With the release of iText 7.2.2 and pdfOffice 2.0.2, we improved the conversion of docx files where a table extends to a new page. Previously, such files would result in converted documents with a table which on subsequent pages did not contain all of the cells in the table header. This has now been resolved.
Below is a Java example demonstrating such a conversion:
You can find the input .docx file and the output PDF file below. If you download both files and compare them, you will find that the table header cells on the second page are identical.