Scientific basis of "Starvation Mode"

There are a number of threads out there on “starvation mode,” but I thought it would be interesting to start one focused on individual scientific papers and experiments.

Here is the thread and part of the original post I’m responding to:

http://www.myfitnesspal.com/topics/show/1188952-advanced-calorie-deficits-fair-warning-long-post

- Whether you want to believe it or not, starvation mode is real
Many people may have the wrong idea of what starvation mode is, but that doesn't make it false, it makes their concept of it false. It's up to us as intelligent human beings to combat misinformation where ever we can. That means when someone spouts something wrong or silly about starvation mode. Speak up. But do it respectfully, and provide not just a response, but proof. There's plenty out there I can give you tons of studies if you need them. But just talking about the metabolic processes surrounding this state should be enough( no studies needed, these are facts about the body, not up for debate). What starvation mode isn't: it has nothing to do with 1200 calories, it doesn't happen after 1 day of not eating, and it is NOT starvation (two completely separate terms), starvation is a state, starvation mode is a process. Yes someone in prolonged, deep starvation mode can eventually starve, but except in the case of complete absence of food, starvation istself is extremely rare and someone giving you case studies on starvation to prove points on starvation mode are only showing their scientific ignorance.
Starvation mode IS, in it's simplest form, your body's metabolic processes recognizing, over the course of days or more likely weeks, that it's receiving far fewer calories than it needs. This triggers two major functions. First, when in starvation mode the body will reduce energy consumption wherever it can, starting with the least important (in it's opinion) functions first. Second the body will use any extra calories it can find and store them as fat. So when someone says starvation mode is a myth, please try to explain that it's not a myth, it's just difficult to prove, and you can't just blithely say someone's in that state without some serious analysis.

First, it is important to define what I mean (and what I think the poster above and people generally mean when they use the term, but it’s totally up for discussion).

Starvation mode is
A metabolic condition (a) whose degree correlates to the size of sustained energy deficit ((b) given adequate vitamin, mineral, and protein intake) that results in

(c) a decrease in basal metabolic rate (BMR) and
(d) an increased propensity to store calories as fat ((e) that is not mediated by behavioral change),

and both effects (f) persist to some degree after switching to a maintenance caloric intake and are beyond which that can be predicted by known factors, including (g) current and (h) previous body composition.

Explanation:
a. Energy deficit meaning (daily Calorie intake)-(total daily energy expenditure(TDEE)). I’m not specifying whether this is achieved by higher exercise or lower intake to keep things broad. I think most people tend to think of it being caused by low intake. By “sustained” I mean more than at least a week or so.
b. These can cause metabolic problems on their own that I don’t think are the concern about “starvation mode”
c. BMR is what people think of as “metabolism”- how many calories per day you burn by sitting still. I think the fear is that this drops significantly in “starvation mode”
d. This was the second claim in the original post, and it seems to be common thought
e. The idea of starvation mode is that metabolism is low and fat sticks, despite diet and exercise. This is why we use MFP- to keep track of and control intake and output.
f. People are generally worried about “starvation mode” causing lasting damage, and (d) makes no sense (there aren’t extra calories to store if you are at a deficit) without this caveat
g. BMR will decrease with weight loss simply because there is less tissue to maintain (as in, people with more fat free mass (FFM) and fat mass (FM) have higher BMRs in general)
h. This is important. In other words, starvation mode effects are not caused by being formerly obese, but how the weight was lost. The appropriate comparisons to make are between subjects currently of the same weight/body composition who lost the same amount of weight either quickly at a high energy deficit or slowly at a moderate energy deficit. Otherwise, you are looking at “used to have more fat mode,” not “starvation mode.”

Now the science. First of all, scientist don’t usually talk about “proving” or “disproving” something: we say things like “Our data support” or “Our data are inconsistent with” or “Our data provide evidence for” etcetera. This is because experiments are really messy. There are the statistical issues of being able to test very limited number of subjects, the high variability across people, inaccuracies of estimation in measurements, etc. More importantly, they are only doing one (or a couple) variations of an experiment with a specific population- meaning the best you can say for sure is that under those particular conditions, your hypothesis held. Scientists are very careful not to overgeneralize, without losing sight of the big picture.

Also, when reading scientific papers, keep in mind that these are written by experts for other experts, and they might use different definitions of words than we do (eg. metabolism) which is part of why it is important to look at what their methods were, what analysis they did, and look at the figures, understand how they are getting the numbers, etc, instead of just reading the title, introduction and discussion. These are generally there for flavor and to sell the paper to the reviewers.

Finally… don’t interpret this as a criticism of these papers. They aren’t out to “prove” starvation mode (which is really a dieting concept and typically not something the researchers are primarily concerned about). I’m just looking for data that is consistent or inconsistent with the existence and degree of “starvation mode,” as defined above. I’m not giving a complete summary of their findings- just what is most relevant to starvation mode, as described above. Some of them seem, as far as I can tell, irrelevant, and it would help if the original poster could explain what he found convincing about them (or I misunderstood what claims he was trying to support with them).

These are papers people posted in the original thread that they believed supported starvation mode:

Redman, Leanne M., et al. “Metabolic and Behavioral Compensations in Response to Caloric Restriction: Implications for the Maintenance of Weight Loss.” PLoS ONE 4, no. 2 (February 9, 2009): e4377.
This study was described in some detail in the original thread (p 4), to which I will respond here. It looks like there was some misunderstanding of what they were measuring and what numbers they were reporting. This paper primarily concerns “non-exercise activity thermogenesis,” which is the calories expended in every-day muscle activity like standing up, fidgiting, etc, while living in the real world. This number is larger in heavier people, because it takes more energy to support a body weighs more. Sometimes when they say they are correcting for FFM and FM, this is what they are doing.
The estimates posted in the original thread (from table 2 of the paper) were derived like this:
They estimated/measured:
TDEE- at baseline, titrating the amount of food to maintain body weight while living at home, doing daily activities
TDEE- estimated during intervention (and baseline) by labeled water (fed to subjects and measured in urine) over 2 weeks. Importantly, they are off in the world, not in a box (this was one of the selling points of this study in the introduction). The TDEE’s they measured this way are based on 2 whole weeks, and includes all the movements of daily life.
Body composition (FM and FFM) at baseline and during experiment
24h-EE- 24 hour sedentary energy expenditure- had people sit (and eat meals) in a room for a day, estimated energy output by measuring oxygen and carbon dioxide (“whole room indirect calorimetry”)
SMR- sleeping metabolic rate. Their estimate of basal metabolic rate, calculated by whole room indirect calorimetry when they were asleep and still
Other information- sex, age.
They built a model to predict TDEE based on age, sex, weight, FFM, FM, 24hEE, and SMR using data across the whole population at baseline. They predicted the TDEE’s based on this model and compared it to what they actually measured from the subject’s urine.
The numbers in Table 2, displayed in figure, are differences between the measured TDEE and predicted TDEE, based on their equations. A person in the LCD category could chose to do more sedentary activitites, like stay home and watch TV rather than more active ones, like shopping, which involves more movement, and that would be reflected in this number! [It also includes the exercise assigned to the CR+EX category- which was 12.5% of baseline calories. Quick calculation- this comes to about 330 Calories burned in exercise- so the TDEE (excluding the assigned exercise) for the CR+EX group is more on par with the CR alone group.]
What causes this difference? Well it’s not a drop in SMR (a standin for BMR, which is what drops in starvation mode). In fact, they are correcting for SMR (as an approximation of FM and FFM, “metabolic body size”) to show that CR/LCD’s cause the “metabolic adaptation” of reducing propensity to move around in daily life.
These numbers are not evidence “for” or “against” starvation mode as defined above, because it is behavioral (and this is what the authors are going for, anyway- see title). They are comparing the energy expended in free-living subjects, not their BMRs.
So what does this paper have in it to support starvation mode? Well, first of all, they don’t show enough raw data to make the comparison between the CR and the LCD (which is the more correct comparison to look for effects of starvation mode). However, they referenced a paper that used the same data and did:

Heilbronn LK et al. “Effect of 6-month Calorie Restriction on Biomarkers of Longevity, Metabolic Adaptation, and Oxidative Stress in Overweight Individuals: A Randomized Controlled Trial.” JAMA 295, no. 13 (April 5, 2006): 1539–1548.
They predicted SMR (based on current FFM (plain old linear regression) using the SMRs and FFMs from baseline) and compared it to measured SMR after 3 months of CR (25% less than TDEE at baseline), LCD (850 Cal/day), or CR+EX (12.5% less than TDEE at baseline and that same amount of calories in exercise). And what did they see? Table 2.
CR- 8% less than than expected (123 Cal)
CR+EX- 5% (78 Cal)
LCD- 3% less (44 Cal)
These numbers are where they got the “6% metabolic adaptation” referred to in the first paper. So this does not support Starvation Mode, because the group that was the most restricted (LCD) did not have a significant drop in BMR on its own or compared to more moderately restricted controls (beyond what is expected by changes in body composition).

Dulloo, A. G., and J. Jacquet. “Adaptive Reduction in Basal Metabolic Rate in Response to Food Deprivation in Humans: a Role for Feedback Signals from Fat Stores.” The American Journal of Clinical Nutrition 68, no. 3 (September 1, 1998): 599–606.
This is a reanalysis of the data from the “Minnesota Starvation Study” from 1950- thanks for posting, this was interesting data to chew on. What they found was that after 24 weeks (6 months) of eating about 50% of what they normally would (subjects went from healthy BMI’s to underweight), BMR dropped beyond which that can be explained by FM and FFM by about 25%, which, for these subjects, was 350 Calories/day! They call this “thermogenic economy.”

Before you think “aha, starvation mode!,” note they weren’t comparing those who lost gradually with those who lost quickly- the purpose of this study was to figure out how best to nourish the starving- the “semistarvation” part was just to achieve a malnourished state, and the experimental variables were the diets they were fed to renourish them. The authors of this paper did, however, look for a correlation between the actual caloric deficit (which varied somewhat among subjects) and thermogenic economy. They did not see it- as in it does not support “starvation mode” as caused by caloric deficit! But it does support (and I believe it was the original evidence for) the idea that losing weight can change your BMR above and beyond that which is expected by changes in body composition.

What’s interesting is why this number (25%) is so much larger than the 6% in the study above (Heilbronn et al). For the sake of argument, I’m going to assume this is a good estimate (even though I suspect that they may have more error than in other studies because they are extrapolating farther outside of the data they used to generate the equation to predict BMR. The authors themselves found the data at this time point unreliable for further statistical analysis, because of errors in body composition due to the extreme emaciation, edema, etc, and used the halfway-point data (average 20% thermogenic economy) instead).

I think the answer might be in the population they were studying. They started with people at a healthy weight/body composition (average 150 lbs, 16% body fat, BMI 21.7), and ended with people who were emaciated. Halfway through the study, they had lost, on average, more than 50% of their body fat (eyeballing figure 2- some had already lost nearly 100%). Whereas in Heilbronn et al, where they saw a much more modest thermogenic economy, the subjects were overweight (180 lbs, 30% fat, BMI 27.7, both men and women) and lost ~24%-32% of their fat mass.

Indeed, the central point of this paper is that the % reduction in fat stores (and the related statistic, starting body fat mass) were the variables that best correlated with the degree of “thermogenic economy.” This correlation persisted during “refeeding,” when the subjects were on experimental diets (surplus calories) to return them to a healthy weight. As in- it was not how much they were currently or previously eating/expending, but how much of their previous fat stores they had lost, that best determined the amount of “thermogenic economy.” Cool.

So in summary- is not consistant with starvation mode defined above, because the degree of thermogenic economy was not correlated to the calorie deficit. It does, however, support a reduction in expected BMR based on previous fat stores.

MacDonald, I A, and J Webber. “Feeding, Fasting and Starvation: Factors Affecting Fuel Utilization.” The Proceedings of the Nutrition Society 54, no. 1 (March 1995): 267–274.
This is a review article. They are describing what happens to the body to generate energy during fasting or underfeeding. They reference the “Minnesota Starvation Experiment,” described above. One of the things I learned here is that there seems to be that a fraction of BMR is variable and under the control of the sympathetic nervous system. As in- noradrenaline can increase BMR by up to ~11% (figure 3). Chronically underfed people (physically active laborers at 1800 Cal/day), underweight people, and normal weight people all have the same maximal response, but the chronically underfed people were less sensitive… in other words, they would have to drink more coffee for the same bump.

Hill, J. O., P. B. Sparling, T. W. Shields, and P. A. Heller. “Effects of Exercise and Food Restriction on Body Composition and Metabolic Rate in Obese Women.” The American Journal of Clinical Nutrition 46, no. 4 (October 1, 1987): 622–630.
They are making a comparison between obese women (only 8 of them total) who are on a very low calorie diet (800 Calories) with and without aerobic exercise. This isn’t the right comparison to make to look for evidence in support of starvation mode. Yes, they measured BMRs and saw a drop along with weight, but they did no corrections for changes in body weight and body composition (also I’m loving the comically large figure 2).

Cahill, G. F. “Survival in Starvation.” The American Journal of Clinical Nutrition 68, no. 1 (July 1, 1998): 1–2.
This is an editorial about how the body produces energy from fat and muscle tissue. The paper it refers to is about protein loss (from muscle) in literally starved obese subjects. Most of the energy from broken down fat tissue cannot be used by the brain (which needs some glucose), so during fasting, after the glycogen stores are used, the body breaks down muscle tissue for protein (which can be used to generate glucose). This is not relevant to starvation mode, which is from a sustained calorie deficit, not starvation.

Zauner, Christian, et al. “Resting Energy Expenditure in Short-term Starvation Is Increased as a Result of an Increase in Serum Norepinephrine.” The American Journal of Clinical Nutrition 71, no. 6 (June 1, 2000): 1511–1515.
This is about literal starvation- fasting- on the short term (5 days). Again, not relevant to starvation mode.

Greenberg, Andrew S., and Martin S. Obin. “Obesity and the Role of Adipose Tissue in Inflammation and Metabolism.” The American Journal of Clinical Nutrition 83, no. 2 (February 1, 2006): 461S–465S.
This is a review. The main point is that we should be thinking about adipose(fat) tissue as serving important roles in the body besides just fat storage. You are going to have to explain what is convincing about this as far as starvation mode. It’s starting to look like you just posted a bunch of papers that were related to the post in general, and not actual support for starvation mode… I may have totally misread/misunderstood
.
Alemany, Marià. “The Defense of Adipose Tissue Against Excess Substrate-Induced Hyperthrophia: Immune System Cell Infiltration and Arrested Metabolic Activity.” The Journal of Clinical Endocrinology & Metabolism 96, no. 1 (January 2011): 66–68. doi:10.1210/jc.2010-2541.
Review article/editorial. This is about metabolic syndrome and adipocytes. It has nothing to do with caloric restriction, weight loss, or fat gain after weight loss- they do talk about weight gain- how fat cells can become insulin resistant when exposed to a constant energy surplus (overeating). The “arrested metabolic activity” in the title refers to the metabolic activity of the fat cells themselves (i.e. converting glucose fat) when they are pushed to their limit.

And here are some papers I posted in the last thread:

http://www.ncbi.nlm.nih.gov/pubmed/23404923
Müller MJ, Bosy-Westphal A. Adaptive thermogenesis with weight loss in humans. Obesity (Silver Spring). 2013 Feb;21(2):218-28.
The authors say the data are still ambiguous to what extent adaptive thermogenesis (changes to the resting energy expenditure as a result of underfeeding) exists, when corrected for loss in fat-free mass and fat mass. Some studies see it, some don't, some see the opposite effect, and some are using different definitions.

http://www.ncbi.nlm.nih.gov/pubmed/19710198
(Same group as above) Contribution of individual organ mass loss to weight loss-associated decline in resting energy expenditure. Am J Clin Nutr. 2009 Oct;90(4):993-1001.
This one you can read for free. They put obese women on very low calorie diets (800-1000 Calories) for 12 weeks and measured changes in their body composition and resting energy expenditures. The amount of adaptive thermogenesis that was not explained by loss in organ/fat/muscle tissue was on average about 50 Calories per day. 50 Cal from the fairly extreme restriction here doesn't seem like a huge concern to me, certainly not worth all the hoopla that we give it.

Given, in these studies, and any study with human participants, the subjects are given diets carefully designed to be low in energy but nutritionally complete. Maybe what you are talking about has to do with people who are not getting enough of certain micronutrients? Protein?

http://www.ncbi.nlm.nih.gov/pubmed/1615886
Prentice AM, Jebb SA, Goldberg GR, Coward WA, Murgatroyd PR, Poppitt SD, Cole TJ. Effects of weight cycling on body composition. Am J Clin Nutr. 1992 Jul;56(1 Suppl):209S-216S.
If very low calorie diets cause rebound fat storage, you would expect that "yo-yo"-ing would cause a net increase in fat, decrease in lean body mass, right? That's not what these authors found in a population of people in a time and place in the world that had annual "hungry seasons." Their body composition as they aged was the same as people who did not have periodic extreme calorie restriction.

http://www.ncbi.nlm.nih.gov/pubmed/21677272
Maclean PS, Bergouignan A, Cornier MA, Jackman MR. Biology's response to dieting: the impetus for weight regain. Am J Physiol Regul Integr Comp Physiol. 2011 Sep;301(3):R581-600.
Also a review that is freely available. Looks like the biochemical signals from the newly-shrunk fat cells are because they used to be bigger. As in, it isn't because of recent calorie restriction directly (what you call "starvation mode"), but because of the fat loss itself (regardless of how it was lost).

Seems to me, if wanted to test if cessation of a very low calorie diet caused preferential fat deposition, one way to do it would be to compare two groups of people- one who ended up in "starvation mode" from severe caloric restriction, and one who lost the same amount of weight otherwise (slower, more exercise, etc), matched for activity level and diet. Then after a period of time, measure their body composition.

I haven't seen an experiment like this- what I did find was:
http://www.ncbi.nlm.nih.gov/pubmed/11010936
Weyer C, Walford RL, Harper IT, Milner M, MacCallum T, Tataranni PA, Ravussin E. Energy metabolism after 2 y of energy restriction: the biosphere 2 experiment. Am J Clin Nutr. 2000 Oct;72(4):946-53.
People were on a ~1800 Calorie diet for 2 years. Which doesn't qualify as "starvation mode," at least according to MFP. They lost ~15% of their body mass. Six months afterward, free to make all their own food choices, they gained the weight back, but it was almost all fat.

So, yeah, it looks like any kind of rapid weight gain results in preferential fat gains, particularly if there used to be more fat tissue at some point. Fat is just easier and quicker to grow than muscle.

There are more papers that were posted on the last thread that I will read and add in the near future. I hope somebody else finds this as interesting/enlightening as I have!

if you want to look for that in scientific papers, you need to look up adaptive thermogenesis

I did look up and read a review on "adaptive thermogeneis" (it's in the section from the old thread), and time permitting, I may track down and read some more of the references in it.

Good point- I agree. I'm definitely not advocating very low calorie diets (unless under medical supervision). People trying to cut calories to extremes on their own are likely to be missing nutrition, not just calories, which can be seriously damaging. One of the papers above points to a propensity to move less caused by low calorie diets, and I bet if I looked I could find data supporting increased food seeking behavior as you are describing it. But when people on MFP think they may be in "starvation mode," they aren't blaming it on too much food or too little exercise, so the point here was to try to figure out if there is a change to BMR. But feel free to post papers about the effects of a low calorie diet on free feeding behavior and locomotion- it's interesting on its own right.

I would also suggest that, while AT is seen in some experiments and absent in others, it does not mean it doesn't exist rather that we don't understand why it is present in some situations.

I agree- that's a better way to put it. I started reading with the idea that there was a consensus about it- that there was a consistent, measurable effect, and I was surprised that researchers weren't able to observe it where they thought they would. I meant to point out that the effect was not as clear-cut as the OP asserted.

Thanks for the info on BAT- the most I ever learned about it was that it's most important in infants and not present in significant amounts in adult humans, but things always get over simplified in introductory classes, and it looks like there have been some advances since then.