Predictors of Long-term Patency after Femoropopliteal Angioplasty: Results from the STAR Registry

Predictors of Long-term Patency after Femoropopliteal Angioplasty: Results from the STAR Registry Timothy W.I. Clark, MD, Jeffrey L. Groffsky, MD, and Michael C. Soulen, MD

PURPOSE: To identify variables predictive of long-term patency after femoropopliteal angioplasty. MATERIALS AND METHODS: The primary patency of 219 limbs in 205 patients from a multicenter registry who underwent femoropopliteal angioplasty between January 1, 1992, and December 31, 1994, was prospectively monitored with a combination of angiography, noninvasive hemodynamic testing, and clinical outcome. Patient demographic, angiographic, and hemodynamic variables were examined alone and in combination to determine effect on long-term primary patency. Each limb was graded as Category 1– 4 according to the American Heart Association (AHA) criteria for arterial lesions, and differences in outcome for each category were examined. Primary patency and intergroup analysis were determined with use of the Kaplan-Meier method and log-rank test, respectively. Cox proportional hazards models were used to calculate relative risks for predictive variables. RESULTS: Primary patency rates for all limbs (on an intent-to-treat basis) at 12, 24, and 36 months were 87% ⴞ 3%, 80% ⴞ 3%, and 69% ⴞ 5%, respectively. Primary patency at 48 and 60 months was 55% ⴞ 7%. Poor tibial runoff (single tibial vessel with 50%–99% stenosis or occlusion) was most predictive of occlusion (relative risk 8.5, P < .0001). The presence of diabetes or renal failure was associated with lower long-term patency (relative risk 5.5 and 4.0, P < .0001 and .0002, respectively). Long-term patency was higher with AHA Category 1 lesions (P ⴝ .006), and no significant difference in patency was observed between Category 2 and 3 lesions (P ⴝ .65). A multivariate Cox proportional hazards model showed only the stratified runoff score and the presence of diabetes to be significant determinants of long-term patency. CONCLUSION: Poor tibial runoff is most predictive of lower long-term patency rates. Diabetes is also independently associated with lower long-term patency rates. The criteria that distinguish Category 2 and 3 lesions do not predict differences in long-term patency, nor do they serve to identify lesions best treated with surgical bypass. This suggests that indications for femoral angioplasty can be extended to include longer and more complex Category 3 lesions. Index terms:

Angioplasty

•

Arteries, femoropopliteal

J Vasc Interv Radiol 2001; 12:923–933 Abbreviations: ABI ⫽ ankle-brachial index, AHA ⫽ American Heart Association, ISCVS ⫽ International Society of Cardiovascular Surgery, PTA ⫽ percutaneous transluminal angioplasty, STAR ⫽ SCVIR Transluminal Angioplasty and Revascularization (Registry), SVS ⫽ Society of Vascular Surgery

PERCUTANEOUS transluminal angioplasty (PTA) has become an established technique in the treatment of From the Department of Radiology, Hospital of the University of Pennsylvania, 3400 Spruce St., Philadelphia, PA 19104. Received January 9, 2001; revision requested February 14; revision received April 3; accepted April 4. The STAR Registry is supported by the Cardiovascular & Interventional Radiology Research and Education Foundation (CIRREF) through grants from Abbott Laboratories, MediTech/Boston Scientific Corporation, and Nycomed. Address correspondence to T.W.I.C.; E-mail: [email protected] © SCVIR, 2001

femoropopliteal disease, with a role that continues to be defined as more studies of long-term outcome become available. We report the long-term primary patency of a cohort of patients from the SCVIR Transluminal Angioplasty and Revascularization (STAR) Registry who underwent angioplasty for femoropopliteal disease. Clinical and angiographic variables were examined alone and in combination to identify those predictive of long-term success. The angiographic categories of femoropopliteal disease adopted by the American Heart Association

(AHA) and SCVIR to determine the utility of these criteria in selecting lesions suited to angioplasty (1) were also examined. The protocol of the STAR Registry enabled uniform definitions, data recording, and follow-up assessment.

PATIENTS AND METHODS Study Protocol The STAR Registry is a multicenter registry of patients undergoing conventional balloon angioplasty and

923

924

•

August 2001

Long-term Patency after Femoropopliteal Angioplasty

other percutaneous interventions for the treatment of lower extremity ischemia. Seven institutions participated in the STAR Registry, ranging from private practices in community hospitals to large academic interventional practices (Alexandria Hospital, Beth Israel Hospital, Brigham and Women’s Hospital, Crozer-Chester Hospital, Hospital of the University of Pennsylvania, Reading Hospital, Thomas Jefferson Hospital, and Western Pennsylvania Hospital). Participating sites encouraged all patients undergoing percutaneous peripheral arterial interventions between January 1, 1992 to December 31, 1994 to enroll in the Registry. Patients signed an Institutional Review Board–approved consent form for release of information to the registry. No absolute exclusion criteria existed other than refusal of the patient to participate. Follow-up of patients in the STAR Registry with prospective collection of data was performed for 5 years and concluded December 31, 1999. The study was designed so that, during the first year after percutaneous intervention, patients were evaluated at 3, 6, and 12 months. Thereafter, patients returned for yearly evaluations. Study Population The initial study group consisted of 397 limbs in 383 patients who underwent femoropopliteal angioplasty during the study period. Patients were excluded from the study group because of concurrent angioplasty above the inguinal ligament (n ⫽ 8 limbs) or below the popliteal artery (n ⫽ 53 limbs), and concurrent femoropopliteal atherectomy (n ⫽ 9 limbs). One hundred eight limbs underwent femoropopliteal angioplasty but had no follow-up recorded in the registry at the time of data collection for this study. The resultant study cohort consisted of 219 limbs in 205 patients. Angioplasty Procedure Angiographic and hemodynamic parameters recorded for each limb are shown in Table 2. Lesions were categorized as stenoses, occlusions, or both. The degree of stenosis was measured by mounting the angiogram on a conventional viewbox and measuring vessel diameter on each angio-

gram with use of a graduated magnifying lens calibrated in 0.3-mm increments. In 11 patients undergoing thrombolysis, lesion length and degree of stenosis were measured based on the appearance of the lesion after thrombolysis but before angioplasty. Lesions were designated as eccentric or concentric and as calcified or noncalcified. Immediate postangioplasty films were assessed for an angiographically discernible intimal flap. Intimal flaps were classified as either confined to or extending beyond the site of angioplasty. The status of the tibial vessels was scored for each limb on the basis of the preprocedural angiogram. Each tibial vessel was assigned a score of 0 to 2 according to the extent of disease (0 ⫽ less than 50% stenosis; 1 ⫽ 50%–99% stenosis; 2 ⫽ occluded), the sum of which formed the total runoff score (0 – 6). Lesions were categorized as involving the superficial femoral artery origin, the distal below-knee popliteal artery, or the intervening femoropopliteal segment, according to the AHA categorization of arterial lesions (1). Because this system was introduced after the STAR Registry had begun, lesions were classified retrospectively according to AHA criteria. Angiograms were initially evaluated by the operators performing the procedure and lesion characteristics were entered into a comprehensive registry data form. Thereafter, these data were entered into a computer spreadsheet by dedicated research assistants at the core lab at Thomas Jefferson University Hospital. Patient angiograms remain in storage at the core lab, and for the purposes of this study, were reevaluated by two authors (T.W.I.C. and J.L.G.). This included classification of AHA criteria and lesion morphology. Follow-up Follow-up clinical assessment included grading of symptoms according to the Society of Vascular Surgery (SVS)/International Society of Cardiovascular Surgery (ISVCS)/SCVIR classification, examination of peripheral pulses and skin changes, noninvasive hemodynamic evaluation (ankle-brachial index [ABI], segmental pressures, pulse volume recordings, tread-

JVIR

mill exercise, and/or color Doppler examination). Follow-up angiography was performed for worsening of symptoms by one clinical category or a drop in ABI of more than 0.15. Study Endpoints and Definitions Success of angioplasty was defined by clinical change (improvement by at least one clinical category, or by at least two categories when tissue loss was present, according to the classification adopted by the SVS/ISCVS/ SCVIR) (2), noninvasive studies (increase in ABI by at least 0.10, or pulsevolume resistance increase of 5 mm Hg in noncompressible vessels), or angiography (residual stenosis no greater than 30%). For the purposes of this study, to be considered patent during follow-up, a treated limb had to meet clinical criteria (no interval percutaneous or surgical intervention involving the femoropopliteal artery and no deterioration in clinical category), noninvasive criteria (ABI maintained within 0.15 of maximum post-PTA), and, when performed, angiographic criteria (no evidence of restenosis exceeding 50% at the site(s) of PTA). Limbs that violated any of these criteria were considered occluded for the purposes of analysis. Only primary patency of limbs was examined. Clinical variables recorded for each patient (Table 1) included age, sex, history of smoking, and comorbidities (hypertension, diabetes mellitus, renal failure, and coronary artery disease). Patient symptoms were graded according to SVS/ISCVS/SCVIR category (2). Mild claudication (Category 1) was present in 2.4% of patients (n ⫽ 5), moderate claudication (Category 2) in 19.5% (n ⫽ 40), and severe claudication (Category 3) in 35.6% (n ⫽ 73). Ischemic rest pain (Category 4) was present in 11.7% of patients (n ⫽ 24) and minor tissue loss (Category 5) in 25.4% (n ⫽ 52). No patients had major tissue loss (Category 6). The SVS/ ISCVS/SCVIR category was not specified in 5.4% of patients (n ⫽ 11). Complications Complications were classified as Minor or Major according to clinical outcome in accordance with the

Volume 12

Number 8

Clark et al

Table 1 Characteristics of 205 Patients Undergoing Femoropopliteal Angioplasty

•

925

Table 2 Angiographic/Hemodynamic Characteristics of 219 Limbs Undergoing Femoropopliteal Angioplasty

Characteristic

n

%

Age (y) Mean Median Range Sex Male Female Diabetes mellitus Hypertension Renal failure Smoking status Smokers Nonsmokers Coronary Artery Disease Clinical Category* 1: Mild claudication 2: Moderate claudication 3: Severe claudication 4: Ischemic rest pain 5: Minor tissue loss 6: Major tissue loss Not specified Aspirin use post-PTA Warfarin therapy post-PTA

n 67.2 68 33–89

%

116 89 89 114 23

56.6 43.4 43.4 55.6 11.2

68 137 100

33.2 66.8 48.8

5 40 73 24 52 0 11 102 18

2.4 19.5 35.6 11.7 25.4 0 5.4 49.8 8.8

* From Reference 2.

SCVIR Standards of Practice Committee (3). Statistical Analysis Clinical risk factors and angiographic/hemodynamic parameters were assessed as variables predictive of outcome (primary patency or occlusion). An exploratory analysis with descriptive statistics was performed for each group of variables, with use of ␹2 analysis for nominal variables and the Student t test for continuous variables. When appropriate, dummy variables were introduced to stratify continuous and categoric variables (such as age, lesion length, or runoff score) into subgroups. After this series of analyses, variables identified as suitable to model fitting and testing the global null hypothesis (lack of effect on primary patency) were examined with univariate Cox proportional hazards models. Relative risks (hazards ratios) were calculated from Cox models to express the magnitude of effect of predictive variables on long-term patency. Variables found to have a significant effect on

Characteristic

N

%

Stenoses Occlusions Both N/A Mean stenosis length (cm) Mean occlusion length (cm) Location SFA origin Distal popliteal Neither N/A Number of sites dilated 1 2 ⱖ3 Eccentric lesions Calcified lesions Degree of stenosis* pre-angioplasty post-angioplasty Intimal Flap Local Extended Runoff Score† 0 1 2 3 4 5 6 ABI* Pre-angioplasty Post-angioplasty Mean ABI increase

172 24 13 10 3.8 4.7

78.5 11.0 5.9 4.6

11 5 182 21

5.0 2.3 83.1 9.6

113 63 43 28 70

51.6 28.8 19.6 12.8 32.0

84 58 26

85.9 ⫾ 13.8 15.3 ⫾ 14.1 38.4 26.5 11.9

70 14 47 12 39 20 17

32.0 6.4 21.5 5.5 17.8 9.1 7.7

0.56 ⫾ 0.17 0.87 ⫾ 0.18 0.31 ⫾ 0.18

* Mean ⫾ SD. † Each tibial vessel is assigned a score (0 ⫽ less than 50% stenosis; 1 ⫽ 50%–99% stenosis; 2 ⫽ occluded), the sum of which is the total runoff score (0 – 6).

long-term patency were later combined in a multivariate Cox model. Long-term vessel patency (time until occlusion) and intergroup analyses were determined with the Kaplan-Meier method and the log-rank test, respectively. SAS 6.1 was used for performing descriptive statistics and Cox models, and StatView was used for Kaplan Meier analysis and log rank tests (both programs from SAS, Cary, NC). For all analyses, a P value less than .05 was considered significant.

RESULTS Immediate Results As shown in Table 2, the majority of lesions treated were stenoses. The

mean stenosis length was less than the mean occlusion length. The AHA categories of femoropopliteal disease distinguish between involvement of the superficial femoral artery ostium, the distal below-knee popliteal artery, and the intervening segment (1). The distribution of angioplasty sites is shown in Table 2. The intervening femoropopliteal vessel was the site of angioplasty in 83.1% of cases (n ⫽ 182). Femoropopliteal angioplasty was performed at one site of the vessel in more than half of cases. Most lesions were concentric and noncalcified. After angioplasty, an intimal flap was visible on immediate postangioplasty arteriograms in 38.4% of limbs (n ⫽ 84). The flap was confined to the site of

926

•

August 2001

Long-term Patency after Femoropopliteal Angioplasty

Table 3 Clinical Improvement After Successful Percutaneous Femoropoliteal Angioplasty (n ⴝ 208) Grade

Description

n

%

⫹3

Markedly improved or symptom free; ABI increased to more than 0.90 Moderately improved: still symptomatic but at least single category improvement; ABI increases by more than 0.10 but not normalized Minimally improved: greater than 0.10 increase in ABI but no categorical improvement No change: no categorical shift and less than 0.10 change in ABI Mildly worse: no categorical shift but ABI decreased more than 0.10, or downward categorical shift with ABI decreased less than 0.10 Moderately worse: one category worse or unexpected minor amputation Markedly worse: more than one category worse or unexpected major amputation Not specified

86

41.3

89

42.8

9

4.3

18

8.7

2

1.0

0

0

0

0

4

1.9

⫹2

⫹1 0 ⫺1

⫺2 ⫺3

Note.—Adapted from Reference 2. * Initial technical failures (n ⫽ 11) excluded.

angioplasty in 26.5% (n ⫽ 58) and extended beyond the site of angioplasty in 11.9% (n ⫽ 26). Angiographic stenosis decreased from a mean of 85.9% ⫾ 13.8% to 15.3% ⫾ 14.1% after angioplasty. ABI increased from a preangioplasty mean of 0.56 to a postangioplasty mean of 0.87 (P ⬍ .001, two-sample t-test). Clinical outcome after angioplasty, excluding 11 immediate technical failures, is shown in Table 3, according to the AHA criteria (1). Marked or moderate improvement in symptoms (Grades 2–3) was observed in 84.1% (n ⫽ 175) of limbs, and mild improvement (Grade 1) was observed in 4.3% (n ⫽ 9). Angioplasty had no clinical benefit (Grade 0) in 8.7% (n ⫽ 18) and worsened symptoms (Grade ⫺1) in 1.0% (n ⫽ 2). Complications Complications occurred in 16 of 219 (7.3%) cases. Minor complications comprised 12 of 16 complications. These included five puncture site hematomas (none of which required transfusion or surgery), one contained vessel rupture at the site of angioplasty managed with local pressure, and one vagal reaction. Five patients

had distal emboli after angioplasty; three were treated with urokinase and two patients required suction embolectomy. None required additional treatment. Major complications occurred in four of 219 cases (1.8%). These included one patient with a puncture site infection that required intravenous antibiotics, and one patient who developed an occlusive dissection and was treated with surgical bypass. Two patients had transient renal failure. No procedure-related deaths occurred and there were no unexpected amputations. Long-term Patency and Predictors of Outcome Clinical risk factors (Table 1) and angiographic/hemodynamic parameters (Table 2) were assessed as variables predictive of patency. The significant variables are summarized with relative risks from Cox proportional analyses in Table 4. Among clinical risk factors, the presence of diabetes mellitus was associated with a significantly lower patency rate (P ⬍ .0001; relative risk 5.5). Renal failure was also a significant risk factor for loss of patency (P ⫽ .0002; relative risk 4.0).

JVIR

Age was stratified into multiple subgroups to determine a threshold age above which patients could be expected to have a lower patency rate; however, no significant threshold was observed (results not shown). Limbs with more than one site of angioplasty had a relative risk of primary patency loss of 2.6 compared to limbs with a single site of angioplasty (P ⫽ .004; Table 4). The presence of an intimal flap was not predictive of outcome (P ⫽ .53), regardless of whether the flap was localized to or extended beyond the site of angioplasty (data not shown). Lesion length was examined as a predictor for long-term patency; lesions exceeding 3 cm in length had a relative risk of patency loss of 2.0 compared to lesions 3 cm or smaller (P ⫽ .022). When this threshold was increased to 10 cm, limbs with lesions greater than 10 cm had a relative risk of 2.8 compared to lesions 10 cm or smaller (P ⫽ .019). The status of tibial runoff was analyzed with respect to long-term patency. No significant relationship to long-term patency was observed when each of the seven possible discrete runoff scores were analyzed (results not shown). However, when the runoff score was stratified, a significant effect on patency was observed. From Cox univariate analysis, limbs with a runoff score of 5– 6 (corresponding to a single tibial vessel with 50%–99% stenosis or occlusion of all three runoff vessels) had a relative risk of patency loss of 8.5 compared to limbs with a runoff score of 0 – 4 (P ⫽ .0001). No other morphologic features had a significant effect on patency. No difference in long-term patency was observed between patients who underwent thrombolysis and those who did not undergo thrombolysis (P ⫽ .14, log-rank test). Each significant predictor variable in Table 4 was combined in a multivariate Cox proportional hazards model (Table 5). The remaining significant variables in the final model were the stratified runoff score (relative risk 5.8, P ⫽ .0001) and diabetes (relative risk 3.5, P ⫽ .0009). Survival analysis with the KaplanMeier method was used to determine long-term patency. Primary patency for all limbs (on an intent-to-treat basis) at 12, 24, and 36 months was 87% ⫾ 3%, 80% ⫾ 3%, and 69 ⫾ 5%,

Volume 12

Number 8

Clark et al

Table 4 Significant Risk Factors for Femoropopliteal Occlusion from Univariate Cox Proportional Hazards Models after Percutaneous Angioplasty Risk Factor

␤

SE

␹2

P Value

Relative Risk

95% CI

Runoff Score* Diabetes Renal failure AHA category† Lesion length ⬎3 cm Lesion length ⬎10 cm Sites of PTA‡

2.14 1.71 1.39 1.29 0.691 1.03 0.934

0.306 0.361 0.372 0.374 0.303 0.441 0.324

48.6 22.1 13.9 11.9 5.21 5.50 8.30

.0001 .0001 .0002 .0006 .022 .019 .004

8.5 5.5 4.0 3.6 2.0 2.8 2.6

4.65–15.4 2.72–11.2 1.93–8.26 1.74–7.55 1.10–3.61 1.19–6.68 1.35–4.81

* Stratified into two groups with combined runoff scores of 5– 6 vs 0 – 4. Each runoff vessel was assigned a score of 0 –2 according to the following: 0 ⫽ no disease to ⱕ50% stenosis; 1 ⫽ 50 –99% stenosis; 2 ⫽ occlusion. † AHA category 2/3 vs AHA Category 1 lesions. For description of AHA categories, see Table 6. ‡ More than one site of femoropopliteal PTA vs one site of femoropopliteal PTA

Table 5 Final Cox Proportional Hazards Model for Femoropopliteal Occlusion after Percutaneous Angioplasty Risk Factor

␤

SE

␹2

P Value

Relative Risk

95% CI

Runoff score* Diabetes

1.76 1.25

0.321 0.378

30.0 11.0

.0001 .0009

5.8 3.5

3.09–10.9 1.67–7.34

* Stratified into two groups with combined runoff scores of 5– 6 vs 0 – 4. Each runoff vessel was assigned a score of 0 –2 according to the following: 0 ⫽ no disease to ⱕ50% stenosis; 1 ⫽ 50%–99% stenosis; 2 ⫽ occlusion.

respectively. Primary patency for all limbs at 5 years was 55% ⫾ 7%, including initial failures (Fig 1). The effect of the significant clinical and angiographic variables on long-term patency after successful angioplasty (excluding initial failures) was compared with the log-rank test, with KaplanMeier curves calculated for diabetes, renal failure, and stratified runoff score (Figs 2– 4). Patients with diabetes had a 36-month primary patency rate of 56% ⫾ 8%, compared to 77% ⫾ 10% for patients without diabetes (P ⬍ .0001). Patients with renal failure had a primary patency rate at 6 months of 51% ⫾ 13%, whereas patients without renal failure had a 6-month patency rate of 91% ⫾ 2% (P ⬍ .0001). After 6 months, only three of 21 patients with renal failure had patent limbs. Patients with at least one patent runoff vessel (runoff score 0 – 4) had a 36-month patency rate of 87% compared to 30% of patients with a single stenotic runoff

vessel or occlusion of all three tibial arteries (runoff score 5– 6, P ⬍ .0001). The AHA criteria for femoropopliteal lesions (Table 6) were not predictive of outcome when combined with other variables in logistic regression analysis (results not shown). However, differences in patency rates exist between categories as shown by Kaplan-Meier analysis (Fig 5), excluding technical failures (n ⫽ 11). Of the 96 limbs with Category 1 lesions, 87% ⫾ 6% were patent at 36 months. No significant difference in patency rates was observed between limbs with Categories 2 (n ⫽ 87) and 3 (n ⫽ 21) lesions (69% ⫾ 7% and 66% ⫾ 12%, respectively). Only three patients with Category 4 lesions underwent angioplasty.

DISCUSSION A cycle of change has occurred in the endovascular treatment of atherosclerotic disease in the femoropopli-

•

927

teal artery since the widespread application of balloon angioplasty in the 1970s and early 1980s. Since that time, mechanical atherectomy and laser atherectomy each enjoyed brief eras of popularity (4 –7). Retrospective case series describing femoropopliteal stents have been encouraging (8 –10), but several randomized, prospective trials have shown no benefit compared to conventional angioplasty (11–13). More recently, stent-grafts (14 –16), vascular brachytherapy (17), and photodynamic therapy (18) have been described for femoropopliteal disease; whether these latter interventions will confer a long-term benefit in patency is unclear. Until then, conventional balloon angioplasty remains the standard of care for endovascular treatment of the femoropopliteal artery. Advantages and Disadvantages of Registry Data We report the experience of the STAR Registry with percutaneous balloon femoropopliteal angioplasty in 219 limbs in 205 patients. A registry affords several advantages. Registries reflect a continuum of current clinical practice. The protocol of the STAR Registry was prospectively designed to use uniform definitions, data recording, and follow-up assessment. The participation of multiple institutions within the registry reduces the effect of referral bias patterns that exist at individual institutions. Multiple operators perform the procedures, avoiding the influence of minor variations in technique. Moreover, a large number of cases could be enrolled in a relatively short period of time. The large number of cases provides a sufficient volume of data for subgroup analysis of many variables, and the relatively narrow accrual window avoids confounding changes in technology or patterns of care. Unlike rigorously controlled clinical trials with narrow inclusion criteria, the allinclusive nature of the registry allows the outcome of angioplasty to be examined for the influence of many clinical, anatomic, and technical variables under “real-life” conditions of practice. Registries have important limitations. Most important is the enormous potential for selection bias during enrollment of patients. Patients deemed

928

•

Long-term Patency after Femoropopliteal Angioplasty

August 2001

JVIR

oropopliteal, or tibial interventions if the remaining segments were patent. Patency Results of STAR Registry

Figure 1. Kaplan-Meier analysis of primary patency; all limbs (intent to treat). KaplanMeier analysis for the entire group of limbs (n ⫽ 219), including initial failures (n ⫽ 11). Numbers shown are the patients at risk at a given time. SE is ⬍10% to 60 months. Primary patency rates at 12, 24, and 36 months are 87%, 80%, and 69%, respectively, and 55% at 60 months.

particularly suited to angioplasty were treated and entered in the registry, whereas those not believed to be candidates for angioplasty were not enrolled. Therefore, the “worst case” examples of the various parameters (for example, lesion length) could have been selectively excluded from analysis. A major concern with this study is the absence of follow-up data in 27% of the initially enrolled patients who underwent angioplasty, compared to an expected attrition rate of 10%–20% for a study of this size. Participating institutions in which interventional radiologists perform their own follow-up or that have dedicated clinical or research coordinators tended to have more complete and timely submission of follow-up data to the registry, compared to community practices in which patients were followed by the referring physicians and follow-up data were retrieved intermittently by participating radiologists. Because the registry is ongoing, not all information had been entered into the data base at the time data were extracted for this study. It is uncertain what the pending

follow-up on this subset of patients might have on the analysis. Another limitation of the registry design was the assumption of independent events; patients who underwent angioplasty of both femoropopliteal arteries, whether during the same or separate settings, introduced potential confounding of what were considered statistically independent events. This confounding is expected to be minor, because each limb was subject to the same clinical predictors (eg, diabetes) but independent anatomic predictors (eg, lesion morphology). The number of patients who underwent treatment of both legs is small (n ⫽ 14; 6.8%) and it is unlikely that the results of the study would be discernibly altered by excluding these patients from the analysis. We excluded patients who underwent concurrent iliac or infrapopliteal PTA; the majority of patients in this study underwent clinical and noninvasive assessment during follow-up. If a limb was no longer considered patent according to the STAR Registry criteria, it would be difficult to distinguish whether this loss of patency should be attributed to the iliac, fem-

The primary patency rates for all limbs as determined by Kaplan-Meier analysis at 1, 2, and 3 years were 87% ⫾ 3%, 80% ⫾ 3%, and 69% ⫾ 5%, respectively. Patency remained stable at 55% ⫾ 7% at 4 and 5 years. This included 11 technical failures (5.0%), which were usually results of inability to cross the lesion. These failures occurred before the widespread use of hydrophilic guide wires and low-profile coated balloon catheters, and therefore may not be representative of current interventional practice. Our patency rates are higher than those reported by Johnston (19), who calculated a patency rate of 36% at 6 years from the University of Toronto series of 254 femoral and popliteal angioplasty procedures. Capek et al (20) reported a 5-year patency rate of 58% among 217 femoropopliteal angioplasty procedures but excluded technical failures. Matsi and colleagues (21) reported a 3-year patency rate of 42% (including initial failures) in 140 limbs treated for a combination of femoropopliteal stenoses and occlusions (Table 7). The reasons for the higher patency rates in our study are likely multifactorial. Although individual operators were not expected to adhere to a strictly uniform angioplasty technique, the registry enrolled patients as technology improvements became available, including hydrophilic guide wires, low-profile noncompliant balloons and smaller introducer systems. Analysis of survival-time data with the Kaplan-Meier technique estimates survival (in this case, limb patency) based on the number of events (failures or censoring) during the period of observation. Patients who were lost to follow-up were censored from the analysis. Ideally, one would expect that the 27% of limbs in which follow-up was not available would have experienced similar rates of failure (loss of primary patency) as the remaining patients in the study. However, if patients did not return for follow-up secondary to a higher rate of failure, this would introduce confounding of the survival estimates and overestimation of patency rates.

Volume 12

Number 8

Clark et al

Figure 2. Effect of diabetes: primary patency according to the presence (䡩) or absence (e) of diabetes. Initial failures (n ⫽ 11) are excluded for the purposes of subgroup analysis. Numbers shown are the patients at risk at a given time. The dashed line indicates an SE exceeding 10%. Patients with diabetes had a 36-month primary patency rate of 56% compared to 77% for patients without diabetes (P ⬍ .0001).

Predictors of Outcome Multiple clinical risk factors were examined as predictors of long-term patency, including age, sex, diabetes, hypertension, renal failure, smoking history, and coronary artery disease. Our data indicate that diabetes is a significant and independent predictor of long-term success; a diabetic patient is more than five times more likely than a nondiabetic patient to have limb occlusion (Table 4) after femoropopliteal angioplasty (P ⫽ .0001). It is also noteworthy that 10 of 11 initial failures occurred in patients with diabetes. Hewes and colleagues (22) reported a 4-year patency rate of 39% in patients with diabetes after femoropopliteal angioplasty, compared to 79% in patients without diabetes. Capek et al (20) also observed a better outcome in patients without diabetes in their analysis of 217 femoropopliteal angioplasty procedures. Matsi and coworkers (21) reported no difference in late outcome between 27 patients with diabetes and 73 patients without diabetes after femoropopliteal angioplasty, although a true difference may have been obscured by more pre-

dictive variables within their multivariate model. Stokes et al (23) treated 40 femoral and 16 popliteal stenoses with angioplasty in patients with diabetes, 71% of whom had gangrene. They observed no difference in patency compared to patients without diabetes when angioplasty was performed for claudication in patients with adequate runoff. The association of diabetes with tibioperoneal disease is well known. In our study, the effect of diabetes remained predictive of longterm outcome when interactions with runoff score were excluded in multivariate analysis (Table 5). Patients with renal failure were four times more likely to have occlusion after angioplasty than patients without renal failure (P ⫽ .0002). This association was significant only when univariate analysis was performed; renal failure was subsequently excluded from the final regression model because of the small magnitude of its effect relative to the runoff score and presence of diabetes. The most predictive angiographic characteristic in our series was tibial runoff score. Limbs with a runoff score

•

929

of 5 or 6 (single tibial vessel with 50%– 99% stenosis or occlusion) had a more than eight-fold greater likelihood of occlusion compared to limbs with a runoff score of 0 to 4 (indicating the presence of at least one patent runoff vessel; Table 3). Runoff score was also the most predictive variable in the study by Johnston (19), who found a 30% 5-year patency rate in limbs with 0 or 1 tibial vessel and a 52% patency rate in limbs with 2–3 tibial vessels. Similar associations of patency with adequate runoff have been reported by others (21,23–25). Gordon et al (26) observed no relation of long-term patency to runoff in angioplasty of 42 superficial femoral artery occlusions but excluded patients without one or more intact tibial arteries. We also found no significant differences in patency among patients with at least one patent tibial artery. The number of sites dilated had a weak correlation with long-term patency; patients who underwent femoropopliteal angioplasty at two or more sites were 2.6 times more likely to have occlusion than those treated at one site (P ⫽ .004). This effect was not significant in multivariate modeling. Age was stratified at 5-year increments from 55 to 75 years to determine if a particular threshold existed above which angioplasty was less successful. With use of univariate models, no effect of age on long-term patency was observed (data not shown). We found that the type of lesion (stenosis or occlusion) did not affect long-term outcome in our study. Krepel and colleagues (24) found no difference in long-term patency between stenoses and occlusions when lesions were less than 3 cm in length. Murray et al (27) reported no difference in long-term outcome when they excluded patients with stenoses more than 7 cm in length from their analysis. Immediate and early success does appear to relate to the type of lesion. Matsi et al (21) reported a lower technical success rate with occlusions compared to stenoses (80% vs 99%; P ⬍ .01). Johnston (19) reported an 81% success rate at 1 month for occlusions compared to 94% for stenoses. Additional variables found not to be predictive of long-term success included sex, aspirin or warfarin use, coexisting hypertension, coronary artery disease, or smoking status. Our

930

•

Long-term Patency after Femoropopliteal Angioplasty

Figure 3. Effect of renal failure: primary patency according to the presence (䡩) or absence (e) of renal failure. Initial failures (n ⫽ 11) are excluded. Numbers shown are the patients at risk at a given time. The dashed line indicates a SE exceeding 10%. Primary patency at 6 months is 51% among patients with renal failure, compared to 91% among patients without renal failure (P ⬍ .0001).

Figure 4. Effect of runoff score: primary patency according to tibial runoff score. Each runoff vessel was assigned a score according to the following: 0, no disease to ⱕ50% stenosis; 1, 50%–99% stenosis; and 2, occlusion. Limbs were stratified into those with runoff scores of 0 – 4 (䡩) and 5– 6 (e). Initial failures (n ⫽ 11) are excluded. Numbers shown are the patients at risk at a given time. SE is ⬍10% to 50 months. Primary patency at 36 months is 87% among limbs with a score of 0 – 4, compared to 30% among limbs with a score of 5– 6 (single stenotic runoff vessel or occlusion of all three tibial arteries) (P ⬍ .0001).

August 2001

JVIR

cohort of patients was not limited to patients with claudication. Ischemic rest pain was present in 12% (n ⫽ 24) of patients and minor tissue loss was present in 25% (n ⫽ 52). The SVS/ ISCVS/SCVIR classification used to grade the clinical severity of patients did not predict long-term patency, although no patients were in Category 6 (major tissue loss extending above transmetatarsal level). The lack of patients with the most severe category of ischemia is not surprising because this analysis was limited to patients undergoing femoropopliteal intervention only; most patients with major tissue loss have multilevel disease. Angioplasty produces a fracture of the intimal–medial border of the vessel, and an intimal flap visible on postangioplasty arteriograms reflects the expected pathophysiology of controlled injury to the vessel. However, an entire era of mechanical and laserassisted atherectomy was based partly on the assumption that a postangioplasty dissection flap was undesirable. Capek et al (20) reported a 2.6% incidence of dissections as “complications” and a 39.2% incidence of “cracking” while acknowledging these would be not considered untoward in present-day angiography. Nevertheless, we were interested in determining if the presence of a dissection flap, arbitrarily designated as limited to or extending beyond the site of angioplasty, was predictive of long-term success. No effect was observed (P ⫽ .53). However, one limb with an occlusive intimal flap was treated with surgical bypass. As a continuous variable, lesion length in our study was predictive of outcome when stratified at various thresholds. Patients with lesions more than 3 cm in length had a relative risk of occlusion of 2.0 compared to those with lesions 3 cm or smaller (P ⫽ .022), and those with lesions greater than 10 cm in length had a relative risk ratio of 2.8 compared to those with lesions 10 cm or smaller (P ⫽ .019). Murray et al (27) reported a 23% patency rate at 2 years for long-segment stenoses compared to 82% for short stenoses, with 7 cm used as a threshold. Capek et al (20) reported a significantly better long-term outcome with lesions less than 2 cm in length (P ⫽ .001). More recently, Murray and colleagues (28) achieved a 93% technical success rate

Volume 12

Number 8

Clark et al

Table 6 American Heart Association Lesion Categories of Femoropopliteal Lesions Category

Description

n

%

1

Single stenosis ⱕ5 cm or single occlusion ⱕ3 cm that does not involve SFA origin or distal popliteal Single stenosis 5–10 cm or single occlusion 3–10 cm not involving distal popliteal; heavily calcified stenoses up to 5 cm; multiple lesions (stenoses or occlusions) each less than 3 cm; lesions with no continuous tibial runoff Single occlusions 3–10 cm involving distal popliteal; multiple lesions each 3–5 cm; single lesions (stenoses or occlusions) ⬎10 cm Complete SFA occlusions; complete popliteal and proximal trifurcation occlusions; severe diffuse disease with no intervening segments Not specified

96

46.1

87

41.8

21

10.1

2

3 4

3 1

14 0.5

Note.—Initial failures (n ⫽ 11) are excluded, which consisted of two Category 1 lesions, eight Category 2 lesions, and one Category 3 lesion. Adapted from Reference 1.

and primary patency rate of 69% at 18 months among 42 patients with longsegment (10 – 40 cm) stenoses. American Heart Association Categories of Femoropopliteal Disease When the type (stenosis or occlusion) and length of lesion were combined into the angiographic categories for femoropopliteal lesions developed by the AHA Task Force (Table 6), differences in long-term outcome were observed. This classification is used to identify lesions suited to angioplasty. By definition, lesions in Categories 1 and 2 can be expected to have a high clinical success rate with complete relief or significant improvement in symptoms after angioplasty. Category 3 lesions (lesions ⬎ 10 cm, 3–10-cm occlusions involving the popliteal artery, or multiple 3–5-cm lesions) are expected to have a significantly lower chance of technical success and longterm benefit compared to surgical bypass. Angioplasty of Category 3 lesions is generally reserved for those patients who are poor surgical candidates. Finally, Category 4 lesions are those with extensive vascular disease for which angioplasty is seldom feasible or appropriate. In our cohort, the majority of patients who underwent successful angioplasty (88%) had Category 1 or 2

lesions. Category 1 lesions were associated with a significantly higher patency rate (87% ⫾ 6%; P ⫽ .006). Twenty-one patients (10%) underwent angioplasty of Category 3 limbs. No significant difference in patency at 36 months was found between lesions in Category 2 and 3 (69% ⫾ 7% vs 66% ⫾ 12%, respectively; P ⫽ .65). This is an important observation, because it suggests that the indications for femoropopliteal angioplasty could be broadened to include the more severe Category 3 lesions, which generally require surgical bypass. Conversely, the lack of difference between Category 2 and 3 lesions could reflect a type II error from insufficient sample size. No conclusions can be drawn from Category 4 lesions because only three such lesions were treated. Bypass Surgery or Angioplasty? Reported 5-year patency rates after bypass surgery for femoropopliteal disease are 60%–70% (performed for claudication) and 21%– 68% (performed for limb salvage) (29). However, few studies have compared angioplasty with surgery. In their early experience with angioplasty, Martin et al (30) compared the patency rate at 2 years between 46 femoropopliteal occlusions treated with angioplasty compared to 133 treated with surgery at the same institution. The surgical

•

931

group had a 42% patency rate compared to a 57% rate in the angioplasty group (30). In the only multicenter, randomized trial comparing angioplasty with surgery, Wolf et al (31) reported a 60% 4-year patency rate among 38 patients with claudication who underwent femoropopliteal angioplasty, compared to a 58% patency rate among 35 patients treated with femoropopliteal bypass. A better outcome was observed among patients with femoropopliteal occlusions who were treated with angioplasty compared to those treated with surgery (76% vs 65% patency at 3 y) (31). Hunink et al (32) used a decision analysis model with reported success rates of surgery and angioplasty to determine optimal treatment in a hypothetical cohort of men with femoropopliteal disease. In patients with femoropopliteal stenosis and claudication or chronic critical ischemia, angioplasty was superior to surgery in increasing quality-adjusted life expectancy and decreasing lifetime expenditures. Surgery was the preferred initial treatment when chronic critical ischemia was caused by an occlusion. However, the authors also noted that angioplasty would always be preferable to surgery when 5-year patency rates of at least 30% could be achieved (32). The STAR Registry results indicate that, with current practice, a 5-year patency rate of better than 50% can be expected.

CONCLUSION Our study indicates that high longterm primary patency can be achieved with femoropopliteal angioplasty. Poor runoff status (single tibial vessel with 50%–99% stenosis or occlusion) was the single most predictive variable of lower long-term patency. We employed a simple and practical method of quantifying tibial runoff. Recent revised reporting standards (33) involve a more complex combination of scores to account for runoff vessels that are occluded for less than half the vessel length, and those with smaller degrees of stenosis (⬍20%). Publication of this new system occurred after the design and data-acquisition phase of our study. Had we used such a system for grading tibial runoff, it is possible more detailed conclusions could have been drawn

932

•

Long-term Patency after Femoropopliteal Angioplasty

August 2001

JVIR

come after angioplasty. The criteria that distinguish Categories 2 and 3 are not reflected in significant differences in long-term patency. Rather than identify lesions best treated with surgical bypass, Category 3 lesions have patency rates similar to or exceeding those reported for surgical bypass (29). Our data suggest that extending the range of lesions amenable to angioplasty in everyday clinical practice to include limbs with more extensive femoropopliteal disease may be warranted if at least one adequate tibial vessel is present.

Figure 5. Kaplan-Meier analysis of primary patency according to AHA Lesion Category: primary patency for limbs in AHA Category 1 (䡩), Category 2 (▫) and Category 3 (). The three limbs that underwent angioplasty of Category 4 lesions are excluded. Initial failures (n ⫽ 11) are excluded for the purposes of subgroup analysis. Numbers shown are the patients at risk at a given time. The dashed line indicates a SE exceeding 10%. No significant difference in patency rates was found between Category 2 and Category 3 lesions (P ⫽ .65). Table 7 Long-term Patency of Femoropopliteal Angioplasty Investigator (Ref. No.)

Year

N

Patency (%)

Follow-up (y)

Initial Technical Failures (%)

Gallino (34) Krepel (23) Hewes (21) Cambria (35) Wilson (15) Capek (19) Johnston (18) Matsi (20)

1984 1985 1986 1987 1989 1991 1992 1994

251 164 137 48 48 217 254 140

67 70 61 33 76 58 38 42

5 5 3 3 3 5 5 3

Excluded (10) Excluded (16) Excluded (14) Included (4) Excluded (16) Excluded (10) Included (11) Included*

* 9% stenoses, 17% occlusions.

for the prediction of long-term patency. As isolated variables, lesion type (stenosis or occlusion) and morphology did not affect outcome. The reasons for this are likely multifactorial. Only 14% of lesions were classified as eccentric, suggesting that some eccentric lesions may have been excluded during enrollment. Calcified lesions are well known to represent more difficulty in achieving optimal balloon dilation; however, no difference in long-term patency was observed after

angioplasty. The presence of an intimal flap after angioplasty was not predictive of primary patency. However, patients who underwent concurrent atherectomy were excluded from the study. Atherectomy was and still is used for certain highly eccentric and/or calcified lesions, and in situations in which a flow-limiting intimal flap is encountered after angioplasty. This may account for the lack of effect of these variables on patency. AHA Category 1 lesions were associated with an excellent long-term out-

Acknowledgments: The authors wish to acknowledge the principal investigators and participating sites of the STAR Registry: Arina van Breda, MD, Alexandria Hospital, Alexandria, VA; Ducksoo Kim, MD, Beth Israel Hospital, Boston, MA; Michael Meyerovitz, MD, Brigham and Women’s Hospital, Boston, MA; Joseph Stock, MD, Crozer-Chester Hospital, Chester, PA; Michael Soulen, MD, Hospital of the University of Pennsylvania, Philadelphia, PA; David Sacks, MD, Reading Hospital, Reading, PA; Gordon McLean, MD, West Pennsylvania Hospital, Pittsburgh, PA; and Geoffrey Gardiner, Jr, MD, Thomas Jefferson Hospital, Philadelphia, PA. References 1. Pentecost M, Criqui M, Dorros G, et al. Guidelines for peripheral percutaneous transluminal angioplasty of the abdominal aorta and lower extremity vessels. Circulation 1994; 89:511–531. 2. Rutherford R, Becker G. Standards for evaluating and reporting the results of surgical and percutaneous therapy for peripheral arterial disease. J Vasc Interv Radiol 1991; 2:169 –174. 3. Aruny JE, Lewis CA, Cardella JF, et al. Quality improvement guidelines for percutaneous management of the thrombosed or dysfunctional dialysis access. J Vasc Interv Radiol 1999; 10: 491– 498. 4. Vroegindeweij D, Tielbeek AV, Buth J, et al. Directional atherectomy versus balloon angioplasty in segmental femoropopliteal artery disease: two-year follow-up with color-flow duplex scanning. J Vasc Surg 1995; 21:255–268. 5. Tielbeek AV, Vroegindeweij D, Buth J, et al. Comparison of balloon angioplasty and Simpson atherectomy for lesions in the femoropopliteal artery: angiographic and clinical results of a prospective randomized trial. J Vasc Interv Radiol 1996; 7:837– 844. 6. Lammer J, Pilger E, Decrinis M, et al. Pulsed excimer laser versus continuous-wave Nd:YAG laser versus con-

Volume 12

7.

8.

9.

10.

11.

12.

13.

14.

15.

Number 8

ventional angioplasty of peripheral arterial occlusions: prospective, controlled, randomised trial. Lancet 1992; 340:1183–1188. Odink HF, de Valois HC, Eikelboom BC. Femoropopliteal artery recanalization: factors affecting clinical outcome of conventional and laser-assisted balloon angioplasty. Cardiovasc Intervent Radiol 1995; 18:162–167. Sapoval MR, Long AL, Raynaud AC, et al. Femoropopliteal stent placement: long-term results. Radiology 1992; 184: 833– 839. Henry M, Amor M, Ethevenot G, et al. Palmaz stent placement in iliac and femoropopliteal arteries: primary and secondary patency in 310 patients with 2– 4-year follow-up. Radiology 1995; 197:167–174. Bergeron P, Pinot JJ, Poyen V, et al. Long-term results with the Palmaz stent in the superficial femoral artery. J Endovasc Surg 1995; 2:161–167. Vroegindeweij D, Vos LD, Tielbeek AV, et al. Balloon angioplasty combined with primary stenting versus balloon angioplasty alone in femoropopliteal obstructions: a comparative randomized study. Cardiovasc Intervent Radiol 1997; 20:420 – 425. Zdanowski Z, Albrechtsson U, Lundin A, et al. Percutaneous transluminal angioplasty with or without stenting for femoropopliteal occlusions? A randomized controlled study. Int Angiol 1999; 18:251–255. Cejna M, Thurnher S, Illiasch H et al. PTA versus Palmaz stent placement in femoropopliteal artery obstructions: a multicenter prospective randomized study. J Vasc Interv Radiol 2001; 12:23– 31. Maynar M, Reyes R, Ferral H, et al. Cragg Endopro System I: early experience. I. Femoral arteries. J Vasc Interv Radiol 1997; 8:203–207. Burger T, Meyer F, Tautenhahn J, et al. Initial expeiences with percutaneous endovascular repair of popliteal artery

Clark et al

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

lesions using a new PTFE stent-graft. J Endovasc Surg 1998; 5:365–372. Lammer J, Dake MD, Bleyn J, et al. Peripheral arterial obstruction: prospective study of treatment with a transluminally placed self-expanding stent-graft. International Trial Study Group. Radiology 2000; 217:95–104. Minar E, Pokrajac B, Ahmadi R, et al. Brachytherapy for prophylaxis of restenosis after long-segment femoropopliteal angioplasty: pilot study. Radiology 1998; 208:173–179. Jenkins MP, Buonaccorsi GA, Raphael M, et al. Clinical study of adjuvant photodynamic therapy to reduce restenosis following femoral angioplasty. Br J Surg 1999; 86:1258 –1263. Johnston K. Femoral and popliteal arteries: reanalysis of results of balloon angioplasty. Radiology 1992; 183:767– 771. Capek P, McLean G, Berkowitz H. Femoropopliteal angioplasty: factors influencing long-term success. Circulation 1991; 83:I70 –I80. Matsi P, Manninen H, Vanninen R, et al. Femoropopliteal angioplasty in patients with claudication: primary and secondary patency in 140 limbs with 1–3 years follow-up. Radiology 1994; 191:727–733. Hewes R, White R Jr, Murray R, et al. Long-term results of superficial femoral artery angioplasty. Am J Roegentenol 1986; 146:1025–1029. Stokes K, Strunk H, Campbell D, et al. Five-year results of iliac and femoropopliteal angioplasty in diabetic patients. Radiology 1990; 174:977–982. Krepel V, van Andel G, van Erp W, et al. Percutaneous transluminal angioplasty of the femoropopliteal artery: initial and long-term results. Radiology 1985; 156:325–328. Matsi P, Manninen H, Suhonen M, et al. Chronic critical lower-limb ischemia: prospective trial of angioplasty with 1–36 months follow-up. Radiology 1993; 188:381–387.

•

933

26. Gordon I, Conroy R, Tobis J, et al. Determinants of patency after percutaneous angioplasty and atherectomy of occluded superficial femoral arteries. Am J Surg 1994; 168:115–119. 27. Murray R, Hewes R, White RJ, et al. Long-segment femoropopliteal stenoses: is angioplasty a boon or a bust? Radiology 1987; 162:473– 476. 28. Murray J, Apthorp L, Wilkins R. Long segment (ⱕ10 cm) femoropopliteal angioplasty: improved technical success and long-term patency. Radiology 1995; 195:158 –162. 29. Weitz J, Byrne J, Clagett G, et al. Diagnosis and treatment of chronic arterial insufficiency of the lower extremities: a critical review. Circulation 1996; 94:3026 –3049. 30. Martin E, Fankuchen E, Karlson K, et al. Angioplasty for femoral artery occlusion: comparison with surgery. Am J Roegentenol 1981; 137:915–919. 31. Wolf G, Wilson S, Cross A, et al. Surgery or balloon angioplasty for peripheral vascular disease: a randomized clinical trial. J Vasc Interv Radiol 1993; 4:639 – 648. 32. Hunink M, Wong J, Donaldson M, et al. Revascularization for femoropopliteal disease: a decision and cost-effectiveness analysis. JAMA 1995; 274:165– 171. 33. Rutherford RB, Baker J, Ernst C, et al. Recommended standards for reports dealing with lower extremity ischemia: Revised version. J Vasc Surg 1997; 26: 517–538. 34. Gallino A, Mahler F, Probst P, Nachbur B. Percutaneous transluminal angioplasty of the arteries of the lower limbs: a 5 year follow-up. Circulation 1984; 70:619 – 623. 35. Cambria R, Faust G, Gusberg R, Tilson M, Zucker K, Modlin I. Percutaneous angioplasty for peripheral arterial occlusive disease. Arch Surg 1987; 122: 283–287.