Journal of Diabetes and Its Complications 33 (2019) 603–609

Contents lists available at ScienceDirect

Journal of Diabetes and Its Complications

journal homepage: WWW.JDCJOURNAL.COM

Periodontal disease, smoking, cardiovascular complications and

mortality in type 1 diabetes
Tumader Khouja a, Rachel G. Miller b, Paul A. Moore c, Trevor J. Orchard b, Tina Costacou b,⁎
a
University of Pittsburgh, Graduate School of Public Health, Department of Health Policy and Management, Pittsburgh, PA, United States
b
University of Pittsburgh, Graduate School of Public Health, Department of Epidemiology, Pittsburgh, PA, United States
c
University of Pittsburgh, School of Dental Medicine, Department of Dental Public Health, Pittsburgh, PA, United States

a r t i c l e i n f o a b s t r a c t

Article history: Aim: To assess the role of periodontal disease (PD) as a predictor of coronary artery disease (CAD) and mortality
Received 14 February 2019 in a prospective type 1 diabetes (T1D) cohort and to evaluate the role of smoking in this relationship.
Received in revised form 21 May 2019 Methods: Data were based on 320 participants of the Pittsburgh Epidemiology of Diabetes Complications study of
Accepted 28 May 2019 T1D who, during 1992–94, received a partial mouth periodontal exam, and who were followed for up to 19 years
Available online 3 June 2019
to ascertain complication incidence. PD was defined as clinical attachment loss of ≥4 mm for at least 10% of the
examined sites. Predictors of all-cause mortality; Hard CAD (CAD death, myocardial infarction or revasculariza-
Keywords:
Type 1 diabetes
tion), and Total CAD (Hard CAD, angina, ischemic ECG) were assessed using Cox models.
Periodontal disease Results: During 19 years of follow-up, 33.7% (97/288) developed CAD, 27.3% (83/304) developed Hard CAD, and
Coronary artery disease 16.9% (54/320) died. Among current smokers, 46.4% (26/56) developed CAD, 42.7% (24/56) developed Hard CAD
Smoking and 29.5% (18/61) died. PD was not associated with all-cause mortality, although it was a significant predictor of
Diabetes complications both CAD (HR = 1.12, CI = 1.01–1.23) and Hard CAD (HR = 1.30, CI = 1.11–1.51). As smoking modified the PD-
CAD and PD-Hard CAD associations, analyses were stratified by smoking status. PD was associated with an in-
creased risk of CAD (HR = 1.25, CI = 1.03–1.50) and Hard CAD (HR = 1.85, CI = 1.17–2.93) only among
smokers.
Conclusion: PD was a significant predictor of CAD and Hard CAD among current smokers with T1D.
© 2019 Elsevier Inc. All rights reserved.

1. Introduction end products (AGE) that activate various proinflammatory mediator

cascades leading to periodontal tissue damage.7,8
Type 1 diabetes (T1D) is one of the most common chronic diseases of PD has been linked to systemic diseases such as cardiovascular dis-
childhood. Recent data from the U.S. suggest that the incidence of T1D ease (CVD), diabetes and chronic kidney disease. 9–11 Although studies
among youth has increased by 1.8% annually from 2001 to 2012. 1 T1D have found a strong association between PD and CVD, a large NHANES
is associated with increased morbidity and mortality with an associated study suggested that smoking plays a significant role in the PD-CVD re-
cost of $14.4 billion per year in direct medical expenses and indirect lationship as an effect modifier,12 while others have found the relation-
costs.2 Periodontal disease (PD) is one of the main oral manifestations ship to be independent of smoking. 13 Both diabetes and PD have been
of T1D.3 individually identified as risk factors for CVD. The combined effect of
PD is a chronic inflammatory disease of the surrounding tooth struc- PD and diabetes on the development of CVD has been studied widely
ture caused by pathogens, leading to tissue destruction and tooth loss. It in type 2 diabetes. 14 Thus, individuals with type 2 diabetes and PD
is estimated that 47.2% of US adults have some form of PD, based on data were shown to have a higher incidence of coronary artery disease,15 ca-
from NHANES 2009–2012.4 Smoking and hyperglycemia are two mod- rotid atherosclerosis 16 and myocardial infarction. 17 Findings from the
ifying risk factors for PD.5 Smoking leads to a strong inflammatory reac- Study of Health in Pomerania further suggested that although measures
tion that has detrimental effects on the periodontium and can increase of PD were independently associated with all cause and CVD mortality,
the risk of periodontitis 2–5 times.5,6 Hyperglycemia in patients with di- there was no evidence of interaction between diabetes and
abetes leads to oxidative stress and the formation of advanced glycation periodontitis. 18 A single study in individuals with T1D suggested that
PD was significantly associated with coronary artery calcium progres-
⁎ Corresponding author at: University of Pittsburgh, Pittsburgh, PA 15213-3310, United
sion, a marker of subclinical atherosclerosis. 19 The aim of this study
States. was therefore to assess the role of PD as a predictor of documented car-
E-mail address: [email protected] (T. Costacou). diovascular complications and mortality in a cohort of individuals with

https://doi.org/10.1016/j.jdiacomp.2019.05.025
1056-8727/© 2019 Elsevier Inc. All rights reserved.
604 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609

childhood-onset T1D and to evaluate the effect of smoking on this a random-zero sphygmomanometer. 30 Hypertension was defined as
relationship. blood pressure N 140/90 mm/Hg or use of blood pressure-lowering
medications.
2. Materials and methods Serum and urinary albumin were measured by
immunonephelometry31,32 and creatinine was assayed by an Ectachem
2.1. Study population 400 Analyzer (Eastman Kodak Co., Rochester, NY). Albumin excretion
rate (AER) was calculated for each of three timed urine samples (24-h,
The Pittsburgh Epidemiology of Diabetes Complications (EDC) study overnight and 4-h collections obtained over a 2-week period); the me-
is a prospective cohort study of childhood-onset (b17 years) T1D. All dian of the three AERs was used in analyses and was natural logarithm
participants of the EDC study were diagnosed, or seen within 1 year of transformed due to its skewed distribution. White blood cell (WBC)
diagnosis, at Children's Hospital of Pittsburgh between 1950 and 1980. count and hemoglobin were measured using a Coulter Counter S-Plus
The cohort has been described in detail elsewhere.20,21 In brief, partici- IV. Height and weight were measured using standard methods to calcu-
pants (n = 658) have been followed with biennial surveys since study late body mass index (BMI).
initiation (1986–1988). Clinical examinations occurred biennially for
the first 10 years and thereafter at 18- and 25-years post baseline. Be-
tween March 1992 and August 1994, of 412 participants scheduled for 2.4. Assessment of outcomes
an EDC clinic visit, 406 enrolled in a dental study. Of these, 16 were
missing all their teeth, two had scheduling conflicts that prevented Participants were followed until October 31, 2014 to ascertain com-
complete oral health assessments, and 68 were excluded for possible plication status (median follow-up time, 19 years). Three main out-
risk of bacteremia, leaving 320 eligible to receive a comprehensive comes were assessed for this analysis. All-cause mortality; Hard
oral health assessment, including a periodontal examination. The meth- Coronary Artery Disease (Hard CAD; CAD death, myocardial infarction
odology of the oral health assessment is described in detail elsewhere. 22 confirmed by Q-waves on electrocardiogram (Minnesota codes 1.1 or
Briefly, a periodontal examination was conducted following the Na- 1.2) or hospital records, or revascularization); and total Coronary Artery
tional Institute of Dental Research (NIDR) adult survey methodology. 23 Disease (CAD; Hard CAD but also including angina, determined by the
Three facial sites (mesial, mid-cervical and distal) of the right maxillary/ EDC study physician, and ischemic electrocardiogram changes (Minne-
left mandibular or left maxillary/right mandibular quadrants were sota codes 1.3, 4.1–4.3, 5.1–5.3, 7.1)). In the EDC study, mortality was
probed, excluding third molars. Clinical attachment loss and pocket ascertained using medical records, death certificates, autopsy reports,
depths were measured using a standard WHO Community Periodontal and/or interview with next of kin.
Index of Treatment Needs (CPITN) pressure-controlled probe by one
of two calibrated dentist examiners. Bleeding on probing and visual as-
sessment of supragingival calculus was assessed as present or absent on 2.5. Statistical analysis
each tooth examined. In addition, all missing teeth excluding the third
molar were recorded using modified criteria from the NIDR adult survey Differences in baseline characteristics were evaluated between
to determine the cause of extraction (disease or orthodontic treatment). PD cases and non-PD cases using the Student's t-test for normally
distributed continuous variables, the Wilcoxon rank sum (Mann-
2.2. Assessment of PD Whitney U) test for non-normally distributed continuous variables
and the chi-square or Fischer's exact test for categorical variables.
Participants who had clinical attachment loss of ≥4 mm in N10% of The Cochran–Armitage test for trend was used for ordinal
periodontal sites examined were defined as having PD. This definition variables.
reflects the Healthy People 2000 and 2010 definition of PD. 24,25 This Kaplan Meier curves were used to assess survival probabilities
parameter was selected to describe a clinically significant amount of between PD cases and non-PD cases for each of the outcomes. Pre-
disease, include an ample sample size for analysis and minimize dictors of CAD, Hard CAD and all-cause mortality were assessed
misclassification of cases due to measurement error.22 using Cox models, excluding prevalent cases of CAD or Hard
CAD, as appropriate, at the time of the oral health exam. Survival
2.3. Assessment of covariates time was defined as the time in years from the oral health exam
to the date of the first event for each outcome studied or censor-
Covariates of interest were selected for analysis from the time of the ship. The proportional hazards assumption was assessed visually
oral health exam. The number of missing teeth was assessed clinically and confirmed by testing time-dependent PD interaction variables.
during the oral health exam. Demographic data, including age, sex, ed- PD violated the proportional hazard assumption for both CAD and
ucational status (used as an indicator of socioeconomic status), and al- Hard CAD and was therefore added to models relating to these
cohol consumption were assessed via survey. Smoking status was two outcomes as a time-varying covariate.
assessed by self-report to the question “Have you smoked at least 100 We assessed the role of current smoking as an effect modifier by in-
cigarettes in your lifetime?” Participants who responded in the affirma- cluding an interaction term between PD and current smoking in the
tive were asked if they currently smoke in a follow-up question. Those models along with the lower order terms. Stratified analyses by current
who responded positively were classified as current smokers, while all smoking status were conducted when effect modification was observed.
others, including former smokers, were considered non-smokers. Cox proportional hazards models stratifying by current smoking status
Fasting blood samples were taken to measure HbA1, lipids, lipopro- were first constructed assessing the association between each potential
teins, serum creatinine and serum albumin. HbA1 values were con- risk factor and the outcome of interest, allowing only for diabetes dura-
verted to DCCT (Diabetes Control and Complications Trial)-aligned tion. Variables that were significantly associated with the outcome were
values HbA1c using a regression equation derived from duplicate assays subsequently included in separate multivariable Cox models for current
[DCCT HbA1c = 0.14 + 0.83 (EDC HbA1)].26 Total cholesterol and tri- smokers and non-smokers. Backward elimination with a significance
glycerides were determined enzymatically.27,28 High density lipopro- level of 0.05 was used to retain significant covariates in the models.
tein (HDL) cholesterol was determined using a modified precipitation All analyses were repeated replacing the dichotomous covariate for hy-
technique. 29 Non-HDL cholesterol (non-HDLc) was calculated by pertension status with a continuous variable for systolic blood pressure.
subtracting HDL from total cholesterol. Blood pressure was measured All statistical analyses were conducted using SAS® 9.4 software (SAS In-
according to the Hypertension Detection and Follow-Up protocol with stitute Inc., Cary, NC, USA.)
 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609 605

Table 1 3. Results
Baseline characteristics of EDC participants who received an oral health exam (1992–
1994) by periodontal disease (PD).
Participants with prevalent CAD at the time of the oral health exam
Non-PD cases PD cases p-Value (n = 32) were excluded from analyses. The prevalence of PD in this co-
(n = 286) (n = 34) hort was 10.6%. Table 1 describes the baseline characteristics of the
Age (years) 31.47 (7.67) 37.61 (6.06) b0.0001 study population based on PD status. PD cases were significantly
Age at onset (years) 8.09 (4.08) 10.71 (3.75) 0.0004 older, with a later age at the onset of T1D, more likely to have less
Duration of diabetes (years) 23.38 (7.34) 26.90 (7.47) 0.009
than a high school education as well as more likely to have more miss-
Female sex 130 (45.45) 12 (35.29) 0.26
More than high school education 208 (72.73) 18 (52.94) 0.02 ing teeth compared with non-PD cases. The prevalence of current
≥7 oz alcohol/wk. (n = 282,33) 199 (70.57) 20 (60.61) 0.24 smoking was significantly higher among PD cases compared with
Current smoker 40 (13.99) 21 (61.76) b0.0001 non-PD cases (current smoker 61.8% vs 14.0%). WBC count was also sig-
Number of missing teeth nificantly higher among cases, although there were no differences in
None 161 (56.29) 6 (17.65)
other biological markers by PD status. PD cases had a higher incidence
1–4 97 (33.92) 10 (29.41) b0.0001
≥5 28 (9.79) 18 (52.94) of CAD, Hard CAD and all-cause mortality.
White blood cell count (x103/μL, 7.11 (1.97) 8.80 (2.92) b0.0001 During 19 years of follow-up, 33.7% (97/288) developed CAD, 27.3%
n = 283,34) (83/304) developed Hard CAD, and 16.9% (54/320) died. Among current
BMI (kg/m2 n = 284,33) 24.66 (3.35) 23.90 (3.26) 0.22
smokers, 46.4% (26/56) developed CAD, 42.7% (24/56) developed Hard
Hypertension 50 (17.48) 10 (29.41) 0.09
HDL cholesterol (mg/dL, 52.07 (12.40) 53.01 (14.03) 0.68 CAD and 29.5% (18/61) died. Table 2 shows the characteristics of the
n = 285,34) study population at the time of the oral health exam by the follow-up
Non-HDL cholesterol (mg/dL, 134.2 (37.07) 145.3 (39.19) 0.10 status of each of the three outcomes of interest. Regardless of outcome
n = 285,34) studied, participants who experienced an event were older, with a lon-
Albumin excretion rate (μg/min) 14.74 23.82 0.44
ger duration of diabetes, more likely to be hypertensive, with higher
(5.18–69.07) (6.27–143.33)
HbA1c (%, n = 284,34) 9.25 (1.41) 9.40 (1.36) 0.53 levels of HbA1c, white blood cell count, non-HDL cholesterol and albu-
CAD incidence (n = 259,29) 79 (30.50) 18 (62.07) 0.0006 min excretion rate. Incident cases were also more likely to be missing
Hard CAD incidence (n = 273,31) 68 (24.91) 15 (48.39) 0.005 a larger number of teeth and a greater proportion of incident cases
All-cause mortality 43 (15.03) 11 (32.35) 0.01
had PD. Fig. 1 shows a clear separation of the Kaplan Meier survival
Data are means (SD), medians (interquartile range) or n (%). curves according to PD status for all outcomes studied: CAD (Fig. 1a),
hard CAD (Fig. 1b) and Mortality (Fig. 1c).

Table 2
Baseline characteristics for all EDC participants who received an oral health exam by incident outcome of interest.

Participant characteristics CAD (N = 288) Hard CAD (N = 304) All-cause mortality (N = 320)

No Yes p-Value No Yes p-Value No Yes p-Value

(n = 191) (n = 97) (n = 221) (n = 83) (n = 266) (n = 54)

Age (years) 29.73 (7.44) 35.33 (6.38) b0.0001 30.38 (7.54) 35.78 (6.40) b0.0001 31.19 (7.51) 36.67 (7.26) b0.0001
Age at onset (years) 8.38 (4.18) 8.21 (4.01) 0.67 8.46 (4.05) 8.34 (4.31) 0.83 8.41 (4.09) 8.15 (4.29) 0.67
Duration of diabetes (years) 21.35 (6.65) 27.11 (7.20) b0.0001 21.93 (6.86) 27.43 (7.03) b0.0001 22.79 (6.98) 28.53 (7.74) b0.0001
Female sex 87 (45.55) 44 (45.36) 0.23 99 (44.80) 38 (45.78) 0.88 122 (45.86) 20 (37.04) 0.23
More than high school education 139 (72.77) 65 (67.01) 0.78 157 (71.04) 57 (68.67) 0.69 187 (70.30) 39 (72.22) 0.78
Alcohol consumption N = 187 N = 96 N = 217 N = 82 N = 261 N = 54
≥7 oz/week 130 (69.52) 67 (69.79) 0.96 149 (68.66) 60 (73.17) 0.45 186 (71.26) 33 (61.11) 0.14
Current smoker 30 (15.71) 26 (26.80) 0.02 32 (14.48) 24 (28.92) 0.01 43 (16.17) 18 (33.33) 0.017
Periodontal disease 11 (5.76) 18 (18.56) 0.01 16 (7.24) 15 (18.07) 0.005 23 (8.64) 11 (20.37) 0.011
Number of missing teeth
None 114 (59.69) 40 (41.24) b0.0001 129 (58.37) 31 0.0004 154 13 b0.0001
1–4 55 40 68 (37.35) (57.89) (24.07)
≥5 (28.80) (41.24) (30.77) 33 82 25
22 17 24 (39.76) (30.83) (46.30)
(11.52) (17.53) (10.86) 19 30 16
(22.89) (11.28) (29.63)
HbA1c (%) N = 190 N = 96 N = 220 N = 82 N = 266 N = 52
9.11 9.58 0.0065 9.12 9.63 0.0031 9.14 9.90 0.0003
(1.42) (1.27) (1.35) (1.35) (1.33) (1.60)
White blood cell count (x103/μL) N = 190 N = 95 N = 220 N = 81 N = 264 N = 53
7.021 7.78 b0.0001 6.98 7.74 0.004 7.06 8.42 b0.0001
(1.98) (2.39) (1.98) (2.10) (2.03) (2.40)
2
BMI (kg/m ) N = 189 N = 96 N = 219 N = 82 N = 263 N = 54
24.14 25.20 0.94 24.42 24.96 0.21 24.58 24.61 0.94
(3.02) (3.70) (3.19) (3.69) (3.35) (3.30)
Hypertension 18 (9.42) 27 (27.84) b0.0001 25 (11.31) 27 (32.53) b0.0001 35 (13.16) 25 (46.30) b0.0001
HDL cholesterol (mg/dL) N = 196 N = 96 N = 221 N = 82 N = 266 N = 53
52.49 52.05 0.79 52.31 (12.50) 52.36 0.98 52.26 (12.46) 51.74 (13.17)
(12.49) (12.73) (12.81) 0.79
Non-HDL cholesterol (mg/dL) N = 191 N = 96 N = 221 N = 82 N = 266 N = 53
126.7 (34.02) 150.6 (38.56) b0.0001 128.6 (33.74) 151.2 (41.13) b0.0001 130.9 (33.50) 158.1 (46.91) b0.0001
Albumin excretion rate (μg/min) N = 190 N = 97 N = 220 N = 83 N = 265 N = 54
11.04 21.30 b0.0001 11.04 39.54 b0.0001 11.50 107.03 b0.0001
(4.36–37.33) (6.79–146.54) (4.36–42.30) (9.33–366.43) (4.42–44.27) (21.15–1209.51)

Data are means (SD), medians (interquartile range) or n (%).

606 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609

Fig. 1. Kaplan Meier survival curves stratified by periodontal disease (PD). a. CAD. b. Hard CAD. c. All-cause mortality (0 = non-PD cases, 1 = PD cases).

Results from Cox proportional hazard models for the risk of CAD and and Hard CAD among current smokers only. This is similar to the finding
Hard CAD are displayed in Tables 3 and 4, respectively. In unadjusted from the Coronary Artery Calcification in Type 1 Diabetes study, where
Cox models, PD was significantly associated with a greater risk of CAD self-reported PD was significantly associated with CAC progression at
(HR = 1.11, CI = 1.01–1.23), Hard CAD (HR = 1.24, CI = 1.08–1.43) 6 years follow-up.19 Our study presents stronger evidence of this asso-
and all-cause mortality (HR = 2.41, CI = 1.24–4.7). However, after ciation as we have verified measure of periodontal disease, a longer fol-
allowing for covariates, PD was no longer a significant predictor of all- low up time and verified clinical outcomes. A recent meta-analysis also
cause mortality (HR = 0.87, CI = 0.41–1.83), although it remained sig- showed that PD was associated with cardiovascular mortality and coro-
nificantly associated with both CAD (HR = 1.12, CI = 1.01–1.23) and nary heart disease among patients with type 2 diabetes, however there
Hard CAD (HR = 1.30, CI = 1.11–1.51) (not shown). was not enough evidence for this association in T1D. 14 Studies in the
Significant effect modification of PD by current smoking status was general population found that PD was associated with all-cause mortal-
observed for both CAD (p-interaction b0.01) and Hard CAD (p-interac- ity and was an independent risk factor for CVD, after adjusting for tradi-
tion b0.001) but not for mortality (p-interaction = 0.65). Indeed, in un- tional risk factors including smoking and diabetes.33,34 We observed a
adjusted analyses stratifying by current smoking status, PD significantly significant association between PD and all-cause mortality in the unad-
predicted the development of CAD and Hard CAD among current justed models, although this association was no longer significant after
smokers (HRCAD = 1.29, CI = 1.07–1.56 and HRHard CAD = 1.93, CI = adjusting for confounders. This different finding could be due to a
1.23–3.05) but not among those not currently smoking (HR CAD = smaller sample size and our study focusing on T1D patients only.
1.10, CI = 0.85–1.42 and HR Hard CAD = 1.19, CI = 0.84–1.68). Adjusting Because smoking is a major risk factor for PD and CVD6 and based on
for covariates did not alter these findings (Tables 3 and 4). When anal- findings from previous studies,12 we tested for effect modification by
yses were repeated replacing the categorical covariate for hypertension current smoking status. We found current smoking to be an effect mod-
with a continuous variable for systolic blood pressure, similar results ifier as the PD – CAD/Hard CAD association was restricted to current
were obtained. smokers only. Both smoking and diabetes are known risk factors for
periodontal disease. 35 Although it is known that compared to controls,
4. Discussion T1D patients have elevated periodontal pro-inflammatory factors, dif-
ferent periodontal pathogen composition 36 and lower salivary flow
In this prospective cohort study of individuals with childhood-onset rates, 37–39 further research is required to investigate how these factors,
T1D, we observed that PD significantly increased the risk of both CAD in addition to genetics, differ in T1D by smoking status. Among the type
 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609 607

Table 3
Cox proportional hazard models for the prediction of CAD.

Crude model Model with Crude model for Crude model for Adjusted model for Adjusted model for
(n = 283, 95 interaction term for smokers (n = 55, 25 non-smokers (n = 228, smokers (n = 55, 25 non-smokers (n = 228, 70
events) PD*smoking events) 70 events) events) events)a

PD (time-dependent) 1.10 (1.002–1.21)b 1.15 (1.04–1.28)c 1.29 (1.07–1.56)c 1.10 (0.85–1.42)

1.25 (1.03–1.50)b 1.09 (0.84–1.40)
PD*smoking 9.45 (1.80–49.74)c
Diabetes duration 1.08 (1.01–1.54)b 1.09 (1.05–1.12)d
HbA1c Not allowed 1.20 (1.01–1.42)b
Non-HDL Not allowed 1.02 (1.01–1.02)d
cholesterol
WBC Not allowed 1.16 (1.01–1.33)b

Data are HR (95%CI).

Multivariable models allowed for univariate predictors of CAD.
a
The model also allowed for log Albumin Excretion Rate and hypertension.
b
p-Value b0.05.
c
p-Value b0.01.
d
p-Value b0.001.

2 diabetes population, a study found that two inhibitors of the osteo- bleeding on probing, which is one of the classical signs of active peri-
blastogenesis pathway, Sclerostin and Dickkopf, were upregulated in odontal disease, is usually masked in smokers due to the vasoconstric-
patients with chronic periodontitis and type 2 diabetes and/or tive effect of nicotine on blood vessels.44,45 Consequently, patients can
smoking.40 Joaquim et al41, found no difference in key periodontal path- be unaware of periodontal problems until the disease progresses to an
ogens between smoking and non-smoking patients with type 2 diabetes advanced stage, which could increase their risk of CAD. Therefore, pa-
compared to smoking and non-smoking non-diabetic patients who had tients with T1D should be advised by their healthcare providers that
generalized chronic periodontitis. In addition, there is no direct mecha- PD, in addition to smoking, places them at increased risk of CVD compli-
nism for how periodontitis affects diabetic complications; suggested cations beyond the traditional risk factors. These patients should be re-
pathways include oxidative stress, dyslipidemia, endothelial dysfunc- ferred to a periodontist and placed on a periodontal treatment regimen.
tion and elevated CRP. 42 There is evidence that periodontal treatment
improves short-term glycemic control and circulating levels of markers 4.1. Strengths and limitations
of inflammation in type 2 diabetes. 14 However, there is insufficient ev-
idence on the effect of PD therapy on HBA1c levels in T1D.14 Our study has some limitations. The use of partial mouth measure
As the risk of CVD associated with PD among patients with diabetes with three facial sites only could have underestimated the prevalence
is significant, screening for PD may provide a cost-effective modality for of PD.46 Partial mouth measures have been shown to bias epidemiolog-
identifying patients at high CVD risk. A recent systematic review ical measures of association between PD and smoking, alcohol, obesity
showed that most patients with diabetes were unaware of the PD- and diabetes. 47 However, when the EDC oral health exam took place
diabetes connection; they were not aware of their risk of PD and did in 1992–94, this partial mouth exam was commonly used as an accept-
not receive information about their oral health risk or advice about able measure of the prevalence of periodontal disease. In addition, we
oral healthcare from their diabetes care provider. 43 This issue has used a definition of periodontal disease (≥4 mm of attachment loss in
been addressed in the recent guidelines developed by the International ≥10% of examined sites). Research shows that the association of PD
Diabetes Federation and the European Federation of Periodontology to with systemic diseases differs according to the classification or defini-
integrate the health care, including oral health care, of patients with di- tion of PD used. 48 Data for this cohort were collected to reflect the
abetes between physicians and dentists. 11 The guidelines state that chil- Healthy People 2000 and 2010 definition of PD 24,25 and do not allow
dren and adolescents with type 1 diabetes should be placed on annual the application of different definitions of PD. However, our main conclu-
oral screenings as soon as possible. 11 It is important to note that al- sion is that patients with T1D who have clinically verified PD and who
though smoking is a significant risk factor for periodontal disease, currently smoke might be at increased risk of developing CVD

Table 4
Cox proportional hazard models for the prediction of Hard CAD.

Crude model Model with Crude model for Crude model for Adjusted model for Adjusted model for
(n = 299, 81 interaction term for smokers (n = 55, 23 non-smokers (n = 244, smokers (n = 55, 23 non-smokers (n = 244, 58
events) PD*smoking events) 58 events) events) events)a

PD (time-dependent) 1.26 (1.09–1.46)c 1.35 (1.15–1.58)d 1.93 (1.23–3.05)c 1.19 (0.84–1.68)

1.85 (1.17–2.93)c 1.18 (0.83–1.70)
b
PD*smoking 7.02 (1.31–37.59)
Diabetes duration 1.13 (1.05–1.21)d 1.09 (1.05–1.13)d
HbA1c Not allowed 1.32 (1.10–1.60)c
Non-HDL Not allowed 1.01 (1.002–1.02)c
cholesterol
Log Albumin Not allowed 1.20 (1.05–1.38)c
excretion rate
WBC Not allowed 1.17 (1.01–1.35)b

Data are HR (95%CI).

Multivariable models allowed for univariate predictors of CAD.
a
The model also allowed for hypertension.
b
p-Value b0.05.
c
p-Value b0.01.
d
p-Value b0.001.
608 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609

complications. Moreover, oral health measures were assessed at one 14. Graziani F, Gennai S, Solini A, Petrini M. A systematic review and meta-analysis of ep-
idemiologic observational evidence on the effect of periodontitis on diabetes an up-
point and we cannot account for changes in PD over time. Smoking sta- date of the EFP-AAP review. J Clin Periodontol 2018;45:167-87.
tus was based on self-report and could be subject to reporting bias. In- 15. Southerland JH, Moss K, Taylor GW, et al. Periodontitis and diabetes associations with
flammation is suspected to play a role in the restriction of PD-CAD/ measures of atherosclerosis and CHD. Atherosclerosis 2012;222:196-201.
16. Zeng XT, Leng WD, Lam YY, et al. Periodontal disease and carotid atherosclerosis: a
HCAD association to smokers. However, we were unable to adjust for meta-analysis of 17,330 participants. Int J Cardiol 2016;203:1044-51.
markers of inflammation beyond WBC count, as they were not collected 17. Xu S, Song M, Xiong Y, Liu X, He Y, Qin Z. The association between periodontal dis-
for this sample. Adding markers of inflammation such as IL-8, TNF and ease and the risk of myocardial infarction: a pooled analysis of observational studies.
BMC Cardiovasc Disord 2017;17:50.
CRP would help further understand the relationship between PD and 18. Kebede TG, Holtfreter B, Kocher T, et al. Association of periodontal destruction and di-
smoking in the development of CVD complications of T1D. Despite abetes with mortality. J Dent Res 2017;96:56-63.
these limitations, this study is unique in that it uses longitudinal data 19. Groves DW, Krantz MJ, Hokanson JE, et al. Comparison of frequency and duration of
periodontal disease with progression of coronary artery calcium in patients with and
from a prospective study with 19 years of follow up with verified CVD
without type 1 diabetes mellitus. Am J Cardiol 2015;116:833-7.
events. This enables us to establish a strong association with the expo- 20. Wagener DK, Sacks JM, LaPorte RE, MaCgregor JM. The Pittsburgh study of insulin-
sure (PD) proceeding the outcome (CVD complications) in T1D, a dependent diabetes mellitus: risk for diabetes among relatives of IDDM. Diabetes
topic that is understudied. 1982;31:136-44.
21. Pambianco G, Costacou T, Ellis D, Becker DJ, Klein R, Orchard TJ. The 30-year natural
history of type 1 diabetes complications: the Pittsburgh epidemiology of diabetes
4.2. Conclusion complications study experience. Diabetes 2006;55:1463-9.
22. Moore PA, Weyant RJ, Mongelluzzo MB, et al. Type 1 diabetes mellitus and oral
health: assessment of periodontal disease. J Periodontol 1999;70:409-17.
PD could be used as an early clinical predictor of CAD complications 23. Miller A, Brunelle J, Carlos J, Brown L, Löe H. The national survey of oral health in US
in TID patients who smoke. In addition, T1D patients who smoke should employed adults and seniors: 1985–1986. Oral Health of United States Adults (NIH
receive coordinated care from both a periodontist and their usual Publication No 87–2868). 1987.
24. Gift HC, Drury TF, Nowjack-Raymer RE, Selwitz RH. The state of the nation's oral
healthcare provider to ensure optimal treatment of both their periodon- health: mid-decade assessment of healthy people 2000. J Public Health Dent
tal disease and CAD risk. Studies that investigate complications of diabe- 1996;56:84-91.
tes should examine PD to further aid our understanding of the PD- 25. US Department of Health and Human Services. Focus area 21 oral health. In: Healthy
People 2010. 2000; 2nd https://www.healthypeople.gov/2010/Document/HTML/
diabetes complications association. Clinical trials to evaluate the effect volume2/21Oral.htm.
of periodontal treatment on outcomes in patients with T1D are 26. Prince C, Becker D, Costacou T, Miller R, Orchard T. Changes in glycaemic control and
warranted. risk of coronary artery disease in type 1 diabetes mellitus: findings from the Pittsburgh
epidemiology of diabetes complications study (EDC). Diabetologia 2007;50:2280-8.
27. Bucolo G, David H. Quantitative determination of serum triglycerides by the use of
Funding enzymes. Clin Chem 1973;19:476-82.
28. Allain CC, Poon LS, Chan CS, Richmond W. Fu PC. Enzymatic determination of total
serum cholesterol. Clin Chem 1974;20:470-5.
This research was funded by NIH grant DK34818 and the Rossi Me- 29. Warnick GR, Albers JJ. Heparin–Mn2+ quantitation of high-density-lipoprotein cho-
morial Fund. lesterol: an ultrafiltration procedure for lipemic samples. Clin Chem 1978;24:900-4.
30. Hypertension detection and follow- up program. The hypertension detection and
follow-up program cooperative group. Prev Med 1976;5:207-15.
Declaration of Competing Interest 31. Ellis D, Buffone G. New approach to evaluation of proteinuric states. Clin Chem
1977;23:666-70.
None. 32. Ellis D, Coonrod BA, Dorman JS, et al. Choice of urine sample predictive of
microalbuminuria in patients with insulin-dependent diabetes mellitus. Am J Kidney
Dis 1989;13:321-8.
References 33. Hansen GM, Egeberg A, Holmstrup P, Hansen PR. Relation of periodontitis to risk of
cardiovascular and all-cause mortality (from a Danish Nationwide cohort study).
1. Mayer-Davis EJ, Lawrence JM, Dabelea D, et al. Incidence trends of type 1 and type 2 Am J Cardiol 2016;118:489-93.
diabetes among youths, 2002-2012. N Engl J Med 2017;376:1419-29. 34. Xu F, Lu B. Prospective association of periodontal disease with cardiovascular and all-
2. Tao B, Pietropaolo M, Atkinson M, Schatz D, Taylor D. Estimating the cost of type 1 cause mortality: NHANES III follow-up study. Atherosclerosis 2011;218:536-42.
diabetes in the U.S.: a propensity score matching method. PLoS One 2010;5, e11501. 35. Genco RJ. Current view of risk factors for periodontal diseases. J Periodontol 1996;67:
3. Mauri-Obradors E, Estrugo-Devesa A, Jane-Salas E, Vinas M, Lopez-Lopez J. Oral man- 1041-9.
ifestations of diabetes mellitus. A systematic review. Med Oral Patol Oral Cir Bucal 36. Polak D, Shapira L. An update on the evidence for pathogenic mechanisms that may
2017;22:e586-94. link periodontitis and diabetes. J Clin Periodontol 2018;45:150-66.
4. Eke PI, Dye BA, Wei L, et al. Update on prevalence of periodontitis in adults in the 37. Coelho A, Paula A, Mota M, et al. Dental caries and bacterial load in saliva and dental
United States: NHANES 2009 to 2012. J Periodontol 2015;86:611-22. biofilm of type 1 diabetics on continuous subcutaneous insulin infusion. J Appl Oral
5. Jepsen S, Caton JG, Albandar JM, et al. Periodontal manifestations of systemic diseases Sci 2018;26, e20170500.
and developmental and acquired conditions: consensus report of workgroup 3 of the 38. Saes Busato IM, Antoni CC, Calcagnotto T, Ignacio SA, Azevedo-Alanis LR. Salivary
2017 world workshop on the classification of periodontal and Peri-implant diseases flow rate, buffer capacity, and urea concentration in adolescents with type 1 diabetes
and conditions. J Periodontol 2018;89:S237-48. mellitus. J Pediatr Endocrinol Metab 2016;29:1359-63.
6. Johannsen A, Susin C, Gustafsson A. Smoking and inflammation: evidence for a syn- 39. Moore PA, Guggenheimer J, Etzel KR, Weyant RJ, Orchard T. Type 1 diabetes mellitus,
ergistic role in chronic disease. Periodontology 2000. 2014;64(1):111–126. xerostomia, and salivary flow rates. Oral Surg Oral Med Oral Pathol Oral Radiol Endod
7. Lappin DF, Robertson D, Hodge P, et al. The influence of glycated hemoglobin on the 2001;92:281-91.
cross susceptibility between type 1 diabetes mellitus and periodontal disease. J 40. Miranda TS, Napimoga MH, Feres M, et al. Antagonists of Wnt/beta-catenin signalling
Periodontol 2015;86:1249-59. in the periodontitis associated with type 2 diabetes and smoking. J Clin Periodontol
8. Chapple ILC, Mealey BL, Van Dyke TE, et al. Periodontal health and gingival diseases 2018;45:293-302.
and conditions on an intact and a reduced periodontium: consensus report of 41. Joaquim CR, Miranda TS, Marins LM, Silva HDP, Feres M, Figueiredo LC, Duarte PM.
workgroup 1 of the 2017 world workshop on the classification of periodontal and The combined and individual impact of diabetes and smoking on key subgingival
Peri-implant diseases and conditions. J Periodontol 2018;89:S74-84. periodontal pathogens in patients with chronic periodontitis. J Periodontal Res
9. Ricardo AC, Athavale A, Chen J, et al. Periodontal disease, chronic kidney disease and 2018;53(3):315-23, https://doi.org/10.1111/jre.12516.
mortality: results from the third National Health and nutrition examination survey. 42. Sanz M, Ceriello A, Buysschaert M, Chapple I, Demmer RT, Graziani F, Herrera D,
BMC Nephrol 2015;16:97. Jepsen S, Lione L, Madianos P, Mathur M, Montanya E, Shapira L, Tonetti M, Vegh
10. Chang JF, Yeh JC, Chiu YL, Liou JC, Hsiung JR, Tung TH. Periodontal pocket depth, hy- D. Scientific evidence on the links between periodontal diseases and diabetes: Con-
perglycemia, and progression of chronic kidney disease: a population-based longitu- sensus report and guidelines of the joint workshop on periodontal diseases and dia-
dinal study. Am J Med 2017;130:61-9. [e61]. betes by the International diabetes Federation and the European Federation of
11. Sanz M, Ceriello A, Buysschaert M, et al. Scientific evidence on the links between peri- Periodontology. Diabetes Res Clin Pract 2018;137:231-41, https://doi.org/10.1016/j.
odontal diseases and diabetes: consensus report and guidelines of the joint workshop diabres.2017.12.001.
on periodontal diseases and diabetes by the international diabetes federation and the 43. Poudel P, Griffiths R, Wong VW, Arora A, Flack JR, Khoo CL, George A. Oral health
European Federation of Periodontology. J Clin Periodontol 2018;45:138-49. knowledge, attitudes and care practices of people with diabetes: a systematic review.
12. Hyman JJ, Winn DM, Reid BC. The role of cigarette smoking in the association be- BMC Public Health 2018;18(1):577, https://doi.org/10.1186/s12889-018-5485-7.
tween periodontal disease and coronary heart disease. J Periodontol 2002;73:988-94. 44. Tarnowski M, Duda-Sobczak A, Lipski J, Zozulinska-Ziolkiewicz D, Wyganowska-
13. Andriankaja OM, Genco RJ, Dorn J, et al. Periodontal disease and risk of myocardial Swiatkowska M. Tobacco smoking decreases clinical symptoms of gingivitis in pa-
infarction: the role of gender and smoking. Eur J Epidemiol 2007;22:699-705. tients with type 1 diabetes - a cross-sectional study. Oral Dis 2018.
 T. Khouja et al. / Journal of Diabetes and Its Complications 33 (2019) 603–609 609

45. Calsina G, Ramon JM, Echeverria JJ. Effects of smoking on periodontal tissues. J Clin severe periodontitis in the atherosclerosis risk in communities study: comparison
Periodontol 2002;29:771-6. of full versus partial-mouth estimates. J Clin Periodontol 2015;42:609-21.
46. Eke PI, Thornton-Evans GO, Wei L, Borgnakke WS, Dye BA. Accuracy of NHANES peri- 48. Beck JD, Moss KL, Morelli T, Offenbacher S. Periodontal profile class is associ-
odontal examination protocols. J Dent Res 2010;89:1208-13. ated with prevalent diabetes, coronary heart disease, stroke, and systemic
47. Akinkugbe AA, Saraiya VM, Preisser JS, Offenbacher S, Beck JD. Bias in estimating the markers of C-reactive protein and interleukin-6. J Periodontol 2018;89:
cross-sectional smoking, alcohol, obesity and diabetes associations with moderate- 157-65.