Can habitual physical activity contribute to reducing the health burden of renal cancer?

Renal pelvic cancers.

There are only 2 published reports concerning physical activity and the risk of renal pelvic cancers (Pan et al., 2006; Wilson et al., 2008). The first of these investigations found no association of renal pelvic cancer with the reported volume of leisure activity, assessed in MET-h/wk. The second report (Wilson et al., 2008) was based on quite a large sample (1,014 pelvic cancers in men and 360 in women) drawn from workers employed in 1960 and/or 1970. Data were adjusted for age and year of diagnosis, and a significant decrease of risk was seen in men with active employment [a standardized incidence ratio of 0.66 (95% confidence limits 0.50-0.88)]. Moreover, the risk for both sexes was increased among those employed in specific trade categories, including sedentary work, indoor work, the machine/electronics industry, and (in men) shop and construction metal industries.

Cohort studies of renal parenchymal cell cancers

Thirteen cohort studies and 1 state-wide correlational analysis (Colli et al., 2009) have examined the incidence of deaths from renal cell cancer in relation to reported habitual physical activity [or in one instance, to an estimate of attained aerobic fitness (Thompson et al, 2008)](Table 2).

Bergstrӧm et al., (1999) conducted an 18-year prospective study of a large sample of Swedes (3.28 million men and 3.35 million women), noting diagnoses of renal cell cancer in relationship to a 4-level categorization of occupational activity. A substantial number of renal tumours (3,291) was seen over this period. In the men, a significant decrease of risk with a dose-response relationship was observed in those with physically more demanding jobs [a relative risk of 0.80 (0.65-0.98) after adjusting for socio-economic status, area of residence and year of follow-up]. However, (perhaps because of lesser job demands or a confounding of data by demanding domestic physical activity), a heavy occupational energy expenditure provided no protection against renal cancer in the women, with a relative risk of 1.25 (0.79-1.96) in the more active individuals. Bergstrӧm, Terry, & Lindblad, (2001) also followed 17,241 twin pairs for a period of 20 years. In this sample, they accumulated only 102 cases of renal cell cancer. They introduced smoking, body mass index and hypertension as co-variates in their second study, and with this adjustment of the data they observed no evidence of any decrease of risk with an active occupation [relative risk 1.25 (0.63-2.50)] or active leisure [relative risk 1.67 (0.83-3.33) in either men or women.

Choi, Jee, Sull, & Nam (2005) followed a sample of 576,562 men initially aged 30 to > 60 years for a total of 6 years. This yielded 92 deaths from renal cancer. The prime focus of the study was on hypertension rather than exercise, and there was a substantial increase of relative risk with a high blood pressure (2.43). However, the risk for exercisers (as classified only by a simple "yes" vs. "no" response) was not significant [0.9 (CI 0.6-1.4)]. In addition to weakness in the assessment of physical activity, this calculation apparently included no co-variates.

Colli et al. (2009) compared the state-wide incidence of renal cancer with the state-wide prevalence of physical inactivity, as reported in the NHANES survey. The 2 indices were significantly correlated in both men (r = 0.48, p <0.001) and women (r = 0.31, p = 0.04).

George et al. (2011) accumulated a substantial sample of 1206 cases of renal cell cancer after following 300,000 adults for 10 years. In contrast to other investigators, they focussed upon sedentary behaviour rather than habitual physical activity. After adjusting data for age, sex, race, diabetes, smoking, alcohol consumption, diet and energy intake, they calculated hazard ratios for the time spent sitting watching television or videos (<1 vs. >7 h/day) and for the total sitting time (<3 vs. >9 h/day), but they found no significant decrease in the risk of renal cancer among the less sedentary individuals, with respective risk ratios of [1.04 (0.72-1.52)] and [0.89 (0.71-1.15)]. One issue with this study is the inclusion of energy intake as a co-variate. Inevitably, this would have claimed some of the variance that would otherwise have been attributed to physical activity.

A 12-year follow-up of 29,133 male smokers found 210 cases of renal cell cancer (Mahabir, Leitzman, & Pietinen, 2004). The risk was unrelated to occupational activity [risk ratio 1.08 (0.54-2.15)], but was strongly related to reported leisure activity, with a dose-related risk ratio of 0.46 (0.18-1.13) favouring the most active individuals in a multivariate analysis.

Moore et al. (2008) questioned 482,386 adults about their leisure activity, both currently and as an adolescent. During an 8.2-year follow-up, a substantial 1,238 of their sample developed a renal cancer. After adjusting data for age, sex, body mass index, height, ethnicity, smoking, diabetes, protein intake, and hypertension, risk was reduced both with frequent current leisure activity [RR 0.77 (0.64-0.92)] and with frequent leisure activity as an adolescent [0.82 (0.68-1.00)].

Nicodemus et al. (2004) followed a sample of 34,637 women for 15 years, accumulating 124 cases of renal cancer. Most of the group reported little physical activity, but nevertheless in an age-adjusted analysis a strong inverse relationship was seen between renal cancer and vigorous leisure activity [relative risk 0.37 (0.14-0.99)]. Relative weight was a major risk factor, and after including this and other co-variates in the model, leisure activity no longer had a significant effect upon the incidence of renal cancer (data not shown).

Paffenbarger et al. (1987) made a retrospective survey of records for 56,683 university alumni over a period of up to 70 years, looking at the incidence of fatal or non-fatal renal cancer in relation to participation in sport while attending university (>5h/week vs. <5h/week). Their sample included 53 renal tumours, and the relative risk in the more active group was 0.95 (0.47-1.94). However, the only co-variate introduced into this analysis was the individual's birth age, and no information was available on the activity patterns of the subjects after they had left university.

Setiawan et al. (2007) completed a prospective study of 161,126 adults over a period averaging 8.3 years, finding 220 cases of renal cancer in men and 127 cases in women. A multivariate analysis demonstrated that substantial relative risks of renal cell cancer were associated with current smoking [2.3 (M), 1.7 (F)], obesity [1.76 (M), 2.27 (F)] and hypertension [1.42 (M), 1.58 (F)]. Questionnaire estimates of physical activity (MET/day), classified by quartiles, also showed a significant inverse relationship to cancer risk in women [the relative risk for the third quartile was 0.52 (0.31-0.89) and that for the fourth quartile was 0.66 (0.40-1.10)], with a dose-related trend (p = 0.027). In contrast, a physically active lifestyle conferred no significant benefit in the men [relative risk for 4th quartile 1.09 (0.75-21.58)]. The authors of this report suggested that the sex difference in response might reflect in part some residual confounding of data by the effects of obesity (which were greater in the women than in the men).

Thompson et al. (2008) examined relationships between a treadmill prediction of aerobic fitness (here considered as a surrogate of habitual aerobic activity) and the risk of dying from various types of cancer in 18,858 men with pre-diabetes or diabetes who were attending the Aerobics Fitness Center in Dallas, TX. Over an average 16.4 year follow-up, this series included 33 deaths from renal tumours, and data adjusted for age, examination year, smoking, alcohol consumption, body mass index, fasting glucose and previous history of cancer showed no significant reduction of risk in those who had achieved at least a moderate level of fitness [(hazard ratio 0.91 (0.45-2.68)].

The cohort study of van Dijk et al. (2004) followed 120,852 adults over a period of 9.3 years. They found 275 histologically-confirmed cases of renal cell cancer over this period. Risk was unrelated to energy intake, but in the men there was a substantial inverse relationship of risk to reported leisure activity. The effect was largest in those taking 30-60 minutes of such activity per day, with a relative risk of 0.52 (0.30-0.91) relative to those who were inactive. There was no evidence of a dose-response gradient, and indeed effects were smaller and non significant in those reporting longer periods of active leisure (at 90 min/d, the relative risk for the men was 0.74 (0.44-1.23), and the risk tended to an increase in the most active women [relative risk 1.13 (0.56-2.29). Moreover, the male participants in this study showed no statistically significant benefit from a higher average rate of occupational energy expenditure [>12 vs. < 8 kJ/min, 0.82 (0.46-1.47)].

In a study focussed primarily upon the effects of cigarette smoking, Washio et al. (2013) followed a sample of 46,395 men and 64,190 women for approximately 20 years, accumulating 88 deaths from kidney cancer. Information was also obtained on sport involvement and daily walking. A non-significant trend favoured those practicing sport for >3 hours/week versus those involved in sport for less than 1 hour/week [hazard ratio 0.64 (0.30-1.37)], while there was statistically significant benefit from walking >60 minutes/day versus <30 minutes/day [hazard ratio 0.57 (0.32-1.00)(p for trend = 0.039)]. Unfortunately, although these ratios were adjusted for age and sex, they were not co-varied for smoking habits or other important variables.

Yun et al. (2008) examined data for 446,927 Korean men aged from 40 to over 60 years. After adjusting data for age, diet, smoking, alcohol consumption, body mass index and fasting blood sugar, the relative risk in those with moderate or high levels of leisure activity did not differ from those with a low level [relative risk 1.01 (0.83-1.23). Data were also sorted by smoking habits; in non-smokers the relative risk was 0.94 (0.68-1.30), and in smokers it was 1.02 (0.78-1.32).

In summary, many cohort investigations have examined quite large samples and followed subjects prospectively over quite long periods; nevertheless, many studies have accumulated only a small number of cases of renal cancer, limiting the likelihood of definitive conclusions. Only 6 of the 13 cohort studies found any relationship between renal cancer and leisure or occupational activity (Bergstrӧm, Moradi, & Lindblad, 1999; Moore, Chow, & Schatzkin, 2008; Nicodemus, Sweeney, & Folsom, 2004; Setiawan, Stranm, & Nomura, 2007; van Dijk, Schouten, & Kiemeney, 2004; Washio et al., 2013). Moreover, the 95% confidence limits of any apparent benefits were broad, and even in these reports, the positive findings were limited to sub-groups determined post-hoc [men only (Bergstrӧm et al., 1999), benefit not seen in a subsequent study with co-variates (Bergstrӧm et al., 2001); benefit seen in leisure but not occupational activity (Moore et al., 2008); benefit not seen after multivariate analysis (Nicodemus et al., 2004 ); benefit seen in women only (Setiawan et al. 2007); benefit limited to leisure activity and seen only with moderate, but not with prolonged activity (van Diyjk et al., 2004); and benefit seen with walking but not with sports involvement (Washio et al., 2014). The 13 studies totalled 7241 cases of renal cancer, and a simple weighting of data by case numbers showed an average risk ratio of 0.88 in the more active individuals. Omitting 5 studies with few co-variates (Bergstrӧm et al., 1999; Choi et al., 2005; Nicodemus et al., 2004; Paffenbarger et al., 1987); Washio et al., 2013), the risk ratio for active individuals (0.96) approached unity. In 7 occupational studies, the average risk for active individuals was again 0.88, but omitting one study without co-variates, the risk ratio rose to 0.98. In the leisure studies, the weighted average for all reports was 0.91, or 0.93 on omitting 4 studies with limited allowance for co-variates.

Case-control studies of renal parenchymal cell cancers

There have been at least 14 unique case-control studies that have included data on the risks of renal cancer in relation to occupational or leisure-time physical activity.

Brownson, Chang, & Davis (1991) made a general survey of cancers in relation to occupational classifications among 17,174 male cancer cases in Missouri. Their sample included 449 individuals with renal tumours. The control group was drawn, somewhat surprisingly, from other individuals with cancer rather than healthy individuals, although cancers of the colon were excluded from the control group. After adjusting data only for age and smoking habits, there was no significant difference in the standardized incidence ratio of renal cancer between those whose occupation demanded a high level of physical activity and those whose occupational activity was low [relative risk 0.77 (0.50-1.11)].

Chiu et al. (2006) compared 406 cases of renal cancer aged 64-68 years with 2,434 controls. In the women, the odds ratio of developing renal cancer among those who reported that they exercised at least once per day versus those who took leisure exercise less than once per month was 0.40 (0.19-0.83), but frequent leisure activity had no significant influence on the risk of renal tumours in the men [odds ratio 0.83 (0.48-1.43)]. There was a substantial risk associated with a high body mass index, but apparently little interaction between body mass index and reported habitual physical activity.

Goodman, Morgernstern, & Wynder (1986) studied 189 men and 79 women who had developed renal cancer with age, sex and race-matched hospital controls. They made a 3-level categorization of leisure activity, ranging from participation in strenuous sports to inactive, and also categorized occupational activity. The risk of renal cell cancer was strongly related to the individual's body mass index, but appeared to be unrelated to either active leisure pursuits [relative risk 1.14 (0.65-2.05) in men, 1.11 (0.44 (0.44-2.97) in women] or occupational physical activity [0.88 (0.48-1.55) in men, 1.20 (0.41-4.06) in women].

In a study that focussed primarily on diet and fibre intake, Hu et al. (2008) compared the reported leisure activities of 617 men and 521 women with renal cancer against the responses of 5,039 age and sex-matched controls. The data were adjusted for age, sex, and residential area only. Questions were asked about the frequency, duration and seasonality of 12 common categories of moderate and of strenuous physical activity. The odds ratio of renal cancer for those with the greatest volume of moderate activity was 0.86 (0.67-1.10), and for those with the greatest volume of strenuous activity (>55 min/week) it was 0.90 (0.71-1.14).

Lindblad et al. (1994) conducted a case-control study that compared renal cell cancer in 207 men and 172 women with findings in 353 controls. A structured interview found a strong inverse and dose-related relationship between relative risk and a 4-level classification of job activity in men, the greatest benefit being associated with vigorous physical activity at the age of 40 years [an odds ratio of 0.37 (0.16-0.89)]. In the men, significant benefit was also associated with the demands of occupation at the age of 30 years. However, occupational activity had no significant impact upon the relative risk of renal cancer in women [odds ratio 0.75 (0.20-2.86)], and leisure activity was said to show no association with the risk of renal cancer in either sex (although the data were not presented).

Mellemgaard et al. (1994) compared 368 histologically-verified cases of renal cancer with 396 controls. They found an association between the incidence of such tumours and relative weight, but no relationship to physical activity either at work or in leisure time. The following year, Mellemgaard et al. (1995) looked at a much larger series of 1,732 cases and 2,309 controls, making a 5-level classification of both occupational and leisure activity at the ages of 20 and 40 years. However, after adjusting data for age, body mass index and smoking, risk ratios showed no significant effect of either occupational [in men 1.11 (0.71-1.67 in women 1.67 (0.91-3.33)] or leisure activity [in men 1.11 (0.56-2.50), in women 1.67 (0.71-3.33)]. Nevertheless, the use of amphetamine was associated with a substantial increase of risk in men.

Menezes et al. (2003) compared 173 male and 133 female cases of renal cancer with matched controls. After adjustments for age, body mass index and smoking habits, reports of strenuous leisure activity were associated with a relative risk of 0.40 (0.22-0.75) in women. However, no reduction of risk was seen among those reporting only moderate physical activity [1.05 (0.55-2.00)]. In the men, both total recreational activity and moderate physical activity yielded risk ratios of 0.49 (0.27-0.89), with the greatest effects being seen from physical activity undertaken as an adolescent. However, any benefit associated with participation in vigorous activity was smaller and not statistically significant [0.72 (0.42-1.23)].

Paffenbarger, Lee, & Wing (1992) examined relationships between fatal and non fatal cancers and prior membership of sports teams in a sample of 56,683 university alumni. There were only 29 cases of renal cancer in this series, and these incidents bore no relationship to the rather tenuous measure of habitual physical activity provided by affiliation to an athletic team as a young adult (a relative risk 0.95 among those who engaged in more than 5 hours of sport per week while at university). Two specific problems with this investigation are that participation in many sports is influenced by body build, and that by middle-age, many of those who were athletes in their youth have become less active and more obese than their "non-athletic" peers (Montoye, Van Huss, Olson, Pierson, & Hudec, 1957).

Pan et al. (2006) compared 810 individuals aged 20-76 years who developed renal cancers with 3,106 controls. Data were adjusted for age, province of residence, educational attainment, body mass intake, total energy intake, smoking, diet, and any exposure to toxic industrial chemicals. A high body mass index was associated with a large increase of risk of renal cancer. Recreational activity was estimated in MET-h/week, ranging from <6 to >37 in men and <6 to >31 in women. The odds ratios for renal cell carcinoma were 1.21(0.86-1.69) and 0.80 (0.54-1.18) for active men and women respectively. Data were also obtained for non-renal cell carcinomas; these were unrelated to the volume of leisure activity.

Parent, Rousseau, & El Zein (2011) reported data for 177 cases of renal cancer aged ~ 59 years and 533 population controls. Data were adjusted for age, socio-economic status, ethnicity, smoking, body mass index, and other potential confounders. The odds ratio for those facing heavy energy demands at work was 0.84 (0.38-1.89). A simple classification of leisure activity ("often" vs. "never/not often") showed no evidence of benefit from physical activity, with an odds ratio of 1.11 (0.76-1.64) for active individuals; further,a combination of occupational and leisure activity did not enhance prognosis [odds ratio 1.02 (0.70-1.49)].

Pukkala et al. (1992) compared the experience of 1,449 physical education teachers with that of 8,619 language teachers. On various measures, the typical physical education teacher had a demonstrably more active lifestyle than the language teachers. The standardized incidence ratio of renal cancer for the physical education teachers was 0.39 (0.01-2.15), whereas that for the language teachers was 1.33 (0.73-2.23), a risk ratio of 0.29, although confidence limits were also broad, since the entire study included only 15 individuals with a renal cancer.

In an article that focussed on leptin levels, Spyridopoulos et al. (2009) examined the minutes per day spent in "physical exercise" between 70 cases of renal cancer and 280 age and sex-matched controls. A chi2 comparison between those taking no activity versus those taking > 90 min/d found a significant advantage in the latter group (p <0.001).

Tavani et al. (2007) studied 767 individuals with renal tumours and 1,534 controls, looking at both occupational and leisure activity as reported for various ages from 15 to 60 years. After controlling for body mass index, smoking, alcohol consumption, hypertension and study centre, a beneficial effect was associated with occupational activity, irrespective of the age when it was undertaken [odds ratio 0.71 (0.55-0.92) p <0.01) p <0.01), but no effect of leisure activity was observed [OR 1.03 (0.78-1.36)].

Xiao et al. (2014) drew a distinction between "white" and "black" subjects, comparing the habitual physical activity of 1217 histologically confirmed cases with that of 1225 population controls when in their 20s and in their 50s. Data were obtained on occupational and leisure activity and on active transportation. After adjustment of the data for study centre, age, sex, educational attainment, smoking habits and family history of cancer, no significant effects of physical activity were seen in the blacks. However, in the whites there was a suggestion of a positive trend for those with a high total activity score both in their 20s [OR 0.74 (0.50-1.10), p for trend = 0.06] and in their 50s [OR 0.67 (0.41-1.05), p for trend = 0.12]. Further, significant benefit was associated with tranportation-related activity in the 20s, and with leisure-time activity in the 50s.

Many of the case-control studies suffer the disadvantage of including relatively few cases of renal cancer. Of 14 unique case-control studies, only 6 showed a statistically significant advantage to an active group, and as with cohort studies this was generally in a sub-group of the data [women only (Chiu, Gapstur, & Chow, 2006), men only, and at work but not in leisure (Lindblad, Wolk, & Bergstrӧm, 1994), in men and women, but a greater benefit for moderate than for vigorous activity (Menezes, Tomlinson, & Kreiger, 2003), a large but non-significant advantage of standardized incidence ratio based on a very small sample (Pukkala, Poskiparta, & Apter, 1992), benefit from occupational but not leisure activity (Tavani, Zucchetto, & Dal, 2007)], and benefit in white but not in black individuals (Xiao et al., 2014). In all, the case-control studies included 7410 cases of renal cell carcinoma, with a weighted average risk for an active lifestyle of 0.93. Omitting 5 studies with limited co-variance adjustment, the risk ratio increased marginally to 0.94. The weighted average for the occupational studies was 0.98, or 0.99 omitting 3 studies with limited co-variates; likewise, the average for leisure studies was 0.98, and omitting 4 reports with limited co-variates it was 1.01.