Prevalence of peripheral arterial disease in subjects with moderate cardiovascular risk: Italian results from the PANDORA study Data from PANDORA (Prevalence of peripheral Arterial disease in subjects with moderate CVD risk, with No overt vascular Diseases nor Diabetes mellitus)

Individuals who are diagnosed with lower extremity peripheral arterial disease (PAD) are associated with a reduced quality of life, in addition to an increased risk of atherosclerosis, stroke, cardiovascular (CV) morbidity and mortality [1–3]. Unfortunately the CV burden associated with the presence of PAD is similar, regardless of whether it is the asymptomatic or symptomatic form of PAD [4, 5]. For this reason, current guidelines recommend the measurement of ankle-brachial index (ABI) to identify asymptomatic patients and allow early and appropriate clinical and therapeutic intervention thus reducing their increased risk of CV-related mortality [1, 6, 7]. ABI is recognised as a method that is accurate and reliable in the diagnosis of PAD, compared to angiography and Doppler ultrasound [8–10]. However, although ABI is easily assessable, this technique is poorly utilised in general practice and PAD is still a disease that continues to be under diagnosed and under treated [11]. Moreover, few studies are actually available with an adequate sample size to accurately describe the prevalence of low ABI in the adult population with moderate CV risk, as defined in NCEP ATP-III guidelines [7, 12].

ABI has recently been evaluated for its use in improving the accuracy in prediction of CV risk in combination with the Framingham Risk Score (FRS), compared to the use of FRS alone [13]. This study demonstrated that a low ABI (≤0.90) was associated with a significant increase in the rate of CV morbidity and mortality compared with the overall rate for FRS. Due to the low prevalence of PAD within this patient group, the U.S. Preventive Services Task Force recommends against routine screening for PAD in asymptomatic individuals [14], as screening for PAD in the general population would have few or no benefits. However, current knowledge concerning the epidemiology of PAD is mainly limited to studies carried out in northern European and northern American populations [15–19], and, only more recently, 2 studies performed in southern European countries [20–22]. Recently, due to the increasing weight of evidence on the use of the ABI to predict cardiovascular morbidity and mortality, current Task Force guidelines now supports the measurement of ABI in patients at risk of PAD [23]. Since epidemiological data for atherosclerotic disease may be markedly affected by a range of factors, including genetics, ethnics, diet, lifestyle/environment etc., it is therefore also plausible that the prevalence and natural history of PAD in Southern European countries may differ from that of Northern European countries. Actually, to date, just one Italian prevalence study has been performed on the symptomatic PAD population [20]. The aim of this study was to examine the prevalence of asymptomatic PAD using ABI, in subjects with moderate risk of cardiovascular disease (CVD) in the Italian population.

The methods and main findings from the PANDORA study have been published elsewhere [24, 25]. In brief, PANDORA was a non-interventional cross sectional study performed across 6 European countries (Italy, Belgium, France, the Netherlands, Greece and Switzerland), that examined the prevalence of PAD in subjects (N = 9816) with moderate cardiovascular risk and the absence of diabetes or overt vascular disease (NCT00689377). The present study is a post-hoc analysis performed on the Italian sub-population of the original PANDORA study [25], comprising 5112 evaluable subjects from 197 Family Medicine Investigation Centers. Inclusion criteria were men aged ≥45 years or women aged ≥55 years (CVD risk factor) and the presence of at least one of the following CVD risk factors: cigarette smoking, hypertension, diagnosed dyslipidemia, family history of early coronary heart disease (CHD) or elevated abdominal circumference. Exclusion criteria included symptoms of PAD, diabetes, established CHD or risk-equivalents and the absence of fatty serum data within the past year. The study was approved by local ethic committees, and informed consent was obtained from all participants before admission to the study.

From a total of 10, 287 subjects admitted to the original PANDORA study, 5, 298 of these subjects were enrolled in Italy (51.6% of total) from May 2007 to July 2008. Among admitted subjects, 96.5% (N = 5, 112) were considered evaluable. The most frequent reasons for exclusion were: a) less than 80% of CRF fields completed (2.45%), b) unmeasurable ABI (2.36%) and c) a lack of serum lipid data collected within the previous 12 months (0.68%).

Among the 6 participating countries of the previous PANDORA study, a higher prevalence of PAD was observed in Greece (28.0%) and Italy (22.9%), rather than in France (12.2%),

Belgium (7.0%), the Netherlands (8.1%) or Switzerland (12.2%). Multiple logistic regression analysis confirmed that after Greek subjects, Italians had a greater risk of PAD (Figure 1).

Distribution of ABI values in all Italian subjects is shown in Figure 2. Overall, 22.9% of subjects had an ABI ≤0.9 and another 23.9% between 0.91-0.99. Approximately half of subjects observed had an ABI > 0.99 (53.23%).

We next evaluated the relationship between PAD and individual subjects determinants (country, age, sex, race, marital status), lifestyle habits (alcohol intake, physical activity), and CVD risk factors (smoking, family history of early CHD, hypertension, elevated waist circumference, dyslipidemia diagnosis and treatment, high LDL-C, low HDL-C, statin treatment). Multiple logistic regression did not yield any difference regarding gender, race, physical activity, elevated waist circumference, dyslipidemia, high LDL-C levels or BMI. Variables significantly associated with PAD are shown in Figure 3

To explore risk factor burden, the cumulative number of CVD risk factors in each subject was analysed. In addition to age, most subjects (with or without PAD) presented 2 or 3 CVD risk factors (Figure 4A). Whenever subjects were evaluated for the presence of ≤2 or > 2 risk factors, the frequency of PAD was higher with the presence of 2 risk factors (Figure 4A, inset). Therefore, the frequency of subjects with PAD and without PAD was greatest in the SCORE 10-year CV death risk class (> 2%) (Table 3). However, frequency of PAD subjects was higher than subjects without PAD for high-risk classes (3-4%, 5-9% and 10-14%).

The association between PAD and the number of CVD risk factors (in addition to age), expressed in terms of OR ± 95% CI estimates of frequency of ≥2 risk factors compared to 1 risk factor is shown in Figure 4B. Subjects presenting ≥2 CVD risk factors in addition to age (i.e. ≥3 CVD risk factors) are more affected by PAD compared to subjects with only 1 additional risk factor (p < 0.0001, Cochran-Mantel-Haenszel test).

Concomitant use of other cardiovascular drugs was similar in subjects with PAD than in those without PAD (5.82% vs 7.23%). The quantity of subjects receiving anti-hypertensive drugs was also similar in both groups (54.5% vs 60.05%). However, statin therapy was significantly associated with absence of PAD (Figure 3).

This post-hoc analysis of the PANDORA Study represents the largest study on the general Italian population to date that has focussed on the prevalence of asymptomatic PAD in the community setting. Previously, another study investigated the prevalence of PAD in Italian subjects, but only patients with symptomatic PAD were evaluated [20]. The present study is in agreement with findings from the original PANDORA study since it confirms the prevalence of a low ABI in patients who would otherwise be classified as intermediate and even low CV risk [25]. The prevalence of asymptomatic PAD within the Italian population was 22.9%. Limited use of statins in the total cohort of Italian subjects (10.5%) was also confirmed within the subgroup diagnosed with dyslipidemia (24%). Other participant countries had a higher percentage of patients on statins. The risk of developing PAD is also increased in subjects with a higher consumption of saturated fatty acids [31], which is common in northern Europe [32]. In contrast, risk is reduced in subjects with higher intake of cereal fiber, Vitamin E and wine consumption [31, 33, 34], which are typical components in the so-called "Mediterranean diet" [32]. These food preferences in Italian people may have a favourable action to decrease cardiovascular risk for PAD. Although gender was equally matched in the present study, this may be a potential factor to consider when comparing across differences in other European countries. However, two separate studies, performed on a Turkish and a US population respectively, have demonstrated that the prevalence of PAD does not vary by gender [22, 35]. Subjects without symptoms of the lower limbs and overt CV disease who are at risk of PAD, including those less than 50 years with diabetes and one additional risk factor (smoking, dyslipidemia, hypertension, or hyperhomocysteinemia), or aged 50 to 69 with a history of smoking or diabetes, or aged 70 and older need to be further screened for PAD

as highlighted in the PARTNERS study [19]. These patients need to be submitted to clinical, diagnostic and instrumental assessment, according to recommendations mainly derived from the PARTNERS study [19]. Another ABI study performed in Germany on 6, 880 unselected outpatients in Primary Care, aged ≥65 years, who underwent ABI testing by family physicians in 344 centres, showed a prevalence of PAD of 18% [36, 37]. This frequency is in agreement with results of the present study, even though in the Get ABI study 2.8% of the subjects with low ABI had PAD symptoms. In addition, risk profile in the Get ABI population showed differences from that of PANDORA, which was due to fewer previous cardiovascular events or the presence of diabetes. Nevertheless, as evidenced by PANDORA, the Get ABI also confirmed the usefulness of expanding measurement of ABI over to other risk categories beyond those indicated by guidelines. A more recent study performed in Turkey showed a 20% prevalence of PAD, corroborating findings observed in our study [22]. Moreover, demographic and clinical characteristics of this Turkish population were similar to the Italian cohort (i.e., age, gender ratio, associated risk factors, etc.), suggesting that a relatively high prevalence of PAD can be observed in other Mediterranean-like lifestyles. Previous data reported from the National Health and Nutrition Examination Study (NHANES; 1999-2004) showed a 3% prevalence of low ABI in the low or intermediate risk population [38]. Furthermore, Sumner et al. analyzed data from three NHANES between the years 1999 -2004 and showed that the prevalence of PAD is consistently increasing in asymptomatic adults in the US population. Prevalence rose from 3.7 to 4.6% (p = 0.001) over a six year period [39]. Marked differences between these results and those from the PANDORA study were observed, where even in countries with minimal prevalence of low ABI, frequencies of over 3% were reported, thus emphasizing the need for periodical measurement of ABI in many risk-groups. There are some notable findings resulting from this study, such as the relationship between restricted use of statins and the prevalence of low ABI. The well recognised beneficial prognostic effect of statin treatment on the prevention of vascular events in patients with symptomatic PAD [40] should be emphasized, with the view to improving their lower extremity motor performance [41]. Our findings from the PANDORA study show a possible protective action of statins against the occurrence of PAD and may assist in re-evaluating the pathogenetic role of dyslipidemia in this condition, especially considering the fact that other characteristics normally altered by statins, such as inflammation, were not taken into account.

It is also worth highlighting the association between marital status and the presence of low ABI, with unmarried or widowed subjects more likely affected by low ABI than married subjects. This shows unexpected evidence on the relationship between marital status and prevalence of PAD, although previous reports have described the role of marital status, as part of a more general social support, as influencing clinical symptoms of PAD [42]. Although we have not directly examined what underlying factors may influence the link between marital status (divorced/separated/unmarried) and prevalence of PAD, it is recognised that the rate of depressive symptoms is increased in these individuals [43, 44]. It is well documented that patients with PAD have a higher rate of depression and that this is also correlated with the reduced activity/sedentary lifestyle that they lead [45]. However, our findings did not reveal any discernible difference in sedentary lifestyle between patients with or without PAD, suggesting that this cause-and-effect relationship between marital status and prevalence of PAD is more complex.

The major interest of psycho-social factors as cardiovascular risk factors is also suggested by these PANDORA findings [46]. The explanation for the different frequency distribution of PAD among countries may be related to the variability of subject characteristics, in terms of both frequency distribution and type of risk factors, and could have been influenced by several lifestyle habits [20] in addition to differences in use of statins between several countries.

This study has some limitations, including the lack of a follow-up period on the occurrence of cardiovascular events, over an appropriate period after ABI assessment, in subjects presenting with a low ABI but otherwise classified as being at low or moderate risk and, as such, not considered candidates for more aggressive preventive treatment. However, the adverse prognostic implications of a low ABI (≤0.90) are widely known and are derived from many previous studies that have definitively confirmed this instrumental examination as an independent predictor of high cardiovascular risk [2, 4]. Therefore, the standard use of Doppler remains indispensable for an adequate ABI measurement in family practice, as reported in healthy subjects, patients at risk and patients clinically suspected for PAD [47]. It should be emphasized that most recent epidemiological studies on PAD have been performed only [21] or almost only [30] in primary care service, placing ABI measurement in the daily clinical practice of the general practitioner. The ABI value may be limited in some patients because of the calcification of tibial arteries that are less compressible, resulting in unusually high ABI values (> 1.40) [1]. Data from a recent study by Cacoub and coworkers [30] in a subset of 2, 077 patients older than 55 years and with 2 or more cardiovascular risk factors (including diabetes) in which the ABI was measured by general practitioners in France, provided a prevalence of low ABI (below 0.90) of 10.4%, which is similar to that of the French population of the PANDORA study, confirming the satisfactory reliability of the data collected [25]. The results reported in the present study are in agreement to other previous studies in the general population [22, 36, 37]. The prevalence of PAD evaluated using the ABI was found to be 18-19% in subjects aged above 55 or 65 years in The Netherlands, Germany, the UK, and at approximately 30% in patients with selected vascular risk factors in North America and France [18, 19, 48–50]. Given the very high prevalence of PAD, we observed in moderate-risk patients and the CV morbidity and mortality associated with PAD [1, 2], undoubtedly better public and health professional awareness would help preserve individual CV health and achieve public health goals.

This study provides further evidence that a low ABI is significantly associated with classical CV risk factors. Compared to other countries (from the original PANDORA study) this Italian population showed a restricted statin medical prescription and treatment, although it is unlikely that this effect may alone account for this increased prevalence. This study indicates the appropriate use of ABI in Family Medicine for the identification of patients at high risk and therefore allowing early treatment for the reduction of CV-related death.