On-site Cytopathologic Analysis of Bronchoscopic Needle Aspiration

Abstract

Bronchoscopic needle aspiration (BNA) through rigid bronchoscopes was first reported in the literature in 1949. The procedure was not widely used, though, until the advent of flexible bronchoscopy. Since 1978, flexible bronchoscopy with BNA has been a standard method for diagnosing hilar and mediastinal adenopathy. The technique is well documented to be safe and effective,1 as well as cost-effective.2,3 The method for this procedure is well described.4 A specially designed needle is passed though the bronchoscope. Under direct visualization the needle is introduced through the bronchial wall into the tissues beneath the bronchial wall. The needle is agitated to separate some of this tissue, and biopsy material is aspirated into a syringe under continuous suction. The collected specimens are put onto microscopic slides and are examined later. BNA can be considered for sampling peripheral nodules or endobronchial lesions visualized on bronchoscopy. More commonly, though, it is used to sample tissue from mediastinal or hilar adenopathy. When lesions cannot be visualized directly, the bronchologist relies on other localizing techniques. Usually aspiration efforts are directed by knowledge of thoracic anatomy and prior computed tomographic imaging. Because there is the possibility of error in picking a site and performing the procedure, multiple needle passes are necessary for each target. This increases the likelihood that at least one sample is diagnostic. The overall yield for the procedure is unclear. Reports range from 0 to 100%.5 BNA can stage 80% or more of patients with a known diagnosis of bronchogenic carcinoma. BNA is a remarkably safe procedure. There are fewer than a dozen serious complications in more than 20 years of published reports. Several authors have suggested procedural variations to improve the yield of BNA. One idea is to have immediate, rather than delayed, cytologic examination. Rapid on-site cytopathologic examination, ROSE, offers theoretical advantages. Bronchologists can plan each needle aspiration based on the results of prior passes. One can continue additional attempts until one is successful, and stop further trials after the diagnosis is known. This spares the patient from the risk of unnecessary additional trials. It gives immediate feedback to the bronchologist. There is some element of chance in every pass of the BNA needle. The chance of success is based on the chance of success with each aspirate and the number of aspirates attempted. The relationship is not a simple multiplication. If each pass has a yield of 20%, the cumulative yield of five passes is 67%, not 100%. This is based on the relationship of independent events and the Law of “At Least One.”6 The chance of at least one of the passes being diagnostic is the opposite of the chance that none of them are diagnostic. If the chance of one success is 20%, then the chance of one failure is 80%. The chance of five successive failures is 80% × 80% × 80% × 80% × 80%, or 33%. The chance of anything else, at least one success in five attempts, is 100% – 33% or 67%. If the yield of each individual pass of the BNA needle were known, it would be possible to calculate the overall yield for the procedure based on the number of passes. The yield is not known. In fact, it is variable and probably depends on the experience and expertise of the bronchologist7–9 as well as the patient population. Table 1 shows calculated yields for different possible numbers of passes and different yields per pass. Review of Table 1 demonstrates two important factors. Even if the yield for each pass is poor, a sufficient number of passes can provide a good overall yield. Also, there are diminishing returns. Each successive trial gives less added benefit. Eventually, additional trials become meaningless. There are no prospective trials designed to confirm the mathematical modeling of Table 1, but a trial by Chin et al.10 does support the model.TABLE 1: Calculated overall yield of BNA procedure with different numbers of needle passes and different yields per trial (calculated to two significant digits)Chin et al.10 studied the number of aspirates needed for BNA in the diagnosing and staging of lung cancer. That trial demonstrated an increased yield with increasing numbers of trials. It also demonstrated the diminishing returns predicted by the mathematical model. The yield after the first pass was 53%. After the second, it was 77%. The yield reached a plateau of 97% after the seventh pass. Fig. 1 shows the yield in that trial and the yield of successive trials based on mathematical modeling with a yield per pass of 53%. The curves are quite similar, suggesting that modeling reflects reality. With a reasonable yield per trial and sufficient needle passes, the procedure has an excellent yield regardless of the presence of ROSE.FIGURE 1.: Yield of bronchoscopic needle aspiration procedures based on a mathematical model of cumulative yield for the procedure with a yield of 53% for each pass and varying numbers of trials. Additionally the yield of successive trials from Chin et al.10 is shown.During the past 22 years, two published studies have concluded that on-site cytopathology can improve the diagnostic yield of BNA.11,12 The first report was published in 1990. Davenport11 improved his diagnostic yield from 31 to 56% when using ROSE of transbronchial aspirates. He also improved his rate of satisfactory specimens from 44 to 82%. He realized that the inadequacy, or conversely the adequacy of specimens, is the best measure of the test, because the diagnostic yield is influenced by the prevalence of disease in the study population. The high yield in that study could be the result of the high prevalence of malignant disease (73%). Currently BNA is used primarily to sample mediastinal and hilar lymphadenopathy, but in that study, needle aspirates were used primarily to diagnose peripheral parenchymal lesions (57%). Most of the patients in Davenport's11 original study had parenchymal disease without adenopathy. ROSE may be valuable for aspiration of peripheral pulmonary nodules, but is not helpful for BNA of hilar or mediastinal adenopathy. Ten years later, Diette et al.12 reported a prospective cohort of 204 cases. They concluded that diagnostic yield was greater when on-site cytopathology was used. The study included patients with adenopathy, parenchymal lesions with adenopathy, or parenchymal lesions alone. Results from BNA were not separated from results of other bronchoscopic techniques. In fact, needle aspiration was only attempted in 54.7% of cases. The authors defined success to include any new diagnoses made by flexible bronchoscopy, even cytopathologic assessment was not involved. For example, some diagnoses were made by direct visualization or cultures. It is not plausible that on-site cytopathology could be responsible for such diagnoses, but the authors credited cytopathology in their results. It is not reasonable to credit successful direct visualization or cultures to the presence of on-site cytopathology, but the authors did not separate diagnoses made by BNA from those made by other means. The trial by Chin et al.10 also examined the effect of ROSE and had some intriguing results. There was an improved yield when ROSE was present compared with when ROSE was absent: 71% compared with 25%. If ROSE was effective then one would expect fewer passes to be necessary to get a diagnosis. Actually, 37% more passes were performed when ROSE was present. Additionally, when BNA was positive for malignancy, 32% fewer trials were necessary until the first positive aspirate. This suggests a different patient population with more accessible disease in the ROSE group. ROSE examines aspirates after they are completed. The technique does nothing to get adequate specimens, only confirms specimen adequacy after the fact. It could be helpful if one were willing to perform an unlimited number of aspiration attempts. In reality, most bronchologists will probably abort the procedure after some number of attempts regardless of the cytologist's opinion. Interestingly, Diette et al.12 had a good diagnostic yield (50%) even when ROSE judged the specimens inadequate. Mathematical modeling suggests diminishing returns with each successive biopsy attempt. Chin et al.10 confirmed this. There is no theoretical reason to believe that ROSE could improve diagnostic yields more than persistence. Trials by Diette et al.12 and Davenport11 did not measure the effect of ROSE on BNA procedures for mediastinal adenopathy. Davenport11 evaluated the effect of ROSE on aspirations of parenchymal lesions as well as adenopathy. Diette et al.12 studied all diagnoses made by many bronchoscopic techniques, even those not involving cytology. Until the appropriate prospective, randomized trial is done, routine use of ROSE cannot be assumed to increase BNA yield. ROSE certainly reduces the number of unnecessary needle passes. However, BNA is a very low-risk procedure, and the benefit of a few less passes is unclear. Immediate feedback may be helpful in training environments, but there is no evidence that training with ROSE is better than training without it. Unlike BNA, ROSE is not cost-effective.13 By any standard, BNA is an excellent technique. Adding an extra step with ROSE makes BNA less attractive, not more. Bronchologists' efforts should advocate proved bronchoscopic procedures rather than unproved ones.