Article Discussed: Fasting-mimicking diet and markers/risk factors for aging, diabetes, cancer, and cardiovascular disease
Science Translational Medicine, Valter Longo et al., University of Southern California, 2017
The One Rep
The fasting-mimicking diet (FMD) is good for the general healthy population, decreasing body weight and waist size, but it is outstanding for unhealthier populations with beneficial effects lasting at least three months.
The Warm Up- For Everyone
What is the fasting mimicking diet?
The FMD is a plant-based restriction diet (See dietary overview page for differences between types of diets) that lasts for five days. This diet is typically repeated once a month to once every six months. This paper focused on completing the diet once a month for three months. The allotted calories and macronutrients for this FMD are 1100 kcal (11% protein, 46% fat, 42% carbohydrate) on the first day and 720 kcal (9% protein, 44% fat, and 47% carbohydrate) on days 2-5. The macronutrient ratios are just as important as the calorie restriction as this is a low protein diet that suppresses insulin-like growth factor 1 (IGF-1). Simplistically, it is believed to be important to cycle IGF-1 for maximum beneficial effect. This topic will be covered in a future write-up. The participants in this study used L-Nutra’s prepackaged food that conforms to FMD.
What impact does the diet have on weight and lab values?
For a healthy population: after three rounds of monthly FMD, weight decreased by 2.6 kg (5.7 lbs) and waist circumference decreased by 4.1 cm (1.6 in). For more unhealthy population, the diet decreased BMI by 1.4, blood pressure by 6.7/ 5.5 mmHg, fasting glucose by 11.8 mg/dl, triglycerides by 25.6 mg/dl, LDL-C by 14.9, and CRP by 1.7.
How long do the effects last?
Many variables continued to be improved three months after the last FMD was completed. Sustained weight loss of -1.4 kg +-2.5 were observed along with lower fasting glucose and blood pressures, especially among unhealthier subjects. It is possible that some subjects may have changed to a healthier diet during this time, thus skewing the data to overstate the long-term effects of the diet.
How “easy” is it?
The drop out rate was 25% during the FMD portion of the study and 10% during the control (normal diet) portion, in which people could eat their regular diet. This is comparable to the dropout rates of other diets. However, out of 100 people consented for this study, only 6 dropped out due to nonadherence or dislike of the diet. Keep in mind that these are highly motivated people that signed up for the diet in the first place.
How safe is it?
First, the authors state that any fast lasting longer than 12 hours should be done under the supervision of a health care professional, preferably one that is familiar with fasting. More studies are needed for safety in subjects who are 70 years and older. I The three most common side effects in this study were minor fatigue (~50%), weakness (~40%), headaches (~25%). There was no impairment of kidney or liver function for any participant.
How does the FMD work?
There are many mechanisms by which intermittent fasting, of which FMD is a subtype, works to improve health that may not be directly measurable by decreases in weight or lab values. See the general comments of the deep cut section below to read more. These claims will be discussed in a future review.
Maximum Effort- For Enthusiasts
The best diet is the one that is used. This study adds yet another tool to the toolbox of a physician or health enthusiast to increase healthspan. It is difficult to compare this diet to other diets because the control group is almost always the Standard American Diet, or SAD, in randomized control studies of diets and any diet is better than the SAD. That being said, I am most excited not by the promises of decreased weight by this diet, but by the numerous positive effects intermittent fasting has on health. Intermittent fasting is associated with up-regulation of CREB, BDNF, and autophagy pathways . There have been mouse studies that have demonstrated reduction of symptoms in autoimmune diseases such as multiple sclerosis via reducing pro-inflammatory cytokines, inducing lymphocyte apoptosis, and oligodendrocyte regeneration along with increases in cognition [2,3]. Additionally, cancer incidence and time to mortality decreased in mice on the FMD . Mice on the FMD had a median increased lifespan of 11%. Mouse studies are great for directing future research but they are terrible for basing interventions on. However, the FMD is a low-risk diet intervention that has massive upside when compared to other diets with proposed physiological mechanisms that just make sense. These claims are will be thoroughly scrutinized in a future review.
Magnitude of effects
As the figure 1 demonstrates, completing 3 rounds of the FMD over 3 months has a beneficial effect on weight, BMI, fat volume as measured by a DEXA scan, IGF-1 levels and systolic blood pressure, but many important disease risk factors are statistically similar to the control group. This similarity is expected as the volunteers for this study were generally healthy and an exclusion criterion was any major medical conditions like hypertension, hyperlipidemia, or diabetes. Essentially, this proves your CRP can only decrease by so much when you’re already healthy. This fact motivated the authors to conduct a post-hoc analysis on the data to stratify patients with higher disease risk factors and identify the FMD’s effect on those individuals as seen in figure 2. For example, the investigators split the systolic blood pressure risk factor into two groups based on their baseline blood pressure; those with systolic blood pressure <120mmHg went into the low risk group and those with systolic blood pressure >120mmHg went into the high risk, or post hoc, group. The cut offs for fasting glucose was >99mg/dl, triglyceride was >100mg/dl, and total cholesterol >199 mg/dl for LDL-C.
Figure 2 above uses the post-hoc data and demonstrates the efficacy of the FMD for patients at higher risk of metabolic disease verses other interventions such as diets and drugs. It is incredibly difficult to compare these different interventions as all the other intervention subjects were significantly unhealthier than the FMD diet subjects. For example, the HCTZ study had subjects with hypertension, the metformin study had subjects with diabetes mellitus, and the HLF/ HLC dietary study had a much higher baseline weight than the FMD study. The FMD study, on the other hand, had no major medical diagnoses even in the post-hoc group. The other studies were included to give you passing knowledge of how the FMD stacks up versus the competition. I think the FMD holds up extremely well when considering that the FMD study’s unhealthiest subjects were healthier in general than these other studies, and thus they had less potential for improvement. I can’t wait for the inevitable FMD-in-hypertensive-diabetic-hyperlipidemic subject study.
Note that post-hoc data analyses are frowned upon in the scientific community. Generally, researchers agree on data analysis methods before the collection of data in order to reduce the risk of bias in a study. Post-hoc data analyses are devised after data is collected. This is suboptimal as two statistical tests are run on one set of data, thereby increasing the risk for type I errors, also known as false positives. Additionally, researchers could examine their own data and develop statistical methods on a subset of that data to create results that would not exist in the original data set. I think the post-hoc analysis in this study was well motivated and the cut off values were appropriate, thus minimizing these risks.
Risks of bias
Response Bias: Unlike other dietary studies, no dietary surveys were given and thus the risk of response bias is minimal. This is incredibly rare in dietary studies and is a major advantage of this study.
Performance Bias: Performance bias is due to either researchers giving extra care to the intervention group or subjects differentially changing their lifestyles during the study. For example, researchers who want the intervention group to succeed may spend more time with them during follow up visits than the control group. For a subject example, subjects may increase exercise during the intervention group. Subjects changing their behavior because they know they are being intervened on is also known as the Hawthorne effect. These biases can be reduced with double blinding both researchers and subjects. This is incredibly difficult in dietary interventions and is almost never done. Analysis of dietary studies always needs to be conducted with these facts in mind.
Conflict of Interest: The lead investigator, Dr. Longo, and another investigator have equity in L-Nutra, the company that provides the prepackaged FMD. Additionally, the University of Southern California receives royalties from this company. Dr. Longo and the other investigator recused themselves from the data gathering and analysis section of this paper. One hundred percent of Dr. Longo’s equity in L-Nutra will be donated to a nonprofit organization. I don’t believe this caused any bias.
Chance Bias: Chance bias exists when the baseline data of the control and intervention groups are not equivocal. The baseline characteristics between the arms were similar except for total cholesterol and LDL-C, which were lower in arm 2 (controlà FMD). This might have made it more difficult for there to be a statistical difference in cholesterol and LDL-C between the control group and the FMD.
Attrition Bias: Attrition bias exists when subjects unequally drop out of a study between the two arms. For example, subjects who are not experiencing results from a diet may get frustrated and differentially drop out when compared to the control group, thus creating data that may make the diet intervention group look better than it actually is. For this study:
Arm 1: N=48 randomized to control diet for 3 months then FMD for 3 months
5 [10%] withdrew during control diet, most commonly for scheduling and personal 10% dropout
11 [25%] withdrew during FMD diet, most commonly for scheduling and diet noncompliance (4 subjects)
16 followed up three months after last FMD
Arm 2: N=52 randomized to FMD for 3 months
13 [25%] withdrew during FMD, most commonly for scheduling (6) and personal (5) issues
34 followed up three months after last FMD
Dropout rate was similar overall between the two arms, thus minimizing attrition bias. The 3-month laboratory follow up for arm 1 was significantly worse. This may be due to a total of 9 months needed for the trial in arm 1 (3 months control + 3 months FMD + 3 months for follow up) while arm 2 was only 6 months. There is a risk of people who were not seeing beneficial results of the diet dropping out at a higher rate than those that were seeing effects. This would increase the effect size of the diet.
Post FMD labs were compared between the arms to determine if the 10% that dropped out during the control phase of arm 1 had any impact on the intervention data. Ideally there would be no difference, which would show that the arms were completely similar. However, HDL and absolute lean body mass decreased more in arm 2. These effects were thought to be minor and the two intervention arms were combined for analysis.
Finally, 82% of the dropouts were female. It is not known why. Female health issues are one potential justification but no data was collected on this aspect.
This study is a randomized nonblinded crossover study. Inclusion criteria include generally healthy adult volunteers with ages ranging from 18-70. Exclusion criteria include major medical conditions. This study is unique in that only one arm underwent the control diet. That is, arm 1 started with three months of the FMD diet with subsequent completion of the study while arm 2 started with three months of control diet followed by three months of the FMD. That is, everyone in the study received the intervention but only half underwent the control diet for a ratio of 2 interventions:1 control. I believe this was reasonable as the arm that underwent the control diet had significantly worse follow up 3 months after the study concluded.