Subpleural Honeycombing on High Resolution Computed Tomography is Risk Factor for Fatal Pneumonitis

Subpleural Honeycombing on High Resolution Computed Tomography is Risk Factor for Fatal Pneumonitis Hiroyuki Ito, MD, Haruhiko Nakayama, MD, Masahiro Tsuboi, MD, MD, Yoichi Kameda, MD, Tomoyuki Yokose, MD, Chikako Hasegawa, MD, and Kouzo Yamada, MD Divisions of Thoracic Surgery, Thoracic Oncology, and Pathology, Kanagawa Cancer Center, Yokohama, Japan Background. Postoperative interstitial pneumonitis is a life-threatening complication after lung cancer surgery. We conducted this study to identify risk factors for postoperative interstitial pneumonitis in patients with no clinical evidence of interstitial lung disease. Methods. We retrospectively studied patients who underwent lung cancer resection. The characteristics of patients were analyzed by reviewing their clinical and surgical records and preoperative chest high-resolution computed tomographic scans. Postoperative interstitial pneumonitis was defined as acute severe hypoxemia accompanied by radiographic diffuse interstitial infiltrates of the lung with no apparent cause within a few weeks after surgery. Results. From 2002 through 2005, 651 patients were evaluated, operated on, and managed by the same team. Postoperative interstitial pneumonia developed in 7 patients (7 of 651, 1.1%). Five of these patients had local, but not diffuse, dorsal subpleural honeycombing occupying three or more segments in both lower lobes on highresolution computed tomography (CT honeycombing). During the same period, 46 patients had CT honeycombing. The incidence of postoperative interstitial pneumonia was 10.9% (5 of 46) among patients with CT honeycombing and 0.3% (2 of 605) among those without CT honeycombing. Four of the 7 (57%) patients with postoperative interstitial pneumonia died of respiratory failure. Mortality among the patients who had postoperative interstitial pneumonia as well as CT honeycombing was 80% (4 of 5); in contrast, none of the patients without CT honeycombing died. Multivariate analyses showed that the presence of CT honeycombing and prolonged operation time were significant risk factors. Conclusions. Subpleural honeycombing on high-resolution computed tomography is a significant predictor of postoperative interstitial pneumonia in asymptomatic patients who undergo resection for lung cancer. (Ann Thorac Surg 2011;91:874 9) 2011 by The Society of Thoracic Surgeons Recent improvements in patient-selection criteria, operative techniques, and postoperative management have contributed to lower mortality from lung cancer surgery. In 1983 the Lung Cancer Study Group reported a mortality rate of 3.7% [1]. In 1999 an analysis of data from a Japanese lung cancer registry estimated that mortality had decreased to 0.9% [2]. Respiratory complications frequently occur after lung cancer resection and about half of all postoperative deaths are attributed to interstitial pneumonia (IP) [3]. Of course, apparent interstitial lung disease (ILD), including idiopathic pulmonary fibrosis (IPF), is an important risk factor for postoperative IP as well as for poor outcomes in patients with lung cancer [4, 5]. The treatment strategy for patients who have lung cancer with apparent ILD must therefore be decided more judiciously than usual [5, 6]. On the other hand, postoperative IP sometimes occurs unexpectedly in patients without apparent ILD who have Accepted for publication Oct 22, 2010. Address correspondence to Dr Ito, Division of Thoracic Surgery, Kanagawa Cancer Center, 1-1-2 Nakao, Asahi-ku, Yokohama, Japan 241-0815; e-mail: h-ito@kcch.jp. acute lung injury-acute respiratory distress syndrome (ALI/ARDS) accompanied by bilateral reticular infiltrations on chest radiography. In many such patients, ILD is not diagnosed preoperatively because their clinical signs and symptoms are atypical of ILD; ie, focal (not diffuse) fibrotic changes on computed tomography (CT) of the chest with no apparent symptoms [7]. High-resolution computed tomography (HR-CT) has been shown to delineate the anatomy of the lungs more clearly than conventional CT. Using HR-CT, it can be easier to check focal fibrosis of the lung and exclude gravitational effect. It is difficult to estimate the incidence of serious but infrequent complications by a multiinstitutional analysis because preoperative evaluations of CT findings as well as operative procedures and postoperative management differ among hospitals. Intraoperative management by anesthesiologists also has a considerable influence on postoperative ALI [8]. To exclude these potential biases, we conducted this study to identify risk factors for postoperative IP in patients with no clinical evidence of ILD before surgical resection who were treated in a single center. 2011 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2010.10.062

Ann Thorac Surg ITO ET AL 2011;91:874 9 SUBPLEURAL HONEYCOMBING IS A RISK FACTOR 875 Abbreviations and Acronyms ALI acute lung injury ARDS acute respiratory distress syndrome CT computed tomography FEV 1 forced expiratory volume in 1 second FVC forced vital capacity HE hematoxylin-eosin stain HR-CT high-resolution computed tomographic HU Hounsfield unit ICU intensive care unit ILD interstitial lung disease IP interstitial pneumonia IPF idiopathic pulmonary fibrosis LDH lactate dehydrogenase Lob. lobectomy NSIP nonspecific interstitial pneumonia Pao 2 arterial oxygen tension Pneu. pneumonectomy Sub. sublober resection UIP usual interstitial pneumonia Patients and Methods Fig 1. Preoperative high-resolution computed tomographic films of computed tomographic honeycombing in three patients. Focal, not diffuse subpleural dorsal honeycombing with focal fibrotic changes can be seen. Our institutional Internal Review Board approved this retrospective study. From January 2002 through December 2005, a total of 651 patients underwent resection of primary lung cancer at Kanagawa Cancer Center. Patients with apparent ILD or obstructive pneumonia before operation were excluded from this study. All patients were closely evaluated by a single team of pulmonologists, radiologists, medical oncologist, and thoracic surgeons together. Surgery and postoperative management were performed by the same surgical team in all patients. After operation, patients were admitted to the intensive care unit for only that day, transfusion was performed at the speed of 1.25 ml kg 1 hour 1, and an oxygen mask (40%) was set in until next morning. From the next day of operation, patients started to walk by themselves with support. In our routine daily check-up of postoperative patients, Spo 2 (oxygen saturation as measured by pulse oximetry) and body weight were checked at least twice a day, and chest X-ray was undertaken on postoperative days 0, 1, 4, and 7. Laboratory values were checked on postoperative days 1, 4, and 7. Cefazolin (1.0 g) was used just before operation to postoperative day 1 twice a day. When we noted a patient with Spo 2 less than 93 in room air and infiltrating shadow on chest X-ray, a chest CT was immediately undertaken. Sputum culture was examined or bronchofiberscopy performed to exclude aspiration and bacterial infection. No patient was given a diagnosis of ILD preoperatively and had no clinical symptoms such as breathlessness, presence of crackles, or finger clubbing except caused by lung cancer. We retrospectively studied patients in whom IP developed postoperatively by reviewing their clinical characteristics (age, sex, smoking history, arterial oxygen tension, % forced vital capacity, forced expiratory volume in 1 second %, coexisting heart disease, tumornodes-metastasis stage), surgical records (operation time, blood loss, operative procedure, blood transfusion), and preoperative HR-CT scans of the chest. Postoperative IP was diagnosed on the basis of acute hypoxia-like ALI- ARDS, an arterial oxygen tension to fraction of inspired oxygen ratio of less than 300 with bilateral infiltrations on chest radiography, and the involvement of both lungs. Patients who had ALI-ARDS caused by bacterial pneumonia or aspiration were excluded. An ultrasound cardiogram was performed to rule out acute heart failure. Both lungs were examined by HR-CT. The conditions of HR-CT were a slice thickness of 3 mm or thinner using the mediastinal (level, 40 Hounsfield unit [HU]; width, 400 HU) and lung (level, 600 HU; width, 1,600 HU) window settings. The HR-CT scans were double-checked by a pulmonologist and a radiologist. We assumed that subpleural localized CT honeycombing may be a predictor of subclinical ILD. The criteria for CT honeycombing in this study were dorsal and focal (not diffuse) subpleural honeycomb formation with or without fibrotic changes occupying three or more segments in both lower lobes (CT honeycombing), without any physical symptoms (Fig 1). Curvilinear shadows and dependent densities indicating gravitational effects were excluded. All patients in whom apparent ILD had been previously diagnosed were excluded. Lung specimens from all patients in whom ALI-ARDS developed during the study period were histopathologically examined to confirm the presence or absence of ILD findings. Each variable was tested by the 2 test, Fisher exact test, and Student t test. Logistic-regression analysis was used for multivariate analysis, performed with Stat View for Windows (version 5.0; SAS Institute Inc, Cary, NC). The p values of less than 0.05 were considered to indicate statistical significance. Results Postoperative IP developed in 7 patients (7 of 651, 1.1%). Four of these patients (4 of 7, 57%) died of respiratory failure. Table 1 shows the clinical characteristics of the patients with postoperative IP. All 7 patients were men, with an average age of 66.5 years (62 to 73). All were smokers, and the average smoking index (cigarettes/day

876 ITO ET AL Ann Thorac Surg SUBPLEURAL HONEYCOMBING IS A RISK FACTOR 2011;91:874 9 Table 1. Patients Characteristics Onset (POD) LDH(Onset) Course Patho. Blood Loss (g) Time (Min.) CT Honeycombing Procedure CT Emphysema Smoking Index Age (Years) Sex Spirometry Case 1,920 lob. 405 390 7 Elevated Died, 27POD UIP 1 73 M Obstructive %FEV 1 67% 2 75 M Normal 1,120 lob. 200 50 4 Elevated Alive NSIP UIP 1,056 lob. 410 150 9 Elevated Died, 275POD 3 71 M Restrictive %FVC 77% 1,000 pneu. 325 355 9 Elevated Died, 19POD UIP 4 68 M Restrictive %FVC 79% 5 72 M Obstructive 900 lob. 335 70 10 Elevated Died, 43POD UIP %FEV 1 67% 6 70 M Normal 1,056 lob. 285 110 9 Elevated Alive IP(-) 7 76 M Obstructive 10 lob. 147 10 10 Not elevated Alive IP(-) %FEV 1 68% CT computed tomography; FEV 1 forced expiratory volume in 1 second; FVC forced vital capacity; IP interstitial pneumonia; LDH lactic dehydrogenase; lob. lobectomy; NSIP nonspecific interstitial pneumonia; Patho. pathologic findings; pneu. pneumonectomy; POD postoperative day; UIP usual interstitial pneumonia. Table 2. Comparison of Patients According to the Presence or Absence of Postoperative Interstitial Pneumonia Variable Postop IP ( ) (n 7) Postop IP ( ) (n 644) p Value Age 73 (68 76) 65 (29 86) 0.04 Sex (male/female) 7/0 367/277 0.02 Smoking (yes/no) 7/0 373/271 0.25 %FVC 97 (69 161) 105 (81 170) 0.22 FEV 1 % 72 (58 84) 75 (72 97) 0.29 Coexisting heart disease 2/5 183/461 0.99 ( /-) TNM stage (I vs II III) 5/2 487/157 0.68 Pao 2 85 (72 100) 88 (55 100) 0.57 CT fibrosis ( / ) 5/2 41/603 0.01 Procedure (pneu lob/sub) 7/0 514/130 0.36 Operation time 301 (147 410) 198 (27 555) 0.01 Operation time 4 hours 5/2 154/490 0.01 Blood transfusion ( / ) 0/7 4/640 0.99 Blood loss (ml) 162 (10 390) 91 (10 1,600) 0.17 Data are presented as mean (range). CT computed tomography; FEV 1 forced expiratory volume in 1 second; FVC forced vital capacity; lob lobectomy; Pao 2 arterial oxygen tension; pneu pneumonectomy; Postop IP postoperative interstitial pneumonia; sub sublober resection; TNM tumor-nodes-metastasis. years) was 1,008. The histologic diagnosis of lung cancer was adenocarcinoma in 3 patients, squamous cell carcinoma in 2, a large neuroendocrine carcinoma in 1, and a small cell carcinoma in 1. The clinical stage was stage IA in 1 patient, stage IB in 3, stage IIB in 2, and stage IIIA in 1. No patient received induction therapy or had a past history of thoracic irradiation. Five of the 7 patients with postoperative IP showed abnormal findings on preoperative spirometry: 3 had obstructive disorders, and 2 had restrictive disorders. The mean operation time was 301 minutes, the mean blood loss was 162 g, and no patient received blood transfusion. When a pneumonectomy patient showed hypoxia and unilateral infiltrations on chest X-ray, ultrasound cardiogram showed slight pulmonary hypertension caused by pneumonectomy but acute heart failure was ruled out. Bronchofiberscopy did not show purulent discharge; the result of bacterial culture was normal flora. On average, the onset day of postoperative IP was 8.3 days (4 to 10) after operation. The most common initial symptoms were fever and hypoxia. The place of postoperative IP was diffuse; there was no tendency in the location of postoperative IP. Six of the seven patients with postoperative IP (86%) had elevated serum lactate dehydrogenase levels at the time of onset. The surgical procedures were pneumonectomy in 21 patients, lobectomy in 500, and sublobar resection in 130. Table 2 compares the clinical characteristics of patients with or without postoperative IP; age, sex, CT honeycombing, and prolonged operation time were risk factors for the development of postoperative IP. Multivariate analyses showed that CT honeycombing (p 0.01) and prolonged operation time were significant in-

Ann Thorac Surg ITO ET AL 2011;91:874 9 SUBPLEURAL HONEYCOMBING IS A RISK FACTOR 877 Table 3. Logistic Regression Analysis Variable p Value Relative Risk 95% Confidence Interval Operation time p 0.04 6.134 1.047 35.939 4 hours CT fibrosis p 0.01 16.967 3.004 95.820 Operation time 4 hours CT fibrosis p 0.01 1607.5 124.65 20730 CT computed tomography. dependent risk factors (p 0.04) (Table 3); combination of CT honeycombing and operation time above 4 hours strongly suggested the risk of postoperative IP. When sublobar resection was excluded from analysis because of a less invasive procedure, age, sex, CT honeycombing, and prolonged operation were risk factors again. The result of multivariate analysis was the same when sublobar resection was excluded. Five of the 7 patients with postoperative IP had CT honeycombing. During the study period, a total of 46 patients had CT honeycombing. The incidence of postoperative IP was 10.9% (5 of 46) among patients with CT honeycombing and 0.3% (2 of 598) among those without CT honeycombing. All 4 patients who died had CT honeycombing and were given a diagnosis of usual interstitial pneumonia (UIP) on histopathologic examinations of surgical specimens (Fig 2). One survivor with CT honeycombing who had postoperative IP showed nonspecific interstitial pneumonia on histopathologic examination. In the other 39 patients with CT honeycombing without developing postoperative IP, not all were histologically examined with a sample from the CT honeycombing area. We examined pathologic specimens from a non-cancer area in 11 patients. Six patients had UIP, 2 had nonspecific interstitial pneumonia, and 3 had tobacco-related emphysema. All patients who had postoperative IP received respiratory support, antibiotics, and steroid therapy. Two patients who had postoperative IP without CT honeycombing recovered successfully. Histopathologic examination showed no evidence of ILD in either of these patients. Mortality among the patients who had postoperative IP as well as CT honeycombing was 80% (4 of 5); in contrast, none of the patients without CT honeycombing died. During the study period, three cases of bacterial pneumonia developed but all responded to treatment with antibiotics. In this period none of the patients without CT honeycombing developed IPF in postoperative observation as outpatients. In 46 patients with CT honey combing, mean postoperative intensive care unit stay was 1.03 days. In 596 patients without CT honeycombing mean postoperative ICU stay was 1.08 days. There was no statistical difference between the two groups (p 0.99). During this period, 4 patients with apparent ILD excluded from this study had a lung cancer operation. Two lobectomies and 2 partial resections were carried out. None of them developed acute postoperative IP. Two of the patients died of respiratory failure 23 months and 38 months later because of gradual deterioration of ILD. The overall operative mortality rate was 1.1%, including 2 patients who died of cerebral infarction and 1 patient who died of bronchopleural arterial fistula. Two thirds of postoperative mortality was caused by postoperative IP. Comment In this study we could precisely reevaluate preoperative chest HR-CT scans in the same setting. A single surgical team performed surgery and postoperative management in a similar manner for all patients with lung cancer in this series. Thus, we could exclude variables of postoperative management as the risk of postoperative IP. Most of the patients with postoperative IP had CT honeycombing. The surgical specimens of some postoperative IP patients in the CT honeycombing group showed UIP on histopathologic examination, but IPF or other types of ILD were not diagnosed clinically before surgery. Typical radiologic findings of IPF on chest CT scans include diffuse and apparent subpleural honeycomb formation, occasionally accompanied by fibrotic changes [7, 9]. High-resolution computed tomography has been shown to delineate the anatomy of the lungs more clearly than conventional CT. The anatomic abnormalities of IPF are progressive, beginning with the microscopic inflammation that precedes fibrosis. In this context, focal (not diffuse) dorsal subpleural honeycombing with or without fibrotic changes occupying three or more segments of both lower lobes on HR-CT, may be able to predict the presence of occult or early-stage IPF owing to the fine definition of the lung anatomy on HR-CT [10]. To this point, CT honeycombing of HR-CT is a simple and useful predictor of the risk of postoperative IP. But not all the patients with this HR-CT finding were pathologically confirmed as UIP in this study, so further prospective studies will be needed to confirm it. But once postoperative IP develops in such patients with occult or early- Fig 2. Histopathologic findings of the resected specimen, showing typical usual interstitial pneumonia in a subpleural lesion (hematoxylin-eosin stain, 100).

878 ITO ET AL Ann Thorac Surg SUBPLEURAL HONEYCOMBING IS A RISK FACTOR 2011;91:874 9 stage IPF, its mortality is extremely high, similar to that in patients with clinically apparent IPF [11, 12]. In clinical practice, surgeons sometimes encounter postoperative patients with fever, acute hypoxia, and bilateral infiltrations on chest radiography but it is difficult to single out the cause. Partial atelectasis, lung edema, intrapulmonary oozing, and pleural fluid aspiration sometimes show radiologic characteristics similar to those of postoperative IP but are not always fatal. In this study, preoperative CT honeycombing and prolonged operation times were independent risk factors of fatal postoperative IP; combination of these two factors strongly suggested high risk. The mechanism of acute exacerbation is still unclear, but occult IPF might be exacerbated by long operation times. Once fever with acute hypoxia and bilateral infiltrations develop, suggesting a potential risk of postoperative IP, the management strategy depends on whether occult IPF progresses to postoperative IP or not. The treatment strategy for postoperative IP and acute exacerbations of ILD is basically the same as that for ARDS (ie, ventilator support plus antibiotics with or without steroid therapy) but mortality remains high [13]. The use of corticosteroids for the treatment of ARDS remains controversial but early low-dose steroid therapy may improve outcomes [14 16]. Our results suggest that patients with postoperative IP who did not have occult IPF may respond to conventional therapy. In contrast, once postoperative IP develops in patients with occult UIP, mortality is high. Our study showed that fatal postoperative IP rarely developed in patients with no apparent IPF before surgery. If patients with occult IPF can be identified preoperatively careful postoperative management and countermeasures against postoperative IP could be implemented. Surgery remains the standard treatment for primary lung cancer even in such high-risk patients [5]. However, about half of all postoperative deaths among patients with primary lung cancer are attributed to postoperative IP. In case of occult IPF patients who developed postoperative IP, there will be a tiny percent of survival. Thus it is important to predict this risk preoperatively in candidates for lung cancer surgery. The detailed pictures of HR-CT were very useful to know the presence of honeycombing and exclude a gravitational effect. Measures to prevent postoperative IP in patients with occult IPF are essential to improving the safety of lung cancer surgery. Subpleural honeycombing with or without fibrotic changes on HR-CT was a significant predictor of both the risk and severity of postoperative IP in asymptomatic patients who underwent resection for lung cancer. Our results and the diagnostic usefulness of HR-CT findings should be confirmed in larger prospective clinical trials, with the ultimate goal of preventing life-threatening complications after lung cancer surgery. References 1. Ginsberg RJ, Hill LD, Eagan RT, et al. Modern thirty-day operative mortality for surgical resections in lung cancer. J Thorac Cardiovasc Surg 1983;86:654 8. 2. Koike T, Yamato Y, Asamura H, et al. Improvements in surgical results for lung cancer from 1989 to 1999 in Japan. J Thorac Oncol 2009;4:1364 9. 3. Watanabe S, Asamura H, Suzuki K, Tsuchiya R. Recent results of postoperative mortality for surgical resections in lung cancer. Ann Thorac Surg 2004;78:999 1002. 4. Kudoh S, Kato H, Nishiwaki Y, et al. Interstitial lung disease in Japanese patients with lung cancer: a cohort and nested case-control study. Am J Respir Crit Care Med 2008;177: 1348 57. 5. Watanabe A, Higami T, Ohori S, Koyanagi T, Nakashima S, Mawatari T. Is lung cancer resection indicated in patients with idiopathic pulmonary fibrosis? J Thorac Cardiovasc Surg 2008;136:1357 63. 6. Tsuboi M, Le Chevalier T. Interstitial lung disease in patients with non-small-cell lung cancer treated with epidermal growth factor receptor inhibitors. Med Oncol 2006;23: 161 70. 7. American Thoracic Society. Idiopathic pulmonary fibrosis. Diagnosis and treatment. International consensus statement. American Thoracic Society (ATS), and the European Respiratory Society (ERS). Am J Respir Crit Care Med 2000;161 (2 Pt 1:646-64. 8. Licker M, Fauconnet P, Villiger Y, Tschopp JM. Acute lung injury and outcomes after thoracic surgery. Curr Opin Anaesthesiol 2009;22:61 7. 9. American Thoracic Society; European Respiratory Society. American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias. Am J Respir Crit Care Med 2002;165:277 304. 10. Orens JB, Kazerooni EA, Martinez FJ, et al. The sensitivity of high-resolution CT in detecting idiopathic pulmonary fibrosis proved by open lung biopsy. A prospective study. Chest 1995;108:109 15. 11. Chida M, Ono S, Hoshikawa Y, Kondo T. Subclinical idiopathic pulmonary fibrosis is also a risk factor of postoperative acute respiratory distress syndrome following thoracic surgery. Eur J Cardiothorac Surg 2008;34:878 81. 12. Tanita T, Chida M, Hoshikawa Y, et al. Experience with fatal interstitial pneumonia after operation for lung cancer. J Cardiovasc Surg (Torino) 2001;42:125 9. 13. Deslauriers J, Mehran R. Handbook of perioperative care in general thoracic surgery. Elsevier Mosby: Philadelphia, PA; 2005:314 7. 14. Lee HS, Lee JM, Kim MS, Kim HY, Hwangbo B, Zo JI. Low-dose steroid therapy at an early phase of postoperative acute respiratory distress syndrome. Ann Thorac Surg 2005; 79:405 10. 15. Kutlu CA, Williams EA, Evans TW, Pastorino U, Goldstraw P. Acute lung injury and acute respiratory distress syndrome after pulmonary resection. Ann Thorac Surg 2000;69:376 80. 16. British Thoracic Society; Society of Cardiothoracic Surgeons of Great Britain, and Ireland Working Party. BTS guidelines: guidelines on the selection of patients with lung cancer for surgery. Thorax 2001;56:89 108. INVITED COMMENTARY The prognosis for idiopathic pulmonary fibrosis (IPF) is very poor, with a median survival of 2 to 4 years after the diagnosis, yet the disease course in individual patients is highly variable. Predicting prognosis in individual patients is challenging, but various clinical and radiologic variables have been identified. Although postoperative 2011 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2010.11.054