Risks and benefits of femoropopliteal percutaneous balloon angioplasty

Risks and benefits of femoropopliteal percutaneous balloon angioplasty Maria G. M. Hunink , MD, PhD, Magruder C. Donaldson, MD, Michael F. Meyerovitz, MD, Joseph F. Polak, MD, An thony D. Whi t temore , MD, Krishna Kandarpa, MD, PhD, Clement J. Grassi, MD, Jobm Artmy, MD, Donald P. Har r ing ton , MD, and John A. Mannick, MD, Boston, Mass., and Stonybrook, N.Y.

Purpose: The purpose of this study was to evaluate the efficacy of angioplasty in the treatment of femoropopliteal arterial disease. Methods: From 1980 to 1991, 126 angioplasty procedures were performed in 131 limbs of 106 patients with 175" femoropopliteal lesions (26 common femoral, 118 superficial femoral, and 31 popliteal). Critical ischemia was present in 55 limbs (42%), and clandication was present in 76 (58%). Angioplasty was performed for a single lesion in 87 limbs (66%) and for multiple lesions in 44 (34%). In 13 limbs (10%) the most severe lesion was an occlusion; in 118 (90%) all lesions were stenoses. Distal runoffwas good (2 or 3 vessels patent) in 72 limbs (55%) and poor (0 or 1 vessel patent) in 59 (45%). Results: Death within 30 days occurred in 0.8%, nonfatal systemic morbidity in 7.1%, and local morbidity in 1.6% of procedures. Multivariate analysis revealed that indication and age were predictive of increased morbidity and mortality rates. Immediate success was achieved in 95% of limbs treated. Mean follow-up time was 2.0 years. The overall 5-year cumulative primary patency rate was 45% (+5%). In a proportional hazards model indication and lesion type were predictive (p < 0.01) of long-term failure, with relative risks of 2.0 (1.2 to 3.3) and 2.7 (1.3 to 5.6), respectively. The 5-year primary patency rate after angioplasty for stenoses and claudication was 55% (-+ 7%), for stenoses and critical ischemia it was 29% ( ± 11%), and for occlusions it was 36% ( + 14%). Conclusion: These results suggest that femoropopliteal angioplasty is a low-risk procedure with acceptable long-term results in patients with claudication and stenoses. (J VAsc SURG 1993;17:183-94.)

Since the introduction of percutaneous translu- minal angioplasty (PTA) in the 1970s for the treatment of peripheral arterial disease, numerous papers have reported the results of this method of revascularization. However, most reports have fo- cused on iliac lesions, and the published experience with femoropopliteal angioplasty is limited? -3 The only published randomized controlled trial compar- ing angioplasty and bypass surgery for the treatment

From the Division of Cardiovascular and Interventional Radiol- ogy, Brigham and Women's Hospital, Harvard Medical School (Drs. Hunink, Meyerovitz, Polak, Kandarpa, Grassi, and Aruny), the Division of Vascular Surgery, Brigham and Women's Hospital, Harvard Medical School (Drs. Donaldson, Whittemore, and Mannick), and the Department of Radiology, State University of New York at Stonybrook (Dr. Harrington).

Presented at the Forty-sixth Annual Meeting of the Society for Vascular Surgery, Chicago, Ill., ~tme 8-9, 1992.

Reprint requests: Maria G. M. Hunink, MD, PhD, Department of Radiology, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115.

24]6/41755

of peripheral arterial disease showed that patients with femoropopliteal artery stenoses or occlusions successfully treated with angioplasty have long-term results similar to those treated with bypass surgery, which included prosthetic and venous bypasses. 4 Johnston s showed by multivariate analysis that the type of lesion and runoff were variables predictive of long-term success after femoropopliteal angioplasty. Hasson et al. 6 demonstrated that the indication for the intervention was the most significant variable predicting procedural morbidity and mortality rates. Largely because of the heterogeneity of patient populations in published studies and the variability of reporting methods, 7,8 the risks and long-term efficacy of femoropopliteal PTA remain incompletely de- fined. Because of the significant potential impact of PTA on morbidity and cost containment, further clarification of the efficacy of angioplasty in the treatment of femoropopliteal arterial disease is likely to be of considerable importance. 9-12

0741-5214/93/$1.00 + .10. 183

184 Hunink et al.

Journal of VASCULAR

SURGERY

METHODS Patients and procedures

All patients referred by a vascular surgeon to vascular radiology for femoropopliteal angioplasty between January 1, 1980, and October 31, 1991, were included in this study. Data were collected prospectively in a computerized registry mainly for the purposes of clinical follow-up. Patients were selected to undergo angioplasty after assessment by clinical examination, noninvasive testing and angiog- raphy, and after consultation between an interven- tional radiologist and vascular surgeon. Although no prospective selection criteria were used, angioplasty was usually performed for focal disease of less than 5 cm in patients with limb-threatening (critical) ischemia considered relatively high risk for systemic complications of surgery and in patients with dis- abling claudication. Variables recorded at the time of the procedure were age, sex, diabetes mellitus requir- ing treatment with insulin or oral hypoglycemic drugs, current smoking, treated or untreated hyper- tension, coronary artery disease (including coronary artery bypass grafting, history of myocardial infarc- tion, or angiographically documented coronary ar- tery disease), indication for the procedure (critical ischemia, including gangrene, ulcers, and rest pain and claudication implying disabling claudication), site of the lesion (common femoral, superficial femoral, or popliteal), lesion type (occlusion or stenosis), distal runoff (poor [0 or 1 vessel patent] versus good [2 or 3 vessels patent]), and the occurrence of procedural compfications.

All angioplasty procedures were performed by or under supervision of staff angiographers. In most cases consultation between the interventional radiol- ogist and vascular surgeon took place immediately after diagnostic angiography had been completed so that angioplasty could be performed at the same appointment. After written informed consent was obtained, the patient was given premedication of oral aspirin (325 mg) and in more recent years nifedipine (10 mg). After vascular access was obtained, a bolus of 5000 units of heparin was administered intrave- nously. The lesion was angiographically documented and then crossed with a straight or J-tipped guide wire, a soft-tipped torquable guide wire, or, more recently, with a Glidewire (Meditech, Boston, Mass.), in most cases with the help of road mapping. The size of the angioplasty balloon was chosen to be equal to the diameter of a normal artery segment adjacent to the diseased segment on a cut-film angiogram without correction for magnification to allow slight overdilation. Inflation was performed for

45 to 60 seconds at a time for one to three inflations while pressure was kept at 6 to 12 atm. Ifvasospasm was observed or anticipated, nitroglycerine (in incre- ments of 100 ~g) or intraarterial tolazoline (30 mg) was administered with blood pressure monitoring. A repeat angiogram was performed, and the angio- plasty was repeated with a balloon larger by 0.5 to 1 mm if a residual stenosis of 50% or greater was present. The sheath or catheter was removed within 2 to 3 hours after administration of the heparin bolus. After satisfactory control of the puncture site, intra- venous heparin was usually restarted and continued for 12 to 24 hours, with adjustment in the dose to maintain a partial thromboplastin time of 1.5 to 2.0 times the patient's normal value. During the latter part of the study we advised patients to take aspirin daily for 2 to 3 months.

The patient's clinical condition and the ankle/bra- chial indexes (ABIs) were documented I day after the procedure. ABIs were not determined if the pressure measurements indicated heavily calcified arteries. A technically successful PTA yielding an angiographi- cally documented lesion of less than 50% and an increase in ABI of more than 0.10 (as long as a meaningful ABI measurement was available) was considered an initial success. 8 The 0.10 value was chosen because it is approximately the noninvasive laboratory error. Follow-up visits to the referring vascular surgeon were scheduled at 1, 3, and 6 months and annually thereafter to assess symptoms and limb status. Patients were evaluated by nonin- vasive laboratory and clinical examination at each follow-up visit. A treated limb was judged patent during follow-up if the ABIs remained improved to within 0.10 of the immediate postprocedure values and there was at least moderate improvement in the patient's symptoms compared with the situation before PTA. 8 Failure of a single lesion after multiple dilations was considered a treatment failure of the entire limb. A computerized registry was used to record all relevant data. The institutional review board approved analysis of the recorded data.

Statistical analysis For this analysis 30-day morbidity and mortality

rate calculations were based on the number of procedures performed, and patency calculations were based on the number of limbs treated. Systemic morbidity was defined as nonfatal cardiac, pulmonary, cerebrovascular, septic, allergic, or renal complica- tions. Local morbidity was defined as nonfatal he- matoma, embolus, pseudoaneurysm, local infection, thrombosis, or dissection extending beyond the

Volume 17 Number 1 January 1993 Risks and benefits of femoropopliteal angioplasty 185

Table I. Complication rates (morbidity plus mortality) by univariate analysis and odds ratios based on multivariate logistic regression for presence versus absence of covariates predictive of complications

Complications (%)

Cova~ate Covariate Odds ratio Covariate present absent (95% CI) *

Age >65 yr 17.5t 2 2.9 6.0 (1.2-30.3)$ 2 Male 6.7 12.1 1.2 (0.3-5.1) Critical ischemia 20.4t 1 1.4 16.0 (2.0-131.2)#; i Occlusion 16.7 8.8 3.1 (0.4-21.8) Poor runoff 17.2t 3 2.9 2.4 (0.4-12.5) Common femoral lesion 9.1 9.6 0.9 (0.2-5.0) PoplJteal lesion 7.7 9.7 0.4 (0.1-2.1) Multiple lesions 6.8 11.0 0.6 (0.1-2.7) Repeat/contralateral PTA 3.9 11.0 1.3 (0.1-14.5) Before Jan. 1, 1986 11.9 8.3 1.3 (0.3-5.0) Smoking 6.3 15.2 1.0 (0.3-4,2) Diabetes mellitus 13.1 6.2 1.9 (0.5-7.2) Coronary artery disease 11.1 7.9 1.3 (0.3-4.8) Hypertension 12.5 4.4 3.4 (0.6-18.2)

*Odds ratio is estimate of relative risk associated with presence as compared with absence of covariate, adjusted for critical ischemia and age; 95% CI given in parentheses. tStatistically significant risk factor in univariate analysis (Pearson chi-square test lp < 0.001, 2p = 0.01, and 3p = 0.01). ¢Statistically significant risk factor in multivariate analysis (Wald test ~p = 0.01 and 2p = 0.03).

angioplasty site. The proportion of immediately successful angioplasty procedures was calculated per limb treated. Immediate failures included lesions that could not be crossed with a guide wire or balloon, lesions after PTA of 50% or greater, or an increase in ABI of 0.10 or less. s Univariate analysis of the association between covariates and complications (morbidity and mortality) was performed with use of the Pearson chi-squared test and Fisher's exact test where appropriate) 3 Multivariate logistic regression analysis was performed to predict the risk of a procedural complication and initial success. 13'~4 The covariates analyzed were age ( > 65 versus - 65 years), sex, indication (critical ischemia versus clan- dication), lesion type (occlusion versus stenosis), distal runoff (poor [0 or i vessel patent] versus good [2 or 3 vessels patent]), common femoral lesion, popliteal lesion, number of lesions treated (multiple versus single lesions), repeat versus primary proce- dure, prior or concurrent contralateral PTA, date the procedure was performed (before or after January 1, 1986, which is the halfway point of the entire time period), and the presence or absence of smoking, diabetes meliitus, coronary artery disease, and hyper- tension (Table I). A stepwise selection of the covariates was performed, including a covariate in the model if statistical significance at the 0.05 level was demonstrated/4 Odds ratios for each significant covariate and their 95% confidence intervals (CI), as well as the predicted morbidity for combinations of

the significant covariates, were calculated with this model. Odds ratios for the nonsignificant covariates were obtained by including them one by one in the model with the significant covariates.

Primary patency was defined as maintenance of hemodynamic and symptomatic improvement after angioplasty without further intervention? Limbs subjected to a repeat angioplasty procedure after failure of the first angioplasty were considered new cases in the analysis. This assttmes that repeat PTA has an equivalent outcome compared with primary procedures, an assumption supported by several authors , i s ' i t In keeping with the definition used by most vascular surgeons we defined lost to follow-up as those cases that are alive and can therefore be expected to return for follow-up but have not been seen for 18 months or longer. We tested for differences in covariates among those lost and those still in follow-up using the Pearson chi-squared test.is Life-table and Kaplan-Meier analyses were per- formed for primary patency of all limbs subjected to angioplasty and for limb salvage for those treated for critical ischemia. 8'17'18 In addition, life-table analysis was performed for survival of all patients who underwent angioplasty. Standard error calculations were performed with the Greenwood formula. 17 Patency was compared in a univariate analysis with the Mantel-Haenszel log-rank test. 17'a8 Multivariate analysis was performed with the Cox proportional hazards model to analyze the effect of the covariates

186 Hunink et al.

Journal of VASCULAR

SURGERY

Table II. Five-year cumulative primary patency by univariate analysis and hazard rate ratios based on multivariate proportional hazards analysis for presence versus absence of covariates predictive of failure after angioplasty

5-year primary patency % (+ SE) * Hazard rate ratio

Covariate Covariate present Covariate absent (95% (2I)

Age >65 yr 54 (~9) 39 (+7) 0.7 (0.4-1.2) Male 47 (+_7) 41 (-+8) 0.8 (0,5-1.4) Critical ischemia 27 ( +- 10)§ 54 ( +- 7) 2.0 (1,2-3.3)11 Occlusion 36 (± 14)§ 46 (+-6) 2.7 (1.3-5.6)] I Poor runoff 27 (+-9)§ 55 (+_7) 1.6 (0,9-2.8) Common femoral lesion 73 ( +_ 11) 40 ( +_ 6) 0.5 (0.2-1.3) Popliteal lesion 47 ( ± 14) 43 ( ± 6) 0.7 (0.2-1.9) Multiple lesions 46 (+ 9) 44 (+_ 7) 1.0 (0.6-1.7) Repeat PTA 42 ( ± 16)$ 45 ( +- 6)$ 1.1 (0.5-2.5) Contralateral PTA 57 ( +- 10) 41 ( + 6) 0.9 (0.4-1.8) Before Jan. 1, 1986 54 (+-8) 39 (+7) 0.7 (0.4-1.2) Smoking 47 ( ± 6) 36 ( + 12) 1.1 (0.6-2.0) Diabetes mellitus 34 ( -~ 9) 52 ( +_ 7) 1.2 (0.7-1.9) Coronary artery disease 36 ( +_ 8) 54 ( ± 7) 1.2 (0.7-2.1) Hypertension 36 ( +_ 7) 59 ( ± 8) 1.4 (0.8-2.5)

*SE, Standard error. ?Hazard rate ratio (relative risk) for a covariate is the ratio of failures per person-year follow-up in patients with the covariate with patients without the covariate, controlling for other significant covariates. Ninety-five percent CI given in parentheses. ~Four-year patency. (Insufficient repeat PTAs existed to report a 5-year patency.) $Statistically significant risk factor in univariate analysis of the patency curves (Mantel-Haenszel log-rank testp = 0.01). IIStatisticaUy significant risk factor in multivariate analysis (Wald testp = 0.01).

compared

with control for confounding (Table II) and to predict patency. 17,~9 A stepwise selection of the covariates was performed, including a covariate in the final model if it demonstrated significance at the 0.05 level. Hazard rate ratios for each statistically signifi- cant covariate and their 95% CI, as well as predicted primary patency curves for combinations of the significant covariates, were computed with use of this model. Hazard rate ratios for the nonsignificant covariates were obtained by including them one by one in the model with the significant covariates.

A small proportion of the procedures were repeat PTAs or prior or concurrent contralateral PTAs or both., Theoretically this violates the assumption of independence of the observations in the statistical analyses. To examine whether this affects the results we (1) added covariates representing this interdepen- dency, (2) reanalyzed complications by patients instead of by procedures, selecting the procedure with complication if multiple procedures with and without complications had occurred, and (3) reana- lyzed patency results by patients instead of by limbs, randomly selecting a limb for patients with bilateral involvement.

RESULTS

From 1980 to 1991, 131 limbs with 175 femoropopliteal lesions were treated during 126

angioplasty procedures in 106 patients (Tables III and IV). In five cases both limbs were treated during the same procedure. Eight lesions in seven limbs had repeat angioplasty for recurrence after previous angioplasty. The indications for angioplasty were nonhealing ulcers or gangrene or both in 30 (23%) limbs, rest pain in 25 (19%), and disabling dandi- cation in 76 (58%). The presence of coronary artery disease, hypertension, or diabetes mellitus did not differ significantly according to indicanon. However, smoking occurred significantly (p < 0.001) more often among patients with claudication. Angioplasty was performed on 26 common femoral, 118 super- ficial femoral, and 31 popliteal artery lesions. Angio- plasty was performed in 87 (66%) limbs on a single lesion and in 44 (34%) limbs on multiple lesions. In 13 (10%) limbs the most severe lesion was an occlusion, and in 118 (90%) all lesions were stenoses. Good runoffto the distal vessels (2 or 3 vessels) was present in 72 (55%) limbs, and poor runoff was present (0 or 1 vessel) in 59 (45%). A statistically significant association between poor runoff and critical ischemia was demonstrated (Pearson chi- squared test p < 0.001, correlation coeffi- cient = 0.50). No other statistically significant asso- ciations among the covariates indication, lesion type, number of lesions, or runoff could be demonstrated (at[ correlation coefficients < 0.08).

Volume 17 Number 1 January 1993 Risks a n d benefits o f femoropopliteal angioplasty 187

Table I lL Characteristics of 106 patients treated with PTA, by indication

Patient characteristics Claudication Critical ischemia (n = 106) (n = 57) (n = 49)

Mean age 63.3 67.5 Male (%) 53 39 Coronary artery disease (%) 46 53 Smoking (%) 77 45 Hypertension (%) 60 67 Diabetes meUitus (%) 44 53

Table ]W. Characteristics of 131 limbs treated with PTA, by indication

Critical Limb characteristics Claudication ischemia

(n = 131) (n = 76) (n = 55)

Stenosis Good runoff, single lesion 35 9 Good runoff, multiple lesions 17 5 Poor runoff, single lesion 10 24 Poor runoff, multiple lesions 7 11

Occlusion Good runoff, single lesion 6 0 Good runoff, multiple lesions 0 0 Poor runoff, single lesion 1 2 Poor runoff, multiple lesions 0 4

Mean ABI before PTA + 0.61 0.43 Mean ABI after PTA + 0.85 0.80

*Excluding heavily calcified noncompressible arteries.

Treatment of six (5%) limbs met with immediate technical and/or clinical failure within 2 days; one patient underwent amputation, one patient was treated medically, and four underwent bypass sur- gery. Multivariate logistic regression analysis failed to demonstrate significant covariates in predicting im- mediate success. The mean ABI increased from 0.61 to 0.85 among patients with claudication and from 0.43 to 0.80 among those treated for critical ischemia (Table IV). Nonfatal systemic morbidity occurred in one (1.4%) procedure performed for claudication and in eight (15%) performed for critical ischemia (Table V). Two procedures performed for critical ischemia were each associated with two systemic complications. Nonfatal local morbidity, excluding technical failure, occurred in two (1.6%) procedures. One patient died after undergoing below-knee am- putation after a failed angioplasty performed for gangrene; the 30-day procedural mortality rate was therefore 0.8%. Univariate analysis demonstrated that the indication for the procedure (critical isch- emia versus claudication), age (>65 versus ___65 years), and runoff (poor versus good) were signifi-

Table V. Thirty-day nonfatal morbidity associated with PTA performed for claudication or critical ischemia

No. (%) @procedures with indicated complication

Critical Claudication ischemia

Complication (n = 72) (n = 54) +

Systemic/remote Myocardial infarction 1 (1.4%) 2 (3.7%) Arthythmia 0 3 (5.6%) Congestive heart failure 0 1 (1.9%) Pneumonia 0 1 (1.9%) Renal failure 0 3 (5.6%)

Local/vascular Hematoma 0 1 (1.9%) Embolus 0 1 (1.9%)

All morbidity 1 (1.4%) 10 (18.5%) +

*Two procedures were associated with two complications each.

cant risk factors in predicting complications (Table I). With control for confounding in multivariate logistic regression analysis, indication and age were the only significant risk factors, with odds ratios of 16.0 and 6.0, respectively. Whether the procedure was a repeat PTA or whether a prior or concurrent contralateral PTA was performed was not significant. Analyzing the results by patients instead of by procedures gave similar results. Comparing the observed and predicted complication rates for the significant risk factors demonstrated a good fit of the logistic model. The predicted procedural complica- tion rate for patients with claudication was 0.5% among patients not older than 65 years and 2.8% for those over 65 years, and for patients with critical ischemia it was 7.2% and 31.8%, respectively.

Mean follow-up time was 2.0 years (range, 1 day to 8.5 years). Twenty-one percent of limbs and 25% of patients were lost to long-term follow-up. Those treated in the first half of the study period were more likely to be lost than those treated in the second half (p < 0.001). No other significant differences were observed between those lost and those still in follow-up. Overall 5-year cumulative primary pa- tency was 45% (+-_ 5%) (Table VI, Fig. 1). Assuming that all limbs lost to follow-up were actually patent gave an upper estimate of 5-year cumulative primary patency of 51% ( -+ 5%), whereas assuming that all limbs lost to follow-up were actually failures gave a lower estimate of 42% (+_ 5%), both of which are well within the 95% CI of the estimated patency. Long-term patency varied depending on the presence of predictive covariates. In univariate analysis the

1 8 8 Hunink et al.

Journal of VASCULAR

SURGERY

1 131

' 0 . 8

E °n

i = .

£).

. ~ 0.4

_a -1 E - i 0 0.2

124

104

l=i= 31

I , , ,7

0 I I 1 I 0 12 24 36 48 60

months

Fig. 1. Cumulative primary patency curves for femoropopliteal angioplasty for all limbs subjected to angioplasty with number at risk for each interval. (Solid line indicates standard error of < 0.10 over entire follow-up period.)

1 64

~ ' 0 . 8 58 44

~ ' 0 . 6 22

E o. I 7

o.4~4 3

E 0 0.2

~ claudication & stenosis (n =as)

occlusion 2 ( n = 1 3 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ischemia & stenosis (n =48)

I 0 I I I 0 12 24 36 48 60

months

Fig. 2. Cumulative primary patency for femoropopliteal angioplasty for stenoses by indication and for all occlusions. (Solid lines indicate standard error of <0.10 and dotted lines >-0.10.)

Mantel-Haenszel log-rank test demonstrated that the indication (critical ischemia versus claudication), the lesion type (occlusion versus stenosis), and the distal runoff (poor versus good) were statistically signifi- cant (p = 0.01) covariates for failure. In a multivari-

ate (Cox proportional hazards) model the covariates shown to be significant were indication and the lesion type (p = 0.01) (Table II). Repeat, prior, or con- current contralateral PTA were not significant in predicting failure. Analyzing the results by patients

Volume 17 Number 1 January 1993 Risks and benefits of femoropopliteal angioplasty 1 8 9

O.8

E

~.~ 11.4

E 0 O.2

o

o - ~ - ~ O t3 0

, %o U L I

daudication & stenosis O 0 0 0 0 0 I~ [] []

ischemia & stenosis LJ

0 I I I I 0 12 24 36 48 60

months

Fig. 3. Cumulative primary Kaplan-Meier patency curves for stenoses by indication, observed (squares) and predicted with a Cox proportional hazards model (solid lines). Regression coefficient for critical ischemia was 0.7 ( + 0.3) and for occlusions 1.0 ( -+ 0.4). (For number at risk during each interval see Fig. 2.)

instead[ of by limbs gave similar results. The observed 5-year cumulative primary patency for clandication and stenoses was 55% (-+7%) and for critical ischemia and stenoses it was 29% ( -+ 11%) (Fig. 2). The corresponding predicted patencies (with the Cox model and Kaplan-Meier approach) are 58% and 34%, respectively (Fig. 3). The observed 5-year patency for occlusions was 36% (_+ 14%) (Fig. 2). The limited number of limbs (n = 13) with occlu- sions did not allow for meaningful life-table analysis stratified by indication and for modeling with the proportional hazards model.

No limbs treated for claudication were ampu- tated. Of the 55 limbs subjected to angioplasty for critical ischemia, four underwent below-knee ampu- tation. The amputations occurred 6, 12, 18, and 54 days ~fter the angioplasty procedure, resulting in a constant cumulative (Kaplan-Meier estimate) limb salvage rate of 91% ( -+ 5%) from 3 months to 5 years. The 5-year cumulative patient survival was 73% (-+7%) (Table VI), corresponding to an average annual mortality rate of 6%.

DISCUSSION

The purpose of this report is to help clarify the role of femoropopliteal angioplasty in the treatment of peripheral arterial disease. Because of the con- fusion created by the variability of reporting meth-

ods used in previous published reports on the subject, this report has tried to conform to the standards recently set for such communications. 7,8 These standards are consistent with those recom- mended for reporting results of surgical bypass, and it is hoped that future comparative analysis of both techniques will be greatly enhanced by such stan- dardization.

The first objective of this study was to identify covariates predictive of an increased risk of compli- cations and immediate failure associated with angio- plasty. A previous multivariate analysis demonstrated an odds ratio for critical ischemia of 2.5 (95% CI 1.4 to 4.7) in predicting morbidity and 4.7 (95% CI 1.2 to 17.9) in predicting 30-day mortality rates. 6 Other studies have found indication and age to be signifi- cant risk factors for complications, z°,21 The multivari- ate analysis presented here supports the hypothesis that the presence of critical ischemia and the patient's age significantly increase the risk of complications associated with angioplasty. The odds ratios for the presence of occlusion, hypertension, and diabetes mellitus indicate that these may be additional risk factors for complications; however, the sample size is too small to demonstrate statistical significance with control for indication and age. The initial success rate for femoropopliteal angioplasty in this series was 95%, comparable to the initial success rates of

190 Hunink et al.

Journal of VASCULAR

SURGERY

Table VI. Life-table analysis* of primary patency for 131 limbs and of survival for 106 patients subjected to femoropopliteal angioplasty

Censored Interval No. at Interval Cumulative Standard

Outcome (mo) risk Died Lost Withdrawn Failures patency/survival patency/survival error

Primary patency

Survival

0-0t 131 0-1 124 1-3 104 3 -6 86 6-12 68

12-24 51 24-36 31 36-48 22 48-60 17 60-

0-1 106 1-3 98 3 -6 84 6-12 73

12-24 60 24-36 39 36-48 27 48-60 20 60-

0 1 6 0.954 0.95 0.02 3 1 16 0.869 0.83 0.03 4 4 8 0.919 0.76 0.04 6 0 11 0.867 0.66 0.04 3 4 9 0.859 0.57 0.05 2 10 5 0.885 0.50 0.05 2 2 3 0.893 0.45 0.05 2 3 0 1.000 0.45 0.05 3 6 0 1.000 0.45 0.05 3 1 3 4 1 0.990 0.99 0.01 4 7 3 0.968 0.96 0.02 4 6 1 0.987 0.95 0.02 4 8 1 0.985 0.93 0.03 3 13 5 0.904 0.84 0.05 2 8 2 0.941 0.79 0.05 2 3 2 0.918 0.73 0.07 2 7 0 0.100 0.73 0.07 3 7

*No., Number at risk at the beginning of the interval. In calculating patency, limbs were censored if the patient died, the patient was lost to follow-up, or the limb was withdrawn patent. In calculating survival, patients were censored if the patient was lost to follow-up or withdrawn alive. tFirst interval (0-0) used to denote technical and immediate clinical failures. The one patient censored in this interval was lost to follow-up 1 day after the procedure.

Table VII. Review of results of femoropopliteal angioplasty from studies reporting primary patency with a life-table or Kaplan-Meier analysis, ordered by decreasing long-term patency

Patency (%) Ischemia Occlusion Poor Morbidity* Mortality*

Reference No. Age (%) (%) (%) (%) (%) Type Initial l yr 2yr 3yr 4yr 5yr _+SET

Walden et al. 22 23 62 65 71 71 16 1.4 Hem:~ 91 69 69 NA NA NA +10 Krepelet al.23 164 63 10 23 NA NA NA S/H:~ 88 71 68 62 62 62 -+4 Gallino et al. z4 289 68 39 41 41 8 0 Hem§ 87 62 61 60 58 58 +3 Wilsonet al. 4 49 61 27 NA NA 38 0 Hem 78 62 59 59 59 NA -+7 BWHI] 131 65 42 10 45 9 0.8 Hem 95 57 50 45 45 45 +5 Capeket al) 6 217 64 26 32 40 10 1.4 Hem§ 90 57 52 49 48 42 _+4 Samson et al.2s 89 70 90 0 NA 25 1.0 S/H 92 50 46 46 NA NA +7 Jeans et al. 26 190 64 49 66 37 NA 2.0 S/H§ 82 48 43 42 41 41 _+4 Henrick et al.27 31 58 0 42 NA 11 0 Hem§ 77 47 41 41 41 NA _+10 Johnston ~ 254 63 20 39 36 13 0.4 Hem 96 63 53 51 44 38 _+4 Jorgenson et al? 8 58 67 100 62 59 11 0 Hem~ 6 4 40 33 25 25 NA _+8

NA, Not available; No., number of procedures; Age, mean/median age; Poor, poor runoff; Hem, maintenance of hemodynamic impro;cement was required for success; S/H, maintenance of either symptomatic or of hemodynamic improvement were considered a S u c c e s s .

*Adjusted so that the denominator is the number of procedures. tStandard error of last patency available, either reported standard error or calculated with Greenwood formula lz from available data. ~zReported patency adjusted to include initial failures. ~Patency was estimated from a published figure specifically of femoi'opopliteal angioplasty, based on life-table or Kaplan-Meier analysis. HResults from the current series.

previous reports (Table VII). Multivariate analysis of initial success did not identify significant risk factors.

The second and main objective of this study was to determine the long-term patency of femo- ropopliteal angioplasty and identify subgroups with

good long-term results. Reported 4- to 5-year primary patency rates from articles with a life-table or Kaplan-Meier analysis range from 25% to 62% (Table VII). The variability in results is probably caused by differences in case mix and criteria for

Volume 17 Number Z[ January 1993 Risks and benefits of femoropopliteal angioplasty 191

success across studies. The best long-term results among these selected series were reported by Krepel et al.23 (62% 5-year primary patency). Note that their criteria for success were maintenance of either symptomatic or hemodynamic improvement, which is more lenient than requiring maintenance of hemodynamic improvement as used in most other series, including our own. Furthermore, most pa- tients in Krepel's series were treated for daudication and stenoses. The lowest long-term patency was reported by Jorgensen et al.28 (25% 4-year primary patency) who required hemodynamic improvement for success and reported results of patients treated for critical ischemia, most with occlusions. Selecting those series that use hemodynamic criteria for suc- cess, with 5-year follow-up, reporting primary pa- tency using life-table or Kaplan-Meier analysis, with a similar case-mix (mean age, 63 to 68 years; 20% to 42% Critical ischemia; 10% to 41% occlusions; 36% to 45% poor runoff) estimates of morbidity range from 8% to 13%, mortality rates range from 0% to 1.4%, and 5-year primary patency ranges from 38% to 5896 (Table VII). s'24'16 Overall 5-year primary patency in the current series was 45% ( _ 5%), with hemodynamic improvement as a criterion for success and a case mix consisting mainly ofstenoses. Our best long-term results were obtained with angioplasty of stenoses for claudication (observed 5-year primary patency of 55% [ +_ 7%]).

The multivariate proportional hazards analysis reported by Johnston s demonstrated that poor runoff and lesion type (occlusion versus stenosis) were significant in predicting long-term failure after fem- oropopliteal angioplasty. Multivariate proportional hazards analysis of the current series identified indication (critical ischemia versus claudication) and lesion type (occlusion versus stenosis) as the signifi- cant covariates in predicting long-term failure. Al- though poor runoff was a significant predictor of failure in the present study's univariate analysis, this was not the case in the multivariate analysis. How- ever, we demonstrated a statistically significant asso- ciation between poor runoff and critical ischemia, suggesting that the deleterious effect of poor runoff may be explained by its association with critical ischemia. This hypothesis also explains the discrep- ancy between Johnston's and our results.

Ewm though reporting methods for patency data have been standardized, methodologic issues that are subject: to controversy still exist. Probably the most pertinent example is choosing the unit of analysis: whether it should be patients, procedures, limbs, arterial segments, or lesions. We think the choice

should be driven by the clinical relevance. With respect to patency, a limb is the clinically relevant unit because it is "a limb" we try to save through our intervention. Furthermore, analyzing angioplasty data with limbs as unit of analysis is analogous to the methods of analysis used for surgical bypass data. With respect to complications, the number of pro- cedures is the clinically relevant trait, because every procedure exposes the patient anew to the risk of a complication, and it is this risk that interests us when we are deciding whether or not to perform the procedure. Because our choice of units of analyses introduces interdependency among a small propor- tion of the observations, we included in our multi- variate models covariates representing repeat PTA and prior or concurrent contralateral PTA. None of these covariates achieved significance in the predic- tive models. Furthermore, we determined the odds and hazard ratios forcing these covariates into the model and, finally, we repeated our analyses includ- ing only one limb per patient and only one procedure per patient. These alternative methods of analysis did not alter the results substantially and did not change the conclusions.

In conclusion, this study suggests that indication and age are significant predictors of morbidity associated with femoropopliteal angioplasty and that indication and lesion type are significant predictors for long-term failure. On the basis of this experience it appears appropriate to consider femoropopliteal angioplasty for focal stenoses associated with dis- abling claudication, in which case angioplasty is a low-risk procedure and has acceptable long-term results. The procedure should be used selectively in the elderly population with critical ischemia and more extensive disease.

We thank Julie Lombara for data entry and retrieval of information on angioplasty procedures performed at our hospital.

R EFER EN C ES

1. Weibull H, Bergqvist D, Jonsson K, Karlsson S, Takolander R. Complications after percutaneous transluminal angioplasty in the iliac, femoral, and popliteal arteries. J VASe SURG 1987;5:681-6.

2. Becker GJ, Katzen BT, Dake MD. Noncoronary angioplasty. Radiology 1989;170:921-40.

3. Adar R, Critchfield GC, Eddy DM. A confidence profile analysis of the results of femoropopliteal percutaneous trans- luminal angioplasty in the treatment of lower-extremity ischemia. J VASC SURG 1989;10:57-67.

4. Wilson SE, Wolf GL, Cross AP, et at. Percutaneous translu- minal angioplasty versus operation for peripheral arterioscle- rosis. Report of a prospective randomized trial in a selected group of patients. J VAsc SURG 1989;9:1-9.

192 Hunink et al.

Journal of VASCULAR

SURGERY

5. Johnston KW. Femoral and popliteal arteries: reanalysis of results of balloon angioplasty. Radiology 1992;183:767~71.

6. Hasson JE, Acher CW, Wojtowycz M, McDermott J, Crummy A, Turnipseed WD. Lower extremity percutaneous transluminal angioplasty: multifactorial analysis of morbidity and mortality. Surgery 1990;108:748-54.

7. Rutherford RB, Flanigan DP, Gupta SK, et al. Suggested standards for reports dealing with lower extremity ischemia. Prepared by the Ad Hoc Committee on Reporting Standards, Society for Vascular Surgery/North American Chapter, inter- national Society for Cardiovascular Surgery, j VAsc SUV, G 1986;4:80-94.

8. Rutherford RB, Becker GJ. Standards for evaluating and reporting the results of surgical and percutaneous therapy for peripheral arterial disease. Radiology 1991;181:277-81.

9. Doubilet P, Abrams HL. The cost of underutilization: percutaneous transluminal angioplasty for peripheral vascular disease. N Engl J Med 1984;310:95-102.

10. Tunis SR, Bass EB, Steinberg EP. The use of angioplasty, bypass surgery, and amputation in the management of peripheral vascular disease. N Engl J Med 1991;325:556-62.

11. Coffman JD. intermittent claudication-be conservative. N Engl J Med 1991;325:577-8.

12. Hunink MGM, Meyerovitz MF [Letter]. The use of angio- plasty, bypass surgery, and amputation in the management of peripheral vascular disease. N Engl J Med 1992;326:414.

13. Models for binary response variables. In: Agresti A. Categor- ical data analysis. New York: john Wiley & Sons, 1990:79- 129.

14. The LOGISTIC procedure. In: SAS Technical Report P-200. Release 6.04. Cary, North Carolina: SAS Institute, 1990: 175-230.

15. Johnston KW, Rae M, Hogg-Johnston SA" et al. 5-year results of a prospective study of percutaneous transluminal angio- plasty. Ann Surg 1987;206:403-413,

16. Capek P, McLean GK, Berkowitz HD. Femoropopliteal angioplasty: Factors influencing long-term success. Circula- tion 1991;83[suppl 1]:1-70-80.

17. Harris EK, Albert A. Survivorship analysis for clinical studies. New York: Marcel Dekker, 1991:1-123.

18. The LIFETEST procedure. In: SAS Technical Report P-179.

Release 6.03. Cary, North Carolina: SAS Institute, 1988:49- 89.

19. Cox proportional hazards model and survival kit. In: STATA reference manual. Release 2.05. Santa Monica, California, 1990:275-86 and 554-68.

20. Belli AM, Cumberland DC, Knox AM, Procter AE, Welsh CL. The complication rate ofpercutaneous peripheral balloon angioplasty. Clin Radiol 1990;41:380-3.

21. Morse MH, Jeans WD, Cole SEA, Grier D, Ndlovu D. Complications in percutaneous transluminal angioplasty: relationships with patient age. Br J Radiol 1991;64:5-9.

22. Walden R, Siegel Y, Rubinstein ZJ, Morag B, Bass A, Adar R. Percutaneous transluminal anagioplasty. A suggested method for analysis of clinical, arteriographic, and hemody- namic factors affecting the results of treatment. J Vasc SUV, G 1986;3:583-90.

23. Krepel VM, van Andel GJ, van Erp WFM, Breslau PJ. Percutaneous transluminal dilatation of the femoropopliteal artery: initial and long-term results. Radiology 1985;156: 325-8.

24. 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;70:619-23.

25. Samson RH, Sprayregen S, Veith FJ, Scher LA" Gupta SK, Ascer E. Management of angioplasty complications, unsuc- cessful procedures and early and late failures. Ann Surg 1984; 199:234-40.

26. Jeans WD, Armstrong S, Cote SEA, Horrocks M, Baird RN. Fate of patients undergoing transluminal angioplasty for lower-limb ischemia. Radiology 1990;177:559-64.

27. Henriksen LO, jorgensen B, Holstein PE, Tonnesen KH, Karle A" Sager P. Percutaneous transluminal angioplasty of infrarenal arteries in intermittent clandication. Acta Clair Scand 1988;154:573-6.

28. Jorgensen B, Henrlksen LO, Karle A" Sager P, Holstein PE, Tormesen KH. Percutaneous ttansluminal angioplasty of iliac and femoral arteries in severe lower-limb ischemia. Acta Clair Scand 1988;154:647-52.

Submitted June 12, 1992; accepted Aug. 12, 1992.

DISCUSSION

Dr. James M. Seeger. I congratulate the authors on a nice presentation and a nice study.

Use of transluminal balloon angioplasty to treat patients with symptomatic femoropopliteal arterial occlu- sive disease remains controversial. Arguments in favor of using angioplasty to treat such lesions include good initial results and low morbidity associated with the procedure. Factors against such therapy include a high rate of recurrent stenosis after angioplasty and the potential for accelerated occlusive disease.

The retrospective review presented here has carefially analyzed 106 patients undergoing femoropopliteal angio- plasty and reports results with use of the standards recommended by our societies. The results of this study

support the high initial success rate and low morbidity of balloon angioplasty for short lesions in the femoropoptiteal arteries previously reported by other authors. In addition, long-term success rates by life-table analysis appear to be acceptable, with 5-year patency at 45%. Multivariate analysis identified elderly patients with critical ischemia to be at highest risk for complications after the procedures, and patients with critical ischemia caused by occlusions to be those at highest risk for long-term failure. Thus on initial review, the conclusions of this study that femoropopliteal balloon angioplasty, particularly in younger patients with claudication caused by arterial stenosis, is a low-risk procedure with acceptable results appears to be justified.

However, before we can accept this conclusion, we

Volume 17 Number 1. January 1993 Risks and benefits of femoropopliteal angioplasty 193

need to consider some other factors. A 5-year cumulative success rate of approximately 50% is acceptable with a low-risk procedure. Unfortunately, in this study 21% of the patients were lost to follow-up, so the 5-year success rate, as the authors point out, may be as low as 42%, Other studies of femoropopliteal angi0plasty have reported con- tinued success at 5 years of between 25% and 60%, usually when on ly patients who had successful procedures were considered. Thus 5-year cumulative success rate 40% or less for super~cial femoral artery balloon angioplasty appears to be accurate.

Second, the natural history of femoropopliteal lesions treated with balloon angioplasty remains unclear. Hemo- dynamic and/or symptomatic failure in these patients may occur fi'om multiple problems including restenosis at the site of angioplasty, progression of the disease, or progres- sion oul:side at a distant site. Follow-up angiography would have been helpful to determine this, but this was not available in this study,

Finally, and probably most important, the best long- term results in this series and in other series examining balloon angioplasty have been achieved in patients with claudication caused by arterial stenosis. However, these patient~; generally have fairly benign courses. The question then remains as to whether balloon angioplasty of femo- ropopliteal lesions is better than medical therapy alone in these patients. In a recent small randomized trial of femoropopliteal balloon angioplasty or exercise therapy in patients with claudication, no difference was seen in walking distance at 15 months between the two groups. Indeed, patients treated with exercise therapy alone gen- erally could walk farther than patients treated with balloon angioplasty.

Regardless, because of the careful analysis of a relatively large number of patients undergoing femoropopliteal angiopiasty, this study is a valuable contribution to our understanding of the role of angioplasty in the treatment of patients with symptomatic femoropopliteal arterial occlu- sive di,;ease,

I have three questions for the authors: First, how were the patients included in this study selected for angioplasty? More important, which patients were excluded, and do you have any information on what happened to the excluded patients?

Second, what degree of hemodynamic improvement was observed? You note that patency was defined as an increase in ABIs of 0.1. What was the actual hemodynamic improvement in this group of patients? In addition, did hemodynamic and/or subjective improvement correlate with objective improvement in walking distance?

And finally, were there any patients with immediate or late failures after angioplasty who were symptomatically worse? Did any of these patients require surgical arterial reconstruction or amputation after angioplasty failure?

Dr. G. M. Maria Htmink. Thank you, Dr. Seeger, for your comments and questions. First, regarding the criteria for selection for angioplasty, we chose patients who had

focal disease of less than 5 cm, where the angioplasty was deemed technically feasible. Usually angioplasty was per- formed if a patient was considered at high risk for systemic complications from bypass surgery, if a patient had disabling claudication and would otherwise not have been considered for revascularization, and in those for whom bypass surgery was considered but a vein was unavailable. The patients who were excluded from angioplasty were the converse of these patients, and they generally underwent bypass surgery if intervention was deemed necessary.

To answer the second question regarding the actual improvement, we defined an improvement as being an 0.10 increase in ABI compared with the ABIs before angio- plasty, and in follow-up both hemodynamic and symptom- atic improvement were used as criteria for success. There- fore our patency results indicate improvement in both. The mean ABI increased from 0.61 to 0.85 among patients with claudication and from 0.43 to 0.80 among those treated for critical ischemia.

We did not actually measure claudicating or maximum walking distance. Therefore we have not yet determined the degree of symptomatic improvement. We think that measuring the change in ABIs, although an objective measurement, is actually only a proxy for what really counts, and that is, is the patient any better? Does the patient feel better? Ideally, I think we would want to determine this with a health status index and a quality of life measure before and after treatment, preferably with a randomized controlled trial comparing patients undergo- ing exercise and undergoing intervention.

That brings me to your comment about the random- ized controlled trial that compared angioplasty and exer- cise. I think you are referring to the study by Creasy et al. There are a number of points to be made with respect to this trial. First, it was a very small trial consisting of 36 randomized patients. Second, the ABI increase after 3 to 9 months was 0.21 with PTA and 0 with exercise. Third, it is true that with exercise the claudicating and maximum walking distance increased progressively, whereas in pa- tients with PTA this was limited at 12 months. However, the limitation was due to contralateral disease in all cases. Although exercise treats bilateral disease, PTA was per- formed unilaterally in this trial. Finally, it should be pointed out that significantly more patients undergoing the exercise protocol stopped smoking, which confounded the results.

In answer to your third question, in general our experience is that I fPTA fails, patients are rarely worse off compared with what may be expected on the basis of their natural history, and a failed PTA does not preclude or negatively influence bypass surgery. As far as our patients with failed PTA grafts are concerned, 12% were treated with repeat angioplasty, 72% went on to bypass surgery, 10% had no treatment, and 7% had an amputation.

Dr. John Porter (Portland, Ore.). I have one question. I am concerned about your reporting methods. I believe it is widely accepted that the important reporting variable is the number of patients, not the number of lesions. I f you

194 Hunink et al.

Journal of VASCULAR

SURGERY

reduce your number to 106 from 170 and accept failure of any single lesion as patient failure, does this have any disastrous effects I suspect it would on your life table for patency?

Dr. Hunink. It is interesting that you bring this up, because we did initially analyze our results with denomi- nator being the number of lesions. However, we repeated the analysis and presented here the results per number of limbs treated. I agree with you that we should not analyze results on the basis of the number of lesions. However, I do not think we should analyze it on the basis of number of patients. I think the limb is the unit of interest here, because of its clinical relevance and because this makes the angioplasty data comparable to the bypass data.

Dr. Henry Berkowitz (Philadelphia, Pa.). Dr. Porter preempted one of my questions. In our study published in 1991 in Circulation, we also encountered the same problem in terms of whether to report data based on limbs or arterial segments versus lesions. Obviously, the question is if multiple lesions were dilated and if there is failure based on hemodynamic parameters, do you count this as one or multiple failures?

Your data agree with ours in that the 5-year patency rate was 45% in your series and 44% in our series. Our study stopped recnliting patients after 1986; therefore we had some 10-year follow-up. The overall patency rate was 35% at 10 years.

The important point your study makes is that only a selective group of patients will get good results. Your results show that the patients -with claudication had the best 5-year patency rate of 54%. In our series the most successful results were obtained when the lesion was less than 2 cm in length, and the patency rate was 57% at 5 years and 53% at 10 years.

I think these data should serve to answer arguments from both sides of the fence, from the overly aggressive interventionist to the overly protective vascular surgeon.

I noticed length of lesion was not a variance you analyzed, yet it was such a significant factor in our data. Did you attempt to analyze it?

Dr. Hunink. Thank you for this question. It is an important point. We dilated only short lesions, and therefore lesion length is of less interest in our study. Unfortunately, we did not document the lesion length starting in 1980. We have started documenting lesion length only during the last few years, so we could not analyze this in detail. I hope we will be dilating lesions of somewhat greater length and then we will report results depending on lesion length. Your point is well taken.

Dr. All AbuRahma (Charleston, W.Va.). I commend you for your excellent presentation.

We had a similar experience in 89 patients on whom we performed PTA between 1985 and 1988, and it may surprise many people that these were performed by surgeons in the operating room. When we classified these according to lesion length, we were very impressed with the results of lesions less than 5 crn. After a mean follow-up of 2.5 years, the less than 5 cm lesions had a patency rate of 81% (the Kaplan-Meier curve analysis). The patency rate dropped to a rate in the low 60s for lesions measuring greater than 5 to 10 cm.

Our criteria were somewhat different from yours. We used a drop of ABI of more than 0.15, not 0.10, as our criterion for failure. Angiography was also done on patients who had a drop in their ABI of 0.15 or more, to confirm failure of the angioplasty.