The overall accuracy (benign and malignant) of 94%, and diagnostic sensitivity, accuracy, and negative predictive value for malignant lesion of 94%, 96%, and 80%, respectively, are within the range reported in the literature.
5-8 In a study of 226 CT-guided lung CNBs, Quint et al
5 reported a diagnostic sensitivity, accuracy, and negative predictive value in the diagnosis of malignancy of 91%, 92%, and 68%, respectively. Loh et al
7 reported a diagnostic sensitivity, accuracy, and negative predictive value of CT-guided thoracic biopsies for malignant lesion of 96%, 97%, and 88%, respectively, in 384 patients who underwent 399 CT-guided thoracic biopsies in Singapore. In the same study,
7 the authors concluded that their high rate of success was partly due to onsite evaluation of FNA specimens, as well as the use of combined FNA and CNBs in some of their patients. Quint et al
5 did not evaluate their specimens onsite, and we believe our higher diagnostic sensitivity, accuracy, and negative predictive value for malignant lesions are at least partly attributable to having a pathologist onsite to evaluate the specimens in real time. Previous studies have reported that immediate onsite evaluation by a trained cytology professional increases the diagnostic ability of image-guided needle biopsies.
7-10
The results of the present study are in keeping with those of previous studies,
5-8,11 indicating that CT-guided biopsies have a relatively high false-negative rate for the diagnosis of malignant lesion. In the present study, 9 of 13 (69%) unsuccessful cases that had subsequent biopsy sampling represented missed cancers. Gelbman et al
11 reported 170 patients who underwent a CT-guided FNA lung biopsy yielding an initial benign result, with adequate follow-up, and found that 18 proved to be false-negatives. Tsukada et al
6 reported that out of 91 true malignant tumors, 21 were diagnosed as benign (false-negative) on initial CT-guided needle biopsy. Thus, a benign nonspecific result cannot be relied on for the exclusion of malignant lesions.
3,5,11 Therefore, it is especially important to have close clinical and imaging follow-up in patients with noncorrelating clinical or imaging and biopsy results with a low threshold for a repeated biopsy.
A limited number of prior studies discuss the ability to subclassify NSCLCs into adenocarcinoma, squamous cell carcinoma, or a mixed carcinoma with an adenocarcinoma component using small biopsies, partly because of the recent clinical need for this distinction. Similar to other reports,
12-14 the current study found that ancillary immunostains were helpful in cases where the cytologic and histomorphologic findings were ambiguous on routine hematoxylin-eosin staining because of poor tumor differentiation or scanty tissue. Several useful immuno-histochemical algorithms using various combinations of TTF-1, napsin A, p63, p40, and CK5/6 have been published, which allow for accurate subclassification of most ambiguous NSCLC cases while conserving tissue for potential molecular marker testing.
3,15,16 Use of special stains as a component of NSCLC subtyping is also recommended by the International Association for the Study of Lung Cancer, the American Thoracic Society, and the European Respiratory Society.
17
Sigel et al
12 compared small tissue biopsy and cytologic specimens in the subtyping of 101 NSCLC cases. These authors reported tissue biopsy categories of definitive vs favored vs unclassified at 71%, 23%, and 6%, respectively.
12 However, when both cytology and biopsy paired specimens were combined, the rate of definitive diagnoses by at least 1 method was increased to 84%, and the unclassified rate decreased to 4%.
12 The authors concluded that optimal results are attained when the 2 modalities are considered jointly.
12 Similar to Sigel et al,
12 we successfully subclassified 86% of the NSCLC cases in the present study and favored a subtype in an additional 8% of cases. In addition, 6% of our NSCLC cases were unclassifiable (NSCLC, not otherwise specified). We believe the use of immunostains along with the combined use of cytology on the TP slides and histomorphology substantially increased our ability to subtype our NSCLC CNB cases, even though we did not specifically assess these parameters in this study.