Hypermethylation of Asc/tms1 Is a Sputum Marker for Late-stage Lung Cancer

DNA hypermethylated gene promoter sequences are extremely promising cancer markers. Their use for risk assessment, early diagnosis, or prognosis depends on the timing of this gene change during tumor progression. We studied this for the proapoptotic gene ASC/TMS1 in lung cancer and used the findings to develop a sputum marker. ASC/TMS1 protein levels are reduced in all lung cancer types (30 of 40; 75%) but not in 10 preinvasive lesions. Hypermethylation of ASC/TMS1 is also associated with invasive cancers (41 of 152 or 27.0% of all lung cancer types) with variation in incidence between histopathologic types including 32.1% (26 of 81) of adenocar-cinomas, 13.2% (7 of 53) of squamous cell carcinomas, 38.5% (5 of 13) of large-cell carcinomas, and 60% (3 of 5) of small-cell lung cancers. The hypermethylation is particularly correlated with late tumor stages being present in only 14% of stage I but 60% of later-stage tumors. The incidence of ASC/TMS1 hypermethylation in sputum DNA fully mimics the tissue findings being present in only 2% (2 of 85) of high-risk, cancer-free smokers, 15% (3 of 18) of patients with stage I non–small-cell lung cancer (NSCLC), but 41% of patients with stage III NSCLC (18 of 44), including 56% (10 of 18) of those with adenocarcinoma. Importantly, sputum is positive for this marker in 24% (10 of 42) of very high risk, clinically cancer-free individuals previously resected for stage I NSCLC. Thus, hypermethylation of ASC/TMS1 is a marker for late-stage lung cancer and, in sputum, could predict prognosis in patients resected for early-stage disease.


Introduction
Hypermethylated gene promoter sequences are extremely promising cancer markers.However, their full potential depends on knowing the timing for the hypermethylation in the progression stages for given tumor types.We now illustrate this point from studies of hypermethylation of the ASC/TMS1 gene in lung cancer.ASC/TMS1 is a candidate tumor suppressor gene which has proapoptotic properties (1)(2)(3).The translated protein is composed of two protein-protein interaction domains, PYD, an NH 2 -terminal PYRIN-domain, and CARD, a COOH-terminal caspase-recruitment domain (1).It binds and activates caspase-1 through CARD-CARD interaction (4)(5)(6).Caspase-1 is necessary for generating the proinflammatory cytokines, interleukin (IL)-1, and IL-18 (IFN-ginducing factor; ref. 7).The PYRIN domain also likely mediates protein-protein interactions in apoptotic and inflammatory signaling pathways (8).ASC/TMS1 is a mediator of nuclear factor nB activation and caspase-8-dependent apoptosis in an IL-1hconverting enzyme protease-activating factor signaling pathway involving interaction of the PYRIN group with caspase-8 (9).Ectopic expression of ASC/TMS1 induces p53-independent apoptosis in human embryonic renal 293 cells via a pathway dependent on caspase-9 and inhibits the growth of breast cancer cells (2,10).A decrease in ASC/TMS1 induced by antisense RNA in the human leukemic cell line HL-60 resulted in reduced sensitivity to anticancer agents (1).Therefore, ASC/TMS1 may play crucial roles in provoking inflammation and cell death.
The potential importance of ASC/TMS1 as a tumor suppressor is illustrated by recent observations for epigenetically mediated, transcriptional silencing of the gene in cancers.Such silencing, associated with aberrant methylation of promoter region CpG islands, is a frequent mechanism for tumor suppressor gene inactivation in human cancers (11,12) and has been observed for ASC/TMS1 in multiple cancer types (2,(13)(14)(15)(16)(17)(18).However, the relationship of this change to progression stages of the tumors, other than its correlation to more aggressive behavior for brain tumors (19), and thus its value as a potential tumor marker, is not known.We have addressed these questions by studying the expression and promoter methylation status of ASC/TMS1 in the progression stages of primary lung cancer.We find that hypermethylation of ASC/TMS1 occurs primarily in late-stage lung cancer and is not present in preinvasive stages of this disease.We use this information for developing ASC/TMS1 hypermethylation in sputum DNA as a potentially potent marker for high risk of developing recurrent and/or second tumors in patients resected for stage I lung cancer.

Materials and Methods
Tissue samples.Primary lung cancers were obtained from 83 patients (mean age, 68 years; range, 59-75 years) from Shinshu University Hospital (Matsumoto, Japan) and 69 (mean, age 69 years; range, 61-73 years) from the Johns Hopkins Medical Institute (Baltimore, MD) and immediately frozen at À80jC after surgical resection.Six normal lung tissues came from individuals without cancer ( five from autopsy and one from lung peripheral to a benign bronchial tumor).Tissue acquisition was conducted under approved guidelines of the institutional review boards of both institutions.
Histologic examination was based on the criteria used for the WHO classification (20).Clinical staging was done according to the new tumornode-metastasis classification criteria (21).All specimens were fixed for 48 hours in 20% phosphate-buffered formalin (pH 7.4) at room temperature, embedded in paraffin, and cut into 10-Am sections for DNA extraction, 7-Am sections for methylation-specific PCR in situ hybridization (MSP-ISH), and 3-Am sections for H&E staining and immunostaining.
Subject enrollment for sputum study.Sputum samples were obtained from individuals at varying degrees of risk for lung cancer from the Albuquerque, New Mexico metropolitan area.Subjects were recruited through advertisement in a local newspaper and are part of a developing cohort of people at risk for lung cancer.Group 1 consisted of current and former smokers (n = 85) with a minimum smoking history of 15 pack years, ages between 40 and 80 years, with good performance status, and no prior diagnosis of cancer in the aerodigestive tract.Group 2 consisted of 42 patients postresection for stage I lung cancer (n = 42) who were enrolled through a Eastern Cooperative Oncology Group (ECOG) phase III chemoprevention trial to evaluate the ability of L-selenomethionine in decreasing the occurrence of secondary primary tumors in patients resected for stage I non-small-cell lung cancer (NSCLC).These individuals must have undergone complete resection of a histologically proven stage IA (pT 1 N 0 ) or stage IB (pT 2 N 0 ) NSCLC, with at least one mediastinal lymph node sampled at resection, be currently disease-free between 6 and 36 months from resection time, z18 years of age, not have received chemotherapy or radiation therapy, and have a negative chest X-ray or computed topography scan V8 weeks before registration on the trial.Group 3 was made up of 18 patients with stage I lung cancer enrolled in the New Mexico Lung Cancer Registry and recruited through the Pulmonary Clinics at the Veterans Administration and the University of New Mexico Medical Centers in Albuquerque.Group 4 consisted of 44 stage III lung cancer patients enrolled through another ECOG trial, approved by The Lovelace Respiratory Research Institute Review Board with all participants giving written informed consent, investigating the ability of thalidomide, in combination with standard chemotherapy and radiation therapy, to improve survival.
All smokers and stage I lung cancer patients enrolled in the Lung Cancer Registry completed a standardized respiratory questionnaire from the American Thoracic Society (22) that describes in detail each subject's complete smoking history, including variability in smoking intensity over time and periodic intervals of cessation.The questionnaire also documents respiratory health (cough, dyspnea, etc.), general health, family history, and occupational exposures.Subjects enrolled on the prevention trial completed a baseline questionnaire that documented tobacco history including pack years, duration, and smoking status.Only age and gender were available for lung cancer patients participating in the thalidomide trial.
Sputum collection and processing.All group 1 participants were asked to provide an induced sputum sample using a variation of the ultrasonic nebulization technique described by Saccomanno et al. (23).Subjects used water or saline to brush tongue, buccal surfaces, teeth, and gingiva gently to remove superficial epithelial cells and bacteria, followed by gargling and rinsing with tap water.Participants then inhaled a nebulized 3% saline solution from an ultrasonic nebulizer for 20 to 30 minutes.Sputum was collected in a sterile specimen cup and an equal volume of Saccomanno solution was added immediately.Subjects in groups 2, 3, and 4, outlined above, used the early morning spontaneous cough technique to collect sputum (24).Participants were provided with a sterile specimen cup containing Saccomanno's fixative in a self-addressed return mailer.To increase the probability that material of deep lung origin was obtained, subjects received detailed verbal instructions by study personnel at the participating institution and written instructions on how to do the technique.For three consecutive mornings, they coughed deeply, and the resulting mucous was expectorated into the cup, which was then placed in a postage-paid mailer for delivery to Lovelace.Sputum samples were defined as adequate by the presence of deep lung macrophages or Curschmann's spiral (23) and, irrespective of adequacy, processed for methylation analysis by extensive vortex mixing, washing once with Saccomanno solution, and storage at room temperature until analysis.In addition, at least two Papanicolaou-stained slides underwent morphologic examination by a certified cytopathologist.
DNA was isolated from sputum as described, modified (1 Ag) with sodium bisulfite for methylation assays (25), and sent for analysis of ASC/TMS1 methylation to investigators at Johns Hopkins Medical Institute, who were blinded to group status.
Bisulfite modification and MSP.Genomic DNA was extracted from frozen tissue by standard proteinase K digestion, phenol/chloroform extraction, and ethanol precipitation as previously described (26).The quality and quantity of the DNA were determined by A 260 /A 280 ratio.
For sputum and paraffin tissues, nested MSP was done.For sputum samples, the primer sequences for the first-stage PCR were (sense) 5 ¶-GGAGTTGGGATTAGAGT- ¶3 and (antisense) 5 ¶-CAACAACTTCAACT-TAAACTTCTTAAACTC- ¶3, yielding a 206-bp product.For paraffin samples, we used a first-stage PCR employing primers which amplified a 146-bp product, and the primer sequences were (sense) 5 ¶-GGAA-GGCGGGGAGTTTAGGTTT- ¶3 and 5 ¶-GGAAGGTGGGGAGTTTAGGTTT- ¶3 and (antisense) 5 ¶-AATCAAATTCTCCAACACATCCAAA- ¶3 and 5 ¶-GAT-CAAATTCTCCAACGCATCCAAA- ¶3.The PCR amplification protocol for stage I was as follows: 95jC for 5 minutes, then denaturation at 95jC for 30 seconds, annealing at 56jC, extension at 72jC for 30 seconds for 40 cycles, followed by a 5-minute final extension.The stage I PCR products were diluted 50-fold and 2 AL were subjected to stage II PCR in which the previously mentioned nonnested primers specific to methylated or unmethylated sequences were used.PCR conditions for sputum samples were 40 cycles at 95jC for 20 seconds, 68jC for 20 seconds, 72jC for 20 seconds, and a final extension at 72jC for 5 minutes.PCR conditions for paraffin tissues were 30 cycles at 95jC for 30 seconds, 68jC for 30 seconds, 72jC for 30 seconds, and a final extension at 72jC for 5 minutes.Negative controls, containing normal peripheral blood lymphocytes and no DNA, and positive controls were done for each PCR reaction.Reaction products were separated by electrophoresis on a 3% agarose gel, stained with ethidium bromide, and photographed.All MSP assays were done in duplicate.
Immunohistochemical staining.Deparaffinized tissue specimens were subjected to immunohistochemical staining for detection of ASC/TMS1 with mouse monoclonal antibodies (1) and using an indirect detection method (28) employing an antimouse immunoglobulin antibody conjugated with horseradish peroxidase (DAKO, Carpinteria, CA) as the secondary antibody.Peroxidase activity was visualized with diaminobenzidine-H 2 O 2 solution and negative controls included omission of primary antibody.We evaluated ASC/TMS1 expression as a percentage of ASC/TMS1-positive cells as previously described (13).
MSP-ISH.MSP-ISH, as previously described ( 29), employed silane-coated slides prepared from 7-Am sections of paraffin-embedded normal and tumor tissues.After deparaffinization, tissue slides were digested in Pepsin Solution (Invitrogen, San Diego, CA) at 50 jC for 10 minutes, washed in 0.1 mol/L Tris-HCl, 0.1 mol/L NaCl for 1 minute, 100% ethanol for 1 minute, air-dried, incubated in 0.2 mol/L NaOH for 10 minutes at 37jC, then placed in the 3 mol/L sodium bisulfite solution and incubated at 55jC for 16 hours, followed by incubation in 0.3 mol/L NaOH for 5 minutes.Primer sequences for ASC/TMS1 were the same as previously described.After denaturing at 95jC for 5 minutes, 35 cycles were conducted at 56jC for 2 minutes and 95jC for 1 minute.
After amplification, ISH was done using an unmethylated-specific or methylated-specific internally digoxigenin-labeled probe (1 Ag/mL; refs.29, 30) prepared as follows.Methylated or unmethylated PCR products, which were previously confirmed by bisulfite sequencing, were cloned into TOPO vector (Invitrogen) and used as templates for construction of a digoxigenin-labeled DNA probe synthesized using the PCR DIG Probe Synthesis Kit (Roche Applied Science, Penzberg, Germany).PCR primers for synthesizing the digoxigenin-labeled DNA probe were, for the methylated ASC/TMS1 gene, (sense) 5 ¶-CGATTTTTTTTTGGTCGGCGGTT- ¶3 and (antisense) 5 ¶-GCGCCCCATAACTCCAAAATC- ¶3, and for the unmethylated allele, (sense) 5 ¶-TGATTTTTTTTTGGTTGGTGGTTG- ¶3 and (antisense) 5 ¶-ACACACCCCATAACTCCAAAAT- ¶3.The amplicon and probe were codenatured at 99jC for 5 minutes, hybridized at 40jC for 16 hours, and washed for 20 minutes in 1Â SSC [0.15 mol/L sodium chloride/0.015mol/L sodium citrate (pH 7)] with 2% bovine serum albumin at 52jC.After washing, immunohistochemistry for detection of hybridized probes employed an alkaline phosphatase-conjugated antidigoxigenin antibody (Roche Applied Science) and then exposure to the chromogen, nitroblue tetrazolium, and 5-bromo-4-chloro-3-indolyl phosphate (Roche Applied Science) at 37jC.The final counterstain, nuclear fast red, stains the negative cells pink in contrast to the blue signal for the ASC/TMS1 probe.A control study omitting Taq polymerase from PCR amplification showed no specific reactivity.
Statistical analysis.The clinical pathologic variables examined in the present study included lymph node metastasis, pulmonary metastasis, pleural invasion, lymphatic invasion, and vessel invasion.Additional study characteristics include age, gender, histology, and stage of disease.Data were summarized using frequencies and percents for discrete variables as well as means and medians for continuous variables.The relationship between ASC/TMS1 protein positive cell expression and lesion type was evaluated by Fisher's exact test.We evaluated the relationship between methylation status in human lung cancers and clinicopathologic characteristics using the m 2 test and Fisher's exact test for independence for dichotomous variables and the Student's t test for continuous variables.In atypical adenomatous hyperplasia, ASC/TMS1 expression was strongly observed in atypical cells (B ).ASC/TMS1 is also detected in bronchioloalveolar adenocarcinoma (C ).In a typical invasive adenocarcinoma, ASC/TMS1 expression is distinctly reduced (D).In normal bronchus (E) and in squamous metaplasia (F ), ASC/ TMS1 is strongly expressed and the protein is also detected in squamous dysplasia (G).However, ASC/TMS1 expression is often reduced in squamous cell carcinoma (H ).Bar, 100 Am.Results were judged to be statistically significant at P V 0.05.The association between methylation of the ASC/TMS1 gene in sputum and group status was assessed using logistic regression with methylation as the outcome variable.Group status was included as an independent variable in the model using indicator variables to identify each group.Smokers were defined as the reference group.The model was adjusted for gender and age.Analyses for the sputum study were conducted using SAS statistical package version 8.02 (SAS Institute, Inc., Cary, NC).All other analyses were done using STATA statistical package version 7.0 (College Station, TX).

Results
ASC/TMS1 is reduced in primary lung cancers.Immunohistochemistry was used to evaluate ASC/TMS1 expression in primary samples of normal lung, precancerous lesions, and cancers.Expression was reduced only in cancers (Fig. 1; Supplementary Table S1).Thus, robust protein expression, appearing in >80% of cells, was observed in multiple cell types from six different normal lung parenchyma samples, including bronchial epithelium (Fig. 1A  and E).Furthermore, this same high expression was seen in six cases of atypical adenomatous hyperplasia, a precancerous lesion for adenocarcinoma (Fig. 1B), and from four cases of squamous dysplasia, a precancerous lesion in the evolution of squamous cell carcinoma (Fig. 1G).Finally, in striking contrast to data for invasive cancers described below, five cases of bronchioloalveolar adenocarcinoma, a noninvasive stage of adenocarcinoma, showed similar robust staining (Fig. 1C) to that found in normal lung parenchyma (81-100% positive cells; Fig. 1A).
In contrast to the robust expression observed in precancerous and non-invasive cancerous lesions, reduced expression was found in 30 of 40 (75%) samples of primary lung carcinomas (Fig. 1D and H).Most adenocarcinomas (23 of 30) showed a mixture of ASC/ TMS1-positive and -negative cells compared with normal lung, and tumors were classified into four groups according to the percentage of ASC/TMS1-positive cells.Four tumor samples showed 0% to 24% positive cells, 11 showed 25% to 50% positive cells, and 8 showed 51% to 80% positive cells.Similar findings of reduced ASC/TMS1 protein were present for four of six squamous cell carcinomas, with three tumor samples showing 25% to 50% positive cells and one showing 50% to 80% positive cells.Two cases of large-cell carcinomas both showed 0% to 24% positive cells, and one of two cases of small cell carcinoma had 25% to 50% positive cells (Supplementary Table S1).Overall, as compared with atypical adenomatous hyperplasia, bronchioloalveolar adenocarcinomas, and normal lung, if reduced expression is defined as <80% positive cells, then ASC expression was significantly reduced in invasive carcinomas (P < 0.01).Taken together, these results indicate that the reduced expression of ASC/TMS1 is confined during lung tumor progression to the invasive stages of the disease.
Hypermethylation of the ASC/TMS1 promoter is a late event in lung cancer progression.We next investigated how the reduced expression of ASC/TMS1 in the progression stages of lung cancer relates to the presence of promoter hypermethylation.First, ASC/TMS1 methylation in the promoter CpG island was seen in 27.0% (41 of 152) of all types of primary lung cancers examined (Fig. 2A).In contrast, this methylation change was absent in all normal lung samples resected from lung cancer-free patients (0 of 6; Fig. 2A).There was a difference in the incidence of methylation between different histopathologic types of cancers.Thus, ASC/ TMS1 methylation was seen in 32.1% (26 of 81) of adenocarcinomas, 13.2% (7 of 53) of squamous cell carcinomas, 38.5% (5 of 13) of large-cell carcinomas, and 60% (3 of 5) of small-cell lung cancers.
Correlation was seen between ASC/TMS1 hypermethylation and reduced expression of the protein, although this epigenetic change seems not to be the sole process associated with low levels of the protein in invasive lung cancers.Twelve of 20 samples analyzed with reduced expression showed promoter hypermethylation whereas none of seven with normal expression showed this change (P = 0.008; Supplementary Table S2).
Other mechanisms accounting for low protein expression in the eight cancers without ASC/TMS1 hypermethylation are not known but could reflect some incidence of mutations or other epigenetic mechanisms.Consistent with the immunohistochemical studies, we rarely found promoter hypermethylation in precancerous lesions.Of 23 such lesions examined (5 lesions of squamous dysplasia and 18 lesions of atypical adenomatous hyperplasia), we observed ASC/TMS1 promoter hypermethylation in only one (Fig. 2B-D).
To further relate the methylation changes to disease progression, a PCR in situ technique (29), MSP-ISH, was used to examine the cell populations harboring the ASC/TMS1 promoter hypermethylation (Fig. 3).For invasive adenocarcinomas, cells with a hypermethylated ASC/TMS1 promoter were easily detected (arrowhead and large arrow) whereas no cells with a hypermethylated promoter Figure 3. ASC/TMS1 methylation in lung cancers and precancerous lesions determined by MSP-ISH.In an invasive adenocarcinoma (T ), which showed hypermethylation in the liquid PCR studies, a positive signal is seen frequently in the cancer cells (large arrows ) and is negative in the adjacent stromal (S) cells (small arrow ) after MSP-ISH with the methylated primer/probe set (A).Strong signals were seen in the stromal cells with the unmethylated primer/probe set (white arrow ; B ).In a bronchioloalveolar adenocarcinoma which did not show hypermethylation in liquid PCR, positive signal is not seen with the methylated primer/probe set (C ), whereas both the cancer cells and normal cells of the alveolus showed a signal after MSP-ISH with the unmethylated primer set (D ).In atypical adenomatous hyperplasia, positive signals are not seen in the preneoplastic cells with the methylated primer/probe set (E) but positive signals are observed with the unmethylated primer/probe set (F ).Bar, 100 Am (F ).
were seen in the stroma (small arrow; Fig. 3A).Strong signals were seen in the stromal cells with primers specific for unmethylated sequences (white arrow; Fig. 3B).In two cases of bronchioloalveolar adenocarcinomas (Fig. 3C and D), the noninvasive stage of adenocarcinoma, only unmethylated ASC/TMS1 alleles were detected and the same was true for two cases of atypical adenomatous hyperplasia (Fig. 3E and F), the precursor lesion for adenocarcinoma.Methylated ASC/TMS1 is correlated with advanced stages of human adenocarcinoma.Tumor samples from sets of patients from both the United States and Japan were used to correlate ASC/ TMS1 hypermethylation with stage and aggressiveness of lung cancers.Because initial analyses showed very similar data for both populations, the findings were pooled from the two groups.For squamous cell carcinomas, ASC/TMS1 hypermethylation showed no correlation to any of the clinical status variables queried.In contrast, several distinct correlations were observed for adenocarcinomas (Table 1).First, the methylation status of ASC/TMS1 was correlated with lymphatic invasion within the resected lung specimens as reported in patients' pathology records.Invasion was observed in 65% of tumors with ASC/TMS1 hypermethylation versus 25% of those without (P = 0.001).Second, 58% of patients with a hypermethylated gene versus only 15% without had metastasis to the lymph nodes (P < 0.001).Finally, advanced stages of adenocarcinoma had a distinctly higher prevalence of ASC/TMS1 hypermethylation (Table 1).Over 80% of tumors which harbored ASC/TMS1 came from patients with stage II disease or higher whereas close to 80% of patients with stage I tumors lacked methylation (P < 0.001).
Interestingly, despite the correlation of ASC/TMS1 hypermethylation with later-stage disease, and intrapulmonary lymphatic invasion, in the tumors from patients with adenocarcinoma, longitudinal follow-up of 96 stage I patients did not reveal a correlation between either of these changes and either incidence of recurrence and/or survival over a 5-year period (data not shown).However, only 18 of the 96 tumors showed ASC/TMS1 methylation and this reduced the power for the longitudinal follow-up.In fact, half of the patients with hypermethylated ASC/TMS1 in their tumors (9 of 18) actually recurred, but this incidence did not achieve significant difference as compared with patients with unmethylated tumors.Larger studies will be required to determine whether the methylation status of ASC/TMS1 may predict the behavior of stage I disease.
The presence of ASC/TMS1 promoter hypermethylation in sputum DNA correlates with very high risk for and/or presence of lung cancer.Based on all of the above studies, and especially because 60% of tumors from patients with stage II to IV adenocarcinomas harbor ASC/TMS1 hypermethylation (Table 1), we hypothesized that the presence of ASC/TMS1 promoter hypermethylation in sputum DNA could be a sign for later stages of lung cancer.This might be so even when such disease is occult with respect to routine clinical staging.Therefore, methylation status of ASC/TMS1 was determined in sputum recovered from four different groups including patients with stage III NSCLC, patients newly diagnosed with stage I NSCLC, patients resected for stage I NSCLC, and smokers who were cancer-free.
The patients with stage III NSCLC served to ask whether hypermethylation of ASC/TMS1 in sputum frequently correlated with the presence of late-stage lung cancer.Seventy-five percent of the sputum samples from these patients were defined as adequate by the presence of deep lung macrophages or Curschmann's spiral (23).However, the detection of methylation of the ASC/TMS1 gene was independent of sputum adequacy and 41% of these patients harbored the hypermethylation change (Fig. 4A and B; Table 2).Most strikingly, the prevalence for methylation in sputum for patients with adenocarcinomas (56%; Table 2) is virtually identical to the incidence (52%) found in direct analyses of the stage III tumors (Table 1).Furthermore, a large number of the 11 patients with a diagnosis of NSCLC, not otherwise specified (NOS in Table 2), would be expected to have undifferentiated adenocarcinomas and 64% of these individuals had the sputum marker.The low incidence for the finding of hypermethylation in patients with squamous cell carcinoma (8.3%) also fits well with the incidence of this change in stage III tumors of this histologic type (15%; Table 1).
In contrast, compared with patients with stage III disease, and consistent with our failure to find the promoter methylation in precancerous and preinvasive cancer stages, methylation of ASC/ TMS1 was detected in sputum from only 2 of 85 (2.4%) smokers without cancer, despite these individuals having a z15 pack year history of smoking (Fig. 4A and B).Ninety-one percent of specimens collected from these cancer-free smokers were judged adequate by cytologic examination.There was an increase in odds ranging from 7.2 to 28.6 for finding the hypermethylation marker in sputum from the cancer patients versus in sputum from heavy, at risk, smokers.Consistent with our hypothesis from the tumor tissue studies that methylated ASC/TMS1 is correlated with advanced stages of lung cancer, the methylation rate in sputum of stage I lung cancer patients was much lower than stage III lung cancer patients (Fig. 4A and B; Table 2).Methylation of ASC/TMS1 was detected in sputum from 3 of 18 (16.7%)stage I lung cancer patients.Importantly, when analyzed for specific tumor types, ASC/TMS1 methylation in sputum closely correlated with disease progression for adenocarcinoma.Methylation of ASC/TMS1 was detected from only 1 of 10 (10%) stage I adenocarcinoma patients versus 9 of 16 (56.2%)stage III adenocarcinoma patients (P = 0.023; Fig. 4C and D).
Finally, we examined the potential that sputum hypermethylation of ASC/TMS1 might predict the presence of occult late-stage lung cancer, or recurrent cancer, in a very high risk population who had no clinical signs of disease.For this study, we examined the marker in sputum samples from individuals who had undergone curative resection for stage I NSCLC and who were disease-free for at least 6 months before sputum sample collection ( from 6 to 36 months).These individuals are known to develop, by 3 years, tumor recurrence and/or second lung tumors at the rate of 20% to 38% postsurgery (31)(32)(33).In these individuals, 79% of specimens were judged cytologically adequate and, strikingly, ASC/TMS1 promoter methylation in sputum was present in 23.8% (10 of 42) of the patients (P V 0.05, compared with the high-risk smokers).In this study, 27 of the patients were resected for adenocarcinoma.Overall, in this group, given that the occurrence of ASC/ TMS1 hypermethylation in late-stage adenocarcinomas is f50% (Table 1), and given that f30%, or 9, of these patients would be expected to have recurrent or second tumors, f15%, or 4 patients, could theoretically be detected by the gene marker.The finding of methylation in 18%, or 5 of the adenocarcinoma cases (Table 2), very well approximates the expected risk for recurrence.Interestingly, although the number of patients (N = 8) is not large, an unexpectedly high percentage (37.5%;Table 2), as compared with actual findings for ASC/TMS1 hypermethylation in tumors of patients with initial squamous cell carcinomas (overall, 15%; Table 1), was also positive in sputum.

Discussion
The results of this study suggest that the loss of ASC/TMS1 function may be a critical event for progression and subsequent metastasis of late stages of lung cancer.Interestingly, loss of function for another proapoptotic gene, DAP kinase, was also associated with metastasis in a mouse model of lung carcinoma (34) and was later found to be methylated in this and other human  (35)(36)(37)(38).In our present study, hypermethylation of ASC/TMS1 is tightly correlated with lymphatic invasion and lymph node metastases in adenocarcinoma, but not squamous cell carcinoma of the lung.The reason for this is unknown but could indicate an important role for ASC/TMS1 in the metastatic process for this particular type of lung cancer.
Another interesting facet of our data is the fact that despite correlation of ASC/TMS1 hypermethylation with intrapulmonary lymphatic invasion and with late-stage disease, these changes, when infrequently present in stage I tumors, did not predict recurrence and/or poor survival outcome for patients with such lesions.This may be because the number of patients available for study with hypermethylation of the gene in stage I tumors was quite small.In fact, half of the patients with hypermethylated ASC/ TMS1 in their tumors (9 of 18) actually recurred, but this incidence did not achieve significant difference as compared with patients with unmethylated tumors.Alternatively, it may then be that whereas silencing of the gene may be important for the progression of later stages of lung cancers once tumors reach this stage, it does not mediate progression per se from localized to distant metastatic disease.Nevertheless, our initial marker studies indicate that the hypermethylation change may be an excellent marker for the presence of metastatic disease once it has begun to occur.In the present study, we focused on this possibility by studying sputum DNA.
To date, most hypermethylation changes seem to be most valuable as molecular markers for cancer detection and/or risk (25,39).In this regard, many important genes that are hypermethylated in cancer harbor this change in precancerous lesions, such as p16, SFRP, and MLH1 (29,(39)(40)(41)(42)(43)(44).Thus, with respect to serving as tumor markers, many tumor suppressor genes may predict risk rather than detect actual cancer.This has proved true; for example, in the use of promoter hypermethylation of genes such as p16 and MGMT where these markers appear frequently in sputum DNA from high-risk smokers who do not have signs of cancer (25,(45)(46)(47)(48).There is then a great need, in the case of sputum, for markers which can be used in samples obtained by noninvasive means to find hypermethylated genes that either mark the presence of cancer and/or correlate with sensitive detection for unrecognized presence of advancing stages of these diseases.Our findings for hypermethylation of ASC/TMS1 with respect to the data we have presented for the biology of lung cancer progression, and the actual initial studies of sputum DNA, indicate that this gene may provide such a marker.
In patients studied who have cancer, the sensitivity for detecting methylation in sputum correlates well with the frequency expected for the tumors to harbor the ASC/TMS1 hypermethylation change.As importantly, the lack of methylation in 98% of sputum samples from individuals who smoke heavily, but have no clinical signs of cancer, suggests a very high specificity for cancer prediction using this marker.This said, precise implication for finding ASC/TMS1 hypermethylation in sputum from 24% of individuals postresection for stage I NSCLC needs further clarification.It is well established that these individuals are at very high risk for recurrent and/or second lung cancers (49).All of our progression data could suggest that, in this setting, invasive late-stage lesions, although not clinically detected, are present in those individuals who had ASC/ TMS1 hypermethylation in the sputum.Longitudinal studies will be essential to document this important possibility and to determine if detection of hypermethylation of this gene in sputum DNA can identify a substantial group of patients at risk for recurrent disease after resection of early-stage lung cancer.These future longitudinal studies seem highly warranted and positive results could greatly help in pinpointing individuals who might benefit most from adjuvant treatment approaches.

Figure 1 .
Figure1.Immunohistochemical staining using the anti-ASC/TMS1 monoclonal antibody.In normal lung, ASC/TMS1 is expressed in the pneumocytes (A).In atypical adenomatous hyperplasia, ASC/TMS1 expression was strongly observed in atypical cells (B ).ASC/TMS1 is also detected in bronchioloalveolar adenocarcinoma (C ).In a typical invasive adenocarcinoma, ASC/TMS1 expression is distinctly reduced (D).In normal bronchus (E) and in squamous metaplasia (F ), ASC/ TMS1 is strongly expressed and the protein is also detected in squamous dysplasia (G).However, ASC/TMS1 expression is often reduced in squamous cell carcinoma (H ).Bar, 100 Am.

Figure 2 .
Figure 2. Methylation analysis of ASC/ TMS1 gene promoter in primary lung cancer and precancerous tissues.A, examples of MSP analysis in primary lung cancer (LC) samples.Numbers identifying individual patient samples are listed on top.M, PCR products for methylation-specific primers; U, PCR products with unmethylated-specific primers; H 2 O, no DNA added, refers to water control.MSP analysis in lung samples from cancer-free individuals (NL1-NL6 ) are also shown.B, examples of MSP analysis in precancerous lesions.The HCT116 colon cancer cell line is shown as a positive control for the fully methylated ASC/TMS1 gene.LY, normal peripheral lymphocyte DNA is shown as a negative control.AAH, atypical adenomatous hyperplasia; SD, squamous dysplasia.C, summary of methylation analysis of ASC/TMS1 in atypical adenomatous hyperplasia and adenocarcinoma.D, summary of methylation analysis of ASC/TMS1 in squamous dysplasia and squamous cell carcinoma.

Figure 4 .
Figure 4.A and B, methylation analysis of ASC/TMS1 promoter in sputum for current and former smokers, stage I lung cancer patients, resected stage I lung cancer patients, and patients with stage III lung cancers.Vertical axis, numbers of sputum samples examined (A) and the percentage of sputum samples examined (B ).C and D, methylation analysis of ASC/TMS1 promoter in sputum from adenocarcinoma patients.Vertical axis, numbers of sputum samples examined (C ) and the percentage of methylation of ASC/TMS1 promoter in sputum (D ).*, P = 0.023, stage I adenocarcinoma patients versus stage III adenocarcinoma patients; **, P = 0.014, resected stage I adenocarcinoma patients versus stage III adenocarcinoma patients.

Table 2 .
Prevalence for promoter hypermethylation of ASC/TMS1 in sputum