How to make a Label Encoder trained on a training dataset transform an unseen value of a test dataset?

During the data preprocessing stage, I decided to apply the Label Encoding on one of the columns because it contained data points in string format. Suppose the column contains the following distinct values : -

pandas_dataframe["types"].values = ["a", "b", "c", "d"]


So I fit label encoder on types column values and the label encoder assigned it values like ("a", 0), ("b", 1), ("c", 2), ("d", 3)

All the above is done with respect to the training dataset.

But while preprocessing the test dataset, when I use the previously trained label encoder I observe that a new value (suppose e) is introduced in the types column on which the label encoder was never trained. In this situation how should I handle this new value e?