Table 1

 Application of the OMERACT filter to the three methods of measurement of synovial membrane immunohistochemical labelling currently in use

Measurement methodValidityReliabilityFeasibilityReferences
Manual counting (MC)Valid measure when the variable being measured is localised to a cell. Less certainty about the validity of this method when measuring variables which are not cell associated or diffuse in their distribution—for example, cytokinesReliability demonstrated between observers in the same research laboratory but not tested between different research centres. Reliability not tested for a wide range of biological variables. Field selection bias could be a problem for RCTsRequires minimal equipment which should be readily available in any research centre in the world. Laborious and time consuming method which is not well suited to large scale studies26, 27, 30, 37, 38, 40
Semiquantitative analysis (SQA)Most subjective of the methods but does attempt to evaluate the tissue as a whole, removing problems of field selection bias. Validated for the measurement of a wide range of biological variables in synovial tissueIntra- and interobserver reliability demonstrated but requires the use of more than one trained observer and standardisation. Use of an atlas of gradings would improve reliability between centresMost feasible of the methods as it requires no special equipment. Only feasibility problems relate to the training and standardisation of observers between centres. This would be facilitated by the development of an atlas of grades for a range of biological variables in the SM22, 35, 37, 40
Digital image analysis (DIA)Demonstrated to be valid for the widest range of biological variables in the SM, including cytokines and metalloproteinases.Intra- and interobserver reliability demonstrated for a wide range of biological variables in the SM. Automatisation of field acquisition and threshold selection would reduce any remaining sources of bias in this methodMost feasible method for measurement of synovial tissue staining in RCTs because of the minimisation of bias, the speed of the method, and the ability to automate many of the steps involved. The significant “learning curve” and the initial set up costs are a major deterrent to wider application of this method outside RCTs and specialised synovial tissue research centres.33, 38, 39, 40