KIT-Negative Gastrointestinal Stromal Tumors
Proof of Concept and Therapeutic Implications
Fabiola Medeiros, MD,* Christopher L. Corless, MD,† Anette Duensing, MD,*
Jason L. Hornick, MD, PhD,* Andre M. Oliveira, MD,* Michael C. Heinrich, MD,‡
Jonathan A. Fletcher, MD,*§ and Christopher D. M. Fletcher, MD, FRCPath*
Abstract: The diagnosis of gastrointestinal stromal tumor (GIST) is
currently based on morphologic features and immunohistochemical
demonstration of KIT (CD117). However, some tumors (in our estimation
approximately 4%) have clinicopathologic features of GIST
but do not express KIT. To determine if these lesions are truly GISTs,
we evaluated 25 tumors with clinical and histologic features typical of
GIST, but with negative KIT immunohistochemistry, for KIT and
PDGFRA mutations using DNA extracted from paraffin-embedded
tissue. Most tumors originated in the stomach (N = 14) or
omentum/mesentery (N = 5). The neoplasms were composed of epithelioid
cells (13 cases), admixed epithelioid and spindle cells (8
cases), or spindle cells (4 cases). Absence of KIT expression was confirmed
by immunoblotting in 5 cases. Tumor karyotypes performed in
4 cases were noncomplex with monosomy 14 or 14q deletion, typical
of GIST. Mutational analysis revealed PDGFRA and KIT mutations
in 18 and 4 tumors, respectively, whereas 3 tumors did not have apparent
KIT or PDGFRA mutations. The PDGFRA mutations primarily
involved exon 18 (N = 15) and included 11 tumors with missense
mutation in codon 842 (PDGFRA D842V or D842Y). In conclusion,
a small subset of GISTs with otherwise typical clinicopathologic and
cytogenetic features do not express detectable KIT protein. When
compared with KIT-positive GISTs, these KIT-negative GISTs are
more likely to have epithelioid cell morphology, contain PDGFRA
oncogenic mutations, and arise in the omentum/peritoneal surface.
Notably, some KIT-negative GISTs contain imatinib-sensitive KIT or
PDGFRA mutations; therefore, patients with KIT-negative GISTs
should not, a priori, be denied imatinib therapy.
Key Words: gastrointestinal stromal tumor, KIT, PDGFRA
(Am J Surg Pathol 2004;28:889–894)
From the *Department of Pathology, Brigham and Women’s Hospital & Harvard
Medical School, Boston, MA; Departments of †Pathology and
‡Medicine, Oregon Health & Science University Cancer Institute and
Portland VA Medical Center, Portland, OR; and §Departments of Pediatric
and Medical Oncology, Dana-Farber Cancer Institute, Boston, MA.
Drs. Medeiros and Corless contributed equally to this publication.
The study was funded in part by a VA Merit Review Grant to Dr. Heinrich.
Reprints: Christopher D. M. Fletcher, MD, FRCPath, Department of Pathology,
Brigham and Women’s Hospital, 75 Francis Street, Boston, MA
02115 (e-mail: email@example.com).
Copyright © 2004 by Lippincott Williams & Wilkins
Gastrointestinal stromal tumor (GIST) is the most common
mesenchymal tumor of the digestive tract and shows lineage
differentiation similar to the interstitial cell of Cajal. 13
Until relatively recently, most GISTs were classified as visceral
leiomyoma or leiomyosarcoma, reflecting the histologic
similarities between these two types of neoplasms. However,
in most cases, GISTs are now distinguished readily from true
smooth muscle tumors based on results of immunohistochemical
staining for KIT (CD117 antigen) and desmin. 5 Most
GISTs have oncogenic KIT mutations that engender constitutive
activation of this receptor tyrosine kinase, resulting in increased
cell proliferation and survival, 17 and such mutations
appear to be early (and even initiating) events in the pathogenesis
of many GISTs.
The central and essential role of KIT in GIST pathogenesis
is underscored by the therapeutic success of the KIT inhibitor
imatinib (Gleevec in the United States, Glivec in the
rest of the world; Novartis, Basel, Switzerland). 4,21 Most patients
with metastatic GIST show a major clinical response after
treatment with single-agent imatinib. 4 KIT positivity by
immunohistochemistry has been considered the gold standard
for GIST diagnosis and, at least for the first clinical trials, this
has been an eligibility criterion for imatinib therapy. 5 However,
in our experience, a subset of tumors that are typical for
GIST both clinically and histologically lack apparent KIT expression,
having low to undetectable KIT protein expression
by immunohistochemistry and Western blot evaluations.
Whether these tumors are indeed GISTs and whether they may
respond to imatinib therapy have been controversial. Recently,
we identified platelet-derived growth factor receptor alpha
(PDGFRA) mutations as an alternative oncogenic mechanism
in a small group of GISTs lacking KIT mutations 9 and this has
been confirmed by others. 11 Because imatinib can also bind
and inhibit PDGFRA, 9,16 responses to this drug have been observed
in some, but not all, patients with PDGFRA-mutant tumors.
8 In this study, we characterize a series of 25 KITnegative
GISTs, which reveal an interesting spectrum of
pathologic and molecular features of potential clinical and
Am J Surg Pathol • Volume 28, Number 7, July 2004 889
Medeiros et al Am J Surg Pathol • Volume 28, Number 7, July 2004
MATERIALS AND METHODS
The study group included 25 tumors that were typical of
GIST clinically and histologically, but were KIT negative by
immunohistochemistry in formalin-fixed, paraffin-embedded
tissue. Among 495 consecutive GISTs diagnosed between
1999 and 2002 at the Brigham and Women’s Hospital, 20 tumors
(4%) were KIT negative by immunohistochemistry. The
remaining 5 tumors were obtained from consultation files of
the Department of Pathology at Oregon Health & Science University.
Five of the 25 cases were included in a prior study. 9
Only surgical excision specimens were included to minimize
nonrepresentative sampling. Patients previously treated with
imatinib mesylate were excluded from the study to avoid treatment
effect as a cause of KIT negativity.
Clinical data, including age, gender, and tumor location,
were obtained in all cases. An average of five hematoxylineosin
archival slides per case was available for histopathologic
review. Paraffin-embedded, formalin-fixed tissue was used for
immunohistochemical analysis. Tissue sections were cut at 4
µm and incubated with the primary antibodies for 40 minutes at
room temperature. Immunohistochemistry was performed for
KIT (Dako Corporation, Carpinteria, CA; polyclonal A4502,
1:250) in all cases without epitope retrieval, as previously described.
12 Selected cases were also stained for CD34 (Dako
Corporation; clone Qbend10, 1:400), desmin (Dako Corporation;
D33, 1:500), S-100 (Dako Corporation; polyclonal,
1:3000), -smooth-muscle actin (Sigma, St. Louis, MO; clone
1A4, 1:20,000), and keratins (Dako Corporation; AE1/AE3;
1:200) with the Envison+ avidin-biotin peroxidase kit (Dako
Corporation) according to manufacturer’s specifications. KITpositive
GISTs were used as positive controls. Negative controls
consisted of substituting normal serum for the primary
antibody, which resulted in no staining of the tissues.
Cytogenetic analysis was performed in 4 cases according
to standard procedures. 7 Immunoblotting was performed
using total cell lysates from snap-frozen GIST specimens, as
described elsewhere. 8,17 Mutational analyses were performed
on DNA extracted from paraffin-embedded tumor tissue using
a combination of PCR amplification, denaturing high performance
liquid chromatography screening, and automated sequencing,
as described previously. 2,9,17
There were 17 males and 8 females (2:1). The median
age at diagnosis was 56 years (range 29–79 years). Most tumors
originated in the stomach (N = 14, 56%), followed by
omentum/mesentery (N = 5) and small bowel (N = 1). In 5
cases (20%), there was an intraabdominal mass, but no distinct
primary site could be identified at the time of presentation,
suggesting possible peritoneal origin. Tumor size ranged from
4 to 38 cm (median, 8.5 cm) (Table 1).
All tumors exhibited classic histologic features for
GIST, being composed of cellular sheets, fascicles, or nests of
cells that lacked significant nuclear pleomorphism. Tumor cell
cytoplasm was eosinophilic and slightly fibrillary with illdefined
cytoplasmic borders, producing a somewhat syncytial
appearance. Nuclei were spindled, ovoid, or rounded and had a
uniform appearance with evenly distributed chromatin. The
majority of cases showed epithelioid cell morphology (N = 13,
52%) (Fig. 1). Eight tumors were of mixed (epithelioid and
spindle) cell type (32%) and four were composed of spindle
cells only (16%). Mitoses ranged from 1 to 52 per 50 high
power fields (median 8) (Table 1). All cases were classified as
intermediate or high risk for aggressive behavior based on tumor
size and mitotic count. 5 All tumors completely lacked KIT
staining by immunohistochemistry (Fig. 1). Complementary
immunostains were performed in 23 cases to exclude other tumors
in the differential diagnosis. Eleven and 10 cases were
positive for CD34 and smooth muscle actin, respectively. Focal
S-100 protein positivity was detected in one tumor. All
cases that were evaluated for desmin and keratin were negative
for these markers.
Cytogenetic analysis revealed noncomplex karyotypes
and a typical loss of chromosome 14 in all four cases evaluated.
One tumor also showed deletion of chromosome 22
(Table 2). Mutational analysis revealed KIT mutations in 4
cases (16%) and PDGFRA mutations in 18 cases (72%) (Table
1). In only 3 cases, no KIT or PDGFRA mutations were identified.
Most PDGFRA mutations involved exon 18 (15 of 18,
83%), including 11 GISTs with missense mutations leading
to a substitution of valine (N = 9) or tyrosine (N = 2) for aspartic
acid 842 (D842V and D842Y, respectively). The D842Y
mutation is a novel mutation, whereas the D842V mutation
has been reported previously. 9 The remaining 4 GISTs with
PDGFRA exon 18 mutations had in-frame deletions. Two
GISTs had PDGFRA mutations in exon 12 (N = 2), encoding
the PDGFRA juxtamembrane region, and one GIST had a previously
undescribed point mutation in PDGFRA exon 14
(N659K). Four GISTs had KIT mutations involving either
exon 11 (N = 3) or exon 9 (N = 1) (Table 1). No tumor had
mutations of both PDGFRA and KIT or more than one mutation
in either of these genes. Immunoblotting, performed in 3
cases with available snap-frozen tissue, confirmed absence of
KIT protein expression (Fig. 2). Two of these GISTs expressed
phosphorylated and total PDGFRA strongly and had PDGFRA
oncogenic mutations (Fig. 2; Table 1). The third case expressed
neither KIT nor PDGFRA but had a KIT exon 11 mutation
(Fig. 2; Table 1).
GIST is a mesenchymal neoplasm that exhibits morphologic
and immunophenotypic features similar to the interstitial
cells of Cajal, which are pacemaker cells regulating gastrointestinal
peristalsis. 13 A characteristic feature of GISTs, similar
to the Cajal cells, is expression of the protein tyrosine kinase
KIT, which is readily detected by immunohistochemistry and
890 © 2004 Lippincott Williams & Wilkins
Am J Surg Pathol • Volume 28, Number 7, July 2004
TABLE 1. Clinicopathologic Features and Mutational Analysis in 25 KIT-Negative GISTs
50 HPF KIT Mutations PDGFRA Mutations
1 76/M Small bowel 15 Epithelioid 20 WT exon 9, 11, 13, 17 Exon 18 delDIMH842-845
2 70/F Stomach 8 Mixed
Medeiros et al Am J Surg Pathol • Volume 28, Number 7, July 2004
FIGURE 1. Epithelioid GIST (A) with negative KIT immunohistochemistry
(B). Note the positive mast cells.
myosarcoma, melanoma, schwannoma, and carcinoma. Notably,
the cytogenetic profiles obtained in four tumors were classic
for GIST. Each of the karyotypes was noncomplex and
featured loss of material from the long arm of chromosome 14,
which is the most frequent cytogenetic aberration in GIST. 10
TABLE 2. Cytogenetic Profiles of Four KIT-Negative GISTs
FIGURE 2. Expression of phosphorylated and total receptor
tyrosine kinases in KIT-positive and KIT-negative GISTs. KITpositive
GISTs, both with KIT exon 11 mutations, are in lanes
1 to 2. KIT-negative GISTs include two with PDGFRA mutations
(lanes 3–4) and one with KIT exon 11 mutation (lane 5). Phosphorylated
and total KIT are expressed strongly in the KITpositive/KIT-mutant
control GISTs, whereas phosphorylated
and total PDGFRA are expressed strongly in the KITnegative/PDGFRA-mutant
GISTs (case nos. 1 and 23). KIT and
PDGFRA are not demonstrably expressed in the KITnegative/KIT-mutant
GIST (case no. 4). The PI3-K stain provides
evidence for approximate equivalency in loading of intracellular
By contrast, leiomyosarcomas, and particularly those of higher
histologic grade, typically have extremely complex karyotypes.
Most GISTs are characterized by oncogenic mutations
of KIT and constitutive activation of the KIT receptor tyrosine
kinase, which seems to drive tumor formation. However, a
small subset of GISTs has been identified that lacks detectable
KIT mutations. 17,20 Recently, activating mutations in the related
receptor tyrosine kinase PDGFRA were reported in 35%
to 67% of GISTs lacking KIT mutations. 9,11 We evaluated
PDGFRA and KIT mutations in the 25 cases in our series and
found that the majority exhibited mutations of the PDGFRA
gene, most frequently involving exon 18. All pure epithelioid
tumors in this series harbored PDGFRA mutations. However,
we also identified four tumors with KIT mutations despite the
892 © 2004 Lippincott Williams & Wilkins
Am J Surg Pathol • Volume 28, Number 7, July 2004
consistent negativity of these tumors for KIT by both immunohistochemistry
and immunoblotting. No tumors had mutations
involving KIT and PDGFRA concomitantly, corroborating
previous evidence that these are mutually exclusive transforming
equivalents in GIST oncogenesis. 9 Three tumors
otherwise indistinguishable from the others in this study had
no detectable KIT or PDGFRA mutations, suggesting that
some KIT-negative GISTs arise through alternative oncogenic
The biologic basis for the absence of KIT expression, in
the subgroup of GISTs identified in our study, is unclear. Most
of the tumors contained PDGFRA mutations, and as discussed
above, such mutations appear to be alternate, or “either/or”
transforming mechanisms, compared with the more common
KIT oncogenic mutations in GISTs. In PDGFRA-mutant tumors,
KIT expression may simply be superfluous and therefore
down-regulated, as mutant PDGFRA appears to act as an
oncogenic substitute for KIT in these cases. It would have been
of interest, if feasible, to assess the PDGFRA-mutant cases immunohistochemically
for evidence of PDGFRA protein expression.
However, multiple trials using a variety of currently
available antibodies (in both our Boston and Portland laboratories)
have failed to identify any commercial antibody that
yields meaningful and reproducible results with a clean background.
Thus, we do not think that these presently available
antibodies have clinical utility. More puzzling are the four tumors
containing KIT gene mutations but in which KIT protein
expression was not demonstrated. It is counterintuitive that the
neoplastic GIST cells would select for a genomic KIT mutation
in the absence of the functional protein product of that gene.
One possibility is that these tumors originated through a KITactivated
pathway and then later became independent as a result
of additional secondary mutations. 6 Another possibility is
that these discordant genomic and protein findings could be
related to technical factors. For example, deletion of the KIT
C-terminus, containing the epitope against which the DAKO
KIT antibody was raised, could lead to false-negative immunostaining
results. On the other hand, C-terminal truncation
cannot account for the negative KIT immunohistochemistry in
all KIT-negative GISTs, because we corroborated absence of
KIT protein expression, using antibodies to a kinase domain
epitope, by immunoblotting in three PDGFRA-mutant or KITmutant
cases with available frozen tissue. Finally, it is conceivable
that routine immunohistochemistry is insufficiently sensitive
to detect the lowest biologically relevant levels of KIT
expression in all tumors. We recently reported a case of a KITmutant
GIST in which KIT expression was virtually absent,
excluding the patient from a clinical trial. The tumor became
PET negative and showed shrinkage on CT scan when the patient
was treated with imatinib off protocol. 1 Based on this anecdotal
example, it is possible that other KIT-mutant GISTs
that lack KIT expression, such as the 4 cases in the present
series, are imatinib sensitive.
In summary, we report that a small subset of GISTs lack
KIT expression but have otherwise typical clinical, histopathologic,
and cytogenetic features. Most of these tumors have at
least partly or totally epithelioid cytomorphology and harbor
PDGFRA mutations, but some examples have KIT mutations
despite the absence of demonstrable KIT protein expression.
Given that most KIT-negative GISTs contain PDGFRA or KIT
mutations, one might expect that patients with such tumors
could potentially respond to therapeutic PDGFRA and KIT inhibition
with imatinib. 8 Even acknowledging that the commonest
PDGFRA D842V mutation is intrinsically imatinib resistant,
8,11 approximately 30% of the PDGFRA mutations
identified are known to be potentially imatinib sensitive, 8 providing
a greater hope of treatment response than with any other
currently approved therapy. Therefore, it is important for pathologists
and oncologists to be aware that, in the context of
otherwise typical morphology, a GIST diagnosis should not be
precluded on the basis of negative immunohistochemical
staining for KIT and that such patients should not a priori be
denied imatinib therapy. Where testing is available, the finding
of KIT or PDGFRA mutations not only may help to confirm or
enable the diagnosis of a KIT-negative GIST, but can also provide
important prognostic information for patients in whom
imatinib therapy is being considered. 8
The authors thank Troy Bainridge, Laura McGreevey,
Andrea Haley, Ajia Town, Maureen Thyne, and Catherine
Quigley for excellent technical assistance.
1. Bauer S, Corless CL, Heinrich MC, et al. Response to imatinib mesylate
of a gastrointestinal stromal tumor with very low expression of KIT. Cancer
Chemother Pharmacol. 2003;51:261–265.
2. Corless CL, McGreevey L, Haley A, et al. KIT mutations are common in
incidental gastrointestinal stromal tumors one centimeter or less in size.
Am J Pathol. 2002;160:1567–1572.
3. DeMatteo RP, Lewis JJ, Leung D, et al. Two hundred gastrointestinal
stromal tumors: recurrence patterns and prognostic factors for survival.
Ann Surg. 2000;231:51–58.
4. Demetri GD, von Mehren M, Blanke CD, et al. Efficacy and safety of
imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J
5. Fletcher CD, Berman JJ, Corless C, et al. Diagnosis of gastrointestinal
stromal tumors: a consensus approach. Hum Pathol. 2002;33:459–465.
6. Fletcher JA, Corless C, Dimitrijevic S, et al. Mechanisms of resistance to
imatinib mesylate (IM) in advanced gastrointestinal stromal tumor
(GIST) [Abstract]. Proc Am Soc Clin Oncol. 2003;22:815.
7. Fletcher JA, Kozakewich HP, Hoffer FA, et al. Diagnostic relevance of
clonal cytogenetic aberrations in malignant soft-tissue tumors. N Engl J
8. Heinrich MC, Corless CL, Demetri GD, et al. Kinase mutations and imatinib
response in patients with metastatic gastrointestinal stromal tumor. J
Clin Oncol. 2003;21:4342–4349.
9. Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating mutations
in gastrointestinal stromal tumors. Science. 2003;299:708–710.
10. Heinrich MC, Rubin BP, Longley BJ, et al. Biology and genetic aspects of
gastrointestinal stromal tumors: KIT activation and cytogenetic alterations.
Hum Pathol. 2002;33:484–495.
11. Hirota S, Ohashi A, Nishida T, et al. Gain-of-function mutations of plate-
© 2004 Lippincott Williams & Wilkins 893
Medeiros et al Am J Surg Pathol • Volume 28, Number 7, July 2004
let-derived growth factor receptor alpha gene in gastrointestinal stromal
tumors. Gastroenterology. 2003;125:660–667.
12. Hornick JL, Fletcher CD. Immunohistochemical staining for KIT
(CD117) in soft tissue sarcomas is very limited in distribution. Am J Clin
13. Huizinga JD, Thuneberg L, Kluppel M, et al. W/kit gene required for
interstitial cells of Cajal and for intestinal pacemaker activity. Nature.
14. Miettinen M, Lasota J. Gastrointestinal stromal tumors: definition, clinical,
histological, immunohistochemical, and molecular genetic features
and differential diagnosis. Virchows Arch. 2001;438:1–12.
15. Reith JD, Goldblum JR, Lyles RH, et al. Extragastrointestinal (soft tissue)
stromal tumors: an analysis of 48 cases with emphasis on histologic predictors
of outcome. Mod Pathol. 2000;13:577–585.
16. Rubin BP, Schuetze SM, Eary JF, et al. Molecular targeting of plateletderived
growth factor B by imatinib mesylate in a patient with metastatic
dermatofibrosarcoma protuberans. J Clin Oncol. 2002;20:3586–3591.
17. Rubin BP, Singer S, Tsao C, et al. KIT activation is a ubiquitous feature of
gastrointestinal stromal tumors. Cancer Res. 2001;61:8118–8121.
18. Sarlomo-Rikala M, Kovatich AJ, Barusevicius A, et al. CD117: a sensitive
marker for gastrointestinal stromal tumors that is more specific than
CD34. Mod Pathol. 1998;11:728–734.
19. Singer S, Rubin BP, Lux ML, et al. Prognostic value of KIT mutation type,
mitotic activity, and histologic subtype in gastrointestinal stromal tumors.
J Clin Oncol. 2002;20:3898–3905.
20. Taniguchi M, Nishida T, Hirota S, et al. Effect of c-kit mutation on prognosis
of gastrointestinal stromal tumors. Cancer Res. 1999;59:4297–
21. Tuveson DA, Willis NA, Jacks T, et al. STI571 inactivation of the gastrointestinal
stromal tumor c-KIT oncoprotein: biological and clinical implications.
22. Wardelmann E, Neidt I, Bierhoff E, et al. c-kit mutations in gastrointestinal
stromal tumors occur preferentially in the spindle rather than in the
epithelioid cell variant. Mod Pathol. 2002;15:125–136.
894 © 2004 Lippincott Williams & Wilkins