Adenoid cystic carcinoma (ACC) is a rare malignancy with 3 components (tubular, cribriform, and solid), and these components often coexist in the same patient. In the past few decades, great strides have been made in the pathological classification of ACC. However, due to the complexity of the pathological morphology, establishing a unified grading/assessment standard for clinical implementation process remains challenging. This study aimed to develop improved prognostic biomarkers to assist clinical pathological assessment at the molecular level. We used laser capture microdissection to isolate tumor microsamples with specific pathological features (solid, cribriform, and tubular component) and paracancerous normal glands from 100 patients, then performed DNA sequencing (n = 955) and RNA sequencing (n = 723) to obtain the genome and transcriptome profile, including copy number variation (CNV) type, gene fusion, and transcriptional expression. We found that CNV and gene fusion patterns differed among the 3 pathological subtypes and correlated with clinical outcome. We also found a mutually exclusive relationship between MYBL1::NFIB fusion and 6q-loss, and they displayed distinct different transcriptional profiles. We discovered that molecular markers 6q-loss and 14q-loss were strongly associated with poor prognosis in ACC. Patients with 6q-loss or 14q-loss had a higher risk of tumor metastasis and recurrence, whereas MYB::NFIB and MYBL1::NFIB fusions showed no adverse prognostic impact. Based on these findings, we propose a new CNV-based molecular classification that stratifies patients into the subgroups of 6q-loss, 14q-loss, and others. The molecular classification method increased the prognostic accuracy from 74.49% (the pathological method) to now 82.65%, especially in cribriform patients (increased from 64% to 80%). Fluorescence in situ hybridization-based detection of 6q-loss and 14q-loss by probe ESR1-6q25 and FOS-14q24 provided a rapid and clinically feasible validation in ACC prognostic assessment. Further multicohort and multifactor analysis demonstrated the robustness and independence of the prognostic value in overall survival and metastasis-free survival of the molecular classification method.
Liu et al. (Sun,) studied this question.