Contents - Previous - Next


Research relating to energy adaptation in man


W.P.T. JAMES*

* The Rowett Research Institute, Greenburn Road, Bucksburn, Aberdeen AB2 9SB, U.K.


1. General introduction
2. The Sukhatme-Margen hypothesis
3. Is energy balance regulated in man?
4. The time basis of energy regulation
5. Altered metabolic rate
6. Other Sukhatme analyses
7. Problems in testing the Sukhatme-Margen hypothesis
8. The reproducibility of metabolic rates in man
9. Adaptation to underfeeding
10. Overfeeding studies
11. Attempts to test the Sukhatme-Margen hypothesis(es)
12. Concluding remarks
References



1. General introduction


This paper deals specifically with the Sukhatme-Margen energy hypothesis set out in 1982. The hypothesis on the autoregulation of energy balance stems from the authors' earlier propositions in relation to protein metabolism (SUKHATME and MARGEN, 1978) and has to be taken seriously for two reasons. First, on a scientific basis we need to recognize that the regulation of energy balance in man is poorly understood and a proposition which claims to be revolutionary deserves serious consideration. The second reason is a political one, but of far greater importance than that assigned to it by many of the scientists involved in research on energy balance. Sukhatme and Margen have claimed that their hypothesis is of immense importance for food policy-makers; furthermore, the policy-makers themselves have attached great significance to it. Thus, the hypothesis formed an appreciable part of the argument in the 5th World Food Survey's assessment of malnutrition in the world (FAO, 1986). Indeed, the idea of "costless adaptations" to low energy intakes seemed so attractive that a novel way of determining a cut-off point in energy intake for categorizing "malnutrition" was devised. This led to a substantial reduction in the number of people assumed to be malnourished.

Those of us who consider that the Sukhatme-Margen hypothesis, in the light of current evidence, has some political features analogous to the proposition of an all-pervading deficiency of vitamin C in Homo sapiens promulgated by Linus Pauling are probably correct. Linus Pauling's views are of great moment because he is a renowned scientist. Similarly, Sukhatme exerts enormous influence in the sphere of food policy because he was an internationally-renowned statistician working in FAO. His statistically-based hypothesis linked to physiological insights from Margen produces a powerful combination from prestigious authors. To dismiss simply the political consequences of the Sukhatme and Margen theory is acceptable for "ivory-tower" scientists, but I consider this approach to be somewhat irresponsible. The challenge for those of us who wish to see an appropriate application of current scientific thinking is how best to proceed.

Is the problem one of persuading planners, economists and community nutritionists that current evidence is already available to refute the hypothesis? If so, why have our arguments failed so far, or are there scientific issues which still remain unresolved? As will become apparent, Beaton and Viteri, in organizing a PAHO/WHO Workshop on the problem (PAHO/WHO/UNU, 1987), judged that the best way to cope with this major issue was to develop a set of protocols for testing the hypothesis. The expertise of international authorities in the field of energy regulation, together with the nutritional involvement of Margen, was used to set out a series of proposals. Before considering these, however, it is appropriate to examine the Sukhatme-Margen hypothesis in some detail in order to ensure that it is given appropriate emphasis. As will become apparent, the Beaton-Viteri proposals deal only with one part of the proposition. This has scientific advantages but some political drawbacks.

2. The Sukhatme-Margen hypothesis


The Sukhatme-Margen hypothesis depends substantially on statistical analyses of the apparent interindividual differences in energy requirements, the proposition being that much of these differences actually depend on the variation in requirement within individuals rather than between individuals. A careful reading of their paper shows that they simply state that WIDDOWSON'S classic data (1947) or those from the EDHOLM (1970) study can be used to suggest that "the body regulates its energy balance by adjusting either intake or expenditure or both, and that consequently the requirements cannot be considered as fixed and equal to the habitual intake". Any dispassionate reading of the literature would allow all experts to agree on the first part of this statement, and the crucial misunderstanding only develops with the final phrase "and equal to the habitual intake". In this and other statements, one finds that much of the difficulty with the Sukhatme-Margen hypothesis simply relates to terminology, and it is their emphasis, later in the paper, on the potential flexibility of energy expenditure by changes in metabolic efficiency through the changing use of different metabolic pathways which seems to present such problems.

3. Is energy balance regulated in man?


Put simply, the Sukhatme-Margen hypothesis, in my view, is not a hypothesis but a statement of fact, if it is taken to signify that body energy is regulated by alterations in intake and/or expenditure. There is now abundant evidence that food intake is regulated in both children and adults (see JAMES, 1985) and that this regulation is apparent within 2-7 days of covert changes in the energy density of the diet. It also seems fairly clear that pre-menopausal women show a cyclical change in both energy intake and expenditure through the course of a normal menstrual cycle so that in addition to "regulation" in the classic sense there is a controlled cyclicity of energy balance. This seems true at least in young women. Given this cyclicity with intakes on average 10-20% higher in the luteal than the follicular phase of the cycle, we know that measurements of weekly food intakes, measured randomly in young women, will lead to an erroneously large standard deviation of intakes, if these intakes are meant to indicate habitual intakes, which I will take to apply to the average of the intake over "several" months. This latter definition may therefore neglect the changes in the average intake from season to season which could reflect, for example in developing countries, the response to a change in the physical demands in the home or work.

Sukhatme and Margen did not make their hypothesis sex-specific; most of their examples relate to studies conducted by others, e.g., by Edholm and Acheson on young men. Almost all of the studies on the control of food intake have been undertaken with young male volunteers, and there is clear evidence of appetite regulation. Thus, in terms of body energy homeostasis, it would not be surprising to observe autocorrelations evolving over time. Sukhatme and Margen neatly show that in ACHESON'S Antarctic subjects (1980) the SD obtained from weekly intakes measured repeatedly are too large for those expected if energy intakes were independent from day to day. They affirm that the customary weekly intake and output have variations of about 12%. This, to me, is evidence of food intake regulation, but Sukhatme and Margen, in noting the discrepancy in Acheson's data on body weight changes and cumulative energy imbalance, simply proceed to neglect error problems cited by Acheson and claim that the variable patterns of intake and expenditure have a rhythm consistent with and indeed characteristic of autoregulation.

In Table 1 are provided unpublished data on "large" and "small" eaters obtained recently by McNeill in Aberdeen. These two groups were selected after 36 young male volunteers had weighed their food for a week, i.e., using the same technique as that employed by WIDDOWSON (1947), EDHOLM et al. (1970) and ACHESON (1980). The groups were from the two extremes of intake but, when detailed studies were conducted in calorimeter chambers under conditions of controlled activity, there was little evidence of any metabolic difference between the groups. The physical activity index was calculated from activity diaries and the extensive literature, and analyses of the published cost of different activities prepared at the Rowett in Aberdeen for FAO as part of the new approach to calculating energy requirements. Only 16% of the different intake could be assigned to differences in time spent on physical activity, implying that the week's weighed food intake did not reflect the long-term energy needs of the men. Either "spontaneous" changes in food intake would lead to a regression to the mean - had intakes been measured for longer - or physiological controls were operating with temporary swings in intake to the more extreme values. Alternatively, the intakes were spuriously high or low because of a Hawthorne effect, whereby the investigation itself affected the result. What does seem apparent, if we accept the data as accurate, is that, as in Edholm's study, interindividual differences in energy expenditure are much less than anticipated. I favour the second of the three explanations for a variable intake, i.e., physiological swings in intake account for the wide range observed. If substantial metabolic intraindividual variation were possible, then we should have seen a greater range in energy expenditure.

Table 1. The energy turnover of young men classified as either small or large eaters



Small eaters n=5

Large eaters n = 5

Significance of Difference

Age

yrs

21.2

22.6

ns

Weight

kg

67.9

67.7

ns

Height

m

1.75

1.81

ns

% body fat*


13.9

10.3

ns

FFM

kg

58.5

60.7

ns

Daily energy intake

MJ/d

85

15.3

<0.01

Basal Metabolic Rate

MJ/d

7.20

7.38

ns


MJ/kg FFM/d

0.123

0.122

ns

24-h energy expenditure

MJ/d

10.59

10.71

ns


MJ/kg FFM/d

0.181

0.176

ns

Physical Activity Index as ratio to BMR


1.38

1.53

<0.05

* Percentage body fat estimated from the regression equations of DURNIN and WOMERSLEY (1974). Fat-free mass was estimated by subtracting the estimated fat content from total body weight. The estimates of the physical activity index and the data on energy intake relate to the free-living condition whereas the 24-hour energy expenditure data and basal metabolic rate are those obtained while subjects eat a fixed diet of standardized composition in amounts related to their measured basal metabolic rate. The calorimetric data are not those reflecting the response to the purported intakes of each individual.



4. The time basis of energy regulation


At this stage it is important to assess whether Sukhatme and Margen go further. Following a mathematical analysis of their own data on successive N balance, the authors point out that "although intake may not be equal to expenditure, even when averaged over one week, humans are probably in balance every day with varying intervals between peaks and troughs and varying amplitudes in daily balance" (p. 359, column 1, paragraph 2). This sentence contains two different ideas, the first of which is straightforward. Given a cycle of daily food intake and night fasting, it is clear that there has to be a time when intake and expenditure are equivalent and the person is in energy balance. The last part of the sentence should presumably have read "... varying amplitudes in daily imbalance" to signify the degree to which, on a daily basis, the difference between 24-hour intake and expenditure will vary.

The authors seem to conform with this reworded concept by claiming in paragraph 4 that it "follows from the evidence that humans possess a physiological regulatory mechanism for controlling appetite and energy expenditure". This subtle change is unwarranted, and the conclusion is in any case based simply on their interpretation of Edholm's data on recruits without regard to the wealth of other data on the regulation of energy balance. There is nothing intrinsic to the EDHOLM (1970) data, which demands that both intake and expenditure are regulated; so the authors begin to slide towards the proposition that energy expenditure must be regulated - a view which is the one which most observers take to be the main theme of the Sukhatme-Margen hypothesis. The authors then maintain that "regulation implies adaptation", a view disputed by many including WATERLOW (1985). The analysis in their Table 3 of the variance in mean balance, i.e., intake minus expenditure, averaged over two, three or more successive days is a new way of specifying that imbalances reduce with time, but this can not only include error measurements but would also apply if intakes, for example, were randomly variable from day to day. It does not necessarily imply that there is regulation of energy expenditure.

5. Altered metabolic rate


Sukhatme and Margen specify that there is "no absolute energy requirement for any day or period". This is self-evident if one chooses not to specify the degree of physical activity and food intake, but simply assesses free-living people undertaking a variable pattern of activity and eating. The authors' sentence seems to be more subtle, however, implying that people can change their metabolic efficiency so that, while engaging in similar activity and eating, they adjust their total energy expenditure. It is this concept which is tackled by Beaton and Viteri, because Sukhatme and Margen clearly prefer regulation of expenditure rather than intake as the mechanism for autocorrelations in energy imbalances (see p. 361, column 2): "it seems more likely that the body regulates its energy balance by varying the efficiency of energy utilization". They then proceed to propose that from their own studies on RMR and the thermic effects of meals the "cost of maintenance cannot be constant as assumed in the literature but that it is regulated".

6. Other Sukhatme analyses


SUKHATME and NARAIN (1983) have extended their analysis of the intraindividual variation in energy requirement and dealt with its implications. Again, they rely exclusively on Edholm's study and are concerned with a detailed statistical analysis, but the references cited include only the published account of this work, rather than a re-analysis of the available primary data. The emphasis in this later paper shifts to the genetic contribution to the control of physiological adaptation, but no novel analysis emerges. In another analysis by SUKHATME (1977), the emphasis is again on the Edholm study and is based on an analysis of variance of daily energy intake and expenditure over time. Unfortunately, as noted by Sukhatme, the Edholm data relate to simultaneous measurements of intake and expenditure in three nonconsecutive weeks, so it is impossible to assess the autocorrelation hypothesis directly. Again, much of the confusion may relate to the definition of a "physiological mechanism for controlling appetite and energy expenditure". The precise meaning of the hypothesis is unclear since the emphasis changes from regulation of intake to suggestions that the metabolic efficiency of the body may vary. An extensive discussion by Sukhatme about enzyme mechanisms for altering efficiency implies that he perceives changes in metabolic efficiency as a key feature.

It should be noted that the policy implications and indeed the basis for the statistical analysis of food supplies which Sukhatme has used to emphasize that households may be less malnourished than expected - these statistical analyses and policy implications have been heavily criticized by DANDEKAR (1982) on both statistical grounds and because of changes in definition and the basis for analysis.

7. Problems in testing the Sukhatme-Margen hypothesis


The principal problem in any attempt to test the Sukhatme-Margen hypothesis is to obtain an agreed specification of the hypothesis.

If the hypothesis is a general proposition about the capacity of man to change metabolic efficiency in response to a sustained change in intake, without changing body energy stores or spontaneous physical activity, then the hypothesis can only be tested with difficulty.

If the hypothesis is a general one, can it be tested simply in young men anywhere? There seems to be an emotional objection, for example, to testing this theory in North America and then applying the findings to Asia. In practice, the hypothesis is changed and the question becomes:

(a) Do women and children behave similarly?

(b) Does preceding childhood malnutrition affect the ability to alter metabolic efficiency?

(c) Do long-standing climatic and nutritional deficiencies alter the capacity or sensitivity of any adaptive mechanisms?

(d) At what point do modest changes in body energy mean that the hypothesis is rejected?

(e) Even if there are reductions in body energy stores and/or physical activity, does it matter? Are not behavioural changes, the ability to take initiatives, sense of well-being and capacity to co-operate in emergencies more important than a purist physiological approach to the sequence of events which occur when energy intakes change?

The assessment which follows considers only the general case of adaptation in normal healthy adult men.

Table 2. The reproducibility of 24-hour energy expenditure in adults studied under standardized conditions in calorimetric chambers or a suit

Subjects

Number

Calorimetric technique

Number of measurements

Time interval Days

Reproducibility

Reference

Difference

%

CV%



Direct

2



2.0


BLAZA (1980)

Male

8

Indirect

2

4-5



1.5

DALLOSSO et al. (1982)



Suit: direct

2

14



3.3

WEBB & ABRAMS (1983)

Male/Female

9

Suit: direct

2

<30



6.0

WEBB & ANNIS (1983)

Male

8

Direct

2

7



2.2

GARBY et al. (1984)

Female

10


3-6*




1.9

DE BOER et al. (1987)

Female

5

Indirect

2

5-7



1.3

MCNEILL et al. (in press)

* Three consecutive 24-hour measurements were repeated 2-24 months later.



8. The reproducibility of metabolic rates in man


Table 2 summarizes a series of studies, mostly conducted on young men, showing that, on a constant diet and with a regulated pattern of physical activity, there is an astonishing reproducibility in 24-hour energy expenditure. This does not, of course, disprove adaptive thermogenesis, but it does suggest that there is little evidence of cyclicity of energy expenditure in men. Furthermore, for "adaptive" changes to occur in energy output in response to the usual fluctuating daily energy intakes prior to the subjects being placed on a standardized diet, these changes must occur very rapidly within a day or two so that further changes were not discriminated on repeated calorimetry. Alternatively, the changes must occur over periods in excess of 2-4 weeks, the periods over which repeated observations have been made.

A recent study by DE BOER et al. (1987) refutes the concept of sudden rapid swings in energy expenditure in response to changes in intake. De Boer had people living in a whole-body calorimeter with intakes altering every day from 50 to 150% of their habitual intakes. Under these circumstances, the 24-hour energy expenditure decreased and increased by about 5% in response to a change in intake ten times larger. The effect was consistent only with the cost of digesting, absorbing and processing the food ingested on each day.

A further manifestation of the limited magnitude of spontaneous changes in thermogenesis comes from observations on men and women studied over 24-hour periods under temporarily standardized conditions of energy balance. Figure 1 shows that, providing account is taken of physical activity and the fat-free mass of the subject, the interindividual variability of 24-hour energy expenditure amounts to only 6-8%. Thus spontaneous interindividual variability in energy metabolism is surprisingly small. These published data accord with McNeill's further observations on "small and large eaters".

None of these sets of observations allows one to conclude that adaptive changes in energy metabolism do not occur in response to sustained changes in energy intake.


Contents - Previous - Next