INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Invasive pulmonary aspergillosis (IPA) is a life-threatening infectious complication in immunocompromised patients and is a consequence of impaired defense function of neutrophils and macrophages. Intact macrophages inhibit the germination of Aspergillus conidia into hyphae, and neutrophils are directed against Aspergillus hyphae (1). In immunocompetent patients with normal function of macrophages and neutrophils, pulmonary infection with airborne Aspergillus sporozoites may lead to development of aspergilloma in preformed pulmonary cavities. Immunosuppressed patients are at high risk to develop invasive pulmonary aspergillosis with angioinvasiveness and tissue destruction by Aaspergillus hyphae.

IPA has the highest incidence in patients with hematologic malignancies after treatment with high-dose chemotherapy or bone marrow transplantation (BMT) and is associated with a high mortality. The risk for developing IPA is directly related to the duration of the neutropenic phase. After BMT the estimated risk of developing IPA is 4 to 6% (2). The overall incidence of IPA seems to have decreased over the past years because of new antifungal treatment strategies and room air filters, but there is still a 55 to 80% mortality rate in patients developing IPA (3, 5). The clinical diagnosis of IPA is based on antibiotic-resistant fever and typical radiological signs. Serological tests are of high specificity, but have a low sensitivity (6). In IPA the diagnostic yield of bronchoscopy with bronchoalveolar lavage (BAL) varies between 21% and 83% (6).

Current treatment for IPA is unsatisfactory, and the best strategy is a matter of debate. Medical treatment is based on amphotericin B in a dosage of 0.5-1.5 mg/kg bodyweight per day, or at higher dosage in a lipid-bounded formula. Combination therapy with other antifungal drugs such as flucytosine or itraconazole is optional. Despite a high in vitro activity against Aspergillus species of these compounds, only half of the patients respond to treatment.

In neutropenic patients receiving amphotericin B in therapeutic dosage for at least 14 d, the reported response rate is 33 to 54% (5, 11). However, the estimated overall mortality of IPA under medical treatment is still very high, and the outcome remains poor even with combination therapy (5, 11).

The prophylactic use of antifungal drugs such as itraconazole, fluconazole, and low-dose intravenous, or inhaled amphotericin B may improve results, but this is still under evaluation (14, 15). The use of high-dose intravenous amphotericin B may be limited by the side effects of this compound, mainly on renal function. Reports initially described emergency lung resection for severe hemoptysis as a complication of IPA. Few series of patients undergoing lung resection for treatment of IPA in immunocompromised patients have been published (10, 16). Fears to perform surgery in neutropenic and thrombopenic patients stem from the risk of uncontrollable bleeding and infection. Therefore the benefit of lung resection for IPA in hematologic patients is still under discussion (18, 19).

This study summarizes a single-center cohort series of neutropenic patients who underwent lung resection for IPA. It describes the risk and benefit of lung resection for IPA including signs of clinical presentation, diagnostic workup, perioperative complications, recurrence of IPA, and long-term follow-up.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Between 1983 and 1997, 65 hematologic patients undergoing stem cell transplantation or high-dose chemotherapy were treated for either proven or highly suspected IPA. Twenty-seven consecutive patients (16 male, 11 female; median age 39 yr, range 8 to 68 yr) underwent lung resection for suspected IPA. All patients were treated for underlying hematologic diseases; four acute lymphoblastic leukemia (ALL), eleven acute myeloid leukemia (AML), three chronic myeloid leukemia (CML), one myelodysplastic syndrome (MDS), one non-Hodgkin's-lymphoma (NHL), and seven aplastic anemia (AA). Six of seven patients with AA were treated with antilymphocyte-globulin. Seventeen patients had received high-dose chemotherapy. Two patients (CML and AA, respectively) underwent allogeneic BMT. One patient (CML) underwent allogeneic and another patient (NHL) autologous peripheral blood stem cell transplantation (PBSCT) (Table 1).

IPA was suspected if localized infiltrates developed on chest X-ray/ thoracic computed tomography (CT) scan in addition to antibiotic resistant fever in neutropenic patients. CT scan was performed in 22 patients with 8-mm slice thickness. In 18 patients, additional 2-mm slices with a high-resolution algorithm (HRCT) were performed. A surgical procedure had been chosen if patients showed localized pulmonary involvement of not more than two nodules at the same side.

Duration of symptoms, duration of neutropenia (neutrophil counts below 0.5 × 10⁹/L), antifungal and antibiotic therapy were recorded. Results of fungal culture of blood, sputum, tracheal secretions, and BAL were analyzed. Bronchoscopy with BAL was performed in 14 patients. Additionally, the following data were collected: preoperative hemoglobin, leukocyte, neutrophil, and platelet count; duration of surgery; intraoperative blood loss; duration of stay in the intensive care unit; duration of chest tube drainage; substitution with erythrocyte packs and platelet packs until 7 d postoperatively. When preoperative platelet count was below 20 × 10⁹/L surgery was performed under current substitution of platelets. Histological results were reviewed in respect to fungal infection and angioinvasion of the fungal hyphae. Postoperative complications, fungal recurrence, and the overall outcome were analyzed up to 1 yr after the operation.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Between 1983 and 1997, 27 patients with hematologic disorders underwent lung resection for treatment of suspected IPA. Patient summary is shown in Table 1.

Preoperative Clinical and Laboratory Findings

Symptoms. Twenty-two patients (81.5%) developed fever with a median duration of 8.5 d (4 to 37 d). Histology of lung resection confirmed invasive fungal infection in 20 of 22 (91%) patients with fever. Further clinical symptoms were thoracic pain (seven patients) and cough (three patients). One patient was on mechanical ventilation prior to surgery because of respiratory failure. Hemoptysis did not occur in our patients.

Laboratory findings. All patients had been neutropenic prior to surgery with a median duration of 17.5 d (2 to 187 d). Twenty-one patients (78%) were neutropenic when lung resection was performed. At day of surgery hemoglobin was 10.3 g% (7.3 to 12.2 g%), and platelet count was 35 × 10⁹/L (5-354 × 10⁹/L). Total white blood cell count was 0.6 × 10⁹/L (0.07-13.6 × 10⁹/L) with a neutrophil count of 0.05 × 10⁹/L (0.002-7.46 × 10⁹/L). Values are shown as median (range).

Radiology. In all patients chest X-rays showed localized infiltrates. Thoracic CT scan was perfomed in 22 cases, and solitary lesions were found in 12 patients. Four patients presented with multiple lesions in more than one lobe. In one patient there were multiple lesions restricted to one lobe. Lobar or segmental consolidation was found in five patients. Four of the five patients without histological evidence of invasive fungal infection presented with extended lobar consolidations, and HRCT of the fifth patient showed peripheral subsegmental consolidation. A classic macronodular pattern was found in 14 patients; 10 showed a typical halo sign. Cavitation with air crescent sign was found only in two patients.

Further diagnostic procedures. Fungal culture of the sputum showed a positive result of fungal infection before surgery in only one patient. The sputum culture was positive for Mucor. Pulmonary mucormycosis was confirmed in the resected lung tissue. Preoperative bronchoscopy with BAL was performed in 14 patients (52%) receiving antifungal agents at the time of bronchoscopy. In all 14 patients, cultures of BAL were negative for mold infection.

Antifungal and antibiotic therapy. The standard procedure for infectious complications in neutropenic patients includes routine administration of antifungal treatment to patients not responding to initial broad-spectrum antibiotics. Therefore 21 patients (78%) received antifungal treatment with a median duration of 11 d (range, 1 to 39 d) before surgery. Nineteen patients received amphotericin B. Three of these patients were treated with amphotericin B in combination with flucytosine or variconazole. Two patients received single-compound treatment with either flucytosine or itraconazole. Twenty-five patients (92.5%) were receiving antibiotic treatment for at least 7 d prior to surgery.

Operative Procedure and Perioperative Management

In all 27 patients, surgery was performed as a diagnostic and therapeutic procedure. Twenty of these patients underwent lobectomy, and seven wedge resection, using a video-assisted thoracoscopic surgical (VATS) approach in two cases.

Twenty-six patients (96%) were extubated immediately after surgery. One patient was ventilated until she died at postoperative Day 10 because of bacterial sepsis.

Duration of surgery varied from 50 to 210 min with a median duration of 120 min. Median intraoperative blood loss was 300 ml (range, < 50 to 1,000 ml). All patients stayed postoperatively overnight in the intensive care unit (median stay, 1 d, range 0 to 10 d).

All patients required chest tube drainage after surgery with a median duration of 2 d (range, 1 to 27 d). Substitution of erythrocytes consisted of a median of two packs on the first postoperative day and an additional three packs from Day 2 to 7. A median of 2 platelet concentrates per patient were transfused within the first week after surgery. Four patients received granulocyte transfusions and five patients were treated with recombinant granulocyte growth factor.

Histology

Invasive fungal infection could be confirmed histologically in 22 of 27 specimen (81.5%). Twenty showed typical features of IPA with one to three yellowish-brown or brick red nodules ranging from 1.0 to 7.5 cm in diameter. In two patients, invasive mucormycosis was found. Angioinvasiveness was histologically documented in all but one case. In this patient, hyphae were found in a bronchiolus with invasion of the neighboring tissue. In five patients (18.5%) there was no fungal infection detectable in the removed lung tissue. Histological examination showed intimal fibrosis of the arteries without fresh thrombi or hyphae in two specimens, nonspecific intra-alveolar fibrosing alveolitis in two specimens, and a bacterial abscess in the fifth case.

Postoperative Complications

Surgery-related complications. Four patients required prolonged pleural drainage due to persistent air leak, recurrent pneumothorax, and hematothorax. One of the two patients with pneumothorax suffered from pleural aspergillosis and eventually died at postoperative Day 56. One patient underwent reoperation at 42 d because of bronchial dehiscence without any signs of Aspergillus infection. She recovered afterwards without further surgical or infectious complications. Minor surgical complications included four pleural effusions without need of drainage, and two seromata at the site of thoracotomy, which resolved spontaneously (Table 2). The amount of perioperative blood loss did not correlate with the preoperative platelet count.

Recurrent fungal infection. All patients with histologically proven aspergillosis received antifungal treatment with amphotericin B immediately after surgery. Antifungal maintenance therapy was continued at least 6 mo postoperatively with oral itraconazole (1,200 mg daily). Two patients with mucormycosis were treated with variconazole (UK 109).

Nineteen of 22 patients (86%) cleared their mold infection. Only three of the 22 patients with histologically proven fungal infection suffered from ongoing or recurrent Aspergillus infection. One patient developed an aspergilloma of the orbita suspected from the CT scan and died with relapsing non-Hodgkin's lymphoma 30 d after lung resection. Histological confirmation of orbital aspergillosis could not be obtained because autopsy was rejected by the relatives. The second patient with underlying aplastic anemia developed a renal aspergilloma 3 wk after surgery. After the nephrectomy she recovered from fungal infection, but died 4 mo later with bacterial sepsis occurring during persistent cytopenia. The third patient developed pleural aspergillosis, requiring chest tube drainage because of a pneumothorax. This patient died with persistent fungal infection despite antifungal treatment, suffering from prolonged aplasia after chemotherapy for CML. The overall fungus-related mortality was 9% (2 of 22 patients).

Stem cell transplantation. After lung resection, seven of the 22 patients (32%) with histologically proven fungal infection underwent BMT or PBSCT for further treatment of their hematologic malignancy. Four patients underwent allogeneic transplantation (1 BMT, 3 PBSCT), and three patients autologous PBSCT. None of these patients developed recurrence of Aspergillus infection.

Follow-up

Survival. Two patients died within the first 30 d after surgery; one patient at Day 21 with recurrence of the NHL and suspicious orbital aspergilloma and the second patient at Day 10 because of bacterial sepsis after cholecystectomy for necrotizing cholecystitis without signs of fungal infection. The overall 90-d mortality was 23% (6 of 27 patients). There were four deaths in the second and third month after surgery (Table 1). One patient died at Day 56 with pleural aspergillosis and pneumothorax. Another patient died at Day 60 because of cerebral hemorrhage during refractory thrombocytopenia. The third patient died of relapsing AML 2 mo after lung resection. The fourth patient developed severe graft versus host disease (GvHD) complicated by bacterial infection and died 3 mo postoperatively.

There was no death due to severe infection, uncontrolled bleeding, or other complications that could be attributed to surgery. Only the death of the patient with pleural aspergillosis might be related to surgery because of the recurrent fungal infection after insertion of the chest tube for postoperative pneumothorax. Within the first year after surgery an additional five patients died with recurrence of the underlying hematological malignancy, and one patient died with bacterial pneumonia during persistent cytopenia. None of these patients suffered from recurrent fungal infection. Overall, 15 of all 27 patients (55.5%) and 12 of 22 patients with proven fungal infection (54.5%), respectively, are currently alive.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Invasive pulmonary aspergillosis is a life-threatening infection occurring mainly in neutropenic patients and is associated with a mortality rate of 40% to 80% (5). Standard medical therapy with long-term administration of amphotericin B also has a high mortality. Lung resection is an alternative approach for treatment of localized IPA. In our series of patients undergoing lung resection for suspected IPA, surgery was associated with a low incidence of major complications, and fungal relapse occurred in 14% of the patients.

The diagnosis of IPA is suspected if antibiotic-resistant fever develops in neutropenic patients and if there are newly developed localized infiltrates on chest X-ray. Thoracic CT scan findings of macronodules, halo signs, and cavitary lesions with air crescent sign are highly indicative of IPA (10, 22). Additionally, if neutropenic patients suffer from thoracic pain or hemoptysis, IPA is very likely to be present. In our series, 20% of patients suffered from thoracic pain.

Detection of Aspergillus species in the sputum has a high specificity but a low sensitivity for the diagnosis of IPA (6, 7, 9). The diagnostic yield of fungal culture and microscopy from BAL or bronchial washings varies considerably in the literature, with an overall sensitivity of 32% and specificity of 99.7% (7). Some investigators describe a higher sensitivity of bronchial washings compared with BAL (6). The use of distinct fungal culture media is maintained to be superior to routine bacterial culture media in detecting mold infection (9). The overall preoperative diagnostic yield of cell culture and microscopy to detect invasive pulmonary fungal infection in patients undergoing lung resection for IPA reported in the literature is approximately 36%, but varied from 5% to 100% (Table 3). In most of these studies BAL has been performed before antifungal treatment was initiated. In the series of Moreau and coworkers (18), patients were only selected for lung resection if the diagnosis of pulmonary fungal infection was confirmed preoperatively by a positive fungal culture of BAL. Therefore, resection was performed only in a selected group of patients.

In our study, 14 of 22 patients with histologically proven pulmonary fungal infection underwent BAL, but no fungi could be cultured from BAL. However, all of these patients received antifungal treatment at the time of bronchoscopy, which might explain the low diagnostic yield of fungal culture in BAL in our study. The decision to perform surgery in our patients was based on clinical findings and radiological findings typical for localized IPA. In 81% of the patients, invasive fungal infection was confirmed by histology. Therefore we recommend lung resection also in patients without positive fungal culture, when the diagnosis of IPA is suspected clinically and radiologically.

In five patients, fungal infection was not found in histological examination; four of these patients (embolic infarction; localized fibrosis) had received antifungal treatment preoperatively. Based on the histological results, antifungal treatment was discontinued. Therefore, surgery contributed in a diagnostic manner to the clinical management of these patients.

Surgical treatment of IPA was first established as an emergency procedure for severe hemoptysis. There are considerable fears of performing thoracotomy in pancytopenic patients with hematologic diseases because of uncontrollable bleeding or infection. Therefore some groups performed lobectomy at the time when normal blood cell counts were achieved (17, 18). In contrast, most of our patients were neutropenic and thrombopenic at the time of surgery. Median intraoperative blood was less than 500 ml, and only one patient required insertion of a chest tube for hematothorax. Infectious complications following surgery were also rare in our patients. One patient died with bacterial sepsis on postoperative Day 10 despite cholecystectomy for necrotizing cholecystitis. There were no other cases of bacterial infection postoperatively. Therefore, neither agranulocytosis nor thrombopenia are contraindications for thoracic surgery in these patients. One of our patients needed reoperation for bronchial dehiscence with no evidence of fungal infection at the site of the anastomosis despite treatment with steroids and cyclosporin A for GvHD. After lung resection for IPA, the overall incidence of major surgical complications reported in the literature is surprisingly low (Table 3). Minor surgical complications included prolonged air leak, pleural effusion without need for drainage, and seroma, and could easily be managed in our patients.

In the literature, the 30-d mortality of patients undergoing lung resection was 17.2%, ranging from 0% to 54% (Table 3). Compared with the results of only medical treatment for IPA, a 30-d survival rate above 80% after lung resection is a strong argument supporting a surgical approach. However, our study can not answer the question of how many patients would have survived under medical treatment without surgery. A randomized trial of medical versus medical combined with surgical treatment would clarify this unsolved question. Because the yield of preoperative diagnostic procedures for histologically proven IPA is low, diagnostic open lung biopsy or VATS biopsy would have to be performed before randomization. However, in our view performing thoracotomy without removing the whole IPA lesion is unethical.

In one series (20) four of five deaths after lung resection occurred in the subgroup of patients with IPA after BMT. Mortality might be higher in allogeneic stem cell recipients than in patients after high-dose chemotherapy because of persistent immunosuppression with cyclosporin A and steroids, and because of concomitant GvHD. However, the number of allogeneic BMT or PBSCT recipients undergoing lung resections for IPA reported in the literature is very low (Table 3).

Persistent or recurrent fungal infection occurred in three of our patients. In one patient, invasive aspergillosis of the orbita was suggested by CT scan but was not histologically proven. The overall incidence of fungal relapse after lung resection reported in the literature varied from no case of fungal relapse in small series (17, 18) to 14% in our study (Table 3), presenting with pulmonary (n = 2) as well as extrapulmonary infection in kidney (n = 1), orbita (n = 1), pleura (n = 2), and as disseminated disease (n = 1).

Patients with localized IPA and therefore qualifying for lung resection may have a better prognosis compared with patients with disseminated disease. However, concerning the high percentage of angioinvasiveness found in histological examination despite previous medical treatment, many of our patients may have developed disseminated IPA, systemic fungal infection, or severe hemoptysis without surgical treatment. Lung resection allowed BMT or PBSCT in seven of our patients without recurrence of IPA. Overall, patient survival after lung resection was mainly influenced by the relapse of the underlying hematologic disease, and not by the surgical procedure.

In conclusion, lung resection for suspected IPA in neutropenic patients has a low incidence of postoperative surgical and infectious complications. The incidence of postoperative fungal relapse is low even in cases with subsequent chemotherapy or stem cell transplantation. Based on the presented data we strongly recommend lung resection for clinically suspected IPA in neutropenic patients.