Study objectives: To evaluate the diagnostic accuracy of bronchial bio
psy (BE) specimens in establishing the specific cell type in primary l
ung cancer (LC), and to study the influence of several factors an this
accuracy. Setting: Tertiary health-care center. Patients: One hundred
forty-six patients with LC diagnosed by BE specimens who underwent th
oracotomy (T). Measurements: We have studied the specific LC cell type
observed in the BE specimen and compared it with the T specimen (refe
rence diagnosis). Age, location and type of bronchial lesion, number a
nd size of the biopsy fragments, tumoral size, sample necrosis, degree
of cell differentiation , tumoral stage, pathologist's experience, an
d the presence of other diagnostic tests with the same cell type were
analyzed to assess their influence on the concordance between the two
diagnoses. Results: The overall concordance between BE and T histologi
c diagnosis was 0.70 (kappa coefficient [K]). Of the different histolo
gic types, the worst result was obtained in large cell carcinoma (LLC)
(K, 0.49). Squamous carcinoma and adenocarcinoma gave similar results
(0.74 and 0.77, respectively), while small cell lung cancer (SCLC) on
ly reached a value of 0.60. The degree of cell differentiation, the ab
sence of necrosis, and presence of other preoperative diagnoses were t
he variables that most influenced the histologic accuracy of BB specim
ens. Therefore, the probability of BB accuracy was 2.7, 7.7, and 25 ti
mes higher in cell-differentiated, than in poorly differentiated, mode
rately differentiated, or undifferentiated carcinomas; 5.2 times highe
r when there was no necrosis in the sample; and 7.43 higher when there
was another preoperative diagnosis. Conclusions: The histologic resul
ts of BE must be examined carefully, especially in cellular subtypes l
ike LLC. The absence of differentiation and presence of necrosis in BE
samples were the factors that require the greatest caution in ascerta
ining the cell type. When they are involved and also in all cases in w
hich identifying the specific cell type has important implications, we
prefer to classify the patients as having SCLC or non-small cell lung
cancer, and then reclassify them later after using a second diagnosti
c technique.