Visual Abstract
Keywords: Cancer; chronic kidney disease; estimated glomerular filtration rate; detection bias; reverse causation; Incidence; glomerular filtration rate; Confidence Intervals; International Classification of Diseases; Follow-Up Studies; Renal Insufficiency, Chronic; Risk; Proportional Hazards Models; Urogenital Neoplasms; Bias; Hematologic Neoplasms; Prostatic Neoplasms
Abstract
Background and objectives
Community-based reports regarding eGFR and the risk of cancer are conflicting. We here explore plausible links between kidney function and cancer incidence in a large Scandinavian population–based cohort.
Design, setting, participants, & measurements
In the Stockholm Creatinine Measurements project, we quantified the associations of baseline eGFR with the incidence of cancer among 719,033 Swedes ages ≥40 years old with no prior history of cancer. Study outcomes were any type and site-specific cancer incidence rates on the basis of International Classification of Diseases-10 codes over a median follow-up of 5 years. To explore the possibility of detection bias and reverse causation, we divided the follow-up time into different time periods (≤12 and >12 months) and estimated risks for each of these intervals.
Results
In total, 64,319 cases of cancer (affecting 9% of participants) were detected throughout 3,338,226 person-years. The relationship between eGFR and cancer incidence was U shaped. Compared with eGFR of 90–104 ml/min, lower eGFR strata associated with higher cancer risk (adjusted hazard ratio, 1.08; 95% confidence interval, 1.05 to 1.11 for eGFR=30–59 ml/min and adjusted hazard ratio, 1.24; 95% confidence interval, 1.15 to 1.35 for eGFR<30 ml/min). Lower eGFR strata were significantly associated with higher risk of skin, urogenital, prostate, and hematologic cancers. Any cancer risk as well as skin (nonmelanoma) and urogenital cancer risks were significantly elevated throughout follow-up time, but they were higher in the first 12 months postregistration. Associations with hematologic and prostate cancers abrogated after the first 12 months of observation, suggesting the presence of detection bias and/or reverse causation.
Conclusions
There is a modestly higher cancer risk in individuals with mild to severe CKD driven primarily by skin and urogenital cancers, and this is only partially explained by bias.
Introduction
CKD is common, with a population prevalence of 5%–15% in most developed countries (1,2). Even mild reductions in kidney function, as depicted by eGFR, associate with a markedly higher risk of comorbidities and complications, such as cardiovascular disease (3), infections (4), anemia (5), fractures (6), and possibly, cancer.
Kidney dysfunction leads to retention of metabolic waste products and results in the disruption of multiple pathways, including disorders of the immune system (7,8), inducing inflammation (9), the activation of the renin-angiotensin system (10), and endothelial abnormalities (11), all of which have been suggested to increase cancer risk (12,13). Various studies have reported higher cancer incidence among patients with ESKD (14–16). For less severe CKD, however, evidence is less robust and at times, conflicting for both any type and site-specific cancer incidence (17–24). Differences among studies may be explained by study characteristics (e.g., differences in outcome ascertainment or length of follow-up) and study populations (e.g., age of participants). It can also be explained by biases that are not adequately addressed. Detection bias is possible, because patients with CKD are likely to be more medicalized and frequently monitored; hence, they are more prone to early cancer discovery. Reverse causality might also exist (particularly in administrative records) due to increased disease surveillance among patients with CKD and nonspecific systemic symptoms from a cancer prompting earlier detection of CKD.
In an attempt to clarify literature inconsistencies, we undertook a comprehensive analysis of the risk of cancer across the full spectrum of kidney function in a large region-representative population of Swedes ages 40 years old and above. We also explored whether these estimates are affected by reverse causation and detection bias.
Materials and Methods
Study Population
We used data from the Stockholm Creatinine Measurements project, a healthcare utilization cohort from the region of Stockholm, Sweden that included all residents undertaking serum creatinine tests during 2006–2011 (25). Laboratory data were linked with regional and national administrative databases for complete information on healthcare utilization, dispensed drugs, validated kidney replacement therapy end points, and follow-up for death, with virtually no loss to follow-up. For this study, we included all community-dwelling participants ages ≥40 years old at their first available outpatient creatinine measurement, which was considered the study baseline. Exclusion criteria were any history of cancer, undergoing kidney replacement therapy (as ascertained by linkage with the Swedish Renal Registry [http://www.medscinet.net/snr/]), or missing information on age or sex (Supplemental Figure 1). The regional ethics committee in Stockholm, Sweden approved the study.
Exposure and Covariates
We selected the first eligible creatinine measurements per patient. Eligible creatinine measurements were those performed in connection with an outpatient visit, with concentrations within the range of 0.5–17.0 mg/dl. All measurements were standardized to isotope dilution mass spectrometry standards. The study exposure was eGFR at cohort entry (baseline), which was calculated from plasma creatinine using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation (26). Data on ethnicity are not available in Sweden by law, and all participants were assumed to be white (27). Five categories of eGFR were studied: eGFR≥105, 90–104, 60–89, 30–59, and <30 ml/min per 1.73 m2. eGFR of 90–104 ml/min per 1.73 m2 served as the reference group, because this category showed the lowest incidence rate (IR) of cancer.
Covariates were defined at baseline, and they included age, sex, and comorbidities (cardiovascular disease [composite of myocardial infarction, congestive heart failure, peripheral vascular disease, and cerebrovascular disease], diabetes, chronic pulmonary disease, rheumatic disease, dementia, peptic ulcer disease, liver disease, hemiplegia or paraplegia, and chronic infections as well as the presence of hypertension). Definition of comorbidities was on the basis of International Classification of Diseases-10 (ICD-10) codes, and diabetes and hypertension were additionally enriched with information on current purchase of related medication (definitions are in Supplemental Table 1). According to study protocol, there are no missing variables in the study.
Follow-Up and Study Outcome
The primary study outcome was diagnosis of any cancer (any ICD-10 of the C chapter). The secondary study outcomes were the diagnosis of site-specific cancers: cancers of oral cavity and pharynx; esophagus and stomach; small intestine; colon and rectum; liver, gall bladder, and bile duct; pancreas; larynx; lung and bronchus; bone and articular cartilage; skin; soft tissue; breast; cervix; unspecified parts of the uterus; ovary; prostate; urogenital area (kidney, ureter, and bladder); central nervous system; thyroid; or unknown origin or hematologic cancers (non-Hodgkin lymphoma, multiple myeloma, and leukemia) (definitions are in Supplemental Table 2). Participants were censored at the end of follow-up (December 31, 2012), death, or migration from the region, whichever occurred first. Death date was obtained from the National Board of Health and Welfare’s Cause-of-Death Register (http://www.socialstyrelsen.se).
Data Analyses
Continuous variables were reported as mean ± SD, and categorical ones were reported as counts and proportions. Baseline characteristics were compared across different eGFR categories by Pearson chi-squared test for proportions and ANOVA for continuous variables. We calculated age- and sex-adjusted IRs with 95% confidence intervals (95% CIs) using the exact method. Multivariable Cox proportional hazard models estimated the association between eGFR categories and the risk of cancer. For site-specific cancer, we a priori decided to focus on the most common ones defined as those with an incidence of >0.5% in our population. Results are reported as hazard ratios (HRs) and 95% CIs. The models were adjusted for age (continuous), sex, and comorbidities (hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease, rheumatoid disease, dementia, peptic ulcer disease, liver disease, hemiplegia/paraplegia, and chronic infections). We repeated the main analyses in men and women and across age categories (40–49, 50–59, 60–69, and ≥70 years old). We also investigated associations using piecewise linear splines, with knots at fixed values of eGFR categories.
To explore the possibility of detection bias and reverse causality, we split the follow-up time into two different time periods (for 12 months and after 12 months) and estimated separate HRs for each of these intervals. Analyses were performed using R (https://www.r-project.org) and Stata, version 15.0 (StataCorp, College Station, TX).
Results
Baseline Characteristics
There were approximately 1.3 million Stockholm citizens accessing healthcare and undergoing creatinine testing during 2006–2011. After applying exclusion criteria (Supplemental Figure 1), the study cohort consisted of 719,033 participants free from cancer history and ages ≥40 years old; 53% were women, and the mean age was 60±14 years old (Table 1). Mean eGFR was 87±18 ml/min per 1.73 m2; 15% of participants had eGFR≥105 ml/min per 1.73 m2, 35% of participants had eGFR of 90–104 ml/min per 1.73 m2, 42% of participants had eGFR of 60–89 ml/min per 1.73 m2, 7% of participants had eGFR of 30–59 ml/min per 1.73 m2, and 1% of participants had eGFR<30 ml/min per 1.73 m2. The most common comorbidities were hypertension (34%) followed by cardiovascular disease (11%), diabetes mellitus (8%), chronic obstructive pulmonary disease (5%), and rheumatoid disease (2%). Although age and the prevalence of comorbidities were higher across worse eGFR categories, chronic infections were less commonly observed (Table 1).
Table 1.
Baseline characteristics of study participants ages ≥40 years old overall and by eGFR strata
| Variable | eGFR Strata, ml/min per 1.73 m2 | |||||
|---|---|---|---|---|---|---|
| Overall, n=719,033 | eGFR>105, n=108,349 | eGFR=90–104, n=252,462 | eGFR=60–89, n=300,939 | eGFR=30–59, n=52,651 | eGFR<30, n=4632 | |
| eGFR, mean ± SD | 87±18 | 111±5 | 97±4 | 78±8 | 50±8 | 23±6 |
| Serum Albumin, g/dl (64% missing) | 3.9±0.4 | 4.0±0.4 | 4.0±0.3 | 3.9±0.4 | 3.7±0.4 | 3.5±0.5 |
| Age, yr, mean ± SD | 60±14 | 46±6 | 56±9 | 65±13 | 78±11 | 80±12 |
| Women | 382,036 (53%) | 57,770 (53%) | 124,229 (49%) | 164,667 (55%) | 32,712 (62%) | 2658 (57%) |
| Comorbidities | ||||||
| Hypertension | 244,247 (34%) | 14,617 (14%) | 65,202 (26%) | 122,045 (41%) | 38,453 (73%) | 3930 (85%) |
| Cardiovascular disease | 79,989 (11%) | 2806 (3%) | 14,379 (6%) | 39,604 (13%) | 20,463 (39%) | 2737 (59%) |
| Myocardial infarction | 23,883 (3%) | 762 (0.7%) | 4594 (2%) | 11,453 (4%) | 6085 (12%) | 989 (21%) |
| Congestive heart failure | 30,705 (4%) | 601 (0.6%) | 3255 (1%) | 13,688 (5%) | 11,230 (21%) | 1931 (42%) |
| Peripheral vascular disease | 14,172 (2%) | 473 (0.4%) | 2357 (0.9%) | 6759 (2%) | 3985 (8%) | 598 (13%) |
| Cerebrovascular disease | 35,974 (5%) | 1353 (1%) | 6672 (3%) | 18,378 (6%) | 8522 (16%) | 1049 (23%) |
| Diabetes mellitus | 56,443 (8%) | 5819 (5%) | 16,400 (7%) | 24,247 (8%) | 8748 (17%) | 1229 (27%) |
| COPD | 38,955 (5%) | 4128 (4%) | 11,376 (5%) | 17,552 (6%) | 5306 (10%) | 593 (13%) |
| Rheumatoid disease | 14,400 (2%) | 1046 (1%) | 3302 (1%) | 6960 (2%) | 2812 (5%) | 280 (6%) |
| Dementia | 10,369 (1%) | 84 (0.1%) | 903 (0.4%) | 5829 (2%) | 3196 (6%) | 357 (8%) |
| Peptic ulcer disease | 10,407 (1%) | 1220 (1%) | 2770 (1%) | 4390 (2%) | 1749 (3%) | 278 (6%) |
| Liver disease | 11,317 (2%) | 2591 (2%) | 4364 (2%) | 3529 (1%) | 728 (1%) | 105 (2%) |
| Hemiplegia/paraplegia | 2565 (0.4%) | 581 (0.5%) | 816 (0.3%) | 876 (0.3%) | 253 (0.5%) | 39 (0.8%) |
| Chronic infections | 8165 (1%) | 2425 (2%) | 3221 (1%) | 2136 (0.7%) | 319 (0.6%) | 64 (1%) |
s-Albumin, Serum Albumin; COPD, chronic obstructive pulmonary disease.
IR of Cancer across eGFR Strata
Over a median follow-up of 5 years, 64,319 cases of cancer (affecting 9% of participants) were detected throughout 3,338,226 person-years. The overall age- and sex-adjusted IR was 14.4/1000 person-years. Slightly higher IRs were noted for participants with CKD stage 3 or higher (15.5/1000 person-years in eGFR of 30–59 ml/min per 1.73 m2 and 19.1/1000 person-years in eGFR<30 ml/min per 1.73 m2) (Table 2).
Table 2.
Adjusted incidence rate of cancer among participants (ages ≥40 years old) overall and site specific during 5 years of follow-up by eGFR strata
| Variable | eGFR Strata, ml/min per 1.73 m2 | |||||
|---|---|---|---|---|---|---|
| Overall, n=719,033 | eGFR>105, n=108,349 | eGFR=90–104, n=252,462 | eGFR=60–89, n=300,939 | eGFR=30–59, n=52,651 | eGFR<30, n=4632 | |
| Any cancer | ||||||
| Any cancer | 64,319 (9%) | 3446 (3%) | 17,506 (7%) | 34,149 (11%) | 8570 (16%) | 648 (14%) |
| Incidence rate (95% CI) per 1000 person-yr | 14.38 (14.23 to 14.53) | 15.33 (14.79 to 15.87) | 14.08 (13.84 to 14.32) | 14.19 (13.97 to 14.41) | 15.47 (15.06 to 15.89) | 19.08 (17.62 to 20.54) |
| Site-specific cancer | ||||||
| Skin | 19,988 (3%) | 764 (0.7%) | 4583 (2%) | 11,191 (4%) | 3204 (6%) | 246 (5%) |
| Incidence rate (95% CI) per 1000 person-yr | 4.04 (3.95 to 4.12) | 3.34 (3.09 to 3.59) | 3.88 (3.76 to 4.01) | 4.33 (4.21 to 4.46) | 4.64 (4.42 to 4.85) | 5.40 (4.70 to 6.10) |
| Colon and rectum | 6077 (0.8%) | 318 (0.3%) | 1575 (0.6%) | 3284 (1%) | 845 (2%) | 55 (1%) |
| Incidence rate (95% CI) per 1000 person-yr | 1.17 (1.12 to 1.22) | 1.57 (1.39 to 1.76) | 1.17 (1.10 to 1.24) | 1.07 (1.01 to 1.13) | 1.08 (0.98 to 1.17) | 1.13 (0.82 to 1.43) |
| Lung and bronchus | 5023 (0.7%) | 291 (0.3%) | 1573 (0.6%) | 2508 (0.8%) | 611 (1%) | 40 (0.9%) |
| Incidence rate (95% CI) per 1000 person-yr | 0.82 (0.78 to 0.87) | 1.62 (1.43 to 1.81) | 0.87 (0.81 to 0.94) | 0.63 (0.59 to 0.68) | 0.70 (0.63 to 0.78) | 0.81 (0.56 to 1.06) |
| Urogenital | 4271 (0.6%) | 182 (0.2%) | 1032 (0.4%) | 2294 (0.8%) | 705 (1%) | 58 (1%) |
| Incidence rate (95% CI) per 1000 person-yr | 0.74 (0.70 to 0.77) | 0.74 (0.63 to 0,85) | 0.67 (0.62 to 0.71) | 0.75 (0.70 to 0.80) | 1.05 (0.95 to 1.16) | 1.35 (0.99 to 1.70) |
| Hematologic | 4258 (0.6%) | 224 (0.2%) | 1163 (0.5%) | 2215 (0.7%) | 603 (1%) | 53 (1%) |
| Incidence rate (95% CI) per 1000 person-yr | 0.93 (0.90 to 0.97) | 0.87 (0.75 to 0.99) | 0.94 (0.88 to 0.99) | 0.92 (0.87 to 0.98) | 1.10 (0.99 to 1.22) | 1.56 (1.14 to 1.99) |
| Prostate (in men; n=336,997) | 11,427 (3%) | 309 (0.6%) | 3318 (3%) | 6434 (5%) | 1272 (6%) | 94 (5%) |
| Incidence rate (95% CI) per 1000 person-yr | 3.52 (3.35 to 3.69) | 3.12 (2.76 to 3.48) | 3.50 (3.30 to 3.70) | 3.69 (3.49 to 3.90) | 3.63 (3.34 to 3.92) | 4.01 (3.19 to 4.84) |
| Breast (in women; n=382,036) | 7341 (2%) | 675 (1%) | 2363 (2%) | 3603 (2%) | 666 (2%) | 34 (1%) |
| Incidence rate (95% CI) per 1000 person-yr | 3.74 (3.65 to 3.83) | 3.48 (3.18 to 3.77) | 3.76 (3.59 to 3.92) | 3.83 (3.68 to 3.97) | 3.72 (3.40 to 4.04) | 3.28 (2.17 to 4.39) |
Incidence rates are adjusted by age and sex. There were 64,319 unique participants who developed cancer, but the sum may exceed this figure due to multiple cancer diagnoses over time. 95% CI, 95% confidence interval.
The most common site-specific cancers were skin (IR=4.0/1000 person-years; 3% of all participants) followed by breast (IR=3.7/1000 person-years; 2% of all women), prostate (IR=3.5/1000 person-years; 3% of all men), colon-rectum (IR=1.2/1000 person-years; 0.8%), hematologic (IR=0.9/1000 person-years; 0.6%), lung (IR=0.8/1000 person-years; 0.7%), and urogenital (IR=0.7/1000 person-years; 0.6%) cancers. IRs of skin, prostate, hematologic, and urogenital cancers were higher across lower eGFR categories. Conversely, IRs of colon-rectum and lung cancer were higher in eGFR≥105 ml/min per 1.73 m2, and they were, in general, flat in other categories. IR of breast cancer was similar across all eGFR categories (Table 2). The proportion of other less common site-specific cancers showed, in general, a tendency to be higher across lower eGFR strata (Supplemental Table 3).
Cancer HRs across eGFR Strata
As a next step, we studied whether the higher IRs across eGFR strata were explained by accompanying comorbidities and characteristics via multivariable Cox adjusted regression, with eGFR=90–104 ml/min per 1.73 m2 as the referent category. After adjusting for identified confounders, a U-shaped association was observed between eGFR and the risk of cancer both when examining eGFR categories (Table 3) and when modeling eGFR as a continuous metric using splines (Figure 1). Compared with participants with eGFR=90–104 ml/min per 1.73 m2, those with eGFR≥105 ml/min per 1.73 m2 had a 9% higher cancer risk (adjusted HR, 1.09; 95% CI, 1.05 to 1.13). Participants with eGFR of 60–89 ml/min per 1.73 m2 did not have a significantly higher cancer risk, but those with eGFR=30–59 and <30 ml/min per 1.73 m2 had 8% (adjusted HR, 1.08; 95% CI, 1.05 to 1.11) and 24% (adjusted HR, 1.24; 95% CI, 1.15 to 1.35) higher risk of cancer, respectively (Table 3). In general, the associations were modest in magnitude.
Table 3.
Hazard ratios overall and for selected site-specific cancers by eGFR strata
| eGFR Strata, ml/min per 1.73 m2 | Total Follow-Up | Splitting Follow-Up Time | ||||||
|---|---|---|---|---|---|---|---|---|
| No. of Participants | No. of Events | HRa | 95% CI | 0–1 yr | >1 yr | |||
| HRa | 95% CI | HRa | 95% CI | |||||
| Any cancer | ||||||||
| <30 | 4632 | 648 | 1.24 | 1.15 to 1.35 | 1.40 | 1.23 to 1.59 | 1.18 | 1.06 to 1.31 |
| 30–59 | 52,651 | 8570 | 1.08 | 1.05 to 1.11 | 1.13 | 1.07 to 1.20 | 1.06 | 1.02 to 1.10 |
| 60–89 | 300,939 | 34,149 | 1.00 | 0.98 to 1.02 | 0.99 | 0.95 to 1.02 | 1.01 | 0.99 to 1.03 |
| 90–104 | 252,462 | 17,506 | Reference | Reference | Reference | |||
| >105 | 108,349 | 3446 | 1.09 | 1.05 to 1.13 | 1.15 | 1.07 to 1.24 | 1.06 | 1.01 to 1.12 |
| Skin cancer | ||||||||
| <30 | 4632 | 246 | 1.40 | 1.23 to 1.60 | 1.29 | 1.02 to 1.64 | 1.45 | 1.23 to 1.70 |
| 30–59 | 52,651 | 3204 | 1.19 | 1.13 to 1.26 | 1.23 | 1.10 to 1.38 | 1.18 | 1.11 to 1.25 |
| 60–89 | 300,939 | 11,191 | 1.11 | 1.07 to 1.15 | 1.12 | 1.03 to 1.22 | 1.11 | 1.06 to 1.16 |
| 90–104 | 252,462 | 4583 | Reference | Reference | Reference | |||
| >105 | 108,349 | 764 | 0.87 | 0.80 to 0.94 | 0.89 | 0.74 to 1.06 | 0.86 | 0.78 to 0.95 |
| Colon and rectum cancer | ||||||||
| <30 | 4632 | 55 | 0.93 | 0.70 to 1.22 | 0.97 | 0.60 to 1.56 | 0.94 | 0.67 to 1.32 |
| 30–59 | 52,651 | 845 | 0.93 | 0.84 to 1.03 | 0.96 | 0.80 to 1.15 | 0.92 | 0.81 to 1.04 |
| 60–89 | 300,939 | 3284 | 0.91 | 0.85 to 0.98 | 0.89 | 0.79 to 1.01 | 0.92 | 0.85 to 1.00 |
| 90–104 | 252,462 | 1575 | Reference | Reference | Reference | |||
| >105 | 108,349 | 318 | 1.34 | 1.17 to 1.53 | 1.59 | 1.27 to 1.99 | 1.23 | 1.03 to 1.45 |
| Lung cancer | ||||||||
| <30 | 4632 | 40 | 0.73 | 0.53 to 1.00 | 0.32 | 0.15 to 0.67 | 1.03 | 0.72 to 1.46 |
| 30–59 | 52,651 | 611 | 0.75 | 0.67 to 0.84 | 0.58 | 0.48 to 0.72 | 0.84 | 0.73 to 0.96 |
| 60–89 | 300,939 | 2508 | 0.73 | 0.68 to 0.78 | 0.62 | 0.55 to 0.70 | 0.79 | 0.72 to 0.86 |
| 90–104 | 252,462 | 1573 | Reference | Reference | Reference | |||
| >105 | 108,349 | 291 | 1.75 | 1.52 to 2.01 | 2.01 | 1.61 to 2.49 | 1.60 | 1.33 to 1.91 |
| Urogenital cancer | ||||||||
| <30 | 4632 | 58 | 1.91 | 1.45 to 2.51 | 3.33 | 2.18 to 5.09 | 1.47 | 1.03 to 2.12 |
| 30–59 | 52,651 | 705 | 1.56 | 1.39 to 1.75 | 2.27 | 1.85 to 2.80 | 1.32 | 1.15 to 1.52 |
| 60–89 | 300,939 | 2294 | 1.13 | 1.04 to 1.22 | 1.33 | 1.14 to 1.54 | 1.04 | 0.95 to 1.15 |
| 90–104 | 252,462 | 1032 | Reference | Reference | Reference | |||
| >105 | 108,349 | 182 | 1.11 | 0.93 to 1.32 | 1.04 | 0.75 to 1.42 | 1.15 | 0.93 to 1.41 |
| Hematologic malignancy | ||||||||
| <30 | 4632 | 53 | 1.71 | 1.28 to 2.28 | 3.68 | 2.49 to 5.43 | 0.98 | 0.63 to 1.52 |
| 30–59 | 52,651 | 603 | 1.23 | 1.09 to 1.38 | 1.79 | 1.45 to 2.20 | 1.03 | 0.89 to 1.19 |
| 60–89 | 300,939 | 2215 | 0.99 | 0.91 to 1.07 | 1.18 | 1.02 to 1.36 | 0.91 | 0.82 to 1.00 |
| 90–104 | 252,462 | 1163 | Reference | Reference | Reference | |||
| >105 | 108,349 | 224 | 0.93 | 0.79 to 1.09 | 0.95 | 0.73 to 1.23 | 0.94 | 0.77 to 1.15 |
| Prostate cancer (in men) | ||||||||
| <30 | 1974 | 94 | 1.24 | 1.01 to 1.53 | 1.96 | 1.47 to 2.61 | 0.87 | 0.64 to 1.19 |
| 30–59 | 19,939 | 1272 | 1.12 | 1.04 to 1.21 | 1.30 | 1.14 to 1.48 | 1.04 | 0.95 to 1.15 |
| 60–89 | 136,272 | 6434 | 1.06 | 1.01 to 1.11 | 1.01 | 0.93 to 1.09 | 1.08 | 1.03 to 1.15 |
| 90–104 | 128,233 | 3318 | Reference | Reference | Reference | |||
| >105 | 50,579 | 309 | 0.92 | 0.81 to 1.04 | 0.85 | 0.67 to 1.08 | 0.94 | 0.82 to 1.09 |
| Breast cancer (in women) | ||||||||
| <30 | 2658 | 34 | 0.90 | 0.64 to 1.27 | 0.71 | 0.35 to 1.45 | 0.98 | 0.66 to 1.45 |
| 30–59 | 32,712 | 666 | 1.00 | 0.91 to 1.11 | 0.99 | 0.80 to 1.22 | 1.01 | 0.90 to 1.13 |
| 60–89 | 164,667 | 3603 | 1.02 | 0.97 to 1.08 | 1.01 | 0.90 to 1.14 | 1.03 | 0.96 to 1.09 |
| 90–104 | 124,229 | 2363 | Reference | Reference | Reference | |||
| >105 | 57,770 | 675 | 0.93 | 0.84 to 1.02 | 0.92 | 0.75 to 1.12 | 0.93 | 0.84 to 1.04 |
HR, hazard ratio; 95% CI, 95% confidence interval.
Model adjusted for age, sex, and comorbidity (hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease, rheumatoid disease, dementia, peptic ulcer disease, liver disease, hemiplegia/paraplegia, and chronic infections).
Figure 1.
U-shape association between eGFR (continuous) and overall cancer risk. Shown are linear splines adjusted for age, sex, and comorbidity (hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease, rheumatoid disease, dementia, peptic ulcer disease, liver disease, hemiplegia/paraplegia, and chronic infections). Data are reported as hazard ratios and 95% confidence intervals (shaded in gray). The reference is set at eGFR of 90 ml/min per 1.73 m2. Histograms represent the eGFR distribution.
HRs for site-specific cancers are shown in Table 3. Compared with participants with eGFR of 90–104 ml/min per 1.73 m2, those with eGFR≥105 ml/min per 1.73 m2 had a higher risk of colon-rectum and lung cancers. Across lower eGFR categories, we observed gradually higher risks of skin (especially nonmelanoma skin cancer), urogenital, and hematologic cancers as well as prostate cancer in men, particularly among participants with eGFR<30 ml/min per 1.73 m2. No association was observed between eGFR and the risk of breast cancer in women.
Sex- and age-stratified analyses are shown in Supplemental Tables 4 and 5. Although the risk of cancer was similar among men and women and a higher in magnitude cancer risk was noted for the younger age category of 40–49 years old, no statistically significant interactions were observed across these strata (all P interaction values >0.05).
Cancer HRs across eGFR Strata within and beyond 1 Year of Follow-Up
To assess potential reverse causation and detection biases, we calculated the HRs over two distinct follow-up periods. Compared with eGFR of 90–104 ml/min per 1.73 m2, there was a clear elevation in the 12-month risk for any cancer as well as skin, urogenital, hematologic, and prostate cancers across lower eGFR strata (Table 3). Over time (>12 months of follow-up), excess risks were lower but remained significantly higher than the reference category for any cancer, skin (nonmelanoma) cancer (Supplemental Table 6), and urogenital cancer. Conversely, no association was observed after 12 months from registration concerning hematologic and prostate cancer risk, suggesting reverse causation/detection bias (Supplemental Figure 2).
Discussion
In a large region-representative cohort of Swedes with ages ≥40 years old followed for 5 years, we report that individuals with reduced eGFR (eGFR<60 ml/min per 1.73 m2) experienced a modestly higher risk of cancer. This higher risk was driven primarily by skin and urogenital cancers. Detection bias and/or reverse causation led to an overestimation of the excess risk of cancer in patients with CKD over the whole follow-up time.
The observed association between reduced kidney function and cancer incidence in our study is consistent with data from the two largest studies to date (17,18). Our estimates of overall cancer incidence within each eGFR stratum are, in general, comparable. There are differences, however, regarding the lowest cancer IR observed (eGFR=90–104 ml/min per 1.73 m2 in our study versus eGFR=60–89 ml/min per 1.73 m2 in the United States [18] and eGFR=45–59 ml/min per 1.73 m2 in Korea [17]), which may be explained by how eGFR categories were defined and the nature of the study sampling. Our study contains data from people accessing healthcare, and those in the high-eGFR range likely represent inaccurate estimation of true GFR in situations with low serum creatinine generation that accompany conditions of reduced muscle mass and/or chronic illness. For this reason, we made the a priori decision to model a separate eGFR category of individuals with eGFR>105 ml/min per 1.73 m2 who, as expanded below, proved to have a significantly higher cancer IR than individuals with eGFR>90–104 ml/min per 1.73 m2. The United States study (18) was a similar healthcare extraction of middle-aged individuals seeking care at Kaiser Permanente. The authors reported a J-shaped association between eGFR and cancer risk (18), but they chose the category of eGFR>90–150 ml/min per 1.73 m2 collectively as the reference. This lumping possibly resulted in the counterintuitive observation that the lowest cancer IR was observed for the category of eGFR=60–89 ml/min per 1.73 m2. Finally, the Korean study was a population-screening sample that selected eGFR>90 ml/min per 1.73 m2 as the reference category (17). The expected lower representation of “diseased” individuals in this cohort may have possibly been linked to a low recruitment of individuals in the high-eGFR range. Because of these differences in study designs and selection of eGFR references, the effect HRs between studies are not directly comparable, but collectively, they provide evidence of a higher (but of modest magnitude) cancer risk with lower kidney function. Other studies have reported no or weak associations between eGFR and cancer, but we note that these studies included participants of all ages (20), participants older than 65 years old (19), or participants with diabetes and CKD (23). Several plausible mechanisms have been proposed to explain the link between reduced kidney function and cancer risk. First, the inflammatory pro-oxidant microenvironment, which often coexists in patients with reduced GFR (9,28), is thought to contribute to cancer development (29). Second, patients with CKD have immune dysfunction (7,8), alterations of the renin-angiotensin system (10), and endothelial abnormalities (11), all of which have also been suggested to increase cancer risk (12,13). Third, some commonly prescribed medications in these individuals, such as antihypertensives, calcium channel blockers (30), and statins (31), have been associated with higher cancer risk. Fourth, uremic toxicity, such as accumulation of indoxyl sulfate, p-cresyl sulfate, or nitric oxide, may additionally create a state of impaired DNA repair, nutritional deficiency, and accumulation of carcinogens (32–34).
An important addition of our study to current evidence is the comprehensive characterization of site-specific cancer risks in relation to eGFR. Furthermore, few studies to date have addressed the temporal relations between eGFR and cancer risk to shed light on reverse causation/detection bias risk. In this context, we noted two patterns of risk when modeling HRs over time. First, we observed that the HR fell over time but remained elevated beyond 1 year from registration for any cancer as well as skin and urogenital cancers, suggesting that reverse causality only explains some of the relationship and reinforcing the notion that low eGFR may be a genuine risk factor for these site-specific cancers. We acknowledge, however, that 1-year follow-up split may still not be enough to assess detection bias; there could still be ascertainment bias due to ongoing follow-up and surveillance of detected masses (ultrasound), hematuria, etc. In any case, these observations confirm and expand previous reports addressing urogenital cancer risk in patients with CKD (17–19). The observation that skin cancer risk (and particularly, nonmelanoma skin cancer risk) is elevated in persons with mild/moderate kidney disease is, however, novel. Although it expands earlier reports in patients undergoing kidney replacement therapy (35–39), possible mechanisms have not been fully elucidated. Nonmelanoma skin cancer in ESKD has been related to chronic inflammation, use of immunosuppressive medication, and uremic pruritus (36,39,40). Furthermore, ultraviolet light exposure is also a known risk factor, especially among whites, which is thought to partly explain the observed higher skin cancer incidence during the last decades in Northern Hemisphere countries, such as those from Scandinavia (41,42).
Second, although our main analysis described an association between eGFR stages and the risk of hematologic and prostate cancers, HRs with splitting follow-up time evidenced this association to exist only during the first-year follow-up, suggesting that the excess risk observed for the whole time period was possibly affected by detection/reverse causation bias. Consistent with this, patients with multiple myeloma release Bence–Jones proteins, the kidney filtration of which is postulated as an important cause of AKI in these patients (43); likewise, kidney failure has been recognized in the initial presentation of myeloma in 50% of patients and reported to appear within 2 months of the diagnosis of myeloma in 75% of patients (44). Finally, although associations of low eGFR with colon cancer, breast cancer (22), and lung cancer (19) have been reported in some studies, these were not confirmed in either our study or larger study samples to date (17,18).
It is interesting that the high end of eGFR in our population (≥105 ml/min per 1.73 m2) presented with higher cancer risk. This same pattern was reported in two previous population-based studies (17,18). We would like to attribute this to inaccuracies of eGFR in diseased populations, but our general description in Table 1 does not reveal striking differences in this stratum other than presenting with younger age and a higher prevalence of chronic infections. Acknowledging that we do not know the reasons behind the outpatient creatinine test that prompted the inclusion in our study, we note that it has been reported that participants with chronic infections have poorer nutritional status, lower immune response, and subsequently, higher risk of cancer (45). Furthermore, it can also be noted that patients with lung cancer and patients with colon-rectum cancer often exhibit features of body weight loss and cachexia (46), resulting in low serum creatinine and hence, a high eGFR. Although we fully acknowledge the lack of clinical information, such as body mass index or weight loss, over time in our study, the important reduction (by 60%) of this risk association after the first year of follow-up in this eGFR category may suggest reverse causation bias and indirectly support our speculations.
Our study has some strengths, such as the region representativeness, the comprehensive characterization of cancer subtypes, and our sensitivity analyses assessing detection and reverse causation bias. Our study has, however, some limitations. Whereas our data are collected in real world healthcare institutions, selected participants may differ from participants without creatinine measurements. Nonetheless, this is unlikely to invalidate our findings given the large population representativeness. We considered one eGFR measure as our exposure, which is subject to measurement variability. Our median follow-up time of 5 years could be considered short when evaluating an outcome like cancer diagnosis, but given our population coverage, we capture a number of events proportional to the reported cancer incidence in Stockholm and Sweden in general for that period (41). It can be also argued that, with a longer follow-up, we would infer misclassification of the exposure (that is, eGFR would change over time). Although the validity of ICD-10 reporting in Swedish inpatient registries is high for many diagnoses, including cancer (47), we would have benefited by confirming our diagnoses with linkage to the Swedish Cancer Register, which contains additional characterization of the cancer event by the pathologists. Nonetheless, we find it unlikely that our results would change, because we focus solely on cancer incidence risk. Some important confounders, such as tobacco use, alcohol use, physical activity, dietary habits, and blood/urine tests, were not available. In Sweden, information on race is not permitted in registries, and we assume in this study that the majority of the population is white. Thus, extrapolation of these findings to other ethnically/racially diverse populations should be done with caution.
In conclusion, this region-representative healthcare study finds a modestly higher cancer risk in individuals with CKD. Detection bias and reverse causality may partly explain the higher risk of cancer during the first year of follow-up, but they less likely explain higher risks in the long term, particularly for all cancers as well as cancers of the skin and the urogenital system. From a research point of view, we suggest future studies on the topic to take into account these biases when studying healthcare data. As a practical application, our findings may help healthcare policy makers to develop and implement appropriate strategies for cancer screening and monitoring in the context of CKD as well as help health service planning.
Disclosures
K.M. reports grants and personal fees from Kyowa Hakko Kirin and personal fees from Akebia outside the submitted work. J.A. reports personal fees from lecturing for Astrazeneca outside the submitted work. B.L. reports grants, personal fees, nonfinancial support, and other support from Baxter Healthcare Corporation during the conduct of the study and grants, personal fees, nonfinancial support, and other support from Baxter Healthcare Corporation outside the submitted work. H.X., G.S., M.T., P.B., C.-G.E., M.L., and J.-J.C. have nothing to disclose.
Supplementary Material
Acknowledgments
The Robert Lundberg Memorial Foundation supported H.X. in performing this study. We acknowledge support from the Stockholm County Council, the Swedish Heart and Lung Foundation, and Martin Rind’s and Westman’s Foundations. Finally, Baxter Healthcare Corporation supports Baxter Novum at Karolinska Institutet.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
Supplemental Material
This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.10820918/-/DCSupplemental.
Supplemental Figure 1. Flow chart of study participants.
Supplemental Figure 2. Hazard ratios for any type and selected site-specific cancers by eGFR strata using a time-varying follow-up.
Supplemental Table 1. List of ICD-10 codes for comorbidity definitions.
Supplemental Table 2. List of ICD-10 codes of cancer diagnosis.
Supplemental Table 3. Diagnosis of site-specific cancer registered during 5 years of follow-up overall and within eGFR strata.
Supplemental Table 4. Subgroup analysis for the most common site-specific cancers by sex.
Supplemental Table 5. Subgroup analysis for the most common site-specific cancers by age categories.
Supplement Table 6. Hazard ratios overall and for skin cancers by eGFR strata.
References
- 1.Eckardt KU, Coresh J, Devuyst O, Johnson RJ, Köttgen A, Levey AS, Levin A: Evolving importance of kidney disease: From subspecialty to global health burden. Lancet 382: 158–169, 2013 [DOI] [PubMed] [Google Scholar]
- 2.Gasparini A, Evans M, Coresh J, Grams ME, Norin O, Qureshi AR, Runesson B, Barany P, Ärnlöv J, Jernberg T, Wettermark B, Elinder CG, Carrero JJ: Prevalence and recognition of chronic kidney disease in Stockholm healthcare. Nephrol Dial Transplant 31: 2086–2094, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gansevoort RT, Correa-Rotter R, Hemmelgarn BR, Jafar TH, Heerspink HJ, Mann JF, Matsushita K, Wen CP: Chronic kidney disease and cardiovascular risk: Epidemiology, mechanisms, and prevention. Lancet 382: 339–352, 2013 [DOI] [PubMed] [Google Scholar]
- 4.Xu H, Gasparini A, Ishigami J, Mzayen K, Su G, Barany P, Ärnlöv J, Lindholm B, Elinder CG, Matsushita K, Carrero JJ: eGFR and the risk of community-acquired infections. Clin J Am Soc Nephrol 12: 1399–1408, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Astor BC, Muntner P, Levin A, Eustace JA, Coresh J: Association of kidney function with anemia: The third national health and nutrition examination survey (1988-1994). Arch Intern Med 162: 1401–1408, 2002 [DOI] [PubMed] [Google Scholar]
- 6.Daya N, Voskertchian A, Schneider ALC, Ballew S, McAdams DeMarco M, Coresh J, Appel LJ, Selvin E, Grams ME: Kidney function and fracture risk: The Atherosclerosis Risk in Communities (ARIC) study. Am J Kidney Dis 67: 218–226, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Eleftheriadis T, Antoniadi G, Liakopoulos V, Kartsios C, Stefanidis I: Disturbances of acquired immunity in hemodialysis patients. Semin Dial 20: 440–451, 2007 [DOI] [PubMed] [Google Scholar]
- 8.Kato S, Chmielewski M, Honda H, Pecoits-Filho R, Matsuo S, Yuzawa Y, Tranaeus A, Stenvinkel P, Lindholm B: Aspects of immune dysfunction in end-stage renal disease. Clin J Am Soc Nephrol 3: 1526–1533, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Carrero JJ, Stenvinkel P: Inflammation in end-stage renal disease--what have we learned in 10 years? Semin Dial 23: 498–509, 2010 [DOI] [PubMed] [Google Scholar]
- 10.Viazzi F, Bonino B, Cappadona F, Pontremoli R: Renin-angiotensin-aldosterone system blockade in chronic kidney disease: Current strategies and a look ahead. Intern Emerg Med 11: 627–635, 2016 [DOI] [PubMed] [Google Scholar]
- 11.Tomiyama H, Yamashina A: Clinical Considerations for the association between vascular damage and chronic kidney disease. Pulse (Basel) 2: 81–94, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wegman-Ostrosky T, Soto-Reyes E, Vidal-Millán S, Sánchez-Corona J: The renin-angiotensin system meets the hallmarks of cancer. J Renin Angiotensin Aldosterone Syst 16: 227–233, 2015 [DOI] [PubMed] [Google Scholar]
- 13.Franses JW, Drosu NC, Gibson WJ, Chitalia VC, Edelman ER: Dysfunctional endothelial cells directly stimulate cancer inflammation and metastasis. Int J Cancer 133: 1334–1344, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Matas AJ, Simmons RL, Kjellstrand CM, Buselmeier TJ, Najarian JS: Increased incidence of malignancy during chronic renal failure. Lancet 1: 883–886, 1975 [DOI] [PubMed] [Google Scholar]
- 15.Birkeland SA, Storm HH, Lamm LU, Barlow L, Blohmé I, Forsberg B, Eklund B, Fjeldborg O, Friedberg M, Frödin L, Glattre E, Halvorsen S, Holm NV, Jakobsen A, Jorgensen HE, Ladefoged J, Lindholm T, Lundgren G, Pukkala E: Cancer risk after renal transplantation in the Nordic countries, 1964-1986. Int J Cancer 60: 183–189, 1995 [DOI] [PubMed] [Google Scholar]
- 16.Vajdic CM, McDonald SP, McCredie MR, van Leeuwen MT, Stewart JH, Law M, Chapman JR, Webster AC, Kaldor JM, Grulich AE: Cancer incidence before and after kidney transplantation. JAMA 296: 2823–2831, 2006 [DOI] [PubMed] [Google Scholar]
- 17.Mok Y, Matsushita K, Ballew SH, Sang Y, Jung KJ, Lee S, Jee SH, Coresh J: Kidney function, proteinuria, and cancer incidence: The Korean heart study. Am J Kidney Dis 70: 512–521, 2017 [DOI] [PubMed] [Google Scholar]
- 18.Lowrance WT, Ordoñez J, Udaltsova N, Russo P, Go AS: CKD and the risk of incident cancer. J Am Soc Nephrol 25: 2327–2334, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wong G, Hayen A, Chapman JR, Webster AC, Wang JJ, Mitchell P, Craig JC: Association of CKD and cancer risk in older people. J Am Soc Nephrol 20: 1341–1350, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Christensson A, Savage C, Sjoberg DD, Cronin AM, O’Brien MF, Lowrance W, Nilsson PM, Vickers AJ, Russo P, Lilja H: Association of cancer with moderately impaired renal function at baseline in a large, representative, population-based cohort followed for up to 30 years. Int J Cancer 133: 1452–1458, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Weng PH, Hung KY, Huang HL, Chen JH, Sung PK, Huang KC: Cancer-specific mortality in chronic kidney disease: Longitudinal follow-up of a large cohort. Clin J Am Soc Nephrol 6: 1121–1128, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Iff S, Craig JC, Turner R, Chapman JR, Wang JJ, Mitchell P, Wong G: Reduced estimated GFR and cancer mortality. Am J Kidney Dis 63: 23–30, 2014 [DOI] [PubMed] [Google Scholar]
- 23.Wong G, Zoungas S, Lo S, Chalmers J, Cass A, Neal B, Woodward M, Perkovic V, Glasziou P, Williams B, Howard K, Chapman JR, Craig JC: The risk of cancer in people with diabetes and chronic kidney disease. Nephrol Dial Transplant 27: 3337–3344, 2012 [DOI] [PubMed] [Google Scholar]
- 24.Shebl FM, Warren JL, Eggers PW, Engels EA: Cancer risk among elderly persons with end-stage renal disease: A population-based case-control study. BMC Nephrol 13: 65, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Runesson B, Gasparini A, Qureshi AR, Norin O, Evans M, Barany P, Wettermark B, Elinder CG, Carrero JJ: The Stockholm CREAtinine Measurements (SCREAM) project: Protocol overview and regional representativeness. Clin Kidney J 9: 119–127, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) : A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Gasparini A, Evans M, Coresh J, Grams ME, Norin O, Qureshi AR, Runesson B, Barany P, Ärnlöv J, Jernberg T, Wettermark B, Elinder CG, Carrero JJ: Prevalence and recognition of chronic kidney disease in Stockholm healthcare. Nephrol Dial Transplant 31: 2086–2094, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Akchurin OM, Kaskel F: Update on inflammation in chronic kidney disease. Blood Purif 39: 84–92, 2015 [DOI] [PubMed] [Google Scholar]
- 29.Fernandes JV, Cobucci RN, Jatobá CA, Fernandes TA, de Azevedo JW, de Araújo JM: The role of the mediators of inflammation in cancer development. Pathol Oncol Res 21: 527–534, 2015 [DOI] [PubMed] [Google Scholar]
- 30.Bangalore S, Kumar S, Kjeldsen SE, Makani H, Grossman E, Wetterslev J, Gupta AK, Sever PS, Gluud C, Messerli FH: Antihypertensive drugs and risk of cancer: Network meta-analyses and trial sequential analyses of 324,168 participants from randomised trials. Lancet Oncol 12: 65–82, 2011 [DOI] [PubMed] [Google Scholar]
- 31.Liu W, Choueiri TK, Cho E: Statin use and the risk of renal cell carcinoma in 2 prospective US cohorts. Cancer 118: 797–803, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Menck CF, Munford V: DNA repair diseases: What do they tell us about cancer and aging? Genet Mol Biol 37[Suppl]: 220–233, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Sun CY, Chang SC, Wu MS: Suppression of Klotho expression by protein-bound uremic toxins is associated with increased DNA methyltransferase expression and DNA hypermethylation. Kidney Int 81: 640–650, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Cheng H, Wang L, Mollica M, Re AT, Wu S, Zuo L: Nitric oxide in cancer metastasis. Cancer Lett 353: 1–7, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Verneuil L, Varna M, Ratajczak P, Leboeuf C, Plassa LF, Elbouchtaoui M, Schneider P, Sandid W, Lebbé C, Peraldi MN, Sigaux F, de Thé H, Janin A: Human skin carcinoma arising from kidney transplant-derived tumor cells. J Clin Invest 123: 3797–3801, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Yanik EL: Skin cancer in the end-stage renal disease population: Unique risk factors for patients on dialysis. Br J Dermatol 175: 1136–1137, 2016 [DOI] [PubMed] [Google Scholar]
- 37.Jensen AO, Svaerke C, Farkas D, Pedersen L, Kragballe K, Sørensen HT: Skin cancer risk among solid organ recipients: A nationwide cohort study in Denmark. Acta Derm Venereol 90: 474–479, 2010 [DOI] [PubMed] [Google Scholar]
- 38.Bretagnol A, Halimi JM, Roland M, Barbet C, Machet L, Al Najjar A, Marlière JF, Badin J, Nivet H, Lebranchu Y, Büchler M: Autosomal dominant polycystic kidney disease: Risk factor for nonmelanoma skin cancer following kidney transplantation. Transpl Int 23: 878–886, 2010 [DOI] [PubMed] [Google Scholar]
- 39.Wang CC, Tang CH, Huang SY, Huang KC, Sue YM: Risk of non-melanoma skin cancer in patients with chronic kidney disease and its relationship to uraemic pruritus. Acta Derm Venereol 97: 1230–1234, 2017 [DOI] [PubMed] [Google Scholar]
- 40.Rangwala S, Tsai KY: Roles of the immune system in skin cancer. Br J Dermatol 165: 953–965, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Åberg A: Statistics – Health and Medical Care: Cancer Incidence in Sweden 2011. December: 22–26, 2012. 21729024 [Google Scholar]
- 42.World Health Organization: World Cancer Report 2014. International Agency for Research on Cancer, Geneva, Switzerland [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Haynes RJ, Read S, Collins GP, Darby SC, Winearls CG: Presentation and survival of patients with severe acute kidney injury and multiple myeloma: A 20-year experience from a single centre. Nephrol Dial Transplant 25: 419–426, 2010 [DOI] [PubMed] [Google Scholar]
- 44.Irish AB, Winearls CG, Littlewood T: Presentation and survival of patients with severe renal failure and myeloma. QJM 90: 773–780, 1997 [DOI] [PubMed] [Google Scholar]
- 45.Fardet A, Druesne-Pecollo N, Touvier M, Latino-Martel P: Do alcoholic beverages, obesity and other nutritional factors modify the risk of familial colorectal cancer? A systematic review. Crit Rev Oncol Hematol 119: 94–112, 2017 [DOI] [PubMed] [Google Scholar]
- 46.Tisdale MJ: Cachexia in cancer patients. Nat Rev Cancer 2: 862–871, 2002 [DOI] [PubMed] [Google Scholar]
- 47.Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL, Reuterwall C, Heurgren M, Olausson PO: External review and validation of the Swedish national inpatient register. BMC Public Health 11: 450, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.