Diffuse Staining for Activated NOTCH1 Correlates With NOTCH1 Mutation Status and Is Associated With Worse Outcome in Adenoid Cystic Carcinoma

Abstract

NOTCH1 is frequently mutated in adenoid cystic carcinoma (ACC). To test the idea that immunohistochemical (IHC) staining can identify ACCs with NOTCH1 mutations, we performed IHC for activated NOTCH1 (NICD1) in 197 cases diagnosed as ACC from 173 patients. NICD1 staining was positive in 194 cases (98%) in 2 major patterns: subset positivity, which correlated with tubular/cribriform histology; and diffuse positivity, which correlated with a solid histology. To determine the relationship between NICD1 staining and NOTCH1 mutational status, targeted exome sequencing data were obtained on 14 diffusely NICD1-positive ACC specimens from 11 patients and 15 subset NICD1-positive ACC specimens from 15 patients. This revealed NOTCH1 gain-of-function mutations in 11 of 14 diffusely NICD1-positive ACC specimens, whereas all subset-positive tumors had wild-type NOTCH1 alleles. Notably, tumors with diffuse NICD1 positivity were associated with significantly worse outcomes (P=0.003). To determine whether NOTCH1 activation is unique among tumors included in the differential diagnosis with ACC, we performed NICD1 IHC on a cohort of diverse salivary gland and head and neck tumors. High fractions of each of these tumor types were positive for NICD1 in a subset of cells, particularly in basaloid squamous cell carcinomas; however, sequencing of basaloid squamous cell carcinomas failed to identify NOTCH1 mutations. These findings indicate that diffuse NICD1 positivity in ACC correlates with solid growth pattern, the presence of NOTCH1 gain-of-function mutations, and unfavorable outcome, and suggest that staining for NICD1 can be helpful in distinguishing ACC with solid growth patterns from other salivary gland and head and neck tumors.

Figure 1

Histologic correlates of NICD1 staining in selected cases diagnosed as ACC. A) NICD1 staining in a tubular/cribriform ACC (hematoxylin counterstain). The field contains several nests of tubular/cribriform ACC showing typical subset NICD1 staining (T) adjacent to normal parotid gland (N), which is NICD1 negative. B) Diffuse NICD1 positivity in a solid ACC. Low-power view demonstrates infiltration of normal salivary gland by an ACC with a solid growth pattern and strong, diffuse NICD1 positivity; note nuclear staining in the high-power inset image. C) Absence of NICD1 staining in a tumor diagnosed as ACC involving the middle turbinate showing a ribbon-like or trabecular to solid growth pattern. Other areas in this tumor showed a tubular/cribriform growth pattern reminiscent of ACC. Note NICD1 staining in endothelial cells, which provide an internal control for immunoreactivity. D) H&E-stained section of a NICD1-negative tumor diagnosed as ACC showing squamoid differentiation and microcystic areas.

Figure 2

NOTCH1 mutations in ACC. A) Positions of point substitutions and in-frame indels in the NOTCH1 extracellular negative regulatory region (NRR). NOTCH1 structure is modeled based on X-ray crystallographic studies of Gordon et al. (34). B) Sequence of juxtamembrane insertional mutations caused by small NOTCH1 duplications in biopsies of ACC involving the trachea and liver in the same ACC patient. Residues in red correspond to duplicated amino acid residues; residues in green correspond to unique intervening amino acid residues; and residues in black correspond to normal, unduplicated amino acid residues. Highlighted AV residues in the sequence from the tracheal biopsy corresponds to the normal site of ADAM cleavage in NOTCH1, while highlighted residues in the sequence from the liver biopsy corresponds to the NOTCH1 transmembrane domain (TM). C) Nanostring analysis of RNA isolated from FFPE ACC tissue. RNA was isolated from two specimens containing ACCs with wild type NOTCH1 alleles; four ACC specimens with NOTCH1 alleles with point substitutions or small indels (NOTCH1 mutated); three ACC specimens with loss of copy number spanning NOTCH1 exons 1-27; and the triple negative breast cancer cell line MB-157, which has homozygous or hemizygous NOTCH1 rearrangements and lacks NOTCH1 exons 1–27 (13). RNA abundance expressed as Nanostring signal strength (Y-axis) for each exon covered by probes spanning the NOTCH1 locus (X-axis) was first normalized using average signal strength internal control housekeeping genes and then was expressed relative to signal strength for a NOTCH1-mutated ACC with no evidence of allelic imbalance, for which signal strength for each exon was arbitrarily set at 1. Note that excess signals for NOTCH1 exons 28-34, which encode the NOTCH1 transmembrane domain and intracellular domain, is seen in the control MB-157 triple negative breast cancer cell line (black) and in the 3 samples (red) prepared from specimens with copy number loss involving exons 1-27.

Figure 3

Kaplan-Meier survival curves for patients with NICD1 diffusely positive (n=12) or NICD1 subset positive (n=15) ACC.

Figure 4

Immunohistochemical staining for NICD1 in a representative case of basaloid squamous cell carcinoma