Pathological Digital Biomarkers: Validation and Application
Can Koyuncu
Round 1
Reviewer 1 Report
The communication paper of Song et al discusses the generalities of different validation steps that are required for complete clinical adoption of pathological digital biomarkers (PDB). The topic is interesting and the authors provide some important points (as regulation, clinical application, and insurance issues), that limits clinical adoption of PDB.
Just please take a look in some misspelled words along the manuscript (some examples: topioc (topic); homone (hormone); etc). Also check the abbreviation use rule (explain the first time they appears in the text).
May be a good definition of what digital biomarker means and a complete list of examples will benefit for a better understanding for non-specialist readers.
Author Response
Just please take a look in some misspelled words along the manuscript (some examples: topioc (topic); homone (hormone); etc). Also check the abbreviation use rule (explain the first time they appears in the text).
- Thank you for the important comment. We corrected typos and unnecessary abbreviation usage.
May be a good definition of what digital biomarker means and a complete list of examples will benefit for a better understanding for non-specialist readers.
: Thank you for the important comment. We described the list of pathological biomarkers as diagnostic, monitoring, pharmacodynamic, predicitive, prognostic, and safety & risk biomarkers and their definitions and examples. Additionally, we added current clinical trials using digital pathology, modetailed contents of validation.
Reviewer 2 Report
The authors reviewed the pathological imaging biomarkers. This is an interesting and important topic. I have the following suggestions:
1) I would expect categorization of three type of biomarkers: diagnostic, prognostic, and predictive. The definition of prognostic and predictive is important and most of the time people confuse them. This paper is quite relevant: https://ascopubs.org/doi/10.1200/JCO.2015.63.3651
2) In the section of "Evaluation for Clinical utility" there are many studies relevant to radiology imaging included but the context of this paper is pathology imaging. I would suggest replace them with pathological imaging studies.
3) I would discuss 3d pathology as well, which is quite relevant and future of digital pathology, such as 10.1158/0008-5472.CAN-21-2843
4) Fda also publish several documents explaining how medical image algorithms should be developed. Some discussion can be borrowed from there.
5) The "Sample collection and processing" section can be elaborated. For example a summary of what type of machine learning algorithms have been utilized in digital pathology. There are very few references included in this section. For example, they are very relevant and can be included either in this section or in the "clinical validation" section https://doi.org/10.1172/JCI145488 and https://doi.org/10.1093/jnci/djab215
Author Response
1) I would expect categorization of three type of biomarkers: diagnostic, prognostic, and predictive. The definition of prognostic and predictive is important and most of the time people confuse them. This paper is quite relevant: https://ascopubs.org/doi/10.1200/JCO.2015.63.3651
- Thank you for the important comment. We added list of pathological biomarkers as diagnostic, monitoring, pharmacodynamic, predictive, prognostic and safety & risk biomarkers and their definitions and examples.
2) In the section of "Evaluation for Clinical utility" there are many studies relevant to radiology imaging included but the context of this paper is pathology imaging. I would suggest replace them with pathological imaging studies.
- Thank you for the important comment. We corrected inappropriate examples and added more detailed examples in the section of various application of pathological biomarkers.
3) I would discuss 3d pathology as well, which is quite relevant and future of digital pathology, such as 10.1158/0008-5472.CAN-21-2843
- Thank you for the precious comment. We added 3D pathology in discussion
4) Fda also publish several documents explaining how medical image algorithms should be developed. Some discussion can be borrowed from there.
- Thank you for the important comment. We added relevant content of CFR from FDA as "The FDA's approval will be required for digital biomarkers to demonstrate their clinical utility and be utilized in clinics. A medical image management and processing system is defined by the FDA as a device that provides one or more capabilities related to the review and digital processing of medical images for the purpose of interpretation by a trained practitioner for disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in medical image interpretation and analysis. Image segmentation, multimodal image registration, and 3D visualization are examples of advanced image manipulation functions. Complex quantitative functions may include semi-automated measurements or time-series measurements"
5) The "Sample collection and processing" section can be elaborated. For example a summary of what type of machine learning algorithms have been utilized in digital pathology. There are very few references included in this section. For example, they are very relevant and can be included either in this section or in the "clinical validation" section https://doi.org/10.1172/JCI145488 and https://doi.org/10.1093/jnci/djab215
Thank you for the important comment, we added relevant references including suggested reference.
Round 2
Reviewer 2 Report
Authors addressed my concerns in the updated version. I don't have any other comments. Thank you for this comprehensive review.
Author Response
Thank you for the precious comments.
We revised typos, abbreviations, and capitals throughly.
