Percutaneous Transthoracic Needle Aspiration Biopsy* A Comprehensive Review of Its Current Role in the Diagnosis and Treatment of Lung Tumors Ryan C. Larscheid, BS; Patricia E. Thorpe, MD; and Walter]. Scott, MD, FCCP
Objective: The purpose of this study is to examine the accuracy and complications of transthoracic needle aspiration biopsy (TINA) to determine its optimal role in the evaluation of patients with lung tumors. Materials and methods: The charts of 130 consecutive patients who had undergone CT-guided TTNA were reviewed retrospectively. Thirty-two (25%) of these patients had subsequent surgery and 5 had subsequent transbronchial biopsy (TBB). Using the final surgical and TBB diagnosis as a reference, the accuracy, sensitivity, specificity, and prevalence of malignancy were calculated. Each case was also examined to determine the presence or absence of complications. Results: Of the 130 biopsy results, 95 (73%) were malignant, 33 (25%) were nonspecific, and only 2 (2%) had a specific benign diagnosis. Thirty-two patients subsequently underwent surgical resection. The overall prevalence of malignancy after surgical diagnosis was 91%. The overall diagnostic accuracy of TINA was 76%. The sensitivity of TINA for the detection of malignancy was 74% and its specificity was 100%. When comparing TINA results of small ( <3 em) and large (~3 em) tumors, the occurrence of nonspecific results was 36% and 16%, respectively. Fifty-six (43%) patients had a pneumothorax subsequent to TINA. Twenty-four (43%) of these patients required a chest tube and remained hospitalized for a mean of 6 days. Conclusion: Patients who are surgical candidates and have a high clinical suspicion for malignancy should undergo surgical biopsy and resection of their lung tumors if indicated. Information gained from TINAs performed on this patient population will rarely result in a change in their clinical management. (CHEST 1998; 114:704-709) Key words: diagnosis; lung cancer; SPN; TINA Abbreviations: TBB =transbronchial biopsy; TINA = transthoracic needle aspiration
ercutaneous transthoracic needle aspiration (TINA) biopsy is a commonly used diagnostic procedure when a lung tumor is diagnosed. It is generally regarded as a safe procedure with limited morbidity and extremely rare mortality. The most common complication is a pneumothorax that will occasionally require a chest tube. Since TINA is highly sensitive for diagnosing malignancy, 1 a diagnosis of malignancy obtained by TTNA can be regarded as having a high predictive value and the patient should be treated accordingly. In some cases, however, the results from TINA can be nonspecific. *From the Departments of Surgery (Mr. Larscheid and Dr. Scott) and Radiology (Dr. Thorpe), Creighton University Medical Center, Omaha, NE. M
A nonspecific diagnosis, however, does not eliminate the possibility of malignancy. Since many patients are not surgical candidates , TTNA is a necessary procedure to obtain a tissue diagnosis before starting radiation therapy or chemotherapy. In patients who are surgical candidates, however, TTNA can have a questionable role in the ultimate decision of how to best treat these patients. We reviewed the records of 130 patients who underwent TTNA to determine the accuracy of the procedure, review the complications of TTNA, and to better define the role of TTNA in the evaluation of patients with suspected intrathoracic malignancy.
MATERIALS AND METHODS
We reviewed the charts of 130 consecutive patients who had a TINA biopsy between January 1989 and May 1995. The patients ranged in age from 12 to 88 years with a m ean of 67 years . Clinical Investigations
Eighty-one of 130 (62%) patients were male and 49 of 130 (38%) of the patients were female. One hundred four of 130 (80%) patients had a history of smoking. Eighty-three of 130 (64%) patients were not considered surgical candidates. Thirty-seven patients had metastatic disease, 18 had poor pulmonary function, 14 were medically high risk, 8 had small cell cancer, and 6 had unresectable tumors. All biopsies were performed by or under the supervision of one of the expe1ienced faculty radiologists. Prior to the TINA, the patient was briefed on the procedure and its possible risks, including infection, bleeding, and pneumothorax. After consent was obtained, the patient was placed in the CT scanner and his or her lung lesion was localized. The skin overlying the lesion was prepared and draped in a sterile fashion. One percent lidocaine without epineph1ine was used for local anesthetic. For most cases, a 22-gauge, 15-cm needle (Chiba; Cook Inc; Bloomington, IN ) and a 23-gauge, 15-cm needle (Sure-Cut; Bauer Medical; Clearwater, FL) were used for aspiration. Between two and six passes were made into the lesion until adequate cellular samples were obtained. The specimens were transferred to the pathology department for analysis. The patient was monitored after the procedure for complications. Chest radiographs were obtained prior to hospital discharge to determine whether a pneumothorax had occurred. In certain cases of pneumothorax, the surgery service was consulted and a chest tube was placed. A pneumothorax of >25% was the main indication for inse rting a chest tube. Other criteria included increasing dyspnea and a progressive or nonresolving pneumothorax. Thirty-two of 130 (25%) patients underwent surgery and 5 other patients underwent transbronchial biopsy (TBB) subsequent to TTNA. The final surgical and TBB diagnoses were used as the reference in calculating the accuracy, sensitivity, and specificity of TINA. The overall prevalence of malignancy after surgical diagnosis was also dete rmined.
TTNA Results Of the 130 biopsy results, 95 (73%) were malignant, 33 (25%) were nonspecific, and only 2 (2%) had a specific benign diagnosis. Nonspecific results were caused by hypocellular samples, lack of specifically identifiable benign or malignant cells, or by specimens that were indistinguishable from normal lung tissue. The cell types of the 95 malignant biopsy specimens were as follows: adenocarcinoma, 34 (36%); squamous cell, 30 (32%); small cell, 8 (8%); nonsmall cell, 15 (16%); and others, 8 (8%). Each of the malignant cell types was also classified as primary or metastatic as shown in Table 1. From the TTNA results, adenocarcinoma was the most prevalent malignancy overall, but squamous cell carcinoma was the most common primary lung tumor. Of the tvvo specific benign diagnoses, one was a benign pleural plaque and the other was a benign fibroma. Surgical and Follow-up Results Thirty-two of 130 (25%) patients underwent surgical resection of their lung tumor. A definitive
Table 1-Results of Malignant TTNAs Cell Type
Adenocarcinoma Squamous cell Small cell Nonsmall cell Others*
20 28 8 13
14 2 0 2
34 30 8 15 8
*Primary: 1 malignant fibrous histiocytoma; metastatic: 1 osteosarcoma, 1 plasmacytoma, 1 fibrosarcoma, l Ewing's sarcoma, 1 testicular germ cell tumor, 1 renal cell carcinoma, and 1 paraganglioma.
diagnosis for the lesion was obtained in all of the surgical cases. Prior to surgery, 25 of 32 (78%) patients had malignant diagnoses and 7 of 32 (22%) were nonspecific. After surgical diagnosis, 29 (91 %) of the nodules were malignant and only 3 of 32 (9%) were benign. When compared with the TTNA diagnosis, 11 of the 32 (34%) surgical diagnoses yielded different results (Table 2). Four of these were only minor differences of cell type that posed relatively no clinical significance. More importantly, there were no errors of differentiation between small cell and nons mall cell types. However, four of seven (57%) of the nonspecific TTNAs yielded malignant results at surgery while three were found to be benign. In addition to the seven patients 'Nith nonspecific TTNA results who underwent surgery, there were five other patients who had nonspecific results and subsequently had a diagnostic TBB specimen. In all five cases, the TBB yielded malignant results. Three tumors were identified as squamous cell carcinoma and the other two as nonsmall cell carcinoma. Of the remaining 21 patients with nonspecific TTNA results, 9 showed cytology consistent with an unspecified benign lesion that was not pursued further. Three patients died before a diagnosis was reached and the remaining nine were unavailable for follow-up . The final results of these data are com-
Table 2-TTNA vs Surgical Diagnosis TTNA Diagnosis l ) Nonsmall cell
2) 3) 4) 5) 6) 7) 8) 9) 10) 11)
Adenocarcinoma Nonsmall cell Adenocarcinoma Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspecific Nonspeci fie
Surgical Diagnosis Squamous cell Squamous cell Squamous cell Squamous cell Osteosarcoma Carcinoid tumor Squamous cell Lymphoblastic lymphoma Benign noncaseating granuloma Benign caseating granuloma Fibrous chronic inflammation CHEST I 114 I 3 I SEPTEMBER, 1998
pared with the TINA diagnoses in Table 3. The final data showed 104 of 130 (80%) malignant diagnoses, 21 of 130 (16%) nonspecific diagnoses, and 5 of 130 (4%) specific benign diagnoses. Smokers vs Nonsmokers Of the 104 patients classified as smokers, 78 of 104 (75%) had malignant TTNA results and 26 of 104 (25%) had nonspecific results. Twenty-three of these patients underwent surgery. Nineteen patients had a malignant TTNA result prior to surgery, while 4 had nonspecific results. All of the malignant TTNA results were verified as such, while two of the nonspecific results were malignant and two were benign. The two malignant tumors were a squamous cell carcinoma and lymphoblastic lymphoma. The two benign tumors were both granulomas. Additionally, five other smokers had TBB after a nonspecific TINA. All of the TBB results showed malignancy. The final outcome of data for this patient group showed 85 of 104 (82%) malignant tumors, 17 of 104 (16%) nonspecific diagnoses, and 2 of 104 (2%) benign tumors. In the nonsmoking patient group, 17 of 26 (65%) TINA results were malignant, 7 of 26 (27%) were nonspecific, and 2 of 26 (8%) were benign. Nine of these patients undenvent surgery. Six patients had a malignant TTNA result prior to surgery, while three had nonspecific results. The six malignancies were verified as such and of the three nonspecific results, two were malignant and one was benign. The malignant tumors were a carcinoid and a metastatic osteosarcoma. The benign result showed fibrous, chronic inflammation. The final outcome of data for this patient group showed 19 of 26 (73%) malignant tumors, 4 of26 (15%) nonspecific, and 3 of26 (12%) benign. Accuracy of TTNA To determine the accuracy, sensitivity, and specificity of TINA, we used the confirmed surgical and TBB diagnoses and compared them with the TTNA diagnoses. All malignant diagnoses are considered "positive" results and both the benign and nonspe-
Table 3-TTNA Results vs Final Results TINA Results Malignant Nonspecific Benign
Final Results 95 33 2
Malignant Nonspecific Benign
104* 21 5t
*The additional malignant results were obtained from TBB (5) and surgery (4). tThe 3 additional benign results were obtained from surgery.
cific diagnoses are "negative" results. Due to the lack of malignant cells on nonspecific results, they are considered "negative" results even though a benign diagnosis is not confirmed. The problem of establishing a specific benign diagnosis is common with TTNAs. Therefore, the final results can be classified under only two categories; they are either malignant or nonmalignant (benign or nonspecific TTNA results). Based on these criteria, there were 25 true positives, 3 true negatives, 9 false negatives, and no false positives. Thus, the sensitivity ofTTNAs for the detection of malignancy is 74% and the overall accuracy of TTNA for diagnosing malignancy was 76%. When the results of the TTNA were malignant, the overall accuracy for diagnosing the correct malignant cell type was 88%. Since there were no false positives, the specificity for diagnosing malignancy is 100%. Accuracy vs Nodule Size For this study, the tumor sizes are broken down into two groups: small tumor (<3 em) and large tumors (2::3 em). In examining the outcome of TTNAs from different-sized lung masses, there are more nonspecific results for the smaller tumors (Table 4). Twenty-two of 61 (36%) small tumors yielded nonspecific results, whereas only ll of 69 (16%) large tumor biopsy specimens were nonspecific. To determine the sensitivity for large and small tumor TTNA biopsy specimens, the TTNA results are again compared with the final surgical and TBB diagnosis. Seventeen surgeries and 3 TBBs were performed on small tumors and 15 surgeries and 2 TBBs were performed on large tumors. There were 12 true positives, 2 true negatives, and 6 false negatives from the small tumor group. This yields a TINA sensitivity of 67% for tumors <3 em . In the large tumor group, there were 13 true positives, 1 true negative, and 3 false negatives. Thus, the sensitivity of TTNA for tumors 2::3 em is 81%. And, as
Table 4-TTNA Results From Small and Large Tumors Tumor (T) Size
T < 1 em 1cm:ST<2cm 2cm:ST<3cm Small tumor totals 3cm:ST<4cm 4 cm:ST
10 26 37 27 15 7 9 58
1 9 12 22 7 3 0
2 20 39 61 34 18 7 10 69
1 2 0 0 0 0 0
before, the specificity for both small and large tumors is 100% due to the absence of any falsepositive results. Additionally, based on final surgical and TBB diagnosis, the prevalence of malignancy was 90% in small tumors and 94% in large tumors. A correct malignant or nonmalignant diagnosis was made for 14 of 20 (70%) of the 20 small tumors and for 14 of 17 (82%) of the large tumors. Table 5 summarizes all of the data on sensitivity, specificity, accuracy, and prevalence of malignancy.
Complications of TTNA The most commonly encountered complication after TTNA was pneumothorax. Fifty-six of 130 patients (43%) suffered a pneumothorax subsequent to TINA. Thirty-two of 56 (57%) pneumothoraces were classified as small (::::;25%) and did not require a chest tube. Twenty-four (43%) of the pneumothoraces were classified as large (>25%) and required chest tube placement. The total incidence of pneumothoraces requiring a chest tube was 18%. The high rate of chest tube insertion (18%) is likely related to the high number of pneumothoraces. Twenty-four of 56 patients (43%) with a pneumothorax required a chest tube. Although this statistic is somewhat high, it does correlate with the rate described in literature in which up to half the patients with a pneumothorax will need a chest tube for resolution. 2 Additionally, we found that the patients in our study with lesions involving the pleura had a lower incidence of pneumothorax than those with intraparenchymallesions. Of the 130 patients, 10 (8%) had pleural lesions. Of these 10, only 1 (10%) had a pneumothorax. This lower rate most likely results from not having to puncture the lung parenchyma to obtain a cellular aspirate. All 24 patients who required a chest tube were admitted to the hospital. The length of their stay ranged from 1 to 14 days with a mean stay of 6.3±7.2 days. The 32 patients who did not require a chest tube were monitored with serial radiographs and were either admitted to the hospital or sent home
Table 5-Sensitivity, Specificity, and Accuracy of
Sensitivity. % Specificity, % Accuracy,% Prevalence of malignancy, %
depending on their condition. Of these, 16 of 32 (50%) were admitted to the hospital for observation. The length of their stay ranged from 1 to 3 days with a mean stay of 1.4±1.2 days. The only complication encountered other than pneumothorax was 4 of 130 (3%) cases of small hemorrhages with hemoptysis. All of these were minor, self-resolving complications that did not require any clinical intervention.
Small Tumors (n=20)
Large Tumors ( n=l7)
67 100 70 90
81 100 82 94
86 100 88 92
From the results of this study, TTNA has been shown to be a sensitive method for diagnosing malignant intrathoracic tumors. Our study also showed that one of every four patients had nonspecific TTNA results. Other studies also demonstrate a high rate of nonspecific TTNAs ranging from 5.5 to 29%.2-6 Although TTNA has a sensitivity of up to 97% for malignant nodules,7 a specific benign diagnosis is much more difficult to obtain. The rate of obtaining a specific benign diagnosis varies from 12 to 68%,1,3 but is more commonly about 20 to 30%. 6 Fortunately, when a specific benign TINA diagnosis is made, there is relatively little risk that malignancy is present. 8 The difficulty of diagnosing a benign tumor is illustrated in our study by the fact that only two of five benign diagnoses were correctly made on the basis ofTTNA. A recent study also demonstrates this point by comparing TTNA and core biopsy results for malignant and benign lesions. Klein et al9 found that TTNA had a sensitivity of 92% for malignant lesions yet only 44% for benign lesions, while core biopsy had sensitivities of 86% and 100%, respectively. Therefore, in patients with benign lesions, TINA will more likely result in a nonspecific diagnosis. Instead, core biopsy was shown to be much more likely to provide an accurate diagnosis if the lesion is benign. The only drawback with this technique is the 54% incidence of pneumothorax.9 Another variable affecting the accuracy of TTNA results is the size of the nodule. Our study showed that 36% of the small tumors ( <3 em) yielded nonspecific TTNA results, while only 16% of the large tumor (;:::::3 em) results were nonspecific. Other researchers have also looked at how tumor size affects accuracy. Each study, however, classified small and large tumors by different criteria and used different imaging techniques for biopsy. Hayata et a110 defined small nodules as <2 em in diameter and used fluoroscopically guided biopsies. They showed an accuracy of 75.0% for small nodules and 89.6% for large nodules. This shows a significant difference in accuracy rates between the two nodule sizes. Li et CHEST I 114 I 3 I SEPTEMBER, 1998
al 11 also show similar differences. Their study used CT guidance for biopsy and defined small nodules as ::Sl.5 em in diameter. Their results demonstrated accuracies of 74% and 96% with a sensitivity of 72% and 94% for small and large nodules, respectively. They also found a significant difference between false-negative rates. The false-negative rate was 19% for small and 6% for large nodules . A more recent study by Westcott et al 12 examined the results of CT-guided biopsies of nodules ::Sl.5 em in diameter. These results showed diagnostic rates similar to those expected for larger nodules. However, even though this study shows similar rates between large and small nodules, our study and others 10 ·11 show that it is more likely that these smaller nodules will have lower diagnostic yield and provide only nonspecific results. This is most likely due to the difficulty of accurately localizing small lesions. In patients with nonspecific TIN A results, a malignant lesion is frequently discovered after surgical diagnosis. Of the six patients with nonspecific TTNAs who underwent surgmy in this study, four (66%) were found to have malignancy. This is a very common problem encountered in other studies as well. One study showed that 100% of its nonspecific TTNAs were malignant, 6 while, more commonly, the range was from 33 to 73%.2-5 Another finding of this study was the 43% rate of pneumothorax. This rate is somewhat high, but does fall within the reported range of 19 to 44%J.3,13-l7 This high rate in our study is most likely due to the number of passes made through the pleura and into the lesion while attempting to obtain an adequate cellular sample. In almost all cases, only one or two passes were made through the pleura. However, many cases required between two and six passes into the lesion before an adequate sample was obtained. These multiple passes through the parenchyma are the likely cause of the higher-than-average pneumothorax rate. Finally, to interpret the sensitivity and specificity of TTNA determined by any study, the pretest probability of malignancy of tl1e patient population under study should be known. The overall prevalence of malignancy is generally accepted to be approximately 40% for solitary pulmonary nodules encountered in a clinical setting. 18 Our patient population included patients with tumors >3 em in size, metastatic disease, history of cancer, presence of multiple tumors, or other coexisting medical conditions that may increase the likelihood of an alternate diagnosis. Therefore, the overall prevalence of malignancy in our patient population was much higher. The decision of whether to perform a surgical procedure as the initial step in the diagnosis of a lung tumor has been widely debated for many years. 708
Westcott et al 12 argue that TINA has become an increasingly accurate and sensitive technique for diagnosing pulmonary masses and that it should be used first instead of immediate surgical removal. However, authors of studies on video-assisted thoracic surgery techniques 5 · 19 contend that the associated costs and risks of TINA can be avoided and that tumors can be surgically diagnosed and resected in an efficient manner with little increase in morbidity. A recent study showed that the cost per correct diagnosis was lowest when open surgical biopsy was used as the initial procedure. 20 Goldberg-Kahn et al2° found that the average cost to correctly diagnose a lesion was $12,888 when open surgical biopsy was used as the initial procedure and $21,543 when TTNA was the initial procedure. This suggests that surgical biopsy can be a more cost -effective approach for determining the malignant or benign nature of a lung lesion in patients who are surgical candidates. In conclusion, although this issue is complex and circumstances vary from patient to patient, the primary objective should be to determine which diagnostic method is most appropriate for a particular patient. Based on the findings of our study, that TINA is associated with a significant rate of pneumothorax, that it not infrequently yields a nonspecific diagnosis, and yet has a very high positive predictive value for malignancy, we recommend the following general diagnostic approach. In patients with a solitary pulmonary nodule, one could calculate the probability of malignancy as described by Cummings et al. 18 Cummings and coauthors 18 used Bayes' theorem to develop a simple method for combining individual clinical features of patients with solitary pulmonary nodules (specifically the size of the nodule, the patient's age and smoking history, and the clinical setting in which the nodule was discovered) into an overall estimate of the probability that the nodule is malignant. For patients who have a low probability of malignancy, TINA is probably not the most ideal diagnostic method since it has a low sensitivity for benign tumors. Surgery would not be the best choice either since there would be an increased likelihood of unnecessarily operating on a benign lesion. Instead, core biopsy could be utilized in these patients since it has been shown to be very effective at diagnosing benign lesions. This technique may be more likely to produce a pneumothorax, but it will avoid the need for a surgical procedure. Conversely, if a patient has a pulmonary tumor with a high likelihood of being malignant, and that patient is a surgical candidate, then immediate surgery is most likely to be the optimum treatment method. If this hypothetical patient underwent TTNA and this resulted in a nonspecific diagnosis, then the patient would probClinical Investigations
ably be referred for open surgical biopsy. If this same patient underwent TINA and was diagnosed as having a malignancy, then the eventual treatment would still be surgical resection. For patients such as this, TTN A adds risk and cost but does not change the treatment plan for the patient. TTNA is most useful for diagnosing pulmonary nodules that are highly likely to be malignant in patients who are not candidates for other surgical procedures.
REFERENCES 1 Khouri NF, Stitik FP, Erozan YS, et a!. Transthoracic needle aspiration biopsy of benign and malignant lung el sions. AJR Am J Roentgenol 1985; 144:281-288 2 Colquhoun SD, Rosenthal DL, Morton DL. Role of percutaneous fine needle aspiration biopsy in suspected intrathoracic malignancy. Ann Thorac Surg 1991; 51:390-393 3 Perlmutt LM, Johnston WVV, Dunnick NR. Percutaneous transthoracic needle aspiration: a review. AJR Am J Roentgenol 1989; 152:451-455 4 Patz EF, Fidler J, Knelson M, et al. Significance of percutaneous needle biopsy in patients with multiple pulmonary nodules and single known primary malignancy. Chest 1995; 107:601-604 5 Mitruka S, Landreneau RJ, Mack MJ, et al. Diagnosing the indeterminate pulmonary nodule: percutaneous biopsy versus thoracoscopy. Surgery 1995; 118:676-684 6 Bergquist TH, Bailey PB, Cortese DA, et al. Transthoracic needle biopsy: accuracy and complications in relation to location and type of lesion. Mayo Clin Proc 1980; 55:475-481 7 Green R, Szyfelbein WM, Isler RJ, et al. Supplementary tissue-core histology from fine-needle transthoracic aspiration biopsy. AJR Am J Roentgenol1985; 144:787-792 8 Calhoun P, Feldman PS, Armstrong P, et al. The clinical
outcome of needle aspirations in the lung when cancer is not diagnosed. Ann Thorac Surg 1986; 41:592-596 Klein JS, Salomon G, Stewart EA. Transthoracic needle biopsy with a coaxially placed 20-gauge automated cutting needle: results in 122 patients. Radiology 1996; 198:715-720 Hayata Y, Oho K, Ichiba M, et al. Percutaneous puncture for cytologic diagnosis: its diagnostic value for small peripheral pulmonary carcinoma. Acta Cytol 1973; 17:469-475 Li H, Boiselle PM, Shepard JA, eta!. Diagnostic accuracy and safety of CT-guided percutaneous needle aspiration biopsy of the lung: comparison of small and large pulmonary nodules. AJR Am J Roentgenol 1996; 167:105-109 Westcott JL, Rao N, Colley DP. Transthoracic needle biopsy of small pulmonary nodules. Radiology 1997; 202:97-103 Westcott JL. Direct percutaneous needle aspiration of localized pulmonary lesions: results in 422 patients. Radiology 1980; 137:31-35 Shepard JO. Complications of percutaneous needle aspiration biopsy of the chest: prevention and management. Semin Intervent Radio! 1994; 11L:181-185 Westcott JL. Percutaneous transthoracic needle biopsy. Radiology 1988; 169:593-601 VanSonnenberg E, Casola G, Ho M, et al. Difficult thoracic lesions: CT-guided biopsy experience in 150 cases. Radiology 1988; 167:457-461 Stanley JH, Fish GD, Andriole JG, et al. Lung lesions: cytologic diagnosis by fine-needle biopsy. Radiology 1987; 162:389-391 Cummings SR, Lillington GA, Richard RJ. Estimating the probability of malignancy in solitary pulmonary nodules. Am Rev Respir Dis 1986; 135:449-452 Kaiser LR, Shrager JB. Video-assisted thoracic surgery: the current state of the art. AJR 1995; 165:1111-1117 Goldberg-Kahn B, Healy JC, Bishop JW. The cost of diagnosis: a comparison of four different strategies in th e workup of solitary radiographic lur.g lesions. Chest 1997; 111:870-876
CHEST I 114 I 3 I SEPTEMBER, 1998