вход по аккаунту


Use of short-term efficacytoxicity tradeoffs to select second-line drugs in rheumatoid arthritis. A metaanalysis of published clinical trials

код для вставкиСкачать
A Metaanalysis of Published Clinical Trials
Objective. Preferred drugs for rheumatoid arthritis (RA) should be those that have maximal efficacy
with the least toxicity. We evaluated the efficacy and
toxicity tradeoffs for drugs frequently used in the treatment of RA.
Methods. We updated 2 metaanalyses of published clinical trials, by adding trials published through
1990 and trials of azathioprine (AZA). We tested 3
different definitions of efficacy, each plotted against 3
different toxicity measures, for antimalarial drugs,
methotrexate (MTX), auranofin, injectable gold, Dpenicillamine, sulfasalazine (SSZ), AZA, and placebo.
Efficacy measures included composite efficacy (a combination of joint count, grip strength, and erythrocyte
sedimentation rate), tender joint count alone, and a
measure based on how many patients dropped out due
to inefficacy. Toxicity measures were the proportion
dropping out due to toxicity, the same dropouts with
side effects weighted for severity using a modification of
a published toxicity index, and the proportion with
severe toxicities (defined as a score of at least 7 of 10 on
the toxicity index). The latter were usually organ toxicities (e.g., cytopenias and renal involvement).
From the Boston University Arthritis Center, and the
Departments of Medicine, Boston City and University Hospitals,
Boston, Massachusetts.
Supported by NIH Multipurpose Arthritis Center grant
David T. Felson, MD, MPH: Associate Professor of Medicine and Public Health; Jennifer J. Anderson, PhD: Associate
Research Professor of Medicine; Robert F. Meenan, MD, MPH,
MBA: Professor of Medicine.
Address reprint requests to David T. Felson, MD, MPH,
A203, 80 East Concord Street, Boston, MA 02118.
Submitted for publication February 21, 1992; accepted in
revised form May 12, 1992.
Arthritis and Rheumatism, Vol. 35, No. 10 (October 1992)
Results. All 9 efficacyltoxicity tradeoff plots suggested that MTX and antimalarial drugs had the highest
efficacy relative to toxicity. MTX scored among the most
efficacious of the drugs and, of these, had the least
toxicity. Antimalarial drugs, though showing only moderate efficacy, had the lowest toxicity rate of all the
drugs. SSZ scored close to MTX but was, in general,
slightly more toxic.
Conclusion. In the short-term context of clinical
trials, antimalarial drugs and MTX have the best
efficacy/toxicity tradeoffs and may, therefore, be the
preferred drugs.
The use of second-line drugs in rheumatoid
arthritis (RA) is currently being reevaluated. The
widely accepted pyramid approach to therapy has
been challenged, and earlier use of second-line drugs
suggested. A key element in the construction of any
treatment scheme is the consideration of which drugs
to use. This requires data on the comparative efficacy
and toxicity of multiple drugs. Unfortunately, many
have never been directly compared, and comparative
trials have often included too few patients to reveal
differences in drug effects (1).
Preferred drugs should be those that provide
maximal efficacy with the least toxicity. Yet both
efficacy and toxicity can be defined in different ways.
Efficacy can be tested over the short term or the long
term. It can take into account the efficacy in all those
who started the therapy or only in those who continue
to take the drug. Evaluation of drug efficacy only in
patients who continue treatment omits information on
those who have dropped out, often because of ineffi-
cacy of the drug. Also, efficacy can be assessed by
combining information from multiple RA outcomes or
on the basis of only a single clinically important
In the same vein, toxicity can be defined several
different ways. It can be viewed as the percentage of
patients who have discontinued treatment because of
side effects, but this approach may miss at least 2 other
facets of toxicity. First, all side effects are not the
same; for example, acute renal failure is usually more
serious than a rash or nausea. Second, only severe
toxicity may affect the desirability of a drug. Drugs
that often cause rash or diarrhea may be modestly
unappealing, but drugs that induce pancytopenia or
proteinuria ought to be avoided, if there are equipotent
alternative therapies.
The goal of the current study was to compare
second-line drugs using clinical trials data, in order to
assess which drugs provide the most efficacy with the
least toxicity. Barring other considerations (e.g., cost,
convenience), those should be the first choices among
second-line drugs.
To provide different perspectives on both efficacy and toxicity, we used 3 different definitions of
each. For efficacy, we selected a composite measure
that incorporates information on improvement in tender joint count, grip strength, and erythrocyte sedimentation rate (ESR). Second, we evaluated improvement in only tender joint count. Finally, we assessed
the proportion of patients in each trial who did nor
drop out of the trial because of drug inefficacy. With
regard to toxicity, we first evaluated the percentage of
patients who dropped out because of drug toxicity.
Also, we modified a recently published toxicity index
(2) and weighted each side effect leading to dropout,
by its severity. Finally, we focused on rates of severe
side effects, consisting mainly of organ toxicities.
We have previously conducted a metaanalysis
evaluating the relative efficacy and toxicity of secondline drugs in RA ( I ) , including auranofin (AUR),
injectable gold, methotrexate (MTX), antimalarial
drugs, D-penicillamine (DP), and sulfasalazine (SSZ).
That metaanalysis evaluated trials published through
August 1989. Data for the current study were based on
an update of our metaanalysis, including trials published through the end of 1990. Also, we added azathioprine (AZA) to the list of drugs studied. With these
additions, 14 trials, including more than 1,000 patients,
were added. These include recent trials of MTX and
SSZ, permitting us to generate more precise estimates
of these drugs’ efficacy and toxicity than before.
The methods of this metaanalysis, including identification of trials, inclusion criteria for selection of trials, and
method of data extraction from the trials, have been described (1). The drugs and minimum dosages were the same
as reported previously. We added AZA at a minimum dosage
of 1 mg/kg/day. All trials published through the end of
December 1990 were included.
Metaanalysis of drug efficacy. As before, to be included in the metaanalysis of drug efficacy, a study must
have had extractable results pertaining to efficacy. Specifically, we required that trial reports contain numerical values
for change in tender joint count (or Ritchie articular index
[3]), ESR, and/or grip strength. Our analysis focused on
treatment arms rather than individual trials. For each trial,
we identified the treatment arms of interest and extracted
those data. We also extracted data for a variety of other
variables that might be related to outcome (including mean
age, mean disease duration, mean initial tender joint count,
length of the study, and its year of publication). We determined that the covariates that were relevant for efficacy
were, for grip strength, the mean disease duration in the
treatment group and the trial length, and for tender joint
count, the initial tender joint count and evaluator blindedness. There were no covariates that affected ESR outcome.
After covariate adjustment, the heterogeneity for each of
these outcomes was nonsignificant ( P > 0.05) as assessed by
the method of DerSimonian and Laird (4).
We computed a combined effect size based on standardized improvements in the tender joint count, ESR, and
grip strength. For each treatment arm, standardized improvement in each of these outcome measures was computed as the change (improvement) in the outcome divided
by the pooled standard deviation at baseline for that particular outcome. Following covariate adjustment (separately
for grip strength and for tender joint count), we computed Z
scores and, for each drug, averaged the Z scores. When data
for 1 or 2 of these outcome measures were absent from a
report, we computed the combined effect size based on
results from the measure(s) that were presented.
We tested 3 different definitions of efficacy. First, we
used the composite measure of efficacy described above.
Second, we evaluated improvement in tender joint count
alone (with covariate adjustment for initial tender joint count
and evaluator blindedness). Fixed effect analyses comparing
the drugs for composite efficacy and tender joint count were
weighted by the number of subjects who completed the trial
in that treatment arm. For the plots, a constant (1.0) was
added to each composite score mean, so that all means were
non-negative. Third, for each treatment group in each trial,
we determined the percentage of patients who had dropped
out because the drug was ineffective. We subtracted this
number from loo%, and the remainder was considered the
percentage nonfailing. Using this definition, we compared
treatments by fixed effects binomial regression, weighted by
number entering the treatment group, and with covariate
adjustment for the mean age of patients in the trial and the
year of publication of the study (the covariates that were
significant predictors of inefficacy-related dropout).
Metaanalysis of drug toxicity. For the toxicity
metaanalysis, methods were identical to those already described (I). To provide different perspectives on drug toxicity in clinical trials, we evaluated 3 different definitions of
toxicity. First, we utilized the definition from our earlier
work, which characterized drug toxicity as the percentage of
patients who dropped out of the trial because of drug
For the second definition, we weighted each toxicity
that caused dropout, using severity weights from a recently
published toxicity index (2) that is based on physician ratings
of the severity of side effects. We modified this index
because it did not weight the severity of mild-to-moderate
laboratory abnormalities and because our preliminary tests
with patients and physicians suggested that the toxicity
rating for diarrhea was excessively high (7.7 on a 10-point
scale); our modifications are explained and shown in Appendix 1. Whenever a patient dropped out because of more than
one toxicity, we chose the most severe toxicity, as designated by the index, as the toxicity that led to the dropout.
The third measure of toxicity was the rate of severe
toxicities experienced by patients during the trials. We
defined a severe toxicity as one with a toxicity index score of
>7 on a 10-point scale. Such toxicity index scores (see
Appendix 1) consisted mostly of organ-based toxicities,
including proteinuria, leukopenia, anemia, elevated liver
function test results (to greater than >1,000 unitdm1 for
serum glutamic oxaloacetic transaminase [SGOT] or serum
glutamic pyruvic transaminase [SGPT]), thrombocytopenia,
and renal failure. Any hospitalization due to drug toxicity
was also characterized as representing severe toxicity. Other
toxicities that were considered severe were wheezing, dyspnea, melena, and jaundice.
Each toxicity outcome was adjusted for the 2 factors
that affected toxicity rates, disease duration and trial length,
and the treatments were compared in a fixed effects analysis
weighted by the number entering the treatment group. We
performed a fixed effects analysis of variance for the toxicity
index outcome, and binomial regression for the 2 toxicity
rate outcomes (dropout and severe toxicity).
Efficacyltoxicity tradeoffs. We plotted each measure
of efficacy versus each measure of toxicity for all drugs, with
efficacy on the vertical axis and toxicity on the horizontal
axis. Drugs closest to the left upper border have maximal
efficacy with the least toxicity. Each drug was plotted as an
oval, with the center of the oval representing the point
estimate of efficacy and toxicity. The upper and lower
bounds of the oval are 1 SEM away from the point estimate
of drug efficacy, while the left and right bounds of the oval
are 1 SEM away from the point estimate of drug toxicity.
The updated metaanalysis with the inclusion of
AZA comprised 79 trials with 147 treatment groups, or
Table 1. Description of studies and patients in the rnetaanalysis
No. of trials
No. of treatment groups
(mean per trial)
No. of patients entering trials
No. of patients completing
trials (%)
Mean no. of patients per
treatment group (range)
Mean no. of weeks of
treatment (range)
Mean no. of weeks to
efficacy evaluation (range)
147 (1.86)
4,904 (75.24)
33.7 (6490)
37.7 (12-104)
26.1 (12-52)
an average of 1.86 per trial (see Table 1). Over 6,500
patients entered these trials, and 4,904 completed
them, for an overall completion rate ofjust over 75%.
Trials lasted, on average, 37.7 weeks, or roughly 9
months. Since our efficacy analyses used 6-month
outcomes when available, the mean number of weeks
to efficacy evaluation was 26.1 (range 12-52).
We identified 15 AZA treatment arms in the
trials, with 278 patients completing this treatment
(Table 2). Some of these trials had been included in our
previous metaanalysis, but the AZA treatment arms
had not been analyzed.
In addition, we identified recently published
and previously unanalyzed trials (see Appendix 2).
Compared with the previous metaanalysis, approximately 100 additional MTX-treated patients and 100
additional SSZ-treated patients were studied, with
several trials added for each of these drugs.
In terms of composite efficacy (Figures 1A-C),
MTX, SSZ, and DP were the 3 most potent drugs and
were approximately equal in efficacy, and a fourth,
Table 2. Number of treatment arms and patients, by drug, in the
updated metaanalysis
No. of
treatment arms
No. of patients
completing treatment
Antimalarial drugs
Petcentage NonfaiEng
Joint Count
Toxidty Index
Toicity Index
Severe Toicity (%)
Figure 1. Plots of efficacy versus toxicity for second-line rheumatoid arthritis drugs. A, Composite efficacy versus dropouts due to toxicity. B, Composite efficacy versus
toxicity index. C, Composite efficacy versus severe toxicity. D,Joint count versus dropouts due to toxicity. E,Joint count versus toxicity index. F, Joint count versus severe
toxicity. G, Percentage nonfailing (calculated as 100 - percent dropping out due to inefficacy) versus dropouts due to toxicity. H, Percentage nonfailing versus toxicity index.
I, Percentage nonfailing versus severe toxicity. (See text for details.) The center of each oval is the best estimate of efficacy and toxicity for each drug. The upper and lower
bounds of the oval are 1 SEM from the point estimate of efficacy. The left and right bounds are 1 SEM from the point estimate of toxicity. AM = antimalarial drugs; AU =
auranofin; AZ = azathioprine; G = injectable gold; MTX = methotrexate; PEN = D-penicillamine; SSZ = sulfasalazine; PL = placebo.
Joint Count
Composite Efficacy
injectable gold, was close in efficacy to these 3.
Scoring lower on efficacy were antimalarial drugs and
AZA. The weakest of the active second-line drugs
evaluated was AUR. The improvement among
placebo-treated patients was minimal,
When we evaluated the ranking of drugs in
terms of improvement in joint count (Figures 1D-F),
we found that the 2 drugs that scored best were MTX
and SSZ. The weakest drugs on this measure were
antimalarial drugs, AUR, and AZA. The scores for DP
and gold were intermediate between those of the
weaker and the stronger drugs. The change in tender
joint count experienced by patients taking placebo was
Our third definition of efficacy was the percentage of subjects entering a trial who did not experience
inefficacy-related dropout. Those taking placebo are
likely to drop out because of inefficacy, and not
surprisingly, the percentage “not failing” was the
lowest for placebo, at -84% (Figures 1G-I). The
second-line drugs had similar scores for percentage
not failing, with the most efficacious drugs by this
definition being MTX, AZA, DP, SSZ, and gold.
Antimalarial drugs scored in an intermediate position,
and the weakest drug from this perspective was AUR.
When we looked at toxicity measures, we focused initially on the percentage of patients who
dropped out because of toxicity (horizontal axis in
Figures l A , D, and G). The drug with the lowest
dropout rate was the group of antimalarial drugs, with
MTX and AUR close behind. The most toxic drug by
this definition was injectable gold.
Next, we looked at the average toxicity index
for each drug based on the severity of each toxicity in
patients who had discontinued the drug because of
side effects (Figures IB, E, and H). The toxicity index
score was highest for injectable gold. Next in order of
decreasing toxicity were DP, SSZ, AZA, MTX, and
AUR. Antimalarial agents scored as the least toxic
drug on the basis of the toxicity index.
Finally, we found overlapping rates of severe
toxicity (Figures lC, F, and I), because the rates of
severe side effects in clinical trials were low, and the
standard errors around our estimates of severe toxicity
were therefore wide. Nonetheless, the rankings of
severe toxicity suggested that DP and gold are closer
to each other on this measure than on the other
toxicity measures, with gold having the highest rate of
severe toxicity. The least toxic with respect to severe
organ effects were the antimalarial drugs. No patients
in clinical trials experienced severe organ toxicity
while receiving antimalarial drugs. Drugs with moderate rates of severe toxicity included AUR, SSZ,
and MTX.
When we plotted different definitions of efficacy
versus toxicity (see Figures 1 A-I), we assessed which
drugs have high efficacy with little toxicity and thus
fell closest to the left upper corner of the plots. In all
plots, the second-line drugs that lay closest to this
corner were the antimalarial drugs and MTX. More
specifically, MTX was consistently one of the strongest drugs and, among the efficacious drugs, had the
most benign toxicity profile. SSZ was also close to the
upper left corner, with relatively high potency and
only modest toxicity (see especially Figures 1A-C).
While antimalarial drugs did not fall among the most
efficacious, their relative safety made their efficacy/
toxicity “ratio” high. Other drugs varied in their
rankings depending upon which measure of efficacy
and toxicity was chosen. Injectable gold remained
consistently among the most toxic medications, while
AUR was consistently among the weakest.
In this study, we evaluated several different
definitions of efficacy and toxicity and plotted these to
assess which second-line drugs conventionally used in
rheumatoid arthritis offer the best tradeoff in terms of
efficacy and toxicity. We found that, regardless of how
we defined efficacy and toxicity, antimalarial drugs
and methotrexate were the best choices. Azathioprine,
newly included in the study, scored in the intermediate
range in terms of both efficacy and toxicity. Injectable
gold was the most toxic drug by 3 different toxicity
measures, and auranofin remained among the weakest
of drugs, regardless of how efficacy was defined.
The short-term treatment time frame we studied
corresponds to those often used in clinical trials, with
efficacy assessed over the initial 6 months of use and
toxicity over the first 9 months. Many important
toxicities (e.g., retinal toxicity with antimalarial drugs)
do not generally occur early in treatment. Nonetheless, long-term observational studies using drug termination as the measure of overall drug appeal ( 5 ) also
indicate that MTX and antimalarial drugs are the 2
agents that patients are the most likely to continue
taking for several years.
If life-threatening adverse reactions occur with
even moderate frequency with any of these second-
line drugs, then such toxicities might affect drug
selection. Specific drugs among those studied have
been reported to cause fatal side effects, although the
precise incidence of these is not known. Examples
include gold-induced thrombocytopenia (6) and pneumonitis (7). Life-threatening side effects due to DP
include bone marrow aplasia, thrombocytopenia (8),
obliterative bronchiolitis (9), and polymyositis (10).
Also, although the frequency is unknown, MTX has
been reported to cause fatal liver complications ( 1 1)
and pneumonitis (12); AZA may induce fatal adverse
reactions, perhaps including secondary malignancies
(13), and SSZ, like other sulfa drugs, may cause lifethreatening bone marrow suppression. The second-line
drugs that are rarely, if ever, associated with fatal
adverse events are antimalarial drugs and AUR.
While our plots of efficacy/toxicity tradeoffs
suggest that MTX and antimalarial drugs are the most
appealing of the second-line drugs, SSZ has a profile
similar to that of MTX. More specifically, our results
place SSZ among the strongest of the traditional
second-line drugs, a finding that may contrast with
general clinical opinion. Our metaanalysis included 5
comparative trials in which SSZ was studied (14-18),
and in all 5 , the average composite improvement with
SSZ treatment was greater than the improvement with
the comparison drug; comparison drugs included injectable gold (18), hydroxychloroquine (17), and DP (3
trials) (14-16).
Because of the large number of interdrug comparisons reported here, we did not compare these
drugs statistically. The number of statistical tests
required to compare 7 active drugs and placebo on 3
different efficacy measures and 3 different toxicity
measures is large, and statistical differences would be
found by chance. Statistical testing was performed in
our earlier metaanalysis in which we evaluated comparative efficacy according to the composite efficacy
measure alone, and toxicity according to the percentage experiencing toxicity-related dropout. AUR was
significantly weaker, and gold had significantly higher
toxicity rates, compared with the other drugs.
Few patients in clinical trials experience severe
organ toxicity. Therefore, our estimates of severe
toxicity rates are based on small numbers of side
effects, resulting in large standard errors, depicted in
Figure 1 as wide ovals. Furthermore, reports of clinical trials often do not include information on the
severity of the organ toxicity that leads to dropout.
Specifically, the degree of SGOT or serum creatinine
abnormality leading to dropout is often not reported.
In such circumstances, we were forced to grade the
laboratory toxicity as indeterminate and scored the
toxicity level as intermediate between mild and
Two toxicities with unspecified severity did not
reach a weighting of 7, liver toxicity and leukopenia.
We performed a sensitivity analysis in which we
classified these also as severe toxicity. When we did
this, the rate of severe toxicity for MTX increased and
lay between the levels for gold and DP. Other drugs
did not change in their rankings. This change for MTX
was due, in large part, to one trial (19), in which MTX
was discontinued and patients dropped out if the
SGOT or SGPT level reached twice the normal limit,
which occurred in 20% of the patients. Other trials did
not adopt this rule, and dropouts related to liver
abnormalities were much less frequent.
An update of our metaanalysis continues to
suggest that there are 4 strong second-line drugs:
injectable gold, DP, MTX, and SSZ. Recent trials of
MTX and SSZ have pushed our estimates of efficacy
for those drugs slightly higher than for the other 2.
AZA, which we had not evaluated before, fell in the
midrange of drug efficacy, at least using our nontraditional metaanalytic method. One reason that AZA may
not score among the strongest drugs is that it has little
effect on the ESR. However, even when we evaluated
the efficacy of AZA on the other outcomes only, it still
did not score among the top 4 drugs.
Our evaluation of drug toxicity corroborated
our previous findings that injectable gold was the most
toxic of these drugs, even though we used a new
method to gauge toxicity, which incorporates more
information on the severity of a given toxicity. While
our rankings of severe toxicity relied on physician
ratings of toxicities, some non-life-threatening side
effects, such as headache or diarrhea, may be severe
to the patient.
Our third measure of efficacy “not failing,” was
based on the percentage of patients who did not drop
out because of inefficacy. As such, it may not measure
efficacy, but rather lack of inefficacy. We believe the
two are strongly correlated, as attested to by the low
rankings of placebo and AUR.
Our metaanalysis did not utilize the traditional
metaanalytic method of subtracting placebo response
from active drug response. Instead, we evaluated the
response for each treatment group within each trial
and averaged these responses across trials, extracting
information from both comparative and placebocontrolled trials. Our method ignores the randomization in each trial that created balanced treatment group
assignments. Instead, we attempted to create comparable groups by adjusting for factors that affected
treatment response. Our success in creating prognostically similar treatment groups was necessarily imperfect, in part because clinical trials data may not
provide all-important prognostic information.
Metaanalysis, in general, remains a highly imperfect way of ranking the efficacy and toxicity of
drugs. Diverse studies are pooled. Also, because reports of trial did not provide intent-to-treat analyses,
we could not perform such analyses. Our toxicity
metaanalysis has additional limitations. Some clinical
trials set rules for patient dropout that may not correspond to rules actually followed by clinicians. Side
effects occurring over the long term were not sampled.
Finally, clinical trials do not necessarily provide an
accurate glimpse of the long-term efficacy and usefulness of second-line drugs in rheumatoid arthritis; the
perspective is necessarily short term.
Despite these limitations, our study provides a
quantitative way to summarize carefully collected
clinical trials data on both the efficacy and toxicity of
RA drugs. Our results parallel those of clinic-based
observational studies that have attempted to rank
these drugs (3,
single large clinical trials in which
differences have been shown (20), and, to a large
degree, clinical impressions of the relative effectiveness of these drugs. In summary, our efficacy/toxicity
tradeoff results suggest that, regardless of how those
parameters are defined, MTX and antimalarial drugs
are, in the short-term context of clinical trials, the
second-line drugs of choice in RA.
We are indebted to Pearl Siew for her secretarial and
technical help.
Felson DT, Anderson JJ, Meenan RF: The comparative
efficacy and toxicity of second-line drugs in rheumatoid
arthritis: results of two metaanalyses. Arthritis Rheum
33: 1449-1461, 1990
Fries JF, Spitz PW, Williams CA, Bloch DA, Singh G,
Hubert HB: A toxicity index for comparison of side
effects among different drugs. Arthritis Rheum 33:121130, 1990
Ritchie DM, Boyle JA, Mclnnes JM, Jasani MK, Dalakos TG, Grieveson P, Buchanan WW: Clinical studies
with an articular index for the assessment of joint
tenderness in patients with rheumatoid arthritis. Q J
Med 147:393406, 1968
DerSimonian R, Laird N: Meta-analysis in clinical
trials. Controlled Clin Trials 7: 177-188, 1986
Wolfe F, Hawley DJ, Cathey MA: Termination of slow
acting antirheumatic therapy in rheumatoid arthritis: a
1Cyear prospective evaluation of 1017 consecutive
starts. J Rheumatol 17:994-1002, 1990
Adachi JD, Bensen WG, Kassam Y , Powers PJ,
Fernando BA, Cividino A, Kean WF, Rooney PJ, Craig
GL, Buchanon WW, Tugwell PX, Gordon DA, Lucarelli
A, Singal DP: Gold induced thrombocytopenia: 12 cases
and a review of the literature. Semin Arthritis Rheum
161287-293, 1987
Levinson ML, Lynch JP 111, Bower JS: Reversal of
progressive, life-threatening gold hypersensitivity pneumonitis by corticosteroids. Am J Med 71:908-912, 1981
O’Brien WM: Toxicity of D-penicillaminein rheumatoid
arthritis. Ann Intern Med 92:120-121, 1980
Wolfe F, Schurle DR, Lin JJ, Polland SM, Smith TW,
Montgomery-Short R, James DL: Upper and lower
airway disease in penicillamine treated patients with
rheumatoid arthritis. J Rheumatol 10:406-409, 1983
Doyle DR, McCurley TL, Sergent JS: Fatal polymyositis in D-penicillamine-treated rheumatoid arthritis. Ann
Intern Med 98:327-330, 1983
Phillips CA, Cera PJ, Mangan TF, Newman ED: Clinical
liver disease in patients with rheumatoid arthritis taking
methotrexate. J Rheumatol 19:229-233, 1992
Searles G, McKendry RJR: Methotrexate pneumonitis
in rheumatoid arthritis: potential risk factors; four case
reports and a review of the literature. J Rheumatol
14:1164-1171, 1987
Kinlen LJ: Incidence of cancer in rheumatoid arthritis
and other disorders after immunosuppressive treatment.
Am J Med 78(suppl lA):44-49, 1985
Neumann VC, Grindulis KA, Hubball S, McConkey B,
Wright V: Comparison between penicillamine and sulphasalazine in rheumatoid arthritis: Leeds-Birmingham
trial. Br Med J 287: 1099-1 102, 1983
Carroll GJ, Will RK, Breidahl PD, Tinsley LM: Sulphasalazine versus penicillamine in the treatment of
rheumatoid arthritis. Rheumatol Int 8:251-255, 1989
Capell HA, Marabani M, Madhok R, Torley H, Hunter
JA: Degree and extent of response to sulphasalazine or
penicillamine therapy for rheumatoid arthritis: results
from a routine clinical environment over a two-year
period. Q J Med 75:335-344, 1990
17. Nuver-Zwart IH, van Riel PLCM, van de Putte LBA,
Gribnau FWJ: A double blind comparative study of
sulphasalazine and hydroxychloroquine in rheumatoid
arthritis: evidence of an earlier effect of sulphasalazine.
Ann Rheum Dis 48:389-395, 1989
18. Williams HJ, Ward JR, Dahl SL, Clegg DO, Willkens
RF, Oglesby T , Weisman MH, Schlegel S, Michaels
RM, Luggen ME, Polisson RP, Singer JZ, Kantor SM,
Shiroky JB, Small RE, Gomez MI, Reading JC, Egger
MJ: A controlled trial comparing sulphasalazine, gold
sodium thiomalate, and placebo in rheumatoid arthritis.
Arthritis Rheum 31:702-713, 1988
19. Williams HJ, Willkens RF, Samuelson CO Jr, Alarc6n
GS, Guttadauria M , Yarboro C, Polisson RP, Weiner
SR, Luggen ME, Billingsley LM, Dahl SL, Egger MJ,
Reading JC, Ward JR: Comparison of low-dose oral
pulse methotrexate and placebo in the treatment of
rheumatoid arthritis: a controlled clinical trial. Arthritis
Rheum 28:721-730, 1985
20. Weinblatt ME, Kaplan H, Germain BF, Merriman RC,
Solomon SD, Wall B, Anderson L, Block S, Irby R,
Wolfe F, Gall E, Torretti D, Biundo J, Small R, Coblyn
J , Polisson R: Low-dose methotrexate compared with
auranofin in adult rheumatoid arthritis: a thirty-sixweek, double-blind trial. Arthritis Rheum 33:330-338,
Class of toxicity
Liver toxicity
Rise in creatinine
Mild toxicity
Hematocrit at least 5 volume %
less than usual for patient,
but 225 volume %
WBC decreased by 50%, but
remaining r l ,500/mm3
Platelets <150,000/mm3, but
SGOT or SGPT 2-3 times
normal, but ~ 1 , 0 0 0unitsiml
Proteinuria I + or 2+ on
dipstick; >500 mg124 hours,
but 5 2 gm/24 hours
Serum creatinine 50% above
usual level for patient, but
5 5 mgidl
Weight for
Severe toxicity
Weight for Weight for
severe unspecified
Hematocrit <25
volume %
WBC < I ,500/mm3
Platelets <100,000/mm3
SGOT > 1.000 unitdm1
Proteinuria s 3 + or
>2 gm/24 hours
Creatinine >5.0 mgidl
* The weights for “mild” laboratory toxicities were determined by surveying 17 board-certified rheumatologists by
mail. As part of the survey, weights for severe toxicities and for symptom toxicities from the previously published
index (ref. 2) were provided. Reports of clinical trials often did not provide data on the severity of laboratory side
effects. To calculate weights for laboratory toxicities of unspecified severity, the frequencies of mild and severe
toxicities for that side effect were examined and the unspecified number was a frequency-weighted average of mild
and severe toxicities [(wtoxm x ptoxm) + (wtoxs x ptoxs) = wtoxu], where wtoxm = weight of mild toxicity: wtoxs
= weight of severe toxicity; and ptoxm and ptoxs = proportions of patients with each, where ptoxm + ptoxs = I.
This rule was altered if the total sum of ptoxm and ptoxs was <10 or if no cases of mild or severe toxicity were
recorded in any trial. In this case, wtoxm was increased by 25% of the difference between mild and severe, or wtoxs
was lowered by this much, depending on whether mild or severe toxicity predominated. If no cases of mild or severe
toxicity were recorded, wtoxu was the unweighted mean of wtoxm and wtoxs.
Weight for diarrhea was reassessed because we believed the published weight (7.7 of a possible 10, higher than the
weights for nausea L3.91, vomiting (6.51, upper abdominal pain [5.8], and rash (4.71) was too high. We separately
surveyed 10 postgraduate physicians, providing weights for other symptom toxicities (e.g., nausea, vomiting) and
asked the respondents to assign an appropriate weight for diarrhea. The mean weight assigned by this group was 4.21.
Values shown are the means. Other weights used in this study but not reported here were taken from the published
index (ref. 2). WBC = white blood cell count: SGOT = serum glutamic oxaloacetic transaminase; SGPT = serum
glutamic pyruvic transaminase.
(For list of trials included in both the updated
metaanalysis and the earlier metaanalysis, see ref. 1. in the
main list of references [Arthritis Rheum 33:144%1461,
1. Urowitz MB, Gordon DA, Smythe HA, Pruzanski W, Ogryzlo
MA: Azathioprine in rheumatoid arthritis: a double-blind, crossover study. Arthritis Rheum 16:411-418, 1973
2. Woodland J, Chaput de Saintonge DM, Evans SJW, Sharman
VL, Currey HLF: Azathioprine in rheumatoid arthritis: doublebind study of full versus half doses versus placebo. Ann Rheum
Dis 40:355-359, 1981
3. F@rre 0,Bjerkhoel F , Salvesen CF, Berg KJ, Rugstad HE,
Saelid G, Mellbye OJ, KBss E: An open, controlled, randomized
comparison of cyclosporine and azathioprine in the treatment of
rheumatoid arthritis: a preliminary report. Arthritis Rheum
30:88-92, 1987
4. Levy J , Paulus HE, Bangert R: Comparison of azathioprine and
cyclophosphamide in the treatment of rheumatoid arthritis
(abstract). Arthritis Rheum W412-413, 1975
5. Levy J, Paulus HE, Barnett EV, Sokoloff M, Bangert R,
Pearson CM: A double-blind controlled evaluation of azathioprine treatment in rheumatoid arthritis and psoriatic arthritis
(abstract). Arthritis Rheum 15: 116-1 17, 1972
6. Carroll GJ, Will RK, Breidahl PD, Tinsley LM: Sulphasalazine
versus penicillamine in the treatment of rheumatoid arthritis.
Rheumatol Int 8:251-255, 1989
7. Weinblatt ME, Kaplan H, Germain BF, Merriman RC, Solomon
SD, Wall B, Anderson L , Block S, Irby R, Wolfe F, Gall E ,
Torretti D, Biundo J , Small R, Coblyn J , Polisson R: Low-dose
methotrexate compared with auranofin in adult rheumatoid
arthritis: a thirty-six-week, double-blind trial. Arthritis Rheum
33:330-338, 1990
Arnold MH, O’Callaghan J, McCredie M, Beller EM, Kelly DE,
Brooks PM: Comparative controlled trial of low-dose weekly
methotrexate versus azathioprine in rheumatoid arthritis: 3-year
prospective study. Br J Rheumatol 29:120-125, 1990
Urowitz MB, Hunter T, Bookman AAM, Gordon DA, Smythe
HA, Ogryzlo MA: Azathioprine in rheumatoid arthritis: a
double-blind study comparing full dose t o half dose. J Rheumato1 1:274-281, 1974
Capell HA, Marabani M, Madhok R, Torley H, Hunter JA:
Degree and extent of response to sulphasalazine or penicillamine therapy for rheumatoid arthritis: results from a routine
clinical environment over a two-year period. Q J Med 75:335344, 1990
Scott DL, Greenwood A, Davies J , Maddison PJ, Maddison
MC, Hall ND: Radiological progression in rheumatoid arthritis:
d o D-penicillamine and hydroxychloroquine have different effects? Br J Rheumatol 29: 126-127, 1990
Johnsen V, Borg G, Trang L E , Berg E , Brodin U: Auranofin
(SK&F) in early rheumatoid arthritis: results from a 24-month
double-blind, placebo-controlled study. Scand J Rheumatol
18:251-260, 1989
Verstraeten A, Sileghem A, Dequeker J: OM-8980 and Dpenicillamine in the treatment of rheumatoid arthritis. Scand J
Rheumatol 19:422431, 1990
Mason M, Currey HLF, Barnes CG, Dunne JF, Hazleman BL,
Strickland ID: Azathioprine in rheumatoid arthritis. Br Med J
1:420-422, 1969
Без категории
Размер файла
810 Кб
second, short, metaanalysis, select, tradeoffs, line, drug, terms, clinical, arthritis, trials, use, efficacytoxicity, rheumatoid, publisher
Пожаловаться на содержимое документа