Background The field of cancer biomarker development has been evolving rapidly. New developments both in the biologic and statistical realms are providing increasing opportunities for evaluation of markers for both early detection and diagnosis of cancer.
Purpose To review the major conceptual and methodological issues in cancer biomarker evaluation, with an emphasis on recent developments in statistical methods together with practical recommendations.
Methods We organized this review by type of study: preliminary performance, retrospective performance, prospective performance and cancer screening evaluation.
Results For each type of study, we discuss methodologic issues, provide examples and discuss strengths and limitations.
Conclusion Preliminary performance studies are useful for quickly winnowing down the number of candidate markers; however their results may not apply to the ultimate target population, asymptomatic subjects. If stored specimens from cohort studies with clinical cancer endpoints are available, retrospective studies provide a quick and valid way to evaluate performance of the markers or changes in the markers prior to the onset of clinical symptoms. Prospective studies have a restricted role because they require large sample sizes, and, if the endpoint is cancer on biopsy, there may be bias due to overdiagnosis. Cancer screening studies require very large sample sizes and long follow-up, but are necessary for evaluating the marker as a trigger of early intervention.
Get full access to this article
View all access options for this article.
References
1.
1. Baker SG. Identifying combinations of cancer markers for further study as triggers of early intervention. Biometrics2000; 56: 1082–1187.
2.
2. Pepe MS, Etzioni R, Feng S et al. Phases of marker development for early detection of cancer. J Nat Cancer Inst2001; 93: 1054–1161.
3.
3. Etzioni R, Urban N, Ramsey S et al.The case for early detection. Nat Rev Cancer2003; 3: 1–10.
4.
4. Baker SG, Kramer BS, Prorok PC. Development tracks for cancer prevention markers. Dis Markers2004; 20: 97–102.
5.
5. Muller Hm, Oberwalder M, Fiegl H et al.Methylation changes in faecal DNA: a marker for colorectal cancer. Lancet2004; 363: 1283–1185.
6.
6. Traverso G, Shuber A, Levin B et al. Detection of APC mutations in fecal DNA from patients with colorectal tumors. N Eng J Med2002; 346: 312–320.
7.
7. Petricoin EF, Ardekani AM, Hitt BA et al. Use of proteomic patterns in serum to identify ovarian cancer. Lancet2002; 59: 572–577.
8.
8. Skates SJ, Horick N, Yu Y et al. Preoperative sensitivity and specificity for early-stage ovarian cancer when combining cancer antigen CA-125 II, CA 15–3, CA 72–4, and macrophage colony-stimulating using mixtures of multivariate normal distributions. J Clin Oncol2004; 22: 4059–4466.
9.
9. McIntosh MW, Drescher C, Karlan B et al.Combining CA 125 and SMR serum markers for diagnosis and early detection of ovarian carcinoma. Gynecol Oncol2004; 95: 9–15.
10.
10. Baker SG, Kramer BS, Srivastava S. Markers for early detection of cancer: statistical issues for nested case-control studies. BMC Med Res Methodol2002; 2: 4–4.
11.
11. Elwood M. Letter to the editor. Lancet2002; 360: 71–71.
12.
12. Michiels S, Koscielny S, Hill C. Prediction of cancer outcome with microarrays: a multiple random validation strategy. Lancet2005; 365: 488–492.
13.
13. Simon R, Radmacher MD, Dobbin K, McShane LM. Pitfalls in the use of DNA microarray data for diagnostic and prognostic classification. J Nat Cancer Inst2003; 95: 14–18.
14.
14. Baker SG. The central role of receiver operating characteristic (ROC) curves in evaluating tests for the early detection of cancer. J Nat Cancer Inst2003; 95: 511–515.
15.
15. Urban N, McIntosh M, Anderson M, Karlan BK.Ovarian cancer screening. Hematol Oncol Clin North Am2003; 17: 4–4.
16.
16. Etzioni R, Kooperberg C, Pepe M, Smith R, Gann PH. Combining biomarkers to detect disease with application to prostate cancer. Biostatistics2003; 4: 523–538.
17.
17. McIntosh MW, Pepe MS. Combining several screening tests: optimality of the risk score. Biometrics2002; 58:657–664.
18.
18. Pepe MS, Thompson ML. Combining diagnostic test results to increase accuracy. Biostatistics2000; 1: 123–140.
19.
19. Pepe MS, Cai T, Longton G. Combining predictors for classification using the area under the receiver operating characteristic curve. Biometrics2005; doi: 10.1111/j.1541-0420.2005.00420.x
20.
20. Liu A, Schisterman EF, Zhu Y. On linear combinations of biomarkers to improve diagnotic accuracy. Stat Med2005; 24: 37–47.
21.
21. Dudoit S, Fridlyand J, Speed TP. Comparison of discrimination methods for the classification of tumors using gene expression data. J Am Stat Assoc2002; 97:77–87.
22.
22. Pepe MS, Longton G, Anderson GL, Schummer M.Selecting differentially expressed genes from microarray experiments. Biometrics2002; 59: 133–142.
23.
23. Radmacher MD, McShane LM, Simon R. A paradigm for class prediction using gene expression profiles. J Comput Biol2002; 9: 505–511.
24.
24. Kapetanovic IM, Rosenfeld S, Izmirlian G.Overview of commonly used bioinformatics methods and their applications. Ann NY Acad Sci2004; 1020: 10–21.
25.
25. Shyr Y, Kim KM. Weighted flexible compound covariate method for classifying microarray data. In A practical approach to microarray data analysis. Kluwer Academic Publishers, 2003: 186–201.
26.
26. Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating efficacy of diagnostic tests. N Eng J Med1978; 299: 926–930.
27.
27. Soto AM, Sonnenschein C. Emergentism as a default: cancer as a problem of tissue organization. J Biosci2005; 30: 103–108.
28.
28. Etzioni R, Pepe M, Longton G, Chengchen H, Goodman G. Incorporating the time dimension receiver operating characteristic curves: a case study of prostate cancer. Med Decis Making1999; 19: 242–251.
29.
29. Crump KC, McIntosh MW, Urban N, Anderson G, Karlan, BY. Ovarian cancer tumor marker behavior in asymptomatic healthy women: implications for screening. Cancer Epidemiol Biomarkers Prev2000; 9: 1107–1111.
31. Morrell CH, Pearson JD, Carter HB, Brant LJ.Estimating unknown transition times using a piecewise nonlinear mixed-effects model in men with prostate cancer. J Am Stat Assoc1995; 90: 45–53.
32.
32. Slate EH, Turnbull BW. Statistical models for longitudinal biomarkers of disease onset. Stat Med2000; 19: 617–637.
33.
33. Skates CJ, Pauler DK, Jacobs I. Screening based on the risk of cancer calculation from Bayesian hierarchical change-point models of longitudinal markers. J Am Stat Assoc2001; 96: 429–439.
34.
34. Skates SJ, Menon U, MacDonal N et al. Calculation of the risk of ovarian cancer from serial CA-125 values for preclinical detection of postmenopausal women. J Clin Oncol2003; 21: 206s–210s.
35.
35. Lin H, Turnbull BW, McCulloch CE, Slate EH.Latent class models for joint analysis of longitudinal biomarker and event process data: application to longitudinal prostate-specific antigen readings and prostate cancer. J Am Stat Assoc2002; 97: 53–65.
36.
36. McIntosh MW, Urban N. A parametric empirical Bayes method for cancer screening using longitudinal observations of a marker. Biostatisitics2003; 4: 27–40.
37.
37. McIntosh MW, Urban N, Karlan B. Generating longitudinal screening algorithms using novel markers for disease. Cancer Epidemiol Biomarkers Prev2002; 11: 159–166.
38.
38. Ransohoff DE. Bias as a threat to the validity of cancer molecular marker research. Nat Rev Cancer2005; 5: 142–149.
39.
39. Baker SG, Tockman MS. Evaluating serial observations of precancerous lesions for further study as a trigger for early intervention. Stat Med2002; 21: 2383–2290.
40.
40. Imperiale TF, Ransohoff DF, Itzkowitz Sh, Turnbull BA, Ross ME for the Colorectal Cancer Study Group. Fecal DNA versus fecal occult blood for colorectal-cancer screening in an average-risk population. N Eng J Med2004; 351: 2704–2214.
41.
41. Thompson IM, Ankerst DP, Chi C et al. Operating characteristics of prostate-specific antigen in men with an initial PSA level of 3.0 ng/mL or lower. J Am Med Assoc2005; 294: 66–70.
42.
42. Barlow WE, Chi C, Carney PA et al. Accuracy of screening mammography interpretation by characteristics of radiologists. J Nat Cancer Inst2004; 96: 1840–1150.
43.
43. Smith-Bindeman R, Chu P, Miglioretti DL et al. Physican predictors of mammography accuracy. J Nat Cancer Inst2005; 97: 358–367.
44.
44. Day NE. Estimating the sensitivity of a screening test. J Epidemiol Community Health1985; 39: 364–366.
46. Dodd LE, Pepe MS. Partial AUC estimation and regression. Biometics2003; 59: 614–623.
47.
47. Schatzkin A, Connor RJ, Taylor PR, Bunnag B.Comparing new and old screening tests when a reference procedure cannot be performed on all screenees. Example of automated cytometry for early detection cervical cancer. Am J Epidemiol1987; 125: 672–678.
48.
48. Baker SG, Pinsky P. A proposed design and analysis for comparing digital and analog mammography: special ROC methods for cancer screening. J Am Stat Assoc2001; 96: 421–428.
49.
49. Alonzo TA, Kittelson JM. A novel design for estimating relative accuracy of screening tests when complete disease verification is not feasible. Biometrics; doi: 10.1111/j.1541-0420.2005.00445.x
50.
50. Miller AB, personal communication.
51.
51. Baker SG, Kramer BS, Prorok PC. Statistical issues in randomized trials of cancer screening. BMC Med Res Methodol2002; 2: 11–11.
52.
52. Gelfand AE, Wang F. Modeling the cumulative risk for a false positive under repeated screening events. Stat Med2000; 19: 1865–1179.
53.
53. Baker SG, Erwin D, Kramer BS. Estimating the cumulative risk of false positive cancer screenings. BMC Med Res Methodol2003; 3: 11–11.
54.
54. Xu J-L, Fagerstrom RM, Prorok PD, Kramer BS.Estimating the cumulative risk of a false-positive test in a repeated screening program. Biometrics2004; 60: 651–660.
55.
55. Warwick J, Duffy SW. A review of cancer screening evaluation techniques with some particular examples in breast cancer screening. J R Stat Soc Series A2005; 168: 657–677.
56.
56. Glaziou P. Meta-analysis adjusting for compliance: the example of screening for breast cancer. J Clin Epidemiol1992; 45; 1251–1156.
57.
57. Baker SG, Kramer BS. Simple maximum likelihood estimates of efficacy in randomized trials and before-and-after studies, with implications for meta-analysis. Stat Methods Med Res2005; 14: 349–367.
58.
58. Baker SG. Analysis of survival data from a randomized trial with all-or-none compliance: estimating the cost-effectiveness of a cancer screening program. J Am Stat Assoc1998; 93: 929–934.
59.
59. Cronin KA, Weed DL, Connor RJ, Prorok PC.Case-control studies of cancer screening: theory and practice. J Nat Cancer Inst1998; 90: 498–504.
60.
60. Connor R, Prorok PC, Weed DL. The case-control design and the assessment of the efficacy of cancer screening. J Clin Epidemiol1991; 44: 1215–1121.
61.
61. Suzuki K-J, Nakaji S, Tokunaga S, Shimoyam T, Umeda T, Sugawara K. Confounding by dietary factors in case-control studies on the efficacy of cancer screening in Japan. Eur J Epidemiol2005; 20: 73–78.
62.
62. Kopec JA, Goel V, Bunting PS et al. Screening with prostate specific antigen and metastatic prostate cancer risk: a population based case-control study. J Urol2005; 174: 495–499.
63.
63. Stevenson CE. Statistical models for cancer screening. Stat Methods Med Res1995; 4: 18–32.
64.
64. Boer R, Plevritis S, Clarke L. Diversity of model approaches for breast cancer screening: a review of model assumptions by The Cancer Intervention and Surveillance Network (CISNET) Breast Cancer Groups. Stat Methods Med Res2004; 13: 525–528.
65.
65. Goldie SJ, Kuntz KM. A potential error in evaluating cancer screening: a comparison of 2 approaches for modeling underlying disease progression. Med Decis Making2003; 23: 232–241.
66.
66. Berry DA, Kronin KA, Plevritis SK et al. For the Cancer Intervention and Surveillance Modeling Network (CISNET) Collaborators. Effect of screening and adjuvant therapy on mortality from breast cancer. N Eng J Med2005; 353: 1784–1192.
67.
67. Shwartz M. A mathematical model used to analyze breast cancer screening strategies. Oper Res1978; 26: 937–955.
68.
68. Chen HH, Duffy SW, Tabar L, Day NE. Markov chain models for the progression of breast cancer. Part II: prediction of outcomes for different screening regimes. J Epidemiol Biostat1997; 2: 25–35.
69.
69. Lee SJ, Zelen M. Modelling the early detection of breast cancer. Ann Oncol2003; 14: 199–202.
70.
70. Henson DE, Ries L, Shambaugh EM. Survival results depends on the staging system. Semin Surg Oncol1992; 8: 57–61.
71.
71. Shen Y, Yang Y, Inoue LYT, Munsell MF, Miller AB, Berry DA. Role of detection method in predicting breast cancer survival: analysis of randomized screening trials. J Nat Cancer Inst2005; 97: 1195–1203.
72.
72. Shen Y, Parmigiani G. A model-based comparison of breast cancer screening strategies: mammograms and clinical breast examinations. Cancer Epidemiol Biomarkers Prev2005; 14: 529–532.
73.
73. Baker SG, Erwin D, Kramer BS, Prorok PC. Using observational data to estimate an upper bound on the reduction in cancer mortality due to periodic screening. BMC Med Res Methodol2003; 3: 4–4.
74.
74. Baker SG, Kramer BS, Prorok PC. Comparing cancer mortality rates before-and-after a change in availability of screening in different regions: extension of the paired availability design. BMC Med Res Methodol2004; 4: 12–12.
75.
75. Baker SG, Lindeman KS. The paired availability design: a proposal for evaluating epidural analgesia during labor. Stat Med1994; 13: 2269–2278.
76.
76. Shaw PA, Etzioni R, Zeliadt SB et al. An ecologic study of prostate-specific antigen screening and prostate cancer mortality in nine geographic areas of the United States. Am J Epidemiol2004; 160: 1059–1169.
