Effect of timing of metastasis/disease recurrence and histologic differentiation on survival of patients with advanced gastric cancer

Machine learning models for predicting one-year survival in patients with metastatic gastric cancer who experienced upfront radical gastrectomy

Tumor metastasis is a common event in patients with gastric cancer (GC) who previously underwent curative gastrectomy. It is meaningful to employ high-volume clinical data for predicting the survival of metastatic GC patients. We aim to establish an improved machine learning (ML) classifier for predicting if a patient with metastatic GC would die within 12 months. Eligible patients were enrolled from a Chinese GC cohort, and the complete detailed information from medical records was extracted to generate a high-dimensional dataset. Appropriate feature engineering and feature filter were conducted before modeling with eight algorithms. A 10-fold cross validation (CV) nested in a holdout CV (8:2) was employed for hyperparameter tuning and model evaluation. Model selection was based on the area under the receiver operating characteristic (AUROC) curve, recall, and precision. The selected model was globally explained using interpretable surrogate models. Of the total 399 cases (median survival of 8.2 months), 242 patients survived less than 12 months. The linear discriminant analysis (LDA), support vector machine (SVM), and random forest (RF) model had the highest AUROC (0.78 ± 0.021), recall (0.93 ± 0.031), and precision (0.80 ± 0.026), respectively. The LDA model created a new function that generally separated the two classes. The predicted probability of the SVM model was interpreted using a linear regression model visualized by a nomogram. The predicted class of the RF model was explained using a decision tree model. In summary, analyzing high-volume medical data by ML is helpful to produce an improved model for predicting the survival in patients with metastatic GC. The algorithm should be carefully selected in different practical scenarios.

Preoperatively Treated Diffuse-Type Gastric Adenocarcinoma: Glucose vs. Other Energy Sources Substantially Influence Prognosis and Therapy Response

Simple Summary

The diffuse type of gastric adenocarcinoma (dGAC) generally confers a poor prognosis compared to intestinal type. Some dGACs are not avid on fluorine-18 fluoro-2-deoxy-D-glucose PET (FDG-PET) while others seem to consume glucose avidly. We analyzed the outcomes based on the avidity of the primary on baseline FDG-PET. Our data suggest that if dGACs used glucose as an energy source then the prognosis was very poor while non-glucose sources improved prognosis. Multi-platform (including metabolomics) profiling of dGACs would yield useful biologic understanding.

Abstract

Diffuse type of gastric adenocarcinoma (dGAC) generally confers a poor prognosis compared to intestinal type. Some dGACs are not avid on fluorine-18 fluoro-2-deoxy-D-glucose PET (FDG-PET) while others seem to consume glucose avidly. We analyzed the outcomes based on the avidity (high with standardized uptake value (SUV) > 3.5 or low with SUV ≤ 3.5) of the primary on baseline FDG-PET. We retrospectively selected 111 localized dGAC patients who had baseline FDG-PET (all were treated with preoperative chemotherapy and chemoradiation). FDG-PET avidity was compared with overall survival (OS) and response to therapy. The mean age was 59.4 years and with many females (47.7%). The high-SUV group (58 (52.3%) patients) and the low-SUV group (53 (47.7%) patients) were equally divided. While the median OS for all patients was 49.5 months (95% CI: 38.5–98.8 months), it was 98.0 months (95% CI: 49.5–NE months) for the low-SUV group and 36.0 months for the high-SUV (p = 0.003). While the median DFS for all patients was 38.2 months (95% CI: 27.7–97.6 months), it was 98.0 (95% CI: 36.9–NE months) months for the low-SUV group was and only 27.0 months (95% CI: 15.2–63.2 months) for the high-SUV group (p = 0.005). Clinical responses before surgery were more common in the low-SUV group but overall we observed only 4 pathologic complete responses in 111 patients. Our unique data suggest that if dGACs used glucose as an energy source then the prognosis was very poor while non-glucose sources improved prognosis. Multi-platform (including metabolomics) profiling of dGACs would yield useful biologic understanding.

Nomograms predicting survival of patients with unresectable or metastatic gastric cancer who receive combination cytotoxic chemotherapy as first-line treatment

BACKGROUND

Some clinicopathological variables are known to influence the survival of patients with advanced gastric cancer. A comprehensive model based on these factors is needed for prediction of an individual's survival and appropriate patient counseling.

METHODS

A nomogram for predicting 1-year survival in patients with advanced gastric cancer in the palliative chemotherapy setting was developed using clinicopathological data from 949 patients with unresectable or metastatic gastric cancer who had received first-line doublet cytotoxic chemotherapy from 2001 to 2006 at the National Cancer Center, Korea (Baseline Nomogram). For 836 patients whose initial response to chemotherapy is known, another nomogram (ChemoResponse-based Nomogram) was constructed using the response to chemotherapy as additional variable. Nomogram performance in terms of discrimination and calibration ability was evaluated using the C statistic and Hosmer-Lemeshow-type χ ² statistics.

RESULTS

Two different nomograms were developed and subjected to internal validation. The baseline nomogram incorporated 13 baseline clinicopathological variables, whereas the chemoresponse-based nomogram was composed of 11 variables including initial response to chemotherapy. Internal validation revealed good performance of the two nomograms in discrimination: C statistics = 0.656 (95% confidence interval, 0.628-0.673) for the baseline and 0.718 (95% confidence interval, 0.694-0.741) for the chemoresponse-based nomogram, which showed significantly better discrimination performance than the baseline nomogram (Z statistics = 3.74, p < 0.01).

CONCLUSION

This study suggests that individual 1-year survival probability of patients receiving first-line doublet cytotoxic chemotherapy for advanced gastric cancer can be reliably predicted by a nomogram-based method incorporating clinicopathological variables and initial response to chemotherapy.

Clinical significance of computed tomography-detected ascites in gastric cancer patients with peritoneal metastases

Patients with peritoneal metastases (PM) are generally considered incurable; therefore, the presence of PM is a critical factor in deciding between palliative surgery and curative resection as a therapeutic strategy. Previous studies have not determined the predictive value of ascites detected on computed tomography (CT) for the presence of PM. We aimed to analyze the factors that are associated with PM in patients with CT-detected ascites.A total of 2207 consecutive patients who were diagnosed with gastric cancer between 2004 and 2013 were identified. Eleven patients with liver cirrhosis or chronic renal insufficiency with ascites and 57 patients who received previous treatment were excluded. Ninety-eight patients who had definite evidence of distant metastasis or PM on CT and 64 patients who did not undergo surgery were excluded. A total of 91 patients were enrolled in the study to analyze the association between CT-detected ascites and surgically confirmed PM.Seventy-six patients underwent curative resection and 15 patients underwent palliative surgery. Twelve patients exhibited peritoneal seeding and 37 patients showed regional lymph node metastasis. Regional lymph node metastasis, advanced gastric cancer, undifferentiated pathology, and the amount of ascites were significantly associated with PM. Multivariable logistic regression analysis identified the amount of ascites to be an independent risk factor for the presence of PM.Regional lymph node metastasis, advanced gastric cancer, undifferentiated pathology, and the amount of ascites were associated with PM. The amount of ascites was found to be an independent risk factor for PM.

Analysis of a cytoskeleton-associated kinase PEAK1 and E-cadherin in gastric cancer

The expression of pseudopodium-enriched atypical kinase 1(PEAK1) has been studied in human cancers. However, their roles in gastric cancer are still unknown. In this study, gastric cancer tissue microarrays were constructed with 159 gastric cancer tissue samples, 150 non-neoplastic gastric epithelium specimens and 152 lymph node samples. Immunohistochemical staining for PEAK1 and E-cadherin was performed. Our study found negative expression of PEAK1 in 113 of 159 (71.1%) gastric cancers, in 46 of 150 (30.7%) non-neoplastic gastric epithelium tissues and in 69 of 94 (73.4%) metastatic lymph nodes. Negative expression of PEAK1 and E-cadherin associated with tumor grading, depth of invasion, lymph node metastases, pTNM stage and macroscopic type. Patients with either positive PEAK1 or E-cadherin expression had a significantly higher survival than those with negative expression. When combined, PEAK1(-)/E-cadherin(-) had a significantly poor prognosis than the rest of the patients. The expression of PEAK1 protein was positively correlated with E-cadherin in cancer tissues. Cox regression analyses showed that PEAK1, E-cadherin and PEAK1(-)/E-cadherin(-) were independent predictors of overall survival. In conclusion, our findings suggest that loss of PEAK1 may play an important role in carcinogenesis and development of gastric cancer through activating epithelial to mesenchymal transition.

Comprehension of readmission after laparoscopy assisted distal gastrectomy: what are the causes?

Purpose

The aim of this study is to evaluate long-term outcomes regarding readmission for laparoscopy-assisted distal subtotal gastrectomy (LADG) compared to conventional open distal subtotal gastrectomy (CODG) for early gastric cancer (EGC).

Methods

Between January 2003 and December 2006, 223 and 106 patients underwent LADG and CODG, respectively, for EGC by one surgeon. The clinicopathologic characteristics, postoperative outcomes, postoperative complications, overall 5-year survival, recurrence, and readmission were retrospectively compared between the two groups.

Results

Multiple readmission rate in LADG was significantly less than that in CODG (0.4% vs. 3.8%, P = 0.039), although the readmission rate, reoperation rate after discharge, and mean readmission days were not significantly different between the two groups. Readmission rates of the LADG and CODG groups were 12.6% and 14.2%, respectively. First flatus time and postoperative hospital stay was significantly shorter in the LADG group. However, there was no significant difference in the complication rates between the two groups. Overall 5-year survival rates of the LADG and CODG group were 100% and 99.1% (P = 0.038), respectively.

Conclusion

Compared to the CODG group, the LADG group has several advantages in surgical short-term outcome and some benefit in terms of readmission in surgical long-term outcome for patients with EGC, even though the oncologic outcome of LADG is similar to that of CODG over 5 years.

Is laparoscopy-assisted total gastrectomy feasible for the treatment of gastric cancer? A case-matched study

BACKGROUND

The aim of this study was to clarify the short- and long-term outcomes of laparoscopy-assisted total gastrectomy (LATG) for gastric cancer compared to conventional open total gastrectomy (COTG) with a case-control study.

METHODS

Between January 2002 and December 2010, a series of 264 patients with gastric cancer underwent R0 total gastrectomy (61 LATG patients and 203 COTG patients). Age, gender, and pathological stage were matched by propensity scoring, and 120 patients (60 LATG and 60 COTG) were selected for analysis.

RESULTS

There were no significant differences in the clinicopathological features between the two groups. Regarding postoperative outcomes, first flatus time was significantly shorter in the LATG group than in the COTG group (p < 0.001), while operation time was significantly longer in the LATG group than in the COTG group (p < 0.001). Postoperative complications occurred in 5 cases (8.3%) of the LATG group and in 11 cases (18.3%) of the COTG group (p = 0.178). There were two recurrences in the LATG group and three recurrences in the COTG group. There was no significant difference in the 5-year survival rate between the two groups (p = 0.667).

CONCLUSIONS

LATG for gastric cancer may be both feasible and efficient compared to COTG for experienced laparoscopic surgeons. This study is valuable for a prospective randomized controlled trial of LATG for gastric cancer in a larger number of patients.

Efficient induction of specific cytotoxic T lymphocytes against gastric adenocarcinoma by a survivin peptide

Survivin has been demonstrated to be an excellent target for immunotherapy in several types of cancer, but little is known of the efficacy of survivin with gastric adenocarcinoma. In this study, a simple method was performed, and relatively high efficacy was shown upon inducing survivin-derived peptide-specific cytotoxic T lymphocytes (CTL) from peripheral blood mononuclear cells of healthy donors. The induced CTLs exhibited specific lysisagainstHLA-A2 matched tumor cells in vitro, and similar results were demonstrated in primary cell cultures isolated from patients with gastric adenocarcinoma. Up to 30% of randomly selected patients could potentially benefit from immunotherapy targeting survivin. These results suggested that this survivin epitope peptide could be a promising vaccine candidate for immunotherapy for patients with gastric adenocarcinoma.

Clinical parameters model for predicting pathologic complete response following preoperative chemoradiation in patients with esophageal cancer

BACKGROUND

Approximately 25% of patients with esophageal cancer (EC) who undergo preoperative chemoradiation, achieve a pathologic complete response (pathCR). We hypothesized that a model based on clinical parameters could predict pathCR with a high (≥60%) probability.

PATIENTS AND METHODS

We analyzed 322 patients with EC who underwent preoperative chemoradiation. All the patients had baseline and postchemoradiation positron emission tomography (PET) and pre- and postchemoradiation endoscopic biopsy. Logistic regression models were used for analysis, and cross-validation via the bootstrap method was carried out to test the model.

RESULTS

The 70 (21.7%) patients who achieved a pathCR lived longer (median overall survival [OS], 79.76 months) than the 252 patients who did not achieve a pathCR (median OS, 39.73 months; OS, P = 0.004; disease-free survival, P = 0.003). In a logistic regression analysis, the following parameters contributed to the prediction model: postchemoradiation PET, postchemoradiation biopsy, sex, histologic tumor grade, and baseline (EUS)T stage. The area under the receiver-operating characteristic curve was 0.72 (95% confidence interval [CI] 0.662-0.787); after the bootstrap validation with 200 repetitions, the bias-corrected AU-ROC was 0.70 (95% CI 0.643-0.728).

CONCLUSION

Our data suggest that the logistic regression model can predict pathCR with a high probability. This clinical model could complement others (biomarkers) to predict pathCR.

A 5 year analysis of readmissions after radical subtotal gastrectomy for early gastric cancer

BACKGROUND

Many patients and surgeons have become interested in the quality of life after surgery for early gastric cancer. However, no reports on readmission rates for gastrectomy patients have been published, even if readmission greatly affects the patient's quality of life.

METHODS

In 530 consecutive early gastric cancer patients who underwent radical subtotal gastrectomy, we determined the incidence, cause, timing, type of treatment, and risk factors for readmission after discharge, and compared the readmission and nonreadmission groups with respect to clinicopathologic features and postoperative outcomes.

RESULTS

Overall, the 5 year and 1 month readmission rates after radical subtotal gastrectomy for early gastric cancer were 13.0% (69 of 530) and 7.5% (40 of 530), respectively. The most common cause for readmission was delayed gastric emptying (17 cases). Among a total of 82 readmissions, 34 patients (41.5%) were readmitted within 2 weeks of surgery. The type of treatment for 82 readmissions included 55 conservative therapies, 15 radiologic or endoscopic interventions, and 12 repeat laparotomies. No significant differences were detected in the clinicopathologic feature and postoperative outcomes between the two groups. The initial hospital stay remained significantly associated with readmission based on multivariate analysis.

CONCLUSIONS

Readmission rate at 1 month after radical subtotal gastrectomy is lower than that after major bowel surgery. Unusual postoperative recovery in a patient with vague symptoms should be managed with greater care before discharge. After subtotal gastrectomy for early gastric cancer, prevention of readmission can improve the patient's quality of life.

The validation of matrix metalloproteinase-9 mRNA gene expression as a predictor of outcome in patients with metastatic gastric cancer

BACKGROUND

The prognostic role of matrix metalloproteinase-9 (MMP-9) in metastatic gastric cancer has not been validated.

PATIENTS AND METHODS

We carried out a molecular analysis in 222 metastatic gastric cancer patients obtained from clinical trials. We assessed the messenger RNA (mRNA) expression of MMP-9, vascular endothelial growth factor receptor-A, and epidermal growth factor receptor in a training cohort of 130 patients and conducted an independent validation in 92 patients. Automated RNA extraction from paraffin and RT-quantitative PCR was used. Immunohistochemistry for MMP-9 and diverse immune cell infiltrates was conducted.

RESULTS

In the training cohort, only MMP-9 significantly correlated with patient's survival. At the cut-off with the highest predictive value, 19% of patients had MMP-9 expression above this cut-off and these showed a median survival of 3.6 months compared with 10.5 months (P=1.7e(-6)) in patients with lower expression. Corresponding 1- and 2-year survivals were 9% and 44% and 0 and 21%, respectively. The application of this cut-off to the validation cohort revealed similar distributions of overall survival according to MMP-9 expression on uni- (P<0.001) and multivariate analyses (P<0.001). No differences in survival according to MMP-9 below best cut-off were found. MMP-9 protein assessed by immunohistochemistry was not prognostic.

CONCLUSION

MMP-9 mRNA expression above a certain cut-off level is associated with dismal survival.

A meta-analysis of prospective randomized trials comparing minimally invasive and open distal gastrectomy for cancer

Current literature suggests that minimally invasive distal gastrectomy (MIDG) may enhance post-operative recovery and decrease morbidity compared to open surgery (ODG) in patients with gastric cancer. A meta-analysis of six Prospective Randomized Trials comparing MIDG (343 patients) and ODG (323 patients) for gastric cancer was conducted. MIDG was associated with increased operative time, reduced blood loss and overall morbidity. There was not sufficient data to draw solid conclusions about the oncologic quality of MIDG.

A meta-analysis of randomized controlled trials that compared laparoscopy-assisted and open distal gastrectomy for early gastric cancer

BACKGROUND

We conducted a meta-analysis to evaluate and compare the advantages of laparoscopy-assisted distal gastrectomy (LADG) over open distal gastrectomy (ODG) for treating early gastric cancer (EGC).

METHODS

We searched MEDLINE, EMBASE, Science Citation Index, and Cochrane Controlled Trial Register for relevant papers published between January 1990 and January 2010 by using the following search terms: laparoscopy-assisted gastrectomy, laparoscopic gastrectomy, and early gastric cancer. The following data were analyzed: operative time, estimated blood loss, number of harvested lymph nodes, time required for resumption of oral intake, duration of hospital stay, frequency of analgesic administration, complications, tumor recurrence, and mortality.

RESULTS

We selected four papers reporting randomized control studies (RCTs) that compared LADG with ODG for EGC. Our meta-analysis included 267 patients with EGC; of these, 134 and 133 had undergone LADG and ODG, respectively. The volume of intraoperative blood loss, frequency of analgesic administration, and rate of complications were significantly lesser for LADG than for ODG. However, the time required for resumption of oral intake and duration of hospital stay did not significantly differ between LADG and ODG. The operative time for LADG was significantly longer than that for ODG; further, the number of harvested lymph nodes was significantly lesser in the LADG group than in the ODG group.

CONCLUSION

LADG is advantageous over ODG because it results in lesser blood loss, is less painful, and is associated with a low risk of complications. Additional RCTs that compare LADG and ODG and investigate the long-term oncological outcomes of LADG are required to determine the advantages of LADG over ODG.

Meta-analysis of laparoscopy-assisted and open distal gastrectomy for gastric cancer

BACKGROUND

Laparoscopy-assisted distal gastrectomy (LADG) for gastric cancer is a minimally invasive technique. We performed a meta-analysis of five randomized clinical trials (RCTs) to evaluate and compare the benefits of LADG with those of open distal gastrectomy (ODG).

METHODS

The present meta-analysis was based on the comparison of LADG with ODG for gastric cancer. The following factors were examined: operative time, estimated blood loss, number of harvested lymph nodes, time to resumption of oral intake, duration of hospital stay, frequency of analgesic administration, complications, tumor recurrence, and mortality.

RESULTS

We selected five RCTs to compare LADG with ODG for gastric cancer. A total of 326 patients with gastric cancer were included in this meta-analysis of whom 164 underwent LADG and 162 underwent ODG. There was a significant difference in the volume of intraoperative blood loss, period of hospital stay, frequency of analgesic administration, and rate of complications between LADG and ODG. There was no difference in the resumption of oral intake, rate of tumor recurrence, and mortality. The operative time was significantly longer and the number of harvested lymph nodes was significantly smaller in LADG than in ODG.

CONCLUSION

LADG is significantly superior to ODG regarding the volume of blood loss, duration of hospital stay, level of pain, and risk of complications. There was no difference in the resumption of oral intake, rate of tumor recurrence, and mortality. However, LADG was significantly inferior to ODG regarding operative time and also had a smaller number of harvested lymph nodes.

Clinical characteristics and treatment outcomes of gastric cancer patients with isolated para-aortic lymph node involvement

BACKGROUND

Although gastric cancer with isolated para-aortic lymph node (PAN) involvement is considered an advanced disease, the clinical characteristics of it have not been comprehensively elucidated.

PATIENTS AND METHODS

We reviewed the medical records of 1,277 patients received palliative chemotherapy with advanced gastric cancer according to metastatic sites: PAN-only metastasis, single organ metastasis other than PAN, and multiple organ metastasis. Time to other organ metastasis (TTOM) was determined only in PAN-only metastasis group as the time interval between initial diagnosis of recurrence or de novo metastasis and confirming distant metastasis beyond PAN area.

RESULTS

The median overall survival (OS) of patients with PAN-only metastasis was significantly longer than that of patients with single organ metastasis other than PAN or multiple organ metastasis (13.8 months vs. 11.4 months vs. 8.4 months; P < 0.001). In the PAN-only metastasis group, patients with recurrent diseases showed longer TTOM beyond the PAN area (10.7 vs. 7.7 months; P = 0.037) and OS (23.8 vs. 12.8 months; P = 0.010) than those with de novo metastatic disease and it was validated by multivariate analysis.

CONCLUSION

Patients with isolated PAN metastasis showed an excellent prognosis compared with patients with metastasis at other sites and it was primarily evident in patients with recurrent PAN metastasis.