Assessment of Mutant Homozygosity in Gastrointestinal
Stromal Tumors
Michelle Wallander1, Carlynn Willmore-Payne1 and Lester Layfield1,2
Institute for Clinical and Experimental Pathology, ARUP Laboratories, Salt Lake City, UT
2Department
of Pathology, University of Utah, Salt Lake City, UT
Introduction
Results
C
G
T
A
30
Homoduplexes
Heterozygote
Melting Curve
T
G
C
Mutant GISTs (%)
Mutant zygosity has not been extensively studied in gastrointestinal
stromal tumors (GISTs), which are characterized by activating
mutations in the receptor tyrosine kinases C-KIT or PDGFRA. We
reviewed a series of mutant GISTs, which were previously analyzed
by high resolution melting analysis (HRMA) and DNA sequencing.
The percentage of mutant homozygosity in C-KIT exons 9, 11, 13,
17 and PDGFRA exons 12 and 18 was determined. Mutant
homozygosity was also correlated with tumor location, size and CKIT (CD117) immunohistochemistry.
Fluorescence
Activated kinases such as EGFR, BRAF and C-KIT are associated
with multiple human malignancies. Gain-of-function deletions,
insertions or point mutations cause constitutive kinase activation,
which stimulates cell proliferation and survival. Given the presumed
dominance of the mutation, it is not surprising that the majority of
activating mutations are heterozygous. However, homozygous CKIT and BRAF mutations have been reported in melanoma. The
zygosity of kinase activating mutations may have important
implications for oncogenesis or treatment response.
100
25
80
60
40
20
0
Heteroduplexes
C-KIT
exon 11
A
C-KIT
exon 9
C-KIT
C-KIT PDGFRA PDGFRA
exon 13 exon 17 exon 18 exon 12
Figure 1. High resolution melting analysis (HRMA) of heterozygous and homozygous DNA.
The heterozygote melting curve is a combination of all four possible duplexes. The base pair
mismatches in the heteroduplexes are unstable, resulting in a lower Tm that is easily
detectable. Distinguishing between the two homoduplexes can be difficult since the Tm is
dependent on the base pair change and nearest neighbor thermodynamics.
Figure 4. Correlation of GIST mutant zygosity with mutation location. Distribution of
heterozygous GIST mutations (blue) and homozygous GIST mutations (red) in all six tested
exons. Homozygous mutations were detected in exon 11 (n = 36), exon 13 (n = 1), exon 18
(n = 3) and exon 12 (n = 2).
5
Homozygotes
Figure 7. Correlation of GIST size with mutant heterozygosity and homozygosity. The mean
heterozygous tumor size (when specified) was 7.84  6.73 cm (n = 34). The mean
homozygous tumor size (when specified) was 11.52  9.63 cm (n = 16). P-Value = 0.181.
Outlier samples = *.
 The majority of GIST mutant homozygosity occurred in C-KIT exon
11 (86%), which harbors the most GIST mutations (Figure 4). No
homozygous mutations were detected in C-KIT exons 9 or 17.
 There was no correlation between CD117 IHC and GIST mutant
homozygosity (Figure 5).
80
Fluorescence
Mutant GISTs (%)
70
 Homozygous GIST mutations were more prevalent than
heterozygous GIST mutations, as a percentage, in metastatic sites
like the liver and pancreas (Figure 6).
60
50
40
30
 There was no significant correlation between tumor size and
mutant homozygosity (Figure 7).
20
10
0
CD117
positive
Temperature
Figure 2. HRMA of a C-KIT K642E heterozygous sample (blue), a K642E homozygous sample
with limited “contaminating” wild-type DNA (red) and a wild-type control (black).
CD117 weak
or focal
CD117
negative
CD117 not
specified
Figure 5. Correlation of CD117 (C-KIT) IHC with GIST mutant heterozygosity (blue) and
homozygosity (red).
Mutant GISTs
(% Each Tumor Site)
# of Mutant GISTs
80
200
150
100
70
60
50
40
30
20
50
Discussion
Our results in 267 mutant GISTs suggests that the majority are
heterozygous. However, homozygous mutations are more difficult
to detect by HRMA and may therefore be slightly underrepresented.
To improve homozygous mutant detection, known mutant genotypes
could be mixed with the sample post-PCR to facilitate heteroduplex
formation.
Homozygous GIST mutants would be predicted to contain 100%
mutant C-KIT or PDGFRA homodimer. This would increase KIT
signaling by four-fold as compared to a heterozygous GIST mutant,
which may result in a more aggressive tumor. We had limited data
on tumor size and no data on patient outcome. We did see a
greater frequency of homozygous mutations located in metastatic
body sites, suggesting that tumors with homozygous GIST
mutations may be more aggressive. Patient follow-up data will be
critical for determining the prognostic value of a GIST homozygous
mutation.
100
250
10
0
References
Figure 3. GIST mutant homozygosity. Forty-two (16%) GISTs demonstrated mutant gene
homozygosity.
ifi
ed
tic
ot
s
Willmore-Payne C, Holden JA, Hirschowitz S, Layfield LJ. BRAF and c-kit gene copy number in
mutation-positive malignant melanoma. Human Pathology, 2006; 37:520-527.
N
et
M
pe
c
as
ta
ol
on
C
tr
a-
G
e
tin
Sm
al
St
li
nt
es
om
 Forty-two (16%) GISTs harbored homozygous mutations as
determined by sequencing (Figure 3).
Ex
Homozygous
h
Heterozygous
I
0
ac
 Homozygous mutant samples with limited “contaminating” normal
DNA can be difficult to distinguish from the wild-type control (Figure
2).
10
Heterozygotes
90
 High resolution melting analysis (depicted in Figure 1) detected
267 mutant GISTs.
15
Temperature
A series of 267 GISTs with known mutations in C-KIT or PDGFRA
were reviewed. Mutational analysis was originally performed at
ARUP Laboratories. Briefly, formalin-fixed paraffin embedded
tumor tissue was microdissected and DNA was extracted by
proteinase K digestion. Crude DNA extract was used directly in
PCR to amplify C-KIT exons 9, 11, 13 and 17 and PDGFRA exons
12 and 18.
After amplification on the LightCycler (Roche
Diagnostics, Indianapolis, IN), samples were briefly heated to 95°C
and then cooled to 40°C. Samples were transferred to the HR-1
high-resolution DNA melting analysis instrument (Idaho Technology,
Salt Lake City, UT) and heated at 0.3°C/sec to generate melting
curves. Normalized and temperature-shifted melting curves were
compared to wildtype DNA, which was used as a negative control.
Results
20
0
Materials and Methods
All samples with an abnormal melting curve were sequenced.
Electropherogram peak height and the percentage of tumor in the
sample were used to classify mutations as heterozygous or
homozygous.
Tumor
location,
size
and
CD117
immunohistochemistry (IHC) were obtained from the surgical
pathology files at the University of Utah. The use of human tissue
for this analysis was approved by the University of Utah Institutional
Review Board (#11903).
GIST Tumor Size (cm)
1ARUP
Figure 6. Correlation of GIST location with mutant heterozygosity (blue) and homozygosity
(red). Extra-GI includes the omentum, soft tissue, mesentary and peritoneum. Metastatic sites
include the liver and pancreas.
Holden JA, Willmore-Payne C, Coppola D, Garrett CR, Layfield LJ. High-resolution melting
amplicon analysis as a method to detect c-kit and platelet-derived growth factor receptor alpha
activating mutations in gastrointestinal stromal tumors. American Journal of Clinical Pathology,
2007; 128:230-238.