Predictability of a Mandibular Corpus Length Multivariate Model Integrating Björk-Jarabak Measurements and the Cephalometric Norm

TL;DR

This study shows that a multivariate model using linear craniofacial measurements predicts mandibular corpus length better than traditional methods in adults.

Contribution

A new multivariate model integrating linear and angular measurements improves MCL prediction accuracy in adults.

Findings

Linear measurements like ACBL, PCBL, and RH significantly correlate with MCL.

The multivariate model outperformed individual models with an R² of 0.8689.

Angular measurements and the cephalometric norm had low predictive value.

Abstract

Introduction

The cephalometric norm of the mandibular corpus length (MCL) or the one-to-one ratio of the MCL to the anterior cranial base length (ACBL) are cephalometric indicators with unknown predictive capacity and clinical utility. Multivariate regression models enable the use of two or more variables to estimate an expected value, in this case, for MCL. This study compares three approaches to predicting MCL in adults by applying Björk-Jarabak measurements: (i) conventional angular norms, which have limited standalone value; (ii) simple linear‐proportion indices of craniofacial structures; and (iii) a multivariate model that integrates both linear and angular measurements.

Methods

A cross-sectional study was conducted using 100 adult cone beam computed tomography (CBCT) scans (63% female, mean age 29.5 ± 8.4 years) who met strict inclusion criteria. Seven simple linear regression models were analyzed for individual cephalometric variables: ACBL, posterior cranial base length (PCBL), ramus height (RH), saddle angle (SA), articular angle (AA) and gonial angle (Gon), and cephalometric norm adjusted for sex and age. Subsequently, a comprehensive multivariate model was developed. Regression coefficients (β), 95% confidence intervals (95% CI), and determination coefficients (R²) were reported for each model.

Results

Linear measurements revealed a statistically significant association with MCL: anterior cranial length (β = 0.86; 95% CI: 0.72-0.99; R² = 0.6462), posterior cranial length (β = 1.16; 95% CI: 0.92-1.40; R² = 0.5331), and RH (β = 0.84; 95% CI: 0.68-0.99; R² = 0.5757). In contrast, the cephalometric norm had low explanatory power (β = 4.72; 95% CI: -0.47--9.93; R² = 0.1091), and the angular measurements were not significant. The final multivariate model, including the three linear variables, showed a superior predictive capacity (R² = 0.8689), with the following coefficients: ACBL (β = 0.41), PCBL (β = 0.49), RH (β = 0.20), and Gon (β = -0.18; all p <0.01).

Conclusion

These findings suggest that linear craniofacial measurements have greater predictive capacity for the MCL than do norms or angles. The multivariate model increased the explanatory capacity by 22.27% relative to the best individual model. The integration of these variables allows more precise and personalized estimates in adults. The use of multivariate models in MCL clinical practice and their validation in other populations is recommended.

Figures