Assessment of the relevance and reliability of alternative toxicological methods requires the application of biostatistical methods. This report contains the recommendations of the ECVAM Task Force on Biostatistics for the appropriate use of biometrical methods, and for closer collaboration between biostatisticians and experimental toxicologists in the development and validation of alternative methods. The main objectives of these recommendations are to improve the compatibility of the results obtained from different studies, and to prevent the misinterpretation of statistical analyses. The following points are discussed: a) the general importance of biostatistics in the development and validation of alternative tests; b) optimisation of study designs; c) statistical characterisation and processing of toxicological data; and d) the analysis of relationships which may exist between data obtained from various tests. In particular, the establishment of appropriate prediction models which are able to predict the toxic effects likely to occur in vivo from the results obtained in an in vitro test system is discussed. The members of the Task Force hope that this report will help to stimulate discussions among biostatisticians and experimental toxicologists about how the proper application of biostatistical methods can improve the design and evaluation of alternative toxicological methods.
BallsM. (1992). In vitro test validation: high hurdling but not pole vaulting.ATLA20, 355–357.
2.
BallsM., BlaauboerB., BrusickD., FrazierJ., LambD., PembertonM., ReinhardtC., RoberfroidM., RosenkranzH., SchmidB., SpielmannH., StammatiA-L., & WalumE. (1990). Report and recommendations of the CAAT/ERGATT workshop on the validation of toxicity test procedures.ATLA18, 313–337.
3.
BallsM., BlaauboerB.J., FentemJ.H., BrunerL., CombesR.D., EkwallB., FielderR.J., GuillouzoA., LewisR.W., LovellD.P., ReinhardtC.A., RepettoG., SladowskiD., SpielmannH., and ZuccoF. (1995). Practical aspects of the validation of toxicity test procedures. The report and recommendations of ECVAM workshop 5.ATLA23, 129–147.
4.
CurrenR.D., SoutheeJ.A., SpielmannH., LiebschM., FentemJ.H., and BallsM. (1995). The role of prevalidation in the development, validation and acceptance of alternative methods.ATLA23, 211–217.
5.
McCabeG.P. (1984). Principal variables.Technometrics26, 137–144.
6.
SjöströmM., and ErikssonL. (1995). Applications of statistical experimental design and PLS modelling in QSAR. In QSAR: Chemometric Methods in Molecular Design (ed. van de WaterbeemdH.), pp. 63–90. Weinheim, Germany: Verlag Chemie.
7.
CohenJ. (1988). Statistical Power Analysis for the Behavioral Sciences, 2nd edn, 567 pp. Hillsdale, NJ, USA: Lawrence Erlbaum Associates.
8.
KirkpatrickS., GelattC., and VecchiM. (1983). Optimization by simulated annealing.Science220, 671–680.
9.
JacksonJ.E. (1991). A User's Guide to Principal Components, 569 pp. New York: John Wiley and Sons.
10.
HöskuldssonA. (1988). PLS regression methods.Journal of Chemometrics2, 211–228.
11.
WoldS. (1995). PLS for multivariate linear modeling. In QSAR: Chemometric Methods in Molecular Design (ed. van de WaterbeemdH.), pp. 195–218. Weinheim, Germany: Verlag Chemie.
12.
HellbergS., ErikssonL., JonssonJ., LindgrenF., SjöströmM., WoldS., EkwallB., Gómez-LechónM.J., ClothierR., AccomandoN.J., GrimesA., BarileF.A., NordinM., TysonC.A., DierickxP., ShrivastavaR.S., Tingsleff-SkaanildM., Garza-OcañasL., & FiskesjöG. (1990). Analogy models for prediction of human toxicity.ATLA18, 103–116.
13.
BoxG.E.P, & DraperN.R. (1987). Empirical Model-Building and Response Surfaces, 669 pp. New York: Wiley.
14.
BoxG.E.P., HunterW.G., and HunterJ.S. (1978). Statistics for Experimenters, 655 pp. New York: Wiley.
15.
DemingS.N., and MorganS.L. (1993). Data Handling in Science and Technology, Vol. 11, Experimental Design: A Chemometric Approach, 2nd edn, 454 pp. Amsterdam: Elsevier.
16.
LindgrenÅ., SjöströmM., and WoldS. (1996). QSAR modelling of the toxicity of some technical non-ionic surfactants towards fairy shrimps.Quantitative Structure Activity Relationships, in press.
17.
BlandM. (1995). An Introduction to Medical Statistics, 396 pp. Oxford: Oxford University Press.
18.
BoxG.E.P., and CoxD.R. (1964). An analysis of transformations.Journal of the Royal Statistical Society B26, 211–252.
19.
EfronB., and TibshiraniR.J. (1993). An Introduction to the Bootstrap, 436 pp. London: Chapman & Hall.
20.
ClemedsonC., McFarlane-AbdullaE., Anders-sonM., BarileF.A., CallejaM.C., ChesnéC., ClothierR., CottinM., CurrenR., DierickxP., FerroM., FiskesjöG., Garza-OcañasL., Gómez-LechónM.J., GuldenM., IsomaaB., JanusJ., JudgeP., KahruA., KempR.B., KerszmanG., KristenU., KunimotoM., KarenlampiS., LavrijsenK., LewanL., LiliusH., MalmstenA., OhnoT., PersooneG., PetterssonR., RoguetR., RomertL., SandbergM., SawyerT.W., SeibertH., ShrivastavaR., SjöströmM., StammatiA., TanakaN., Torres-AlanisO., VossJ.U., WakuriS., WalumE., WangX., ZuccoF., and EkwallB. (1996). MEIC evaluation of acute systemic toxicity. Part II. In vitro results from 68 methods used to test the first 30 reference chemicals and a comparative cytotoxicity analysis.ATLA24, Suppl. 1, 273–311.
21.
LinhardH., and ZucchiniW. (1986). Model Selection, 301 pp. New York: Wiley.
22.
HolzhütterH.G., and QuedenauJ. (1995). Mathematical modelling of cellular responses to external signals.Journal of Biological Systems3, 127–138.
23.
GuardabassoV., RodbardD., and MunsonP.J. (1987). A model-free approach to estimate relative potency in dose-response curve analysis.American Journal of Physiology252, E357–364.
24.
SnedecorG.W., and CochranW.G. (1980). Statistical Methods, 7th edn, 507 pp. Iowa, USA: Iowa State University Press.
25.
GelmanA., CarlinJ.B., SternH.S., and RubinD.B. (1995). Bayesian Data Analysis, 526 pp. New York: Chapman & Hall.
26.
FederP.I., LordoR.A., DiPasqualeL.C., BagleyD.M., ChudkowskyM., DemetruliasJ.L., HintzeK.L., MarenusK.D., PapeW., RoddyM., SchnetzingerR., SilberP.M., TealJ.J., WeiseS.L., and GettingsS.D. (1991). The CTFA evaluation of the Draize primary eye irritation test (Phase I), Hydroalcoholic formulations. Part 1, Statistical Methods.In Vitro Toxicology4, 231–246.
27.
ChatfieldC., and CollinsA.J. (1980). Introduction to Multivariate Statistics, 246 pp. London: Chapman & Hall.
28.
ManleyB.F.J. (1986). Multivariate Statistical Methods, a Primer, 232 pp. London: Chapman & Hall.
29.
DraperN., and SmithH. (1981). Applied Regression Analysis, 2nd edn, 736 pp. New York: John Wiley and Sons.
30.
GeorgeE.I., and McCullochR.E. (1993). Variable selection via Gibbs sampling.Journal of the American Statistical Association88, 881–889.
31.
SilvermanB.W. (1985). Some aspects of the spline smoothing approach to non-parametric regression curve fitting (with discussion).Journal of the Royal Statistical Society B47, 1–52.
32.
HastieT., and TibshiraniR. (1987). Generalised additive models: some applications.Journal of the American Statistical Association82, 371–386.
33.
AtkinsG.L. (1976). Tests for the goodness of fit of models.Biochemical Society Transactions4, 357–361.
34.
DraperD. (1995). Assessment and propagation of model uncertainty (with discussion).Journal of the Royal Statistical Society B57, 45–98.
