Why Scientists Accept It but Physicians Do Not
(OMNS Feb 6 2019)
The UVB-vitamin D-cancer hypothesis is nearly 40 years old [Garland, 1980]. There are 5293 publications with cancer and vitamin D or 25-hydroxyvitamin D [25(OH)D] in the title or abstract listed at pubmed.gov as of January 30, 2019. Nonetheless, this hypothesis has not been widely accepted; in fact, since the publication of the results for vitamin D supplementation for cancer in the VITamin D and OmegA-3 TriaL (VITAL) [Manson, 2019], support has been eroded further. As will be discussed here, the problem does not seem to be lack of evidence but, instead, the difference in how two cultures, the 'scientific' and the 'medical' communities, evaluate evidence.
Scientific evidence can take many forms. For vitamin D and cancer, there are several main types of evidence. The type first used to develop the hypothesis and extend it was the geographical ecological approach. In this approach, populations are defined by geographical location and cancer outcomes averaged for each population are compared statistically with averaged indices for solar UVB dose as well as other cancer risk-modifying factors. This approach has identified about 20 types of cancer for which solar UVB doses are inversely correlated with mortality rates, primarily in countries near the equator [Moukayed, Grant, 2013]. But since this approach cannot prove causality, critics question whether other factors might explain the findings.
Another approach is called "observational." Generally, participants are enrolled in cohorts with blood drawn at baseline and cancer incidences noted prospectively during a several-year follow-up period. Such studies have found very strong inverse correlations between baseline serum 25(OH)D concentration and incidence of colorectal cancer [Garland, Gorham, 2017]. However, for breast cancer, such prospective studies have not found similar results. As explained in 2015, the problem with studying breast cancer incidence in this way is that it develops very rapidly, so that baseline 25(OH)D after a long follow-up period may not be as relevant [Grant, 2015]. However, case-control studies, in which 25(OH)D is measured near the time of cancer diagnosis, find very strong inverse correlations between 25(OH)D and breast cancer incidence [Grant, 2015], [Grant and Boucher, 2017]. Critics of this approach question whether having cancer affects 25(OH)D, i.e. "reverse causation." There is no evidence that it does in studies made near the time of diagnosis, but it could as cancer progresses. However, concern over "reverse causation" has been raised to explain why some vitamin D randomized controlled trials (RCTs) do not agree with most observational studies [Autier, 2017].
Studies of biological mechanisms
A third approach is to study mechanisms whereby vitamin D reduces risk of cancer. The mechanisms can be grouped into those that reduce incidence, angiogenesis, and metastasis -- and are well known [Moukayed, Grant, 2013], [Moukayed, Grant, 2017].
A fourth approach is to use Mendelian randomization (MR) studies on gene variants that regulate serum 25(OH)D concentrations [Zheng et al. 2017]. This approach is thought to be independent of UVB exposure and oral vitamin D intake, and so should be able to assess causality. Unfortunately, the mutations studied so far only affect 25(OH)D by small amounts, so that to be definitive, this approach would require a large cohort of participants, perhaps 100,000. While there have been many MR studies of vitamin D and cancer, only one, for ovarian cancer [Ong, 2016], found a beneficial effect of serum 25(OH)D elevated by gene variants, and a subsequent study by the same author did not [Ong, 2018].
Randomized Controlled Trials
This brings us to the fifth approach, one favored by the medical profession -- the randomized controlled trial (RCT). The two basic assumptions of RCTs for pharmaceutical drugs are that the trial is the only source of the agent and that there is a linear dose-response relationship. Neither assumption is satisfied for vitamin D; an increment in the dose of vitamin D induces a smaller change in serum 25(OH)D with higher baseline 25(OH)D levels, so that at higher baseline 25(OH)D levels the same dose produces smaller reduction in cancer risk. Robert Heaney pointed out in 2014 that the proper way to conduct RCTs for nutrients such as vitamin D required assessment of vitamin D status [Heaney, 2014], an approach that was further developed in 2018 [Grant et al., 2018]. With this background, we can review the vitamin D-cancer RCTs reported to date.
The first reported vitamin D-cancer RCT showing benefits of supplementation was from Creighton University [Lappe, 2007]. This study involved postmenopausal women living in Nebraska who were on average slightly overweight (body mass index 29), with a baseline 25(OH)D of 29 ng/mL. Those in one treatment group were given 1000 IU/d vitamin D3 plus 1450 mg/d calcium, those in a second treatment group were given 1450 mg/d calcium and placebo, while those in the control group were given only the placebos. This study found that the reduction of incident cancer in the Ca + D group was 60% and in Ca-only group was 47%. But when analysis was performed for cancers diagnosed after the first 12 months, the reduction for the Ca + D group jumped to 77% while it did not change significantly for the Ca-only group. In this study, both vitamin D treatment and basal 25(OH)D were found to be significant and independent predictors of cancer risk. [Lappe, 2007].
The second report of a successful vitamin D RCT was a reanalysis of data from the Women's Health Initiative. This study found, in a group of more than15,000 women, that for those (43%) who were not taking personal calcium or vitamin D supplements at the beginning of the trial, CaD supplements given to a random selection of them non-significantly reduced the risk of colorectal cancer by 17%, but significantly decreased the risk of total, breast, and invasive breast cancers by 14-20% . In the women who were taking personal supplements of calcium or vitamin D, the additional supplements did not alter cancer risk. [Bolland, 2011]. These results are consistent with the women not taking vitamin D supplements prior to entry into the study having had low baseline serum 25(OH)D levels.
The third major successful vitamin D RCT was another one from Creighton University, again with people on average slightly overweight (BMI 30) with a baseline 25(OH)D level of 33 ng/mL. Those in the treatment group were given 2000 IU/d vitamin D3 plus 1500 mg/d calcium, while those in the control group were given placebos; this study found that cancer incidence over 4 years was 4.2 percent in the vitamin D3 + calcium group and 6 percent in the placebo group, a non-significant difference. [Lappe, 2017] However, observational data from this RCT, reported in the online supplement, showed that the achieved serum 25(OH)D was significantly inversely associated with cancer incidence, because compared to those with a baseline serum 25(OH)D level < 30 ng/ml, the cancer incidence for those with a 25(OH)D level between 30 and 55 ng/ml was reduced by 35 percent.
The vitamin D-cancer RCT with the most recent results is VITAL [Manson, 2019]. This RCT had more than 25,000 participants including 5106 black participants. Half were given 2000 IU/d vitamin D3 and half were given a placebo. Again, most were slightly overweight (BMI 28). Mean baseline 25(OH)D in the treatment group was 28 ng/ml for males (based on 395 participants) and 32 ng/ml for females (based on 441 participants). Over a 5.3year follow-up the study found cancer in 1617 participants (vitamin D: 793, placebo: 824).The vitamin D group had a 17% lower risk of death from cancer (341 deaths); a slightly higher risk (2%) for breast cancer; a slightly higher risk (9%) for colorectal cancer; and a 12% lower risk for prostate cancer. However, it also reported that, excluding the first two years of follow-up, the risk of death from cancer was reduced by 25%, and that for those who were borderline overweight (BMI < 27), the risk was reduced by 14%, while those who were not overweight had a 26% lower risk of invasive cancer! In subjects with dark skin, the risk was non-significantly lower by 13%. Since a given vitamin D dose will increase serum 25(OH)D more in thin than in fat people, those BMI results were predictable, and since those with dark skin have average serum 25(OH)D levels lower than those with light skin, those differences were also predictable.
While it is understandable that the New England Journal of Medicine has a policy of reporting only one major result for an RCT, what is not understandable is that the non-significant results in the above-mentioned study for cancer were reported but not the significant ones. Unfortunately, the vast majority of news reports only referred to results mentioned in the abstract, often quoting the first author, but without mentioning the beneficial effects of supplementation. A contributing factor to the errant reporting was likely that the VITAL study was designed prior to 2012 [Manson, 2012], when the importance of assessing vitamin D status by measuring 25(OH)D first became widely appreciated. In addition, concern had been raised during the Institute of Medicine's review on vitamin D that there might be a U-shaped 25(OH)D-health outcome relationship [Ross, 2011], based on some observational studies. Thus, instead of using a higher vitamin D3 dose, such as 4000 IU/d, only 2000 IU/d was given. It was subsequently pointed out that the poorer outcomes noted with high 25(OH)D levels were analyzed without ascertaining when those subjects started supplementation [Grant, 2016]. However, if such subjects had only recently started supplementation, either for a common condition such as osteoporosis, or for non-specific symptoms associated with their illness, the resulting delay in the rise of their 25(OH)D level could skew the result.
GrassrootsHealth.net has taken the initiative and is conducting vitamin D supplementation studies based on measurements of 25(OH)D. In their first such study, they pooled results for 1135 women in the GrassrootsHealth volunteer cohort plus 1169 women in the first Creighton University RCT [Lappe, 2017], and found a large and significant reduction in risk (67%) for all-cancer incidence for serum 25(OH)D levels > 40 ng/ml vs. < 20 ng/ml [McDonnell, 2016]. In their second study, based on 3325 women from the two Creighton University cohorts and 1713 women from the GrassrootsHealth cohort, they found reductions in breast cancer incidence that were associated with higher levels of baseline serum 25(OH)D. Women with 25(OH)D concentrations ≥ 60 ng/ml had an 80% lower risk of breast cancer than women with concentrations < 20 ng/ml after adjustment for age, BMI, smoking status, calcium supplement intake, and study of origin. [McDonnell, 2018]
In retrospect, the ecological studies were based primarily on cancer mortality rates, for which the evidence of beneficial effects of vitamin D is stronger than for cancer incidence rates. However, combining all the different types of studies does suggest that having a serum 25(OH)D in the range of 40-60 ng/ml significantly reduces the risk of, and increases survival from, many common types of cancer. One exception is prostate cancer, for which a higher serum 25(OH)D appears to predict increased incidence rate, [Gao, 2018] but also increases survival rates [Song, 2018]. It should also be noted that dark skin lowers cancer-specific survival rates compared to light skin, even after adjustment for socioeconomic status, stage at diagnosis, and treatment, which is likely attributable to those with dark skin having lower 25(OH)D levels [Grant, Peiris, 2012].
To maintain 25(OH)D in the range 40-60 ng/ml may require doses of 2000-5000 IU/d of vitamin D3 or more, depending on many factors.
Both scientists and physicians claim to have a scientific approach, considering a broad range of evidence. For example, the use of Hill's criteria for systematizing evidence on causality in a biological system [Hill, 1965] can include strength of association, consistency, temporality, biological gradient, plausibility, coherence, experimental evidence (e.g., RCT), and analogy. These criteria have been examined for vitamin D and cancer with the finding that they are generally satisfied [Grant, 2009; Mohr, 2012]. Physicians, however, are accustomed to judging pharmaceutical agents on RCT findings. However, RCTs of vitamins or other essential nutrients cannot be judged as if they were medical agents, since, as we have seen above, this is usually inappropriate. In addition, medical training rarely provides much information on nutrition. Time for physicians to invest on basic sciences may be limited, so many have little incentive to read more than the abstracts of even quite important papers. Thus, to those working on nutrition, and in particular those working on the health benefits of vitamin D and on identifying the many mechanisms of action of hormonal vitamin D, the medical system appears to view vitamin D almost as a threat to clinical practice, perhaps even to health care income generation. Further, the persistent disregard of the known actions of vitamin D3 and its potential health benefits suggests that the well-known methods for discouraging acceptance of change, as identified in the 'Disinformation Playbook' are being used deliberately to discourage the provision of vitamin D for the large numbers of people with deficiency. That would not be tolerated for other basic nutrients such as iron or calcium [Grant, 2018].
To bridge the gulf between the scientific and medical communities regarding vitamin D and cancer may need a well-designed vitamin D3 clinical trial, based on measuring baseline and achieved 25(OH)D, and supplementing individuals with various vitamin D3 doses as required to achieve repletion, possibly at up to 5000 IU/d. However, physicians can, and often do, add vitamin D treatment to their management of cancer treatment, and individuals can still take personal supplements of vitamin D3.