Blunt Truth or False Hope?

Inflammatory BC

Inflammatory Breast Cancer — diagnosis is based on the clinical picture that accompanies the underlying cancer

In the days of “surgery alone” for breast cancer, the diagnosis of inflammatory carcinoma was a death sentence, usually in the range of 6 months. But then, reports emerged of 5-year survivors if chemotherapy was used up front, then surgery, then radiation.

And this was the hope in 1994 when I met Ruth (pseudonym), a 34-year-old who was in her first trimester of pregnancy when she presented with classic inflammatory breast cancer. Skin punch biopsy was performed, along with random core needle biopsies that confirmed the diagnosis. Then she underwent chemotherapy, beginning in the 2nd trimester, stopping prior to delivery of a healthy boy. One week after delivery, I performed modified radical mastectomy, which was then followed by radiation therapy.

Inflammatory dermal lymphatics involved

Skin punch biopsy — skin surface to the left, with a focus of dermal lymphatic invasion on the right.  This finding on pathology is not a requirement for the “inflammatory” designation, but is frequently present.

Pathology on the mastectomy specimen was not encouraging. Although the breast had responded both clinically and with only focal areas of invasion on microscopy, she still had 16 of 24 nodes positive. The outlook became even more grim when, one year later, I excised a nodule in her mastectomy scar, and recurrent cancer was confirmed.

 What do you tell the patient then regarding expectations for her future?

Younger physicians might not be aware that the standard of care for centuries was to tell the patient a lie, never disclosing that cancer was present. Plato said (in Greek, of course): “A lie may prevent the occurrence of undesirable views, beliefs or actions.” Although not covered in the Hippocratic Oath, the writings of Hippocrates place him in the same camp as Plato. And this was the prevailing practice for a long, long time, based on the notion that the patient’s attitude was critical for even a temporary recovery. Thus, false hope reigned supreme over the truth to give the patient the best possible odds.

In 1847, the AMA Code of Ethics followed suit by directing physicians to avoid making “gloomy prognostications to the patient,” (oddly, however, the physician was instructed to be completely honest with friends and relatives). “Only if absolutely necessary should the truth be given to the terminal patient.” The authority figure that led to continuation of this policy was Thomas Percival (1740-1803), “codifier of medical ethics,” whose influence extended to the AMA from beyond his grave in the U.K.

This practice of false hope was not without its detractors, however. One Rev. Thomas Gisborne wrote that physicians should be honest with patients on the grounds of conscience and the observation that “lies fail to convince patients anyway.” Instilling hope should be encouraged only “as far as truth and sincerity will admit.”

As for William Osler, apparently, he waffled on the controversy, claiming that the choice about blunt truth vs. false hope depends on context.

Remarkably, this (innocently deceptive) practice continued in the U.S. well into the 1950s and early 1960s, confirmed through several large surveys wherein the majority of doctors were still not honest with patients after a diagnosis of cancer. The rapid and dramatic shift to honesty came in the late 1960s and early 1970s in the U.S. where repeat surveys using the same questions now revealed nearly all physicians were honest about a diagnosis of cancer.

But the U.S. is not the norm. Many countries continue this deception as standard practice today, and a study in U.K. revealed that 37% of physicians still sometimes withhold the true diagnosis (I’m taking this from a 2006 reference, so it may no longer be the case – after all, it’s hard to imagine that in our current era where patients can access their own lab results online, that any deception still occurs in those countries with electronic medical records). Still, for many around the world, this practice of hiding the truth from the terminally ill continues, unchanged from Plato’s time.

At the other extreme, neurosurgeons (and occasionally, all physicians) are notorious for “hanging crepe,” that is, presenting a worse picture than probable, for a variety of reasons, not the least of which is the gross inability to predict the human brain (and many other diseases as well). And, of course, any outcome better than expected generates special praise for the user of the crepe-hanging approach.

As it turns out, breast cancer can be a lot like a head injury or a brain tumor when trying to predict the future.

For Ruth, my patient of 1994-1996, I felt the prognosis was grim, and I can’t recall how I couched the chance of survival, or if I avoided it altogether, leaving the topic for the medical oncologist. But this is my guess: I probably gave her a slightly optimistic outlook, while saying something like, “5-year survival is becoming more the norm and some patients are actually making it to 10 years.”

If I recall the actual numbers from the mid-1990s, I believe it was something like 5-10% were making it to 10 years if there were no distant mets. But for Ruth, 16 positive nodes and a chest wall recurrence so soon after completion of therapy was ominous.

With a newborn son, I have to assume Ruth was hoping for more than 10 years, even though we considered 10 years as a major triumph for a disease that had been universally fatal within months, just a few decades earlier. In truth, however, our prognostications were guesswork. Along with her pastor husband, Ruth would clearly be double-checking our estimated prognosis with the Almighty’s prescience, and would be leaning heavily on miracles of God, rather than the miracles of modern medicine.

The family moved away from Oklahoma City shortly after Ruth’s chest wall recurrence in 1996, and she was lost to follow-up even though I wondered about her on many occasions.

I’m going to pause here and give the reader a final chance to guess the outcome…

 Now, the follow-up…

One week ago (Jan 2019), Ruth’s medical oncologist and I each received a Friend Request on Facebook. It was from Ruth. 24 years had passed since her diagnosis. Disease free. Her 24 y/o son who had chemo in utero was also perfectly healthy.

Medical miracle – or – Miracle miracle?

As it turned out, it didn’t matter whether our approach was the “blunt truth” or “false hope.”  Ruth had her own success firmly arranged all along.


Like Rats Fleeing a Sinking Hypothesis

Megan handling DMBA

I claim to be an expert on rat mammary gland anatomy, a surprisingly complex topic that might seem to be of limited use in today’s world. And like many who claim expertise in a tiny niche, the truth is that it was my assistant in the lab (that is, a medical student) who did all the hard work in discovering nuances in rat anatomy that are not in the textbooks. Still, in accordance with the unwritten rules of academia, I claim all the credit.

You might think I’m kidding, since I often dabble in semi-comedic hyperbole. What could rat mammary gland anatomy possibly have to do with a clinical controversy? So, to avoid burying the lead, here’s where I’m headed – For over 30 years, researchers and clinicians have been quoting the studies of preventive mastectomies performed on female Sprague-Dawley rats in the DMBA (7,12-Dimethylbenz[a]anthracene) carcinogenesis model, wherein the number of cancers emerging after exposure to DMBA is – amazingly – the same whether or not mastectomy is performed. Of several articles on this topic, one of the more commonly referenced publications is from Wong JH et al in Surgery 1986; 99:67-71. Inexplicably, in this and similar studies, preventive mastectomies did not even make a dent in the number of cancers, roughly 5 breast cancers per rat, with or without “mastectomies.”

Then comes a non sequitur – as one prominent breast surgeon stated in the years leading up to the discovery, sequencing, and commercialization of BRCA-1 testing in humans: “Because the BRCA mutation will be present in every cell, it is possible that preventive mastectomy will have no effect whatsoever, as in the case of DMBA-induced tumors that occur with the same frequency whether or not the rat has undergone preventive mastectomy.”

Even at the time (circa 1992), that statement sounded like it contained a logical fallacy, though I couldn’t pinpoint the error from my short experience in Philosophy 101. Yet, there is an enormous difference between a “mutation in every cell,” versus “every cell will become cancer,” or even “every cell is equally at very high risk for cancer.” The human body is estimated to be composed of 32 trillion cells. If we assign 1% of those cells to breast epithelium, we’re talking about 300 billion cells. I attempt this rough calculation to avoid co-opting the phrase, “billions and billions,” attributed to Carl Sagan (but only adopted as mantra after Johnny Carson used the phrase on his TV show). Now, with 300 billion breast epithelial cells already carrying a BRCA-1 mutation, why is it that without preventive surgery, breast cancer will arise from only one or two or three clones over a lifetime? Alternatively stated, 99.999999999% of these cells never become cancer.

From the pre-BRCA era, we knew that patients with very strong hereditary risks were most likely to develop “only” one or two cancers during their lifetimes, the primary difference being earlier age of onset, far more impressive than the actual number of cancers over time. Even without a BRCA mutation, breast cells accumulate many somatic mutations over the course of one’s lifetime, sometimes generating a growth advantage to each clone, so it remains an oddity as to why only one or two clones of cancer cells emerge clinically. Perhaps the first cancer fires up the immune system to keep all the other “premalignant,” mutated cells in check. Perhaps it doesn’t even require a full-blown cancer to accomplish this. Maybe the body recognizes pre-malignant clones and acts accordingly for self-preservation. I don’t know. I only know that the mathematics don’t add up. 300 billion breast epithelial cells primed for cancer, yet only one tumor in most cases, two in some, and three or more in a few?

Many years ago (circa 1989), when I had the delusion that I was going to build a benign breast tissue research empire at the University of Oklahoma, I attended a basic science conference on the topic of early carcinogenesis in breast cancer. In my naiveté, when I checked in, I asked at the registration desk how I would get my CME. The registrar was caught off guard and seemed puzzled as I explained the meaning of CME. But she maintained composure as she broke the news to me – there wasn’t such a thing as CME at a basic science conference. I was the only physician attending.

Anyway, after I settled in, I realized that I was barely able to follow the presentations beyond the introductory 35mm slides. However, one talk was both understandable and memorable – a presentation on the somatic mutations in breast cells having normal morphology under light microscopy. Specifically, in breast cancer patients, when looking at adjacent normal tissue, many of the mutations were identical to those in the tumor, though not quite as many. But even more remarkable, some of these same mutations were present in tissue far away from the tumor, including the opposite breast. After all, what environmental insult (other than focal radiation) generates genetic “hits” in only one breast? The cells might look normal, but they are not. They accumulate somatic mutations long before changes occur under the microscope.

Then, the speaker drove the point home – “Halsted was right for the wrong reasons when he described his “field effect” as the justification for mastectomy. In fact, the field effect is real, but it’s a bilateral field effect at the molecular biologic level that only rarely translates to the clinic. And this is the case whether one is talking about accumulating somatic mutations in breast cells over a lifetime, or having germline mutations at birth. Having germline mutations in every cell certainly increases long-term risk of breast cancer, but the eventual crossover to malignancy occurs in surprisingly few cells, be it from either somatic or germline mutations…or both.

Coming at this from another angle, let’s apply the crystal ball to a 25 year-old BRCA-positive patient, and condense her future into the present. Yes, we know there are about 300 billion mutated cells, yet in the crystal ball, we see only 3 cancers emerging over 50 years, one in the RUOQ in the year 2026, one in the RLIQ in the year 2037, and one in the LLOQ in the year 2049. All 3 future cancers, however, are located within the tissue that is removed with bilateral preventive mastectomies. Thus, it does not matter that microscopic tissue containing mutated cells is left in this patient if she opts for preventive surgery, and this seems to be the case in 90-95% of patients who opt for preventive mastectomies.

Whether or not there is a direct correlation between percentage of tissue removed to relative risk reduction is unknown. I suspect those pesky residual cells in BRCA+ patients do, in fact, keep the correlation from being exact, but it’s close. My guess is something like this: If 99% of the breast epithelium is removed, there will be a 90-95% risk reduction. Since it’s a relative risk reduction, the BRCA+ patients will have a higher absolute risk of future cancer after preventive mastectomies than someone at lower risk (e.g., 90% risk reduction applied to 80% absolute risk leaves 8% lifetime risk post-mastectomies in a BRCA+ patient, whereas a 90% risk reduction applied to someone with a 30% risk leaves a 3% remaining risk).

Other variables prevent strict adherence to mathematical probabilities, however. Some patients have well-defined boundaries to their breast parenchyma while others are poorly defined. Then, there are varying degrees of precision from one surgeon to the next in carefully removing all grossly evident breast parenchyma. As for leaving the nipple-areola complex, this should be a moot point – not only are there fewer TDLUs in this area, but also if we resort to our crystal ball again, how many cancers are going to occur directly beneath the nipple? A few, but not many.

In 2017, the oncologic safety of nipple-sparing mastectomy in women with breast cancer was published in the Journal of the American College of Surgeons by Smith BL et al (Vol 225: 361-365). In 311 patients with established cancer, only 3.7% recurred locally, but there were ZERO recurrences near the retained nipple-areola complex. (We await their data on the 2,000 mastectomies performed for risk reduction in patients without cancer.) More pertinent to this blogatorial is the performance of risk-reducing nipple-sparing mastectomies in BRCA-positive patients before cancer occurs, addressed in “Oncologic Safety of Prophylactic Nipple-Sparing Mastectomy in a Population with BRCA Mutations” that appeared in JAMA Surg. 2018; 153:123-129 (and was my prompt to write this article).

In the short-term, 22 cancers were expected in BRCA1 and BRCA2 mutation carriers undergoing preventive surgery, yet no cancers actually developed. That’s ZERO cancers after 548 nipple-sparing mastectomies! In the invited commentary, there it was again, after all these years, just like in 1986: “Although it seems intuitive that reducing the volume of breast tissue would likely reduce the risk of developing breast cancer, BRCA carriers have germline mutations. Any residual breast tissue remains at the same inherent risk of developing breast cancer.”

So, we’re back to the making the distinction between “every cell has the mutation” as distinct from “every cell has an equally high potential to become cancer.” First of all, the only time each cell has perfectly equal propensity for cancer is at conception. After that, somatic mutations begin to accumulate in women (yes, even in utero) with or without germline mutations, and certain clones with a growth advantage will emerge in focal regions, not equally scattered throughout the breast tissue. Within those focal regions, further mutations will provide yet another growth advantage in even fewer focal areas, until there is eventually crossover to malignancy in only a tiny fraction of the original 300 billion cells. (Yes, the “two-hit” hypothesis for tumor suppressor genes is currently the rule, but my guess is that “clinical emergence of cancer” is more complicated than two-hits, as immune surveillance enters the picture.)

Now, back to the rats. It is helpful to note that the 2018 commentary (above) to the JAMA Surg article, reminding us of the “same inherent risk” in residual breast tissue, was written by the lead author of the DMBA article from 32 years ago wherein preventive surgery had zero impact on the rate of development of cancers. One can readily appreciate his skepticism about the short-term clinical data. Nevertheless, it takes working with the DMBA model to understand how it is analogous to the human situation, but more importantly – how the DMBA rats are different.

There are two reasons why the DMBA model is not a good way to study preventive surgery – 1) rat mammary anatomy, and 2) the DMBA carcinogen turns many cells malignant within a very short time frame.

Rat mammary anatomy – speaking as a bona fide expert now, there are indeed some analogies between humans and rats, e.g., breast tissue close to the skin predisposes to residual epithelium, very few TEBs beneath the nipple (TEBs are terminal end buds that are analogous to human TDLUs), etc., but it’s the differences that are important here. In spite of anatomic discussions describing the rat mammary “fat pad” where all the action is, there is no breast mound, or discrete parenchymal cone that one can call a “breast.” The rat has 12 nipples, each with a single duct opening to the outside, but the supporting breast parenchyma is a diffuse sheet that covers nearly the entire ventral surface of the animal, extending from lower jaw to anus and even wrapping around to the dorsal surface of the rat in some places. Diagrams of the extent of this parenchymal sheet indicate a formidable task for the surgeon who believes he or she can remove the diffuse breast tissue associated with 12 nipples. But in “our” experience, the extent of the parenchymal sheet is even more impressive, such that one is talking about removing the breast tissue from approximately one-third of the surface area of the entire animal. And in the words of my medical student assistant: “…a striking feature is the lack of boundaries in the mammary tissue.”

But that’s not where the problems end. As “we” discovered in “our” meticulous dissections, there are some mighty forces that keep the surgeon from a truly extirpative procedure. Again, in the words of James Banta, MS-2 on summer fellowship (who, after his experience with me, became an ophthalmologist): “Approaching the axilla, the most difficult part of the procedure, one encounters the cutaneous trunci muscle. It originates on the lesser tubercle of the humerus and inserts directly into the skin forming a broad, thin sheet that thickens as it approaches the axilla. This muscle penetrates the second and third mammary glands and separates them into superficial and deep layers. While the deep layer is easily removed, the remnant of breast tissue in the superficial layer cannot be properly excised without removing the thickened portion of the cutaneous trunci, which along with the breast tissue remnant, is tightly adherent to the dermis. Removing this muscle, with its breast tissue remnant, requires ligature of numerous tributaries of the dorsal branch of the lateral thoracic artery, resulting in necrosis of the skin flap.”

Yes, we have a difficult time getting rid of all the microscopic breast tissue in humans, too, but if you use Google Images for “DMBA tumors in rats,” you’ll see tumors under the animal’s jaw, or on its back, or near its tail, and it will give the term “residual breast tissue” a whole new meaning.

Add to this the second reason why this particular model is not a good one for surgical prevention — the DMBA carcinogen turns many cells fully malignant within a very short time frame.

 The DMBA model is sometimes used as a good example in distinguishing “initiation” from “promotion,” a basic principle behind carcinogenesis. DMBA does the initiating (mutations), while hormones (esp. estrogen and prolactin) do the promotion. But not so fast. As early as 1962, the prominent breast cancer researcher at Roswell Park, Dr. Thomas Dao, made the case that the hormonal milieu in the DMBA model is part of the initiation of tumor cells. So what? Well, it means that cancer cells are “created” in one lockstep, unlike human carcinogenesis. In addition, these malignant cells are widely scattered throughout all the (diffuse) breast tissue.

On the average, if you give DMBA to female, virgin Sprague-Dawley rats at the age of sexual maturity, you’ll see 100% of them develop 3-5 breast cancers after a short latency of 8 to 22 weeks. And to show how powerful the hormonal contribution is, if you perform oophorectomy on these animals 4 weeks prior to the DMBA, you’ll get zero cancers. As Dr. Dao pointed out, initiation with prompt carcinogenesis (without true promotion) is thus accomplished through the combination of DMBA and hormones together. Hormones might have some additional promoter aspects, but this is secondary. Cancer cells are created at the git-go, and they are widely scattered.

In another departure from human counterparts, these histologically malignant DMBA tumors metastasize only rarely. If it were not for euthanasia, the tumors would kill through their bulk and local effects, draining the animal of all resources for life. So, most studies end by counting the initial wave of cancers, then putting the animal to sleep. So, what happens if you remove these cancers as quickly as they appear? They just keep coming. So many malignant cells are created by DMBA that there’s seemingly no end to the number of cancers.

And this is the scenario that some conceptualize for the BRCA-positive patient when they say “cells are just one step away from being cancer, so the risk is not lowered if any cells remain.” But this ignores our crystal ball for humans where we can see 3 or 4 cancers (max.) spread out over a 50-year time frame. (Yes, I know, we’ve all seen 5 or more simultaneous separate cancers in one breast, but this is a rare exception and raises questions if these are truly 5 different clones or, more likely, a pre-existing widespread DCIS, or intramammary lymphatic spread, or….but I digress.)

Pathology in the DMBA model is usually drawn from palpable tumors. But if you sample anywhere in the diffuse sheet of breast tissue, you’ll find wildly atypical cells that are lying in wait to emerge later. In contrast, when a BRCA-positive patient undergoes preventive mastectomy, most of the tissue is completely normal, maybe with too many lymphocytes in the lobules (perhaps keeping those billions of pre-malignant cells in check).

Yes, there is a fairly high rate of focal high-risk lesions in preventive mastectomy specimens, as well as 2-4% with invasive cancer in BRCA+ patients. But compare the reported pathology findings in BRCA-positive mastectomy specimens to the much-lower-risk preventive mastectomy patients, and you won’t find a great deal of difference in the incidence of ADH, ALH/LCIS, borderline lesions, or even occult DCIS (highly dependent on sampling technique, of course). A few articles even describe the BRCA patients with a lower incidence of high-risk lesions. But nothing compares to the DMBA model where it’s hard to find normal breast tissue anywhere. The point is that, in humans, there is a mismatch between the “molecular biologic field effect” that is present with either somatic mutations or germline mutations versus what emerges clinically. Therefore, we should rest our cases on clinical observations and not the DMBA model.

The primary benefit of the DMBA model is to test various hormonal strategies in the prevention and treatment of breast cancer, given that the malignant cells created are usually hormonally responsive. And this was why I had my team at the University of Oklahoma adopt the DMBA model, which is no easy feat when you’re starting from scratch (lab space, funding, animal care and protection regulations, handling of the highly carcinogenic DMBA, etc.). From this, we published on the prevention of rat mammary carcinoma with leuprolide compared to oophorectomy, then again with leuprolide compared to tamoxifen, plus we dabbled in experimental GnRH agonists and melatonin prevention as well.

The summer after the first year of medical school used to be wide open, and I made the decision to offer a (competitive) summer fellowship that was one-half clinical and one-half research. Funded by a small army of women that helped me at the time, the fellowship became very popular and was always filled with top students, such that we tried to take on more every year (we peaked at 4 students one summer). My purpose was not altogether altruistic, as I hoped to make an early impression on the impressionable, developing the future personnel who would fill the multidisciplinary clinical spots as well as the multidisciplinary research spots at OU. It worked well. In fact, our first student in the fellowship, Elizabeth Jett, subsequently became a breast radiologist and is currently the Director of the OU Breast Institute, where I had served as the founding medical director in 1993. One of my favorite photos from that era is Betsy holding a Sprague-Dawley rat while attempting a smile through her disgust.

But of all my prior students, it was poor James Banta, future ophthalmologist, who got caught in a hair-brained scheme I had at the time – laser photodynamic therapy (could we tag indocyanine green to cytokeratin?) to eradicate the residual breast tissue after mastectomy in the Sprague-Dawley rat (with human applications, of course). It seemed like the natural thing to do, that is, remove all the breast tissue you can surgically, then obliterate the remaining epithelium non-surgically. We were not total hacks when it came to the laser approach. We had the expertise and advice from Wei Chen, PhD who, 20 years later, would be awarded an R-01 grant from the NCI for his “laser photodynamic therapy with combined immunologic boost” approach in a variety of cancers. But in the poor Sprague-Dawley rats, it was just too much, even though we applied the laser to just one side of the animal (see photo below for back-of-the-envelope planning as we designed our study groups).

DMBA research


In spite of our best intentions, we had an unacceptable mortality rate, and our never-published paper that followed was a treatise on surgical technique used for preventive mastectomies in rats, coupled with an extensive discussion of aggressive post-op care of the animal to avoid mortality. Like rats fleeing a sinking hypothesis, we ended the laser project. And from that experience, an ophthalmologist was born.

Returning to the question at hand – what is the future breast cancer risk for patients who undergo bilateral preventive mastectomies for BRCA-positivity, or for that matter, any of the strong genetic predispositions? We already know short-term risk is dramatically reduced in several studies. And, longer-term risk is also apparently reduced, as evidenced by the Mayo Clinic data where the mastectomies were done many years ago, then BRCA tested later. Numbers are small (26 with BRCA mutations), but zero cancers occurred after a median follow-up of 13.4 years. Importantly, 90% of the women in the Mayo Clinic series had the old-fashioned “subcutaneous mastectomy” wherein more tissue was left behind than today’s iteration of “nipple-sparing mastectomy.” Perhaps, modern results will be even better. The meta-analysis of De Felice F, et al (Ann Surg Oncol 2015; 22:2876-2880) suggests a 93% relative risk reduction, although a number of caveats exist here (starting with the admitted possibility of some patients from different studies being counted twice).

Since we aren’t certain about lifetime risks (or even 20-year risks) after preventive mastectomies, how do we counsel patients as to future risk? While some are confident that no cancers will occur, and their patients are told “no need for imaging follow-up,” I am biased by a small group of patients in my care who underwent subcutaneous mastectomies (old school technique) many years prior to BRCA testing, then were found later to harbor the mutation (similar to the Mayo Clinic sub-group). Their preventive surgeries were performed on the basis of family history, but then later developed breast cancer in their skin flaps (with the longest interval between surgery and cancer being 37 years in a BRCA1 positive patient – surgery at 40, then triple-negative cancer arising beneath the skin flap at 77). I consider these patients to remain at lifelong risk (probably a linear risk), prompting my policy of ongoing imaging. Interestingly, none of the cancers in my patients arose beneath the salvaged nipple-areolar complex.

As we await better long-term data, I offer another example as to how I handle counseling: If we’re talking about a 40 year-old BRCA+ patient who has 40 more years of life expectancy, and whose strong family history places her at the high end of the wide range of risk (80%), and then she undergoes bilateral salpingo-oophorectomy, her risk will be reduced to an estimated 40% lifetime (50% relative reduction from an absolute 80% to 40%). This remaining risk can also be stated as “1% per year.” Then, for further risk reduction, she undergoes bilateral preventive mastectomies, which takes her from 40% to 4% (lifetime). In this case, the 90% relative risk reduction is applied to the absolute 40%, leaving 4% (or 0.1% per year).

If, however, someone is diagnosed with a BRCA mutation later in life, age 50, after the benefit of early-age BSO is lost with regard to breast cancer, then she has higher risk (though not the “full” 80% that she had at age 25). If she has a strong family history to support the mutation, then she has an approximate 60% remaining lifetime risk spread out over 30 years, or 2% per year. A 90% relative reduction of this 60% leaves her with a 6% lifetime risk of breast cancer after bilateral preventive mastectomies. This 6% is only slightly less than the average risk patient at the same age who has never undergone breast surgery. This is my mathematical justification for continued monitoring and screening in most gene-positive patients who have undergone preventive surgery. In reality, it’s a matter of logic pending data, using lower school math.

For a quick tutorial on the controversies surrounding our current attempts to project risk and counsel patients, let me recommend the invited editorial by David Euhus, MD (Ann Surg Oncol 2015; 22:2807-2809), written in response to the 2015 meta-analysis mentioned above.

And please, no more extracting DMBA data to apply to human preventive surgical approaches. Let the rats off the ship. We have enough clinical evidence to know that preventive surgery in humans lowers risk substantially. Although our data is largely short-term when compared to “lifetime,” there is no reason to believe that these dramatic reductions in risk are only temporary and that the genetically-primed cells are going to “catch up” later on, rendering surgery a waste of time. No, the most pressing data is going to be the impact of surgical prevention on survival. That data is, in fact, starting to trickle in, and things are looking good so far.

Lifetime Risks are Lame – Let Me Count the Ways

Along with the 2007 announcement from the American Cancer Society proclaiming that women at high risk for breast cancer should be screened with MRI came an unfortunate by-product – the golden calf of LIFETIME RISKS.

My choice of “lame” is not without some thought, the implication being that one can still walk if lame, but it can be a struggle.

The often-ignored definition of risk assessment is the calculation of “absolute risk over a defined period of time.” The problem is: “lifetime risk” is poorly defined. There is a world of difference between cumulative lifetime risk (what many conceptualize) and remaining lifetime risk (what the models actually calculate).

In addition, the certainty about the power and persistence of risk grows less and less over time. Many of our lifetime risks are simply linear extensions of shorter studies. For instance, the Tyrer-Cuzick model will calculate almost 70% lifetime risk when a 35 year-old is diagnosed with LCIS, but most follow-up studies are limited to 20 or 30 years. No cohort of LCIS patients has been published with a mean follow-up of 50 years, which is what the T-C model is calculating in a 35 year-old.

Then there is the question about calculating lifetime risks to identify patients to apply screening technology that is unlikely to be in use in 50 years. And the bigger question is whether or not screening for breast cancer will be required at all in 50 years. Eventually, with an effective “cure” for all stages of disease, there will be no need to screen.

Lifetime risk calculations are not without serious hazard. By generating high numbers (esp. to get insurance coverage for MRI screening), there are some women who will “jump ship” and move ahead toward preventive mastectomies based on inflated figures. Lifetime risks have the curious tendency of “piling up” and weighing the patient down as if the entire load is going to come spilling down on her “any day now.” The addition of SNPs to risk modeling has the potential to make that scenario more common, with inflated values for risk that have never been prospectively validated after consolidation (grouping individual SNP risks into a whole).

I was challenged to learn more about risk assessment in 1991 by Dr. David Page who cautioned back then: “Limit risk assessment to a 20-year maximum calculation” (for all the reasons above). It took many years to fully understand this sage advice, now applicable to my longstanding criticism of our MRI screening guidelines.

Admittedly, it was a stretch for the American Cancer Society (ACS) to endorse MRI without mortality reduction data (and with relatively small studies), but they correctly understood that proof of a mortality reduction by adding MRI to mammography would be difficult to come by within a reasonable time frame. Here we are, 11 years later, still with no mortality reduction data for MRI screening, a benefit that will only be confirmed through a prospective, randomized trial.

At the same time, there was incontrovertible evidence that MRI was detecting many more cancers than mammography (double to triple the number). Without confirmation of a mortality reduction for MRI, the next best thing is the surrogate of Sensitivity. (Specificity only addresses the practical aspects of screening, not mortality reduction).

Invoking deductive reasoning, the syllogism works something like this:

Major Premise: Early detection with mammography reduces breast cancer mortality by 20-30% with only 40% Sensitivity, indicating that breast cancer biology is quite vulnerable to early detection.

Minor Premise: Screening MRI has a 90% Sensitivity, and Sensitivity and Biology are the only variables involved in mortality reduction.

Conclusion: Screening MRI will result in a mortality reduction well in excess of what is achieved with mammography alone.


Had there not been the foundation of a proven mortality reduction with screening mammography, then a proposal to screen with MRI would have floundered. And if you winced while reading the “40% Sensitivity” for mammography, I did not pull that out of thin air. In fact, that’s the sensitivity level for mammography when compared to MRI in the combined analysis of 5 international MRI screening trials. (Sardanelli F, Podo F. Eur Radiol 2007;17:873-887). If you don’t like this 40%, the ACS weighed in as well, basing their recommendations on 6 international trials (the 5 above plus one more) wherein mammographic sensitivity ranged from 16% to 40% (Saslow D, et al. CA Cancer J Clin 2007; 57:75-89).

So, going out on a long limb, the ACS opted to create guidelines for patient selection that approximated how 6 international MRI screening trials had been designed. And that’s when the problems began. Relying on the strategies used for patient inclusion in those trials laid a faulty foundation. The MRI screening trials were not focused on ideal patient selection, but on proving the benefit of MRI. Lifetime risk was the norm for all studies. This skews the experience to favor younger women where lifetime risks are higher, based on the key word – remaining.

All our mathematical models calculate remaining lifetime risk, not total cumulative lifetime risk. As we age, we “pass through” our various risks, until we finally meet up with that 100% risk of death, wherein the remaining lifetime risk for any new disease is finally 0. Thus, lifetime risks DECLINE over time, while short-term breast cancer incidence INCREASES over time.

By focusing exclusively on empirical data to the exclusion of rational thought, we got our risk strategies perfectly backwards. Consider tamoxifen prevention. Entry requirements for the NSABP P-01 trial were based on short-term risks even though the effect of tamoxifen is durable over the long-term. Here, we should be using 20-year risk calculations, but by sticking to guidelines that duplicate P-01, we use 5-year risks primarily. In contrast, with MRI screening – where we want to know the probability of a mammographically occult cancer in the short-term, specifically on a given day – we use long-term risks that lamely reflect the chance of a cancer being found on an MRI in the short-term.

Granted, when we use lifetime risks, we are increasing cancer detection rates (CDRs) over the long term due to the higher rate of disease incidence. And if we were using MRI to screen only BRCA-positive patients, a substantial difference in CDRs would exist. But when we move down to the 35 year-old at 21% lifetime risk for breast cancer vs. the 35 year-old at 12% general population risk, the difference between CDRs over the next 20 to 30 years is negligible. This is what keeps “precision screening” from being precise. When you convert risk levels to the actual differences in yield, there’s really not that much difference between high risk and normal risk, with the exception of patients at “very high” risk.

Even the initial prevalence screen in high risk vs. baseline risk does not generate as much difference in CDR as one might expect. Check out the prevalence screen data from Dr. Christiane Kuhl’s MRI screening study in the general population (Radiology 2017; 283:361-370) – it’s a comparable CDR on that first screen (22.6 per 1,000) to what one finds in the high-risk international screening trials (i.e., 22, 22, 23, 29, 30 and 36 per 1,000).

Returning to the main reason lifetime risks bomb when put into practical use – the impact of age on remaining lifetime risk – let’s walk through the two different ways in which “lifetime risk” can be conceptualized:

The oft-quoted “12% lifetime” is total risk over the course of an entire lifetime, from birth to age 90 or beyond. This is a cumulative lifetime risk. (And if we take away those women with known risk factors for breast cancer, it’s not 12% — but more like 7% to 8%.) However, this is NOT what the mathematical models calculate. All models calculate remaining lifetime risk, which is directly related to the patient’s age, that is, the remaining number of years anticipated.


powerpoint #1


In the diagram above, the top graph demonstrates how we tend to conceptualize lifetime risks where the solid line indicates lifetime cumulative risk for the general population (12%), starting at age 20, with the patient icon at the end of that long accumulation. The dotted line below the solid one represents the “lifetime risk” that a 60 year-old is facing for her remaining years (7%). These cumulative lifetime risk graphs are deceiving in that one senses a persistent rise in risk over time. In reality, however, the lifetime risk as viewed with the patient icon looking forward in time (bottom graph), reveals that remaining lifetime risk is actually declining.

But that’s only the first step in clearing the confusion. Once you realize we’re talking about a declining number over time, how do you reconcile increasing short-term incidence for breast cancer to a peak age at 55-60? I fashioned the diagram below for my book – Mammography and Early Breast Cancer Detection: How Screening Saves Lives (McFarland, 2016) – trying to illustrate why we have the paradoxical situation of high short-term risk (in terms of rate/100,000) in the face of declining lifetime risk. That is, lifetime risks are going down, while short-term incidence is rising.

Powerpoint #2


To try and explain this paradox, I used two different y-axes. The dotted lines related to the y-axis on the left represent remaining lifetime risk, the top line for a high-risk patient, the bottom line for a baseline risk patient. The solid line relates to short-term incidence with the y-axis on the right. Since there are different units of measurement for the 2 y-axes, the design was created for illustrative purposes to make the point that these two oft-quoted numbers are paradoxically at odds with each other. As one ages, their remaining lifetime risk is in constant decline, whereas the short-term incidence peaks around 55-60, then slightly declines.

As a result of this paradox, the use of lifetime risks is highly discriminatory to the older age groups (and don’t forget the net benefit of mammographic screening, in general, is found from ages 60 to 69). A young woman with risk factors can easily qualify for MRI even though her short-term probabilities of breast cancer might be low, while the older woman with the same risks and high short-term probability fails to meet the Golden Calf standard of “20% or greater lifetime risk,” according to American Cancer Society guidelines (with NCCN and others joining in later using very similar guidelines, endorsing the 20% threshold).

I’ve used the following example of age consequences and discrimination since 2007 in multiple publications and presentations, and it still holds true today:

Powerpoint #3


When it comes to screening MRI, the question we’re asking with selective screening is “How can we maximize cancer detection rates (CDRs) to make this cost-effective? These CDRs are directly related to disease prevalence and incidence in the screened population, and it has been the specious conclusion since 2007 that the best way to do this is through “remaining lifetime risk.” But look what happens in the example above, where the 30 y/o easily qualifies for MRI, but her risk over the next 10 years is only 3.5% (Claus model). Because the 60 y/o with the same risk factors is discriminated against through the use of lifetime risks, she fails to qualify for breast MRI even though her chance of having a mammographically occult breast cancer over the next 10 years is TRIPLE the patient who does qualify for MRI.

And in another twist of the same principle, if a 30 year-old has no risk factors other than a biopsy showing ordinary hyperplasia, the T-C model will calculate a 23% lifetime risk, which is based on 55 years of remaining risk. So this 30 y/o qualifies for MRI, while our 60 y/o in the example above with two first-degree, premenopausal relatives with breast cancer does NOT qualify? And we’ve lived with this since 2007?

In the last scenario on the Powerpoint slide above, if we look at a 60 y/o patient with NO risk factors, her 10-year risk is nearly identical to our “very high risk” 30 year-old. Yet, try to order a screening MRI on a 60 y/o with no risk factors and then watch the brouhaha that follows. Remarkably, the ACS guidelines include the specific admonition that MRI is “not recommended” for women with lifetime risk under 15% — which, through age discrimination, excludes many women with occult cancers at a short-term rate higher than younger patients who qualify for MRI. An active statement against MRI screening based on lifetime risks is emblematic of a serious misunderstanding of the long-term-short-term paradox described above.

If you ever wondered why the NSABP P-01 trial for tamoxifen prevention used “risk of a 60 y/o woman without other risks” as their threshold for inclusion – in effect, turning an average patient into a “high risk” patient – it’s because of the paradox noted above, wherein the NSABP needed quick answers so they focused on short-term incidence rather than remaining lifetime risks. In contrast, the MRI studies and subsequent guidelines did just the opposite, focusing on long-term risk to the exclusion of short-term incidence.

Here’s yet another variation on how age discrimination is an inherent feature of remaining lifetime risk: At the individual level, a woman who barely qualifies for MRI at a young age will be unqualified (or disqualified) later, perhaps within a mere 5 years. Remember, remaining lifetime risks decline over time. And if you’re not updating previously calculated risk every 5 years or so, then you’re quoting a number higher than reality allows. 100% of our patients have declining lifetime risks, and it takes some effort to recalculate those risks periodically. I try to do this every 5 years, though I still encounter patients whose risk calculation is fossilized.

Example: Take the typical patient with one first-degree relative with breast cancer, her mother diagnosed with breast cancer at age 60. When the patient is age 40, the Tyrer-Cuzick model will calculate a 22% lifetime risk for breast cancer, qualifying her for screening MRI. But as time goes on, and the patient reaches age 55, just at the point when short-term incidence is peaking (and closing in on her mother’s age when diagnosed), she has passed through enough risk that the T-C model now calculates 18%. Too bad. No MRI.

In the ACS publication that announced the 2007 guidelines, Table 4 shows the fairly wide variation from one model to the next, using 5 different risk scenarios applied to the “preferred models” as advised by the American Cancer Society (BRCAPRO, Claus, Tyrer-Cuzick). The variation is concerning, yes, but here’s the kicker – All 5 clinical scenarios begin with the patient (or proband) being 35 years old. Why wasn’t there a Table that showed the much wider variation imparted by different age groups? Did the authors consider the difference between “cumulative” and “remaining?” Or, were they so fixated on the starting age for MRI screening, they forgot that many women enter high-risk programs at 50 or older, when the incidence peaks (and where it is harder to qualify for MRI)?

How difficult is it to fix the problem? Even though countless women have been denied breast MRI for the past 11 years due to age discrimination imparted by the faulty guidelines, the “fix” is so simple as to defy logic as to why it has not been done already (new guidelines are due any day now, I’m told). You simply add the option of a short-term risk calculation in addition to the lifetime option. Introduce a 5-year risk number and the problem is fixed (unless the threshold is too high).

As it stands now, we have patients qualifying for MRI but not SERM risk reduction, while others qualify for SERM risk reduction but not MRI. That makes no sense. “Here, take this pill every day for 5 years after reviewing the long list of side effects, including death from DVT….but sorry, you’re not at high enough risk to qualify for MRI screening.”

The trial design of ACRIN 6666 indicates that there are “some out there” who understand that CDRs are boosted through the use of both short-term and long-term risks, the former designed for older women, while the latter for younger women. It’s the only way to handle the paradox of short-term vs. long-term risk. In the ACRIN 6666 trial of screening ultrasound (with a subgroup also getting screened with a single MRI), patients had to have breast density (a complex definition) as well as a single risk factor in addition to the density. The single risk could be a Gail calculation of lifetime risk…..or a 5-year Gail calculation.

Trial design took place prior to widespread adoption of the T-C model, but the rationale used displayed a deep understanding of how to improve yields based on equitable criteria. For instance, the requisite degree of risk was lessened as density increased. Think about it. Two parameters – risk and density – intimately bound when the endgame is the probability of a mammographically occult cancer. Like the high risk patient, the high density patient is more apt to have a mammographically occult cancer than a low density patient. So when the density level is higher, the requisite risk level was relaxed in ACRIN 6666. It’s rational and insightful.

We are doing something similar at Mercy Breast Center in OKC in a NCI-funded study of 4,000 normal mammograms (NCI R01CA197150), using a computer analysis system with machine learning that converts density patterns to a Risk Score, comparing left to right, and year-over-year, a computer program developed by Drs. Bin Zheng and Hong Liu at the University of Oklahoma Advanced Cancer Imaging Lab in Norman. We use a sliding scale, wherein a Risk Score of 0.80 prompts a screening MRI, but if density is Level C, then a score of 0.75 qualifies, and for Level D, a score of 0.70 qualifies.

When the 2007 guidelines for MRI screening were released, there were so many inconsistencies and oddities, I assumed that corrections and modifications would be prompt. That has not been the case. As noted above, the guidelines were largely dependent on the international trials where the focus was on MRI performance, not risk assessment strategies. In fact, if you read the inclusion criteria of the international MRI screening trials, it’s not always clear in some of the studies how patients were selected.

In our 2014 publication that challenged current guidelines for MRI screening (The Breast Journal 2014; 20:192-197), we performed risk calculations using Gail, Claus, and Tyrer-Cuzick on all patients who had their cancer discovered through routine asymptomatic screening with MRI. Most of our MRI discoveries would have never happened had we relied on the ACS guidelines due to the fact that we had incorporated breast density levels into patient selection. This point system was first proposed prior to the 2007 ACS guidelines, although not in print until 2008 (Hollingsworth AB, Stough RG. Breast MRI screening for high risk patients. Semin Breast Dis 2008; 11:67-75.)

At the time when we introduced our point system (2008 in print, 2004 first use), we had only diagnosed 7 patients with MRI screening. None were identified with the Gail model, none with Claus, and only 3 of 7 with Tyrer-Cuzick. The impact of using breast density as an equal parameter to calculated risk was apparent to us early on. Our strategy also took into account the wide variation in the different models (by not relying on their illusion of certainty) and avoiding the paradox of declining lifetime risk in the face of rising incidence by simply placing patients in one of 3 risk level categories: Baseline Risk, High Risk and Very High Risk.

This might seem reactionary, but I’ve never fully embraced the mathematical models. Why? Their merging of risk factors is based on accepted statistical modeling applicable to industry in general, but with no accounting for the biologic interaction between risks. As such, the Gail told us that Atypical Hyperplasia and Family History were synergistic, consistent with the original work of Page and Dupont. But then (with the blessing of Dupont) the Mayo Clinic data now indicates that family history contributes nothing in this situation – the risk of Atypical Hyperplasia trumps family history and imparts the same level of absolute risk regardless of other factors. On a bigger scale, this is biology trumping mathematics.

And to support my skepticism of the mathematical models, look at the c-stats for the various models that reflect “discrimination” at the individual level. They’re not pretty. Better than flipping a coin, but nothing to brag about and well below predictive models in other diseases. The c-stats are comparable to “accuracy” in the statistical sense, and while the original Gail had an embarrassing 0.58, we’re still not above 0.70 with the latest and greatest models. Yes, you might see the word “excellent” associated with various models, but they will be talking about “calibration,” not discrimination. Calibration is the predicted-to-observed ratio, that is, how many cancers will develop in a cohort. And therein lies the unequivocal benefit of mathematical modeling – that is, in the design of clinical trials where investigators need to predict the number of breast cancers that will occur. But at the individual level (discrimination), not so good.

It is probably no surprise that I still like my original scoring system more than any other option. With 4 levels of density and 3 levels of risk, we generated a total that converted to a recommendation of 1) annual MRI, 2) biennial MRI, 3) triennial MRI, or 4) No MRI. In this model, patient age has no impact at all on selection for MRI. As a result, the age distribution in our MRI-discovered cancers closely reflects age-at-diagnosis in the mammographically screened population – that is, 80% of our MRI-discovered cancers are in patients over the age of 50.

Powerpoint #4

By the time of our 2014 publication, we had diagnosed 33 patients with MRI screening. Had we used the Gail model, only 9 of 33 cancers would have been discovered. Had we used the Tyrer-Cuzick model (where calculations are consistently higher), still only 12 of 33 cancers would have been discovered. And poor Claus – originally, the “preferred” model as it was the only model used in the international trials (of the 3 that used modeling) – here, only 1 of 33 patients would have qualified for MRI screening. Using all 3 models and opting for the highest calculated risk, and adding BRCA positivity, we still would have identified only 16 of 33 cancers (48.5%) in this loose interpretation of ACS guidelines.

Clearly, there are problems with the current guidelines. And it’s really quite simple – if you are going to use a second line of defense (MRI, in this case), then its use ought to be predicated on the probability that the first line of defense is going to fail. That first line is mammography, and the probability of failure is based entirely on breast density. Using risk factors alone (without density) to select patients for MRI screening does not address that first line of defense in any fashion whatsoever, an incomprehensible deficiency.

In fact, even though risk and density were equally weighted in our 2008 “point system,” one can make the case to jettison risk calculations entirely, and base MRI screening on density levels alone. Witness what is going on in The Netherlands with the DENSE Trial of MRI screening. This is a prospective, randomized trial of mammography every 2 years versus mammography plus MRI every two years in women aged 50 to 75 (over the course of 3 screens), using a single entry criterion – Level D density.

 Risk levels have been tossed out (other than the inherent risk of Level D mammograms) and the entire study is predicated on the idea that this group of patients will harbor a substantial number of mammographically occult cancers (comparable to the yields in the high risk MRI screening trials). If the ACRIN 6666 subgroup that underwent a single MRI is any indicator, the Dutch will generate a landmark study that, by the way, includes mortality reduction as one of the endpoints.

I began this blogatorial with the intent to focus on the paradox of long-term risks versus short-term incidence, but before I knew it, I had slipped into my chronic, ongoing rant about the treatment of mammographic density as some sort of isolated risk factor that “needs more research” in the 2007 MRI screening guidelines.

But there is so much more in the current guidelines to whine about – such as the disconnect between risk of breast cancer and risk of gene-positivity (addressed by Kevin Hughes, MD and his team in Cancer 2008; 113:3116-3120). Then, there’s the odd approach to tissue risks by the ACS wherein suddenly “lifetime risk” with ADH/ALH/LCIS is tossed out the window and the risks are described with short-term values, such as “12-year follow-up.” So, a young woman at 40% lifetime risk after a diagnosis of ADH does not qualify for MRI while the same woman at 21% lifetime risk due to family history will qualify (thankfully, peer reviewers don’t follow the letter of the law, and the ADH patient will usually qualify.) This is a good thing. We are now up to 52 MRI-discovered cancers, and ADH was the dominant risk factor in 12 of the 52.

Okay, I’m clearly rambling now, and the next thing you know, I’ll be discussing the exclusion of patients with prior breast cancer in the “needs more research category” where I guess we’re supposed to be comfortable with 40% sensitivity for mammography.

The revised MRI screening guidelines – 2nd Edition – were targeted for release several years ago, and I’m not sure what happened. When the ACS released their revised mammography screening guidelines for the general population in 2015, it was stated that the high-risk guidelines would be next. I can’t complain that I didn’t get the opportunity to make my case. Although no one invited me to the party that will decide the new guidelines, I did have the opportunity at a committee meeting to chat with one of the key policymakers at the American Cancer Society who will be guiding the new recommendations. As I discussed the age discrimination problem, it was clear that all the ramifications of our current system had not been considered the first time around, especially as pertains to remaining lifetime risk.

So, we await the new guidelines. After what I’ve seen so far, given the precious few who have devoted careers to the nuances of risk assessment, I’ve got to raise this skeptical toast to the policy-makers: “May you not make things worse than they already are.”


Invasive Lobular Carcinoma Doesn’t Make the Grade



The use of histologic grading in breast cancer is a time-honored tradition grounded in decades of controversy. To wit…

Bloom-Richardson (1957) duked it out with Black (nuclear parameters only), then Scarff and Torloni added their modifications in 1968, which then became the Scarff-Bloom-Richardson system (Torloni apparently didn’t make the cut). Then, in 1991, Elston-Ellis weighed in with their modification of Scarff-Bloom-Richardson, and while this was “abbreviated” for a while as the Elston modification of Bloom-Richardson, we are all thrilled that, instead of the rambling “Elston-Ellis modification of Scarff-Bloom-Richardson with a nod to Torloni,” we can rest in the diminutive institutional eponym – the Nottingham Grading System.

Having witnessed the 1991 Elston-Ellis modification (now Nottingham), it was “peace at last” as both reproducibility and reasonable prognostic information were confirmed…at least to an acceptable degree at the time (What pathologist doesn’t waffle between Grade 2 & 3 on occasion?)

Tubule formation, mitotic count and nuclear pleomorphism – three headers, each generating 1 to 3 points. When the 3 scores are added, Grade 1 totals 3-5, Grade 2 is 6-7. Grade 3 is 8-9.

Left out in the cold, however, with the adoption of the Nottingham system was invasive lobular carcinoma (ILC). By definition, the histology of ILC does not allow tubule formation as a feature. So, a low grade ILC gets penalized with 3 points before one ever gets to the true distinguishing features – mitoses and nuclear pleomorphism. This creates the bizarre situation wherein invasive ductal grading starts with a minimum total of 3 points, whereas an ILC (a slightly more favorable histology) starts with a minimum of 5 points, ready to leap into the Grade 2 category. One simple, subjective, slip-up here (a 2 for either mitoses or pleomorphism), and you’ve turned a truly Grade 1 cancer into Grade 2 with a total score of 6.

Note that the long-discredited Black Grading System is custom-made for ILC, as the Black system does not count tubule formation in the first place. In this era of software medicine and synoptic reporting, most pathologists reluctantly comply and slap a “3” onto ILC from the git-go due to the lack of tubule formation, rationalizing that it really doesn’t make that much difference. And in this era where systemic treatments are based on other markers, this appears to be correct.

Some pathologists prefer the purist approach and will avoid grading ILC using the Nottingham system altogether, while others have taken an active stance of resistance, pointing out the need for a 2-tiered system applicable to ILC only (e.g., Adams Al, et al. Histologic grading of invasive lobular carcinoma: does use of a 2-tiered nuclear grading system improve interobserver variability? Ann Diag Pathol 2009; 13:223-225).

In general, however, most go with the flow, and as stated in one major medical center’s guidelines for pathologists – “Although classic lobular carcinoma by definition is scored as 3 for lack of tubule formation, cases will receive scores of 1 for nuclear pleomorphism and the mitotic count will usually be low, resulting in an overall score of grade 1.” Not quite – I’ve seen (in print), the remarkable claim that “classic ILC is a Grade 2 tumor,” demonstrating how error can disguise itself as the norm. In fact, classic ILC has cells that are so bland they can appear to be lymphocytes, such that the slightest pleomorphism can add the dreaded 2-point burden, thereby converting a Grade 1 to Grade 2. When I reviewed our classic invasive lobulars for the past 3 years, it was an even split – half were Grade 1 and half were Grade 2. From the biologic standpoint, and if put on equivalent footing with ductal, nearly all would be Grade 1.

(As an aside, I still hear comments that “invasive lobular is more aggressive than ductal,” which is not the case. Instead, ILC is much more likely to be missed on clinical exam and mammography such that we often encounter insidious, advanced stage tumors. The clinical impression comes from delayed diagnosis, not inherent biology…with the exception of pleomorphic lobular.)

To be fair, one rare variation of ILC forms tubules – a tubulolobular. As a result, nearly all tubulolobulars will end up as Grade 1. However, the other variants – classic, solid, mixed and alveolar (pleomorphic addressed below) – have no tubule formation whatsoever as part of the diagnostic criteria.

In fact, most prefer to call tubulobulars a distinct histologic sub-type rather than an ILC variant. In this scenario, the next step up in Grade will yield a diagnosis of “ducto-lobular” or “mixed ductal and lobular” with a variable e-cadherin pattern just adding to the confusion.

Until recently, the issue of flawed grading in ILC rarely altered therapy, so “who cares if we practice sloppy grading?” After all, treatment is based on tumor size, nodal status, ER/PR, HER2, and multi-gene panels.

But in January 2018, the AJCC announced version 8.0 of the breast cancer staging criteria, and with one fell swoop, histologic grade (formerly a distinct “marker”) was merged into the staging system. Finally, I thought, this serious defect in a grading system for ILC will get some attention and be corrected. If not, then we’re going to have a faulty grade convert to a faulty stage which converts to faulty therapy.

So, the first thing I did when I finally figured out how to calculate 8.0 Stage was walk through all possibilities using Grade 1 versus Grade 2, as this is where aberrant use of the Nottingham system for ILC will hurt – converting what should be Grade 1 to Grade 2. Amazingly, there were no alterations in Stage! Be it Grade 1 or Grade 2, the Stage was the same.

This was not true for Grade 2 vs. Grade 3 where one can demonstrate some differences imparted to Stage, but this is toward the pleomorphic ILC end of things where overtreatment is unlikely to occur. For an ILC to receive Grade 3, even with tubule formation getting penalized to a 3, mitotic count and pleomorphism have to add up to 5 or 6 – by definition, not “classic” ILC – and by then, the alert will sound that this is “pleomorphic ILC.” Here, the resultant higher stage supports the need for aggressive systemic therapy.

The tougher issue, addressed herein, is whether or not the 3 points for “no tubules” might take someone from Grade 1 to Grade 2 with an advance in Stage that prompts systemic therapy that would not otherwise be recommended. And, after all my belly-aching above, it does not appear this “over-grading” will happen with the new 8.0 staging system. Multi-gene tumor profiles make it even less likely.

So, in the end, we can all return to our deep slumber and ignore the fact that invasive lobular carcinoma is irrationally graded as a matter of routine, with a 3-stroke penalty before the game even begins.

The Premature Burial of Breast Cancer Histology


The conference room could have held 8,000, but many of the seats were empty. So I’m guessing that at least 3,000 witnesses at the 2017 San Antonio Breast Cancer Symposium heard the same presentation about triple-negative breast cancer that I did.

Investigators had discovered an unusual and unexplained finding – triple-negative cancers arising in BRCA-1 gene-positive patients had a better prognosis than triple-negatives in the general population. At the end of the presentation, curious audience members approached the microphones to question the presenter and offer scientific challenges.

The answers went something like this, no matter what the question: “We simply don’t have an explanation for our findings, so more work needs to be done on this.” No thoughts from the podium. None from the floor. Yet, the answer was easy, obvious and available 20 years ago, before this study was even conceived.

However, the answer has been lost in a frenzy of molecular biology. We no longer impress our scientific colleagues by using “old” terms that emanated from such clunky tools as the light microscope. We prefer terms like Luminal A, Luminal B, triple negative, etc., to describe our cancers being studied. Problem is, we have 50-75 years of quality information that is based on the light microscope, and those truths did not disappear just because DNA microarray was invented. Yet, histologic sub-types have become passé, even buried, prematurely.

The key word that was left out of the San Antonio presentation was “medullary.” No one mentioned it, and I wouldn’t be surprised if some of the basic scientists had never heard of it. But when BRCA-1 was first identified, a wave of articles soon appeared revealing a disproportionate rate of medullary carcinomas occurring in BRCA-1 positive patients. Even those cancers that did not qualify for “pure medullary” met some of the requirements and were called “atypical medullary carcinomas,” again in disproportionate numbers in BRCA-1 positive women. Were they triple negative? Most of them, yes, only we didn’t use that particular phrase, even though the 3 markers were routinely tested.

Not all triple negatives are medullary (or atypical medullary), nor are all medullary cancers triple negative. But when a classic medullary was diagnosed in a BRCA-1 positive patient, it didn’t take long for researchers to figure out that the prognosis was better in these patients, consistent with the favorable prognosis in cases of medullary carcinoma in the general population. Stated alternatively, the 2017 results being presented were actually known under the guise of “medullary” 20 years ago.

In spite of their ominous appearance (the worst cytologic features anywhere in breast pathology), medullary carcinomas paradoxically have a much better prognosis than non-medullary triple-negatives. This was the missing link at San Antonio. The researchers were likely encountering medullary cancers without recognizing them as such. After all, they were focused entirely on the molecular biology of “triple negative” breast cancer using highly sophisticated DNA arrays, while the light microscope was buried somewhere in another department.

In my view, medullary carcinoma is the most intriguing histology in breast cancer, bar none. Zero glandular formation, wild pleomorphism, large and multiple nucleoli, bizarre mitotic figures everywhere, so that the tumor looks as if it might be growing in front of your eyes in spite of formalin fixation. Amazingly, these features are somehow countered by the mystery of the lymphocytes that have corralled the nests of highly malignant cells into small groups, locking them inside the tumor, so to speak.

This “corral effect” is seen at the perimeter of the tumor as well, with a sharply circumscribed, pushing border, once again as if the lymphocytes have wadded up a piece of trash into a tightly compacted ball so that no cells can escape. Oddly enough, we do not see large necrotic areas in medullary from its rapid growth (doubling times of one month have been reported), indicating the cells are held in check by lymphocytes only temporarily. In fact, there is evidence to support that notion in that the “favorable” outlook of medullary tends to wane as tumors increase in size, especially after node positivity.

Nearly 30 years ago, I recall telling medical students who rotated with me on the breast service at the University of Oklahoma that if someone can figure out what those lymphocytes are doing in medullary carcinoma, then they’ll find a cure for breast cancer. Without claiming supernatural prescience, today we have intense investigations into TILs (tumor infiltrating lymphocytes), and ways in which those lymphocytes can be activated with immune therapies.

And here’s a wild thought – what if the unusual appearance of medullary cancer is due to the host’s unique immune system, rather than the specific, pseudo-aggressive tumor biology? In such a case, one person’s triple-negative tumor is another person’s medullary carcinoma, even though both individuals have identical tumor molecular biology. In fact, medullaries routinely test as super-aggressive, using proliferative markers or whatever, indistinguishable from the profiles of other triple negatives. The difference in natural history can only be told using light microscopy.

 In my wild conjecture, the tumor morphology would be the result of the immune system, NOT the inherent biology of the tumor. If this “host” stuff sounds familiar, Dr. Bernie Fisher first proposed host immunity as being of equal importance as tumor biology in the formulation of his “alternative theory” in the 1950s. Although he was referencing all invasive cancers, perhaps medullary histology is revealing how the optimal immune response can occur naturally. (As an aside, Dr. Fisher is still living, having turned 100 on August 23, 2018.)

Today, researchers know that cancer cells have a powerful mechanism that cripples the immune system, that is, the presence of PD-L1 on the cancer cell surface that binds to a PD-1 receptor on T-cells to prevent their activation. What if a medullary is nothing more than someone’s immune system that blocks this binding, allowing the T-cells to do what they should – corral the cancer?

The lymphocytic distribution in invasive cancer was recognized over 75 years ago, with the degree of inflammatory response proposed as part of the routine prognostic information. The idea was a good one, but it didn’t quite pan out enough for clinical utility. But go to any breast cancer meeting today, and someone will be talking about TILs (tumor-infiltrating lymphocytes), or other inflammatory cells/responses that might be of critical importance in outcomes, completely separate from inherent tumor biology. DNA microarrays might turn out to be only HALF the story.

In the BRCA-1 positive population that was discussed at San Antonio, not all the tumors were medullary, of course, but it wouldn’t take many to tilt survival in favor of the BRCA-1 positive patients with triple-negative cancer, given the large prognostic edge to medullary. The remarkable feature was that the word “medullary” was never used. The only language spoken was “molecular biology.”

I’ve noticed a trend wherein pathologists rarely use histologic sub-types as the final diagnosis. Instead, a tumor will be called “Invasive ductal carcinoma, with medullary features.” Such a soft call doesn’t allow for customized prognostic information to be given to the patient. Instead, “medullary” becomes a curiosity rather than a histologic sub-type with specific, associated behavior. And with this shift in attitude, hundreds of articles on the clinical behavior of medullary carcinoma have gone up in smoke, while researchers stand at the podium, mystified at their own results.


Another example of histology denied – Tubular Carcinoma


Largely a product of mammographic screening, the use of the term “tubular” has fallen by the wayside, lying in the gutter with “medullary.” Yet, there is a large body of evidence that tells us that this tumor is different than the standard Grade 1 breast cancer. Tubulars are not simply “well-differentiated,” they are “super-well-differentiated,” so much so that even the experts have a hard time distinguishing certain benign sclerosing lesions from tubulars.

What difference does it make? Well, given a pure tubular (90% minimum surface area meets classic tubular criteria, while remaining 10% is no worse than Grade 1), if it’s under 1.0cm in size, I’m not sure anyone has ever died in recorded medical history. One publication was just ambiguous enough that I can’t be 100% certain about the 100% survival rate, but it’s no worse than 99% (better than some DCIS). As for node-positivity with a pure tubular under 1.0cm, it hovers between 1% and 2%, less than the complication rate of sentinel node biopsy and well below 10-20% node positivity seen with garden variety breast cancer of the same size. Even when larger tubular cancers spread to lymph nodes, the outlook is still remarkably good, better than your non-tubular Grade 1.

As for multi-gene tumor assays that claim to trump histology and tumor size, I would say, “Not so fast.” While the majority of tubular carcinomas score low on these assays, some tubulars make it into the intermediate range. Granted, TAILORx takes the pressure off, but if someone is contemplating chemotherapy for a classic, pure tubular under 1.0cm, I’d say, “If you can improve upon 99% survival, then have at it.”

Virtually all pure tubulars are strongly positive for ER/PR, and endocrine therapy is often employed (although for the pure tubulars under 1.0cm, the systemic benefit is close to nil, while risk reduction for a new primary might be substantial). But what if you get back a report claiming that a pure tubular is ER/PR negative/negative? Odds are that you have a quality control issue. I’ve seen a triple-negative tubular once – and when I asked pathology to repeat the study, there was groaning and complaining. This was a number of years ago when the methodology was cytosol measurement rather than IHC. But when the case was analyzed, the original values had used a piece of the lumpectomy specimen (pre-core-bx-era) that had no tumor (normal breast cells are triple-negative, after all). When the study was repeated on actual tumor, it was strongly ER/PR positive/positive.

Other explanations are out there for ER/PR being low or negative, including the “tubular” designation being incorrect (due to high grade nuclei). However, a negative ER/PR for tubular should be a red flag for a problem somewhere.

Now, once you go below 90% purity, or increase the size above 1.0cm, all bets are off. But one of the most intriguing things about tubular carcinoma is the relationship between purity and tumor size.

Tubulars are known for their small size. Given a 0.5cm tumor discovered on mammography, it’s likely to be a tubular. And that brings up one of the most important facts in the tubular literature, mostly forgotten today (though I try to resurrect it from the dead at every available opportunity) – pure tubulars morph into routine invasive ductal as they grow. Okay, maybe I shouldn’t be so dogmatic, so let me defend that statement, as the implications here are huge – not for clinical management per se, but in the epidemiologic debate surrounding screening mammography and overdiagnosis.

Let me back up. When was the last time you encountered a 3.0cm pure tubular carcinoma? For me, the answer is “Never.” In fact, pathology studies were done long ago that correlated “purity” with “size.” That is, a direct correlation exists between the percentage of tubular histology and tumor diameter. To me, the conclusion is obvious and, importantly, without an alternative explanation — as tubulars gradually increase in size (slowly at first), they become less and less “pure,” eventually losing the “tubular” moniker. So, the pathology report might not even mention tubular features once the purity drops, let’s say, below 50%. The larger the tumor, the lower the percentage of tubular features.

I believe tubulars de-differentiate over time as part of their natural history. Hard to prove, since one can’t follow a single tumor to measure purity over time. But from the studies of multiple tubulars, the trend is impressive when purity and size are correlated. De-differentiation is the pathology term for what the basic scientists call tumor “progression,” that is, malignancies becoming “more malignant” over time with additional somatic mutations. (Note: the anti-screening epidemiologists abhor the concept of tumor progression.)

Tumor progression is at the heart of the debate on “overdiagnosis” of invasive cancer, a criticism of screening mammography. One cannot measure or recognize overdiagnosis as it is happening. Its existence is theorized entirely through indirect observations. One of the most common of these observations is the “excess” number of cancers that occur in the screened population in any of the historical studies. The attempt is then made to point out that these cancers would never result in the death of the patient as they do not progress. If they did, the number of cancers in the unscreened group should eventually catch up to the number found in the screened group.

However, in spite of the anti-screeners focusing on “excess, non-progressing cancers” drudged up through mammography, their indirect calculations force the necessity of a direct observation, a critical question rarely discussed – Where do these excess cancers go in the unscreened population? There are only two possibilities, if these “non-cancers” (as they’ve been called) are not found through screening – 1) dormancy – tumors reach a certain size then stop growing, or 2) they must regress, that is, they form, but then self-destruct, so to speak, without anyone ever knowing they existed. No matter how sophisticated and convincing the indirect evidence for “overdiagnosis” might be, one still has to account for those “excess” cancers being found on mammography.

And here’s where we circle back to tubular carcinoma, an ideal candidate to label as “overdiagnosis.” Even the lowly tubular, with its indolent course, will grow over time and, with additional mutations, eventually progress into a potentially life-threatening tumor. If tubulars can de-differentiate as they grow, one has to be very concerned about any invasive cancer having the same potential. The reason a tubular can be a functional “overdiagnosis” is because the age of the patient does not allow enough time for the lesion to progress, which can take many years. (Epidemiologists try to demonstrate that there are excess cancers with screening even in younger patients, but I don’t want to digress any further in this blogatorial).

Returning to our TWO (and only two) POSSIBLE OUTCOMES for “excess breast cancers” – dormancy or regression – theorized to explain the claims of the anti-screening epidemiologists, let’s address dormancy first. If tumors form, but then stagnate and quit growing, reaching a peaceful settlement with the host, then they should be found a-plenty at autopsy. Yet, from the pre-mammographic era of unscreened women, an analysis of all available autopsy studies (Ann Intern Med 1997; 127:1023-1028) showed only a 1.4% incidence of invasive carcinoma at autopsy (range = 0% to 1.8%), a strong indictment against the concept of tumor dormancy as 1% is exactly what one would expect from a single prevalence screen using mammography. (DCIS is another matter, not addressed here.)

As for tumor regression, the idea was so outlandish that those of us at the front lines tried to brush it away. But the “overdiagnosis voices” got stronger and stronger, apparently realizing that dormancy wasn’t going to fly, so they should focus on tumor regression. After all, it’s the perfect explanation, since the evidence has “disappeared.” Yet, as any breast radiologist will tell you, “I’ve never seen a breast cancer regress on serial mammograms.”

How would they know? They know from “untreated” cancers generated by: 1) refusal to proceed with treatment after diagnosis, 2) comparisons of newly diagnosed cancers to imaging done in prior years, and 3) “missed” cancers on mammography, where the radiologist gets to see the tumor develop (grow) year-over-year.

But these cavalier dismissals of tumor regression were anecdotal and couldn’t start to influence the evidence-driven epidemiologists. Fortunately, the Society of Breast Imaging put together a formal study (J Am Coll Radiol 2017;14:863-867) to look at “untreated cancers,” that is, diagnosed by core biopsy, but then no treatment, for whatever reason. At any single facility, this sequence of events is relatively rare, so in this study, engineered by Dr. Ed Sickles, it took 42 sites to pool their data, and – lo and behold – out of 240 cases of “untreated” invasive breast cancer (and 239 DCIS cases) there was not a single case of tumors shrinking or disappearinga strong indictment against the concept of tumor regression.

 Without dormancy and without regression, “overdiagnosis” becomes a relatively rare phenomenon, exaggerated largely by failure to account for enough lead time in the historical screening studies. Yes, the elderly patient with a 0.8cm pure tubular has probably been overdiagnosed, but not due to her tumor biology, as much as her life expectancy. In contrast to the relatively small rate of overdiagnosis (5-10% is reasonable), the outrageous estimates of 50% and higher that are being spoon-fed as fact to the U.S. media are beyond the pale.

And this entire diatribe is brought forth just because pure tubular carcinoma de-differentiates as it grows. It’s always fascinating how there can be so many implications from a single, arcane fact. The bigger question, of course, is how to reconcile tumor progression as promulgated by basic scientists and light microscopy versus the undying belief in tumor regression as proposed by anti-screeners.

Returning to my primary point: With today’s research techniques and strategies, ALL pure tubular carcinomas – recognized only through old-fashioned light microscopy — are going to be tossed into the larger batch of Grade 1 breast cancers, or Luminal A cancers, where the critically important phenomenon of tumor progression/de-differentiation would never have been seen.

Yet in spite of direct evidence against dormancy or regression, empowering tumor progression as very real, this is what we hear mostly today: “This implies that many small cancers are not destined to progress to large cancers; instead, their detection results in overdiagnosis.” (New England Journal of Medicine, June 8, 2017). Epidemiologists do not look through the microscope; they only look at numbers.

This is a cautionary blogatorial — tubular and medullary are only examples. We’ve not only buried the light microscope in breast cancer research, but we’re also starting to forget that it ever existed.

Mathematical Risk Models — where Independent Risks Become Dependent

2004 risk working group

In the early 2000s, breast cancer risk assessment began to ease into the clinic, specifically for guiding SERM risk reduction strategies.  Our Komen-sponsored Working Group (above), under the leadership of Victor Vogel, MD, cobbled some guidelines together over the course of one year before publishing in 2004.  But risk assessment didn’t hit full force until the 2007 American Cancer Society guidelines were announced for screening high risk women with MRI.  (After 11 years, these high-risk guidelines are currently in the process of revision.)

WARNING: The following discussion of quasi-precision medicine is not for those faint of heart when it comes to numbers and mathematical models of risk.

In the pre-Gail days, the only predictive models were esoteric family history tables from the epidemiologic literature that were virtually unknown to clinicians. No attempt was made to incorporate other risk factors, and maybe that was a good thing.

Having started one of the earliest risk assessment programs in the U.S., pre-Gail, I leaned heavily on the Dupont tables that converted relative risks to absolute risks over defined time periods, preferably 20 years max. The problem was determining a reasonable overall relative risk (RR) when more than one risk factor was present. Anticipating this problem, investigators had begun coupling risk factors and calculating single RRs. For instance, “nulliparity plus a first degree relative with BC” carried an RR of 2.7 in one study. Plug that into the 20-year risk table from Dupont, and for a 50 year-old, you’ll get a 12% risk of breast cancer over the next 20 years (thru age 70). This calculation is a little higher than what the Gail predicts today, and about the same as what the Tyrer-Cuzick (TC) model predicts.

The best-known coupling of risk factors was “atypical hyperplasia and a first degree relative with breast cancer,” with the Page/Dupont team assigning a synergistic RR=9.0 (later nullified by the Mayo Clinic data). Although Page and Dupont introduced their seminal work in the 1970s, it was a landmark paper published in 1985 by the New England Journal of Medicine that captured the attention of what was then precious few breast specialists. Nevertheless, multi-factorial risk assessment was born.

This dual-risk approach was sufficient for risk factors in couplets, but what about 3 or more risk factors? And what about those continuums that are applicable to all, such as age?

Enter Gail (1989 publication, although common use lagged until the P-01 trial), where risks were not directly studied as couplets, triplets, or beyond. The Gail approach didn’t care about biologic relationships. Rather, relative risks (RRs) were merged into absolute risk mathematically (with multiplication of RRs being at the heart of the statistical formulas). The user never sees the RRs, only the final tally in the form of an absolute risk over a defined period of time.

Validation of the Gail came through the Texas Breast Screening Project, then from the 6,700 women in the placebo arm of the NSABP P-01 trial (then CASH validation, then NHS validation), where it was accurate in predicting the number of cancers that would occur in a large cohort (called “calibration” – #predicted vs. #observed). But prediction at the individual level (called “discrimination”)? Well, not so good. With a concordance-statistic (c-stat, which is comparable to statistical “accuracy” expressed as AUC) of 0.58, the original Gail was not much better than flipping a coin for an individual. This should not be surprising given that the majority of patients who develop breast cancer do not have the risk factors included in the Gail model, if they have any known risks at all.

After the Tyrer-Cuzick (IBIS) model emerged and bypassed the Gail in popularity (for reasons I won’t bother with here), one could easily fiddle with many more variables to see the impact that individual risks had on the final calculated risk. C-stats crept above 0.60, but realize that .70 is a minimum standard for “fairly good performance,” and it would be far better to reach 0.80 where one will find the predictive models used for diabetes and cardiovascular disease.

In order to be included in the mathematical models, risk factors must be “independent.” That is, they exert their effect with or without other risk factors. Yet, after they are merged mathematically, strange things happen biologically.

Let’s look at a patient who undergoes risk assessment with the Tyrer-Cuzick model at age 50, having no other risk factors other than her plan to take Prempro™ HRT for the next 20 years (yes, it’s an extreme example to make a point).

Tyrer-Cuzick generates a baseline risk of 11.4% lifetime that will increase to 16.3% with 20 years of intended E+P. So, the risk of E+P is an additional 4.9% absolute risk over a 35-year lifespan.

Now do the same calculations of E+P for a woman who has 2 first-degree relatives with breast cancer in their 30s, is nulliparous, and has had a benign biopsy with specific results unknown. Baseline risk is 40%, and if she adds 20 years of intended HRT use with E+P, her absolute risk is now 63%. This is a 23% absolute increase in risk over a 35-year period with E+P in this patient.

So, the T-C model is telling us that this high-risk patient is far more susceptible to E+P than the average-risk patient.

Does this strike you as odd? Is this precision medicine? Are we to believe that E+P is nearly 5 times more powerful in risk potential if taken by someone who is already at high risk? If so, the independent risk of HRT with E+P is, in fact, dependent on other risk factors with the T-C model.

This is because mathematics drive these models, not biology. Mitchell Gail did not invent from scratch the mathematics he used in 1989. There were statistical rules already established for the merger of risk factors independent of medical science. They can be applied to industry or commerce or any other discipline that is analyzing risk. Biology has a way of fouling things up, however.

Does the Women’s Health Initiative help with this issue? Not really. Theoretically, if the Hazard Ratio is the same HR = 1.24 for all levels of risk, then yes, there will be a disproportionate increase in risk for E+P in the higher risk patients. In that study, however,  higher risk women had lower HRs (1.13), which would support my contention that absolute risk is fixed across the different levels of underlying risk.  That said, none of the calculations reached statistical significance, so we don’t know.

Let’s move on to the newly added breast density feature of the Tyrer-Cuzick model version 8.0. What’s the absolute risk of having Level D mammograms, a so-called independent risk factor? There is no single absolute risk. Like we just saw with E+P, the independent density risk factor is dependent on the other risks.

Using version 8.0 of the T-C model, start with a 40 y/o, but add no other risk factors, leaving breast density “unknown.” (Tricky point – if you try to use v.8.0 on patients younger than 40, the density feature will not work at all.)

Lifetime risk will = 11.3%, but now add Level D density and you’ll get 17.5%. So, Level D density carries with it an absolute risk 6.2% in this patient. Fair enough. Sounds about right.

Now enter our high risk parameters above – two 1st degree relatives diagnosed at 35, prior biopsy and nulliparity, generating a lifetime risk of 40%. Level D density takes on a whole new level of absolute power, raising absolute risk to 54%, or an absolute increase of 14%, compared to the 6.2% in the baseline risk patient (an even bigger disparity will be calculated if you use the no-risk patient rather than the patient with general population risk).

As an aside, if you fiddle with various densities, you might be surprised to learn that the referent (RR=1.0) in this new addition to the T-C model is Level C density. So, in our example above where the baseline risk was 11.3%, if you then add Level C, the calculated risk barely changes to 11.6%. This is in sharp contrast to what the high-density grassroots movement is doing when women are told (by legislation in most states) that they have Level C or D density and that this raises risk 2-fold. When it comes to the T-C model, only Level D counts, and even then, it’s dependent on other risks. Incidentally, both Level A & B will generate lifetime risks well below what you’ll get using version 7.0 of the T-C model.

Another feature to point out – even though Level D density might be considered a 2-fold risk factor (RR) when compared to the average patient (which should be the patient at the border between B & C with 50% density, not Level C), the model does not multiply baseline risk X2. Again, there are complex statistical models used here, and even though multiplication is at the heart of those models, it’s the accepted statistical formulas used for merging risk factors that keep that from happening.

Moving now to SNPs where we add virtually every known SNP studied in breast cancer to the T-C model, as currently being done by Myriad and Ambry. I already addressed SNPs in the January 2018 blog, where the heart of merging risks is multiplication. If there are 86 SNPs, all with RRs under 1.26, then it’s this: 1.10 X 1.02 X 1.18 X 0.9 X 1.10 X 0.88 and on and on until all SNPs are included, with a final RR usually close to 1.0 (no effect). That said, it is possible to generate cumulative risks (mathematically) that reach RR=2.5, more powerful than some of the predisposition genes being tested at the same time.

And not only are the SNPs impacting our final results more than I anticipated, they have more power on the protective side as well, so with a SNP tally having an RR below 1.0, I’ve seen 10 absolute %-age points shaved off of one of my patient’s calculated risk, dragging her below the threshold for MRI testing, whereas without SNPs, she easily qualified (now I ask for SNP panels only in my patients who do NOT currently qualify for MRI).

But as for the power of the SNPs as related to the calculations without SNPs, we have the same phenomenon as with E+P and breast density above. That is, for the baseline risk patient, the SNPs have little impact. But take the same SNP results and apply them to someone at 40% lifetime risk, and suddenly the SNPs are far more powerful. Same SNPs. Different power.

My question that I’m tossing out to experts around the country is this – If RRs are DERIVED from patients without the risk in question (that is, baseline risk individuals that serve as the denominator), then why are they APPLIED to patients after other risks are included? It works fine for the baseline risk patient. But when it comes to the patient already at 40% risk, the newly added risk factor was not measured in women already at 40% risk. Its power was calculated without other risks.

In contrast, witness what has happened to the combination of atypical hyperplasia and a first-degree relative with breast cancer. Originally, Page and Dupont measured a 9-fold risk, and this synergism (termed interactive risks) was considered solid. Recently, the Mayo Clinic data (including the assistance of Dupont with his Nashville data) has confirmed that the family history adds next to nothing once atypical hyperplasia is diagnosed. It’s the same 4-fold risk with or without FH that Page and Dupont had calculated years ago without family history.  So, the key point in question — i.e., independent risks being dependent on other risks — no longer applies to the combination of ADH and FH (using either the T-C model, or the “benign breast disease” model (BBDAH-BC) generated by Mayo).

The T-C model has therefore made adjustments to this new information (the Gail has not), and if you enter a patient who has ADH and no other risk factors, the ADH prevails no matter what else you add, with regard to family history, that is.  Jump over to breast density or SNPs or other risk factors, and in sharp contrast to what happened with the addition to a positive FH, the calculated risk will be augmented, sometimes dramatically.  Thus, the patching of the T-C model causes some remarkable incongruities, with ADH fixing the risk at a single point in spite of FH, but totally vulnerable to exaggeration with other co-existing risk factors.

Furthermore, try the same exercise using plain old hyperplasia, rather than ADH.  You’ll find that we’re back to synergism with FH, and the calculations begin to zoom once again.  In other words, only ADH fixes the risk at a value that is impermeable to family history.  This is what happens when you try to fix the model based on direct observation of two risks working (or not working) together.  Although a more accurate calculation s derived, it creates an inconsistency because the remainder of the model is still using statistical mergers that don’t care about biology.  Thus, hyperplasia and a first degree relative will generate the same risk as if that same patient has ADH and a first degree relative (wherein the FH won’t count in the latter instance)!

The point is that direct measurement of risk factors as couplets or triplets is going to be more accurate than our mathematical models that take independent risk factors and turn them mathematically into factors dependent on each other, without any prospective validation.

My intuitive approach would be blasphemy to the makers of these models. But just to give the statisticians a good laugh, here is how I would handle the addition of SNPs:

A 40 year-old woman is already at 40% risk of breast cancer due to other factors, and with her SNPs having been found to impart a 2-fold risk, her final calculation is a 63% risk for breast cancer – according the T-C model, enlisting the SNP feature (under Tools) Note: Even the overestimated risk, using approved statistics, don’t multiple 40% X 2 to get 80%, or else it wouldn’t take much to exceed 100%.

But the SNPs were not measured in women already at 40% risk. Could we be counting the same risks twice? What if the SNP risk is the same risk imparted by FH, or proliferative changes on a biopsy? (After all, it turned out that positive family history added nothing to the diagnosis of ADH. In the past, we were in effect counting the same risk twice.) Instead of deriving SNPs from women at 40% risk, the SNPs were derived from women without other risk factors (with the exception of some work in patients with positive family history). So it seems (to me) that if SNPs are operating independently, they should be converted to an absolute risk increase based on the general population risk from which they were derived, then added to the absolute risk already calculated. Same risk factor = same absolute imparted risk, regardless of other factors (true independence).

That gibberish is difficult to follow, so here’s how my (antithetical) methodology based on apparently faulty intuition would work: The 2-fold SNP risk in the baseline 40 y/o patient generates a risk of 22% (11% X 2), so this relatively powerful SNP combination has a power of 11 additional percentage points of absolute risk. If our patient is already at 40% risk, then we add 11% to reach 51%, not the 63% generated by the T-C model, and not the 80% generated by those who simply multiply SNP scores by calculated risk.

It seems as though it would be even more accurate to start out – not with the 11-12% baseline risk – but with a 7% lifetime risk (assuming full life expectancy in a young woman) since this is the true “no-risk” baseline upon which RRs are usually calculated. Thus, a 2-fold SNP result would convert 7% to 14%, or an absolute 7% increase (as opposed to the 11% increase above).

Would it be so weird to convert the well-documented risk factors to absolute risk elevations based on a 7% baseline, then add them together? (yes, it would, if we include all 100+ alleged risk factors that have been described over the years).

The statisticians are screaming foul right now, if they didn’t stop reading already. Nonetheless, I’m trying my best to understand what is happening here, as we pile risks on top of risks on top of risks, multiplying RRs all the way. I have ongoing e-mail discussions with statisticians and experts and “model designers,” and I’m not progressing very well as a student, still trying to explain mathematics at the biologic/philosophic/logical level wherein independent risk factors become so dependent on each other.

Help is welcome. If you can reconcile this conundrum, please contact me. This is a black and white issue only for the statisticians. And if you want to defend these models to the death, remember this: excellent calibration (# of predicted cases vs. # of observed cases) can co-exist with bad discrimination (predicting who is going to get cancer at the individual level). Even with density and SNPs added in, the models are still in the 0.60s on c-stats. That’s not good, and that is not precision medicine.

To be frank, I don’t think the science of breast cancer risk assessment has caught up with our technologies. We have the tools needed to find virtually all cancers early, but due to cost and inconvenience, we are limited to risk stratification based on marginal models. As a confirmed cynic and skeptic (as commonly occurs with age), I think the strategy outlined by our Risk Assessment Working Group sponsored by Komen many years ago (2004 – photo at top) was on the right track by simply assigning women to one of 3 groups – baseline, high risk and very high risk. Although that article was intended to refine risk assessment by identifying patients for ductal lavage and the search for atypia, we mentioned the possibility of MRI screening (prior to most results from the international MRI screening trials).

But no one today is interested in grouping patients into risk levels based on categorical risks. The trend is to convert anything that moves into a number wherein empiricism can take over from there on. After all, how do you add digits to the right of the decimal point if you’re using such archaic terms as “very high risk”? We are much more comfortable with the illusion of certainty provided by 29.7%, or 11.6%, or an improbable 91.3%.

In debates about the propriety of “risk-based screening,” both for routine mammography (doing less) and for multimodality imaging (doing more), remember we are dealing with mathematical modeling applicable to industry and commerce that might not reflect true underlying biologic risk at all, not to mention c-stats that are sub-optimal.

So what’s the big deal if we’re off a bit, as long as we ballpark it? Coupla things. First, insurers are not playing in the ballpark; they have canonized the precarious (and age discriminatory) 20% threshold, so there’s a huge difference between 19% and 21% now that we’ve created this quagmire for ourselves. But at the higher end of risk – take a patient at modest risk elevation, and now add E+P HRT, Level D density and bad SNPs, and in the quest to beat 20%, you end up with an inflated risk value that drives her to preventive mastectomies. It seems odd that we are to counsel breast cancer patients out of contralateral prevention because the risks of cancer are considered to be so low (sometimes quoting 10-year risks for a lifetime potential), while at the same time, we have no hesitancy to jack up risks in asymptomatic patients (quoting lifetime risks) and, as a result, drive them beyond MRI screening or SERM risk reduction to bilateral preventive mastectomies.

Alas, I’m like everyone else who battles insurers to pay for screening MRI – “Get that 20% even if it means reverting to version 7.0 for density level A, B & C patients.” I’m just trying to avoid that unwieldy trap of fooling myself. As Richard Feynman said in his 1974 Caltech commencement address: The first principle is that you must not fool yourself — and you are the easiest person to fool.

The Book That Keeps Writing Itself

Although I finished writing Killing Albert Berch in 2016, the book had not finished writing itself. After release by Pelican Publishing (Gretna, LA) in 2017, things began to happen. First of all, the children and grandchildren of characters in the book contacted me with their stories as to how they were connected to this 1923 true crime saga. This, I expected.

But the biggest surprise did not come until the book had been out for 6 months. If you’ve read the book, then you might remember a chapter title, “Where Have You Gone, Robert Johnigan?” This chapter was an account of my attempts to discover more about the African-American porter who was murdered along with my grandfather.

In the manuscript draft that was accepted for publication, I had hit a dead end, though I proposed a Johnigan family as Robert’s relatives, mostly living in Ardmore, OK at the time of the murders. But shortly before the book was finalized for the printer, I made a serendipitous discovery through re-checking where a “private story” had been generated about Robert Johnigan, a murder victim in Marlow, OK.

Chasing this clue led me – not to the Johnigan family – but to a friend of a Johnigan who had found the old newspaper article about the 1924 trials. He had been asked by the Johnigan family member to keep it quiet, as the descendants, mostly in Minneapolis and Kansas City, did not know about the murder.

Then, I stumbled into their world with my research, prompting the friend to tell the family members that a book was being written about their ancestor. Indeed, I had found the correct Johnigan family and awaited all the details. But then, in my arranged telephone interview with the spokesperson for the Minneapolis branch of Johnigan descendants, she informed me that the family would be opting out of participation.  The story of the murder had come as quite a shock to many family members. The only interesting tidbit that I would learn was this: Even though the descendants hadn’t known about the murder, they had always been told, “Never live in a small town.”

And that’s how I ended my revised chapter in Killing Albert Berch, mere days before the printer’s deadline. I never expected to hear from the Johnigans again.

Then, in May 2018, six months after the book’s release, I received an e-mail stating this: “I am the 2X great-grandson of Robert Johnigan.” Christopher Harris informed me that he had been given my book as a gift, something he now cherished as much as any gift he’d ever received. As the self-described family historian (based in Kansas City), he had an inkling about the murder, but Killing Albert Berch confirmed it, plus all the bizarre details that had been lost to history. Then, he said, in response to my chapter title, “Where Have You Gone, Robert Johnigan?”…We are still here!

Christopher (photo above) is a project manager for a global medical technology company, a family man with wife and two young daughters. His brother is a pastor, and he used his brother’s Facebook account to contact me. Christopher gave me a wealth of new information. For readers of the book who recall Robert’s wife Lizzie whose legal paper trail led 90 years later to the lost trial transcripts, it turns out that Lizzie was Robert’s second wife, and step-mother to the four children. Robert and Lizzie Shaw were married November 13, 1919 in Ardmore, Oklahoma (4 years before the murders). Robert’s first wife and the biologic mother to four children was Willie Boyd, who died of unknown causes. Only one of Robert’s four children (Josephine, age 7 at the time of the murders) had children, but there were many descendants after that.

Christopher then sent me family pictures (siblings and children of Robert) that are now on the book’s web site, although we still don’t have a picture of Robert or Lizzie (Christopher is searching). You can visit this new web page by clicking here:

All this occurred just a few weeks before my visit to Marlow, a town event held at the public library (photos above), where citizens were invited to hear me discuss the book. About 100 people gathered, and we talked for 2 hours. It was certainly a highlight of the post-publication signing/speaking schedule. And the new Johnigan information was of great interest. There were many questions that I was able to answer by virtue of information courtesy of 2X great grandson of Robert.

Many of the attendees had strong connections to the story – a Gandy descendant, a friend of Lula’s sister-in-law Aunt Mittie, a daughter of the Johnson Hotel Café owner who later bought another café where Marvin Kincannon drank beer daily after his release from prison, and on and on. My key Marlow resources were there as well – Debbe Ridley, D.B. Green, Janet Loveless, while the 4th resource – 96 year-old Dr. Jack Gregston – passed away shortly after the book’s release. Sheila Gregston, his widow, wrote me a nice thank-you note, letting me know that she had read him the book, finishing shortly before he died and that he had enjoyed it immensely.

It was an eerie feeling discussing the murder of my grandfather, knowing the powerful impact it had on our family, and with the location of the event only a few blocks away from where the murders had occurred. Still overwhelmed by the shock of the Johnigan family coming forward only two weeks earlier, with Christopher noting the book was his best gift ever, I have to admit that the surrealism of the moment choked me up a bit.

One loose end – if the identity of the self-liberated slave girl on page 263 ever comes to light, I will be able to say, “My work is done.” But now that I think about it, it’s already done. It’s just that the book keeps writing itself.