One perennial topic among coding consultants (and, I suspect, from time to time, Medicare fraud attorneys) is whether only the CPT code text is binding on providers, or whether the AMA definitions, preambles, and remarks are also binding.
In the past, I've definitely seen CMS communications that only the "codes" themselves are binding on CMS and MACs, allowing CMS and MACs to easily have different interpretations than AMA regarding the actual service. That is, CMS wanted the minimum amount of influence of AMA writings on CMS interpretations, and that minimum was defined by the code text itself. (For example, CMS's NCCI manual instructions may be confusing in regard to, or even contradictory to, AMA definitions).
Here's an example where the ball bounces in favor of AMA definitions. CMS released annual physician fee schedule rulemaking on August 3, published in Federal Register August 17.
From page 50117 to 50120, CMS discusses its interpretation of correct use of remote physiological monitoring codes (e.g. 99453, 99454, and others).
What I noted is that CMS conspicuously, and repeatedly, relies on the exact word choices AMA makes in its various remarks and definitions regarding these codes. So in this case, everything AMA remarked about these codes seems to be viewed by CMS as influential and pertinent.
Remarkably, as of August 2020, a CPT code search at CMS yields no articles or LCDs defining acceptable or unacceptable uses of 99454 (remote 30 day monitoring, paying $62-plus per month depending on geography).
I've provided two cloud documents. One is my version of the 50 mb, 600-page rule that is editable and highlightable and reduced to 25 mb, here. The other is a 5-page extract specific to the remote physiological monitoring codes, here.