GI Bleeding: Diagnosis and Treatment
Interventional radiology plays an important role in the diagnosis and treatment of GI bleeding.21–36 The overall management of GI hemorrhage is discussed elsewhere (see Chap. 82). This section will focus on angiographic localization and treatment of upper and lower GI bleeding.
Indications and Patient Selection
GI bleeding typically occurs intermittently, not continuously. Angiography should be performed during active bleeding because often the only angiographic abnormality to guide treatment is the presence of contrast agent extravasation. Unfortunately, the clinical determination of active bleeding can be difficult. A decline in the hematocrit reflects previous bleeding, and large quantities of blood can accumulate within the intestines and slowly be expelled over hours following a bleeding event. Therefore, hemodynamic parameters such as tachycardia and hypotension often are the most accurate indicators of active bleeding.
Fortunately, the widespread use of endoscopy and nuclear medicine bleeding scans has improved the ability to document active bleeding and localize the source. Endoscopy should be the initial diagnostic modality in the workup of GI hemorrhage, particularly in the upper tract where it also may be therapeutic. Even if the source is not visualized, secondary findings usually can confirm the presence of active bleeding.
Nuclear medicine bleeding scans with technetium-labeled red blood cells are most useful in the evaluation of bleeding distal to the ligament of Treitz. Radionuclide studies can confirm active bleeding and localize intermittent lower intestinal hemorrhage because the radiopharmaceutical can be monitored for up to 24 hours. These studies are noninvasive and can be performed portably in the ICU. Angiography, on the other hand, is an invasive test that requires transport to the special procedures suite. For diagnosis, there must be active bleeding during the 10-second angiographic injection at a rate sufficient for detection. It is estimated that a bleeding rate of at least 0.5 mL/min is needed for angiographic visualization, whereas a rate of only 0.1 mL/min is required for radionuclide localization.
Preprocedure assessment includes confirmation of probable active bleeding and an estimation of the rate and severity of the bleeding. It is important to document the presence of known cardiovascular disease that would preclude vasoconstrictor administration and establish if there is a history of intestinal or intraabdominal vascular surgeries that may have altered the vascular anatomy. Resuscitation with blood products should be ongoing. In particular, accurate assessment of the coagulation profile is necessary because all endovascular therapies rely on normal coagulation to form a stable clot. A treatment plan should be selected prior to angiography, and surgical consultation should be sought early because failure to control bleeding by endovascular methods may necessitate emergent surgery.
Once active bleeding is confirmed and angiography is deemed beneficial, the patient should be transported promptly to the angiographic suite. Portable fluoroscopy units are inadequate for these complex examinations.
In upper GI bleeding, initial celiac axis and subsequent selective left gastric and gastroduodenal arteriography are performed. In lower GI bleeding, the superior and inferior mesenteric arteries, as well as the celiac artery, must be studied. The nuclear medicine bleeding scan is useful to guide selective injections.
The diagnosis of active bleeding is confirmed by identifying extravasated contrast material within the bowel lumen (Fig. 101-4A). In suspected lower intestinal bleeding, bladder catheterization is important so that extravasated contrast material is not obscured in the lower pelvis and rectum.
This elderly woman presented with severe lower GI bleeding. A. Digitial substraction angiography with selective injection of the superior mesenteric artery demonstrates a focus of contrast extravasation (short arrow) within the ileocolic region. B. After coil embolization (long arrow) of the bleeding vessel, repeat contrast injection demonstrates occlusion of the coiled vessel and no further extravasation.
Transcatheter embolization or selective intraarterial vasoconstrictor infusion techniques are endovascular alternatives to surgery in these critically ill patients. Therapy is individualized to the patient's overall medical condition, the cause and location of the bleeding, and the experience of the interventional radiologist.
For lower GI bleeding, the nonoperative therapy of choice traditionally has been selective vasoconstrictor infusion. However, embolotherapy, which has long been valuable in upper GI bleeding, is being chosen more frequently in the treatment of lower GI bleeding as well. Embolotherapy typically has been considered safe in the upper GI tract owing to its rich collateral arterial network. Embolization of the lower GI tract, with its lesser degree of collateralization, raised concerns for precipitating bowel ischemia and infarction, suppositions that were proven true in the past.21–23 However, with the development of microcatheters, there is renewed interest in embolization as the primary treatment for lower GI tract bleeding. Microcatheters allow selective catheterization of a peripheral bleeding vessel, thus limiting the region of intestine affected by embolization. Superselective embolization is proving to have high technical success rates with lower rebleeding rates than vasopressin infusion and low rates of ischemia and infarction25,26 (see Fig. 101-4B). When superselective catheterization is not possible, vasoconstrictor infusion remains the best endovascular therapy.
Vasopressin is a potent vasoconstrictor that is the pressor component of the active hormone of the posterior pituitary gland. It produces constriction of the splanchnic vascular bed and contraction of the smooth muscle of the gut. Direct intraarterial injection of vasopressin results in fewer systemic side effects than intravenous administration because a lower dose is required.
Vasopressin therapy is started immediately after selective angiography with direct infusion into the major artery supplying the bleeding site. In an area with a dual blood supply (e.g., the duodenum), vasopressin must be infused through both vascular beds (celiac and superior mesenteric artery) and is less effective. Initial infusion rates generally range from 0.1 to 0.4 unit/min, with angiography repeated at 20-minute intervals to assess the vasoconstrictive effect. If the bleeding stops, the patient is returned to the ICU on a continuous infusion. If bleeding persists, the vasopressin infusion is increased. In general, an infusion of more than 0.4 unit/min results in significantly more systemic side effects. Therefore, if bleeding is not controlled at 0.4 unit/min, alternative therapies should be considered.
The vasopressin infusion is continued for 12 to 24 hours and then tapered gradually over 24 to 48 hours while the patient's condition is monitored carefully. An infusion of saline for the final 12 to 24 hours is helpful in the event of rebleeding. If rebleeding does occur, vasopressin infusion can be repeated.
Transcatheter embolotherapy carries a higher rate of success without rebleeding and eliminates the need for prolonged indwelling arterial catheters. In some cases, embolotherapy serves as definitive treatment of bleeding. Embolotherapy is most helpful in areas with rich collateral arterial networks (i.e., esophagus, stomach, duodenum, and rectum), where multiple blood supplies minimize the risk of ischemic injury. As discussed earlier, because of the high success rates without rebleeding, embolization is being selected more frequently as therapy for bleeding in the small intestine and colon, regions without the same degree of collateralization. Owing to the risk of ischemia and infarction, embolotherapy typically should be avoided in the postoperative gut, especially near anastomoses.
Multiple agents have been used for embolization in the GI tract, and the agent of choice varies by bleeding location and experience of the operator. Surgical gelatin (Gelfoam), polyvinyl alcohol (Ivalon), and microcoils all have been used with success. Gelfoam and coils are the most commonly used agents. Many authors favor Gelfoam because it is a resorbable agent that provides temporary occlusion (approximately 5 to 10 days), allowing the bowel lesion to heal. Injected as particles intermixed with contrast material, it carries a small risk of nontarget embolization. Many authors prefer coil embolization because, when superselective catheterization is possible, the coils can be deployed precisely without affecting nearby vessels.
Certain locations are best treated using a particular agent. For example, the left gastric artery can be embolized with Gelfoam for treatment of a Mallory-Weiss tear, even when endoscopy has suggested a bleeding site near the gastroesophageal junction but angiography has failed to document active extravasation. Coil embolization of this vessel should be avoided because permanent occlusion would preclude treatment via this vessel should rebleeding occur. The morbidity of Gelfoam embolization of the left gastric artery is extremely low, and the method decreases rebleeding. In contrast, coils are much preferred in the small bowel and colon because they can be deployed precisely with permanent occlusion of the bleeding vessel.
Immediate Postprocedure Care
Embolotherapy and vasoconstrictive infusions both work by decreasing the pulse pressure at the bleeding site. This allows hemostasis and healing of the bleeding lesion. This approach requires a functioning hemostatic mechanism. Frequently, patients with massive GI bleeding have multisystem organ failure or sepsis or have received many units of transfused blood. Any of these can result in an underlying coagulopathy, which should be corrected as rapidly as possible.
Patients on vasopressin infusion must be monitored carefully. During the initial infusion, many patients experience abdominal cramps secondary to the effects on intestinal smooth muscle. This may result in dramatic emptying of accumulated blood from the bowel. If cramps fail to subside within 30 to 60 minutes, or if they recur, significant gut ischemia may be present. The vasopressin dose should be decreased slowly until the pain is relieved. Patients also should be monitored carefully for signs of myocardial and peripheral ischemia, dysrhythmia, and hyponatremia secondary to the antidiuretic effect of vasopressin. Hyponatremia and decreased urine output usually are treated easily with diuretics.
Results and Complications
Results with transcatheter therapy for acute GI bleeding are highly variable depending on several factors, including the underlying medical condition, the cause of bleeding, and the bleeding site. In our experience, patients with extensive multisystem organ failure and GI bleeding rarely benefit from intervention. Unfortunately, it is difficult to determine beforehand which individuals may benefit.
In patients without multisystem organ failure, control of upper GI bleeding with embolotherapy usually is successful in 70% to 80% of cases.28,30 In lower GI bleeding, vasopressin infusion is effective in 60% to 90% of patients, but the overall recurrence rate can be as high as 40%. Despite rebleeding, these patients can benefit from this temporary control, allowing elective surgical intervention under more favorable conditions. Embolization for lower GI tract bleeding has resulted in clinical success rates of 60% to 100%, with recent series limited to the colon reporting greater than 85% success. Rebleeding rates following embolization are approximately 10%.24
Complications of embolotherapy in the upper GI tract are rare28 owing to the rich collateral network around the stomach and duodenum that protects against infarction. Care must be taken in postoperative patients, in whom collateral vessels have been ligated. Backflow of embolic material into the liver and spleen occurs occasionally and generally is well tolerated. Clinically significant hepatic or splenic infarctions are rare.
Complications of vasopressin treatment typically are related to diffuse vasoconstriction, which results in significant ischemia to the heart or extremities. Other complications include dysrhythmia and hyponatremia. Thrombosis of the femoral or visceral arteries may result from prolonged placement of the intraarterial catheter. Significant complications related to catheter-directed vasopressin therapy are infrequent, generally less than 10%.31
Despite multiple scintigrams, endoscopies, and/or arteriograms, some patients with lower GI tract hemorrhage will have no bleeding site identified. For this subset with obscure bleeding, elective arteriography is indicated to evaluate for the presence of a structural abnormality that may bleed intermittently and guide surgical resection of this offending lesion.32 In addition, heparin, intraarterial vasodilators, and/or intravenous or intraarterial thrombolytics have been administered in an attempt to provoke bleeding. Several studies have evaluated one or a combination of these agents and have documented a contribution to detection and treatment of bleeding in between 20% and 50% of patients without any adverse events or complications.33–36 Thus the diagnostic yield of provocative arteriography may be somewhat disappointing, but for a subset of patients with chronic, repetitive, nonlocalized bleeding, it may serve a role when all other measures have failed.
Transjugular Intrahepatic Portosystemic Shunt
Indications and Patient Selection
Patients with portal hypertension represent a diverse group in the ICU. Typically, patients fall into one of two broad categories with respect to their symptoms. The most urgently ill are those with uncontrollable bleeding from esophagogastric varices. The second group, consisting of those with fluid management problems, includes patients with intractable ascites, hydrothorax, or hepatorenal syndrome.
Portal hypertension with variceal bleeding is a special form of upper GI hemorrhage that requires a different approach than arterial bleeding. Endoscopy with sclerotherapy/banding remains the initial diagnostic and therapeutic procedure of choice. However, endoscopic treatment is not always successful, and recurrent bleeding occurs in about 30% to 50% of patients.
Portal hypertension with ascites or hydrothorax typically is managed with diuretics and paracentesis/thoracentesis. However, when the fluid reaccumulates in a short time period requiring frequent repeat percutaneous removal or therapy results in the development of renal failure, electrolyte imbalances, or encephalopathy, alternate therapy may need to be employed.
Historically, the definitive therapeutic intervention was the surgical creation of a shunt from the high-pressure portal system to the central circulation, allowing decompression of the portal system. Interventional radiologists now play a significant role in the treatment of these complex patients with percutaneous creation of transjugular intrahepatic portosystemic shunts (TIPS).37–41 TIPS is a percutaneous portocaval shunt through the liver parenchyma connecting a portal vein and a hepatic vein. Its function is identical to a surgically created portosystemic shunt, but morbidity and mortality are reduced markedly with this less invasive alternative.
Indications for TIPS include multiple episodes of esophagogastric variceal hemorrhage refractory to sclerotherapy, presence of gastric varices in the absence of splenic vein thrombosis, intractable ascites, hepatic hydrothorax, hepatorenal syndrome, Budd-Chiari syndrome, and preoperative portal decompression in patients awaiting liver transplantation.
Preprocedure evaluation includes a determination of the indication and urgency for TIPS, review of hepatic imaging, and assessment of patient candidacy. Patients with decompensated portal hypertension typically are very sick, and for some, TIPS will improve their symptomatology and may be lifesaving. However, for others, TIPS may result in fulminant liver failure, new or worsening encephalopathy, or premature death. Several parameters have been used in an attempt to stratify patient risk and predict survival after TIPS. The Child-Pugh classification considers the presence and degree of ascites and encephalopathy, as well as the serum bilirubin and albumin concentrations and the INR. A newer measurement, the MELD score, is a mathematical model based on the serum creatinine and bilirubin concentrations, the INR, and the etiology of the underlying liver disease. Although neither can predict survival with complete accuracy, recent data support that the MELD score is a better predictor of 3-month outcome than the Child-Pugh score37 and that a MELD risk score less than 1.8 is associated with improved survival.38
Fluid resuscitation to treat ongoing bleeding and correction of an existing coagulopathy are critical prior to and during the procedure. For patients with ascites, paracentesis just prior to the procedure can facilitate safe creation of a TIPS. Finally, for some patients, general anesthesia may be necessary and helpful because many of these patients have low baseline blood pressures that preclude sedation at levels necessary to provide patient comfort.
Prior to TIPS placement, it is useful to confirm portal vein patency. This is typically done by US evaluation or other noninvasive imaging but, if necessary, can be determined via superior mesenteric arteriography. In the emergent situation, this preprocedure imaging may not be feasible.
When the decision to perform a TIPS procedure has been made, the patient is transported to the interventional suite. The right internal jugular vein is accessed, and a venous sheath is inserted into the right atrium. The right hepatic vein is catheterized. Wedged hepatic venography typically is performed using carbon dioxide. This gas diffuses rapidly through the liver retrogradely, providing an image of the portal vein and its branches. Using this venogram for guidance, a needle is directed through the hepatic parenchyma into the portal vein. Pressure measurements between the portal vein and the right atrium are obtained, and the portosystemic gradient is calculated. Direct portal venography is then performed to document the anatomy and the presence of varices (Fig. 101-5A). The newly created tract is dilated and held open by one or more self-expanding metal stents. After stent placement, pressures are again obtained to confirm reduction of the portosystemic gradient, and venography is performed to document stent patency and assess the degree of variceal filling (see Fig. 101-5B). A gradient of less than 12 mm Hg is desired because pressure gradients above this level are associated with recurrent variceal hemorrhage. If initial shunt placement is unsuccessful at reducing the gradient adequately and varices continue to opacify, variceal embolization with Gianturco coils or ethanol injection can be performed as an adjunct.
This 40-year-old woman with a history of cirrhosis due to ethanol abuse developed intractable ascites not responsive to medical management. Creation of a transjugular intrahepatic portosystemic shunt (TIPS) was requested. A. Initial venography after passage of a needle through the hepatic parenchyma from the right hepatic vein (arrow) into a portal vein (P) demonstrates filling of varices (arrowheads). B. Following TIPS creation, contrast agent injection demonstrates brisk flow through the shunt and no filling of the varices.
Immediate Postprocedure Care
Following portal decompression and stabilization, patients should be monitored carefully for at least 24 hours and should continue to receive any needed blood products. A color duplex US of the shunt, as well as the hepatic and portal veins, should be obtained the following day to assess adequacy of flow through the shunt and to provide a baseline for subsequent follow-up.
Results and Complications
Technical success of TIPS placement is greater than 90% when the portal vein is patent. TIPS have been created in the presence of a thrombosed portal vein, but technical success rates are lower. Primary shunt patency at 6 months ranges from 75% to 90%,39,40 with recurrent bleeding occurring in 4% to 17%. A 30-day mortality rate of approximately 15% and a 6-month mortality rate of 30% have been reported. The overall mortality rate varies by the Child-Pugh classification and patient stability at the time of the shunt procedure. A series by Encarnacion and coworkers41 broke down the cumulative 30-day survival rates for Child-Pugh classes A and B (91%) and class C (71%). These figures compare favorably with surgically placed portosystemic shunts in terms of overall morbidity and mortality and the length of postprocedure survival. In terms of rebleeding, the recurrence rate is lower with TIPS when compared with sclerotherapy. It is, of course, understood that many of the patients undergoing percutaneous shunt creation have failed sclerotherapy.
TIPS creation has been very successful in acutely relieving portal hypertension and variceal bleeding, but shunt stenosis may occur in up to 70%.40 Routine surveillance with Doppler US is necessary to identify shunt problems requiring further venographic examination and intervention. TIPS venography and revision can be performed on an outpatient basis in most cases.
Complications of TIPS include neck hematoma, liver capsule puncture with or without abdominal hemorrhage, encephalopathy, hemobilia, or worsening liver failure. Occasionally, patients develop multisystem organ failure with no evident source of sepsis. This may emphasize the importance of a functioning liver for the modulation of systemic inflammation.
Bronchial Artery Embolization
Life-threatening hemoptysis, although uncommon, is associated with an exceedingly high mortality rate (50% to 70%) in patients treated by conservative means (see Chap. 41). Since the early 1970s, interventions in the management of massive hemoptysis have gravitated toward transcatheter embolization. The mortality rate for surgery in the setting of acute hemorrhage remains high (33%). Embolization can act as palliative therapy or can temporize patients until surgery is feasible.44–47
Indications and Patient Selection
Massive hemoptysis is defined as bleeding in excess of 300 mL during a 24-hour period. However, smaller amounts may be life threatening due to airway compromise, so intervention may be indicated with lesser amounts of hemorrhage. Bleeding arises from the systemic circulation (bronchial arteries and transpleural collaterals) in about 95% of patients. Chronic inflammatory diseases such as tuberculosis, bronchiectasis, aspergillosis, and cystic fibrosis and neoplastic diseases such as bronchogenic carcinoma all may predispose to bronchial bleeding. In about 5% of patients, hemoptysis relates to lesions of the pulmonary arterial circulation (Rasmussen/mycotic aneurysm or pulmonary arteriovenous fistula). A particular cause of massive hemoptysis in the ICU patient is an iatrogenic injury of the pulmonary artery following pulmonary artery catheter insertion. Both systemic and pulmonary arterial causes of hemoptysis are amenable to transcatheter embolization.
Attempts to localize or lateralize the bleeding to one or more lobes should be undertaken prior to angiography. Chest radiography, chest CT, and/or bronchoscopy can be used. Bronchoscopy has long been considered the primary method of diagnosis and localization, but CT has proven to be very helpful in preprocedure evaluation.44 CT has higher rates of localization, may detect causative vascular lesions, may suggest a specific diagnosis, and may identify bronchial and nonbronchial systemic feeder arteries.45,46 Protective measures prior to transport to angiography include insertion of a double-lumen endotracheal tube to allow selective ventilation and protection of the contralateral lung and/or insertion of a balloon occlusion catheter at the bleeding site to tamponade the source until embolization.
As with other complicated vascular diagnostic and therapeutic interventional radiologic procedures, patients must be transported to the angiographic suite for optimal angiographic imaging and intervention.
There is marked variability in bronchial arterial anatomy, so evaluation begins with a nonselective descending thoracic aortogram. After anatomic mapping, selective arteriography is performed in the bronchial artery corresponding to the radiographic and/or bronchoscopic abnormality. In situations where the underlying process is related to chronic inflammation, transpleural collateral vessels from the subclavian, axillary, internal mammary, intercostal, and phrenic arteries may be recruited. Depending on the location of the bleeding site, evaluation of one or more of these potential pathways via selective arteriography may be necessary.
Unlike arteriographic evaluation of GI bleeding, extravasation of contrast material is not identified commonly. Angiography more often will reveal a hypertrophied and abnormal bronchial artery with neovascularity, hypervascularity, aneurysm formation, or shunting of blood into the pulmonary artery or vein. Prior to transcatheter embolization, the diagnostic arteriogram must be reviewed carefully for the presence of arterial branches supplying the spinal cord. This anatomic configuration occurs more commonly on the right, where 90% of patients have a common bronchial-intercostal artery trunk, 5% to 10% of which have branches supplying the spinal cord. Embolization should not be performed if a spinal branch could be embolized inadvertently as well. In some cases, the selective catheter or a coaxial microcatheter can be advanced beyond a spinal branch origin, allowing safe embolization. If no abnormal bronchial or other systemic transpleural collateral source is found, pulmonary arteriography should follow.
For bronchial and nonbronchial systemic sources of massive hemoptysis, polyvinyl alcohol foam (Ivalon) and gelatin sponge (Gelfoam) are the embolic agents used most commonly. In selected cases of hemoptysis secondary to an arteriovenous malformation (AVM) arising from the pulmonary artery, coils or detachable balloons can be used to occlude the large single feeding vessel with good results. This latter situation is seen more commonly in patients with hereditary hemorrhagic telangiectasia, where multiple AVMs may be present.
Immediate Postprocedure Care
As with any embolization procedure performed to treat bleeding, patients should be monitored closely over at least 24 hours for signs of rebleeding. Considering the potential risk of nontarget embolization to the spinal cord, neuromuscular checks should be obtained at frequent intervals following embolization.
Results and Complications
Bronchial artery embolization for the immediate control of massive hemoptysis is successful in more than 75% of patients, with recurrence of bleeding in approximately 20%.46,47 In one report, embolization was unsuccessful and hemoptysis continued to death in approximately 9% of patients.47 Long-term success rates are lower, likely owing in part to progression or continuation of the underlying disease. Recurrent bleeding following embolization may be the result of partial embolization, recanalization, revascularization by collateral channels, inadequate treatment, development of systemic nonbronchial collaterals, or progression of the lung disease. Repeat embolization has been shown to improve outcomes in patients with recurrent hemoptysis, particularly when repeat embolization entails embolization of previously overlooked supplying vessels.47 It is important to note that bronchial artery embolization should be considered a palliative, not a curative, procedure. It allows time to plan elective surgery or to treat the patient with antimicrobials.
Procedure-related complications of diagnostic arteriography are uncommon but include chest pain due to transient ischemia, temporary dysphagia due to embolization of esophageal branches, and arterial dissection. A rare additional risk is transverse myelitis following contrast agent injection into a spinal artery, particularly ionic agents. The major feared complication of bronchial artery embolization is spinal cord ischemia due to inadvertent embolization of one or more spinal arteries. Careful evaluation for bronchial or intercostal arteries that provide branches to the anterior spinal artery is critical because embolization should be avoided in such situations.
Foreign Body Retrieval and Catheter Repositioning
Indications and Patient Selection
ICU patients often require one or more intravascular infusion catheters and/or monitoring devices. On occasion, catheters are malpositioned or fractured and may lodge in the vascular system—usually the central veins, heart, or pulmonary arteries. This typically occurs during catheter introduction or repositioning and rarely is related to a defect in the material itself. Other types of intravascular foreign bodies may include broken pacemaker leads or migrated caval filters.
The presence of an intravascular foreign body usually is an indication for its removal. Complications have been reported in up to 70% of patients in whom catheter fragments were left in place.48 Major complications of a retained catheter fragment include vascular or cardiac perforation, sepsis, dysrhythmia, and pulmonary embolism. The medicolegal implications also should be considered.
Percutaneous retrieval of the intravascular foreign body is the treatment of choice. It generally can be accomplished without an incision or general anesthesia. The alternative is a major surgical procedure that can pose high risks. Standard monitoring is necessary, with special attention to ECG monitoring given the risk of inducing arrhythmias when manipulating catheters or fragments within the heart.
Access into the vascular system typically is chosen to facilitate easy, minimally traumatic removal. A venogram is performed to confirm the intravascular position of the fragment and to select a suitable site of engagement. A number of retrieval devices are available, but the snare device is selected most commonly. Once the fragment has been engaged and trapped, it is pulled taut and folded on itself so that it can be withdrawn at the venous access site. Rarely, a surgical cutdown is necessary if the fragment is large or not pliable. Alternative retrieval devices include baskets, graspers, and balloon catheters.
A number of maneuvers can redirect malpositioned central venous catheters. When standard replacement is unsuccessful at the patient's bedside, fluoroscopic guidance can facilitate repositioning. Prior to catheter manipulation, venography should be performed to define the anatomy and document any potential problems or occlusions. In many cases, redirection under direct visualization is all that is required. In other cases, the venous system can be entered from a separate site and the end of the catheter snared and pulled into the desired orientation.
Results and Complications
Percutaneous foreign body retrieval is very effective, with success rates ranging from 80% to 95%.49 Unsuccessful retrieval often is due to the lack of a free end to grasp, incorporation of the fragment into the vessel wall, or location in an extravascular compartment.
Complications of percutaneous retrieval and catheter repositioning are rare. Transient dysrhythmias can occur during catheter manipulation within the heart. Most resolve by withdrawing all devices from the heart. If the foreign body is firmly adherent to the endothelium, endocardium, or a cardiac valve, traction must be applied gently to avoid vascular damage.
Percutaneous Caval Filter Placement
Indications and Patient Selection
Pulmonary embolism is a common cause of morbidity and mortality in the ICU patient. Anticoagulation with intravenous heparin, oral warfarin, or subcutaneous low-molecular-weight heparin remains the first-line therapy for treatment. However, in some patients, anticoagulation is contraindicated or ineffective. This has led to the development of a number of mechanical devices to “filter” emboli from the pelvis and lower extremities.50–58
Until the mid-1980s, these devices required a surgical cutdown in the internal jugular vein owing to the large size of the deployment sheath. In recent years, the profile of caval filters has decreased from 24F (Greenfield filter) to between 6F and 12F, making percutaneous placement feasible. This approach is convenient and can be combined with pulmonary arteriography and/or central venous catheter insertion.
As insertion techniques evolved over the past 10 years, many new devices have been developed. Several excellent reviews of caval filters are available.51–53 Currently, a number of permanent low-profile caval filters are available for patient use in the United States. These include the Titanium Greenfield, Bird's Nest, VenaTech, Simon Nitinol, TrapEase, Optease, and Gunther tulip vena caval filters.
Temporary or retrievable filters are the newest development in caval filtration devices. Temporary filters remain tethered to a transcutaneous catheter or wire at the insertion site so that early removal can be accomplished to prevent infection. Retrievable filters are designed with a hook at one end that can be engaged using a snare device, allowing removal through a long sheath. The Gunther tulip and TrapEase vena caval filters have been deployed and retrieved successfully, although each is approved by the Food and Drug Administration (FDA) for permanent deployment only, and this represents an off-label use of these products52,53 (Fig. 101-6). Retrieval within 10 to 14 days of insertion is recommended owing to “endothelialization” of the filter legs and the potential for vein damage with later removal. Venography is required at the time of retrieval to ensure the absence of thrombus within or around the filter because removal of the filter would result in dislodgment of this clot and pulmonary embolism.
Retrieval of a Gunther tulip inferior vena caval filter. A. Fluoroscopy demonstrates a long vascular sheath through which a snare device is used to engage the hook of the filter (arrow). B. The hook is engaged and the sheath advanced over the snare and hook. C.As the sheath is further advanced over the filter, the legs collapse and pull away from the vessel wall. D. The filter is covered with fibrinous debris and blood clot.
Indications for caval filter placement generally fall into three categories: (1) contraindication to or complication of anticoagulation in a patient with documented pulmonary emboli, (2) recurrent pulmonary embolism despite anticoagulation, and (3) prophylaxis. Although the first two are well-accepted indications, prophylaxis is controversial.53 Indications for prophylactic vena caval interruption have become more liberal as the procedure has become simpler and safer and retrievable filters have been developed. Nevertheless, many physicians remain unfamiliar with filter placement or consider it too invasive for common use. Caval filtration is greatly underused in the management of critically ill patients.
Prophylactic filter placement (in addition to heparin therapy) is warranted in patients with proven venous thromboembolism and cardiovascular compromise or severe chronic obstructive pulmonary disease, as well as in all patients with free-floating pelvic thrombi. In addition, prophylactic caval filter placement should be considered in certain high-risk surgical (orthopedic, trauma, neurosurgical) and oncology patients. If retrievable filters become more accepted and gain FDA approval, prophylactic filter placement may become more common.
Inferior vena caval filter placement generally requires transport to the interventional radiology suite. Rarely, in highly unstable patients, filter placement may be done at the bedside with portable fluoroscopy equipment. However, the lack of high-quality fluoroscopy equipment makes performance of an adequate inferior vena cavagram difficult (see below).
The femoral approach generally is preferred for percutaneous caval filter placement because it is easily accessible. The internal jugular approach may have advantages in patients with extensive pelvic and caval thrombosis, tortuous iliac veins, or multiple femoral lines.
An inferior vena cavagram should be performed before filter deployment to demonstrate the level of the renal veins, the presence of thrombus in the inferior vena cava, the diameter of the inferior vena cava, and the presence of congenital venous anomalies. After the inferior vena cavagram is performed, the puncture site is dilated to a suitable size, and a sheath is placed. The filter is introduced through the sheath and discharged below the renal veins. Infrarenal placement is preferred to maintain renal vein patency in the event of caval thrombosis. If thrombus extends to the level of the renal veins, suprarenal filter placement may be used.53 A follow-up cavagram is done to document filter placement.
Immediate Postprocedure Care
The patient should be kept at bed rest with close observation of the puncture site for 4 to 6 hours. If indicated, heparin can be restarted immediately after local hemostasis is obtained. Continued heparin therapy can be helpful to prevent extension of thrombus in the legs or pelvis.
Results and Complications
The reported rate of recurrent pulmonary emboli varies with each particular filter design but ranges from 0.5% to 3%. In most cases, however, documentation of recurrent pulmonary emboli is difficult to obtain.
Complications of inferior vena cava filter placement include caval thrombosis, filter migration, and trauma at the insertion site. Caval thrombosis varies with filter design and ranges from 2% to 9%. In recent years, the currently available filters have been modified such that filter migration is a rare event.
A potential complication with the percutaneous approach is puncture-site thrombosis. Significant access-site thrombosis with the newer low-profile filters ranges from 2% to 10%, with an overall (nonocclusive and occlusive) thrombosis frequency of up to 35%. Between 4% and 6% of patients display clinical symptoms.58
Catheter-Directed Fibrinolytic Therapy
Local catheter-directed fibrinolytic therapy is an important addition to the nonsurgical treatment of acute and subacute arterial occlusion. Catheter-directed fibrinolytic infusions infuse the thrombus directly and therefore concentrate the fibrinolytic agent's action at the site of the vascular occlusion. This is favored over systemic therapy because it permits lower doses and minimizes systemic side effects.59–64
Indications and Patient Selection
Catheter-directed fibrinolytic therapy achieves the best results with acute thrombosis. Therefore, the best indication is thrombosis that occurs during angiography or angioplasty. The arterial catheter is already in place, and the clot is fresh.
Fibrinolytic therapy also is beneficial in native-artery and bypass-graft occlusions. Thrombosis in these cases usually is a manifestation of underlying stenotic disease. Frequently, these stenoses can be treated with interventional radiologic techniques such as angioplasty, sparing the patient a surgical bypass. Although the exact age of the clot can be difficult to determine, best results are seen with occlusions that are less than 6 to 10 weeks old.
Another indication for fibrinolytic therapy is embolic occlusion. In large-vessel emboli (e.g., femoral or brachial artery), a surgical Fogarty embolectomy is extremely effective. Small-vessel emboli, however, are difficult to remove mechanically owing to vessel size, so catheter-directed fibrinolysis frequently is the best treatment available.
Despite the effectiveness of fibrinolytic therapy, not all patients are candidates. Patients with acute thrombosis and limb-threatening ischemia should be treated with surgery because catheter-directed thrombolysis is a slow process, requiring 12 or more hours of treatment for reperfusion. In addition, although the fibrinolytic agent is directed into the clot, a significant amount of it enters the systemic circulation as it diffuses through the clot. Therefore, patients at risk of significant bleeding are not candidates. Specifically, contraindications to thrombolysis include recent intracranial, thoracic, or abdominal surgery; recent GI bleeding; recent stroke or known intracranial neoplasm; recent major trauma; current pregnancy; severe hypertension; known bleeding diathesis; and infected thrombus.
In catheter-directed fibrinolytic therapy, the goal is to deliver the drug directly into the thrombus in order to activate plasminogen bound to the fibrin clot directly without activating significant amounts of circulating plasminogen. Various coaxial catheters and multisidehole catheters are available that allow maximum permeation of the fibrinolytic agent into the thrombus. Typically, the infusion catheter is placed through an arterial sheath that facilitates the catheter exchanges necessary for repeat arteriograms.
After a complete diagnostic arteriogram, a multisidehole catheter is embedded into the thrombus and the infusion started. Traditionally, urokinase was the agent of choice. However, when sales of urokinase were suspended in 1998 owing to production problems, several other fibrinolytic agents, including tissue-type plasminogen activator (t-PA) and recombinant tissue-type plasminogen activator (rt-PA), became more widely used. Streptokinase, the first lytic agent used, is currently in disfavor because of its antigenicity and inferior results in clinical studies.60 Although urokinase has now returned to the American market, its use remains limited in some areas because of cost.
No standard doses for these agents have been agreed on. Most involve a steady infusion over a period of hours with or without a loading dose. For urokinase, a typical protocol is to lace the clot with 250,000 units and then initiate infusion with 250,000 units/h for 2 hours. After the initial high-dose infusion, the dose is decreased to 80,000 to 100,000 units/h for the duration of the therapy, which may last up to 24 to 36 hours. A recent review of rt-PA indicates that protocols for rt-PA have been more variable, with both weight-based and non-weight-based dosing, dosing with or without an initial bolus, and/or the use of high-dose short-duration pulse-spray thrombolysis.59 Typical doses range from 0.02 to 0.1 mg/kg per hour for weight-based and 0.25 to 10 mg/h for non-weight-based regimens to a maximum of 100 mg.
Immediate Postprocedure Care
Patients undergoing catheter-directed fibrinolytic infusions are placed in the ICU. The puncture site, the affected extremity, and the patient's bleeding parameters are monitored frequently. Patients are kept at strict bed rest, with frequent Doppler examinations of the ischemic extremity.
Low-dose intravenous heparin (500 to 1000 units/h) is frequently instilled through a sheath to decrease the incidence of pericatheter thrombosis (see below). Anticoagulation with heparin also is helpful to prevent rebound thrombosis after the fibrinolytic infusion is completed.
Laboratory monitoring tests should include baseline fibrinogen, partial thromboplastin time, and hematocrit. Every 4 hours, the fibrinogen level should be checked to monitor for systemic effects of the fibrinolytic agent. If the fibrinogen level falls below 100 mg/dL, the infusion should be slowed or discontinued. Experience has shown that a fibrinogen level below 100 mg/dL is associated with a significantly increased incidence of remote bleeding.60 However, fibrinogen levels rarely fall this low during catheter-directed urokinase infusions.
Periodic angiographic follow-up is essential and must be performed in the angiographic suite. High-quality fluoroscopic observation is necessary to evaluate progression of lysis. Frequently, the position of the catheter must be changed to keep it embedded in clot.
Successful lysis generally is achieved after 12 to 18 hours of infusion. At that point, underlying stenotic lesions are treated with angioplasty, and the catheter is removed. If no progress is noted over 6 to 12 hours, the infusion is stopped. It is critical that all decisions regarding institution of fibrinolytic therapy, continuation of the therapy, and termination of the infusion be made in conjunction with the vascular surgeon caring for the patient.
Results and Complications
A successful fibrinolytic infusion generally is one that yields either complete lysis or lysis with significant clinical improvement. Success rates reported in the literature vary from 40% to 90%, with an average of 60% to 70%61,62 Comparison of various series and fibrinolytic agents is extremely difficult because of heterogeneous patient populations and protocols. Several studies, however, have evaluated the effectiveness of catheter-directed fibrinolysis versus surgery. The Rochester study found similar limb salvage rates for both groups (82%) but improved overall survival in the thrombolysis group at 1 year (84% versus 58% for surgery, p = 0.01).63 The Surgery versus Thrombolysis for Ischemia of the Lower Extremity (STILE) trial comparing thrombolysis (urokinase or rt-PA) and surgery found no difference in morbidity or mortality between the two groups based on intent to treat but noted better amputation-free survival for the thrombolysis patients at 6 months (85% versus 62% for surgery, p = 0.01) and reduction in the extent of surgery needed for patients with acute limb ischemia, whereas patients with chronic limb ischemia had lower amputation rates with surgery (4% versus 14% for lysis).47 Generally, success correlates best with age of the thrombus and status of the distal runoff vessels.
Complications are related primarily to bleeding. Most bleeding occurs at arterial puncture sites and is relatively minor. However, life-threatening bleeding at remote sites (intracranial, gastrointestinal, retroperitoneal) has been reported and can result in significant morbidity. When serious bleeding occurs, therapy should be discontinued unless the bleeding can be controlled by manual compression. Fresh frozen plasma can be given to replace coagulation factors. Significant bleeding generally is reported in 10% to 12% of patients. The risk of bleeding increases with the length of infusion, so the duration of the infusion should be kept under 24 hours if possible. In addition, bleeding risk increases when fibrinogen levels drop below 100 mg/dL, and the infusion should be reduced or stopped if levels fall. Finally, the use of systemic heparin adds to the risk of bleeding. Therefore, the partial thromboplastin time should be monitored carefully when concomitant heparin therapy is given.
Patients may experience dramatic worsening of their ischemic symptoms during the fibrinolytic infusion. This “trashing” of the extremity is usually due to fragmentation of the thrombus and distal embolization. Generally, the symptoms will resolve rapidly by continuing the fibrinolytic infusion and treating the patient with narcotic analgesia. Surgery usually is ineffective for such embolization. However, if significant clinical deterioration persists longer than an hour, angiography should be repeated, and emergent surgery should be considered.
Pericatheter thrombosis is an occasional problem that usually is caused by poor blood flow along the path of the catheter. To prevent this, the intravascular length of catheter must be kept to a minimum, and the smallest possible catheter system should be used. Pericatheter thrombosis can be minimized by using low-dose systemic heparin.