1. |
Risks to patients |
No stroma or tumor cells in biopsy |
Discomfort and risks associated with a sampling intervention |
Take multiple biopsies from one lesion at the same time (a minimum of three biopsies per lesion9) and check amount of tumor cells before analysis of sTILs and inclusion in the trial |
2. |
Risks to patients |
Loss of data confidentiality |
Patient samples sent to multiple institutions and reviewers |
Pseudonymization should be applied, hide slide labels, implement strict access control, ensure no metadata is linked to a slide |
3. |
Risks to biomarker development |
Inter-laboratory variability and interobserver variability |
Different methodologies used to score slides, interrater variability |
Use of international guidelines for scoring and training, use consensus score of four expert pathologists from three institutes |
4. |
Operational risks |
Failure of sample collection, processing, and quality |
Missing or poor-quality samples resulting in poor consensus scoring |
Standardized tissue processing, workflow management |
5. |
Operational risks |
Inadequate image quality or no ability to access image |
Missing, poor, or inaccurate scoring |
Track the scores of all pathologists and notify when scores are inconsistent |
6. |
Operational risks/risks to patients |
Long turnaround time |
Delay in patient randomization and treatment |
Timeline tracking incorporated in workflow |
7. |
Operational risks |
Data management failure |
Errors in collecting manual scores, typos, data conversion issues |
Structured digital scores from pathologist to the analyst |