Series

Obesity 3
Metabolically healthy obesity: the low-hanging fruit in
obesity treatment?
Norbert Stefan, Hans-Ulrich Häring, Matthias B Schulze

Obesity increases the risk of several other chronic diseases and, because of its epidemic proportions, has become a Lancet Diabetes Endocrinol 2017
major public health problem worldwide. Alarmingly, a lower proportion of adults have tried to lose weight during the Published Online
past decade than during the mid-1980s to 1990s. The first-line treatment option for obesity is lifestyle intervention. September 14, 2017
http://dx.doi.org/10.1016/
Although this approach can decrease fat mass in the short term, these beneficial effects typically do not persist. If a
S2213-8587(17)30292-9
large amount of weight loss is not an easily achievable goal, other goals that might motivate people with obesity to
See Online/Comment
adopt a healthy lifestyle should be considered. In this setting, the concept of metabolically healthy obesity is useful. http://dx.doi.org/10.1016/
Accumulating evidence suggests that, although the risk of all-cause mortality and cardiovascular events might be S2213-8587(17)30312-1
higher in people with metabolically healthy obesity compared with metabolically healthy people of a normal weight, This is the third in a Series of
the risk is substantially lower than in individuals with metabolically unhealthy obesity. Therefore, every person with three papers about obesity
obesity should be motivated to achieve a normal weight in the long term, but more moderate weight loss sufficient for Department of Internal
the transition from metabolically unhealthy obesity to metabolically healthy obesity might also lower the risk of Medicine IV, University
Hospital Tübingen, Tübingen,
adverse outcomes. However, how much weight needs to be lost for this transition to occur is under debate. This Germany (Prof N Stefan MD,
transition might be supported by lifestyle factors—such as the Mediterranean diet—that affect cardiovascular risk, Prof H-U Häring MD); Institute
independent of body fat. In this Series paper, we summarise available information about the concept of metabolically of Diabetes Research and
healthy obesity, highlight gaps in research, and discuss how this concept can be implemented in clinical care. Metabolic Diseases of the
Helmholtz Centre Munich,
Tübingen, Germany (N Stefan,
Introduction lifestyle intervention in the Look AHEAD trial,13 which was H-U Häring); German Center for
The prevalence of overweight and obesity has increased done in patients with type 2 diabetes, of whom 85·1% also Diabetes Research,
globally during the past few decades, and increased fat had obesity, 68% of participants lost 5% of their initial Neuherberg, Germany
(N Stefan, H-U Häring,
mass promotes morbidity and mortality.1 In the USA, weight, and 37% lost 10%. After a median follow-up of Prof M B Schulze DrPH); and
obesity is closely associated with several of the top ten 9·6 years, mean weight loss in the intensively treated Department of Molecular
leading causes of death, including heart disease, stroke, participants in the Look AHEAD study was 6·0%, still Epidemiology, German
cancer, and type 2 diabetes.2,3 Medical organisations are within the range recommended by guidelines for reducing Institute of Human Nutrition
Potsdam-Rehbruecke,
beginning to classify obesity, defined as a BMI of 30 cardiovascular risk. However, mean BMI only declined Nuthetal, Germany
kg/m² or higher, as a disease.4 There is an ongoing debate from 35·9 kg/m² (SD 6·0) at the start of the trial to (M B Schulze)
about whether the lowest risk of mortality is found in the 33·6 kg/m² by the end.14 Therefore, although a decrease in Correspondence to:
overweight state (defined by WHO as a BMI of bodyweight is achievable for obese people with lifestyle Prof Norbert Stefan, Department
25·0 to <30·0 kg/m²).5–7 However, the largest study to intervention, this intervention does not reduce BMI to the of Internal Medicine IV,
University Hospital Tübingen,
date with the most rigorous criteria to account for normal range in the long term. Similarly, in another large Otfried-Müller-Straße 10,
confounding factors showed that a BMI in the normal study (which included participants with elevated fasting Tübingen 72076, Germany
weight range (defined by WHO as 18·5 to <25·0 kg/m²), and post-load plasma glucose concentrations in the [email protected]
and specifically a BMI of 20·0–25·0 kg/m², was prediabetic range, of whom 68% were obese, with a tuebingen.de

associated with the lowest all-cause mortality.8 These data baseline BMI of 34·0 kg/m² [SD 6·7]) that involved
suggest that maintaining BMI in this range or reducing intensified lifestyle intervention—the Diabetes Prevention
an increased bodyweight to this range could effectively Program Outcomes Study15—weight loss after 10 years was
reduce the risk of early death. about 1 kg in the control group and only 2 kg in the lifestyle
Most clinical guidelines recommend an initial loss intervention group. BMI at follow-up was greater than
of 5–8% of total bodyweight for overweight and obese 33 kg/m² in both groups.
individuals, instead of the achievement of normal weight, Although modest weight loss of 5–8%, which is in the
to prevent cardiometabolic diseases.9–12 Lifestyle range recommended by most clinical guidelines,
intervention that targets eating behaviour and physical improves measures of diabetes, hypertension, and
activity should always be the first option for weight disability without major risks, its effect on cardiovascular
management, because it is associated with relatively low outcomes remains unclear.16–21 In the Look AHEAD trial,14
costs and minimal risk of complications. Such lifestyle intensive lifestyle intervention did not reduce the risk of
intervention programmes usually include intensive cardiovascular morbidity or mortality compared with a
behavioural counselling with about 14 visits in 6 months; control programme of diabetes support and education
results show that 60–65% of patients lose 5% or more of among overweight or obese patients with type 2 diabetes.
their initial weight.9–12 For example, 1 year after intensive Notably, the subgroup of individuals who lost at

www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9 1

 Series

health literacy.24 Thus, the concept of metabolically

Cutoff points
healthy obesity could facilitate and motivate doctor–
Normal triglyceride <150 mg/dL or <1·7 mmol/L and not on drug patient communication. Communication with the
concentrations treatment for elevated triglyceride concentration
patient that they have reached the status of metabolically
Normal HDL cholesterol concentrations
healthy obesity, or not, might be more easily
Men ≥40 mg/dL or ≥1·05 mmol/L
understandable for the patient than giving them
Women ≥50 mg/dL or ≥1·25 mmol/L and not on drug
treatment to increase HDL cholesterol
information about each of the continuous parameters
Normal blood pressure <130/85 mm Hg and not on antihypertensive
and their cutoff values. Although the effectiveness of this
drug treatment communication approach has not been tested, the
Normal fasting glucose <100 mg/dL or <5·6 mmol/L and not on drug concept of using more easily understandable targets is
treatment for hyperglycaemia similar to the ongoing design of nutritional labels on
food. Research indicates that well designed, simple food
These measures are those most frequently used to define metabolic syndrome in
studies that evaluate subgroups of obese people with regard to cardiovascular labels greatly improve consumers’ ability to assess the
risk, according to a systematic review.22 Cutoff points refer to definitions of healthfulness of foods.25
metabolic syndrome components according to the National Expert Panel on
Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
(NCEP-ATP-III) update from 2005.23
Parameters and cutoffs to define metabolic health
There is still no consensus among scientists and
Table: Cutoff points for triglycerides, HDL cholesterol, blood pressure, clinicians about the parameters and cutoff values to
and fasting glucose to define metabolic health
define metabolic health. Most studies of metabolically
healthy obesity used the definition of having less than
least 10% of their bodyweight (mean weight loss 15·4% two criteria of metabolic syndrome, based on the
[SD 5·0], mean BMI at follow-up close to overweight parameters and their cutoff values (table). Studies
range, estimated as approximately 30·5 kg/m²) over a evaluating several metabolic syndrome components
median follow-up of 10·2 years had a 21% lower risk of differ in the number of components used and the
cardiovascular morbidity or mortality than did individuals definition of metabolic health. Most studies included in
with stable weight or weight gain.16 our systematic review22 used the absence of high blood
In this Series paper, we address the question of how pressure, high levels of triglycerides and glucose, and
See Online for appendix much weight loss might be necessary to protect from low levels of HDL cholesterol (appendix), usually using
cardiometabolic diseases in people with obesity; in effect, cutoffs for these parameters outlined in the NCEP-ATP-
how much weight loss is necessary to move from a state III criteria23 for defining metabolic syndrome (table).
of metabolically unhealthy obesity to metabolically Although most studies investigated the absence of
healthy obesity. Furthermore, we discuss whether the metabolic syndrome, a more strict definition of
concept of metabolically healthy obesity could help to metabolic health (ie, the absence of all of the above
stratify people with obesity as low or high cardiometabolic components) could be useful.22 A recent analysis of the
risk despite similar weight loss. Finally, we highlight EPIC-CVD cohort26 suggests that people with this
future research approaches that could help to better obesity phenotype might indeed have a cardiovascular
understand the mechanisms underlying the transition risk profile comparable to that of healthy individuals of
from metabolically unhealthy to metabolically healthy normal weight (relative risk [RR] of coronary heart
obesity, and discuss how this knowledge can be disease 1·21, 95% CI 0·76–1·92). More strict criteria
implemented in clinical care. also classify a considerably smaller proportion of obese
people as “healthy”. By contrast, central obesity defined
The concept of metabolically healthy obesity by high waist circumference is unlikely to be a useful
Relevance for the patient parameter to stratify individuals with obesity as
For clinicians, the goals of a structured lifestyle metabolically healthy or unhealthy, because most people
intervention in addition to weight loss are well defined. with a BMI of ≥30 kg/m² are also centrally obese, owing
These goals include normalisation of the parameters that to the strong correlation between BMI and waist
are used to define metabolic syndrome (table), such circumference. There is still little evidence to suggest
as hypertension (systolic/diastolic blood pressure that other cardiometabolic risk markers, such as
≥130/85 mm Hg), low HDL cholesterol levels (<40 mg/dL C-reactive protein concentrations, could be used to
in men and <50 mg/dL in women), hyperglycaemia better define metabolic health.22
(fasting glucose level ≥100 mg/dL), visceral obesity (waist What mechanisms could predispose to metabolically
circumference ≥102 cm in men and ≥88 cm in healthy obesity? A 2017 study27 supports the hypothesis
women), and insulin resistance (eg, homoeostasis model that metabolically healthy versus unhealthy obesity
assessment of insulin resistance ≥90th percentile or phenotypes could be related to differences in total fat
≤2·5 mU/L × mg/dL). However, the normal ranges of all mass and regional fat accumulation. Lotta and colleagues27
of these parameters might be difficult to appraise for identified 53 genomic regions associated with insulin
some patients, particularly in patient groups with poor resistance phenotypes (fasting hyperinsulinaemia

2 www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9

 Series

[adjusted for BMI], hypertriglyceridaemia, and low HDL A significantly increased risk for people with
cholesterol levels) by use of genome-wide association metabolically healthy obesity compared with
studies. An increased number of risk alleles at these metabolically healthy people of normal weight was only
53 loci was also strongly associated with higher insulin observed when the analysis was restricted to studies
resistance measured during a euglycaemic and with a follow-up of longer than 10 years (RR
hyperinsulinaemic clamp or estimated from the oral 1⋅24, 1⋅02–1⋅55). A recent pan-European cohort study
glucose tolerance test, and with higher risk of type 2 (EPIC-CVD)26 further supports this notion: people with
diabetes and coronary heart disease. Although insulin obesity without metabolic syndrome were still at a
resistance is often considered secondary to higher higher risk of coronary heart disease than healthy people
adiposity, the 53 loci were associated with lower body-fat of normal weight (RR 1⋅28, 1⋅03–1⋅58). There is also
percentage, BMI, and hip circumference; among the evidence that individuals with metabolically healthy
adipose compartments measured by dual-energy X-ray obesity have a substantially higher risk of type 2 diabetes
absorptiometry, the loci were most strongly associated than do metabolically healthy individuals of normal
with lower gynoid and leg fat mass.27 These data indicate weight (RR 4·03, 2·66–6·09). Nevertheless, the RR was
that impaired expansion of subcutaneous fat mass, substantially higher in individuals with metabolically
particularly of fat mass in the lower body, might unhealthy obesity (8·93, 6·86–11·62).43 These data
predispose people to a metabolically unhealthy obesity support the hypothesis that people with metabolically
phenotype. healthy obesity are not protected from cardiometabolic
diseases in general, and that they should adopt a healthy
Metabolic health and risk of cardiometabolic lifestyle and try to reach and maintain a healthy weight.
disease Nevertheless, achieving or maintaining a status of
Although the concept of metabolically healthy obesity metabolically healthy obesity could be a valid interim
has been widely studied in research settings, its target in the prevention of cardiometabolic diseases. A
relevance to clinical practice is still unclear.28–42 The lack study by Appleton and colleagues40 suggests that risk of
of a consensus on its relevance is not only due to the diabetes, cardiovascular disease, or stroke is not increased
absence of an established definition of the phenotype of in people with metabolically healthy obesity compared
metabolically healthy obesity, but also because with metabolically healthy individuals of normal weight
epidemiological evidence that link obesity with cardio­ if the metabolically healthy obesity phenotype is
metabolic risk is heterogeneous. In a meta-analysis,22 we maintained during a natural follow-up over several
summarised results of studies that stratified individuals years—ie, with no interventions, such as weight-loss
with obesity by one or more cardiometabolic risk factors programmes. However, the same study also showed that
and evaluated cardiovascular risk in prospective settings. the metabolically healthy obesity phenotype is a transient
We pooled findings from the included studies into six condition for many participants; 30·6% of people who
different subgroups, namely absence of metabolic intially had metabolically healthy obesity developed
syndrome, high insulin sensitivity, absence of hyper­ metabolically unhealthy obesity over a 10-year period.40
tension, absence of diabetes, absence of hyperlipidaemia, Appleton and colleagues’ classification of metabolically
or absence of any of these metabolic factors. None of at-risk was defined as an individual matching any two of
these approaches clearly showed a subgroup of people the following criteria (consistent with the International
with obesity that was not at an increased risk of Diabetes Federation metabolic syndrome criteria18):
cardiovascular events compared with metabolically triglyceride concentration equal to or higher than
healthy individuals of normal weight. Our analysis of 1·7 mmol/L; HDL cholesterol concentration lower than
data from 13 studies suggested that individuals 1·0 mmol/L in men, or 1·3 mmol/L in women, or lipid-
with metabolically healthy obesity are still at higher lowering medication use; blood pressure equal to or
risk of cardiovascular events than metabolically greater than 130/85 mm Hg, or antihypertensive
healthy individuals of normal weight (RR 1·45, medication use; or fasting glucose concentration equal to
95% CI 1·20–1·70).22 However, this risk is considerably or higher than 5·6 mmol/L or self-reported diabetes.
lower than that of people with metabolically unhealthy Participants with one risk factor or less were classified as
obesity (compared with metabolically healthy individuals metabolically healthy. They were then classified according
of normal weight, RR 2·31, 95% CI 1·99–2·69).22 to their BMI and the WHO BMI cutoffs for normal
Similarly, a meta-analysis by Kramer and colleagues,39 weight, overweight, and obesity.
which included studies on all-cause mortality and The study by Appleton and colleagues40 also highlights
cardiovascular events, suggests a substantially elevated potential factors that are associated with maintenance of
risk in people with metabolically unhealthy obesity metabolic health (younger age and lower waist
compared with people who are metabolically healthy of a circumference). However, it is still unclear whether
normal weight (RR 2⋅65, 2⋅18–3⋅12). Metabolic health there are factors that could be altered to make
was defined by the absence of metabolic syndrome or bodyweight gain resilient to cardiometabolic risk.
fewer than two signs of metabolic syndrome. Despite these unknowns, the data strongly suggest that

www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9 3

 Series

individuals converting from metabolically unhealthy to Transition from metabolically unhealthy to

metabolically healthy obesity would substantially healthy obesity
decrease, but not completely eliminate, their risk of Weight loss
cardiometabolic disease. Because people with metabolically healthy obesity have a
Further support—although indirect, because lower risk of cardiometabolic diseases than people with
metabolically healthy obesity was not defined—that the metabolically unhealthy obesity do, the question arises:
concept of metabolically healthy obesity might help to how can a transition from metabolically unhealthy to
stratify cardiometabolic risk has been provided by data healthy obesity be achieved? Lifestyle intervention-
from the Swedish Obese Subjects (SOS) study.44 This induced weight loss should always be the first target in
study was the first to show that, compared with the treatment of obesity, irrespective of metabolic health
conventional treatment, bariatric surgery reduces phenotypes.9–12 Weight loss reduces cardiometabolic risk
mortality by 24%, (adjusted hazard ratio [HR] 0·76, and provides other health benefits, such as improved
95% CI 0·59–0·99),45 risk of cardiovascular death by osteoarthritis and sleep apnoea, decreased cancer risk
53% (0·47, 0·29–0·76), and total first-time (fatal or non- and depression episodes, and improved wellbeing.9–12 But
fatal) cardiovascular events (myocardial infarction or how much weight loss is required to transition from
stroke, whichever came first) by 43% (0·67, 0·54–0·83).44 metabolically unhealthy to healthy obesity?
However, bariatric surgery is an invasive procedure with In a study of 19 men and women with obesity, in which
notable side-effects, and the number needed to treat to participants were not analysed according to their metabolic
prevent one cardiovascular event (myocardial infarction health phenotypes, a 5% weight loss through a lifestyle
or stroke) was relatively high (50) in the SOS study.44 intervention involving behaviour intervention and dietary
Subgroup analysis showed that the effectiveness of the counselling improved insulin sensitivity and reduced
intervention depended on fasting insulin levels at intra-abdominal fat mass, liver fat content, fasting
baseline: reduced cardiovascular risk with surgery glycaemia, insulinaemia, and triglyceride concentrations,
compared with controls was only observed in compared with a weight maintenance programme (mean
participants with baseline insulin levels above the bodyweight at baseline 106·6 kg [15·0], mean bodyweight
median (HR 0·69, 95% CI 0·54–0·87), and was not seen at follow-up 106·7 kg [14·7]).48 However, because
in people with lower levels (0·93, 0·67–1·28). The participants were not analysed according to metabolic
number needed to treat was also substantially lower for health phenotypes, no conclusions about the effects of the
people with high insulin values (an indication of insulin intervention in the context of the respective phenotypes
resistance) than for those with lower insulin values (an can be made. Such characterisation was performed in six
indication of insulin sensitivity; 21 vs 173). Baseline other studies involving a lifestyle intervention,49–54
fasting insulin levels also modified the treatment effect including the TULIP study done by our group.52 In these
in terms of mortality (p for interaction=0·013) and studies, cardiometabolic risk parameters improved in
incidence of diabetes (p=0·007), and lower fasting participants with metabolically unhealthy obesity. Only
glycaemia also interacted with the treatment effect in one study53 investigated the proportion of participants who
terms of risk of developing diabetes (p=0·002).46 These converted from the metabolically unhealthy to the
data indicate that stratification of people with obesity metabolically healthy obesity phenotype. In that study,
based on insulin resistance might be helpful for 78 premenopausal women with obesity were given a
clinicians focusing on the prevention of cardiometabolic calorie-restricted diet for 12 weeks; 53 of these women had
diseases. Insulin resistance could explain most, if not metabolically unhealthy obesity (mean BMI at baseline
all, risks associated with metabolic syndrome,47 and lack 34·3 kg/m² [SD 3·0]), and 25 had metabolically healthy
of insulin resistance might be a hallmark of obesity (33·4 kg/m² [2·4]).53 Metabolically healthy obesity
metabolically healthy obesity.28 Although these data was defined as having less than two criteria of metabolic
highlight the role of insulin resistance as a strong syndrome (according to the American Heart Association
predictor of cardiometabolic diseases, there are definition23), although a waist circumference of greater
limitations, including that insulin levels are difficult to than 88 cm was allowed. Mean percentage weight loss was
compare between different labs and there is no similar in both groups (−8·8% [3·1] for metabolically
universally accepted cutoff value for insulin resistance. unhealthy obesity and −8·3% [3·7] for metabolically
Thus, it would be useful to investigate in a post-hoc healthy obesity). 13 (25%) women with metabolically
analysis of the SOS study data whether determination of unhealthy obesity converted to metabolically healthy
metabolically healthy obesity—using blood pressure, obesity during the intervention. It is not known whether
dyslipidaemia, and hyperglycaemia as parameters— this conversion takes place similarly in men.
could also help to stratify individuals with obesity into For this Series paper, we did a similar analysis in the
those with a large (people with metabolically unhealthy TULIP study,52 which evaluated a 9-month intensive
obesity) or a small (people with metabolically healthy lifestyle intervention with diet modification and increased
obesity) potential treatment benefit in terms of reducing physical activity, similar to the lifestyle intervention in the
cardiometabolic or mortality risk. Finnish Diabetes Prevention Study.18 In the TULIP study,

4 www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9

 Series

which only investigated white people, individuals were

included if they fulfilled at least one of the following
criteria: family history of type 2 diabetes; BMI higher than
10
27 kg/m²; or a diagnosis of impaired glucose tolerance,
previous gestational diabetes, or both. After baseline
measurements were taken, all individuals had dietary
counselling and up to ten sessions with a dietician.
Through individual counselling, the aim was to reduce fat 0
intake to less than 30% of total calorie intake, reduce
saturated fat intake to amounts lower than 10% of total fat
intake, increase fibre intake to 15 g per 1000 kcal, and
achieve at least 5% weight loss during the study.52 Of –10
357 people, 150 had obesity (mean BMI 35·2 kg/m² [4·5]
Percentage change
for the subgroup with obesity) at baseline. Of those with
obesity, 41 had metabolically unhealthy obesity—defined
as having two or more criteria of metabolic syndrome23—
–20
and lost more than 5% bodyweight during the intervention
(BMI at baseline 35·3 kg/m² [3·5]). 11 (27%) of these
participants (BMI at baseline 33·0 kg/m² [2·5]) converted
to metabolic health, while 30 participants (BMI at baseline
36·0 kg/m² [3·5]) still were metabolically unhealthy at –30
9-month follow-up. The median weight loss in those who
converted was 9·2 kg (9·1%, IQR 1·7–12·9), while it was
7·9 kg (8·0%, 6·4–9·6) in those who did not convert. Of
the 11 who converted to metabolic health, 5 (45%)
–40
remained obese, whereas 24 (80%) of the 30 non-
converters were still obese at follow-up. With regards to
the parameters of metabolic syndrome, decreased serum
triglycerides was the greatest change seen in participants
who converted, followed by decreased blood pressure. –50
e

es

se

re

re

ht

es

se

re

re

ht
ro

ro
nc

nc

Those who did not convert showed small changes in

su

su

su

su
id

id
co

co
ig

ig
te

te
re

re
er

er
we

we
es

es

es

es
lu

lu
les

les
fe

fe
lyc

lyc
gg

pr

pr

gg

pr

pr
dy

dy
um

ho

um

these and some other parameters (figure 1). When we ho

ig

ig
od

od

od

od
tin

tin
Bo

Bo
Lc

Lc
Tr

Tr
irc

irc
blo

blo

blo

blo
s

s
HD

HD
tc

Fa

tc

Fa
tested which baseline parameters predicted the
lic

lic

lic

lic
ais

ais
sto

to

sto

to
W

W
as

as
Sy

Sy

conversion from metabolically unhealthy obesity to

Di

Di
Converters Non-converters
metabolically healthy obesity (considering age, sex, BMI,
waist circumference, systolic and diastolic blood pressure, Figure 1: Changes in parameters of metabolic syndrome in the TULIP study
fasting glucose, HDL cholesterol and triglycerides, and Changes (medians and IQRs) of the parameters of metabolic syndrome (unpublished data), as defined by the
American Heart Association, in the TULIP study.52 Data are separately presented for people who converted from
MRI and spectroscopy-derived measurements of visceral metabolically unhealthy obesity to metabolic health during the 9-month lifestyle intervention (n=11; baseline
fat mass and liver fat content), only BMI and liver fat BMI 33·0 kg/m2) and those who remained with metabolically unhealthy obesity (n=30; baseline BMI 36·0 kg/m2).
content remained independent predictors. This finding
suggests that more weight loss might be needed when
the baseline BMI is higher. Future studies should evaluate low in fruits (7·5%), and high in sugar-sweetened
the effect of different levels of weight loss across different beverages (7·4%). There is strong evidence from
levels of obesity on metabolic health. landmark studies, such as DASH56 and OmniHeart,57 that
diet quality can substantially affect cardiometabolic risk
Healthy lifestyle factors without alterations of bodyweight. Similarly, the
Weight loss is not the only target that can reduce PREDIMED trial,58,59 a study of 7447 individuals (age range
cardiometabolic risk in individuals with metabolically 55–80 years) with high cardiovascular risk (either type 2
unhealthy obesity. There is substantial evidence that a diabetes or at least three of the following risk factors:
healthy lifestyle can reduce the risk of cardiometabolic smoking, hypertension, elevated LDL cholesterol con­
morbidity and mortality, independent of effects on centrations, low HDL cholesterol concentrations,
bodyweight. As Micha and colleagues55 estimated, almost overweight or obesity, or a family history of premature
half of all cardiometabolic deaths in the USA are coronary heart disease), showed that eating an adapted ad
attributable to suboptimal diets, including diets high in libitum Mediterranean diet over a median follow-up of
sodium (9·5% of all cardiometabolic deaths), low in nuts 4·8 years reduces risk of cardiovascular events by
or seeds (8·5%), high in processed meats (8·2%), low in about 30%, compared with a control diet, despite having
seafood omega-3 fats (7·8%), low in vegetables (7·6%), little effect on bodyweight. This effect appeared to be

www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9 5

 Series

stronger among participants with obesity, hypertension, message from the literature from weight-loss
and dyslipidaemia. An analysis of the PREDIMED study60 interventions needs to be clarified. In the study by Ruiz
also showed that a Mediterranean diet was associated and colleagues53 and in our TULIP study,52 only about a
with reversion of metabolic syndrome among individuals quarter of individuals with metabolically unhealthy
who had the syndrome at baseline. If a post-hoc analysis obesity converted to metabolically healthy obesity,
of the PREDIMED study showed an early conversion despite losing more than 5% of their bodyweight.
from metabolically unhealthy to metabolically healthy Although this outcome could be discouraging for health-
obesity to predict reduced mortality and cardiometabolic care providers and most patients, there could be positive
risk, this would support the notion that people with aspects for non-converters. First, patients can quickly
metabolically unhealthy obesity should be targeted early assess that their weight loss was not sufficient to achieve
with a Mediterranean diet. The PREDIMED-Plus study61 metabolic health. Although patients might not see
might provide evidence in the future, if such a diet has smaller changes in cardiometabolic risk factors as a
additional benefits for weight loss when accompanied by success, these changes could reduce cardiometabolic
energy restriction, physical activity promotion, and risk. Knowledge of metabolic health could help to inform
structured behaviour support. By contrast, the Women’s patients about the larger achievable metabolic benefits
Health Initiative Randomized Controlled Dietary of a lifestyle intervention. Second, if non-converters
Modification Trial suggests that ad libitum low-fat diets know that, on average, converters only lost a few kg
are unlikely to have a beneficial effect on cardiovascular more than them, this information might motivate the
events.62 There was little change in weight between the non-converters to carry on their lifestyle intervention to
intervention and control groups over 3 years (<1 kg). achieve a weight loss with metabolic benefit. Given that
Taken together, diets that emphasise plant-based foods more weight loss might be required in people with a
play an important role in prevention of cardiovascular higher starting BMI, physicians should start weight-loss
diseases, although randomised trials of dietary intervention as early as possible. For those who convert
interventions with cardio­vascular endpoints are rare, with from metabolically unhealthy to metabolically healthy
the exception of PREDIMED. obesity, the results are probably motivating. In the
TULIP study,52 the achieved weight loss of 9·1% was
Applying the concept of metabolically healthy sufficient for the patients to take an intermediate step
obesity to treatment towards better cardiometabolic health, for which
Implications for doctor–patient communication substantial protection from cardiometabolic diseases
To understand how the concept of metabolically healthy can be expected. Patients who achieve metabolically
obesity can be used to treat obesity, the take-home healthy obesity should maintain their lower bodyweight
over time and ideally try to lose more bodyweight to
High cardiometabolic risk-placebo group (baseline BMI 35·2 kg/m², weight change –1·9%)
further reduce cardiometabolic risk.
High cardiometabolic risk-treatment group (baseline BMI 34·5 kg/m², weight change –9·4%) From a practical viewpoint, the amount of weight loss
Low cardiometabolic risk-placebo group (baseline BMI 41·2 kg/m², weight change –1·2%) achieved during a weight-loss programme should not be
Low cardiometabolic risk-treatment group (baseline BMI 41·0 kg/m², weight change –10·2%)
the primary focus of doctor–patient communications.
Instead, clinicians should inform patients about the
effectiveness of the achieved weight loss in the context of
changes of metabolic parameters and blood pressure—
Cumulative diabetes incidence

NNT: 24
ie, whether they reached a status of metabolic health.
Therefore, metabolic health can be a meaningful goal
agreed between clinicians and patients.11

Implications for medical guidelines

Based on results of the Look AHEAD trial,16 the SOS
study,45 and studies investigating improvements of
cardiometabolic risk factors during weight loss in
patients with metabolically unhealthy obesity,49–54 it is
important to address in the medical guidelines that, for
many obese individuals, a weight loss of 5–8% might not
NNT: 120 be sufficient to prevent cardiometabolic diseases and
reduce the risk of early mortality. Increased weight loss
Time might be necessary to lower risk, particularly for people
with more severe obesity (class II and III obesity).
Figure 2: Expected effect of weight loss on risk of incident type 2 diabetes by metabolic health subgroup
In addition to weight loss, the type of dietary
Weight-loss intervention with phentermine/topiramate with subgroup analyses stratified by the Cardiometabolic
Disease Staging score.67 This score incorporates parameters of metabolic syndrome, thus is similar to the concept intervention should be considered in the context of
of metabolic health. NNT=number needed to treat. cardiometabolic benefits of a lifestyle intervention.

6 www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9

 Series

Systematic reviews and meta-analyses63–66 indicate that a

A
variety of diets, widely varying in their macronutrient 45 SOS gastric bypass
composition, can be effective for weight loss when SOS LI
adhered to, and thus alternatives to low-fat weight-loss Look AHEAD
control
diets, as was incorporated in the intervention in the Look 40 Look AHEAD
AHEAD trial,16 exist. To encourage low-fat diets now intensive LI
DPP control
seems out of step with current evidence, as such diets DPP intensive LI
have not been shown to provide cardiovascular benefits 35

BMI (kg/m2)
on their own,62 by contrast with a Mediterranean diet.58
Thus, based on current, albeit limited, evidence from Obesity
30
randomised trials on prevention of cardiovascular
disease, a Mediterranean diet should be incorprated into
weight-loss programmes for people with metabolically 25
Overweight
unhealthy obesity. However, no trials have specifically
addressed whether interventions for weight loss or diet
Normal weight
modification can promote the transition from 18·5
Baseline 10-year follow-up
metabolically unhealthy to metabolically healthy obesity,
and further studies are warranted. B
Regardless of metabolic status, all people with obesity
should aim for metabolic health and normal weight. MUHO MHO
However, on the basis of the scarce resources that are • Dyslipidaemia Absence of:
• Hypertension • Dyslipidaemia
available for weight-loss programmes, it might be 40
• Hyperglycaemia • Hypertension
reasonable to allocate more resources to people with • Insulin resistance • Hyperglycaemia
metabolically unhealthy obesity. A study67 investigated • Fatty liver • Insulin resistance
• Visceral obesity • Fatty liver
the effect of phentermine/topiramate-induced weight 35
• Visceral obesity
BMI (kg/m2)

Cardiometabolic
loss, compared with placebo, on the prevention of type 2 risk: high Cardiometabolic
diabetes in people who were stratified according to the risk: moderate
Weight loss: ~10%
Cardiometabolic Disease Staging score into subgroups 30
of high or low cardiometabolic risk. This score includes Weight loss: ~20%
the parameters of metabolic syndrome and, thus, is
Metabolically healthy
similar to the concept of metabolically healthy versus 25
normal weight
unhealthy obesity; however, metabolic health was not Cardiometabolic
determined on the basis of the currently most widely risk: low

used definitions (ie, fewer than two parameters of 18·5

Time
metabolic syndrome, or fewer than two of the following
abnormalities: elevated concentrations of triglyceride, Figure 3: Changes in BMI over 10 years in three large clinical trials, and associated risk of cardiometabolic
glucose, and C-reactive protein; decreased HDL diseases with weight loss based on the presence or absence of metabolic health in obesity
(A) Baseline and 10-year BMI values in the Look AHEAD trial,14 the Diabetes Prevention Program (DPP) Outcomes
cholesterol concentrations; insulin resistance; and study,15 and the Swedish Obese Subjects (SOS) study,45 according to intervention. (B) Characteristics of people with
hypertension).28,29,39,40,68,69 Furthermore, in this study,67 the MUHO and MHO by BMI and cardiometabolic risk. Weight loss of about 10% in people with MUHO might result in
baseline mean BMI ranged from 34·5 kg/m² to conversion to MHO, on the basis of data from the Look AHEAD trial.16 Once an individual has converted to MHO,
41·2 kg/m² among subgroups (mean baseline BMI of only a further weight loss of about 20% will result in a BMI that is close to the normal healthy weight range, where
the risk of cardiometabolic diseases is lowest. LI=lifestyle intervention. MUHO=metabolically unhealthy obesity.
each subgroup is given in figure 2) and the weight loss MHO=metabolically healthy obesity.
was similar between the high cardiometabolic
risk (9·4%) and low cardiometabolic risk (10·2%)
treatment groups (figure 2). There were two main Conclusions
observations. First, although the treatment reduced Lifestyle intervention in people with obesity to reduce
diabetes risk compared with placebo in the high-risk bodyweight is considered safe and effective at decreasing
group, the risk in this group did not drop as low as the cardiometabolic risk. However, results from several
baseline risk in the low-risk group. Thus, in high-risk studies, especially from the Look AHEAD trial13 and the
individuals, substantially more weight loss might be Diabetes Prevention Program Outcomes Study,15 showed
needed to achieve a diabetes risk as low as that in the that it is very difficult to reach and maintain a bodyweight
low-risk group. Second, the numbers needed to treat to in the normal or overweight range with lifestyle
prevent one case of diabetes over about 1 year were 120 in intervention alone. As was shown in the SOS studies,44–46
the low-risk group, but only 24 in the high-risk group.67 for many people with obesity, only bariatric surgery
Therefore, targeting patients at high risk with specific appears to be effective to reach this goal (figure 3A).
weight-loss strategies might improve the cost-to-benefit Furthermore, the chronicity of obesity and infrequent
ratio of such interventions. discussions about weight between primary care clinicians

www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9 7

 Series

Contributors
Search strategy and selection criteria NS and MBS reviewed the literature and wrote the manuscript. H-UH
reviewed the literature and contributed to discussion design.
We searched PubMed for full-text original studies and review
Declaration of interests
articles published in English between Jan 1, 1990, and
We declare no competing interests.
Aug 31, 2017, to identify reports on metabolic parameters
Acknowledgments
and mortality in people with obesity. Search terms were
We are very grateful for the constructive comments provided by
“metabolically healthy obesity” OR “metabolically unhealthy Frank B Hu during the drafting process. This work was aided in part by
obesity” OR “metabolically benign obesity” OR “metabolic funding from the German Research Foundation (KFO 114 and STE
syndrome” OR “insulin sensitivity” OR “insulin resistance” OR 1096/1-3) and from the German Federal Ministry of Education and
Research to the German Center of Diabetes Research.
“bariatric surgery” AND “lifestyle intervention” AND
“mortality”. The reference lists of the identified papers were References
1 NCD Risk Factor Collaboration (NCD-RisC). Trends in adult body
used to identify additional relevant papers. The final reference mass index in 200 countries from 1975 to 2014: a pooled analysis of
list was selected on the basis of relevance to the subject of 1698 population-based measurement studies with 19·2 million
this Series paper. We included studies on mortality if they participants. Lancet 2016; 387: 1377–96.
2 National Center for Health Statistics (US). Health, United States,
were prospective, observational cohort studies that provided 2015: with special feature on racial and ethnic health disparities.
estimates of risk associated with BMI for subgroups of obese Hyattsville, MD: National Center for Health Statistics (US), 2016.
individuals in comparison to non-obese individuals. 3 Heymsfield SB, Wadden TA. Mechanisms, pathophysiology, and
management of obesity. N Engl J Med 2017; 376: 254–66.
4 American Medical Association House of Delegates. Relevant AMA
policy. http://www.npr.org/documents/2013/jun/ama-resolution-
and their patients might have resulted in fewer adults obesity.pdf (accessed Sept 4, 2017).
trying to lose weight nowadays than in previous decades.70 5 Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause
mortality with overweight and obesity using standard body mass
Thus, it is important to identify easily achievable targets index categories: a systematic review and meta-analysis.
(ie, the low-hanging fruit) that can motivate people with JAMA 2013; 309: 71–82.
obesity towards their goal of lower bodyweight and, 6 Hughes V. The big fat truth. Nature 2013; 497: 428–30.
ultimately, protection from cardiometabolic diseases and 7 Ahima RS, Lazar MA. Physiology. The health risk of obesity—better
metrics imperative. Science 2013; 341: 856–58.
early death. In this respect, the concept of metabolically 8 Global BMI Mortality Collaboration. Body-mass index and all-cause
healthy obesity might be an appropriate first goal for mortality: individual-participant-data meta-analysis of
initial weight loss.11 For people with obesity, it might be 239 prospective studies in four continents. Lancet 2016;
388: 776–86.
more easily understandable that they should continue to 9 Eckel RH, Jakicic JM, Ard JD, et al. 2013 AHA/ACC guideline on
follow a healthy lifestyle to reach the first step lifestyle management to reduce cardiovascular risk: a report of the
(metabolically healthy obesity), rather than to appreciate American College of Cardiology/American Heart Association Task
Force on Practice Guidelines. Circulation 2014; 129: S76–99.
the multiple targets of metabolic syndrome. Conversion
10 American Diabetes Association. Foundations of care and
to metabolically healthy obesity might be a valid medical comprehensive medical evaluation. Diabetes Care 2016;
reward for individuals with obesity, although this could 39 (suppl 1): S23–35.
be particularly hard to achieve if they start from a very 11 American College of Cardiology/American Heart Association Task
Force on Practice Guidelines, Obesity Expert Panel, 2013. Executive
high BMI. Importantly, all patients need to be encouraged summary: guidelines (2013) for the management of overweight and
to have a healthy diet and lifestyle and to lose more obesity in adults: a report of the American College of Cardiology/
American Heart Association Task Force on Practice Guidelines and
bodyweight because, in the long term, the risk of the Obesity Society published by the Obesity Society and American
cardiometabolic diseases and early death is still higher in College of Cardiology/American Heart Association Task Force on
people with metabolically healthy obesity than in Practice Guidelines. Based on a systematic review from the The
Obesity Expert Panel, 2013. Obesity (Silver Spring) 2014;
metabolically healthy people of normal weight (figure 3B). 22 (suppl 2): S5–39.
Implementing the concept of metabolic health into 12 Jensen MD, Ryan DH, Apovian CM, et al. 2013 AHA/ACC/TOS
management strategies will remain difficult, however, as guideline for the management of overweight and obesity in adults:
a report of the American College of Cardiology/American Heart
there is no universally accepted diagnosis for metabolic Association Task Force on Practice Guidelines and The Obesity
health. There is immense and ongoing interest in Society. J Am Coll Cardiol 2014; 63: 2985–3023.
defining a universally accepted definition of metabolic 13 Pi-Sunyer X, Blackburn G, Brancati FL, et al. Reduction in weight
and cardiovascular disease risk factors in individuals with type 2
health for adults and for adolescents and children. Thus, diabetes: one-year results of the look AHEAD trial. Diabetes Care
although the definition of metabolic health as having less 2007; 30: 1374–83.
than two components of metabolic syndrome might be 14 Wing RR, Bolin P, Brancati FL, et al. Cardiovascular effects of
the most practical approach in the clinical setting at this intensive lifestyle intervention in type 2 diabetes. N Engl J Med 2013;
369: 145–54.
stage of research, a different and widely accepted, 15 Knowler WC, Fowler SE, Hamman RF, et al. 10-year follow-up of
definition might be agreed upon in the future. Also, diabetes incidence and weight loss in the Diabetes Prevention
whether the goal of metabolic health, which does not Program Outcomes Study. Lancet 2009; 374: 1677–86.
16 Gregg EW, Jakicic JM, Blackburn G, et al. Association of the
acknowledge the potential benefit of smaller changes in magnitude of weight loss and changes in physical fitness with
cardiometabolic risk factors, enables more effective long-term cardiovascular disease outcomes in overweight or obese
physician–patient communication and greater motivation people with type 2 diabetes: a post-hoc analysis of the Look AHEAD
randomised clinical trial. Lancet Diabetes Endocrinol 2016; 4: 913–21.
for patients is yet to be proven.

8 www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9

 Series

17 Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the 40 Appleton SL, Seaborn CJ, Visvanathan R, et al. Diabetes and
incidence of type 2 diabetes with lifestyle intervention or cardiovascular disease outcomes in the metabolically healthy obese
metformin. N Engl J Med 2001; 346: 393–403. phenotype: a cohort study. Diabetes Care 2013; 36: 2388–94.
18 Tuomilehto J, Lindstrom J, Eriksson JG, et al. Prevention of type 2 41 Hamer M, Bell JA, Sabia S, Batty GD, Kivimäki M. Stability of
diabetes mellitus by changes in lifestyle among subjects with metabolically healthy obesity over 8 years: the English Longitudinal
impaired glucose tolerance. N Engl J Med 2001; 344: 1343–50. Study of Ageing. Eur J Endocrinol 2015; 173: 703–08.
19 WJ Rejeski, EH Ip, AG Bertoni, et al. Lifestyle change and mobility 42 Phillips CM. Metabolically healthy obesity across the life course:
in obese adults with type 2 diabetes N Engl J Med 2012; 366: 1209–17. epidemiology, determinants, and implications. Ann N Y Acad Sci 2017;
20 The Trials of Hypertension Prevention Collaborative Research 1391: 85–100.
Group. Effects of weight loss and sodium reduction intervention on 43 Bell JA, Kivimaki M, Hamer M. Metabolically healthy obesity and
blood pressure and hypertension incidence in overweight people risk of incident type 2 diabetes: a meta-analysis of prospective
with high-normal blood pressure. The Trials of Hypertension cohort studies. Obes Rev 2014; 15: 504–15.
Prevention, phase II. Arch Intern Med 1997; 157: 657–67. 44 Sjostrom L, Peltonen M, Jacobson P, et al. Bariatric surgery and
21 Wing RR. Long-term effects of a lifestyle intervention on weight long-term cardiovascular events. JAMA 2012; 307: 56–65.
and cardiovascular risk factors in individuals with type 2 diabetes 45 Sjostrom L, Narbro K, Sjostrom CD, et al. Effects of bariatric surgery
mellitus: four-year results of the Look AHEAD trial. Arch Intern Med on mortality in Swedish obese subjects. N Engl J Med 2007;
2010; 170: 1566–75. 357: 741–52.
22 Eckel N, Meidtner K, Kalle-Uhlmann T, Stefan N, Schulze MB. 46 Sjostrom L. Review of the key results from the Swedish Obese
Metabolically healthy obesity and cardiovascular events: a systematic Subjects (SOS) trial—a prospective controlled intervention study of
review and meta-analysis. Eur J Prev Cardiol 2016; 23: 956–66. bariatric surgery. J Intern Med 2013; 273: 219–34.
23 Grundy SM, Cleeman JI, Daniels SR, et al. Diagnosis and 47 Eckel RH, Alberti KG, Grundy SM, Zimmet PZ. The metabolic
management of the metabolic syndrome: an American Heart syndrome. Lancet 2010 375: 181–83.
Association/National Heart, Lung, and Blood Institute Scientific 48 Magkos F, Fraterrigo G, Yoshino J, et al. Effects of moderate and
Statement. Circulation 2005; 112: 2735–52. subsequent progressive weight loss on metabolic function and
24 Williams MV, Davis T, Parker RM, Weiss BD. The role of health adipose tissue biology in humans with obesity. Cell Metab 2016;
literacy in patient-physician communication. Fam Med 2002; 23: 591–601.
34: 383–89. 49 Shin MJ, Hyun YJ, Kim OY, Kim JY, Jang Y, Lee JH. Weight loss
25 Temple NJ, Fraser J. Food labels: a critical assessment. Nutrition effect on inflammation and LDL oxidation in metabolically healthy
2014; 30: 257–60. but obese (MHO) individuals: low inflammation and LDL oxidation
26 Lassale C, Tzoulaki I, Moons KGM, et al. Separate and combined in MHO women. Int J Obes (Lond) 2006; 30: 1529–34.
associations of obesity and metabolic health with coronary heart 50 Karelis AD, Messier V, Brochu M, Rabasa-Lhoret R. Metabolically
disease: a pan-European case-cohort analysis. Eur Heart J 2017; healthy but obese women: effect of an energy-restricted diet.
published online Aug 14. DOI:10.1093/eurheartj/ehx448. Diabetologia 2008; 51: 1752–54.
27 Lotta LA, Gulati P, Day FR, et al. Integrative genomic analysis 51 Janiszewski PM, Ross R. Effects of weight loss among
implicates limited peripheral adipose storage capacity in the metabolically healthy obese men and women. Diabetes Care 2010;
pathogenesis of human insulin resistance. Nat Genet 2017; 49: 17–26. 33: 1957–59.
28 Stefan N, Häring H-U, Hu FB, Schulze MB. Metabolically healthy 52 Kantartzis K, Machann J, Schick F, et al. Effects of a lifestyle
obesity: epidemiology, mechanisms, and clinical implications. intervention in metabolically benign and malign obesity.
Lancet Diabetes Endocrinol 2013; 1: 152–62. Diabetologia 2011; 54: 864–68.
29 Wildman RP, Muntner P, Reynolds K, et al. The obese without 53 Ruiz JR, Ortega FB, Labayen I. A weight loss diet intervention has a
cardiometabolic risk factor clustering and the normal weight with similar beneficial effect on both metabolically abnormal obese and
cardiometabolic risk factor clustering: prevalence and correlates of metabolically healthy but obese premenopausal women.
2 phenotypes among the US population (NHANES 1999–2004). Ann Nutr Metab 2013; 62: 223–30.
Arch Intern Med 2008; 168: 1617–24. 54 Dalzill C, Nigam A, Juneau M, et al. Intensive lifestyle intervention
30 Stefan N, Kantartzis K, Machann J, et al. Identification and improves cardiometabolic and exercise parameters in metabolically
characterization of metabolically benign obesity in humans. healthy obese and metabolically unhealthy obese individuals.
Arch Intern Med 2008; 168: 1609–16. Can J Cardiol 2014; 30: 434–40.
31 McLaughlin T, Abbasi F, Lamendola C, Reaven G. Heterogeneity in 55 Micha R, Peñalvo JL, Cudhea F, Imamura F, Rehm CD,
the prevalence of risk factors for cardiovascular disease and type 2 Mozaffarian D. Association between dietary factors and mortality
diabetes mellitus in obese individuals: effect of differences in from heart disease, stroke, and type 2 diabetes in the United States.
insulin sensitivity. Arch Intern Med 2007; 167: 642–48. JAMA 2017; 317: 912–24.
32 Karelis AD, St-Pierre DH, Conus F, Rabasa-Lhoret R, Poehlman ET. 56 Appel LJ, Moore TJ, Obarzanek E, et al. A clinical trial of the effects
Metabolic and body composition factors in subgroups of obesity: of dietary patterns on blood pressure. DASH Collaborative Research
what do we know? J Clin Endocrinol Metab 2004; 89: 2569–75. Group. N Engl J Med. 1997; 336: 1117–24.
33 Meigs JB, Wilson PW, Fox CS, et al. Body mass index, metabolic 57 Appel LJ, Sacks FM, Carey VJ, et al. Effects of protein,
syndrome, and risk of type 2 diabetes or cardiovascular disease. monounsaturated fat, and carbohydrate intake on blood pressure
J Clin Endocrinol Metab 2006; 91: 2906–12. and serum lipids: results of the OmniHeart randomized trial. JAMA
34 Karelis AD, Rabasa-Lhoret R. Inclusion of C-reactive protein in the 2005; 294: 2455–64.
identification of metabolically healthy but obese (MHO) individuals. 58 Estruch R, Ros E, Salas-Salvadó J, et al. Primary prevention of
Diabetes Metab 2008; 34: 183–84. cardiovascular disease with a Mediterranean diet. N Engl J Med
35 Blüher M. Are metabolically healthy obese individuals really 2013; 368: 1279–90.
healthy? Eur J Endocrinol 2014; 171: 209–19. 59 Estruch R, Martínez-González MA, et al. Effect of a high-fat
36 Samocha-Bonet D, Dixit VD, Kahn CR, et al. Metabolically healthy Mediterranean diet on bodyweight and waist circumference:
and unhealthy obese--the 2013 Stock Conference report. a prespecified secondary outcomes analysis of the PREDIMED
Obes Rev 2014; 15: 697–708. randomised controlled trial. Lancet Diabetes Endocrinol 2016;
37 Lotta LA, Abbasi A, Sharp SJ, et al. Definitions of metabolic health 4: 666–76.
and risk of future type 2 diabetes in BMI categories: a systematic 60 Salas-Salvadó J, Fernández-Ballart J, Ros E, et al. Effect of a
review and network meta-analysis. Diabetes Care 2015; 38: 2177–87. Mediterranean diet supplemented with nuts on metabolic syndrome
38 Mathew H, Farr OM, Mantzoros CS. Metabolic health and weight: status: one-year results of the PREDIMED randomized trial.
understanding metabolically unhealthy normal weight or Arch Intern Med 2008; 168: 2449–58.
metabolically healthy obese patients. Metabolism 2016; 65: 73–80. 61 The Spanish Biomedical Research Centre in Physiopathology of
39 Kramer CK, Zinman B, Retnakaran R. Are metabolically healthy Obesity and Nutrition (CIBERobn). http://www.predimedplus.com/
overweight and obesity benign conditions?: a systematic review and en/project (accessed on May 31, 2017).
meta-analysis. Ann Intern Med 2013; 159: 758–69.

www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9 9

 Series

62 Howard BV, Van Horn L, Hsia J, et al. Low-fat dietary pattern and risk 67 Guo F, Garvey WT. Cardiometabolic Disease Staging predicts
of cardiovascular disease: the Women’s Health Initiative Randomized effectiveness of weight loss therapy to prevent type 2 diabetes:
Controlled Dietary Modification Trial. JAMA 2006; 295: 655–66. pooled results from phase III clinical trials assessing phentermine/
63 Bray GA, Siri-Tarino PW. The role of macronutrient content in the topiramate extended release. Diabetes Care 2017; 40: 856–62.
diet for weight management. Endocrinol Metab Clin North Am 2016; 68 Karelis AD. Metabolically healthy but obese individuals. Lancet
45: 581–604. 2008; 372: 1281–83.
64 Nordmann AJ, Nordmann A, Briel M, et al. Effects of low- 69 Stefan N, Schick F, Häring H-U. Causes, characteristics, and
carbohydrate vs low-fat diets on weight loss and cardiovascular risk consequences of metabolically unhealthy normal weight in
factors: a meta-analysis of randomized controlled trials. humans. Cell Metab 2017; 26: 292–300.
Arch Intern Med 2006; 166: 285–93. 70 Snook KR, Hansen AR, Duke CH, Finch KC, Hackney AA, Zhang J.
65 Johnston BC, Kanters S, Bandayrel K, et al. Comparison of weight Change in percentages of adults with overweight or obesity trying to
loss among named diet programs in overweight and obese adults: a lose weight, 1988–2014. JAMA 2017; 317: 971–73.
meta-analysis. JAMA 2014; 312: 923–33.
66 Wycherley TP, Moran LJ, Clifton PM, Noakes M, Brinkworth GD.
Effects of energy-restricted high-protein, low-fat compared with
standard-protein, low-fat diets: a meta-analysis of randomized
controlled trials. Am J Clin Nutr 2012; 96: 1281–98.

10 www.thelancet.com/diabetes-endocrinology Published online September 14, 2017 http://dx.doi.org/10.1016/S2213-8587(17)30292-9