Content uploaded by Leila Eshraghi
Author content
All content in this area was uploaded by Leila Eshraghi on Oct 16, 2019
Content may be subject to copyright.
The Genomic Landscape of Sporadic Prolactinomas
Sunita M. C. De Sousa
1,2,3
&Paul P. S. Wang
4
&Stephen Santoreneos
5
&Angeline Shen
6,7
&Christopher J. Yates
6,7
&
Milena Babic
2
&Leila Eshraghi
2,4,8
&Jinghua Feng
4,8
&Barbara Koszyca
9
&Samuel Roberts-Thomson
10
&
Andreas W. Schreiber
4,8,11
&David J. Torpy
1,3
&Hamish S. Scott
2,3,4,8
#Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
Somatic GNAS and USP8 mutations have been implicated in sporadic somatotrophinomas and corticotrophinomas, respectively.
However, no genes are known to be recurrently mutated in sporadic prolactinomas. The prevalence of copy number variants
(CNV), which is emerging as a mechanism of tumorigenesis in sporadic pituitary adenomas in general, is also unclear in
prolactinomas. To characterize the genetic events underpinning sporadic prolactinomas, we performed whole exome sequencing
of paired tumor and germline DNA from 12 prolactinoma patients. We observed recurrent large-scale CNV, most commonly in
the form of copy number gains. We also identified sequence variants of interest in 15 genes. This included the DRD2,PRL,
TMEM67,andMLH3 genes with plausible links to prolactinoma formation. Of the 15 genes of interest, CNV was seen at the gene
locus in the corresponding tumor in 10 cases, and pituitary expression of eight genes was in the top 10% of tissues. However,
none of our shortlisted somatic variants appeared to be classical driver mutations as no variant was found in more than one tumor.
Future directions of research include mechanistic studies to investigate how CNV may contribute to prolactinoma formation,
larger studies of relevant prolactinoma subsets according to clinical characteristics, and additional genetic investigations for
aberrations not captured by whole exome sequencing.
Keywords Prolactinoma .Pituitaryadenoma .Wholeexome sequencing .Copynumbervariation .Lossof heterozygosity .Driver
mutation
Introduction
The genetic basis of sporadic prolactinomas is currently un-
known. This is in contrast to well-described somatic events in
other pituitary tumors, namely: GNAS mutations in
somatotrophinomas and occasional non-functioning pituitary
adenomas [1–3]; USP8 [3,4] and rarely NR3C1 [3,6]muta-
tions in corticotrophinomas; and CTNNB1 (encoding β-
catenin) and BRAF mutations in the vast majority of
adamantinomatous and papillary craniopharyngiomas, respec-
tively [5]. Some of these somatic events recapitulate multisys-
tem disorders, including McCune-Albright syndrome due to
*Sunita M. C. De Sousa
Sunita.DeSousa@sa.gov.au
1
Endocrine and Metabolic Unit, Royal Adelaide Hospital,
Adelaide, Australia
2
Department of Genetics and Molecular Pathology, Centre for Cancer
Biology, an SA Pathology and University of South Australia
Alliance, Adelaide, Australia
3
School of Medicine, University of Adelaide, Adelaide, Australia
4
ACRF Cancer Genomics Facility, Centre for Cancer Biology, an SA
Pathology and University of South Australia Alliance,
Adelaide, Australia
5
Department of Neurosurgery, Royal Adelaide Hospital,
Adelaide, Australia
6
Department of Diabetes and Endocrinology, Royal Melbourne
Hospital, Melbourne, Australia
7
Department of Medicine, University of Melbourne,
Melbourne, Australia
8
School of Pharmacy and Medical Sciences, University of South
Australia, Adelaide, Australia
9
Department of Anatomical Pathology, Royal Adelaide Hospital,
Adelaide, Australia
10
Department of Anatomical Pathology, Royal Melbourne Hospital,
Melbourne, Australia
11
School of Biological Sciences, University of Adelaide,
Adelaide, Australia
Endocrine Pathology
https://doi.org/10.1007/s12022-019-09587-0
GNAS somatic mosaicism [6], and a newly described
syndromic disorder including pediatric Cushing’s disease
due to a germline heterozygous USP8 mutation [7].
Gene-specific assessment of sporadic prolactinomas has
shown nil or only rare somatic variants in biologically plausi-
ble genes. This includes genes where germline variants cause
familial pituitary tumor syndromes (FPTS), such as MEN1 [8]
and AIP [9], and genes implicated in sporadic pituitary ade-
nomas, including GNAS [1], USP8 [4], and TP53 [10,11]. A
recurrent germline gain-of-function PRLR mutation has been
identified in prolactinoma patients, but no such variants have
been found in the somatic setting [12]. A somatic HRAS var-
iant has been identified in an aggressive prolactinoma, but this
association was not borne out in an extension study including
72 prolactinomas [13].
A limitation of single gene studies is the reliance on
existing knowledge to select candidate genes. “Orphan”genes
of hitherto unknown function could be contributory to
prolactinomas akin to other genetic discoveries, such as the
roles of GPR101 in X-linked acrogigantism [14]andARMC5
in bilateral macronodular adrenal hyperplasia [15]. Whole ex-
ome or genome sequencing offers an unbiased approach to
novel gene discovery. Only three pangenomic studies of
prolactinomas have been performed to date, all employing
whole exome sequencing (WES). Wang et al. [16] focused
on point variants conferring bromocriptine resistance in a co-
hort of 12 prolactinomas and identified 11 candidate genes
between initial and follow-up [17] studies. Bi et al. [18]inves-
tigated 41 pituitary macroadenomas, including three
prolactinomas. Six genes were mutated in more than one tu-
mor, but none of these were prolactinomas. Song et al. [3]
examined 125 pituitary adenomas, including 20
prolactinomas. Two genes were considered to be potential
tumorigenesis genes but only one prolactinoma harbored a
variant in these genes. Overall, these studies did not find re-
current sequence variants amongst the prolactinomas that
could constitute driver mutations. There is, however, emerg-
ing evidence of recurrent copy number variants (CNVs) in
sporadic pituitary adenomas, including the small number of
prolactinomas thus far studied [2,3,18,19].
The aim of the present study was to perform WES in a pure
prolactinoma cohort to identify recurrent somatic genetic
events. We hypothesized that, like other pituitary tumors, so-
matic driver mutations and/or CNVs might also underlie the
development of prolactinomas.
Materials and Methods
Patients
Twelve patients with clinically evident prolactinomas that had
been surgically resected were recruited from two tertiary
referral pituitary centers in Australia. Clinical data were col-
lated using medical records.
DNA Extraction
Patients provided fresh blood samples for germline DNA ex-
traction. Operative tumor specimens were retrieved for somat-
ic DNA extraction. Tumor specimens had either been stored as
fresh frozen (n= 6) or formalin-fixed paraffin-embedded
(FFPE; n= 6) tissue. Duration of tumor storage ranged from
7 months to 8 years. DNA was extracted using commercially
available kits (Qiagen and Bioline) according to manufacturer
protocols. FFPE samples were deparaffinized and additional
DNA repair steps were performed using uracil-N-glycosylase
to enzymatically remove formalin-induced cytosine deamina-
tion artifacts.
Whole Exome Sequencing
WES of germline and tumor DNA samples was performed
using the Roche NimbleGen SeqCap EZ MedExome v3.0
target enrichment kit, and the Illumina NextSeq 500 sequenc-
ing platform. The average of mean depth of coverage amongst
all samples was 129x, and 97% oftarget bases were covered ≥
20x.
Filtration of Sequence Variants
Bioinformatic analysis was performed in the ACRF Cancer
Genomics Facility of the Centre for Cancer Biology, SA
Pathology (Adelaide, Australia). BWA-MEM was used to
align short reads to GRCh37/hg19 (version b37+decoy).
Small variants (typically < 50 bp) were called using Genome
Analysis Toolkit (GATK) HaplotypeCaller package version
3.4. Raw WES data were initially filtered for variants that
were: high quality (by GATK internal filters); very rare (<
0.2% population); potentially functional (by snpEFF impact,
branching/binding predictions, GERP, or CADD); and not in
regions of segmental duplication.
Germline variants were considered further if they had a
GATK genotype quality (GQ) score > 50 and depth of cover-
age > 30x, and were not situated in a low complexity region.
Drawing on existing literature, we searched for germline var-
iants in known FPTS genes: AIP,CDH23,CDKN1B,
DICER1,GPR101,MAX,MEN1,PRKAR1A,SDHA,SDHB,
SDHC,andSDHD [14,20–23].
Artifact was observed in tumor DNA results due to reasons
including presumed normal tissue admixture and DNA degra-
dation in FFPE specimens. Raw data from tumor DNA were
thus re-analyzed by a dedicated in-house somatic variant call-
ing pipeline to identify variants present in tumor DNA and
absent in germline DNA. To increase the reliability of somatic
variant calls, this pipeline integrates four variant callers that
Endocr Pathol
detect insertions/deletions (“indels”) and single-nucleotide
variants (SNVs): Shimmer (v e5bafb4), Seurat (v 2.6),
Strelka2 (v 2.9.0), and VarScan2 (v 2.4.0); and three callers
that detect SNVs only: MuTect (v 1.1.4), SomaticSniper (v
1.0.5), and Virmid (v 1.1.1). Only somatic indels and SNVs
that were detected by at least two or five variant callers, re-
spectively, were considered to be candidate somatic sequence
variants.
These candidate somatic variants were shortlisted to a final
list of somatic variants of interest that were absent in popula-
tion genomic databases (dbSNP, 1KG, UK10K, gnomAD,
ExAC, and ESP) with evidence of being highly damaging
(high snpEFF impact). Pituitary expression of these final
genes of interest was determined using the Genotype-Tissue
Expression (GTEx) project database (https://gtexportal.org),
comprising 53 non-diseased tissue sites including 183 pitui-
tary samples.
All germline variants in known FPTS genes and somatic
variants of interest were verified by inspection of raw se-
quencing data in Integrated Genomics Viewer (IGV).
Identification of Somatic Copy Number Variants
Raw WES data were interrogated for copy number variation
(CNV) via in-house scripts, with calculation of copy number
using a normalized read depth of coverage against control
samples and correlation with minor allele frequency.
Coverage plots of sequence read depth and minor allele fre-
quency was manually inspected to identify chromosomal and
arm level copy number gains and losses as well as copy num-
ber neutral loss of heterozygosity (LOH).
Statistical Analysis
IBM SPSS Statistics 25.0 was used for statistical analysis. The
Mann-Whitney Utest was used to assess differences in the
median numbers of candidate somatic variants and chromo-
somes affected by CNV or copy neutral LOH per tumor ac-
cording to relevant categorical clinical characteristics. P
values < 0.05 were considered statistically significant.
Results
Clinical Characteristics
The study cohort consisted of six women and six men aged
16–65 years at prolactinoma diagnosis. The tumors
hypersecreted prolactin alone. Apart from one patient who
presented with pituitary apoplexy, all patients were treated
with dopamine agonists (DA) preoperatively. In all cases, sur-
gical resection was by the trans-sphenoidal route and histopa-
thology confirmed pituitary adenomas with positive
immunostaining for prolactin. Postoperative tumor remnants
or recurrences were observed in ten patients, all of whom had
macroadenomas or giant adenomas at the baseline scan. The
remaining two patients had microadenomas that were resected
because of DA intolerance, with gross total resection achieved
and no evidence of tumor recurrence to date. No patients had
received other medical therapies or radiotherapy at study en-
rollment. Other clinical characteristics of the patient cohort are
given in Table 1.
Pathological Characteristics
Pathological characteristics are described in Table 2. Most
tumors were densely granulated lactotroph adenomas. Ki-67
index was only available in a minority of tumors. Few or no
mitoses were observed in the remaining tumors, arguing
against a significant degree of proliferation [24]. Histological
invasion was found in only three tumors. Despite DA pretreat-
ment in all but one patient, fibrosis was only observed in 4/11
DA-treated tumors (Fig. 1).
Germline Sequence Variants
The only known FPTS gene with germline sequence variants
after filtration was CDH23, with missense variants observed
in Patient 6 (c.1103G>A (p.Arg368His), population preva-
lence in gnomAD 0.06%, CADD 26.2, GERP 4.5) and
Patient 11 (c.4510G>T (p.Ala1504Ser), gnomAD 0.01%,
CADD 25, GERP 5.06; and c.4907C>T (p.Ala1636Val),
gnomAD 0.07%, CADD 22.5, GERP 5.75).
Somatic Sequence Variants
Filtration of WES data revealed 138 candidate somatic vari-
ants, none of which were found in more than one tumor. Only
one gene (PHTF1) was mutated in more than one tumor.
Another two genes (NBEAL2,TMEM67) were each mutated
twice in the same tumor from Patient 1. Of the 135 different
genes containing candidate somatic variants, there was no
overlap with genes implicated in FPTS or sporadic pituitary
adenomas (i.e., AIP,CDH23,CDKN1B,DICER1,GNAS,
GPR101,MAX,MEN1,NR3C1,PRKAR1A,PRLR,SDHA,
SDHB,SDHC,SDHD,TP53,USP8).
Each tumor harbored multiple candidate somatic variants
(median 9.5 per tumor, range 3–23). There was no significant
difference in the median number of variants according to gen-
der (male 7.5 vs female 15.0, P= 0.107), indication for sur-
gery (DA intolerance 9.5 vs other indications 14.0, P=0.624),
tumor consistency (no cystic component 9.0 vs cystic compo-
nent 15.0, P= 0.114), or extent of resection (no remnant 9.5 vs
remnant 14.0, P=0.780).
The 138 candidate somatic variants were shortlisted to 15
variants of interest (Table 3;Fig.2)thatwereabsentin
Endocr Pathol
population genomic databases and highly damaging (n=14),
or situated in a gene with another candidate somatic variant in
another tumor (n=1, PHTF1). The shortlist included non-
sense or frameshift variants in three genes (DRD2,PRL,
TMEM67) with known associations with the pituitary gland
and the MLH3 gene which is a tumorigenesis gene in other
tissues. Using the GTEx database, we observed that the pitu-
itary was in the top 10% of expressing tissues for 8/15
shortlisted genes of interest. We next used the STRING data-
base (https://string-db.org) of known and predicted protein-
protein interactions to look for interactions between the 15
genes of interest. The only interaction was the known link
between PRL encoding prolactin and DRD2 encoding the
D2 dopamine receptor (D2R), which are co-expressed in mul-
tiple species and co-mentioned in medical literature.
No patient had a germline variant in the same gene con-
taining a somatic variant of interest in their corresponding
tumor. Conversely, no candidate somatic variants were found
in CDH23 in the two patients with germline CDH23 variants.
Somatic Copy Number Variants
All but one tumor contained chromosomal or arm level CNVs
and/or copy neutral LOH (median 10.5 chromosomes affected
per tumor, range 0–21). There was no significant difference in
the median number of chromosomes affected according to
gender (male 10.0 vs female 12.0, P= 0.377), surgical indica-
tion (DA intolerance 8.0 vs other indications 12.0, P=0.514),
tumor consistency (no cystic component 10.0 vs cystic
component 17.0, P= 0.266), or extent of resection (no rem-
nant 10.0 vs remnant 10.0, P= 0.926). CNV most commonly
manifested as whole or partial chromosomal gain (median 10
chromosomes per tumor, range 0–20), and occasionally as
whole or partial chromosomal loss (median 0 chromosomes
per tumor, range 0–2). Copy neutral LOH was also seen (me-
dian 0.5 chromosomes per tumor, range 0–6).
Recurrent and single cases of chromosomal gain, loss, and
copy neutral LOH are shown in Table 4. Examples of CNV
calling are depicted in Fig. 3. The most frequent chromosomes
affected were Chr 8, 9, and 14 followed by Chr 3, 7, 12 and 20
for gains, and 1 and 15 for copy neutral LOH. No chromo-
somes showed recurrent losses.
Each tumor was assessed for regional overlap between its
CNV results and any observed variant of interest. Copy num-
ber gain waspresentin the corresponding tumor at the locus of
10/15 genes of interest and corresponding monosomy was
observed for 1/15 genes.
Discussion
The major somatic event in our cohort of 12 patients with
prolactinomas was large-scale CNV, most commonly in the
form of copy number gains. We also observed sequence var-
iants of interest in 15 genes, including genes of putative inter-
est in prolactinoma tumorigenesis. Although we found that
pituitary expression is in the top 10% of tissues for over half
of our genes of interest, these somatic variants do not appear to
Table 1 Clinical characteristics at time of diagnosis
Patient Age (yr),
gender
Tumor
maximum
diameter (mm)
Hardy’s
score
Tumo r
consistency
PRL
(xULN)
Surgical
indication
Postoperative
remnant
#
Tumo r
recurrence
*
140F 16
^
3
^
Solid
^
33
^
DA resistance Yes N/A
2 16 F 8 1 Cystic 9 DA intolerance No N
3 56 M 8 3 Solid 10 DA intolerance No N
4 42 F 11 3 Solid 5 DA intolerance Yes N/A
5 28 F 60 3 Solid 278 DA resistance Yes N/A
6 53 M 18 3 Solid 145 DA intolerance Yes N/A
7 32 M 26 3 Solid 20 DA resistance Yes N/A
8 64 M 52 3 Mixed 576 Apoplexy at Dx Yes N/A
9 65 M 37 3 Solid 67 DA resistance Yes N/A
10 32 F 16^ 2^ Solid
^
28^ DA intolerance No Y
11 40 M 41 3 Solid 215 DA resistance Yes N/A
12 61 F 46 3 Mixed 72 DA resistance Yes N/A
DA, dopamine agonist; Dx,diagnosis;F,female;M, male; N/A, not applicable; N/S, not stated in report and images not available for review; PRL,
prolactin; xULN, absolute level divided by upper limit of normal; yr,year
#
Based on postoperative imaging and serum prolactin results
*
Only applies to tumors that were completely resected
^
Preoperative results used as results at initial diagnosis unavailable
Endocr Pathol
be classical driver mutations as none were found in more than
one tumor. We also found rare missense germline variants in
the recently recognized FPTS gene, CDH23. However, somat-
ic second-hits were not found in the corresponding tumors and
CDH23 is a notably large gene with 69 exons, which may
increase the propensity for variants of uncertain significance.
Our results recapitulate the findings of the few systematic
genomic studies of prolactinomas that have been performed
to date (Table 5), whereby CNV is common and recurrently
mutated genes are rare.
We observed several recurrent copy number gains and copy
neutral LOH. This is in keeping with the pituitary
macroadenoma WES study by Bi et al. [18], and their
follow-up study of 114 pituitary adenomas including 14
prolactinomas [2], that indicated two patterns of CNV: a high-
ly disrupted group mostly consisting of functional adenomas
(including prolactinomas) or atypical null cell adenomas with
CNV involving a mean of 39% of the genome; and a less
disrupted group mostly consisting of non-functioning adeno-
mas with CNVinvolving a mean of 0.5% of the genome. Song
et al. [3] also found a high degree of CNV in their mixed
cohort of pituitary adenomas, with almost one-third of pitui-
tary adenomas showing CNVinvolving > 80% of the genome.
Our patients’tumors demonstrated recurrent gains in Chr 7, 9,
and 14 similar to Bi et al. [18] and in Chr 1, 3, 7, 16, and 20
similar to Song et al. [3]. In contradistinction to these former
studies, we observed additional recurrent gains in Chr 5–7, 10,
12, 17–19, 21, 22, and X. We also found no recurrent chro-
mosomal losses, whereas Bi et al. [18] found losses to be
particularly common in Chr 1p and 11 in hormonally active
adenomas. While we found recurrent copy neutral LOH in
Chr 1, 4, and 15, Bi et al. [18] only found Chr 11q LOH.
Discrepancies between the different cohorts are at least partly
explained by the heterogenous mixes of different pituitary
adenoma subtypes in previous studies. Our pure prolactinoma
cohort should be considered separately to these previous stud-
ies because of the potential for specific DA treatment effects in
11 of our 12 patients who were treated with a DA preopera-
tively. DA resistance may have led to the observed high rate of
CNV or vice versa.
Table 2 Pathological characteristics of patient cohort at time of diagnosis
Patient Positive hormone
IHC
Granulation pattern Mitoses Ki-67 index Histological invasion Fibrosis
1 PRL Undetermined Scant U/A No No
2 PRL Densely granulated < 1/10hpf U/A No No
3 PRL Densely granulated Nil U/A No Yes
4 PRL Undetermined Scant U/A No No
5 PRL, LH Densely granulated Scant U/A Yes—sphenoid, nasopharynx Yes
6 PRL Densely granulated < 1/10hpf U/A Yes—dura No
7 PRL, TSH, LH, FSH Undetermined Scant U/A No No
8 PRL Sparsely granulated Nil U/A Yes—sphenoid No
9 PRL Densely granulated Nil <1% N Yes
10 PRL Densely granulated < 1/10hpf 3% No No
11 PRL Densely granulated Nil < 1% No Yes
12 PRL Sparsely granulated < 1/10hpf U/A* No No
FSH, follicle-stimulating hormone; IHC, immunohistochemistry; LH, luteinizing hormone; PRL, prolactin; TSH, thyroid-stimulating hormone; U/A,
unavailable
*Topoisomerase index 5%
Fig. 1 Hematoxylin & eosin
appearance of prolactinomas at
medium power in a case
demonstrating no fibrosis (a
Patient 12) and another
demonstrating marked fibrosis (b
Patient 11)
Endocr Pathol
By contrast to the high burden of CNV, we found relatively
few sequence variants per tumor and a lack of recurrent se-
quence variants between tumors. This argues against a major
role of driver mutations in the pathogenesis of prolactinomas.
This differs from the experience of studying
somatotrophinomas, corticotrophinomas, and
craniopharyngiomas, but mimics findings in other pituitary
tumor subtypes. Paired tumor-normal WES studies of seven
patients with non-functioning pituitary adenomas in 2013 [25]
and of four patients with TSHomas in 2016 [26]alsofoundno
recurrent variants that could be considered driver mutations. A
low number of somatic mutations have been observed in pre-
vious studies of pituitary adenomas compared with other neo-
plasms [3,18]. However, direct comparison of absolute mu-
tation numbers between studies is limited by differing
methods of variant filtration. Within studies, we and others
[3] have found no association between prolactinoma clinical
characteristics and number of sequence variants.
Some variants within individual tumors were of interest
due to their location in genes with a plausible connection to
prolactinoma formation. This includes the truncating DRD2
variant found in a patient with a 40-year history of
prolactinoma that showed DA escape over the last 3 years
despite increasing doses of cabergoline, necessitating surgery
and radiotherapy. We speculate that this variant, found in 19%
of tumor DNA, may reflect a subclone that is driving the
patient’s recent DA resistance. Indeed, D2R expression is typ-
ically high in prolactinomas, and downregulation has been
hypothesized as a mechanism of DA resistance [27].
Compared with DA-responsive prolactinomas, resistant tu-
mors demonstrate decreased D2R density, overall reduction
in D2R mRNA production, and altered expression of D2R
mRNA isoforms with lower expression of the more efficient
short isoform [28]. In female mouse models, D2R deficiency
induces lactotroph hyperplasia [29]. However, we did not find
DRD2 variants of interest in our other five patients with DA
Table 3 Somatic sequence variants of interest
Gene, ID
a
Chr locus Pituitary rank
amongst all
tissue expression
b
CNV at locus
c
Pt: variant VAF in tumor DNA
ANKS3, ENSG00000168096 16: 4780135 3 Trisomy Pt 1: c.15delC
(p.Ser5fs)
33%
C19orf25, ENSG00000119559 19: 1475427 3 Trisomy Pt 7: c.216dupG
(p.Ile73fs)
12%
C9orf163, ENSG00000196366 9: 139379389 2 Trisomy Pt 1: c.491delC
(p.Pro164fs)
15%
CAST, ENSG00000153113 5: 96077063 35 Tetrasomy (2:2) Pt 5: c.888+1G>T 30%
DCAF10, ENSG00000122741 9: 37860080 31 Tetrasomy (2:2) Pt 12: c.1202_1203delCT
(p.Pro401fs)
26%
DRD2, ENSG00000149295 11: 113283323 1 Monosomy Pt 1: c.1093C>T
(p.Gln365*)
19%
KLRD1, ENSG00000134539 12: 10460684 4 Trisomy Pt 1: c.7+1G>C 21%
LDB2, ENSG00000169744 4: 16597359 34 Nil Pt 9: c.3G>A
(p.Met1?)
37%
MLH3, ENSG00000119684 14: 75514552 12 Trisomy Pt 8: c.1806delA
(p.Lys602fs)
27%
NBEAL2, ENSG00000160796 3: 47036629 25 Tetrasomy (2:2) Pt 1:
p.Phe470fs, c.1407_1408delCT
41%
PHTF1, ENSG00000116793 1: 114242872 5 Nil Pt 4: n.468-7insA
Pt 7: n.468-7delA
Pt 4: 33%
Pt 7: 16%
PRL, ENSG00000172179 6: 22290411 1 Trisomy Pt 4: c.483dupA
(p.Val162fs)
13%
SKIDA1, ENSG00000180592 10: 21805663 10 Nil Pt 5: c.1088delC
(p.Pro363fs)
16%
SPTBN2, ENSG00000173898 11: 66453356 13 Nil Pt 4: c.7159A>T
(p.Lys2387*)
35%
TMEM67, ENSG00000164953 8: 94797512 2 Trisomy Pt 1: c.1194C>A
(p.Tyr398*)
40%
Chr, chromosomal; CNV, copy number variation; Pt, Patient number corresponding to tumor in which variant detected; VAF, variant allele frequency
a
HGNC gene symbol, gene ID by snpEFF
b
Pituitary rank after sorting all 53 non diseased tissue types by median TPM(transcripts per million) in GTEx
c
Corresponding CNVat the gene locus in the same tumor
Endocr Pathol
resistance. Wang et al. [16] also reported an absence of DRD2
sequence variants, although the sensitivity of this study was
reduced by its overall low depth of coverage with only 10 x
coverage in 80% of the exome. Another tumor in our cohort
harbored a frameshift variant in PRL, which is well known to
be highly expressed in lactotrophs [27]. Autocrine signaling
between prolactin and the abundant prolactin receptors on
lactotrophs has been postulated as the explanation for the sex-
ual dimorphism in lactotroph hyperplasia in D2R knockout
mice [29]. By this theory, male mice lacking the D2R do not
reach the prolactin threshold required for the feed-forward
loop to activate and trigger lactotroph hyperplasia [29]. We
also found isolated somatic variants in TMEM67 where
biallelic inactivating variants have been implicated in hypopi-
tuitarism [30], and in MLH3 which is a mismatch repair gene
with a possible role in Lynch syndrome [31]. Although a
Lynch syndrome registry study found an increased risk of
pituitary adenomas [32], there is currently no evidence of a
specific role for MLH3 in pituitary tumorigenesis.
The remaining variants of interest were located in genes
with no currently known associations with the pituitary
gland. Comparison with the previously published genomic
studies including prolactinomas showed little overlap:
Song et al. [3] found one ANKS3 frameshift variant and
two SKIDA1 variants; and Bi et al. [18] found a KLRD1
missense variant. None of these variants were seen in the
prolactinoma subsets of these studies. In addition, none of
our cases fulfilled Knudson’s two-hit model of tumor sup-
pressor genes as no patients had germline variants in the 15
genes harboring somatic variants of interest and the two
patients with germline CDH23 variants had no candidate
somatic variants in CDH23. CNV might have arguably
been the second-hit in some of these tumors as 11/15
(73%) of variants of interest were in regions of CNV in a
given tumor. Trisomy and tetrasomy could be especially
relevant as increased mutant dosage can amplify a domi-
nant negative effect by a sequence variant, thereby contrib-
uting to tumorigenesis. On the other hand, the
maximum VAF was 41% among the 15 variants of interest
despite the frequent coexistence of CNV. Furthermore,
most prolactinomas in our study harbored multiple CNV
and the CNVs were large; thus, it is unlikely that any single
gene in these regions of CNV can explain the pathogenesis
of prolactinomas.
Fig. 2 Examples of somatic point variants as visualized in Integrated Genomics Viewer. aCAST missense variant in Patient 5: c.888+1G>T. bC9orf163
deletion in Patient 1: c.491delC (p.Pro164fs). cPRL insertion in Patient 4: c.483dupA (p.Val162fs)
Table 4 Somatic CNV analysis
summary showing chromosomes
with whole or partial gains, losses
and copy neutral LOH in either
multiple or single tumors
Copy neutral LOH Gain Loss Mixed CNV
and copy neutral
LOH
Recurrent Chr 1, 4, 15 Chr 1, 3, 5–10, 12, 14,
16–22, X
nil nil
Single Chr 5, 6, 10, 11, 16,
20
Chr 2, 4, 11, 13, 15 Chr 11, 13, 15, 18,
X
Chr 1, 4, 11
Chr, chromosome; CNV, copy number variants; LOH, loss of heterozygosity
Endocr Pathol
A key limitation of our study is that WES does not
detect deep intronic and intergenic variants, balanced
translocations, fusion genes, or epigenetic changes.
Integrative genomic analyses of both DNA and RNA
[33] as well as the emerging technology of long-read se-
quencing with real-time analysis of nucleotide binding [34]
may elucidate some of these possibilities. Furthermore,
half of our tumor samples were FFPE, although we
employed an optimized DNA extraction protocol to limit
artifactual results because of this. Another limitation of our
study is that the identification of CNV and copy neutral
LOH was based on broad patterns in VAF and ploidy
based on depth of coverage. We were thus only able to
categorize CNV and copy neutral LOH at the arm or
chromosomal level. Smaller CNVs may have been missed,
although other data support the predominance of large-
scale CNV, as observed in our tumors, over smaller
CNVs [18]. The relatively low allele frequencies of our
variants of interest are also noteworthy. This may seem-
ingly contradict the known monoclonal origin of pituitary
adenomas [27]. However, the < 50% VAFs seen in our
tumor DNA results may reflect CNV, which was a com-
mon finding in our tumors, and/or normal tissue admix-
ture, particularly as pituitary adenomas are rarely resected
en bloc and interspersed normal pituitary tissue is a com-
mon microscopic finding. VAFs < 50% may alternatively
represent the presence of multiple tumor clones. The
possibility of this may be assessed in future studies through
spatial transcriptomics whereby sequencing results are
overlaid with tissue sections to compare the transcriptomes
of different tumor regions [35]. Finally, the small size of
this pilot study limited our ability to identify clinical pre-
dictors of the number of somatic variants and the number
of chromosomes affected by CNV or copy neutral LOH.
An independent validation set of another group of
prolactinomas using the same platforms employed in this
study would have been ideal to further explore these pu-
tative clinicopathological correlations and our identified
genes of interest; however, surgery is rarely performed
for prolactinomas and thus, further tumors were not avail-
able for investigation.
Larger studies involving sufficient numbers of different
prolactinoma subsets (e.g., cystic prolactinomas or young
onset prolactinomas in males) with use of fresh frozen
tumor samples may better elucidate the genetic drivers
of tumorigenesis. Our findings of suspicious albeit isolat-
ed somatic variants in strong candidate genes such as
DRD2 may be a function of the heterogenous patient
and tumor case mix in the prolactinoma studies to date.
We kept our inclusion criteria at a minimum in order to
capture sufficient numbers of prolactinomas, which other-
wise tend to be medically managed with DAs. Routine
biobanking of pituitary adenomas will help facilitate fu-
ture studies, although resected tumor tissue is often
Fig. 3 Examples of tumor DNA calls of normal diploid status,
chromosomal gain, chromosomal loss, and copy neutral loss of
heterozygosity (LOH) based on somatic heterozygous variant allele fre-
quency (VAF) (top panels) and ploidy estimates using depth of coverage
(bottom panels). aDisomic baseline in Chr 2 in Patient 6 represented by
the usual 0.5 heterozygous VAF and ploidy count of 2. bChr 3 trisomy
(2:1) in Patient 10 represented by separation of heterozygous VAF into
VAFs of approximately 0.4 and 0.6 and increased ploidy count at 3. cChr
9 tetrasomy (2:2) in Patient 10 represented by usual 0.5 heterozygous
VAF but increased ploidy count at 4. dChr X monosomy in Patient 1
(female) represented by separation of heterozygous VAF and decreased
ploidy count at 1. eChr 4 copy neutral LOH in Patient 1 represented by
separation of heterozygous VAF but normal ploidy count of 2
Endocr Pathol
piecemeal, in small quantity and potentially damaged by
intraoperative cauterisation.
In conclusion, this systematic genomic study of all coding
genes in a pure prolactinoma cohort demonstrated variants in
genes of biologically plausible interest within individual tu-
mors, without overlap between prolactinomas in this study or
with the few other published pangenomic studies [2,3,16,
17]. We instead found a high degree of CNV, corroborating
other preliminary studies of sporadic prolactinomas [19]and
larger studies of mixed pituitary adenoma subtypes [2,3,18].
Further research is required to determine how CNV may con-
tribute to prolactinoma formation and ways in which this
could be therapeutically targeted.
Funding SMCD is supported by an A.R. Clarkson Scholarship from the
Royal Adelaide Hospital. This study was produced with support from a
Royal Adelaide Hospital Health Services Charitable Gifts Board grant,
and with the financial and other support of the Cancer Council SA’sBeat
Cancer Project on behalf of its donors and the State Government of South
Australia through the Department of Health.
Compliance with Ethical Standards
Conflict of Interest The authors declare that they have no conflict of
interest.
Ethical Approval All procedures performed in studies involving human
participants were in accordance with the ethical standards of institutional
research committees (Melbourne Health: HREC/16/MH/132; Royal
Table 5 Pangenomic studies of paired tumor and germline DNA from prolactinoma patients. All studies employed whole exome sequencing
Study Cohort Filter for GOI GOI Recurrent CNV
Wang 2014
[16],
Gao 2015
[17]
DA responsive vs resistant PRLoma
(n=12)
Variants differing between
responsive
vs resistant PRLoma
C1orf170
DPCR1
DSPP
KRTAP10–3
MUC4
MX2
POTEF
PRB3*
PRDM2*
PRG4
RP1L1
N/T
Song 2016
[3]
Pituitary adenoma (n=120,incl20
PRLoma)
Recurrently mutated in multiple
PA
GRB10*
IARS
KIF5A*
SP100
TRIP12
Gains:
Chr 1p13.2, 1q31.3, 3p22.3, 7q21.11,
16q12.2, 20p13,
20q13.33
Losses:
Chr 1p36.31, 3p21.31, 9q34.11, 11q13.2,
11p15.5, 16p13.3
Bi 2017 [18] Pituitary macroadenomas (n=42,incl
3 PRLoma)
Recurrently mutated in multiple
PA
ATAD3B
BHLHE22
KDM2B
OR5M1
TTN*
VPS13B
Gains:
Chr 7, 9q, 14q
Losses:
Chr1,2,4,10,11,15q,18
Copy neutral LOH:
Chr 11q
Present
study
PRLoma
(n=12)
Absent in population databases
and strong
in silico prediction for
pathogenicity, or
recurrently mutated in
multiple PRLoma
ANKS3
C19orf25
C9orf163
CAST
DCAF10
DRD2*
KLRD1
LDB2
MLH3
NBEAL2
PHTF1
PRL*
SKIDA1
SPTBN2
TMEM67*
Gains:
Chr1,3,5–10, 12, 14, 16–22, X
Losses:
nil
Copy neutral LOH:
Chr1,4,15
Chr, chromosome; CNV, copy number variants; DA, dopamine agonist; GOI, genes of interest; incl, including; LOH, loss of heterozygosity; n,number of
cases that underwent whole exome sequencing; N/T, not tested; PA, pituitary adenoma; PRLoma, prolactinoma
*particular genes of interest
Endocr Pathol
Adelaide Hospital: SSA/18/CALHN/445) and with the National Health
and Medical Research Council guidelines.
Informed Consent Informed consent was obtained from all individual
participants included in the study.
References
1. Tordjman K, Stern N, Ouaknine G, Yossiphov Y, Razon N,
Nordenskjold M, Friedman E (1993) Activating mutations of the
Gs alpha-gene in nonfunctioning pituitary tumors. J Clin
Endocrinol Metab 77 (3):765–769. https://doi.org/10.1210/jcem.
77.3.8396579
2. Bi WL, Greenwald NF, Ramkissoon SH, Abedalthagafi M, Coy
SM,LigonKL,MeiY,MacConaill L, Ducar M, Min L,
Santagata S, Kaiser UB, Beroukhim R, Laws ER, Jr., Dunn IF
(2017) Clinical Identification of Oncogenic Drivers and Copy-
Number Alterations in Pituitary Tumors. Endocrinology 158 (7):
2284–2291. https://doi.org/10.1210/en.2016-1967
3. Song ZJ, Reitman ZJ, Ma ZY, Chen JH, Zhang QL, Shou XF,
Huang CX, Wang YF, Li SQ, Mao Y, Zhou LF, Lian BF, Yan H,
Shi YY, Zhao Y (2016) The genome-wide mutational landscape of
pituitary adenomas. Cell Res 26 (11):1255–1259. https://doi.org/
10.1038/cr.2016.114
4. Reincke M, Sbiera S, Hayakawa A, Theodoropoulou M, Osswald
A, Beuschlein F (2015) Mutations in the deubiquitinase gene USP8
cause Cushing's disease. Nat Genet 47 (1):31–38. https://doi.org/
10.1038/ng.3166
5. Brastianos PK, Taylor-Weiner A, Manley PE, Jones RT, Dias-
Santagata D, Thorner AR, Lawrence MS, Rodriguez FJ, Bernardo
LA, Schubert L, Sunkavalli A, Shillingford N, Calicchio ML,
Lidov HG, Taha H, Martinez-Lage M, Santi M, Storm PB, Lee
JY, Palmer JN, Adappa ND, Scott RM, Dunn IF, Laws ER, Jr.,
Stewart C, Ligon KL, Hoang MP, Van Hummelen P, Hahn WC,
Louis DN, Resnick AC, Kieran MW, Getz G, Santagata S (2014)
Exome sequencing identifies BRAF mutations in papillary
craniopharyngiomas. Nat Genet 46 (2):161–165. https://doi.org/
10.1038/ng.2868
6. Bi WL, Larsen AG, Dunn IF (2018) Genomic Alterations in
Sporadic Pituitary Tumors. Curr Neurol Neurosci Rep 18 (1):4.
https://doi.org/10.1007/s11910-018-0811-0
7. Cohen M, Persky R, Stegemann R, Hernandez-Ramirez LC, Zeltser
D, Lodish MB, Chen A, Keil MF, Tatsi C, Faucz F, Buchner D,
Stratakis CA, Tiosano D (2019) Germline USP8 mutation associ-
ated with pediatric Cushing disease and other clinical features: a
new syndrome. J Clin Endocrinol Metab. https://doi.org/10.1210/
jc.2019-00697
8. Poncin J, Stevenaert A, Beckers A (1999) Somatic MEN1 gene
mutation does not contribute significantly to sporadic pituitary tu-
morigenesis. Eur J Endocrinol 140 (6):573–576
9. Raitila A, Georgitsi M, Karhu A, Tuppurainen K, Makinen MJ,
Birkenkamp-Demtroder K, Salmenkivi K, Orntoft TF, Arola J,
Launonen V, Vahteristo P, Aaltonen LA (2007) No evidence of
somatic aryl hydrocarbon receptor interacting protein mutations in
sporadic endocrine neoplasia. Endocr Relat Cancer 14 (3):901–906.
https://doi.org/10.1677/erc-07-0025
10. Tanizaki Y, Jin L, Scheithauer BW, Kovacs K, Roncaroli F, Lloyd
RV (2007) P53 gene mutations in pituitary carcinomas. Endocr
Pathol 18 (4):217–222. https://doi.org/10.1007/s12022-007-9006-y
11. Yagnik G, Jahangiri A, Chen R, Wagner JR, Aghi MK (2017) Role
of a p53 polymorphism in the development of nonfunctional pitu-
itary adenomas. Mol Cell Endocrinol 446:81–90. https://doi.org/10.
1016/j.mce.2017.02.017
12. Gorvin CM, Newey PJ, Rogers A, Stokes V, Neville MJ, Lines KE,
Ntali G, Lees P, Morrison PJ, Singhellakis PN, Malandrinou FC,
Karavitaki N, Grossman AB, Karpe F, Thakker RV (2018)
Association of prolactin receptor (PRLR) variants with
prolactinomas. Hum Mol Genet. https://doi.org/10.1093/hmg/
ddy396
13. Cai WY, Alexander JM, Hedley-Whyte ET, Scheithauer BW,
Jameson JL, Zervas NT, Klibanski A (1994) ras mutations in hu-
man prolactinomas and pituitary carcinomas. J Clin Endocrinol
Metab 78 (1):89–93. https://doi.org/10.1210/jcem.78.1.8288721
14. Trivellin G, Daly AF, Faucz FR, Yuan B, Rostomyan L, Larco DO,
Schernthaner-Reiter MH, Szarek E, Leal LF, Caberg J-H,
Castermans E, Villa C, Dimopoulos A, Chittiboina P, Xekouki P,
Shah N, Metzger D, Lysy P, Ferrante E, Strebkova N, Mazerkina N,
Zatelli M, Lodish M, Horvath A, de Alexandre R, Manning A,
Levy I, Keil M, Sierra Mde L, Palmeira L, Coppieters W,
Georges M, Naves L, Jamar M, Bours V, Wu T, Choong C,
Bertherat J, Chanson P, Kamenický P, Farrell W, Barlier A,
Quezado M, Bjelobaba I, Stojilkovic S, Wess J, Costanzi S, Liu P,
Lupski J, Beckers A, Stratakis C (2014) Gigantism andacromegaly
due to Xq26 microduplications and GPR101 mutation. N Engl J
Med 371 (25):2363–2374
15. Assié G, Libé R, Espiard S, Rizk-Rabin M, Guimier A, Luscap W,
Barreau O, Lefèvre L, Sibony M, Guignat L, Rodriguez S,
Perlemoine K, René-Corail F, Letourneur F, Trabulsi B, Poussier
A, Chabbert-Buffet N, Borson-Chazot F, Groussin L, Bertagna X,
Stratakis C, Ragazzon B, Bertherat J (2013) ARMC5 mutations in
macronodular adrenal hyperplasia with Cushing's syndrome. N
Engl J Med 369 (22):2105–2114
16. Wang F, Gao H, Li C, Bai J, Lu R, Cao L, Wu Y, Hong L, Wu Y,
Lan X, Zhang Y (2014) Low levels of PRB3 mRNA are associated
with dopamine-agonist resistance and tumor recurrence in
prolactinomas. J Neurooncol 116 (1):83–88. https://doi.org/10.
1007/s11060-013-1276-2
17. Gao H, Wang F, Lan X, Li C, Feng J, Bai J, Cao L, Gui S, Hong L,
Zhang Y (2015) Lower PRDM2 expression is associated with
dopamine-agonist resistance and tumor recurrence in
prolactinomas. BMC Cancer 15:272. https://doi.org/10.1186/
s12885-015-1267-0
18. Bi WL, Horowitz P, Greenwald NF, Abedalthagafi M, Agarwalla
PK, Gibson WJ, Mei Y, Schumacher SE, Ben-David U, Chevalier
A, Carter S, Tiao G, Brastianos PK, Ligon AH, Ducar M,
MacConaill L, Laws ER, Jr., Santagata S, Beroukhim R, Dunn IF
(2017) Landscape of Genomic Alterations in Pituitary Adenomas.
Clin Cancer Res 23 (7):1841–1851. https://doi.org/10.1158/1078-
0432.ccr-16-0790
19. Wierinckx A, Delgrange E, Bertolino P, François P, Chanson P,
Jouanneau E, Lachuer J, Trouillas J, Raverot G (2018) Sex-
Related Differences in Lactotroph Tumor Aggressiveness Are
Associated With a Specific Gene-Expression Signature and
Genome Instability. Front Endocrinol (Lausanne) 9:706–706.
https://doi.org/10.3389/fendo.2018.00706
20. De Sousa SM, McCabe MJ, Wu K, Roscioli T, Gayevskiy V, Brook
K, Rawlings L, Scott HS, Thompson TJ, Earls P, Gill AJ, Cowley
MJ, Dinger ME, McCormack AI (2017) Germline variants in fa-
milial pituitary tumour syndrome genes are common in young pa-
tients and families with additional endocrine tumours. Eur J
Endocrinol 176 (5):635–644. https://doi.org/10.1530/eje-16-0944
21. Zhang Q, Peng C, Song J,Zhang Y, Chen J,SongZ, Shou X, Ma Z,
Peng H, Jian X, He W, Ye Z, Li Z, Wang Y, Ye H, Zhang Z, Shen M,
Tang F, Chen H, Shi Z, Chen C, Chen Z, Shen Y, Wang Y, Lu S,
Zhang J, Li Y, Li S, Mao Y, Zhou L, Yan H, Shi Y, Huang C, Zhao
Y (2017) Germline Mutations in CDH23, Encoding Cadherin-
Related 23, Are Associated with Both Familial and Sporadic
Pituitary Adenomas. Am J Hum Genet 100 (5):817–823. https://
doi.org/10.1016/j.ajhg.2017.03.011
Endocr Pathol
22. Rostomyan L, Daly AF, Petrossians P, Nachev E, Lila AR, Lecoq
AL,LecumberriB,TrivellinG,SalvatoriR,MoraitisAG,
Holdaway I, Kranenburg-van Klaveren DJ, Chiara Zatelli M,
Palacios N, Nozieres C, Zacharin M, Ebeling T, Ojaniemi M,
Rozhinskaya L, Verrua E, Jaffrain-Rea ML, Filipponi S,
Gusakova D, Pronin V, Bertherat J, Belaya Z, Ilovayskaya I,
Sahnoun-Fathallah M, Sievers C, Stalla GK, Castermans E,
Caberg JH, Sorkina E, Auriemma RS, Mittal S, Kareva M, Lysy
PA, Emy P, De Menis E, Choong CS, Mantovani G, Bours V, De
Herder W, Brue T, Barlier A, Neggers SJ,Zacharieva S, Chanson P,
Shah NS, Stratakis CA, Naves LA, Beckers A (2015) Clinical and
genetic characterization of pituitary gigantism: an international col-
laborative study in 208 patients. Endocr Relat Cancer 22 (5):745–
757. https://doi.org/10.1530/erc-15-0320
23. Pepe S, Korbonits M, Iacovazzo D (2019) Germline and mosaic
mutations causing pituitary tumours: genetic and molecular aspects.
J Endocrinol 240 (2):R21-r45. https://doi.org/10.1530/joe-18-0446
24. De Sousa SMC, McCormack AI (2019) Aggressive Pituitary
Tumors and Pituitary Carcinomas. In: Feingold KR, Anawalt B,
Boyce A et al. (eds) Endotext. MDText.com, Inc., South
Dartmouth (MA),
25. Newey PJ, Nesbit MA, Rimmer AJ, Head RA, Gorvin CM, Attar
M, Gregory L, Wass JA, Buck D, Karavitaki N, Grossman AB,
McVean G, Ansorge O, Thakker RV (2013) Whole-exome se-
quencing studies of nonfunctioning pituitary adenomas. J Clin
Endocrinol Metab 98 (4):E796–E800. https://doi.org/10.1210/jc.
2012-4028
26. Sapkota S, Horiguchi K, Tosaka M, Yamada S, Yamada M (2017)
Whole-Exome Sequencing Study of Thyrotropin-Secreting
Pituitary Adenomas. J Clin Endocrinol Metab 102 (2):566–575.
https://doi.org/10.1210/jc.2016-2261
27. Evans CO, Moreno CS, Zhan X, McCabe MT, Vertino PM,
Desiderio DM, Oyesiku NM (2008) Molecular pathogenesis of
human prolactinomas identified by gene expression profiling, RT-
qPCR, and proteomic analyses. Pituitary 11 (3):231–245. https://
doi.org/10.1007/s11102-007-0082-2
28. Wu ZB, Zheng WM, Su ZP, Chen Y, Wu JS, Wang CD, Lin C, Zeng
YJ, Zhuge QC (2010) Expression of D2RmRNA isoforms and
ERmRNA isoforms in prolactinomas: correlation with the response
to bromocriptine and with tumor biological behavior. J Neurooncol
99 (1):25–32. https://doi.org/10.1007/s11060-009-0107-y
29. Kelly MA, Rubinstein M, Asa SL, Zhang G, Saez C, Bunzow JR,
Allen RG, Hnasko R, Ben-Jonathan N, Grandy DK, Low MJ
(1997) Pituitary lactotroph hyperplasia and chronic
hyperprolactinemia in dopamine D2 receptor-deficient mice.
Neuron 19 (1):103–113
30. Brancati F, Camerota L, Colao E, Vega-Warner V, Zhao X, Zhang
R, Bottillo I (2018) Biallelic variants in the ciliary gene TMEM67
cause RHYNS syndrome. Eur J Hum Genet 26 (9):1266–1271.
https://doi.org/10.1038/s41431-018-0183-6
31. Wu Y, Berends MJ, Sijmons RH, Mensink RG, Verlind E, Kooi
KA, van der Sluis T, Kempinga C, van dDer Zee AG, Hollema H,
Buys CH, Kleibeuker JH, Hofstra RM (2001) A role for MLH3 in
hereditary nonpolyposis colorectal cancer. Nat Genet 29 (2):137–
138. https://doi.org/10.1038/ng1001-137
32. Bengtsson D, Joost P, Aravidis C, Askmalm Stenmark M, Backman
AS, Melin B,von Salome J, Zagoras T, Gebre-Medhin S, Burman P
(2017) Corticotroph Pituitary Carcinoma in a Patient With Lynch
Syndrome (LS) and Pituitary Tumors in a Nationwide LS Cohort. J
Clin Endocrinol Metab 102 (11):3928–3932. https://doi.org/10.
1210/jc.2017-01401
33. Branford S, Wang P, Yeung DT, Thomson D, Purins A, Wadham C,
Shahrin NH, Marum JE, Nataren N, Parker WT, Geoghegan J, Feng
J, Shanmuganathan N, Mueller MC, Dietz C, Stangl D, Donaldson
Z, Altamura H, Georgievski J, Braley J, Brown A, Hahn C, Walker
I, Kim SH, Choi SY, Park SH, Kim DW, White DL, Yong ASM,
Ross DM, Scott HS, Schreiber AW, Hughes TP (2018) Integrative
genomic analysis reveals cancer-associated mutations at diagnosis
of CML in patients with high-risk disease. Blood 132 (9):948–961.
https://doi.org/10.1182/blood-2018-02-832253
34. De Sousa SM, Hardy TS, Scott HS, Torpy DJ (2018) Genetic
Testing in Endocrinology. Clin Biochem Rev 39 (1):17–28
35. Thrane K, Eriksson H, Maaskola J, Hansson J, Lundeberg J (2018)
Spatially Resolved Transcriptomics Enables Dissection of Genetic
Heterogeneity in StageIII Cutaneous Malignant Melanoma. Cancer
Res 78 (20):5970–5979. https://doi.org/10.1158/0008-5472.can-
18-0747
Publisher’sNote Springer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
Endocr Pathol
A preview of this full-text is provided by Springer Nature.
Content available from Endocrine Pathology
This content is subject to copyright. Terms and conditions apply.