Shades of gray in gastrointestinal testing
The matter of methodology
William Check, PhD
In the field of gestalt psychology there is a famous drawing of a human
face that can be interpreted as a beautiful young woman or an old woman, depending
on the viewer’s perception. A similar perceptual discrepancy appears to exist
among physicians in their views of two laboratory tests being proposed as part
of the workup of inflammatory bowel disease-perinuclear antineutrophil cytoplasmic
antibody (pANCA) and anti-Saccharomyces cerevisiae antibodies (ASCA).
Although all physicians are looking at largely the same data, they interpret
the clinical utility of the two tests differently. It is not so much an issue
of whether the glass is half-empty or half-full, but more a question of whether
the glass holds a vintage Lafitte-Rothschild or a grapey Beaujolais nouveau.
Jonathan Braun, MD, chairman of pathology and laboratory medicine at UCLA, sees
a twofold value to the tests in the diagnostic evaluation of inflammatory bowel
disease, as both diagnostic and therapeutic aids. First, in screening, he says,
"For the primary care physician, they provide a relatively low-cost tool to
decide whether the individual is at low or high risk for IBD versus IBS [inflammatory
bowel syndrome], a common but poorly defined condition. Results on these tests
allow a primary care physician to make a more informed choice on whether to
manage by wait and see or whether that patient should go on to full workup and
management by a gastroenterologist."
He points out that the first option is low cost but runs the risk
of providing suboptimal care, while the second choice poses the
risk of overusing costly health care services.
A typical IBD patient has had symptoms for three years before
being biopsied endoscopically, Dr. Braun notes. "The value of serologic
testing is not to avoid endoscopic workup," he says, "but to help
decide which among many patients with abdominal complaints are at
higher likelihood of IBD and hence would be good candidates for
more extensive workup." If this approach works, it would allow clinicians
to bring appropriate patients to diagnosis earlier while sparing
the overall cost of managing patients with abdominal pain.
In the second application, Dr. Braun adds, "This information also
helps the gastroenterologist, because details of the serology can
guide choices for therapeutic management."
Stephan Targan, MD, professor of medicine and director of the
division of gastroenterology at UCLA and of the IBD Center at Cedars-Sinai
Hospital, expresses a similar view. "As a combination," Dr. Targan
says, "the tests as they are now run are quite specific for IBD.
I see them being used both as a screen and to assist in separation
of ulcerative colitis from Crohn’s disease." Ulcerative colitis
and Crohn’s disease are the two main subtypes of IBD, differing
in their histopathology, distribution in the gastrointestinal tract,
For the first application, and using a three-step approach (see
"The matter of methodology," page 62), "These tests provide an excellent
way of screening and allow you to rule out IBD" in most patients
with persistent GI symptoms, Dr. Targan says. "To my mind, they
provide a screening option in someone with persistent abdominal
pain or diarrhea," he says. "A primary care physician wouldn’t order
them the first time the patient comes to the office, but would in
someone with subacute or chronic complaints." In Dr. Targan’s view,
endoscopic biopsy is indicated for persons who test positive in
Data supporting this application are from a prospective study
that is being reviewed for publication, according to Dr. Targan.
For the second application, patterns of reactivity are "rather specific"
for Crohn’s disease or ulcerative colitis, Dr. Targan says. Patients
with UC are more likely to be pANCA-positive and ASCA-negative, while
the reverse holds for CD. "These patterns are particularly important
when a physician wants to differentiate the two conditions because
of emerging therapies, such as anti-TNF, that are more specific for
one form of IBD," Dr. Targan says. Anti-TNF (Remicade) is a therapeutic
monoclonal antibody against tumor necrosis factor-alpha (TNF).
Drs. Braun and Targan are the co-founders of Prometheus, a private
corporation that offers pANCA and ASCA testing on a reference basis.
Other physicians, both pathologists and gastroenterologists,
aren’t yet convinced about these applications. "I’m almost reluctant
to talk about it," says Henry Homburger, MD, director of the clinical
immunology laboratory at the Mayo Clinic. "I’m just not sure how
many clinicians are going to find this testing to be helpful. I
don’t think the predictive values are high enough either for one
form or another of IBD, or even as a sort of rule-out usage, where
if you get a negative result you think of something else. So I don’t
share the enthusiasm for pANCA or ASCA for these purposes."
Dr. Homburger notes that ulcerative colitis carries with it an
increased long-term risk of colon cancer. And Crohn’s disease, because
of its relapsing nature in some patients, can be an extremely serious,
even life-threatening, illness. "With diseases of this serious nature,"
he continues, "one is obligated to obtain a definitive result. And
at this time that means tissue biopsy diagnosis and/or conclusive
radiographic finding. I don’t think you make these diagnoses on
the basis of a test that has a predictive value of maybe 70 percent."
Dr. Homburger acknowledges that "By the time a patient is referred
to a gastroenterologist who is considering the differential diagnosis
between Crohn’s disease and ulcerative colitis, testing may have
a role. But certainly not at the screening end."
Charles Elson, MD, director of the division of gastroenterology
and hepatology at the University of Alabama, Birmingham, believes
the combination of pANCA and ASCA "may be useful as an adjunctive
test" to standard endoscopy and radiology. "But," he emphasizes,
"this is not a test-unfortunately it has been marketed this way-to
distinguish IBS from IBD." Even if specificity is 90 percent, Dr.
Elson points out, "When you apply it to a large group, false positives
will greatly outnumber true positives."
Although data have been published on the combined assays, in Dr.
Elson’s view, "The marketing seems to be beyond where the publications
are. The data that support those applications are not published,
except as abstracts," he notes. "They need to publish those results
so we can all share those data."
For distinguishing Crohn’s disease from ulcerative colitis, antibody
tests are not needed when symptoms are typical, Dr. Elson says.
For indeterminate cases, he believes, the issue is still unresolved.
"So I don’t routinely get these tests."
William Sandborn, MD, associate professor of medicine and head
of inflammatory bowel disease research at the Mayo Clinic, headed
a recent multicenter study on these two markers in IBD. He, too,
holds a limited view of their utility. "At this time, they would
play an adjunctive role to endoscopy and radiology in patients with
known IBD, where you are trying to distinguish between ulcerative
colitis and Crohn’s disease," he says. "In patients whom we now
call indeterminate colitis, this set of tests might be useful."
The question that has not been answered definitively, he says,
is, "in patients presenting with digestive symptoms who haven’t
been evaluated with endoscopic or radiological studies and who don’t
have known IBD, what are the sensitivity and specificity and positive
and negative predictive values of these tests in that group? That
has not been studied, at least in adults."
Accurately distinguishing IBS from IBD has become more important
with the approval of a drug, losatron, to treat IBS, Dr. Sandborn
points out. A set of symptom-based parameters called the Rome criteria
can help to make a diagnosis of IBS, but that is a diagnosis of
exclusion. Being able to distinguish IBD from IBS definitively on
the basis of laboratory tests without endoscopy would be attractive,
but has not yet been proven to be feasible, in Dr. Sandborn’s estimation.
Dr. Sandborn sums up: "In patients in whom you are having a hard time
figuring out what is going on, these tests add one more piece of evidence.
But they are adjunctive, not a replacement for endoscopy." He adds,
"Screening remains an open question."
Paul Rutgeerts, MD, a gastroenterologist
and professor of medicine at the Catholic University in Leuven,
Belgium, has been following this field for several years. He wrote
in a 1998 editorial, "It is unlikely that pANCA and ASCA measurement
will prove useful in the screening of diarrheal disease. However,
they would be very valuable if they enabled identification of indeterminate
colitis as CD or UC" (Gastroenterology. 1998;115:1007). Dr. Rutgeerts’
evaluation remains essentially unchanged.
"These serological markers are very interesting," he now says,
"but I don’t think they are ready for clinical use yet. Their sensitivity
and specificity need confirmation, although it seems that a combination
of ASCA and pANCA might be helpful in diagnosis of subtypes of IBD.
But they are certainly not useful for screening," he states, even
though "Sometimes this is already advocated by commercial firms."
Dr. Rutgeerts, too, sees the most likely application of the test
combination in patients with indeterminate colitis (IC), which occurs
in about 10 percent of IBD cases. "In these patients," he says,
"in whom the differential diagnosis [between ulcerative colitis
and Crohn’s disease] cannot be made with certainty, measurement
of serological antibodies might be suitable. But even this is not
really proven yet.
"We cannot rely on [this test combination] as a diagnostic tool,"
he concludes. "We need more data. It is not what you would use in
patients presenting with diarrhea. In these patients, we should
make a more intensive investigation, including tissue biopsy."
What are the data on which these evaluations are based? In Dr.
Rutgeerts’ 1998 editorial, he summarized several surveys by writing
that pANCA is positive in about 60 to 70 percent of patients with
ulcerative colitis and five to 10 percent of those with Crohn’s
disease, while ASCA is positive in 60 to 70 percent of CD patients,
10 to 15 percent of UC patients, and up to five percent of controls.
When results from the two tests are combined, the pattern of pANCA-positive/ASCA-negative
is 57 percent sensitive and 97 percent specific for UC, while a
pattern of pANCA-negative/ASCA-positive is 49 percent sensitive
and 97 percent specific for CD. Most CD patients who were positive
for pANCA "had a UC-like presentation," Dr. Rutgeerts wrote.
In Dr. Sandborn’s multicenter study, presented at this year’s
meeting of the American Gastroenterological Association, figures
based on assays done at Prometheus’ laboratory were slightly, but
not substantially, different. Most notably, specificity for UC was
16 percent lower (81 percent) and sensitivity for CD was 11 percent
lower (38 percent) than cited by Dr. Rutgeerts. Positive predictive
values of the relevant patterns were 86 percent for CD and 75 percent
for UC; negative predictive values were 61 percent for CD and 63
percent for UC. (Of course, predictive values vary with prevalence
in the population studied.)
In this trial, Dr. Sandborn and his colleagues were assessing the
validity of trying to differentiate UC from CD using pANCA and ASCA
testing when they already knew the patients had some form of IBD.
"I think that is the setting in which it will be useful," he comments.
"To know for certain that the patient has IBD presupposes that you
have done an endoscopy and perhaps a small bowel x-ray. So at this
time testing for serologic markers would play an adjunctive role."
Subjects in this survey were 296 patients who had been diagnosed
with inflammatory bowel disease in Olmsted County, Minnesota, between
1970 and 1993, so the evaluation of the serological tests was necessarily
retrospective. A prospective study of this sort would not be feasible,
Dr. Sandborn points out. Olmsted County has 125,000 persons and
the incidence for UC and CD each is in the range of seven per 100,000.
"So we will see maybe 15 new cases per year in Olmsted County. Nationally,
there are maybe 15,000 new cases of each per year. So a prospective
study would be a challenge," he says.
Turning to the question of differentiating ulcerative colitis
from Crohn’s disease for therapeutic purposes, Dr. Elson notes
there is a surgical procedure a surgeon would not perform on CD
patients but would do for UC. This operation is total colectomy
with ileo-anal anastomosis, in which the total colon is removed
except for a rim of rectal wall and the ileum is formed into a pouch
in the pelvis and connected to that rectal channel. With this extensive
two- to three-stage operation, patients stay continent and defecate
in the normal way.
Distinguishing accurately between CD and UC becomes important
in this situation because, while CD is mostly a small bowel disease,
in some instances it is limited to the colon. (UC is always limited
to the colon.) In these cases it is not possible to differentiate
CD and UC by standard means. "It would be very interesting to have
a serologic test for this," Dr. Rutgeerts says. "Certainly when
surgery is being considered, it is of the utmost importance to know
whether a patient has UC or CD." In UC, if you resect the colon,
you usually cure the patient. In CD, if you resect the colon, recurrence
elsewhere in the bowel is possible.
Another consideration when contemplating ileal pouch anal anastomosis
surgery, Dr. Targan says, is the minority of UC patients who develop
inflammation in the pouch, or "pouchitis." Over time, particularly
during the first 12 to 18 months after surgery, some patients get
diarrhea and urgency again. In contrast to the original symptoms,
this recurrence is probably due to hypersensitivity to bacterial
growth in the pouch and usually responds to antibiotics. Dr. Targan
says the level of pANCA in UC patients before surgery is predictive
of the risk of developing chronic pouchitis in the neo-rectum that
does not respond to antibiotics.
Finally, there is the question of whether serological testing
can predict response to anti-TNF therapy in CD patients. Dr. Braun
cites Prometheus data showing that, while about half of CD patients
respond at least short term to Remicade, of those CD patients who
are pANCA positive with high titers, hardly any respond to anti-TNF.
"This is a very costly drug," Dr. Braun notes, "so using the test
for this purpose can potentially save considerable money."
But, at this year’s meeting of the American Gastroenterological
Association, Dr. Rutgeerts presented results from a study of 85
patients receiving anti-TNF therapy for refractory or fistulizing
disease that arrived at a different conclusion: It was titled, "ASCA
and pANCA do not predict response to anti-TNF treatment."
"Crohn’s disease patients who are ASCA-negative and pANCA-positive
do seem to do worse after Remicade treatment," Dr. Rutgeerts told
CAPTODAY. But only eight patients fell into this group. "The true
predictive value of these serologic parameters has not been defined
yet," Dr. Rutgeerts concluded.
When assessing a new technology, including a laboratory test, one criterion is whether the technology
is restricted to one or only a few locations (and therefore not
proven in more general settings) or whether it is likely that the
results would be achieved at a variety of appropriate facilities.
Dr. Homburger notes that, until recently, ASCA was available only
from Prometheus, so no one has been able to evaluate it (or the
combination of ASCA and pANCA) in their own laboratory with their
own patient population. With the Inova kit for ASCA now available,
Dr. Homburger says, "Some large commercial laboratories may elect
to add it to their repertoire and let clinicians sort out the results.
Others-and we are in this second group-will want to know how it
performs in their own laboratories."
In the meantime, until definitive data become available, Dr. Homburger’s
advice to clinical pathologists is always to have a dialogue with
their clinicians. "When clinicians suggest that they have heard
that this may be useful, sit down with those individuals and explore
their rationale for coming to their conclusion," he says.
When clinicians persist in wanting a test that is not available
in-house, such as ASCA, in all likelihood pathologists will have
to send the test out, except for large volume institutions.
"I don’t think it will be economically worthwhile to do it in
one’s own laboratory," Dr. Homburger says. "So you need to see how
it performs at reference laboratories. But the clinical pathologist
is always in the middle between the clinician and whoever is providing
the test. Even if it is done in-house," he notes, "you need to find
the published outcomes and put them together with the expectations
of the clinicians to make sure there is not a major disconnect there."
William Check is a freelance medical writer in Wilmette, Ill.