Time to move beyond clinical and pathologic classification of BAC.

Although the World Health Organization defines BAC as a distinct entity, it is now clear that BAC in fact represents a heterogeneous group of diseases with shared pathologic characteristics but distinct molecular aberrations and varying clinical courses and outcomes. Historically, studies have lumped all adenocarcinomas with lepidic growth patterns into the same category, including both pure BAC, which grows along alveolar septa with no evidence of vascular, lymphatic, or pleural invasion, as well as mixed adenocarcinoma with BAC features.[1] Pure BAC, which represents less than 5% of adenocarcinomas, has been shown to have a very low risk of lymph node or distant metastasis,[2] and in fact represents carcinoma in situ rather than true invasive cancer. In contrast, the much more common mixed adenocarcinomas with BAC features are true invasive cancers; their prognosis is similar to or perhaps slightly better than that of typical non–small-cell lung cancer (NSCLC).[3] “Pure” BAC can be further subdivided into mucinous and nonmucinous subtypes; these subtypes have distinct cells of origin, different immunohistochemical staining patterns, and significantly different clinical characteristics.[4] Molecular Testing Defines Different BAC Phenotypes Despite shared pathologic characteristics, molecular testing has uncovered distinct oncogenic pathways that differentiate subsets of BAC. For example, the majority of mucinous BACs have mutations in the K-ras oncogene,[5,6] which are commonly associated with tobacco smoking. There is evidence that atypical adenomatous hyperplasia, a premalignant condition, shares many of these same molecular abnormalities and may be a precursor lesion to mucinous BAC.[7] Mucinous BAC is also infrequently associated with p53 mutations, which are common in invasive adenocarcinomas of the lung.[8,9] One particularly interesting concept in BAC biology is the viral oncogene hypothesis, which proposes that some BACs may have an underlying viral cause. The hypothesis is based on the striking resemblance between mucinous BAC and sheep pulmonary adenomatosis, a malignancy caused by the Jaagsiekte sheep retrovirus.[10] So far, however, there is no evidence that a retrovirus is the initiating agent in human BAC. In contrast, a relatively high rate of activating EGFR mutations have been observed in nonmucinous BAC, although it is unclear whether the BAC classification is a stronger predictor of EGFR mutation than clinical characteristics such as never-smoking status and female sex. However, EGFR mutations are found in only 30% to 50% of nonmucinous BACs, which is just slightly higher than the incidence of K-ras mutations in these lesions (14% to 23%).[5,6] Because EGFR and K-ras mutations are both believed to be oncogenic drivers, and because the two mutations tend to be mutually exclusive, it is clear that pathologic classification alone is an inadequate basis on which to select patients for targeted therapies. BAC-Specific Clinical Trials and Treatment Recommendations There have been a number of phase II clinical trials testing either chemotherapy or EGFR inhibitors in patients with advanced BAC; these are well summarized in the review by Levy and colleagues. The authors correctly emphasize that these BAC-specific trials have included a heterogeneous mixture of mucinous BAC, nonmucinous BAC, and adenocarcinomas with BAC features;[11] the heterogeneity of