Dear Colleagues, I looked at your interesting preprint (10.64898/2026.09.16.751033) about adding Ab multiplexing to a Xenium-processed FFPE and I found quite some confusion. You named “4i” the multiplexing protocol you use (iterative indirect immunofluorescence imaging): this is a specific and precise reference to a totally different multiplexing protocol, and the definition is extracted verbatim from a paper published in 2018 (Gut, G., et al. Science 361, eaar7042 (2018) doi:10.1126/science.aar7042). Furthermore, there is no reference to that method in the References list and in the Methods section.

You use and reference instead a method we published in 2014 (your reference 16) and perfectioned in 2017: Bolognesi, M. M. et al. Multiplex Staining by Sequential Immunostaining and Antibody Removal on Routine Tissue Sections. J. Histochem. Cytochem. 65, 431–444 (2017). doi:10.1369/0022155417719419. We named that method MILAN: Multiple Iterative Labeling by Antibody Neodeposition. We also published a protocol: Giorgio Cattoretti, Francesca Maria Bosisio, Lukas Marcelis, Maddalena Maria Bolognesi 2019. Multiple Iterative Labeling by Antibody Neodeposition (MILAN) . Protocol exchange https://dx.doi.org/10.21203/rs.2.1646/v5 (https://www.protocols.io/view/multiple-iterative-labeling-by-antibody-neodeposit-dm6gp9dnjvzp/v5 )

While we thank you for quoting and using our methods, albeit an early version, we ask you to remove any mention of the “4i” protocol - in your manuscript, abstract included - in your GitHub URL - in any other public record where this confusion may arise. You are welcome to quote the 2017 JoHC paper and the term “MILAN”. Thank you

We acknowledge the excellent, long-standing contribution to the field of multiplex immunostaining of these authors. There is however a premise in the introduction of this preprint, with which we do not agree at all: “..However, many cyCIF techniques remain limited by the harsh conditions required for signal removal (e.g., fluorophore photobleaching) which compromise sample integrity and complicate image co-registration as cycle number increases. …” We published in 2017 a paper (https://journals.sagepub.com/doi/10.1369/0022155417719419) and a method, MILAN (https://dx.doi.org/10.21203/rs.2.1646/v5) where an excess of 30 (thirty) cycles of staining and stripping FFPE sections in indirect IF results in negligible cell loss and staining intensity variations of less than 15%. See supplementary Fig. 1 in https://doi.org/10.1038/s41598-020-57627-1. The JoHC paper has been quoted 127 times and viewed 26,512 times by many but you. We built from this method and published a cell and landscape classification of the human lymph node (https://doi.org/10.1371/journal.pone.0346693) using ~80 antibodies. We suggest to revise incorrect statements which do not reflect previously established better targets in multiplexing. Thank you.

Dear Colleague, It is curious to notice that for this very interesting, complex study on tonsils you use the term CyCIF (see refs 35 and 36) but the cyclic staining and stripping method is by no mean corresponding to those references: you use the MILAN method (https://www.protocols.io/view/multiple-iterative-labeling-by-antibody-neodeposit-dm6gp9dnjvzp/v5) after Bolognesi MM et al (https://doi.org/10.1369/0022155417719419) without acknowledging the method source. Please correct according to good, reproducible and responsible scientific referencing. Thank you

It is curious to notice that for this very interesting, complex study you use the term CycIF (see refs 19 and 20) but the cyclic staining and stripping method is by no mean corresponding to those references: you use the MILAN method (https://www.protocols.io/view/multiple-iterative-labeling-by-antibody-neodeposit-dm6gp9dnjvzp/v5) after Bolognesi MM et al (https://doi.org/10.1369/0022155417719419) without acknowledging the method source. Please correct accordingly.

Congratulations for the comparative investigation on LLM tools which will progressively change pathology diagnoses. What makes me skeptic however is having a pathologist diagnosis as a gold standard. For lymphatic invasion, Nottingham score and cancer subtype (ductal, lobular) a comparison with agreement scores among pathologists may be a better gauge of LLM performance. I do disagree for all IHC-based parameters; not only we have shown how unreliable is the human eye to evaluate immunohistology results (Biological Imaging , Volume 4 , 2024 , e10; DOI: https://doi.org/10.1017/S2633903X24000138), but, as reported by Johnston L. et al (npj Precis. Oncol. 8, 203 (2024); doi: 10.1038/s41698-024-00696-6) 12% of the breast cancers with the lowest amount of Her2 (≤ 3,000 amol/µg) and 34% of the ones with the next lowest amount (3,000-6,000 amol/µg) are scored Her2 3+ with IHC by pathologists, the gold reference standard most people use. Machines may be better than pathologists at evaluating reality. PS I have been a surgical pathologist most of my working life.