The pitfalls of ultrasubspecialization

If you haven’t yet seen the new PBS documentary on Cancer, do it as soon as possible. A free stream is available on the PBS website but it is well worth the $15 on iTunes.

It makes many good points, one of which is the siliness of viewing cancer in general, or of any particular organ, as being a single entity. Genitourinary and GYN malignancies are sill fresh in my mind after this last rotation, so an example that comes first is prostate cancer. Most have your standard testosterone-dependent, androgen deprivation therapy-sensitive cells. Once they stop responding to hormonal therapy, treatment is still targeted towards the (now mutated) androgen receptor. Small cell prostate cancer, however, looks and behaves differently—tending to be bulkier, more aggressive, and having earlier visceral organ metastases. Ultimately, we treat it more like its namesake in the lung, with cisplatin and etoposide.

That was an easy distinction to make, since small cell prostate cancer looks nothing like adenocarcinoma under a microscope. Not so for breast cancer. We now know that it is at least four diseases which are at first glance all the same: luminal A (hormone receptor-positive, Her2-negaitve); luminal B (HR-positive, Her2-positive); HR-negative, Her2-positive; and triple-negative (also called basal-like, though definitions of basal-like breast cancer vary). The first three, which we are now able to distinguish with immunohistochemistry and FISH, have different behaviour, treatment, and prognosis. The fourth is a catch-all category that probably contains many different diseases we don’t know about yet. Some of those triple-negatives may have more in common with colon or lung cancer than they do with other malignancies of the breast.

Which organ the cancer is in should be important to a surgeon or a radiation oncologist, who have to deal with the anatomy. But should medical oncologists subspecialize by organ, or by cell? Why is a neuro-oncologist better suited to treat primary CNS lymphoma than a hematologist whose main interest are aggresive lymphomas? Does a GI oncologist have a better skillset and knowledge base for dealing with neuroendocrine tumors of the pancreas than an oncologist who deals with endocrine gland malignancies? Are there other, not so obvious connections between different cancers that we are missing because of ultrasubspecialization?

I don’t know enough oncology to answer any of these questions, but they are interesting questions to make.

more ...


Male breast cancer, a slide show

As the year winds down, these tumor board presentations will get less frequent. For now, though, it is still once a month. My latest, on breast cancer in men, seemed to be well-received. I suspect it was because, unlike most rare cancers, this one was easy to fit into a preexisting pattern: it is just like female breast cancer, except for… And voilà—you get quick and easily understood knowledge about a whole new disease entity.

more ...

Apple’s App Store rules, Dosegate edition

First they came for the nerds.

Now they are coming for the doctors (see What’s New in Version 3.0.5). The makers of MedCalc, the best medical calculator app out there, explained what happend in detail1. This was the rule they were supposedly infringing:

22.9 Apps that calculate medicinal dosages must be submitted by the manufacturer of those medications or recognized institutions such as hospitals, insurance companies, and universities.

Nevermind that many doctors view themselves as institutions—this is an idiotic rule. Is University of Baltimore, which has no biomedical science courses or programs, allowed to publish a drug dose calculator? Is GEICO?

The FDA has issued guidance for mobile medical apps. It specificaly allows calculators that use generally available formulas, and forbids apps which calculate radiation dosage, but does not mention drugs. Where, then, did this rule come from?

It is, of course, the same App store rules that allowed these pearls of quackery.

It’s madness, and it’s maddening.


  1. Seeing that URL made me appreciate the developers even more. 

more ...

Opasnosti vrebaju

Pacijent u ordinaciji ima pedesetak godina, izgleda mlađe. Oćelaveo od terapije, mada bih po konstituciji, odeći, i stavu—dva metra i mišićav, tri krsta i brojanica, vojnički i za dva decibila preglasan—pretpostavio da nikada nije imao bujne lokne.

more ...



The more you know…

If the unstated goal of these rotation post-mortems was to summarize what I had learned, a single post may not be enough for breast cancer. Six weeks ago, I knew that it was common, maybe overdiagnosed, possibly overtreated, and beating all other cancers for research funding by a vast margin. All this was a vague sense of being informed—like a NYT reader may feel after reading the Sunday Magazine feature—rather than actual knowledge.

Having talked to a good number of women with breast cancer, and worked with a few attendings dedicated to the field, I know it enough to know that I need to know more; but also enough to keep me interested. What from the outside looks like cookbook this-marker-means-she’s-getting-that-treatment medicine is in fact an intricate work of knowing your patient, figuring out where she stands in the heaps of data generated by decades-long studies following thousands of women on different protocols, discussing the options, and coming to a mutualy agreed decision1. Hard work, all of it.

Harder still is working on those data-generating trials. Anyone can think of a clinicaly relevant question, but can they make it into a feasable protocol? Can they gather a team to manage all the patients in the center, and all the different centers? Can they manage that team? Looking at a recent set of trials you will hear more about soon, the scale boggles the mind.

Side note: “We don’t have a crystal ball” is common oncspeak for “I don’t know what your prognosis is”2, but if a person has breast cancer what are the Gail model or Oncotype DX if not (developing, imperfect) tellers of fortune? And wouldn’t it be great to have a similar set of tools and statistics for all cancers?

So, not going into the field, but thoroughly impressed.


  1. Which is—truth be told—what we should do for any cancer, or illness. Alas, most diseases lack data, options, or both. 

  2. Or in the paternalistic dialect of the language, “I don’t want to tell you that you prognosis is poor”.  

more ...

A tale of two interleukins

Another month, another slide show. This one is about two things I had no experience with prior to fellowship—interleukin 15 and chondrosarcoma.

more ...

In the ivory tower

The rotation is called malignant hematology, but between Thanksgiving, Christmas, New Year’s, and everyone being away for ASH, it was a joke. The only memorable part were the first two weeks—the oncology consults at Georgetown—which reminded me that the NIH was indeed an ivory tower. This is compared to a well-known academic center. Compared to a community hospital—my last employer being a good example, though you may find much better—the Institutes in general and the Clinical Center in particular are a tower of crystalized angel tears sitting high on top of a mountain range on Mars. This is neither criticism nor praise, but a statement of fact2.

Having only spent four days on Georgetown consult service, I would rather not comment1. But here are ten things about the inpatient side of my home institution that might interest the outside world:

  • There is only one EMR and no paper charts;
  • Someone else, most commonly the research nurse, will obtain the outside records and have them scanned in—so the EMR is the only place you need to look for anything;
  • There are usually no medical students, interns, or residents—it’s the NP/PA, fellow, and the attending running the show;
  • Remember the 3am page from the nurse asking for a Tylenol order? If it is for someone on a phase 1 trial, you’d better check the protocol and call the attending before giving it;
  • Patients can be “out on pass”—meaning they can leave the campus for up to 20-something days and still count as an inpatient, without needing to be discharged and readmitted;
  • If a patient needs to stay an extra day or two because of transport issues, nobody blinks an eye;
  • For three to to five days you will get weird looks if it is your patient, but there will be no passive-aggresive emails about the hospital not getting reimbursement;
  • If it is six days or more it is likely they are homeless—social work will be on it;
  • Ordering most imaging—ultrasounds, CTs, MRIs, etc—requires two steps: getting a time-slot from central scheduling, or the tech on call if it is a same-day scan, and putting in the order. Think about the implications and let it sink in;
  • All discharge medications are dispensed from the outpatient pharmacy free of charge to the patient. Yes, all of them;

Most people treated at the Clinical Center are trial participants who are sick, rather than “just” patients. There are no administrative or financial pressures, no dealings with insurance companies, and not much concern with disposition. As you can imagine, this makes for a beautiful work environment.


  1. This makes it sound worse than it was. I don’t want to comment because n=4, not because I’m holding back the criticism. 

  2. Angel tears are a pain to clean up when they melt. 

more ...