Principles and Concepts of Cancer Screening Lorelei Mucci, ScD Epidemiology of Cancer – Lecture 3 February 1, 2011
Jan 04, 2016
Principles and Concepts of Cancer Screening
Lorelei Mucci, ScD
Epidemiology of Cancer – Lecture 3
February 1, 2011
Cancer screening - definition
The examination of asymptomatic individuals to classify them as likely or unlikely to have the disease of interest
Goal is to reduce the morbidity and mortality associated with a cancer by diagnosing the disease at an earlier stage Early treatment is less intrusive, less prone to risk or
complications, and more effective than delayed
Cancer screening in the US
US Preventive Services Task Forces: Supports screening for:
Breast cancer (2009): biennial mammogram, women 50 – 74 years
Cervical cancer (2003): pap, women 18* – 64 years
Colorectal cancer (2008): colonoscopy/FOBT/sigmoidoscopy, men and women 50-74 years
http://www.uspreventiveservicestaskforce.org/uspstopics.htm#Ctopics
Cancer screening in the US
US Preventive Services Task Forces: Not enough evidence for:
Skin (2009)
Ovarian (2004)
Prostate (2011)
Lung (2004)
http://www.uspreventiveservicestaskforce.org/uspstopics.htm#Ctopics
Success of cancer screening
Effectiveness of screening for a specific cancer relies on a number of key attributes:
I. Suitability of the disease II. Suitability of the screening testIII. Suitability of the screening program in the population
selected for testingIV. Effectiveness of screening program
I. Suitable disease
A’ A B C Dbiologic detectable by clinical diagnosis severe illness death fromonset screening test due to symptoms develops (metastases) the disease
|-------|------------------|-------------|----------------| Preclinical phase Clinical phase
Preclinical phase
- Cancer is not symptomatic but is detectable
- Screening is not useful if:
• Cannot detect prior to onset of symptoms
• Treatment during preclinical phase is no more effective
- Length of preclinical phase depends on screening tool (clinical exam vs. biomarkers)
I. Suitable disease
Cancer prevalence preclinical phase Prevalence ≈ Incidence * Duration Cancer may not be suitable for screening if low incidence
or short duration in preclinical phase Heterogeneity in duration of preclinical phase Incidence of disease may be low in general population,
higher among subgroups E.g. smokers, genetic susceptibility Targeted screening
I. Suitable disease
Cancer mortality Expected rates of death/metastases should be sufficiently
high, AND Earlier intervention during preclinical phase leads to
substantial reductions in morbidity from the cancer and its treatment, and mortality
II. Suitable screening test
Relatively simple to administer and perform Rapid, both in its conduct and in its turn-around time to
obtain results Inexpensive: low cost to benefit ratio
E.g. mammography $25000 USD per life year saved Safe, and should cause as little discomfort or potential harm
Important since most screened will likely not have the cancer of interest
Acceptable to target population E.g. Colonoscopy
Balance of benefit and harm from false positive
II. Suitable screening test
Correctly identify as positive individuals who have cancer and as negative those without cancer: reliability and validity
Reliability Ability of test to give same result on repeated applications
in the same person Depends on the intrinsic variability of the factor being
measured, the variability of the method used, the skill of performing the measurement, and the accuracy of interpretation of its value
Validity: The 2x2 Table
Gold Standard
Disease Present
Disease Absent
Total
Screening Test Result
Positive a b a + b
Negative c d c + d
Total a + c b + d
II. Suitable screening test
False positive
False negative
Gold Standard
Disease Present
Disease Absent
Total
Screening Test Result
Positive a b a + b
Negative c d c + d
Total a + c b + d
Sensitivity
Sensitivity is “positivity” of the test in the presence of disease; indicates % individuals with disease correctly classified by screening test as having disease
Drives number of false negatives
Estimate of sensitivity should be regarded as the sensitivity of the screening test relative to the diagnostic test
Gold Standard
Disease Present
Disease Absent
Total
Screening Test Result
Positive a b a + b
Negative c d c + d
Total a + c b + d
Specificity
Specificity is the probability that individuals without the disease will be correctly classified by the screening test as being disease-free
Drives number of false positives
Sensitivity and Specificity
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Screening test result
Nu
mb
er
of
ind
ivid
ua
ls
Disease-free
Cancer
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1
1 - Specificity
Se
ns
itiv
ity
Receiver Operator Curve
(False positive)
Gold Standard
Disease Present
Disease Absent
Total
Screening Test Result
Positive a b a + b
Negative c d c + d
Total a + c b + d
Positive Predictive Value
Positive Predictive Value (PPV) is proportion of individuals correctly classified as having cancer among total who test positive
Specificity drives false positives because number of non-diseased individuals greatly exceeds number with cancer. Small decreases in specificity may lead to large increase in false positives and a large decrease in the PPV
Many influence acceptability of test in population
Gold Standard
Disease Present
Disease Absent
Total
Screening Test Result
Positive a b a + b
Negative c d c + d
Total a + c b + d
Negative Predictive Value
Negative Predictive Value (PPV) is proportion of individuals classified as being disease-free who do not have the disease
Sensitivity drives false negatives
If NPV < 1 is similar to prevalence of preclinical disease, then screen test may miss most of preclinical disease
PPV and NPV are driven by sensitivity and specificity and prevalence of disease
III. Suitable screening program
Organized application of screening, early diagnosis and treatment activities in large groups is mass screening or population screening
Free of discomfort and attendant risks
Convenient and attractive to the target population
Economical, both to the individual and to society
High PPV and NPV
III. Suitable screening program
Improve suitability even if test characteristics fixed:I. PPV by increasing prevalence
Limit screening to high risk population Decrease frequency of screening interval
II. False positives Raise threshold for positivity Screen with two sequential tests, and “positive” only if positive on both
tests
III. False negatives lower the criteria for positivity Screen at more frequent intervals Screen with two different tests and consider “positive” if positive on at
least one
III. Suitable screening program
Critical that there be follow-up for individuals who test positive on screening
Protocol for quick follow-up with diagnostic testing Subsequent therapeutic intervention if the individual is
indeed found to have the disease For screening program to be successful at reducing
mortality, cases must be detected during the preclinical phase with time for treatment to be more effective than if treatment had been given at a later time availability and timeliness of effective treatment are critical components
III. Suitable screening programOptimal screening interval
During initial screen, cancer incidence will be high as prevalent pool is diagnosed
After one screen, incidence will drop to level lower than expected (without screening)
After 2nd screen, incidence will again rise Rate of increase will depend
on the interval since 1st screen If timing between tests too
short, preclinical cases will not be replaced
Cancer Rate
0
10
20
30
40
50
60
70
80
90
0 5 10 15
Time
Screening
III. Suitable screening program
PseudodiseaseA lesion that becomes known only as a result of screening; it would not be discovered otherwise
1) Cases that would never progress to a symptomatic state
2) Cases that would progress but are interrupted by death from an unrelated diseases before symptoms develop
25% of mammographically detected breast cancers, 50% of chest x-ray and/or sputum-detected lung cancers, and 60% of prostate-specific antigen–detected prostate cancers (Welch and Black, JNCI 2010)
Heterogeneity of cancer progression.
Welch H G , Black W C JNCI J Natl Cancer Inst 2010;102:605-613
Published by Oxford University Press 2010.
III. Suitable screening program
Pseudodisease
If screening is continued indefinitely, incidence rate may increase above baseline
Detection of non-progressive cancer that would not have otherwise come to light clinically and only arise as a result of additional screening
IV. Evaluation of screening effectiveness
Study design: Randomized trial Outcomes:
Improvements in overall survival Cancer-specific survival* Quality of life
Length of observation period: must account for the natural history of the cancer, time by which most screen-detected cases would die of cancer if they were not treated early.
Random allocation can take place at the individual or community level
Intention to treat analysis
IV. Evaluation of screening effectiveness
Study design: Observational studies Case – control study
Cases – Individuals who die of cancer of interest Controls – represent source population that gave rise to cases:
should they include only individuals with cancer? Controls should be sampled independently of whether they have
been screened Exposure window can be defined to approximate the detectable
preclinical period
IV. Evaluation of screening effectiveness
Number needed to screen (NNS): number need to be screened for a given duration to prevent one death NNS for haemoccult screening to prevent a death from colon
cancer is 1,374
NNS for mammography to prevent a death from breast cancer is 2451 among women aged 50-59 (50).
IV. Evaluation of screening effectiveness
There are possible sources of bias when evaluating a screening program that may result in a false picture of its efficacy
Lead-time bias Length-time bias
IV. Evaluation of screening effectiveness
Lead time bias: interval from disease detection with screening to the time at which diagnosis would have been made without screening; e.g. the amount of time by which the diagnosis was advanced due to screening
Distribution of lead times in a screening program depends on:--the sensitivity of the screening test--the frequency of the screening test--the incidence rate of disease--the distribution of the duration of preclinical disease
IV. Evaluation of screening effectivenessLength bias: Phenomenon by which screen-detected cases may not be representative of all cases
Screen-detected cases have longer preclinical phases, biologically slower progression, and better prognosis than cases detected through diagnostic evaluation
Screening program may appear artificially good because cases detected by screening may be less aggressive and higher proportion of indolent tumors found in screened group. Less aggressive forms of a disease are more likely to be picked up in a screening program because they have a longer detectable preclinical phase. Less aggressive forms of disease usually have better survival