The aim of the present article is to introduce and discuss the problem of optimal modelling of the prognostic information provided by putative prognostic variables, possibly measured on a quantitative scale. A number of methodological aspects will be treated, with particular reference to the role of spline functions and artificial neural networks, which will be discussed in the context of the analysis of survival data. The problem of the evaluation and the choice of the optimal statistical models will be examined, with particular attention to the critical aspects related to the definition of prognostic indexes on the basis of the results of the selected models. Clinical examples in breast cancer on the evaluation of the prognostic impact of several tumor markers are provided. This paper is addressed to all researchers who are interested in the evaluation of the prognostic role of tumor markers, therefore we will stress the necessity of integrating the methodologies of biological, clinical and statistical research in the assessment of prognosis.
VerweijP.J.M., Van HouwelingenH.C.Cross-validation in survival analysis.Stat Med1993; 12: 2305–14.
4.
HilsenbeckS.G., ClarkG.M., McGuireW.Why do so many prognostic factors fail to pan out?Breast Cancer Res Treat1992; 22: 197–206.
5.
HillC.Valeur prognostique d'une variable continue et point de cèsure optimal.Bull Cancer1993; 80: 649–52.
6.
FaraggiD., SimonR.A simulation study of cross-validation for selecting an optimal cutpoint in univariate survival analysis.Stat Med1996; 15: 2203–13.
7.
SilvestriniR., DaidoneM.G., BeniniE., FarandaA., TomasichG., BoracchiP., SalvadoriB., VeronesiU.Validation of p53 accumulation as a predictor of distant metastasis at 10 years of follow-up in 1400 node-negative breast cancers.Clin Cancer Res1996; 2: 2007–13.
8.
HastieT., SleeperL., TibshiraniR.Flexible covariate effects in the proportional hazards model.Breast Cancer Res Treat1992; 22: 241–50.
9.
MarianiL., CoradiniD., BiganzoliE., BoracchiP., MarubiniE., PilottiS., SalvadoriB., SilvestriniR., VeronesiU., ZucaliR., RilkeF.Prognostic factors for metachronous contralateral breast cancer: a comparison of the Cox regression model and its artificial neural network extension.Breast Cancer Res Treat1997; 44: 167–78.
10.
GaspariniG., BarbareschiM., BoracchiP., VerderioP., CaffoO., MeliS., Dalla PalmaP., MarubiniE., BevilacquaP.Tumor angiogenesis predicts clinical outcome of node-positive breast cancer patients treated with adjuvant hormone therapy or chemotherapy.The Cancer Journal from Scientific American1995; 1: 131–41.
11.
GaspariniG., BrooksP.C., BiganzoliE., VermeulenP.B., BonoldiE., DirixL.Y., RanieriG., MiceliR., ChereshD.A.Vascular integrin αvβ3: A new prognostic indicator in breast cancer.Clin Cancer Res1998; 4: 2625–34.
12.
SigurdssonH., BaldetorpB., BorgA., DalbergM., FernöM., KillanderD., OlssonH.Indicators of prognosis in node-negative breast cancer.New Engl J Med1990; 322: 1045–53.
13.
MarubiniE., ValsecchiM.G.Analysing survival data from clinical trials and observational studies.Chichester: John Wiley & Sons, 1995.
14.
BoracchiP., MezzanotteG., MarianiL., ValagussaP., MarubiniE.Clinically useful measures to assess the effectiveness of treatments: hints for a proper choice with special emphasis on cancer research.Tumori1990; 76: 1–9.
15.
TherneauT.M., GrambschP.M., FlemingT.R.Martingale-based residuals for survival models.Biometrika1990; 77: 147–60.
16.
GaleaM.H., BlameyR.W., ElstonC.E., EllisI.O.The Nottingham prognostic index in primary cancer.Breast Cancer Res Treat1992; 22: 207–19.
17.
Van HouwelingenH.C., ten Bokkel HuininkW.W., van der BurgM.E.L., van OosteromA.T., NeijtJ.P.Predictability of the survival of patients with advanced ovarian cancer.J Clin Oncol1989; 7: 769–73.
18.
MarubiniE., MorabitoA., ValsecchiM.G.Prognostic factors and risk groups: some results given by using an algorithm suitable for censored survival data.Stat Med1983; 2: 295–303.
19.
IntratorO., KooperbergC.Trees and splines in survival analysis.Stat Methods Med Res1995; 4: 237–61.
20.
HarrellF.E., LeeK.L., MarkD.B.Tutorial in biostatistics. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.Stat Med1996; 15: 361–87.
21.
DurrlemanS., SimonR.Flexible regression models with cubic splines.Stat Med1989; 8: 551–61.
22.
CoradiniD., BiganzoliE., BoracchiP., BombardieriE., SeregniE., De PaloG., MartelliG., Di FronzoG.Effect of steroid receptors, pS2 and cathepsin D on the outcome of elderly breast cancer patients: an exploratory investigation.Int J Cancer (Pred Oncol)1998; 79: 305–11.
23.
DittadiR., BiganzoliE., BoracchiP., SalbeC., MioneR., GattiC., GionM.Impact of steroid receptors, pS2 and cathepsin D on the outcome of N+ postmenopausal breast cancer patients treated with tamoxifen.Int J Biol Markers1998; 13: 30–41.
24.
ToiM., GionM., BiganzoliE., DittadiR., BoracchiP., MiceliR., MeliS., MoriK., TominagaT., GaspariniG.Co-determination of angiogenic factors thymidine phosphorylase and vascular endothelial growth factor in node-negative breast cancer: Prognostic implications.Angiogenesis1997; 1: 71–83.
25.
FaraggiD., SimonR.A neural network model for survival data.Stat Med1995; 14: 73–82.
26.
HertzJ., KroghA., PalmerR.G.Introduction to the theory of neural computation.Redwood City, CA: Addison-Wesley, 1991.
27.
RumelhartD.E., McClellandJ.L.Parallel distributed processing: explorations in the microstructure of cognition. Vol. 1, Foundations.Cambridge, Massachussets: MIT press, 1986.
28.
RavdinP.M., ClarkG.M.A practical application of neural network analysis for predicting outcome of individual breast cancer patients.Breast Cancer Res Treat1992; 22: 285–93.
BurkeH.B.Artificial neural networks for cancer research: Outcome prediction.Semin Surg Oncol1994; 10: 73–9.
31.
SnowP.B., SmithD.S., CatalonaW.J.Artificial neural networks in the diagnosis and prognosis of prostate cancer: A pilot study.J Urol1994; 152: 1923–6.
32.
LiestolK., AndersenP.K., AndersenU.Survival analysis and neural nets.Stat Med1994; 13: 1189–200.
33.
BiganzoliE., BoracchiP., MarianiL., MarubiniE.Feed forward neural networks for the analysis of censored survival data: A partial logistic regression approach.Stat Med1998; 17: 1169–86.
34.
RipleyB.D., RipleyR.M.Neural networks as statistical methods in survival analysis. To appear in Dybowski R and Gant V. Artificial neural networks: Prospects for medicine. Landes Biosciences Publishers.
35.
BishopC.M.Neural networks for pattern recognition.Oxford: Clarendon Press, 1995.
36.
AkaikeH.Information theory and an extension of the maximum likelihood principle. In: PatrovB.N., CsakiF., eds. Proceedings of the Second International Symposium on Information Theory, Akademia Kiado, Budapest, 1973.
37.
VerweijP.J.M., Van HouwelingenH.C.Penalized likelihood in Cox regression.Stat Med1994; 13: 2427–36.
