Suk-Hwan Lee

Accuracy of Endorectal Ultrasonography in Preoperative Staging of Rectal Tumors

PURPOSE: Preoperative staging of rectal tumors is considered essential to tailor treatment for individual patients. The aim of the present study was to evaluate the accuracy of endorectal ultrasonography in preoperative staging of rectal tumors. METHODS: Eleven hundred eighty-four patients with rectal adenocarcinoma or villous adenoma underwent endorectal ultrasonography evaluation at a single institution during a ten-year period. We compared the endorectal ultrasonography staging with the pathology findings based on the surgical specimens in 545 patients who had surgery (307 by transanal excision, 238 by radical proctectomy) without adjuvant preoperative chemoradiation. Comparisons between groups were performed using chi-squared tests and logistic regression analysis. RESULTS: Overall accuracy in assessing the level of rectal wall invasion was 69 percent, with 18 percent of the tumors overstaged and 13 percent understaged. Accuracy depended on the tumor stage and on the ultrasonographer. Overall accuracy in assessing nodal involvement in the 238 patients treated with radical surgery was 64 percent, with 25 percent overstaged and 11 percent understaged. CONCLUSION: The accuracy of endorectal ultrasonography in assessing the depth of tumor invasion, particularly for early cancers, is lower than previously reported. The technique is more precise in distinguishing between benign tumors and invasive cancers and between tumors localized to the rectal wall and tumors with transmural invasion. Differences in image interpretation may in part explain discrepancies in accuracy between studies.

A Pathologic Complete Response to Preoperative Chemoradiation Is Associated With Lower Local Recurrence and Improved Survival in Rectal Cancer Patients Treated by Mesorectal Excision

AbstractPURPOSE: Preoperative chemoradiation reduces tumor size and nodal metastasis in patients with rectal cancer. Tumor downstaging has been associated with an increased probability of a sphincter-saving procedure and with improved local control. However, pathologic complete response to chemoradiation has not been correlated with local control and patient survival. We studied the prognostic value of pathologic complete response to preoperative chemoradiation in rectal cancer patients. METHODS: We have prospectively followed up 168 consecutive patients with ultrasound Stages II (46) and III (122) rectal cancer treated by preoperative chemoradiation followed by radical resection with mesorectal excision; 161 had a curative resection. Recurrence and survival were compared with tumor characteristics and pathologic complete response. Average follow-up was 37 months. RESULTS: Tumor downstaging occurred in 97 (58 percent) patients, including 21 (13 percent) patients who had a pathologic complete response. None of the clinical or pathologic variables was associated with pathologic complete response. The estimated 5-year rate of local recurrence was 5 percent; of distant metastasis, 14 percent. None of the patients with pathologic complete response has developed disease recurrence. We found no difference in survival among patients with pathologic Stages I, II, or III tumors. CONCLUSIONS: A pathologic complete response to preoperative chemoradiation is associated with improved local control and patient survival. For patients without pathologic complete response, the pathology stage does not have prognostic significance.

Treatment of locally recurrent rectal cancer.

PURPOSE: This study was designed to analyze the outcome for patients with isolated local recurrence after radical treatment of rectal cancer and to identify predictors of curative resection. METHODS: The medical records of 87 patients who developed isolated local recurrence after curative radical surgery for primary rectal cancer were retrospectively reviewed. Survival rates from the time of recurrence were calculated using the Kaplan-Meier method. Tumor stage and histology, patient characteristics, and treatment variables were analyzed using logistic regression to identify predictors of curative surgery. RESULTS: Symptomatic treatment alone or chemotherapy and/or radiation therapy was provided to 23 patients (26 percent), and surgical exploration was performed in 64 patients. In 22 patients (25 percent), the tumor was considered unresectable at surgery (n=13) or was resected for palliation with gross or microscopic positive margins (n=9). In 42 patients (48 percent), curative-intent resection was performed. The only independent predictors of resectability were younger age at diagnosis, earlier stage of the primary tumor, and initial treatment by sphincter-saving procedure. There was no difference in survival between patients who had no surgery and those who had palliative surgery. The estimated five-year survival rate for patients who had curative-intent resection was better than for those who had no surgery or palliative surgery (35vs. 7 percent;P=0.01). Of the 42 patients who underwent curative-intent resection, 14 (33 percent) developed a second recurrence at a mean of 15±11 months after reoperation. Twenty-five percent of patients developed major complications. CONCLUSIONS: Salvage surgery for locally recurrent rectal cancer may be helpful in a selected group of patients. The stage and treatment of the primary tumor may help to identify patients with the best chance for curative-intent resection.

Clinicopathologic characteristics and outcomes of gastric cancers with the MSI-H phenotype.

We examined the correlation between microsatellite instability (MSI) status and the clinicopathological features and prognostic value in gastric cancer and compared the efficacy of immunohistochemical staining for hMLH1 and hMSH2 with a polymerase chain reaction (PCR)‐based test.

Vector watermarking scheme for GIS vector map management

A geographical information services (GIS) can be provided on the basis of a digital map, which is the fundamental form of representation of data in a GIS. Because the process of producing a digital map is considerably complex and the maintenance of a digital map requires substantial monetary and human resources, a digital map is very valuable and requires copyright protection. A digital map consists of a number of layers that are categorized in terms of topographical features and landmarks. Therefore, any unauthorized person can forge either an entire digital map or the feature layers of the map. In this paper, we present a robust and invisible watermarking scheme based on polylines and polygons for the copyright protection of a GIS digital map. The proposed scheme clusters all polylines and polygons in the feature layers of the map on the basis of the polyline length and the polygon area. And then a watermark is embedded in GIS vector data on the basis of the distribution of polyline length and polygon area in each group by moving all vertices in polylines and polygons within a specified tolerance. Experimental results confirm that the proposed scheme is more robust against geometric attacks, such as rotation, scaling, and translation (RST) transformations, data addition, cropping, breaking, and filleting attacks, and layer attacks with rearrangement and cropping, when compared with conventional schemes. Moreover, the scheme also satisfies data position accuracy.

Cytoplasmic expression of HuR is related to cyclooxygenase-2 expression in colon cancer.

PURPOSE HuR, human family embryonic-lethal abnormal vision-like protein, can bind to mRNA and stabilizes the nucleic acid in the cytoplasm, resulting in more efficient translation. HuR is predominantly present in the nucleus and shuttles between the nucleus and cytoplasm. HuR stabilizes cyclooxygenase-2 (Cox-2) mRNA in several cancers, including breast, stomach, lung and brain cancer. MATERIALS AND METHODS We investigated the expression and cellular location of HuR, as well as evaluated Cox-2 expression in 79 colorectal cancer patients with the use of immunohistochemical methods. The biological implications of HuR localization and Cox-2 expression in colorectal carcinoma were evaluated. RESULTS Nuclear HuR expression was observed in 59 (74.7%) tumors and cytoplasmic HuR expression was seen in 25 (31.6%) tumors. Cox-2 immunoreactivity was noted in 42 (53%) tumors. The expression of cytoplasmic HuR was significantly associated with Cox-2 expression (p=0.004). Cytoplasmic expression of HuR showed a correlation with lymphatic invasion (p=0.025) and the presence of a lymph node metastasis (p=0.027). The presence of nuclear HuR showed no correlation with Cox-2 expression or any other of the clinicopathological parameters that were examined. CONCLUSION These results suggest that cytoplasmic translocation of HuR is associated with Cox-2 expression for some colorectal carcinomas.

Mechanisms of Microsatellite Instability in Colorectal Cancer Patients in Different Age Groups

PURPOSEThe proportion of colorectal cancers located proximal to the splenic flexure increases with age. Colorectal cancers of the microsatellite instability phenotype are preferentially located in the proximal colon. We investigated the location of colorectal cancer with this phenotype in different age groups to determine whether different molecular mechanisms could account for the changes in distribution of colorectal cancers.METHODSA representative sample of 230 colorectal cancers from three age groups (<45 years, 60–70 years, >87 years) was selected from a subset of The Upper Midwest Oncology Medical Registries database. Microsatellite instability was determined by polymerase chain reaction using a panel of five microsatellite markers. The presence of new microsatellite alleles at two or more loci was scored as microsatellite instability. Tumors were otherwise considered microsatellite stable. MLH1 and MSH2 expression was determined by immunohistochemistry. Methylation of the MLH1 gene promotor was determined by methylation-specific polymerase chain reaction assay.RESULTSThe proportion of tumors of the microsatellite instability phenotype was 21 percent in the young group, 15 percent in the middle group, and 33 percent in the old group. More tumors of the microsatellite instability phenotype were proximal compared with microsatellite-stable tumors in all three age groups, but the differences were significant only for the old group. Tumors of the microsatellite instability phenotype in the older group were associated with MLH1 inactivation (24/29 or 83 percent), MLH1 promoter methylation (18/29 or 62 percent), and proximal location (25/29 or 86 percent), while tumors in the young group were associated with MSH2 inactivation (8/18 or 44 percent) and distal location (11/18 or 62 percent).CONCLUSIONThe age-related proximal shift of colorectal cancers is associated with the microsatellite instability phenotype, MLH1 inactivation, and MLH1 promoter hypermethylation.

The mechanism of microsatellite instability is different in synchronous and metachronous colorectal cancer.

MLH1 promoter hypermethylation has been described as the primary mechanism for high-frequency microsatellite instability (MSI-H) in sporadic colorectal cancers (CRCs). The underlying molecular mechanism for microsatellite instability (MSI) in synchronous and metachronous CRCs is not well described. A total of 33 metachronous CRC patients and 77 synchronous CRC patients were identified from 2884 consecutive patients undergoing cancer surgery in an academic center. Evaluable tumors were tested for MSI, immunohistochemistry for MLH1 and MSH2 protein expression, and hypermethylation of the MLH1 promoter. MSI-H tumors were found in 12 (36%) metachronous CRC patients and 29 (38%) synchronous CRC patients. MSI-H metachronous CRC patients were younger at index cancer diagnosis (64 vs. 76 years, P = 0.01) and more often were diagnosed before 50 years of age (4 of 12 vs. 0 of 29, P = 0.005). Loss of MLH1 expression associated with promoter hypermethylation was common in all patients, although more common in MSI-H synchronous patients (50% metachronous vs. 83% synchronous, P = 0.03). Overall, MLH1 promoter hypermethylation was seen in 7 of 17 (41%) metachronous and 44 of 54 (81%) synchronous MSI-H CRCs tested (P = 0.004). Although MSI occurred with equal frequency among patients with synchronous and metachronous CRCs, the underlying mechanism for MSI was different. Observed differences in MLH1 promoter hypermethylation and patient characteristics suggest most MSI-H synchronous CRCs in our population were sporadic in origin. In contrast, more MSI-H metachronous CRCs were associated with patient and tumor characteristics suggestive of underlying hereditary nonpolyposis CRC.

Mesh watermarking based projection onto two convex sets

This paper proposes 3D-mesh watermarking based on projection onto convex sets (POCS). As such, a 3D-mesh model is iteratively projected onto two constraint convex sets until the convergence condition is satisfied. The sets consist of a robustness set and invisibility set designed to embed the watermark. The watermark can be extracted without the original mesh model using the decision values and index that the watermark was embedded with. Experimental results verify that the watermarked mesh model has both robustness against mesh simplification, cropping, scaling, rotation, translation, and vertex randomization and invisibility.

CAD drawing watermarking scheme

A CAD (computer-aided design) drawing based on vector data is very important art work in industrial fields. It is considered to be content for which copyright protection is urgently required. This paper presents a watermarking scheme based on k-means++ for CAD drawings. A CAD drawing consists of several layers, and each layer consists of various geometric objects such as LINE, POLYLINE, CIRCLE, ARC, 3DFACE, and POLYGON. POLYLINE along with LINE, 3DFACE, and ARC objects are the most commonly used fundamental objects in a CAD drawing. The proposed scheme embeds the watermark into the geometric distribution of POLYLINE, 3DFACE, and ARC objects in the main layers. Hence, in the proposed scheme, we select the target object with a high distribution among POLYLINE, 3DFACE, and ARC objects in a CAD drawing and then select layers that include the maximum number of instances of the target object. Then, we cluster the target objects in the selected layers by using k-means++ and embed the watermark into the geometric distribution of each group. The geometric distribution is the normalized length distribution in a POLYLINE object, the normalized area distribution in a 3DFACE object, and the angle distribution in an ARC object. Experimental results have verified that the proposed scheme is robust against file format conversion, layer attacks, and various types of geometric editing carried out using the CAD editing tools.