J Dent Res 83(Spec Iss C):C122-C124, 2004
© 2004 International and American Associations for Dental Research
Potential Modern Alternative Designs for Caries Clinical Trials (CCTs) and How These can be Validated against the Conventional Model
R.K. Chesters1,*,
R.P. Ellwood2,
A.R. Biesbrock3, and
S.R. Smith4
1 Unilever Dental Research, Port Sunlight, Bebington, UK;
2 Colgate Palmolive, Dental Health Unit, Manchester, UK;
3 Procter & Gamble Co., Cincinnati, OH, USA; and
4 GlaxoSmithKline, Weybridge, UK;
* corresponding author, Richard_chesters{at}colpal.com
 |
ABSTRACT
|
|---|
The main reasons that industry runs caries clinical trials (CCTs) are to provide proof of efficacy and to collect in vivo safety data on new products. In recent years, predominantly due to declining caries levels and the use of positive controls, the cost of performing these CCTs has escalated. It is now reaching the stage where it is becoming commercially prohibitive to conduct new studies. This is likely to stifle innovation of new anticaries products, and we now need new, more discriminatory, faster, and less expensive study designs. There are many ways in which the design of CCTs may be changed, such as improving diagnostic efficiency, improving data handling/statistical modeling, and using high-risk populations. However, it is paramount that the overriding principle behind CCT design validation must be that the results/conclusions from any new design are in line with those shown previously by ‘conventional’ CCTs, to ensure the maintenance of standards for both efficacy and safety. It is suggested that the validation of any new trial design must involve comparisons with regimens previously shown in conventional CCTs to have different anticaries efficacies. For example, since several clinical trials have shown convincing evidence for a monotonic dose response for fluoride at least up to levels of 2500 ppm F, one could choose two products, differing solely in their fluoride level. One aim for this workshop is to identify and agree on validation principles for new clinical trial designs. This will facilitate general international acceptance of novel smaller/faster CCTs designs both now and in the future. We recognize that any new design must not compromise the standard of proof of either efficacy or safety. In addition, any principles will need to take account of current understanding of the caries process, while recognizing the need for change to match future developments in cariology. Finally, the mechanism of action of the test product must be considered, in assessments of the acceptability of novel designs, if this differs markedly from the regimens used to validate the design.
KEY WORDS: caries • clinical trials • study design • validation • model
|
INTRODUCTION
|
|---|
The main reason the oral care industry runs Caries Clinical Trials (CCT) is to demonstrate proof of efficacy to obtain pre-marketing approval and support advertising claims. Another important reason, however, is to collect safety data on products from well-monitored studies on relatively large numbers of subjects. Such safety data might include, for example, soft-tissue evaluations, staining, and effects on supragingival calculus.
Clearly, no manufacturer will invest time and resources in a CCT without first collecting evidence about the new product’s quality, safety, and efficacy. Thus, a hierarchy of testing is carried out to determine whether a CCT can be justified on both ethical and economic grounds. This hierarchy typically includes in vitro laboratory studies, mode of action studies, simple clinical evaluation and caries models, such as in situ testing, plus safety and quality testing. However, our current state of knowledge has meant that many companies have experienced the situation where this pre-testing with in vitro or other artificial models has failed to predict the anti-caries efficacy observed in the subsequent CCT. Thus, the proof of efficacy will remain, at least for novel product formulations, a caries clinical trial measuring the effects in humans on their natural teeth.
The decline in dental caries in the last 20 years (Fig. 1
) (Todd and Dodd, 1985; O’Brien, 1994), together with the need to make comparisons against effective anticaries control agents, has meant that the size of caries clinical trials have increased dramatically. This in turn has increased their cost and complexity. Fig. 2 illustrates the effect of increment on cell size to give constant discriminatory power (Chesters, 2001). These cell sizes were calculated by 16*variance/delta/delta, where variance is based on actual size of increment via 5 x mean, delta is size of effect (as a fraction, i.e., 10% = 0.1*mean), and 16 is a constant allowing for 80% power to detect a two-tailed difference at the 0.05 level of significance. In an attempt to mitigate the effects of falling caries prevalence, manufacturers have sought to reduce cell sizes by carrying out CCTs in regions or populations with higher caries rates, such as Eastern Europe, Asia, and Latin America. The overall effect of these changes has been to lead to spiraling costs, which, in turn, makes it increasingly difficult to justify commercially the expense of developing more effective products.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 2. Cell size as a function of predicted caries increment for control group. (Data supplied by Unilever Dental Research.)
|
|
There is, therefore, a need to develop designs that will allow studies for demonstrating the anticaries efficacy of dental products to be done both more quickly and on fewer subjects, thereby ensuring that limited resources are used in the most effective way. Clearly, any new design must be conservative, in the sense that it must not lower the standard of proof of efficacy or safety.
|
WHERE WE ARE NOW
|
|---|
The dental profession’s view of caries at the historic American Dental Association (ADA) Chicago meeting in October, 1968, focused on caries at a level that would, at that time, be likely to initiate restorative intervention. However, recently there has been a change in emphasis from a restorative approach to the management of dental caries to a more preventive philosophy, leading to a recognition of the importance of non-cavitated lesions—for example, white- or brown-spot caries lesions (NIH Consensus Development Conference, 2001). It therefore seems logical that these changes are reflected in an alteration in the threshold used in CCTs to define decay. Evidence of early demineralization, such as white spots, is now widely taken to be an indication for a need for preventive and therapeutic interventions (Pitts, 2000). The recognition of the importance of these early lesions provides an opportunity to reconsider the way we measure dental caries lesions. Changes in lesion state can be potentially measured at smaller intervals, meaning that less time is required for a lesion to change in state. In addition, positive clinical outcomes could then include non-progression or reversal of lesions.
This change in emphasis has also been recognized by the International Dental Federation (FDI, 1999) implying that CCTs could be carried out in shorter time periods given more sensitive caries diagnostic methods. However, other anticaries testing guidelines, such as those outlined by the American Dental Association (1988), have not been revised in recent years.
|
ALTERNATIVE DESIGNS FOR CARIES CLINICAL TRIALS
|
|---|
There are many potential ways of improving the discrimination of CCTs (Table). The utilization of high-caries-risk subjects and/or high-caries-risk populations is a strategy that has been frequently used to aid discrimination. A significant problem with the use of high-caries-risk populations is the difficulty in extrapolating the results to a more general population. Understanding the underlying etiologic factors related to the observed higher caries rates is crucial in assessment of the generalizability of the population and observed effects.
Subject recruitment has historically been mainly in schools, because children are readily accessible. Also, the most caries-susceptible teeth, the 1st and 2nd permanent molars, erupt at this time. We can see, from the cross-sectional data in Fig. 3, that the maximum two-year caries increment occurs shortly after eruption of these teeth. One strategy for improving discrimination would be to enroll children into caries clinical trials only when either the 1st or 2nd molars are in the process of erupting into the oral cavity. Alternatively, children would undergo a dental assessment, and only those with erupting molars would be recruited into the trial. Logistical and cost impacts of such strategies need to be considered.
View larger version (13K):
[in this window]
[in a new window]
|
Figure 3. Predicted two-year caries increment (DMFT) for different ages at baseline, derived from cross-sectional data.
|
|
The use of indicator teeth has already been shown to improve product discrimination. Huntington (1985) showed that applying logistic modeling to second permanent molar data could increase the sensitivity of the study by up to 5 times that obtained with traditional methods. This observation is perhaps less surprising if we consider what a small proportion of surfaces actually develops caries, even in a three-year time period. The large difference in caries susceptibility of different teeth is well-known by dentists. It is therefore not surprising that restricting an analysis to the more susceptible teeth might increase statistical power. The power of this approach, as with all designs, is clearly influenced by the caries incidence in the trial population.
It is also possible that a significant improvement in CCT design could be realized by improving subject compliance, since this would tend to reduce variance of the data. However, the situation is probably quite complex, since supervised brushing, regardless of treatment, will undoubtedly reduce caries incidence rate in a given study population. Designs involving supervision may require higher background caries increment rates or longer study durations relative to unsupervised studies, to offset the reductions in caries that accompany supervised brushing. In addition, it is unclear from the existing evidence whether supervised brushing studies will lead to enhanced sensitivity to discriminate between treatments. Clearly, at one extreme, if subjects fail to use the test products, then we cannot expect to observe product effectiveness or significant discrimination between groups. Supervised brushing in caries studies is a viable approach, if we recognize that optimal discrimination of product effectiveness may fall somewhere between these two extremes of compliance (unsupervised vs. supervised). It also might be argued that supervised brushing has the potential to reduce the generalizability of conclusions.
One reason why caries clinical trials have historically lasted from two to four years has been the time taken for significant numbers of lesions to develop at the cavitation level. Thus, measuring caries at an earlier stage in the caries continuum, provided that this is done according to well-validated and reproducible caries diagnostic methods, could shorten the duration of CCTs. In addition, it is important to consider, and possibly report, all clinically feasible caries-related events, including arrest, progression, and regression. This approach would also have an important advantage in that it would greatly reduce the impact/influence of dental treatment on the trial result, although it does not eliminate the issue completely. For example, the reasons for placing fissure sealants cannot be assigned, and therefore these surfaces are lost from any analysis.
Monitoring the quality of the data being collected is essential. Direct data entry onto computers eliminates the need for subsequent data entry and also can improve the quality of data by excluding invalid codes and reducing the quantity of missing data. The use of direct data entry also makes it easy to carry forward information from earlier examinations—for example, on missing teeth—thereby possibly reducing the number of longitudinal errors. As always, the training and calibration of the clinical examiners are vital to success, particularly when asking clinicians to use techniques with which they were not previously familiar.
|
NEW DESIGN VALIDATION AGAINST CONVENTIONAL MODEL(S)
|
|---|
Probably the easiest way to validate any new trial design would be to carry out a comparison of two products previously demonstrated in a superiority CCT, to show a difference in efficacy. Several clinical trials have shown convincing evidence for a monotonic dose response for fluoride, at least up to levels of 2500 ppm F (Fig. 4).
Thus, an obvious choice would be two products differing solely in their fluoride levels. However, testing with any other products, which have been proven previously to have different anticaries efficacies, could also be used to validate a new design. Another, but probably more risky, approach would be to apply both the new and a conventional approach (assuming that they are compatible) in a ‘superiority’ CCT, which is a more stringent test. The danger of this approach is that validation of the new design would be dependent upon both the new and conventional approaches giving a statistically significant difference between treatments.
Clearly, it is inappropriate to validate a new design in an equivalence study, where ‘no statistically significant difference’ would be expected. There are two further caveats that apply to these propositions. First, we must be confident that the mechanism of action of the new active system is comparable with that of the reference system. For example, there could be some concern about whether the short-term efficacy of antibacterial agents and sugar substitutes would be sustained in longer-term use, or whether long-term resistant or adaptation of the oral flora might take place.
Second, it is also important that new methods are not abused to identify clinically trivial differences between products by increasing factors such as their duration and panel size.
|
FUTURE
|
|---|
Currently available diagnostic methods are far from ideal. Consequently, there is a need for new quantitative, objective, reproducible methods for caries diagnosis, which are both more sensitive and more specific. It is expected that, in due course, the use of such methods might allow both the number of subjects and the duration of CCTs to be reduced.
In addition, most of the diagnostic methods commercially available cannot be applied to all sites (surfaces/teeth). For example, the Electrical Caries Monitor (Lode) is used only on occlusal surfaces, bitewing radiographs on approximal surfaces of posterior teeth, and DIAGNOdent® on occlusal and (more recently) on approximal surfaces (Chesters et al., 2002), and even the caries activity visual method (Ekstrand et al., 1998) is intended only for occlusal surfaces.
Finally, it is important that the consensus reached during this meeting acknowledge that alternative designs, which make use of new diagnostic methods, are acceptable provided that they meet the validation criteria agreed upon at this meeting (ICW). There is a need to bring the guidelines for running caries clinical trials into line with modern practices and to encourage the innovation of new, more effective anticaries products by reducing the duration of the innovation cycle.
|
FOOTNOTES
|
|---|
Presented at the International Consensus Workshop on Caries Clinical Trials, Glasgow, Scotland, January 7–10, 2002
|
REFERENCES
|
|---|
American Dental Association (1988). Report of workshop aimed at defining guidelines for caries clinical trials: superiority and equivalency claims for anticaries dentrifrices. Council on Dental Therapeutics. J Am Dent Assoc 117:663–665.[ISI][Medline]
Chesters RK (2001). Personal communication.
Chesters RK, Pitts NB, Matuliene G, Kvedariene A, Huntington E, Bendinskaite R, et al. (2002). An abbreviated caries clinical trial design validated over 24 months. J Dent Res 81:637–640.[Abstract/Free Full Text]
Clarkson JE, Ellwood RP, Chandler RE (1993). A comprehensive summary of fluoride dentifrice caries clinical trials. Am J Dent 6:S59–S106.
Ekstrand KR, Ricketts DNJ, Kidd EAM, Qvist V, Schou S (1998). Detection, diagnosing, monitoring and logical treatment of occlusal caries in relation to lesion activity and severity: an in vivo examination with histological validation. Caries Res 32:247–254.[ISI][Medline]
FDI (1999). Guidance on the assessment of the efficacy of toothpastes. FDI Commission. Work Project (8-95). Int Dent J 49:311–316.[ISI][Medline]
Huntington E (1985). Increasing the sensitivity of caries clinical-trials by applying logistic modeling to specific teeth. Community Dent Oral Epidemiol 13:264–267.[ISI][Medline]
National Institutes of Health (NIH) Consensus Development Conference (2001). Diagnosis and management of dental caries throughout life. NIH Consensus Statement March 26–28, 2001. 18(1):1–24.
O’Brien (1994). Children’s dental health in the United Kingdom 1993. London: Her Majesty’s Stationery Office.
Pitts NB (2000). Need for early caries detection methods: a European perspective. In: Proceedings of the 4th Indiana Conference on Early Detection of Dental Caries II. Stookey GK, editor. Indianapolis: Indiana University School of Dentistry, ISBN 0-9655 149-2-7.
Todd JE, Dodd T (1985). Children’s dental health in the United Kingdom 1983. London: Her Majesty’s Stationery Office.
This article has been cited by other articles:
|
|
|
|
 
H. Whelton
Overview of the Impact of Changing Global Patterns of Dental Caries Experience on Caries Clinical Trials
J. Dent. Res.,
July 1, 2004;
83(suppl_1):
C29 - C34.
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
