Results

One hundred and seventy one patients were randomised into the study (90 in group A and 81 in group B). The groups were comparable in age (mean 67.0 years (range 41–85) in group A, mean 67.8 years (range 38–88) in group B) and sex distribution (62% men in both groups). Overall, 134 patients (78%) were diagnosed as having a malignancy (72 of 90 (80%) in group A; 62 of 81 (76.5%) in group B) and there was no significant difference between the groups in the type of malignant disease (table 1, p = 0.389).

In group A 68 of 90 patients (76%) underwent bronchoscopy as their initial investigation. The CT appearances suggested that an alternative investigation would be more appropriate in 16 patients (figs 1 and 2). Six patients in group A had no invasive investigations: two with presumed malignancy (one had a positive sputum and one had widespread cancer) and in four patients the CT scan was either normal or showed benign disease (such as bronchiectasis).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Group A: patients with malignancy.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Group A: patients with no malignancy.

Of the 68 patients in group A who underwent bronchoscopy as their initial diagnostic procedure, a tissue diagnosis was obtained in 50 (73%) compared with 44 of 81 patients (54%) in group B (figs 3 and4). Overall, a diagnosis was made from the initial invasive investigation in 64 of 84 patients (76%) in group A and 45 of 81 (55%) in group B (table 2, p = 0.005). All patients in group B underwent bronchoscopy in accordance with the agreed protocol.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Group B: patients with malignancy.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Group B: patients with no malignancy.

In patients with confirmed malignancy, bronchoscopy was diagnostic in 50 of 56 patients (89%) in group A (fig 1) compared with only 44 of 62 (71%) in group B (fig 3, table 3, p = 0.012). There was also a significant increase in the number of patients with malignant disease receiving a histological diagnosis on initial invasive investigation (63 of 70 (90%) in group A and 44 of 62 (71%) in group B; p = 0.004).

Of the 50 diagnostic bronchoscopies in patients with malignancy in group A, 46 patients were diagnosed by positive biopsy specimens of a visible tumour (31 also had positive washings, 12 also had positive brushings), one patient was diagnosed by a positive brushing alone of a visible tumour, and three patients were diagnosed on the basis of a “blind” washing. In the six patients in whom bronchoscopy was non-diagnostic, tumour was seen in two patients, biopsy specimens were taken in five, brushings in two, and washings in all six.

Of the 44 diagnostic bronchoscopies in patients with malignancy in group B, 38 patients were diagnosed by positive biopsy specimens of a visible tumour (28 also had positive washings, seven also had positive brushings), two were diagnosed by a positive washing of a visible tumour (one also had a positive brushing), and four were diagnosed on the basis of a “blind” washing (one of these also had a positive “blind” brushing). In the 18 patients with a negative bronchoscopic examination tumour was visible in four patients, biopsies were attempted in 15, brushings in nine, and washings in all 18 patients.

Overall, the improved diagnostic accuracy of bronchoscopy in group A appeared to relate primarily to a higher proportion of visible tumours, which is to be expected since patients in group A with no endobronchial component to their tumour on CT scanning would have had an alternative procedure. However, there was a higher rate of non-diagnosis when tumour was visible in group B (four of 44, 9%) than in group A (two of 49, 4%). In patients with no visible tumour there was no difference between the proportion of patients undergoing bronchial biopsies (five of seven in group A; 14 of 18 in group B) or bronchial brushings (three of seven in group A; 10 of 18 in group B).

Patients in whom malignancy was not found have been followed up for a period of at least 24 months and, as yet, no new evidence of malignancy has emerged.

Complications were rare and the rates were comparable in the two groups (table 4). There were no deaths or major haemorrhages secondary to the investigations in either group. A minor pneumothorax (classified as one that was asymptomatic, less than 30%, and did not require any specific treatment) occurred in three patients in group A and four in group B. A major pneumothorax requiring intercostal tube drainage occurred in one patient in group B.

Table 5 shows the average cost of investigations at Papworth for patients in group A, with universal CT scanning before bronchoscopy, compared with an estimated cost for patients in group B assuming a projected CT rate after bronchoscopy of 60%. The costs do not take into account overnight stays. At Papworth hospital both FOB and percutaneous biopsies are performed on a day care basis provided the patient is neither frail nor returning to an empty home.

Discussion

There have been a number of studies correlating CT scanning with bronchoscopic findings. In 1985 Colice et alreported that CT scans were moderately accurate in predicting airway abnormalities using 10 mm collimation (sensitivity 63–85%; specificity 61–77%),11 but they were inaccurate in defining the type of mucosal abnormality as they could not distinguish between localised mucosal abnormalities, endobronchial mass, or extrinsic compression. Similar conclusions were reached by Naidichet al.6

Improved accuracy was reported by Mayr et alusing 5 mm collimation.12 Their study of 142 patients with suspected endobronchial tumour showed sensitivities of 91–94% and a specificity of 99% in the diagnosis of a normal or narrowed bronchial lumen. They recommended CT scanning prior to bronchoscopy in patients with suspected carcinoma on the chest radiograph to guide the bronchoscopist and in those with a normal chest radiograph and normal or equivocal FOB if a bronchial tumour was still clinically suspected. With diagnostic performance improved by thin slice collimation (2–5 mm), the main advantage of spiral CT scanning is improved delineation of the extent of the tumour and its relationship to adjacent structures, together with enhanced detection of minor bronchial abnormalities. Further advantages of spiral CT scanning include multiplanar reformatting of volume data sets and three dimensional surface displays which are complementary to bronchoscopy.13 Our CT protocol also allowed formal staging of the tumour.

Despite the advantages of modern scanning techniques, FOB is required to detect subtle mucosal abnormalities and to obtain a tissue diagnosis. For these reasons it remains the investigation of choice for the inspection of major and segmental bronchi.14

The role of CT scanning in the investigation of haemoptysis has also been assessed in a number of studies.15-17 In this particular clinical setting CT scanning improves the diagnostic yield from bronchoscopy and should be the primary imaging investigation.

Our study is unique in that it is the only prospective investigation of the use of CT scanning prior to FOB in all patients with suspected endobronchial carcinoma and of its impact on management. Our results show that CT scanning before bronchoscopy benefits patients in a number of ways. A small number (7% of our patients) avoided the need for any invasive investigations as the CT scan was either normal or considered diagnostic. When bronchoscopy was performed a diagnosis was more likely when the bronchoscopist had the benefit of additional information from the CT scan on the site and characteristics of the presumptive tumour, allowing better targeting of the lesion. In patients without malignancy a CT scan allowed better targeting of abnormal areas. Finally, the number of investigations and days spent in hospital were reduced in patients undergoing CT scanning prior to bronchoscopy.

For ethical reasons the results of the CT scans in group B were reviewed immediately after bronchoscopy in all patients, thereby identifying the most appropriate second invasive investigation in patients with a non-diagnostic examination. This obviously contributed to the high histological confirmation rate in both groups, and is also likely to have reduced the apparent cost benefit of performing CT scans routinely before FOB (which is based on a projected CT scan rate of only 60% in group B).

During the period of our study we were not using transbronchial (Wang) needle aspiration routinely, but since this study we have begun using the technique to sample subcarinal nodes and/or extrabronchial masses at the time of the initial diagnostic bronchoscopy, and this is likely to increase further the cost effectiveness of CT scanning prior to bronchoscopy.

In a significant number of patients the CT appearances indicated either that there was unlikely to be an endobronchial component to a tumour and/or that a different invasive procedure (such as percutaneous biopsy) was more likely to be diagnostic than FOB, so reducing the proportion of patients at bronchoscopy without a visible tumour. In our study we attempted to exclude patients with an obvious peripheral mass amenable to percutaneous needle biopsy. In institutions where needle biopsy is less freely available, a higher proportion of patients are likely to undergo a bronchoscopic examination as their initial invasive investigation, and the benefit of prior CT scanning is therefore likely to be higher in this situation.

Although the theoretical risk of pneumothorax is greater in patients who undergo percutaneous biopsy than FOB biopsy, the complication rates were comparable in our two groups of patients. However, to detect a significant difference in complication rates similar to those shown in table 4 we would have needed to study 5550 patients.

The cost benefit analysis of performing CT scanning before FOB in all patients with suspected endobronchial carcinoma must allow for the fact that many patients with lung cancer undergo a CT examination after histological diagnosis to stage their disease, although the proportion of patients proceeding to a staging CT scan varies from centre to centre. The average cost of obtaining a histological diagnosis in group A was £525 per patient. Such a cost would equate to that of a projected cost of performing CT scans after bronchoscopy in approximately 60% of patients (table 5). In centres where CT scans are performed in about 60% of patients with proven or suspected lung cancer, it would appear to be cost effective to change to universal CT scanning before bronchoscopy since this significantly improves the diagnostic accuracy of bronchoscopy, allows about 7% of patients to avoid an unnecessary invasive procedure, and a further 10% to avoid the requirement for a second invasive investigation. In units where the named CT rate is higher or lower than 60%, cost effectiveness would be correspondingly higher or lower. Similarly, where the CT scan was used to guide additional staging techniques such as Wang needle aspiration of enlarged lymph glands, cost effectiveness would be increased further.

Our oncology service is characterised by a desire to provide a histological diagnosis and subsequent treatment plan for patients with suspected lung cancer in the shortest possible time and with the least inconvenience to patients. CT scanning before bronchoscopy in patients with suspected endobronchial carcinoma improves the diagnostic accuracy of the first invasive test with little increase in cost and no change in patient risk. While this approach may not be transferable to all other units, in units such as ours, where an average of 8–10 patients with suspected lung cancer are investigated each week, the benefits of running a concurrent CT scan/needle biopsy/bronchoscopy list are considerable, allowing flexibility and developing the multidisciplinary approach to lung cancer.