Funding

Osteoarthritis Initiative

Updated July 12, 2001

Steering Group
Draft #9, 2/17/00
Longitudinal Studies of Osteoarthritis:
State-of-The Science Evaluation

Prepared by Marc C. Hochberg, M.D., M.P.H.,1 and Cori Vanchieri, M.A..2

1 Professor of Medicine and Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, MD.

2 Vanchieri Communications, Silver Spring, MD.

The Osteoarthritis Initiative:

The Osteoarthritis (OA) Initiative was formulated to develop a project plan to facilitate scientific evaluation of biomarkers of bone and cartilage turnover for OA (http://www.niams.nih.gov/Funding/Funded_Research/Osteoarthritis_Initiative/mission.asp). These biomarkers, including biochemical parameters of bone and cartilage turnover, and anatomic characteristics of the joint as visualized by imaging techniques, are to be considered as possible surrogate endpoints for clinical trials of novel interventions.

An exploratory meeting on biomarkers for OA was held on April 16, 1999 (http://www.niams.nih.gov/Funding/Funded_Research/Osteoarthritis_Initiative/summary416.asp). At that meeting, the suggestion was endorsed to examine existing databases for their possible use in identifying and validating biomarkers. The Osteoarthritis Initiative Steering Group met on June 24, 1999 to further discuss plans for the initiative. The whole group and the subgroup on epidemiology/biostatistics/genetics noted that while "ongoing studies of OA are useful, [they] all have weaknesses in clinical data or the samples collected" (http://www.niams.nih.gov/Funding/Funded_Research/Osteoarthritis_Initiative/biomarkersum62499.asp). Specifically, the subcommittee discussed several focus questions including "Do existing population studies and clinical trials provide sources of relevant information, or starting points for expansion?" (http://www.niams.nih.gov/Funding/Funded_Research/Osteoarthritis_Initiative/epidsubgroup.asp). It was recognized that several ongoing cohorts could yield information relevant to the goals of the initiative. A list of existing cohorts was developed among the subcommittee members and included within the subcommittee's report. The subcommittee stated that "[a] review of existing potential sources of relevant information and samples [is] strongly encouraged, ..."

The Epidemiology "White Paper" Project:

A "State of the Science" evaluation of osteoarthritis natural history studies was proposed to include a review of the literature and protocols for each of the existing OA natural history studies with development of a summary report that would provide an overview, document major findings, and identify gaps in knowledge. The goals of this "white paper" were to 1) develop a descriptive summary of each of the existing OA cohorts, 2) develop from the literature review a summary of the major findings from the cohort studies, 3) provide a descriptive summary of the specimens collected in each cohort, 4) address the types of imaging studies conducted in the cohorts, and 5) summarize the strengths and weaknesses of the currently existing cohorts to provide a basis for identifying and validating imaging and biochemical markers.

Principal investigators of the cohort studies identified by the epidemiology/biostatistics/genetics subcommittee were contacted by a scientific writer (Ms. Cori Vanchieri) and asked to respond to a structured questionnaire providing information on the cohorts and a list of citations reporting the major findings from these cohorts. It was decided not to obtain such information on clinical trials nor to include information on clinical trials in this analysis. The following summary provides the subcommittee with information to make recommendations of the feasibility of using existing cohorts for validation studies of imaging modalities and biochemical measures as surrogate markers in OA studies. If the existing cohorts are not felt to be sufficient for this activity, the subcommittee will propose the construction of one or more new cohorts for this activity.

Summary of Existing Cohorts:

This systematic review identified a total of 25 longitudinal cohort studies of OA. Of these 25 cohorts, 16 are population-based and the remaining 9 are patient-based. All but 2 of the population-based and 1 of the patient-based cohorts are active. Of the 16 population-based cohorts, 12 are conducted in the United States and 4 in Europe. Of the 9 patient-based cohorts, 7 are conducted in Europe and 2 in the United States. The summary of the U.S. cohorts is provided in Appendix A and the European cohorts are described in Appendix B. Relevant material is included in Table 1 and Table 2.

The population-based cohorts:

The population-based cohorts represent a mix of studies, some of which have recruited representative samples of regions while others have recruited volunteers: the unifying factor is that subjects were not selected on the basis of the presence of existing symptomatic OA. These cohorts thus offer the possibility of identifying factors associated with the development of anatomic changes of OA as identified primarily by radiograph (magnetic resonance imaging is available in one population-based study [HABC]) as well as examining the relationship between symptoms and activity limitation and anatomic changes of OA in persons from the general population. These cohorts range in size from as little as 300 participants with longitudinal data (BLSA with serial knee radiographs) to over 6000 participants (SOF with serial hip radiographs). Most cohorts are comprised of North American and European Caucasians; 4 U.S. cohorts, however, were oversampled to contain African Americans. Most cohorts are comprised of both men and women; however, 5 cohorts are comprised of exclusively women. One of these includes young women (MBHS) while two others include perimenopausal women (SWAN, Chingford); these cohorts provide the unique opportunity to examine the relationship between changes in sex hormone levels that accompany the perimenopause and the development of anatomic changes and symptoms of OA.

Almost all of the cohorts, except SOF, have data on anatomic changes of OA involving the knee. The imaging technique is predominantly the weight-bearing anteroposterior knee radiograph not taken in either the semiflexed or Schuss position. Lateral views of the knee and sunrise views of the patellofemoral joint are not routinely available. Hence, there is concern about the validity of the finding of joint space narrowing (decrease in interbone distance) and the assessment of progression based on a decline in interbone distance. However, for the diagnosis of OA based on the Kellgren-Lawrence grading system, which relies primarily on the presence of definite osteophytes at the tibiofemoral joint, these radiographs can be used for case identification. Hip and hand radiographs are available in approximately half of the cohorts. Because of the low prevalence and older age of onset of hip OA, only the large cohorts with hip radiographs (e.g., JCOP, SOF, and Rotterdam) would seem suitable for studies of hip OA.

Virtually all of these cohorts collected and stored serum samples, and most collected and stored urine samples, which theoretically could be used for studies of biochemical markers of bone and cartilage turnover. It is not clear whether samples were collected at the same visit as the imaging procedure in all studies, and further information needs to be obtained relevant to this issue. In addition, the techniques for collection and storage of the specimens need to be further determined as diurnal variation in levels of some of the potentially informative markers has been identified (see review by Dr. Poole). It is useful to note that all samples are available for the initiative either with approval or collaboration of the investigators of the specific cohort studies.

Hence, it appears that the existing longitudinal population-based cohorts will be useful for the conduct of studies, probably of the nested case-control or case-cohort design, to validate the association of biochemical markers of bone and cartilage turnover with the onset of knee and possibly hip OA, as defined by standard radiograph imaging techniques. It is unlikely that these cohorts will provide acceptable information on disease progression because of the relatively small numbers of cases of OA in each cohort at baseline and the lack of use of currently recommended techniques for obtaining images to assess progression as reflected in the loss of interbone distance (a surrogate [?] for cartilage loss).

The patient-based cohorts:

The patient-based cohorts, by definition, include patients with OA. All these cohorts are comprised of North American and Northern European Caucasians of both sexes and range in size from approximately 100 to 500 patients. The patients range in age from the fourth through ninth decade and almost all have knee OA except for patients followed after participation in the ECHODIAH study who have hip OA. Again, as in the population-based cohorts, only one study has performed magnetic resonance imaging of the knee (Spenshult); all other studies have performed radiography.

All studies have stored serum and urine samples, and three have stored synovial fluid samples, which theoretically could be used for studies of biochemical markers of bone and cartilage turnover. The techniques for collecting and storage of the specimens need to be further determined as diurnal variation in levels of some of the potentially informative markers has been identified (see review by Dr. Poole). It is useful to note that all samples are available for the initiative either with approval or collaboration of the investigators of the specific cohort studies.

The patient-based cohorts complement the population-based cohorts in that their greatest utility is in identifying biomarkers associated with progression of OA. Publications from the now inactive Bristol OA 500 study have described the longitudinal course of patients with knee OA and the lack of correlation between radiographic and clinical changes (Osteoarthritis Cart 1997;5:87-97), while others have examined the relationship between serum biomarkers and disease progression as defined by radiographic changes (Arthritis Rheum 1995;38: 760-7; Br J Rheumatol 1995;34:306-10; Ann Rheum Dis 2000;59:71-4).

Summary:

Existing longitudinal cohorts may be able to contribute to the goals of the OA initiative; specifically, to facilitate the scientific evaluation of biomarkers for OA. However, there are substantial drawbacks to most of these studies that lead to the need for generation of new cohorts for biomarker validation. Regarding the evaluation of imaging techniques as surrogate markers for anatomic changes of OA, it is unlikely that the existing cohorts will be able to provide definitive information about the role of magnetic resonance imaging as a surrogate marker of either the presence of disease or progression of disease. Furthermore, other studies have suggested newer techniques for obtaining standard radiographic images of joints for evaluation of bone and cartilage changes in OA of the knee and hip, and these techniques have not been employed in most of these cohorts. Hence, validation of imaging techniques as surrogate endpoints will require additional studies.

With regard to biochemical markers of bone and cartilage turnover, it is possible that the existing cohorts, especially the population-based cohorts, can be used to evaluate the relationship of biomarkers with the development of disease, especially for knee OA and possibly also for hip OA. This would require the use of older radiographs for disease definition; however, this does not appear to be an insurmountable proposition. In addition, the clinical patient-based cohorts may be useful for validating biomarkers for disease progression, as suggested by preliminary data from the Bristol OA 500 cohort concerning serum levels of C reactive protein and hyaluronic acid and changes in serum levels of cartilage oligomeric matrix protein. Clearly, use of data and specimens from these cohorts should be considered, in addition to the analysis of data from new cohorts designed as part of this initiative. Indeed, results of such analyses could help with the rational choice of potential biomarkers to be examined in new studies.