AJCP / Original Article
INSM1
A Novel Immunohistochemical and Molecular Marker for
Neuroendocrine and Neuroepithelial Neoplasms
Jason N. Rosenbaum, MD, Zhenying Guo, MD, Rebecca M. Baus, Helen Werner,
William M. Rehrauer, PhD, and Ricardo V. Lloyd, MD, PhD
From the Department of Pathology and Laboratory Medicine, University of Wisconsin Hospital and Clinics, Madison.
Key Words: INSM1; Neuroendocrine; Neuroepithelial; IHC; Cancer; NEN
Am J Clin Pathol  October 2015;144:579-591
DOI: 10.1309/AJCPGZWXXBSNL4VD
ABSTRACT
Objectives: Neuroendocrine neoplasms (NENs) are
heterogeneous neoplasms, which are sometimes malignant,
although predicting metastasis is difficult. INSM1 is a
transcription factor expressed transiently in embryonic
neuroendocrine (NE) tissue, thought to coordinate
termination of cell division with differentiation of NE and
neuroepithelial cells. In adult tissues, INSM1 has been
identified in multiple tumors of NE or neuroepithelial origin
but has not been thoroughly investigated as a potential
neoplastic marker.
Methods: We evaluated INSM1 as a semiquantitative
immunohistochemical (IHC) marker for NE and
neuroepithelial neoplasms and as a quantitative reverse
transcriptase polymerase chain reaction (qRT-PCR) marker
for gastrointestinal NENs (GI-NENs).
Results: Using IHC, we found in normal adult tissue that
INSM1 expression was highly restricted to nuclei of NE
cells and tissues. INSM1 was not detected in any adult
nonneoplastic, non-NE tissue. In neoplastic tissue, INSM1
was detectable by IHC in 88.3% of 129 NEN specimens.
In contrast, INSM1 was detected by IHC in only one of 27
neoplasms without a neuroepithelial or NE component.
Using qRT-PCR, we evaluated INSM1 gene expression in
113 GI-NEN specimens.
Conclusions: INSM1 expression was significantly increased
in neoplastic vs nonneoplastic tissue. Furthermore, among
midgut GI-NENs, neoplasms with known metastases showed
significantly higher expression than those that had not yet
metastasized.
© American Society for Clinical Pathology
Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
Upon completion of this activity you will be able to:
• describe INSM1 and its role in normal cellular development.
• apply INSM1 immunohistochemistry as a diagnostic aid.
• correlate the level of INSM1 expression with other clinicopathologic
properties.
The ASCP is accredited by the Accreditation Council for Continuing
Medical Education to provide continuing medical education for physicians.
The ASCP designates this journal-based CME activity for a maximum of
1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only
the credit commensurate with the extent of their participation in the activ-
ity. This activity qualifies as an American Board of Pathology Maintenance
of Certification Part II Self-Assessment Module.
The authors of this article and the planning committee members and staff
have no relevant financial relationships with commercial interests to disclose.
Exam is located at www.ascp.org/ajcpcme.
Neuroendocrine neoplasms (NENs) are a heterogeneous
group of neoplasms, derived from neuroendocrine (NE) cells
in diverse tissues. Collectively, NENs have an estimated
incidence of 2.5 to 5 per 100,000 people per year,1 with
gastrointestinal NENs (GI-NENs, formerly carcinoids) com-
prising roughly two-thirds of the total. It is often difficult to
predict which tumors will go on to metastasize.
NENs share histologic and biochemical features with
each other but also with neuroepithelial tumors exhibiting
neural (rather than glial) features (eg, medulloblastoma).
Neuroepithelial tissues and their neoplastic derivatives (eg,
medulloblastoma, retinoblastoma, and olfactory neuroblas-
toma) share many developmental and biochemical proper-
ties with their NE counterparts, often expressing common
markers such as chromogranin and synaptophysin, and
regulated by similar transcription factors, such as ASCL1,
NGN3, and NEUROD1.
INSM1 is a transcription factor involved in the terminal
steps of these developmental pathways. Originally isolated
Am J Clin Pathol  2015;144:579-591 579
DOI: 10.1309/AJCPGZWXXBSNL4VD
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms

via a subtraction library of human insulinoma and gluca- ❚Table 1❚

gonoma samples, from which it took its name (insulinoma INSM1 Is a Sensitive Immunohistochemical Marker for NE
and Neuroepithelial Neoplasmsa
associated protein 1), INSM1 has since been detected in
Proportion IHC Positive,
several tumor types in humans, mice, and rats and is thought Neoplasm No./Total No.
to be largely absent from healthy adult tissues.2-12
Neuroepithelial and NE neoplasms
The human INSM1 protein binds both to DNA and Carcinoid (lung) 6/6
to other proteins, most notably cyclin D1.13-15 By binding Esthesioneuroblastoma 1/1
cyclin D1, INSM1 directly arrests progression of the cell GI-NEN (GI carcinoid) 40b/42
Large cell NE carcinoma 2/2
cycle. The binding of cyclin D1 also directly mediates the
Medullary thyroid carcinoma 2/3
transcriptional effects of INSM1. Thus, INSM1 function- Medulloblastoma 2/2
ally links transcriptional activity to cell cycle arrest. Studies Merkel cell carcinoma 6/6
in knockout mice and in human cells altering or ablating EMPSGC 1/1
NE carcinoma of breast 1/1
expression of INSM1 have reaffirmed a critical role in the
Neuroblastoma 3b/4
appropriate development of neural and NE tissues.16 Effects Pan-NEN 19b/21
on both terminal cellular differentiation and cellular prolif- Paraganglioma 9/9
eration have been observed in the pancreas,4,6,16-18 enteroen- Parathyroid adenoma 0/4
Parathyroid carcinoma 0/2
docrine cells,6 autonomic nervous system,8 central nervous Pheochromocytoma 7/7
system,9,10 olfactory epithelium,19 and pituitary.11 Pituitary adenoma 4/6
Laboratory studies have associated INSM1 with a number Pituitary carcinoma 3/3
of NE and neuroepithelial tumors: insulinomas (for which the PNET 1c/2
Retinoblastoma 2/2
gene is named), pituitary adenomas, medullary thyroid carci- Small cell carcinoma (lung) 3/3
nomas,2 pheochromocytomas,20 hypothalamic hamartomas,17 Teratoma, immature 2c/2
small cell lung cancers (SCLCs),2,21-23 and medulloblasto- Total 114/129
Non-NE neoplasms
mas.24-26 Lan et al21 reported that INSM1 was expressed in
Adrenal cortical neoplasms 0/3
97% of SCLC cell lines, with a high degree of concordance Breast adenocarcinoma 1/4
with the NE markers chromogranin A and l-3,4-dihydroxy- Colonic adenocarcinoma 0/2
phenylalanine decarboxylase. In non-SCLC, INSM1 was Lung adenocarcinoma 0/2
Melanoma 0/4
found only in 13% of lines and only in those that also express
Pancreatic carcinoma 0/3
other NE markers. INSM1 was highly associated with medul- Prostate adenocarcinoma 0/2
loblastoma,26 with highest expression noted in the desmoplas- Thyroid carcinoma 0/4
tic variant. INSM1 was also detected in a screen for genes Total 1/24
Neoplasms with NE differentiation
with increased expression in malignant pheochromocytomas.20 recognized on H&E
A recent reanalysis of microarray expression data identified Colonic adenocarcinoma 1/1
upregulation of INSM1 in small intestinal GI-NENs.27 Despite Endometrioid carcinoma 1c/2
Prostate adenocarcinoma 2/2
these prior studies, INSM1 has not been evaluated as a broad-
Total 4/5
spectrum marker for clinical use in human disease.
EMPSGC, endocrine mucin-producing sweat gland carcinoma; GI, gastrointestinal;
We examined a large series of NE tissues and NENs for GI-NEN, gastrointestinal neuroendocrine neoplasm; IHC, immunohistochemistry;
INSM1 expression and found that it was highly expressed NE, neuroendocrine; Pan-NEN, pancreatic neuroendocrine neoplasm; PNET,
primitive neuroectodermal tumor.
in NE cells and NENs. Expression of INSM1 was not pres- a Neuroectodermal neoplasms are generally diffusely positive for INSM1, with

ent in most adult non-NE tissue. Furthermore, quantitative a handful of specimens showing focal positivity. INSM1 is detected in both
aggressive neoplasms and their less aggressive counterparts (eg, pituitary
reverse transcriptase polymerase chain reaction (qRT-PCR) carcinoma and adenoma). INSM1 has not been detected in parathyroid neoplasms
(n = 6) or in hyperplastic parathyroids (n = 2; data not shown). Control neoplasms
analysis showed that INSM1 messenger RNA (mRNA) from a variety of tissues do not express INSM1 at levels detectable by IHC,
expression correlated with metastatic disease in GI-NENs. with the exception of ductal lesions of the breast. Five non-NE neoplasms with
an NE component recognizable on H&E are also included (one colorectal, two
endometrioid endometrial, and two prostatic adenocarcinomas). Four of these
five specimens demonstrate strong (3+) staining in the NE component, consistent
with INSM1 expression being predominantly restricted to NE tissue. Likewise,
most nonneoplastic tissues do not express INSM1 (Table 2). INSM1 detects
Materials and Methods neuroepithelial neoplasms with a sensitivity of 88.3% (92.7% if parathyroid
neoplasms are excluded). INSM1 is very specific to NE and neuroepithelial tissue.
b Two specimens in these categories showed focal staining.
c Focal staining (2%-19% of nuclei positive).
Immunohistochemistry and Western Blot
The study comprised 129 cases of NE and neuroepithe-
lial neoplasms, 27 control neoplasms, and five neoplasms

580 Am J Clin Pathol  2015;144:579-591 © American Society for Clinical Pathology

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 AJCP / Original Article

with known NE differentiation. Cases were retrieved from ❚Table 2❚

the archives of the University of Wisconsin Hospital and Normal Adult Tissues Lacking Expression of INSM1 by
Immunohistochemistrya
Clinics (UWHC) Department of Pathology and evaluated
Tissue/Cell Type No. of Slides Reviewed
by immunohistochemistry (IHC) ❚Table 1❚ and ❚Table 2❚.
Cases were retrieved and histories reviewed with appropriate Adnexa of skin 7
Adrenal cortex 7
institutional review board approval (2011-0712, approved Bone 6
February 13, 2012) in accord with the ethical standards Breast ductal epithelium 6
established by the UWHC. Cases span the years 1995 to Brunner’s glands 4
2014. INSM1 was detected by IHC on 5-mm sections using a Cardiac muscle 1
Cartilage 5
LabVision Autostainer (Thermo Fisher Scientific, Waltham, Cerebellum 1
MA). Monoclonal anti-INSM1 antibody (SC 271408; Santa Cerebral cortex 2
Cruz Biotechnology, Santa Cruz, CA) was applied at a Dermis 6
dilution of 1:5,000, followed by Mach3 Mouse Probe Endocardium 1
Endometrial glands 2
(MP530L; BioCare Medical, Concord, CA) and Mach3 Endometrial stroma 2
Mouse horseradish peroxidase (HRP) (MH530L; BioCare Epithelium, unspecified 5
Medical). The IHC protocol was validated on fetal autopsy Exocrine pancreas 14
Glomeruli 2
tissue ❚Image 1A❚, ❚Image 1B❚, and ❚Image 1C❚. Western blot
Hair follicles 7
was performed according to standard laboratory protocol, as Liver 2
previously reported,28 with the anti-INSM1 antibody applied Lymphoid tissue 66
at a concentration of 1:10,000, incubated at 4°C overnight Olfactory epithelium 3
Optic nerve 2
and an HRP-conjugated goat anti–mouse secondary anti-
Ovarian stroma 3
body applied at a concentration of 1:2,000. The specimen Pancreatic ductal epithelium 13
was considered negative if fewer than 1% of cells showed Parathyroid 3
positive nuclear staining, focally positive if 2% to 19% of Pleura 5
Pneumocytes 10
cells showed positive nuclear staining, and frankly positive Prostate parenchyma 4
if 20% of cells or greater showed positive nuclear staining. Renal tubular epithelium 2
GI-NEN specimens were also semiquantitatively evaluated Retina 2
for intensity of staining on a scale of 0 to 3+. Sclera 2
Seminiferous tubules 1
Serosa 2
qRT-PCR Skeletal muscle 10
The qRT-PCR assay for INSM1 expression was devel- Squamous epithelium 9
oped on control tissue from the UWHC autopsy archives. Sustentacular cells 8
Thyroid 5
Infant cerebellum was used as a positive control, and kid- Urothelium 2
ney from the same patient was used as a negative control. a Expression of INSM1 was evaluated on nonneoplastic tissues, exposed to the
Samples were prepared using established protocols for RNA immunohistochemistry stain in the course of evaluating neoplasms on the same
slides. INSM1 could not be detected in any of the listed tissues or cell types.
extraction from formalin-fixed, paraffin-embedded (FFPE) Blood vessels, smooth muscle, fibroconnective tissue, nerves, and adipose were
tissue. Following deparaffinization using CitriSolv (Thermo consistently negative for INSM1 expression.
Fisher Scientific) and ethanol, total RNA was extracted
from a total of three 12-mm sections of FFPE tissue using Technologies, Coralville, IA), yielding a 139–base pair
the RNEasy FFPE Kit (73504; Qiagen, Valencia, CA). product. Peptidylprolyl isomerase A (PPIA) and glyceralde-
Samples were treated with DNAse using RQ1 RNase Free hyde 3-phosphate dehydrogenase (GAPDH) mRNA levels
DNase (M6101; Promega, Madison, WI). Aliquots (5 mL) were used to normalize the relative quantity of the INSM1
from total RNA were reverse transcribed and real-time PCR transcript in different FFPE tissue samples (PPIA and
amplified concurrently using the SuperScript III Platinum GAPDH primers from Qiagen). Each specimen was tested
SYBR Green One-Step qRT-PCR Kit (11736-051; Life for DNA contamination by performing a direct amplification
Technologies, Carlsbad, CA) following the manufacturer's on the purified mRNA without reverse transcription. Sam-
instructions. qRT-PCR reactions were 20 mL in volume and ples that amplified were presumed to be contaminated with
run on a LightCycler 2.0 instrument (03531414001; Roche, DNA and were treated with DNAse and reevaluated. Once
Indianapolis, IN) using an INSM1 gene-specific primer set samples were determined to be free of DNA contamination,
(forward, 5′-AGTGCCACCTGTGCCCAGTGT-3′; reverse, they were amplified by qRT-PCR according to standard
5′-AGGCCGGGCGAGCTGTAGAA-3′; Integrated DNA protocols and methods.

© American Society for Clinical Pathology Am J Clin Pathol  2015;144:579-591 581

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593 DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms
A B
C
Cases evaluated by qRT-PCR included GI-NENs (car-
cinoids) spanning the years 2006 to 2014 from the Univer-
sity of Wisconsin surgical pathology archives. Appropriate
FFPE blocks were selected for analysis based on primary
location, known metastases, and quality of specimen. A total
of 113 tissue samples were collected from 79 patients (38
male, 41 female; mean age, 55.6 years [range, 21-90 years]).
RNA was purified and evaluated as described above.
INSM1 fold-changes (DDCPs) were calculated in dupli-
cate vs both GAPDH and PPIA. Pearson’s correlation coef-
ficient (R) was calculated for each data set to assess repro-
ducibility of results. GAPDH was slightly more reproducible
(R = 0.89 vs R = 0.88) and more reliably amplified (data not
shown). Subsequent analyses calculated INSM1 DDCP with
respect to GAPDH. All calculations and statistical analyses
were performed using Microsoft Excel (Microsoft, Red-
mond, WA).
582 Am J Clin Pathol  2015;144:579-591
Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
❚Image 1❚ Anti-INSM1 antibody detects INSM1 expression.
INSM1 exhibits robust nuclear expression in developing (A
and B, 17 weeks gestational age; C, 21 weeks) human neural
and neuroendocrine tissue. High levels of expression are
appreciated in enteroendocrine cells of the gastrointestinal
epithelium (A, top, ×200), cells of the enteric nervous
system (A, bottom), fetal pancreas (B, ×400), and developing
cerebellum (C, ×1,000).
Results
To demonstrate the utility of IHC for detecting INSM1
in FFPE human tissue, we first confirmed that INSM1 is
detected in FFPE tissue types in which INSM1 expression is
recognized or suspected, which occur predominantly during
embryonic and fetal development. Using autopsy tissue, we
confirmed INSM1 expression in developing neuroepithelial
and NE tissues. In fetal tissues, INSM1 was detected in GI
epithelium and enteric nervous system (Image 1A), pancreas
(Image 1B), cerebellum (Image 1C), thyroid, respiratory
epithelium, and thymus. The cellular expression pattern was
nuclear, and background staining was minimal. In parallel,
we used postnatal cerebellum to demonstrate quantitative
detection of INSM1 by qRT-PCR in FFPE tissue. INSM1
amplified reliably from cerebellum but not from kidney of
the same decedent.
© American Society for Clinical Pathology
 AJCP / Original Article

D E

F
❚Image 1❚ (cont) Lower but still detectable levels of
expression are seen in individual C cells of the thyroid,
Kulchitsky cells of the respiratory epithelium, and
neuroendocrine cells within the thymus (data not shown).
Western blot confirms specificity of the anti-INSM1
antibody with clear, strong bands of the expected sizes in
a gastrointestinal neuroendocrine neoplasm (GI-NEN) from
the ileum (approximately 58 kD) and rat pancreatic cell line
(Pc12) but not in normal human tissue (D). The same GI-NEN
was used to confirm the immunohistochemistry (IHC) in
adult neoplastic tissue (E, H&E, ×100; F, INSM1 IHC, ×100).
Neoplastic cells show strong nuclear expression with little to
no background or cytoplasmic staining. Intensity of staining
can vary from cell to cell, often between adjacent cells (see
inset in F). Insets are ×400.

We next confirmed the specificity of the anti-INSM1
antibody by performing Western blot on fresh tissue from
a GI-NEN of the ileum ❚Image 1D❚. Anti-INSM1 antibody
detected robust expression of a single protein of the expected
size (58 kD) in the GI-NEN, as well as a slightly larger
protein in a rat pheochromocytoma cell line (PC12), a find-
ing consistent with the slightly longer INSM1 sequence in
mouse (PubMed GeneID 53626) and rat (unpublished data,
J.N.R. and J. García-Añoveros, 2008). Western blot did not
detect protein in human, nonneoplastic, nonneuroectodermal
tissue. The same GI-NEN specimen was used to confirm
the detection of INSM1 in FFPE tissue ❚Image 1E❚ and
❚Image 1F❚.
To determine the sensitivity and specificity of IHC
for INSM1, we examined a spectrum of neuroectodermal
neoplasms (Table 1) ❚Image 2❚. INSM1 was detected in
88.3% of NE and neuroepithelial neoplasms examined.
Notably, INSM1 was not detected in normal, hyperplastic,
or neoplastic parathyroid tissue (Table 1). INSM1 was gen-
erally not detected in nonneuroectodermal neoplasms. Four
of five non-NE neoplasms with a component of NE differ-
entiation (one colonic, one endometrial, and two prostatic
adenocarcinomas; Table 1) did exhibit strong (3+) INSM1
expression within the majority of the NE component.
Among NE neoplasms lacking any histologic sign of neural
or NE differentiation, INSM1 was detected only in breast
carcinoma (Table 1).
We surveyed normal tissue from all slides examined by
IHC for INSM1 expression, and in the nonneoplastic, non-
NE tissues, we detected no expression of INSM1 (Table 2).
Notable tissues in which INSM1 was not detected included
lymphoid tissue (n = 66 specimens), nerve and central ner-
vous system tissue (n = 30), exocrine pancreas (n = 14),
pneumocytes (n = 10), sustentacular cells of adrenal medulla
and paraganglia (n = 8), thyroid (n = 8), and adrenal cortex
(n = 7). In a handful of nonneoplastic adult cell types, we did

© American Society for Clinical Pathology Am J Clin Pathol  2015;144:579-591 583

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593 DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms

A B

C D

❚Image 2❚ INSM1 is expressed in neuroendocrine (NE) and neuroepithelial neoplasms. Most neoplasms of NE origin show at
least focal immunohistochemistry (IHC) positivity for INSM1. INSM1 is strongly expressed in a representative pancreatic NE
neoplasm but not in adjacent nerve, vessels, or exocrine pancreas (A, ×40). In pheochromocytoma (B, ×100) and paraganglioma
(C, ×400), INSM1 is detected in neoplastic NE nuclei but not in sustentacular cells or nonneoplastic adrenal cortex (B, left side
of panel). Medullary thyroid carcinoma (D, ×400) expresses INSM1 in neoplastic cells.

detect INSM1 expression. These are all NE tissues, including
adrenal medulla, pancreatic islets, and GI enterochromaffin
cells ❚Image 3❚ and ❚Image 4❚. We also identified INSM1-
expressing cells in occasional small clusters within normal
bronchial epithelium and in occasional individual cells in
nonneoplastic prostate glands (data not shown). INSM1 was
not detected in normal ductal or glandular epithelia from
other sites (breast, urothelium, pancreas, endometrium, skin
adnexa, renal epithelium; n = 37).
We compared IHC for INSM1 directly with the tra-
ditional, cytoplasmic NE markers chromogranin and syn-
aptophysin (Image 4). In a case with roughly comparable
staining from all three markers, the distinct, nuclear pattern
of INSM1 clearly distinguishes it from the other two mark-
ers (Image 4). In addition, INSM1 appears more specific
to NE cells than synaptophysin (compare the epithelial and
stromal staining in Image 4D with that in Image 4A). In our
data set (n = 30) of GI-NENs stained for all three markers,
INSM1 is quite comparable in sensitivity and specificity to
both chromogranin and synaptophysin ❚Table 3❚.
Expression of INSM1 was detected by qRT-PCR in
tissue from GI-NENs, with the average GI-NEN showing a
14.69-fold increase in expression of INSM1 relative to non-
neoplastic tissue (P < .005). With a threshold level taken as
1-fold expression (ie, no increase in expression compared
to the negative control), qRT-PCR detects GI-NENs with a

584 Am J Clin Pathol  2015;144:579-591 © American Society for Clinical Pathology

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 AJCP / Original Article

E F

G H

I ❚Image 2❚ (cont) Expression in adult nonneoplastic thyroid,

including C cells, has not been observed. In NE neoplasms
of the lung (representative small cell carcinoma depicted in
E [×40]), INSM1 is always detected in at least some cells.
Nonneoplastic lung tissue does not appear to express INSM1
(E, right side), but isolated expression in nonneoplastic NE cells
cannot be definitively ruled out. Staining of nuclei is robust and
nearly uniform in most Merkel cell carcinomas (representative
in F [×40]). Notably, elements of unaffected skin, including
glandular elements, do not express INSM1. Most neoplasms
with immature neural or neuroepithelial components show at
least focal IHC positivity for INSM1. Esthesioneuroblastoma
(olfactory neuroblastoma), a rare neoplasm of the olfactory
epithelium, exhibits robust expression of INSM1 (G,
×200). Medulloblastomas are already known to express
INSM1, a feature that may correlate with the desmoplastic
subtype. In the present study, IHC detects INSM1 in classic
subtype medulloblastomas (H, ×40; see also Table 1). The
neuroectodermal component of an immature teratoma is
clearly highlighted by IHC for INSM1 (I, ×100). Insets are ×400.

© American Society for Clinical Pathology Am J Clin Pathol  2015;144:579-591 585

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593 DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms
A B
❚Image 3❚ Immunohistochemistry for INSM1 is largely specific for neuroendocrine and neuroepithelial neoplasms. INSM1 is not
detectable in most nonneoplastic adult tissues, with a few notable exceptions: pancreatic islets (A, ×200), adrenal medulla (B,
×200), and enterochromaffin cells of the gastrointestinal epithelium (see Image 4B and Image 5).
sensitivity of 80.6% and a specificity of 92.3% ❚Figure 1A❚.
Although higher expression levels of INSM1 appear to cor-
relate with metastatic potential of GI-NENs ❚Figure 1B❚, this
trend was not statistically significant for GI-NENs, when
no other clinical data are considered (P = .17 by analysis of
variance [ANOVA]).
Several specimens had INSM1 expression levels
orders of magnitude higher than the mean. These speci-
mens were all found in derivatives of the embryonic
foregut (n = 4 specimens, all from the stomach or esopha-
gus). This prompted evaluation of INSM1 expression with
respect to the embryonic site of origin. Expression levels
of INSM1 in the foregut were consistently and signifi-
cantly higher than those in the midgut or hindgut (43-fold
increase in expression vs 4- and 6-fold increases in expres-
sion, respectively; P < .01) ❚Figure 1C❚. When the embry-
onic sites are then subdivided based on malignant behav-
ior, tissue from the embryonic foregut expresses INSM1 at
dramatically higher levels in every category, including in
nonneoplastic tissue; the average fold increase in expres-
sion of INSM1 in nonneoplastic tissue of the foregut,
midgut, and hindgut is 1.29-fold, 0.26-fold, and 0.30-fold,
respectively. Moreover, the number of specimens in the
foregut and hindgut (n = 24 and 23, respectively) was
not sufficient for these categories to achieve statistical
significance on their own (P = .22 and .15 by ANOVA,
respectively). Only the pattern of INSM1 expression in
the midgut achieves statistical significance (n = 60, P <
.05 by ANOVA) ❚Figure 1D❚. The low numbers of foregut
specimens along with the significantly higher expression
586 Am J Clin Pathol  2015;144:579-591
Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
of INSM1 suggest that there may be underlying biologic
differences between sites within the GI tract, and these
differences may skew our analysis.
Since the numbers of specimens from the foregut and
hindgut were too low to achieve significance, we focused
specifically on GI-NENs arising from the midgut to deter-
mine whether level of expression of INSM1 correlates with
malignant behavior. Among GI-NENs of the midgut, the
average fold-change of INSM1 in primaries with known
metastases was nearly twice that of primaries without known
metastases (Figure 1D; P < .05 by t test). To determine
whether this difference achieved clinical significance, we
directly compared the expression level of INSM1 for the
samples. Among primary lesions of the midgut, the mean
increase in INSM1 expression is 4.8-fold ❚Figure 1E❚. Using
this as a threshold, expression of INSM1 above 4.8-fold pre-
dicts nodal or distant metastasis with 81.3% specificity and
47.6% sensitivity.
Semiquantitative evaluation of IHC for INSM1 cor-
related with average fold-change of INSM1 mRNA expres-
sion by qRT-PCR ❚Image 5❚ and ❚Figure 2❚. When evalu-
ated on a scale of 0 to 3+, this correlation is not statistically
significant (correlation coefficient of 0.46; P = .26 by
ANOVA). However, when evaluated as weak staining (0 or
1+; mean INSM1 fold-increase of 1.03) vs strong staining
(2+ or 3+; mean INSM1 fold-increase of 7.23), the differ-
ence in the mean fold-increase in expression of INSM1 is
highly significant (P < .005 by t test). Among the weakly
staining specimens, no specimen had an increase greater
than 3.25-fold.
© American Society for Clinical Pathology
 AJCP / Original Article

A B

C D

❚Image 4❚ Immunohistochemistry (IHC) for INSM1 compares well with traditional neuroendocrine (NE) markers. Comparative
staining of a rectal gastrointestinal NE neoplasm (A, H&E, ×100) reveals robust expression of INSM1 (B, ×200), along with
both chromogranin A (C, ×200) and synaptophysin (D, ×200), the more traditional markers of neoplastic NE tissue. In this case,
INSM1 appears coexpressed with chromogranin and synaptophysin in nearly all neoplastic cells. In contrast to other NE IHC
markers, INSM1 expression is specifically nuclear.

Discussion which agrees with previous reports.3,6,8,11,19,29 Regarding

We found that INSM1 was consistently expressed in
NENs and in related neuroepithelial neoplasms (such as
medulloblastoma). This expression was highly restricted—
in the adult, we detected INSM1 in only a handful of normal
NE tissues and not at all in tissues outside the NE system.
In neoplasms, INSM1 expression was similarly restricted. In
addition, qRT-PCR analysis correlated quantitative INSM1
expression with metastatic disease in GI-NENs.
We readily detected INSM1 by IHC in the nuclei of
fetal neural and NE tissues, including pancreas, thyroid, GI
and tracheal epithelium, enteric neurons, and cerebellum,
expression of INSM1 in adults, most of the previous litera-
ture has been restricted to animal models or cell culture.2-4,6,8
By IHC, we confirmed expression of INSM1 in a handful
of nonneoplastic adult cell types in humans: NE cells of the
gastrointestinal tract, pancreatic islets, adrenal medulla, and
rarely small nests in bronchial epithelium and individual
cells of prostatic glandular epithelium. We did not detect
INSM1 expression in any nonneoplastic, non-NE adult
tissue.
Applying anti-INSM1 IHC to neoplasms, we detected
INSM1 in some proportion of cells in nearly all NENs
and all neuroepithelial neoplasms examined. The notable

© American Society for Clinical Pathology Am J Clin Pathol  2015;144:579-591 587

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593 DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms
❚Table 3❚
Comparison of Staining for INSM1, Chromogranin, and
Synaptophysin in GI-NENsa
No. of Cases (of 30) INSM1 Chromogranin Synaptophysin
18 + + + ing, and more accurately quantifiable. Quantification may
1 + + –
8 + – +
2 – + +
0 + – –
0 – + –
1 – – +
0 – – –
27/30 21/30 29/30
a Gastrointestinal neuroendocrine neoplasm (GI-NEN) cases show variable staining
for INSM1, chromogranin, and synaptophysin, and the proportion of cases lacking
staining for each antibody does not entirely overlap. Of 30 cases in which INSM1,
chromogranin, and synaptophysin staining is performed, all three markers are
positive in only 18 (60.0%) of cases. No case lacked staining for all three markers.
INSM1 is positive in a total of 27 (90.0%) cases, chromogranin is positive in
21 (70.0%) cases, and synaptophysin is positive in 29 (96.7%) cases. The three
specimens that did not stain for INSM1 are from the esophagus, stomach, and
rectum. Five of the nine chromogranin-negative specimens are rectal, with the
remaining cases arising in the stomach, duodenum, ascending colon, and sigmoid
colon, respectively. The synaptophysin-negative specimen is appendiceal. The case
negative for both INSM1 and chromogranin is gastric. Cases positive for a given
marker are denoted with +; cases showing less than 1% of neoplastic cells with
expression of the marker are denoted with –.
exceptions are neoplasms of the parathyroid (both adenomas
and carcinomas), which do not show detectable expression
of INSM1. Furthermore, neither hyperplastic nor normal
adult parathyroids express INSM1. It is not clear why
parathyroids and their derivative neoplasms did not express
INSM1—they are known to express other NE markers, such
as synaptophysin, chromogranin, and CD56.30 INSM1 has
not been specifically investigated in parathyroid tissue; per-
haps the specialized endodermal derivation of this NE tissue
makes use of an alternative transcriptional mechanism. With
parathyroid neoplasms excluded, INSM1 was detectable by
IHC in 92.7% of NE and neuroepithelial neoplasms gener-
ally, with possible variation according to specific category
of neoplasm.
Importantly, INSM1 was not detectable in most non-
NE, nonneuroepithelial neoplasms. Indeed, outside of NE
and neuroepithelial neoplasms, INSM1 was detected in only
two scenarios: neoplasms with NE differentiation recog-
nized by other means (four of five such specimens) and in
breast carcinoma from one patient. For practical purposes,
IHC for INSM1 distinguishes neoplastic NE cells from
nearby tissue that may have a similar appearance on H&E,
such as lymphoid tissue. In addition, although we do detect
INSM1 expression in a handful of nonneoplastic adult NE
tissues, these are infrequent and usually readily distinguish-
able from neoplasms of these tissues.
We found that INSM1 compares well with traditional
markers of NE tissue. In contrast with most other markers
of NE tissue, INSM1 is a transcription factor with nuclear
expression. This feature offers advantages over cytoplasmic
588 Am J Clin Pathol  2015;144:579-591
Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
markers such as chromogranin and synaptophysin. Even in
cases exhibiting comparable expression of all three markers,
the clear nuclear pattern of INSM1 is generally easier to
interpret, more readily distinguished from artifactual stain-
aid in the evaluation and staging of immature teratomas,
for example, allowing an accurate determination of the
neuroepithelial component of the tumor. No significant asso-
ciation with prognosis in NE or neuroepithelial tumors has
been demonstrated for other NE markers. INSM1 is known
to coordinate terminal differentiation of these tissues with
termination of cell division. Because of its relationship with
the cell cycle, quantification of INSM1 may offer prognostic
as well as diagnostic utility.
Direct quantification of the INSM1 gene product offers
a more direct assay with which to correlate the clinicopatho-
logic properties of individual neoplasms. To quantify the
expression level of INSM1, we developed a qRT-PCR assay
for expression of INSM1 mRNA. GI-NENs are the most
common of the NENs but lack good markers correlating
with propensity for distant or nodal metastases. To deter-
mine if the quantitative expression of INSM1 correlates with
malignancy, we applied our assay to GI-NENs. We found
that expression levels were not consistent between different
portions of the GI tract. In both neoplasms and control tis-
sue, INSM1 is expressed at much higher levels in derivatives
of the embryonic foregut. It is likely that some combination
of greater numbers, higher turnover, or more NE activity
in the enterochromaffin cells of the foregut contributes to
the higher levels of expression. Importantly, despite com-
paratively low numbers of specimens from the foregut, the
difference in expression is enough to skew a more general
analysis of GI-NENs.
Restricting our analysis to GI-NENs that arise in
derivatives of the midgut, we found that overexpression of
INSM1 correlated with the propensity of the primary tumor
to metastasize. Furthermore, our data suggest that increased
expression of INSM1 by a level of 4.8-fold or greater is
81.3% specific to neoplasms with a known metastasis.
From a prognostic standpoint, this was likely a conservative
estimate; one of the primary specimens with expression of
INSM1 elevated above this threshold shows histologic evi-
dence of aggressive behavior (involvement of the muscularis
propria) on H&E, despite not having a proven metastasis. In
addition, the patients in this study have not been followed
prospectively, and it is likely that some of the patients
may develop metastases over time. It is not clear whether
increased expression of INSM1 is a cause or a consequence
of pathology. No oncogenic mutations within the INSM1
gene have been identified to date. Nevertheless, elevated
expression of INSM1 in several tumor types, now including
GI-NENs, suggests potential prognostic utility for INSM1.
© American Society for Clinical Pathology
 AJCP / Original Article

A B C
50 50 80

INSM1 Fold-Change
40 40
INSM1 Fold-Change

INSM1 Fold-Change
60

30 30
40
20 20
20
10 10

0 0 0

ut

ut
gu
ss tic

ss tic

si t

t a it h

s
t a ou

dg
m

se
ue

ue

re

id
Ti las

Ti la s

as W
as

si
as th

in
ta

M
Fo
et s

H
et i
op

pl

op

as
M sW

M a rie
eo

et
ne

ne

ie
N

M
im
on

on

ar

Pr
N

im
Pr

D 8 E 18

16

6
14
INSM1 Fold-Change

INSM1 Fold-Change
12
4
10

8
2
6

4
0
ss tic

si t

ta ith

s
ta ou

se
ue

s
Ti las

as W

2
si
as th

ta
et s
et i
op

as
M sW

M rie

et
ne

a
ie

M
im
on

0
ar

Pr
N

im

Primaries Without Primaries Without

Pr

Metastases Metastases
Midgut

Midgut

❚Figure 1❚ Overexpression of INSM1 messenger RNA in gastrointestinal neuroendocrine neoplasms (GI-NENs) correlates with
metastasis. INSM1 is overexpressed in GI-NENs but not in nonneoplastic tissue (A). The mean DDCP for GI-NENs (fold-change
of 14.69) is 34 times greater than the mean DDCP for nonneoplastic tissue (fold-change of 0.43; P < .02 by t test). Using a DDCP
threshold of 1.00 to define overexpression, use of quantitative reverse transcriptase polymerase chain reaction for INSM1 to
detect GI-NENs has a sensitivity of 80.6% and specificity of 92.3%. When specimens are categorized by degree of malignancy,
a trend emerges directly correlating overexpression of INSM1 with metastatic potential, although this pattern does not achieve
statistical significance (B; P = .17 by analysis of variance [ANOVA]). Expression of INSM1 in GI-NENs varies significantly
between those that arise from derivatives of the embryonic foregut, midgut, and hindgut (C; P < .01 by ANOVA). Average
fold-change of INSM1 in the foregut is highest, at nearly 43-fold, despite consisting of only 24 specimens. In contrast, GI-NENs
derived from the midgut showed a mean increase in INSM1 expression of 4.12-fold (65 specimens). When specimens are
subdivided based on metastatic behavior in addition to embryonic derivation, only the midgut specimens (D) achieve statistical
significance (P = .22 in the foregut; P < .05 in the midgut; P = .15 in the hindgut by ANOVA). The average fold-increase of
INSM1 in midgut GI-NENs with known metastasis is significantly higher than that of GI-NENs without known metastasis (6.11
vs 3.19, respectively; P < .05 by t test). Scatterplot of the actual fold-change for individual specimens (E) reveals that midgut
GI-NENs without metastasis rarely (three of 16 specimens) express INSM1 at greater than a 4.8-fold increase in expression
(dashed line). In contrast, half of primaries with metastasis (10 of 21 specimens) demonstrate increased expression at greater
than 4.8-fold. Increased expression of INSM1 by greater than 4.8-fold has a sensitivity for predicting metastasis of 47.6% and a
specificity of 81.3%. Error bars reflect SEM.

© American Society for Clinical Pathology Am J Clin Pathol  2015;144:579-591 589

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593 DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
 Rosenbaum et al / INSM1 in Neuroendocrine Neoplasms

A B C D

❚Image 5❚ Quantified INSM1 messenger RNA expression correlates with immunohistochemistry for INSM1 protein.
Gastrointestinal neuroendocrine neoplasms often (95%, n = 42) stain with anti-INSM1 antibody, but the intensity can vary (A-D,
×200; see also Table 1). We grade semiquantitative expression on a scale of 0 (A) to 3+ (D). In the colon (B), nonneoplastic
neuroendocrine cells strongly expressing INSM1 can be seen at the top of the image.

In summary, INSM1 is a useful immunohistochemical

INSM1 Fold-Change

15 and molecular marker for neoplasms of neuroectodermal

12
9 epithelial neoplasms and may provide a useful adjunct to
6 existing markers. In addition, in the most common NENs,
3
0
0 1+ 2+ 3+
❚Figure 2❚ Quantitative expression is evaluated by
quantitative reverse transcriptase polymerase chain reaction
and correlated with histology. Semiquantitative interpretation
of anti-INSM1 immunohistochemistry correlates with
average fold-change of INSM1 messenger RNA among those
specimens. Although the overall pattern does not reach
statistical significance (P = 0.26 by analysis of variance; n =
42; correlation coefficient = 0.46), the difference between
INSM1 expression between weakly staining (0/1+) and
intermediate to strongly staining (2+/3+) specimens is
significant (P < .005 by t test). Error bars reflect SEM.
Semiquantitative assessment of INSM1 expression by
IHC comports with quantitative analysis of INSM1 expres-
sion by qRT-PCR, with higher fold-increase being associat-
ed with stronger IHC staining. Our data suggest (although do
not confirm) there may be prognostic value in either IHC or
qRT-PCR for INSM1, or perhaps a combination of the two.
Further investigation with larger numbers of specimens and
follow-up of patient outcomes will be necessary. Additional
studies in NENs arising from other tissues may provide more
clarity and additional clinical utility.
origin. It aids in the primary diagnosis of NE and neuro-
quantitative evaluation of INSM1 mRNA expression cor-
related with malignant behavior and may prove prognosti-
cally useful. Further investigation of INSM1 in NE and
neuroepithelial neoplasms may yield additional useful
information.
Corresponding author: Jason N. Rosenbaum, MD, Barnes
Jewish Hospital, Washington University, 425 S Euclid Ave,
Campus Box 8118, St Louis, MO 63110; JRosenbaum@path.
wustl.edu.
The authors thank the University of Wisconsin Hospital and
Clinics Department of Pathology and Laboratory Medicine for
support, including a resident research and development grant.
The authors also thank the University of Wisconsin Translational
Research Initiatives in Pathology laboratory, in part supported
by the University of Wisconsin Department of Pathology and
Laboratory Medicine and UWCCC grant P30 CA014520 for use
of its facilities and services.
References
1. Öberg K, Castellano D. Current knowledge on diagnosis and
staging of neuroendocrine tumors. Cancer Metastasis Rev.
2011;30:3-7.
2. Goto Y, De Silva MG, Toscani A, et al. A novel human
insulinoma-associated cDNA, IA-1, encodes a protein
with “zinc-finger” DNA-binding motifs. J Biol Chem.
1992;267:15252-15257.

590 Am J Clin Pathol  2015;144:579-591 © American Society for Clinical Pathology

Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
DOI: 10.1309/AJCPGZWXXBSNL4VD
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017

3. Xie J, Cai T, Zhang H, et al. The zinc-finger transcription
factor INSM1 is expressed during embryo development
and interacts with the Cbl-associated protein. Genomics.
2002;80:54-61.
4. Zhu M, Breslin MB, Lan MS. Expression of a novel
zinc-finger cDNA, IA-1, is associated with rat AR42J
cells differentiation into insulin-positive cells. Pancreas.
2002;24:139-145.
5. Breslin MB, Zhu M, Lan MS. NeuroD1/E47 regulates
the E-box element of a novel zinc finger transcription
factor, IA-1, in developing nervous system. J Biol Chem.
2003;278:38991-38997.
6. Gierl MS, Karoulias N, Wende H, et al. The zinc-finger
factor Insm1 (IA-1) is essential for the development of
pancreatic beta cells and intestinal endocrine cells. Genes
Dev. 2006;20:2465-2478.
7. Duggan A, Madathany T, de Castro SC, et al. Transient
expression of the conserved zinc finger gene INSM1 in
progenitors and nascent neurons throughout embryonic and
adult neurogenesis. J Comp Neurol. 2008;507:1497-1520.
8. Wildner H, Gierl MS, Strehle M, et al. Insm1 (IA-1) is
a crucial component of the transcriptional network that
controls differentiation of the sympatho-adrenal lineage.
Development. 2008;135:473-481.
9. Farkas LM, Haffner C, Giger T, et al. Insulinoma-associated
1 has a panneurogenic role and promotes the generation
and expansion of basal progenitors in the developing mouse
neocortex. Neuron. 2008;60:40-55.
10. Jacob J, Storm R, Castro DS, et al. Insm1 (IA-1) is an
essential component of the regulatory network that specifies
monoaminergic neuronal phenotypes in the vertebrate
hindbrain. Development. 2009;136:2477-2485.
11. Welcker JE, Hernandez-Miranda LR, Paul FE, et al. Insm1
controls development of pituitary endocrine cells and requires
a SNAG domain for function and for recruitment of histone-
modifying factors. Development. 2013;140:4947-4958.
12. Kawaguchi A, Ikawa T, Kasukawa T, et al. Single-cell
gene profiling defines differential progenitor subclasses in
mammalian neurogenesis. Development. 2008;135:3113-3124.
13. Liu WD, Wang HW, Muguira M, et al. INSM1 functions as
a transcriptional repressor of the neuroD/beta2 gene through
the recruitment of cyclin D1 and histone deacetylases.
Biochem J. 2006;397:169-177.
14. Zhang T, Liu WD, Saunee NA, et al. Zinc finger
transcription factor INSM1 interrupts cyclin D1 and
CDK4 binding and induces cell cycle arrest. J Biol Chem.
2009;284:5574-5581.
15. Wang HW, Muguira M, Liu WD, et al. Identification of
an INSM1-binding site in the insulin promoter: negative
regulation of the insulin gene transcription. J Endocrinol.
2008;198:29-39.
© American Society for Clinical Pathology
Downloaded from https://academic.oup.com/ajcp/article-abstract/144/4/579/1766593
by LIBRARY SERVICES - CHRISTIAN MEDICAL COLLEGE - VELLORE user
on 01 December 2017
AJCP / Original Article
16. Osipovich AB, Long Q, Manduchi E, et al. Insm1 promotes
endocrine cell differentiation by modulating the expression
of a network of genes that includes Neurog3 and Ripply3.
Development. 2014;141:2939-2949.
17. Parent AS, Matagne V, Westphal M, et al. Gene expression
profiling of hypothalamic hamartomas: a search for genes
associated with central precocious puberty. Horm Res.
2008;69:114-123.
18. Zhang T, Saunee NA, Breslin MB, et al. Functional role of an
islet transcription factor, INSM1/IA-1, on pancreatic acinar
cell trans-differentiation. J Cell Physiol. 2012;227:2470-2479.
19. Rosenbaum JN, Duggan A, Garcia-Anoveros J. Insm1
promotes the transition of olfactory progenitors from
apical and proliferative to basal, terminally dividing and
neuronogenic. Neural Dev. 2011;6:6.
20. Sandgren J, Andersson R, Rada-Iglesias A, et al.
Integrative epigenomic and genomic analysis of malignant
pheochromocytoma. Exp Mol Med. 2010;42:484-502.
21. Lan MS, Russell EK, Lu J, et al. IA-1, a new marker for
neuroendocrine differentiation in human lung cancer cell
lines. Cancer Res. 1993;53:4169-4171.
22. Taniwaki M, Daigo Y, Ishikawa N, et al. Gene expression
profiles of small-cell lung cancers: molecular signatures of
lung cancer. Int J Oncol. 2006;29:567-575.
23. Amelung JT, Buhrens R, Beshay M, et al. Key genes in lung
cancer translational research: a meta-analysis. Pathobiology.
2010;77:53-63.
24. Gilbertson RJ. Medulloblastoma: signalling a change in
treatment. Lancet Oncol. 2004;5:209-218.
25. De Smaele E, Fragomeli C, Ferretti E, et al. An
integrated approach identifies Nhlh1 and Insm1 as Sonic
Hedgehog–regulated genes in developing cerebellum and
medulloblastoma. Neoplasia. 2008;10:89-98.
26. Pomeroy SL, Tamayo P, Gaasenbeek M, et al. Prediction of
central nervous system embryonal tumour outcome based on
gene expression. Nature. 2002;415:436-442.
27. Kidd M, Modlin IM, Drozdov I. Gene network-based
analysis identifies two potential subtypes of small intestinal
neuroendocrine tumors. BMC Genomics. 2014;15:595.
28. Hardin H, Guo Z, Shan W, et al. The roles of the epithelial-
mesenchymal transition marker PRRX1 and microRNA
146b-5p in papillary thyroid carcinoma progression. Am J
Pathol. 2014;184:2342-2354.
29. Mellitzer G, Bonne S, Luco RF, et al. IA1 is NGN3-
dependent and essential for differentiation of the endocrine
pancreas. EMBO J. 2006;25:1344-1352.
30. Chu PG, Weiss LM. Modern Immunohistochemistry. 2nd ed.
New York, NY: Cambridge University Press; 2014.
Am J Clin Pathol  2015;144:579-591 591
DOI: 10.1309/AJCPGZWXXBSNL4VD