Relationship of Carbon Monoxide Pulmonary Diffusing Capacity to Postoperative Cardiopulmonary Complications in Patients Undergoing Pneumonectomy

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

DLCO and postpneumonectomy complications

RELATIONSHIP OF CARBON MONOXIDE PULMONARY DIFFUSING CAPACITY TO POSTOPERATIVE CARDIOPULMONARY COMPLICATIONS IN PATIENTS UNDERGOING PNEUMONECTOMY

Jeng-Shing Wang Respiratory Medicine Section, Vancouver General Hospital and University of British Columbia, Vancouver, British Columbia, Canada.

This retrospective analytic study evaluated whether abnormal diffusing capacity for carbon monoxide (DLCO) is a predictor of postoperative morbidity and mortality in patients undergoing pneumonectomy for lung cancer. The medical records of patients undergoing pneumonectomy at Vancouver General Hospital between January 1992 and December 1997 were reviewed. Postoperative complications occurring within 30 days of resection were classified into mortality, and cardiovascular, pulmonary, and technical complications. A total of 151 pneumonectomy cases were reviewed. There were 100 men (66%) and 51 women (34%) with a mean age of 61 years. Complications occurred in 73 patients (48%), including mortality in eight (5%), cardiovascular morbidity in 50 (33%), pulmonary morbidity in 30 (20%), and technical morbidity in 22 (15%). Arrhythmia (21%) and pulmonary edema (13%) were the two major cardiovascular complications. Patients with complications had a greater smoking history, a longer hospital stay, a lower

forced expiratory volume in 1 second (FEV1), a lower FEV1/forced vital capacity (FVC) ratio, a lower DLCO, and a lower DLCO/alveolar volume (VA) ratio than patients without complications. A DLCO of 70% predicted was the best functional predictor of postoperative complications, with a complication rate of 94% in patients with a DLCO of less than 70% predicted compared with 27% in patients with a DLCO of at least 70% predicted (sensitivity, 62%; specificity, 96%). However, technical morbidity was not related to preoperative lung function variables, including DLCO. Patients with a DLCO of at least 70% predicted had a low postpneumonectomy complication rate. Although cardiac arrhythmia was the major cause of morbidity, pulmonary edema was the major cause of mortality.

Key Words: diffusing capacity, postoperative complications, preoperative assessment (Kaohsiung J Med Sci 2003;19:437–46)

Lung cancer is the most common cause of death due to the curative treatment for non-small-cell lung cancer without cancer in men, and has become the most common cancer in metastasis is lung resection. The removal of lung women. The prognosis in untreated cases is poor; at present, parenchyma from patients, who are usually smokers and have chronic obstructive pulmonary disease [1] and may have compromised cardiovascular or pulmonary status, may lead to cardiopulmonary complications or death. The functional loss resulting from lung resection varies with the extent of resection, the relative function of the tissue removed Received: March 4, 2003 Accepted: August 5, 2003 compared with that of the remaining lung tissue, and the Address correspondence and reprint requests to: Dr. Jeng-Shing degree of functional impairment prior to surgery. Currently, Wang, 166 Min-Shiang Street, Kaohsiung 800, Taiwan. E-mail: [email protected] about 30% of patients undergoing lung resection develop

Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 437 © 2003 Elsevier. All rights reserved. J.S. Wang cardiopulmonary complications, with a 30 day mortality cytology of the lung lesion. The final diagnosis was based varying between 0.6% and 5% [2–4] depending on the on the pathology of the resected lung. Mediastinal extent of lung resection. The mortality in one study was involvement was generally excluded by mediastinoscopy 6.8% following pneumonectomy and 3.9% following or CT scan of the chest, and metastasis by CT scan of the lobectomy [2], while that in another study was 5.7% after brain, liver and bone if clinically warranted. pneumonectomy, 4.4% after bilobectomy, and 1.4% after From the medical records, we obtained general data lesser resections [4]. including age, sex, height, weight, smoking habit, and the Many parameters have been used to assess the risk of presence of diabetes, prior thoracic operation, chronic postoperative outcome, but it is not clear which preoperative obstructive pulmonary disease, and heart disease (including pulmonary function tests are the best predictors of hypertension, old myocardial infarction, and New York postoperative complications. Preoperative pulmonary Heart Association class). We also noted the duration of function testing usually includes spirometry, lung volumes, hospitalization and of the postoperative period, the year of diffusing capacity, oximetry, and arterial blood gases, and surgery, the operation site and position used for surgery, may include, in selected cases, radionuclide lung scanning, and if pericardial dissection was performed. Laboratory exercise testing, invasive pulmonary hemodynamic data including blood hemoglobin, serum albumin, measurements, and risk stratification analysis. The single creatinine, and glutamine oxaloacetic transaminase (GOT) breath diffusing capacity of the lung for carbon monoxide were recorded. We reviewed cardiac evaluation (electro- (DLCO) has recently been shown to be an independent cardiogram, echocardiogram, and cardiac stress test), predictor of postoperative outcome, including mortality and radionuclide lung ventilation/perfusion scan, and radiology postoperative complications [5–8]. Patients with low DLCO findings including chest roentgenography and chest CT (= 60% predicted for pneumonectomy or bilobectomy, = 50% scan. Lung function data, including forced expiratory predicted for lobectomy) underwent major pulmonary volume in 1 second (FEV1), forced vital capacity (FVC), resection with an increased respiratory complication rate [9]. DLCO, and DLCO/alveolar volume (DLCO/VA), were A low preoperative DLCO may identify those patients who expressed as % predicted value. We used the prediction have emphysema and a reduced pulmonary capillary bed. equations of Crapo et al for spirometry and lung volumes [10, They may be prone to postoperative cardiopulmonary 11], and those of Miller et al for nonsmokers for DLCO [12]. complications because of reductions in the pulmonary Spirometry was performed using a computerized dry capillary bed and reduced gas capacity for exchange, rolling seal spirometer (either Model “Transfer Test USA”, especially after pneumonectomy. The purpose of this PK Morgan, Chatham, Kent, UK, or Model No. 922, retrospective study was to evaluate whether abnormal Sensormedics, Anaheim, CA, USA) according to American DLCO is specifically useful in predicting postoperative Thoracic Society (ATS) criteria [13]. The highest values of morbidity and mortality following pneumonectomy. FEV1 and FVC were taken from the two best tests that agreed with each other to within 5%. Lung volumes were determined using the helium dilution technique on the MATERIALS AND METHODS Morgan equipment. Each patient rebreathed a 10% helium mixture in a closed circuit until equilibrium was reached, The study was approved by the University of British and the helium concentration was continuously monitored Columbia Ethics Committee. We reviewed the hospital using a helium analyzer. DLCO was determined using the records of all patients who had undergone pneumonectomy single breath technique [14] with an automated valve and for lung cancer at Vancouver General Hospital between timing device and a bag in a box system with the Morgan January 1992 and December 1997; we excluded patients equipment, as previously described [15,16]. The subject who had received radiation treatment or chemotherapy inhaled a volume of test gas containing 10% helium, 0.3% prior to surgery. A total of 151 charts were reviewed to carbon monoxide, 21% oxygen, and the balance nitrogen. A determine whether preoperative lung function data breath-holding time of about 10 seconds was used and predicted postoperative cardiopulmonary complications. determined by the Jones-Meade method [17]. The dead Diagnosis and staging were based on one or more of the space washout and alveolar sample volumes were set at following: chest roentgenography and computed 900 mL. All assessments of DLCO conformed to ATS tomography (CT) scan, sputum cytology, bronchoscopy standards [18]. The mean of two DLCO measurements with bronchial brushing and/or biopsy, and biopsy or within 5% of each other was taken to be the measured

438 Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 DLCO and postpneumonectomy complications

DLCO, and a maximum of four DLCO maneuvers, each (34%). A total of 136 patients (90%) were smokers or ex- separated by at least 4 minutes, were performed. smokers, with a mean smoking history of 46 pack-years; 15 Complications occurring within 30 days after resection patients had never smoked. The mean duration of were considered to be postoperative complications. hospitalization was 14 ± 7 days, with a mean postoperative Complications were classified into mortality, and stay of 9 ± 3 days. Most patients (139, 92%) had primary lung cardiovascular, pulmonary, and technical morbidity. cancer. The most common type was squamous carcinoma Cardiovascular morbidity included myocardial infarction, (73 patients, 48%), followed by adenocarcinoma (46 patients, congestive heart failure, pulmonary edema, shock, 30%), and poorly differentiated carcinoma (13 patients, 9%). arrhythmia, and cerebrovascular accident. Pulmonary Four patients had adenosquamous carcinoma, and in three morbidity included ventilatory support greater than patients, the final pathology of the resected lung showed 48 hours, re-intubation, pulmonary embolism, pneumonia, small-cell carcinoma (the preoperative diagnosis in all three atelectasis, and respiratory insufficiency (PaCO2 > 45 mmHg cases was non-small-cell carcinoma). There were 12 cases of or PaO2 < 55 mmHg, without supplemental oxygen). Tech- metastatic cancer of various types. nical morbidity included bronchopleural fistula, recurrent Lung function tests showed an FEV1 (mean ± SD) of laryngeal nerve injury, wound infection, empyema, 2.39 ± 0.68 L, FEV1 % predicted of 78 ± 18%, FVC of 3.53 ± hematoma, and miscellaneous morbidities such as 0.90 L, FVC % predicted of 90 ± 18%, FEV1/FVC of 68 ± 11%, mediastinitis, pneumomediastinum, and bleeding. residual volume (RV)/total lung capacity (TLC) of 35 ± 8%, Data analysis was done with Microsoft Excel™ 97 DLCO of 20.08 ± 5.78 mL/min/mmHg, DLCO % predicted (Microsoft, Santa Rosa, CA, USA) and SPSS™ 8.0 (SPSS, of 80 ± 20%, DLCO/VA of 3.77 ± 0.97 mL/min/mmHg/L, Chicago, IL, USA). We determined the means and standard and DLCO/VA % predicted of 87 ± 22%. deviations (SDs) for the different variables in patients with Complications occurred in 73 patients (48%), and and without complications. Comparisons between different included mortality (5%), cardiovascular morbidity (33%), groups for continuous variables were made using the two- pulmonary morbidity (20%), and technical morbidity (15%) tailed Student’s t test. The Chi-squared test was used for (Table 2). The causes of death were pulmonary edema in six categorical variables. Multiple variable analysis using patients, cerebrovascular accident in one, and acute stepwise logistic regression [19] was performed to myocardial infarction in one. Arrhythmia and pulmonary investigate the relative usefulness of the combination of edema were the two major causes of cardiovascular variables for the prediction of postoperative complications. morbidity, occurring in 21% and 13%, respectively, of all A p value of less than 0.05 was considered statistically patients. Arrhythmias included atrial fibrillation in 28 significant. Receiver operating characteristic (ROC) curves patients, atrial flutter in two, frequent ventricular premature [20] and Fisher’s exact test [21] were used to define the best contractions in one, and ventricular bigeminy in one. cut-off limits of the different variables in relation to Pneumonia was the major cause of pulmonary morbidity postoperative complications, and sensitivity, specificity, (11%), and recurrent laryngeal nerve injury was the most positive predictive value, and negative predictive value for common cause of technical morbidity (3%). Miscellaneous each variable were determined [20]. The area under the technical morbidity included mediastinitis (1 patient), ROC curve (AURC) was estimated using the following pneumomediastinum (1), bleeding (1), septicemia (1), renal algorithm: AURC between 2 successive points = mean failure (1), enteritis (1), benign prostatic hypertrophy (1), sensitivity × difference in (1 – specificity). ischemic heart disease (1), chest wall infection (1), and The total AURC is the sum of successive individual pleurisy (1). areas. The relative risk and risk difference for postoperative Patients with complications had a significantly greater complications, obtained using a given cut-off limit for a smoking history and a longer hospital stay than patients preoperative variable, were calculated as A(C+D)/C(A+B) without complications, and also had significantly lower and A/(A+B) – C/(C+D) (Table 1). FEV1 % predicted, FEV1/FVC, DLCO % predicted, and DLCO/VA % predicted than those without complications (Table 3). There was no significant difference between the RESULTS patients with technical complications and patients without technical complications for any preoperative variable, The 151 pneumonectomy patients had a mean age of 61 ± 10 including hypertension, previous myocardial infarction, (mean ± SD) years; 100 were men (66%) and 51 women New York Heart Association class, or abnormal

Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 439 J.S. Wang

Table 1. Definition of symbols used to calculate different Table 2. Prevalence of postoperative complications cut-off limits from Fisher’s exact test following pneumonectomy, n = 151 Complications No complications n (%)

< Cut-off limits A B Overall complications 73 (48) ≥ Cut-off limits C D Mortality 8 (5)

Cardiovascular morbidity* 50 (33) Myocardial infarction 3 (2) electrocardiogram, echocardiogram, or cardiac stress test. Congestive heart failure 7 (5) Table 4 compares preoperative variables between the Pulmonary edema 20 (13) Shock 3 (2) mortality cases and the remaining patients. There were Arrhythmia 31 (21) significant differences in age and FEV1/FVC between the Cerebrovascular accident 4 (3) two groups. Patients who died were significantly older by Pulmonary morbidity* 30 (20) a mean of 9 years. The unexpectedly higher FEV1/FVC in the mortality group might have been due to a slightly lower Ventilatory support > 48 hr 7 (5) Re-intubation 7 (5) FVC in this group, leading to a higher FEV1/FVC ratio. No Pulmonary embolism 2 (1) other preoperative variable (including hypertension, Pneumonia 17 (11) previous myocardial infarction, New York Heart Association Lobar collapse 11 (7) class, and abnormal electrocardiogram, echocardiogram, Respiratory insufficiency 10 (7) or cardiac stress test) was significantly different in the Technical morbidity* 22 (15) mortality cases and the remaining patients. Bronchopleural fistula 2 (1) Using multiple variable stepwise logistic regression, no Recurrent laryngeal nerve injury 5 (3) model was found that, after combining different variables, Wound infection 4 (3) significantly improved the prediction of overall com- Hematoma 2 (1) Empyema 2 (1) plications, mortality, cardiovascular morbidity, and Miscellaneous 10 (7) pulmonary morbidity as compared with the single variable * More than one type of morbidity may be found in any one patient. analysis. The best predictor was DLCO % predicted for overall complications (p < 0.001), cardiovascular morbidity

(p < 0.001), and pulmonary morbidity (p < 0.001). For DLCO/VA % predicted, FEV1 % predicted, and FEV1/FVC. mortality, the only predictor was age (p < 0.05). The incidence The ROC curves of these four variables for overall of overall complications, cardiac morbidity, pulmonary complications, cardiovascular morbidity, and pulmonary morbidity, technical morbidity, and mortality were morbidity were determined and the AURC was calculated. calculated for six intervals of DLCO % predicted (< 40, < 50, The largest AURC was 0.84, calculated from the ROC curve < 60, < 70, < 80, and ≥ 80%), which was the best predictive of DLCO % predicted for overall complications (Figure 3), preoperative variable (Figure 1). Overall complications, confirming that DLCO % predicted was the best predictor. cardiac morbidity, and pulmonary morbidity were related Furthermore, the incidence of mortality was calculated for to DLCO % predicted, but technical morbidity and mortality six intervals for age (≤ 55, > 55, > 60, > 65, > 70, and > 75 were not. The incidence of overall complications, cardiac years), the ROC curve of age for mortality was determined, morbidity, pulmonary morbidity, technical morbidity, and and the AURC was calculated (Figure 4). The AURC was mortality was calculated for six intervals of age (≤ 55, > 55, 0.77, confirming that age was the best predictor for mortality. > 60, > 65, > 70, and > 75 years) (Figure 2). There was no The best cut-off limit was defined by the point closest to mortality among patients aged 55 years or less, while the left upper corner, and it was 70% in the ROC curve of increased mortality was observed among patients more DLCO % predicted for overall complications (p < 0.001) than 70 years in age. (Figure 3). Forty-five of 73 patients with overall To evaluate the level of preoperative variables that complications had a DLCO of less than 70% predicted correlated with complications, the incidence of (sensitivity, 62%), and 75 of 78 patients without overall complications was calculated for six intervals (< 40, < 50, complications had a DLCO of at least 70% predicted < 60, < 70, < 80, and ≥ 80%) for DLCO % predicted, (specificity, 96%). This gave a positive predictive value of

440 Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 DLCO and postpneumonectomy complications

Table 3. Preoperative variables examined in relation to pneumonectomy

No complications (n = 78) Complications (n = 73) p Age (yr) 61 ± 11 62 ± 10 NS Height (cm) 169 ± 10 169 ± 9NS Weight (kg) 74 ± 14 72 ± 14 NS Smoking (pack-years) 40 ± 19 (n = 68) 53 ± 27 (n = 68) < 0.01 Hemoglobin (g/L) 130 ± 16 127 ± 18 NS Albumin (g/L) 38 ± 5(n = 46) 37 ± 7(n = 51) NS Creatinine (µmol/L) 82 ± 16 83 ± 20 NS GOT (U/L) 22 ± 6(n = 62) 24 ± 13 (n = 62) NS Hospital stay (d) 13 ± 616± 8 < 0.01 Postoperative stay (d) 9 ± 29± 4NS

FEV1 (L) 2.52 ± 0.69 2.22 ± 0.63 < 0.01 FVC (L) 3.60 ± 0.88 3.42 ± 0.89 NS

FEV1 % predicted 83 ± 17 74 ± 17 < 0.001 FVC % predicted 93 ± 17 88 ± 19 NS

FEV1/FVC (%) 70 ± 10 66 ± 12 < 0.01 RV/TLC (%) 34 ± 736± 9NS DLCO (mL/min/mmHg) 22.56 ± 4.98 17.42 ± 5.41 < 0.001 DLCO % predicted 90 ± 16 69 ± 18 < 0.001 DLCO/VA (mL/min/mmHg/L) 4.12 ± 0.73 3.40 ± 1.06 < 0.001 DLCO/VA % predicted 96 ± 16 79 ± 23 < 0.001

NS = not significant; GOT = glutamine oxaloacetic transaminase; FEV1 = forced expiratory volume in 1 second; FVC = forced vital capacity; RV = residual volume; TLC = total lung capacity; DLCO = diffusing capacity for carbon monoxide; VA = alveolar volume.

Table 4. Comparison of variables between patients with mortality or without mortality

Mortality (n = 8) No mortality (n = 143) p Age (yr) 70 ± 761± 10 < 0.01 Height (cm) 168 ± 5 169 ± 10 NS Weight (kg) 72 ± 11 73 ± 14 NS Smoking (pack-years) 58 ± 42 (n = 7) 45 ± 23 (n = 129) NS Hemoglobin (g/L) 126 ± 23 128 ± 16 NS Albumin (g/L) 31 ± 9(n = 7) 38 ± 5(n = 90) NS Creatinine (µmol/L) 80 ± 21 83 ± 18 NS GOT (U/L) 38 ± 33 (n = 7) 22 ± 6(n = 117) NS Hospital stay (d) 20 ± 12 14 ± 7NS Postoperative stay (d) 8 ± 49± 3NS

FEV1 (L) 2.29 ± 0.59 2.40 ± 0.68 NS FVC (L) 3.20 ± 0.94 3.55 ± 0.90 NS

FEV1 % predicted 78 ± 16 78 ± 18 NS FVC % predicted 83 ± 18 91 ± 17 NS

FEV1/FVC (%) 74 ± 668± 11 < 0.05 RV/TLC (%) 36 ± 635± 9NS DLCO (mL/min/mmHg) 18.58 ± 6.27 20.16 ± 5.77 NS DLCO % predicted 76 ± 20 80 ± 20 NS DLCO/VA (mL/min/mmHg/L) 3.98 ± 1.00 3.76 ± 0.97 NS DLCO/VA % predicted 95 ± 23 87 ± 22 NS

77%, a negative predictive value of 81%, a relative risk of limit. Five of the eight mortality cases were over 70 years 3.4 and a risk difference of 67% (p < 0.001, Fisher’s exact old (sensitivity, 63%), and 119 of 143 patients without test). For age (Figure 4), 70 years was the best cut-off mortality were at or below this cut-off (specificity, 83%).

Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 441 J.S. Wang

This gave a positive predictive value of 17%, a negative cardiovascular and pulmonary morbidity, after predictive value of 98%, a relative risk of 7.0 and a pneumonectomy. The strong correlation between diffusing risk difference of 15% (p < 0.01, Fisher’s exact test). capacity and complications is probably due to the Postoperative complications were compared using these contribution of a reduced pulmonary capillary bed and preoperative variables. Again, the best predictor for alveolar capillary membrane to cardiopulmonary complications was DLCO % predicted and the best complications. Our results suggested that a DLCO of at predictor for mortality was age. least 70% predicted would be associated with a low postpneumonectomy complication rate. DLCO/VA was not as good a predictor as DLCO in relation to postoperative DISCUSSION complications. A normal DLCO/VA can be misleading, implying that a low DLCO with a reduced single VA and a Our study showed that patients with overall complications normal DLCO/VA reflect a normal pulmonary capillary had a greater smoking history, longer hospital stay, lower bed [22].

FEV1 % predicted, lower FEV1/FVC, lower DLCO % In patients with technical complications, none of the predicted, and lower DLCO/VA % predicted than patients variables used for preoperative evaluation were significantly without overall complications (Table 3). Both DLCO and different from those in patients without technical spirometry data were useful in discriminating postoperative complications. This finding indicates that technical complications. Further analyses using logistic regression morbidity was not related to patients’ lung function, models and ROC curves with AURC determinations (Figure including DLCO (Figure 1), but was due to other factors 3) showed that DLCO % predicted was the best predictor such as surgical technique, anesthetic technique, nursing for overall complications, followed, in decreasing order, by care, and perioperative care. The lack of relation between

DLCO/VA % predicted, FEV1 % predicted, and FEV1/FVC. diffusing capacity and technical morbidity indicates the The same analysis showed that these variables predicted specific value of a low diffusing capacity in predicting overall complications best, followed, in decreasing order, cardiopulmonary complications following pneumo- by cardiovascular morbidity, pulmonary morbidity, and nectomy. mortality, as shown for DLCO % predicted in Figure 1. The mortality group was significantly different from the Our results indicated that diffusing capacity was a group without mortality, being older by a mean of about strong predictor of risk of complications, including 9 years, but did not have lower lung function values (Table

100 90 80 70 60 60

50 50 Overall 40 complications 40 Cardiovascular 30 30 morbidity Overall complications 20 Pulmonary morbidity Incidence of events (%) 20 Cardiovascular morbidity Technical morbidity 10 Pulmonary morbidity 10 Mortality 0 Technical morbidity 0 Mortality Incidence of events (%) 55 80 > 75 > 70 > 65 > 60 > 55 ≤ ≥ < 80 < 70 < 60 < 50 < 40 Age (yr) DLCO % predicted

Figure 1. Incidence of mortality, technical morbidity, pulmonary Figure 2. Incidence of mortality, technical morbidity, pulmonary morbidity, cardiovascular morbidity, and overall complications in relation morbidity, cardiovascular morbidity, and overall complications in relation to diffusing capacity for carbon monoxide (DLCO) % predicted. Mortality to age. Mortality was related to age. and technical morbidity were not related to DLCO % predicted.

442 Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 DLCO and postpneumonectomy complications

1 1 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4

Sensitivity

Sensitivity 0.3 0.3 0.2 0.2 0.1 0.1 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

1 – Specificity 1 – Specificity

Figure 3. Receiver operating characteristic (ROC) curve of diffusing Figure 4. Receiver operating characteristic (ROC) curve of age for capacity for carbon monoxide (DLCO) % predicted for prediction of overall prediction of mortality. The solid line is the line of identity for a test complications. The solid line is the line of identity for a test without any without any discrimination. The area under the ROC curve was 0.77; the discrimination. The area under the ROC curve was 0.84; the best best cut-off point was 70 years, with a sensitivity of 63% and specificity cut-off point was 70%, with a sensitivity of 62% and specificity of 96%. of 83%.

4). Analyses using logistic regression models and ROC published in 1955 [24], many attempts have been made to curves with AURC determinations (Figure 4) showed that establish criteria to predict postoperative mortality and age was the best predictor for mortality. Greater age may cardiopulmonary complications after lung resection. The have increased the risk of underlying heart disease criteria used to select patients for major pulmonary resections predisposing to mortality, since pulmonary edema was the are based on clinical data, spirometry, more detailed major cause of mortality. Our results suggested that an age pulmonary functional assessment, and cardiac evaluation. of 70 years or less would be associated with a relatively low Severe abnormalities detected by spirometry indicate an postpneumonectomy mortality rate. Mortality was increased risk after pulmonary resection and should prompt predictable by age (Figure 2), but not by DLCO, which was further preoperative evaluation and a critical assessment of the best preoperative variable predictive of the patient’s overall condition [25]. Patients with hypoxemia cardiopulmonary complications (Figure 1). Our patients or hypercapnia are at increased risk for morbidity or with low DLCO usually had well-preserved spirometry mortality after thoracotomy [26]. Lung volume values (70 ± 11% predicted FEV1 and 84 ± 23% predicted determinations may be helpful, because an increase in the FVC in patients with DLCO < 60% predicted). Modern RV/TLC ratio is associated with a high incidence of surgical, anesthetic, and critical care techniques used postoperative pulmonary complications [27]. Tests routinely in the perioperative period may have reduced the previously used to evaluate differential function of each mortality rate in patients with poor lung function [9] and lung, such as bronchospirometry and the lateral position resulted in the lack of relationship between preoperative test, have been replaced by radionuclide lung scanning, variables and mortality [23]. In other studies, there was with quantitative measurement of the contribution of each disagreement about the relationship of a low DLCO to lung to pulmonary ventilation and blood flow [28]. The postoperative mortality. Four studies showed a relationship attractive features of radionuclide lung scanning include its between impaired DLCO and postoperative mortality [5– ready availability in general hospitals, negligible risk to 8], while two did not show such a relationship [9,23]. This patients, and a fairly high degree of accuracy in the prediction difference may be due to the different patient populations of postoperative pulmonary function [28]. Pulmonary studied, the differences in extent of resection, and possibly hemodynamic measurements may be useful in selected the differences in surgical, anesthetic, and critical care patients; a right ventricular ejection fraction of at least 35%, techniques used in the different centers. a pulmonary vascular resistance of less than 200 dyne × sec Since the first well-performed study of pulmonary × cm-5, and a ratio of pulmonary vascular resistance to right function testing in candidates for lung surgery was ventricular ejection fraction of less than 5.0 should be

Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 443 J.S. Wang associated with low postpneumonectomy morbidity and values for preoperative variables should be applied mortality [29]. Exercise testing stresses the entire cautiously, especially since lung volume resection surgery cardiopulmonary and oxygen delivery systems and assesses for emphysema [34], applied at the same time as surgery for the reserve that can be expected and may be needed after lung cancer, may allow patients with marginal lung function, surgery [30]. A resting saturation of less than 90% or previously considered inoperable, to undergo lung resection desaturation of at least 4% during exercise are significantly [35]. predictive of increased mortality and morbidity [31]. Risk stratification analysis, using a multifactorial cardiopulmonary risk index based on conventional cardiac ACKNOWLEDGMENTS and pulmonary clinical data, was highly predictive of postoperative cardiopulmonary complications [32]. The author was supported in part by a fellowship from the Despite the need for accurate estimation of the risk of British Columbia Lung Association and by the Vancouver complications after major lung resection, the use of General Hospital Foundation. He is grateful for the guidance traditional methods of assessing operative risk provides and support provided by Dr. Raja Abboud in helping him only a modest ability to predict postoperative morbidity to review and revise the manuscript. The author also thanks and mortality in patients with significant impairment. Newer Dr. Sunny Dong and Dr. Peter Pare for their advice and tests such as pulmonary hemodynamic measurements are help, and Dr. Harry Joe for his help with the statistical not widely used because they are expensive and labor analysis and graphical presentation of data. intensive, and few data are available to assess their accuracy. The recent literature clearly demonstrates that DLCO is an important predictor of morbidity and mortality [5–8]. DLCO REFERENCES reflects pulmonary alveolar-capillary surface area and pulmonary capillary blood volume, provided that it is 1. Lange P, Nyboe J, Appleyard M, et al. Ventilatory function corrected for any decrease in hemoglobin content. A low and chronic mucus hypersecretion as predictors of death from DLCO indicates the presence of significant emphysematous lung cancer. Am Rev Respir Dis 1990;141:613–7. changes in the lung [15,16,33] and a reduced pulmonary 2. Kadri MA, Dussek JE. Survival and prognosis following capillary bed [16]. The use of DLCO in addition to clinical resection of primary non-small cell bronchogenic carcinoma. data, spirometry, and lung volume assessments will result Eur J Cardiovasc Surg 1991;5:132–6. 3. Miller JI. Physiologic evaluation of pulmonary function in the in improved prediction of the outcome following lung candidate for lung resection. J Thorac Cardiovasc Surg 1993;105: resection, especially pneumonectomy. 347–52. Untreated lung cancer is a disease with 100% mortality. 4. Damhuis RA, Schutte PR. Resection rates and postoperative Lung resection, the best and potentially curative therapy mortality in 7,899 patients with lung cancer. Eur Respir J 1996; for non-small-cell lung cancer, carries risks, especially when 9:7–10. performed on elderly patients with coexistent 5. Ferguson MK, Little L, Rizzo L, et al. Diffusing capacity predicts morbidity and mortality after pulmonary resection. cardiopulmonary disease. We advocate the routine use of J Thorac Cardiovasc Surg 1988;96:894–900. diffusing capacity in preoperative assessment prior to lung 6. Markos J, Mullan BP, Hillman DR, et al. Preoperative resection as an additional test to help identify patients at assessment as a predictor of mortality and morbidity after increased risk, for whom increased efforts in postoperative lung resection. Am Rev Respir Dis 1989;139:902–10. care may help to reduce morbidity and mortality. 7. Pierce RJ, Copland JM, Sharpe K, Barter CE. Preoperative risk Patients whose lung function is below the safe evaluation for lung cancer resection: predicted postoperative product as a predictor of surgical mortality. Am J Respir Crit preoperative limits for lung resection should not be rejected Care Med 1994;150:947–55. for surgical therapy, because without surgery, the outlook 8. Ferguson MK, Reeder LB, Mick R. Optimizing selection of for lung cancer is poor. They should be given intensive patients for major lung resection. J Thorac Cardiovasc Surg therapy to improve lung function. They should also be 1995;109:275–83. nd counseled as to the risks of death or prolonged disability 9. Bousamra M 2 , Presberg KW, Chammas JH, et al. Early and late morbidity in patients undergoing pulmonary resection from surgery, given other treatment options, and offered with low diffusion capacity. Ann Thorac Surg 1996;62:968–75. the option of resection using all the postoperative 10. Crapo RO, Morris AH, Gardner RM. Reference spirometric management skills if the risk for surgery is not considered values for spirometry using techniques and equipment that meet prohibitive. However, the previous concept of threshold ATS recommendations. Am Rev Respir Dis 1981;123:659–64.

444 Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 DLCO and postpneumonectomy complications

11. Crapo RO, Morris AH, Clayton PD, Nixon CR. Lung volumes in 24. Gaensler EA, Cugell DW, Lindgren I, et al. The role of healthy non-smoking adults. Bull Eur Phys Resp 1982;18:419–25. pulmonary insufficiency in mortality and invalidism following 12. Miller A, Thornton JC, Warshaw R, et al. Single breath diffusing surgery for pulmonary tuberculosis. J Thorac Cardiovasc Surg capacity in a representative sample of the population of 1955;29:163–87. Michigan, a large industrial state. Am Rev Respir Dis 1983;127: 25. Zibrak JD, O’Donnell CR, Marton K. Indications for pulmonary 270–7. function testing. Ann Intern Med 1990;112:763–71. 13. American Thoracic Society. Standardization of spirometry: 26. Drings P. Preoperative assessment of lung cancer. Chest 1989; 1987 update. Am Rev Respir Dis 1987;136:1286–96. 96(1 Suppl):S42–4. 14. Ogilvie CM, Forster RE, Blakemore WS, Morton JW. A 27. Mittman C. Assessment of operative risk in thoracic surgery. standardized breathholding technique for the clinical Am Rev Respir Dis 1961;84:197–207. measurement of the diffusing capacity of the lung for carbon 28. Marshall MC, Olsen GN. The physiological evaluation of the monoxide. J Clin Invest 1957;36:1–17. lung resection candidate. Clin Chest Med 1993;14:305–20. 15. Morrison NJ, Abboud RT, Ramadan F, et al. Comparison of 29. Lewis JW Jr, Bastanfar M, Gabriel F, Mascha E. Right heart single breath carbon monoxide diffusing capacity and function and prediction of respiratory morbidity in patients pressure-volume curve in detecting emphysema. Am Rev undergoing pneumonectomy with moderate severe Respir Dis 1989;139:1179–87. cardiopulmonary dysfunction. J Thorac Cadiovasc Surg 1994; 16. Morrison NJ, Abboud RT, Muller NL, et al. Pulmonary capillary 108:169–75. blood volume in emphysema. Am Rev Respir Dis 1990;141:53–61. 30. Olsen GN, Weiman DS, Bolton JW, et al. Submaximal invasive 17. Jones RS, Meade F. A theoretical and experimental analysis of exercise testing and quantitative lung scanning in the anomalies in the estimation of pulmonary diffusing capacity evaluation for tolerance of lung resection. Chest 1989;95: by the single breath method. Q J Exp Physiol 1961;46:131–43. 267–73. 18. American Thoracic Society. Single breath carbon monoxide 31. Ninan M, Sommers KE, Landreneau RJ, et al. Standard exercise diffusing capacity (transfer factor): recommendations for a oximetry predicts postpneumonectomy outcome. Ann Thorac standard technique. Am Rev Respir Dis 1987;136:1299–307. Surg 1997;64:328–33. 19. Daniel WW. Regression analysis: some additional techniques. In: 32. Epstein SK, Faling LJ, Daly BDT, Celli BR. Predicting Daniel WW, ed. Biostatistics: A Foundation for Analysis in the complications after pulmonary resection. Preoperative exercise Health Sciences, 7th ed. New York: John Wiley & Sons, 1999:536–52. testing vs a multifactorial cardiopulmonary risk index. Chest 20. Mould RF. Sensitivity and specificity. In: Mould RF, ed. 1993;104:694–700. Introductory Medical Statistics, 3rd ed. Bristol and Philadelphia: 33. Gelb AF, Gold WM, Wright RR, et al. Physiologic diagnosis of IOP publishing, 1998:232–8. subclinical emphysema. Am Rev Respir Dis 1973;107:50–63. 21. Zar JH. One-sample hypotheses. In: Zar JH, ed. Biostatistical 34. Cooper JD, Trulock EP, Triantafillou AN, et al. Bilateral Analysis, 4th ed. New Jersey: Prentice Hall, 1999:92. pneumectomy (volume reduction) for chronic obstructive 22. MacIntyre NR. Diffusion capacity of the lung for carbon pulmonary disease. J Thorac Cardiovasc Surg 1995;109: monoxide. Respir Care Clin North Am 1997;3:221–33. 106–19. 23. Cerfolio RJ, Allen MS, Trastek VF, et al. Lung resection in 35. McKenna RJ Jr, Fischel RJ, Brenner M, Gelb AF. Combined patients with compromised pulmonary function. Ann Thorac operations for lung volume reduction surgery and lung cancer. Surg 1996;62:348–51. Chest 1996;110:885–8.

Kaohsiung J Med Sci September 2003 • Vol 19 • No 9 445