The paper by Knigge et al., uses ARM as an approach to datamnine biomarker and clinical data with the aim of identifying patients who will respond or progress. This approach has identified some "themes" that predict progression e.g., female gender and elevated CgA, but this is not invariant. This approach has the potential to provide a more personalized approach for evaluating individual patients. Overall, the paper is well-written and presented, but there are several weaknesses in the current presentaiton.

Major issues.
1. Biomarker plots: Although the authors include a plot for CPE, no other plots for the other informative biomarkers, e.g., CgA etc are included. A critical unmet need is appropriate plots for all markers that were found to be informative. Including the data from the controls would be most useful.
2. It would be very useful information to include AUROCs for each of the markers at the levels the authors identify as relevant, e.g., WISP-1 >3.73 etc. Some simple metrics would enhance the value of these markers and provide a better understanding of their potential utility.
3. Were the authors able to evaluate other factors, e.g., TGR, in their model? It would be appropriate in the discussion to include whether other imaging or biomarkers may have utility.
4. It would also be appropriate to define the strengths and outline the weaknesses of the study in the discussion.

Minor comments:
None.

The authors have appropriately undertaken all requests.