[ed. note: Michael Jorrin, who I dubbed “Doc Gumshoe” years ago, writes for us a couple times a month on medicine and health issues. He is not a physician, but explains the science and the data well, and helps provide some background understanding and perspective that often benefits both investors and folks who are attentive to their own health. He doesn’t focus on investment ideas, but follows our trading rules when investments are featured in his articles. You can see his past articles and comments here.]
What made me think about those fundamentally different procedures was Travis’s piece about the “Biggest medical breakthrough of the decade.” In case you missed that entry, this turns out to be the Trovagene (TROV) urine test that detects the specific DNA of cancer cells. From the standpoint of the patient, nothing is simpler than a urine test. From the standpoint of the technology that would identify the DNA, that screening procedure, although simple, could lead to a diagnostic process that was in fact highly complex and required the most advanced technology. Doc Gumshoe is not a stock analyst, and I would not dream of assessing the likelihood that Trovagene stock would zoom and make a bundle.
At this point, only one clinical trial seems in any way to be connected with Trovagene, and that one is an observational study looking at DNA mutations in patients already diagnosed with lung or colorectal cancer. That is probably not the cancer screening test touted as the “biggest medical breakthrough of the decade.” I pointed out in a comment on that post that identifying the cancer cell DNA would not likely point to the location of the cancer, nor the stage. To arrive at a diagnosis would likely require imaging, and perhaps biopsy or exploratory surgery. The urine test is merely a screening procedure. If it’s positive, what’s the next step?
That’s always the question with screening. When we go for a routine physical, many of the things the health-care provider/doctor carries out are screening procedures. “Screening” implies that just about every patient, or every patient in a certain group, will undergo that procedure. We get weighed and measured, and the doctor calculates our body-mass index, or BMI. That’s a screening procedure, because there’s an assumption that a BMI over a certain value predisposes us to some health problems. But it’s not a diagnosis of those health problems. Our heart rate and blood pressure are measured. We have an electrocardiogram (EKG). A blood test provides information on a great many important parameters, including lipids, blood glucose, and liver functions. Our blood-oxygen level is assayed. And those are only a few of the screening procedures we go through.
None of those procedures are diagnostic. A diagnosis presumes that the patient presents with symptoms or other indications of a disease or disorder, perhaps discovered by means of screening. The purpose of the diagnosis is to determine the cause of these indications, and – we hope! – to point the way to a treatment that cures or at least manages the underlying illness. Screening alerts the clinician that further investigation may be required.
There is a certain amount of ambiguity here, of course. For example, a blood pressure test lets the clinician know what our blood pressure is – at that moment, at least. Typically, the clinician is not going to arrive at a diagnosis of hypertension based on a single elevated blood pressure reading. Blood pressure evaluations are screenings because everybody gets them, and they could be considered as diagnoses because chronically elevated blood pressure is indeed a pathology. But a single markedly elevated BP reading would call for repeated testing before the patient is evaluated as “hypertensive.” Thus, the initial test is a screening and the follow-up might be thought of as part of the diagnosis.
Screening procedures, especially if they are fairly quick and inexpensive, are highly favored by that contingent in the health-care community that focuses primarily not on individual patient care but on measures that benefit the population as a whole. This plays out in lots of ways, big and little. Advanced technology can tell us whether we have exercised enough, and it can monitor our blood sugar and our heart rate, and no doubt will soon be able to keep track of a number of other vital functions, and perhaps transmit them in real time to an all-knowing automated monitor that translates these observations into health guidance, or into something approximating a diagnosis. The techies like that, and the health insurance companies favor quick, easy, low-cost devices that they hope will keep you out of the doctor’s office or the hospital – for your own good, of course, but for their own benefit as well.
Several recent findings and developments are related to this topic. One that I have referred to before has to do with the relevance of the BMI to overall health.
How useful is the body-mass index?
The particular findings that prompted me to revisit this subject was an analysis published in JAMA this past month, which found that the patient cohort with the lowest mortality had a BMI squarely in the middle of the “overweight” category. This is contrary to what continues to be the advice of many people that dole out health wisdom. What’s particularly interesting is that the BMI values associated with the lowest mortality have actually gone up over time – whereas, back in the mid 1970s the lowest mortality was associated with BMI values that are labeled “normal,” in the decade from 2003 to 2013 those BMI values went by several notches to the range labeled “overweight.”
The analysis compared three studies done in Denmark, in three different cohorts, and during three different time spans. The first cohort consisted of 13,704 persons were enrolled between 1976 and 1978. The second cohort, 9,482 persons, were enrolled between 1991 and 1994. And the third cohort, 97,362 persons, were enrolled between 2003 and 2013. All subjects were followed until November 2014, emigration, or death.
The BMI values associated with the lowest all-cause and cardiovascular mortality were those of the third cohort: 27.0 for all-cause mortality, and 26.4 for cardiovascular mortality. The BMI values associated with lowest mortality increased by 3.3 points from the earliest-enrolled subjects to the most recently-enrolled. The JAMA study did not provide a reason why that increase in the optimal BMI should have taken place. Speculation is that changes in risk factors or in the standards of cardiac care between the 1970s and the first decade of the 21st century might have affected persons with differing physical characteristics in different ways. It should also be noted that the third cohort, enrolled between 2003 and 2013, was much larger than the previous two cohorts.
The authors pointed out that the mortality curve followed what they called the “U-curve;” in previous Doc Gumshoe pieces I have referred to it as the “J-curve.” The meaning of the term is that mortality drops to a low at a certain point on the curve and then rises again. Thus, when BMI rises from the 27.0 value that corresponds with the lowest all-cause mortality, mortality increases proportionately, so that the mildly obese and some normal-weight cohorts may have similar mortality rates, and the morbidly obese and underweight cohorts may also have similar mortality rates.
Similar “U” or “J” curves describe other associations, for example, the association between alcohol consumption and life expectancy. In that case, the best life- expectancy data was associated with persons who consumed between one and two alcoholic beverages per day; when alcohol consumption increased beyond that low point in the curve, life expectancy decreased slightly, such that life expectancy in the four drinks per day cohort was about equal to that in teetotalers, and as consumption increased beyond four drinks per day, life expectancy decreased further.
The flaw in the usual BMI guidance was no surprise to Doc Gumshoe, who wrote the following in a blog posted back in July, 2013:
In case you’ve forgotten, the official BMI ranges are: below 18.5, underweight; 18.4 – 24.9, normal; 25.0 – 29.9, overweight; over 30.0, obese. Taking yours truly as an example, I am 6’ 3” and I weigh about 217 pounds, so my BMI is 27.1 – in the middle of the overweight category. If I weighed 180 pounds, my BMI would be 22.5 – normal. To be underweight, I would have to weigh about 140 pounds. I haven’t been that skinny since I was a sophomore in high school, and I was really, really skinny then.
However, there’s pretty convincing evidence that those BMI ranges are flat wrong. An analysis of 40 studies looked at BMI and calculated the risk for both overall mortality and cardiovascular mortality, compared with normal weight subjects, which for this study was defined as BMI 20 – 24.9. No surprise – very obese individuals, those with a BMI over 35, were at very high mortality risk – about 10% higher total mortality, and 88% higher cardiovascular mortality. But persons with BMI values below 20 were also at very high mortality risk – 37% higher total mortality (higher than the very obese!) and 45% higher cardiovascular mortality.
What about the overweight and the obese? The merely overweight were at the lowest risk for both total (13% lower) and cardiovascular mortality (12% lower). And the obese in the 30 – 34.9 BMI range were essentially at the same risk as the normal weight subjects.
So, based on that study, which, by the way, came out of the Mayo Clinic, and included data on more than a quarter of a million subjects, the healthiest weight range is what the BMI classifies as “overweight.” The authors of the study suggested that it might be time to scrap the BMI as a risk factor. Applying the Goldilocks Principle, a bit overweight might be Just Right.
The BMI is a crude screening tool, and maybe, if adjusted, it might be useful. But although the defects have long been acknowledged, the advocates don’t seem to be the least willing to go back and revise the guidelines.
Celiac disease: whether or not to screen?
The U. S. Preventive Services Task Force (USPSTF) said, a couple of weeks ago, that the evidence did not support screening for celiac disease. This was somewhat surprising, because, in contrast with other screening procedures, the tests for celiac disease are fairly accurate.
The most commonly used test for celiac disease is the tissue transglutaminase antibody test (tTG-IgA). This test is both highly sensitive and highly specific. The sensitivity is rated at about 98%, meaning that it will miss only about 2% of persons who have celiac disease. In other words, the false negative rate for the tTG-IgA test is only 2%. And the specificity is rated at 95%, meaning that at most about 5% of people without celiac disease will test positive – a 5% false positive rate.
Some other tests are used to test for celiac disease, including a test for endomysial antibodies (EMA), for immunoglobulin A (IgA), and for antibodies to deaminated gliadin peptide. These tests would likely be used only in individuals in whom there was a high level of suspicion of celiac disease, but who had tested negative on the tTG-IgA test.
The USPSTF draft recommendation regarding celiac disease screening applies to adults, adolescents, and children who do not have signs or symptoms of celiac disease. The recommendation states:
“The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of screening for celiac disease in asymptomatic persons.”
As I read the USPSTF non-recommendation, which is written with the utmost care, it appears that the bases for their conclusion are two-fold: one, that the prevalence of celiac disease is very low, and, two, that in persons who do not have the symptoms of celiac disease, the sensitivity and specificity of celiac screening that does not include the examination of tissue samples is not perfect. In the US, prevalence of celiac disease in adults ranges from about 0.40% to 0.95%; thus, celiac screening would employ a serologic test with a 2% false negative rate and a 5% false positive rate to a population with at most a 1% prevalence of the target disease. In other words, you need a finer net to catch fish that uncommon.
The USPSTF is careful to avoid playing up the potential “harms” of celiac screening, although they do mention the adverse “life-style” consequences of a false positive, which might persuade non-celiac individuals to adopt the gluten-free diet. Celiac screening is recommended by the American College of Gastroenterology for asymptomatic persons with a first-degree relative with a diagnosis of celiac disease, or for those with type 1 diabetes (T1DM) with suggestive signs or symptoms. Children with T1DM, first-degree relatives with celiac disease, and several other conditions such as autoimmune thyroid disease, Turner syndrome, and Down syndrome should similarly be tested.
My guess is that at least part of the USPSTF’s motivation in issuing this non-recommendation is based on economics. It would cost a lot of money if widespread screening for celiac disease became the norm, and many of the task force’s members are economically motivated. However, if we consider those people out there who think they might have celiac disease, and have consequently adopted a gluten-free diet, with attendant increased costs and inconvenience, it might be useful for them to have this celiac screening test, and have it confirmed that they do not have celiac disease.
There is, of course, much discussion out there about non-celiac gluten sensitivity, and the USPSTF does no more than mention it, avoiding any clear statement of any kind. But it is surely the case that persons with digestive symptoms similar to those in celiac disease – malabsorption, diarrhea, pain, and weight loss – may have other pathologies at work in their system. It would certainly benefit them to rule out celiac disease, and then for a physician to move on to determine the actual cause of their digestive distress. Ruling out potential etiologies is an important part of the diagnostic process.
So, while Doc Gumshoe agrees with the USPSTF that universal celiac screening is unnecessary, screening should be available whenever there’s a bona fide reason.
Let’s take a look at another USPSTF pronouncement and see where it has led.
What happened after the USPSTF’s anti-recommendation of PSA screening?
In comparison with the USPSTF’s stand-offish position regarding celiac screening, their position on widespread screening for prostate-specific antigen (PSA) is uncompromising opposition. Reduced to essentials, the basis for the recommendation not to perform PSA screening as a routine part of a physical examination was that in most men in whom prostate cancer was discovered as a result of this test, the cancer would grow so slowly that the men would die of something else first. The usual way this was stated is that men died with prostate cancer, not of prostate cancer. The USPSTF’s circumspect way of justifying the anti-recommendation was to state that the benefits of PSA screening did not outweigh the potential harms, which included everything from psychological anxiety if the test came back positive, risk of infection associated with biopsy, and adverse effects from treatment such as incontinence and poor sexual performance.
A quick look at prostate cancer statistics would likely dim the rosy picture painted by the “with – not of – prostate cancer” scenario. Prostate cancer causes more than 26,000 deaths in the US every year, and, according to the American Cancer Society, kills one man in 39. Although the five-year survival rate for prostate cancer overall is nearly 100%, it drops to less than 30% if the cancer metastasizes to distant sites. Not such an optimistic picture. And how would one detect and treat prostate cancer without PSA screening? You tell me.
All the same, the USPSTF has persisted in pushing their anti-recommendation stance. In fact, late last year they initiated a proposal to the Centers for Medicare and Medicaid (CMS) that would have penalized physicians for performing “non-recommended” PSA screening. Eminent oncologists and urologists protested vehemently, and this spring CMS walked it back a little way, while still stating that measures to ensure appropriate use of the PSA test were still being investigated.
The real motivation for this stuff, in my jaundiced view, is economic. It’s not the cost of the PSA assay itself that’s of concern. As part of a routine blood test, the cost of the PSA test is trivial, and kits for testing PSA levels at home are available for around $50. What concerns observers who are focused on health-care costs in the economy is what happens after a positive result on the PSA test. And, by the way, authorities are not in total agreement on what constitutes a “positive result” on that test. It’s generally agreed that a level of less than 1 ng/mL is not a positive result, and, depending on age, might suggest another PSA test in a year or two. A level of 2.5 or 3.0 ng/mL (depending on which authority is consulted) might mean that discussion with the physician and follow-up is recommended.
And that’s where the health-care economists start to worry. Lots of men will have PSA levels higher than that cut-point, and not just because they have incipient prostate cancers. Furthermore, lots of men will actually have incipient prostate cancers – about one in seven, according to the National Cancer Institute. So, having had the PSA test, and having learned that there is a possibility that they have prostate cancer, many men want to proceed to the next step, which will likely be a needle biopsy of the prostate. At this point, it starts to get a bit more expensive.
One of the factors on which the more economically-focused in the health-care community zero in is the “C-word scare factor.” If the results of the biopsy indicate that, yes indeed, there are cancer cells lurking in the prostate gland, a common – and completely understandable – reaction in many men is, “I want it OUT!!!” That reaction persists even in cases where, based on the results of the biopsy as well as the PSA level, the risk that the cancer will progress to the stage where it poses a real threat to life and health is low. For example, the data suggest that in a man aged around 70 with a Gleason score of 6 or less and a PSA level of 10 or less, the risk is fairly low. Nonetheless, the C-word scare factor plays a part, and many such men opt for treatment. This is an example of what the USPSTF refers to as “unnecessary treatment.”
But low risk is not the same as no risk. As I said a bit further up in this piece, the American Cancer Society estimates that one in 39 men in the US dies of prostate cancer. The ACS also estimates that one in seven men in the US will be diagnosed with prostate cancer. So, out of 100 men, about 14 will be diagnosed with prostate cancer, and between two and three of those unfortunate individuals will die of that disease. Yes, perhaps based on scrupulous evaluation of the biopsy results, a decision not to treat some of those 14 men might be justified, and treatment might be reserved for those men in whom the cancer is considered to be most threatening. However, I fail to see how a decision of this life-or-death importance could be arrived at without that initial PSA screening.
Some “experts” have even suggested simply avoiding the C-word. Call this usually indolent and seldom fatal disease by another name, they say, and fewer men will opt for unnecessary treatment. What a dandy way to avoid “unnecessary treatment!” The doctor deceives you, you go home soothed, and when the undertaker carts you off, the death certificate lists “natural causes.” Was this suggestion perhaps imported from North Korea?
A more honest and perhaps more productive strategy than “let’s just not call it cancer” is what is now termed “active surveillance.” That means, in men who have received a positive PSA reading, a range of interventions including follow-up PSA tests and in some cases annual biopsies. Active surveillance is not at all the same thing as “watchful waiting,” which implies a specific decision to forego definitive treatment and, instead, to resort to palliative treatment in the event that metastasis of some form occurs. Palliative treatment can include such options as hormonal therapy or some forms of radiation therapy – interventions that are essentially not curative. The watchful waiting option assumes that a cancer has been found in the prostate, but that the patient will not benefit from curative treatment at that point.
The active surveillance option assumes that a reasonably accurate distinction can be made between men who will benefit from curative treatment such as radical prostatectomy and those who will probably not benefit from such treatment. This means identifying men with localized cancers that are likely to progress, based on such factors as the rate of change in PSA levels and Gleason scores. Here’s what the American Urological Association (AUA) says:
“… men considered appropriate for active surveillance are those with lower grade, localized prostate cancer who have a lower risk for clinical progression within the first 10 to 15 years after the diagnosis. Thus, this treatment strategy may be best suited for men with a shorter life expectancy.”
The AUA goes on to say that some men with a longer life expectancy may opt for active surveillance if they are reluctant to accept the potential side effects of curative therapy. But, they add:
“This can be a difficult decision since signs of progression must be identified before the cancer evolves to a stage (or grade) where therapy is no longer curative. Currently, providing evidence-based recommendations for when to intervene in patients with a long life expectance are not possible since markers of disease progression are poorly validated.”
The USPSTF in part based their anti-recommendation on the European Randomized Study of Screening for Prostate Cancer (ERSPC), which has followed 162,388 men over 13 years. The USPFTS did not deny that in some cases, the PSA test led to diagnosis and treatment of prostate cancer, resulting in saving some men’s lives. Their objection, boiled down to essentials, is that it didn’t save enough lives to justify the costs, both in terms of adverse effects and the economic costs.
The more recent news is that, as the observation period has continued, greater absolute benefit from PSA screening has emerged. Specifically, after 7 years, 41 cases of prostate cancer had to be identified to save one life. But the more recent report found that after 13 years of follow-up, one life could be saved when 27 cases were identified.
In my previous Doc Gumshoe piece, I pointed out to a huge defect in this study, and I’ll repeat it now: it was supposed to compare two cohorts – one that had PSA tests and one that did not have PSA tests…. but almost half of the men in the group that wasn’t supposed to have PSA tests went ahead and had the tests anyway, which presumably also lowered the mortality in the control group and reduced the difference between the two groups.
I think the USPSTF picked the ERSPC because it was the study that most closely supported their view. Most other studies at that point had strongly endorsed PSA testing. For example, a study from the Duke University Prostate Center underscored the very strong association between a man’s initial PSA score and his risk of death from prostate cancer. If a man’s first PSA score is between 4.0 and 9.9, his risk of death from prostate cancer is three times higher than the risk in a man whose first PSA score was less than 2.5. And if the first PSA score was 10.0 or higher, the risk of death is eleven times higher. This study points to the desirability of starting to track the PSA score earlier in life, so that men with rapidly rising PSA scores can be identified. One fundamental anti-PSA-testing argument is that it is difficult to distinguish aggressive from indolent prostate cancer, but the association between the first PSA score and prostate cancer death strongly suggests that rapid increases in the PSA point to aggressive cancers.
Circling back to our principal topic …
Screening is not the same thing as diagnosis, although some screening procedures go a good way to establishing a diagnosis. For example, a fasting blood glucose level of 126 mg/dL on two consecutive tests is considered to be diagnostic of diabetes. In that case, the screening test and the diagnostic test are one and the same, although responsible clinicians would certainly want to employ further diagnostic strategies, like the hemoglobin A1c assay, which calculates blood glucose over a period of several weeks, in order to make sure that the results were reliable before initiating treatment. On the other end of the reliability scale, the results of weighing and measuring the height of a person in order to calculate BMI don’t come within leagues of constituting a diagnosis of any disease or condition, especially since at this point most docs are aware that the correlation between BMI and general health is skewed. For most people in the “overweight” category, the most a physician would be likely to say is, “You might think about losing a few pounds.”
But population-wide screening does have real implications. Some employer-funded health insurance plans strongly encourage covered employees to participate in “health-maintenance” programs. These programs collect data on such matters as exercise, diet, and – yes! – BMI. Based on those data, some employees may be eligible for discounts on their share of the insurance premiums. And the health-insurance community has so far not quite twigged to the serious flaws in BMI as a predictor of general health.
Blood pressure measurement is an almost universal screening procedure. It is cheap, quick, non-invasive, and certainly does provide essential information. But blood pressure is greatly variable. Anecdote: I used to drive into New York City for my checkup with my former excellent PCP (now retired). I would park my car on the street and dash into his office, where immediately his excellent nurse, bossy Lena, would command me to strip, weigh me, and take my BP. She would raise her eyebrows. Elevated! Then, after a bit, my doctor would come in and examine me, calmly and leisurely, and discuss the results of the tests I had undergone. Then he would take my BP again. He would smile. Normal! Lena’s initial BP test was the screening. It did what screening is supposed to do – tell my PCP that another BP test was indicated.
In the case of celiac screening, there is uncertainty as to what the test results might mean. The expectation is that in the huge majority of cases, the result would be negative. And even in that small percentage of positive results, doubt might still persist. What to do in those cases? It is those doubts that led the USPSTF essentially to refuse to render a verdict – neither recommend screening, nor, as in the PSA test, recommend against screening. But some persons, on an individual basis, might want the celiac screen, and in those cases it should be available.
Population-wide screening is an extremely important public health measure. Even with widely-available BP tests, millions of people are not aware that they are chronically hypertensive. This is also true of diabetes and elevated cholesterol. Screening for those conditions is valuable, and it is unambiguous: a positive result points clearly to the next step. That should be one of the criteria for a screening procedure. A screening test should not leave the physician wondering what to do next. And it should not leave the patient in a quandary.
* * * * * * * *
I had promised to do one more blog about treatment of viral diseases, especially about the great progress in treating hepatitis C. But in the sleuthing process, I found myself engulfed in the viral hepatitis spectrum in general. I know a tiny fraction of what Dr KSS knows about this treatment area, and I want to avoid egregious blunders, but I do think a wider look might be valuable. In addition, there are hints that broad-spectrum antiviral drugs might be on the horizon, and I’ll squint at those. Best to all, Michael Jorrin (aka Doc Gumshoe).