Limits...
A nomogram for predicting the likelihood of lymph node metastasis in early gastric signet ring cell carcinoma

View Article: PubMed Central - PubMed

ABSTRACT

Treatment algorithm has not been established for early gastric cancer with signet ring cell carcinoma (SRC), which has a reported low rate of lymph node metastasis (LNM) similar to differentiated cancer. A cohort of 256 patients with early gastric SRC at our center between January 2002 and December 2015 were retrospectively reviewed. Multivariate logistic regression analysis was used to determine the independent factors of LNM. A nomogram for predicting LNM was constructed and internally validated. Additional external validation was performed using the database from Cancer Institute Ariake Hospital in Tokyo (n = 1273). Clinical performance of the model was assessed by decision analysis of curve. The overall LNM incidence was 12.9% (33/256). The multivariate logistic model identified sex, tumor size, and LVI as covariates associated with LNM. Subsequently, a nomogram consisted of sex, tumor size, and depth of invasion was established. The model showed qualified discrimination ability both in internal validation (area under curve, 0.801; 95% confidence interval [CI], 0.729–0.873) and in external dataset (area under curve, 0.707; 95% CI, 0.657–0.758). Based on the nomogram, treatment algorithm for early gastric SRC was proposed to assist clinicians in making better decisions. We developed a nomogram predicting risk of LNM for early gastric SRC, which should be helpful for patient counseling and surgical decision-making.

No MeSH data available.


Validation of nomogram in training set and validation set. (A) Calibration plot of nomogram in training set. After 500 repetitions of bootstrap, the bias-corrected plot showed a good agreement between the predicted probability and actual probability (Mean absolute error = 0.021). (B) The AUC of nomogram in training set was 0.801 (95% CI, 0.729–0.873) after 500 repetitions of bootstrap (Delong). (C) Calibration plot of nomogram in validation set (mean absolute error = 0.007). (D) AUC of nomogram in validation set was 0.707 (95% CI, 0.657–0.758). AUC = area under curve.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5120931&req=5

Figure 2: Validation of nomogram in training set and validation set. (A) Calibration plot of nomogram in training set. After 500 repetitions of bootstrap, the bias-corrected plot showed a good agreement between the predicted probability and actual probability (Mean absolute error = 0.021). (B) The AUC of nomogram in training set was 0.801 (95% CI, 0.729–0.873) after 500 repetitions of bootstrap (Delong). (C) Calibration plot of nomogram in validation set (mean absolute error = 0.007). (D) AUC of nomogram in validation set was 0.707 (95% CI, 0.657–0.758). AUC = area under curve.

Mentions: Though obviously overestimated when probability was more than 30% in training set, bias-corrected calibration plot of the nomogram predicted LNM corresponding closely with the actual probability in both datasets. The mean absolute error in training set and validation set was 0.021 and 0.007, respectively. The AUC was 0.801 (95% CI, 0.729–0.873) in training set, and 0.707 (95% CI, 0.657–0.758) in validation set (Fig. 2).


A nomogram for predicting the likelihood of lymph node metastasis in early gastric signet ring cell carcinoma
Validation of nomogram in training set and validation set. (A) Calibration plot of nomogram in training set. After 500 repetitions of bootstrap, the bias-corrected plot showed a good agreement between the predicted probability and actual probability (Mean absolute error = 0.021). (B) The AUC of nomogram in training set was 0.801 (95% CI, 0.729–0.873) after 500 repetitions of bootstrap (Delong). (C) Calibration plot of nomogram in validation set (mean absolute error = 0.007). (D) AUC of nomogram in validation set was 0.707 (95% CI, 0.657–0.758). AUC = area under curve.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5120931&req=5

Figure 2: Validation of nomogram in training set and validation set. (A) Calibration plot of nomogram in training set. After 500 repetitions of bootstrap, the bias-corrected plot showed a good agreement between the predicted probability and actual probability (Mean absolute error = 0.021). (B) The AUC of nomogram in training set was 0.801 (95% CI, 0.729–0.873) after 500 repetitions of bootstrap (Delong). (C) Calibration plot of nomogram in validation set (mean absolute error = 0.007). (D) AUC of nomogram in validation set was 0.707 (95% CI, 0.657–0.758). AUC = area under curve.
Mentions: Though obviously overestimated when probability was more than 30% in training set, bias-corrected calibration plot of the nomogram predicted LNM corresponding closely with the actual probability in both datasets. The mean absolute error in training set and validation set was 0.021 and 0.007, respectively. The AUC was 0.801 (95% CI, 0.729–0.873) in training set, and 0.707 (95% CI, 0.657–0.758) in validation set (Fig. 2).

View Article: PubMed Central - PubMed

ABSTRACT

Treatment algorithm has not been established for early gastric cancer with signet ring cell carcinoma (SRC), which has a reported low rate of lymph node metastasis (LNM) similar to differentiated cancer. A cohort of 256 patients with early gastric SRC at our center between January 2002 and December 2015 were retrospectively reviewed. Multivariate logistic regression analysis was used to determine the independent factors of LNM. A nomogram for predicting LNM was constructed and internally validated. Additional external validation was performed using the database from Cancer Institute Ariake Hospital in Tokyo (n = 1273). Clinical performance of the model was assessed by decision analysis of curve. The overall LNM incidence was 12.9% (33/256). The multivariate logistic model identified sex, tumor size, and LVI as covariates associated with LNM. Subsequently, a nomogram consisted of sex, tumor size, and depth of invasion was established. The model showed qualified discrimination ability both in internal validation (area under curve, 0.801; 95% confidence interval [CI], 0.729–0.873) and in external dataset (area under curve, 0.707; 95% CI, 0.657–0.758). Based on the nomogram, treatment algorithm for early gastric SRC was proposed to assist clinicians in making better decisions. We developed a nomogram predicting risk of LNM for early gastric SRC, which should be helpful for patient counseling and surgical decision-making.

No MeSH data available.