A new home for intercalated duct lesion?

Intercalated duct lesions are a forgotten member of the salivary gland tumour family. They are rarely targeted on their own, and are usually found in the shadows of other more consequential lesions.

I was taught that they canonically occur in the background of basal cell adenoma or epithelial-myoepithelial carcinoma. But recently, cases have been found to occur in association with unexpected tumor types.

The group where I am training recently published a case arising in association with mucoepidermoid carcinoma; mucoepidermoid carcinoma does not originate from intercalated ducts, so such a situation was not expected to occur [1].

We came across another unexpected intercalated duct lesion in an adenoid cystic carcinoma, the occurrence of which had not yet been reported. This was already a surprising case because the tumor was grossly well-circumscribed and the diagnosis of adenoid cystic carcinoma was not originally suspected.

Hiding in the middle of the tumor were a few small intercalated duct lesions, which looked completely different from the adenoid cystic carcinoma; but there was still a transition point or "blending" between the two. The question was then whether the two lesion types were related or coincidentally next to each other. Immunostaining looked the same in both the intercalated duct lesion and the adenoid cystic carcinoma, which supported the former hypothesis. We felt this was novel enough to warrant reporting in the scientific literature!

To really prove a link between the two, we attempted fluorescence in situ hybridization (FISH) to see if the same MYB rearrangement was present in both components. Unfortunately, this adenoid cystic carcinoma was part of the minority that have rearrangements in MYBL1 rather than MYB; we did not have FISH probes validated for MYBL1, which would have allowed us to demonstrate the same fusion in both the adenoid cystic carcinoma and the intercalated duct lesions.

MYBL1::NFIB fusion can also be detected by molecular genetic techniques, which we performed and which did indeed confirm the fusion was present. But this method doesn't allow to see the actual cells with gene rearrangements like FISH does, so it's not possible to know for sure if the translocation was present in both the intercalated duct lesion and the adenoid cystic carcinoma.

This is a limitation of our report that was rightfully highlighted during peer review, but we think that on balance, given the morphologic transition between the two tumors and the shared immunophenotype, the tumors are more likely related than not.

Read our "Image" report [2] in Head and Neck Pathology (gift link).

On another note, this is the first time I attempted to prepare a manuscript in Typst, which is a typesetting language similar in objective to LaTeX, but with lighter syntax for content and a less arcane programming style. This is great for someone like me who enjoys versioning (very difficult with DOCX/ODT files) and programmatic variable management (so that numbers drawn from data are always up-to-date in manuscript contents).

However, it will likely be a long time before the publishing cartels accept Typst documents as a method of submission. This is especially true as the language continues to evolve. Preprint servers like arxiv may come around sooner but there are still challenges.

Fortunately, most publishers do accept submissions in LaTeX, even though LaTeX is rarely used in health sciences clinical research! This meant I could use ttt, a fairly basic Typst to LaTeX converter that automatically interpolates Typst content into the journal's LaTeX template. With minimal configuration, I could iteratively and reproducibly rebuild the LaTeX "source" based on changes in the Typst manuscript, without having to wrestle with Pandoc, or copy-paste raw .tex into the journal template. I did have to add some minor functionality to satisfy the template requirements, but my changes were accepted upstream.

This was a fairly simple paper to try this out. I am working on another project with tables that require me to write raw LaTeX (for now), but I still think this is a tolerable compromise given all of the benefits of Typst.

References