J C E M
O N L I N E
A d v a n c e s
i n
G e n e t i c s — E n d o c r i n e
R e s e a r c h
A Comprehensive Next Generation Sequencing–Based
Genetic Testing Strategy To Improve Diagnosis of
Inherited Pheochromocytoma and Paraganglioma
Eleanor Rattenberry, Lindsey Vialard, Anna Yeung, Hayley Bair, Kirsten McKay,
Mariam Jafri, Natalie Canham, Trevor R. Cole, Judit Denes, Shirley V. Hodgson,
Richard Irving, Louise Izatt, Márta Korbonits, Ajith V. Kumar, Fiona Lalloo,
Patrick J. Morrison, Emma R. Woodward, Fiona Macdonald, Yvonne Wallis,
and Eamonn R. Maher*
Context: Pheochromocytomas and paragangliomas are notable for a high frequency of inherited
cases, many of which present as apparently sporadic tumors.
Objective: The objective of this study was to establish a comprehensive next generation sequencing
(NGS)– based strategy for the diagnosis of patients with pheochromocytoma and paraganglioma
by testing simultaneously for mutations in MAX, RET, SDHA, SDHB, SDHC, SDHD, SDHAF2,
TMEM127, and VHL.
Design: After the methodology for the assay was designed and established, it was validated on DNA
samples with known genotype and then patients were studied prospectively.
Setting: The study was performed in a diagnostic genetics laboratory.
Patients: DNA samples from 205 individuals affected with adrenal or extraadrenal pheochromocytoma/head and neck paraganglioma (PPGL/HNPGL) were analyzed. A proof-of-principle study
was performed using 85 samples known to contain a variant in 1 or more of the genes to be tested,
followed by prospective analysis of an additional 120 samples.
Main Outcome Measures: We assessed the ability to use an NGS-based method to perform comprehensive analysis of genes implicated in inherited PPGL/HNPGL.
Results: The proof-of-principle study showed that the NGS assay and analysis gave a sensitivity of
98.7%. A pathogenic mutation was identified in 16.6% of the prospective analysis cohort of 120
patients.
Conclusions: A comprehensive NGS-based strategy for the analysis of genes associated with predisposition to PPGL and HNPGL was established, validated, and introduced into diagnostic service.
The new assay provides simultaneous analysis of 9 genes and allows more rapid and cost-effective
mutation detection than the previously used conventional Sanger sequencing– based
methodology. (J Clin Endocrinol Metab 98: E1248 –E1256, 2013)
heochromocytomas and paragangliomas are rare,
mostly benign, tumors that are notable for a high
frequency of inherited cases (1–5). Both pheochromocytomas (which arise from the adrenal medulla) and paragangliomas (also known as extraadrenal pheochromocytomas) (together referred to as PPGLs) usually present
P
with symptoms caused by the cardiovascular effects of
excess catecholamine secretion. Head and neck paragangliomas (HNPGLs) are generally nonfunctional and are
derived from parasympathetic ganglia. Before 2000,
about 10% of PPGLs were thought to be inherited (6) and
mostly associated with defined clinical syndromes: von
ISSN Print 0021-972X ISSN Online 1945-7197
Printed in U.S.A.
Copyright © 2013 by The Endocrine Society
Received February 5, 2013. Accepted April 26, 2013.
First Published Online May 10, 2013
* Author affiliations are shown at the bottom of the next page
Abbreviations: HNPGL, head and neck paraganglioma; MLPA, multiplex ligation-dependent probe amplification; NGS, next-generation sequencing; PPGL, pheochromocytoma
and paraganglioma; VFT, variant filtering threshold; VUS, variant of undetermined
significance.
E1248
jcem.endojournals.org
J Clin Endocrinol Metab, July 2013, 98(7):E1248 –E1256
doi: 10.1210/jc.2013-1319
doi: 10.1210/jc.2013-1319
Hippel-Lindau disease, multiple endocrine neoplasia 2A,
and neurofibromatosis type I (caused by VHL, RET, and
NF1 gene mutations, respectively) (2). Subsequently, mutations in the genes encoding the B and D subunits of
succinate dehydrogenase (SDHB and SDHD) were shown
to cause inherited PPGLs and HNPGLs (7–9). Furthermore, in a population-based cohort of individuals with
apparently nonsyndromic sporadic PPGL, almost onefourth of patients had a detectable mutation in RET,
SDHB, SDHD, or VHL, suggesting that about one-third
of all patients with PPGLs have an inherited cause for the
tumor (4). Currently, at least 10 genes (MAX, NF1, RET,
SDHA, SDHB, SDHC, SDHD, SDHAF2, TMEM127,
and VHL) are accepted as causes of inherited PPGL/HNPGL (10 –14), and further genes may be identified. Although a diagnosis of pheochromocytoma in the setting of
neurofibromatosis type I can usually be made on clinical
grounds alone, the phenotypes of other genetic causes are
variable and overlapping, and detection of all patients
with inherited PPGLs/HNPGLs requires molecular genetic testing. Because only about one-third of patients with
a detectable mutation have a positive family history, it was
initially proposed that all patients with PPGLs/HNPGLs
should be offered genetic testing (3, 15). However, as the
number of causative genes has increased, most centers
have adopted a targeted testing strategy to reduce costs (1,
2, 16). For example, testing for mutations in RET, SDHB,
SDHD, and VHL, is usually restricted to patients with 1
or more of the following: a family history of PPGL/HNPGL, age ⬍45 years, multiple tumors, extraadrenal location, and previous HNPGL (1). However, although such
targeted testing strategies have a high mutation detection
rate (percentage of tests that detect a mutation), a significant fraction of mutation-positive individuals remain undetected (17). Therefore, to maximize detection of patients with PPGLs/HNPGLs who harbor a germline
mutation in MAX, RET, SDHA, SDHB, SDHC, SDHD,
SDHAF2, TMEM127, or VHL, it is essential to develop a
comprehensive mutation screening test.
The advent of massively parallel next generation sequencing (NGS) technologies has provided opportunities
to radically alter strategies for diagnostic genetic testing.
Compared with conventional Sanger sequencing, NGS
provides increased capacity and speed at a significantly
jcem.endojournals.org
E1249
reduced cost (18). The use of molecular identifiers (“barcodes”) allows multiplexing of samples, increasing
throughput and efficiency (19, 20). Therefore, NGS could
facilitate a transition from targeted, sequential analysis of
individual PPGL/HNPGL genes in selected high-risk individuals to a strategy of simultaneously testing multiple
predisposition genes in all at-risk individuals. Here we
describe the successful implementation of a comprehensive 9-gene NGS-based screening strategy for the improved diagnosis of PPGLs/HNPGLs.
Materials and Methods
DNA samples
The study comprised 2 phases during which a total of 205
DNA samples were analyzed. In phase 1, 85 DNA samples containing known variants detected previously by Sanger sequencing were evaluated. In total, 182 variants were interrogated, of
which 84 were unique (59 single base substitutions, 14 deletions,
6 duplications, 3 insertions, and 2 indels).
In phase 2, 120 DNA samples from patients referred with
PPGLs/HNPGLs were prospectively tested. Of these patients, 42
(35%) had previously had 1 or more genes on the panel tested by
Sanger sequencing, and 78 (65%) had had no previous genetic
studies. These patients were referred from centers around the
United Kingdom.
Assay design
The assay included 9 genes known to be associated with a inherited predisposition to PPGL/HNPGL: MAX (NM_002382.3),
RET (NM_020975.4), SDHA (NM_004168.2), SDHAF2
(NM_017841.2), SDHB (NM_003000.2), SDHC (NM_003001.3),
SDHD (NM_003002.2), TMEM127 (NM_017849.3), and VHL
(NM_000551.3). The 48.48 Access Array system (Fluidigm Corporation, San Francisco, California) for amplicon-based target enrichment, was used in combination with the GS Junior NGS sequencer
(454 Life Sciences, a Roche Company, Branford, Connecticut), a 454
pyrosequencing-based technology (21). Primer pairs were designed
using Primer3 (22) to amplify the full coding sequence of all genes,
except for the RET gene for which only exons 10, 11, and 13 to 16
were included. Amplicons ranged in size from 310 to 460 bp and had
melting temperatures of 59 to 61°C, and primers had no more than 3
of the same base in a run, whenever possible. To avoid amplification
of SDHA pseudogenes, primers were designed so that, where possible,
the 3⬘ nucleotide of both primers was specific to SDHA.
Target enrichment, DNA indexing, and sample
pooling
Genomic DNA was amplified using the 48.48 Access Array
system, which facilitates the simultaneous amplification of at
Centre for Rare Diseases and Personalised Medicine (E.R., M.J., E.R.W., E.R.M.), School of Clinical and Experimental Medicine, College of Medical and Dental Sciences, University of
Birmingham, Edgbaston, Birmingham B15 2TH, United Kingdom; West Midlands Regional Genetics Service (E.R., L.V., A.Y., H.B., K.M., T.R.C., E.R.W., F.M., Y.W., E.R.M.), Birmingham
Women’s Hospital NHS Trust, Edgbaston, Birmingham B15 2TG, United Kingdom; Northwest Thames Regional Genetics Service (N.C.), Northwick Park Hospital, Harrow, Middlesex HA1
3UJ, United Kingdom; Department of Endocrinology (J.D., M.K.), William Harvey Research Institute, Bart’s and the London School of Medicine, Queen Mary University of London, London
EC1M 6BQ, United Kingdom; Department of Clinical Genetics (S.V.H.), St George’s Hospital, University of London, London SW17 0RE, United Kingdom; Department of ENT Surgery (R.I.),
University Hospital Birmingham Foundation Trust, Edgbaston, Birmingham B15 2TH, United Kingdom; Department of Clinical Genetics (L.I.), Guy’s and St Thomas’ Foundation Trust, Guy’s
Hospital, London SE1 9RT, United Kingdom; Department of Clinical Genetics (A.V.K.), Great Ormond Street Foundation Trust, Great Ormond Street Hospital, London WC1N 3JN, United
Kingdom; Genetic Medicine (F.L.), Central Manchester Hospitals NHS Foundation Trust, Manchester Academic Health Sciences Centre, Manchester M13 9WL, United Kingdom; and
Department of Medical Genetics (P.J.M.), Queen’s University Belfast, Belfast HSC Trust, Belfast BT9 7AB, United Kingdom
E1250
Rattenberry et al
Molecular Diagnosis of Pheochromocytoma & Paraganglioma
least 48 targets for up to 48 patient DNA samples. The inclusion
of common 5⬘ sequences within each primer enabled the subsequent incorporation of patient-specific barcodes and the 454
sequencing adaptor during a second round of PCR. All processes
were performed according to the manufacturer’s protocol “Multiplex Amplicon Tagging for 454 Titanium Sequencing on the
48.48 Access Array IFC”. The patient samples were then pooled
in equal volumes before sequencing.
J Clin Endocrinol Metab, July 2013, 98(7):E1248–E1256
performed using Alamut (Interactive Biosoftware, Rouen,
France) as an interface and involved (1) interrogation of locusspecific databases, Human Gene Mutation Database, and inhouse databases, (2) a literature review, (3) a search of the dbSNP, National Heart, Lung, and Blood Institute Exome
Sequencing Project Exome Variant Server, and 1000 genomes
project databases, (4) species conservation, and (5) in silico prediction for the impact of missense and splicing variants, as
appropriate.
NGS
NGS was performed using the GS Junior system. Emulsion
PCR (Lib-A kit with 0.5:1 molecule to bead ratio), bead enrichment, and pyrosequencing were performed as per the manufacturer’s instructions.
NGS data analysis
Data analysis was performed using NextGENe version 2.16
sequence analysis software (SoftGenetics, State College, Pennsylvania) and a semiautomated in-house bioinformatics pipeline.
After removal of poor quality reads (median score threshold
⬍20, uncalled bases ⬎2, called base number of each read ⬍100
and/or ⬎2 bases with score of ⬍17) samples were demultiplexed,
and reads were mapped against gene-specific GenBank reference
files. Variants were selected to be called when ⬎15% in ⬎15
total reads for the base being interrogated. A 15% mutation filter
was chosen because a heterozygous variant had been statistically
determined to be identified 99.995% of the time when there was
30-fold coverage (23). Regions with less than 30-fold coverage
were analyzed by Sanger sequencing. (NextGENe was set at the
lower threshold of ⬎15-fold coverage to identify likely variants.)
To improve test specificity, the variants identified for each patient were compared, using the NextGENe variant comparison
tool to classify false-positive results seen in multiple samples
across the same run. The data from the variant comparison tool,
the individual variant reports, and the coverage data were imported into a Microsoft Excel template and analyzed using the
in-house pipeline that flagged patients with regions requiring
Sanger sequencing because of (1) ⬍30-fold coverage, (2) a substitution variant or (3) a called deletion/duplication variant that
was not within 2 SDs of the mean percentage call for that variant
for all patients. Known benign polymorphisms were automatically excluded.
Sanger sequencing
Bidirectional Sanger sequencing of PCR products was performed using the ABI BigDye 3.1 Terminator Cycle Kit (Applied
Biosystems, Foster City, California) and were analyzed using an
ABI Prism 3730 (Applied Biosystems) and Mutation Surveyor
version 3.3 (SoftGenetics) program.
Multiplex ligation-dependent probe amplification
(MLPA)
MLPA was performed using the kits SALSA P226 SDH and
SALSA P016 VHL (MRC Holland, Amsterdam, The Netherlands), and data were analyzed using GeneMarker version 1.70
(SoftGenetics) program.
Variant assessment
Variant pathogenicity was assessed according to previously
published best practice guidelines (24). Much of this analysis was
Results
Optimization
Primers for the Access Array system were tested using
scaled-up volumes of the reactions; 6 amplicons were redesigned to optimize amplification. Despite redesign, RET
exon 14 did not produce an amplicon of sufficient quality,
and it was removed from the Access Array setup and analyzed by Sanger sequencing.
Initial NGS runs were performed to test assay design
and sequencing depth per fragment and to optimize the
number of patients per run to give ⬎30-fold coverage for
all amplicons. Four amplicons producing consistently low
reads were removed from the assay and analyzed by
Sanger sequencing. These amplicons had ⬎66% GC content: MAX exon 1 (67%); SDHA exon 1 (75%); SDHAF2
exon 1 (68%); and TMEM127 exon 2 (72%).
The Access Array system was adjusted to allow the
most uniform NGS coverage possible. Amplicons with
consistently high read numbers were diplexed on the
48.48 Access Array system, and those with low read numbers were amplified in duplicate. During the optimization
and validation phase, a mean number of ⬃82 000 passed
filter reads was achieved. The optimal number of patients
sequenced per run to attain ⬎30-fold coverage for all amplicons was determined to be 20.
NGS assay validation
Sensitivity
The NGS assay was validated using 85 patient DNAs
containing 182 variants (84 unique variants) (see Supplemental Table 1 published on The Endocrine Society’s
Journals Online web site at http://jcem.endojournals.org).
A total of 171 variants met the required minimum coverage threshold set, and of these 170 were detected (76 of 77
unique variants), giving an overall sensitivity of 98.7%.
The single undetected variant (with ⬎15-fold coverage)
was a 6-bp duplication in SDHB intron 4, c.424-19_42414dupTTCTTC. This is a polymorphic tract where the
major allele contains 8 TTC repeats. The duplication allele
was correctly identified in 7 of 49 (14.3%) reads. A number of sequencing/alignment errors had occurred, leading
doi: 10.1210/jc.2013-1319
jcem.endojournals.org
E1251
was close to that expected for all
variant types. However, there were
some outliers; for example, 5 of 26
homozygous substitutions were not
identified in 100% of reads. This was
caused by low-level sequencing
errors.
Prospective patient analysis
Phase 2 involved prospective
analysis of 120 patient samples. The
mean number of passed filter reads
was ⬃108 000 per run higher than
the average of the validation runs
Figure 1. Observed variant frequency during validation.
(⬃82 000) and increasing amplicon
coverage. This was probably the reto the incorrect calling. This polymorphism was se- sult of a Roche software upgrade in combination with
quenced on an additional 5 occasions and detected each increased user experience. The complete genetic workflow
time (mean ⫽ 26% of reads).
(Figure 2) identified 44 variants (excluding benign polymorphisms of no clinical significance), of which 36 were
Specificity
unique: 12 were classed as pathogenic; 3 as probably
False variant calls resulting from errors in determining
pathogenic; and 21 as a variant of undetermined signifithe number of bases within homopolymer tracts is a reccance (VUS) (5 patients had multiple variants). The variognized problem with the 454 sequencing chemistry (25).
ants are summarized in Table 1. An example of a variant
Interrun variability was observed for homopolymer erdetected by NGS and confirmed by Sanger sequencing is
rors, with recurrent errors observed within a run varying
shown in Figure 3.
between runs. During the validation period, numerous
As expected, the detection rate for novel pathogenic
false-positive calls were identified for every DNA sample.
variants was dependent on the extent of previous molecIn one sequencing run, there were 164 unique variants,
ular genetic analysis. One or more genes had been seand each DNA sample had an average of 65 variants
called, 46 of which were probably homopolymer-related quenced previously in 35% of patients, of whom 6 had a
artifacts. To limit the Sanger sequencing confirmations VUS identified (all detected by NGS). Ten additional varirequired and to increase specificity, 2 filtering steps were ants in 7 patients (3 of whom had a previously diagnosed
introduced. (1) The region of interest was limited to the VUS) were identified. All 10 additional variants were idencoding sequence ⫾ 5 bp (with the exception of 5⬘ and 3⬘ tified in genes that had not previously been analyzed, and
untranslated regions). This reduced the number of unique 3 were considered likely to be pathogenic.
No previous genetic analysis had been performed in
variants identified to 27 and only 4 were probable artifacts
associated with homopolymer tracts. These 4 variant calls 65% of patients and a total of 28 variants were identified
comprised 36% of all variants observed. (2) The in-house in 27 individuals. Of these, 46% were considered to be
bioinformatics pipeline was set to filter out homopolymer- pathogenic mutations (10 in SDHB, SDHD, and VHL and
related deletions/duplications that were within 2 SDs of 3 in genes that were not previously part of the routine
the mean for that variant for the patients on that run. It diagnostic service). In addition, 12 VUSs were identified (2
also filtered benign polymorphisms, leaving an average of of which were identified in 1 patient) and 3 likely patho0.5 variant per DNA sample, which required Sanger con- genic changes. In summary, a variant of interest was idenfirmation and pathogenicity analysis. Performing this fil- tified in 34.6% of new referrals and a definite pathogenic
tering on the validation runs did not cause any known mutation in 16.6%.
pathogenic mutations to be missed.
Variant frequency
This is calculated by dividing the number of mutant
reads by the total number of reads. Figure 1 provides details of the variant frequency observed for the 171 variants
detectable by NextGENe. The average variant frequency
Discussion
Although a strong a priori case can be made for genetic
testing to be offered to all patients with PPGL and HNPGL, financial considerations have led to a more restric-
E1252
Rattenberry et al
Molecular Diagnosis of Pheochromocytoma & Paraganglioma
J Clin Endocrinol Metab, July 2013, 98(7):E1248–E1256
NGS assay of 9 genes
Pathogenic mutation
Confirmed by
SS
Not confirmed
by SS
VUS
Confirmed by
SS
No variants
Not confirmed
by SS
SS of non-NGS exons and exons with <30x coverage
Pathogenic mutation
VUS
No variants
MLPA
Pathogenic mutation
VUS
No variants
Report findings
Figure 2. Testing workflow for NGS-based reporting diagnostic patients. SS, Sanger sequencing.
tive genetic testing policy in most centers. The price
charged locally for testing VHL, RET, SDHB, and SDHD
by conventional sequential (Sanger) sequencing technology and MLPA is ⬃£1800 (⬃$2700) (17). Various strategies for prioritizing testing have been suggested, including the presence or absence of clinical criteria and/or
immunohistochemistry for SDHB protein expression (1,
15). We hypothesized that an NGS-based genetic testing
strategy might provide a cost-effective approach to increase detection of individuals with inherited PPGL and
HNPGL. The diagnostic test described here is locally
priced at £500 (⬃$750) per sample to test for 9 PPGL/
HNPGL genes (MAX, RET, SDHA, SDHB, SDHC,
SDHD, SDHAF2, TMEM127, and VHL). However, to
determine whether this strategy was likely to be a feasible
innovation for clinical practice, it is necessary to establish
(1) the possibility of whether NGS could be used as a
reliable alternative to Sanger sequencing to interrogate all
relevant parts of the genes of interest and (2) the sensitivity
and specificity of the NGS test.
The Access Array system was found to be an effective
system to produce an amplicon library for the GS Junior,
and its advantage is that additional amplicons can be included when further PPGL/HNPGL genes are identified.
However, the validation process identified some exons
unsuitable for analysis in this workflow. In particular,
those with a GC content greater than 66% tend to produce
poor results, and this fact should be considered in the
design of assays using this system.
Because this was a novel assay, validation for diagnostic use was essential. Analytical sensitivity was experimentally determined to be 98.7%. Only 1 of 77 unique vari-
ants was not detected during phase 1. However, this
undetected variant was identified on 5 other occasions
(although in a lower average percentage of reads than
might be expected for a heterozygous change). This was
probably the result of sequence context, because the technology appeared to have difficulty sequencing this region.
The variant in question was a benign polymorphism, and
the failure to detect it was not of clinical significance, but
it is important to consider that in other circumstances a
pathogenic mutation on an underrepresented allele may be
overlooked. It is notable that nonhomopolymer repeat
tracts, such as a TTC repeat, can also lead to sequencing
errors when Roche NGS technology is used.
As yet there is no consensus regarding the minimum
coverage and variant filtering threshold (VFT) required to
reliably detect a heterozygous variant. Although de Leeneer et al (23) theoretically calculated that a 15% VFT
with 30-fold coverage would allow detection of 99.995%
of heterozygous variants, this assumes that neither allele is
preferentially sequenced. If this is not the case (and the
mutant allele is sequenced less efficiently), then higher coverage would be required. Several groups have used a similar enrichment strategy and sequencing chemistry and set
locally derived limits to detect heterozygous mutations.
For example, Hollants et al (26) proposed that a VFT of
20% and a minimum coverage of 25-fold was sufficient to
detect heterozygotes in familial hypercholesterolemia
genes, whereas for BRCA1 and BRCA2 screening, Michils
et al (27) suggested that 10% VFT and 27-fold coverage
was required to detect all heterozygous changes. Our findings suggest that a 15% VFT with 30-fold coverage, which
is required to optimize variant detection while attenuating
doi: 10.1210/jc.2013-1319
jcem.endojournals.org
E1253
Table 1. Variants identified/confirmed during prospective analysis
Gene
Exon
Variant Type
Variant (Coding DNA)
Variant (Protein)
Times Observed
Variant Class
Tumor Type
MAX
MAX
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHA
SDHAF2
SDHB
SDHB
SDHB
SDHB
SDHB
SDHB
4
5
1
2
2
5
7
8
8
10
12
13
13
3
1
1
1
2
2
5
Nonsense
Missense
Frameshift
Missense
Nonsense
Synonymous
Synonymous
Synonymous
Missense
Missense
Synonymous
Missense
Synonymous
Missense
Splicing
Splicing
Deletion
Missense
Deletion
Missense
c.[223C⬎T]; [⫽]
c.[425C⬎T]; [⫽]
c.[1338delA]; [⫽]
c.[136A⬎G]; [⫽]
c.[91C⬎T]; [⫽]
c.[549C⬎T]; [⫽]
c.[822C⬎T]; [⫽]
c.[1002G⬎A]; [⫽]
c.[923C⬎T]; [⫽]
c.[1273G⬎A]; [⫽]
c.[1623G⬎A]; [⫽]
c.[1753C⬎T]; [⫽]
c.[1776T⬎C]; [⫽]
c.[319C⬎T]; [⫽]
c.[423⫹1G⬎A]; [⫽]
c.[72⫹1G⬎T]; [⫽]
c.[-?_200⫹?del];[⫽]
c.[118A⬎G]; [⫽]
c.[73-?_200⫹?del];[⫽]
c.[487T⬎C]; [⫽]
p.(Arg75X)
p.(Ser142Leu)
p.(His447Metfs*23)
p.(Lys46Glu)
p.(Arg31X)
p.(⫽)
p.(⫽)
p.(⫽)
p.(Thr308Met)
p.(Val425Met)
p.(⫽)
p.(Arg585Trp)
p.(⫽)
p.(Arg107Cys)
1
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
5
Pathogenic
VUS
VUS
VUS
Likely pathogenic
VUS
VUS
VUS
VUS
VUS
VUS
Likely pathogenic
VUS
VUS
Pathogenic
Pathogenic
Pathogenic
Pathogenic
Pathogenic
VUS
SDHB
SDHB
SDHB
SDHB
SDHC
SDHC
SDHC
SDHC
SDHC
SDHD
6
7
8
2–7
1
2
3
4
5
1
Missense
Missense
Frameshift
Deletion
5’UTR
Splicing
Missense
Missense
Missense
Missense
c.[587G⬎A]; [⫽]
c.[725G⬎A]; [⫽]
c.[770dupT]; [⫽]
c.[73-?_765⫹?del];[⫽]
c.[-118_-117delAG]; [⫽]
c.[77⫹2dupT]; [⫽]
c.[148C⬎T]; [⫽]
c.[214C⬎T]; [⫽]
c.[380A⬎G]; [⫽]
c.[34G⬎A]; [⫽]
p.(Arg50Cys)
p.(Arg72Cys)
p.(His127Arg)
p.(Gly12Ser)
1
1
1
1
1
1
1
1
1
3
Pathogenic
Pathogenic
Likely pathogenic
Pathogenic
VUS
Pathogenic
VUS
VUS
VUS
VUS
SDHD
3
Missense
c.[242C⬎T]; [⫽]
p.(Pro81Leu)
2
Pathogenic
SDHD
TMEM127
TMEM127
TMEM127
TMEM127
3
3
3
3
4
Synonymous
Missense
Frameshift
Synonymous
Synonymous
c.[312C⬎T]; [⫽]
c.[268G⬎A]; [⫽]
c.[512delTinsGCC]; [⫽]
c.[534C⬎T]; [⫽]
c.[411A⬎T]; [⫽]
p.(⫽)
p.(Val90Met)
p.(Val171Glyfs*137)
p.(⫽)
p.(⫽)
1
1
1
1
1
VUS
VUS
Pathogenic
VUS
VUS
PHEO
PHEO
HNPGL
PGL
1. HNPGL 2. PGL
HNPGL
HNPGL
HNPGL
PGL
PHEO/PGL
HNPGL
HNPGL
PGL
PGL
PHEO
PHEO
PGL
PGL
PGL
1. PHEO
2. PHEO
3. PHEO
4. PHEO
5. HNPGL
HNPGL
HNPGL
PHEO
PGL
HNPGL
HNPGL
HNPGL
HNPGL
HNPGL
1. HNPGL
2. PHEO
3. HNPGL
1. HNPGL
2. HNPGL
PGL
PGL
HNPGL
HNPGL
PHEO
p.(0?)
p.(Lys40Glu)
p.?
p.(Ser163Pro)
p.(Cys196Tyr)
p.(Arg242His)
p.(Asn258GlufsX17)
p.?
Abbreviations: HNPGL ⫽ head and neck paraganglioma; PGL ⫽ paraganglioma (abdominal); PHEO ⫽ pheochromocytoma (adrenal); VUS ⫽ variant
of uncertain significance.
the concomitant increase in false-positive results, for this
methodology. In theory, this assay could be applied to the
analysis of tumor samples but would probably require the
analysis of fewer samples so that coverage could be increased to detect clonal variants.
It is well established that homopolymer tract sequencing errors are problematic when the Roche 454 chemistry
is used (25). To reduce the number of variants requiring
Sanger sequencing confirmation, we decreased the regions
analyzed to the coding sequence ⫾ 5 bp of flanking intronic sequence for all genes. In addition, a variant comparison strategy was implemented to simultaneously interrogate all variants in all patient DNAs on a single NGS
run. Any deletion/duplication called was compared with
the data from the same base for all samples sequenced on
that run. Only variants for which the percentage deletion
or duplication call was greater than 2 SDs from the mean
were confirmed by Sanger sequencing. When these filters
were applied to the validation data, all known mutations
were detected, and the number of false-positives calls was
dramatically reduced. Not all NGS chemistries have inherent issues with homopolymer tract sequencing, and
platforms such as the Illumina MiSeq (Illumina, Inc, San
Diego, California) offer some advantages in this respect.
The largest deletion identified during the validation phase
was 7 bp in SDHD exon 1, and the largest duplication was
13 bp in SDHB exon 7. Although whole-exon deletions were
input during the validation phase, the normalized read number did not accurately detect them and thus MLPA was used
as an adjunct to the NGS screen during phase 2.
As reported previously for similar assays (27–29), we
found that the variant frequency was quite variable for
both heterozygous and homozygous changes. This phenomenon may reflect random sampling of the PCR product. For a direct amplicon-based sequencing strategy duplicate reads cannot be identified and removed (as would
E1254
Rattenberry et al
Molecular Diagnosis of Pheochromocytoma & Paraganglioma
J Clin Endocrinol Metab, July 2013, 98(7):E1248–E1256
Figure 3. Example of a pathogenic mutation identified in MAX (c.[223C⬎T];[⫽]). A, Image of the first 40 reads from the NextGENe output of the
NGS data. B, Image of the Mutation Surveyor output for the confirmatory Sanger sequencing.
occur when a method using fragmentation is performed),
which may lead to bias for 1 allele and hence the requirement for a lower VFT. An alternative target enrichment
methodology could be tested in a trial to determine
whether a capture and fragmentation approach would
lead to less allelic bias.
After establishing a diagnostic testing workflow, we
undertook a prospective analysis of samples referred for
doi: 10.1210/jc.2013-1319
diagnostic testing. In addition to the reduced cost of the
NGS test, a further major advantage was that 9 genes were
analyzed simultaneously, whereas conventional sequential molecular genetic testing strategies analyze 1 gene and
then proceed to testing another (if no mutation is detected). However, an inevitable consequence of testing
more genes is the identification of additional VUSs. It was
found that 61% of variants detected were in this class. In
a diagnostic laboratory, accurate variant classification is
essential because it directly affects patient care and family
management. In the short term, the identification of a VUS
can cause uncertainty for the patient and additional work
for the laboratory staff and clinicians. In this study, 51.9%
of the VUSs (but only 15.4% of pathogenic mutations)
were in MAX, SDHA, SDHAF2, or TMEM127, genes not
analyzed previously in the laboratory. These genes are all
relatively newly identified, and, thus, there is a paucity of
literature. We anticipate that as research and diagnostic
testing of these genes expands, more VUSs will be assigned
to pathogenic or benign classes. Nevertheless, we also anticipate that some VUSs will remain difficult to classify and
will require in depth clinical, in silico, and in vitro investigation. Testing multiple genes in parallel can aid VUS
classification such that if a VUS is identified with a known
pathogenic mutation in the same or another gene, then,
unless there is digenic inheritance, the likelihood of the
VUS being pathogenic is reduced (although genetic information should always be interpreted in the context of clinical and pathological findings [including immunohistochemical studies]).
In summary, we developed and validated a novel diagnostic assay that increases the detection of germline mutations in patients with PPGL and HNPGL at a substantially lower cost and reduced processing time than
conventional (Sanger-based) molecular genetic analysis.
In addition to a 70% cost reduction and more genes analyzed, the time taken to report a result to a clinician was
reduced from 160 days for sequential testing of 4 genes to
⬍60 days for 9 genes. In a cohort of 120 prospectively
analyzed patients, we detected 10 pathogenic germline
mutations in the genes (SDHB, SDHD, RET, and VHL)
analyzed in our previous “standard” mutation analysis
strategy and a further 3 mutations by analyzing the less
frequently analyzed PPGL/HNPGL genes, a 33% increase
in diagnostic yield compared with that for our previous
strategy. Longer term, it is predicted that better classification of potential VUSs will further enhance the utility of
this assay and enable the advantages of precise molecular
diagnosis to be offered to larger cohorts of patients.
jcem.endojournals.org
E1255
Acknowledgments
We are grateful to all the referring clinicians, patients, and staff
at the West Midlands Regional Genetics Laboratory.
Address all correspondence and requests for reprints to: Professor E. R. Maher, Department of Medical Genetics, University
of Cambridge, Addenbrooke’s Treatment Centre, Cambridge
Biomedical Campus, Cambridge, CB2 0QQ, United Kingdom.
E-mail: [email protected].
This report is independent research arising from a Doctoral
Research Fellowship (to E.R.) supported by the National Institute for Health Research and the CSO. The views expressed in
this publication are those of the author(s) and not necessarily
those of the NHS, the National Institute for Health Research or
the Department of Health.
This work was supported by VHL Alliance, Medical Research
Council (clinical research fellowship to M.J.), and British Heart
Foundation.
Disclosure Summary: The authors have nothing to disclose.
References
1. Erlic Z, Rybicki L, Peczkowska M, et al. Clinical predictors and
algorithm for the genetic diagnosis of pheochromocytoma patients.
Clin Cancer Res. 2009;15:6378 – 6385.
2. Jafri M, Maher ER. The genetics of phaeochromocytoma: using
clinical features to guide genetic testing. Eur J Endocrinol. 2012;
166:151–158.
3. Gimenez-Roqueplo AP, Lehnert H, Mannelli M, et al. Phaeochromocytoma, new genes and screening strategies. Clin Endocrinol
(Oxf). 2006;65:699 –705.
4. Neumann HP, Bausch B, McWhinney SR, et al. Germ-line mutations in nonsyndromic pheochromocytoma. N Engl J Med. 2002;
346:1459 –1466.
5. Gimenez-Roqueplo AP, Dahia PL, Robledo M. An update on the
genetics of paraganglioma, pheochromocytoma, and associated hereditary syndromes. Horm Metab Res. 2012;44:328 –333.
6. Bravo EL. Evolving concepts in the pathophysiology, diagnosis, and
treatment of pheochromocytoma. Endocr Rev. 1994;15:356 –368.
7. Baysal BE, Ferrell RE, Willett-Brozick JE, et al. Mutations in SDHD,
a mitochondrial complex II gene, in hereditary paraganglioma. Science. 2000;287:848 – 851.
8. Astuti D, Latif F, Dallol A, et al. Gene mutations in the succinate
dehydrogenase subunit SDHB cause susceptibility to familial pheochromocytoma and to familial paraganglioma. Am J Hum Genet.
2001;69:49 –54.
9. Astuti D, Douglas F, Lennard TW, et al. Germline SDHD mutation
in familial phaeochromocytoma. Lancet. 2001;357:1181–1182.
10. Comino-Méndez I, Gracia-Aznárez FJ, Schiavi F, et al. Exome sequencing identifies MAX mutations as a cause of hereditary pheochromocytoma. Nat Genet. 2011;43:663– 667.
11. Burnichon N, Brière JJ, Libé R, et al. SDHA is a tumor suppressor
gene causing paraganglioma. Hum Mol Genet. 2010;19:3011–
3020.
12. Niemann S, Muller U. Mutations in SDHC cause autosomal dominant paraganglioma, type 3. Nat Genet. 2000;26:268 –270.
13. Bayley JP, Kunst HP, Cascon A, et al. SDHAF2 mutations in familial
and sporadic paraganglioma and phaeochromocytoma. Lancet Oncol. 2010;11:366 –372.
14. Qin Y, Yao L, King EE, et al. Germline mutations in TMEM127
confer susceptibility to pheochromocytoma. Nat Genet. 2010;42:
229 –233.
E1256
Rattenberry et al
Molecular Diagnosis of Pheochromocytoma & Paraganglioma
15. Karasek D, Frysak Z, Pacak K. Genetic testing for pheochromocytoma. Curr Hypertens Rep. 2010;12:456 – 464.
16. Neumann HP, Erlic Z, Boedeker CC, et al. Clinical predictors for
germline mutations in head and neck paraganglioma patients: cost
reduction strategy in genetic diagnostic process as fall-out. Cancer
Res. 2009;69:3650 –3656.
17. Jafri M, Whitworth J, Rattenberry E, et al. Evaluation of SDHB,
SDHD and VHL gene susceptibility testing in the assessment of
individuals with non-syndromic phaeochromocytoma and paraganglioma and head and neck paraganglioma (HNPGL). Clin Endocrinol (Oxf). 2013;78:898 –906.
18. Meldrum C, Doyle MA, Tothill RW. Next-generation sequencing
for cancer diagnostics: a practical perspective. Clin Biochem Rev.
2011;32:177–195.
19. Morgan JE, Carr IM, Sheridan E, et al. Genetic diagnosis of familial
breast cancer using clonal sequencing. Hum Mutat. 2010;31:484 –
491.
20. O’Sullivan J, Mullaney BG, Bhaskar SS, et al. A paradigm shift in the
delivery of services for diagnosis of inherited retinal disease. J Med
Genet. 2012;49:322–326.
21. Margulies M, Egholm M, Altman WE, et al. Genome sequencing in
microfabricated high-density picolitre reactors. Nature. 2005;437:
376 –380.
22. Rozen S, Skaletsky H. Primer3 on the WWW for general users and
for biologist programmers. Methods Mol Biol. 2000;132:365–386.
J Clin Endocrinol Metab, July 2013, 98(7):E1248–E1256
23. de Leeneer K, De Schrijver J, Clement L, et al. Practical tools to
implement massive parallel pyrosequencing of PCR products in next
generation molecular diagnostics. PLoS One. 2011;6:e25531.
24. Bell JA, Bodmer D, Sistermans E, Ramsden SC. Practice guidelines for the
interpretation and reporting of unclassified variants (UVs) in clinical molecular genetics. Clinical Molecular Genetics Society, October 22,
2007. http://www.cmgs.org/BPGs/pdfs%20current%20bpgs/UV%20
GUIDELINES%20ratified.pdf. Published October 10, 2007. Accessed
February 5, 2013.
25. Huse SM, Huber JA, Morrison HG, Sogin ML, Welch DM. Accuracy and quality of massively parallel DNA pyrosequencing. Genome Biol. 2007;8:R143.
26. Hollants S, Redeker EJ, Matthijs G. Microfluidic amplification as a
tool for massive parallel sequencing of the familial hypercholesterolemia genes. Clin Chem. 2012;58:717–724.
27. Michils G, Hollants S, Dehaspe L, et al. Molecular analysis of the
breast cancer genes BRCA1 and BRCA2 using amplicon-based massive parallel pyrosequencing. J Mol Diagn. 2012;14:623– 630.
28. Bonnal RJ, Severgnini M, Castaldi A, et al. Reliable resequencing of
the human dystrophin locus by universal long polymerase chain
reaction and massive pyrosequencing. Anal Biochem. 2010;406:
176 –184.
29. de Leeneer K, Hellemans J, De Schrijver J, et al. Massive parallel
amplicon sequencing of the breast cancer genes BRCA1 and BRCA2:
opportunities, challenges, and limitations. Hum Mutat. 2011;
32:335–344.
Members can search for endocrinology conferences,
meetings and webinars on the Worldwide Events Calendar.
www.endo-society.org/calendar