High-Protein Diets Linked to Cancer: Should You Be Concerned?


    Protein is as dangerous as smoking?

    A recent study on high protein has been making the media rounds lately, with some news sites using headlines such as “Eating large amounts of meat, cheese may be as deadly as smoking.”

    That’s quite the claim. As is our habit, we decided to delve more deeply into the study and its actual findings (we’ve previously covered fish oil and your prostate and the truth about vitamin D).

    Note: We’re about to get science-heavy. You can scroll to the bottom for the takeaways. We’ve also summarized important points along the way.

    Lets look at the study

    The study itself was part epidemiological and part rodent intervention, and can be found here. In assessing the role of IGF-1 in cancer and aging, they took data from a pre-existing survey (NHANES III; one of the largest American-based national surveys of health and nutrition), assessing 6381 adults over the age of 50.

    After getting baseline data on protein intake (averaging 16% of caloric intake; 69% of this being animal-based protein), they then divided people into three groups: the highest third (above 20% protein), lowest (below 10% protein), and a middle group (so they could compare the categories).

    After dividing everyone into one of these three groups, they looked at the data for these people over the course of the next 18 years to see if mortality or disease in any of the groups differed.

    The individual pieces of data they found from dietary analysis were:

    • A positive correlation between moderate and high protein intakes and diabetes-related mortality, relative to the lowest intake. This persisted, albeit to a lesser degree, when looking at people over the age of 65.
    • No relation between higher protein intake with all-cause mortality, cancer-related mortality, or cardiovascular mortality overall. A small increase in risk was seen when looking only at people between the ages of 50-65. This risk was reversed for people above the age of 65, where dietary protein had a protective effect against all forms of mortality (excluding diabetes-related).

    When investigating IGF-1 levels (this is separate because the sample is now reduced to 2,253) they found:

    • The increases in IGF-1, which correlated with meat intake, predicted overall mortality seen in the 50-65 cohort as well as the 66+ cohort.
    • The protective effects of protein intake in the elderly (66+) persisted if IGF-1 was lower, but higher levels of IGF-1 were not associated with protective effects, although they were not associated with increased mortality either.

    The controls in place to reach this conclusion were restricting the age to people over the age of 50 only, and after assessing protein intake, they controlled for overall calories and other macronutrient intake as well as the sources of protein (either animal or plant). Further controls mentioned were age, race/ethnicity, education, sex, disease status, smoking, dietary changes, and attempted weight loss.

    NHANES III gets its nutritional data via 24-hour dietary recall.

    A quick note on epidemiological studies - they take data after-the-fact and try to make connections. This is not the same as a randomized clinical trial. This is a very important distinction to make, as epidemiological information cannot be used to state that “A causes B.” It is meant to spur further research.

    The above bits of data are all that was available from the epidemiological evidence. A separate rodent study was then conducted to attempt to verify causation. It found:

    • When feeding mice either high (18%) or low (4-7%) protein diets for one week prior to tumor implantation (B16 melanoma cells) and throughout the study, it was noted that the majority of mice in both the high (100%) and low (90%) protein groups bore tumors but the size of the tumors in the low protein group was smaller (78% smaller relative to the high protein group).
    • They did find elevated IGF-1 levels in the high protein group and higher levels of its inhibitor protein (IGFBP-1) in the low protein group.
    • When doing the study again in mice highly deficient in growth hormone (due to their genetic composition), tumor growth associated with high protein was decreased (estimate from graph is to around 10-20% of what was seen in normal mice given that dose of protein)
    • When replicating the study with breast cancer cells rather than B16 melanoma, they found similar results.
    • Replacing animal protein (casein) with plant (soy) did not change the results around much, and serum IGF parameters were similar.
    • When investigating weight (not tumor implantation) into older mice and assessing their growth, it was noted that the expected negative effects of high protein were turned positive

    There do not appear to be any significant listed conflicts of interest in this study, with the only disclosed affiliation being the corresponding author (one author having equity in L-Nutra; a provider of medical foods). Funding was provided by NIH/NIA which had no role in producing the methodology.

    What does this tell us?

    First, it should be mentioned that to fully appreciate this study we must view it as two studies. There is an epidemiological study and there is a mouse intervention study; anytime tumor growth is mentioned, it refers to the mouse study, and causation can only be applied to the mouse study. It cannot be applied to the human study (as it is an epidemiological study).

    As such, we’ll address each one independently:

    Epidemiology study in humans

    This study noted a 2 to 4-fold increase in mortality. This may sound large, but is too small to be directly compared with smoking. That range in smoking has been noted in thorough analysis of studies when referring to overall mortality, yet it is still much less than the 20-fold expected increased risk for lung mortality. Not only do repeated studies on a topic refine and lessen the initial estimates (the first estimates are almost always too high, which are lessened with subsequent meta-analysis), but information on smoking controls biochemical confounds that predict mortality. This study did not. A direct comparison between the two, despite having similar risk ratios, is misleading.

    No participant was under the age of 50, and people were divided into either 50-65 or 65+.

    Beyond that, there do seem to be interesting results. There was an increased risk of diabetes-related death. This is unique, since the other mortalities and overall mortality were only higher in those between the ages of 50 and 65. For those older than 65, it was reduced. Mortality seemed to be associated with IGF-1 concentrations in all cases, and IGF-1 activity seemed to be negatively associated with the protective effects seen in those over 65.

    Higher protein seemed to mean higher mortality between the ages of 50-65. Once over 65, it actually reversed (unless you were diabetic)!

    In regards to the data collection and controls, it seems quite adequate. 24-hour dietary recollections from NHANES III are acceptable (self-reported dietary recollection will always be flawed to a degree) and this was supported by serum IGF-1 concentrations. It should also be stated that the sample size was reduced to about a third when it pertains to IGF-1 (due to not having measurements in everyone).

    Quite a few unanswered questions

    Many good controls were in place, but there are a few questions that readers may have that this study cannot address:

    • “Does exercise exert a protective effect?” - this study did not measure exercise.
    • “What about sources of protein?” - this study did not measure anything beyond whether it came from an animal or plant. Chicken breast is not comparable to processed bologna meat.
    • “Do these results apply to persons younger than 50?” - we don’t know
    • “What about eating your veggies and fruits?” - vegetable and fruit servings were not counted, just carbohydrates and fats.
    • “What did they eat in place of protein in the low protein group?” - protein as a percent of caloric intake was assessed, but not levels of carbs or fats (beyond group averages before division into thirds for analysis), so we don’t know if the macronutrient used to replace protein had a protective effect or not.
    • “Fast food versus home cooked?” - not measured.
    • “Could a healthy lifestyle completely circumvent these effects?” - perhaps, but the study never looked into ‘a healthy lifestyle’ beyond macronutrients
    • “What about weight loss?” - the study controlled for ‘attempted weight loss’, so it would be prudent to assume that no significant weight loss occurred in most subjects. Due to this, any questions pertaining to whether weight loss exerted a protective effect (usually does in other studies, if the weight loss occurs in obese persons) cannot be answered by this study

    Many questions arise from an epidemiological study, which is common. It is important to remember these that this is a function of how they are run, and thus why they cannot prove causation.

    Ultimately though, this study is good evidence to suggest that IGF-1 is not only playing a role in the aging process and mortality, but that it is also tied into overall protein intake, and this initially harmful effect may eventually turn protective in the elderly. It is a good study if you omit the grandiose claim of meat being as harmful as smoking.

    Intervention in mice

    This study proves causation in mice, and was used indirectly to support the role of IGF-1 in humans. The study was quite interesting in regards to tumor size (a 78% reduction in size from dietary modification is nothing to scoff at), but there are a few things to address before delving into the study:

    • It was investigating tumor growth, not mortality.
    • All the mice were implanted with tumors, which is totally different than what happened in the epidemiological study where all persons were cancer-free at baseline. This is important since IGF-1 is thought to promote growth of pre-existing tumors, not to cause them
    • Due to investigating tumors exclusively, this data can supplement the cancer-related mortality seen in the epidemiological study but doesn’t provide much evidence for the diabetes-related mortality or other types.
    • Any conclusion in the epidemiological study and the mouse study should be independent from one another, and if using one to support the other it must be clearly stated; the human component of this study do not state that protein accelerates tumor growth.

    With that out of the way, this mouse study noted that protein didn’t necessarily have a protective effect (the vast majority of mice, be it on low protein or high protein, experienced tumor implantation and the tumors stayed there) but it suggests an ‘anti-growth’ effect of lower protein intake or, conversely, a pro-growth effect of higher protein intakes. The type of protein did not matter, since soy also augmented tumor growth to a similar degree as casein.

    It demonstrated, perhaps via a correlation again, that tumor growth was associated with IGF-1 activity. This makes logical sense. IGF-1 is an anabolic agent to most cells, be they skeletal or tumor cells! While a GH inhibitor was not used, they did use a GH-deficient line which produces less IGF-1 and found significantly less effects of protein on tumor growth suggesting a causal role.

    Higher protein means more IGF-1. IGF-1 is an anabolic agent that causes cell-growth, regardless of type. Skeletal tissue (good) or tumor cells (bad), it does not differentiate.

    Other considerations for those that want to know more

    In regards to dietary protein overall, the general observation of increased mortality has been noted in other epidemiological research in healthy cohorts, although the magnitude of this increased risk is lesser than observed in this study, with a nonsignificant increased risk of 1.02 (95% CI of 0.98-1.07) and elsewhere a conclusion that increasing protein by a decile increased overall mortality risk by 2% (95% CI of -1% to 5%); to clarify, if the CI (confidence interval) crosses the 0 point the data is not seen as statistically significant.

    Interestingly, both studies noted that carbohydrates had a protective effect on mortality (specifically cardiovascular mortality), leading to a hypothesis that perhaps increased protein seen in this study is merely a proxy measurement of reduced carbohydrate intake; plausible due to food sources nor carbohydrate intake in the protein groups being measured, but at this time that hypothesis cannot be addressed with the study in question. If true, however, it would pinpoint protein as accidentally being a red herring in the mortality issue.

    Finally, nutritional epidemiology such as this is known to have larger effects whenever a study on a new topic (i.e. mortality in otherwise healthy people) is introduced which the magnitude being significantly reduced when subject to future meta-analysis. This refers to the 2 to 4-fold increased risk (hazard ratios) noted in the study though as the 23 to 73-fold risk being compared to smoking was statistically misleading (and noted by the authors as such), and when paired with the previously cited other studies on mortality it should serve to show that the degree of risk suggested by this study may be an overstatement.

    In single-study situations, it seems like everything we eat is associated with cancer. One must look at the full body of evidence.

    Any evidence that protein is good for diabetics?

    It is always important to look at the full body of evidence, not just one paper. To that end, as the paper showed that high protein was potentially an issue for diabetics, the question to ask is:

    Are there any studies done that directly look at diabetics and protein-intake?

    This leads us to a Systematic review and meta-analysis of different dietary approaches to the management of type 2 diabetes.

    high-protein diets are effective in improving various markers of cardiovascular risk in people with diabetes and should be considered in the overall strategy of diabetes management.

    This systematic review and meta-analysis showed that interventions with higher protein diets can actually improve glycemic control and cardiovascular risk in diabetics. This doesn’t prove that higher protein diets decrease mortality in diabetics, it does show discordance between epidemiological and intervention studies. It does remind us to keep an eye on the bigger picture.

    Common questions that I need answered

    Should I stop eating meat?

    Whether you do or not, it shouldn’t be because of this study. If you eat a lot of heavily smoked and processed meat, it may be a smart choice to eat more slow cooked and non-smoked meat

    This study was also done on people over 50 years old. In fact, it found higher protein was beneficial for those older than 65 (unless they had diabetes). Unless you fit that bill, don’t be too concerned.

    Should I switch to plant based sources of protein?

    Not if you don’t want to. Beyond the hype you may have read, the actual data found that they were either equal (in mice) or that was slightly better (epidemiological).

    Should I be worried about IGF-1?

    IGF-1 causes cell growth. It doesn’t care if it it’s a muscle cell or an organ cell or a cancerous cell. This same scary-sounding hormone is what helps you build more muscle when you lift weights. Eating protein seems to increase IGF-1. If your underlying diet is poor (too much processed meat, not enough exercise), then high protein intake may be an issue.

    What should I know?

    Yes, dietary protein was implicated in mortality. No, it isn’t as harmful as smoking. The number which the comparison was borne from (diabetes-related mortality) does exist in the study and is immediately followed by:

    “We underline that our hazard ratios and confidence intervals may be inflated due to our sample size and the extremely low incidence of diabetes mortality in the low protein group. Overall, there were only 21 diabetes deaths among persons without diabetes at baseline, only 1 of which was from the low protein group”

    To even suggest that eating protein is as bad as smoking is pure sensationalism.

    Human nutrition is a complex topic, and isolating one area can be complicated. Often times, two disparate issues can be combined together incorrectly:

    Issue 1: as we explained, IGF-1 is responsible for cell growth, and it does not care what kind of cells it is growing. For those who exercise regularly, IGF-1 is a boon. For those that have poor diet without exercise, it is not.

    Issue 2: poor diet is a ubiquitous problem. Processed meats are a staple of poor diets. Processed meats also tend to have protein. The study made no differentiation between different animal sources. As previously stated, chicken breast is not comparable to processed bologna meat.

    When you combine Issue 1 and Issue 2 without looking at the bigger picture, one can mistakenly infer that “high protein is bad for you.”

    A more accurate headline for this study would have been “High protein for those between 50 years to 65 years old who have poor diet and lifestyle habits may be associated with increased cancer risk.”

    Want to know how much protein you should be taking? We looked at the scientific evidence on the recommended intake.