A Nomogram Model for Predicting the Postoperative Recurrence of Localized Laryngeal Amyloidosis

Abstract

Objective:

To analyze the factors related to postoperative recurrence in patients with localized laryngeal amyloidosis (LocLA) and to construct a nomogram prediction model (NPM).

Methods:

We collected the data for LocLA patients diagnosed from March 2000 to May 2019 and clinical characteristics data were extracted. Factors related to recurrence were analyzed using multivariate logistic regression. The NPM was constructed for predicting the recurrence risk of LocLA. The receiver operating characteristic (ROC) curve evaluated the distinguishing ability using the area under curve (AUC). The calibration curve was created to evaluate the consistency of the NPM.

Results:

A total of 226 confirmed LocLA cases were included. One hundred seventy-five cases (77.4%) had localized single nodule, and 51 cases had more than one lesions. Sixty-three (27.9%) cases had no multinucleated giant cell (MGC) around amyloid, and 163 (72.1%) cases had MGC around amyloid. Multivariate logistic regression analysis showed that more than one lesions (odds ratio [OR] = 3.206 and 95% confidence interval [CI]: 1.492-6.888; P value: .003), subglottic involvement (OR = 2.926 and 95% CI: 1.300-6.585; P = .010), and no multinucleated giant cell (MGC) around amyloid (OR = 2.503 and 95% CI: 1.173-5.342; P = .018) had a statistically significant effect on postoperative LocLA recurrence (P < .05). The AUC of the ROC curve was 0.753 (95% CI: 0.667-0.832). The bias-corrected curve approached the ideal curve, with an average absolute error of 0.037.

Conclusions:

More than one lesions, subglottic involvement, and no MGC around amyloid are risk factors for postoperative recurrence of LocLA. The NPM constructed has good applicability.