Vitamin D and cancer risk: cohort study
2 recent articles assessed the effect of vitamin D and cancer risk, both suggesting some benefit: one a prospective observational study and the other an interventional one.
1. A long-term Japanese prospective study found a decreased overall cancer risk with increasing vitamin D levels (see vit d and cancer japan bmj2018 in dropbox, or doi.org/10.1136/bmj.k671)
Details:
--nested case-cohort study within the Japan Public Health Center-based Prospective Study cohort, a study in 9 public health centers across Japan, started in 1990 and expanded in 1993, with 140,420 participants
--mean age 54, 36% male, BMI 24, 18% physical activity >1x/week, 72% never smoker/9% <20 pack-yrs/12% 20-40 pack-yrs/7% >40 pack-yrs, 62% never drinker/9% occasional/13% <150g per wk/8% 150-300gm/8% >300g, 4% diabetes, 21% fam history of cancer, 16% supplemental vitamins
--there were some impressive differences in the groups: cancer cases were somewhat older (56 vs 54), more often male (52% vs 34%), more likely to smoke >20 pack-yrs, more likely to drink any amount, more diabetes (7% vs 4%) [there was no analysis of statistical significance]
--higher vitamin D levels were found in older (1-2 years), more physically active in leisure time (1-2% more), less likely to have diabetes (0.5%) or family history of cancer (4%), less likely to smoke much (1-2%) but more likely to drink a lot (1%) [no analysis of statistical significance]
--3301 people with incident cases of cancer were compared to 4044 randomly selected participants (from base cohort of 33,736) for plasma 25(OH)D levels over median follow-up of 15.9 years
Results:
--inverse association between 25(OH)D levels and risk for total cancer:
--multivariable adjusted hazard ratio, as compared to the lowest quartile (15 ng/ml) (p=0.001 for the trend):
--second quartile (19 ng/ml): 19% decreased risk; HR 0.81 (0.70-0.94)
--third quartile (23 ng/ml): 25% decreased risk; HR 0.75 (0.65-0.87)
--fourth quartile (29 ng/ml): 22% decreased risk; HR 0.78 (0.67-0.91)
--inverse association between 25(OH)D levels and risk for liver cancer (the only specific cancer with a trend for statistical signficance):
--multivariable adjusted hazard ratio, as compared to the lowest quartile (p=0.006 for the trend):
--second quartile: 30% decreased risk; HR 0.70 (0.44-1.13)
--third quartile: 35% decreased risk; HR 0.65 (0.40-1.06)
--fourth quartile: 55% decreased risk; HR 0.45 (0.26-0.79)
--further adjustment for dietary factors (total energy, fruits/veges, meat, fish/shellfish, isoflavone, green tea, coffee): risk slightly attenuated, but still significant with p=0.02 for trend
--and even further adjustment for hepatitis B and C and ALT levels did not affect the results
--inverse association for premenopausal breast cancer (86 cases), p=0.03 for trend. but no overall relationship for breast cancer (though p=0.08, the confidence intervals were 0.51-1.11, so they are skewed to more likely benefit than harm)
--inverse relation with prostate cancer but of borderline statistical significance: multivariable adjusted decease with p=0.07 for trend, but conf intervals of 0.41-1.02, also making benefit statistically highly more likely than harm
--sensitivity analysis: no overall change if eliminate cancers at any one specific site; no difference by sex, occupational status, excluding cancers in first 3 years of the study, excluding those using supplementary vitamins
Commentary:
--there are some of the usual limitations of such a study as this one:
--one cannot prove causation in a cohort study (were there unknown/unaccounted for factors which differentiated those developing cancer vs not, where vitamin D might be a marker of such a factor, perhaps outdoors exercise which might also be associated with higher vitamin D levels??).
--there was only one baseline vitamin D level (in such a long-term study, were there significant changes in 25(OH)D levels related to changes in vitamin D added to foods, cultural shifts in foods/supplements over time, sun exposure, etc??)
--we are looking at many different cancers clumped together. these cancers have different known risk factors and mechanisms, making it a little strange to lump them together. is the cancer incidence related to changes in risk factors independent of vitamin D, because of higher levels of population awareness of the risk factors, and perhaps those with higher 25(OH)D levels overall were more health conscious? on the other hand, we do have some observational data that exercise, for example, does seem to have some general effect on many but not all cancers (some of which, like melanoma and perhaps prostate cancer may be worse with exercise) see http://gmodestmedblogs.blogspot.com/2017/01/leisure-time-activity-and-lower-cancer.html . and vitamin D receptors, as noted in the next blog, are generalized throughout the body, including in the part of the immune system devoted to cancer surveillance, which might explain a general benefit of 25(OH)D on cancer (see https://www.cambridge.org/core/services/aop-cambridge-core/content/view/302152110AEE222430F44164E53FEA90/S0029665111001650a.pdf/vitamin_d_and_immune_function_an_overview.pdf for some details)
--some of the cancers were quite infrequent (would these results be generalizable to areas of the world with higher relative incidences of those cancers which were underrepresented in this study?)
--there may be genetic differences in vitamin D receptors which might limit the generalizability of such a single-population cohort as in this Japanese study (does the higher frequency of a specific change found in Asians, the Fokl polymorphism in vitamin D receptors (associated with perhaps less vitamin D action), limit the generalizabillity of results to Westerners? And even Westerners may also have significant genetic variabilities also within different groups; for example African-American women more commonly have differences in vitamin D-binding protein, which is associated with higher bone density but lower 25(OH)D levels than white women--see Powe CD. N ENgl J Med 2013; 1991.)
--this study had the strong benefit of having lots of people and following them a long time.
--one other long-term large study, the Copenhagen City Heart Study, found 2400 cancers over 28 years of follow-up, revealing a 6% increased cancer risk associated with a 50% decrease in 25(OH)D levels
--many of the other studies finding no association between 25(OH)D levels and cancer were small, had few cancers identified, or had shorter follow-up; a meta-analysis found an 11% decrease in overall cancers with 20 ng/ml increase in 25(OH)D levels (see Yin L. Prev Med 2013; 57:753)
-- and some of the studies finding no benefit for vitamin D had very different baseline 25(OH)D levels (this study had a pretty common spread of 25(OH)D levels, certainly comparable to what we find in our community, not the high baseline levels in some of the negative studies; this latter point is supported in the above study, where the "protective effect" of higher 25(OH)D levels decreased going from the 3rd to 4th quartiles, ie suggesting a possible ceiling effect
see next blog for the randomized control trial and an overall perspective on vitamin D/cancer
1. A long-term Japanese prospective study found a decreased overall cancer risk with increasing vitamin D levels (see vit d and cancer japan bmj2018 in dropbox, or doi.org/10.1136/bmj.k671)
Details:
--nested case-cohort study within the Japan Public Health Center-based Prospective Study cohort, a study in 9 public health centers across Japan, started in 1990 and expanded in 1993, with 140,420 participants
--mean age 54, 36% male, BMI 24, 18% physical activity >1x/week, 72% never smoker/9% <20 pack-yrs/12% 20-40 pack-yrs/7% >40 pack-yrs, 62% never drinker/9% occasional/13% <150g per wk/8% 150-300gm/8% >300g, 4% diabetes, 21% fam history of cancer, 16% supplemental vitamins
--there were some impressive differences in the groups: cancer cases were somewhat older (56 vs 54), more often male (52% vs 34%), more likely to smoke >20 pack-yrs, more likely to drink any amount, more diabetes (7% vs 4%) [there was no analysis of statistical significance]
--higher vitamin D levels were found in older (1-2 years), more physically active in leisure time (1-2% more), less likely to have diabetes (0.5%) or family history of cancer (4%), less likely to smoke much (1-2%) but more likely to drink a lot (1%) [no analysis of statistical significance]
--3301 people with incident cases of cancer were compared to 4044 randomly selected participants (from base cohort of 33,736) for plasma 25(OH)D levels over median follow-up of 15.9 years
Results:
--inverse association between 25(OH)D levels and risk for total cancer:
--multivariable adjusted hazard ratio, as compared to the lowest quartile (15 ng/ml) (p=0.001 for the trend):
--second quartile (19 ng/ml): 19% decreased risk; HR 0.81 (0.70-0.94)
--third quartile (23 ng/ml): 25% decreased risk; HR 0.75 (0.65-0.87)
--fourth quartile (29 ng/ml): 22% decreased risk; HR 0.78 (0.67-0.91)
--inverse association between 25(OH)D levels and risk for liver cancer (the only specific cancer with a trend for statistical signficance):
--multivariable adjusted hazard ratio, as compared to the lowest quartile (p=0.006 for the trend):
--second quartile: 30% decreased risk; HR 0.70 (0.44-1.13)
--third quartile: 35% decreased risk; HR 0.65 (0.40-1.06)
--fourth quartile: 55% decreased risk; HR 0.45 (0.26-0.79)
--further adjustment for dietary factors (total energy, fruits/veges, meat, fish/shellfish, isoflavone, green tea, coffee): risk slightly attenuated, but still significant with p=0.02 for trend
--and even further adjustment for hepatitis B and C and ALT levels did not affect the results
--inverse association for premenopausal breast cancer (86 cases), p=0.03 for trend. but no overall relationship for breast cancer (though p=0.08, the confidence intervals were 0.51-1.11, so they are skewed to more likely benefit than harm)
--inverse relation with prostate cancer but of borderline statistical significance: multivariable adjusted decease with p=0.07 for trend, but conf intervals of 0.41-1.02, also making benefit statistically highly more likely than harm
--sensitivity analysis: no overall change if eliminate cancers at any one specific site; no difference by sex, occupational status, excluding cancers in first 3 years of the study, excluding those using supplementary vitamins
Commentary:
--there are some of the usual limitations of such a study as this one:
--one cannot prove causation in a cohort study (were there unknown/unaccounted for factors which differentiated those developing cancer vs not, where vitamin D might be a marker of such a factor, perhaps outdoors exercise which might also be associated with higher vitamin D levels??).
--there was only one baseline vitamin D level (in such a long-term study, were there significant changes in 25(OH)D levels related to changes in vitamin D added to foods, cultural shifts in foods/supplements over time, sun exposure, etc??)
--we are looking at many different cancers clumped together. these cancers have different known risk factors and mechanisms, making it a little strange to lump them together. is the cancer incidence related to changes in risk factors independent of vitamin D, because of higher levels of population awareness of the risk factors, and perhaps those with higher 25(OH)D levels overall were more health conscious? on the other hand, we do have some observational data that exercise, for example, does seem to have some general effect on many but not all cancers (some of which, like melanoma and perhaps prostate cancer may be worse with exercise) see http://gmodestmedblogs.blogspot.com/2017/01/leisure-time-activity-and-lower-cancer.html . and vitamin D receptors, as noted in the next blog, are generalized throughout the body, including in the part of the immune system devoted to cancer surveillance, which might explain a general benefit of 25(OH)D on cancer (see https://www.cambridge.org/core/services/aop-cambridge-core/content/view/302152110AEE222430F44164E53FEA90/S0029665111001650a.pdf/vitamin_d_and_immune_function_an_overview.pdf for some details)
--some of the cancers were quite infrequent (would these results be generalizable to areas of the world with higher relative incidences of those cancers which were underrepresented in this study?)
--there may be genetic differences in vitamin D receptors which might limit the generalizability of such a single-population cohort as in this Japanese study (does the higher frequency of a specific change found in Asians, the Fokl polymorphism in vitamin D receptors (associated with perhaps less vitamin D action), limit the generalizabillity of results to Westerners? And even Westerners may also have significant genetic variabilities also within different groups; for example African-American women more commonly have differences in vitamin D-binding protein, which is associated with higher bone density but lower 25(OH)D levels than white women--see Powe CD. N ENgl J Med 2013; 1991.)
--this study had the strong benefit of having lots of people and following them a long time.
--one other long-term large study, the Copenhagen City Heart Study, found 2400 cancers over 28 years of follow-up, revealing a 6% increased cancer risk associated with a 50% decrease in 25(OH)D levels
--many of the other studies finding no association between 25(OH)D levels and cancer were small, had few cancers identified, or had shorter follow-up; a meta-analysis found an 11% decrease in overall cancers with 20 ng/ml increase in 25(OH)D levels (see Yin L. Prev Med 2013; 57:753)
-- and some of the studies finding no benefit for vitamin D had very different baseline 25(OH)D levels (this study had a pretty common spread of 25(OH)D levels, certainly comparable to what we find in our community, not the high baseline levels in some of the negative studies; this latter point is supported in the above study, where the "protective effect" of higher 25(OH)D levels decreased going from the 3rd to 4th quartiles, ie suggesting a possible ceiling effect
see next blog for the randomized control trial and an overall perspective on vitamin D/cancer
Comments
Post a Comment
if you would like to receive the near-daily emails regularly, please email me at gmodest@uphams.org