- Email: [email protected]
Use of radiographs to measure outcome in rheumatoid arthritis
Use of Radiographs to Measure Outcome in Rheumatoid Arthritis
MICHAEL
H. WEISMAN,
San Diego,
California
Currently, radiographic analysis of the hands and wrists provides the most practical, useful, and objective biologic endpoint for measurement of outcome in rheumatoid arthritis. Semiquantitative scoring methods for cartilage loss and osseous erosions developed by Sharp (Arthritis Rheum 1971; 14: 706-720) and Larsen (Stand J Flheum 1973; 2: 136-138) have established standards for sensitivity and interobserver reliability. Sharp (Arthritis Rheum 1985; 28: 1624) has shown that in order to reliably measure change in erosion score, a certain degree of worsening (i.e., progression) must take place. At least one year is necessary, preferably two. Others have shown that rheumatoid arthritis erosions take place and progress early (Years 1 to 3) in most patients, and once destruction is established it may not be possible to adequately assess further change radiographically. Therefore, in order to perform a clinical trial of “disease modifying” agents with radiographic analysis as endpoint, the selections of a proper study population and time interval are most critical. Although it may be difficult to perform, an ideal clinical study for a therapeutic intervention would use patients with rheumatoid arthritis of less than five years’ disease duration and perform radiographic analysis at a 24-month minimum interval. The selection of a proper control group may be a near impossibility under these circumstances.
M.D.
From the Division of Rheumatology, Department of Medicine, University of California San Diego School of Medicine, San Diego, California. Requests for reprints should be addressed to Dr. Michael H. Weisman, UCSD Medical Center H-81 1 -G, 225 Dickinson Street, San Diego, California 92103.
96
October
30, 1967
The American
Journal
The measurement of disease outcome in rheumatoid arthritis has been and continues to be one of the most vexing and challenging problems rheumatologists face. For any disease with a host of striking clinical manifestations often containing major components of either functional loss or gross anatomic deformities, it would be expected that measurement of change would be an easy task. Frequently discussed and cited to underscore this apparent dilemma in rheumatoid arthritis are highly variable clinical courses, spontaneous remissions, or mistaken diagnoses. Yet anyone who takes care of rheumatoid arthritis patients surely knows many patients whose disease is relentless and progressive. It is plainly obvious to any observer that the radiographic examination reveals an unfortunate but accurate image of the patient’s problem. The use of radiographic techniques to measure the progression of osseous erosions has been the standard for outcome measurement of rheumatoid arthritis. Therefore, clinical decision making as well as therapeutic trials have relied upon the assumption that regardless of symptom relief, if therapy
does
not
change
progressive
joint
destruction,
it really
has
not
affected the basic disease process. The term, disease modifying antirheumatic agent, had its inception from this notion of an agent that truly
of Medicine
Volume
83
(suppl 46)
SYMPOSIUM
of these measurements for clinical trials of therapeutic agents have not always gone smoothly. Despite many refinements of these techniques and experimental as well as clinical observations over the years, there remain several unanswered questions and a few serious dilemmas regarding the satisfactory employment of radiographic analysis to the outcome issue. The evolution of radiographic methods to assess osseous change has recently been reviewed in detail. Genant [I] concluded that fine-detailed radiography in rheumatoid arthritis, with optical magnification, provides the most consistent and reliable technique to evaluate articular abnormalities. Sharp et al [2] and Larsen [3] in the early 1970s developed in parallel the most widely used methods for semiquantitative scoring of radiographic changes in the hands and wrists of patients with rheumatoid arthritis. The Larsen method employs a composite evaluation scale that includes erosion, joint space narrowing, and osteoporosis based upon reference films; when these methods are compared head-to-head (Sharp versus Larsen), there is little difference to choose between them [4]. Nevertheless, the use of these techniques from their inception raised several questions that have subsequently been addressed and largely answered in the decade and a half that followed: (1) When and where do erosions first occur? (2) What is the rate of development of erosive disease in the hands and wrists? (3) What is the correlation of hand erosions to either functional status or large joint disease? (4) How reliable and sensitive are the different methods to detect meaningful changes in a clinical trial? (5) Are there refinements of these techniques to make them less expensive and more widely applicable to studies of therapeutic agents? Several important problems remain unanswered or unaddressed, however. These questions concern the “healing” or regression of erosions, the differing sensitivities of radiologists and rheumatologists to detect normal from abnormal, and the contribution of small variations in the actual technique of film-taking such as slight differences in rotation. In addition, there still exists the everpresent question of whether there is a threshold effect of joint damage beyond which it is technically impossible to assign scores of additional osseous change. Further, the process of erosive damage may begin with multiple small erosions that later coalesce, the number may diminish, and the end result may produce a single large defect. How are these difficulties overcome? Brook and his associates [5,6] have attempted to deal with the first of these issues. In two prospective studies, they followed almost 100 patients from disease onset for a mean of five years. Erosions in many cases developed independently from joint space loss. This observation has supported the notion that the two anatomic findings of osseous defects and cartilage loss are not simply a continuum or different phases of the same process. There-
October
30, 1987
ON NABUMETONE-WEISMAN
fore, these two items should be considered separately in any radiographic analysis. This observation is also consistent with the proposed pathogenetic mechanisms for joint destruction in rheumatoid arthritis; discrete steps or sequences (some of which involve cells that produce compounds with bone destructive potential) may be involved [7]. These investigators also observed that foot or metatarsal head erosions occurred earlier than erosions in the hands and were most frequently noted overall. Most important, however, was the finding that over the period of observation (onset to five years), in 71.3 percent of patients, “diagnostic” erosions developed but 23.4 percent had no erosive disease. Of those in whom erosions developed, 90 percent did so within two years of disease onset. Half of the erosive cases progressed and half remained static during the five years of follow-up. Not surprisingly, the severity of the erosions at outset correlated with a poor outcome at follow-up. Therefore, it can be assumed from these carefully done studies with a radiographic examination performed every six months that if a patient is going to get erosions they will do so early, and their subsequent course is charted as static versus progressive within the first two years. For comparison, Masi and coworkers [8] observed similar clinical findings from earlydiagnosed rheumatoid arthritis. They concluded that different clinical patterns emerged early and remained established or static within the early (defined as five years or less) course of the disease. Hence, the radiographic observations of Brook et al [5,6] appear to have a clinical correlation from the studies of Masi et al [8]. A number of extensive studies by Scott and co-workers [9-l 21 attempted to increase the accuracy and assess the validity of using hand and wrist radiograms as an assessment tool for disease outcome. They confirmed the earlier findings of others [13,14] that the rate of joint damage was greatest in early years with a gradual reduction in the speed of progression occurring after five years. Not surprisingly, a poor correlation was observed between degree of erosion and hand function. Similarly, a lack of correlation between radiographic findings of hand damage and other large joint destruction was observed. The authors conclude that the rate of change in radiographic results may be unique to an individual patient and is not constant over time. They also observed that control of the acute synovitis as manifest by joint swelling, tenderness, or by acute phase reactants (erythrocyte sedimentation rate, C-reactive protein) may take place without an apparent change in erosion progression. This apparent dichotomy between control of clinical variables and progression of erosions becomes less apparent as patients are followed beyond six months or one year. However, the studies from Scott and co-workers [9-l 21 do not have a suitable control group for comparison to allow for any conclusions to be drawn about intervention. Nevertheless,
The
American
Journal
of Medicine
Volume
83
(suppl 48)
97
SYMPOSIUM
TABLE
l l
l
l l l l l
ON NABUMETONE-WEISMAN
I
Tentative Conclusions Rheumatoid Arthritis
Reached by This Author Clinical Trials
about
to Measure
Outcome
in
Interval between films probably should be at least one year, preferably two, in order to provide measurable differences Methods for scoring changes appear not to be critical as long as certain well-established principles are followed (i.e., number of readers, experience, etc.) Radiologists may be better suited to provide radiographic assessments. They are more qualified to detect normal variants and other minor variations Disagreement exists regarding blinding films to sequence to avoid bias Number of new erosions is easier to assess than progression of well-established disease; some patients never get erosions Early disease (Years 1-5) may provide the best window from a methodologic standpoint to measure change in osseous erosions There may be a “threshold” effect of disease progression radiologically beyond which it may not be possible to assess change Selection of a proper study and control population appears to be more critical than any other factor in determining the validity of a radiographic outcome trial
ing rates of deterioration of erosions despite disease control, performing a power calculation led these authors to conclude that 120 patients are needed in each arm of an active versus placebo trial for two years to disclose a significant difference. If patient withdrawal from the study was expected because of the specific agent used, perhaps three or four times this amount of patients would be required. It is not possible, according to the authors, to maintain a placebo group for the length of time necessary (at least one year or perhaps two years) in order to document a real difference in erosion progression. Since benefit is expected at six months from most of these agents, factors will come into play including.expectations of alternate treatment after six months to create a differential patient-withdrawal rate from both active and placebo arms of any study to greatly contaminate both the active as well as the comparison placebo group. It is not without irony, the authors suggest [18], that the positive effects expected by patients and physicians from “second-line” therapeutic agents in the medium-term situation may represent an obstacle to performing long-term assessments for two years and beyond. It is worth noting that at a time when medium-term options were far fewer or less available, the Medical Research Council of Great Britain [20,21] was able to give maintenance therapy to patients with rheumatoid arthritis using prednisolone 20 mg per day for two years and retard progression of rheumatoid arthritis as evidenced by radiographic results in comparison with an analgesic or aspirin group. Although radiographic analysis of the hand and wrist as an outcome measure for clinical trials in rheumatoid arthritis is a much maligned concept, it is clear that there is little else to replace it at the present time. It has been suggested [15] that in comparison with other alternatives for outcome measures, radiographic evidence of abnormalities is a true biologic endpoint and can be objectively assessed, probably will precede functional loss, and likely remain independent from social or personal forces that influence disability. It is also evident that the methods do not need further refinement or testing. Furthermore, the
Scott’s studies reinforce caution and concern about the interpretation of osseous erosions in the hand and wrist as a simple outcome measure. Two very energetic and carefully done studies by Sharp et al [4] and Fries and co-workers [15] attempted to finetune the reproducibility and reliability as well as many technical/methodologic aspects concerning the use of serial radiographs for clinical trials. Inter-observer variability was generally noted to be very good, yet a standardization of scores was necessary in order to compare results from different readers since readers have differing levels of sensitivity. The analysis of sequential films from the same patient established a quantification for what heretofore had been largely an intuitive concept: in order for films to be reliably characterized by the different readers as progressive, an increase in osseous erosion standardized score of greater than 15 units was required. These 15 units translated into disease duration of one year or more. Hence, these investigators established a limit of reproducibility for radiologic observations that are necessary to reliably characterize erosions as progressive. Fries’ [15] studies carried these observations a step further and made very specific recommendations about the use of techniques to eliminate time without sacrificing information and reiterated a reminder that trained, experienced readers are extremely important for accurate assessments of disease progression. Despite methodologic refinements and an extraordinary amount of experience using hand and wrist radiograms in clinical trials, the published data providing evidence that drug therapy can retard radiographic erosions in rheumatoid arthritis is very weak [I 61. Of 17 studies with a suitable control group and a rigorous methodologic approach, only two showed strong support for disease control by any intervention. Provocative questions concerning this failure of documentation by radiographic analysis have been provided by Pullar and co-workers from Glasgow [17-191. These investigators suggest that radiologic confirmation of a therapeutic effect caused by a “second-line” agent may be an unattainable goal. From their own data show-
98
the Use of Radiographs
October
30, 1987
The
American
Journal
of Medicine
Volume
83
(suppl 4B)
SYMPOSIUM
greatest obstacle to their satisfactory employment in clinical trials will be the design of the study itself and the proper selection of patients and controls for analysis. From the data and speculations already discussed; this author has reached several conclusions concerning the use of radiographic examination to measure outcome in rheumatoid arthritis clinical trials (Table I). The optimum interval between radiographic films, in order to reliably assess real change, should be at least 12 months and very likely should be 24 months. Experienced readers, preferably three of them, should be employed and radiologists may be better suited to interpret the minor alterations and normal variants that occur in small bones of the hand. It remains to be determined whether it is better for a skilled skeletal radiologist to view sequential films and quickly assess the progress of rheumatoid arthritis or to rely upon the more tedious methods of radiologic evaluation of the hands and wrists. Disagreement still exists as to whether it is necessary to prevent bias by blinding films as to sequence when multiple readers are performing assessments. It seems clear that early disease (less than five years) may be easier to measure when we apply radiographic tools. This is not to say that our drugs do not work in established or persistent disease; rather, our measurements are less useful or accurate under those circumstances. It may be suggested in a schematic model (Figure 1) that there is a window of disease duration through which our currently available measurements provide the sharpest focus for determining whether osseous erosions may be influenced by therapy. In addition, the selection of the population for study appears to be the most critical factor for a radiographic outcome therapeutic trial. Limitations set by the selection of a suitable control group have been discussed and indeed are imposing, but not, hopefully, impossible to overcome.
ON NABUMETONE-WEISMAN
1 Radiographic Progression
I-l Disease
Duration
Figure 1. Schematic model for radiographic cieterminatioi of progression in rheumatoid arthritis. Time interval for the steep rise in erosion score would be the optimum period for an intervention trial.
Finally, it remains to be determined if the radiogram is actually depicting the clinical results of our current therapeutic efforts. At least two possibilities exist: (1) the radiogram is simply out of phase and lags behind adequate control of clinical variables and, hence, we need more time with the drugs we have; or (2) we do not yet have the right agent to affect the critical pathogenetic step or steps in bone and cartilage destruction. Hopefully, the answers will be forthcoming. ACKNOWLEDGMENT I am much appreciative of the inspiration given by John Hatter, M.D., who provided the stimulus for this analysis. Gratitude is also expressed to Don Resnick, M.D., Jose Guerra, M.D., and Nathan Zvailfer, M.D., for many helpful discussions, and to Denise Smith for her expert typing skills.
REFERENCES 1. 2.
3. 4.
5. 6.
7.
Genant HK: Methods of assessing radiographic change in rheumatoid arthritis. Am J Med 1983; 75 (6A): 35-47. Sharp JT, Lidsky MD, Collins LS, Moreland J: Methods of scoring the progression of radiologic changes in rheumatoid arthritis. Arthritis Rheum 1971; 14: 706-720. Larsen A: Radiological grading of rheumatoid arthritis. Stand J Rheum 1973; 2: 1366138. Sharp JT, Bluhm GB, Brook A, et al: Reproducibility of multiple observer scoring of radiologic abnormalities in the hands and wrists of patients with rheumatoid arthritis. Arthritis Rheum ‘1985; 28: 16-24. Brook A, Corbett M: Radiographic changes in early rheumatoid disease. Ann Fiheum Dis 1977; 36: 71-73. Brook A, Fleming A, Corbett M: Relationship of radiological change to clinical outcome in rheumatoid arthritis. Ann Rheum Dis 1977; 36: 274-275. Zvaifler NJ: Overview of etiology and pathogenesis. In: Utsinger PD, Zvaifler NJ, Ehrlick GE, eds. Rheumatoid arthritis. Philadelphia: Lippincott, 1985; 151-l 58.
October
30, 1987
8.
9.
10.
11.
12.
13. 14.
The
Masi AT, Feigenhaum SL, Kaplan SB: Articular patterns in the early course of rheumatoid arthritis. Am J Med 1983; 75 (6A): 16-26. Scott DL, Grindulis KA, Struthers GR, Coulton BL, Popert AJ, Bacon PA: Progression of radiological changes in rheumatoid arthritis. Ann Rheum Dis 1984; 43 (1): B-17. Scott DL, Coulton BL, Bacon PA, Popert AJ: Methods of x-ray assessment in rheumatoid arthritis: a re-evaluation. Br J Rheumatol 1985; 24: 31-39. Scott DL, Coulton DL, Popert AJ: Long term progression of joint damage in rheumatoid arthritis. Ann Rheum Dis 1986; 45: 373-378. Dawes PT, Fowler PD, Jackson R, et al: Prediction of progressive joint damage in patients with rheumatoid arthritis viewing gold or d-penicillamine therapy. Ann Rheum Dis 1986; 45: 945-949. Sharp JT: Radiographic evaluations of the course of articular disease. Clin Rheum Dis 1983; 9: 541-557. De Carvalho A, Grandal H, Jorgensen B: Radiologic evaluation
American
Journal
of Medicine
Volume
83
(suppl 4B)
99
SYMPOSIUM
15.
16.
17.
18.
100
ON NABUMETONE-WEISMAN
of the progression of rheumatoid arthritis. Acta Radio1 [Diagn] (Stockh) 1980; 21: 115-121. Fries JF, Block DA, Sharp JT, McShane DJ, Spitz P, et al: Assessment of radiologic progression in rheumatoid arthritis: a randomized, controlled trial. Arthritis Rheum 1986; 29: 1-9. lannuzzi L, Dawson N, Zein N, Kushner I: Does drug therapy slow radiographic deterioration in rheumatoid arthritis? N Engl J Med 1983; 309 (17): 1023-1028. Pullar T, Hunter JA, Capell HA: Does second-line therapy affect the radiological progression of rheumatoid arthritis? Ann Rheum Dis 1984; 43 (1): 18-23. Pullar T, Capell HA: A rheumatological dilemma: Is it possible to
October
30, 1987
The
American
Journal
of Medicine
19.
20.
21.
Volume
alter the course of rheumatoid arthritis? Can we answer the question? Ann Rheum Dis 1985; 44: 134-140. Pullar T, Capell HA: Can treatment really influence the radiological progression of rheumatoid arthritis? Br J Rheum 1986; 25: 2-6. Joint Committee of the Medical Research Council and the Nuffield Foundation: A comparison of prenisolone with aspirin or other anaelgesics in the treatment of rheumatoid arthritis. Ann Rheum Dis 1959; 18: 173-186. Joint Committee of the Medical Research Council and the Nuffield Foundation: A comparison of prenisolone with aspirin or other analgesics in the treatment of rheumatoid arthritis. Ann Rheum Dis 1960; 19: 331-337.
83 (suppl
48)
MICHAEL
H. WEISMAN,
San Diego,
California
Currently, radiographic analysis of the hands and wrists provides the most practical, useful, and objective biologic endpoint for measurement of outcome in rheumatoid arthritis. Semiquantitative scoring methods for cartilage loss and osseous erosions developed by Sharp (Arthritis Rheum 1971; 14: 706-720) and Larsen (Stand J Flheum 1973; 2: 136-138) have established standards for sensitivity and interobserver reliability. Sharp (Arthritis Rheum 1985; 28: 1624) has shown that in order to reliably measure change in erosion score, a certain degree of worsening (i.e., progression) must take place. At least one year is necessary, preferably two. Others have shown that rheumatoid arthritis erosions take place and progress early (Years 1 to 3) in most patients, and once destruction is established it may not be possible to adequately assess further change radiographically. Therefore, in order to perform a clinical trial of “disease modifying” agents with radiographic analysis as endpoint, the selections of a proper study population and time interval are most critical. Although it may be difficult to perform, an ideal clinical study for a therapeutic intervention would use patients with rheumatoid arthritis of less than five years’ disease duration and perform radiographic analysis at a 24-month minimum interval. The selection of a proper control group may be a near impossibility under these circumstances.
M.D.
From the Division of Rheumatology, Department of Medicine, University of California San Diego School of Medicine, San Diego, California. Requests for reprints should be addressed to Dr. Michael H. Weisman, UCSD Medical Center H-81 1 -G, 225 Dickinson Street, San Diego, California 92103.
96
October
30, 1967
The American
Journal
The measurement of disease outcome in rheumatoid arthritis has been and continues to be one of the most vexing and challenging problems rheumatologists face. For any disease with a host of striking clinical manifestations often containing major components of either functional loss or gross anatomic deformities, it would be expected that measurement of change would be an easy task. Frequently discussed and cited to underscore this apparent dilemma in rheumatoid arthritis are highly variable clinical courses, spontaneous remissions, or mistaken diagnoses. Yet anyone who takes care of rheumatoid arthritis patients surely knows many patients whose disease is relentless and progressive. It is plainly obvious to any observer that the radiographic examination reveals an unfortunate but accurate image of the patient’s problem. The use of radiographic techniques to measure the progression of osseous erosions has been the standard for outcome measurement of rheumatoid arthritis. Therefore, clinical decision making as well as therapeutic trials have relied upon the assumption that regardless of symptom relief, if therapy
does
not
change
progressive
joint
destruction,
it really
has
not
affected the basic disease process. The term, disease modifying antirheumatic agent, had its inception from this notion of an agent that truly
of Medicine
Volume
83
(suppl 46)
SYMPOSIUM
of these measurements for clinical trials of therapeutic agents have not always gone smoothly. Despite many refinements of these techniques and experimental as well as clinical observations over the years, there remain several unanswered questions and a few serious dilemmas regarding the satisfactory employment of radiographic analysis to the outcome issue. The evolution of radiographic methods to assess osseous change has recently been reviewed in detail. Genant [I] concluded that fine-detailed radiography in rheumatoid arthritis, with optical magnification, provides the most consistent and reliable technique to evaluate articular abnormalities. Sharp et al [2] and Larsen [3] in the early 1970s developed in parallel the most widely used methods for semiquantitative scoring of radiographic changes in the hands and wrists of patients with rheumatoid arthritis. The Larsen method employs a composite evaluation scale that includes erosion, joint space narrowing, and osteoporosis based upon reference films; when these methods are compared head-to-head (Sharp versus Larsen), there is little difference to choose between them [4]. Nevertheless, the use of these techniques from their inception raised several questions that have subsequently been addressed and largely answered in the decade and a half that followed: (1) When and where do erosions first occur? (2) What is the rate of development of erosive disease in the hands and wrists? (3) What is the correlation of hand erosions to either functional status or large joint disease? (4) How reliable and sensitive are the different methods to detect meaningful changes in a clinical trial? (5) Are there refinements of these techniques to make them less expensive and more widely applicable to studies of therapeutic agents? Several important problems remain unanswered or unaddressed, however. These questions concern the “healing” or regression of erosions, the differing sensitivities of radiologists and rheumatologists to detect normal from abnormal, and the contribution of small variations in the actual technique of film-taking such as slight differences in rotation. In addition, there still exists the everpresent question of whether there is a threshold effect of joint damage beyond which it is technically impossible to assign scores of additional osseous change. Further, the process of erosive damage may begin with multiple small erosions that later coalesce, the number may diminish, and the end result may produce a single large defect. How are these difficulties overcome? Brook and his associates [5,6] have attempted to deal with the first of these issues. In two prospective studies, they followed almost 100 patients from disease onset for a mean of five years. Erosions in many cases developed independently from joint space loss. This observation has supported the notion that the two anatomic findings of osseous defects and cartilage loss are not simply a continuum or different phases of the same process. There-
October
30, 1987
ON NABUMETONE-WEISMAN
fore, these two items should be considered separately in any radiographic analysis. This observation is also consistent with the proposed pathogenetic mechanisms for joint destruction in rheumatoid arthritis; discrete steps or sequences (some of which involve cells that produce compounds with bone destructive potential) may be involved [7]. These investigators also observed that foot or metatarsal head erosions occurred earlier than erosions in the hands and were most frequently noted overall. Most important, however, was the finding that over the period of observation (onset to five years), in 71.3 percent of patients, “diagnostic” erosions developed but 23.4 percent had no erosive disease. Of those in whom erosions developed, 90 percent did so within two years of disease onset. Half of the erosive cases progressed and half remained static during the five years of follow-up. Not surprisingly, the severity of the erosions at outset correlated with a poor outcome at follow-up. Therefore, it can be assumed from these carefully done studies with a radiographic examination performed every six months that if a patient is going to get erosions they will do so early, and their subsequent course is charted as static versus progressive within the first two years. For comparison, Masi and coworkers [8] observed similar clinical findings from earlydiagnosed rheumatoid arthritis. They concluded that different clinical patterns emerged early and remained established or static within the early (defined as five years or less) course of the disease. Hence, the radiographic observations of Brook et al [5,6] appear to have a clinical correlation from the studies of Masi et al [8]. A number of extensive studies by Scott and co-workers [9-l 21 attempted to increase the accuracy and assess the validity of using hand and wrist radiograms as an assessment tool for disease outcome. They confirmed the earlier findings of others [13,14] that the rate of joint damage was greatest in early years with a gradual reduction in the speed of progression occurring after five years. Not surprisingly, a poor correlation was observed between degree of erosion and hand function. Similarly, a lack of correlation between radiographic findings of hand damage and other large joint destruction was observed. The authors conclude that the rate of change in radiographic results may be unique to an individual patient and is not constant over time. They also observed that control of the acute synovitis as manifest by joint swelling, tenderness, or by acute phase reactants (erythrocyte sedimentation rate, C-reactive protein) may take place without an apparent change in erosion progression. This apparent dichotomy between control of clinical variables and progression of erosions becomes less apparent as patients are followed beyond six months or one year. However, the studies from Scott and co-workers [9-l 21 do not have a suitable control group for comparison to allow for any conclusions to be drawn about intervention. Nevertheless,
The
American
Journal
of Medicine
Volume
83
(suppl 48)
97
SYMPOSIUM
TABLE
l l
l
l l l l l
ON NABUMETONE-WEISMAN
I
Tentative Conclusions Rheumatoid Arthritis
Reached by This Author Clinical Trials
about
to Measure
Outcome
in
Interval between films probably should be at least one year, preferably two, in order to provide measurable differences Methods for scoring changes appear not to be critical as long as certain well-established principles are followed (i.e., number of readers, experience, etc.) Radiologists may be better suited to provide radiographic assessments. They are more qualified to detect normal variants and other minor variations Disagreement exists regarding blinding films to sequence to avoid bias Number of new erosions is easier to assess than progression of well-established disease; some patients never get erosions Early disease (Years 1-5) may provide the best window from a methodologic standpoint to measure change in osseous erosions There may be a “threshold” effect of disease progression radiologically beyond which it may not be possible to assess change Selection of a proper study and control population appears to be more critical than any other factor in determining the validity of a radiographic outcome trial
ing rates of deterioration of erosions despite disease control, performing a power calculation led these authors to conclude that 120 patients are needed in each arm of an active versus placebo trial for two years to disclose a significant difference. If patient withdrawal from the study was expected because of the specific agent used, perhaps three or four times this amount of patients would be required. It is not possible, according to the authors, to maintain a placebo group for the length of time necessary (at least one year or perhaps two years) in order to document a real difference in erosion progression. Since benefit is expected at six months from most of these agents, factors will come into play including.expectations of alternate treatment after six months to create a differential patient-withdrawal rate from both active and placebo arms of any study to greatly contaminate both the active as well as the comparison placebo group. It is not without irony, the authors suggest [18], that the positive effects expected by patients and physicians from “second-line” therapeutic agents in the medium-term situation may represent an obstacle to performing long-term assessments for two years and beyond. It is worth noting that at a time when medium-term options were far fewer or less available, the Medical Research Council of Great Britain [20,21] was able to give maintenance therapy to patients with rheumatoid arthritis using prednisolone 20 mg per day for two years and retard progression of rheumatoid arthritis as evidenced by radiographic results in comparison with an analgesic or aspirin group. Although radiographic analysis of the hand and wrist as an outcome measure for clinical trials in rheumatoid arthritis is a much maligned concept, it is clear that there is little else to replace it at the present time. It has been suggested [15] that in comparison with other alternatives for outcome measures, radiographic evidence of abnormalities is a true biologic endpoint and can be objectively assessed, probably will precede functional loss, and likely remain independent from social or personal forces that influence disability. It is also evident that the methods do not need further refinement or testing. Furthermore, the
Scott’s studies reinforce caution and concern about the interpretation of osseous erosions in the hand and wrist as a simple outcome measure. Two very energetic and carefully done studies by Sharp et al [4] and Fries and co-workers [15] attempted to finetune the reproducibility and reliability as well as many technical/methodologic aspects concerning the use of serial radiographs for clinical trials. Inter-observer variability was generally noted to be very good, yet a standardization of scores was necessary in order to compare results from different readers since readers have differing levels of sensitivity. The analysis of sequential films from the same patient established a quantification for what heretofore had been largely an intuitive concept: in order for films to be reliably characterized by the different readers as progressive, an increase in osseous erosion standardized score of greater than 15 units was required. These 15 units translated into disease duration of one year or more. Hence, these investigators established a limit of reproducibility for radiologic observations that are necessary to reliably characterize erosions as progressive. Fries’ [15] studies carried these observations a step further and made very specific recommendations about the use of techniques to eliminate time without sacrificing information and reiterated a reminder that trained, experienced readers are extremely important for accurate assessments of disease progression. Despite methodologic refinements and an extraordinary amount of experience using hand and wrist radiograms in clinical trials, the published data providing evidence that drug therapy can retard radiographic erosions in rheumatoid arthritis is very weak [I 61. Of 17 studies with a suitable control group and a rigorous methodologic approach, only two showed strong support for disease control by any intervention. Provocative questions concerning this failure of documentation by radiographic analysis have been provided by Pullar and co-workers from Glasgow [17-191. These investigators suggest that radiologic confirmation of a therapeutic effect caused by a “second-line” agent may be an unattainable goal. From their own data show-
98
the Use of Radiographs
October
30, 1987
The
American
Journal
of Medicine
Volume
83
(suppl 4B)
SYMPOSIUM
greatest obstacle to their satisfactory employment in clinical trials will be the design of the study itself and the proper selection of patients and controls for analysis. From the data and speculations already discussed; this author has reached several conclusions concerning the use of radiographic examination to measure outcome in rheumatoid arthritis clinical trials (Table I). The optimum interval between radiographic films, in order to reliably assess real change, should be at least 12 months and very likely should be 24 months. Experienced readers, preferably three of them, should be employed and radiologists may be better suited to interpret the minor alterations and normal variants that occur in small bones of the hand. It remains to be determined whether it is better for a skilled skeletal radiologist to view sequential films and quickly assess the progress of rheumatoid arthritis or to rely upon the more tedious methods of radiologic evaluation of the hands and wrists. Disagreement still exists as to whether it is necessary to prevent bias by blinding films as to sequence when multiple readers are performing assessments. It seems clear that early disease (less than five years) may be easier to measure when we apply radiographic tools. This is not to say that our drugs do not work in established or persistent disease; rather, our measurements are less useful or accurate under those circumstances. It may be suggested in a schematic model (Figure 1) that there is a window of disease duration through which our currently available measurements provide the sharpest focus for determining whether osseous erosions may be influenced by therapy. In addition, the selection of the population for study appears to be the most critical factor for a radiographic outcome therapeutic trial. Limitations set by the selection of a suitable control group have been discussed and indeed are imposing, but not, hopefully, impossible to overcome.
ON NABUMETONE-WEISMAN
1 Radiographic Progression
I-l Disease
Duration
Figure 1. Schematic model for radiographic cieterminatioi of progression in rheumatoid arthritis. Time interval for the steep rise in erosion score would be the optimum period for an intervention trial.
Finally, it remains to be determined if the radiogram is actually depicting the clinical results of our current therapeutic efforts. At least two possibilities exist: (1) the radiogram is simply out of phase and lags behind adequate control of clinical variables and, hence, we need more time with the drugs we have; or (2) we do not yet have the right agent to affect the critical pathogenetic step or steps in bone and cartilage destruction. Hopefully, the answers will be forthcoming. ACKNOWLEDGMENT I am much appreciative of the inspiration given by John Hatter, M.D., who provided the stimulus for this analysis. Gratitude is also expressed to Don Resnick, M.D., Jose Guerra, M.D., and Nathan Zvailfer, M.D., for many helpful discussions, and to Denise Smith for her expert typing skills.
REFERENCES 1. 2.
3. 4.
5. 6.
7.
Genant HK: Methods of assessing radiographic change in rheumatoid arthritis. Am J Med 1983; 75 (6A): 35-47. Sharp JT, Lidsky MD, Collins LS, Moreland J: Methods of scoring the progression of radiologic changes in rheumatoid arthritis. Arthritis Rheum 1971; 14: 706-720. Larsen A: Radiological grading of rheumatoid arthritis. Stand J Rheum 1973; 2: 1366138. Sharp JT, Bluhm GB, Brook A, et al: Reproducibility of multiple observer scoring of radiologic abnormalities in the hands and wrists of patients with rheumatoid arthritis. Arthritis Rheum ‘1985; 28: 16-24. Brook A, Corbett M: Radiographic changes in early rheumatoid disease. Ann Fiheum Dis 1977; 36: 71-73. Brook A, Fleming A, Corbett M: Relationship of radiological change to clinical outcome in rheumatoid arthritis. Ann Rheum Dis 1977; 36: 274-275. Zvaifler NJ: Overview of etiology and pathogenesis. In: Utsinger PD, Zvaifler NJ, Ehrlick GE, eds. Rheumatoid arthritis. Philadelphia: Lippincott, 1985; 151-l 58.
October
30, 1987
8.
9.
10.
11.
12.
13. 14.
The
Masi AT, Feigenhaum SL, Kaplan SB: Articular patterns in the early course of rheumatoid arthritis. Am J Med 1983; 75 (6A): 16-26. Scott DL, Grindulis KA, Struthers GR, Coulton BL, Popert AJ, Bacon PA: Progression of radiological changes in rheumatoid arthritis. Ann Rheum Dis 1984; 43 (1): B-17. Scott DL, Coulton BL, Bacon PA, Popert AJ: Methods of x-ray assessment in rheumatoid arthritis: a re-evaluation. Br J Rheumatol 1985; 24: 31-39. Scott DL, Coulton DL, Popert AJ: Long term progression of joint damage in rheumatoid arthritis. Ann Rheum Dis 1986; 45: 373-378. Dawes PT, Fowler PD, Jackson R, et al: Prediction of progressive joint damage in patients with rheumatoid arthritis viewing gold or d-penicillamine therapy. Ann Rheum Dis 1986; 45: 945-949. Sharp JT: Radiographic evaluations of the course of articular disease. Clin Rheum Dis 1983; 9: 541-557. De Carvalho A, Grandal H, Jorgensen B: Radiologic evaluation
American
Journal
of Medicine
Volume
83
(suppl 4B)
99
SYMPOSIUM
15.
16.
17.
18.
100
ON NABUMETONE-WEISMAN
of the progression of rheumatoid arthritis. Acta Radio1 [Diagn] (Stockh) 1980; 21: 115-121. Fries JF, Block DA, Sharp JT, McShane DJ, Spitz P, et al: Assessment of radiologic progression in rheumatoid arthritis: a randomized, controlled trial. Arthritis Rheum 1986; 29: 1-9. lannuzzi L, Dawson N, Zein N, Kushner I: Does drug therapy slow radiographic deterioration in rheumatoid arthritis? N Engl J Med 1983; 309 (17): 1023-1028. Pullar T, Hunter JA, Capell HA: Does second-line therapy affect the radiological progression of rheumatoid arthritis? Ann Rheum Dis 1984; 43 (1): 18-23. Pullar T, Capell HA: A rheumatological dilemma: Is it possible to
October
30, 1987
The
American
Journal
of Medicine
19.
20.
21.
Volume
alter the course of rheumatoid arthritis? Can we answer the question? Ann Rheum Dis 1985; 44: 134-140. Pullar T, Capell HA: Can treatment really influence the radiological progression of rheumatoid arthritis? Br J Rheum 1986; 25: 2-6. Joint Committee of the Medical Research Council and the Nuffield Foundation: A comparison of prenisolone with aspirin or other anaelgesics in the treatment of rheumatoid arthritis. Ann Rheum Dis 1959; 18: 173-186. Joint Committee of the Medical Research Council and the Nuffield Foundation: A comparison of prenisolone with aspirin or other analgesics in the treatment of rheumatoid arthritis. Ann Rheum Dis 1960; 19: 331-337.
83 (suppl
48)