Oncogene (2002) 21, 664 ± 673 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc
Analysis of telomerase as a diagnostic biomarker of cervical dysplasia and carcinoma Elke A Jarboe1, Kai-Li Liaw3, L Chesney Thompson2, David E Heinz1, Paige L Baker1, Jamie A McGregor4, Terry Dunn4, Jan E Woods5 and Kenneth R Shroyer*,1 1
Department of Pathology, University of Colorado Health Sciences Center, Denver, Colorado, CO 80262, USA; 2Department of Obstetrics and Gynecology, University of Colorado Health Sciences Center, Denver, Colorado, CO 80262, USA; 3Department of Epidemiology, Merck Research Laboratories, West Point, Pennsylvania, PA 19486, USA; 4Department of Obstetrics and Gynecology, Denver Health Medical Center, Denver, Colorado, CO 80204, USA; 5Department of Pathology, Denver Health Medical Center, Denver, Colorado, CO 80204, USA
Telomerase expression is a potentially important marker of high-grade cervical dysplasia and squamous cell carcinoma (SCC). The routine practice of cervical cytology is limited by problems of false negative diagnoses as well as by poor speci®city for clinically signi®cant lesions in patients with low-grade cytologic abnormalities. Telomerase is widely expressed in most SCCs as well as in a high proportion of high-grade squamous intraepithelial lesions. Histochemical studies have con®rmed that telomerase is expressed in the lower portions of normal or metaplastic squamous mucosa but that telomerase positive cells extend into the upper epithelial layers in cases of high-grade dysplasia. Since the cervical smear samples the uppermost cell layers of the cervical mucosa, but does not normally include cells derived from the lower layers of the squamous mucosa, the detection of telomerase in exfoliated cells of the cervical smear may have speci®city for clinically signi®cant lesions. The analysis of hTR, hTERT, and telomerase activity are complicated by a number of technical factors that may lead to either false negative or false positive test results. Thus, the practical application of telomerase analysis as a diagnostic adjunct for cervical cytopathology may depend on the development of more reliable and sensitive assay systems, possibly formatted for cytochemical applications. Oncogene (2002) 21, 664 ± 673. DOI: 10.1038/sj/onc/ 1205073 Keywords: telomerase; hTR; hTERT; TRAP; cervix neoplasia Introduction The diagnosis of cervical dysplasia and carcinoma is based on both the objective evaluation of well-de®ned microscopic features and on the more subjective analysis of often subtle parameters to result in a ®nal
*Correspondence: KR Shroyer, Department of Pathology, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Denver, CO 80262, USA; E-mail: [email protected]
diagnosis. The subjective component of this process results in intra-observer and inter-observer diagnostic variability that may compromise patient care and increase liability and costs. As a result, there has been intense interest in the development of diagnostic adjunctive techniques to help enhance the accuracy of the objective component of the diagnostic processes. Telomerase is a fundamental marker of neoplastic transformation that is widely expressed in both premalignant intraepithelial lesions and in most malignant lesions of the uterine cervix and therefore, is a potential candidate for this purpose. Background Cervical cancer is the most common fatal malignancy of women in many developing countries. In the United States and other industrialized nations, however, cervical neoplasia currently ranks about tenth, due in large part to the success of cervical cytology as a screening program for the detection of intramucosal precursor lesions (Cancer Facts and Figures, 2001; Guidozzi, 1996; Birley, 1995; Tkeshelashvili et al., 1993). Invasive squamous cell carcinoma (SCC) comprises 75 ± 77% of cases of malignant epithelial neoplasms of the cervix. Less common histologic types include endocervical adenocarcinomas, adenosquamous carcinomas, mucoepidermoid carcinomas, neuroendocrine tumors, and other rare types (Wright et al., 1994a). Although these tumors are important causes of patient morbidity and mortality, the potential role for biomarkers as diagnostic adjuncts of these tumors is beyond the scope of the current review. Ideally, however, innovative adjunctive means of detecting cervical squamous neoplasia would be of assistance in the diagnosis of these tumors as well. Cytologic diagnosis of cervical SCC and precursor lesions SCC is the most thoroughly investigated of cervical cancers and the pre-neoplastic spectrum of its
Telomerase in cervical dysplasia and cancer EA Jarboe et al
associated precursor lesions is relatively well de®ned. Malignant precursor lesions of the cervical mucosa are detected by cytologic examination of the Papanicolaou smear (Pap smear). In the United States, approximately 50 million Pap smears are performed yearly (DeMay, 1996b). Cytologic ®ndings are classi®ed by the Bethesda system as normal/benign reactive changes, squamous cell abnormalities, and glandular cell abnormalities (Kurman and Solomon, 1994). Squamous cell abnormalities include atypical squamous cells of undetermined signi®cance (ASCUS), low grade squamous intraepithelial lesions (LSILs), encompassing evidence of human papillomavirus (HPV) infection and/or mild dysplasia, and high grade intraepithelial lesions (HSILs), including moderate dysplasia (CIN II) and severe dysplasia/carcinoma in situ (CIN III) (Kurman and Solomon, 1994). A revision of the Bethesda system of cytologic classi®cation will be released in the fall of 2001. Routine cervical cytology is the most eective screening test for cancer that has ever been devised. Despite its dramatic successes, however, the practice of cervical cytopathology has come under increasing criticism due both to problems of false negative diagnoses for HSIL and SCC and because of poor speci®city for clinically signi®cant lesions in cases that are diagnosed as ASCUS or LSIL. False negative Pap smears may be a consequence of sampling errors, screening errors, and errors of interpretation (where cytologically abnormal cells were present on the microscope slide but were misdiagnosed by the cytopathologist as normal). Reports of false negative rates in cervical cytology vary widely, from 1.6 to almost 28%, in the diagnosis of cervical cytology specimens (Naryshkin, 1997; Krieger and Naryshkin, 1994; Tabbara and Sidawy, 1996). One source of false negative results in diagnostic cervical cytology is the presence of only rare abnormal cells in some Pap smears. Although Pap smears may contain 500 000 or more epithelial cells, some cervical lesions may exfoliate less than 100 cytologically abnormal cells (DeMay, 1996a,b). Furthermore, cervical smears from patients with invasive cervical carcinoma may have false negative cytologic diagnoses due to the presence of only rare abnormal cells, obscuring necrotic debris, in¯ammation, or bleeding (DeMay, 1996a,b; Benoit et al., 1984; Berkeley et al., 1980; Gay et al., 1985; Pairwuti, 1991; Richart, 1993; Robertson and Woodend, 1993). As a result, cervical cytology may have a false negative rate of 50 ± 75% in patients with SCC (Gay et al., 1985; Pairwuti, 1991; van der Graaf et al., 1987; van der Graaf and Vooijs, 1987). The clinical management of patients with ASCUS or LSIL remains unsettled and is frequently expensive. Although a signi®cant portion of patients with these low grade abnormalities of the cervical smear have an underlying HSIL on cervical biopsy, the majority of these patients do not have clinically signi®cant lesions (Wright et al., 1994b, 1995a; Hatch et al., 1995). However, essentially all patients with ASCUS or LSIL on a screening Pap smear require further evaluation by
repeat Pap smear or colposcopic examination to rule out the possibility of underlying HSIL. The application of molecular adjunctive tests could provide enhanced objective bases for triage of patients with clinically signi®cant disease or reduce unnecessary procedures and expenditures.
Human papillomavirus in cervical carcinoma and precursor lesions HPV infection is associated with the overwhelming majority of cases of invasive SCC and has also been found in a high proportion of cases of LSIL and HSIL. In addition, the majority of cases of endocervical adenocarcinoma are associated with HPV infection. Over 100 dierent HPV types have now been characterized based on nucleotide sequence dierences of the viral genome, including over 40 that have been found to infect the cervical mucosa (zur Hausen, 1996). Epidemiologic studies have divided HPVs into `low risk' types, including HPV types 6, 11, 42, 43 and 44 and `high risk' types, including HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68, that have been associated with a higher risk for malignant transformation (zur Hausen, 1996). HPV oncoproteins are directly involved in the pathogenesis of cervical dysplasia and carcinoma, E6 through its interaction with the tumorsuppressor cellular protein p53 and E7 through interaction with hypophosphorylated Rb proteins p107 and p130 (Wolf and Ramirez, 2001; zur Hausen, 1996, 1999). HPV DNA detection has been extensively evaluated for the triage of patients with low grade cytologic abnormalities in order to identify patients that are at greatest risk for underlying HSIL or SCC. Recent data from the national ALTS (ASCUS/LSIL Triage Study) trial indicated that HPV detection was not useful for the triage of patients with LSIL due to the high rate of positivity in that cytologic subgroup of patients (Manos et al., 1999). Although only about 24% of LSIL cases under the age of 35 have an underlying HSIL, 83% of this group tested positive for HPV by the Hybrid Capture II assay. By contrast, a negative HPV test result excluded underlying HSIL in over 99% of women with ASCUS. A positive test result, however, did not reliably predict clinically signi®cant disease as demonstrated by the presence of HPV in over a third of cases that were negative for HSIL or SCC on colposcopic biopsy (Solomon et al., 2001). Telomerase in HPV-mediated malignant transformation Activation of telomerase expression may be a central mechanism by which HPV infection results in malignant transformation of the cervical mucosa. Anderson et al. (1997) evaluated cervical biopsies by L1 consensus sequence PCR and found HPV 16 DNA in three and HPV 18 DNA in four of 10 telomerase positive cases of cervical carcinoma. More recently, Oncogene
Telomerase in cervical dysplasia and cancer EA Jarboe et al
using GP5/GP6 consensus sequence PCR, Yashima et al. (1998) detected HPV types 16/18 in 11 of 13 HSIL cervical biopsies, including four biopsies that were positive for both telomerase expression and HPV DNA. Furthermore, the expression of telomerase correlated with the intensity of HPV 16/18 E6/E7 oncogene expression, as determined by in situ hybridization in high grade dysplasia and SCC (Riethdorf et al., 2001). In cultured human keratinocytes, the E6 oncoprotein of high risk HPVs activates telomerase expression through a p53 independent mechanism (Klingelhutz et al., 1996; Stoppler et al., 1997). Telomerase activation appears to result from E6 interaction with the hTERT promoter, proximal to the ATG initiation codon (Oh et al., 2001). Coexpression of both E6 and E7 oncoproteins results in a higher level of telomerase expression than in keratinocytes that express only E6 (Oh et al., 2001). Telomerase expression is a marker of cervical dysplasia and carcinoma There is abundant laboratory evidence to support the conclusion that telomerase expression contributes to cellular immortalization, as a fundamental component of the processes of malignant transformation (reviewed in Shay et al., 2001). The role of telomerase expression in premalignant or dysplastic processes, however, is less well de®ned. Carcinomas are composed of clonal populations of malignant epithelial cells that have gained the ability to invade surrounding tissues and to penetrate lymphovascular spaces. The nuclear and cytologic features of carcinoma cells, however, are similar to those of dysplastic cells that comprise their associated precursor lesions. Molecular studies have demonstrated that most HSILs, as well as a signi®cant proportion of low grade dysplasias, are composed of clonal cell populations, consistent with the concept that they result from the intraepithelial proliferation of neoplastic cell populations (Park et al., 1995, 1996; Enomoto et al., 1997). The mitotic rates of HSILs re¯ect the arrest of dierentiation of the dysplastic squamous mucosa and focally equal that which is observed in associated SCCs. If telomerase expression is required to support an unlimited replicative phenotype, telomerase should be expressed in both SCCs and in high grade dysplastic lesions of the cervical mucosa. If on the other hand, telomerase expression plays some fundamental role in phenotypic features that distinguish malignant from premalignant cell populations, telomerase would not be predicted to be activated in HSILs. The analysis of telomerase as a potential biomarker of cervical dysplasia has been a focus of intense study for several years. The telomerase holoenzyme consists of an RNA component, designated hTR, which contains a template region that is complementary to the telomeric DNA repeat (TTAGGG)n , and a single catalytic protein subunit, designated hTERT (Kim et al., 1994; Feng et al., 1995; Nakamura et al., 1997;
Meyerson et al., 1997). Takakura et al. (1998) found frequent correlation between the detection of telomerase activity and the detection of hTERT mRNA by reverse transcriptase ± PCR (RT ± PCR). Among the 23 cervical cancers that were evaluated for both telomerase activity and hTERT, 18 (78%) of cases showed concordant results but ®ve (22%) were discordant, including three cases that tested positive for TERT but were negative for telomerase activity and two cases that tested negative for hTERT but paradoxically, were positive for telomerase activity. By contrast, tests for telomerase activity and hTERT were uniformly negative in all 14 samples of normal cervical mucosa. Although both telomerase activity and hTERT mRNA were associated with cervical cancer, hTR was also detected in benign cervical mucosa (Takakura et al., 1998; Wisman et al., 2000, 2001). Thus, the detection of hTR by RT ± PCR appears to be non-speci®c for clinically signi®cant cervical disease. Similarly, the telomerase-associated protein (TP1/TLP1) has been found to be widely expressed in both cervical cancer and in normal cervical mucosa and is therefore, not a potential biomarker for lesions of the cervical mucosa (Takakura et al., 1998). The evaluation of hTERT RNA by RT ± PCR is complicated by the existence of several splice variants, including four insertion variants and two deletion site variants (Yi et al., 2000; Ulaner et al., 1998; Colgin et al., 2000; Kilian et al., 1997; Wick et al., 1999). The bdeletion variant and all four of the insertion variants result in premature termination of the hTERT translation products (Yi et al., 2000; Ulaner et al., 1998). The protein product of the a-deletion variant may be a dominant negative inhibitor of telomerase expression and its expression in transfected cell lines has been reported to result in an 85 ± 95% reduction in the endogenous expression of telomerase activity (Colgin et al., 2000; Yi et al., 2000). Thus, expression of the a-deletion variant could result in discordance between telomerase activity levels and the levels of hTERT mRNAs. Most approaches for hTERT RT ± PCR result in the ampli®cation of not only the fulllength hTERT transcript but also all six of the known insertion/deletion variants. Although multiplex RT ± PCR protocols could be designed to encompass fulllength hTERT mRNA and all splice variants, quantitation of the relative levels of expression of the individual splice variants may be impractical in a clinical laboratory diagnostic setting. As a result, hTERT RT ± PCR may not necessarily correlate with telomerase activity levels or re¯ect underlying cervical neoplasia. Histologic localization of telomerase Marked dierences in the distribution of hTR in benign versus malignant cervical tissues have been found by histochemical approaches. In situ hybridization has demonstrated that hTR is largely restricted to the basal and parabasal cell layers of ectocervical
Telomerase in cervical dysplasia and cancer EA Jarboe et al
mucosa, metaplastic squamous epithelium, or LSIL, although focal expression has also been observed in some histologically normal endocervical glands (Yashima et al., 1998; Frost et al., 2000). By contrast, hTR was detected throughout all layers of the lesional mucosa in moderate dysplasia, severe dysplasia, and in carcinoma in situ (Yashima et al., 1998; Frost et al., 2000; Wisman et al., 2000). Nakano et al. (1998) demonstrated that hTR and hTERT mRNA are colocalized in cervical SCCs as well as in benign and dysplastic cervical mucosa biopsy specimens. In contrast to the report from Yashima et al. (1998) that found hTR in lymphoid follicles but not in stromal lymphocytes, however, Nakano et al (1998) noted a positive signal for both hTR and hTERT mRNA in stromal lymphocytes. The expression of hTR appears to be heterogeneously distributed in SCCs and is related to the degree of tumor cell dierentiation. In general, terminally dierentiated components of SCCs, including squamous pearls, are negative for hTR (Yashima et al., 1998; Frost et al., 2000) and the distribution of hTERT mRNA has also been shown to have regional variability in some cases (Nakano et al., 1998). Thus, tumor heterogeneity in telomerase expression could in¯uence the analysis of telomerase. The immunohistochemical localization of the hTERT protein has been challenging due to the low copy number of expression in most tissues and due to limited availability of high titer speci®c antibodies. Recent studies have detected the hTERT protein in archival formalin-®xed cervical mucosal biopsy specimens using a highly sensitive tyramide-based indirect immunoperoxidase method (Frost et al., 2000). Less sensitive methods have failed to reveal speci®c localization of telomerase in our hands, although other groups have reported success in some tissues using conventional avidin ± biotin-based immunoperoxidase methods (Hiyama et al., 2001). Studies in our laboratory showed that hTERT protein is predominantly localized to lower portions of the suprabasal squamous cells of normal or benign-reactive cervical mucosa (Frost et al., 2000). By contrast, almost fullthickness distribution of nuclear staining was observed in nearly all cases of moderate and severe dysplasia (Frost et al., 2000). In SCC, there was a paradoxically decreased intensity of hTERT nuclear staining and hTERT protein appeared to be diusely distributed throughout the cells, rather than showing the discrete nuclear pattern of localization that was characteristic of benign and premalignant cervical tissues (Frost et al., 2000). Thus, malignant transformation of the cervical squamous mucosa may be associated with disruption of normal hTERT nuclear translocation processes. TRAP assay for telomerase The TRAP assay remains the gold standard for the evaluation of telomerase expression in clinical samples. The TRAP assay is an extraordinarily sensitive and
speci®c PCR-based functional enzyme assay. Telomerase, present in crude protein extracts of cellular or tissue samples catalyzes the repetitive addition of six base telomeric repeat sequences (TTAGGG) onto the 3' end of the substrate oliogonucleotide (TS). The extension products are subsequently ampli®ed by PCR and are detected by either polyacrylamide gel electrophoresis or less commonly, by hybridization with probes that are complementary to the TTAGGG sequence, followed by ELISA (Wu et al., 2000). Both isotopic and non-isotopic detection formats are widely used for the analysis of telomerase expression. The most commonly used formats, however, rely on the incorporation of 32P end-labeled TS primers and visualization of the six base pair ampli®cation products by autoradiography or phosphorimaging (Figure 1). The original TRAP assay, as described by Kim et al. (1994) was reported to detect telomerase activity in as few as 100 cancer cells in a background of up to 106 normal somatic cells. Subsequent modi®cations of the TRAP protocol have included utilization of an internal assay standard that permits semi-quantitative analysis of the telomerase products and can also be used to evaluate for the presence of PCR inhibitors that could result in false negative TRAP results (Wright et al., 1995b). The sensitivity of the TRAP assay can now be routinely extended to 10 HeLa cell equivalents, using commercially available systems (Figure 2).
Telomerase expression in cervical biopsies Although numerous studies have shown that telomerase expression can be detected by TRAP assay in most
Figure 1 TRAP assay for telomerase. Serial dilutions of HeLa cells, 10, 102, and 103 cell equivalents; lanes A-G, cervical cytology specimens; lane H, reagent negative control; Arrow, internal telomerase assay standard. Phosphorimager analysis (Storm 860, Molecular Dynamics). Lanes A and B showed faint non-speci®c ampli®cation patterns including primer dimer bands. HeLa cell positive controls and lanes C ± G demonstrated progressive six base pair ampli®cation products, consistent with telomerase activity Oncogene
Telomerase in cervical dysplasia and cancer EA Jarboe et al
cervical carcinomas, the proportion of cases of HSIL, LSIL, and benign cervical mucosa that have tested positive for telomerase activity has varied widely between dierent studies (Kyo et al., 1996; Anderson et al., 1997; Pao et al., 1997; Mutirangura et al., 1998; Shroyer et al., 1998; Snijders et al., 1998; Wisman et al., 1998; Yashima et al., 1998; Yokoyama et al., 1998; Zhang et al., 1999) (Table 1). These dierences are undoubtedly related, at least in part, to variability in the TRAP assay protocols but other factors could also have been involved. Most studies reported a low rate of telomerase detection in samples of normal cervical mucosa with only 4/107 (3.7%) of normal cervical biopsies that were TRAP positive among ®ve published studies (Pao et al., 1997; Kawai et al., 1998; Mutirangura et al., 1998; Snijders et al., 1998; Wisman et al., 1998; Zhang et al., 1999). By contrast, two other studies reported TRAP positivity in a much higher proportion of normal specimens, including a total of 27/169 (16%) cervical biopsies (Shroyer et al., 1998;
Figure 2 TRAP assay of HeLa cells, serial dilution. Telomerase extension products were evaluated by polyacrylamide gel electrophoresis and phosphorimager analysis (ImageQuant version 5.2, Molecular Dynamics). Results are expressed as the ratio of total product generated/internal telomerase ampli®cation standard (TPG/ITAS). Values represent summary data of six separate experiments
Yashima et al., 1998). Based on histochemical observations of the presence of hTR and hTERT protein in the lower portions of the normal cervical mucosa, it would be predicted that a much higher proportion of normal cervical biopsy specimens should test positive for telomerase activity than has been observed in most studies. Negative TRAP assay results in normal biopsy specimens must, therefore, re¯ect either ineciency of extraction, the presence of telomerase or PCR inhibitors, or the expression of telomerase in the basal and parabasal cell populations at levels that are below the threshold of detection. In review of published studies, it is apparent that there is a relationship between telomerase test results in HSILs and normal biopsy specimens. The two studies that had the highest reported rates of detection of telomerase in normal cervical biopsies also reported the highest proportion of cases of HSIL that tested positive for telomerase (Table 1). Among studies with a low rate of telomerase detection in normal cervical biopsies, telomerase activity was detected in only 31/93 (33%) of HSIL biopsies. By contrast, for the two studies that found the highest rates of telomerase detection in normal biopsies, telomerase was detected in 38/47 (81%) of HSILs. Thus, it appears that telomerase is widely expressed in a very high proportion of HSILs, but that the proportion of cases that test positive may be in¯uenced by the test sensitivity, eciency of tissue extraction, or other technical factors. The observation that telomerase activity is commonly expressed in both HSILs and in cervical SCCs indicates that telomerase is required to support an unlimited replicative phenotype of either intramucosal or invasive neoplastic epithelial cells. Therefore, the activation of telomerase expression is an early event in the processes that result in malignant transformation of the cervical mucosa. Several studies have reported that there are quantitative dierences in the level of telomerase expression in premalignant versus malignant lesions of the cervical mucosa (Kyo et al., 1996, 1998; Zheng et al., 1997; Nagai et al., 1999). These studies have found that among telomerase positive cases, there are relatively low levels of expression in normal cervical mucosa, higher levels in cervical dysplasia, and much higher levels in invasive SCC. The quantitation of
Table 1 Detection of telomerase expression by TRAP in cervical tissues, summary of representative reports Histologic diagnosis Kyo et al. (1996) Anderson et al. (1997) Pao et al. (1997) Kawai et al. (1998) Mutirangura et al. (1998) Shroyer et al. (1998) Snijders et al. (1998) Wisman et al. (1998) Yashima et al. (1998) Yokoyama et al. (1998) Zhang et al. (1999) Oncogene
0/35 (0%) 1/7 (14%) 1/9 (11%) 9/50 (18%) 0/8 (0%) 0/21 (0%) 18/138 (13%) 2/27 (7.4%)
2/6 (33%) 14/25 (56%) 0/10 (0%) 0/2 (0%) 7/21 (33%)
8/20 (40%) 25/26 (96%) 6/23 (26%) 12/43 (28%) 12/21 (62%)
10/12 (83%) 10/10 (100%) 22/24 (91.7%) 12/13 (92%) 33/34 (97%) 18/18 (100%) 23/24 (96%) 18/21 (86%) 19/20 (95%) 46/50 (92%)
Telomerase in cervical dysplasia and cancer EA Jarboe et al
telomerase activity in clinical samples, however, may be of limited signi®cance unless the proportion of lesional epithelial cells is de®ned in the test samples. Cervical biopsies with dysplasia often include benign epithelial components, including normal ectocervical squamous cells, metaplastic squamous cells, endocervical cells, and numerous other cellular components that are not an intrinsic component of the dysplastic process. By contrast, biopsies of cervical SCC usually are composed of a relatively higher proportion of lesional epithelial cells. Thus, quantitative dierences in telomerase expression levels must be standardized to the number of lesional cells that were included in any ¯uid-based assay. For the evaluation of cervical biopsy specimens, this would require utilization of highly speci®c microdissection techniques, and normalization of test results to the number of lesional cells that are tested. Telomerase expression in cervical cytology A rapidly expanding body of data suggests that the detection of telomerase expression could play a useful role as a diagnostic adjunct in the practice of cervical cytopathology. Studies of normal cervical smears have shown telomerase expression in from 0 to 11% of samples (Table 2). With increasing severity of cytologic abnormality, there appears to be a corresponding increase rate of detection of telomerase activity. Although Gorham et al. (1997) and Wisman et al. (1998) found telomerase activity in only a relatively small proportion of cases of HSIL and SCC, several other published studies have shown the detection of telomerase activity in from 46 to 83% of cases of HSIL and in from 88 to 100% of cases of SCC (Table 2). The majority of these studies of telomerase expression in cervical cytology samples have utilized modi®cations of the original approach described by Kim et al. (1994). These modi®cations included variability in the number of cycles of PCR ampli®cation, the use of isotopic versus non-isotopic, and variability in the protein concentration of the assay sample (Gorham et al., 1997; Yashima et al., 1998; Zheng et al., 1997; Iwasaka et al., 1998; Kyo et al., 1997). In addition, some studies used an assay system that involved fullycompetitive co-ampli®cation of a 150 bp internal telomerase assay standard (ITAS) (Yashima et al.,
Table 2 Cytologic diagnosis Gorham et al. (1997) Iwasaka et al. (1998) Kyo et al. (1997) Wisman et al. (1998) Kawai et al. (1998) Yashima et al. (1998) Zheng et al. (1997) Reddy et al. (2001)
1998). The utilization of an ITAS that fully competes with ampli®cation of the telomerase extension product may decrease the sensitivity of the telomerase assay. Excluding the study by Zheng et al. (1997) none of these studies de®ned the sensitivity of the TRAP assay, as applied in their study. Furthermore, the lesional or epithelial cell component of the test samples were generally unde®ned.
Sources of error in telomerase test results Various factors may result in discordance between the results of telomerase analysis and the ®nal clinicopathologic diagnosis (Table 3). Cervical cancer tissue biopsies generally include a high proportion of lesional epithelial cells and as a result, variability in the eciency of extraction and in the sensitivity of the assay is unlikely to result in false negative test results. By contrast, dysplastic lesions of the cervix may consist of only minute foci of abnormal cells and often represent only a small component of the total biopsy sample. The processes of tissue extraction for telomerase analysis disrupt the architectural integrity of the biopsy specimen and in general, the precise morphologic diagnosis of tissues that are processed for telomerase analysis is not possible. Although some papers have reported successful detection of telomerase activity from frozen sections, which facilitates a precise morphologic correlation (Wisman et al., 1998), the use of frozen sections results in air-drying of the histologic preparation and could result in loss of telomerase activity. As an alternative approach, most investigators
Table 3 Factors that may interfere with telomerase detection as a biomarker for cervical neoplasia False negative results
False positive results
1. Sampling errors or sample degradation 2. Incomplete telomerase extraction 3. Presence of only rare lesional cells 4. False positive cytologic/ histologic diagnosis 5. Telomerase degradation 6. Telomerase inhibitors 7. PCR inhibitors
1. Follicular cervicitis 2. Presence of endometrial cells 3. False negative cytologic/ histologic diagnosis
Detection of telomerase expression by TRAP in cervical smears, summary of representative reports Normal
0/10 (0%) 5/62 (8%) 3/33 (9%) 1/9 (11%) 2/36 (6%) 13/119 (11%) 3/41 (7%) 0/18 (0%)
7/27 (26%) 5/8 (63%) 3/26 (12%) 10/32 (31%) 4/17 (24%) 4/13 (31%) 1/2 (50%)
1/22 (4.5%) 19/36 (53%) 14/24 (58%) 26/97 (27%) 19/26 (73%) 6/13 (46%) 6/12 (50%) 10/12 (83%)
1/2 (50%) 29/30 (97%) 15/17 (88%) 4/13 (31%) 7/8 (88%) 20/20 (100%) Oncogene
Telomerase in cervical dysplasia and cancer EA Jarboe et al
have utilized split samples, where one portion of a tissue fragment is processed for histologic assessment and a corresponding portion is tested for telomerase expression. Under ideal circumstances, telomerase expression re¯ects the histologic diagnosis in the mirror image specimen that was processed for microscopic evaluation. The lesional epithelium in the portion of tissue that was processed for microscopic examination, however, is not necessarily representative of the sample that was tested for telomerase activity. One important cause of false negative results for telomerase expression may be inecient extraction of activity from the biopsy or cytology specimen. Cervical biopsy specimens consist of a relatively thin layer of mucosal epithelial cells, overlying a densely collagenized ®brous stroma. Although relatively non-keratinized malignant epithelial cells may be readily lysed by mechanical disruption in the presence of non-ionic detergents such as CHAPS buer or NP-40, the keratinized epithelium of a cervical biopsy appears to be highly resistant to mechanical disruption. Studies in our laboratory found a low rate of detection of telomerase activity from benign and malignant cervical biopsies that were subjected to mechanical disruption and extraction with CHAPS or NP-40 buer. However, a high proportion of cases of cervical dysplasia tested positive for telomerase when the tissues were crushed under liquid nitrogen, prior to extraction by conventional methods. Using this approach, telomerase activity was detected by TRAP assay in 25 of 26 (96%) samples of HSIL, including 10 of 10 samples of moderate dysplasia, 12 of 13 samples of severe dysplasia, and three of three samples of carcinoma in situ (Shroyer et al., 1998). Telomerase activity was also detected in 14 of 25 cases of LSIL (56%), and 10 of 18 (56%) of samples with reactive atypia. The expression of telomerase however, was not strictly limited to potentially premalignant epithelial lesions. In our study, nine of 50 (18%) samples of histologically normal squamous mucosa tested positive for telomerase activity, including three of seven cases that were suggestive but not diagnostic of HPV, six of 40 (15%) samples of histologically unremarkable squamous mucosa and 0 of three samples of normal endocervical mucosa (Shroyer et al., 1998). The detection of telomerase activity in a relatively high proportion of morphologically normal cervical biopsy specimens could result from the presence of basal and parabasal cells. Less ecient methods that do not result in complete disruption of the full thickness of the cervical mucosal epithelium, however, may not detect telomerase activity derived from the deepest layers of the epithelium. The Pap smear is a complex biologic sample that includes not only super®cial and intermediate squamous cells but also variable numbers of endocervical cells, endometrial cells, acute and chronic in¯ammatory cells, red blood cells and various microbes. The proportion of lesional cells is unpredictable, cannot be standardized, and is not necessarily re¯ective of the extent or severity of disease. Many cases of cervical
dysplasia include abnormal cells with prominent nuclear and cytoplasmic aberrations that are readily recognized in the conventional cervical smear. In a high proportion of cases, however, the lesional cell component is either sparse or is manifest by subtle cytologic features that may render diagnosis dicult. Telomerase analysis has the potential to play an important role as a diagnostic adjunct in cases that have only rare lesional cells that could be missed by routine microscopic examination. However, in those cases, it may be dicult to dierentiate between the possibility of a false positive telomerase result versus a false negative cytologic diagnosis. In colposcopy cases where a cervical biopsy is available, correlation of telomerase results with the histopathologic diagnosis may help de®ne basis of discordance between telomerase and cytologic test results. The cervical biopsy itself, however, may also be subject to false negative diagnoses due to the presence of equivocal histologic freatures or due to colposcopic failure to sample the appropriate lesional area of the cervical mucosa. Thus, there is no absolute gold standard to allow evaluation of the sensitivity and speci®city of any potential biomarker of cervical dysplasia or carcinoma. The ®nal evaluation of telomerase test performance must rely on the synthesis of all available data, including test parameters, cytologic diagnosis, histologic diagnosis, and clinical outcome correlation. The presence of endometrial cells or tubal epithelial cells from premenopausal women or activated lymphocytes from patients with follicular cervicitis could result in false positive test results (Brien et al., 1997; Kyo et al., 1997; Gorham et al., 1997; Shroyer et al., 1997; Yashima et al., 1998; Yokoyama et al., 1998). Although the proportion of endometrial cells in the cervical smear after day 10 of the normal menstrual phase is generally negligible, at earlier stages in the cycle, it is possible that contamination of the cervical sample by endometrial cells could result in the detection of telomerase activity in the absence of cytologic abnormalities. Review of cervical biopsies from patients with telomerase positive but cytologically negative cervical smears often reveals follicular cervicitis. In those cases, the activated B-cell component of the lymphoid follicles is the presumptive source of telomerase activity. Clinical samples may be susceptible to degradation of telomerase activity, which could lead to false negative results by TRAP. The cervical mucosa is normally treated with acetic acid at the time of colposcopy to highlight areas of the cervical mucosa that are suspicious for intraepithelial neoplasia. This treatment, however, can interfere with the detection of telomerase activity by the TRAP smear, possibly as a result of denaturation of the telomerase holoenzyme complex (Chang Chien et al., 1998; Gullu and Kurdoglu, 1999). Thus, it is important that cytologic specimens be collected from the mucosal surface prior to the application of acetic acid. Ideally, the cervical biopsy should also be collected prior to the application of acetic acid although this approach could interfere
Telomerase in cervical dysplasia and cancer EA Jarboe et al
with the identi®cation of the most clinically signi®cant lesions by colposcopic examination. The cervical cytologic specimen could contain inhibitors of telomerase activity that could lead to false negative test results. The addition of EDTA to the collection buer stabilizes telomerase activity in urinary cytology specimens (Yahata et al., 1999). In addition, resin extraction of the telomerase extension products, prior to PCR results in an increase in the proportion of urine specimens that have detectable telomerase activity (Yahata et al., 1999). However, resin extraction of residual TS extension products from HSIL Pap smears did not detect telomerase activity in cases that were previously TRAP negative (unpublished personal observations). Thus, false negative TRAP results in cervical cytologic specimens may re¯ect loss of telomerase activity prior to initial processing for the TRAP assay. Serial incubation studies in our lab have shown that telomerase activity is stable in HeLa cells stored in wash buer at 48C for up to 5 days, although there is a noticeable decrease in signal intensity after 7 days incubation. Since the time since desquamation of cervical cells is not known, it is likely that the stability of clinical samples is less than that of the ideal observed in cell culture models. As a result, the rate of telomerase degradation in clinical samples cannot be predicted. Attempts to correlate telomerase activity with the cytologic diagnosis and the ®nal histopathologic diagnosis may be confounded by several diagnostic factors. For reasons that are not well understood, the Pap smear that is collected at the time of colposcopic examination has a higher false negative rate than the referral Pap smear that revealed the initial cytologic abnormality. Thus, TRAP assay test results should be compared with the cytologic ®ndings on both the colposcopic Pap and the referral Pap smear specimens. Conversely, some smears may have a false positive cytologic diagnosis as a result of misinterpretation of cellular components derived from atypical metaplastic changes or atypical reparative changes. In addition, false positive cytologic diagnoses can result from misinterpretation of degenerative changes in exfoliated endometrial epithelial or stromal cells. In many cases, an apparent false negative test result can be evaluated by microscopic review of both the cervical smear and the corresponding cervical biopsy. Some samples, however, have equivocal histologic features on the cervical biopsy specimen that confound attempts to assign a ®nal cytologic/histopathologic diagnosis.
Telomerase is a potentially useful biomarker of high grade premalignant and malignant lesions of the cervical mucosa. Although telomerase analysis might play a limited diagnostic or prognostic role in the evaluation of cervical biopsy specimens, it shows greater promise as an adjunctive marker for the triage of patients with low grade cytologic abnormalities. It is unlikely that telomerase analysis or the analysis of any other biomarker will ever completely replace the Pap smear as a screening system for cervical dysplasia and cancer. Telomerase, however, is one of several markers that could be used as a diagnostic adjunct for the triage of patients with ASCUS or LSIL Pap smears for subsequent colposcopic examination and biopsy. Thus, the greatest role for telomerase analysis may be to defer treatment of patients that do not require further evaluation, rather than to serve as a method to increase the already high level of sensitivity of the Pap smear. The practical application of telomerase analysis as a diagnostic marker will necessitate the development of highly reproducible and sensitive assay systems that can be conducted on a large scale in the routine clinical laboratory. This will require the implementation of automated systems of analysis with non-radioisotopic detection systems. Although the analysis of hTERT by RT ± PCR is an attractive approach for this purpose, the potential complications induced by the existence of hTERT splice variants will need to be addressed. Alternatively, automated systems that are based on the TRAP assay could be designed for implementation in the diagnostic laboratory. Either of these general approaches, however, are likely to be relatively expensive and may be subject to problems related to sampling errors, ineciency of protein or RNA extraction, telomerase inhibitors, PCR inhibitors, or sample degradation. Thus, in order for telomerase to come into widespread application as a diagnostic adjunct, it may be necessary to develop sensitive and speci®c cytochemical methods for the localization of telomerase in cervical cytologic specimens. Whether or not this will become technically feasible remains to be determined. Acknowledgments Research related to this review was supported in part by grant R01 CA78442-01A1 from the National Cancer Institute.
References Anderson S, Shera K, Ihle J, Billman L, Go B, Greer B, Tamimi H, McDougall J and Klingelhutz A. (1997). Am. J. Pathol., 151, 25 ± 31. Benoit AG, Krepart GV and Lotocki RJ. (1984). Am. J. Obstet. Gynecol., 148, 690 ± 694. Berkeley AS, LiVolsi VA and Schwartz PE. (1980). Lancet, ii, 375 ± 376. Birley HD. (1995). Ann. Trop. Med. Parasit., 89, 453 ± 463.
Brien TP, Kallakury BVS, Lowry CV, Ambros RA, Muraca PJ, Malfetano JH and Ross JS. (1997). Cancer Res., 57, 2760 ± 2764. Cancer Facts & Figures, 2001. (2001). American Cancer Society, Atlanta, GA. Chang Chien CC, Lin H, Leung SW, Hsu CY and Cho CL. (1998). Gynecol. Oncol., 71, 99 ± 103. Oncogene
Telomerase in cervical dysplasia and cancer EA Jarboe et al
Colgin LM, Wilkinson C, Englezou A, Kilian A, Robinson MO and Reddel RR. (2000). Neoplasia, 2, 426 ± 432. DeMay RM. (1996a). Am. J. Obstet. Gynecol., 175, 1110 ± 1113. DeMay RM. (1996b). Exfoliative cytology. Chicago: ASCP Press, pp. 61 ± 205. Enomoto T, Haba T, Fujita M, Hamada T, Yoshino K, Nakashima R, Wada H, Kurachi H, Wakasa K, Sakarai M, Murata Y and Shroyer KR. (1997). Int. J. Cancer, 73, 1 ± 6. Feng J, Funk WD, Wang S-S, Weinrich SL, Avilion AA, Chiu CP, Adams RR, Chang E, Allsopp RC, Yu J, Le S, West MD, Harley CB, Andrews WH, Greider CW and Villeponteau B. (1995). Science, 269, 1236 ± 1241. Frost M, Bobak J, Gianani R, Kim N, Weinrich S, Spaulding DC, Cass LG, Thompson LC, Enomoto T, Lopez-Uribe D and Shroyer KR. (2000). Am. J. Clin. Path., 114, 726 ± 734. Gay JO, Donaldson LD, and Goellner JR. (1985). Acta Cytol., 28, 1043 ± 1046. Gorham H, Yoshida K, Sugino T, Marsh G, Manek S, Charnock M, Tarin D and Goodison S. (1997). J. Clin. Pathol., 50, 501 ± 504. Guidozzi F. (1996). Obstet. Gynecol. Sur., 51, 247 ± 252. Gullu IH and Kurdoglu M. (1999). Br. J. Cancer, 81, 186. Hatch KD, Schneider A, and Abdel-Noar MW. (1995). Am. J. Obstet. Gynecol., 172, 1150 ± 1157. Hiyama E, Hiyama K, Yokoyama T and Shay JW. (2001). Neoplasia, 3, 17 ± 26. Iwasaka T, Zheng PS, Yokoyama M, Fukuda K, Nakao Y and Sugimori H. (1998). Obstet. Gynecol., 91, 260 ± 262. Kawai K, Yaginuma Y, Tsuruoaka H, Grin M, Hayashi H and Ishikawa M. (1998). Eur. J. Cancer, 34, 2082 ± 2086. Kilian A, Bowtell DD, Abud HE, Hime GR, Venter DJ, Keese PK, Duncan EL, Reddel RR and Jeerson RA. (1997). Hum. Mol. Genet., 6, 2011 ± 2019. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PL, Coviello GM, Wright WE, Weinrich SL and Shay JW. (1994). Science, 266, 2011 ± 2015. Klingelhutz AJ, Foster AJ and McDougall JK. (1996). Nature, 380, 79 ± 82. Krieger P and Naryshkin S. (1994). Acta Cytol., 38, 291 ± 298. Kurman RJ and Solomon D. (1994). The Bethesda System for Reporting Cervical/Vaginal Cytologic Diagnoses. New York: Springer-Verlag, pp. 9 ± 80. Kyo S, Kanaya T, Ishikawa H, Ueno H and Inoue M. (1996). Clin. Cancer Res., 2, 2023 ± 2028. Kyo S, Takakura M, Kohama T and Inoue M. (1997). Cancer Res., 57, 610 ± 614. Kyo S, Takakura M, Tanaka M, Kanaya T and Inoue M. (1998). Int J Cancer, 79, 66 ± 70. Manos MM, Kinney WK, Hurley LB, Sherman ME, ShiehNgai J, Kurman RJ, Ransley JE, Fetterman BJ, Hartinger JS, McIntosh KM, Pawlick GF and Hiatt RA. (1999). JAMA, 281, 1605 ± 1610. Meyerson M, Counter CM, Eaton EN, Ellisen LW, Steiner P, Caddle SD, Ziaugra L, Beyersbergen RL, Davido MJ, Liu Q, Bacchetti S, Haber DA and Weinberg RA. (1997). Cell, 90, 785 ± 795. Mutirangura A, Sriuranpong V, Termrunggraunglert W, Tresukosol D, Lertsaguansinchai P, Voravud N and Niruthisard S. (1998). Br. J. Cancer, 78, 933 ± 939. Nagai N, Oshita T, Murakami J and Ohama K. (1999). Oncol. Rep., 6, 325 ± 328. Nakamura TM, Morin GB, Chapman KB, Weinrich SL, Andrews WH, Linger J, Harley CB and Cech TR. (1997). Science, 277, 955 ± 959.
Nakano K, Watney E and McDougal JK. (1998). Am. J. Pathol., 153, 857 ± 864. Naryshkin S. (1997). Arch. Pathol. Lab. Med., 121, 270 ± 272. Oh ST, Kyo S and Laimins LA. (2001). J. Virol., 75, 5559 ± 5566. Pairwuti S. (1991). Acta Cytol., 35, 40 ± 46. Pao CC, Tseng CJ, Lin CY, Yang FP, Hor JJ, Yao DS and Hsueh S. (1997). J. Clin. Oncol., 15, 1932 ± 1937. Park T-W, Fujiwara H and Wright TC. (1995). Cancer Suppl., 76, 1902 ± 1913. Park TW, Richart RM, Sun X W and Wright Jr TC. (1996). J. Natl. Cancer Inst., 88, 355 ± 358. Reddy VG, Khanna N, Jain SK, Das BC and Singh N. (2001). Int. J. Cancer, 11, 100 ± 106. Richart RM. (1993). Pathol Ann., 8, 301 ± 328. Riethdorf S, Riethdorf L, Schulz G, Ikenberg H, Janicke F, Loning T and Park TW. (2001). Int. J. Gynecol., 20, 177 ± 185. Robertson JH and Woodend B. (1993). J. Clin. Pathol., 46, 700 ± 702. Shay JW, Zou Y, Hiyama E and Wright WE. (2001). Hum. Mol. Genet., 10, 677 ± 685. Shroyer KR, Stephens JK, Silverberg SG, Markham N, Shroyer AL, Wilson ML and Enomoto T. (1997). Int. J. Gynecol. Pathol., 16, 225 ± 232. Shroyer KR, Thompson C, Enomoto T, Eskens J, Shroyer AL and McGregor JA. (1998). Am. J. Clin. Pathol., 109, 153 ± 162. Snijders PJ, van Duin M, Walboomers JM, Steenbergen RD, Risse EK, Helmhorst TJ, Verheijen RH and Meijer CJ. (1998). Cancer Res., 58, 3812 ± 3818. Solomon D, Schiman M, Tarone R for the ALTS Group. (2001). J. Natl. Cancer Inst., 93, 293 ± 299. Stoppler H, Hartman DP, Sherman L and Schlegel R. (1997). J. Biol. Chem., 272, 13332 ± 13337. Tabbara SO and Sidawy MK. (1996). Diagn. Cytopathol., 14, 84 ± 86. Takakura M, Kyo S, Kanaya T, Tanaka M and Inoue M. (1998). Cancer Res., 58, 1558 ± 1561. Tkeshelashvili VT, Bokhan JV, Kuznetzov VV, Maximov SJ and Chkuaseli GT. (1993). Eur. J. Gynaec. Onc., 14, 89 ± 94. Ulaner GA, Hu JF, Vu TH, Giudice LC and Homan AR. (1998). Cancer Res., 58, 4168 ± 4172. van der Graaf Y and Vooijs GP. (1987). J. Clin. Pathol., 40, 438 ± 442. van der Graaf Y, Vooijs GP, Gaillard HLJ and Go DM. (1987). Acta Cytol., 31, 434 ± 438. Wick M, Zubov D and Hagen G. (1999). Gene, 17, 97 ± 106. Wisman GB, Hollema H, de Jong S, ter Scheggert J, TjongHung SP, Ruiters MH, Krans M, de Vries EG and van der Zee AG. (1998). J. Clin. Oncol., 16, 2238 ± 2245. Wisman GB, De Jong S, Meersma GJ, Helder MN, Hollema H, de Vries EG, Keith WN and van der Zee AG. (2000). Hum. Pathol., 31, 1304 ± 1312. Wisman GB, Knol AJ, Helder MN, Krans M, de Vries EG, Hollema H, de Jong S and van der Zee AG. (2001). Int. J. Cancer, 91, 658 ± 664. Wolf JK and Ramirez PT. (2001). Cancer Invest., 19, 621 ± 629. Wright TC, Kurman RJ and Ferenczy A. (1994a). Blaustein's Pathology of the Female Genital Tract. 4th edn. Kurman RJ. (ed.). New York: Springer-Verlag, pp. 287 ± 289. Wright TC, Kurman RJ and Ferenczy A. (1994b). Blaustein's Pathology of the Female Genital Tract. 4th edn. Kurman RJ. (ed.). New York: Springer-Verlag, pp. 229 ± 232.
Telomerase in cervical dysplasia and cancer EA Jarboe et al
Wright TC, Sun XW and Koulos J. (1995a). Obstet. Gynecol., 85, 202 ± 210. Wright WE, Shay JW and Piatyszek MA. (1995b). Nucl. Acids Res., 23, 3794 ± 3795. Wu Y-Y, Hruszkewycz AM, Delgado RM, Yang A, Vortmeyer AO, Moon Y-W, Weil RJ, Zhuang Z and Remaley AT. (2000). Clin. Chim. Acta, 293, 199 ± 212. Yahata N, Nagao K, Ohyashiki JH, Hisatomi H and Ohyashiki K. (1999). Int. J. Oncol., 14, 709 ± 712. Yashima K, Ashfaq R, Nowak J, Von Gruenigen V, Milchgrub S, Rathi A, Albores-Saavedra J, Shay JW and Gazdar AF. (1998). Cancer, 82, 1319 ± 1327.
Yi X, White DM, Aisner DL, Baur JA, Wright WE and Shay JW. (2000). Neoplasia, 2, 433 ± 440. Yokoyama Y, Takahashi Y, Shinohara A, Lian Z and Tamaya T. (1998). Gynecol. Oncol., 68, 145 ± 149. Zhang DK, Ngan HY, Cheng AN, Cheung AN, Liu SS and Tsao SW. (1999). Eur. J. Cancer, 35, 154 ± 160. Zheng P-S, Iwasaka T, Yokoyama M, Nakao Y, Pater A and Sugimori H. (1997). Gynecol. Oncol., 66, 222 ± 226. zur Hausen H. (1996). Biochim. Biophys. Acta, 1288, F55 ± F78. zur Hausen H. (1999). Proc. Assoc. Am. Physic., 111, 581 ± 587.