This past March, as in years past, the United States and Canadian Academy of Pathology (USCAP) held its annual meeting in the beautiful city of Vancouver. I think this was my third USCAP in Vancouver and at least 20th  USCAP meeting in the past 25 years.

USCAP remains the largest gathering of surgical pathologists on the planet and the most academic/educational in terms of pure hours of CME. Pathologists from around the world gather for a week to learn and reunite with colleagues.

One of the mainstays of the meeting are the presentations by pathologists in training, residents and fellows, either in the form of oral presentations (proffered papers) and poster sessions. Training programs pride themselves on sending trainees to the meeting to showcase their research efforts and offer new ideas or concepts or data on clinical behavior, outcome or classification of diseases. Many of the faculty for the courses come from the largest programs who send the most residents and fellows to the meeting.

The well-known medical schools demonstrate their long-standing tradition of mentorship to their house staff and to attending pathologists alike, training legions of pathologists from all the world for decades.

Some of the faculty I recall from my first USCAP meeting are still active in the organization, serving on the executive board still teaching classes or offering interesting case reviews in the evening hours after a full day of posters, podium presentations and courses.

You can learn pathology from 7 AM to 10 PM while trying to attend companion meetings, alumni reunions and business gatherings. I get much of my 50 hours of CME at the meeting by the end of March by attending this meeting.

USCAP is the equivalent of RSNA for radiologists and ASCO for oncologists, to name a few, for the sheer volume of content that is available, on a much smaller scale in terms of vendor, given our numbers. We have microscopes and whole slide scanners and books. Those other meetings have hundreds of pharmaceutical companies and large CT and MRI machines displayed, respectively. Given our numbers, it is unlikely USCAP will ever sell out McCormick Place, one of the largest convention centers in the world, like these meetings do annually. However, for many academic pathologists and vendors it is the go-to meeting of the year. I never get tired of watching trainees defend their research conclusions and even sitting through 3 hours of case discussion until 10 PM.

However, there is something missing in academic pathology and what and how we are training the pathologists of tomorrow, in my opinion. For all the hard work and preparation of abstracts and presentations, I think there is a large gap between what residents are learning and what they will need to practice in the future.

It seems training programs are still firmly founded in our roots, morphology and classification of disease. Techniques of histology, morphologic comparisons and increasingly outcomes-based findings based on morphology combined with molecular pathology make up most of the courses and presentations, both poster and oral. Increasingly, morphometry and next generation sequencing to further classify human disease and understand its pathobiology and implications for clinical management are presented and the exhibit hall reflects this market.

But these are still largely based on slides, tried and true formalin-fixed, paraffin embedded (FFPE) sections placed on a 1×3 inch piece of glass to stain with dyes or micro dissection for further study and comparison. I bet the very term “FFPE” is referenced hundreds of times a day at the meeting.

But there was extremely little discussion on machine learning, deep learning or artificial intelligence (AI), collectively computational pathology. Outside of the informatics sessions I recall very few presentations and course discussions on the topic of AI, in my opinion, the single greatest opportunity for surgical pathology since the invention of the microscope.

Are the pathologists of tomorrow learning today what will be needed? Are training programs adding computational pathology to their curriculums? There are core competencies that have been established for pathology informatics (PIER) but are these keeping up with more than some basics on laboratory information systems or databases?

The sessions and posters on breast focus on ADH versus DCIS and the liver sessions are still consumed by morphology of hepatitis C and non-alcoholic steatohepatitis, to name a few examples. Papillary thyroid cancer still has nuclei on cytology that are considered characteristic for the disease. Urine cytology still has a broad “atypical” category. And fibromyxoid lesions can still be difficult to sort out for a novice, particularly if you know nothing of the location, clinical duration of symptoms and imaging studies, if available, to name a few more examples. But little discussion on using techniques other than sheer dyes and reviewing cytology or biopsy material with the resected specimen for correlation and quality assurance.

In my training, I learned very little of immunohistochemistry (IHC). My attendings didn’t receive it in their training and didn’t use it. Ditto for molecular techniques and imaging technologies. We focused on the morphology and discussion with the clinicians and surgeons, gross and microscopic findings and going down to radiology to review the films on a light box with a real radiologist. If it weren’t for the AFIP I would have learned little about IHC and when I did, discovered the IHC only confirmed what my attendings thought on H&E. With the introduction of ER, PR and HER2 in breast cancer diagnostics and initially a panel of CD15, CD30, CD20 and CD45 for Hodgkin’s disease as it was called at the time, one had to learn IHC by the time I finished my residency. I likely learned how to do so, at least in part, at USCAP meetings.

But now it seems we are stuck on morphology and IHC and some molecular correlations, that overtime, like IHC may prove to be less specific than we once thought. I had one attending who said CD99 stained 99 things. The last time I searched “CD99” in Immunoquery there were 99 tumors that have been reported as being “positive for CD99”. NSE was not used after it was discovered that it “Now Stains Everything”.

Perhaps we are taught what our teachers know and can discuss. They teach us what has worked for them and their understanding of the topic. And the more experienced folks don’t talk about what they don’t know. As the saying goes, you can’t teach an old dog new tricks.

If it weren’t for my younger attendings in residency, those closer to their own board examinations than not, IHC and molecular would have been a self-taught exercise entirely.

No doubt there is still a role to spend full days at meetings such as USCAP reviewing the histomorpology of follicular adenoma versus follicular carcinoma or variations of prostate cancer or the histogenesis of collagenous colitis and lymphocytic colitis. Review of grading of dysplasia in Barrett’s esophagus always seems timely.

But I also think that our training programs need to evolve, to use our foundation in morphology in conjunction with computational pathology. There is no doubt that what we can see on those 4 micron thick tissue sections stained with dyes sandwiched between a piece of glass and a another piece of glass with chemicals that allow the tissue to be magnified represent the tip of the iceberg in terms of the content we can see. We recognize the patterns but there are other patterns to be discovered beyond the 40x objective with tools that can process more information than we could have possibly imagined even a few years ago.

I hope that the pathologists of tomorrow are ready for tomorrow. I recall hearing about “immunotherapy” at ASCO many years ago. Now it is on billboards everywhere. RSNA had discussions on AI a decade ago. Maybe it was called “computer assisted diagnosis” but they had it.

We seem to still be having the same discussions about nuclear and cytoplasmic features that I recall hearing about in Vancover in 2004 and a decade before that.

It is my understanding that USCAP last year did not “allow” or “accept” an oral presentation on the results of the Philips clinical trial for primary diagnosis that was used for FDA clearance. A poster was presented. A significant study for many reasons, including but not limited to, establishing a “baseline” concordance rate among pathologists to establish a threshold by which to compare the digital diagnosis with the optical one. That now establishes a predicate by which other studies can be measured against for clinical validation. A large study on many tissue types comprising nearly 2000 cases and 16 pathologists. And because of the commercial support, USCAP had some “concerns” about an oral presentation. Our colleagues in ASCO and RSNA manage this very well.  Thousands of people attend these talks with appropriate disclosures and decide for themselves what may be applicable in their respective practices.

I hope that our training programs, our academic organizations and their companion societies can evolve and change our behaviors.

OR

Comments (2)

  1. Emilia Andersson MD, PhD

    So well captured!
    As pathologist we need to take the lead on digital pathology and AI and make sure that it really serves our patients the best. We need to work together with computer scientists to ensure that our diagnosis reflects the prognostic or predictive information that is needed for informed decisions on treatment modalities.

platinum partners

gold partners

Silver Partners

Media Partners