This source https://www.hq.nasa.gov/alsj/alsj-average_g.html states that:
From time to time, Houston could update the onboard SV using tracking and other available date; [sic] but, generally, it was the AGC's job to update the SV at about two second intervals
Although highly unlikely, let's imagine that communication with MCC disappeared entirely.
During powered flight, the Average-G Routine was used in solving the equations of motion using (1)integrations of three-component dynamic acceleration provided by IMU Pulsed Integrating Pendulous Accelerometers (PIPAs) and (2) three-component values of gravitational acceleration.
During coasting flight, with the engines off, the AGC used the Coasting Integration Routine (Section 5.2 in the GSOP) to update the SV. Although Earth was the dominant gravitational influence in earth orbit and, likewise the Moon in lunar orbit, the effects of both had to be included to update the SV with sufficient accuracy. The spacecraft orbit was similar to a Keplerian ellipse, but not quite identical. The Coasting Integration Routine made use of a venerable, 19th-century perturbation technique called Encke's Method.
Now, the Apollo Guidance Navigation and Control, R-567, section 5, p.5.2-11 (http://www.ibiblio.org/apollo/NARA-SW/R-567-sec5-rev8-5.1-5.2.pdf) has an interesting remark regarding the Coasting Integration Routine:
The LGC Coasting Integration Routine is restricted to earth or lunar orbit and is not to be used in cislunar-midcource space
Question: How was the state vector to be updated in this unfortunate case (no communications with earth and coasting to/from the moon)? Or am I missing something obvious here (i.e. it must've been planned, perhaps, to keep updating it somehow between MCC uploads anyway)?
Bonus question: As for any numerical integration, AGC would inevitably accumulate errors; was there a star sighting procedure, similar to P52 (IMU align), but for obtaining precise state vector from the stars (or any other procedure), without MCC uploading it?