Chapter 22 – Evaluation of predictive genetic tests for common diseases: bridging epidemiological, clinical, and public health measures Tables

Human Genome Epidemiology (2nd ed.): Building the evidence for using genetic information to improve health and prevent disease

“The findings and conclusions in this book are those of the author(s) and do not
necessarily represent the views of the funding agency.”
These chapters were published with modifications by Oxford University Press (2010)

A. Cecile J. W. Janssens, Marta Gwinn, and Muin J. Khoury

 

Measure Description Formula
Table 22-1Overview of epidemiological, clinical, and public health measures in the evaluation of predictive genetic tests
Epidemiological evaluation
Genotype frequency Frequency of genotype in total population P(g) = g/N
Population risk Disease risk in total population P(D) = D/N
Penetrance Disease risk conditional on genotype status Formula 1
Relative risk Ratio of disease risks of carriers and noncarriers RR = P(D | G) / P(D | n G)
Odds ratio Ratio of odds of disease of carriers and noncarriers Formula 2
Risk difference Difference between disease risks of carriers and noncarriers RD = P(D | G)−P(D | n G)
Clinical validity
Sensitivity Proportion of carriers among affected Se = P(G|D)
Specificity Proportion of noncarriers among unaffected Sp = P(nG|nD)
False positive rate Proportion of carriers among unaffected FPR = 1-Sp = P(G|nD)
False negative rate Proportion of noncarriers among affected FNR = 1-Se = P(nG|D)
Positive predictive value Proportion of affected among carriers PPV = P(D|G)
Negative predictive value Proportion of unaffected among noncarriers NPV = P(nD|nG)
Clinical or public health utility
Likelihood ratio Ratio of the genotype frequency in affected and the genotype frequency in unaffected LRg = P(g|D)/P(g|nD)
Population attributable fraction Proportion of cases that is attributable to the genetic variant Formula 3
Number needed to treat Number needed to treat to prevent one case Formula 4
Number needed to screen Number of cases needed to screen to prevent one case Formula 5

Persons with the risk genotype are called “carriers”; those without the risk genotype are “noncarriers.” Persons who will develop the disease are called “affected”; those who will not are called “unaffected.” The letter “g” represents the number of persons with a given genotype, which can be either the risk genotype (G) or the referent genotype (nG). N = the total number of persons in a population; D = the number of persons who will develop the disease; nD = the number of persons who will not develop the disease; P = probability. The symbol “|” stands for “conditional on”: for example, P(D|G) means “the probability of disease conditional on the risk genotype,” or “the proportion of persons with the risk genotype who will develop disease.” The symbol “∩” denotes “and.” Although the examples refer to predictive testing for future disease, all measures can also be calculated for diagnostic tests that aim to identify persons with or without the disease.

[back to chapter]

 

Will develop disease Will not develop disease Total
Table 22-2Summary of studies of the ALDH2 Glu487Lys polymorphism and cancers of the colon and/or rectum
Carriers of risk genotype True positive False positive G
Noncarriers False negative True negative nG
Total D nD N

Persons with the risk genotype are called “carriers”; those without the risk genotype are “noncarriers.” The letter “g” represents the number of persons with a given genotype, which can be either the risk genotype (G) or the referent genotype (nG). N = the total number of individuals in a population; D = the number of individuals who develop the disease; nD = the number of individuals who will not develop the disease.

[back to chapter]

 

Disease Huntington disease Breast cancer Colorectal cancer Type 2 diabetes
Clinical scenario* Offspring of patients Offspring of mutation carriers General population
Table 22-3Examples of the evaluation of predictive genetic tests for monogenic and multifactorial diseases
Gene 4p16.3 BRCA1/2 MLH1/MSH2 PPARG CAPN10 TCF7L2
Marker CAG repeats P12A SNP44 rs7903147
Genotype definition At-risk Mutations Mutations Mutations PP TT TT
Referent PA/AA CC/CT CC/CT
Epidemiological evaluation
Genotype frequency At-risk 50% 50% 50% 73% (18) 62% (18) 7% (13)
Referent 50% 50% 50% 26% 38% 93%
Disease risk 50% 39% 38% 33% (14) 33% (14) 33% (14)
Penetrance At-risk 100% 65% (19) 70% (9) 36% 36% 49%
Referent 0% 13% (20) 6% (20) 24% 28% 32%
Odds ratio 12.9 40.1 1.77 (18) 1.45 (18) 2.05 (13)
Relative risk 5.13 12.6 1.49 1.29 1.54
Risk difference 100% 52% 64% 12% 8% 17%
Clinical validity and utility
Sensitivity 100% 84% 93% 80% 68% 10%
Specificity 100% 72% 76% 31% 41% 95%
False negative rate 0% 29% 24% 70% 59% 90%
False positive rate 0% 16% 7% 20% 32% 5%
Positive predictive value 100% 65% 70% 36% 36% 49%
Negative predictive value 100% 87% 94% 76% 72% 68%
Likelihood ratio At-risk 2.94 3.88 1.15 1.15 1.94
Referent 0 0.23 0.1 0.65 0.79 0.95
Clinical or public health impact
Population attributable fraction 100% 67% 85% 26% 15% 4%
Number needed to treat 1 2 2 8 12 6
Number needed to screen 2 4 3 11 20 84

[back to chapter]

Numbers are for illustration purposes only. Specific risk estimates may vary among populations. All calculations were performed according to the formulas from Table 22.2, by using the Risk Translator of the HuGE Navigator.
*Clinical scenario specifies the target population for the genetic testing.
Disease risks for offspring of mutation carriers, calculated as the average of the penetrances.

back to Chapter 22