Searching for a Punctuated Equilibrium in Pathology
I am a pathology resident at a well-known, highly regarded program. I have neither the experience nor expertise to consider myself an expert diagnostician, though I feel this does not preclude me from commenting on what I believe to be a systemic problem in my field. It may also be useful to mention a second disclaimer before I begin- the ideas put forth here are in no way meant to diminish or belittle what occurs every day in Pathology departments around the country. Pathology, and more specifically surgical pathology, is charged with accurately diagnosing disease, a charge on which almost all clinical care relies upon. The pathologists that I know exercise this duty with the utmost diligence, critical thought, and painstaking accuracy as humanly possible.
Currently, the practice of surgical pathology rests upon pattern recognition. Disease, luckily for us, forms a fairly “consistent” pattern, both of itself and surrounding stroma, that when artifactually fixed in formalin allows us to classify it into a subcategory that is clinically meaningful. Disease can thus be diagnosed, prognosis can be interpolated, and therapeutic decisions made on a cross linked piece of tissue, that is placed in a bed of melted wax, put onto a slide, a subjective “looks-like” score is mentally taken by the practicing pathologist, and barring any further analysis the diagnosis is rendered. Does this sound like the best we can do? Isn't this the same "cutting-edge" technolgy used in the 19th and 20th century? Where is the innovation?
The flow through of a general pathology specimen in an academic medical center, which is actually a mystery to most clinicians I have found, occurs on some iteration of this: a specimen is received (day 0), “grossed in” (day 0, hopefully), slides are processed and made (day 1), the slides are reviewed by an attending pathologist (day 2) and a diagnosis is rendered, or stains/ancillary tests are ordered and the diagnostic process continues (days 4, 5, ...etc). The National Cancer Insititue (NCI), when describing what a pathology report is on its website states that “The pathologist sends a pathology report to the doctor within 10 days after the biopsy or surgery is performed.” 10 days. Really? Patel et. al at Emory, did perhaps the most thorough analysis on turnaround time (TAT),
A total of 713 specimens were analyzed, 551 (77%) were verified within 2 days and 162 (23%) in 3 days or more. Lung, gastrointestinal, breast, and genitourinary specimens showed the highest percentage of cases being signed out in over 3 days. Diagnosis of malignancy (including staging of the neoplasia), consultation with other pathologists, having had a frozen section, and use of immunohistochemical stains were significantly associated with increased turnaround time in univariate analysis. Decalcification was not associated with increased turnaround time. In multivariate analysis, consultation with other pathologists, use of immunohistochemistry, diagnosis of malignancy, and the number of slides studied continued to be significantly associated with prolonged turnaround time. Our findings suggest that diagnosis of malignancy is central to significantly prolonging the turnaround time for surgical pathology specimens, thus institutions that serve cancer centers will have longer turnaround time than those that do not.
In an additional report, Cromier et. al. report a turnaround time for difficult cases sent out for diagnostic consultation of 32.8 days.
The issue is really that we are given fresh tissue on which we could make a rapid molecular diagnosis, only to then destroy it to give a less precise, more time-consuming morphologic diagnosis. The change has to come as a revolutionary paradigm shift- if it has X Y Z mutation what are the prognostic vs diagnostic implications vs what is the morphological diagnosis. Our diagnostic lines currently read something like this: Clear Cell of the Kidney, pT2 pN0. Although this places the patient into a pre-defined though non-precise sub-group, on which the therapy he or she recieves will be based, It is time we redefine these generalities with "precision diagnostics". We are starting to be able to give discrete prognostic information, based on objective as opposed to subjective variables, coupled with actionable molecular information. Aren't the molecular abnormalities the drivers of the disease anyways?
This will take a sizeable effort in its least, because the issues are countless. I will not touch on those here, but will instead introduce a couple of opportunities/ideas of what the "future" might hold. Briefly though, all arguments of current limitations can be solved with data. The power is in the data, and the quicker and more voluminous amounts we produce, we can use techniques taught to us from other indusustries to distill it.
While current molecular diagnostics have focused only on “back-end” resolutions, it is time that we face the problem with a “front-end” solution. Consumer genomics firms (ie Foundation Medicine and Genomic Health) and public databases such as The Cancer Genome Atlasare seeking to impact precision medicine by annotating and exploiting genomics for the sole purpose of therapeutic intervention. I argue that there is just as much of a role for genomic exploitation in diagnostics, not just in the form of a post-hoc SNaPshot tumor analysis. Instead, the information should be the mechanism by which we diagnose, providing a relevant “bin” for classification, prognostication, and therapeutic intervention in one comprehensive, rapid test.
I foresee a future in which a specimen is recieved, the margins are evaluated intraoperatively via mass spectrometry, and the main specimen (biopsy or resection) is then "aspirated" in multiple areas (to account for tumor heterogeneity) and a "real-time" (~3 hour) molecular analysis of the tissue is performed. The output of this workflow would be tumor type, margin status, prognostic -omic signatures, and actionable therapuetic information. Rapid, objective, precise data for a world where we practice presicion medicine.
Patel, S , Smith, J, Kurbatova, E, Gaurner, J. Factors that impact turnaround time of surgical pathology specimens in an academic institution Human Pathology, Volume 43, Issue 9, September 2012, Pages 1501–1505
Cormier, B., Bonneau, C., Kerdraon, R., Heitzmann, A., Garnaud, S., Michenet, P. Request of second opinion for difficult diagnosis in surgical pathology. Assessment of a one year activity in a general hospital. .Annales de pathologie, Volume 27, Issue 5, 2007, pg:345
L. V. Sequist, R. S. Heist, A. T. Shaw, P. Fidias, R. Rosovsky, J. S. Temel, I. T. Lennes, S. Digumarthy, B. A. Waltman, E. Bast, S. Tammireddy, L. Morrissey, A. Muzikansky, S. B. Goldberg, J. Gainor, C. L. Channick, J. C. Wain, H. Gaissert, D. M. Donahue, A. Muniappan, C. Wright, H. Willers, D. J. Mathisen, N. C. Choi, J. Baselga, T. J. Lynch, L. W. Ellisen, M. Mino-Kenudson, M. Lanuti, D. R. Borger, A. J. Iafrate, J. A. Engelman, and D. Dias-Santagata .Implementing multiplexed genotyping of non-small-cell lung cancers into routine clinical practiceAnn Oncol (2011) 22 (12): 2616-2624