If the company happens to be a biotech/pharma then they can only have data in humans from clinical trials.
With a technology, chemistry etc. company, things are easier since tests can be run to collect data and measurements to verify an invention actually works as claimed at least to an extend before a patent is filed.
To gain a patent the invention must not have been publicly disclosed prior to the patent filing, because then it would no longer be novel.
This presents a conundrum since a drug/treatment intervention must be described in detail to several parties before a clinical trial can be approved, initiated and eventually provide data that may support the invention.
In general a drug without prior data in one or more indications can only rely on prophetic examples that appear reasonable and unambiguously stated. The patent examiner, or anyone else for that matter, will not be able to question let alone refute whether a treatment works. If that was possible with a level of certainty that rivals controlled clinical trials, well then controlled clinical trials would not be required.
Science is hard, very useful and often fascinating and engaging when done well! Patents must be secured while the understanding ongoing science and empirical tests can yield is still in progress and before the details are publicly disclosed.