Telomerase activity in precancerous hepatic nodules

Recent studies have demonstrated that telomerase, a reverse transcriptase linked to cellular "immortalization," is activated in a variety of malignant human tumors. This study was conducted to determine whether telomerase activity represents a marker of malignant transformation in precancerous (dysplastic) nodules arising in patients with cirrhosis.

T he telomeres are special components of chromosomal ends containing repeated six nucleotide sequences, which protect the chromosomes from sticking to each other and undergoing structural changes. [1][2][3][4] When normal somatic cells replicate, the telomeres are shortened, because DNA polymerases cannot finalize the copying process to the very end of the DNA strands. It has been proposed that decreasing telomere length over time signals cellular senescence or programmed cell death, thus protecting the genome from further changes that could lead to malignant transformation.
As opposed to normal cells, ''immortalized'' cells in culture and cancer cells in vivo exhibit short but stable telomere lengths, which are maintained by the action of telomerase, a reverse transcriptase with an RNA component. [5][6][7] Recent studies have demonstrated significant telomerase activity in a variety of human carcinomas, including breast, liver, colon, prostate, and thyroid primary tumors. 6,[8][9][10][11] size, color, texture, or degree of bulging at the cut panel's opinion and our own experience that microscopic examination cannot afford sufficient distinc-surface. Histologic examination is necessary for definitive identification of DNs. However, these lesions tion between these two entities. The etiology of cirrhosis was determined on the basis of clinical, serologic, cover a broad spectrum of histologic appearances, ranging from features indistinguishable from those of and histologic data. large regenerative nodules to features reminiscent of well differentiated hepatocellular carcinoma (HCC).

Telomerase Activity Assay
Telomerase activity was assayed blindly (without Differential diagnosis may be difficult or impossible at both ends of the spectrum, especially in small biopsy knowledge of the histologic findings) by using the telomeric repeat amplification protocol (TRAP assay) 6 samples. 19 We have hypothesized that activation of te-with some modifications. 20,21 Forty-six frozen samples, approximately 0.5 cm 1 0.5 cm 1 0.2 cm, were homog-lomerase expression in precancerous nodules of cirrhotic livers may represent a marker of progression in enized with a drill and disposable pestles in 100 mL ice-cold lysis buffer containing 0.5% CHAPS (Pierce, the multistep process of carcinogenesis. We conducted this study to assess telomerase activity levels Rockford, IL), 10 mM Tris-HCl, 1 mM MgCl 2 , 1 mM ethyleneglycotetraacetic acid, 0.1 mM phenylmethyl-in precancerous lesions and to consider the possibility of using telomerase activity determination for diag-sulphonylfluoride, 5 mM 2b-mercaptoethanol and 10% glycerol. The samples were incubated on ice for nostic purposes. We followed the terminology for nodular hepatocellular lesions, which was proposed by an 30 minutes and subsequently centrifuged at 14,000 revolutions per minute for 30 minutes (4 ЊC). Superna-International Working Party that was sponsored by the 1994 World Congresses of Gastroenterology. 19 tant fluids then were transferred into new tubes and snap-frozen on dry ice. Protein content was measured by the BCA assay (Pierce), working dilutions of speci-

Tissues and Histologic Techniques
mens were equalized to 3 mg of protein/mL, and the TRAP assay was performed using 6 mg of protein in Fourteen cirrhotic liver specimens were selected from the files of the Mount Sinai Medical Center in New each reaction. The initial TS-upstream primer extension step was followed by heating at 99 ЊC for 5 min-York City. Selection was based purely on availability of snap-frozen tissue samples of large nodular lesions.
utes to stop telomerase activity. Immediately thereafter, 2 U of Taq-polymerase (Boehringer-Mannheim Thirteen specimens were livers removed on orthotopic transplantation, and one was a left lateral segment Biochemicals, Indianapolis, IN), 2 mCi of [a-32 P]-dCTP (Amersham, Arlington Heights, IL), and 0.1 mg of CX-resected for HCC. Each specimen was sectioned serially in 0.5 to 0.7-cm thick slices, and carefully exam-downstream primer were added to each tube in a polymerase chain reaction (PCR) mix containing 11 PCR ined for the presence of large nodules measuring at least 1 cm in greatest dimension and appearing dis-buffer with MgCl 2 . PCR was performed for 30 cycles at 94 ЊC for 30 seconds, 50 ЊC for 30 seconds, and tinct from the surrounding cirrhotic parenchyma in terms of color, texture, or degree of bulging at the cut 72 ЊC for 90 seconds. Fifteen microliters of the PCR product of each sample then was visualized by electro-surface. Fresh samples of all cirrhotic livers and 30 large nodules contained therein were snap-frozen on phoresis on 12% nondenaturing polyacrylamide gels followed by autoradiography at 080 ЊC for 8 and 48 dry ice and stored at 070 ЊC until use. Additional sections from the same areas were submitted for routine hours (short and long exposure, respectively). The reaction series was performed three times. For histologic examination. Two normal liver samples also were snap-frozen and included in the study as ''nor-a semiquantitative assessment of telomerase activity, dilutions of 11, 0.11, and 0.011 were used (6.0, 0.6, mal tissue controls.'' These were derived from donor livers with essentially normal histology, which were and 0.06 mg of protein per reaction, respectively), as described in the literature. 8 Specimens were evaluated not used for transplantation.
Four-micron thick, hematoxylin and eosin (H & as: 1) negative, when bands did not appear in any dilution; 2) weakly positive, when bands appeared E) stained sections of paraffin blocks were used for diagnostic classification, which was made into four after 48 hours' exposure in the 11 diluted samples; 3) positive, when bands appeared after 8 hours' exposure groups: 1) cirrhotic liver; 2) large regenerative nodule (LRN)/low grade DN; 3) high grade DN; and 4) HCC.
in the 11 samples; and 4) strongly positive, when bands appeared in the 0.11 or more diluted samples. We followed the nomenclature and classification scheme proposed by the International Working Lysis buffer was used as a negative control. True positivity was checked by adding RNase A into the Party 19 ; however, we decided to combine LRNs and low grade DNs in one group because it is both this samples. 6  of telomerase, which then becomes inactive. A series grade DNs came from liver specimens with multiple (uncountable) large nodules (one each from Cases 5 of all 11 samples were treated with 1 mg RNase A (Boehringer-Mannheim) each during the initial TS-and 8). All other lesions were derived from specimens with one to six large nodules in each. Six HCCs were primer extension step. To verify true negative results, 0.5 mL (1.5 mg protein) of a known positive specimen well differentiated and one was moderately differentiated. All nodular lesions measured 1 -2 cm in greatest (from IMR-32, a neuroblastoma cell line) was added to the reactions as an ''internal control,'' 21 as follows.
dimension, with the exception of 3 cases of HCC (2 well differentiated and 1 moderately differentiated), All specimens were processed blindly at 11 and 0.11 dilutions, together with a parallel series of 0.11 diluted which measured 3.8 -5 cm.
On microscopic examination, the large noncan-samples containing the ''internal control.'' In every case there was successful amplification of the ''inter-cerous nodules were well circumscribed and surrounded by a rim of fibrous tissue, similar to that ob-nal control''; thus, samples shown to be negative or weakly positive at the 11 and negative at the 0.11 served around cirrhotic nodules. LRNs/low grade DNs were comprised of normal or near-normal-appearing dilution while simultaneously positive for the ''internal control'' were not processed further to the 0.011 hepatocytes (Fig. 1A). Portal tracts were observed in a fairly regular distribution. In three lesions (one from dilution. Samples shown to be positive at the 11 and negative at the 0.11 dilution while positive for the Case 8 and two from Case 12), large cell change of hepatocytes was present. High grade DNs demon-''internal control'' also were not processed further to the 0.011 dilution. Samples positive at the 0.11 dilu-strated evidence of cytologic or architectural atypia (Fig. 1B). Cytologic atypia, in the form of irregular nu-tion were processed to 0.011, together with a parallel series containing the ''internal control.'' clear contour, nuclear hyperchromasia, or increased nuclear-cytoplasmic ratio, was observed in all lesions, at least focally. Portal tracts were present in small

Clinical Data and Pathologic Findings
numbers and irregular distribution. Architectural atypia in the form of a nodule-in-nodule growth pat-Eight patients were men and six were women. Their ages ranged from 20 -70 years (mean, 48 years). Demo-tern was observed in four lesions (one each from Cases 2 and 11 and two from Case 4), focal pseudogland graphic data, the etiology of cirrhosis, and classification of large nodular lesions are provided in Table 1.
formation in two lesions (one each from Cases 11 and 14), and focal unpaired (nontriadal) arteries in two Overall, 13 nodules were found to be LRNs/low grade DNs, 10 high grade DNs, and 7 HCCs. Two LRNs/low lesions (one each from Cases 4 and 11).  The nodules of HCC lacked portal tracts, con-sis to HCC. In these studies, normal liver specimens lacked telomerase activity and specimens with chronic tained scattered unpaired arteries, and demonstrated more prominent cytologic atypia than DNs. Five of 6 liver disease showed weak activity in 38 -55% of cases, whereas HCC specimens, with few exceptions, were well differentiated HCCs had trabecular architecture with pseudogland formation and bile production ( Fig. either positive or strongly positive for telomerase activity. Telomerase activity thus was proposed to be a 1C). In addition, one of these lesions (Case 9) showed dense hyalinized collagen among the tumor cell tra-useful marker for the early diagnosis of HCC. 25,27 In our study, there was no overlap in telomerase beculae. The sixth well differentiated HCC (Case 10) featured thin, adenoma-like trabeculae. The single activity levels between specimens of cirrhotic parenchyma and those of HCC; none of the cirrhotic speci-moderately differentiated HCC (Case 14) had trabecular architecture and included a large number of fat-mens was positive, and no HCC was either weakly positive or negative. These findings further support containing tumor cells.
the widely accepted view that telomerase is activated at a high level in HCC. At this point, it should be noted

Telomerase Activity Assay Findings
The results of the TRAP assay are presented in Table  that the level of positivity in our specimens generally appeared to be lower than that already described in 1. The assay was successfully performed in 44 of the 46 frozen tissue samples, with the appropriate controls HCC by other authors. 8,26,27 This may be due to the modifications in the TRAP method we used, which ruling out false-positive and false-negative results. The assay was unsuccessful in one sample of cirrhotic liver included: 1) a higher temperature to stop the action of telomerase (99 ЊC vs. 90 ЊC); and 2) ''hot start'' PCR and one sample of high grade DN, apparently because the tissues had degraded before processing. In these afterward, which made the amplification of the telomerase-extended TS primer products more specific. two cases, band-ladders were detected throughout the lanes and without regular distances among the bands Nevertheless, our data are in good correlation with those of these previous studies. 8,26,27 This is particularly (data not shown).
No telomerase activity was detected in the nor-evident in our case of moderately differentiated HCC, in which the strongest positivity was detected. 27 mal liver specimens. Five of 13 cirrhotic specimens were negative, and 8 exhibited weak activity. In the The focus of our study was the assessment of telomerase activity in precancerous hepatic nodules. majority of large nodular lesions, including 10 of 13 LRNs/low grade DNs, 7 of 9 high grade DNs, and 4 Only three such nodules have been evaluated previously, and all showed telomerase activity of a level of 7 HCCs, telomerase activity was positive. Two LRNs/low grade DNs were weakly positive, whereas similar to HCC. 26 We evaluated a total of 22 large noncancerous nodules, 19 of which (86%) exhibited a de-another one, in a 23-year-old patient (Case 8), was strongly positive (Fig. 2A). Telomerase activity was gree of telomerase activity similar to HCC. The cirrhotic parenchyma surrounding the lesions was either weakly positive in one high grade DN, and strongly positive in another. Three HCCs also were strongly negative or weakly positive for telomerase activity. These findings suggest that in hepatocarcinogenesis positive for telomerase activity. One (Case 14; moderately differentiated HCC) showed faint bands, even telomerase activation occurs before nodules with microscopic features of HCC are recognizable. A similar in the 0.011 dilution after 48 hours (Fig. 2B). A summary of telomerase activity in the various groups of phenomenon recently has been described in a rat model of chemical carcinogenesis. 29 lesions is provided in Figure 3.
The terminology of nodular hepatocellular lesions recently has been standardized in the consensus arti-

DISCUSSION
In the last few years, the literature regarding telomere cle of the International Working Party. 19 A review article that soon followed 30 emphasized the distinguishing stabilization and telomerase activation in cancer has been growing rapidly, fueled to great extent by the microscopic features of the various types of nodules. High grade DNs are differentiated from low grade DNs hope of using telomerase activity as a marker of malignant behavior. Indeed, with the exception of germ cells on the basis of cytologic or architectural atypia. The atypical features may either be diffuse or focal. Evi-and leukocytes, normal human cells do not express telomerase, whereas various malignant tumor cells dence of cytologic atypia includes irregular nuclear contour, nuclear hyperchromasia, increased nuclear-do. 6,[8][9][10][11] Since 1995, eight studies, 8,22 -28 all from Japan, have assessed telomere length and telomerase activity cytoplasmic ratio, cytoplasmic basophilia or clear cell change, rare mitotic figures, and resistance to iron ac-in patients with chronic liver disease and HCC. As a rule, telomere length was found to decrease progres-cumulation. In high grade DNs, portal tracts are present in smaller numbers and more irregular distribu-sively from normal liver to chronic hepatitis to cirrho-   Cancer tion compared with low grade DNs; liver cell plates tivity was classified as positive and strongly positive in these two nodules, respectively. These findings indi-may be up to three cells in thickness; focal pseudogland formation may be observed; and expansile, cate that significant telomerase activity may be detected in some lesions of regenerative nature. On the clone-like populations of hepatocytes often are present that compress the adjacent liver cell plates or por-basis of current morphologic definitions of precancerous nodules, 19 this is a substantial limitation to utiliz-tal tracts in a nodule-in-nodule growth pattern.
HCC is distinguished from high grade DN on the ing telomerase activity as a marker of malignant transformation in the liver. basis of the following microscopic features 19 : absence of portal tracts, presence of stromal invasion, cell The majority of DNs and LRNs (86%) exhibited telomerase activity levels similar to those of HCCs; plates three or more cells in thickness, unattached ''floating'' cross sections of trabeculae, diffuse pseudo-however, the levels of three lesions were similar to those encountered in a subset of cirrhotic liver sam-glandular structures, numerous unpaired arteries, moderately irregular nuclear contour, high nuclear-ples (weak positivity). The use of internal telomerase assay standards 11,26,27 may provide more detailed infor-cytoplasmic ratio with significantly increased cell density, reduction of reticulin fibers, and more than rare mation that could be useful in such areas of overlap, although true quantitation cannot yet be performed. mitotic figures. However, it has been emphasized that the features of malignancy may be focal, and differen-Regenerating hepatocytes, small numbers of ''immortalized'' liver cells, and lymphocytes all have been pro-tial diagnosis between HCC and high grade DN occasionally is impossible. 19 posed as possible sources of the weak telomerase activity observed in chronic liver disease. 8 Nevertheless, Separation of low grade DNs from LRNs is even more difficult. According to the International Working weak telomerase activity in cirrhotic liver specimens should be considered to derive from a limited number Party, the most certain way to distinguish between these two types of nodules will be the application of of cells, 27 which is unlikely to be the case in DNs. Further research in this direction is warranted; in situ molecular genetic techniques, which is a prospect for the future. 19 However, the following histologic features TRAP assay may clarify the issue. 31 In conclusion, we found similar telomerase activ-have been considered to favor a diagnosis of LRN over one of low grade DN: presence of a large (uncountable) ity levels in HCC and the majority of large hepatic nodules arising in cirrhosis, including lesions with ei-number of lesions, presence of sinusoidal dilatation, and absence of cytologic atypia or clone-like fea-ther a precancerous (DNs) or regenerative nature (LRNs). There was a substantial difference in telo-tures. 18,19,30 In our study, 9 of the 22 large noncancerous nod-merase activity levels between large nodular lesions and the surrounding cirrhotic parenchyma, although ules assayed for telomerase activity were high grade DNs. Eight of 9 such nodules were either positive (n a minority of lesions exhibited overlapping levels. Our findings suggest that telomerase activation is an early Å 7), or strongly positive (n Å 1) for telomerase activity. However, it is interesting to note that a similar inci-event in large nodule formation in cirrhosis, which may facilitate the action of other factors in the process dence of positivity was detected in the category of LRNs/low grade DNs, with 11 of 13 lesions being either of carcinogenesis. However, telomerase activity cannot be used as a marker of malignant transformation. positive (n Å 10), or strongly positive (n Å 1). This category included lesions of either a regenerative Although the majority of HCCs and precancerous nodules exhibit high telomerase activity levels, some large (LRNs) or precancerous nature (low grade DNs). The high incidence of positivity in this category indicates regenerative nodules also do. that assessment of the telomerase activity level cannot be of help in distinguishing between these two types