Molecular Analysis to Demonstrate That Odontogenic Keratocysts Are Neoplastic

Narasimhan P. Agaram, MD; Bobby M. Collins, DDS; Leon Barnes, MD; Deren Lomago, BS; Dalal Aldeeb, MD; Patricia Swalsky, BS; Sydney Finkelstein, MD; Jennifer L. Hunt, MD

Context.Odontogenic keratocysts (OKCs) are unique odontogenic lesions that have the potential to behave aggressively, that can recur, and that can be associated with the nevoid basal cell carcinoma syndrome. Whether they are developmental or neoplastic continues to be debated. Objectives.To identify loss of heterozygosity of tumor suppressor genes in OKCs and to suggest a pathogenetic origin for these lesions. Design.We examined 10 OKCs for loss of heterozygosity of tumor suppressor genes, using a microdissection and semiquantitative genotyping analysis. The genes analyzed included 10 common tumor suppressor genes, as well as the PTCH gene, which is mutated in nevoid basal cell carcinoma syndrome. dontogenic keratocysts (OKCs) are unusual cystic lesions that arise from the dental laminal epithelium and usually occur in the mandible or the maxilla.1 Odontogenic keratocysts can arise sporadically or in association with the nevoid basal cell carcinoma syndrome (NBCCS or Gorlin syndrome), in which patients present with multiple abnormalities, especially basal cell carcinomas, OKCs, skeletal changes, and palmar pits.2 Odontogenic keratocysts are known for their potentially aggressive clinical behavior. They can present with extensive local invasion into adjacent structures and frequently recur after simple surgical enucleation or even after more extensive treatments.3 A few may even undergo malignant transformation into squamous cell carcinoma.4,5 Despite their aggressive behavior, the leading theory of the origin of OKCs has remained one of a developmental anomaly.6 Histologically, OKCs are lined by a thin squamous epithelium, approximately 5 to 8 cells thick, covered by a thin corrugated layer of parakeratin. The epithelium can show budding of the basal layer into the underlying stroma, and there can also be smaller adjacent detached cysts in the wall, which have been termed daughter cysts.6 These leAccepted for publication October 29, 2003. From the Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, Pa (Drs Agaram, Barnes, Aldeeb, Finkelstein, and Hunt; Mr Lomago; and Ms Swalsky); and the Department of Pathology, University of Pittsburgh Dental School, Pittsburgh, Pa (Dr Collins). The authors have no relevant nancial interest in the products or companies described in this article. Reprints: Jennifer L. Hunt, MD, Department of Pathology, University of Pittsburgh Medical Center, Presbyterian Hospital A616.3, 200 Lothrop St, Pittsburgh, PA 15213 (e-mail: huntjl@upmc.edu). Arch Pathol Lab MedVol 128, March 2004

Results.Loss of heterozygosity was seen in 7 of 10 cases, with a frequency between 11% and 80% of the genes studied. The genes that exhibited the most frequent allelic losses were p16, p53, PTCH, and MCC (75%, 66%, 60%, and 60%, respectively). Daughter cysts were associated .02), but epiwith a higher frequency of allelic loss (P thelial budding was not. Conclusions.Our study indicates that a signicant number of OKCs show clonal loss of heterozygosity of common tumor suppressor genes. The nding of clonal deletion mutations of genomic DNA in these cysts supports the hypothesis that they are neoplastic rather than developmental in origin. (Arch Pathol Lab Med. 2004;128:313317)
sional elements that are removed from the main cyst have been implicated as the cause for recurrence of the lesion after simple enucleation.1,7 Another theory to explain the high recurrence rate assumes that OKCs are neoplastic.2,8 Sporadic OKCs have been shown in several studies to harbor mutations in the PTCH gene on chromosome 9q22. PTCH is the gene responsible for NBCCS.9,10 However, to our knowledge no studies have examined OKCs for DNA damage in other known tumor suppressor genes or tumor oncogenes. To explore the possibility that OKCs might represent a neoplastic process, we performed a genotypic analysis using a panel of tumor suppressor genes on a series of these lesions. Clonal allelic losses of tumor suppressor genes are known to be present in neoplastic conditions and to accumulate in a temporal fashion in many types of benign and malignant neoplasms.11 Therefore, our nding of signicant clonal loss of heterozygosity in tumor suppressor genes lends further support to the hypothesis OKCs are neoplastic. METHODS
Cases of OKCs were selected from the les of the Division of Anatomic Pathology, University of Pittsburgh PresbyterianShadyside Hospital, and the Division of Oral Medicine and Pathology, University of Pittsburgh School of Dental Medicine, Pittsburgh, Pa. Cases that had undergone decalcication were not included. The hematoxylin-eosinstained slides were reviewed, and the diagnosis conrmed. Pathology data were obtained from the original report and from review of the slides. Clinical follow-up was obtained from the patients electronic medical record. A parafn block was selected from each case to include areas of the lesion and normal adjacent stromal tissue. Slides were cut from the block (1 hematoxylin-eosin, 6 unstained deparafnized
Neoplastic Status of Odontogenic KeratocystsAgaram et al 313

Table 1. Tumor Suppressor Genes, Their Chromosomal Location, the Microsatellite Marker Used, Primer Sequences, and the Rate of Allelic Loss
Tumor Suppressor Gene Chromosome Location Forward Primer (5 3 ) Reverse Primer (5 3 ) Rate of Allelic Loss, % (No. of Cases)

Marker

L-MYC CMM VHL HOGG1 APC MCC PTCH MTS1/p16 (a) MTS1/p16 (b) Pten p53 (a) p53 (b)

1p34.2 1p36.2 3p26 3p26 5q21 5q21 9q22.1 9p21 9p21 10q23 17p13 17p13

MYCL.5NT D1s.407 D3s.1539 D3s.2303 D5s.615 D5s.592 D9s.252 D9s.254 D9s.251 D10s.1173 D17s.974 D17s.1289

CCGTAGCCTGGCGAGACTCC GCCTGTGCTAACCACATGG CCATTACTCTCTCCATAGCTAG GTTAGTATCCCAAGGGGAGC GAGTCAGGGTTTTACAGAG GGCCTAACTGGAATGTGT TTGTCAACTCCTAATATGGAC TATCCTGGGTAATAACTGCC TATTCTGCATGTTTTATGTG GGAGGTGGAGGTTGCAGTG AGCCTGGGTGAGAGTGAGAC GCATGGTCTTTTTCCATTCC

GAAAATTCGACGTTGTTAAAG GACCCGCCCCCTATCTTCC GAAAATACATTACTTCGAC GGACTGGGACAGAGGTCTCG GGAAAAGTAAACAGATACA CGGGTCTATTTGTGGTCG CATACTCTTGAACCCCTATAG CTCCTATTTGGACGAAGTGAG GAAAAATTCCGAAACATC CTGACAAAATGGATATTAC GTGGTTGGACAATTGTTACCG GAATTTACTGACGAATCTCCGTC

29 0 20 20 40 60 60 75 17 14 66 0

(2/7) (0/3) (1/5) (1/5) (2/5) (3/5) (3/5) (3/4) (1/6) (1/7) (2/3) (0/5)

Table 2. Clinical and Pathologic Characteristics of Cases, and the Frequency of Allelic Loss (FAL) for Each Case
Case No. Age, y/Sex Site Size, cm Inammation Buds Daughter Cysts FAL, %

1 2 3 4 5 6 7 8 9 10

67/M 76/M 53/M 75/M 7/M 19/F 20/M 68/F 40/F 64/F

Mandible Mandible Mandible Mandible Mandible Mandible Mandible Maxilla Mandible Maxilla

1 2.5 Unknown 4 1 2 2.5 Unknown Unknown Unknown

Yes Yes Yes No Yes No Yes Yes No Yes

Yes No Yes Yes No No No Yes Yes No

No Yes No Yes Yes No Yes No No Yes

80 57 0 20 33 0 11 33 0 33

slides, and 1 hematoxylin-eosin). The hematoxylin-eosinstained slides were examined, and targets that were present throughout the series were selected and marked for microdissection. Tissue was microdissected from the unstained slides under direct visualization using a stereoscopic microscope and a beveled surgical blade. DNA was extracted from the tissue fragments using a standard proteinase K digestion and extraction protocol for microdissected tissue fragments. Primers designed to yield polymerase chain reaction (PCR) products smaller than 150 base pairs were used, which is optimal for parafn-embedded tissue DNA. The genetic loci selected for PCR included short tandem repeats that are known to be at or near specic tumor suppressor genes (Table 1). The short tandem repeats were selected for their high polymorphism rates ( 75%) in the general population. Polymerase chain reactions were performed for tumor and normal tissue under the same conditions. The PCR products were analyzed using capillary electrophoresis (Applied Biosystems, Inc, Prism 310, Foster City, Calif, and Genescan software). The PCR products from the normal tissue were analyzed rst to determine whether the patients DNA was informative at each specic marker locus. The software generates a different peak for each differently sized PCR product when the patient sample is informative at that genetic locus (ie, heterozygous for the short tandem repeat unit). The ratio of the 2 different peaks in normal tissue is approximately 1:1, as the 2 chromosomal copies of the locus are present in equal amounts. For informative loci, the PCR products from the lesion were then analyzed. When the ratio of the 2 peaks is greater than 1.5 or less than 0.6, allelic imbalance is present in a clonal manner in more than 50% of the cells sampled, and this is designated as loss of heterozygosity. The genotyping results were collected for each case, and a statistical analysis comparing the clinical, histologic, and genotyping results was performed using 2 analysis and the Student t
314 Arch Pathol Lab MedVol 128, March 2004

test (SPSS software, version 11.0, Chicago, Ill). A correlation was designated as statistically signicant at P .05. This study was approved by the Institutional Review Board of the University of Pittsburgh.

RESULTS The study included 10 OKCs, which were surgically resected from 4 female and 6 male patients. The mean age of the patients was 49 years (range, 776 years). The lesions had an average size of 2.2 cm (range, 14 cm). One of the cases was a recurrent OKC, while the remainder were primary lesions. No patients in this series were known to have NBCCS. All of the tumors were clinically completely excised. The clinical, pathologic, and molecular results are presented in Table 2. The review of each case showed the typical histologic features of OKC, including a cystic lesion with a wavy parakeratinized squamous epithelium, polarization of the basal epithelial cells, and underlying brous stroma (Figure 1). Focal signicant chronic inammation, consisting of dense lymphoid and plasma cell inltrates, was noted in 7 of 10 cases (Figure 2). Epithelial buds were present in 5 of 10 cases (Figure 3), and daughter cysts were seen in 5 of 10 cases (Figure 4). None of the cases had evidence of malignant transformation. DNA was obtained and adequately amplied in all cases. The average number of informative genetic loci per case was 7. Normal tissues did not show loss of heterozygosity in the gene loci studied. Loss of heterozygosity was seen in at least 1 tumor suppressor gene locus in 7 of 10 cases studied in the epithelial component that was samNeoplastic Status of Odontogenic KeratocystsAgaram et al

Figure 1. Typical thin epithelial lining of an odontogenic keratocyst with a wavy (corrugated) surface and supercial keratinization (hematoxylineosin, original magnication 200). Figure 2. Lining of an odontogenic keratocyst that shows signicant inammation, consisting of lymphoplasmacytic inltrates (hematoxylin-eosin, original magnication 20). Figure 3. Epithelium of an odontogenic keratocyst that shows epithelial budding into the underlying stroma (arrowheads) (hematoxylin-eosin, original magnication 100). Figure 4. Daughter cyst (arrowheads) separated from the main lesion of this odontogenic keratocyst (arrows) (hematoxylin-eosin, original magnication 100).

pled (Table 2). The frequency of allelic loss (FAL) averaged 27% among all cases studied (range, 0%80%). The genes that most commonly had allelic losses were p16, p53, PTCH, and MCC, which showed FALs of 75%, 66%, 60%, and 60%, respectively. The allelic loss rates for each gene studied are shown in Table 1. The presence of daughter cysts signicantly correlated with the FAL; 5 of 5 cases with daughter cysts demonstrated at least 1 mutation, while only 2 of 5 without daughter cysts had mutations (P .02). Budding did not correlate with mutations, but did correlate with the age of the patient and the size of the lesion. Buds were found in an older age group (mean age, 61 vs 37 years; P .02) and in larger cysts (2.5 vs 2.0 cm; P .03). The FAL correlated with the presence of inammation (35% FAL with inammation vs 7% FAL without inammation; P .05). The overall mutation rate did not correlate with site, age, sex, or size of the tumor.
Arch Pathol Lab MedVol 128, March 2004

Although follow-up periods were limited, there were no known recurrences after the date of primary surgery. One of the lesions studied was a recurrent OKC (patient 9 in Table 2), and this lesion did not show any mutations. COMMENT Odontogenic keratocyst is a distinctive odontogenic cyst with a potential for aggressive behavior, local recurrence, and an association with NBCCS. The term OKC was rst used in 1956 by Philipsen12 to describe an odontogenic cyst with a keratinous epithelial lining. These cysts represent between 5% and 15% of all odontogenic cysts.6 Odontogenic keratocysts, like other odontogenic cysts, have been hypothesized to be developmental, arising from the dental lamina.13 Odontogenic keratocysts usually present in the second and third decades, with an average age of presentation of 30.8 years.14 There is a second peak in the fourth to fth
Neoplastic Status of Odontogenic KeratocystsAgaram et al 315

decades. A slight male predominance has been noted in most studies. Odontogenic keratocysts are most often located in the mandible (65%), although they can also occur in the maxilla (35%). Clinically, an OKC will present as a localized swelling with possible spontaneous drainage of cyst uid. Radiographs show a well-dened lytic lesion, which can be unilocular or multilocular. In one fourth of the cases they are associated with the crown of the adjacent tooth and 10% to 15% are reported to erode the adjacent roots.6 Microscopically, OKCs are lined by a stratied squamous lining with a thin, corrugated, parakeratotic layer. The basal cells of the epithelium are characteristically tall and show prominent palisading. Budding of the basal layer and daughter cyst formation are frequent ndings. There can also be an associated inammatory response. About 5% of all OKCs are associated with the NBCCS. This inherited autosomal dominant condition presents with numerous manifestations, the chief of which are multiple basal cell carcinomas, OKCs, skeletal anomalies, and palmar pits.15,16 Odontogenic keratocysts may behave aggressively and can penetrate cortical bone, extending into the surrounding soft tissue or into the maxillary sinuses.1 They are also particularly prone to recurrence, with reported local recurrence rates ranging from 3% to 60%.1,13 The recurrence rate is higher for tumors associated with NBCCS and in lesions arising from the mandible.2,14 The presence of daughter cysts is also considered to be a factor for the high recurrence rates.14 The type of therapy used is known to inuence the recurrence rate, with simple enucleation showing the highest recurrence rates.3,13 Application of adjuvant local therapy, such as Carnoy solution or cryotherapy, may decrease recurrence rates.3 Despite the known high recurrence rate and rare reports of OKCs that transform into squamous cell carcinomas,5 the etiology of these tumors is still generally considered to be a developmental anomaly as opposed to a neoplasm.6 However, the aggressive behavior of this lesion has given rise to the hypothesis that OKCs may represent a low-grade neoplasm.2,8 Although there are some similarities between OKCs and other types of odontogenic cysts, signicant differences suggest a different biological origin. One study found that OKCs have a weak and discontinuous linear staining for laminin and collagen IV, which suggests unusual interactions between the epithelium and the adjacent connective tissue.17,18 In addition, OKCs have greater suprabasal staining with markers of proliferation, such as Ki-67 and proliferating cell nuclear antigen,19 and more signicant staining with p53 as compared to the other odontogenic cysts.2,20 Clearly, the OKCs that arise in association with NBCCS have a genetic contribution to their origin and are unlikely to be purely developmental. The gene for NBCCS has been localized to chromosome 9q22 and is known to be a tumor suppressor gene related to the PATCHED gene (PTCH).2 A study of sporadic OKCs showed that the cysts could also demonstrate germline mutations in the PTCH gene or loss of heterozygosity at 9q22.3-q31.9,10 In our study, we assessed 10 different sporadic OKCs for a panel of tumor suppressor genes to determine the frequency of loss of heterozygosity. Overall, loss of heterozygosity was seen in 7 of 10 cases. The FAL, which is a calculated rate of allelic imbalance, ranged from 11% to
316 Arch Pathol Lab MedVol 128, March 2004

80%. The genes with the most frequent allelic losses were p16 (75%), p53 (66%), PTCH (60%), and MCC (60%). Of particular note was the high FAL at the 9q22 locus, which contains the PTCH gene, corroborating prior reports of DNA damage in this gene in sporadic OKCs. Because daughter cysts and budding have been clinically associated with recurrence in the past, we examined the correlation between the rate of mutations and the presence of these histologic elements. All 5 cases with daughter cysts demonstrated at least 1 gene locus with allelic loss, whereas only 2 of 5 cases without daughter cysts had allelic loss (P .02). Budding did not correlate with allelic loss, but did correlate with an older age group (mean age, 61 vs 37 years; P .02) and the size of the cysts (2.5 vs 2.0 cm; P .03). Although inammation is considered to arise secondary to cyst irritation, it is possible that it represents a response to tumor as well. In our series, inammation was associated with a higher FAL (35% vs 7%; P .05). The rate of allelic loss did not correlate with site, age, sex, or size of the tumor. Allelic imbalance in microdissected areas can be used as a measure of clonality within an area of tissue studied, because the sampling includes hundreds of cells. According to Knudsons hypothesis of tumor suppressor genes,21 these measurable DNA losstype mutations are presumed to be accompanied by an immeasurable mutation (point mutation, methylation mutation, etc) in the copy of the gene that remains. The presence of allelic imbalance of tumor suppressor genes in the majority of tumors studied is strong evidence that these lesions are neoplastic, rather than developmental anomalies. A neoplastic origin explains the high recurrence rate, the infrequent transformation to squamous cell carcinoma, and the association with the known carcinogenic syndrome, NBCCS. The association with daughter cysts and higher FAL in our series is intriguing and suggests the possibility that daughter cysts are histologic evidence of extension or local invasion of the lesional tissue. Our results also conrm the presence of allelic loss in the PTCH gene, which was demonstrated in a prior study of these lesions. In conclusion, this study shows that there is signicant loss of heterozygosity of tumor suppressor genes in sporadic OKCs. The presence of these allelic losses lends substantial support to the theory that OKCs are neoplastic in origin, rather than developmental. Because our study had only one case of recurrence, additional work is needed to determine whether the pattern or number of tumor suppressor genes with allelic loss can predict behavior of the tumors. Furthermore, sporadic tumors can be compared to those arising in patients with NBCCS to determine if the genotyping prole suggests any differences in pathogenesis.
References
1. Shear M. The aggressive nature of the odontogenic keratocyst: is it a benign cystic neoplasm? Part 1: clinical and early experimental evidence of aggressive behaviour. Oral Oncol. 2002;38:219226. 2. Shear M. The aggressive nature of the odontogenic keratocyst: is it a benign cystic neoplasm? Part 2: proliferation and genetic studies. Oral Oncol. 2002;38: 323331. 3. Blanas N, Freund B, Schwartz M, Furst IM. Systematic review of the treatment and prognosis of the odontogenic keratocyst. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2000;90:553558. 4. Makowski GJ, McGuff S, Van Sickels JE. Squamous cell carcinoma in a maxillary odontogenic keratocyst. J Oral Maxillofac Surg. 2001;59:7680. 5. Anand VK, Arrowood JP Jr, Krolls SO. Malignant potential of the odontogenic keratocyst. Otolaryngol Head Neck Surg. 1994;111:124129. 6. Regezi JA. Odontogenic cysts, odontogenic tumors, broosseous, and giant cell lesions of the jaws. Mod Pathol. 2002;15:331341.

Neoplastic Status of Odontogenic KeratocystsAgaram et al

7. Chow HT. Odontogenic keratocyst: a clinical experience in Singapore. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1998;86:573577. 8. Barreto DC, Chimenos Kustner E. New considerations about the diagnosis of odontogenic keratocyst. Med Oral. 2001;6:350357. 9. Barreto DC, Gomez RS, Bale AE, Boson WL, De Marco L. PTCH gene mutations in odontogenic keratocysts. J Dent Res. 2000;79:14181422. 10. Lench NJ, High AS, Markham AF, Hume WJ, Robinson PA. Investigation of chromosome 9q22.3-q31 DNA marker loss in odontogenic keratocysts. Eur J Cancer B Oral Oncol. 1996;32B:202206. 11. Thiagalingam S, Foy RL, Cheng KH, Lee HJ, Thiagalingam A, Ponte JF. Loss of heterozygosity as a predictor to map tumor suppressor genes in cancer: molecular basis of its occurrence. Curr Opin Oncol. 2002;14:6572. 12. Philipsen HP. Om Keratocyster (kolesteatoma) I Kaeberbe. Tandlaegebladet. 1956. 13. Meara JG, Shah S, Li KK, Cunningham MJ. The odontogenic keratocyst: a 20-year clinicopathologic review. Laryngoscope. 1998;108:280283. 14. Myoung H, Hong SP, Hong SD, et al. Odontogenic keratocyst: review of

256 cases for recurrence and clinicopathologic parameters. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2001;91:328333. 15. Cohen MM Jr. Nevoid basal cell carcinoma syndrome: molecular biology and new hypotheses. Int J Oral Maxillofac Surg. 1999;28:216223. 16. Wicking C, Bale AE. Molecular basis of the nevoid basal cell carcinoma syndrome. Curr Opin Pediatr. 1997;9:630635. 17. Kubota Y, Oka S, Nakagawa S, Shirasuna K. Interleukin-1alpha enhances type I collagen-induced activation of matrix metalloproteinase-2 in odontogenic keratocyst broblasts. J Dent Res. 2002;81:2327. 18. Oliveira MD, Souza LB, Pinto LP, Freitas RA. Immunohistochemical study of components of the basement membrane in odontogenic cysts [in Portuguese]. Pesqui Odontol Bras. 2002;16:157162. 19. de Paula AM, Carvalhais JN, Domingues MG, Barreto DC, Mesquita RA. Cell proliferation markers in the odontogenic keratocyst: effect of inammation. J Oral Pathol Med. 2000;29:477482. 20. Piattelli A, Fioroni M, Santinelli A, Rubini C. P53 protein expression in odontogenic cysts. J Endod. 2001;27:459461. 21. Knudson AG Jr. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci U S A. 1971;68:820823.

Arch Pathol Lab MedVol 128, March 2004

Neoplastic Status of Odontogenic KeratocystsAgaram et al 317