Principles and Concepts of Cancer Screening

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