Thrombectomy

Pulmonary Embolism Following Hemodialysis Access Thrombolysisl Thrombectomyl Timothy L. Swan, MD Stephen H. Smyth, MD Stephen J. Ruffenach, MD Scott 5. Berman, MD Gerald D. Pond, MD Index terms: Dialysis, shunts, 91.44 ' Embolism, pulmonary, 60.72,91.458 Thrombectomy, 91.1274 Thrombolysis, 91.1274 Urokinase, 91.1274

JVIR 1995;6:683-686 Abbreviations:PE = pulmonary embolism, PTA = percutaneous transluminal angioplasty, FTFE = polytetrafluoroethylene, UK = urokinase

From the Department of Radiology, University 0fArizonaHealth Sciences Center, 1501 N Campbell, Tucson, AZ 85724. From the 1994 SCVIR annual meeting. Received October 13, 1994; revision requested December 5; revision received March 20, 1995; accepted April 3. ~ d dress reprint requests to G.D.P. 63 SCVIR, 1995

PURPOSE: The increased use of thrombectomy with deliberate pulmonary embolization of thrombus following initial thrombolysis for occluded hemodialysis fistulas prompted the authors to measure the prevalence of pulmonary embolism (PE) due to the procedure. PATIENTS AND METHODS: Thirty-one patients with 43 acutely thrombosed polytetrafluoroethylene hemodialysis fistulas were treated with thrombolysis/thrombectomy. Perfusion lung scans were obtained in 22 patients. Patients were also continuously monitored for clinical signs or symptoms of PE. RESULTS: Perfusion scans were interpreted as consistent with PE in 59%of those studied, but no clinical signs or symptoms were present in 41 of the 43 cases (95%).However, two patients developed both signs and symptoms of acute PE in the postprocedural period and died. One had underlying pulmonary disease and had undergone thrombectomy before. The other had chronic heart disease. CONCLUSION: Thrombolysis/thrombectomyis usually safe and effective, even though many patients develop subclinical PE. The authors urge extreme caution in patients who have underlying pulmonary or cardiac disease andor have undergone the procedure before. S I N C E the introduction of extracorporeal hemodialysis by Kolff and Berk in 1944 (I),maintenance of vascular access has been a constant challenge. Polytetrafluoroethylene (PTFE) grafts are favored by vascular surgeons largely because they do not require as much maturation as do native arteriovenous fistulas (Cimino-Brescia).Unfortunately, these grafts fail with almost predictable regularity. The mean duration of primary patency is about 6 months, and the combined primary and secondary patency is as short as 9 months (2,3). Access failures are usually caused by stenoses, principally at the venous anastomosis (3). Arterial anastomoses develop stenoses less often, but either can compromise flow enough to cause thrombosis. Central stenoses at prior subclavian catheter insertion sites may also contribute to access site occlusion.

Previously, failed dialysis fistulas were always surgically revised by means of thrombectomy and anastomotic revision, extension of the graft, or complete replacement. Unfortunately, surgery consumes vascular access sites quickly. Since prolonged survival of dialysis patients is not uncommon, 20 or more operative procedures may be necessary during their lifetime. Radiologists offer an alternative to surgical revision with several reported techniques (4-10). The most common method is thrombolysis followed by angioplasty, if indicated. The first technique uses urokinase (UK) injection into the thrombus, maceration of the thrombus, and endovascular thrombectomy. This results in pulmonary embolization of residual thrombus but permits prompt resumption of dialysis, sometimes in less than 1hour. The

683

684

Journal of Vascular and Interventional Radiology

September-October 1995

Scientific, Watertown, Mass), with the end hole occluded by a guide wire, was placed across each anastomosis. Heparin (5,000 IU) was administered intravenously. By occluding the end hole with the guide wire and power injecting 3-5-mL aliquots just below burst pressure for the catheter, high-velocity jets of a mixture of contrast material, saline, and UK were delivered into the thrombus, causing fragrnentation and initiating lysis. Each graft was injected with a total of 250,000 IU of UK in 20-mL aliquots in this fashion. Aliquots were fluoroscopically observed dissecting around, into, and through the thrombus and then dissipated. Each burst was given after the prior one had dissipated. The thrombus was then further macerated with a 5-F Fogarty thrombectomy balloon (11mm) catheter (Baxter Healthcare, Santa Ana, Calif), with multiple inflations and a to-and-fro manipulation of the catheter. The entire process required approximately 10-20 minutes. Percutaneous transluminal PATIENTS AND METHODS angioplasty (PTA) of the venous anastomosis was usually necessary Thirty-five grafts in 31 patients (16 men and 15 women) were evalu- to extract thrombus that had not lysed. Small to moderate amounts of gted for hemodialysis access thromresidual thrombus were evident in bosis. Four patients in this group all cases and were pushed or pulled were treated for thrombosis of two through the venous anastomosis different sites of hemodialysis fistuwith the Fogarty catheter or the anlas that were placed at different gioplasty balloon. Arterial anastotimes. Ages ranged from 14 to 97 motic angioplasty andlor central veyears (mean, 59 years). These patients underwent treatment with the nous angioplasty was performed when necessary. We placed Wallthrombolysis/thrombectomy techstents (Schneider, Minneapolis, nique on 43 separate occasions for Minn) in two ~ a t i e n t to s maintain acutely thrombosed PTFE grafts patency of venous angioplasty sites. (Goretex; Gore & Associates, FlagPatients underwent dialysis after staff, Ariz, or Impra; Impra, Tempe, successful restoration of flow. Ariz). The 33 upper extremity and During each procedure, patients two lower extremity grafts had ocwere continuously monitored for cluded within 72 hours before treatclinical signs andlor symptoms of ment. PE. Twenty-two technetium-99m Crossed catheters were used to MAA (macroaggregates of human access each graft from both the veserum albumin) perfusion lung nous and arterial sides. Each limb scans were obtained, all within 24 was punctured approximately 5 cm hours of thrombolysis/thrombecfrom the anastomosis and the cathtomy. Xenon ventilation scans were eter directed toward the opposite also obtained in 13 of the 22 paanastomosis. A Mewissen multipleside-hole catheter (Medi-techBoston tients. A current chest radiograph

second technique depends entirely on continuous drip infusion of UK, which should reduce residual thrombus but may take many hours. Patients treated with continuous drip infusion require close observation including hospital admission and occasionally a stay in the intensive care unit during treatment. A central catheter may be required for dialysis before lysis is complete. This technique is now generally rejected as too slow, resource intensive, and costly. The newest, and probably fastest, technique uses mechanical thrombectomy alone, without UK (8). We have recently adopted thrombolysis/thrombectomy, prompted by economic and logistic pressures coupled with reports of safety and success at other institutions. Nonetheless, we were concerned about pulmonary embolism (PE) as a result of the technique and undertook this study.

was available in 15 patients. Scans and chest radiographs were requested in all cases. However, referring physicians denied the requests or considerations such as the next scheduled dialysis prevented the studies in some patients. The nuclear medicine studies and chest radiographs were reviewed independently by five senior radiologists who were asked to determine whether there were perfusion abnormalities consistent with PE that were not explained by findings on the chest radiograph or ventilation scan. Disagreements were resolved by majority rule.

RESULTS Thrombolysis/thrombectomy was successful in 38 of 43 attempts (88%). A total of 59 stenoses were encountered among the 38 successful attempts: 30 venous anastomotic, 11arterial anastomotic, 13 peripheral venous (basilic or cephalic), and five central venous (axillary, subclavian, innominate, or superior vena cava). Of these, 44 were considered causative stenoses. No stenosis was identified in four cases. Failure due to anatomic, not technical, reasons (multiple complex venous stenoses recognized after initial, partial, or complete lysis) resulted in abandonment of the procedure in favor of surgical replacement in five of 43 (12%).No arterial-side failures occurred. In the others, PTA (n = 32) with or without local surgical revision (n = 7) was successful. These patients were able to return for dialysis within 48 hours. None required interim central venous catheterization for dialysis. Perfusion defects judged to be consistent with PE and not explained by the findings on chest radiographs or ventilation scans were identified in 13 of 22 patients (59%). Often, scans were obtained in patients immediately following direct visualization of embolizing fragments fluoroscopically. Nonetheless,

Swan et a1

685

Volume 6 Number 5

nine of the lung scans were normal or showed no convincing evidence of PE. Forty-one of 43 treatments (95%) resulted in no clinical suspicion of PE. However, two patients developed signs and symptoms of acute PE in the postprocedural period; both died. The first patient had known pulmonary fibrosis secondary to scleroderma. In addition, she had undergone thrombectomy for dialysis access thrombosis 40 days earlier. She became dyspneic and agitated immediately following thrombectomy. Tachypnea, bradycardia, and hypotension followed along with a precipitous drop in 0, saturation. The patient died 2 days later. Permission for postmortem examination was denied. Perfusion defects were noted after the initial thrombolysis/thrombectomy 40 days earlier (a baseline ventilatiordperfusion scan was normal). After the second thrombolysis/ thrombectomy procedure, additional defects consistent with PE were identified. The second patient had chronic congestive heart failure. He underwent thrombolysis/thrombectomy for the first time, followed by venous PTA with no immediate symptoms of PE. There was no change in his usual level of dyspnea during or following the procedure. However, the patient experienced a cardiopulmonary arrest 36 hours following the procedure and died 4 days later. The patient's postprocedural ventilation/ perfusion scan was grossly abnormal with multiple large segmental and subsegrnental perfusion defects and near normal ventilation. The original interpretation was "high probability of PE." Overall mortality was 6% (two of 31 patients). Other complications included minor hemorrhage a t previous dialysis puncture sites in two cases, neither of which required surgical intervention or transfusion. In two cases, embolization of thrombus into a native artery occurred; this was treated successfully with continued infusion of UK.

DISCUSSION Thrombolysis for thromboembolic occlusions of the vascular system is one of the mainstays of the interventional radiologist. Transcatheter lytic therapy has recently been adapted for use on acutely occluded dialysis fistulas, thus avoiding surgical revision or replacement in the vast majority of cases. The goal of percutaneous transcatheter therapy is to preserve the functional life of these fistulas because there is an anatomically limited number of hemodialvsis access sites. In ah effort to reduce procedure time, speed the resumption of dialysis, eliminate the need for temporary dialysis catheters, and minimize admissions and cost, percutaneous thrombectomy following thrombolysis has been popularized. Some even advocate mechanical thrombectomy alone for occluded access grafts (8-10). Beathard reviewed 650 cases of thrombolysis, including his experience of 103 cases, and found no reports of clinically significant sequelae, even among 55 of his own patients who were treated with mechanical thrombectomy only (10). Dialysis centers and nephrologists are enthusiastic about this technique because patients return for dialysis within hours. There is no wait for a vascular surgery consultation or an available operating room. Surgical morbidity and mortality are eliminated, and graft site maturation is no longer a concern. In addition, an interim central venous catheter for dialysis is unnecessary. Compared with infusion, thrombolysis/thrombectomy offers enormous potential advantages. Although some reports suggest that rapid resolution with UK only can be achieved routinely (6,7),infusion may require hospitalization. Monitoring of the patient in the intensive care unit is mandatory in many institutions for thrombolytic therapy. Infusion often requires larger amounts of UK, further increasing

costs. Finally, infusion may fail owing to hemorrhagic morbidity (12). Thrombolysis/thrombectomy was adopted with reservation, due especially to concerns about possible PE. This study revealed that in 95% of the procedures, no signs or symptoms of PE occurred even though 59% of the scans were abnormal and consistent with PE. From the literature and personal communications, few symptomatic patients were expected, although several authors had mentioned the possibility of PE (10-14). Intuitively, a high frequency of positive perfusion scans was anticipated even if the patients remained asymptomatic, especially when passage of macerated thrombus from the fistula was observed fluoroscopically. To our surprise, PE detectable with scans occurred in only 59% of cases. Nonetheless, the prevalence of clinically significant PE was 5% in this series. This prevalence has not been reflected in the literature, although one clinically apparent PE was noted by Schilling et a1 (11). Dolmatch et a1 (15) responded to the report by Trerotola et a1 (8) on thrombectomy without UK that the risks of embolization are not yet fully understood and may be much more significant than reported. Dolmatch et a1 questioned the risk of the procedure in patients with cardiopulmonary disease and in those who underwent repeated thrombectomies. The two patients in this series who died following thrombolysis/ thrombectomy suggest that underlying cardiopulmonary disease does indeed seem to increase the risk of this procedure. Such patients may not tolerate even a modest burden of embolic material. Dolmatch et a1 also asked, "How effective is the pulmonary circulation in handling repeated episodes of embolization?" Multiple thrombectomies within a short interval hypothetically place patients at greater risk. One of the two deaths occurred in a patient who underwent two procedures and in whom the scans indicated a progression of thromboembolism.

686

Journal of Vascular and Interventional Radiology

September-October 1995

We believe that thrombolysis/ thrombectomy is generally safe and effective for management of occluded dialysis access sites. It offers enormous cost and time savings. However, we caution extreme care in the treatment of patients with underlying cardiac or pulmonary disease, or patients who have undergone the procedure previously. Perhaps these patients should undergo a trial of thrombolysis alone or surgical revision. Performance of preprocedural pulmonary perfusion scanning andl or pulmonary function studies might also be considered in high-risk cases. References 1. Kolff WJ, Berk HThJ. The artificial kidney: a dialyser with great area. Acta Med Scand 1944; 117:121-131. 2. Roberts AC, Valji K, Bookstein J J , Hye RJ. Pulse-spray pharmacomechanical thrombolysis for treatment of thrombosed dialysis access grafts. Am J Surg 1993; 166:221226. 3. Mehta, S. Statistical summary of clinical results of vascular access procedures for hemodialysis. In: Gore WL, et al, eds. Vascular access

4.

5.

6.

7.

8.

9.

for hemodialysis 11. Flagstaff, Ariz, Precept Press, 1991; 145-155. Davis GB, Dowd CF, Bookstein JJ, Maroney TP, Lang EV, Halasz N. Thrombosed dialysis grafts: efficacy of intrathrombic deposition of concentrated urokinase, clot maceration, and angioplasty. AJR 1987; 149:177-181. Smith TP, Cragg AH, Castaiieda F, Hunter DW. Thrombosed polytetrafluoroethylene hemodialysis fistulas: salvage with combined thrombectomy and angioplasty. Radiology 1989; 171:507-508. Valji K, Bookstein J J , Roberts AC, Davis GB. Pharmacomechanical thrombolysis and angioplasty in the management of clotted hemodialysis grafts: early and late clinical results. Radiology 1991; 178:243-247. Brunner MC, Matalon TS, Pate1 SK, McDonald V, Jensik SC. Ultrarapid urokinase in hemodialysis access occlusion. JVIR 1991; 2:503506. Trerotola SO, Lund GB, Scheel J, Savader SJ, Venbrux AC, Osterman FA. Thrombosed dialysis access drafts: percutaneous mechanical declotting without urokinase. Radiology 1994; 191:721-726. Middlebrook MR, Soulen MC, Cope C, Shlansky-Goldberg RD, Haskal ZJ, Pentecost MJ. Thrombosed dialysis grafts: percutaneous balloon

10.

11.

12.

13.

14.

15.

thrombectomy versus thrombolysis (abstr). JVIR 1994; 5%. Beathard GA. Mechanical versus pharmacomechanical thrombolysis for the treatment of thrombosed dialysis access grafts. Kidney Int 1994; 45:1401-1406. Schilling J J , Eiser AR, Slifkin RF, Whitney JT, Neff MS. The role of thrombolysis in hemodialysis access occlusion. Am J Kidney Dis 1987; 10:92-97. Young AT, Hunter DW, CastanedaZuniga WR, et al. Thrombosed synthetic hemodialysis access fistulas: failure of fibrinolytic therapy. Radiology 1985; 154:639-642. Kumpe DA, Cohen MAH. Angioplasty/thrombolytic therapy of failing and failed hemodialysis access sites: comparison with surgical treatment. Prog Cardiovasc Dis 1992; 10:263-278. Bookstein J J , Fellmeth B, Roberts A, Valji K, Davis G, Machado T. Pulsed spray pharmacomechanical thrombolysis: preliminary clinical results. AJR 1989; 152:1097-1100. Dolmatch BL, Gray RJ, Horton KM. Will iatrogenic pulmonary embolization be our pulmonary embarrassment? Radiology 1994; 191:615-617.