Contemporary Treatment of Popliteal Artery Aneurysms in 14 Countries: A Vascunet Report

a Department of Surgical Sciences, Section of Vascular Surgery, Uppsala, Sweden b Department of Vascular Surgery, St Olavs Hospital, Trondheim, Norway c Centro Hospitalar Universitário de Lisboa Central e Hospital de Santa Marta, Lisbon, Portugal d Australian and New Zealand Society for Vascular Surgery, Melbourne, Australia e Vascular Unit, Department of Surgery, Mater Dei Hospital, Malta f Clinic for Vascular and Endovascular Surgery, Serbian Clinical Centre, Belgrade, Serbia g Department of Vascular Surgery, Copenhagen University Hospital Rigshospitalet, Denmark h Clinic of Vascular Surgery, Cantonal Hospital, Winterthur, Switzerland i Department of Vascular Surgery, Landspitalinn University Hospital, Reykjavik, Iceland j Department of Vascular Surgery Medical Centre, Pecs University, Pecs, Hungary k Università degli Studi di Siena, Siena, Italy l Department of Vascular and Endovascular Surgery, Nancy University Hospital, University of Lorraine, Nancy, France Department of Surgical Sciences, Otago University, Dunedin, New Zealand n Department of Vascular Surgery, Helsinki University Central Hospital, Helsinki, Finland


INTRODUCTION
Popliteal artery aneurysms (PAAs) are the most common peripheral arterial aneurysms. Even so, the prevalence is relatively low. Unlike aortic aneurysms, the main clinical presentation is not rupture, which is quite uncommon, 1 but thrombosis and embolism leading to acute or chronic limb ischaemia. 2,3 The number of operations for PAA was estimated at 9.6/ million person years, but varied considerably between countries. 4 This low incidence of surgery makes the disease difficult to study, and management remains controversial and differs between institutions, regions, and countries. 4e6 In many centres, open surgical treatment is the gold standard. The preferred open surgical technique, posterior or medial approach, using vein or a synthetic graft, is controversial. 7,8 Endovascular treatment has emerged as an alternative treatment, and has been used increasingly often more recently. 3,6 A minimally invasive procedure with a short hospital stay is attractive, but questions remain about its durability. 3,9e11 Vascunet, a collaboration of registries for vascular surgery in Europe, Australia, New Zealand, and Brazil, started in 1997. 12,13 It reported on PAA treatment in eight countries between 2009 e 2012, 4 describing great intercountry variability in incidence, indications, and choice of surgical techniques. The authors recommended an update of vascular registries introducing new variables to improve future studies, 4 which took place and created this new extended analysis.
The aim was to evaluate indications, treatment strategies, and outcome of PAA repair in 14 countries in a contemporary setting.

MATERIALS AND METHODS
In April 2019 all national and regional registries collaborating in Vascunet were invited to participate.
Representative surgeons from all 14 countries that accepted this invitation discussed and agreed upon a set of variables, and definition of these, to include in the study (given in Table S1). The variables were chosen based on previous work with PAA, 4 as well as two previous Delphi consensus processes on chronic lower limb ischaemia 14 and acute limb ischaemia. 15 The authors of this paper are responsible for the different registries studied and the accuracy of the included variables from respective countries. All data from contributing countries were merged in to one database that was analysed. Only data from inpatient treatments were included. Not all registries could provide all variables (see Table 2).
Data from 14 countries were included (Australia, Denmark, Finland, France, Hungary, Iceland, Italy, Malta, New Zealand, Norway, Portugal, Serbia, Sweden, and Switzerland). Finland submitted population based data from the Helsinki region. France submitted data exclusively from Nancy, Lorraine. Portugal contributed with already merged data, not with individual cases. Thus, the Portuguese data were only included in descriptive tables, not in the statistical analyses. All data were prospectively registered in a registry devoted to quality improvement and research in vascular surgery, covering both open and endovascular surgery in a defined population. Registries not covering endovascular procedures, or a defined population, were not invited to participate in this project.
Only four countries registered a hybrid procedure (open and endovascular surgery performed simultaneously); these 160 cases were classified as open surgery. The Norwegian registry provided age in five year groups rather than exact age. In Hungary it was estimated that 65% of all operations were captured in the registry. France-Lorraine was not included in the calculation of incidence of PAA repairs, as it was impossible to estimate the proportion of captured operations in the French registry due to private institutions in that region that did not contribute to the registry. The registries in Denmark, Finland, Italy, Iceland, Malta, and Sweden registered follow up at 30 days; in Serbia and Switzerland it was a mix depending on centre and time period; and in the remaining countries early outcome was registered at discharge. One year follow up data were provided by Denmark, Finland, France-Lorraine, Iceland, Italy, Malta, Serbia, and Sweden. Data were presented according to the STROBE statement. 16 The SPSS software package version 25.0 (IBM, Armonk, New York, USA) was used for statistical analysis. Statistical comparisons were performed with cross tabulation with the chi square test for dichotomous variables with different degrees of freedom and analysis of variance for continuous variables. Each individual country was tested against the sum of all the other countries. A p value < .01 was considered significant, adjusting for multiple comparisons. Correlation was tested with the Pearson correlation coefficient. Time trends were not analysed continuously but by comparing entire years. Survival and amputation over time was compared using KaplaneMeier curves and the log rank test, presented with 99% confidence intervals. Cases with missing data on surgical techniques or mode of admission (acute/elective) were omitted (20 cases, 0.2%). No imputation procedure was performed.

RESULTS
During 2012e2018 a total of 9 425 cases of definitive surgery for PAA were identified and included from 11 countries. Denmark and Switzerland contributed with 384 cases from 2017 to 2018. This resulted in 9 809 cases of PAA included in the common database from 13 different countries and regions. The Portuguese patients (n ¼ 955) were included when comparing countries in the tables, and 10 764 procedures were studied including those.
The largest number of procedures were submitted from Italy, Australia, and Sweden ( Table 1). The overall incidence of PAA repair was 10.4 operations/million/year but varied more than eightfold among countries.
Most operations were elective. Among the 26.8% emergency procedures, 2 405 (91.0%) were performed for aneurysm thrombosis and consequent acute limb ischaemia (ALI), and the remaining 236 cases for rupture.  The rupture cases were excluded and analysed separately in all of the following analyses.

Elective cases
Patient characteristics of 7 182 elective operations are presented in . In 70.0% the diameter of the aneurysm was 30 mm. In the Swedvasc registry the proportion of asymptomatic patients having elective repair was 73.9% and they had a mean diameter of 29.7 mm compared with symptomatic patients with a mean diameter of 29.8 mm. The information regarding risk factors and comorbidities was almost complete (>90% available information), showing preoperative differences in study populations between countries that were all statistically significant (p < .001).
Ten of the 14 countries registered the surgical approach used in cases treated by OR. A medial approach was used more often (77.7%) than a posterior one (Table 3). In the 5 350 cases with available information on graft type, 3 415 (63.8%) had a vein graft, 1 894 (35.4%) a synthetic graft, and 41 (0.8%) a composite graft.

Acute limb ischaemia
A total of 2 405 cases (24.5%) were operated on for ALI. Patient characteristics are presented in Table 2. The incidence varied from 0.4/million inhabitants/year in New Zealand to 5.5/million inhabitants/year in Sweden. The mean age was 71.0 years and the mean diameter 35.5 mm. In 84.1% the diameter of the aneurysm was 30 mm or more. Patients presenting with ALI were more often active smokers (46.3% vs. 40.0%, p < .001), and had a lower frequency of cardiac history (25.2% vs. 29.6%, p < .001) than elective cases. Other risk factors and patient characteristics were similar when comparing emergency and elective surgery. OR dominated even more in the emergency cases (83.2%).
The use of pre-or peri-operative thrombolysis was registered in six countries. In those, 21.9% of emergency cases received thrombolysis compared with 0.7% of elective cases. The use of thrombolysis varied considerably between countries, being most frequent in Sweden (39.7% in ALI; 2.7% in elective cases, p < .001).

Ruptured popliteal artery aneurysms
Patients operated on for ruptured PAA (n ¼ 236, 2.4%) were older (74.7 years vs. 71.3, p < .001) and had larger aneurysms (49.2 mm vs. 32.7 mm, p < .001) than those operated on for other indications. Other pre-operative characteristics were similar. Ruptured PAAs were treated by open surgery in 78.5%; and 100% were operated on with a medial approach. During the hospital stay or within 30 days from surgery, 9.1% of the patients had a major amputation and 8.1% died.

In hospital and 30 days outcome
The frequencies of major complications, including death and amputation, were difficult to compare among countries, since some only reported events during the in hospital episode; others reported follow up 30 days after surgery. The results after elective treatment, and for ALI, are given in Table 4. Complication rates were similar if the countries that were outliers in terms of number of patients (Italy, Iceland, and Malta) were excluded. As expected, many complications were more common after operation for ALI than after elective surgery. After elective surgery, graft occlusions were registered in 2.3% of cases, amputations in 0.7% and death in 0.5%, compared with 9.6%, 5.0%, and 1.9% after treatment for ALI (p < .001 for all comparisons).
The amputation and mortality rates after elective surgery were stable during the study period. In the ALI group, however, an increase in the amputation rate at the end of the study period (2017e2018) was identified, log rank p ¼ .007 (Fig. 1). This increase in amputation rate in the ALI group remained when outliers in terms of patient numbers (Italy, Iceland, and Malta) were excluded. This time trend in amputation rate remained also when open and ERs were analysed separately. The proportion of patients treated for ALI decreased during the study period, from 29.7% in 2012 to 23.8% in 2017, p < .001.

One year outcomes
One year follow up data were provided by eight of the registries (Denmark, Finland, France-Lorraine, Iceland, Italy, Malta, Serbia, and Sweden). One year data on amputation were available for 42.6%e100% (mean 85.0%) of the procedures in those countries. In total, information on amputation within one year was available in 3 439 of the PAA repairs, and was 1.0% after elective repair and 8.5% after ALI. Information on one year patency was available in 3 314 cases, and was 83.1% after elective and 74.4% after ALI treatment. Elective OR had a one year amputation rate similar to endovascular treatment (1.2% vs. 0.2%; p ¼ .095), although better one year patency (84.0% vs. 78.4%; p ¼ .005). Patients operated on with vein grafts had higher patency and lower amputation rates at one year than those operated on with synthetic grafts (86.8% vs. 72.3% and 1.8% vs. 5.2%; both p < .001). In subgroup analysis of surgical technique in elective cases (data available for 1551 repairs) the posterior approach had a lower amputation rate (0.0% vs. 1.6%, p ¼ .009) than the medial approach and a trend towards better patency at one year (84.0% vs. 78.7%, p ¼ .021).
Mortality data at one year were available after 1 814 repairs (all repairs from Denmark, Finland, Iceland, Malta,  Serbia, and Sweden) and were 1.4% after elective repair and 6.1% after treatment for ALI (p < .001).

DISCUSSION
The Vascunet collaboration made it possible to amalgamate contemporary data on treatment of PAA across several countries and regions, enabling geographical comparisons and study of time trends. The present study represents by far the largest cohort on PAA ever reported. The number of operations per million inhabitants per year varied more than eightfold between the studied countries. In the previous Vascunet PAA report, 4 the highest incidence of treatment by population for PAA was found in Sweden, verified in this report (Table 1). New Zealand had the lowest incidence of PAA repair, and the lowest proportion of active smokers. An increase in incidence over time was noted in Sweden: 8.3/million/year from the same registry (1994e 2001), 2 17.6/million/year (2009e2011), 4 and 19.3/million/ year in this report. In Norway the incidence increased from 11.9/million/year (2009e2012) to 16.1 (2012e2018), and in Switzerland from 5.2 (2009e2011) to 17.2 (2017e2018). The great increase in Switzerland is partly thought to be explained by improved coverage of the registry. The Swissvasc registry was rebuilt in 2016 and more units joined. It is likely that with this revision of Swissvasc a more accurate prevalence of PAA has been captured.
In Finland and New Zealand a falling incidence was observed: in Finland from 13.9 (2009e2011) to 7.1 (2012e 2018), and in New Zealand from 7.0 (2010e2012) to 2.4 (2012e2018). This great variability in incidence and time trends of PAA repair between countries can be explained by a true difference in prevalence of the disease, differences in diagnostic activity, different indications for PAA treatment, and also by differences in how well the registries capture these particular procedures. In Denmark a more precise reporting of the anatomy of peripheral aneurysms was introduced in 2017. Otherwise, the authors of this paper, who are also responsible for the different registries studied, report no great changes in how the registries capture PAA repair during the studied time period. A national AAA screening program may increase the detection of PAAs since the prevalence of PAA among patients with AAA is high. This may partially explain the highest incidence of elective procedures in Sweden, 17 the only country with such a programme among those contributing to this study. The proportion of emergency surgery is also high in Sweden (30%), contradicting this possible explanation.
Some controversy in indication for treatment of asymptomatic PAAs exists and may affect the incidence of PAA repairs. The primary aim in the management of asymptomatic PAA is to prevent thrombo-embolism, acute ischaemia and subsequent risk of amputation. Approximately 30% of patients treated for PAAs have ALI, 18 and they are known to have poorer outcomes, 19 confirmed in the present study. The present study, however, includes no data on the natural history of PAAs, since only treated patients were studied. Some advocate that all PAAs should be repaired, regardless of size, because of the high complication and amputation rates after ALI. 20 Even when PAAs are initially asymptomatic, patients will develop symptoms at a mean rate of 14% per year (range, 5%e24%), 21 and one third will develop ALI within five years. 22 In patients selected for anticoagulation and/or routine surveillance due to small aneurysm size (2e3 cm) or coexisting cardiovascular or malignant disease, 33%e45% eventually need surgical management anyway. 22 Others suggest that asymptomatic PAAs can safely be observed. 23,24 The presence of thrombus in a PAA, however, appears to increase the risk of developing symptoms and the rate of expansion. 25 Consequently, no international consensus regarding the indications for treatment of asymptomatic PAAs exists. Few registries in this study included information on whether the elective repair was performed for an asymptomatic patient or not, but in the Swedish registry the majority of elective cases were asymptomatic (73.9%). The majority of elective cases in this study were probably asymptomatic, although the exact proportion is unknown. The mean pre-operative diameter of elective PAAs in this report varied from 27.3 mm in Malta to 38.3 mm in Hungary.
The proportion of emergency surgery ranged from 14% in Iceland to 40% in France-Lorraine. No correlation between incidences of PAA repair and the frequency of elective/ emergency treatment for PAA was found, however, suggesting a true difference in prevalence of PAA in the studied populations. The fact that PAA is often associated with multiple aneurysmal disease, and that a family history is common, suggests genetic mechanisms, explaining why ethnic differences may play a role. 26 The only variable that existed to enable classification of the pre-operative degree of ischaemia was the ankle brachial index, and that was only available in eight of 14 registries. There was a great variability, however, since the mean value ranged between 0.26 in Serbia and 0.63 in Denmark (Table 2). Although Rutherford classification would probably have been better, one of the conclusions of the 2020 Clinical Practice Guidelines on the management of Acute Limb Ischaemia was that the classification of ALI needs to be updated and revised. 5 Elective cases were more often treated by ER. Although ER of PAAs decreases length of hospital stay and peri-operative morbidity, its durability is inferior compared with OR. 27,28 The results from the present study show that ER in elective cases had a lower frequency of wound complications, acute coronary events, renal failure and early graft occlusions during hospital stay or 30 day follow up. At one year the endovascular group had a similar amputation rate (0.2% vs. 1.2%; p ¼ .095) but inferior patency (78.4% vs. 84.0%; p ¼ .005). As the primary concern regarding ER is durability, adding long term follow up to the registries is important.
The type of open surgical approach was documented in ten countries ( Table 3). The medial approach dominated for both elective and emergency surgery (77.7% and 91.3%, respectively), but varied greatly between countries (50%e100%). The medial approach has the advantages of being familiar to all vascular surgeons, providing easy access to the entire great saphenous vein (without turning the patient during the procedure) and being the only logical option for bypass grafts that must extend to the distal tibial or pedal vessels. 29 However, multiple studies have reported that late expansion is common after this technique. The aneurysm continues to enlarge if collateral blood flow into the aneurysm sac persists, a situation analogous to that of a type II endoleak with endovascular aneurysm repair. 30,31 Continued expansion can result in pain, swelling or thrombosis due to vein compression, and even rupture. 32 Ravn et al. 7 reported late expansion in 33% of PAA repairs (57/174 cases) after a mean of 7.2 years when the medial approach was used, which was symptomatic in most cases. Late expansion was almost non-existent after an operation with a posterior approach. Thus, follow up beyond 30 days is recommended in patients operated with a medial approach to exclude late sac expansion. 33 In the present study the posterior approach was associated with higher rates of wound complications and early graft occlusion, but with superior patency and a lower amputation rate at one year. Thus in this study, the posterior approach was associated with more early complications but better long term outcome. The groups are not quite comparable, however, since patients with aneurysms extending above the adductor hiatus, or below the origin of the anterior tibial artery, cannot be operated on from behind. The graft material differed greatly between countries (Table 3), similar to a previous Vascunet report on infrainguinal bypass surgery. 34 Since a venous graft is associated with better long term outcome, countries using more prosthetic grafts should review their practice.
This great variation in choice of treatment among countries regarding open or endovascular treatment, and the choice of open surgical approach, emphasises the lack of consensus recommendations for treatment. When comparing outcomes for different countries and after different surgical techniques, there are residual confounders that were not possible to address in this observational study. While multiple randomised control trials (RCTs) were performed to guide the choice of open or ER of AAA, there are no similar data on how to treat PAA. The need of RCTs in the future to answer the question of preferred treatment strategy for PAAs is highlighted. International collaboration in this field is needed, given the relatively low frequency of these procedures. Although there is a lack of randomised data, the recently published ESVS Guidelines on ALI issued a strong recommendation against using ER for PAAs with ALI (Class III, Level B). 5 When analysing time trends, stable amputation and mortality rates were found after elective surgery. The amputation rate increased during 2017e2018 after ALI, however, and simultaneously the proportion of ALI cases decreased. Few previous studies report the results separately for PAAs that present with ALI, but in those that do, the amputation rate varies between 5% and 28%. 3,18,29 In the present study, the amputation rate was 3.6% in 2012 and 7.2% in 2018, so the reported results still compare favourably. This increased amputation risk after treatment for ALI remains a matter of concern. It should be emphasised, however, that the registries only report treated patients, and those undergoing primary amputation are not reported. Thus, a possible explanation for the increased proportion being amputated could be that patients are being treated more aggressively, even those with the most severe ischaemia that previously would have been treated by primary amputation. However, data are lacking on patients who underwent primary amputation without a prior revascularisation attempt.

Limitations
A potential limitation with all registry studies is the risk of selection bias due to insufficient external validity. Many of the included registries have been validated (Table S3), 35e39 but a specific validation of PAA treatment has rarely taken place.The Swedish registry used the fact that many patients have bilateral PAA and found that among 146 bilateral procedures, 141 (96.6 %) had reported the contralateral operation to the registry. 2 Comorbidities are defined slightly differently in the various registries, affecting the internal validity.The fact that patients from 14 different countries were studied results in inhomogeneity, but also that results may be relevant for patients worldwide. The fact that a small proportion of patients (30%, although it was 85.0% among the eight countries that reported one year follow up) had one year follow up makes it difficult to assess medium term outcome, and outcome beyond one year is unknown. Since the registries only report patients operated on, it was not possible to investigate at what threshold diameter a PAA should be repaired.

Conclusions
This report on definitive repair of PAA from 14 countries sheds light on a great variability and the lack of consensus recommendations to guide treatment of PAA. OR dominates in both the elective and emergency scenarios, and the results of this study support this strategy. There is a great need for future RCTs and consensus recommendations.