INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient selection criteria for lung volume reduction surgery (LVRS) have evolved steadily since the reintroduction of LVRS in 1994 by Cooper and coworkers (1), and currently focus on two objectives: (1) identification of patients with emphysema who are likely to survive the rigors of major thoracic surgery, despite having severe underlying lung disease; and (2) identification of that subset of patients who are likely to benefit significantly from the procedure. Those criteria initially proposed by Yusen and coworkers (2) were developed to ensure that LVRS is offered to candidates who are an acceptable operative risk, and for the most part, have been universally embraced. Criteria developed to identify "optimal LVRS candidates," those most likely to experience the largest improvements in lung function in response to surgery, have been more controversial (3).

Initial selection criteria of this type have focused on identifying patients with heterogeneous disease principally involving the upper lung fields (4). The rationale for selecting patients on the basis of such criteria is that the presence of "target regions" that contain damaged, overinflated lung (1) reduces overall lung recoil pressure and allows the chest wall to assume an elevated resting volume, (2) "compresses" surrounding, less diseased lung, reducing the volume of functional lung and lowering the vital capacity, and (3) increases resistance to expiratory airflow both by decreasing airway tethering, and by directly compressing airways in adjacent lung regions (3, 4).

Using an approach to patient selection based on preoperative radiological testing, several investigators have reported favorable outcomes 3 and 6 mo after LVRS (6, 8, 9). Improvements in both FEV₁ and FVC have consistently been greater among individuals with upper lobe predominant disease than among patients with more homogeneous or lower lobe disease, identified by either computed tomography scanning or ventilation-perfusion scanning (3, 4). Unfortunately, only a minority of patients with end-stage emphysema who satisfy other selection criteria for LVRS also demonstrate upper lobe predominant disease with identifiable target lesions amenable to resection. In some studies, application of these radiological criteria excludes up to 80% of patients with end-stage emphysema who might otherwise be potential LVRS candidates (2).

Although patient selection based on a requirement for focal upper lobe destructive changes has a sound rationale, physiological arguments, and more recent clinical data, suggest that many patients who do not possess upper lobe target lesions can still benefit significantly from LVRS. Hoppin has argued that by changing the relationship between recoil pressure and lung volume, LVRS improves expiratory flow rates at a given volume, resulting in improved flow dynamics independent of whether target lesions are present (10). Fessler and Permutt have further argued that potential responsiveness to LVRS is determined by the degree to which gas trapping occurs, as reflected in an elevated residual volume (RV)/TLC ratio, independent of whether this occurs diffusely throughout the lung, or focally within bullous lesions (11). Although, on average, patients with heterogeneous disease improve after LVRS to a greater extent than those with more homogeneous disease, significant overlap in responses has been noted among these two groups (5, 7). In a cohort of patients with end-stage emphysema, many of whom had homogeneous disease, we have previously shown that a physiological criterion, resistance to airflow on inspiration during resting breathing, correlated closely with the magnitude of improvement in lung function measured 6 mo postoperatively. Patients with relatively preserved airway structures, but reduced FEV₁ due to loss of elastic recoil and airway tethering, should have more normal resistance to airflow during inspiration when intrathoracic pressures act to distend, rather than compress, the airways. We hypothesized that these individuals should respond more favorably to LVRS than those with marked intrinsic airway narrowing due to bronchitis or bronchiolitis, or with extensive loss of parallel conducting pathways due to widespread tissue destruction. Although a low inspiratory resistance did correlate with greater improvement in 6-mo spirometry after LVRS in our original cohort of 29 patients, the radiographic pattern of disease distribution was not examined. Thus, it is unclear whether physiological measurements, such as inspiratory resistance, predict favorable outcomes among the same group of patients identified by radiological criteria, or whether they help identify a distinct group of patients who would be excluded by conventional radiological criteria, but who, as suggested by the theoretical arguments of Hoppin and Fessler, might still benefit from LVRS.

The present retrospective longitudinal study compares the ability of radiological and physiological criteria to predict outcomes to LVRS among patients with end-stage emphysema. The clinical utility of the two criteria was compared among patients with both heterogeneous and homogeneous disease, with the end point being improvement in FEV₁ at 6-mo follow-up. The sensitivity and specificity of the two criteria were considered individually, and then together in a multivariate model developed using stepwise linear regression analysis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection and Medical Characteristics

Fifty-two patients with end-stage emphysema underwent bilateral lung volume reduction surgery at our institution between October 1994 and May 1998. Fifty patients were available for 6-mo follow-up and agreed to participate in this study. Two patients with emphysema associated with ₁ antiprotease deficiency were excluded from this analysis. The study protocol, which involved preoperative and 6-mo postoperative spirometry and lung physiology assessments, and access to hospital medical records, was approved by our human subjects institutional review board. A participating physician obtained consent from patients at the time of initial screening for LVRS. Six-month follow-up was available with all 50 initial participants.

Patient ages ranged from 42 to 73 yr (mean, 56 ± 8 yr); 26 subjects were male. Baseline functional capacity assessed by Karnofsky scale was consistent with moderate to severe compromise (71 ± 6%). All patients reported a significant smoking history (mean, 68 ± 29 pack-years).

Three-quarters (37) of participating patients were oxygen dependent preoperatively. An additional two used oxygen with activity and sleep. All patients were receiving an inhaled -agonist, and the majority (47 of 48) also received an inhaled anticholinergic and inhaled steroid (43 of 48). More than half were also receiving an oral theophylline preparation (27 of 48), but only 12 patients were using oral steroids at the time of study enrollment. Forty-six patients completed a formal 6- to 8-wk pulmonary rehabilitation program prior to surgery.

Measurements of Lung Physiology

Preoperative spirometry was performed within 3 mo of surgery and at 6-mo follow-up. The follow-up time point was designated as that corresponding to 6 mo after completion of bilateral surgery, because the majority (40) of patients had sequential staged procedures. All pulmonary function studies were performed on a Morgan transfer factor volume displacement spirometer within 1 h of administration of bronchodilator therapy in accordance with American Thoracic Society (ATS) guidelines (12).

Measurements of inspiratory conductance were performed in our human subjects physiology laboratory as previously described (6). In brief, an esophageal balloon was passed through the nares to a distance of 40 cm. Appropriate intrathoracic positioning was demonstrated by observing transpulmonary pressure (Ptp) fluctuations during panting, deep inspiration, and Valsalva maneuver. Ptp was recorded using a ± 100-cm H₂O pressure transducer. Flow at the mouth was recorded with a Fleisch No. 1 pneumotachograph. Lung conductance during inspiration (G_Li) was determined as the "best fit" of recorded transpulmonary pressure (Ptp), volume (V), and flow () data to the equation of motion during the inspiratory phase of breathing:

where EL,I is inspiratory lung elastance.

For each subject, results for G_Li were determined for between 4 and 10 breaths during spontaneous breathing. The single final reported value represents the average of multiple runs. Only values for which the fit between experimental data and model fit was good (r² > 0.9 by multivariate regression analysis) were used in determining the final G_Li result.

Radiological Studies

Standard quantitative ventilation-perfusion scanning was performed after injection with macroaggregated technetium-labeled albumin. Only the perfusion scan results were utilized in the present study for grading disease distribution. Counts were apportioned to the upper, middle, and lower lung zones by dividing right and left lungs into three equally sized regions, and measuring total counts in each of these regions.

Several different methods were initially considered for identifying patients with heterogeneous upper lobe predominant disease. Each method evaluated the ratio of perfusion of upper to lower lung fields, generating an upper/lower perfusion ratio (ULPR) index. Method 1 utilized the ratio of upper lobe perfusion to lower lobe perfusion (U/L), neglecting the magnitude of perfusion within the middle lung field. Method 2 utilized the ratio of upper lung field perfusion to the average perfusion of the middle and lower lung fields (U/[0.5 × (M + L)]). Method 3 considered absolute hypoperfusion of the upper lung field, without regard to perfusion in other lung fields, by assuming that in the supine position, approximately one-third of the total perfusion should go to the upper lung field. By Method 3, this index is calculated as (U/33.3). Although we subsequently observed that each of these methods had some merit and correlated with improved FEV₁, the best correlations with outcome were observed with Method 2, particularly among patients with heterogeneously distributed emphysema. Therefore, for the purpose of this study, patients were classified as having homogeneously distributed disease if their upper to lower region perfusion ratio (ULPR) as defined by method 2 was between 0.75 and 1.25. Patients with ULPR indices outside this range were classified as having heterogeneous disease.

Surgical Methodology

Volume reduction surgery among this cohort was performed by five different surgeons. Forty-two of 48 patients underwent volume reduction using the no-cut autologous buttress plication method of Swanson and coworkers (13). Six additional patients underwent LVRS in which both staple plication and resection were performed. Forty-two underwent sequential thoracoscopic procedures separated by 4.2 ± 5.0 mo (range, 2 to 16 mo). Seven patients underwent a single bilateral thoracoscopic plication procedure. Average length of postoperative stay was 6.7 d (per visit) for sequential unilateral procedures, and 8.7 d for bilateral procedures.

Statistical Analysis

Results are reported as means ± standard deviations. Comparisons of preoperative and 6-mo postoperative data were performed by paired t test analysis. Comparisons between different subgroups of the cohort at a given time point were performed by Student t test. Comparisons between multiple groups were performed by one-way analysis of variance (ANOVA). Univariate and multivariate regression analyses were performed by a least-squares approach (StatMost software; Dataxiom, Los Angeles, CA). Statistical significance was defined as p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lung Function before and 6 mo after Lung Volume Reduction Surgery

Preoperative spirometry showed evidence of severe expiratory flow obstruction, with FEV₁ (0.66 ± 0.22 L, 23 ± 7% predicted), FVC (2.09 ± 0.79 L, 57 ± 16% predicted), and FEV₁/ FVC (0.33 ± 0.08) ratio significantly reduced compared with predicted values. Preoperative inspiratory conductance (G_Li) during spontaneous breathing was also markedly reduced (0.13 ± 0.06 L/s/cm H₂O) compared with normal (normal values range between 0.5 and 2 L/s/cm H₂O) indicating a significant component of intrinsic airways disease among this cohort of patients.

Statistically significant improvements in FEV₁, FVC, and FEV₁/FVC ratio were observed 6 mo after lung volume reduction surgery. FEV₁ increased 35 ± 40% (0.88 ± 0.37 L, p = 0.0005), FVC increased 25 ± 33% (2.53 ± 0.93 L, p = 0.0004), and FEV₁/FVC ratio increased 9 ± 3% (0.36 ± 0.11, p = 0.021). The distribution of responses among the cohort is summarized in Figure 1. Fifty percent of patients improved their FEV₁ by greater than 25% in response to surgery. Results reported here are similar to those observed in our original smaller cohort, as well as those reported by other investigators (3).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Distribution of responses to LVRS at 6-mo follow-up. Response is measured as percent improvement in FEV1. Forty-eight percent of patients (23 of 48) in this cohort demonstrated a greater than 25% increase in FEV1 after LVRS. Nine of 10 patients who experienced the largest improvements in spirometry after LVRS (a greater than 75% increase in FEV1 at 6 mo) had heterogeneous upper lobe disease by / ULPR criterion. Nevertheless, many patients with homogeneous disease experienced significant improvements in FEV 1. Solid column, homogeneous; open column, heterogeneous.

Ventilation-Perfusion Scan ULPR Scores

Ventilation-perfusion scan upper to lower perfusion ratios (ULPRs) were determined as described above. Results for each of the three methods of assessment are summarized in Table 1. Average scores were similar for each of the three methods used to assess ULPR (no significant difference; p > 0.3 by ANOVA). All methods of ULPR determination demonstrated, on average, a reduction in perfusion to the upper lung fields relative to lower lung fields (i.e., average ULPR was < 1.0 for each method). If homogeneous disease is defined as that associated with an ULPR between 0.75 and 1.25, Method 1 identified 12 emphysema patients as having homogeneous disease, Method 2 identified 21 patients, and Method 3 identified 14 patients.

Relationship between Preoperative ULPR Index and Improvement in Function 6 mo after LVRS

Relationships between each of the ULPR indices and 6-mo postoperative improvement in spirometry after LVRS are included in Table 1. For the cohort as a whole, ULPR scores determined by Method 3 (U/33.3) (r = 0.42, n = 48, p = 0.002) and Method 2 (r = 0.38, n = 48, p = 0.008; shown in Figure 3) correlated with lung function in a statistically significant, although weak, fashion. The relatively weak relationships between / assessment of disease distribution and improvement in lung function after LVRS are similar to findings of previous investigators (2).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Relationship between ULPR score and change in FEV1 in response to LVRS measured 6 mo postoperatively. Responses have been categorized as heterogeneous upper lobe, homogeneous, and heterogeneous lower lobe. Overall, the relationship between ULPR heterogeneity and FEV1 was poor (r = 0.38, n = 48, p = 0.008). Patients with upper lobe heterogeneous disease tended to show larger increases in FEV1 than did patients with either homogeneous disease or lower lobe predominant disease, although there was significant overlap among all groups.

Among individuals designated by / scanning as having a heterogeneous distribution of disease, a significant correlation remained between ULPR and change in FEV₁ 6 mo postoperatively (r = 0.40, p = 0.035, n = 27). By contrast, the correlation with 6-mo improvement in FEV₁ among patients with homogeneous disease was poor (n = 21, r = 0.15, p > 0.15).

Relationship between Inspiratory Conductance during Spontaneous Breathing and Improvement in Lung Function 6 mo after Lung Volume Reduction Surgery

Inspiratory conductance was significantly correlated with 6-mo improvement in FEV₁ for the cohort as a whole (r = 0.53, n = 48, p < 0.001; Figure 2). The correlation with postoperative improvement in FEV₁ was slightly better among patients with homogeneous disease (r = 0.56, n = 21, p = 0.008), although the correlation between G_Li and FEV₁ among patients with heterogeneous disease was also significant (r = 0.51, n = 27, p = 0.02).

View larger version (19K): [in this window] [in a new window]   Figure 2.   Relationship between inspiratory conductance and improvement in FEV1 at 6 mo after LVRS. Linear regression analysis for this cohort of 48 patients demonstrated a significant relationship between GLi and FEV1; r = 0.53 and p < 0.001. The majority of patients with improvements in FEV1 >0.4 L had heterogeneous disease (squares), although a significant overlap in responses is noted, with many patients classified as having homogeneous disease showing significant increases in FEV1 6 mo after LVRS.

Mean improvements in FEV₁ after LVRS were significantly larger among patients with heterogeneous disease (FEV₁ = 0.28 ± 0.27 L, n = 26; 41% improvement) than among those with homogeneous disease (FEV₁ = 0.14 ± 0.14 L, n = 26; 26% improvement; p = 0.017 by Student t test). For each group, the improvements were statistically different from 0 (paired t test; homogeneous [Ho] group, p = 0.00014; heterogeneous [He] group, p = 4 × 10⁵).

Among the 22 patients with heterogeneous disease that was confined largely to the upper lobes (ULPR < 0.75), response to LVRS was even more pronounced (FEV₁ = 0.34 ± 0.26 L; percent increase, 50%), consistent with findings reported by centers that utilize this criterion in their patient selection process (1, 5, 6, 8, 9).

Multivariate Regression Analysis Using both ULPR and GL,I as Independent Variables to Predict FEV₁ 6 mo after LVRS

Our results suggest that among patients with heterogeneously distributed changes of emphysema, / scan results correlated best with 6-mo postoperative improvement in FEV₁, whereas among those with homogeneous disease, G_Li correlated best with 6-mo improvement. On the basis of these observations, we hypothesized that these two independent predictors of outcome could be combined into a single algorithm, which could be applied to identify potential responders to LVRS, independent of disease distribution.

Multivariate stepwise linear regression analysis demonstrated that FEV₁ correlated significantly with both ULPR (defined by Method 2) and G_Li according to the relationship:

(1)

where p = 0.0001 for the G_Li correlation, and p = 0.019 for the ULPR correlation. Together, these criteria accounted for 33% of the variance in postoperative FEV₁ among this cohort (r² = 0.331).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study compares the ability of preoperative radiological and physiological screening criteria to predict lung function improvement 6 mo after LVRS in a cohort of 48 patients with severe end-stage smoking-related emphysema (preoperative FEV₁ = 0.66 ± 0.22 L) and limited functional capacity (Karnofsky functional status, 71 ± 6%). All patients had been screened preoperatively for LVRS in standard fashion, except that the presence of homogenous disease was not used to exclude potential candidates from surgical consideration. Improvement in lung function at 6 mo among the cohort as a whole was similar to what has been previously reported (5). FEV₁ increased 35 ± 40% (0.66 ± 0.22 versus 0.88 ± 0.37 L), and FVC increased 25 ± 33% (2.09 ± 0.79 to 2.52 ± 0.93 L). On the basis of results of quantitative ventilation perfusion scanning, 21 patients were identified as having homogeneous disease, and 27 as having heterogeneous disease. In 22 of 27 patients, the emphysema was upper lobe predominant. Preoperatively, patients with homogeneous and heterogeneous disease were clinically indistinguishable. FEV₁ values were also nearly identical in these two subsets of patients preoperatively (FEV₁ = 0.63 ± 0.19 L [homogeneous] versus 0.68 ± 0.26 L [heterogeneous]). However, improvements in lung function were, on average, greater among patients with heterogeneous disease (FEV₁ = 0.28 ± 0.27, 41% improvement) than among those with homogeneous disease (FEV₁ = 0.14 ± 0.14 L, 22% improvement). Among those patients with heterogeneous disease that was upper lobe predominant (ULPR < 0.75), improvements were even larger (FEV₁ = 0.34 ± 0.26 L, 50%). These findings confirm results previously reported by other investigators regarding the relationship between disease distribution and response to LVRS (1, 4), and argue that the presence of upper lobe predominant emphysema is, in fact, a preoperative indicator of favorable outcome.

Despite this, significant overlap was observed in the extent to which lung function improved after LVRS between patients with homogeneous and heterogeneous disease (Figures 2 and 3). Eighteen of 27 (67%) patients with heterogeneous disease (17 of 18 had upper lobe predominant disease) experienced a  15% improvement in FEV₁, whereas 11 of 21 (52%) patients with homogeneous disease also experienced similar physiological improvements. Radiological screening criteria applied so as to restrict surgery to patients with heterogeneous, upper lobe disease would have reduced the total number of LVRS responders by 38% (29 responders to 18 responders), and excluded 54% of patients (26 of 48) from surgical consideration. These results argue that although use of radiological criteria may help identify patients most likely to respond with large improvements in FEV₁ after LVRS, these criteria restrict LVRS to a small subset of the potentially larger group of emphysema patients who could still benefit from it.

We have previously shown that measurements of airflow conductance during inspiration can identify potential responders to LVRS, independent of disease distribution (5). Results presented here, which include data from the 29 patients in our original study as well as an additional 19 patients, confirm our previous observations in a larger cohort. Current findings further suggest that inspiratory conductance helps to specifically identify those patients with homogeneous disease who can benefit from LVRS, the same group that would be excluded from surgical consideration on the basis of standard radiological screening tests. G_Li correlated well with improvements in FEV₁ at 6 mo for the group as a whole (r = 0.53, n = 48, p < 0.001), and correlated equally well with physiological response among individuals with homogeneous (r = 0.56, n = 21, p = 0.008) and heterogeneous disease FEV₁ (r = 0.50, n = 27, p = 0.015). By comparison, ULPR assessed by ventilation-perfusion scanning correlated poorly with FEV₁ among patients with homogeneous disease (r = 0.15, n = 21, p > 0.15), but better with surgical response among patients with heterogeneous disease (r = 0.39, n = 27, p = 0.04).

Together these results suggest that radiological and physiological criteria help identify distinct populations of emphysema candidates who are likely to respond to LVRS with an improvement in FEV₁. Strictly objective measures from quantitative ventilation-perfusion scanning can be used to identify a significant percentage of potential LVRS responders (18 responders with heterogeneous disease among 29 total LVRS responders) on the basis of findings suggesting upper lobe predominant disease. Seventy-seven percent of patients (17 of 22) defined as having heterogeneous upper lobe predominant disease who underwent LVRS improved their FEV₁ by at least 15% at 6-mo follow-up. By applying physiological selection criterion to the 21 patients with homogeneous disease, and utilizing inspiratory resistance criteria defined in our original study (R_Li < 10 cm H₂O/L/s, or G_Li > 0.1 L/s/cm H₂O), an additional 11 patients were identified who improved their FEV₁ after LVRS.

Multivariate regression analysis was used to combine these two distinct evaluation techniques into a composite preoperative screening algorithm [Eq. (1)]. The resulting expression accounts for 33% of the variance in postoperative improvement in lung function. In this model, both preoperative G_Li (p = 0.0001) and ULPR index (p = 0.019) correlated significantly with FEV₁ at 6 mo. The model suggests that for each 0.1 L/s/cm H₂O unit change in preoperative G_Li that is observed between any two patients, the anticipated postoperative difference in FEV₁ improvement would be 200 ml (i.e., 0.1 × 2.159). For each 1-unit difference in preoperative ULPR index that is observed between patients, the anticipated postoperative difference in FEV₁ improvement would be 129 ml (i.e., 1.0 × 0.129). A higher preoperative inspiratory conductance is associated with a larger anticipated improvement in FEV₁. A higher preoperative ULPR index score (less upper lobe emphysema) is associated with a lower anticipated improvement. By way of example, a patient with a preoperative FEV₁ of 0.75 L, inspiratory conductance of 0.12 L/s/cm H₂O, and ULPR of 0.75 would be expected to experience a 25% improvement in FEV₁ at 6-mo follow-up. A patient with the same ULPR index but an inspiratory conductance of 0.05 L/s/cm H₂O would be predicted to experience only a 7% improvement in FEV₁. Finally, reduced G_Li of 0.05 L/s/cm H₂O but a UPRL indicative of upper lobe predominant disease (i.e., value of 0.4), would be expected to improve lung function by 15%.

Several characteristics of this multivariate model are noteworthy, and potentially controversial. It tends to weight G_Li to a greater extent than ULPR, whereas in practice, and in our own data set, evidence of upper lobe predominant disease predicts the largest response to LVRS. This property of the model is due to two factors other than the simple scaling differences that exist between the G_Li index and ULPR index. First, G_Li correlated better with the dependent variable FEV₁ than did the ULPR index, and therefore it receives a stronger weighting statistically with respect to its contribution to change in FEV₁. Second, our results suggest that the majority of patients with upper lobe predominant disease also have higher values of inspiratory conductance compared with those with homogeneous or lower lobe disease. Thus, it was unusual in our cohort to observe patients with upper lobe predominant disease who had a reduced inspiratory conductance and yet still improved their FEV₁ significantly at 6 mo (seen in only 4 of 48 patients). These findings suggest that the G_Li criterion identifies a significant portion of patients with heterogeneous disease likely to respond to LVRS, thus overlapping with the patient cohort identified by ULPR. However, G_Li also identifies patients with homogeneous disease who are likely to respond, but cannot be identified radiologically.

The / criteria established here are distinct from most criteria previously reported. Prior studies have primarily used assigned readers who apply semiquantitative scoring systems to interpret either / scanning or chest computed tomography (CT) scans (2, 3). Although this approach has the advantage of incorporating reader experience into the interpretive process, it is subject to the limitations of interreader variability. From a clinical perspective, it requires the availability of a radiology staff who are willing and able to apply such interpretive strategies. The method proposed here is strictly objective, and based on quantitative ventilation-perfusion scan results. Thus, it can be used by any clinician without special training or knowledge. This objective approach compares favorably with results reported using more subjective scoring systems. Wang and coworkers (3) have previously observed a significant correlation between a semiquantitative index of upper lobe predominant disease and 6-mo improvement in FEV₁ among a cohort of 96 subjects with end-stage emphysema (r = 0.38, p < 0.001). We observed a similar relatively weak, and less significant correlation in our group of 48 patients (r = 0.38, p = 0.04).

On the basis of the findings of this study, we have developed a simple, clinical screening approach, summarized in Figure 4, that can be used to help evaluate patients being considered for LVRS. Patients with end-stage emphysema demonstrating evidence of heterogeneous, upper lobe predominant disease who satisfy other conventional selection criteria can be considered as appropriate candidates for LVRS, independent of G_Li. Thus, detailed lung mechanics studies are unnecessary in the screening process. By contrast, patients with homogeneously distributed emphysema should not, a priori, be excluded from consideration for LVRS, but rather should undergo assessment of lung mechanics to determine G_Li. If conductance during spontaneous breathing is greater than 0.1 L/s/cm H₂O, then patients could be considered candidates for LVRS despite the radiographic evidence of homogeneous disease. In the present study, this criterion had a sensitivity of 85% (11 of 13) for identifying patients with homogeneous emphysema who improved after LVRS. By contrast, patients in this group who had a G_Li less than 0.1 L/s/cm H₂O were unlikely to benefit from LVRS, and were not good candidates for this procedure. In fact, only one patient (1 of 8; specificity, 88%) with homogeneous disease and reduced inspiratory conductance experienced improvement in FEV₁ at 6 mo.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Suggested simple approach combining radiological and physiological screening criteria in the assessment of patients being considered for lung volume reduction surgery. Although this approach does not accurately predict the magnitude of improvement in FEV1 at 6-mo follow-up (r2 = 0.331), it does help discriminate between patients likely to have a favorable response to LVRS at 6-mo follow-up and those who are unlikely to do so. Among patients with heterogeneous upper lobe disease identified by this approach, 77% had a greater than 15% improvement in FEV1 after LVRS. Only 20% of patients (one of five) with heterogeneous lower lobe disease improved their FEV1 values after LVRS. Among patients with homogeneous disease and GLi > 0.1, 85% improved their FEV1 by a similar amount. Finally, among patients with homogeneous disease and GLi < 0.1, 12% (one of eight) improved their FEV1 in response to LVRS. Thus, this method identifies significantly more patients who can benefit from LVRS than radiological criteria alone, while also identifying the majority of individuals who are unlikely to benefit from this procedure.

Although this algorithm is helpful for identifying specific patient traits associated with potential responsiveness to LVRS, out data clearly indicate that factors other than those considered here are important in determining physiological responses to LVRS. Combined preoperative radiological and physiological indices accounted for only 33% of the variance in FEV₁. Surgeon experience, intraoperative and postoperative complications, pulmonary vascular reserve, degree of preoperative hyperexpansion, and preoperative nutritional status, as well as other factors, could have contributed to the large percentage of variability in response not accounted for by the independent variables considered in the present study.

Despite these limitations, the results presented here suggest that objective radiological and physiological criteria can be used together to help evaluate patients preoperatively for consideration of LVRS. Radiological criteria identify patients with more heterogeneous, upper lobe predominant disease that responds best to LVRS. Good airflow during inspiration, as assessed by measurement of inspiratory conductance, identifies some of the same patients because upper lobe predominant disease was frequently associated with this more favorable physiological pattern. Furthermore, G_Li criteria identified 11 additional patients with homogeneous disease who responded favorably to LVRS, and who would have been excluded from LVRS on the basis of radiological screening alone. Combining objective radiological and physiological criteria can help identify potential candidates for LVRS and extend this beneficial procedure safely to as many subjects as possible.