Prognostic Significance of DNA Ploidy, S-Phase Fraction, and Tissue Levels of Aspartic, Cysteine, and Serine Proteases in Operable Gastric Carcinoma¹

The traditional factors of GC³ are still inadequate for the prognostic characterization because patients with identical clinical or pathological stages may differ widely in the clinical evolution (1). It is therefore extremely important to define the prognostic factors that may help to recognize the more aggressive types of such tumors. Reliable and reproducible prognostic indicators are being investigated to help clinicians identify high-risk groups and address more rational treatment decisions. Thus, the identification of factors more strictly related to tumor biology may be useful in characterizing patients with a different prognosis. Recently, more and more attention has been focused on the cell kinetics of many human tumors with possible prognostic significance, including gastrointestinal neoplasias (2). At the present time, one of the most widely used approaches is cellular proliferation rate measured by flow cytometric analysis (SPF), proposed as a rapid, simple, and particularly versatile method applicable to fresh, frozen, and paraffin-embedded tissues. In fact, some reports have observed that SPF was an independent prognostic variable in GC (3, 4, 5). Previous studies have also identified a DI and/or DNA ploidy as a useful biological factor, which may predict the clinical outcome of GC (4, 5, 6, 7, 8). Furthermore, a consistent bulk of investigations has shown that the expression levels of lysosomal aspartic and cysteine proteinase CD, CB, and CL and serine protease uPA may be significantly altered in tumor cells and tissues as compared to their normal counterpart (9, 10). These alterations appear to be associated with the malignant progression of several human neoplastic diseases including GC (10, 11, 12, 13, 14, 15, 16). However, recent studies indicate that at least for CB and CL, this phenomenon may very likely be due to an altered regulation and/or decreased affinity of specific intracellular inhibitors of these proteases (17, 18) or also, as recently reported for CB,the amplification of the gene encoding for these enzymes (19, 20), rather than to an increased enzyme activity per se.

The aims of this paper were, firstly, to compare DNA ploidy and SPF,their possible relations to the proteinases CD, CB, CL, and uPA, and that of SA, a specific intracellular inhibitor of cysteine proteinases and clinicopathological features; and secondly, to evaluate the possible prognostic role of either DNA ploidy and SPF or proteinases in GC.

Paired tumor and normal tissue samples collected (January 1992 to June 1996) by the Laboratory of Oncobiology of the University of Palermo from a consecutive series of 63 patients undergoing resective surgery for primary operable GC at a single institution (Department of Anatomy,Surgery, and Oncology) were prospectively studied. Briefly, exclusion criteria included: (a) history of previous neoplasias, and(b) prior chemotherapy or radiation therapy. Resection of the primary GC was performed in all cases. All patients underwent potentially curative tumor resection, including radical lymph node dissection. To avoid evaluator variability, all resection specimens and microscopic slides had been meticulously examined by a single pathologist (R. M. T.). In addition, the same pathologist assessed tumor size and site, depth of invasion, TNM stage, tumor grade(histological differentiation), and lymph node metastasis. The depth of tumor invasion was also evaluated histologically as reported by Watanabe et al. (13). Tumors were staged according to the latest TNM classification (21). Finally,metastatic cases were identified by clinical and histopathological analyses of neoplastic cells in organs such as lymph nodes, liver, and so forth along with the primary tumor. Clinicopathological data of all patients were available (Table 1). Postoperatively, all patients were checked at 3-month intervals for the first 2 years, at 6-month intervals for the next 2 years, and annually thereafter. The follow-up program included a clinical examination,routine check of circulating tumor markers (CEA, TPA, CA19.9, and CA72.4 assay), annual chest radiography, and endoscopy. Abdominopelvic computed tomography scan was also performed every year for the first 2 years. Disease relapse (local recurrence or distant metastasis) was confirmed histologically where possible. All of the patients with relapse disease after surgery received a standard chemotherapeutic regimen for GC (5-fluorouracil, epirubicin,methotrexate, etoposide, doxorubicin, cisplatin).

Multiple samples of the primary tumor tissue were taken from different representative areas and processed within 30 min of surgical resection or biopsy. All tissues were carefully trimmed to remove as much nonneoplastic tissue as possible, avoiding the nonviable areas. Furthermore, from each patient, multiple samples of normal mucosa (as confirmed by histology) were taken in a corresponding nontumor area as far as possible from the tumor site to be used as a control for biochemical/cytochemical and flow cytometric analysis. The tissues were bisected, one-half of each sample was processed for pathological examination, and the remaining half of the sample pool was immediately frozen and stored at −80°C until analyzed. The adequacy of the material was checked on frozen tissue sections, and only tissue samples with >80% tumor content were used in subsequent biochemical/cytochemical and flow-cytometric analysis.

DNA flow cytometry was performed on mechanically disaggregated samples of frozen tumor tissue as in previously described protocols (22). DNA histogram analysis was carried out by the Multicycle Software Program (Phoenix Flow Systems, San Diego, CA),including systematic background substraction (23, 24). DNA ploidy, DI, and SPF were determined as previously reported (22). Briefly, healthy gastric mucosa was used as an internal DNA-diploid control for each sample. Tumors with a DI = 1 were defined as DNA-diploid, and tumors with lower or higher DI values were considered DNA-aneuploid if they contained >10% aneuploid cells.

Paired sets of tumor and normal tissue specimens (70–210 mg) were thawed, minced, and suspended at 25 mg/ml in a tissue homogenization buffer [50 mm Tris-HCl, 5 mm EDTA, 0.5 mm DTT, and 0.2% v/v Triton X-100 (pH 6.9)]. Samples were homogenized at 2°C in an ice slurry in a Willelms Polytron PT-10 homogenizer (Kinematica, Luzern, Switzerland) set to rotate at 9, with two bursts of 30 s and a 1-min intercooling period. The homogenate was centrifuged at 500 g for 10 min. The supernatant was respun at 13,000 g for 15 min at 4°C in a preparative ultracentrifuge (Beckman Model L80). The 13,000-g supernatant (i.e., cytosol) was stored at −80°C until assays. Total CD content was determined, as previously reported (25), by an antibody-based immunoenzymatic assay kit (Triton, Ciba Corning, Diagnostics, Alameda,CA) according to the manufacturer’s instructions. The detection limit was 0.018 pmol/ml. Total CB, CL, and SA tissue levels were determined by commercially available solid phase ELISA kits (KRKA d.d.,Novo mesto, Slovenia) in cytosol diluted 1:2 or 1:5 (v/v) according to the manufacturer’s protocol. Antibodies used for CB and CL assays recognized mature forms, precursor molecules, and inhibitor-cathepsin complexes. For the determination of SA, MoAbs used in sandwich ELISA were obtained by hybridoma cell lines different from those used to produce other cysteine protease inhibitor MoAbs (26). The minimal detectable concentrations were 0.9 ng/ml for CB, 1.6 ng/ml for CL, and 1.04 ng/ml for SA. uPA levels were determined by a commercially available immunoluminometric assay kit (Byk Sangtec, Cormano, Italy). The MoAbs used bind either to the proenzyme or to the active form as well as to the enzyme bound to its receptor (uPA-R) or to its endogenous inhibitor (PAI-1). The detection limit was reported to be <0.005 ng/ml. The cytosol protein content was determined by a commercially available colorimetric micromethod kit(Sigma, St. Louis, MO).

Because of their asymmetric distribution, the Mann-Whitney Utest was used to evaluate differences in proteinases and SPF values between the two categories of DNA ploidy. The nonparametric Spearman correlation method was used to evaluate the correlation between the proteinases variables examined. DFS was measured from the day of primary surgery to the date of first relapse (locoregional or metastatic), and OS was measured from the day of surgery to the date of death specifically due to the tumor. Clinical and morphobiological variables were examined univariately by means of the Kaplan-Meier method (27), and significance of differences for each prognostic factor was assessed by the log-rank test and Wilcoxon test or test for trend when appropriate (28). Multivariate analysis was carried out by means of the Cox’s logistic regression model using a backward procedure (29). The null hypothesisβ = 0 was tested by the Wald statistic. The relative contribution of clinicopathological and biological variables was assessed by means of the likelihood ratio test. For the prognostic variables contributing significantly to the model, the effect was calculated in terms of RR and the associated 95% confidence limits. The 0.05 level of probability was taken as significant.

Adequate DNA histograms were obtained for all normal and tumoral gastric tissues by means of flow-cytometry. The coefficients of variation of the diploid G₀/G₁ peak ranged from 2.6% to 4.8% (mean, 3.6%). DNA aneuploidy was found in 71% of the cases (45/63), while 9% of these (4/45) showed multiclonality. The SPF ranged from 3.5% to 37.2% (median, 15.2% and interquartile range,12.1–19.1%). By using the SPF median value as the cutoff point,tumors were accordingly divided into low (≤15.2%) and high (>15.2%)SPF tumors. Table 2 gives a summary of the significant associations of DNA ploidy and SPF to the clinicopathological variables for GC. Both DNA ploidy and SPF were associated with TNM stage and node status, whereas only DNA ploidy was related with depth of invasion. Furthermore, a significant association was found between flow cytometric variables (DNA ploidy and SPF). No association was found between DNA ploidy and SPF and any of the following factors: age, sex, histological grade, tumor size, and tumor site.

Table 3 reports the tissue levels of CD,CB, CL, uPA, and SA determined in GC and in the corresponding normal mucosa. Only CB, CL, and uPA levels were significantly increased in tumor tissue as compared to the paired normal counterpart(P < .05 for CL and P < .01 for CB and uPA; Table 3). Moreover, the SA levels were lower in tumor tissue as compared to normal tissue; however, this decrease was not statistically significant. CD tumor levels were significantly correlated with CB, uPA, and SA (P < .05), whereas CB levels were correlated with CD, uPA, and SA but not with CL (Table 4). SA levels correlate with all of the enzymes considered, but not with CL. Moreover, CB levels were significantly associated with TNM stage [I (median) = 93.3;II = 142.0; III-IV = 159.8; P < .05], node status [node negative (median) = 91.5; node positive =159.7; P < .06], histological grade[G₁ (median) = 63.2;G₂ = 121.7; G₃ = 159.8; P < .01] and DNA ploidy [Diploid (median) =92.4; Aneuploid monoclonal = 156.3; Aneuploid multiclonal = 250.9; P < .01].

The median follow-up of patients was 73 months (range, 12–108 months). At the time of analysis (April 1998), 31 patients had relapsed (2 had local-regional recurrence and 29 had distant metastases) and 30 had died of a tumor-related cause. The overall 5-year survival was 45.8%(SE, 6.9) for the whole series. At the univariate analysis, high stage,lymph-node positivity, DNA aneuploidy, and high SPF proved to be significantly related to quicker relapse and to shorter OS (Table 5), whereas depth of invasion was related only to survival. Univariate analysis failed to reveal any significant association between age, sex, histological grade, tumor location, tumor size, CB, CD, CL, uPA, or SA, and relapse, or survival. Fig. 1 shows the probability of disease-free interval in relation to SPF; Fig. 2 shows the probability of OS by SPF and DNA ploidy. In Cox’s model for multivariate analysis, only the significant variables at univariate analysis were considered. The estimated hazards ratio was higher for SPF (>15.2%), which appeared to be the most relevant indicator of relapse (RR = 8.50), whereas multiple regression analysis carried out for OS showed that SPF and DNA ploidy provided independent information for predicting risk of death (Table 6).

Many attempts have been made to find new prognostic indicators for patients with GC to identify the subgroup with highly aggressive tumors and a high likelihood of disease relapse or death. Aggressive surgical approaches, such as extended lymph node dissection or at least intensive postoperative therapy, might be options for these poor prognosis patients. There is still some controversy as to whether or not the DNA content and SPF are significant prognostic factors in GC. Several reports have shown that patients with DNA-aneuploid and/or high SPF tumors had a worse prognosis than those with DNA-diploid and/or low SPF tumors (3, 4, 5, 6, 7). Recently, in a study performed on fresh tumor specimens taken from 76 patients with GC, Abad et al. (6) found that DNA aneuploidy, histological type,and the presence of extranodal metastases were independent predictors of OS. By contrast, SPF did not predict patient outcome. These results agree with those of Setala et al. (7) and Victorzon et al. (8), whose respective analysis of 289 and 242 GC cases led to the observation that DNA ploidy and TNM stage significantly predicted OS, whereas no significant difference was found between SPF and prognosis. In contrast, Lee et al.(3), in a 5-year prospective study of 217 patients with GC, found that SPF was an independent prognostic variable, whereas DNA ploidy was not associated with patient survival. Finally, Yonemura et al. (4) and Ohyama et al.(5), in their respective series of 493 and 117 patients with GC using multivariate analysis, found that DNA content and SPF(measured by in vivo bromodeoxyuridine) were independent prognostic factors for OS. In our prospective study, which was based on univariate and multivariate analyses with established prognostic parameters (depth of invasion, TNM stage, node status, histological grade), we found that SPF was the only predictor of disease relapse. Furthermore, the present study clearly shows that both DNA ploidy and SPF are significant and independent prognostic factors for OS in patients with GC who have undergone surgical resection. Our data also show that DNA multiploidy represents the most powerful predictive indicator for survival. These results suggest that studies of pathological prognostic factors should include these biological variables. Moreover, because DNA analysis of endoscopic biopsy specimens correlates well with surgical specimen DNA analysis (30), preoperative evaluation of flow cytometric variables may be useful for predicting prognosis and choosing an adequate therapeutic modality in patients with GC. The frequency of DNA aneuploidy has been reported in a variable percentage of GC, from 36%to 71% (3, 4, 5, 6, 7, 8, 31, 32). In the present series, the DNA-aneuploid rate (71%) was among the highest published thus far. This is probably due to the use of multiple tissue sampling of all cases studied, which greatly reduces the probability of missing DNA-aneuploid clones, and to the method chosen to preserve the samples(freezing at −80°C), which makes it possible to obtain a higher histogram resolution compared with paraffin-embedded samples. The present study shows that DNA ploidy and/or SPF are significantly associated to depth of invasion, advanced clinical stage, and the presence of lymph node metastases. Histopathological features are universally considered bad prognostic factors (14). Thus,the association with these variables suggests that flow cytometric indicators are related to tumor aggressiveness. This is in accordance with previous studies (6, 7, 8). Moreover, the perturbance of cell cycle due to an increased proliferative activity of aneuploid clones or to a prolongation of the S phase may enhance the risk of genetic alterations. This phenomenon may result in the onset of DNA-aneuploid cells, which would explain the significant relationship between high SPF and aneuploidy observed in the present study. Finally,unlike other investigators (7, 31), we did not observe a significant association between DNA ploidy and tumor site (fundus or cardia) as noted in three other series (3, 8, 33,). The present results confirm, in part, the data obtained from other studies,which have reported elevated content or activity levels of CB, CL, and uPA in GC tissues as compared to normal gastric mucosa (11, 12, 13, 15, 16, 32).

However, only CB proved to be significantly associated with most of the clinicobiological parameters of progression of this tumor. Moreover,the intracellular cysteine proteinase inhibitor SA levels were lower in tumor tissue as compared to those in its normal counterpart. These results indicate that the alterated regulation of cysteine proteinases/SA ratio at the tumor level may facilitate the progression of GC. In fact, this phenomenon has been observed “in vitro ” in human cancer cell lines, and it has proved to be associated with a more malignant phenotype (34, 35). Furthermore, decreased levels of SA have also been seen “in vivo ” in different human tumors of epithelial origin and have proved to be associated with the metastatic potential and differentiation of these tumors (11, 18, 19). These data suggest that the increased level of CB, together with other proteolytic enzyme (metalloproteases), which appear to determine the malignant phenotype of GC, may play a major role in the progression of this tumor (34, 36, 37). However, the recent observations,which showed that proteinase inhibitors at subinhibitor concentrations may act as mitogens thus facilitating cell proliferation (19, 35), indicate that SA may also directly share the malignant progression of GC with a mechanism not related to its proteinase-inhibiting activity. This may in part explain the direct and not the inverse correlation, which might be expected,observed between CB and SA at the tumor level. On the other hand,conflicting results have been obtained regarding the distribution of CD. In fact, unlike the results reported in other studies (14, 16), we did not observe any significant difference in the levels of this protease between GC and normal mucosa or the association with most of the clinicobiological parameters considered, including disease relapse and death. These conflicting results may be partly explained by the different methodological approaches used to assess this parameter(i.e., immunohistochemistry versus immunoenzyme assay). Our investigations also showed significantly higher levels of uPA content in tumor tissue as compared to those in normal tissue; no significant association was found, however, between the uPA system and the clinicobiological parameters of GC. These results are partly in agreement with those from other studies (12, 16, 37) and further indicate that the evaluation of the single components of the uPA system, namely the uPA receptor or uPA inhibitors PAI-1 and PAI-2,may be more appropriate for the assessment of the prognostic significance of this serine protease in human cancer (15, 16). However, it cannot be ruled out that the different results we obtained on the prognostic significance of the uPA complex may also be due to the different number of patients evaluated in our study.

In conclusion, DNA ploidy and SPF may be reasonably considered as indicators of biological and clinical aggressiveness. If our results are confirmed in larger prospective studies, it might be possible to add such variables to other prognostic factors for more aggressive strategies such as extended lymph node dissection or intensive postoperative therapy in high risk groups (patients with high SPF and/or DNA-aneuploid tumors) to prevent disease relapse or death. In addition, the clinical impact of the CB/SA system as biochemical prognostic parameters in this tumor has still to be sufficiently assessed.