Comparative study of angiographic changes in diabetic and non-diabetic patients with peripheral arterial disease

Background: Diabetics are at 5-15 times greater risk of developing peripheral arterial disease (PAD) and few studies have compared risk factors and distribution and severity of arterial changes in diabetics compared with non-diabetics. Objectives: To compare angiographic changes between diabetic and non-diabetic patients with advanced PAD and correlate them with risk factors. Methods: A retrospective cross-sectional study was conducted of consecutive patients undergoing lower limb arteriography for PAD (Rutherford 3-6) using TASC II and Bollinger et al. angiographic scores. Exclusion criteria were upper limb angiographies, unclear images, incomplete laboratory test results, and previous arterial surgeries. Statistical analyses included chi-square tests, Fisher’s test for discrete data, and Student’s t test for continuous data (significance level: p < 0.05). Results: We studied 153 patients with a mean age of 67 years, 50.9% female and 58.2% diabetics. A total of 91 patients (59%) had trophic lesions (Rutherford 5 or 6) and 62 (41%) had resting pain or limiting claudication (Rutherford 3 and 4). Among diabetics, 81.7% were hypertensive, 29.4% had never smoked, and 14% had a history of acute myocardial infarction. According to the Bollinger et al. score, infra-popliteal arteries were more affected in diabetics, especially the anterior tibial artery (p = 0.005), while the superficial femoral artery was more affected in non-diabetics (p = 0.008). According to TASC II, the most severe angiographic changes in the femoral-popliteal segment occurred in non-diabetic patients (p = 0.019). Conclusions: The most frequently affected sectors were the infra-popliteal sectors in diabetics and the femoral sector in non-diabetics.


INTRODUCTION
Atherosclerosis is the principal cause of cardiovascular diseases, 1,2 a group of conditions that includes coronary artery disease, peripheral arterial disease (PAD), and cerebrovascular disease. 3If all of its manifestations were considered a single pathological entity, atherosclerosis would be considered the number one cause of death worldwide, estimated at 1/3 of all deaths both globally 4 and in Brazil. 5[8][9] Peripheral arterial disease primarily affects elderly patients with risk factors for atherosclerosis 6,9 and manifests insidiously with progressive intermittent claudication.7 Development of pain at rest, ischemic ulcers, and/or gangrene characterizes critical limb ischemia (CLI), which occurs in around 5-10% of cases.9 The best treatment option in these cases tends to be revascularization of the limb involved, except in patients with severe comorbidities or very limited prognosis of successful revascularization, for whom amputation is the most appropriate treatment.[10][11][12][13][14][15] More than half of PAD patients have an anklebrachial index (ABI) of < 0.9 and are asymptomatic.16 Diabetics have around 5 to 15 times greater risk of developing PAD than people who are not diabetic.12,14,17 In patients with CLI, an arterial imaging study is recommended to best characterize the lesion and for planning surgery.9][20][21] The few studies that have compared arterial injuries in patients with PAD and DM have shown that among diabetics arterial injuries tend to be concentrated in smaller caliber arteries (infrapopliteal).14,18,19,[22][23][24] However, none of these studies used angiographic scores to classify lesions and correlate them with risk factors and demographic data.
Our objective, therefore, was to quantify and compare angiographic changes in diabetic and nondiabetic patients with advanced PAD using scores and correlate them with other risk factors.

MATERIALS AND METHODS
A retrospective, cross-sectional, observational study was conducted of a consecutive series of cases, analyzing arteriographies conducted from 2012 to December of 2016 at a single center, comparing diabetic and non-diabetic patients.The study was approved by the local Ethics Committee (decision number 1.578.037).
Consecutive patients with advanced PAD (Rutherford 3 to 6) with and without diabetes were enrolled.Briefly, the Rutherford classification 25 consists of the following categories: 0 = asymptomatic; 1 = mild claudication; 2 = moderate claudication; 3 = severe claudication; 4 = pain at rest; 5 = small trophic lesion; 6 = extensive necrosis.Categories from 3 to 6 were arbitrarily defined as advanced for the present study.Exclusion criteria included angiographs of upper limbs, patients who had undergone surgical revascularization interventions, incomplete laboratory test results, and unclear angiographs.
Demographic data were collected from the hospital's MV electronic patient record system, including age, ethnicity self-declared on the patient record, comorbidities, and laboratory test results (urea, creatinine, glycemia, and lipid profile).Clinical complaints reported by the patients and confirmed by the physicians were as follows: pain at rest, ischemic or mixed ulcer, edema, cyanosis, and gangrene or infection of limbs.Finally, the Rutherford clinical classification was used to stratify patients with severe PAD (Rutherford grades I, II, and III and classes 1 to 6). 25 Diabetes was defined by the criteria of two fasting glycemia results over 126, glycemia over 200 2 h after a glucose challenge, or casual glycemia over 200 with associated symptoms, whether or not the patient takes insulin, and glycated hemoglobin.Systemic arterial hypertension was defined as systolic arterial blood pressure over 139 mmHg or diastolic arterial blood pressure over 89 mmHg (regardless of treatment), according to the 7th Brazilian Arterial Hypertension Guidelines of 2016, 26 and renal dysfunction was assessed by creatinine result > normal reference value (1.2 and 1.3 mg/dL for women and men, respectively).Patients' lipid profiles were defined as total cholesterol (high density lipoprotein [HDL], low density lipoprotein [LDL], very low density lipoprotein, and triglycerides) and vascular physical examination and patient history were used to classify patients' clinical vascular involvement. 27ll arteriographies are stored in complete form in DICOM format on the server for the hospital's digital records system (an MV system).Images were analyzed by an undergraduate scholarship student, always supervised by one of the team's vascular surgeons.Images were assessed on a workstation, using image optimization tools such as brightness and contrast enhancement, image magnification, digital rulers, and automatic and manual stenosis analyses, among others.
The sample was selected consecutively over the course of the study period, applying the inclusion and exclusion criteria defined.Palpation of pulses, clinical symptoms, and Rutherford classification were the main criteria used when deciding to order angiographs.Although measurements for ABI were taken for these patients, ABI was not included in the study because it is subject to serious limitations in diabetic patients and because we do not routinely calculate the ratio between the pressures at the hallux and arm.Arteriographies confirmed clinical suspicion and were used as the basis for sample selection.
The angiographic images were classified using three different scores: the Inter-Society Consensus for the Management of Peripheral Arterial Disease II (TASC II), 9  The TASC II classification is subdivided into the following sectors: aorto-femoral, femoropopliteal, and infrapopliteal.Four types of vascular involvement are scored in each sector, classified as type A -single stenosis; B -mild stenosis or occlusions; C -moderate stenosis or occlusions; and D -extensive stenosis or occlusions.
The aortoiliac and femoropopliteal sectors were classified by the TASC II 29 score into four classes (TASC II A, B, C, and D). 9 In the absence of TASC II criteria of arterial injury in a given sector, a value of TASC II = 0 was attributed arbitrarily.
The Bollinger et al. score 28 was used to conduct angiographic assessments of 10 arteries: 1) abdominal aorta; 2) common iliac artery; 3) external iliac artery (up to the femoral bifurcation); 4) internal iliac artery (up to the first bifurcation); 5) deep femoral artery (the proximal 15 cm of its main branch); 6) superficial femoral artery (up to where it crosses the medial margin of the femur); 7) popliteal artery (up to the bifurcation of the fibular and anterior tibial arteries, excluding the tibial-fibular trunk); 8) anterior tibial artery (up to the proximal 3 cm); 9) fibular artery; and 10) posterior tibial artery (both up to the proximal 5 cm). 29The original Bollinger et al. score (Table 1) is a matrix on which occlusions or stenoses (columns) are scored according to the extent of the injury (rows).For example, a > 50% lesion of the adductor canal is scored as 6 points, but if there are also < 25% lesions involving more than half of the artery, a further 3 points are added (total = 9).This score therefore provides a semi-quantitative analysis of the severity of the lesions involving a specific artery in a group of patients and also enables vector analysis (evaluating an arterial segment), which widens the study spectrum (Figure 1).
In addition to these scores, an adapted version of the Bollinger et al. score was also created, including scores for the entire distal arterial segment -since the original Bollinger et al. score only covers the proximal 5 cm of the infrapatellar arteries.The authors adopted this modification in order to obtain a score for the entire infrapatellar segment, attributing arbitrary values for the whole of this distal arterial segment, as shown in Table 2. Stenosis classes < 50% were excluded in this segment, because the smaller caliber of the distal arteries prevents assessment of smaller stenoses with the same level of detail as in the proximal arteries.Thus, the arteries of the leg were  classified as infrapatellar popliteal artery (segment P3) -up to the emergence of the anterior tibial artery; tibial-fibular trunk -up to the bifurcation of the posterior tibial and fibular arteries; and anterior tibial, fibular, and posterior tibial arteries -all along their entire course, up to the initial formation of the plantar arch (Table 2).The arteriographic assessments were conducted using Centricity DICOM Viewer 3.0 software, which is incorporated into the GE Healthcare angiography machine (Figure 1).Scores were allocated by the lead author (GOT), under direct supervision, and confirmed by vascular surgery specialists.
All patients who underwent diagnostic angiography or endovascular intervention with concurrent angiography were analyzed and, as such, the sample was selected by convenience.Since this is a comparative study of patterns of lower limb arterial involvement, a calculation for 95% reliability and 10% margin of error estimated a minimum of 150 patient records, divided between diabetic and non-diabetic groups.
Statistical analysis started with descriptive statistics, calculating frequencies and percentages for qualitative variables and means, medians, standard deviations, and minimum and maximum values for quantitative variables.
To test for associations between the variable DM and explanatory variables of interest, the chi-square test or Fisher's exact test, when necessary, were used.A generalized linear model with Poisson distribution was used to determine the influence of DM in relation to the sectors and lesions mentioned.For clinical variables, a test of normality was conducted to verify the distribution of data.Student's t test was used to compare means of variables that exhibited symmetrical distribution between groups of diabetic and non-diabetic patients.For variables that exhibited asymmetrical distribution, a generalized linear model with Gamma distribution was estimated.The significance level was set at p < 0.05.
Analyses were performed using SAS version 9.3, by a statistician from the institution's Research Support Office.

RESULTS
A total of 353 angiographs performed from 2012 to 2016 were initially selected for the study, 200 of which were excluded (n = 153).A total of 1,530 segments were analyzed using the Bollinger et al. score, 306 with the TASC II, and 765 with the modified Bollinger et al. score.Eighty-nine angiographic examinations were of diabetic patients and 64 were of non-diabetic patients (Figure 2).
In the aortoiliac sector (Figure 3), mean total Bollinger et al. scores (stenosis + occlusions) were significantly higher in non-diabetics for the internal iliac artery only (mean scores of 4.4 vs. 2.9 points; p = 0.001).In turn, the TASC II assessment of the aortoiliac sector revealed no significant differences between the two groups (p = 0.051), although the difference was close to attaining significance.
In the femoropopliteal sector (Figure 4), nondiabetics had higher mean total Bollinger et al. scores for the deep femoral artery (mean score of 3.7 vs. 4.8; p = 0.04) and the superficial femoral artery (mean score of 8.3 vs. 10.9;p = 0.008).In turn, the TASC II assessment revealed a higher frequency of severe infrapatellar classifications (TASC II C and D) among non-diabetics than diabetics (p = 0.019).
In the infrapatellar sector (Figure 5), the Bollinger et al. score showed means for > 50% stenosis in the anterior tibial artery (mean scores of 0.7 vs. 0.4; p = 0.005) and 25-50% in the fibular arteries (mean score of 0.8 vs. 0.5; p = 0.04) and posterior tibial arteries (mean score of 0.5 vs. 0.3; p = 0.04) in the diabetic patients, but with no significant differences in the total scores for these arteries.Figure 5 shows the total scores for each artery, summing stenoses and occlusions.
The only statistically significant difference in the modified Bollinger et al. scores (Figure 5) was for popliteal involvement, which was more accentuated among diabetic patients (mean score of 5.0 vs. 3.3; p = 0.02).The anterior tibial artery scored higher among diabetics, but without statistical significance (p = 0.06).

DISCUSSION
With regard to the objective of this study, in general, distal arterial involvement was more frequent among the diabetics and proximal lesions were more frequent among non-diabetics.Among the principal risk factors, smoking was less frequent among diabetics, dyslipidemia was more frequent among diabetics, and SAH was similar in both groups.1][32] The proportions of certain types and ethnicities and their relationship with cardiovascular diseases vary greatly by continent and region. 32,33moking was less frequent among diabetic patients, but the tobacco load of smokers and ex-smokers was similar.These data show that smoking was not of itself an important risk factor for development of PAD in the patient sample as a whole.The association between smoking and DM, SAH, and dyslipidemia is cumulative in development of vascular disease. 9This study is in agreement with data in the literature in terms of demographic data.Nonetheless, Santos et al. 18 found a larger proportion of smokers in a non-diabetic group, which is a subset that needs greater attention to lifestyle changes and because smoking is an independent risk factor for development of PAD.
Dyslipidemia was more frequent among diabetics (HDL cholesterol and triglycerides).However, all patients with PAD treated at this center are given prescriptions for statins, irrespective of whether they have some type of dyslipidemia, which could interfere with this type of analysis.The present study was also unable to determine the patients' degree of compliance with their medication.In general, dyslipidemia in diabetics tends to be characterized by increased triglycerides and reduced HDL cholesterol, while LDL cholesterol concentrations generally do not differ between diabetics and non-diabetics. 30When they studied the morphology of atherosclerotic plaques in diabetics and non-diabetics, He et al. 31 also found that the diabetics had significantly higher levels of triglycerides and, obviously, of glycemia.In contrast to the present study, however, total cholesterol was also elevated in the diabetics, although HDL levels were similar.All of these data are in line with the systemic nature of involvement in DM, causing imbalances in the lipid profile that are most obvious in regard to triglycerides. 32nfections and wet gangrene tend to be more common in diabetic patients, 16,33 especially in our setting, because of access problems, low socioeconomic status, and lack of information, 34 making these patients more susceptible to developing these injuries. 16In this study, data on these conditions are merely for reference, because the study is focused on arterial damage.
The aortoiliac TASC II scores revealed no significant differences between the two groups.This sector encompasses larger caliber arteries and it is possible that the calcium deposits that are more common among diabetics could have provoked greater local patency.This analysis may not have been significant because there really was no difference between the groups or possibly because the sample was not large enough to detect differences.In general, the Bollinger et al. score for this sector is similar to the TASC II, but it is a little more sensitive for the internal iliac artery, with higher scores (significant) among the nondiabetics.We do not see any explanation for this peculiarity.Few previous studies have conducted comparative assessments of this sector in diabetics and non-diabetics, since the incidence of lesions in these sites is low, which makes it difficult to conduct precise statistical assessments.
Previous studies have shown more severe arterial involvement in the femoropopliteal than the aortoiliac sector in both groups, 18,19,23,24,35 with a higher frequency of occlusions than of stenoses.This characteristic was observed in studies by Graziani et al. 23 (who only investigated diabetics) and Bradbury et al. 35 (who only investigated non-diabetics), while both studies revealed higher frequencies of severe stenoses and occlusions in this sector, although with an even higher frequency in the infrapatellar sector.In turn, a comparative angiographic study of the two groups 18 observed that a sample of 117 patients (87 diabetics and 74 nondiabetics) with femoropopliteal occlusion exhibited no difference between diabetics and non-diabetics in terms of popliteal artery refilling.In the present study, the TASC II score for the femoropopliteal sector showed a higher frequency of severe classifications (TASC II C and D) among non-diabetics than among diabetics, which further confirms the statement that non-diabetics exhibit more accentuated proximal involvement than distal, in contrast to what is seen in diabetics.The Bollinger et al. score also showed a predominance of lesions in deep femoral, superficial femoral, and popliteal arteries among non-diabetics.Arterial narrowing in the adductor canal and arterial encapsulation inside this muscular aponeurotic canal have been identified as factors predisposing to plaques in this area. 36However, there is no explanation for the fact that diabetics exhibit different behavior in this sector.
Previous studies 18,19,23,24,35 have also reported a greater frequency of lesions in the infrapatellar sector in diabetic patients.The present study corroborates these findings with a significantly higher frequency of > 50% stenosis in the anterior tibial artery and 25-50% stenosis in the fibular and posterior tibial arteries.However, total scores (stenosis + occlusions) were similar in both groups (Figure 4).With relation to distal involvement, Santos et al. 18 found a similar frequency of opacification of popliteal, fibular, and anterior tibial arteries in diabetics and non-diabetics, although there was a higher frequency of posterior tibial artery occlusion among diabetics (p = 0.008).In contrast, the modified Bollinger et al. score revealed greater popliteal involvement among the diabetics.Additionally, in a study by Santos et al., 18 the diabetics were predominantly female, hypertensive, and nonsmokers, and logistic regression showed that only female sex was a risk factor for non-opacification of the posterior tibial artery.In an analogous study, Jude et al. 24 showed that arterial injuries in diabetics were predominantly in the deep femoral artery and all infrapatellar arteries, but these authors did not conduct qualitative stratification of the injuries found.Thus, in the present study, presence of stenosis in the three major arteries of the leg was significant, but there was only a significant association with > 50% stenosis for the anterior tibial artery.Multiple lesions in distal arteries can limit the success of vascular reconstruction because of deficient runoff. 37ertain limitations of the present study should be acknowledged.It is a retrospective study and, as such, patient records tend to have a considerable proportion of missing data, which leads to a large number of exclusions and a high probability of biases, thereby weakening the quality of the sample and the study conclusions.Another limitation was the inability to analyze damage to the extremities using the Wagner, WIFi, or Glass classifications, 38 also because of the study's retrospective design.Although scores were used, and checked by specialists, a certain degree of subjectivity may have affected assessment of the arteriographies.The dyslipidemia data may be subject to bias because the patients were taking statins.Nonsignificant results may have been because of true absence of significance or because the sample was too small after the excessive losses.A large-scale multicenter sample could clear up this doubt.
Randomized prospective studies, and particularly multicenter designs, are needed to improve the quality of evidence on the differential profile of peripheral artery involvement in diabetics.

CONCLUSIONS
Among the patients in the present study, the Bollinger et al. score revealed higher scores for arterial lesions to the internal iliac, deep femoral, and superficial femoral arteries among the non-diabetics.However, popliteal artery scores were higher among diabetics.The TASC II classification revealed similar frequencies of aortoiliac classifications in diabetics and non-diabetics and higher frequencies of C and D classifications for the femoropopliteal sector among non-diabetics.
In the infrapatellar segment, the Bollinger et al. stenosis scores were significantly higher among the diabetics for all three distal arteries (anterior tibial, posterior tibial, and fibular), but there was no difference in total score (occlusions + stenosis).The modified Bollinger et al. scores were only different for the popliteal artery, with greater involvement among the diabetics.
the Bollinger et al. score, 28 and a modified Bollinger et al. score.

Figure 1 .
Figure 1.Method used to calculate stenosis for the Bollinger et al. score.Two reference lines were drawn tangential to the lateral arterial walls.Stenosis was measured by comparing the comparative distances between the point of narrowing and the normal reference.

Figure 2 .
Figure 2. Flow diagram showing recruitment of cases, exclusions, and arterial segments analyzed.

Figure 3 .Figure 4 .
Figure 3.Comparison of results for aortoiliac sector between diabetics and non-diabetics using the TASCII score and the Bollinger et al. score, by arteries analyzed.*= p < 0.05

Figure 5 .
Figure 5. Results for the Bollinger et al. and modified Bollinger et al. scores for the infrapatellar sector between diabetics and nondiabetics, by arteries analyzed.*= p < 0.05.

Table 1 .
Bollinger et al. score table. 1 point is added for occlusions that are longer than 2 cm.Where there is occlusion, other stenoses or plaques are not counted and where > 50% or 25-50% stenosis is present, plaques are not counted.

Table 2 .
Modified Bollinger et al. score -proposed scores for the distal sector

Table 3 .
Demographic data and risk factors, comparison between groups.