On 2014 Jun 03, David Keller commented:

Two large, well-designed randomized trials agree: Multivitamins reduce cancer risk in men

Guallar and colleagues dismiss the clinical trial data demonstrating that multivitamin and mineral supplements (MVM’s) reduce cancer risk in men, stating: “ Because the observed possible benefits were limited to men, were modest (as in PHS II), or were evident only in subgroup analyses (as in the SU.VI.MAX study) and did not consistently extend to reductions in mortality, these findings are only weak signals compatible with small or no benefit.” (1)

I will challenge each clause of their compound statement separately, but first, for the record, I quote the abstract of the Su.Vi.Max study (2):

"However, a significant interaction between sex and group effects on cancer incidence was found (P = .004). Sex-stratified analysis showed a protective effect of antioxidants in men (relative risk, 0.69 [95% confidence interval {CI}, 0.53-0.91]) but not in women (relative risk, 1.04 [95% CI, 0.85-1.29]). A similar trend was observed for all-cause mortality (relative risk, 0.63 [95% CI, 0.42-0.93] in men vs 1.03 [95% CI, 0.64-1.63] in women; P = .11 for interaction)."

Simply stated: Men who took multivitamins had significantly fewer cancers and a lower overall death rate than men who did not. These were not small effects: the cancer rate for men was reduced by 31% and the overall death rate by 37%.

First, let us discuss the fact that these benefits “were evident only in subgroup analyses”. Multiple post-hoc subgroup analyses are frowned on because they can be used to dredge the data for statistical flukes, such as finding that the tested intervention benefits only persons born under a certain astrological sign, or some other biologically implausible subgroup. The criteria for an acceptable subgroup analysis are: the subgroup should be rational, predefined in the experimental protocol, based on pre-randomization patient characteristics, large enough to retain statistical power, one of only a small number of such analyses, and the subgroup effect should be evident in other studies (3,4). The Su.vi.max subgroup analysis met these requirements; it was conducted on men, a large, natural subgroup which derived benefits in another study and was pre-specified in the study protocol. This rigorous analysis for the male subgroup yielded unambiguous results, the integrity of which withstands methodological scrutiny.

Guallar and colleagues also cast doubt on the data because the benefits of MVM’s were limited to men. The Su.vi.max investigators explained this by noting that women generally have better nutritional status than men; they are both more likely to eat fruits and vegetables, and also more likely to take MVM’s at baseline. Women, with good baseline nutrition, have less to gain by taking MVM’s and therefore exhibit no benefit in these studies.

There are many differences between men and women, both biological and cultural, which result, for one thing, in the fact that women outlive men by several years on average. The Su.vi.max and PHS-II study results indicate that relative dietary deficiencies of micro-nutrients may help explain the premature mortality of the male half of the population, as compared with the female.

As I pointed out in another comment (5), the five antioxidants included in the Su.vi.max supplement were a subset of the 30 nutrients in Centrum Silver, but at significantly higher doses. Thus, there was a dose-response effect evident for these antioxidants, with the PHS-II men who took MVM’s displaying a small but significant reduction in cancer rates, and the Su.vi.max men exhibiting a larger reduction in cancer rates, plus a decrease in all-cause mortality, consistent with the higher nutrient levels in the Su.vi.max supplement. The dose-response effect explains why the men in Su.vi.max had lower cancer rates compared with the men in PHS-II, and a significant reduction in overall mortality: they were taking higher doses of the beneficial micronutrients. Guallar’s complaint that the reduction in mortality was “not consistent” is thus refuted.

The USPSTF update again gave MVM supplements a grade of “I” (insufficient evidence to make a recommendation for the general population) for cancer prevention. However, NIH regulations mandate that clinical studies be analyzed separately for men and women when the response to an intervention differs by sex (6). Given the quality of the data supporting MVM use for men, USPSTF should change their rating of MVM supplements to at least a “C” level recommendation for men - meaning that men should discuss MVM supplements with their doctors in the context of their individual diet quality and nutritional status.

Unless they are consuming a diet rich in fruits and vegetables, men should consider adding a standard multivitamin & mineral supplement to reduce their risk of cancer, according to the best available evidence from two large, prospective randomized trials.

References:

1: Guallar E. Enough Is Enough, Annals of Int Med. June 3 2014; Vol 160, No. 9

2: Hercberg S, Galan P, Preziosi P, Bertrais S, Mennen L, Malvy D, et al. The SU.VI.MAX Study: a randomized, placebo-controlled trial of the health effects of antioxidant vitamins and minerals. Arch Intern Med. 2004; 164:2335-42.

3: Wittes J. On Looking at Subgroups. Circulation.2009; 119: 912-915 doi: 10.1161/CIRCULATIONAHA.108.836601

4: Dijkman B, et al. How to work with a subgroup analysis. Can J Surg. Dec 2009; 52(6): 515–522. PMCID: PMC2792383

5: Keller DL. Open letter to the USPSTF: the evidence shows multivitamins reduce cancers in men, PubMed Commons, accessed on 6/3/2014 http://www.ncbi.nlm.nih.gov/pubmed/24566474#cm24566474_4093

6: NIH website, accessed on 6/29/2014 http://grants.nih.gov/grants/funding/women_min/guidelines_update.htm

On 2014 Jun 03, David Keller commented:

Two large, well-designed randomized trials agree: Multivitamins reduce cancer risk in men

Guallar and colleagues dismiss the clinical trial data demonstrating that multivitamin and mineral supplements (MVM’s) reduce cancer risk in men, stating: “ Because the observed possible benefits were limited to men, were modest (as in PHS II), or were evident only in subgroup analyses (as in the SU.VI.MAX study) and did not consistently extend to reductions in mortality, these findings are only weak signals compatible with small or no benefit.” (1)

I will challenge each clause of their compound statement separately, but first, for the record, I quote the abstract of the Su.Vi.Max study (2):

"However, a significant interaction between sex and group effects on cancer incidence was found (P = .004). Sex-stratified analysis showed a protective effect of antioxidants in men (relative risk, 0.69 [95% confidence interval {CI}, 0.53-0.91]) but not in women (relative risk, 1.04 [95% CI, 0.85-1.29]). A similar trend was observed for all-cause mortality (relative risk, 0.63 [95% CI, 0.42-0.93] in men vs 1.03 [95% CI, 0.64-1.63] in women; P = .11 for interaction)."

Simply stated: Men who took multivitamins had significantly fewer cancers and a lower overall death rate than men who did not. These were not small effects: the cancer rate for men was reduced by 31% and the overall death rate by 37%.

First, let us discuss the fact that these benefits “were evident only in subgroup analyses”. Multiple post-hoc subgroup analyses are frowned on because they can be used to dredge the data for statistical flukes, such as finding that the tested intervention benefits only persons born under a certain astrological sign, or some other biologically implausible subgroup. The criteria for an acceptable subgroup analysis are: the subgroup should be rational, predefined in the experimental protocol, based on pre-randomization patient characteristics, large enough to retain statistical power, one of only a small number of such analyses, and the subgroup effect should be evident in other studies (3,4). The Su.vi.max subgroup analysis met these requirements; it was conducted on men, a large, natural subgroup which derived benefits in another study and was pre-specified in the study protocol. This rigorous analysis for the male subgroup yielded unambiguous results, the integrity of which withstands methodological scrutiny.

Guallar and colleagues also cast doubt on the data because the benefits of MVM’s were limited to men. The Su.vi.max investigators explained this by noting that women generally have better nutritional status than men; they are both more likely to eat fruits and vegetables, and also more likely to take MVM’s at baseline. Women, with good baseline nutrition, have less to gain by taking MVM’s and therefore exhibit no benefit in these studies.

There are many differences between men and women, both biological and cultural, which result, for one thing, in the fact that women outlive men by several years on average. The Su.vi.max and PHS-II study results indicate that relative dietary deficiencies of micro-nutrients may help explain the premature mortality of the male half of the population, as compared with the female.

As I pointed out in another comment (5), the five antioxidants included in the Su.vi.max supplement were a subset of the 30 nutrients in Centrum Silver, but at significantly higher doses. Thus, there was a dose-response effect evident for these antioxidants, with the PHS-II men who took MVM’s displaying a small but significant reduction in cancer rates, and the Su.vi.max men exhibiting a larger reduction in cancer rates, plus a decrease in all-cause mortality, consistent with the higher nutrient levels in the Su.vi.max supplement. The dose-response effect explains why the men in Su.vi.max had lower cancer rates compared with the men in PHS-II, and a significant reduction in overall mortality: they were taking higher doses of the beneficial micronutrients. Guallar’s complaint that the reduction in mortality was “not consistent” is thus refuted.

The USPSTF update again gave MVM supplements a grade of “I” (insufficient evidence to make a recommendation for the general population) for cancer prevention. However, NIH regulations mandate that clinical studies be analyzed separately for men and women when the response to an intervention differs by sex (6). Given the quality of the data supporting MVM use for men, USPSTF should change their rating of MVM supplements to at least a “C” level recommendation for men - meaning that men should discuss MVM supplements with their doctors in the context of their individual diet quality and nutritional status.

Unless they are consuming a diet rich in fruits and vegetables, men should consider adding a standard multivitamin & mineral supplement to reduce their risk of cancer, according to the best available evidence from two large, prospective randomized trials.

References:

1: Guallar E. Enough Is Enough, Annals of Int Med. June 3 2014; Vol 160, No. 9

2: Hercberg S, Galan P, Preziosi P, Bertrais S, Mennen L, Malvy D, et al. The SU.VI.MAX Study: a randomized, placebo-controlled trial of the health effects of antioxidant vitamins and minerals. Arch Intern Med. 2004; 164:2335-42.

3: Wittes J. On Looking at Subgroups. Circulation.2009; 119: 912-915 doi: 10.1161/CIRCULATIONAHA.108.836601

4: Dijkman B, et al. How to work with a subgroup analysis. Can J Surg. Dec 2009; 52(6): 515–522. PMCID: PMC2792383

5: Keller DL. Open letter to the USPSTF: the evidence shows multivitamins reduce cancers in men, PubMed Commons, accessed on 6/3/2014 http://www.ncbi.nlm.nih.gov/pubmed/24566474#cm24566474_4093

6: NIH website, accessed on 6/29/2014 http://grants.nih.gov/grants/funding/women_min/guidelines_update.htm