| Sign In to gain access to subscriptions and/or personal tools. |
The Challenges of Validating Diagnostic Methods Relative to a Conventional Two-year Caries Clinical Trial
1 Procter & Gamble Company, Cincinnati, OH, USA; Correspondence: corresponding author, Biesbrock.ar{at}pg.com
ABSTRACT This paper is directed to the question, "What are the appropriate validation criteria for the use of a new clinical trial methodology as a replacement for a conventional two- to three-year caries study?" It is important to recognize that the objective of a two- to three-year randomized, controlled caries trial is to test a precisely framed hypothesis, regarding an experimental product’s efficacy relative to a control product. The external validity of conventional two- to three-year caries clinical studies in determining the efficacy and safety of anti-caries products is well-accepted. However, caries clinical trials are not without limitations and have increasingly been viewed as inefficient with respect to measuring the disease process in a holistic manner. The endpoint of a caries lesion with loss of enamel integrity (cavitation) focuses on one end of the caries progression continuum at the expense of early caries initiation and progression. Several early caries detection methods have been developed that correlate with mineral loss of the tooth surface. These diagnostics differ from conventional visual-tactile and radiographic methods in that they are capable of detecting early non-cavitated lesions, and this can generate continuous data. As diagnostic methods become accepted, they will lead to study designs that diverge from the conventional two- to three-year caries studies. Modification of the existing two- to three-year conventional caries design for assessment of product effectiveness, whether by the introduction of a new diagnostic method or by modification of the overall clinical design, must result in a clinical design that is able to differentiate known treatments on the basis of caries prevention efficacy. Given that the fluoride dose response has been characterized in the literature, this should form the basis of any validation package for new methodologies. In conclusion, a minimum expectation for acceptance as a replacement to conventional testing should be that the method or design can differentiate products of known efficacy from one another and that the efficacy relationship observed in a two- to three-year conventional study can be observed with the new method or design.
Key Words: caries • clinical trials • validation • early caries detection • diagnostics INTRODUCTION The question of the validation of diagnostic methods and/or alternative clinical designs as replacements for conventional long-term caries prevention studies is a timely subject, given the current interest in early caries detection methodologies. In many respects, this should be viewed scientifically as two separate questions: (1) What are appropriate validation criteria for early caries diagnostic methods? and (2) When implemented into a clinical study protocol, what are the appropriate validation criteria for the use of this new methodology as a replacement for conventional two- to three-year caries prevention studies? Importantly, these are fundamentally different questions that need to be addressed through distinct approaches. The first question is the topic of a separate paper at this workshop; therefore, this paper will focus on the second question. In 1968, the American Dental Association sponsored a Conference on Clinical Testing of Cariostatic Agents. The conference proceedings led to publication of consensus guidelines for caries prevention studies that have remained a standard in caries testing in the United States over the past 30–40 years, with very little evolution from the original design elements. Specifically, a randomized, controlled clinical study with an ideal duration of two to three years was advocated. The clinical endpoint was a carious tooth surface (as opposed to sound) as detected by visual-tactile examination supplemented with radiographs. The effectiveness of this design as a tool to differentiate products based on cariostatic activity has been remarkably long-lived, since the current ADA guidelines for caries testing still recognize a conventional two-year study as a necessity for proof of anti-caries activity. Current FDI guidelines also recognize conventional two-year studies as the standard, but also discuss provisions for shorter studies (Reich, 1999). During this same timeframe, significant advances have been made with respect to our understanding of the molecular basis of caries and our ability to measure early enamel demineralization and caries progression. Many believe that the science has reached a juncture which may justify alternative modern clinical methods and designs as replacements for conventional two- to three-year caries prevention studies.
To entertain the replacement of the accepted standard, we need to fully understand the strengths and weaknesses of the two- to three-year caries study design. Recognize that the primary objective of two- to three-year randomized, controlled caries trials is to test a precisely framed hypothesis (O’Mullane, 1976). In the case of caries studies, that hypothesis centers around the determination of an experimental product’s efficacy relative to a control product. The external validity of conventional two- to three-year caries clinical studies, with respect to their value as a tool to differentiate cariostatic products from one another, is well-accepted. Fluoride dose-response effects have been repeatedly demonstrated with different product forms, by different examiners, and in diverse populations (Table
Over the last 40 years, subtle variations in study design approaches have evolved. Global variations in approaches to caries diagnosis in clinical studies and practice represent one of the primary differences. One approach (visual-tactile) involves a visual-tactile examination supplemented with radiographic examination for diagnosis of interproximal surfaces, while a second approach (clinical-visual assessment) involves an expanded visual examination without the tactile component, supplemented on occasion with fiber-optic transillumination examination for diagnosis of interproximal surfaces. Recognize also that hybrids of these models have been conducted. The clinical-visual assessment approach readily allows early visual changes in enamel appearance (white-spot lesions) to be recorded, while the visual-tactile approach tends to exclude these lesions in the absence of a tactile change. The diagnostic management of white-spot lesions in caries clinical trials has been historically a problem: Research which examined the long-term prognosis of white-spot demineralization found that the vast majority fail to progress to overt caries (Backer-Dirks, 1966). Based on this research, it has been reported that the probability that a white-spot demineralization will progress to overt caries over a seven-year period is P = 0.13, while the probability that it will regress to a sound surface is P = 0.87 (Beiswanger, 1996). These probabilities are likely dependent on several factors, including levels of carbohydrate exposure and fluoride concentration of the drinking water. To a certain extent, the variations in caries diagnosis in caries trials have evolved as distinct approaches to manage the signal-to-noise ratio in caries clinical studies and the changing presentation of the disease over the last 40 years. In addition, there are regional variations in accepted clinical practice that govern the use of explorers and radiographs for caries diagnosis in the conduct of dentistry. In practical application, the philosophical differences between the two approaches are often likely blurred. Based on results in published caries clinical studies, strong arguments can be made that both approaches are valid and proven methods for differentiating treatment effects in caries clinical trials. However, the current design of caries clinical trials is not without its limitations and has increasingly been viewed as inefficient with respect to measuring the disease process in a holistic manner. The endpoint of a caries lesion with loss of enamel integrity (cavitation), which is most frequently used in conventional caries studies, focuses on one end of the caries progression continuum at the expense of early caries initiation and progression (demineralization which occurs at a level that cannot always be detected by the naked eye). This early mineral loss is part of the caries process and an absolute necessity to reach the eventual clinical determination of cavitation at the enamel surface. Several early caries detection methods have been developed that correlate with mineral loss of the tooth surface (including light fluorescence, electroconductivity, etc.). These diagnostic methods fundamentally differ from conventional visual-tactile and radiographic methods in that they are capable of detecting subclinical lesions (lesions undetectable to the naked eye) and generate continuous as opposed to categorical (dichotomous) data. Clearly, these appear to be attractive features for a disease diagnostic. Invariably, as modern diagnostic methods become accepted, they will lead to novel, more efficient study designs that diverge from the standard conventional two- to three-year caries prevention studies. Alternatively, interest in the field of caries research may continue to drive efficient study designs based on variations of conventional measurement endpoints that diverge from the standard conventional two- to three-year caries prevention studies. A measurement diagnostic that demonstrates a strong correlation with demineralization as measured by an accepted standard (histology, polarized light, TMR, etc.) becomes a candidate for replacing or supplementing visual-based techniques in the cross-sectional diagnosis of caries. However, this finding by itself, although essential, does not justify replacing two- to three-year clinical trials to differentiate among products based on cariostatic activity. A diagnostic method that is capable of detecting caries lesions in dental practice may not bring greater efficiency to a clinical study with respect to examination time and/or balancing the signal-to-noise ratio. Evidence of enhanced diagnostic sensitivity does not address this issue. Modification of the existing two- to three-year conventional caries design for the assessment of product effectiveness, whether by the introduction of a new diagnostic method or by modification of the overall clinical design, must result in a clinical design that is able to differentiate between and among known treatments on the basis of caries-prevention efficacy. This concept has been observed with other diseases, such as osteoporosis, which is an excessive, proportional reduction in trabecular and cortical bone mineral and matrix, leading to increased skeletal fragility. This disease shares several similarities with dental caries, highlighted by the fact that both involve a compromise in the structural integrity of calcified tissue, which ultimately leads to a loss of function. The disease typically presents after the age of 45 in women and 55 in men (Courpron, 1981). The disease is largely asymptomatic until clinical symptoms that include pain, skeletal deformity, and spontaneous overt fracture manifest. The vertebrae, hip, pelvis, ribs, and femoral neck are at greatest risk of fracture. The pathogenesis of the disease centers around a shift in the metabolic equilibrium between normal rates of bone loss and formation. The primary endpoint of the majority of clinical studies has been osseous fracture (vertebral, hip, femur). Development of additional primary endpoints has been an arduous process, with various radiography-based measurements, clinical biomarkers, and quantitative bone mass measurements examined. It seems intuitive that increases in bone mass should correlate well to reductions in bone fracture, and in fact, measurement technologies which measure bone mass and product interventions which promote bone mass have been heavily explored for the management of osteoporosis. For example, multiple studies have demonstrated that sodium fluoride stimulates bone formation in women with osteoporosis by 8% per year and prevents vertebral fracture by 4% per year (Riggs et al., 1990; Riggs and Melton, 1992). Radiography measurements clearly demonstrate increased bone mass following sodium fluoride, which intuitively suggests a benefit for osteoporotic patients. However, sodium fluoride therapy has also been associated with increased femoral neck and hip fracture (Gutteridge et al., 1984; Hedlund and Gallagher, 1989). Excessive fluoride content potentially leads to the development of qualitatively abnormal bone structure and increased fragility; thus, increased bone mass is not necessarily equivalent to increased bone strength and resistance to fracture (Carter and Beaupre, 1990; Riggs and Melton, 1992). For this reason, efforts to replace osseous fracture as the primary efficacy measurement in osteoporosis studies with bone mass and density measurements have been gradual. The analogy from this example to dental caries is that methods that measure changes in mineral content, without illuminating the quality of mineral structure, may or may not correlate well to the caries status of the tooth. This is a substantive risk associated with accepting a novel caries diagnostic as a replacement for conventional measures in a conventional (or modified) two- to three-year caries study. Ultimately, if alternative detection methods or designs are to be used as tools to differentiate between and among products based on cariostatic activity, they should be required to demonstrate external validity similar to that demonstrated by conventional two- to three-year caries clinical studies. In this context, a minimum expectation for acceptance as a replacement for conventional testing should be that the method or design can differentiate products of known efficacy from one another, and that the efficacy relationship observed in a two- to three-year conventional study can be observed with the new method or design. It is desirable that the results be replicated in at least two studies to demonstrate the robustness of the methodology. If possible, populations in which validation studies are performed should have caries experience comparable with that of populations in which earlier caries clinical studies were performed. Given that the fluoride dose response has been characterized in the literature, this should form the basis of any validation package for new methodologies. This concept is wholly consistent with the standards that have been used in the validation of other caries measurement tools designed to differentiate product effectiveness, including in vitro pH cycling and in situ caries models. Importantly, during the validation process, results from the new methodology should be benchmarked relative to conventional two- to three-year caries clinicals to help determine the relative clinical significance of observed results with new methods, the primary concern being that product differences that are relatively small according to current study designs may appear more robust as methods evolve.
FOOTNOTES Presented at the International Consensus Workshop on Caries Clinical Trials, Glasgow, Scotland, January 7–10, 2002 REFERENCES
Journal of Dental Research, Vol. 83, No. suppl 1,
C53-C55 (2004) This article has been cited by other articles:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
