INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients with lung cancer who undergo evaluation for lung resection are usually smokers and therefore often have concomitant heart and lung disease. This puts them at increased risk for the development of postoperative complications, including permanent respiratory disability and death. Therefore, the functional assessment of their cardiorespiratory reserves is very important. This assessment has recently undergone major changes; apart from preoperative measurement of pulmonary function (FEV₁, diffusing capacity of the lungs for carbon monoxide [DL_CO]), exercise testing with the determination of maximal oxygen uptake (O₂max) has gained increasing importance. Recently the emphasis has shifted to the prediction of postoperative function (predicted postoperative = ppo). The parameter, most firmly established, is the FEV₁-ppo, but also DL_CO-ppo (1), and most recently O₂max-ppo (3) have been suggested.

Despite increasing enthusiasm for exercise testing and split function studies, one has to remember that many patients undergoing lung resections up to pneumonectomy do not need extensive tests for the assessment of their cardiorespiratory reserves. We therefore developed an algorithm for the preoperative functional evaluation (Figure 1) incorporating the cardiac history including an electrocardiogram (ECG), and the three parameters FEV₁, DL_CO, and CO₂max with their ppo values. The choice of these parameters of cardiorespiratory function and their respective cutoff values for operability were based on our previous investigations as described in METHODS (3, 4), and on the recent literature (1, 5).

View larger version (26K): [in this window] [in a new window]   Figure 1.   Algorithm used for the assessment of functional operability of lung resection candidates. The numbers indicate the number of patients having undergone different functional tests, the figures at the bottom indicate their respective extent of potential pulmonary resection. The two patients who underwent cardiac evaluation after exercise testing had shown ischemic heart disease, are indicated in italic and in parentheses to avoid double counting. O 2max = maximal oxygen uptake on exercise.

The goal of the current study was to prospectively assess the impact of this algorithm on postoperative morbidity and mortality in a consecutive series of lung resection candidates scheduled for standard thoracotomy. The validation of the algorithm was done by comparing the postoperative complication rates with the ones obtained in our previous series (4) and with other reports in the literature.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

All patients enrolled in this study were referred to the Respiratory Division of our hospital for investigation and treatment of suspected or proven lung cancer. A total of 137 consecutive patients (102 males and 35 females) with resectable pulmonary lesions scheduled for lung resection by thoracotomy underwent functional assessment between January 1993 and September 1996.

Algorithm

The following algorithm was used in the preoperative functional evaluation (Figure 1): Any patient with active or suspected cardiac disease had first to undergo a thorough cardiac workup and if necessary even cardiac bypass surgery in case of ischemic heart disease. Only patients whose cardiac condition was amenable to treatment could undergo further investigation for pulmonary resection. Patients who had a negative cardiac history (including a normal ECG) and whose FEV₁ and DL_CO were both  80% of predicted normal could undergo lung resection up to a pneumonectomy without any further tests. The cutoff values of 80% of predicted were chosen because the respective postoperative values are usually > 40% of predictedidentified as safe (1, 2)even after pneumonectomy as the diseased lung generally contributes < 50% of the total lung function (8). If either FEV₁ or DL_CO was < 80% of predicted, exercise testing with the measurement of O₂max was performed. If O₂max was > 75% of predicted (4) or > 20 ml/kg/min (5, 6), patients again qualified for resection up to pneumonectomy. A O₂max < 10 ml/kg/min (3, 6) or < 40% of predicted (4) was generally considered prohibitive for any resection. All patients with O₂max values in between, that is 40 to 75% of predicted as well as 10 to 20 ml/kg/min, underwent split function studies to determine the ppo values for FEV₁, DL_CO, and O₂max. For the calculation of the remaining pulmonary function for each of these three parameters, the fractional contribution of the lung tissue to be resected with curative intent was deducted from overall preoperative lung function. This calculation was based on lung perfusion scans (see also SPLIT FUNCTION STUDIES).

If both FEV₁-ppo and DL_CO-ppo were < 40% of predicted the patient was deemed inoperable (1, 2); if either one or both was  40% of predicted, O₂max-ppo became the decisive factor. Patients with a O₂max-ppo of < 10ml/kg/min as well as < 40% of predicted were deemed inoperable (3); if it was either  40% of predicted or  10 ml/ kg/min, the patient was deemed operable up to the calculated extent of resection.

Pulmonary Function Tests

Pulmonary function tests (PFT) were performed while the patient was at rest in a seated upright position. These tests consisted of spirometry and body plethysmography using the Masterlab (Jäger, Würzburg, Germany). The measurement of the flow-volume loops was recorded according to American Thoracic Society (ATS) criteria (9). Of all recorded parameters the following two were used for the assessment of operability: the best expiratory volume in the first second (FEV₁) and DL_CO. In patients with obstructive airflow limitation, the PFT were repeated after 1 to 2 wk of intensive bronchodilator therapy including a steroid trial, and the best performance of each patient was used.

The term "percent predicted" used throughout the article refers to the relationship of the value obtained to the normal value for the patient's sex, age, and height. The term "postoperative predicted" refers to the value predicted for the patient after operation and can be expressed in absolute or in percentage of predicted values.

Exercise Testing

If either FEV₁ or DL_CO was < 80% of predicted, symptom-limited cycle ergometry was performed according to our algorithm (cycle: ER 900 L, Jäger; cardiopulmonary stress testing unit: EOS Sprint, Jäger). Baseline measurements were recorded after a minimal resting period of 3 min on the bicycle. The patients then started exercising at constant speed (60 rpm). After a 2 min warm-up period at 20 watts a ramp protocol with a 20 watt per minute workload increase was started. The exercise test was stopped when the patients were exhausted, a plateau in O₂ uptake appeared or at any signs (ECG) or symptoms of myocardial ischemia, including a fall in blood pressure. All parameters were recorded until the end of a 6-min recovery period. Continuous measurements of ventilation (E), oxygen consumption (O₂), carbon dioxide production (VCO₂), and pulse rate were averaged every 15 sec. Blood pressure was measured manually (Riva-Rocci) every minute. A precordial ECG (equivalent leads V₁, II, and V₅) was monitored continuously (Cardiotest EK 53 R; Hellige, Freiburg, Germany) and hard copies written at rest, at peak exercise, at the end of the recovery period, and additionally when arrhythmias or changes in the ST-T segments occurred. Arterial blood samples were drawn from the radial or brachial artery at rest and at peak exercise and analyzed immediately. Maximal oxygen consumption was recorded in L/min, ml/kg/min, and as a percentage of predicted normal values according to Jones and colleagues (10).

Split Function Studies

Split function studies of regional pulmonary function were performed while the patient was in a seated upright position and breathing normally at rest. All patients with O₂max values between 10 to 20 ml/kg/ min and 40 to 75% of predicted underwent additional radioisotope ventilation-perfusion studies to estimate the postoperative FEV₁ (FEV₁-ppo) according to the method described by Markos and coworkers (1). Using the perfusion lung scans (technetium-99m-labeled macroaggregates), we calculated the fractional contribution of the lung tissue to be resected to overall lung function. Thereby the following formulas were used: pneumonectomy: postoperative FEV₁ (FEV₁-ppo) = preoperative FEV₁ × (1  fractional contribution of the affected lung); lobectomy: postoperative FEV₁ (FEV₁-ppo) = preoperative FEV₁ × (1  lobar fractional contribution). The same technique was used to predict the postoperative diffusion capacity (DL_CO-ppo) (1) and the postoperative exercise capacity (O₂max-ppo) (3). Calculations were made both for absolute and for percentage of predicted values of FEV₁-ppo, DL_CO-ppo, and O₂max-ppo. For FEV₁-ppo and DL_CO-ppo only the percentage of predicted values were used in the algorithm.

Surgical Procedures, Postoperative Care, and Complications

All pulmonary resections from wedge resection to pneumonectomy were performed with an open thoracotomy; the same team of four thoracic surgeons performed all operations. All patients received standardized anticoagulant therapy with low-molecular-weight heparin (3,000 IU subcutaneously/day) perioperatively. After the operation all patients were admitted to the intensive care unit (ICU) for the first postoperative night. They were extubated either in the theater at the end of the operation or in the ICU within 2 to 4 h. If no complications occurred the patients were transferred to a normal thoracic surgery ward not later than 24 h postoperatively. The duration of hospital stay was calculated as 1 d prior to surgery until discharge. The decision to transfer a patient from the ICU to the normal ward and from there to discharge him or her from hospital was taken by the anesthetists and the thoracic surgeons respectively according to their clinical assessment. The postoperative period (30 d) was recorded for complications as defined in Table 1. Only moderate and severe complications expected to occur as a result of poor cardiopulmonary reserve were included in the list. For comparison's sake it contained the same criteria that were used in our previous reports (3, 4) and in some important earlier studies (1, 5, 6).

Statistical Analysis

The comparison of all variables between the current and our previous study results was made with the use of the chi-square test for categorical variables, and Student's t test for continuous variables. A significance level of p < 0.05 was chosen.

Functional Assessment in Our Previous Series

The equipment and test analyses for PFT, exercise testing, and split function studies were identical in our previous (4) and in the current study. The number of investigations per patient, however, differed between the two studies. In the previous series all patients underwent PFT and exercise testing; further, all patients with either an FEV₁ < 2 L or a DL_CO < 50% underwent additional radioisotope ventilation-perfusion studies to estimate the postoperative FEV₁ and DL_CO. All patients who were able to perform all investigations then underwent pulmonary resection irrespective of the functional test results. This was essential to analyze each test parameter for its predictive value of postoperative complications, and for the elaboration of cutoff values for safe resection.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Functional Evaluation

Six (4%) of the 137 patients (Figure 1) underwent first a thorough cardiac workup because of known or suspected cardiac disease; four of them were deemed operable from a cardiac point of view after radionuclide heart studies had ruled out active disease and confirmed a normal left ventricular ejection fraction (LVEF). Two were considered inoperable as their cardiac condition was not controllable by medical or surgical treatment. Sixty-seven patients (49%) who had a negative cardiac history (including a normal ECG) and whose FEV₁ and DL_CO were both  80% were deemed operable for resections up to a pneumonectomy without any further testing.

The remaining 68 patients underwent a symptom-limited cycle ergometry; thereafter 51 patients (37% of the initial 137, 75% of the 68) qualified for resection up to a pneumonectomy. Two patients with sufficient PFT reserves (italic numbers in Figure 1) showed marked ischemic ECG changes occurring during maximal exercise. One of them had known coronary artery disease and previous aortic valve replacement. A subsequent radionuclide ventriculography showed a stunned area but normal LVEF; he was therefore deemed operable but died of pneumonia and consecutive multiple organ failure 9 d after a left upper lobectomy. The second patient underwent coronary angiography, which revealed serious three-vessel disease; he underwent coronary artery bypass surgery, and subsequently a lobectomy with an uncomplicated postoperative course.

Seventeen patients (12%) with O₂max values between 10-20 ml/kg/min underwent split function studies to determine their ppo function according to our algorithm. In eight patients postoperative predicted lung function parameters were good enough for pneumonectomy; six patients could undergo resection up to a lobectomy only. In three patients the O₂max-ppo was < 10 ml/kg/min as well as < 40%, therefore they were deemed inoperable (Patients 1, 2, and 5 of Table 2).

Thus, of the consecutive group of 137 lung resection candidates, who were functionally assessed with our algorithm, five patients (4%)all with lung cancerwere deemed inoperable because of inadequate cardiopulmonary functional reserve (Table 2). Two of these five patients would have been scheduled for pneumonectomy, three for lobectomy. The parameters for which they were excluded from surgery are indicated in italic in Table 2. Two patients (Patients 1 and 2) had sufficient pulmonary function results, but very poor cardiopulmonary reserves reflecting a marginal cardiac output (low O₂max-ppo, and low LVEF on echocardiography). Three were excluded because of poor pulmonary function results (Patients 3-5); two of these three patients were unable to perform the exercise test, one of them would also have been excluded because of a low O₂max (Patient 5). The 1-yr follow-up of the spontaneous course of four of these five patients was complicated by recurring episodes of either acute respiratory failure or of congestive heart failure which necessitated hospitalization in each case; the respiratory failures were not directly tumor-related, but resulted from an exacerbation of the patient's chronic obstructive pulmonary disease (COPD). One patient had an uneventful course up to 1 yr after radiotherapy.

Procedures

Finally, 132 patients were considered operable and underwent lung resection by thoracotomy. The mean age was 62 yr (range 23 to 81). The baseline characteristics are summarized in Table 3; they were comparable to the ones obtained in our previous study group (4) (no differences on Student's t test). There were 99 (75%) men and 33 (25%) women. The majority of patients (n = 123) had clinically resectable malignancies whereas a few had benign disorders (n = 3) or carcinoid tumors (n = 6). The operations performed were 38 pneumonectomies, 3 bilobectomies, 82 lobectomies, and 9 segmental or wedge resections. There were no operative deaths. All patients were extubated within 24 h.

Postoperative Course

One hundred seventeen patients had an uneventful postoperative course. Postoperative complications (within 30 d) occurred in 15 patients (11%), of whom two died, which reflects an overall mortality rate of 1.5%. Of the 15 patients with complications, four had undergone a pneumonectomy, 10 patients a lobectomy, and one patient a segmental resection. No patient died after pneumonectomy, bilobectomy, or a resection of less than a lobe; the mortality of lobectomies was 2.5% (2 of 82). The two patients who died had a O₂max-ppo of < 10 ml/ kg/min (7.5 and 8.2 ml/kg/min respectively), but > 40% of predicted (41 and 46%). Both had undergone a lobectomy; one died of intractable respiratory failure and the other one of pneumonia and multiple organ failure. The complications are summarized in Table 4. In comparison with our previous study (4) the complication rate was reduced by 49% (from 20 to 11%) and the mortality decreased from 3.8 to 1.5% (Table 5). Neither of these decreases reached statistical significance.

All but three survivors were discharged by Day 30. In two patients the postoperative period was complicated by pulmonary embolism and in one patient by prolonged air leak. The mean stay in the ICU was 1.4 d (SD 1.8) and the total hospital stay 14.6 d (SD 5).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We tested in a prospective study the feasibility of an algorithm established for the functional evaluation of lung resection candidates. The results showed that adherence to the cutoff points of the various parameters reduced the postoperative complicationsmorbidity and mortalityby half (49%) in comparison to our previous series (4). This improvement was statistically not significant because of the low number of complications, especially the low mortality rate. Clinically, however, we consider a decrease in mortality from 3.8% to 1.5% very relevant, as it is very hard to lower an already low figure further while extending the limits of operability at the same time. An overall morbidity of 11% with a mortality of 1.5% compares favorably with most of the larger published series. Two large studies show even lower mortality than ours; Miller (11) had an overall mortality of 0.64% in a large series of 2,340 patients. Their mortality for pneumonectomy was 4.97%, for lobectomies 0.39%, for wedge resections 0.13, and for exploratory thoracotomies 0.59%. These excellent results may have been achieved, in part, through fairly rigorous patient selection, as for a pneumonectomy the FEV₁ had to be > 2 L and the FEV₁-ppo > 1 L. Another very low mortality rate of 0.9% was reported by Kearney and coworkers in a series of 331 patients (7). The bulk of the literature shows that there is a clear correlation between the extent of resection and postoperative morbidity and mortality. Segmental or wedge resections have the lowest and pneumonectomies the highest risk (1, 4, 11). In a large series by Loddenkemper comprising 433 patients, the mortality rates after pneumonectomy were 9.4% and after lobectomy 5% (12). In Markos's study the rates were 16.7% and 0% (1). At our institution an early series had respective rates of 7 and 4% for the years 1972-1978 (13), and 10 and 2.2% for the years 1991-1992 (4). In recent years, the mortality rates after lung resection have generally decreased although the limits of operability have been extended. In our opinion, these results have been achieved through better functional evaluation before lung resection, but also through improved perioperative care.

The two patients who died after a lobectomy in our study were known high-risk candidates with borderline operability. These two patients, as well as the three patients in our previous study (3) who died, all had an absolute O₂max-ppo value of < 10 ml/kg/min. We therefore slightly amended our algorithm in excluding patients with an absolute O₂max-ppo value of < 10 ml/kg/min, even if their percentage of predicted value was > 40%. On the other hand, we could lower the percentage of predicted O₂max-ppo value from 40% to 35%, as some patients had survived resections with values between 35-40% of predicted, provided that the absolute value was  10 ml/kg/min. In practice, this change makes it more difficult for elderly patients to qualify for operation. With this alteration we theoretically should have obtained a zero mortality without increasing the exclusion rate. Since the introduction of the amended algorithm in late 1996 (Figure 2 [15]), no patient has died from cardiorespiratory complications.

View larger version (28K): [in this window] [in a new window]   Figure 2.   Proposed algorithm for the assessment of the cardiore-spiratory reserves of lung resection candidates. Patients undergo successive steps from top to bottom, until they qualify for varying extents of resection or are deemed inoperable. The "safety loop" for patients with cardiac problems is indicated in the upper left-hand corner. The dashed line leading from exercise testing back to the cardiac workup is for patients with a negative cardiac history and a normal ECG, who show symptoms or signs of ischemia during exercise. Tl = thallium; Tc = technetium. (With permission from Reference 15.)

A very low mortality rate might of course reflect too rigorous patient selection. In our current study only 5 (4%) of all evaluated patients were deemed functionally inoperable, similar to 5% in the previous one (4). Further, all of these five patients were carefully followed up, and with the spontaneously occurring acute episodes of acute respiratory failure in three and congestive heart failure in one patient, it is probably fair to speculate that the stress of a lobectomy or pneumonectomy might have been fatal to them. We therefore think that the introduction of our algorithm did not unnecessarily exclude patients from surgery. The comparison with our previous study should be largely unbiased, as the baseline patient data were comparable to the current study, and the surgical team and technique as well as the perioperative management were identical. Thus the better outcome reflects the impact of the algorithm.

An important aspect is the cost involved in the preoperative investigation. Half of all patients had a negative cardiac history, a normal ECG, and an FEV₁ and DL_CO of  80% of predicted and thus qualified for a resection of up to a pneumonectomy without further investigations. Another 51 of the remaining 68 also qualified for a resection up to that extent after exercise testing. Thus 118 (86%) patients were potentially operable for a pneumonectomy after noninvasive cardiopulmonary testing. The need for radionuclide studies was reduced to 17 (12%) patients only. In our institution, exercise testing is about 20% cheaper than a radionuclide study and can be conducted in the pulmonary function laboratory. We therefore prefer to use it before radionuclide studies. This is in contrast to algorithms proposed by Bechard (16) and Nakahara and colleagues (17), where radionuclide studies are done before exercise testing. Overall, we think that our algorithm is economical especially when taking into account the spared costs due to the improved complication rates. Video-assisted thoracoscopic surgery (VATS) has been reported to be associated with lower complication rates than open thoracotomy (18), and might therefore help in further lowering the threshold of functional operability. Whether VATS can provide adequate intraoperative cancer staging is uncertain. Therefore, this procedure cannot become established as routine practice for lung cancer surgery as long as long-term survival rates are not available.

The latest development in the field of thoracic surgery is lung volume reduction surgery (LVRS) to improve pulmonary function in patients with severe emphysema (19). Such patientsdeemed inoperable according to our algorithmmight become eligible again through a combination of LVRS and simultaneous cancer surgery. First reports by Gonzalez Munoz and coworkers (20), Cooper and coworkers (21), and McKenna and coworkers (22) have shown the feasibility of this approach by removing functionally poor "target zones" of the lung with LVRS and simultaneously pulmonary nodules which were either within or outside these target zones. Some of these nodules were stage I lung cancer, which were mostly removed by wedge resection. With this new concept the limits of functional operability will potentially be taken another step further. It must be emphasized, however, that most of the reported patients (22, 23) had been scheduled for LVRS and their cancers were chance discoveries on preoperative computed tomographic (CT) scans; patients who primarily present with lung cancer and are secondarily found to be functionally inoperable may have more advanced cancer, which might be less amenable to minimal resections. The long-term follow-up will therefore have to carefully assess whether sublobar resections, which have been shown to be fraught with clearly higher local recurrence rates (24), will remain feasible in such patients. One good argument in favor of limited resections is that "functionally inoperable" patients might have a better outcome in terms of pulmonary function and even survival when offered LVRS with limited resection in comparison with a nonsurgical treatment modality, such as radiotherapy, chemotherapy, or just best supportive care.

In summary, our results have shown that adherence to our algorithm for preoperative functional evaluation before lung resection reduced morbidity and mortality by half without unnecessarily excluding patients from surgery. For patients with normal or only slightly abnormal PFT (FEV₁ and DL_CO) and a negative cardiac history, lung resections up to a pneumonectomy are clearly feasible without exercise or split function studies.