Typical allelic profiles of microsatellite stability in constitutional DNA samples (A) and microsatellite instability in tumoral samples (B). bp Indicates base pair; N, lymph nodes; and T, tumor size.
Kaplan-Meier survival estimates according to microsatellite instability (MSI) status in 250 patients who underwent gastrectomy. The median (95% confidence interval) survival was 26.7 (20.8-32.5) months for patients with microsatellite-stable (MSS) tumors, and the median was beyond the observation period for patients with MSI tumors (P < .001).
Kaplan-Meier survival estimates according to microsatellite instability (MSI) in 250 patients who underwent gastrectomy. The median (95% confidence interval) survival was 26.7 (20.8-32.5) months for patients with microsatellite-stable (MSS) tumors and 18.5 (0.0-50.1) months for patients with MSI tumors having instability at 2 to 4 markers (MSI/<5), and the median was beyond the observation period for patients with MSI tumors having instability at all 5 markers (MSI/5) (P < . 001).
Kaplan-Meier survival estimates according to the number of microsatellite instability (MSI) markers in 180 patients with M0 disease who underwent potentially curative (R0) gastrectomy. The median (95% confidence interval) survival was 48.9 (31.0-86.8) months for patients with microsatellite-stable (MSS) tumors, and the median was beyond the observation period for patients with MSI tumors having instability at 2 to 4 markers (MSI/<5) or with MSI tumors having instability at all 5 markers (MSI/5) (P = . 004).
Giovanni Corso, Corrado Pedrazzani, Daniele Marrelli, Valeria Pascale, Enrico Pinto, Franco Roviello. Correlation of Microsatellite Instability at Multiple Loci With Long-term Survival in Advanced Gastric Carcinoma. Arch Surg. 2009;144(8):722–727. doi:10.1001/archsurg.2009.42
Microsatellite instability (MSI) correlates with clinicopathologic characteristics and long-term prognosis in patients having gastric carcinoma.
Analysis of prospectively collected data and biologic material.
Tertiary University Hospital, Policlinico “Le Scotte,” Siena, Italy.
Two hundred fifty patients with gastric carcinoma.
Main Outcome Measures
Five mononucleotide repeats (BAT-26, BAT-25, NR-24, NR-21, and NR-27) were analyzed in these patients.
An MSI phenotype was identified in 63 patients (25.2%) and correlated with specific clinicopathologic characteristics. Favorable prognosis was confirmed for patients with an MSI phenotype in univariate (P < .001) and multivariate (P = .05) analyses. Significant differences in clinicopathologic characteristics and long-term prognoses were observed among patients with microsatellite-stable tumors, tumors having instability at 2 to 4 markers, and tumors having instability at all 5 markers (MSI/5). The MSI/5 phenotype was associated with older age (P < .001), female sex (P = .001), antral tumor location (P = .04), intestinal histotype (P = .003), and less infiltration of the serosa (P = .006); lymph node involvement was rare (P < .001) and was limited to few (median, 3) metastatic lymph nodes (P = .001). Long-term survival of patients with the MSI/5 phenotype is favorable and was confirmed in multivariate analysis (relative risk vs patients with stable tumors, 0.32; 95% confidence interval, 0.16-0.63; P = .002).
Compared with stable tumors, MSI tumors have distinct clinicopathologic features and are associated with a better prognosis. Patients with the MSI/5 phenotype have a very good prognosis.
Despite a decreasing incidence during the last few decades, gastric carcinoma (GC) remains the fourth most common cancer worldwide and the second leading cause of cancer-related death.1 The overall mortality has decreased during the same period, although the long-term prognosis remains poor, with 5-year survival rates that rarely exceed 40%.2
As with most other tumors, GC develops through the accumulation of genetic and epigenetic alterations affecting oncogenes and tumor suppressor genes. It is widely accepted that microsatellite instability (MSI) and chromosome instability are 2 major genomic instability pathways involved in GC pathogenesis.3 Microsatellite instability has been described as a frequent genetic alteration in colon cancer4 and in several other cancer types.5,6 Although microsatellite alterations have been widely investigated in familial and sporadic colon cancer,7 limited and inconclusive data are available for stomach cancer.
In the present study, we analyzed a large cohort of patients with GC using a panel of 5 quasimonomorphic mononucleotide repeats (BAT-26, BAT-25, NR-24, NR-21, and NR-27).8 We assessed clinicopathologic characteristics and long-term prognoses of patients having MSI tumors.
The study analyzed 250 patients with primary GC and operated on at the Division of Surgical Oncology, Tertiary University Hospital, Policlinico “Le Scotte,” University of Siena, Siena, Italy, between January 1, 1990, and December 31, 2004. All patients underwent resection, including 183 patients with potentially curative (R0) tumors and 35 patients with microscopic (R1) and 32 patients with macroscopic (R2) residual tumors; 20 patients (17 with R1 and R2 residual tumors) had metastatic disease at the time of operation. Clinicopathologic and follow-up data were prospectively collected for the whole cohort. The median follow-up period for surviving patients was 92.1 months (range, 36.3-260.9 months).
Informed consent was obtained from all patients. The study was approved by the appropriate local ethics committees.
The objective of surgery was complete resection of the tumor, although gastrectomy was performed even in patients with untreatable disease when required for symptom palliation. Gastric resection was performed based on the extent of tumor in the stomach. Distal subtotal gastrectomy was preferred for tumors located in the lower and middle thirds of the stomach provided that the proximal resection margin remained at least 5 cm from the tumor edge; otherwise, total gastrectomy was performed.9 Extended (D2) and superextended (D3) lymphadenectomy (in 157 patients [62.8%]) was performed in most R0 resections and has been routinely used since 1995, except for patients deemed unfit for this procedure.10 None of the patients received preoperative chemotherapy or chemoradiotherapy.
Tumor size was calculated by measuring the largest diameter of the neoplasm. The most recent version of the TNM classification by the American Joint Committee on Cancer11 and the International Union Against Cancer was used for histopathologic staging. Level and number of metastatic lymph nodes were also recorded for each patient. Histologic classification followed the criteria by Lauren12; mixed-type tumors were considered together with diffuse (nonintestinal) type. Lymphatic and vascular invasion was assessed for 176 and 184 patients, respectively.
Frozen samples of normal mucosa and tumor tissue were available for each patient in our biologic material bank. Tumoral mucosa and corresponding normal mucosa of the stomach were flash-frozen in liquid nitrogen. Tumoral and constitutional DNA were extracted after histopathologic confirmation using a commercially available DNA purification kit (Puregene; Gentra Systems, Inc, Minneapolis, Minnesota) and following the manufacturer's instructions.
Microsatellite analysis was evaluated using 5 quasimonomorphic mononucleotide repeats, namely, BAT-26, BAT-25, NR-24, NR-21, and NR-27.8 The 5′ antisense primer was labeled with a fluorescent dye using FAM for BAT-26 and NR-21, NED for BAT-25 and NR-27, and VIC for NR-24 (ABI PRISM Primer Pairs; Applied Biosystems, Foster City, California).
The 5 mononucleotide repeats were coamplified in one multiplex (pentaplex) polymerase chain reaction (PCR) using the protocol for amplification of microsatellite loci (Multiplex PCR; Qiagen, Studio City, California). The pentaplex PCR was performed on tumoral and matched constitutional DNA of the same patient with GC. The allelic profiles of these 5 mononucleotides were detected on an automated DNA sequencer (ABI PRISM 3100 Genetic Analyzer, Applied Biosystems) according to the manufacturer's protocol (Figure 1).
According to the definition of the National Cancer Institute5 workshop on MSI for cancer detection and familial predisposition, we considered a tumor as having MSI whenever 2 or more markers showed instability on 5 loci. The clinicopathologic characteristics and survival rates associated with tumors having low instability (one locus involved) are similar to those of stable tumors5,13 and were considered together (microsatellite-stable [MSS] phenotype). DNA samples showing an abnormal allelic shift were confirmed by a second PCR and gene scan.
Analyses were performed using commercially available statistical software (SPSS 14.0 for Windows; SPSS Inc, Chicago, Illinois). Statistical associations between clinicopathologic characteristics and MSI status were assessed by χ2 test for categorical variables and by t test or analysis of variance for continuous variables. Survival curves were estimated using the Kaplan-Meier method and were compared using log-rank test. Multivariate analysis was performed using a Cox proportional hazards regression model by considering the following risk factors: sex, age (older vs the median or younger), tumor location (other vs antrum), Lauren12 histotype (nonintestinal vs intestinal), depth of tumor invasion (pT2, pT3, or pT4 vs pT1), lymph node involvement (pN1, pN2, or pN3 vs pN0), presence of systemic metastasis (M1 vs M0), and R category (R1 or R2 vs R0). Postoperative mortality was assessed, with deaths unrelated to tumor recurrence considered censored observations at the time of death. Because differences in clinicopathologic features and survival rates were observed within the MSI phenotype for tumors with instability at 2 to 4 markers (MSI/<5) vs tumors with instability at all 5 markers (MSI/5), an analysis was performed to separately investigate MSS, MSI/<5, and MSI/5 tumors. P < .05 was considered statically significant.
Among 250 patients with GC, MSI was identified in 63 GCs (25.2%). The MSI phenotype was significantly associated with older age (mean [SD], 72.8 [10.7] vs 66.7 [10.6] years; P < .001), and it was more frequent in women (55.6% vs 37.4%, P = .02) with intestinal histotype tumors (81% vs 59.9%, P = .002). The MSI phenotype was rarely associated with systemic metastases (1.6% vs 10.2%, P = .03) or with pT3 or pT4 tumors (38.1% vs 60.4%, P = .002). Compared with the MSS phenotype, pN+ lymph node involvement was less frequent in patients having the MSI phenotype (57.1% vs 75.9%, P < .001), with fewer metastatic lymph nodes (mean [SD], 3.6 [7.1] vs 9.4 [12.1]; P = .001). Conversely, no differences in tumor size (mean [SD], 58.8 [2.9] vs 57.8 [3.2] mm; P = .82) or percentage of R1 or R2 resections (19% vs 29.4%, P = .14) were observed.
The presence of lymphovascular invasion was also evaluated. Both vascular invasion (21.4% vs 53.5%) and lymphatic invasion (33.3% vs 69.3%) were rarely associated with MSI tumors compared with MSS tumors (P < .001).
Kaplan-Meier survival estimates according to MSI phenotype are plotted in Figure 2. In patients with M0 and R0 disease, 5-year survival was 46% (median, 48.9 months; 95% confidence interval, 31.0-66.8 months) for patients with MSS tumors and 71% (the median was beyond the observation period) for patients with MSI tumors (P = .002).
Multivariate analysis of Cox proportional hazards regression models confirmed MSI as an independent predictor of survival (relative risk vs patients with MSS, 0.62; 95% confidence interval, 0.38-1.00; P = .05). The models controlled for age (P = .74), sex (P = .09), tumor location (P = .70), Lauren12 histotype (P = .10), depth of tumor invasion (P < .001), lymph node involvement (P < .001), presence of systemic metastasis (P = .005), and R category (P < .001).
Table 1 shows differences in clinicopathologic characteristics associated with the types of tumors. These include MSS, MSI/<5, and MSI/5 tumors.
The MSI/5 phenotype correlated with older age (P < .001), female sex (P = .001), antral tumor location (P = .04), and intestinal histotype (P = .003). Invasion of the serosa (P = .006) and lymph node metastasis (P < .001) were rarely observed. Among 22 patients having MSI/5 tumors with lymph node involvement, deposits were observed in few (median, 3; range, 1-8) metastatic lymph nodes (P = .001).
Lymphovascular MSI invasion was rare compared with that in MSS tumors (P < .001). Vascular invasion was observed in 33.3% of MSI/<5 tumors and in 14.8% of MSI/5 tumors (P = .24). Lymphatic invasion was equally frequent (33.3%) in MSI/<5 and MSI/5 tumors.
Figure 3 shows Kaplan-Meier curves for patients with MSS, MSI/<5, and MSI/5 tumors (P < .001). Survival rates for patients with M0 disease who underwent potentially curative (R0) resections are shown in Figure 4 (P = .004). Limiting the analysis to patients who underwent noncurative (R1 or R2) resections, the median (95% confidence interval) survival was 9.0 months (5.6-12.4 months) for 55 patients with MSS tumors, 5.2 months (2.1-8.3 months) for 7 patients with MSI/<5 tumors, and 58.4 months (0.0-132.5 months) for 5 patients with MSI/5 tumors (P < .009).
Table 2 summarizes results of the multivariate analyses using Cox proportional hazards regression models controlling for other risk factors. The MSI/5 phenotype was confirmed to be an independent predictor of long-term survival (relative risk vs patients with the MSS phenotype, 0.32; 95% confidence interval, 0.16-0.63; P = .002).
Microsatellite instability is defined as the presence of replication errors resulting in insertions or deletions of bases within nucleotide repeats, known as microsatellite regions. Although a consensus panel for determination of MSI phenotype has been proposed and accepted for colorectal cancer,5 several markers with differing results have been used for determination of MSI in GC.14- 19 Based on evaluation using several markers, the MSI phenotype generally characterizes 15% to 25% of patients with GC tumors, more frequently in older women.3,18 Most MSI tumors are of intestinal histotype located in the distal part of the stomach, with limited lymph node involvement. Our data regarding MSI phenotype are consistent with other findings reported to date.3,18
Several observations of unexpectedly good prognoses in patients with advanced tumors and instability at all 5 markers led us to look at this subset of patients. The case of a 31-year-old woman who had undergone palliative subtotal gastrectomy with macroscopic residual (R2) tumor 63 months previously for a nonintestinal pT4N2 tumor is still alive. The favorable survival in patients with the MSI/5 phenotype was also evident among patients who underwent palliative resection (median survival, 58.4 months; 95% confidence interval, 0.0-132.5 months).
Older age, female sex, antral tumor location, intestinal histotype, less invasion of the serosa, and involvement of fewer lymph nodes significantly correlated with MSI/5 phenotype. The number of involved lymph nodes was particularly low in patients having tumors with instability at all 5 markers. Among 22 patients with MSI/5 pN+ disease, 21 had fewer than 6 positive lymph nodes, while the 31-year-old patient already mentioned had 8 positive lymph nodes.
Significantly lower rates of lymphovascular invasion were consistently observed for MSI/5 and MSI/<5 tumors compared with MSS tumors. These data should correlate with less tumor aggression, lower disease stage, and older age in patients with MSI/5 tumors. Some of these characteristics have been reported in MSI tumors,13 but they seem to be particularly evident in MSI/5 tumors.
The MSI/5 phenotype delineates a subset of tumors with a similar growth pattern but with less propensity to invade serosal layers and spread systemically or via lymph nodes. Considering the locoregional growth of this class of tumors, one could argue for an aggressive surgical approach even in very advanced cases.
Survival analysis confirms high survival rates for patients with MSI/5 tumors (Figures 3 and 4). Survival of patients with the MSI/5 phenotype was significantly better than that of patients with the MSI/<5 phenotype, and this difference was confirmed in multivariate analysis. These novel findings identify a particular subset of MSI tumors with less biologic aggression and with a favorable prognosis. Other findings similarly correlate tumor characteristics and differentiation with tumor invasion and genetic changes.20
In conclusion, our data confirm previous evidence that tumors with MSI demonstrate distinct clinicopathologic features and better prognosis compared with patients having stable tumors. A subset of tumors was observed with instability at 5 quasimonomorphic mononucleotide repeats (BAT-26, BAT-25, NR-24, NR-21, and NR-27) . These tumors have a low predilection to spread systemically or to involve lymph nodes, and they are associated with favorable long-term prognosis even in patients with advanced disease. Further studies among more patients are necessary to confirm our results and to render them reproducible and useful in clinical practice.
Correspondence: Corrado Pedrazzani, MD, Sezione di Chirurgia Oncologica, Dipartimento di Patologia Umana e Oncologia, Policlinico “Le Scotte,” University of Siena, 53100 Siena, Italy (firstname.lastname@example.org).
Accepted for Publication: May 28, 2008.
Author Contributions: All authors take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Corso, Pedrazzani, Pinto, and Roviello. Acquisition of data: Marrelli. Analysis and interpretation of data: Corso, Pedrazzani, Marrelli, and Pascale. Drafting of the manuscript: Corso, Pedrazzani, Marrelli, and Pascale. Critical revision of the manuscript for important intellectual content: Pedrazzani, Pinto, and Roviello. Statistical analysis: Pedrazzani and Marrelli. Obtained funding: Pinto and Roviello. Administrative, technical, and material support: Corso, Marrelli, and Pascale. Study supervision: Roviello.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant 2005064071_004PRIN (Progetti di Ricerca di Interesse Nazionale) 2005, “Fattori genetici ed eredofamiliari nel carcinoma gastrico,” from Ministero Universitá e Ricerca.
Additional Contributions: Lorenzo Garosi provided valuable help in data collection, material storage, and technical support.