As mentioned in the other answers, there is no digital way to be absolutely sure that a document won't be copied, edited, or transferred in some way using digital means.
You could add a layer of protection like loading a remote image in your PDF with an identifier of that particular client in the URL, which would help you know when that PDF is loaded. Continued use from different IPs might signal the document was copied by this client and released improperly.
However, measures like this can be circumvented if the client knows what to look for. You will always be susceptible to someone copying the data in an offline way (taking a photo, printing and re-scanning), or removing metadata digitally (reprinting the PDF digitally to remove metadata).
In the end it is a game of setting up reasonable safeguards to catch casual offenders and setting up other policies to protect from determined leakers.
It is a bit tangential to your question, but if you are interested in a less digital solution you may want to consider implementing a Canary of some kind to help track down a leak after the fact. Since these rely on more human factors it may be less likely for them to be noticed. This is called a Canary trap (https://en.wikipedia.org/wiki/Canary_trap), and is when you give a different variation of the document to each signer of an NDA, so that when information is leaked you can use these variations to determine who leaked the data.
These should be small to prevent it from being noticed, but if recipients of different document versions are unlikely to compare the data, it gives you the possibility for recourse in case of a leak.