Statistics from Altmetric.com
Any patient presenting to a respiratory physician with a possible diagnosis of lung cancer requires a rapid and accurate histological diagnosis, together with enough staging information to enable a correct management plan to be arranged. Standards for these processes have been suggested.1 In practice it is incumbent upon physicians to assess each case and to determine the optimum combination of sampling and imaging tests that will be likely to achieve a firm diagnosis and staging at the minimum inconvenience to his or her patients, and with a minimum of delay which is known to be very distressing to them.2
Since the advent of fibreoptic bronchoscopy (FOB) in 1974, and with its current very wide availability, most physicians would consider this as their first investigation after a clinical assessment and plain radiology. Selection would be influenced by the latter, so that lesions clearly falling into the category of small solitary pulmonary nodules would be far more likely to be investigated by computed tomographic (CT) scanning and fine needle aspiration biopsy (FNA). For lesions of less than 2 cm in diameter FNA is superior to bronchoscopy even if imaging is used to guide the transbronchial biopsy or transbronchial needle aspiration.3 4
The probability that a lesion, thought by a physician to be accessible to bronchoscopy, can actually be diagnosed in this way is not easy to ascertain. However, a recent UK multicentre prospective study of 1660 consecutive cases investigated by FOB because of a prior likelihood of lung cancer showed that a definite tumour was seen in 57%.5 In a further 20% the appearances were very suggestive of a tumour. Thus, overall, one in five of these tests was negative. The proportion with a positive histological examination at bronchoscopy is likely to have been between 75% (diagnosis within seven days of bronchoscopy) and 85% (diagnosis up to 14 days). Only one in eight patients (15%) had had a prior CT scan, and whether or not this guided the bronchoscopist at all is not known.
This large study with a sensitivity for bronchoscopy of about 77% and a definite histological diagnosis rate of about 75% or better seems to suggest that the traditional way of assessing lung cancer should continue to be by bronchoscopy first, followed by a CT scan when indicated. However, a paper in this issue ofThorax by Laroche et al 6 from the Oncology Unit at Papworth suggests strongly that, where the facilities and organisation exist, there may be advantages in reversing this sequence at no greater cost and with a reduction in the number of invasive tests needed to make a firm histological diagnosis. This possibility has been suggested in several retrospective series7-9 but theirs is the first prospective study.
The authors studied a consecutive series of 171 patients thought on the basis of their basic examination and/or plain radiographs to have a high probability of tumour accessible to bronchoscopy. They showed that a prior spiral CT scan in the randomly allocated “test” population prevented any further tests in six of 90 patients (7%), increased the diagnostic yield of subsequent bronchoscopy to 75% (compared with 54% in the control group in whom bronchoscopy was performed before the result of the spiral CT scan was known), and increased the percentage of patients diagnosed after a single invasive test from 55% to 76%. If the diagnosis was eventually confirmed as lung cancer, 89% of patients were correctly sampled and diagnosed when bronchoscopy was done in the knowledge of the scan result compared with 71% when bronchoscopy was performed before the CT scan. The additional cost of performing spiral CT scans on each patient (given as £121 or US$195) was offset by the need for fewer other invasive tests as a result of the information available from the CT scan, even though they were more expensive—for example, the cost of an FOB was given as £387 (US$620) per case.
The important question then for all cancer units is whether this evidence is good enough to justify a change in routine practice and also whether it is generally practicable to do so.
The technical advances in fibreoptic bronchoscopes since 1974 have been essentially to reduce their diameter, increase their flexibility, and improve their angle of vision and optics. It is unlikely, however, that further changes will alter the performance of these instruments significantly. The application of fluorescence bronchoscopy is still a research tool for early diagnosis. Additional techniques such as perbronchoscopic needle biopsy have been studied intensively but are still not in widespread use because they are technically difficult and have not been shown conclusively to increase the sensitivity of the test, as surgeons still rightly prefer to stage patients preoperatively by mediastinal sampling.10 11
By contrast, there have been definite and continuing advances in imaging technology. The time taken to scan patients has reduced, and reconstruction technology has changed in a number of ways. Although conventional scanners are being replaced by spiral/helical scanners, not all of these machines have the same reconstruction ability. This has led to the use of various scanning protocols, although all the recommended techniques involve thinner sections through the main airways. Most units use 5 mm collimation rather than the 3 mm collimation used in the study by Laroche et al. The thinner section protocols allow the bronchial anatomy to be visualised very well and this is most vividly demonstrated when the reconstructions allow “virtual bronchoscopy”.
The practical issues for most units will inevitably be whether the putative cancer workload could be reorganised to allow same day CT scans and bronchoscopic examinations with no loss of CT or bronchoscopy “slots”. There is some preliminary evidence that this can be organised with benefit, even when the referral rate is fairly low—for example, 54 patients in 31 weeks in the study by Williamset al.12 Many units in the UK, and possibly in other countries, do have very busy scanning departments where 2–3 week delays in staging CT scans are not uncommon.13 The introduction of helical scanners could change this picture because of their greater throughput, but new techniques such as scanning for pulmonary emboli add more cases to the overall CT workload. Yet most patients with lung cancer do have a CT scan, so the challenge of providing early CT scanning is one of organisation.
A second general point, not specifically considered in the paper by Laroche et al, is whether CT scanning is better than bronchoscopy in the further investigation of a patient with significant unexplained haemoptysis and a normal examination and radiograph. The evidence here is more clear cut; several series, admittedly retrospective,14 15 have suggested that CT scanning is more sensitive than routine bronchoscopy for these patients, although at present many patients are referred from primary care specifically for bronchoscopy. Prior CT scanning should aid the bronchoscopist, particularly in the less straightforward case, but it will not completely obviate the need for bronchoscopy as studies have shown that endobronchial disease is missed by spiral CT scanning. This concerns not only in situ disease but includes endobronchial lesions, particularly in subsegmental airways.
The study by Laroche et al is important because it shows once again the advantage of assessing and managing patients with cancer in a multidisciplinary way. Although confirmation of these results is necessary, it is highly likely that patients referred to a chest physician with a clinical suspicion of cancer and a compatible radiograph will, in due course, proceed with an initial spiral CT scan before routine bronchoscopic examination. As the paper by Laroche et al has shown, a number of these patients will be fully diagnosed by the imaging investigation and the success rate of bronchoscopy may be improved in the others.