Cigarette smoking has been recognized as a causal risk factor with moderate effect on RCC. As early in the 1960s, one hospital-based case-control study has linked tobacco use to RCC risk for both genders [6]. The association was significant in all types of tobacco users. Since then, various studies have confirmed the association in both case-control and prospective cohort studies. A large meta-analysis of 5 cohort and 19 case-control studies reported a 50 % increase in RCC risk among male smokers and a 20 % increase among female smokers. A dose–response relationship between the number of cigarettes smoked and RCC risk was reported for both genders. Smoking cessation should be promoted because an approximate 15–30 % reduced risk was observed in both men and women who stopped smoking for more than 10 years [7]. The attributable risk percentage (AR%) of smoking calculated from population-based case-control and cohort studies is comparable (21 vs. 23 %), which indicates that more than one fifth of RCC could be prevented in the general population if they had no exposure to smoking. Although smoking is primarily a lifestyle factor in ever smokers, it is an environmental exposure to never smokers. One study showed that patients who were never smokers were more likely to report exposure to environmental tobacco smoking (ETS) either at home or in public [8]. However, interpretation of these results requires caution due to the likelihood of recall bias and the relatively small sample size.

A number of potential mechanisms linking smoking to the development of RCC have been proposed [1]. Cigarette smoke contains more than 60 carcinogens and most of them are formed during combustion [9]. One crucial mechanism of linking tobacco carcinogens with cancer is through the formation of DNA adducts. Benzo[α]pyrene (BaP), a polycyclic aromatic hydrocarbon (PAH), was the first carcinogen found in cigarette smoke, and its carcinogenicity has been confirmed in animals and humans. However, the level of BaP is much lower in cigarette products nowadays, and its carcinogenicity is weaker compared to some other PAH compounds [10]. Considerable evidence has linked cigarette-smoke nitrosamines such as 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), N′-nitrosonornicotine (NNN), and aromatic amines such as 4-aminobiphenyl to several cancer types. However, little supporting evidence has been gained from studies of kidney cancer. Only two studies have been conducted, and these data suggest that peripheral lymphocytes derived from patients with RCC are more sensitive to NNK or benzo[α]pyrene diolepoxide (BPDE) than those derived from healthy controls [11, 12]. In addition, smoking-induced chronic hypoxia of renal tissue caused by carbon monoxide exposure, tubulotoxicity, increased oxidative stress, and endothelial cell dysfunction are all hypothesized to increase RCC risk [1].

Due to the great efforts in the campaign to target smoking cessation and changes of policy in public health, the overall prevalence of cigarette smoking has declined over the past several decades in the United States. By contrast, incidence rates of kidney cancer have steadily increased at more than 2 % per year since the 1970s. Thus, it is presumed that smoking is unlikely to be a major contributor to the rising RCC incidence trends observed in this country.

The prevalence of hypertension in US adults has remained quite stable since the 1990s [13]. Over this period, there has been considerable improvement in awareness and control of the disease, and this may translate into benefits of reduced comorbidity for hypertensive patients in the future. Hypertension is a confirmed risk factor for development of RCC. Although hypertension is highly correlated with obesity, sufficient data supports its independent effect. A cohort study found that elevated systolic and diastolic blood pressure may predispose to development of RCC in a clear dose–response manner [14, 15]. A meta-analysis of 18 studies showed a significant 1.6-fold increased risk of RCC associated with hypertension [16]. It is difficult to isolate the contribution of antihypertensive medication use from history of hypertension because of the solid correlation between the two. Many observational studies have reported the hazardous impact on kidney cancer risk by taking antihypertensive medicine [16–18]. However, results from other studies indicated that instead of medication use, the excess risk is attributable to history of hypertension. For example, both systolic and diastolic blood pressure were associated with elevated risk of RCC when the analyses were restricted to non-antihypertensive medication users [19]. Clinical trials also showed no effect of taking antihypertensive medicine on cancer incidence [20]. The biological mechanisms that relate hypertension to RCC are still not fully understood. The renin–angiotensin system (RAS) plays an essential role in regulation of blood pressure and convincing evidence indicates that angiotensin II (Ang II) is involved in processes of cell proliferation, migration, angiogenesis, and inflammation [21]. Interestingly, slower development of tumor was observed in a xenograft mouse model of RCC when treated with captopril, an angiotensin-converting enzyme (ACE) inhibitor [22]. Furthermore, it is hypothesized that renal injury and metabolic or functional changes resulting from hypertension-induced chronic renal hypoxia and lipid peroxidation may subsequently raise renal susceptibility to carcinogens.

Type II diabetes mellitus (DM) could be a risk factor for RCC independent of obesity/BMI. A meta-analysis that adjusted for obesity/BMI, alcohol consumption, and smoking has shown that a history of type II DM significantly confers higher predisposition of RCC [60]. However, only a few of the individual studies were further adjusted for history of hypertension that residual confounding may distort the association. In women, parity increases disease risk when compared to nulliparous, and a dose–response effect for parity number was significant in a meta-analysis [61]. Inflammation and elevated oxidative stress through pregnancy-induced physiologic changes, pregnancy-associated weight gain, and high levels of circulating estrogens have all been hypothesized as underlying mechanisms. There are other reported risk factors with inconclusive associations, such as end-stage renal disease, long-term hemodialysis, acquired renal cystic disease, use of statin and aspirin, and occupational and other environmental exposures [1, 62].

Compelling evidence supports genetic susceptibility to RCC. The risk of RCC is two to three times higher in individuals who have first-degree relatives with kidney cancer [63]. In addition, familial aggregation is seen in approximately 3 % of kidney cancer patients with inherited kidney cancer syndromes. Of these, the von Hippel–Lindau syndrome (VHL) is by far the most commonly recognized familial cancer syndrome associated with kidney cancer [64]. Other major familial kidney cancer syndromes include hereditary papillary renal cell carcinoma (HPRC), hereditary leiomyomatosis renal cell carcinoma (HLRCC), and Birt–Hogg–Dubé syndrome (BHD) [65], which are caused by mutations in the c-Met proto-oncogene, FH (fumarate hydratase), and FLCN (folliculin), respectively. Interestingly, VHL, FH, c-Met, and FLCN are all involved in the pathways that respond to nutrient stimulation and cell metabolism, indicating kidney cancer is a metabolic disease [66].

Most of the early candidate gene studies involved small numbers of cases and controls, and very few of the initially reported positive susceptibility alleles have been replicated in subsequent validation studies [67]. With regard to RCC, candidate gene studies reported many positive associations with single nucleotide polymorphisms (SNPs) in genes involved in xenobiotic metabolism, DNA repair, cell growth/apoptosis, inflammation, and other pathways [1]. None of the previously reported candidate SNPs has been replicated in large independent studies.

The advent of GWAS in recent years has revolutionized the study of cancer association. Unlike the hypothesis-driven candidate gene approach, GWAS is a discovery-driven, agnostic approach that does not depend on prior knowledge of SNPs and genes. It thoroughly screens up to millions of common SNPs across the entire genome. Due to the multiple testing of SNPs in the screening phase, to increase validity of the results, stringent Bonferroni correction (P-value < 5×10⁻⁸) and multistage follow-up validations with large sample sizes are required.

Three recent GWASs identified four novel genetic susceptibility loci that mapped to 2p21 (EPAS1), 11q13.3 (a CCND1 transcriptional-enhancer site), 12p11.23 (ITPR2), and 2q22.3 (ZEB2) [68–70]. The observed effect size of each genetic locus is relatively small, which is expected for common variants. Two validated SNPs are located in intron 1 of EPAS1 (endothelial PAS domain-containing protein 1) on chromosome 2p21. However, the putative function of these SNPs is still unclear. EPAS1 encodes HIF-2α, which is a biologically plausible causal gene in the VHL/HIF pathway. Thus, GWAS results provide further support for the involvement of EPAS1 in RCC etiology. Interestingly, although the SNP found in 11q13.3 is not close to any gene with known function (>50 kb), the locus is hypothesized to be a transcriptional-enhancer site of CCND1 (encoding cyclin D1) [71]. Another locus maps to ITPR2 (inositol 1,4,5-trisphosphate receptor, type 2) on 12p11.23. Interestingly, the same SNP has also been identified as being associated with waist–hip ratio in another GWAS [72]. Finally, the locus on 2q22.3 is mapped to ZEB2 (zinc finger E-box-binding homeobox 2) that functions as a DNA-binding transcriptional repressor. Future pooled analysis of GWAS data with larger sample size will undoubtedly identify additional common RCC susceptibility SNPs.

The emergence of next-generation sequencing (NGS) provides a unique opportunity to discover rare variants that may explain some extent of the missing heritability in complex diseases. However, current NGS studies of kidney cancer have been mostly focused on profiling somatic mutations, while large-scale NGS studies using germline DNA have not been conducted. For example, a whole-exome sequencing (WES) study of tumor tissues has identified the SWI/SNF chromatin remodeling complex gene PBRM1 (polybromo 1) as the second most frequently mutated gene in ccRCC [73]. Another study reported that the tumor suppressor gene BAP1 (BRCA1 associated protein-1) could be used to define a new class of RCC [74]. With more than 400 tumor samples, The Cancer Genome Atlas (TCGA) identified 19 significantly mutated genes. Integrative analyses highlighted the importance of previously well-known pathways such as the VHL/HIF pathway, chromatin remodeling pathway, and PI3K/AKT/mTOR pathway [75]. Future NGS study using germline DNA may focus on identifying rare mutations, which can play critical role(s) in kidney cancer development.

Intermediate phenotypic biomarkers have the advantage of measuring aggregated effects of genetic variations and have larger effect size than individual SNPs. The suboptimal DNA damage/repair capacity in PBLs that were challenged with different mutagens was shown to be associated with increased risk of RCC [11, 12, 76]. One study found BPDE (benzo[α]pyrene diol epoxide) induced lymphocytic chromosome 3p deletion was associated with RCC risk [11]. Two other studies used comet assays to show that the high sensitivity of PBLs to NNK (nicotine-derived nitrosamine ketone)- and BPDE-induced DNA damage was associated with increased risk of RCC [12, 76]. Another interesting phenotypic biomarker is telomere length. Telomeres protect chromosomes from degradation and end-to-end fusion. Short telomere length in PBLs was associated with increased risk of a variety of cancers [77]. However, the association in RCC is inconsistent. Two case-control studies found that shorter telomeres in PBLs to be associated with an increased risk of RCC [78, 79] and one prospective nested case-control study failed to confirm the association [80]. Mitochondrial DNA (mtDNA) copy number is recently attracting great interest as a potential cancer susceptibility marker. Decreased mtDNA copy number was shown to be associated with multiple types of cancer. In two studies, lower mtDNA copy number in PBLs conferred an increased risk of RCC [81, 82]. By contrast, a recently published prospective study showed that a high copy number of mtDNA increased risk [83].