Assignment Brief
Excelsior College PBH 321
Page 1
PRO MISING NEW D RUG FOR ARTH RITIS P AIN – OR SAME OLD STORY?
Use this commentary as an example for how you are expected to respond to questions critiquing the article.
Side notes from the instructor to you, which are simply informational, are in [* ].
Q: What were the study’s rationale/hypothesis and objectives?
This study aimed to test the safety, efficacy, and possible side effects of a new drug for treatment of
osteoarthritis of the knee. Given the limited existing treatment options for the condition, and a small Phase 1
trial which suggested possible benefits of tanezumab, the investigators aimed to compare treatment of
arthritis pain with different injected doses of the drug to treatment with a placebo injection.
Q: What were the assigned treatment and control groups, and how were they defined?
Individuals were randomly allocated to treatment with either 10, 25, 50, 100 or 200 ug of tanezumab
via injection, or a placebo (control group), also injected.
Q: Was there any way that the blinding of treatment status could have been compromised (in other words,
if the investigator somehow figures out treatment status)? If so, what might have been its potential impact?
The pharmacist preparing the dosage (either the drug or placebo) was aware of the subject’s
randomization to treatment or placebo. It is unlikely the pharmacist would have (accidentally or purposefully)
changed someone’s treatment from their randomly assigned one, and since the pharmacist presumably played
no part in the analysis of study data, their awareness of treatment status did not likely affect the study’s
findings. Similarly, the statistician’s knowledge would not be expected to have an impact, barring unethical
research practices from this individual. An opportunity for un-blinding may have been presented if individuals
assigned to placebo experienced significantly more knee pain, or if individuals assigned to tanezumab
experienced many side effects – but this is really only likely to occur when the investigator ascertains the
outcome directly (such as by interviewing or examining the patient). Because participants self-reported all of
their outcome information (i.e., instead of being directly assessed by a physician) throughout the study, and
this information was summarized later by the investigators, compromise of the investigator’s blinding to
subject treatment status was unlikely and probably did not constitute a major source of bias in this study.
Excelsior College PBH 321
Page 2
Q: What was the outcome of interest for this study? How was it measured? Do you see any problems with
the way the outcome of interest was measured? If so, suggest some alternatives.
Pain while walking and overall knee pain was recorded by the patient in a daily diary. The patient’s
overall assessment of their pain during the trial (called “global ass ...
1. Assignment Brief
Excelsior College PBH 321
Page 1
PRO MISING NEW D RUG FOR ARTH RITIS P AIN – OR
SAME OLD STORY?
Use this commentary as an example for how you are expected to
respond to questions critiquing the article.
Side notes from the instructor to you, which are simply
informational, are in [* ].
Q: What were the study’s rationale/hypothesis and objectives?
This study aimed to test the safety, efficacy, and possible side
effects of a new drug for treatment of
osteoarthritis of the knee. Given the limited existing treatment
options for the condition, and a small Phase 1
trial which suggested possible benefits of tanezumab, the
investigators aimed to compare treatment of
2. arthritis pain with different injected doses of the drug to
treatment with a placebo injection.
Q: What were the assigned treatment and control groups, and
how were they defined?
Individuals were randomly allocated to treatment with either 10,
25, 50, 100 or 200 ug of tanezumab
via injection, or a placebo (control group), also injected.
Q: Was there any way that the blinding of treatment status could
have been compromised (in other words,
if the investigator somehow figures out treatment status)? If so,
what might have been its potential impact?
The pharmacist preparing the dosage (either the drug or
placebo) was aware of the subject’s
randomization to treatment or placebo. It is unlikely the
pharmacist would have (accidentally or purposefully)
changed someone’s treatment from their randomly assigned one,
and since the pharmacist presumably played
no part in the analysis of study data, their awareness of
treatment status did not likely affect the study’s
findings. Similarly, the statistician’s knowledge would not be
expected to have an impact, barring unethical
research practices from this individual. An opportunity for un-
3. blinding may have been presented if individuals
assigned to placebo experienced significantly more knee pain,
or if individuals assigned to tanezumab
experienced many side effects – but this is really only likely to
occur when the investigator ascertains the
outcome directly (such as by interviewing or examining the
patient). Because participants self-reported all of
their outcome information (i.e., instead of being directly
assessed by a physician) throughout the study, and
this information was summarized later by the investigators,
compromise of the investigator’s blinding to
subject treatment status was unlikely and probably did not
constitute a major source of bias in this study.
Excelsior College PBH 321
Page 2
Q: What was the outcome of interest for this study? How was it
measured? Do you see any problems with
the way the outcome of interest was measured? If so, suggest
some alternatives.
Pain while walking and overall knee pain was recorded by the
4. patient in a daily diary. The patient’s
overall assessment of their pain during the trial (called “global
assessment” by the authors) was recorded at
the time of 7 study visits (on days 14, 28, 70, 84, 112, 136, and
182). Adverse events were evaluated by
telephone at week 6. Since the outcome data was provided
directly by the participant (i.e., self-reported
discomfort), and categorized afterward by the investigator based
on previously established clinical guidelines
(OMERACT-OARSI), we should not be overly concerned about
possible investigator influence on the way
outcome was measured. The way participant’s report pain scales
on a daily basis may change over time, as
participants tolerance for pain may change over time, regardless
of treatment status. However, it is hard to
imagine a more objective way to measure pain, as pain is a
rather subjective endpoint. [Side note: one reason
that scores and scales are used to evaluate subjective outcomes
like pain is because they can better deal with
the variation in pain tolerance and other differences in how
individuals might characterize pain.]
Q: Are there any possible concerns about patient compliance
with their assigned treatment? What impact
5. might non-compliance have on study findings?
Data were analyzed using a modified intent-to-treat approach,
despite some individuals being lost to
follow-up. We can see from Figure 1 that many patients chose
to enter the open-label extension of the study
(offered at day 112), [*this is just a continuation of the existing
study that examines long-term safety of the
drug], and that some had an adverse event, did not respond to
contact, or discontinued the study for some
other reason. However, the modified intent-to-treat analysis still
included the majority of the individuals
randomized to treatment or placebo [*it is called a “modified”
intent-to-treat because it doesn’t include
everyone regardless of compliance, as we would do in a “true”
intent-to-treat analysis].
Since we know there were quite a few people who discontinued
the study for some reason, we have to
be concerned about the potential impact of non-compliance on
study findings. Non-compliance reduces the
benefit of randomization – making our treated and placebo
groups similar with respect to confounding factors.
Table 1 shows that randomization was initially (i.e., at the start
of the study) effective at distributing those
factors equally between treatment groups. If we compare the
6. proportion of participants having no response or
no contact (i.e., lost to follow-up) to the numbers analyzed in
the intent-to-treat approach, the major concern
is the loss of this benefit from randomization. [For example,
Figure 1: of the 75 individuals assigned to placebo,
Excelsior College PBH 321
Page 3
41 (54.6%) were lost by the study’s end.} We also don’t know
to what extent people who remained in the study
actually complied with their assigned treatment regimen.
The authors collected data on “rescue” treatments in order to
determine if use of rescue medication
also declined over time. However, we might also consider the
possibility that taking acetaminophen or
tramadol may have impacted the efficacy of the tanezumab
treatment. The authors do not discuss this possible
influence of rescue medications on the study findings.
Q: Why do you think there was an entire paragraph dedicated to
disclosure of who conducted, oversaw, and
analyzed the study’s findings? What implications does this
7. information have on the study?
In order for research to be ethical, the persons funding
research needs to be fully disclosed, or else there
is the possibility for a conflict of interest. The pharmaceutical
industry is a multi-billion dollar industry. Knowing
the Pfizer conducted the study does not imply there was
anything wrong with regard to the study’s conduct,
but it does demonstrate the likelihood of a financial interest in
the study’s findings. To avoid conflicts of
interest, safety and operations were monitored by an external
(i.e., not Pfizer) monitoring board.
Q: To what population(s) do you think this study’s findings may
or may not be generalizable, and why?
Since this study was conducted among individuals aged 40 to 75
years old with existing knee pain, this
same population is the group to which study findings are most
easily applied. [*In other words, we may have
some confidence that the findings would apply on average to
other people of the same age with the same
problem.] Notably, the drug is intended for arthritis pain,
specifically. The findings of this study don’t
necessarily apply to individuals with other types of pain, since
the mechanism for how these drugs work may
8. differ in various parts of the body. Therefore we also cannot
know for sure if the drug would show the same
benefit among younger people with knee pain (i.e., younger than
40 years), perhaps pain that is not associated
with arthritis.
Q. This trial may have some relevance to clinical practice, but
there were adverse events associated with
some of the treatment levels. Based on the frequency and type
of adverse events described in the article,
what advice might you want to provide physicians regarding the
use of tanezumab for osteoarthritis pain?
At the very end of the paper we learn that there were enough
severe adverse events in a separate study
to at least temporarily put a hold on further examination of
tanezumab, specifically the development of
arthritis in other joints. This is obviously of some concern. The
concept of equipoise states that in order for us to
Excelsior College PBH 321
Page 4
experimentally test a new drug on a human population, we must
be certain that it is no less effective than the
9. current standard of care (normal treatments), and we have a
reason to believe it might be of benefit. Certainly
the earlier, smaller trials suggested some benefit, so conducting
the trial was appropriate. However, at least
one other study suggested that there may be more than the
acceptable level of harm coming from treatment
with this drug. The adverse events occurring in the current
study appear to have been concentrated among the
higher dose treatment groups, but the authors admit they were
unable to fully evaluate that possibility.
Additionally, the adverse events were mild in most cases, but
since this was a short-term study, the risk and
severity of adverse events with long-term treatment remain
unknown. My advice to physicians at this point
might be to continue to prescribe standard treatments but to be
aware that additional, more long-term studies
of tanezumab proving benefit with minimal risks may make it a
possible future treatment option.
Promising New Drug For Arthritis Pain – or Same Old Story?
Excelsior College PBH 321
Page 1
10. CRITICAL EYE ON RESEARCH IN EPIDEMIOLOGY
By the end of this activity, you will be able to illustrate the pur
poses, designs, weaknesses, and relevance of a
randomized drug trial and of a cross‐sectional study.
In this module, you will learn to analyze a research article on an
epidemiological topic. It is understood that you may
have minimal background in some of the medical terminology a
nd statistical methods described in the paper. However,
comprehending these components fully is not critical to evaluati
ng the experimental study design. After reading the
article, you should attempt to answer the questions posed for thi
s research. Once you have answered these questions to
the best of your ability, you can read commentary provided by y
our instructor so that you can see how critical analysis is
applied to a research article.
In the graded portion of this activity, you will apply this same s
trategy to another article, answering questions with a
critical eye on the research. The instructions below will help yo
u complete this activity.
Step 1
Read the article, Lane et al., Tanezumab for the Treatment of Pa
in from Osteoarthritis of the Knee. New
England Journal of Medicine, 2010; 363; 16. You will find a lin
k for this article in the assignment area.
In reading the article, relate its contents to the faculty notes and
readings, paying particular attention to the
components of the experimental study design. As you read, thin
k about the investigator’s choices in:
• developing the study hypothesis
• selecting the treatment and comparison groups, and
11. •
how they conducted randomization and blinding of treatment sta
tus
Also, consider alternatives to their choices that may have impro
ved the study or perhaps made it even more
prone to bias.
Questions to Consider: Think about the design of this randomize
d controlled trial. What was its purpose? What
were the treatment and control assignments, and are they suffici
ent to answer the study hypothesis? Studies
such as this are frequently used to directly inform clinical practi
ce. How might a trial such as this one provide
useful information for clinicians treating a patient with osteoart
hritis? Perhaps most important, are both the
efficacy of the drug and the risk of side effects sufficient to rec
ommend it to patients?
Step 2
After completing Step 1, answer the following questions to the
best of your ability. You may want to record
your answers in a Word document. Note: Step 2 will not be grad
ed – it is for your practice and understanding:
a. What were the study’s rationale/hypothesis and objectives?
b.
Was there any way that the blinding of treatment status could ha
ve be compromised (in other
words, the investigator somehow figures out treatment status)? I
f so, what might have been its
potential impact?
c.
Do you see any problems with the way the outcome of interest
12. was measured? If so, suggest some
alternatives.
d.
Are there any possible concerns about patient compliance with t
heir assigned treatment? What
impact might non‐compliance have on study findings?
e.
Why do you think there was an entire paragraph dedicated to dis
closure of who conducted,
oversaw, and analyzed the study’s findings? What implications
does this information have on the
study?
f.
To what population(s) do you think this study’s findings may or
may not be generalizable, and why?
g.
This trial may have some relevance to clinical practice, but ther
e were adverse events associated
Excelsior College PBH 321
Page 2
with some of the treatment levels. Based on the frequency and t
ype of adverse events described in
the article, what advice might you want to provide physicians re
garding the use of Tanezumab for
osteoarthritis pain?
13. Step 3
Read the commentary I have provided and compare your answer
s to my analysis. At this point, you may want
to review the article to reinforce some of the points I have made
in my critical examination of the research
article. You will find a link to my commentary in the assignmen
t area.
Step 4
Now it is your turn to try analyzing epidemiological research! P
roceed to read the article, “Mussolino et al.,
Jogging and Bone Mineral Density in Men: Results From NHAN
ES III, American Journal of Public Health, 2001;
1056‐1059.” You will find a link for this article in the assignme
nt area.
As you read the article, I would like you to relate the contents o
f the article to the faculty notes and readings,
paying particular attention to the components of the cross‐sectio
nal study design. Think about the strengths
and limitations of the cross‐sectional study design. What are the
problems that arise from using data collected
for another purpose? Based on what you understand about the s
tudy design, what are the limitations of the
investigator’s conclusions?
Consider how useful the cross‐sectional study design is when re
sources are limited and a tentative public
health hypothesis needs to be evaluated. How does the availabili
ty of the NHANES database make conducting
this type of study feasible? There are also some disadvantages t
o the cross‐sectional approach. Could this
same study have been conducted using an experimental design?
14. Use the example response for the research
article by Lane et al. to guide you as you develop your answers.
Step 5
After reading the article, answer the following questions by rec
ording your answers in a Word document. As
this is a different type of research study, the questions are slight
ly different from the ones you answered in the
previous practice session. Note: your answers for this second st
udy will be graded. Submit your Word
document in the dropbox for this activity. Remember to answer
these questions in your own words. Do not
copy entire phrases from the article without attribution.
a. What were the study’s rationale/hypothesis and objectives?
b.
Was the use of an existing data source potentially problematic?
What are the strengths and
weaknesses?
c.
What were the inclusion/exclusion criteria for the study analyse
s?
d.
What was the outcome of interest? How was it measured, with r
espect to the timing of the
exposure (jogging)? Can you think of any advantages/disadvant
ages of the way either the exposure
or outcome was measured?
e.
To what population(s) do you think this study’s findings may or
may not be generalizable, and why?
15. f.
Are there any hypotheses that can be generated from the major f
indings?
g.
Could this same study have been conducted using an experiment
al design? Why or why not?
Step 6
Once you have answered the questions in Step 5 and created a
Word document, submit it to the dropbox for
this activity.
CRITICAL EYE ON RESEARCH IN EPIDEMIOLOGY
A B S T R A C T
July 2001, Vol. 91, No. 71056 American Journal of Public
Health
Jogging and Bone Mineral Density in
Men: Results From NHANES III
Michael E. Mussolino, MA, Anne C. Looker, PhD, and Eric S.
Orwoll, MD
Michael E. Mussolino and Anne C. Looker are with
the National Center for Health Statistics, Centers for
Disease Control and Prevention, Hyattsville, Md.
Eric S. Orwoll is with Oregon Health Sciences Uni-
versity, Portland.
16. Requests for reprints should be sent to Michael
E. Mussolino, MA, Division of Epidemiology, Na-
tional Center for Health Statistics, 6525 Belcrest
Rd, Suite 730, Hyattsville, MD 20782 (e-mail:
[email protected]).
This article was accepted January 12, 2001.
Objectives. This cross-sectional
population-based study assessed the as-
sociation of jogging with femoral bone
mineral density (BMD) in men.
Methods. Data are from a nationally
representative sample of 4254 men aged
20 to 59 years from the Third National
Health and Nutrition Examination Sur-
vey (NHANES III). Total femoral BMD
was measured by dual energy x-ray ab-
sorptiometry. Jogging was self-reported.
Results. Jogging (any vs none) was
strongly associated with higher BMD in
multivariate models (P < .01) for both
young and middle-aged men. Men who
jogged 9 or more times per month had
higher BMD levels than those who
jogged only 1 to 8 times per month
(P = .01).
Conclusions. Jogging is associated
with higher femoral neck BMD in men.
Additional large-scale studies that mea-
sure all aspects of jogging are war-
ranted. (Am J Public Health. 2001;91:
17. 1056–1059)
Jogging is enjoyed by millions of peo-
ple in the United States. Individuals who un-
dertake high-impact activities like jogging on
a regular basis may have higher bone mineral
density (BMD), particularly at the femoral
neck.1 Higher levels of weight-bearing or vig-
orous physical activity have been shown to
be associated with a lower incidence of hip
fractures.2,3 A number of studies have inves-
tigated the effect of jogging on BMD, with
many focusing on women.4–9 Among men,
the results for jogging and BMD have been
inconsistent. Some studies have found higher
BMD among joggers in the femoral neck,
lumbar spine, or total body,1,10–12 while others
have found an association only at lower limb
sites.13,14 Other studies have found that BMD
among runners is the same as or lower than in
nonrunners.15,16
A problem of the studies of both men and
women has been the use of self-selected ath-
letes, which, coupled with small sample sizes,
likely limits generalizability. Other analyses of
BMD and physical activity have combined jog-
ging with other strenuous activities.17 The Third
National Health and Nutrition Examination
Survey (NHANES III), a large-scale national
study, provides an opportunity to investigate
the relation between jogging and BMD in a
representative sample of young and middle-
aged men with a wide variation in self-reported
jogging frequency that likely encompasses the
casual jogger as well as the more serious run-
18. ner. Thus, our results may be more broadly gen-
eralizable than those of previous studies.
Methods
Data for these analyses were taken from
NHANES III, which collected data from a na-
tional probability sample of the civilian non-
institutionalized US population from 1988 to
1994.18 Analyses were limited to 4603 non-
Hispanic White, non-Hispanic Black, and Mex-
ican American men whose BMD levels were
measured and who were aged 20 to 59 years at
the time of the NHANES III examination, be-
cause relatively few study respondents 60 years
or older reported jogging. Men of other race/
ethnicity groups were omitted because of the
small number of observations. Also excluded
were 3 men with missing values on the jog-
ging variables and 346 men with unknown val-
ues on any of the other variables assessed in
the study. A total of 4254 men were included
in the analysis (954 joggers and 3300 nonjog-
gers). Because nonjoggers in the main analy-
sis may have participated in other leisure ac-
tivities, we also did a subanalysis in which
nonjoggers were restricted to those who re-
ported no leisure activities (n = 577).
Measurement of BMD
BMD was measured by trained examin-
ers in mobile examination centers. Total
femoral bone density was measured by dual
energy x-ray absorptiometry (Hologic QDR-
19. 1000; Hologic, Inc, Waltham, Mass). Scans
were reviewed by consultants at the Mayo
Clinic, Rochester, Minn, for quality control.19
Jogging Assessment
For data on self-reported jogging, the fol-
lowing questions were asked: “In the past
month, did you jog or run?” and “In the past
month, how often did you jog or run?” Jog-
ging was categorized as 9 or more times per
month, 1 to 8 times per month, and none.
July 2001, Vol. 91, No. 7 American Journal of Public Health
1057
TABLE 1—Characteristics of 4254 Male Joggers and
Nonjoggers Aged 20 to 59
Yearsa: Third National Health and Nutrition Examination
Survey
Jogging Status
Any Noneb
N 954 3300
Age at interview, y 33.4* 38.2
Weight, kg 81.4* 83.2
Height, cm 176.9 176.8
Body mass index, kg/m2 26.0* 26.6
Total femoral bone mineral density, g/cm2 1.069* 1.018
Smoking status, %
Current 23.2* 40.0
Former 28.1 25.4
20. Never 48.6* 34.6
Alcohol, drinks/mo 11.6 13.4
Any chronic conditions, % 4.2* 8.7
Weight loss from maximum ≥ 10%, % 26.1 24.9
Dietary variables
Food energy, kcal/day 2840.7 2828.5
Calcium, mg/day 1016.5 1036.4
Protein, g/day 105.0 104.7
Self-assessed health status, %
Excellent 34.2* 20.1
Very good 39.3 34.6
Good 21.7* 34.4
Fair 4.4* 9.5
Poor 0.4* 1.4
aControlling for age at interview.
bOf the 3300 nonjoggers, 2723 were “active nonjoggers” (i.e.,
reported other physical
activity) and 577 were “sedentary nonjoggers” (i.e., did not
report any physical activity).
*P < .05 (t test).
Other Baseline Variables
Information on smoking status (current,
former, or never), alcohol consumption (drinks
per month), self-assessed health status, food
energy, calcium consumption, protein intake,
weight history (including maximum weight),
and chronic conditions related to secondary
osteoporosis were obtained by interview. We
calculated body mass index (BMI) from mea-
21. surements of the subjects’ height and weight.
The nutritional variables were determined on
the basis of what the subject recalled eating in
the past 24 hours. The chronic-conditions vari-
able was based on self-reported doctor’s diag-
noses of congestive heart failure, stroke,
chronic bronchitis, goiter, other thyroid dis-
ease, and diabetes.
Statistical Analysis
Statistical analyses were performed with
linear regression procedures in SAS20 and SU-
DAAN.21 We used sample weights when cal-
culating point estimates, so estimates are rep-
resentative of the civilian noninstitutionalized
US population at the time of NHANES III.
Weighted multivariate regression analyses were
performed for the total sample and stratified
by age (20–39 and 40–59 years). All models
were adjusted for age at interview, race/eth-
nicity, BMI, food energy, calcium consump-
tion, protein intake, smoking status, alcohol
consumption, chronic conditions, and weight
change, unless otherwise indicated.
Results
Baseline characteristics of men by jog-
ging status (any vs none) are shown in Table 1.
Joggers tended to be younger, so all subsequent
comparisons were made after age was con-
trolled for. Joggers weighed less than nonjog-
gers and were more likely to have never
smoked, have no chronic conditions, and be in
22. excellent health. Overall, 22.3% of men re-
ported jogging in the past month. Mean femoral
BMD was 5.0% higher among joggers than
among nonjoggers; when joggers were com-
pared with sedentary nonjoggers, the figure
rose to 7.7% (1.069 g/cm2 vs 0.993 g/cm2).
In the multivariate models, men who re-
ported any jogging had a significantly higher
BMD than those who did not jog (P < .001).
Results of categorizing jogging status by fre-
quency are shown in Table 2. Men who re-
ported jogging 9 or more times per month or
1 to 8 times per month had higher BMD lev-
els than men who did not jog (P < .01). Fur-
thermore, men who jogged 9 or more times
per month had significantly higher BMD lev-
els than those who jogged 1 to 8 times per
month (P=.01) (data not shown). Age-specific
analyses revealed similar results for ages 20 to
39, while only the category of 9 or more times
per month was significant for ages 40 to 59.
The relatively small number of joggers in this
age group may have led to the reduced P val-
ues in the category of 1 to 8 times per month.
The interaction of race/ethnicity and jog-
ging group was not significant (P = .56), sug-
gesting that the effect of jogging frequency is
the same in all 3 race/ethnicity groups.
In a subanalysis to assess the effect of jog-
ging frequency on BMD, we found that the
BMD levels of those who jogged frequently
(i.e., >20 times/month) were similar to the lev-
23. els of those who jogged less frequently, sug-
gesting that there may be a ceiling for the
graded effect beyond which no further benefit
occurs.
When treated as a continuous variable in
the multivariate analyses, jogging remained a
strong predictor of higher BMD (P<.01) (data
not shown). When the self-assessed health sta-
tus variable was added to the regression mod-
els, jogging remained a statistically significant
predictor of BMD (P < .01).
We also investigated the impact of total
number of leisure activities on the relationship
between jogging and BMD. Leisure activities
varied, but walking and gardening/yard work
were the most reported activities among both
joggers (68.0% and 61.4%, respectively) and
physically active nonjoggers (56.8% and
66.1%, respectively). Joggers reported a larger
number of leisure activities than nonjoggers
(mean = 4.8 vs 2.6). Joggers also reported a
greater number of weight-bearing activities
(mean = 2.9 vs 1.1). In regression models ad-
justed for age and total number of leisure ac-
tivities, BMD remained significantly higher
for joggers (P < .01). Results were similar after
age and number of weight-bearing activities
were adjusted for (P < .01).
Discussion
Our results from this large nationally rep-
resentative sample are consistent with previ-
ous research suggesting that jogging is asso-
24. ciated with higher femoral BMD in men.
Adjustment for established BMD risk factors
did not appreciably alter the conclusions. Fur-
thermore, among the 954 joggers in the study,
47.3% reported running 8 or fewer times per
month, suggesting that even infrequent jog-
ging may be beneficial to BMD.
In our study, we were able to examine one
aspect of jogging on BMD: frequency. Other
aspects of jogging, such as distance or pace,
could not be assessed because NHANES III
did not include questions on these items. To
July 2001, Vol. 91, No. 71058 American Journal of Public
Health
TABLE 2—Frequency of Jogging and Bone Mineral Density
(BMD) for Men
Aged 20 to 59 Yearsa: Third National Health and Nutrition
Examination Survey
Jogging Frequency n BMD (g/cm2) β P
Ages 20–39
≥9 times/mo 350 1.113 .0705 .0000
1–8 times/mo 405 1.083 .0401 .0015
None 1796 1.043 (Reference) (Reference)
Ages 40–59
≥9 times/mo 98 1.076 .0651 .0021
1–8 times/mo 101 1.034 .0236 .1487
None 1504 1.011 (Reference) (Reference)
25. Ages 20–59
≥9 times/mo 448 1.104 .0685 .0000
1–8 times/mo 506 1.071 .0354 .0010
None 3300 1.036 (Reference) (Reference)
aModels were adjusted for age at interview, race/ethnicity, body
mass index, food energy,
calcium consumption, protein intake, smoking status, alcohol
consumption, chronic
conditions, and weight change.
our knowledge, jogging pace has not been ex-
amined in previous studies. Several studies
have examined the effect of jogging distance on
BMD in men, and all but one report22 sug-
gested that there may be a ceiling beyond which
additional distance does not improve BMD.
For example, one study reported that men who
ran more than 64 km per week had signifi-
cantly lower vertebral BMD than nonrunners,23
and another concluded that male long-distance
runners had reduced BMD and increased bone
turnover compared with controls.16 Mac-
Dougall et al.14 found no further increase in
lower-leg BMD in male runners who ran more
than 20 miles per week compared with those
who ran less; in fact, increased distance was
potentially detrimental (i.e., mean BMD in
those who ran 60–75 miles per week was sim-
ilar to that in controls). Our results showing
similar femoral BMD in those who jogged fre-
quently (i.e., >20 times/month) and those who
jogged less frequently, while exploratory in na-
ture, were consistent with these studies. The
mechanism underlying this ceiling effect is
26. probably multifactorial and may include factors
such as body weight14 or sex hormone status.24
Our study focused on the proximal femur,
which is a skeletal site that receives increased
loading from jogging. The effect of jogging on
other skeletal sites, which may not be as di-
rectly loaded, has varied. One longitudinal
study found significantly higher lumbar spine
BMD levels in older male runners,11 but other
studies have found similar lumbar spine BMD
levels in runners and controls.1,15 Running also
does not appear to affect BMD in the forearm.3
Our study has the advantage of including
joggers with widely diverse jogging frequency.
Thus, our results may be more applicable to
the general population than those of previous
studies, which tended to focus on more dedi-
cated or elite athletes. However, our study has
several limitations. It is cross-sectional in na-
ture and therefore cannot provide definitive ev-
idence that jogging caused the higher BMD
levels observed in male joggers. The dose–
response seen for jogging frequency is sug-
gestive in this regard, but it is also possible that
self-selection played a role (i.e., those who
chose to jog had higher BMD before com-
mencing to jog). Results did not change when
we included in the analysis self-reported health
status or the presence of chronic conditions,
which suggests that the relationship between
jogging and BMD is not due to differences in
health status between joggers and nonjoggers.
However, self-selection could also occur be-
27. cause those who choose to jog are better
adapted physically to perform this activity (i.e.,
their musculoskeletal system, including BMD,
is better suited to jogging). This would not be
addressed by the overall health status or chronic
condition variables.
We also used a self-reported assessment
of jogging in our study rather than a direct mea-
surement. However, other data suggest that
self-reported physical activity is reasonably
valid. For example, in the predictive models of
cardiorespiratory fitness that also included
weight, resting heart rate, and current smok-
ing, self-reported physical activity was found
to be the principal contributor.25 The associa-
tions consistently found between self-reported
physical activity and lower risk of several dis-
eases also support the validity of the measure.25
Our study looked at leisure-time physical ac-
tivity only, so it could underestimate total phys-
ical activity. However, most people do not have
jobs requiring regular physical exertion.25
We conclude that jogging is associated
with significantly higher femoral BMD in men.
This effect was graded in nature and was ob-
served in both younger and middle-aged men.
This finding may have public health signifi-
cance, since femoral BMD is a strong predic-
tor of hip fracture, which is the most devastat-
ing consequence of osteoporosis from a public
health standpoint. Most important, our data
suggest that even a modest frequency of jog-
28. ging is associated with higher femoral BMD
and thus may provide some protection against
osteoporosis and fracture. Additional large-
scale studies that measure all aspects of jog-
ging are warranted.
Contributors
M. E. Mussolino was primarily responsible for the
study design, writing, and data analyses. A. C. Looker
provided input into the initial study design, wrote parts
of the “Discussion” section, and revised other parts
of the manuscript. E. S. Orwoll provided guidance
on interpreting data and contributed to the analysis of
the impact of other leisure activities on the relation-
ship between jogging and BMD. All authors partici-
pated in preparing the manuscript.
References
1. Brahm H, Strom H, Piehl-Aulin K, Mallmin H,
Ljunghall S. Bone metabolism in endurance
trained athletes: a comparison to population-
based controls based on DXA, SXA, quantitative
ultrasound, and biochemical markers. Calcif Tis-
sue Int. 1997;61:448–454.
2. Kujala UM, Kaprio J, Kannus P, Sarna S,
Koskenvuo M. Physical activity and osteoporotic
hip fracture risk in men. Arch Intern Med. 2000;
160:705–708.
3. Joakimsen RM, Magnus JH, Fonnebo V. Phys-
ical activity and predisposition for hip fractures:
a review. Osteoporos Int. 1997;7:503–513.
4. Petit MA, Prior JC, Barr SI. Running and ovu-
29. lation positively change cancellous bone in pre-
menopausal women. Med Sci Sports Exerc.
1999;31:780–787.
5. Emslander HC, Sinaki M, Muhs JM, et al. Bone
mass and muscle strength in female college ath-
letes (runners and swimmers). Mayo Clin Proc.
1998;73:1151–1160.
6. Taaffe DR, Robinson TL, Snow CM, Marcus R.
High-impact exercise promotes bone gain in
well-trained female athletes. J Bone Miner Res.
1997;12:255–260.
7. Etherington J, Harris PA, Nandra D, et al. The
effect of weight-bearing exercise on bone min-
eral density: a study of female ex-elite athletes
and the general population. J Bone Miner Res.
1996;11:1333–1338.
8. Snow-Harter C, Bouxsein ML, Lewis BT, Carter
DR, Marcus R. Effects of resistance and en-
durance exercise on bone mineral status of
young women: a randomized exercise interven-
tion trial. J Bone Miner Res. 1992;7:761–769.
9. Jonsson B, Ringsberg K, Josefsson PO, Johnell
O, Birch-Jensen M. Effects of physical activity
on bone mineral content and muscle strength in
women: a cross-sectional study. Bone. 1992;13:
191–195.
10. Need AG, Wishart JM, Scopacasa F, Horowitz
M, Morris HA, Nordin BEC. Effect of physical
30. July 2001, Vol. 91, No. 7 American Journal of Public Health
1059
activity on femoral bone density in men. BMJ.
1995;310:1501–1502.
11. Lane NE, Oehlert JW, Bloch DA, Fries JF. The
relationship of running to osteoarthritis of the
knee and hip and bone mineral density of the
lumbar spine: a 9 year longitudinal study. J
Rheumatol. 1998;25:334–341.
12. Stewart AD, Hannan J. Total and regional bone
density in male runners, cyclists, and controls.
Med Sci Sports Exerc. 2000;32:1373–1377.
13. Bennell KL, Malcolm SA, Khan KM, et al.
Bone mass and bone turnover in power athletes,
endurance athletes and controls: a 12-month lon-
gitudinal study. Bone. 1997;20:477–484.
14. MacDougall JD, Webber CE, Martin J, et al. Re-
lationship among running mileage, bone den-
sity, and serum testosterone in male runners. J
Appl Physiol. 1992;73:1165–1170.
15. Goodpaster BH, Costill DL, Trappe SW, Hughes
GM. The relationship of sustained exercise train-
ing and bone mineral density in aging male run-
ners. Scand J Med Sci Sports. 1996;6:216–221.
16. Hetland ML, Haarbo J, Christiansen C. Low
bone mass and high bone turnover in male long
distance runners. J Clin Endocrinol Metab.
31. 1993;77:770–775.
17. Greendale GA, Barrett-Connor E, Edelstein S,
Ingles S, Haile R. Lifetime leisure exercise and
osteoporosis. The Rancho Bernardo Study. Am
J Epidemiol. 1995;141:951–959.
18. Plan and operation of the Third National Health
and Nutrition Examination Survey, 1988–94.
National Center for Health Statistics. Vital
Health Stat 1. 1994;No. 32.
19. Wahner HW, Looker A, Dunn WL, Hauser MF,
Walters LC, Novak C. Quality control of bone
densitometry in a national health survey
(NHANES III) using three mobile examination
centers. J Bone Miner Res. 1994;9:951–960.
20. SAS/STAT User’s Guide, Version 6. 4th ed. Cary,
NC: SAS Institute Inc; 1990.
21. Shah BV, Barnwell BG, Bieler GS. SUDAAN
User’s Manual: Software for Analysis of Corre-
lated Data, Release 6.40. Research Triangle
Park, NC: Research Triangle Institute; 1995.
22. Lane NE, Bloch DA, Jones HH, Marshall WH
Jr, Wood PD, Fries JF. Long-distance running,
bone density, and osteoarthritis. JAMA. 1986;
255:1147–1151.
23. Bilanin JE, Blanchard MS, Russek-Cohen E.
Lower vertebral bone density in male long dis-
tance runners. Med Sci Sports Exerc. 1989;21:
66–70.
32. 24. Cooper CS, Taaffe DR, Guido D, Packer E, Hol-
loway L, Marcus R. Relationship of chronic en-
durance exercise to the somatotropic and sex
hormone status of older men. Eur J Endocrinol.
1998;138:517–523.
25. Physical Activity and Health: A Report of the
Surgeon General. Atlanta, Ga: National Center
for Chronic Disease Prevention and Health Pro-
motion; 1996.
T h e n e w e n g l a n d j o u r n a l o f m e d i c i n e
n engl j med 363;16 nejm.org october 14, 2010 1521
original article
Tanezumab for the Treatment of Pain
from Osteoarthritis of the Knee
Nancy E. Lane, M.D., Thomas J. Schnitzer, M.D., Ph.D.,
Charles A. Birbara, M.D.,
Masoud Mokhtarani, M.D., David L. Shelton, Ph.D., Mike D.
Smith, Ph.D.,
and Mark T. Brown, M.D.
From the University of California at Davis
Medical School, Sacramento (N.E.L.);
Northwestern University Feinberg
School of Medicine, Chicago (T.J.S.); Uni-
versity of Massachusetts School of Medi-
cine, Worcester (C.A.B.); Rinat Neurosci-
34. weight) or placebo on days 1 and 56. The primary efficacy
measures were knee pain
while walking and the patient’s global assessment of response
to therapy. We also
assessed pain, stiffness, and physical function using the
Western Ontario and Mc-
Master Universities Osteoarthritis Index (WOMAC); the rate of
response using the
criteria of the Outcome Measures for Rheumatology Committee
and Osteoarthritis
Research Society International Standing Committee for Clinical
Trials Response
Criteria Initiative (OMERACT–OARSI); and safety.
Results
When averaged over weeks 1 through 16, the mean reductions
from baseline in knee
pain while walking ranged from 45 to 62% with various doses
of tanezumab, as
compared with 22% with placebo (P<0.001). Tanezumab, as
compared with placebo,
was also associated with significantly greater improvements in
the response to ther-
apy as assessed with the use of the patients’ global assessment
measure (mean in-
creases in score of 29 to 47% with various doses of tanezumab,
as compared with
19% with placebo; P≤0.001). The rate of response according to
the OMERACT–OARSI
criteria ranged from 74 to 93% with tanezumab treatment, as
compared with 44%
with placebo (P<0.001). The rates of adverse events were 68%
and 55% in the tan-
ezumab and placebo groups, respectively. The most common
adverse events among
36. pain either locally or systemically, depending on
the dose and the route of administration.8,9 In-
creased expression of nerve growth factor is found
in inflamed tissues from patients with conditions
such as arthritis, pancreatitis, and prostatitis.10-12
Levels of nerve growth factor are also elevated in
animal models of inflammatory pain, and phar-
macologic inhibition of the activity of nerve growth
factor in these models reduces or blocks signs of
pain. Therefore, nerve growth factor appears to
have a role in causing and augmenting pain in
these models.1,13-15 The development of therapeu-
tic interventions that are based on antagonism of
nerve growth factor is of interest.16,17
The treatment options for patients with pain-
ful osteoarthritis of the knee are inadequate. Non-
steroidal antiinf lammatory drugs and narcotic
analgesics are commonly used18,19; however, these
medications have well-described gastrointestinal
and cardiorenal side effects,20,21 and the response
to them is unsatisfactory in some patients.22,23
Potent analgesic medications with acceptable side-
effect profiles may help to avoid or delay surgi-
cal intervention.24
Tanezumab is a humanized IgG2 monoclonal
antibody directed against nerve growth factor that
blocks the interaction of nerve growth factor with
its receptors, TrkA and p75.25 A small phase 1
clinical trial showed that a single intravenous in-
jection of tanezumab substantially reduced pain
in patients with osteoarthritis of the knee.26,27
We report the results of a proof-of-concept study
of tanezumab in patients with advanced osteoar-
thritis of the knee who did not have a satisfactory
37. response to nonopiate pain medications or who
were considered to be candidates for invasive in-
tervention. We compared the safety, side-effect
profile, and efficacy of repeat doses of tanezu-
mab as compared with placebo.
M e t h o d s
Study Population
We enrolled patients, 40 to 75 years of age, who
had osteoarthritis of the knee as diagnosed on
the basis of American College of Rheumatology
criteria,28 with radiographic confirmation (Kell-
gren–Lawrence grade 2 or higher, on a scale of
0 to 4, with higher numbers indicating more se-
vere signs of osteoarthritis). Patients were eligible
only if they were unwilling to take nonopiate pain
medications or had had an unsatisfactory response
to them or if they were candidates for or seeking
invasive interventions such as intraarticular in-
jections or total knee replacement. All pain med-
ications except the “rescue” medications, acet-
aminophen and tramadol, were discontinued at the
screening visit. At the time of randomization, pa-
tients had to have pain while walking on a flat
surface (the walking-pain measure of the Western
Ontario and McMaster Universities Osteoarthritis
Index [WOMAC]) that they rated between 50 and
90 on a visual-analogue scale that ranged from
0 to 100, with 100 indicating maximal pain. In
addition, among patients who discontinued pain
medication during the screening period, an in-
crease in the walking-pain score of 10 or more was
required between screening and randomization.
38. The exclusion criteria were pregnancy, a history
of or current symptoms of an autoimmune disor-
der, cancer within the previous 5 years except for
cutaneous basal-cell or squamous-cell cancer re-
solved by excision, allergic reaction to monoclo-
nal antibodies or IgG-fusion proteins, infection
with hepatitis B or hepatitis C virus or the human
immunodeficiency virus, drug abuse, fibromyal-
gia, clinically significant cardiac disease, diabetes
mellitus requiring oral treatment or insulin, clini-
cally significant neurologic disease, or a clinically
significant psychiatric disorder. All participants
provided written informed consent.
Study Design and Oversight
Patients were recruited between March 30, 2006,
and May 3, 2007, at 46 study centers in the United
States and were screened within 30 days before
randomization. Eligible patients who were taking
pain medication other than acetaminophen and
tramadol underwent a washout period (of at least
5 half-lives of the medication). Patients rated their
knee pain and recorded the score in an electronic
diary every day for 3 days before randomization
to establish their baseline pain score. Eligible pa-
tients were randomly assigned on day 1, with the
use of an interactive voice-response system, to pla-
cebo or to tanezumab at a dose of 10, 25, 50, 100,
or 200 µg per kilogram of body weight, such that
there were equal numbers in each study group.
The New England Journal of Medicine
Downloaded from nejm.org at NEW YORK STATE LIB
ACQUISITIONS on January 9, 2011. For personal use only. No
40. could enter an open-label extension of the trial
(ClinicalTrials.gov number, NCT00399490) at week
16 (day 112) if they had received two doses of
the study drug and had been followed for at
least 8 weeks after the last dose.
The study was designed and coordinated by
Rinat Neuroscience, a subsidiary of Pfizer. An
external data and safety monitoring board mon-
itored safety, and day-to-day study operations, in-
cluding data management, were overseen by PPDI
(a contract research organization contracted by
Pfizer). The data were analyzed by Pfizer. The first
author wrote the first draft of the manuscript.
All the authors were involved in the design of the
study and interpretation of the data, contributed
to the writing of the manuscript, made the deci-
sion to submit the manuscript for publication,
attest that the study was performed in accordance
with the protocol and the statistical analysis plan,
and vouch for the accuracy and completeness of
the reported results. Editorial support was provid-
ed by UBC Scientific
Solution
s and was funded by
Pfizer. The study protocol was approved by the
local ethics committee at each study center be-
fore patient enrollment began. The protocol, in-
cluding the statistical analysis plan, is available
41. with the full text of this article at NEJM.org.
Efficacy Assessments
The primary efficacy outcomes were the change
from baseline in the pain the patient felt in the
index knee while walking on a flat surface and in
the patient’s global assessment of response to ther-
apy, averaged over weeks 1 through 16. Second-
ary efficacy outcomes included the change from
baseline in overall knee pain and in scores on the
WOMAC subscales for pain, stiffness, and physi-
cal function. Pain while walking and overall knee
pain were recorded daily in an electronic diary,
whereas the patient’s global assessment of response
to therapy and scores on the WOMAC subscales
were recorded on study-visit days. Pain, the pa-
tient’s global assessment, and scores on the
WOMAC subscales were assessed with the use of
a visual-analogue scale that ranged from 0 to 100.
In the case of pain and WOMAC scores, a lower
score indicated improvement (i.e., less pain, less
stiffness, and less limitation of physical function),
whereas in the case of the patient’s global assess-
ment, a higher score indicated improvement (i.e.,
42. a better response to therapy). Another secondary
outcome was the response to therapy on the ba-
sis of the criteria of the Outcome Measures for
Rheumatology Committee and Osteoarthritis Re-
search Society International Standing Committee
for Clinical Trials Response Criteria Initiative
(OMERACT–OARSI).29 Patients were classified as
having had a response if the WOMAC pain or phys-
ical-function score decreased by 50% or more and
by 20 or more points on the visual-analogue scale
or if two of the following three findings were re-
corded: a decrease in the WOMAC pain score by
20% or more and by 10 or more points on the vi-
sual-analogue scale, a decrease in the WOMAC
physical-function score by 20% or more and by
10 or more points on the scale, or an increase in
the score on the patient’s global assessment by 20%
or more and by 10 or more points on the scale.
Rescue medication use, also a prespecified second-
ary outcome, was recorded daily in the patients’
diaries.
Safety Assessments
The nature, onset, duration, severity, and outcome
44. logic examination performed by the investigator
or adverse events suggestive of peripheral neurop-
athy were further evaluated by an independent
neurologist.
Statistical Analysis
We estimated that we would need to enroll 75
patients in each group for the study to have 80%
power to detect a difference between the tanezu-
mab groups and the placebo group of 15 points
or more on the visual-analogue scale for the pri-
mary outcomes (the average change from base-
line through week 16 in knee pain while walking
and in the patient’s global assessment of re-
sponse to therapy), with an effect size of 0.5 (in-
dicating a moderate difference).31 With respect to
the intensity of pain, decreases of 10 or more
points on a visual-analogue scale that ranges
from 1 to 100 are considered to be minimally
important improvements, and decreases of 20 or
more points are considered to be moderately im-
portant improvements.32
Changes from baseline in all the measures that
45. were assessed with the use of a visual-analogue
scale were determined with a mixed-model, re-
peated-measures analysis, with model terms for
study site, study group, study week, and the in-
teraction between study medication and study
week, and with the baseline score on the visual-
analogue scale as a covariate, with no imputation
for missing data. A repeated-measures analysis
was also used to assess the number of rescue
medication pills taken. For the analysis of rates
of response according to OMERACT–OARSI crite-
ria (calculated on the basis of the average change
from baseline to week 16), we used the Cochran–
Mantel–Haenszel test, stratified according to study
site, to compare the proportions of patients in the
tanezumab groups who had a response with the
proportion of those in the placebo group who had
a response.
R e s u l t s
Baseline Characteristics of the Patients
Of the 450 patients who underwent randomiza-
tion, 440 received at least one dose of the study
46. medication and underwent at least one efficacy
assessment (the modified intention-to-treat pop-
ulation) (Fig. 1). Most patients in the modified
intention-to-treat population had a Kellgren–Law-
rence grade of 3 (52%) or 4 (17%) and severe pain33
both at the time of screening (mean [±SD] score
on the visual-analogue scale, 58±12) and at the
time of randomization, after washout of previous
medications (score on the visual-analogue scale,
71±11). A total of 87% of the patients reported
taking pain medication for knee pain before they
enrolled in the study. The baseline characteristics
of the patients were similar across study groups
(Table 1).
Efficacy
As compared with placebo, tanezumab, at all the
doses studied, was associated with an improve-
ment in the primary efficacy measures. The
mean reduction from baseline in the score on the
visual-analogue scale for knee pain while walk-
ing, averaged over weeks 1 through 16, ranged
from 31.0 to 45.2 points with various doses of
47. tanezu mab, as compared with 15.5 points with
placebo (a reduction of 45 to 62% with tanezum-
ab vs. 22% with placebo, P<0.001 for the com-
parison of all doses of tanezumab with placebo)
(Fig. 2A). We observed significant improvements
among patients receiving tanezumab as com-
pared with those receiving placebo by the end of
the first week, and significant improvements
continued to be seen throughout the remainder
of the treatment period. The mean increase from
baseline in the score on the patient’s global assess-
ment of response to therapy, averaged over weeks
1 through 16, ranged from 16.3 to 23.7 points
with various doses of tanezumab, as compared
with 9.2 points with placebo (an increase of 29 to
47% with tanezumab vs. 19% with placebo,
P≤0.001 for the comparison of all doses of tan-
ezumab with placebo) (Fig. 2B). By week 2, the
scores on the patient’s global assessment had im-
proved in the group receiving 25 µg of tanezum-
ab per kilogram, as compared with placebo
(P = 0.002); by week 4, the scores had improved in
the groups receiving 50 µg, 100 µg, and 200 µg of
tanezumab per kilogram, as compared with pla-
cebo (P = 0.01, P<0.001, and P<0.001 for the three
49. were similar in magnitude to those reported for
knee pain while walking (reductions of 43 to
62% with tanezumab vs. 23% with placebo,
P<0.001 for the comparison of all doses of tan-
ezumab with placebo). Treatment with tanezumab,
as compared with placebo, was also associated
with reductions in the mean WOMAC scores for
pain (reductions of 46 to 64% vs. 23%), stiffness
(48 to 65% vs. 22%), and physical function (47 to
65% vs. 22%) over the same period (P<0.001 for
all comparisons) (Table 2). The percentage of
patients who had a response to therapy accord-
ing to OMERACT–OARSI criteria, averaged over
weeks 1 through 16, was significantly higher with
tanezumab treatment than with placebo (74 to
93% vs. 44%, P<0.001 for the comparison of all
doses of tanezumab with placebo) (Table 2). Res-
cue medications that were allowed per protocol
were used less frequently by tanezumab-treated
patients than by placebo-treated patients during
weeks 1 through 16 (odds ratio with tanezumab,
0.50; 95% confidence interval [CI], 0.24 to 1.02;
P = 0.05) and was significantly lower during weeks
1 through 4 (odds ratio, 0.49; 95% CI, 0.24 to 0.99;
50. P = 0.04).
Safety
Among patients in the tanezumab groups, the
three most common adverse events were head-
ache, upper respiratory tract infection, and par-
esthesia (Table 3). The incidence of treatment-
related adverse events was higher among patients
treated with 100 µg or 200 µg of tanezumab per
kilogram than among patients who received lower
doses (28% and 35% in the groups receiving 100 µg
and 200 µg per kilogram, respectively, vs. 11 to
18% in the groups receiving other doses).
Peripheral sensory symptoms, including par-
esthesia, were reported in 14% of the patients re-
ceiving tanezumab and in 4% of those receiving
placebo (Table 3); the severity of these adverse
450 Patients underwent randomization
75 Were assigned
to receive placebo
51. 65 Discontinued study
by wk 26
24 Entered OLE
28 Had no response
13 Had other reason
10 Completed study
73 Were included in
modified intention-
to-treat analysis
74 Were included in
safety analysis
75 Were assigned to
receive tanezumab,
10 µg/kg
56 Discontinued study
by wk 26
36 Entered OLE
52. 6 Had an adverse
event
7 Had no response
7 Had other reason
19 Completed study
74 Were included in
modified intention-
to-treat analysis
74 Were included in
safety analysis
75 Were assigned to
receive tanezumab,
25 µg/kg
53 Discontinued study
by wk 26
34 Entered OLE
1 Had an adverse
53. event
12 Had no response
6 Had other reason
22 Completed study
75 Were included in
modified intention-
to-treat analysis
74 Were included in
safety analysis
75 Were assigned to
receive tanezumab,
50 µg/kg
62 Discontinued study
by wk 26
38 Entered OLE
4 Had an adverse
54. event
10 Had no response
10 Had other reason
13 Completed study
72 Were included in
modified intention-
to-treat analysis
74 Were included in
safety analysis
75 Were assigned to
receive tanezumab,
100 µg/kg
50 Discontinued study
by wk 26
33 Entered OLE
3 Had an adverse
event
55. 4 Had no response
10 Had other reason
25 Completed study
74 Were included in
modified intention-
to-treat analysis
74 Were included in
safety analysis
75 Were assigned to
receive tanezumab,
200 µg/kg
60 Discontinued study
by wk 26
38 Entered OLE
8 Had an adverse
event
5 Had no response
57. T h e n e w e n g l a n d j o u r n a l o f m e d i c i n e
n engl j med 363;16 nejm.org october 14, 20101526
events was mild in 56 of the 70 patients with
these symptoms in the tanezumab groups and in
all 3 patients with these symptoms in the placebo
group and was moderate in the remaining 14 pa-
tients in the tanezumab groups. The results of
neurologic examinations of these patients were
predominantly normal; in patients with clinically
significant changes, abnormalities were minor
and consisted mainly of changes in sensation in
the extremities and diminished deep-tendon re-
flexes. A total of 15 patients had abnormal periph-
eral sensation that was unresolved at the final
visit. Of these 15 patients, 6 had clinically sig-
nificant findings on neurologic examination: de-
creased ankle ref lexes in 1 patient who was
receiving 100 µg of tanezumab per kilogram;
decreased temperature and a sensation of sharp
pain, vibration, or both, in the toes or feet in
4 patients (1 each in the groups receiving 25 and
100 µg of tanezumab per kilogram and 2 in the
58. group receiving 200 µg per kilogram); and bilat-
eral decreased sensation in pain, fine touch, and
temperature in a radial-nerve distribution in 1 pa-
tient who was receiving 100 µg of tanezumab per
kilogram; all other aspects of the neurologic ex-
amination were normal.
Additional follow-up data were available for
7 patients with abnormal peripheral sensation
who subsequently participated in the open-label
extension study, and in each case, the adverse
event resolved before the patient’s entry into the
extension study. Similarly, follow-up data were
available for 1 of the 6 patients with neurologic
findings, and in the case of this patient, the neu-
rologic findings had normalized by the time of the
baseline visit in the open-label extension study.
The mean time to the onset of abnormal sen-
sory symptoms was 33 days (median, 14) after the
first dose of tanezumab, and the mean duration
of symptoms was 18 days (median, 4) across tan-
Table 1. Baseline Characteristics of the Study Patients.*
61. WOMAC score‖
Pain 69.0±11.9 65.8±13.9 69.2±12.5 62.1±12.3 68.3±13.2
68.4±12.0
Stiffness 74.4±13.5 69.7±13.1 75.0±12.4 66.7±17.5 71.2±17.9
73.3±13.1
Physical function 69.0±12.5 63.8±13.6 69.2±14.6 62.6±12.3
67.4±14.8 67.8±14.0
* Plus–minus values are means ±SD.
† Race was self-reported.
‡ A Kellgren–Lawrence score of 2 (minimal signs of
osteoarthritis) indicates definite osteophytes without reduction
of the joint space; a score
of 3 (moderate signs of osteoarthritis) indicates diminished joint
space; and a score of 4 (severe signs of osteoarthritis) indicates
greatly re-
duced joint space. Data are from the intention-to-treat
population, and missing data are excluded.
§ Knee pain while walking was assessed with the use of a
68. 16.3±1.7
9.2±1.8
Mean Change from
Baseline over
Weeks 1–16
Dose 1 Dose 2
Figure 2. Change from Baseline in Patients’ Assessment of
Knee Pain while Walking and in Patients’ Global
Assessment of Response to Therapy.
The patient’s assessment of knee pain while walking and the
patient’s global assessment of response to therapy
were obtained at baseline and at the indicated times with the use
of a visual-analogue scale that ranged from 0 to 100.
In the case of knee pain, a decrease in the score indicates
improvement (i.e., less pain); in the case of the patient’s
global assessment, an increase in the score indicates
improvement (i.e., a better response to therapy). Changes are
reported as least-squares means ±SE. P<0.001 for the
comparisons of all doses of tanezumab with placebo in the
70. were more variable.
Serious adverse events were reported in 6 pa-
tients (2%) receiving tanezumab (appendicitis,
bacterial arthritis, cellulitis, spinal stenosis, breast
cancer, and syncope) and in 1 patient (1%) re-
ceiving placebo (noncardiac chest pain). A total of
6% of tanezumab-treated patients withdrew from
the study because of adverse events; no placebo-
treated patients withdrew because of adverse
events (Fig. 1). We observed no clinically impor-
tant changes in electrocardiographic findings,
postural vital signs, or mental status or cognition
in any of the study participants, and we did not
detect the presence of antitanezumab antibodies
in any of the patients assigned to a tanezumab
group.
One site in the current study also participated
in a subsequent phase 3 trial and was closed by
the sponsor owing to substantial noncompliance
with Good Clinical Practice guidelines and with
the protocol in that phase 3 study. Therefore, all
analyses from the current study were repeated,
with the 23 patients from that site excluded; only
71. very small changes in the efficacy and safety re-
sults and slight increases in P values were seen
when the patients from that site were not included
in the analyses (see the Supplementary Appendix,
available at NEJM.org).
D i s c u s s i o n
Two injections of tanezumab — a monoclonal an-
tibody that inhibits nerve growth factor — 8 weeks
apart at doses ranging from 10 to 200 µg per
kilogram resulted in clinically significant reduc-
tions in knee pain, stiffness, and limitations of
physical function in patients with moderate-to-
severe knee osteoarthritis. Entries in daily pain
diaries indicated that differences between tanez-
umab therapy and placebo were apparent within
days after the first injection, and the efficacy per-
sisted throughout the 4-month treatment period.
Although the study was not powered to assess
dose response, and no formal dose–response
analy sis was performed, the reductions in pain
appeared to be greater among patients taking
higher doses of tanezumab (100 or 200 µg per
72. kilogram) than among those taking lower doses,
with no clear benefit of the 200-µg dose over the
100-µg dose. Clinically meaningful pain relief is
often described as a reduction in pain intensity of
approximately 30% from the baseline level,34,35
and in this study, reductions ranged from 45 to
62% with tanezumab. Furthermore, reductions in
Table 2. Secondary Efficacy Outcomes.*
Outcome
Placebo
(N = 73)
Tanezumab,
10 µg/kg
(N = 74)
Tanezumab,
25 µg/kg
(N = 75)†
Tanezumab,
50 µg/kg
(N = 72)
73. Tanezumab,
100 µg/kg
(N = 74)
Tanezumab,
200 µg/kg
(N = 72)
Change in WOMAC score from baseline
through week 16
Pain subscale –16.2±2.4 –30.1±2.3 –36.0±2.2 –29.0±2.4 –
39.6±2.2 –43.5±2.3
Stiffness subscale –16.3±2.4 –33.5±2.3 –37.7±2.2 –34.5±2.4 –
42.7±2.2 –47.8±2.4
Physical-function subscale –15.2±2.3 –30.1±2.3 –34.9±2.2 –
30.8±2.4 –40.5±2.2 –43.8±2.3
Response to therapy according to
OMERACT–OARSI criteria
74. by week 16 (% of patients)‡
43.8 74.3 84.0 75.0 93.2 93.1
* Plus–minus values are means ±SE. P<0.001 for all
comparisons of the five doses of tanezumab with placebo. These
analyses were per-
formed on data from the modified intention-to-treat population.
† One patient who was randomly assigned to receive 25 μg of
tanezumab per kilogram of body weight instead received 50 μg
per kilogram.
‡ According to the criteria of the Outcome Measures for
Rheumatology Committee and Osteoarthritis Research Society
International Standing
Committee for Clinical Trials Response Criteria Initiative
(OMERACT–OARSI), patients were classified as having had a
response if the
WOMAC pain or physical-function score decreased by 50% or
more and by 20 or more points on the visual-analogue scale or
if two of the
following three findings were observed: a decrease in the
WOMAC pain score by 20% or more and by 10 or more points
on the scale, a de-
76. The majority of adverse events that were re-
ported by patients taking tanezumab, including
abnormal peripheral sensations, were mild to
moderate in severity. The occurrence of adverse
events appeared to be dose-dependent. Since nerve
growth factor is thought to act on small-diameter
sensory afferents, the occurrence of paresthesia
and other signs associated with large-fiber sensory
function is interesting. The more frequent occur-
rence of these events within a short time after
administration of the first dose suggests that there
may be transient changes in sensitivity or “tone”
of different afferent fiber populations, leading to
altered sensations. Owing to their largely transient
nature, it is unlikely that these adverse events are
indicative of neurodegenerative changes. The tran-
sient nature of these events is consistent with
Table 3. Frequency of Adverse Events.
Variable
Placebo
(N = 74)
81. reserved.
T h e n e w e n g l a n d j o u r n a l o f m e d i c i n e
n engl j med 363;16 nejm.org october 14, 20101530
findings in long-term studies of the safety of high-
dose tanezumab in nonhuman primates.36 Eluci-
dation of the mechanisms underlying these ef-
fects and any potential long-term consequences
require further investigation. The assessed mea-
sures of mental status and cognition were un-
changed, suggesting that the effects of tanezumab
were limited to the peripheral nervous system.
The limitations of this study include the lack
of a comparison group receiving a different ac-
tive treatment, a study population that was too
small for a statistical comparison of efficacy ac-
cording to dose, and short-term exposure to
tanezumab (since knee pain from osteoarthritis
usually requires long-term treatment). However,
our study shows efficacy in patients with more
82. severe osteoarthritis than those in other trials.
For example, two studies37,38 of glucosamine and
chondroitin in patients with osteoarthritis of the
knee excluded patients who had the highest de-
gree of severity on radiography (a Kellgren–Law-
rence grade of 4), whereas 17% of our study popu-
lation had this degree of severity. Baseline scores
for knee pain on the visual-analogue scale in these
two other studies were 54 and 57, respectively,
whereas our population had a mean score for knee
pain on the day of randomization of 71±11.
Since the completion of this study and through
May 24, 2010, progressively worsening osteoar-
thritis associated with radiographic evidence of
bone necrosis developed in 16 subjects participat-
ing in 1 of 13 phase 3 studies of tanezumab for
osteoarthritis of the hip and knee; all 16 subjects
required total joint replacements. The affected
joints were the knee, hip, or shoulder (predomi-
nantly unilateral involvement), with more than
half the cases occurring in a joint other than the
index joint under evaluation in the study. These
16 events led the Food and Drug Administration
83. (FDA) on June 22, 2010, to put the osteoarthritis
clinical program for tanezumab on clinical hold
until more information can be obtained to de-
termine the true incidence and the causality of
these events. More recently, the FDA requested the
suspension of two additional trials of tanezumab,
one involving patients with low back pain and
the other involving patients with diabetic neu-
ropathy.
Our proof-of-concept study showed that tan-
ezumab had a favorable efficacy profile for the
treatment of moderate-to-severe knee pain associ-
ated with osteoarthritis. Longer trials involving
larger samples are needed to better understand
safety and tolerability issues and explore the clini-
cal potential of tanezumab as an alternative to
current pharmacologic treatments.
Supported by Rinat Neuroscience, now a subsidiary of Pfizer.
Editorial support was provided by Papia Das and Elizabeth
Young of UBC Scientific