This paper presents a new line based 6-DOF monocular algorithm that uses the iSAM2, a point-based Graph SLAM approach. We extend iSAM2 to minimize the reprojection error of the line features to solve the line-based SLAM problem. A specific line representation is exploited that combines the Plucker Coordinates and the Cayley representation. The Plucker Coordinates are used for the 3D line projection function, and the Cayley representation helps to update the lines parameters during the non-linear optimization process. An undelayed initialization method with inverse depth parameters is also adopted. Both simulation and real experiments are carried out showing that the approach achieves high accuracy and consistency, and outperforms EKF-based SLAM method.