Epidemiology/Population

Hypertension and Diabetes Mellitus

Coprediction and Time Trajectories
Vasilis Tsimihodimos, Clicerio Gonzalez-Villalpando, James B. Meigs, Ele Ferrannini

Abstract—Type 2 diabetes mellitus and hypertension overlap in the population. In many subjects, development of diabetes
mellitus is characterized by a relatively rapid increase in plasma glucose values. Whether a similar phenomenon occurs
during the development of hypertension is not known. We analyzed the pattern of blood pressure (BP) changes during
the development of hypertension in patients with or without diabetes mellitus using data from the MCDS (Mexico
City Diabetes Study; a population-based study of diabetes mellitus in Hispanic whites) and in the FOS (Framingham
Offspring Study, a community-based study in non-Hispanic whites) during a 7-year follow-up. Diabetes mellitus at
baseline was a significant predictor of incident hypertension (in FOS, odds ratio, 3.14; 95% confidence interval, 2.17–
4.54) independently of sex, age, body mass index, and familial diabetes mellitus. Conversely, hypertension at baseline
was an independent predictor of incident diabetes mellitus (in FOS, odds ratio, 3.33; 95% CI, 2.50–4.44). In >60% of the
converters, progression from normotension to hypertension was characterized by a steep increase in BP values, averaging
20 mm Hg for systolic BP within 3.5 years (in MCDS). In comparison with the nonconverters group, hypertension and
diabetes mellitus converters shared a metabolic syndrome phenotype (hyperinsulinemia, higher body mass index, waist
girth, BP, heart rate and pulse pressure, and dyslipidemia). Overall, results were similar in the 2 ethnic groups. We conclude
that (1) development of hypertension and diabetes mellitus track each other over time, (2) transition from normotension
to hypertension is characterized by a sharp increase in BP values, and (3) insulin resistance is one common feature of
both prediabetes and prehypertension and an antecedent of progression to 2 respective disease states.  (Hypertension.
2018;71:422-428. DOI: 10.1161/HYPERTENSIONAHA.117.10546.) Online Data Supplement •
Key Words: blood pressure ◼ diabetes mellitus ◼ heart rate ◼ hypertension ◼ insulin resistance
Downloaded from http://ahajournals.org by on November 15, 2024

D iabetes mellitus and hypertension are among the most

common diseases and cardiovascular risk factors, respec-
tively, worldwide, and their frequency increases with increas-
hypertension. This observation may reflect not only delayed
recognition of the presence of hypertension, clinical iner-
tia, and poor adherence to the prescribed regimen but also
ing age.1 Elevated blood pressure (BP) values are a common uncertainty regarding the treatment targets and pathogenic
finding in patients with type 2 diabetes mellitus (T2D) and are correlation.
thought to reflect, at least in part, the impact of the underly- A previous report from the MCDS (Mexico City Diabetes
ing insulin resistance on the vasculature and kidney.1 On the Study) showed that, in ≈2/3 of patients with either normo-
contrary, accumulating evidence suggests that disturbances in glycemia or impaired glucose tolerance, the development of
carbohydrate metabolism are more common in hypertensive overt diabetes mellitus is characterized by an abrupt (within
individuals,2,3 thereby indicating that the pathogenic relation- ≈3.5 years) increase in plasma glucose values by ≈50 mg/dL.7
ship between diabetes mellitus and hypertension is actually Whether a similar phenomenon is seen during the development
bidirectional. of hypertension is not known. Therefore, the first aim of the
The development of hypertension in diabetic individuals present analysis was to determine the pattern of BP changes
not only complicates treatment strategy and increases health- during the development of hypertension in patients with or
care costs but also heightens the risk for macrovascular and without diabetes mellitus in MCDS. The second aim was to
microvascular complications considerably.2,4 Although BP quantify the longitudinal association of T2D and hypertension
lowering is followed by a significant reduction in cardiovas- in this population-based study during the follow-up period of
cular and microvascular morbidity and mortality,5,6 a large 7 years. Within this scope, we tried to identify clinical and
proportion of diabetic subjects exhibit poorly controlled laboratory characteristics that may reflect an increased risk for

Received October 27, 2017; first decision November 7, 2017; revision accepted December 19, 2017.
From the Department of Clinical and Experimental Medicine, University of Pisa, Italy (V.T.); Centro de Estudios en Diabetes, Unidad de Investigacion
en Diabetes y Riesgo Cardiovascular, Centro de Investigacion en Salud Poblacional, Instituto Nacional de Salud Publica, Mexico City, Mexico (C.G.-V.);
Division of General Internal Medicine Division, Massachusetts General Hospital, Boston (J.B.M.); Department of Medicine, Harvard Medical School,
Boston, MA (J.B.M.); and CNR Institute of Clinical Physiology, Pisa, Italy (E.F.).
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.
117.10546/-/DC1.
Correspondence to Vasilis Tsimihodimos, Department of Clinical and Experimental Medicine, University of Pisa, Via Roma 67, Pisa 56100, Italy. E-mail
[email protected]
© 2018 American Heart Association, Inc.
Hypertension is available at http://hyper.ahajournals.org DOI: 10.1161/HYPERTENSIONAHA.117.10546

422
 Tsimihodimos et al   Time Trajectory of Incident Hypertension   423

the development of diabetes mellitus, hypertension, or both. and hip circumferences, and systolic and diastolic BP were measured;
Because the population of MCDS included Hispanic individu- pulse pressure was calculated as the difference between systolic and
diastolic BP and mean BP as the sum of diastolic BP and one third
als from low-income areas with a high risk for the develop-
of pulse pressure.
ment of diabetes mellitus, we explored the generalizability of
any results by asking the same questions in the non-Hispanic Biochemical Measurements
white population of the FOS (Framingham Offspring Study). Blood samples were obtained in the fasting state and 2 hours after a
standard 75-g oral glucose load. Serum samples were centrifuged,
Methods divided into aliquots, and stored at −70°C until assayed. Fasting con-
The authors declare that all supporting data are available within the centrations of serum insulin, proinsulin, plasma glucose, total cho-
article (and its online-only Data Supplement). lesterol, low-density lipoprotein cholesterol, high-density lipoprotein
cholesterol, triglycerides, and plasma glucose and insulin concentra-
Study Populations tions 2 hours after an oral glucose load were determined as described
elsewhere7 at baseline and at follow-up.
The MCDS is a population-based cohort participating in a longitu-
dinal survey of incident diabetes mellitus and cardiovascular risk
factors. Low-income neighborhoods in Mexico City were identified, Statistical Analysis
and a complete enumeration of these was performed from November The study design is that of a longitudinal cohort study of incident
1989 to October 1992. Among the 15 532 inhabitants of these neigh- diabetes mellitus as a function of baseline hypertension or of inci-
borhoods, 2280 men and women (aged 35–64 years) were randomly dent hypertension as a function of baseline diabetes mellitus, both
selected from 6 low-income colonias examined between 1990 and with confounder covariate adjustment. We also assessed the rate of
1992 and invited to return for 2 follow-up examinations, the first con- change of BP as hypertension developed. Baseline data are presented
ducted between 1993 and 1995 and the second between 1997 and as mean±SD; median and (interquartile range) are reported for vari-
1999. Of the 1770 subjects participating in the first follow-up (at 3.25 ables with non-normal distribution, which were log transformed for
years), 1753 returned for the second follow-up (at 7 years). The clini- use in statistical analyses. Categorical variables were compared by
cal characteristics of the subjects not returning for the second follow- the χ2 test, continuous variables by ANOVA, with Bonferroni–Dunn
up were essentially superimposable on those of the subjects who did testing of multiple post hoc comparisons. Logistic regression for inci-
(data not shown). dent events was performed by defining response as a diagnosis of
Examinations were standardized and included interviews, anthro- hypertension (or diabetes mellitus) at either of the 2 follow-up visits
pometry, BP measurements, a fasting blood draw, and a 75-g oral and nonresponse as no hypertension (or diabetes mellitus) at the last
glucose tolerance test. Trained interviewers obtained information on visit; results are expressed as the odds ratio (OR) with 95% confi-
medical history, medication use, and smoking status. dence intervals (95% CI) as a function of baseline exposures. For
The protocol was approved by the Ethics Committee of the each continuous variable in a multivariate model, OR was calculated
Centro de Estudios en Diabetes, Centro de Investigacion en Salud for 1 SD of the population value of that variable. A P value ≤0.05 was
Poblacional, Instituto Nacional de Salud Publica, Mexico City, and considered statistically significant for the test of the hypothesis that
Downloaded from http://ahajournals.org by on November 15, 2024

all subjects gave informed consent. diabetes mellitus predicts hypertension or vice versa.
The FOS (Framingham Offspring Study) is a community-based
cohort including 3754 men and women who attended the fifth clinic
examination (1989–1992) of the FHS. Participants were followed up Results
from baseline to the sixth (1995–1998) and seventh (1998–2001) off-
spring exams, for an average period of 7 years. We used the exam Development of Hypertension
visit date when a new case of diabetes mellitus or hypertension was At the 3 examinations, 16% to 46% of the study subjects were
identified as the date of diagnosis; otherwise, follow-up was censored hypertensive; among them, the prevalence of diabetes mellitus
at last follow-up (examination 6 or 7) for participants remaining non-
diabetic or nonhypertensive. The protocol was approved by the Ethics (20%–39%) was significantly higher than that among normo-
Committee of Boston University Medical Center, and all subjects tensive subjects (P<0.0001 for all 3 data sets; Table 1). Among
gave written informed consent. subjects who were normotensive at baseline (n=1876), 108
In both cohorts, hypertension was defined as a systolic BP ≥140 became hypertensive at 3.25 years; another 107 subjects who
mm Hg or a diastolic BP ≥90 mm Hg or current antihypertensive were normotensive at both baseline and 3.25 years were found
treatment. In both studies, subjects whose BP was <140/90 mm Hg at
baseline and both follow-up visits were classified as normotensives, to be hypertensive at 7 years, and 28 other subjects who were
those whose BP was <140/90 mm Hg at the first visit who became normotensive at baseline and missed examination 2 were
hypertensive at the second or third visit were classified as converters. hypertensive at examination 3. Thus, a total of 243 subjects
T2D was classified as a fasting plasma glucose concentration ≥126 converted to hypertension during the 7-year follow-up, yield-
mg/dL or a 2-hour plasma glucose concentration ≥200 mg/dL on a
standard 75-g oral glucose tolerance test. Subjects who gave a his-
ing a crude conversion rate of ≈2% per year.
tory of diabetes mellitus and who at the time of their clinical exami- In comparison with subjects who were seen and found
nation were taking oral antidiabetic agents were also considered to to be normotensive at all 3 examinations (nonconverters),
have T2D regardless of their plasma glucose values. Insulin-taking converters to hypertension were older, heavier with a more
diabetic subjects whose age of onset was ≥40 years or whose body central fat distribution, and had higher systolic and diastolic
mass index (BMI) was >30 kg/m2 were also considered to have T2D.
Subjects with type 1 diabetes mellitus were excluded. Subjects who BP values and higher pulse rate at baseline regardless of
developed diabetes mellitus at the first or second follow-up were de- their time of conversion. Diabetes mellitus was more preva-
noted as converters. Subjects who tested normal on the oral glucose lent among either group of converters than in nonconverters
tolerance test on all 3 examinations were considered to be bona fide (Table 2). Moreover, among normotensive individuals, diabe-
nonconverters during the observation period.
tes mellitus at baseline was a significant predictor of incident
hypertension (in FOS, OR, 3.14; 95% CI, 2.17–4.54) inde-
Anthropometric Measurements
Diabetes mellitus in at least one parent or sibling was coded as a posi- pendently of age, BMI, and family history of diabetes mellitus
tive family history of diabetes mellitus. Before examinations, all par- (Figure 1). Of note, when the baseline mean BP was included
ticipants were asked to fast for at least 12 hours. Height, weight, waist in the model, the predictive value of diabetes mellitus was
 424  Hypertension  March 2018

Table 1. The MCDS (Mexico City Diabetes Study) and the FOS (Framingham
Offspring Study)*
Baseline First Follow-Up Second Follow-Up
Patients MCDS FOS MCDS FOS MCDS FOS
n 2280 3754 1770 3353 1753 3132
NT, n (%) 1876 (82) 2461 (66) 1487 (84) 1947 (58) 1381 (79) 1678 (54)
 ND 1656 (88) 2321 (94) 1198 (81) 1814 (93) 1102 (80) 1541 (92)
 D 220 (12) 140 (6) 289 (19) 133 (7) 279 (20) 137 (8)
HT, n (%) 404 (18) 1293 (34 283 (16) 1406 (42) 372 (21) 1454 (46)
 ND 310 (77) 1040 (80) 199 (70) 1406 (80) 227 (61) 1097 (75)
 D 94 (23) 253 (20) 84 (30) 283 (20) 145 (39) 357 (25)
D indicates diabetic; HT, hypertensive; ND, nondiabetic; and NT, normotensive.
*Number (%) of subjects examined at baseline, first follow-up (3.25 y in MCDS and 4 y in FOS), and
second follow-up (7 y in MCDS and 9 y in FOS) by blood pressure (NT and HT) and glucose tolerance status
(ND and D).

attenuated and became nonsignificant (in MCDS, OR, 1.44; The presence of diabetes mellitus did not consistently affect
95% CI, 0.97–2.14). the pattern of BP change in patients developing hypertension
All subject groups exhibited weight gain during the during the follow-up. Thus, in MCDS patients not receiving
observation period independently of the conversion status or antihypertensive medications, the increase in systolic BP in
the time of conversion. In MCDS, the increase in BMI was those converting to hypertension at examination 2 was 18
a significant independent predictor of incident hypertension mm Hg if nondiabetic and 20 mm Hg if diabetic. On the con-
(the hazard ratio for 1 SD change in BMI was 1.31; 95% CI, trary, the corresponding changes in systolic BP for patients
1.12–1.55 and in the same model, the corresponding hazard converting at examination 3 were 27 mm Hg in diabetic versus
ratio for the presence of diabetes mellitus at baseline was 1.79; 17 mm Hg in nondiabetic patients (P<0.05).
95% CI, 1.14–2.77). On conversion, both systolic and dia-
stolic BP values rose markedly and similarly in both groups Development of Diabetes Mellitus
Downloaded from http://ahajournals.org by on November 15, 2024

of hypertension converters (Figure 2). Using only the data Among subjects who were nondiabetic at baseline (n=1966),
of MCDS subjects not receiving antihypertensive treatment, 89 had developed diabetes mellitus by 3.25 years; another 71
the rise in systolic BP was 19 (14) mm Hg in subjects con- subjects who were nondiabetic at both baseline and 3.25 years
verting at examination 2 (n=65) and 19 (17) mm Hg in those were found to be diabetic at 7 years, and 10 other subjects
(n=60) converting at examination 3. Values higher than the who were nondiabetic at baseline and missed examination 2
90th percentile of the changes in systolic BP observed in non- were diabetic at examination 3. Thus, a total of 170 subjects
converters were found in 70% of the subjects converting at converted to diabetes mellitus during the 7-year follow-up,
examination 2 and in 58% of those converting at examination yielding a crude conversion rate of 1.2% per year. Among
3. Similar changes were observed in the converters of FOS. nondiabetic individuals, hypertension at baseline was more

Table 2. Clinical Phenotype of Normotensive Subjects Converting to Hypertension at the First or the
Second Follow-Up and in Subjects Remaining Normotensive at All 3 Examinations (Nonconverters)
First Follow-Up Second Follow-Up Nonconverters
Data Set MCDS FOS MCDS FOS MCDS FOS P Value*
N 108 436 107 235 1154 1443 …
Sex (M/F) 39/69 218/218* 38/69 113/122 479/675 606/837 <0.01
Age, y 51±8† 61±9† 50±8† 62±9† 45±8 58±9 <0.0001
BMI, kg/m2 29.3±4.4† 28.3±5.1† 28.6±4.4† 28.7±4.9† 27.6±4.2 26.9±4.6 <0.0001
Waist, cm 100±11† 99±13† 98±11† 102±13† 95±12 97±13 <0.0001
Systolic BP, mm Hg 124±10† 126±5† 119±9† 124±7† 111±11 116±11 <0.0001
Diastolic BP, mm Hg 76±7† 80±10† 74±9† 79±10† 70±8 71±8 <0.0001
Pulse rate, bpm 76±10† … 74±9† … 72±9 … <0.0001
Diabetes mellitus, % 50† 13† 41† 16† 34 7 <0.0001
BMI indicates body mass index; BP, blood pressure; FOS, Framingham Offspring Study; and MCDS, Mexico City Diabetes Study.
*ANOVA or χ2 for the 3 groups for each data set.
†P≤0.05 vs nonconverters by Bonferroni–Dunn test for each data set.
 Tsimihodimos et al   Time Trajectory of Incident Hypertension   425

Figure 3. Multivariate logistic analysis of incident diabetes

Figure 1. Multivariate logistic analysis of incident hypertension
mellitus in nondiabetic individuals in the MCDS (Mexico City
in normotensive individuals in the MCDS (Mexico City Diabetes
Diabetes Study) and FOS (Framingham Offspring Study). Plots are
Study) and FOS (Framingham Offspring Study). Plots are odds
odds ratios and 95% confidence intervals (CI; calculated for 1 SD
ratios and 95% confidence intervals (CI; calculated for 1 SD of
of continuous predictor variable; data are adjusted for smoking
continuous predictor variable). BMI indicates body mass index.
categorized as ever or not). BMI indicates body mass index.

prevalent among diabetes mellitus converters than noncon-

verters (25% versus 15%; P=0.001) and was a significant at baseline in the same model was 1.80; 95% CI, 1.03–3.04).
predictor of incident diabetes mellitus (in FOS, OR, 3.33; Among the 1656 participants who were normotensive and
95% CI, 2.50–4.44) independently of sex, age, BMI, and nondiabetic at baseline, 104 had converted to diabetes mellitus
familial diabetes mellitus (Figure 3). Again, the increase in at 7 years, 165 to hypertension, and 24 to both diabetes mel-
BMI during the observation period was a significant predic- litus and hypertension. In comparison with the nonconverters
tor of incident diabetes mellitus (in MCDS, the hazard ratio group, hypertension and diabetes mellitus converters shared
for 1 SD change in BMI was 1.36; 95% CI, 1.16–1.60 and most phenotypic traits, namely, higher BMI, waist girth, BP,
the corresponding hazard ratio for presence of hypertension heart rate and pulse pressure values, serum triglycerides, and
Downloaded from http://ahajournals.org by on November 15, 2024

plasma insulin concentrations (both fasting and postload;

Table S1 in the online-only Data Supplement). In addition, in
MCDS, age was higher in hypertension converters but not in
diabetes mellitus converters (in FOS, age was higher in dia-
betes mellitus converters too), whereas familial diabetes mel-
litus, fasting and 2-hour plasma glucose, and serum proinsulin
concentrations were higher in diabetes mellitus converters but
not in hypertension converters. All these anthropometric and
metabolic differences from the control group were accentu-
ated in the double converters (Table S1 in the online-only
Data Supplement). In multivariate analysis, age, BMI, fasting
plasma glucose, mean BP, and 2-hour plasma insulin concen-
trations were independent risk factors for the development of
either hypertension or T2D (Figure 4).

Discussion
The first main finding of the present study is that not only
does the presence of hypertension predict future diabetes
mellitus, in agreement with earlier epidemiological observa-
tions,2,3,8,9 but also the incidence of hypertension increases
significantly in the presence of diabetes mellitus. During
the 7 years of follow-up, BP behaved as a tracking variable
as individuals who converted to hypertension (at the first or
second follow-up visit) had increased baseline BP values
Figure 2. Systolic and diastolic blood pressure (BP) values in compared with nonconverters, although still within the nor-
normotensive subjects who remained normotensive at both follow- mal range.10 Indeed, baseline BP was the strongest predictor
up examinations (black dashed lines), became hypertensive at of incident hypertension, and its inclusion in the statistical
the first examination (gray lines), or became hypertensive at the model significantly attenuated the predictive value of diabetes
second examination (black solid lines). Plots are mean±SD. The
dotted gray lines indicate that the apparent decline in BP values mellitus. More strikingly, hypertension and diabetes mellitus
after diagnosis are likely because of antihypertensive treatment. tracked each other consistently (Figures 1 and 3), and people
 426  Hypertension  March 2018

one common feature of both prediabetes and prehypertension,

and one antecedent of progression to the 2 respective disease
states.
Apart from the detrimental effects that disturbed insulin
signaling exerts on carbohydrate metabolism, the hyperinsu-
linemia that characterizes insulin resistance states leads to vas-
cular smooth muscle cell proliferation and increased vascular
stiffness, which predispose to the development of hyperten-
sion.11 Additionally, insulin may directly or indirectly impair
vasodilation and increase oxidative stress and the inflamma-
tory process in the vascular wall.12,13 The sum of these effects
is the impaired autoregulation of vascular tone, increased vas-
cular resistance, and BP elevation. Finally, the antinatriuretic
properties of insulin increase renal retention of sodium and
Figure 4. Multivariate logistic analysis of incident hypertension
or diabetes mellitus in the Mexico City Diabetes Study. Plots are water leading to volume overload, thereby predisposing to the
odds ratios and 95% confidence intervals (CI; calculated for 1 SD development of hypertension.14
of continuous predictor variable). BMI indicates body mass index. A novel finding from both study cohorts is that, in indi-
viduals who ultimately develop T2D, hypertension, or both,
at high risk for the development of either hypertension or dia- the time trajectory of plasma glucose7 and BP values is not
betes mellitus share common metabolic abnormalities, that is, a progressive slow increase but—in the majority of cases—
abdominal obesity, hyperinsulinemia, and hypertriglyceride- a steep elevation several-fold larger compared with changes
mia (even more prominent in those destined to develop both observed in nonconverters (or, in the case of patients con-
abnormalities). Thus, the general population contains a pool verting at the second follow-up, compared with the changes
of individuals with the phenotype of the metabolic (or insulin observed in the same patients between baseline and the first
resistance) syndrome from which new hypertension or diabe- follow-up visit). Although the pathophysiological basis of
tes mellitus (or both) emerge over time. Importantly, weight this relatively acute decompensation remains indeterminate,
gain may be one factor that contributes to the development of it could be hypothesized that it may be related to sympathetic
both hypertension and diabetes mellitus. Parenthetically, the excitation. The sequence of events that lead to activation
increased incidence of hypertension in patients with diabetes of the sympathetic nervous system is unknown. However,
Downloaded from http://ahajournals.org by on November 15, 2024

mellitus may also reflect the closer surveillance of these indi- in healthy volunteers, insulin dose dependently stimulates
viduals (ie, a small detection bias). The second, and possibly norepinephrine release, particularly in skeletal muscle, and
the most important, finding of this study is that the progres- enhances sympathetic neuronal discharge.15 In subjects
sion from normotension to hypertension in individuals des- with uncomplicated obesity monitored for 24 hours, there
tined to become hypertensive is marked by a steep increase is episodic sympathetic dominance in phase with postpran-
in BP values averaging 20 mm Hg for systolic BP within 3.5 dial hyperinsulinemia, which abates after weight loss.16 The
years. In >60% of the converters, the increase in BP values autonomic contribution to BP is greater in obesity, and gan-
during the period that preceded conversion was greater than glionic blockade of the autonomic nervous system results in
the 90th percentile of the changes in systolic BP observed BP decrease that is more pronounced in obese individuals.17
in nonconverters. This biphasic BP pattern is similar to that Obese subjects with hypertension display increased sympa-
previously described for blood glucose values in MCDS indi- thetic nerve activity, an abnormality that is partially corrected
viduals developing diabetes mellitus.7 Finally, both the copre- after diet-induced weight loss.18 Leptin, an adipokine that has
diction of hypertension and diabetes mellitus and this biphasic been found to circulate in increased concentrations in obese
pattern of progression are not unique to Hispanic individuals and insulin resistant subjects, can act centrally to activate the
because essentially the same findings were observed in the sympathetic nervous system19; not all studies have confirmed
non-Hispanic white population of FOS. this hypothesis.20 In addition, experimental models suggest
One potential factor responsible for the covariance of dia- that leptin may also contribute to the pathogenesis of hyper-
betes mellitus and hypertension is insulin resistance.1 Of note, tension via aldosterone-dependent mechanisms.21 In line with
in a subcohort of FOS with a shorter follow-up, an inverse these suggestions, in our population, BMI values at baseline
association between incident hypertension (or BP progres- and weight gain during the observation period were signifi-
sion) and a proxy of insulin resistance was seen principally in cant predictors of both incident hypertension and diabetes
younger people.10 Here, however, both fasting plasma insulin mellitus, whereas heart rate and pulse pressure, both raw
(a typical proxy for insulin resistance in epidemiological stud- indices of sympathetic nervous system activity, were found
ies) and plasma insulin concentrations 2 hours after glucose to be elevated in patients who converted to hypertension.
ingestion were consistently higher at baseline in both hyper- Finally, obese individuals with or without diabetes mellitus
tension and diabetes mellitus converters. Furthermore, base- have been shown to have reduced concentrations of circulat-
line insulin levels copredicted both hypertension and diabetes ing natriuretic peptides. Because these molecules favorably
mellitus after controlling for age and BMI and also for base- affect intravascular volume status and vascular tone, this
line BP and plasma glucose values (Figure 4). This pattern of mechanism may be involved in the pathogenesis of hyperten-
results lends support to the notion that insulin resistance is sion in patients with diabetes mellitus.22
 Tsimihodimos et al   Time Trajectory of Incident Hypertension   427

Our findings may have implications in the everyday care Disclosures

of patients with diabetes mellitus. Thus, diabetic patients with None.
BP values near the upper limit of normal should be monitored
for the development of hypertension, especially if they have References
a positive family history of hypertension and the phenotypic 1. Ferrannini E, Cushman WC. Diabetes and hypertension: the bad compan-
features of the metabolic syndrome. Because development of ions. Lancet. 2012;380:601–610. doi: 10.1016/S0140-6736(12)60987-8.
2. Perreault L, Pan Q, Aroda VR, Barrett-Connor E, Dabelea D, Dagogo-Jack
hypertension in patients with diabetes mellitus is marked by a S, Hamman RF, Kahn SE, Mather KJ, Knowler WC; Diabetes Prevention
significant increase in macrovascular and microvascular risk,2,23 Program Research Group. Exploring residual risk for diabetes and micro-
efforts should be made to delay or ideally prevent the increase in vascular disease in the Diabetes Prevention Program Outcomes Study
(DPPOS). Diabet Med. 2017;34:1747–1755. doi: 10.1111/dme.13453.
BP. Obviously, the follow-up scheme of both MCDS and FOS
3. Wei GS, Coady SA, Goff DC Jr, Brancati FL, Levy D, Selvin E, Vasan
does not reflect everyday clinical practice, as is generally true of RS, Fox CS. Blood pressure and the risk of developing diabetes in African
observational population-based studies. Under ideal conditions, Americans and whites: ARIC, CARDIA, and the Framingham Heart
patients with diabetes mellitus or hypertension are seen 2 or 3 Study. Diabetes Care. 2011;34:873–879. doi: 10.2337/dc10-1786.
4. Aroda VR, Knowler WC, Crandall JP, Perreault L, Edelstein SL, Jeffries SL,
times per year. However, the time pattern of BP progression we Molitch ME, Pi-Sunyer X, Darwin C, Heckman-Stoddard BM, Temprosa
describe here may still emerge from more frequent follow-up M, Kahn SE, Nathan DM; Diabetes Prevention Program Research Group.
visits. On the contrary, in an era of continuously increasing pres- Metformin for diabetes prevention: insights gained from the Diabetes
sures on healthcare systems, understanding the factors that pre- Prevention Program/Diabetes Prevention Program Outcomes Study.
Diabetologia. 2017;60:1601–1611. doi: 10.1007/s00125-017-4361-9.
dispose to, or precipitate, the development of an outcome should 5. Hansson L, Zanchetti A, Carruthers SG, Dahlöf B, Elmfeldt D, Julius S,
increase clinicians’ awareness and may facilitate the timely diag- Ménard J, Rahn KH, Wedel H, Westerling S. Effects of intensive blood-
nosis of conditions that might otherwise go unnoticed. pressure lowering and low-dose aspirin in patients with hypertension:
Apart from lifestyle modification, several classes of anti- principal results of the Hypertension Optimal Treatment (HOT) ran-
domised trial. HOT Study Group. Lancet. 1998;351:1755–1762.
diabetic drugs such as SGLT2 (sodium-glucose cotransporter 6. Zoungas S, de Galan BE, Ninomiya T, et al; ADVANCE Collaborative
2) inhibitors and GLP-1 (glucagon-like peptide 1) receptor ago- Group. Combined effects of routine blood pressure lowering and intensive
nists have been shown to lower BP24,25 (although the data for glucose control on macrovascular and microvascular outcomes in patients
with type 2 diabetes: new results from the ADVANCE trial. Diabetes
liraglutide is less convincing26) and reduce cardiovascular events
Care. 2009;32:2068–2074. doi: 10.2337/dc09-0959.
in secondary prevention.25,27 Thus, use of these drugs might be 7. Ferrannini E, Nannipieri M, Williams K, Gonzales C, Haffner SM, Stern
prioritized in diabetic patients at high risk for the development of MP. Mode of onset of type 2 diabetes from normal or impaired glucose
hypertension, although the clinical value of this strategy in terms tolerance. Diabetes. 2004;53:160–165.
8. Stamler J, Stamler R, Rhomberg P, Dyer A, Berkson DM, Reedus W,
of hard end point reduction has been unequivocally proven only Wannamaker J. Multivariate analysis of the relationship of six variables to
Downloaded from http://ahajournals.org by on November 15, 2024

in individuals with established cardiovascular disease. blood pressure: findings from Chicago community surveys, 1965–1971. J
Chronic Dis. 1975;28:499–525.
Perspectives 9. Emdin CA, Anderson SG, Woodward M, Rahimi K. Usual blood pressure
and risk of new-onset diabetes: evidence from 4.1 million adults and a
The development of hypertension and diabetes mellitus pre-
meta-analysis of prospective studies. J Am Coll Cardiol. 2015;66:1552–
dicts each other over time. The transition from normoten- 1562. doi: 10.1016/j.jacc.2015.07.059.
sion to hypertension is characterized by a sharp increase in 10. Arnlöv J, Pencina MJ, Nam BH, Meigs JB, Fox CS, Levy D, D’Agostino
BP values. Insulin resistance is one common feature of both RB, Vasan RS. Relations of insulin sensitivity to longitudinal blood
pressure tracking: variations with baseline age, body mass index,
prediabetes and prehypertension, and an antecedent of pro- and blood pressure. Circulation. 2005;112:1719–1727. doi: 10.1161/
gression to 2 respective disease states, especially in indi- CIRCULATIONAHA.105.535039.
viduals who gain weight over time. Because development of 11. McEniery CM, Wilkinson IB, Johansen NB, Witte DR, Singh-Manoux A,
hypertension in patients with diabetes mellitus is marked by a Kivimaki M, Tabak AG, Brunner EJ, Shipley MJ. Nondiabetic glucometa-
bolic status and progression of aortic stiffness: the Whitehall II Study.
significant increase in macrovascular and microvascular risk, Diabetes Care. 2017;40:599–606. doi: 10.2337/dc16-1773.
efforts should be made to delay or ideally prevent the increase 12. Potenza MA, Addabbo F, Montagnani M. Vascular actions of insulin with
in BP. In this context, the prioritization of antidiabetic drugs implications for endothelial dysfunction. Am J Physiol Endocrinol Metab.
that reduce BP (such as SGLT2 inhibitors or GLP-1 receptor 2009;297:E568–E577. doi: 10.1152/ajpendo.00297.2009.
13. Anfossi G, Russo I, Doronzo G, Trovati M. Contribution of insulin resis-
agonists) in patients with diabetes mellitus at high risk for the tance to vascular dysfunction. Arch Physiol Biochem. 2009;115:199–217.
development of hypertension may be of clinical value. doi: 10.1080/13813450903136791.
14. Nakamura M, Satoh N, Suzuki M, Kume H, Homma Y, Seki G, Horita S.
Sources of Funding Stimulatory effect of insulin on renal proximal tubule sodium transport
J.B. Meigs is supported in part by National Institutes of Health is preserved in type 2 diabetes with nephropathy. Biochem Biophys Res
Commun. 2015;461:154–158. doi: 10.1016/j.bbrc.2015.04.005.
(NIH) National Institute of Diabetes and Digestive and Kidney
15. Lembo G, Napoli R, Capaldo B, Rendina V, Iaccarino G, Volpe M,
Diseases K24 DK080140; partial support was provided by funds
Trimarco B, Saccà L. Abnormal sympathetic overactivity evoked by insu-
from a European Medical Information Framework grant (IMI JU GA lin in the skeletal muscle of patients with essential hypertension. J Clin
115372-2). The Framingham Offspring Study is supported by NIH Invest. 1992;90:24–29. doi: 10.1172/JCI115842.
National Heart, Lung, and Blood Institute Framingham Heart Study 16. Emdin M, Gastaldelli A, Muscelli E, Macerata A, Natali A, Camastra S,
contracts N01-HC-25195 and HHSN268201500001I. The MCDS Ferrannini E. Hyperinsulinemia and autonomic nervous system dysfunc-
(Mexico City Diabetes Study) has been supported by the following tion in obesity: effects of weight loss. Circulation. 2001;103:513–519.
grants: RO1HL 24799 from the National Heart, Lung, and Blood 17. Shibao C, Gamboa A, Diedrich A, Ertl AC, Chen KY, Byrne DW, Farley
Institute; Consejo Nacional de Ciencia y Tecnología 2092, M9303, G, Paranjape SY, Davis SN, Biaggioni I. Autonomic contribution to blood
F677-M9407, 251M, 2005-C01-14502, and SALUD 2010-2151165; pressure and metabolism in obesity. Hypertension. 2007;49:27–33. doi:
and Consejo Nacional de Ciencia y Tecnología (CONACyT) [Fondo 10.1161/01.HYP.0000251679.87348.05.
de Cooperación Internacional en Ciencia y Tecnología (FONCICYT)] 18. Lambert E, Straznicky N, Eikelis N, Esler M, Dawood T, Masuo K,
C0012-2014-01-247974. Schlaich M, Lambert G. Gender differences in sympathetic nervous
 428  Hypertension  March 2018

activity: influence of body mass and blood pressure. J Hypertens. 24. Chilton R, Tikkanen I, Hehnke U, Woerle HJ, Johansen OE. Impact
2007;25:1411–1419. doi: 10.1097/HJH.0b013e3281053af4. of empagliflozin on blood pressure in dipper and non-dipper patients
19. Han C, Wu W, Ale A, Kim MS, Cai D. Central leptin and tumor necrosis with type 2 diabetes mellitus and hypertension. Diabetes Obes Metab.
factor-α (TNFα) in diurnal control of blood pressure and hypertension. 2017;19:1620–1624. doi: 10.1111/dom.12962.
J Biol Chem. 2016;291:15131–15142. doi: 10.1074/jbc.M116.730408. 25. Marso SP, Daniels GH, Brown-Frandsen K, Kristensen P, Mann JF,
20. Brown RJ, Meehan CA, Gorden P. Leptin does not mediate hypertension Nauck MA, Nissen SE, Pocock S, Poulter NR, Ravn LS, Steinberg WM,
associated with human obesity. Cell. 2015;162:465–466. doi: 10.1016/j. Stockner M, Zinman B, Bergenstal RM, Buse JB; LEADER Steering
cell.2015.07.007. Committee; LEADER Trial Investigators. Liraglutide and cardiovascu-
21. Huby AC, Otvos L Jr, Belin de Chantemèle EJ. Leptin induces hyperten- lar outcomes in type 2 diabetes. N Engl J Med. 2016;375:311–322. doi:
sion and endothelial dysfunction via aldosterone-dependent mechanisms 10.1056/NEJMoa1603827.
in obese female mice. Hypertension. 2016;67:1020–1028. doi: 10.1161/ 26. Kumarathurai P, Anholm C, Fabricius-Bjerre A, Nielsen OW, Kristiansen
HYPERTENSIONAHA.115.06642. O, Madsbad S, Haugaard SB, Sajadieh A. Effects of the glucagon-like
22. Zois NE, Bartels ED, Hunter I, Kousholt BS, Olsen LH, Goetze JP. peptide-1 receptor agonist liraglutide on 24-h ambulatory blood pressure
Natriuretic peptides in cardiometabolic regulation and disease. Nat Rev in patients with type 2 diabetes and stable coronary artery disease: a ran-
Cardiol. 2014;11:403–412. domized, double-blind, placebo-controlled, crossover study. J Hypertens.
23. Navar AM, Gallup DS, Lokhnygina Y, Green JB, McGuire DK, Armstrong 2017;35:1070–1078. doi: 10.1097/HJH.0000000000001275.
PW, Buse JB, Engel SS, Lachin JM, Standl E, Van de Werf F, Holman RR, 27. Zinman B, Wanner C, Lachin JM, Fitchett D, Bluhmki E, Hantel S,
Peterson ED; TECOS Study Group. Hypertension control in adults with Mattheus M, Devins T, Johansen OE, Woerle HJ, Broedl UC, Inzucchi
diabetes mellitus and recurrent cardiovascular events: global results from SE; EMPA-REG OUTCOME Investigators. Empagliflozin, cardio-
the trial evaluating cardiovascular outcomes with sitagliptin. Hypertension. vascular outcomes, and mortality in type 2 diabetes. N Engl J Med.
2017;70:907–914. doi: 10.1161/HYPERTENSIONAHA.117.09482. 2015;373:2117–2128. doi: 10.1056/NEJMoa1504720.

Novelty and Significance

What Is New? Summary
• Diabetes mellitus and hypertension copredict each other. The development of hypertension and diabetes mellitus predict
• The progression to hypertension is marked by a steep increase in blood
each other over time. The transition from normotension to hyper-
pressure values.
tension is characterized by a sharp increase in blood pressure val-
What Is Relevant? ues. Insulin resistance is one common feature of both prediabetes
and prehypertension and an antecedent of progression to 2 respec-
• Diabetic patients with blood pressure values near the upper limit of nor-
mal should be monitored for the development of hypertension. tive disease states.
• Antidiabetic drugs that reduce blood pressure should be prioritized in
Downloaded from http://ahajournals.org by on November 15, 2024

diabetic patients at high risk for the development of hypertension.