6
The Lyme Times
Continued on page 26
NIH Clinical trials
What we learned from four famous trials funded by the National Institutes of Health
By Brian Fallon, MD, MPH
The groundbreaking clinical trials investigating post-treatment
Lyme disease syndrome (PTLDS) contributed to our knowledge
of tick-borne diseases for several reasons.
All the trials showed that a proportion of patients got better
with time (even without medication), i.e., the placebo response.
That’s good news for patients as it demonstrates that not everyone
needs to receive additional antibiotics to get better.
Response to treatment might not be demonstrated immediately
after treatment. In the Krupp trial, there was no difference in the
percentage of responders to IV ceftriaxone vs IV placebo imme-
diately after treatment; but when these same patients were as-
sessed five months later (off antibiotics), patients receiving IV
ceftriaxone were 3.5 times more likely to have meaningful im-
provement in fatigue compared to those who initially received IV
placebo. While antibiotic retreatment can be effective, the impact
may not be seen immediately. This finding was later demonstrated
again in a European study.
The Krupp trial suggested that a much greater proportion of pa-
tients respond to treatment if they are IgG Western blot positive
at the time. In this trial, the responder rate was 80% for drug-ran-
domized patients who were IgG Western blot positive vs 46% for
those who were not; in other words, patients are almost twice as
likely to respond to retreatment if their serologic tests are positive.
This finding highlights the importance of careful study design.
Had the Krupp trial only recruited individuals who were sero-
negative, that study may not have shown the positive treatment
effect on the primary outcome measure of fatigue observed at six
months.
While some have argued that the beneficial response in the
Krupp trial should be disregarded, our Lyme encephalopathy
clinical trial (analyzed in the same fashion as the Krupp trial)
found nearly identical benefits on the fatigue measure at six
months favoring drug over placebo. Finding a nearly identical
treatment effect from an independent study conducted at a dif-
ferent university is compelling evidence in support of the validity
of the original Krupp results.
Thankfully, the NIH funded these four clinical trials, as without
them much less would be known. The NIH should be encouraged
to support new clinical trials, however, to draw firm and clear con-
clusions about the beneficial effects of antibiotic therapy. It is pre-
mature to draw conclusions from such a small sample of Lyme
patients (37 in the Fallon trial, 55 in Krupp, 51 in Klempner’s se-
ronegative trial, and 78 in Klempner’s seropositive trial).
Significance of study design
To get a meaningful outcome, it is important to en roll indi-
viduals with significant impairment – those who meet a severity
cutoff. Too often this key design feature gets violated. For example,
the Krupp trial recruited patients who met a severity cutoff on
fatigue – and convincingly showed a benefit to drug over placebo.
For the Columbia Encephalopathy trial, only patients who met a
severity cutoff on cognition were allowed to enroll. While we did
not see a sustained improvement in cognition to six months for
drug vs placebo, we did see a significantly greater improvement
at 12 weeks.
A preset severity cutoff ensures that a true treatment effect
can be observed. Simply stating that all who entered the trial re-
ported functional impairment is not the same as saying that all
who entered the trial scored at an “impaired level” on the primary
outcome measure of interest. Understandably researchers face
a huge push to enroll as many people as possible as quickly as
possible because such trials cost a lot of money. For the sake of
drawing meaningful conclusions, however, impairment must
meet a predefined standard.
Among the four NIH trials, the two Klempner studies had the
largest enrollment, (a strong positive) but they were the only trials
without a “severity cutoff ” for enrollment (a striking negative). Al-
though these trials showed no drug-placebo difference, we don’t
know what proportion of enrollees would have met a meaningful
cutoff for impairment at the start of the study. Future trials must
include a severity threshold to maximize the likelihood that their
hypothesis is adequately tested.
Columbia encephalopathy trial
We examined whether certain laboratory biomarkers might
predict good or poor outcomes, including the CD 57 marker,
inflammatory markers (CRP, ESR), and presence of antibodies
against
Babesia
or
Bartonella
. Unfortunately, none of these
markers proved helpful in predicting treatment response. Inter-
estingly, the only markers associated with treatment response were
clinical markers — not lab test markers. In our trial, we found
that the physical exam was one of the best predictors of treatment
Brian Fallon and a team of researchers at Columbia University Medical
Center conducted one of the four famous NIH clinical trials in 2003:
From left: Dexterrie Ramirez, Kathy Corbera, MD, Megan Romano, and
Iordan Slavov, PhD.