Title: KIT codon 558 insertions in gastrointestinal stromal tumors (GISTs). Analysis of 17 rare KIT mutants.*

Elsevier Editorial System(tm) for Human Pathology Manuscript Draft Manuscript Number: YHUPA-D-07-00544R1 Title: KIT codon 558 insertions in gastrointestinal stromal tumors (GISTs). Analysis of 17 rare KIT mutants.* Article Type: Original Article Keywords: GIST; KIT exon 11; insertion; mutation; imatinib Corresponding Author: Dr. Jerzy Lasota, M.D. Corresponding Author's Institution: Armed Forces Institute of Pathology First Author: Jerzy Lasota, M.D. Order of Authors: Jerzy Lasota, M.D.; Ewa Wardelmann, MD; Maria Debiec-Rychter, MD; Sabine Merkelbach-Bruse, PhD; Raf Sciot, MD; Janusz Rys, MD; Sonja E Steigen, MD; Katarzyna Iwanik, MD; Joseph A Holden, MD; Anna Jerzak vel Dobosz, MS; Hans-Ulrich Schildhaus, MD; Markku Miettinen, MD; Wojciech Kuban, Ph.D. Abstract: Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal neoplasms of gastrointestinal tract often driven by oncogenic KIT exon 11 mutations. Although deletions and substitutions are most frequent KIT exon 11 mutations, duplications and insertions have been reported as well. In contrast to duplications, which cluster in 3'KIT exon 11, insertions affect 5'KIT, particularly codon 558. Clinicopathologic profile of GISTs with insertions in codon 558 is not known. In this study, 17 GISTs with codon 558 insertions is reported. Fifteen (88.2%) KIT codon 558 insertions consisted of 1694_1695insTCC leading to Lys558delinsAsnPro. However, 2 variant mutants Lys558delinsAsnGln and Lys558delinsAsnAsn were also identified. Based on analysis of insertions and flanking sequences, the insertions contain inverted DNA sequences of the opposite strand. Therefore, these insertions may developed due to a DNA strand switch during replication by DNA polymerases and by the effects of several different DNA repair processes. Patient median age was 61 years, and male to female ratio was 1:1.8. GISTs were diagnosed in stomach (n=4), small intestine (n=7) and rectum (n=3). Three tumors were disseminated and primary location could not be established. Fourteen tumors had spindle cell morphology and epithelioid cell features were seen in 2

intestinal and 1 disseminated GIST. Based on size and mitotic activity, 2 of 4 (50%) gastric and 3 of 7 (48.9%) small intestinal GISTs had >50% risk of metastases, based on previous studies of GIST prognosis. All 3 rectal GISTs were malignant. Metastases were verified in 8 of 12 (66.7%) of patients with known clinical and follow-up data. In summary, KIT codon 558 insertions are rare mutations accounting for < 1% of all KIT mutants. GISTs with these mutations appear to have predilection to female patients and intestinal location. Moreover, KIT codon 558 insertions might indicate an increased risk of malignant behavior for gastric GISTs

Letter of Submission May 13, 2008 Editors Human Pathology Dear Editor, We greatly appreciate the review of our manuscript entitled KIT codon 558 insertions in gastrointestinal stromal tumors (GISTs). Analysis of 17 rare KIT mutants. We have carefully evaluated the reviewer comments, and our point-by-point response to these comments is enclosed together with the revised version of our paper (all changes are in red). We hope that this version is acceptable for publication in Human Pathology. This manuscript has not been published previously, neither has it been considered concurrently by another publisher. If there will be publication costs other than those indicated in your Instruction for Authors, kindly consult us, as authorization from the government must be obtained before the manuscript is typeset. Under Title 17 of the Code, Section 105, copyright protection is not available for any work of the United States Government. Thank you for your consideration. Sincerely, Jerzy Lasota, MD, PhD Department of Soft Tissue Pathology Armed Forces Institute of Pathology 6825 16 th Street, N.W. Washington D.C. 20306-6000 Fax: 202-782-9182; Phone: 202-782-2792 E-mail: lasota@afip.osd.mil or jurek@erols.com

Response to Reviewers Point-by-point response to reviewer comments Reviewer #1 1. Would it be possible for the authors to elaborate on the location of this domain in the tertiary structure of this protein? Although we share reviewer s interest in the tertiary structure of the mutant protein, crystallographic studies would be required to elaborate scientifically on this matter. Since our study is based on FFPE tissues, the latter is unfeasible. However, we have redeveloped the paragraph in the Discussion addressing functional consequences of mutation-related structural changes of the KIT juxtamembrane domain. 2. It would be helpful if the authors included the treatment and clinical follow-up in the results section. Currently, theses data are addressed in the discussion, but not previously mentioned. Although, treatment data have already been included in Table 2 in the previous version, we have added an extra paragraph to the Results addressing this issue. 3. The results section is not well organized and the information flow is not entirely logical. Furthermore, additional emphasis is warranted regarding predicted risk (higher, it seems, than other mutational subsets). Following reviewer suggestion, we have redeveloped the Results section. 4. Although the overall numbers prohibit detailed comment on the prognostic importance/clinical behavior of GISTs with these mutations, are there any trends when compared to different mutational subsets/locations? Table comparing a risk of progressive disease among GIST with different type of KIT exon 11 mutations has been added. Following reviewer comment, we added an additional table comparing the risk of progressive disease among small intestinal and gastric GISTs with different types of KIT exon 11 mutations. 5. Minor copy editing is recommended. We have carefully reviewed the final version of the manuscript for copy editing errors. Reviewer #2 1. In the manuscript, there is no table that summarizes the information of the tumors originally analyzed in this study. Table 1 comprises the data published in the literature,

but some data are included in this study (Table 1.a). In other words, this paper includes the data that have already been published in the literature. On the other hand, Table 2 summarizes the data of this study, but lacks the genetic information in question. These shortcomings of the tables make it extremely difficult to understand the authors observations. Because almost all mutations are identical, we feel it is not necessary to include them into the Table. 2. The authors conclude that GISTs with the insertion mutations are inclined to occur in the small intestine and in female patients and to exhibit more malignant behaviors. However, none of these conclusions is statistically confirmed. Although they have, in fact, mentioned the lack of statistical evidences, it is obviously difficult to obtain statistical significance in analyses of the subjects in such a limited number. The authors might as well tone down their claims somewhat and describe their observations as they are. We have repeatedly stressed in the previous and new version of the manuscript that our observations have not been statistical confirmed. However, to support our hypothesis, a table comparing a risk of progressive disease among GISTs with different types of KIT exon 11 mutations has been added in the new version of our paper. Also, a recently published clinical data of two tumors with such insertions have been incorporated into the current version of the manuscript. Minor points: 1. The data of sequencing analyzes and the histopathological observations in Figure 1 should be separated. The former may as well be added to Figure 2, together with the information of the codon alignment. Following the reviewer s suggestion, we have combined data from Figure 1 and 2 and developed a new Figure (Figure 1 in the current version). 2. The sequence of the mutant allele in Figure 2 is incorrect and lacks one G in the first position of codon 559 (GTT). The information about amino acids may help understanding, if added to this figure. Following the reviewer s observation, sequence in the former Figure 2 (now Figure 1) has been corrected. 3. The most intriguing observation in this study is that the inserted sequences were found in the opposite, i.e. negative, strand. These inversion-like insertions strongly suggest their causative origins. The authors mentioned errors in DNA replication. What are the detailed mechanisms?

The most likely explanation is based on formation of a hairpin structure in a DNA strand, and this has been added to the discussion. However, the detailed mechanisms leading to such mutations are not known and their understanding will require experimental studies, which are beyond the scope of this report. We have developed separate sections in the Results and Discussion focused on molecular mechanisms leading to such mutations. Moreover, additional references have been added as well. 4. There seems to be confusion in the genetic terminology. What does the term, duplication, imply? This term normally refers to gene duplication, a large-scaled chromosome rearrangement to duplicate genes. A mutation nomenclature used in this study is based on the recommendations of Human Genome Variation Society (www.hgvs.org). The duplication definition provided by this genetic society is quoted below. Duplications are designated by "dup" after an indication of the first and last nucleotide(s) duplicated. Example: c.77_79dup (or c.77_79dupctg, c.77_79dup3) denotes that the three nucleotides 77 to 79 are duplicated.

* Revised Manuscript KIT codon 558 insertions in gastrointestinal stromal tumors (GISTs). Analysis of 17 rare KIT mutants.* 1 Jerzy Lasota, M.D., 2 Wojciech Kuban 3 Eva Wardelmann, M.D., 4 Maria Debiec-Rychter, M.D., 3 Sabine Merkelbach-Bruse, Ph.D., 5 Raf Sciot M.D., 6 Janusz Rys, M.D., 7 Sonja E. Steigen, M.D., 8 Katarzyna Iwanik M.D., 9 Joseph A. Holden M.D., 10 Anna Jerzak vel Dobosz M.S., 3 Hans-Ulrich Schildhaus, M.D. 1 Markku Miettinen, M.D. 1 Department of Soft Tissue Pathology, Armed Forces Institute of Pathology, Washington, DC 2 Laboratory of Genomic Integrity, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD 3 Department of Pathology, University of Bonn Medical Center, Bonn, Germany 4 Department of Human Genetics, Catholic University of Leuven, Belgium 5 Department of Pathology, Catholic University of Leuven, Belgium 6 Department of Pathology, M. Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Krakow, Poland. 7 Department of Pathology, University of Tromsø, Tromsø, Norway 8 Department of Pathomorphology, Karol Marcinkowski Medical University, Poznań, Poland 9 Department of Pathology, University of Utah Health Sciences Center, Salt Lake City 10 Department of Molecula Biology, M. Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Warszawa, Poland. Running title: KIT exon 11 insertions in GISTs Key words: GIST, KIT, exon 11, insertion, mutation Address for correspondence and requests for reprints: Jerzy Lasota M.D. Department of Soft Tissue Pathology Armed Forces Institute of Pathology 6825 16 th Street, N.W., Bldg. 54 Washington, DC 20306-6000 Fax: 202-782-9182, Phone: 202-782-2841 E-mail: lasota@afip.osd.mil, jurek@erols.com *The opinions and assertions contained herein are the expressed views of the authors and are not to be construed as official or reflecting the views of the Departments of the Army or Defense. This study was partially supported by the grant from the Polish Committee for the Scientific Research (PBZ-KBN-091/2003).

2 ABSTRACT Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal neoplasms of gastrointestinal tract often driven by oncogenic KIT exon 11 mutations. Although deletions and substitutions are most frequent KIT exon 11 mutations, duplications and insertions have been reported as well. In contrast to duplications, which cluster in 3 KIT exon 11, insertions affect 5 KIT, particularly codon 558. Clinicopathologic profile of GISTs with insertions in codon 558 is not known. In this study, 17 GISTs with codon 558 insertions is reported. Fifteen (88.2%) KIT codon 558 insertions consisted of 1694_1695insTCC leading to Lys558delinsAsnPro. However, 2 variant mutants Lys558delinsAsnGln and Lys558delinsAsnAsn were also identified. Based on analysis of insertions and flanking sequences, the insertions contain inverted DNA sequences of the opposite strand. Therefore, these insertions may developed due to a DNA strand switch during replication by DNA polymerases and by the effects of several different DNA repair processes. Patient median age was 61 years, and male to female ratio was 1:1.8. GISTs were diagnosed in stomach (n=4), small intestine (n=7) and rectum (n=3). Three tumors were disseminated and primary location could not be established. Fourteen tumors had spindle cell morphology and epithelioid cell features were seen in 2 intestinal and 1 disseminated GIST. Based on size and mitotic activity, 2 of 4 (50%) gastric and 3 of 7 (48.9%) small intestinal GISTs had >50% risk of metastases, based on previous studies of GIST prognosis. All 3 rectal GISTs were malignant. Metastases were verified in 8 of 12 (66.7%) of patients with known clinical and follow-up data. In summary, KIT codon 558 insertions are rare mutations accounting

3 for < 1% of all KIT mutants. GISTs with these mutations appear to have predilection to female patients and intestinal location. Moreover, KIT codon 558 insertions might indicate an increased risk of malignant behavior for gastric GISTs. INTRODUCTION Gastrointestinal stromal tumors (GISTs) are the most common gastrointestinal (GI) mesenchymal neoplasms diagnosed in different parts of GI-tract. Majority of GISTs show spindle cell morphology, however, tumors with epithelioid or pleomorphic cell features can also occur. Biological potential of GISTs vary from indolent, benign tumors to high grade sarcomas (1). GISTs are believed to develop from interstitial cells of Cajal or their precursors that are located around the myenteric plexus and dispersed in the muscularis propria (2). These cells act as intermediaries between the GI autonomic nervous and smooth muscle systems, generate pacemaker activity, and are involved in neurotransmission (3). Interstitial cells of Cajal share a phenotypic similarity with GISTs including KIT expression (2, 4). Activation of KIT, a type III tyrosine kinase receptor and KITdependent signaling pathways is crucial for the development and maintenance of different cell populations, including those of interstitial cells of Cajal (4, 5). Oncogenic, mutational KIT activation is a major driving force in GIST pathogenesis (6). In GISTs, gain-of-function KIT mutations occur predominantly in the juxtamembrane domain (exon 11). Less common mutational hot spots are located in the extracellular (exon 9) and tyrosine kinase (exon 13 and 17) domains (6, 7). Although

4 a great majority of GISTs have KIT mutations, gain-of-function mutations in plateletderived growth factor receptor alpha (PDGFRA), another member of type III tyrosine kinase receptor family, occur in a subset of tumors (8). A great majority of KIT exon 11 mutations consist of deletions, substitutions and duplications. However, in-frame insertions disrupting KIT codon 558 have also been reported (9-19). Most of these mutations consisted of 1694_1695insTCC leading to Lys558delinsAsnPro at the protein level. Molecular mechanisms underlying the formation of such a mutation are not known. Some types of KIT or PDGFRA mutations have been linked to specific tumor location and clinical behavior (7). GISTs with KIT exon 9 and exon 11 duplications show spindle cell morphology and occur in intestinal and gastric locations, respectively (13, 20, 21). In contrast, most of PDGFRA mutants have epithelioid morphology and are of gastric origin (22-24). Gastric GISTs with KIT exon 11 deletions tend to follow more malignant course of disease than those with substitutions (25). The clinicopathologic profile of GISTs with KIT codon 558 insertions is unknown. Although 14 tumors with such mutations have been mentioned in various GIST studies, incomplete clinicopathologic data have been published only in a few cases (Table 1). Thus, the purpose of this study was to evaluate 17 GISTs with KIT codon 558 insertions and define their clinicopathologic profile. MATERIALS AND METHODS

5 Seventeen primary or metastatic GISTs with KIT exon 11 insertions diagnosed prior to imatinib therapy were obtained from the files of the Armed Forces Institute of Pathology (AFIP), Washington, D.C., and Department of Pathology, University of Bonn Medical Center, Bonn, Germany, and, Department of Human Genetics, Catholic University of Leuven, Belgium, and Department of Pathology, University of Tromsø, Norway, and Department of Pathomorphology, Karol Marcinkowski Medical University, Poznań, Poland, Department of Pathology, M. Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Krakow, Poland and Department of Pathology, University of Utah Health Sciences Center, Salt Lake City All tumors were diagnosed based on previously published criteria including morphologic and immunohistochemical features. In addition, 9 tumors were evaluated immunohistochemically for keratin 18 (K18) expression following a previously published procedure (26). Mitoses were counted in 50 high power fields (HPFs), a total area of 5 square millimeters. Clinical follow-up data were obtained from the AFIP and contributor files. Based on tumor size and mitotic activity, GISTs were classified into 8 clinicopathologic prognostic groups as previously reported (1). KIT mutation status was evaluated using previously published methodologies (20 19). Nomenclature of the KIT mutations identified at the DNA level and deduced KIT-mutant sequences at the protein level are based on the recommendations of Human Genome Variation Society (www.hgvs.org). KIT mrna reference sequence XO6182 was obtained from GeneBank at www.ncbi.nlm.nih.gov. RESULTS

6 Molecular studies. There were 14 primary and 3 metastatic tumors with KIT codon 558 insertions. Fifteen of 17 (88.2%) of these consisted of 1694_1695insTCC leading to Lys558delinsAsnPro at the protein level. However, 2 variant insertions (1694_1695insCCA and 1694_1695insCAAC, the latter coexisting with 1695delG) leading to Lys558delinsAsnGln (Case 10) and Lys558delinsAsnAsn (Case 8) were also identified. Representative examples of direct sequencing of PCR products are shown in Figure 1. In Case 15, 2 samples were evaluated, 1 from a 6 cm large rectal GIST diagnosed 57 months prior to clinical dissemination, and another from intraabdominal metastatic lesion. While the rectal tumor was KIT exon 11 wild-type (KIT-WT), the metastatic lesion showed 1694_1695insTCC mutation. Analysis of KIT exon 11 sequence revealed that short TCC, CCA and CAAC insertions might contain sequences corresponding to inverted complementary DNA sequences involving or flanking codon 558 (Figure 1). To better understand the possible molecular mechanisms for these mutations, including possible formation of hairpin structures, KIT exon 11 seqence was evaluated using DINAMelt web server (27) and 4 melting profiles were predicted. Figure 2 shows predicted melting profiles for KIT exon 11 sequence. Demographic and clinicopathologic profile of GISTs with KIT codon 558 insertions.

7 The demographic and clinicopathologic data of GISTs with KIT codon 558 insertions analyzed in this study are listed in Table 2. Patient median age was 61 years, and male to female ratio was 1:1.8. Among 14 primary GIST, there were 4 (28.6%) gastric, 7 (50%) small intestinal and 3 (21.4%) rectal tumors. Three GISTs were disseminated, and primary location could not be established. Gastric tumors varied from 1.5 to 20 cm (average 10.6 cm), while the small intestinal ones varied from 2.1 to 11 cm (average 7.1 cm). Three of 4 gastric GISTs and 5 of 7 small intestinal GISTs were >5 cm. Two gastric and 2 small intestinal tumors, had >5 mitoses per 50 HPFs. Based on tumor size and mitotic activity 2 gastric, 3 small intestinal and 3 rectal GISTs were considered to have a high risk (>50%) of developing metastatic disease. Comparison of risk of progressive disease among gastric and intestinal GISTs characterized by different types of KIT exon 11 mutations is shown in Table 3. An increased frequency of tumors with high risk of progressive disease was seen among the GISTs with insertions compared to tumors with single nucleotide substitutions. Clinical follow-up data confirmed malignant course of disease in 1 gastric, 2 small intestinal and 2 rectal cases. Patient with the gastric tumor died of disease 54 months, while patients with disseminated small intestinal and rectal tumors were treated with imatinib mesylate and were alive from 2 to 186 months after primary diagnosis. All gastric GISTs and a majority (12 of 14) of intestinal tumors had spindle cell morphology. Representative histological images are shown in Figure 3 (A, B). In case 15, primary rectal GIST showed epithelioid morphology, while intraabdominal metastasis diagnosed 57 months after primary tumor had a spindle cell pattern. Histological images of both tumors are shown in Figure 3 (C, D).

8 Immunohistochemically all tumors were KIT (CD117) positive. K18 expression was not detected in any of analyzed tumors. DISCUSSION An oncogenic, mutational activation of KIT is a major driving force in GIST pathogenesis (6). A great majority of KIT mutations occur in juxtamembrane domain (exon 11) and consist of deletion/deletion-insertions, substitutions, duplications and insertions. KIT codon 558 is one of the most commonly deleted codons and extremely rarely harbors single nucleotide substitutions. Also, no duplications were found involving this codon. In contrast, insertions affect exclusively codon 558, however they are infrequent (7). A complex mutation consisting of deletions and insertions of inverted complementary DNA sequences was reported in 5 part of KIT exon 11 and involved codon 558 as well (28). In this study, 17 GISTs with insertions affecting KIT codon 558 were analyzed. Previously, 11 tumors with the same or similar insertions have been reported. However, only incomplete clinicopathologic parameters are available of these cases (9-19). 88.2% of KIT insertion mutants reported in this study had identical TCC insertions occurring between A1694_G1695 in the KIT codon 558. A 1694_1695insTCC leads to deletion-insertion, Lys558delinsAsnPro at the protein level. Such mutation was previously reported in 11 GISTs (9-12, 16-19). Moreover, this mutation was shown to cause constitutive KIT tyrosine phosphorylation (29 26).

9 In this study, 2 variant insertions affecting codon 558 were also identified: a CCA insertion (1694_1695insCCA) leading to Lys558delinsAsnGln, and a CAAC insertion coexisting with a loss of G from the AAG codon 558 triplet (1694_1696delinsCAAC) leading to Lys558delinsAsnAsn at the protein level. A Lys558delinsAsnGln mutation has been reported previously (14-15). Also, another variant insertion leading to Lys558delinsGlnPro mutation at the protein level was reported once (13). However, genomic sequence of the latter has not been published and cannot be easily predicted. It seems that besides insertion of 3 nucleotides, additional substitutions in codon 558 are necessary to create such a mutant. The pathogenesis of the insertions can be approached by inspection of the inserted sequences in comparison with the flanking sequences and the melting profiles and bioenergetic status of the corresponding DNA sequences. Short insertions affecting KIT codon 558 are probably a result of a DNA replication error. There are at least three possible scenarios for molecular events leading to such mutations. However, an initial step of intrastrand base pairing to form a so-called hairpin structure, an abnormal loopformation in the normally linear DNA-strand, on at least one of the DNA strands, is involved in all three scenarios. Figure 2 shows prediction of melting profiles and formation of the hairpin structures for KIT exon 11 sequence. Model A appears to be most stable and feasible than the other models, as it spends the minimum free energy to form the palindromic hairpin loop. If DNA polymerase is prevented from replicating through the palindromic hairpin, the terminal bases of the growing strand dissociate from the template and anneal to a short stretch of paired-complementary sequence on the other side of the hairpin. The hairpin is

10 thus bypassed, and DNA replication proceeds unhindered. In these cases, replication would proceed well into one side of the inverted and then slip to the other side of the repeat. Template switching between pseudo-inverted repeats causes not only sequence substitution but also base substitution and single-base deletion. However, the molecular mechanisms of template switching are still poorly understood (30). Also, one-sided mutation within complementary inverted sequence could possibly by event of illegitimate recombination be localized at the palindromic hairpin tip. The other molecular mechanism that could be postulated is an endonucleolytic nick. DNA break repair after nucleolytic digestion would result in an opposite inverted sequence and small deletions in the context of palindromic sequence, from one or both sides, with the potential to protect the locus from further rearrangement (31). The illegitimate rejoining of double-strand breaks in mammalian cells is a relatively efficient process (32). Also, it is possible that besides palindromic sequences, other unknown factors might commit a DNA-molecule to undergo this type of mutation and determine the insertion endpoints. The frequency of KIT codon 558 insertions among other KIT mutants varies between the studies from 1.2% to 10.5% (9, 10, 13, 18, 19). This most likely might represent random statistical variations related to relatively small number of analyzed cases. Based on AFIP data based of 700 KIT exon 11 mutants, insertion in KIT codon 558 represent <1% of all KIT mutations in GISTs. KIT exon 11 mutations have been found in GISTs from different locations. However, duplications in 3 KIT exon 11 have shown a predilection to gastric tumors with spindle cell morphology (13, 2019). Also, various PDGFRA mutations have been found mostly in gastric GISTs, especially those with epithelioid morphology (22-24 21-23). In contrast,

11 a great majority of Ala502_Tyr503dup mutants have been intestinal tumors (13, 21 20). It is possible, that GISTs showing predilection to different parts of GI-tract might develop from different subsets of interstitial cell of Cajal (3). Although gastric GISTs with KIT codon 558 insertions were more frequent among previously published cases (11, 14-17), tumors with such mutation clearly had predilection to intestinal location in this study. Large clinicopathologic and population-based studies showed that gastric GISTs are the most frequent (60%) followed by small intestinal (30%) and rectal (4%) tumors, respectively (1). In contrast to this finding, 50% of GISTs with KIT codon 558 insertions were from small intestine, and only 28.6% were diagnosed in stomach. Moreover, frequency of rectal GISTs was 5 times higher (21.4% versus 4%) than predicted. Majority of 1694_1695insTCC GIST mutants had spindle cell morphology, however, epithelioid and mixed spindle/epithelioid cell features were seen in 2 intestinal and 1 disseminated GIST. A mixed morphology was previously reported in 1 small intestinal tumor (11). Recent study based on 906 GISTs suggested that epithelioid cell features seen in small intestinal tumors might represent malignant transformation (33 29). There was only 1 pure epithelioid GISTs with 1694_1695insTCC mutation among five gastric tumors from this and previous study (16). This malignant GISTs immunohistochemically showed strong keratin expression (16). None of 9 tumors evaluated in this study expressed keratin. Thus, presence of insertions affecting KIT codon 558 can not be linked to such an immunophenotype. However, a previous study documented keratins expression in a subset of gastric and intestinal GISTs and linked this feature to malignant behavior (26 25).

12 In Case 15, the primary tumor was diagnosed in rectum 57 months prior to metastases. This primary tumor was KIT-WT, in contrast to the metastasis, which had insertion in KIT codon 558. Because detection of KIT mutations in GISTs decline substantially in older paraffin blocks (18, 25 24, 34 28), there is possibility that negative result of the primary tumor, in fact represents a false negative finding. However, a similar finding of a primary tumor with Val559Asp substitution and metastatic tumor with Lys558delinsAsnPro, diagnosed more than 3 years after primary surgery, has also been reported (11). Moreover, the morphology of the primary and metastatic tumor differed substantially in this case. The rectal GIST had purely epithelioid morphology, while the metastatic lesion was of spindle cell type (Figure 3). Taking into consideration the morphology of primary and metastatic lesions, different genotype and recently reported frequent occurrence of minimal GISTs in humans (35 30), it is possible that metastatic tumor in this case in fact represented a second primary GIST. Previous studies have shown that certain types of KIT exon 11 mutations could be linked to malignant behavior. For example, gastric GISTs with KIT exon 11 deletions follow more malignant course of disease than the ones with point mutations (25 24). Also, the presence of Tyr557_Lys558del was linked to unfavorable GIST clinical outcome (36, 37 31, 32). Similarly, gastric GISTs with insertions involving codon 558 showed a tendency for a higher risk of progressive disease than tumors characterized by single nucleotide substitutions. However, a link between insertions in KIT codon 558 and malignant clinical behavior of gastric GISTs could not be established statistically, because of relatively small number of analyzed cases. Also, three of four previously reported KIT codon 558 mutants were malignant GISTs of gastric origin (16, 17, 19).

13 Based on previous studies, it appears that the mutation profile does not correlate with clinical behavior of small intestinal GISTs (33). Also, in this study, no substantial difference between clinical behavior of small intestinal GISTs with insertions and GISTs defined by other KIT mutations was identified. Imatinib mesylate, a KIT tyrosine kinase inhibitor, has been successfully used for treatment of metastatic and advanced GISTs (38, 39 33, 34). In general, KIT exon 11 mutants have a good response to imatinib inhibition in vitro and in vivo studies (40, 41 35, 36). However, a recent in vitro study showed that a rare KIT substitution Val559Ile induces in contrast to common Val559Asp mutation, imatinib-resistant constitutive KIT activation (42 37). Thus, inhibitory effect of imatinib might differ substantially even among mutants involving the same codon. A Lys558delinsAsnPro has been shown to cause constitutive KIT tyrosine phosphorylation and was sensitive to imatinib in vitro (29 26). GISTs with Lys558delinsAsnPro and Lys558delinsGlnPro mutations showed partial response to imatinib treatment for almost eight and two months, respectively (15, 17). Although a Lys558delinsAsnPro mutant developed secondary resistance to imatinib treatment, no secondary KIT or PDGFRA mutations were identified in progressive lesion (17). In this study, all four Lys558delinsAsnPro mutants showed primary response to the imatinib treatment. The biological potential and sensitivity to imatinib of rare Lys558delinsAsnAsn mutant identified in this study requires further study. In summary, we have examined the pathogenesis and defined the clinicopathologic profile of GISTs with KIT codon 558 insertions. These mutations contained inverted DNA sequences of the opposite strand and may originate through DNA replication error following a hairpin formation in a DNA strand. The frequency of such tumors among

14 other KIT mutants is < 1%. Tumors with these mutations were more frequent in females, and show predilection to intestinal location and spindle cell morphology. Presence of KIT codon 558 insertions might indicate an increased risk for malignant behavior among gastric GISTs. However, more cases should be studied to confirm this observation.

15 REFERENCES 1. Miettinen M, Lasota J. Pathology and prognosis of gastrointestinal stromal tumors at different sites. Semin in Diagn Pathol 2006;23:70-83. 2. Kitamura Y, Hirota S. Kit as human oncogenic tyrosine kinase. Cell Molec Life Sci 2004;61:2924-31. 3. Huizinga JD. Gastrointestinal peristalsis: joint action of enteric nerves, smooth muscle, and interstitial cells of Cajal. Microsc Res Tech 1999;47:239-247. 4. Kindblom LG, Remotti HE, Aldenborg F, Meis-Kindblom JM. Gastrointestinal pacemaker cell tumor (GIPACT): gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 1998; 152; 1259-1269. 5. Pawson T. Regulation and targets of receptor tyrosine kinases. Eur J Cancer 2002;38: S3-S10. 6. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998;279:577-80. 7. Lasota J, Miettinen. KIT and PDGFRA mutations in gastrointestinal stromal tumors (GISTs). Seminars in Diagnostic Pathology 2006;23:91-102. 8. Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating mutations in gastrointestinal stromal tumors. Science 2003;299:708-710. 9. Taniguchi M, Nishida T, Hirota S, et al. Effect of c-kit mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 1999;59:4297-4300. 10. Rubin BP, Singer S, Tsao C, et al. KIT activation is a ubiquitous feature of gastrointestinal stromal tumors. Cancer Res 2001;61:8118-8121. 11. Andersson J, Sjogren H, Meis-Kindblom JM, et al. The complexity of KIT gene mutations and chromosome rearrangements and their clinical correlation in gastrointestinal stromal (pacemaker cell) tumors. Am J Pathol 2002;160:15-22. 12. 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. 13. Antonescu CR, Sommer G, Sarran L, et al. Association of KIT exon 9 mutation with nongastric primary site and aggressive behavior: KIT mutation analysis and clinical correlates of 120 gastrointestinal stromal tumors. Clin Cancer Res 2003;9:3329-3337.

16 14. Subramanian S, West R, Corless CL, et al. Gastrointestinal stromal tumors (GISTs) wit KIT and PDGFRA mutations have distinct gene expression profiles. Oncogene 2004; 23:7780-90. 15. Antonescu CR, Besmer P, Guo T, et al. Acquired resistance to imatinib in gastrointestinal stromal tumors occurs through secondary gene mutation. Clin Cancer Res 2005;11:4182-4190. 16. Rossi G, Sartori G, Valli R, Bertolini F, Bigiani N, Schirosi L, Cavazza A, Luppi G. The value of c-kit mutational analysis in a cytokeratin positive gastrointestinal stromal tumour. J Clin Pathol 2005;58:991-993. 17. Heinrich MC, Corless CL, Blanke CD, et al. Molecular correlates of imatinib resistance in gastrointestinal stromal tumors. J Clin Onco 2006;24:4764-4774. 18. Steigen SE, Eide TJ, Wasag B, et al. Mutations in gastrointestinal stromal tumorsapopulation-based study from Northern Norway. APMIS 2007;115:289-298. 19. Kontogianni-Katsarou K, Dimitriadis E, Lariou C, et al. KIT exon 11 codon 557/558 deletion/insertion mutations define a subset of gastrointestinal stromal tumors with malignant potential. World J Gastroenterol 2008;14:1891-1897. 20. Lasota J, Dansonka-Mieszkowska A, Stachura T, et al. Gastrointestinal stromal tumors with internal tandem duplications in 3 end of KIT juxtamembrane domain occur predominantly in stomach and generally seem to have a favorable course. Mod Pathol 2003;16:1257-64. 21. Lasota J, Kopczynski J, Sarlomo-Rikala M, et al. KIT 1530ins6 mutation defines a subset of predominantly malignant gastrointestinal stromal tumors of intestinal origin. Human Pathol 2003;34:1306-12. 22. Lasota J, Dansonka-Mieszkowska A, Sobin LH, et al. A great majority of GISTs with PDGFRA mutations represents gastric tumors of low or no malignant potential. Lab Invest 2004;84:874-83. 23. Wardelmann E, Hrychyk A, Markelbach-Bruse S, et al. Association of plateletderived growth factor receptor α mutations with gastric primary site and epithelioid or mixed cell morphology in gastrointestinal stromal tumors. J Molec Diagnostics 2004; 6:197-204. 24. Wasag B, Debiec-Rychter M, Pauwels P, et al. Differential expression of KIT/PDGFRA mutant isoforms in epithelioid and mixed variants of gastrointestinal stromal tumors depends predominantly on the tumor site. Mod Pathol 2004;17:889-94.

17 25. Miettinen M, Sobin LH, Lasota J. Gastrointestinal stromal tumors (GISTs) of the stomach - a clinicopathologic, immunohistochemical and molecular genetic study of 1756 cases with long-term follow-up. Am J Surg Pathol 2005;29:52-68. 26. Sarlomo-Rikala M, Tsujimura T, Lendahl U, et al. Patterns of nestin and other intermediate filament expression distinguish between gastrointestinal stromal tumors, leiomyomas and schwannomas. APMIS 2002;110:499-507. 27. Markham, N. R. & Zuker, M. DINAMelt web server for nucleic acid melting prediction. Nucleic Acids Res 2005; 33:W577-W581. 28. Lasota J, Miettinen M. KIT exon 11 deletion-inversions represent complex mutations in gastrointestinal stromal tumors. Cancer Genet Cytogenet 2007;175:69-72. 29. Tuveson DA, Willis NA, Jacks T et al. STI571 inactivation of the gastrointestinal stromal tumor c-kit oncoprotein: biological and clinical implications. Oncogene 2001;20:5054-8. 30. Maki H. Origins of spontaneous mutations: specificity and directionality of basesubstitution, frameshift, and sequence-substitution mutageneses. Annu Rev Genet. 2002;36:279-303. 31. Akgün E, Zahn J, Baumes S, et al. Palindrome resolution and recombination in the mammalian germ line. Mol Cell Biol 1997; 17:5559-70. 32. Roth DB, Wilson JH. Illegitimate recombination in mammalian cells. In Genetic recombination; Kucherlapati R, Smith GR (ed.); American Society for Microbiology, Washington, D.C. 1988; p.621 653. 33. Miettinen M, Makhlouf H, Sobin LH, et al. Gastrointestinal stromal tumors (GISTs) of the jejunum and ileum: a clinicopathologic, immunohistochemical and molecular genetic study of 906 cases before imatinib with long-term follow-up. Am J Surg Pathol 2006;30:477-89. 34. Andersson J, Bumming P, Meis-Kindblom JM, et al. Gastrointestinal stromal tumors with KIT exon 11 deletions are associated with poor prognosis. Gastroenterology 2006; 130:1573-1581. 35. Agaimy A, Wunsch PH, Hofstaedter F et al. Minute gastric sclerosing stromal tumors (GIST tumorlets) are common in adults and frequently show c-kit mutations. Am J Surg Pathol 2007; 31:113-120. 36. Wardelmann E Losen I, Hans V, et al. Deletion of Trp-557 and Lys-558 in the juxtamembrane domain of the c-kit protooncogene is associated with metastatic behavior of gastrointestinal stromal tumors. Int J Cancer 2003;106:887-95.

18 37. Martin J, Poveda J, Llombart-Bosch A, et al. Deletions affecting codons 557-558 of the c-kit gene indicate a poor prognosis in patients with completely resected gastrointestinal stromal tumors: a study by the Spanish Group for Sarcoma Research (GEIS). J Clin Oncol 2005;23:6190-8. 38. Joensuu H, Roberts PJ, Sarlomo-Rikala M et al. Effect of tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. N Engl J Med 2001;344:1052-6. 39. Demetri GD. Identification and treatment of chemoresistant inoperable or metastatic GIST: experience with the selective tyrosine kinase inhibitor imatinib mesylate (STI571). Eur J Cancer 2002; 38 Suppl 5:S52-59. 40. 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. 41. Debiec-Rychter M, Sciot R, Le Cesne A, et al. KIT mutations and dose selection for imatinib in patients with advanced gastrointestinal stromal tumors. Eur J Cancer 2006;42:1093-1103. 42. Nakagomi N, Hirota S. Juxtamembrane-type c-kit gene mutation found in aggressive systematic mastocytosis induces imatinib-resistant constitutive KIT activation. Lab Invest 2007;87:365-371.

19 FIGURE LEGENDS Figure 1. Genomic sequences of three types of KIT codon 558 insertions (A-C). A 1694_1695insTCC leading to Lys558delinsAsnPro (A), 1694_1695insCCA leading to Lys558delinsAsnGln (B), and 1694_1695insCAAC leading to Lys558delinsAsnAsn (C). KIT wild-type and mutant sequences and KIT amino acids are shown schematically above. Mutations are in red. KIT wild-type sequences homologous to inverted insertions are boxed. Figure 2. Four melting profiles for KIT exon 11 sequences predicted using DINAMelt web server for nucleic acid melting prediction (27). Nucleotides are numbered according to KIT mrna reference sequence XO6182, obtained from GeneBank at www.ncbi.nlm.nih.gov. Figure 3. Examples of histological images of GISTs with KIT codon 558 insertions (A,B). Case 3, malignant gastric GIST with 100 mitosis/50hpf (A). Case 8, benign small intestinal GIST with 0 mitosis/50hpf (B). Histological images of tumors diagnosed in Case 15 (C, D). A rectal epithelioid GIST (C) diagnosed 57 months prior to intraabdominally disseminated spindle cell tumor (D).

Table 1 Ref. Table 1. Summary of clinicopathologic data of previously reported GISTs with KIT codon 558 insertions. KIT 558 ins among exon 11 mutants (%) Type of mutation Age/Sex Tumor location Additional information 9 2/71 (2.8%) Lys558delinsAsnPro ND ND Preimatinib study 10 2/34 (5.9%) Lys558delinsAsnPro ND ND Preimatinib study 11 1 b Lys558delinsAsnPro 50/M IAB MET Small bowel GIST with Val559Asp 3 years prior to diagnosis 12 1 a /19 (5.3%) Lys558delinsAsnPro ND ND Preimatinib study 13 1/81(1.2%) Lys558delinsGlnPro ND ND Preimatinib study 14 1 c Lys558delinsAsnGln 53/F Stomach Tumor size 10.5 cm 15 1/21 (4.8%) Lys558delinsAsnGln ND Small bowel Malignant spindle cell GIST from imatinib treatment trial 16 Case report Lys558delinsAsnPro 32/F Stomach Malignant epithelioid GIST; IAB MET after 48 months; imatinib treatment 17 1/26 (3.9%) Lys558delinsAsnPro 60/M Stomach Malignant GIST from imatinib treatment trial 18 1 a /58 (1.7%) Lys558delinsAsnPro ND ND Preimatinib study 19 2/19 (10.5%) Lys558delinsAsnPro 79/M Stomach Malignant epihtelioid GIST; tumor size16cm; 9 mitoses in 50HPF; no follow-up data Lys558delinsAsnPro 60/F Small bowell Malignant epithelioid GIST; tumor size 17cm; 3 mitoses in 50HPF; no follow-up data a included in this study; b, c from GIST cohorts preselected for genetic and gene expression profile studies. Abbreviations: ND - no data, IAB-intraabdominal, METmetastases

Table 2 Table 2. Summary of clinicaopathologic data from the cases analyzed in this study. Case Age/Sex Location Histology Size (cm) Mitoses/ Group No. 50HPF 1 43/M Stomach Spindle 20 20 6b UNK 2 68/F Stomach Spindle 12 0 3b UNK 3 52/M Stomach Spindle 9 100 6a DOD (54) 4 54/F Stomach Spindle 1.5 2 1 UNK Follow-up (months) 5 42/F Small bowel Spindle/Epithelioid 11 5 3b AWD (102) Gleevec 6 42/F Small bowel Spindle 10 8 6b AWD (182) Gleevec 7 57/M Small bowel Spindle 9 9 6a ATSU (5) 8 72/F Small bowel Spindle 8 0 3a ANED (170) 9 55/M Small bowel Spindle 6 1 3a DUNK (11) 10 85/F Small bowel Spindle 3.5 0 2 UNK 11 57/F Small bowel Spindle 2.1 1 2 UNK 12 61/M Rectum Spindle 6 + 3.5 41 6a UNK 13 63/F Rectum Spindle 7.5 17 6a LIVER MET; AWD (2) Gleevec 14 65/M Rectum Spindle/Epithelioid UNK 125 ND AWD (186) Gleevec 15a* 63/F Rectum Epithelioid 6 4 3a IAB MET (57) 15b** 68/F Intraabdominal Spindle UNK 44 ND DOD (4) 16 78/F Intraabdominal Spindle UNK 8 ND UNK 17 67/F Intraabdominal Epithelioid UNK 5 ND IAB and LIVER MET * 14a KIT/PDGFRA-WT GIST; **14b GIST with Lys558delinsAsnPro; Abbreviations: ATSU-alive tumor status unknown, AWD- alive with disease, DOD-died of disease, DUNK-died of unknown causes, IAB-intraabdominal, MET-metastases, ND-not done, UNK-unknown

Table 3 Table 3. Comparison of risk of progressive disease among gastric and small intestinal GISTs with different types of KIT exon 11 mutations. Group Gastric GISTs Small intestinal GISTs All* Del/delins* Point Insertions in All** Del/delins** Point Insertions in mutations* codon 558 mutations** codon 558 1 116 (7.5%) 4 (5.8%) 1 (2.9%) 1 (25%) 69 (8.7%) 6 (10%) 1 (3.6%) 0 2 456 (29.4%) 15 (21.7%) 12 (35.3%) 0 174 (22%) 6 (10%) 7 (25%) 2 (28.6%) 3a 301 (19.4%) 10 (14.5%) 15 (44.1%) 0 177 (22.4%) 18 (30%) 4 (14.3%) 2 (28.6%) 3b 193 (12.4%) 7 (10.1%) 2 (5.9%) 1 (25%) 99 (12.5%) 11 (18.3%) 1 (3.6%) 1 (14.3%) 4 8 (0.5%) 0 0 0 2 (0.3%) 0 1 (3.6%) 0 5 137 (8.8%) 7 (10.1%) 1 (2.9%) 0 37 (4.7%) 3 (5%) 0 0 6a 153 (9.9%) 9 (13%) 3 (8.8%) 1 (25%) 108 (13.7%) 7 (11.7%) 3 (10.7%) 1 (14.3%) 6b 188 (12.1%) 17 (24.6%) 0 1 (25%) 125 (15.8%) 9 (15%) 5 (17.9%) 1 (14.3%) Total: 1552 69 34 4 791 60 28 7 * and ** based on previous studies on gastric (25) and small intestinal GISTs (21).

Figure 1 A Trp Lys Val Val Codon No. 557 558 559 560 KIT-WT 5' T G G A A G G T T G T T 3' 3' A C C T T C C A A C A A 5' Trp Asn Pro Val KIT-MT 5' T G G A A T C C G G T T 3' B KIT-WT 5' T G G A A G G T T G T T 3' 3' A C C T T C C A A C A A 5' Trp Asn Gln Val KIT-MT 5' T G G A A C C A G G T T 3' C KIT-WT 5' T G G A A G G T T G T T 3' 3' A C C T T C C A A C A A 5' Trp Asn Asn Trp KIT-MT 5' T G G A A C A A C G T T 3'

Figure 2 Click here to download high resolution image

Figure 3 Click here to download high resolution image