Clinical features of Parkinson disease when
onset of diabetes came first
A case-control study
E. Cereda, MD, PhD
M. Barichella, MD
E. Cassani, MD
R. Caccialanza, MD
G. Pezzoli, MD
Objective: Recent literature suggests that diabetes is a risk factor for Parkinson disease (PD). We
investigated the clinical features of patients with idiopathic PD (IPD) in whom the onset of diabe-
tes came first.
Methods: We designed a case-control study. From the cohort of all new patients with IPD free of
vascular disease (n ? 783) admitted and evaluated at our institute over a 3-year period (2007–
2010), we included all the patients with a diagnosis of diabetes prior to PD onset (n ? 89) and a
control group (n ? 89) matched (1:1) for gender, body mass index (?1 kg/m2), and duration of PD
(?1 year). The Unified Parkinson’s Disease Rating Scale (UPDRS) motor score was the primary
Results: At study entry, patients with diabetes were similar to controls in terms of most demo-
graphic, lifestyle, and general medical features with exception of statins (18% vs 3.4%; p ?
0.003). However, diabetes was associated with higher UPDRS motor (22.3 ? 9.0 vs 19.3 ? 7.9;
p ? 0.019) and activities of daily living (9.7 ? 5.1 vs 8.3 ? 4.3; p ? 0.049) scores, more severe
Hoehn & Yahr staging (p ? 0.009), and higher treatment doses of levodopa (mg/day, 448 ? 265
vs 300 ? 213; p ? 0.0001; mg/kg/day, 5.8 ? 4.0 vs 3.8 ? 2.9; p ? 0.0001).
Conclusions: Onset of diabetes before the onset of PD appears to be a risk factor for more severe
PD symptoms. These findings support the hypothesis that diabetes has a role in the etiopatho-
genesis of PD. Neurologists should be aware of the potential impact of diabetes on overall PD
ANOVA ? analysis of variance; BMI ? body mass index; IPD ? idiopathic Parkinson disease; PD ? Parkinson disease;
UPDRS ? Unified Parkinson’s Disease Rating Scale.
Current research suggests that chronic neurodegenerative disorders share similar pathogenic
pathways to some extent.1,2Diabetes is now considered a risk factor for dementia,3or more in
general for cognitive decline, and prospective studies suggest that it may contribute also to the
development of idiopathic Parkinson disease (IPD).4–7Along with this, diabetes has been
associated with dementia in PD8and with the progression of parkinsonian signs, particularly
rigidity and gait disturbances, in elderly people.9In this regard, there is some evidence suggest-
ing that atherosclerosis contributes to neurodegeneration in these cases.10,11However, the asso-
ciation between diabetes and Parkinson disease (PD) seems to be significant even in the absence
of established vascular disease.4
We designed the present case-control study to determine whether prior onset of diabetes
contributes to the severity of PD symptoms and their progression when causes of secondary
parkinsonism have been systematically excluded.
METHODS Data source. We conducted a population-based retrospective case-control study. We used the Parkinson Institute
research database, which contains computerized demographic (age, sex, education, weight, height, body mass index [calculated as the
ratio between weight and height squared, Kg/m2]), general medical (family history, diagnoses, and drug prescriptions), and disease-
From the Nutrition and Dietetics Service (E. Cereda, R.C.), Fondazione IRCCS Policlinico San Matteo, Pavia; and Parkinson Institute–Istituti
Clinici di Perfezionamento (M.B., E. Cassani, G.P.), Milan, Italy.
Study funding: Supported by the “Fondazione Grigioni per il Morbo di Parkinson” and the Fondazione IRCCS Policlinico San Matteo (to E.C.).
Go to Neurology.org for full disclosures. Disclosures deemed relevant by the authors, if any, are provided at the end of this article.
Correspondence & reprint
requests to Dr. Cereda:
Copyright © 2012 by AAN Enterprises, Inc.
specific records (disease onset and related features, disease pro-
gression, and pharmacologic therapy), as well as lifestyle
information (smoking status, physical activity, and hydrocarbon
exposure) on all patients admitted to the institute for disease
assessment. Data of all the patients attending the Parkinson In-
stitute for disease assessment between November 1, 2007, and
December 31, 2010, were reviewed for this study.
Subject selection. All new cases of IPD, diagnosed according
to UK Parkinson’s Disease Society Brain Bank criteria,12were
initially eligible for inclusion. Accordingly, from the cohort of
patients attending our Institute for disease assessment (parkinso-
nian syndrome) we excluded case subjects diagnosed with pro-
gressive supranuclear palsy, multiple system atrophy, or any type
of dementia preceding PD diagnosis; those having cognitive im-
pairment (Mini-Mental State Examination score ?25 points);
or those reporting the use of neuroleptic drugs, with sudden
onset of symptoms, with repeated head injury, or with a history
of encephalitis. We also excluded vascular parkinsonism on the
basis of medical history (positive for any cardiovascular or cere-
brovascular disease) and brain imaging evaluation by CT, MRI,
SPECT, or DaTSCAN.
Ascertainment of diabetes was performed by means of a self-
administered questionnaire and confirmed through direct inter-
Table 1Types of diabetes medications in
diabetic patients with PDa
Biguanides ? sulfonylureas
Insulin ? Biguanides
Abbreviation: PD ? Parkinson disease.
aSingle drugs (e.g., glinides or thiazolidinediones) or other
Table 2Characteristics of the patients included in the study according to the presence or absence of
diabetes preceding PD diagnosisa
Diabetes (n ? 89) No diabetes (n ? 89)p Valueb
Male, n (%)
58 (65.1)58 (65.1) 1.000
Previous or current smoking, n (%)
23 (25.8) 20 (22.5)0.726
Sedentary, n (%)
65 (73) 61 (68.5)0.621
Education, y, mean (SD)
8.4 (3.7)9.2 (3.9) 0.162
Body weight, Kg, mean (SD)
77.1 (12.4)78.4 (12.3) 0.483
Body mass index, Kg/m2, mean (SD)
27.7 (3.9)27.6 (3.3)0.854
Age, y, mean (SD)
70.7 (7.7) 69 (7.9) 0.123
At onset of disease
66.9 (8.0)65.1 (8.2)0.140
Length of disease, y, mean (SD)
3.8 (3.5)3.9 (3.4)0.847
Positive family history for PD, n (%)
12 (13.5)9 (10.1)0.643
Positive history for hydrocarbon exposure, n (%)
21 (23.6) 16 (18.0)0.460
Main symptom at diagnosis, n (%)
45 (50.6) 52 (58.4)
30 (33.7)27 (30.3)
4 (4.5) 4 (4.5)
4 (4.5) 3 (3.4)
6 (6.7) 3 (3.4)
Levodopa at inclusion, mg/d, mean (SD)c
448 (265) 300 (213)
mg/Kg/d, mean (SD)
5.8 (4.0)3.8 (2.9)
Inception of levodopa therapy, y, mean (SD)
2.3 (2.3)2.1 (2.2) 0.554
Dopamine-agonist therapy at inclusion, n (%)
38 (42.7) 43 (48.3) 0.547
Total levodopa equivalent at inclusion, mg/d, mean (SD)d
503 (292) 350 (206)
mg/Kg/day, mean (SD)
6.5 (4.2) 4.5 (2.8)
Hypertension, n (%)
32 (36) 27 (30.3) 0.524
Use of statins, n (%)
16 (18)3 (3.4)0.003
Abbreviation: PD ? Parkinson disease.
aPercentages are calculated within single groups.
bContinuous and categorical variables were compared between groups with Student t test or ?2test or Fisher exact test.
cFrom levodopa-containing medications.
dIncluding total levodopa equivalent dose from other antiparkinsonian medications.17
Neurology 78May 8, 2012
viewing. We also excluded patients with onset of diabetes after
diagnosis of PD. Moreover, venous blood samples were collected
for fasting glucose assessment. Information on the use of diabetes
medications was also obtained.
Clinical presentation. To investigate any potential difference
in clinical presentation we used only data collected at the first
visit and focusing on positive family history of PD, age and
symptoms at diagnosis, levodopa therapy (in mg/day and mg/kg/
day; excluding and including equivalent dose of other antipar-
kinsonian medications13) and its initiation after diagnosis, use of
dopamine agonists, and clinical rating scales (Hoehn & Yahr and
Unified Parkinson’s Disease Rating Scale [UPDRS]).
Ethics. The study was performed in agreement with the princi-
ples of the Declaration of Helsinki and the protocol was ap-
proved by the local Ethics Committee. We obtained written
informed consent from every patient recruited.
Statistical analyses. Based on a 2-tailed test with a 5% signif-
icance level (type I error [?]) and a power of 80% (type II error
[?]), we calculated that at least 80 patients in each group (dia-
betic vs nondiabetic) were required to detect a 20% difference in
part III (motor score) of the UPDRS.
The first step in data management was the extraction of
patients with a diagnosis of diabetes prior to PD onset. There-
after, cases were matched (1:1) to nondiabetic patients with
IPD by gender, body mass index (?1 kg/m2; BMI), and dis-
ease duration (?1 year).
Continuous variables were reported as mean and SD and
analyzed using the Student t test or Mann-Whitney U test (2-
group comparisons) or analysis of variance (ANOVA) (multiple-
group comparisons). In particular, 2-way ANOVA analysis was
used to investigate the effect of diabetes and the duration of
disease (independent variables) on both UPDRS part III score
and levodopa doses. Categorical variables were presented as
count and percentage and were analyzed by Fisher exact test or
the ?2test as appropriate. Finally, the use of multivariable mod-
els was considered for the adjustment of potential differences
between the study groups.
All data were analyzed using MEDCALC®for Windows
version 18.104.22.168 (MedCalc Software, Mariakerke, Belgium), set-
ting the level of significance at a 2-tailed p value of ?0.05.
RESULTS Over a 3-year period 1,510 patients
presenting parkinsonian symptoms were referred
to our Institute. A clinical diagnosis of IPD was
made or confirmed in 851 cases. Of these 783 had
complete data on diabetes history and enough in-
formation to calculate BMI. Prevalence of diabetes
(type 2, 97.8%) preceding PD onset in this cohort
was 11.4%. The final study population consisted
of 178 patients (89 cases of diabetes and 89 con-
trols matched for gender, BMI, and duration of
PD). The pharmacologic treatment for diabetes in
use at study entry is presented in table 1. Diabetic
patients were similar to controls in terms of most
demographic, lifestyle, and general medical fea-
tures; an exception was the higher use of
cholesterol-lowering medications (table 2). How-
ever, regarding clinical features of PD, we ob-
served that diabetic patients not only required
treatment with higher doses of levodopa (table 2)
but they also experienced more severe symptoms.
In particular, the Hoehn & Yahr staging and the
scores of the first 3 parts of UPDRS were signifi-
cantly higher (table 3). Linear regression models,
adjusted for age, gender, and statin use, confirmed
the differences found in doses of levodopa and se-
verity of symptoms, while lipid-lowering medica-
tions themselves were not associated with outcome
Grouping the patients by duration of disease
(0–1 [n ? 48], 2–3 [n ? 60], 4–6 [n ? 38], 6 years
and longer [n ? 32]) showed that levodopa require-
ments (figure, A and B) and UPDRS part III score
(in the best “on” phase for those with a disease dura-
tion ?6 years; figure, C) differed significantly ac-
cording to the prior onset of diabetes, particularly in
the short term (up to 6 years of disease).
DISCUSSION In the present study we showed that
patients in whom the onset of diabetes occurs before
the onset of PD experience more severe PD symp-
toms and reduced efficacy of levodopa therapy.
These findings provide further support to the hy-
pothesis that diabetes is a risk factor for PD.4–7The
evidence appeared to be stronger in patients with
shorter duration of PD, but it is likely that the lack of
significance found for longer disease duration was
due to the limited statistical power secondary to
small sample size.
To our knowledge, this is the first report address-
ing the influence of prior onset of diabetes on the
clinical features of PD. The strength of our results
rests on the choice of inclusion and exclusion criteria.
Table 3Clinical rating scales in patients with PD at entry to the study
Diabetes (n ? 89) No diabetes (n ? 89)p Valuea
Hoehn & Yahr stage, n (%)
18 (20.2) 29 (32.6)
51 (57.4)54 (60.7)
18 (20.2)4 (4.5)
2 (2.3) 2 (2.2)
0 (0) 0 (0)
UPDRS score, mean (SD)
1.7 (1.8)1.2 (1.5) 0.046
9.7 (5.1) 8.3 (4.3)0.049
22.3 (9.0) 19.3 (7.9) 0.019
33.7 (15.0)28.8 (13.8) 0.024
1.1 (1.7)0.8 (1.3) 0.188
Abbreviations: PD ? Parkinson disease; UPDRS ? Unified Parkinson’s Disease Rating Scale
(part I: mentation, behavior, and mood; part II: activities of daily living; part III: motor exami-
nation; total: part I ? part II ? part III; part IV: complications of therapy).
aAccording to ?2test (distribution of patients among Hoehn & Yahr stages) or Mann-
Whitney U test.
Neurology 78May 8, 2012
We matched study groups not only for gender and
duration of PD, but also for BMI. Although the role
of adiposity as risk factor for PD is still limited and
needs confirmation,14it is well demonstrated that
obesity is associated with a higher prevalence of dia-
betes. Moreover, PD progression is associated with
progressive weight loss.14Diabetes is also an impor-
tant risk factor for cardiovascular disease and it is
more frequent in patients with vascular parkinson-
ism.15In the present study, we tried to include only
patients with IPD on the basis of both clinical and
imaging correlates. Vascular parkinsonism appears to
be characterized by particular neuroimaging find-
ings, higher age at onset, smoking status, cardiovas-
cular disease, and higher prevalence of lower body
predominance of motor involvement, postural insta-
bility, and gait disturbances.11,15In our study groups,
prevalence of current or ex-smokers and main motor
symptoms at diagnosis were similar among diabetic
and nondiabetic patients. Moreover, we found no
difference in age at diagnosis of PD. Being that dia-
betes is a risk factor for PD, disease onset would be
expected at a younger age in diabetic patients. How-
ever, diabetes and PD appear in some cases to cluster
with age6while in others the risk of PD appears to be
higher when diabetes had developed at least 10 years
earlier.7Unfortunately, we were not able to explore
this issue thoroughly and we admit that the lack of
information on the duration of preceding diabetes is
a limitation of our study. Moreover, we cannot ex-
clude survival bias due to high mortality at early ages
among diabetic patients.
The mechanisms directly responsible for dysfunc-
tion of the nigro-striatal pathway (pre- and postsyn-
aptic areas) are still unclear. Hyperglycemia
contributes to endothelial dysfunction,16and im-
paired insulin action, due to insulin resistance or de-
ficiency, could also lead to neurodegeneration by
modulating dopamine concentrations in the brain
and neuron survival.1,2Moreover, some evidence
suggests that chronic low-grade inflammation and
oxidative stress may play a role.17,18
Other limitations deserve a more extensive dis-
cussion. The main limitation is the case-control
design that allows for the inference of weak caus-
ative relationships.19We tried to exclude cases of
secondary parkinsonism, particularly those of vas-
cular origin, not only on the basis of general clini-
cal criteria, but also of neuroimaging findings.11,15
However, internationally accepted diagnostic cri-
teria are not yet available and the responsibility of
microvascular complications could not be ex-
cluded. Self-reporting of clinical conditions, such
as a history of diabetes, could be another source of
bias.19However, all diabetic patients reported the
use of antidiabetic drugs. We also assessed fasting
glucose in every patient. However, the use of more
sensitive methods, such as an oral glucose toler-
ance test, would have excluded also asymptomatic
of motor symptoms according to Unified Parkinson’s Disease Rating Scale (UPDRS) part III.
Data are presented as mean ? SD bars. For all the analyses (analysis of variance) a statisti-
cally significant (p ? 0.05) effect for diabetes was detected (*p ? 0.05 for matched groups;
comparisons by post hoc test).
Neurology 78May 8, 2012
diabetic patients from the nondiabetic control Download full-text
group. Within the spectrum of the metabolic syn-
drome diabetes may be associated with higher rates
of other cardiovascular risk factors, such as high
blood pressure, hypercholesterolemia, and hyper-
uricemia. Both positive and null associations be-
tween PD andhypertension
reported,20,21while high serum cholesterol and
uric acid have been found to be protective to some
extent.21–24Prevalence of hypertension was similar
in our study groups, but we did not assess the
prevalence or the levels of the other factors. We
did find that the use of statins was more common
among diabetic patients, but this variable was not
associated with clinical presentation. No informa-
tion on glucose control was collected. However,
all the patients were under antidiabetic therapy.
It would be interesting to know whether better
glucose control is associated with less severe symp-
toms or not.
Although the present findings need to be con-
firmed, neurologists should be aware of the potential
impact of diabetes on overall PD patient manage-
ment. Future studies should investigate the patho-
genic pathways involved, the possible existence of a
genetic link between diabetes and PD, as well as the
role of glucose control in PD progression.
All the authors significantly contributed to the work and approve the
manuscript for submission. E. Cereda: data collection, data analysis, data
interpretation, and manuscript drafting. M.B.: data interpretation, re-
view, and critique. E. Cassani: data collection. R.C.: review and critique.
G.P.: data interpretation, supervision, review, and critique.
The authors thank Jennifer S. Hartwig, MD, for assistance in editing the
The authors report no disclosures relevant to the manuscript. Go to
Neurology.org for full disclosures.
Received October 18, 2011. Accepted in final form January 17, 2012.
1.Craft S, Watson GS. Insulin and neurodegenerative dis-
ease: shared and specific mechanisms. Lancet Neurol
2.Cardoso S, Correia S, Santos RX, et al. Insulin is a two-edged
knife on the brain. J Alzheimer Dis 2009;18:483–507.
3.Lu FP, Lin KP, Kuo HK. Diabetes and the risk of multi-
system aging phenotypes: a systematic review and meta-
analysis. PLoS One 2009;4:e4144.
4.Cereda E, Barichella M, Pedrolli C, Pezzoli G. Diabetes
and risk of Parkinson’s disease: a systematic review and
meta-analysis. Diabetes Care 2011;34:2614–2623.
5.Hu G, Jousilahti P, Bidel S, Antikainen R, Tuomilehto J.
Type 2 diabetes and the risk of Parkinson’s disease. Diabe-
tes Care 2007;30:842–827.
6.Driver JA, Smith A, Buring JE, Gaziano JM, Kurth T, Logros-
Xu Q, Park Y, Huang X, Hollenbeck A, Blair A, Schatzkin
A, Chen H. Diabetes and risk of Parkinson’s disease. Dia-
betes Care 2011;34:910–915.
Sandyk R, Awerbuch GI. The association of diabetes mel-
litus with dementia in Parkinson’s disease. Int J Neurosci
Arvanitakis Z, Wilson RS, Schneider JA, Bienias JL, Evans
DA, Bennett DA. Diabetes mellitus and progression of ri-
gidity and gait disturbance in older persons. Neurology
Arvanitakis Z, Wilson RS, Bienias JL, Bennett DA. Diabe-
tes and parkinsonian signs in older persons. Alzheimer Dis
Assoc Disord 2007;21:144–149.
Kalra S, Grosset DG, Benamer HT. Differentiating vascu-
lar parkinsonism from idiopathic Parkinson’s disease: a
systematic review. Mov Disord 2010;25:149–156.
Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What fea-
tures improves the accuracy of clinical diagnosis in Parkin-
son’s disease: a clinicopathologic study. Neurology 2001;
Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke
CE. Systematic review of levodopa dose equivalency re-
porting in Parkinson’s disease. Mov Disord 2010;25:
Barichella M, Cereda E, Pezzoli G. Major nutritional
issues in the management of Parkinson’s disease. Mov Dis-
Winikates J, Jankovic J. Clinical correlates of vascular par-
kinsonism. Arch Neurol 1999;56:98–102.
Haratz S, Tanne D. Diabetes, hyperglycemia and the
management of cerebrovascular disease. Curr Opin Neurol
lin resistance. J Clin Invest 2003;112:1821–1830.
Chen H, O’Reilly EJ, Schwarzschild MA, Ascherio A. Pe-
ripheral inflammatory biomarkers and risk of Parkinson’s
disease. Am J Epidemiol 2008;167:90–95.
Bruzzi P, Green SB, Byar DP, Brinton LA, Schairer C.
Estimating the population attributable risk for multiple
risk factors using case-control data. Am J Epidemiol 1985;
Qiu C, Hu G, Kivipelto M, et al. Association of blood
pressure and hypertension with the risk of Parkinson dis-
ease: the National FINRISK Study. Hypertension 2011;
Simon KC, Chen H, Schwarzschild M, Ascherio A. Hy-
pertension, hypercholesterolemia, diabetes, and risk of Par-
kinson disease. Neurology 2007;69:1688–1695.
Hu G, Antikainen R, Jousilahti P, Kivipelto M,
Tuomilehto J. Total cholesterol and the risk of Parkinson
disease. Neurology 2008;70:1972–1979.
Chen H, Mosley TH, Alonso A, Huang X. Plasma urate
and Parkinson’s disease in the Atherosclerosis Risk in
Communities (ARIC) study. Am J Epidemiol 2009;169:
Weisskopf MG, O’Reilly E, Chen H, Schwarzschild MA,
Ascherio A. Plasma urate and risk of Parkinson’s disease.
Am J Epidemiol 2007;166:561–567.
Neurology 78 May 8, 2012