for all-cause cancer mortality associated with albuminuria decreased from RR, 1.83 (95% CI, 1.36-2.47; P ≤ 0.001) to 1.38 (95% CI, 0.99-1.91; P = 0.06), and became marginally significant in men (Table 2). Indeed, age and smoking are two of the most important factors in cancer deaths. In this cohort, age was significantly associated with increased cancer mortality for all-cause and cause-specific cancers regardless of sex except lung cancer in women (P = 0.10; data not shown). On the other hand, smoking is a significant risk factor for all-cause and lung cancer for both sexes, and other types of cancer deaths for men. Of note, neither albuminuria nor smoking were related to colorectal cancer death for men or women. Likewise, no association was found between smoking and prostate cancer death (RR, 1.44; 95% CI, 0.66-3.14; P = 0.35; data not shown).

Comparable RRs were generated from the multivariate-adjusted analyses using log-transformed UACR as predictor. Logarithmically transformed UACR was significantly correlated with all of the outcomes of interest except colorectal cancer in men, suggesting a dose-response relationship between albuminuria and cancer mortality (Table 3). Of note, there is a positive association between log-transformed UACR and prostate cancer in men (RR, 1.40; 95% CI, 1.01-1.95; P = 0.04). Despite the consistent positive associations between log-transformed UACR and cancer deaths observed in women, none of the results are significant (Table 3).

The results of the present study showed that albuminuria was associated with increased risk of cancer deaths from all-cause, lung, and prostate cancers in men from the United States ages 50 and older. The association was not significant for any type of cancer in women of the same age group.

One prior report from the Tromsø Study showed that UACR at baseline was correlated with subsequent cancer incidence. A total 590 out of 5,425 participants without diabetes or previous cancer had the first diagnosis of cancer in 10.3 years of follow-up. Each SD-higher rise in the log-transformed UACR was associated with a RR of 1.17 for cancer (P < 0.001). Participants with UACR in the highest quintile were 8.3- and 2.4-fold more likely to receive a diagnosis of bladder and lung cancer, respectively (6). Our study again examined the relationship between UACR and a variety of cancer mortalities, with similar results.

Despite the growing regard for albuminuria as a paraneoplastic and inflammatory phenomenon, the underlying mechanisms for the association between albuminuria and cancer incidence/mortalities are largely unknown (21-24). Chronic inflammation triggers endothelial dysfunction leading to albuminuria (25) and may provide a common link because inflammation is associated with both increased cancer (26) and cardiovascular risk (4). The renin-angiotensin system may participate in generating the increased cancer risk associated with albuminuria. Angiotensin II has been implicated in the development or invasion of several kinds of cancer (27). Angiotensin could exert mitogenic activity through angiotensin II type I (ATII-I) receptors and ATII-I receptor expression is higher in cancerous prostate (28).

The current data analysis showed a positive association between albuminuria and all-cancer mortality in men, mainly attributed to death from lung and prostate cancer. There was a dose-responsive relationship between albuminuria and prostate cancer death in men. The RR of prostate cancer deaths increased by 1.40-fold when the log-scale of albuminuria increased by 1. In stark contrast with male counterparts, there was no significant association between albuminuria with both all-cause and cause-specific cancer deaths in women. The moderating influence of sex could be attributed to prostate cancer in men and the differential risks of lung cancer between men and women. Although the prevalence of lung cancer in men has been decreasing, it has been increasing in women. The biology of lung cancer differs between the sexes, attributed in part to differences in growth factor receptors, endocrine influences, enzymatic clearance of toxins, the susceptibility to DNA damage, and the capacity for DNA repair, all of which may contribute to the observed differences in treatment outcomes (29). The finding in the current study of an association between albuminuria and lung cancer mortalities in men, but not in women, further shows a sex difference in the development and prognosis of lung cancer.

Albuminuria is a marker of many underlying diseases, including cardiovascular disease. In the study group, the estimated prevalence of pre-existing cardiovascular disease (positive response to the questions: “Doctor ever told you had a heart attack” or “Doctor told: congestive heart failure”) was 10.2% (12.8% for men and 8.0% for women). After further adjustment for cardiovascular disease, the associations between albuminuria and cancer deaths were generally reduced 5% to 10% and there was no positive association between pre-existing cardiovascular disease with cancer mortalities. Similarly, although serum cholesterol has been found to be positively associated with albuminuria (30), baseline cholesterol was not associated with cancer mortalities.

One of the major limitations of this study is that despite biological plausibility, the current evidence supports only an association, and not causation, between albuminuria and cancer mortality. In addition, the use of single albumin measurements may not adequately assess normal within-person variations and may bias our findings toward the null (31). The small sample size for many cancer case subpopulations limits the ability to determine whether the association between albuminuria and cancer is a general oncogenic effect or is attributed to one or more specific cancers. Thus, it is also difficult to develop a predictive model based on albuminuria unless other markers (e.g., prostate-specific antigen and serum calcium levels) are also incorporated (32).

In conclusion, the presence of albuminuria increases the risk of cancer death, especially all-cause, lung, and prostate cancer in men over 50 years of age. The association was independent of other conventional risk factors, such as smoking, diabetes, hypertension, and chronic kidney disease. The association was not apparent in women and the sex specificity of the observation clearly needs to be resolved. Further study is also needed to explore whether albuminuria contributes to prostate cancer or is merely a predictive marker of its subsequent incidence and/or progression. Understanding the underlying biological mechanisms in this regard will determine the prospect of using renal interventions in the management of prostate cancer.

The authors indicated no potential conflicts of interest.

The authors appreciate the statistical assistance of Dr. Gordon G. Brown from RTI International and Dr. Chia-Hsuin Chang from National Taiwan University Hospital.

Grant Support: The G62024 Interdisciplinary Research Grant from the University of North Texas Health Science Center at Fort Worth and by the grants from National Taiwan University Hospital Yun-Lin Branch (98.X002 and 99.X004).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.