INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cystic fibrosis (CF), with an incidence of ~ 1 in 2,500 live births, is the most common inherited genetic disease among Caucasians (1). In recent years the median survival age of patients with CF has increased to 31.3 yr (2). As patients grow older, a number of complications of CF lung disease may occur. One of the more serious life-threatening complications associated with CF lung disease is hemoptysis, which may range in severity from minor to massive bleeding (3). The presence of occasional minor blood-streaking in the sputum is common in patients with CF, and although not acutely life-threatening, recurrent episodes of minor hemoptysis can preclude effective airway clearance maneuvers and adversely affect a patient's quality of life (4). Major hemoptysis (bleeding greater than 240 ml/24 h) has a yearly incidence of 1% and may carry a high mortality, particularly with conservative therapies, as in the era prior to bronchial artery embolization (BAE) (5). Although later studies suggest that conservative therapy may be a reasonable approach (6), most physicians and patients currently opt for therapies that result in immediate control of bleeding, given the potential for a poor outcome when bleeding is massive. Although no controlled studies have been performed to demonstrate its efficacy, the technique of BAE for hemoptysis in CF has become accepted in recent years (3, 4, 7).

Major hemoptysis usually originates from the systemic arterial supply to the lung. The chronic inflammation present in the lungs of patients with CF results in hypertrophy of the bronchial arteries, together with new vessel formation (angiogenesis). This hypervascularity, coupled with extensively damaged pulmonary parenchyma, results in a propensity toward hemorrhage, particularly during an exacerbation of pulmonary infection. Major hemoptysis can lead to asphyxiation and airway obstruction, shock, and exsanguination, and should therefore be treated promptly. Episodes of hemoptysis may be managed by several approaches, depending on the urgency of the situation, ranging from medical management and BAE to surgery. We report our experience from 1987 to 1997 at the University of North Carolina in the treatment and follow-up of 18 patients with CF lung disease who underwent BAE for hemoptysis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Database Review Methods

A search of the Vascular and Interventional Radiology sectional database from 1987 to 1997 identified 18 patients with CF who underwent 36 BAE procedures for the control of hemoptysis, performed during 29 separate hospitalizations. A retrospective review of all available medical records and radiographs was conducted. The data obtained included lung function testing at the time of the procedure. We used the FEV₁ to assess the severity of lung disease, along the guidelines published by the American Thoracic Society (8). Therefore, for the purposes of the study, the clinical severity of pulmonary disease was assessed by the degree of FEV₁ impairment as follows: Mild (FEV₁, 65 to 79% predicted), moderate (FEV₁, 50 to 64% predicted), severe (FEV₁, 35 to 49% predicted), or very severe (FEV₁ < 35% predicted). If available from the medical record, the patient's symptoms at the time of the hemoptysis with respect to localization of the bleed were noted. The severity of the hemoptysis was characterized by estimating the amount and frequency of bleeding, also as noted in the medical record. Bleeding was considered major when there was greater than 240 ml of hemoptysis within a 24-h period of greater than 100 ml of hemoptysis per day persisting for 3 d or more (3). Sputum microbiology at the time of the hemoptysis was reviewed in all cases, including the antibiotic sensitivity of any bacteria cultured. Multidrug resistance was defined as resistance to all agents in two of the following classes of antibiotics: Class I, -lactams (including ceftazidime, imipenem, and aztreonam); Class II, aminoglycosides (specifically tobramycin); Class III, quinolones (ciprofloxacin) (9). The results of any bronchoscopies performed were noted and reviewed. Chest radiographs and any computed tomograms (CT) performed were reviewed in a blinded fashion. Attempts were made during the review to localize the site of bleeding by careful analysis of the preprocedure radiologic studies and the patient's symptoms (for example, a sensation of gurgling in one specific area). With respect to chest radiographs, areas with a focal infiltrate were designated as the most likely site of bleeding. Also, if a patient had asymmetric chronic disease involvement, the lung most severely affected was designated as the most likely source of bleeding. The observations made during review of the bronchoscopies, radiologic studies, and presenting symptoms were correlated with the ultimate results of the BAE procedures. Symptoms and investigations to localize the site of bleeding were defined as being accurate when embolization of arteries supplying the section of lung in question resulted in cessation of bleeding. Films and reports from the BAE procedures were reviewed for bronchial artery anatomy, vessels embolized, embolic materials used, and complications. Follow-up after the BAE procedure was also reviewed from the records. Particular note was made of recurrent hemoptysis, requirement for repeat BAE procedures, surgical intervention (including lobectomy and transplant), or death.

Initial Medical Management

All patients were admitted to the University of North Carolina Hospital Cystic Fibrosis Center under the care of the Pulmonary and Critical Care Medicine service. Episodes of hemoptysis were treated according to the guidelines presented in the Cystic Fibrosis Foundation Consensus Conference Report on Pulmonary Complications of Cystic Fibrosis (3). Sputum cultures were obtained and patients were given antibiotics intravenously if a pulmonary exacerbation was suspected. The choice of therapy was guided by the most recent sputum microbiology data, usually to include two synergistic antipseudomonal antibiotics. A complete blood count with platelet count was performed, and blood was typed and cross-matched in the event transfusion was required. Liver function tests and coagulation studies were obtained, and any drugs that might interfere with coagulation were discontinued. Any coagulation defects were corrected with vitamin K or fresh frozen plasma as indicated. In the case of massive hemoptysis, patients were generally admitted to the intensive care unit and nursed with the bleeding lung in the dependent position to avoid contamination of the nonbleeding lung with blood. Chest physical therapy was discontinued until bleeding had ceased for 24 to 48 h. A member of the Vascular and Interventional Radiology Service evaluated patients on admission with massive hemoptysis. When it was determined that BAE was clinically indicated, generally when bleeding persisted despite medical management, or if the bleeding was considered so urgent as to require immediate intervention, informed consent was obtained and the patient was transferred to the angiographic suite.

Embolization Procedure

Patients initially received an intramuscular injection of 50 to 75 mg demerol hydrochloride and 25 to 50 mg vistaril. Sedation was then titrated with intravenous doses of 25 mg fentanyl and 1 to 2 mg midazolam hydrochloride. Efforts were made to avoid respiratory compromise and to keep neurologic depression to a minimum so that the patient's status could be monitored throughout the procedure. All procedures were performed via a common femoral artery approach with the placement of a vascular sheath (6 Fr). For patients undergoing their first BAE procedure, selective bronchial artery catheterization was immediately performed. In patients who had prior BAE procedures, thoracic aortography and selective subclavian arteriograms were initially performed prior to selective catheterization. In patients with lower lobe disease, the inferior phrenic artery was also catheterized. Aortograms were performed with 5 or 6 Fr pigtail catheters, and selective bronchial artery catheterization was performed primarily with 4 or 5 Fr reverse-curved catheters and less commonly with Cobra or multipurpose-type catheters. The primary site of interest, as determined by patient history and findings of available radiographic and/or bronchoscopic results, was catheterized first. All abnormal vessels supplying the area of interest were embolized if technically possible. Nonbronchial systemic arterial collateral supply from the subclavian arteries was also approached via the femoral approach using H1H, multipurpose, or steamed-shaped catheters.

After successful bronchial artery catheterization, arteriograms were performed with digital subtraction technique and nonionic contrast material. A critical search for any spinal artery contribution was made. Oblique arteriograms were obtained for questionable midline arterial branches to differentiate true spinal arteries from tracheal and esophageal branches. If an adequate and stable catheter position could not be accomplished with the original catheter, a conventional catheter exchange for a more suitable shape/configuration or coaxial catheterization was performed. A variety of microcatheters has been used.

The primary embolic material used was polyvinyl alcohol (PVA) particles, 300 to 500 µm in size. Some larger vessels required particles as large as 2 mm. When large bronchial artery to pulmonary artery/ pulmonary vein shunts were identified, larger particles were used to avoid distal embolization. With enormous feeding vessels, large shunts, or when distal embolization was to be prevented, stainless steel or platinum coil embolization was performed. In the 36 procedures, PVA alone was used in 25 cases, gelfoam alone was used in three cases, and platinum-tipped metallic coils (2 and 3 mm) alone were used in two cases. PVA and coils were used in three cases, and PVA and gelfoam were used in one case. In one procedure, performed for recurrent bleeding several hours after an earlier embolization procedure, the vessel to be embolized could not be catheterized; therefore, no embolic material was used. Typically, 100 mg of the PVA particles were suspended in a 20-ml syringe with 10 ml of contrast and 10 ml of saline. This reservoir syringe was connected to a side of a three-way stopcock, which was connected to the catheter and to the delivery syringe. The delivery syringe was either a 1- or a 3-ml luer lock syringe. The particles would be vigorously resuspended by injection between the two syringes just prior to administration. All injections were monitored fluoroscopically. When the movement of the opacified embolic solution became pulsatile, the catheter was cleared with saline and contrast injection performed. When forward flow was virtually halted, no further particles were injected.

When all obvious branches supplying the region of interest were embolized, the procedure was terminated. A brief neurologic examination was intermittently performed throughout the procedure and at completion.

Successful imaging and embolization of a large, hypertrophied right bronchial artery are illustrated in Figure 1.

View larger version (145K): [in this window] [in a new window]   View larger version (121K): [in this window] [in a new window]   Figure 1.   (Top panel ) Angiogram of a hypertrophied right bronchial artery supplying the right upper lobe. (Bottom panel ) Angiogram of the same bronchial artery after embolization with polyvinyl alcohol particles.

Follow-up

BAE was considered to have successfully controlled the hemoptysis if the bleeding was clinically arrested by one or more embolization sessions during that hospital stay. After discharge from the hospital, patients were seen and evaluated in follow-up at the UNC Cystic Fibrosis Clinic. Recurrence was defined as hemoptysis after discharge significant enough to warrant repeat BAE or lobectomy.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Between 1987 and 1997, 18 patients had 29 separate hospitalizations for hemoptysis treated by BAE. This group of patients consisted of 11 men and seven women 17 to 36 yr of age (mean, 27 yr). Fifteen patients had one hospitalization, and one patient each had three, four, and seven separate hospitalizations during which BAE was performed. The mean FEV₁ at the time of embolization for the 18 patients was 40% of predicted (range, 13 to 84%). Using the criteria previously defined, three patients were categorized as having mild disease, three as having moderate disease, one as having severe disease, and 11 as having very severe disease. Sputum cultures demonstrated 16 patients with Pseudomonas aeruginosa, two with Staphylococcus aureus, two with Burkholderia cepacia, and one each with Stenotrophomonas maltophilia, Aspergillus fumigatus, and Burkholderia gladioli. The presence of nontuberculous mycobacteria (Mycobacterium avium complex) in the sputum was recorded on a repeated basis in one patient. Sputum cultures showed the presence of multi-drug-resistant organisms (see above for definitions) in nine patients (50%) (six with P. aeruginosa, one with S. aureus, one with B. gladioli, and one with B. cepacia). Admission coagulation studies were available in 27 presentations, with mild elevation of prothrombin time (PT) in eight (27%) (range, 12.9 to 13.5 s). One patient had a PT of 15.4 and a partial thromboplastin time (PTT) of 41.7, whereas another had a PT of 16.4 and a PTT of 33.4. All other PTs and PTTs were within normal limits.

Symptoms and investigations to localize the site of bleeding were defined as being accurate when embolization of vessels feeding the area of lung in question resulted in cessation of bleeding. On initial evaluation, the patients were able to symptomatically localize the site of bleeding in nine of 24 presentations (47%). In six of these nine, the chest radiography showed bilateral patchy disease and was unhelpful in localizing the site of bleeding. There were no cases where patients suggested the wrong side with respect to the site of bleeding.

We have classified the radiographic data according to presentation (n = 29), rather than by patient, because of the possibility for variability over time in the chest radiographs. The chest radiographs and/or radiographic reports were available for review in all presentations. As might be expected in patients with CF, the most common site of focal infiltrate or area of severe disease was the right upper lobe (n = 11, 38%), followed by right lower lobe (n = 2, 7%), left upper lobe (n = 1, 3%), right middle lobe (n = 1, 3%), and right middle lobe and right upper lobe (n = 1, 3%). The remaining 13 presentations (45%) demonstrated bilateral disease with no focal infiltrate. When a focal infiltrate or focal area of severe disease was present, this area was judged the most likely source for the bleeding. Using these criteria, the chest radiography suggested a site for the bleeding in 16 presentations (13 patients). When these data were compared with the actual site of bleeding (from the angiographic records), this site was found to be the source of bleeding in 13 (81%) of these 16 presentations. Using these criteria, the chest radiograph reasonably accurately localized the site of bleeding when focal disease was present. Chest CTs were obtained in three patients. The CT was useful in localizing the site of bleeding by delineating focal disease not readily apparent on the plain radiograph in two of the three patients. In one of these two, the chest radiograph was also able to suggest a site for the bleeding. In the third patient, the chest radiograph showed bilateral patchy disease, whereas the CT clearly demonstrated a more focal process in the left upper lobe.

Bronchoscopy was performed during seven of the hospitalizations and demonstrated active bleeding or blood confined to a certain area of lung in five patients. In all five, bleeding was seen in the right upper lobe. The chest radiograph showed right upper lobe disease in three of these patients. In the other two, the chest radiograph showed bilateral symmetric disease and was unhelpful in suggesting a site of bleeding. In one of the patients, selective bronchial washes were performed to absolutely confirm the site of the bleeding. This was done because the patient continued to have recurrent bleeding despite several BAE procedures in that area.

Of the 18 patients, 15 presented on a single occasion for embolization and one each presented three, four, and seven times. This resulted in a total of 29 separate hospitalizations. Nine of these were for chronic minor hemoptysis judged severe enough to warrant intervention, 18 were for major hemoptysis, and two were for chronic hemoptysis occurring during the month prior to admission. During these 29 hospitalizations, 23 required only one embolization session for initial control of hemoptysis, five required two sessions, and one required three sessions. Overall, the efficacy of BAE for initial control of hemoptysis was 75% (n = 22) after one session, 89% (n = 26) after two sessions, and 93% (n = 17) after three sessions. One patient, with very severe lung disease (FEV₁ = 24%), failed BAE after having undergone several prior BAE procedures at an outside institution. In this first session at UNC, he underwent BAE of an accessory bronchial artery arising from the aortic arch and a right inferior phrenic artery arising from the celiac trunk. The patient experienced hemoptysis of 150 ml several hours later and was taken back to the angiographic suite for a second procedure. At this time a large left bronchial artery arising from the aortic arch near the origin of the left subclavian artery was identified, but adequate catheter position for embolization could not be obtained. Hemoptysis continued, and the patient died 6 d later of refractory respiratory failure. A second patient failed BAE after presenting with 800 ml of hemoptysis in a 24-h period. The patient underwent BAE of a right intercostal artery supplying the right lung and a left bronchial artery. Two days later, while in the intensive care unit, he experienced an episode of massive hemoptysis associated with simultaneous bilateral pneumothoraces thought to be secondary to a ball-valve phenomenon created by blood within the airways. Autopsy revealed a laceration in the right lower lobe with hemorrhage into the right lower and the right upper lobes.

Most patients tolerated the procedure without complications. However, one patient experienced transient agitation, blurred vision, and short-term amnesia after BAE. In this patient branches of the thyrocervical and costocervical trunk were embolized. Another patient experienced transient confusion after the procedure. In this patient a right bronchial artery was embolized. In another patient, focal neurologic symptoms and signs were noted while angiography of the right subclavian artery was being performed. These resolved when the catheter was withdrawn. Although some symptoms may have been attributable to anesthesia, the procedure was likely to have been responsible. However, no residual abnormalities were identified in any patient 24 h after the procedure. In another patient, the formation of a subadventitial hematoma of the left bronchial artery was noted during the procedure, likely secondary to guide wire trauma. The BAE procedure proceeded uneventfully despite this complication.

Recurrence was defined as hemoptysis after discharge significant enough to warrant intervention. Of the 16 patients who had successful BAE procedures, 12 did not experience a documented recurrence after embolization. However, it should be noted that three of these patients were lost to follow-up. The average follow-up period for these patients was 10.4 mo prior to transplant (n = 3) or death from respiratory failure (n = 2).

Of the four patients who experienced recurrent bouts of hemoptysis, one experienced a solitary recurrence, one had two recurrent bleeds, one had three recurrent bleeds, and one experienced six recurrent bleeds. Nine recurrences were with major hemoptysis, whereas three were for minor hemoptysis. The mean time for recurrence ranged form 10 d to 38 mo, with an average time to recurrence of 12.1 mo. In nine of the 10 episodes of recurrent hemoptysis, the hemoptysis was successfully controlled with BAE. In the remaining one, the patient returned with recurrent massive hemoptysis 1 mo after a hospitalization during which three BAEs were performed. After discussion between the clinicians, radiologists, and surgeons, the patient underwent a right upper lobe lobectomy. At surgery, a plethora of transpleural vessels was noted supplying a heavily diseased right upper lobe, which would have been difficult to treat with embolotherapy (Figure 2). There was no recurrence of the bleeding after surgery. In another patient who presented with recurrent massive hemoptysis, the patient was noted to have a large tortuous right bronchial artery upon angiography. On review of the record of this patient, it became evident that this side had been embolized 8 yr previously with a coil. the vessel had apparently recanalized. It was reembolized with PVA as outlined above.

View larger version (137K): [in this window] [in a new window]   Figure 2.   Angiogram of the thyrocervical trunk in a patient who experienced massive hemoptysis. A plethora of transpleural collateral vessels is noted supplying the right upper lobe. Also, bronchial artery to pulmonary vein shunting is visible (arrow). These vessels were embolized, but the patient later required a right upper lobe lobectomy for control of recurrent bleeding.

The incidence of bleeding from nonbronchial systemic collateral vessels in cases of recurrent bleeding was 75% (n = 12 of 16 presentations). This includes those patients who had BAEs at outside institutions prior to their presentations at UNC. In contrast, in the 13 presentations where patients were undergoing their first embolization, the incidence of bleeding from systemic collateral vessels was only 8% (n = 1 of 13 presentations).

Fifteen patients were followed for a mean of 21.6 mo (range, 2 d to 91 mo) after embolization, with three patients lost to follow-up. Four patients died during the follow-up period at 2, 6, and 20 d, and at 36.5 mo after embolization. Three of these patients had very severe pulmonary disease, whereas the other patient had excellent lung function (FEV₁ = 84%). This patient died 2 d after embolization of massive hemoptysis with bilateral pneumothoraces as previously described. The patient who died 6 d after an embolization procedure had an unsuccessful procedure and died with uncontrolled hemoptysis and severe clinical disease (FEV₁ of 24% predicted prior to admission). The patient who died at 20 d had his hemorrhage controlled but died from respiratory failure. The final death was secondary to respiratory failure, despite control of bleeding. The time-scale of events and any other significant factors associated with each patient's follow-up period is presented in Figure 3.

View larger version (27K): [in this window] [in a new window]   Figure 3.   Graph illustrating the length of follow-up for each individual patient and the significant events that occurred during this time period.

One patient not included in this study underwent bronchial artery angiography after recently being treated at an outside hospital for an episode of massive hemoptysis. This episode was successfully treated with conservative therapy. However, the patient was admitted to UNC hospitals 1 mo later for a pulmonary exacerbation without hemoptysis. After a discussion involving radiologists, surgeons, and pulmonologists, it was decided to evaluate the patient's bronchial arteries for possible pathology. The angiogram revealed a bronchial artery with a common trunk supplying both lungs. Branches supplying the right upper lobe and left upper lobe were only slightly hypertrophied, whereas those supplying the left lung were normal. The interventional radiologist decided to forego embolization at that time since the patient was not actively bleeding and the vessels were not remarkably abnormal.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Recent literature supports the use of bronchial artery embolization for hemoptysis in cystic fibrosis (4, 7, 10). Prior to the development of this procedure, patients with massive hemoptysis were managed conservatively, or underwent emergency surgery (5, 6). The mortality in the earlier reports was quite high (5), although later data suggest that at least in the immediate time around bleeding, patients could survive despite expectoration of large amounts of blood (6). Comparisons involving these older data and more recent reports are very difficult. The patient population has changed as patients with CF survive longer, there is a more aggressive approach to treatment in general with the advent of lung transplantation, and multi-drug-resistant bacteria have become more prevalent (1, 13). Several recent reports support the current practice using BAE in hemoptysis in CF. A consensus conference convened by the Cystic Fibrosis Foundation examined the available data in relation to the management of hemoptysis (3). This panel of experts concluded that, although the data in relation to the use of BAE were not supported by rigorous scientific data, nonetheless the procedure did appear to be successful in the experience of the clinicians participating in the consensus conference, as well as suggested by case reports, published studies, and theoretical considerations derived from the literature.

The current study also demonstrates that BAE is a relatively safe and effective method for the treatment of hemoptysis in patients with CF. BAE was useful for hemoptysis of varying severity with generally successful results. The data from this study are useful for several reasons. It has allowed an examination of the patient population with massive hemoptysis requiring this procedure. The data has allowed a review of the safety and efficacy of the procedure in these patients. Finally, the data make possible some attempt at defining the clinical characteristics, which may guide the investigation and treatment to optimize outcome while minimizing risk.

The majority of the patients in this study had very severe pulmonary disease, defined for the purposes of the study as those patients with an FEV₁ less than 35% predicted (8). An episode of major hemoptysis in this population is clearly life-threatening, and efforts should be made to stop the bleeding early in its course. Also, mild hemoptysis may serve as a warning of impending major hemoptysis. Therefore, at our institution, we often elected to proceed with BAE directly, although intensive medical management may occasionally be successful (11). The criteria for embolization follow those set by the Cystic Fibrosis Consensus Committee on Pulmonary Complications of Cystic Fibrosis, and those of Cohen and colleagues (11): (1) major hemoptysis as previously defined, (2) mild or chronic hemoptysis that interferes with lifestyle, and (3) hemoptysis that interferes significantly with airway clearance maneuvers.

All patients receive standard medical therapy on admission, as previously outlined, pending a decision regarding the necessity for BAE. A main focus of this medical therapy is the treatment of pulmonary exacerbations. It is usually safer to withhold airway clearance physical therapy maneuvers within 24 to 48 h of a bleeding episode. However, dual antibiotic therapy to cover P. aeruginosa is usually reasonable. A good choice of drugs would be an aminoglycoside and a -lactam antibiotic, but probably not one of the semisynthetic penicillins because of the potential for platelet dysfunction secondary to this class of agents (14). If S. aureus is isolated from the sputum, it should also be treated (13).

It is interesting to note that patients in this study had a high incidence of colonization/infection with multi-drug-resistant bacteria. In addition, one patient was considered to have active infection with M. avium complex. The presence of these organisms in this patient population may simply be a marker of more severe disease. On the other hand, a causative relationship between the presence of resistant organisms and hemoptysis cannot be excluded. In addition, the presence of such microorganisms in these patients implies that the interventional radiology staff should take measures to minimize the risk of transmission when patients with CF are undergoing a BAE procedure, particularly when organisms such as B. cepacia are present in the sputum.

Malabsorption of fat-soluble vitamins, specifically vitamin K, can predispose patients with CF to coagulation abnormalities. Therefore, it is important to check the PT and PTT on all patients with CF admitted for hemoptysis. Interestingly, in this study, coagulation studies in these patients were for the most part not significantly abnormal. However, it is still important to address the possibility that coagulation defects may be present and treat such abnormalities as appropriate. Treatment with oral or parenteral vitamin K or fresh-frozen plasma may be considered in such circumstances. Another point of note is drug effect on the coagulation system, particularly platelet function. The penicillins, including the antipseudomonal semisynthetic penicillins, are noteworthy in this regard (14).

Localizing the site of the bleeding is helpful prior to the BAE procedure. Perhaps the most reliable method, as noted in this and in other studies, for localizing an area of bleeding is the patient's report of "gurgling" or discomfort in a specific area of the thorax (4, 11). In the current study, when a focal infiltrate or area of most severe disease was visible on the chest radiograph, its presence was fairly accurate in suggesting this area as the site of bleeding. The chest CT was performed in a minority of patients, but it was useful only in one patient for localizing hemorrhage. Therefore, the CT scan may be considered in those cases where the symptoms and the chest radiograph are not helpful. Bronchoscopy may also be useful to localize the site of bleeding, as in seven of the patients in our series. Although some earlier studies advocated the use of bronchoscopy as the primary method for localizing the site of bleeding, more recent literature has placed less importance on this modality (10, 17, 18). Certainly if there is a large amount of blood in the airways, flexible bronchoscopy plays a limited role. We suggest that bronchoscopy has a role similar to that of CT, i.e., that it be reserved for those patients where the symptoms and chest radiograph are not useful. Clearly, the decision to employ either study should be made on a case-by-case basis.

Some investigators advocate that in the setting of acute hemoptysis, all angiographically visible bronchial arteries to both lungs should be embolized (7, 10, 12). Others promote a more selective approach, with embolization being directed toward the lung most clinically suspected of being the origin of the hemoptysis (11). In this study, only unilateral embolizations were performed in seven of the 20 patients. As discussed previously, our approach is to pursue a more selective approach to BAE. Initial embolization is performed on the side most suspected as the source of the hemoptysis, based on the information as described in the preceding section. Only unilateral embolization was performed in the patients from our series who localized the symptoms to one side, and BAE was successful in arresting the bleeding in all of them. In the series from Cohen and colleagues (11), seven of 20 patients were able to localize symptoms to one lung. In the series from Tonkin and colleagues (4), five of the 11 patients were able to localize their symptoms consisting of a "gurgling" or "funny feeling" in one lung. In all 12 of these patients, a unilateral embolization was performed, and there was no further bleeding after the initial procedure. Including our patients, of these 21 cases where the patients could localize the bleeding site themselves, unilateral embolization resulted in control of the hemoptysis. Thus, if a patient can sufficiently localize the bleeding to one side or an area of one lung, it appears that targeted unilateral embolization will be successful in controlling the bleeding.

In those patients in whom no localizing signs or symptoms are present and diagnostic studies do not suggest one side over the other, all visible bronchial arteries are embolized. Only if the bronchial arteries are thought to be too small to account for the patient's hemoptysis do we proceed to extensive angiographic evaluation, including descending thoracic aortography and selective subclavian arteriograms. Alternately, this more extensive search for nonbronchial systemic collaterals is carried out if the patient continues to hemorrhage. Although this graduated approach to embolization in the setting of hemoptysis may require repeat procedures on a select number of patients, the risks associated with subclavian angiography and embolization of subclavian arterial branches significantly increases the overall risks of the procedure as manifested by the several episodes of transient neurologic complications in our patients (see below). As described previously in our report, cessation of the hemorrhage occurred in 81% of the cases where initial selective embolization was performed. This compares quite favorably with the 88% rate of bleeding control achieved with aggressive embolization of all visible bronchial artery supply in both lungs (12).

It is rare to identify frank arterial extravasation as the definite source of the hemorrhage during angiography for hemoptysis. In trying to determine the site of hemorrhage, the angiographer relies on indirect signs such as size of the vessel, degree of hypervascularity, intensity of the vascular blush, and any evidence of vascular shunting, all of which influence the decision as to whether or not to embolize the particular vessel. It is difficult to be sure that the vessel that is embolized is actually the source of hemorrhage, and theoretically every possible vessel contributing to bronchial artery supply should be embolized. However, each additional vessel that is catheterized and embolized increases the risk of the procedure. Additionally, searching for the multiple vessels that potentially can provide bronchial supply may be an exhausting prolonged procedure for both the patient and the radiologist. The contrast volumes may also become significant. In one series, the procedure was stopped when a total volume of 350 ml has been reached and procedure resumed in 24 to 48 h (11). At our institution, we embolize the most likely sources realizing that a majority of the patients will respond well, but also appreciating that a few patients may require a repeat procedure. If the initial embolization is unsuccessful, we repeat the procedure with a more exhaustive search for any other vessels that may be contributing to the bleeding.

The results of this study support the previous reports that BAE is effective in achieving initial control of hemoptysis (4, 7, 10). When data from this study and these other studies are pooled, BAE initially controlled the hemoptysis in 92 of 102 (90%) cases. The data from this study suggest that control of bleeding occasionally requires repeated sessions of BAE during a single hospitalization. In this setting, a patient can experience continued bleeding as a result of one or two scenarios. Either the patient continues to bleed from an inadequately embolized vessel or the patient is bleeding from a vessel not embolized during the initial BAE procedure. In all five cases requiring repeat embolization during a single hospitalization, the patient was bleeding from vessels not previously embolized. In three of these cases, bleeding was from nonbronchial systemic collateral vessels not previously embolized. Additional bronchial arteries were responsible for the persistent bleeding in the remaining two patients. In one patient who had had several previous BAE procedures, a large right bronchial artery was noted at the most recent angiogram. This vessel had been embolized 8 yr previously with a platinum coil and had evidently recanalized. These data highlight the importance of searching carefully for all possible bleeding sites during the angiogram and embolizing these vessels if necessary. In addition, even if a patient has previously had BAE to a particular vessel with a coil, repeat angiograms should include an evaluation of the possibility of recanalization and rebleeding from the same vessel.

The increased incidence of nonbronchial systemic collateral vessels supplying the lung in inflammatory lung diseases such as CF has been noted previously (19, 20). In our experience, patients with recurrent bouts of hemoptysis were 10 times more likely to be bleeding from nonbronchial systemic collateral vessels than were patients undergoing their first BAE. The one patient in this study with systemic collateral vessels on initial presentation had an area of severe disease in the apex of his right lung with marked pleural thickening, which may have played a role in the formation of transpleural collateral vessels. This patient had several BAE procedures, and eventually required a right upper lobectomy to control the bleeding. Thus, these data suggest that patients presenting with recurrent hemoptysis with previous BAE, or those refractory to repeated BAE, should have a careful search performed during the angiography for nonbronchial systemic vessels. The reason for the high incidence of bleeding from nonbronchial systemic collateral vessels in patients with recurrent hemoptysis and previous BAE is unclear. Patients with recurrent hemoptysis may have more severe disease, with increased nonbronchial systemic arterial supply to the lung, and thus increased frequency of recurrent bleeding. The severity of disease may therefore be related to the development of new blood vessels. Alternatively, recurrent BAE procedures may play a role in the formation of nonbronchial systemic collateral vessels. If the latter statement is true, perhaps a reevaluation of the use of BAE for less severe episodes of hemoptysis might be necessary.

Embolization of bronchial arteries has been associated with numerous complications, including severe chest pain requiring analgesia, dysphagia, bronchial necrosis, bowel ischemia, and paraplegia (4, 21). However, more recent reports have demonstrated that, with the use of nonionic contrast and digital subtraction angiography in the hands of a skilled interventional radiologist, the actual incidence of these complications is quite low (11, 12, 18). There were no complications of note in the series of Cohen and colleagues (11), and the complications in the other series occurred in ~ 10 to 30% of patients, and were generally transient. Of all the complications, transient chest pain and dysphagia appear to be the most reported, as much as 50% of patients having chest pain in one series (12). The data from the current series also supports the notion that in experienced hands, the complication rate is very low, with transient neurologic events predominating.

Perhaps the most feared complication is paraplegia resulting from embolization of a spinal artery. The angiographer should perform a diligent search for such vessels, but the presence of such a vessel should not be viewed as an absolute contraindication to BAE (3, 4, 25). In the current study, one spinal artery was identified, but BAE proceeded without event. Three patients did experience neurologic deficiencies, but these were transient, with no permanent focal deficits. Although the data from this study support the notion that BAE as currently performed is associated with few side effects, nonetheless, the possibility of such complications is worrisome. Each situation must be evaluated in terms of the importance of bleeding control versus possible neurologic side effects during the procedure. With the proper care, the risks of complications are outweighed by the potential benefits of immediate cessation of bleeding.

An algorithm for use in the management of patients with CF presenting with hemoptysis is outlined in Figure 4. In patients presenting with hemoptysis who have not had a previous bronchial artery embolization procedure, we suggest first attempting to localize the bleeding with the patient's history and a chest radiograph, paying particular attention to the area most diseased. In some cases, a bronchoscopy or CT may be useful adjunctive investigations. If these studies are able to localize an area of bleeding or severe disease, the BAE begins with a direct search for bronchial arteries supplying this region. Any abnormal vessels supplying this site are embolized. If the likely site of hemoptysis cannot be localized, then all abnormal vessels supplying both lungs should be embolized. In patients with recurrent bleeding after prior BAE, particularly those who have had several BAE procedures in the past, we attempt to isolate the obvious site of bleeding. Once the site is identified, a vigilant search for all vessels supplying the area, including nonbronchial systemic collaterals, is performed and all abnormal vessels are embolized. If the site cannot be localized, then all abnormal vessels supplying both lungs are embolized. If an initial BAE procedure fails to adequately control hemoptysis, then other methods for localizing the site of bleeding (bronchoscopy, with selective sequential bronchial washes, or chest CT) may be employed, together with a more vigilant search for abnormal vessels at angiography.

View larger version (35K): [in this window] [in a new window]   Figure 4.   Proposed algorithm for the treatment of hemoptysis in patients with cystic fibrosis.

In summary, this study supports previous data showing BAE to be a relatively safe and effective method of treating hemoptysis in patients with CF. In this series, patients requiring BAE have been those with severe lung disease and with a high incidence of multi-drug-resistant bacteria in the sputum. A chest radiograph showing an area of focal severe disease is a good predictor of the site of bleeding. Despite a moderately high recurrence rate, BAE can provide relief of symptoms for a significant period of time. BAE should be employed in cases of major hemoptysis when conservative therapy fails to provide prompt relief. BAE can also play a significant role in improving the quality of life in patients with minor or recurrent hemoptysis, as previously suggested (4). A careful search for all abnormal vessels, including previously embolized vessels and nonbronchial systemic collateral vessels, should be made in patients undergoing repeat BAE. Clinicians who treat patients with CF should familiarize themselves with BAE. Evaluation of these patients should be made by an experienced team of physicians, including a pulmonologist experienced in CF, together with an experienced interventional radiologist and a thoracic surgeon.