Ceriello 

and Prattichizzo Cardiovasc Diabetol

https://doi.org/10.1186/s12933-021-01289-4
(2021) 20:101
Cardiovascular Diabetology

REVIEW Open Access

Variability of risk factors and diabetes

complications
Antonio Ceriello* and Francesco Prattichizzo*   

Abstract 
Several studies suggest that, together with glucose variability, the variability of other risk factors, as blood pressure,
plasma lipids, heart rate, body weight, and serum uric acid, might play a role in the development of diabetes compli-
cations. Moreover, the variability of each risk factor, when contemporarily present, may have additive effects. However,
the question is whether variability is causal or a marker. Evidence shows that the quality of care and the attainment
of the target impact on the variability of all risk factors. On the other hand, for some of them causality may be con-
sidered. Although specific studies are still lacking, it should be useful checking the variability of a risk factor, together
with its magnitude out of the normal range, in clinical practice. This can lead to an improvement of the quality of care,
which, in turn, could further hesitate in an improvement of risk factors variability.
Keywords:  Diabetes mellitus, Glucose variability, Blood pressure variability, Lipids variability, Body weight variability,
Uric acid variability, Heart rate variability, Oxidative stress, Cardiovascular complications, Microvascular complications

Background [8] confirm that in type 2 diabetes (T2D) long-term GV is

Growing attention has been recently paid to the possible correlated with an increased risk of both CV and micro-
the role of glucose variability (GV) in the development of vascular complications. However, we need to underline
diabetic complications, particularly cardiovascular (CV) that while we pay much attention to GV, [1] emerging
ones [1]. evidence suggests that also the long-term variability of
Many observational [1] and some interventional stud- other risk factors may be involved in the development of
ies [2], as well as post-hoc analyses of trials such as the diabetes complications.
“Action in Diabetes and Vascular Disease: Preterax and Long-term variability is defined as the fluctuations of
Diamicron MR Controlled Evaluation” (ADVANCE), a certain risk factor outside the recommended range in
[3] the “Trial Comparing Cardiovascular Safety of Insu- successive measures [1]. While a number of metrics has
lin Degludec vs. Insulin Glargine in Patients With Type been proposed to assess long-term variability, the most
2 Diabetes at High Risk of Cardiovascular Events” commonly used is the standard deviation (SD) of the col-
(DEVOTE), [4] the “Veterans Affairs Diabetes Trial” lected values [1].
(VADT), [5] the “Antihypertensive and Lipid-Lowering Common risk factors for the development of diabetes
Treatment to Prevent Heart Attack Trial” (HALLHAT), complications include blood pressure, lipid parameters
[6] the “Action to Control Cardiovascular Risk in Diabe- (i.e. total, HDL-, and LDL-cholesterol and triglycerides),
tes” (ACCORD) trial, [7] and the “Empagliflozin Cardio- heart rate, body weight, and uric acid. Here we summa-
vascular Outcome Event (EMPA-REG OUTCOME) trial rize the key available literature describing the variabil-
ity of such risk factors in relation to the development of
complications in patients with diabetes. Then, we syn-
*Correspondence: antonio.ceriello@hotmail.it; francesco. thesize the therapeutic options that seem available at the
prattichizzo@multimedica.it moment to face this new emerging challenge of diabetes
IRCCS MultiMedica, Via Gaudenzio Fantoli, 16/15, 20138 Milan, Italy management.

© The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and
the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material
in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material
is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the
permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​
mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​mmons.​org/​publi​cdoma​in/​
zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 2 of 11

Variability of risk factors and diabetic the “The Rio de Janeiro Type 2 Diabetes Cohort Study”,
complications SBP-visit-to-visit variability emerged as an independent
Blood pressure predictor of MACE (hazard ratio: 1.25, 95% CI 1.03–1.51
Grove et al. [9] originally observed variability in visit-to- for a 1-SD increase in 24-month SD), but not of total CV
visit systolic blood pressure (SBP) was related to inci- events, CV and all-cause mortality, and of any microvas-
dence of coronary heart disease in a population with or cular outcome [14].
without hypertension. Very recently, pooled data from the ACCORD and
Several studies, after this finding, have reported that a VADT Trials have shown that both systolic and diastolic
blood pressure (both systolic and diastolic) variability in blood pressure variability can favor the development of
people with diabetes is an independent predictor of both heart failure in T2D, an effect likely mediated by dips, not
macro- and microvascular complications, particularly elevations, of blood pressure [15].
nephropathy. A post-hoc analysis pooling data from the Irbesartan
The ADVANCE was a factorial randomized controlled Diabetic Nephropathy Trial (IDNT) and the “Reduc-
trial to test the effect of a tight control of both glucose tion of End Points in Non-Insulin-Dependent Diabetes
and blood pressure in patients with T2D on major With the Angiotensin II Antagonist Losartan” (RENAAL)
adverse cardiovascular events (MACE), i.e. myocardial Study explored the effect of long-term blood pressure
infarction, stroke, or CV death and microvascular end- variability on CV and renal endpoints in 2739 partici-
points, i.e. new or worsening nephropathy or retinopa- pants with T2D and nephropathy [16]. The renal end-
thy [10]. SBP was measured at six successive visits for point was a composite of time to confirmed doubling of
24 months after randomization and was used to estimate serum creatinine level, development of end-stage renal
its variability, defined as SD. During a median 2.4 years of disease, or death, while the CV outcome was defined as
follow-up, the SD of SBP variability was associated with CV death, myocardial infarction, stroke, hospitaliza-
the incidence of both MACE and microvascular events tion for heart failure, or revascularization [16]. Larger
despite multiple adjustments for a plethora of variables SBP visit-to-visit variability was independently asso-
including mean SBP [10]. ciated with an increased risk for the composite renal
The ADVANCE-ON (ADVANCE-Observational) fol- endpoint, but not with the CV outcome [16]. Similarly,
lowed-up 9114 patients not experiencing MACE, renal another study taking advantage of clinical records from
events, or death during the active phase of the trial for 30,851 T2D patients with hypertension and normal
an additional observational follow-up of 7.6  years after estimated glomerular filtration rate (eGFR) at baseline
termination of the treatments. The SD of SBP, measured found that an increased long-term blood pressure vari-
during the 24 months of active phase of the trial, was log- ability predicted kidney disease, defined as a composite
linearly and independently associated with an increased of eGFR less than 60 and/or a decrease in eGFR at least
risk of the primary outcome, i.e. MACE, renal events, 30% from baseline levels during a 4-year follow-up [17].
or death, during the protracted follow-up, extending the Of note, SBP variability was measured by three metrics:
previous findings to a longer-term range [11]. coefficient of variation, SD of the mean SBP and aver-
A retrospective cohort study was conducted in primary age absolute difference of successive values in each indi-
care and analyzed 124,105 Chinese adults with T2D and vidual, all providing similar results [17]. Several clinical
without prior diagnosed CV disease (CVD) [12]. Dur- characteristics (older age, male sex, SBP, diastolic blood
ing a median follow-up of 39.5 months, a positive linear pressure, albuminuria, glycated hemoglobin, insulin
relationship was observed between the SBP variability treatment) were related to intra-individual SBP variabil-
and the incidence of both newly developed CVD and all- ity [17]. More recently, the role of visit-to-visit variability
cause mortality, irrespectively of the mean SBP [12]. In of blood pressure on the development of hypertension
particular, patients with an SD of SBP of < 5 mmHg had and changes in renal function in patients with T2D dia-
the lowest risks while patients with an SD of ≥ 10 mmHg betes and normal blood pressure has been evaluated in a
had the highest risk [12]. Another, retrospective cohort real-life clinical setting [18]. After a mean follow-up time
study enrolling 10,163 patients with T2D and free of of 3.5 ± 2.8 years, an increase of 5 mmHg of visit-to-visit
CVD at baseline showed that the variability of SBP was variability of blood pressure was associated with a 19%
associated with an increased risk of CVD, independently (P < 0.0001) and a 5% (P = 0.008) independent increased
of the mean SBP level [13]. Of note, five different metrics, risk of developing hypertension and worsening of albu-
i.e. SD, coefficient of variation, variation independent minuria, respectively [18].
of mean, average real variability, and successive vari- Overall, a number of studies suggests that variability of
ability of measurements, measured during 24-months of SBP is independently associated with the development
observation, were used to perform the analysis [13]. In of a range of diabetes complications, while less data are
 Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 3 of 11

available for diastolic blood pressure variability. The stud- death, myocardial infarction, resuscitated cardiac arrest,
ies showing an association of blood pressure variability coronary revascularization, and unstable or new-onset
with diabetes complications are summarized in Table 1. angina as the primary endpoint [29]. When used as a
time-dependent covariate, BW variability, measured as
Lipids average successive variability, was linearly and indepen-
High-density-lipoprotein-cholesterol (HDL), triglyc- dently associated with an increased risk of any coronary
erides, and low-density-lipoprotein-cholesterol (LDL) event, major coronary event, any CV event, and death.
visit-to-visit variability has been associated with CVD in In particular, when comparing the highest with the low-
non-diabetic subjects [19]. est quintile of BW variability, the increased risk for any
Patients with diabetes are known to suffer from a component of the composite outcome was substantially
marked variability of the lipids plasma levels [20]. More higher [29]. These results suggest that among subjects
recently, in 162 T2D patients followed for 1  year, it has with T2D, fluctuation in BW is associated with higher
been shown that the LDL variability was as independent mortality and a higher rate of CV events, independent of
determinant of carotid maximum IMT, independently of traditional CV risk factors [29].
a number of variables including mean LDL levels, body The ACCORD trial participants’ weight was docu-
mass index (BMI), waist circumference, diabetes therapy mented annually during the trial [30].
and duration, means and SD of glycaemic and other lipid Out of the 10,251 ACCORD participants, 911 (8.9%)
variables, as well as the use of hypolipidemic and anti- has normal weight, 2985 (29.1%) were overweight, and
hypertensive drugs [21]. A more recent study correlates 6355 (62%) were obese. During a mean of 3.5 years of fol-
high LDL variability with an increased risk of CVD in low-up, BW variability was associated with the primary
T2D [22]. outcome MACE, but also with heart failure, death, and
In another study, data from 1792 subjects who under- microvascular events, an observation independent of CV
went percutaneous coronary intervention were ana- risk factors and BMI [30].
lyzed [23]. During a median follow-up of 65 months, 114 The “Verona Diabetes Study” explored the impact of
patients (6.4%) had a MACE [23]. Visit-to-visit variability variability of fasting glycaemia, BMI, and pulse pressure,
in LDL-, HDL-, and non-HDL-cholesterol was signifi- measured as coefficient of variation, on all-cause mortal-
cantly higher in the group suffering a CV event compared ity in 1319 subjects with T2D followed for 10 years [31].
to those not experiencing the endpoint [23]. In the mul- When analyzing data according to age subgroups, the
tiple regression analysis, LDL-, HDL-, and non-HDL- variability of glycaemia, BMI and pulse pressure indepen-
cholesterol variability parameters were all independent dently predicted all-cause mortality in patients > 65 years,
predictors of CV events after adjusting for a number of but not in younger subjects, suggesting a possible role for
confounding variables. These relationships were also the variability of these risk factors in determining mortal-
observed in the subgroup with diabetes [23]. ity in older patients with T2D [31].
Of note, HDL variability has been reported as risk fac-
tor for the appearance also of albuminuria in T2D [24].
Finally, while there are no data linking visit-to-visit vari- Uric acid
ability of triglycerides to macrovascular complications High level of uric acid has been reported to be a risk fac-
of diabetes, two studies show that both high levels of tor for both CVD and nephropathy in diabetes [32, 33].
postprandial (i.e. short-term variability) and visit-to-visit There are not specific studies relative to uric acid vari-
variability of triglycerides predict the development or ability in relation to CVD development in patients with
worsening of nephropathy in diabetes [25–27]. The main diabetes. However, in 8822 non-diabetic men, aged
characteristics of the studies showing an association of 40–65, followed for 5  years, coronary heart disease and
lipids variability with diabetes complications are summa- all-cause mortality yielded a significant association with
rized in Table 2. the variability of uric acid [34]. More recently, high uric
acid variability has been found associated with a higher
Body weight risk of developing future CV events in patient with coro-
Several years ago, the evidence that variability in body nary artery disease [35].
weight (BW) could be related to high risk of CVD A recent study showed that the variability of uric acid
emerged from the Framingham Study [28]. was predictive of kidney alteration, particularly of eGRF
Data from three clinical trials were pooled and used decline, in T2D [36]. Of note, serum uric acid levels are
to evaluate the impact of BW variability in 6408 patients related to insulin resistance and BMI, rather than insu-
with T2D on the development of macrovascular end- lin levels, suggesting that obesity-driven metabolic syn-
points, using a composite of coronary heart disease drome as a major determinant of its levels [37].
Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 4 of 11

Table 1  Summary of the studies showing an effect of blood pressure variability on the development of complications in patients with
diabetes
Risk factor Type of Short or Metrics used Type of Sample size Significantly Follow-up length References
variability long term study associated
assessed variability outcomes

Systolic blood Visit-to-visit Long-term Standard Post-hoc 9114 MACE; micro- 2.4 years [10]
pressure deviation analysis of vascular
trial outomes
Post-hoc 9114 MACE, renal 7.6 years [11]
analysis of events, or
trial death;
Retrospec- 124105 Newly devel- 39.5 months [12]
tive cohort oped CVD;
study all-cause
mortality
Retrospec- 10163 Risk of CVD 24 months [13]
tive cohort
study
Prospective 632 MACE 11.3 years [14]
cohort
study
Post-hoc 2739 Composite 2.6/3.4 years [16]
analysis of 2 renal out-
trials come
Retrospec- 30851 Composite 4 years [17]
tive cohort renal out-
study come
Coefficient of Post-hoc 9383 plus Heart failure 56.6 months/59.5 months [15]
variation analysis of 2 1550 develop-
trials ment
Retrospec- 10163 Risk of CVD 24 months [13]
tive cohort
study
Retrospec- 30851 Composite 4 years [17]
tive cohort renal out-
study come
Variation Retrospec- 10163 Risk of CVD 24 months [13]
independ- tive cohort
ent of mean study
Average real Retrospec- 10163 risk of CVD 24 months [13]
variability tive cohort
study
Post-hoc 9383 plus Heart failure 56.6 months/59.5 months [15]
analysis of 2 1550 related
trials event
Retrospec- 30851 Composite 4 years [17]
tive cohort renal out-
study come
Successive Retrospec- 10163 Risk of CVD 24 months [13]
variability of tive cohort
measure- study
ments
Diastolic Visit-to-visit Long-term Coefficient Post-hoc 9383 plus Heart failure 56.6 months/59.5 months [15]
blood pres- of variation analysis of 2 1550 develop-
sure AND Aver- trials ment
age real
variability
 Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 5 of 11

Table 2  Summary of the studies showing an effect of lipids variability on the development of complications in patients with diabetes
Risk factor Type of Short or Metrics used Type of study Sample size Significantly Follow-up References
variability long term associated length
assessed variability outcomes

LDL Visit-to-visit Long-term Standard devia- Observational 162 Carotid intima- 1 year [21]
tion cohort media thick-
ness
Observational 5354 Cardiovascular 3.2 years [22]
cohort events
Observational 1792 Cardiovascular 65 months [23]
cohort events
HDL Visit-to-visit Long-term Standard devia- Observational 1792 Cardiovascular 65 months [23]
tion cohort events
Observational 864 Appearance of 3 years [24]
cohort albuminuria
Non-HDL cho- Visit-to-visit Long-term Standard devia- Observational 1792 Cardiovascular 65 months [23]
lesterol tion cohort events
Triglycerides Visit-to-visit Long-term Standard devia- Observational 457 Incident micro- 6.8-years [27]
tion cohort albuminuria
Post-prandial Short-term Coefficient of Observational 168 eGFR decline 6.0 years [26]
variation cohort

Heart rate electrocardiograms [42]. The SD of consecutive RR inter-

Differently from the other discussed risk factors, heart vals was used as a measure of HR variability. Coronary
rate (HR) variability is often studied by a resting ECG artery calcium was measured at two visits 6 years apart.
or with a 24-h Holter, thus it refers to short-term vari- Reduced HR variability was associated in both T1D and
ability and not to visit-to-visit variability. In particular, healthy controls with older age, higher HbA1c, elevated
low-frequency (0.04–0.15 Hz) oscillations are a surrogate albuminuria, coronary artery calcium volume, as well
measure of HR variability ascribable to both sympathetic as higher fibrinogen levels at baseline. Higher HR vari-
and parasympathetic innervation [38–41]. Cardiac auto- ability at baseline was independently associated with a
nomic neuropathy affects the parasympathetic nervous decreased probability of coronary artery calcium pro-
system, leading to reduced HR variability [38]. In people gression, even after adjusting for a range of CV risk fac-
without diabetes, decreased HR variability is associated tors including inflammatory parameters [42].
with an increased incidence of cardiac events [39]. Finally, two studies, both with a follow-up of about
Low HR variability has been reported in both type [1] 5  years, were able to show that low HR variability is
(T1D) and T2D and, intriguingly, in pre-diabetes [40]. related to a high risk of cardiac and total mortality in
A preliminary study explored the impact of autonomic T2D [43, 44]. The main characteristics of the studies
neuropathy on HR variability and on the progression of showing an association between body weight, uric acid,
atherosclerosis in 61 T2D patients followed for 8  years and heart variability and diabetes complications are sum-
[41]. Results suggested that patients with autonomic marized in Table 3.
neuropathy had decreased low frequency HR variability,
which was correlated with both a reduced carotid artery Interaction between the variability of the risk
diameter and an increased atherosclerotic intima-media factors
thickness (IMT). HR variability further decreased dur- It is well known that there is a cumulative effect of the
ing follow-up, while patients with lower frequency of various risk factors in producing serious CV complica-
HR at baseline had a more relevant enlargement in the tions. This evidence originally came out from the histori-
thickness of the carotid bulb intima-media at follow-up, cal Framingham Heart Study [45]. In the same study it
possibly suggesting that a low frequency HR variability was also clear that the presence of diabetes contributes to
might foresee the degree of atherosclerosis progression amplify the additive effect of the risk factors in the devel-
in patients with T2D [41]. opment of the CV complications [45].
Subjects with T1D or those without diabetes from As above reported, the variability of a risk factor
the “Coronary Artery Calcification in Type 1 Diabe- contributes an additive risk, independently from the
tes” study underwent supine deep breathing 12-lead magnitude and the duration of the abnormal level of
Table 3  Summary of the studies showing an effect of body weight, uric acid, and heart rate variability on the development of complications in patients with diabetes
Risk factor Type of variability Short or Metrics used Type of study Sample size Significantly Follow-up length References
assessed long term associated outcomes
variability
Ceriello and Prattichizzo Cardiovasc Diabetol

Body weight Visit-to-visit Long-term Average successive vari- Post-hoc analysis of 3 6408 Composite of cardiovas- 3.9/4/4.9 years [29]
ability trials cular events
Post-hoc analysis of trial 10251 MACE, heart failure, 3.5 years [30]
death, and microvas-
cular events
(2021) 20:101

Coefficient of variation Observational cohort 1319 All-cause mortality in 10 years [31]

patients > 65 years
Uric acid Visit-to-visit Long-term Standard deviation Retrospective cohort 10163 eGRF decline 24 months [36]
study
Heart rate 24-h ECG Holter, expira- Short-term Low frequency Observational cohort 61 Intima-media thickness 8 years [41]
tion/inspiration (E/I)
ratio during deep
breathing, accelera-
tion index (AI) of R-R
interval in response to
head-up tilt
Supine deep breathing Short-term Standard deviation Observational cohort 1416 Coronary artery calcium 6.0 ± 0.5 years [42]
12-lead electrocardio- of consecutive RR
grams intervals
Coefficient of variance Observational cohort 8917 (3089 with dia- Sudden cardiac death 5.2 years [43]
for 100 R–R intervals betes)
24-h ECG Holter Short-term Standard deviation of NN Observational cohort 240 All-cause mortality 15.5 years [44]
intervals
Page 6 of 11
 Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 7 of 11

the risk factor. The research in this field is relatively Is it time to treat?
new, however, some evidence already exists. Studies The major issue is whether risk factors variability is
have explored the effect of the simultaneous variabil- causal or a marker, and, while with glucose the case for
ity of several risk factors on a target complication. In causality can be made [1], it would be more difficult with
non-diabetic subjects the combined variability of sev- uric acid, for example.
eral risk factors contributes to the risk for both CVD The evidence in favour to be just a marker is that the
and end stage renal disease [46, 47]. An additive effect overall quality of care seems conditioning the variability
of HbA1c and blood pressure variability on the risk of of all these risk factors [52]. The quality-of-care summary
mortality has been reported in T1D patients followed- score (Q-score) is a surrogate, validated measure of qual-
up for 1430 days [48]. The hazard ratio for high HbA1c ity of care as a whole. The frequency of risk factors meas-
variability was 1.78 ± 0.36. The hazard ratio for high urement, their values, and the relative pharmacological
SBP variability was 1.69 ± 0.33. The hazard ratio for therapy contribute to produce the Q-score, the values
high HbA1c and high SBP variability together was of which can range between zero and forty with higher
2.37 ± 0.32 [48]. In older T2D patients the variabil- scores being descriptive of better quality of care [52]. In a
ity of fasting glycaemia, BW and blood pressure were study including 273,888 people with diabetes the variabil-
independently associated with an increased risk of all- ity of HbA1c, systolic and diastolic blood pressure, total
cause mortality during the 10  years of follow-up [31]. cholesterol, LDL, HDL, and uric acid was inversely corre-
In people with T2D an inverse association has been lated with the Q-score value [52]. In a multivariate linear
reported between glucose fluctuations and heart rate regression analysis a Q-score > 25 was associated with a
variability [49]. Another study reported that glucose significantly minor variation in HbA1c, systolic and dias-
and blood pressure variability were associated with tolic blood pressure, uric acid, total cholesterol, HDL, and
endothelial and CV damage in diabetic patients with LDL cholesterol, when compared to a score < 15 [52]. The
optimal metabolic control [50]. However, in a popula- analysis of standardized β coefficients evidenced that the
tion of patients with T2D and no history of CVD, in Q-score had a higher impact on the variability of ­HbA1c,
which other CV risk factors were within or near to the SBP, total cholesterol, and LDL cholesterol [52]. This
recommended targets, only 2-h post breakfast blood finding suggests that when a quality-of-care improve-
glucose level, but not the variability of blood pres- ment is pursued to increase the achievement of the rec-
sure, lipids and creatinine, was associated with inci- ommended targets, this action might be accompanied by
dent CVD, as observed during the follow-up period of a reduction of risk factors variability. This seems to be the
5–8 years [51]. case. In a recent study, the attainment of HbA1c, blood
In T2D the additive effect of HbA1c, fasting glucose, pressure, and LDL-cholesterol goals was associated with
systolic and diastolic blood pressure, total-cholesterol, a significant improvement of their variability [53].
HDL, LDL, triglycerides, and uric acid variability on However, it also true that, according to Hill definition
the appearance of kidney disease has been reported of causality, the variability of several risk factors does fit
[36]. with this concept [54]. Indeed, the findings relative to
A recent study using data from a clinical database the association with the quality of care intuitively sug-
reporting at least 5 measures of multiple risk factors gest that risk factors variability might result from clinical
from 4231 patients with T2D followed up for a median inertia and drug non-adherence, two pervasive phenom-
of 3.4  years showed that a significantly higher risk of ena in real-world settings of diabetes treatment [55].
developing albuminuria was associated with variability Therefore, more studies are needed to explore both the
in HbA1c, while the variability in systolic and diastolic relevance of these two phenomena in determining vari-
blood pressure, HDL, LDL and uric acid predicted the ability and to assess if variability of risk factors per se rep-
decline in eGFR, with the association with uric acid resents an additive driver of diabetes complications.
variability being particularly strong [36]. In addition, The current strategy for the management of a risk fac-
the concomitance of high variability in HbA1c and tor is to try to maintain its levels within the normal range
HDL conferred the highest risk of developing albumi- as long as possible. The evidence that the variability in
nuria (Fig.  1), while a high variability in uric acid or time of a risk factor has a potential harmful effect on the
diastolic blood pressure conferred the highest risk of development of diabetic complications opens the discus-
decline in eGFR (Fig.  2). This novel evidence suggests sion of whether a new therapeutic strategy is needed: the
that the contribution of the variability of each single normalization of the level of a risk factor might be not
factor might have higher or lesser impact according to enough without reducing its variability. Whether it seems
the specific complication studied [36]. This new field of more than acceptable to improve the quality of care and
research needs more studies. to reach the optimal target, hoping that these actions
Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 8 of 11

Fig. 1  Recursive partitioning techniques (RECPAM) analysis of developing albuminuria in a cohort of 4231 patients with T2D followed up for a
median of 3.4 years and with 5 subsequent measurements of risk factors [36]. The tree-growing algorithm resumes the hazard of developing
albuminuria according to a multivariable Cox regression analysis. At each step, the method proceeds forward using the covariate with the highest
difference in risk. The algorithm proceeds until user-defined conditions are met. Variables used to build the model were quartiles of variability in
HbA1c, systolic blood pressure (SBP) and diastolic blood pressure (DBP), serum uric acid (UA), total, high-density lipoprotein (HDL), low-density
lipoprotein (LDL) cholesterol and triglycerides., while additional baseline parameters were considered in the model as global variables, i.e. age,
gender, duration of diabetes, smoking, hypertension, baseline HbA1c, blood pressure, UA, lipid parameters and estimated glomerular filtration rate
(eGFR) values. The variable determining patient’s assignment to the subsequent group is evidenced on the branch proceeding to the following
subgroup, while rectangles represent the REPCAM class. The numbers in the circles and rectangles represent the patients who develop albuminuria
compared with the total number of patients in the subgroup, respectively (Reproduced with permission from Ref. [36])

may also improve the variability of the risk factors [52, helpful in reducing not only blood pressure but also its
53], the key issue is whether it is also the case to plan variability [58].
more specific interventions to improve the variability of Although statin therapy is widely known to substan-
a risk factor. This seems, for example, the case for glucose tially decrease LDL, variability in LDL levels during the
variability [56]. treatment has been observed. Although visit-to-visit
The question is certainly not only a scientific curiosity. variability in LDL is largely attributable to statin non-
There is also the possibility of intervening on the variabil- adherence, it is worthy to remember that Bangalore et al.
ity of a specific risk factor. showed that LDL variability was consistently less pro-
Diverse anti-hypertensive drugs have different effects nounced with the atorvastatin dose of 80  mg/day com-
on reducing blood pressure variability. Indeed, a meta- pared to the 10 mg/day dose [59, 60].
analysis suggested that variability of SBP was decreased Moderate weight loss has been preliminary reported to
by calcium-channel blockers and thiazides, while it was improve HR variability in T2D [61]. Finally, weight loss
increased by angiotensin-converting enzyme inhibi- strategies that minimize weight cycling probably should
tors, angiotensin-receptor blockers, and beta-blockers be given preference over those that are prone to cause a
[57]. Similarly, the use of a SGLT-2 inhibitor may also be “yoyo effect” [62].
 Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 9 of 11

Fig. 2  RECPAM analysis of developing a decrease in glomerular filtration rate (GFR). The RECPAM tree-growing algorithm models the hazard of
developing GFR < 60 mL/min/1.73 ­m2 using the same approach and the same variables described for Fig. 1 in the same population (Reproduced
with permission from Ref. [36])

Conclusions On the other side, no trial has clearly proven that a reduc-
In most of the intervention studies in settings of T2D, tion in variability per se, independently of the risk factor
a reduction in the magnitude of diverse CV risk factors magnitude, is beneficial. However, evidence is growing
have translated into tangible benefits on hard endpoints. suggesting that the variability of a risk factor might be
Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 10 of 11

not less dangerous than its magnitude and/or the time 9. Grove JS, Reed DM, Yano K, Hwang LJ. Variability in systolic blood
pressure: a risk factor for coronary heart disease? Am J Epidemiol.
spent having it in an abnormal level. Moreover, it is also 1997;145:771–6.
enough clear that the variability of each factor is additive 10. Hata J, ADVANCE Collaborative Group, et al. Effects of visit-to-visit vari-
in leading to the final picture of a diabetic complication. ability in systolic blood pressure on macrovascular and microvascular
complications in patients with type 2 diabetes mellitus: the ADVANCE
Therefore, from a therapeutic viewpoint, is it time to try trial. Circulation. 2013;128:1325–34.
to reduce both magnitude and variability of CV risk fac- 11. Ohkuma T, ADVANCE Collaborative Group, et al. Prognostic value of vari-
tors? Because an improvement of the quality of care [52] ability in systolic blood pressure related to vascular events and premature
death in type 2 diabetes mellitus: the ADVANCE-ON study. Hypertension.
and the attainment of their target [53] seem to be the best 2017;70:461–8.
strategy to reach this goal, we believe that the answer is 12. Wan EY, Fung CS, Yu EY, Fong DY, Chen JY, Lam CL. Association of visit-
not complicated. to-visit variability of systolic blood pressure with cardiovascular disease
and mortality in primary care Chinese patients with type 2 diabetes: a ret-
Acknowledgements rospective population-based cohort study. Diabetes Care. 2017;40:270–9.
Not applicable. 13. Yu ZB, et al. Association of visit-to-visit variability of blood pressure with
cardiovascular disease among type 2 diabetes mellitus patients: a cohort
Authors’ contributions study. Diabetes Metab J. 2019;43:350–67.
AC: writing—original draft preparation; FP: writing-review and editing. Both 14. Cardoso CRL, Leite NC, Salles GF. Prognostic importance of visit-to-visit
authors read and approved the final manuscript. blood pressure variability for micro- and macrovascular outcomes in
patients with type 2 diabetes: the Rio de Janeiro type 2 diabetes cohort
Funding study. Cardiovasc Diabetol. 2020;19:50.
This work has been supported by the Italian Ministry of Health—Ricerca 15. Nuyujukian DS, Koska J, Bahn G, Reaven PD, Zhou JJ, VADT Investigators.
Corrente to IRCCS MultiMedica and by Ricerca Finalizzata RF-2016-02364513 Blood pressure variability and risk of heart failure in ACCORD and the
to AC. VADT. Diabetes Care. 2020;43:1471–8.
16. McMullan CJ, Lambers Heerspink HJ, Parving HH, Dwyer JP, Forman JP, de
Availability of data and materials Zeeuw D. Visit-to-visit variability in blood pressure and kidney and car-
Not applicable. diovascular outcomes in patients with type 2 diabetes and nephropathy:
a post hoc analysis from the RENAAL study and the Irbesartan diabetic
nephropathy trial. Am J Kidney Dis. 2014;64:714–22.
Declarations 17. Viazzi F, AMD-Annals Study Group, et al. Long-term blood pressure vari-
ability and development of chronic kidney disease in type 2 diabetes. J
Ethics approval and consent to participate Hypertens. 2019;37:805–13.
Not applicable. 18. Viazzi F, AMD-Annals Study Group, et al. Long-term blood pressure vari-
ability, incidence of hypertension and changes in renal function in type 2
Consent for publication diabetes. J Hypertens. 2020;38:2279–86.
Not applicable. 19. Waters DD, et al. Visit-to-visit variability of lipid measurements as predic-
tors of cardiovascular events. J Clin Lipidol. 2018;12:356–66.
Competing interests 20. Hulse M, Gershberg H. Variability in blood cholesterol, triglycerides, free
Authors do not have conflicts of interest to declare. fatty acids, glucose and body weight in maturity-onset diabetics. Am J
Med Sci. 1969;258:114–20.
Received: 16 March 2021 Accepted: 27 April 2021 21. Takenouchi A, et al. Visit-to-visit low-density lipoprotein cholesterol vari-
ability is an independent determinant of carotid intima-media thickness
in patients with type 2 diabetes. J Clin Med Res. 2017;9:310–6.
22. Hsu WH, et al. Greater low-density lipoprotein cholesterol variability
increases the risk of cardiovascular events in patients with type 2 diabe-
References tes mellitus. Endocr Pract. 2019;25:918–25.
1. Ceriello A, Monnier L, Owens D. Glycaemic variability in diabetes: clinical 23. Lee EY, et al. Effect of visit-to-visit LDL-, HDL-, and non-HDL-cholesterol
and therapeutic implications. Lancet Diabetes Endocrinol. 2019;7:221–30. variability on mortality and cardiovascular outcomes after percutaneous
2. Ranjan AG, Rosenlund SV, Hansen TW, Rossing P, Andersen S, Nørgaard K. coronary intervention. Atherosclerosis. 2018;279:1–9.
Improved time in range over 1 year is associated with reduced albumi- 24. Chang YH, Chang DM, Lin KC, Hsieh CH, Lee YJ. High-density lipo-protein
nuria in individuals with sensor-augmented insulin pump-treated type 1 cholesterol and the risk of nephropathy in type 2 diabetic patients. Nutr
diabetes. Diabetes Care. 2020;43(11):2882–5. Metab Cardiovasc Dis. 2013;23:751–7.
3. Hirakawa Y, et al. Impact of visit-to-visit glycemic variability on the risks of 25. Ceriello A, Genovese S. Atherogenicity of postprandial hyperglycemia
macrovascular and microvascular events and all-cause mortality in type 2 and lipotoxicity. Rev Endocr Metab Disord. 2016;17:111–6.
diabetes: the ADVANCE trial. Diabetes Care. 2014;37:2359–65. 26. Tsuboi A, Takenouchi A, Kurata M, Fukuo K, Kazumi T. Postmeal triglyceri-
4. Zinman B, et al. Day-to-day fasting glycaemic variability in DEVOTE: demia and variability of HbA1c and postmeal glycemia were predictors
associations with severe hypoglycaemia and cardiovascular outcomes of annual decline in estimated glomerular filtration rate in type 2 diabetic
(DEVOTE 2). Diabetologia. 2018;61:48–57. patients with different stages of nephropathy. J Diabetes Metab Disord.
5. Zhou JJ, Schwenke DC, Bahn G, Reaven P, VADT Investigators. Glycemic 2017;16:1.
variation and cardiovascular risk in the Veterans Affairs Diabetes Trial. 27. Bardini G, Innocenti M, Rotella CM, Giannini S, Mannucci E. Variability of
Diabetes Care. 2018;41:2187–94. triglyceride levels and incidence of microalbuminuria in type 2 diabetes. J
6. Echouffo-Tcheugui JB, Zhao S, Brock G, Matsouaka RA, Kline D, Joseph JJ. Clin Lipidol. 2016;10:109–15.
Visit-to-visit glycemic variability and risks of cardiovascular events and 28. Lissner L, et al. Variability of body weight and health outcomes in the
all-cause mortality: the ALLHAT study. Diabetes Care. 2019;42:486–93. Framingham population. N Engl J Med. 1991;324:1839–44.
7. Sheng CS, et al. Prognostic significance of long-term HbA1c variability for 29. Bangalore S, Fayyad R, DeMicco DA, Colhoun HM, Waters DD. Body
all-cause mortality in the ACCORD trial. Diabetes Care. 2020;43:1185–90. weight variability and cardiovascular outcomes in patients with type 2
8. Ceriello A, Ofstad AP, Zwiener I, Kaspers S, George J, Nicolucci A. diabetes mellitus. Circ Cardiovasc Qual Outcomes. 2018;11:e004724.
Empagliflozin reduced long-term HbA1c variability and cardiovascular 30. Yeboah P, Hsu FC, Bertoni AG, Yeboah J. Body Mass Index, change in
death: insights from the EMPA-REG OUTCOME trial. Cardiovasc Diabetol. weight, body weight variability and outcomes in type 2 diabetes mellitus
2020;19:176. (from the ACCORD trial). Am J Cardiol. 2019;123:576–81.
 Ceriello and Prattichizzo Cardiovasc Diabetol (2021) 20:101 Page 11 of 11

31. Zoppini G, Verlato G, Targher G, Bonora E, Trombetta M, Muggeo M. 48. Wightman SS, Sainsbury CAR, Jones GC. Visit-to-visit HbA1c variability
Variability of body weight, pulse pressure and glycaemia strongly predict and systolic blood pressure (SBP) variability are significantly and addi-
total mortality in elderly type 2 diabetic patients. The Verona diabetes tively associated with mortality in individuals with type 1 diabetes: an
study. Diabetes Metab Res Rev. 2008;24:624–8. observational study. Diabetes Obes Metab. 2018;20:1014–7.
32. Viazzi F, et al. Natural history and risk factors for diabetic kidney 49. Chen Y, Jia T, Yan X, Dai L. Blood glucose fluctuations in patients with
disease in patients with T2D: lessons from the AMD-annals. J Nephrol. coronary heart disease and diabetes mellitus correlates with heart
2019;32:517–25. rate variability: a retrospective analysis of 210 cases. Niger J Clin Pract.
33. Verma S. Association between uric acid levels and cardio-renal outcomes 2020;23:1194–200.
and death in patients with type 2 diabetes: a subanalysis of EMPA-REG 50. Di Flaviani A, et al. Impact of glycemic and blood pressure variability
OUTCOME. Diabetes Obes Metab. 2020;22:1207–14. on surrogate measures of cardiovascular outcomes in type 2 diabetic
34. Grossman C, Grossman E, Goldbourt U. Uric acid variability at midlife as patients. Diabetes Care. 2011;34:1605–9.
an independent predictor of coronary heart disease and all-cause mortal- 51. Hashemi Madani N, Ismail-Beigi F, Khamseh ME, Malek M, Ebrahimi
ity. PLoS ONE. 2019;14:e0220532. Valojerdi A. Predictive and explanatory factors of cardiovascular disease
35. Lim SS, et al. Association of variability in uric acid and future clinical out- in people with adequately controlled type 2 diabetes. Eur J Prev Cardiol.
comes of patient with coronary artery disease undergoing percutaneous 2017;24:1181–9.
coronary intervention. Atherosclerosis. 2020;297:40–6. 52. Ceriello A, AMD-Annals Study Group, et al. Overall quality of care predicts
36. Ceriello A, AMD-Annals Study Group, et al. Variability in HbA1c, blood the variability of key risk factors for complications in type 2 diabetes:
pressure, lipid parameters and serum uric acid, and risk of development an observational, longitudinal retrospective study. Diabetes Care.
of chronic kidney disease in type 2 diabetes. Diabetes Obes Metab. 2019;42:514–9.
2017;19:1570–8. 53. Kitaoka K, et al. Association of ABC (HbA1c, blood pressure and LDL-
37. Bonora E, et al. Relationship of uric acid concentration to cardiovascular cholesterol) goal achievement with visit-to-visit ABC variability and
risk factors in young men. Role of obesity and central fat distribution. postprandial dysmetabolism in type 2 diabetic patients. Asia Pac J Clin
The Verona young men atherosclerosis risk factors study. Int J Obes Relat Nutr. 2020;29:476–82.
Metab Disord. 1996;20(11):975–80. 54. Hill AB. The environment and disease: association or causation? 1965. J R
38. Proceedings of a consensus development conference on standardized Soc Med. 2015;108:32–7.
measures in diabetic neuropathy. Summary and recommendations. 55. Reach G, Pechtner V, Gentilella R, Corcos A, Ceriello A. Clinical inertia and
Diabetes Care 1992;15:1104–1107. its impact on treatment intensification in people with type 2 diabetes
39. Tsuji H, et al. Impact of reduced heart rate variability on risk for cardiac mellitus. Diabetes Metab. 2017;43(6):501–11.
events: the Framingham heart study. Circulation. 1996;94:2850–5. 56. Ceriello A. Glucose variability and diabetic complications: is it time to
40. Coopmans C, et al. Both prediabetes and type 2 diabetes are associated treat? Diabetes Care. 2020;43:1169–71.
with lower heart rate variability: the Maastricht study. Diabetes Care. 57. Webb AJ, Fischer U, Mehta Z, Rothwell PM. Effects of antihypertensive-
2020;43:1126–33. drug class on interindividual variation in blood pressure and risk of stroke:
41. Gottsäter A, Ahlgren AR, Taimour S, Sundkvist G. Decreased heart rate a systematic review and meta-analysis. Lancet. 2010;375:906–15.
variability may predict the progression of carotid atherosclerosis in type 2 58. Kario K, Ferdinand KC, O’Keefe JH. Control of 24-hour blood pressure with
diabetes. Clin Auton Res. 2006;16:228–34. SGLT2 inhibitors to prevent cardiovascular disease. Prog Cardiovasc Dis.
42. Rodrigues TC, Ehrlich J, Hunter CM, Kinney GL, Rewers M, Snell-Bergeon 2020;63:249–62.
JK. Reduced heart rate variability predicts progression of coronary artery 59. Mann DM, et al. A pilot study identifying statin nonadherence with visit-
calcification in adults with type 1 diabetes and controls without diabetes. to-visit variability of low-density lipoprotein cholesterol. Am J Cardiol.
Diabetes Technol Ther. 2010;12:963–9. 2013;15:1437–42.
43. Kataoka M, Ito C, Sasaki H, Yamane K, Kohno N. Low heart rate variability 60. Bangalore S, et al. Relation of variability of low-density lipoprotein choles-
is a risk factor for sudden cardiac death in type 2 diabetes. Diabetes Res terol and blood pressure to events in patients with previous myocardial
Clin Pract. 2004;64:51–8. infarction from the IDEAL trial. Am J Cardiol. 2017;119:379–87.
44. May O, Arildsen H. Long-term predictive power of heart rate variabil- 61. Sjoberg N, Brinkworth GD, Wycherley TP, Noakes M, Saint DA. Moder-
ity on all-cause mortality in the diabetic population. Acta Diabetol. ate weight loss improves heart rate variability in overweight and obese
2011;48:55–9. adults with type 2 diabetes. J Appl Physiol. 2011;110:1060–4.
45. Mahmood SS, Levy D, Vasan RS, Wang TJ. The Framingham heart study 62. Contreras RE, Schriever SC, Pfluger PT. Physiological and epigenetic
and the epidemiology of cardiovascular disease: a historical perspective. features of yoyo dieting and weight control. Front Genet. 2019;10:1015.
Lancet. 2014;383:999–1008.
46. Kim MK, et al. Associations of variability in blood pressure, glucose
and cholesterol concentrations, and body mass index with mortality Publisher’s Note
and cardiovascular outcomes in the general population. Circulation. Springer Nature remains neutral with regard to jurisdictional claims in pub-
2018;138:2627–3263. lished maps and institutional affiliations.
47. Kim MK, et al. Effects of variability in blood pressure, glucose, and choles-
terol concentrations, and body mass index on end-stage renal disease in
the general population of Korea. J Clin Med. 2019;8:755.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission

• thorough peer review by experienced researchers in your field
• rapid publication on acceptance
• support for research data, including large and complex data types
• gold Open Access which fosters wider collaboration and increased citations
• maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions