Course Guide
Clinical trial essentials for
medical imagers
Sponsoring Organizations:
Radiological Society of North America, Oak Brook, IL
National Cancer Institute, NIH, Bethesda, MD
Compiled and edited by
C. Carl Jaffe, M.D. NCI
Click here for 2011 full participating faculty list
Concepts addressed in this course
How does a screening trial differ from a therapeutic trial?
How does a randomized control trial (RCT) differ from an observational trial?
Must the control arm of a therapeutic RCT always be a placebo?
What distinguishes Phase 1, 2, 3 and 4 trials?
What are ‘Phase 0’ trials and what role would imaging have in them?
What data do you need to determine a statistically valid study size?
When does a trial need a Data Safety Monitoring (DSM) board?
What functions do interim and futility analyses serve?
What are the definitions of type 1 and type 2 errors?
How do you constitute an imaging central reading review panel and how does it function?
What is the regulatory definition (e.g., FDA) of a “surrogate’ biomarker?
How do “prognostic” biomarkers differ from ‘predictive’ markers?
What happens if an ‘adverse event’ occurs to a study subject?
TABLE OF CONTENTS
Introduction................................................................................................................................................... 3
Trials: their types .......................................................................................................................................... 4
Diagnostic Technology Assessment Trials .................................................................................................... 5
Introduction to Therapy Clinical Trials ........................................................................................................ 6
Investigational Drugs and Devices ............................................................................................................. 14
Biases and Their Control ............................................................................................................................ 15
Assessing predictive value of tests .............................................................................................................. 18
Imaging as a Measure of Therapeutic Response ........................................................................................ 18
Ethics Considerations ................................................................................................................................. 18
Monitoring of Clinical Trials: IRB, DSM, CTEP ...................................................................................... 21
Sponsorship and Economics of Imaging Trials .......................................................................................... 23
Informatics Tools for Protocol Development.............................................................................................. 25
Bibliography ............................................................................................................................................... 27
Required reading, in advance of the workshop........................................................................................... 28
2
Introduction
What are clinical trials?
A clinical trial is a carefully designed research process. Studies are done to find out whether
promising approaches to disease screening, diagnosis, and treatment are safe and effective
<http://www.cancer.gov/clinicaltrials/learning/what-is-a-clinical-trial>.
Information Resources
The National Cancer Institute’s (NCI) Web site, cancer.gov, provides access to a wealth of information on
clinical cancer care. The site contains information from PDQ®, including the latest information about
cancer treatment, screening, prevention, genetics, supportive care, and complementary and alternative
medicine, as well as a registry of cancer clinical trials that is coordinated with clinicaltrials.gov. The latter
provides a listing of trials that the key medical publishers have agreed must have been pre-specified and
announced, without which the trial results will not be published.
At the cancer.gov and the National Library of Medicine (NLM) websites clinical specialists review current
literature from the medical journals, evaluate its relevance, and synthesize it into clear summaries, which
are reviewed monthly and updated based on new information. Most information summaries appear in
two versions: 1) a technical version for the health professional and 2) a non-technical version for patients,
their families, and the public. Many of the summaries are also available in Spanish. NCI also maintains a
telephone helpdesk for patients seeking cancer care and clinical trials information at 1-800-4-CANCER.
The NCI Website also includes approximately 100 fact sheets on various cancer-related topics,
information on ordering NCI publications, and educational features and news summaries concerning the
latest results from cancer clinical trials.
NCI’s clinical trials registry (PDQ) contains more than 1,800 ongoing clinical trials, including information
about studies around the world. The trials listed in the NCI PDQ site are transferred and harmonized
weekly with the clinicaltrials.gov website maintained by NLM. The latter site includes trials on a variety of
diseases besides cancer. Although no single resource lists every cancer clinical trial being conducted in the
United States and abroad, PDQ is the most comprehensive cancer clinical trials registry. It contains
information about trials sponsored by NCI, the pharmaceutical industry, and some international groups.
Users can narrow their search by multiple parameters, such as stage of disease, phase of trial, treatment
modality, and geographic location. PDQ also contains an archival file of more than 11,000 clinical trials
that are no longer accepting patients, including contact information for the principal investigators of trials
that may not yet be published in the biomedical literature. Demographics and incidence of cancer by state
and other parameters are accessible from the NCI SEER database.
The American Cancer Society (ACS) has a website that also provides information on cancer prevalence
and other information that can be helpful in determining the import and relevance during the planning of
a clinical trial.
3
Trials: their types
Trial types can be classified as:
 Treatment
 Prevention
 Screening
 Diagnostic
 Genetics
 Quality-of-life
Screening Trials
The ultimate goal of screening is to reduce the burden of advanced disease. For many types of disease,
detection and treatment of the disease at an early stage can result in an improved outcome. Single arm
trials can be useful for determining if a new screening test can detect disease earlier, e.g., before it is
symptomatic, and for determining the feasibility of screening. Non-invasive ways to detect disease at an
early stage include body fluids and imaging. Randomized control trials (RCT) composed of a screened
group and a control group is the most reliable method for determining whether earlier detection translates
into a reduction in disease morbidity or mortality.
Diagnostic Trials
Diagnostic tests are conducted to determine whether cancer is present, where it is located and its stage.
Some diagnostic trials compare two or more techniques to diagnose cancer and determine whether one
technique is superior. Genetic tests are being evaluated as diagnostic tools to further classify cancers so as
to direct therapies or improve treatments for people with specific genetic changes.
The Clinical Trial Protocol
Clinical trials follow strict scientific guidelines. These guidelines clearly state the study’s design and who
will be able to participate in the study (patient eligibility). Every trial has principal investigator(s). The
principal investigator prepares a plan for the study (a protocol) which describes the clinical trial in
complete detail. The protocol explains what the trial will do, how the study will be carried out, and why
each part of the study is necessary. It includes information on:
 The importance and rationale that justifies the study
 A statistically justified cohort size (for each arm)
 Eligibility criteria (requirements may involve type of cancer, general health, age, prior therapy)
 The intervention or therapy being tested (e.g. drug dosage, etc.)
 What tests participants will have and how often (a ‘patient calendar’)
 What information will be gathered about the participants and how their privacy will be protected
 The participants options if they choose to withdraw
 A readable consent form (at an eighth-grade readability level)
 The endpoints (primary and secondary) and correlative studies
 Methods of analysis
 Regulatory requirements and reporting obligations
4
Diagnostic Technology Assessment Trials
Some trials are primarily focused on imaging technology endpoints such as sensitivity for disease detection
or staging, etc. The trial design and its elements with thus encompass:
Scope of the diagnostic technology assessment process:
– Developmental level of technology
– Performance of technology, as measured by specific metrics.
Developmental stages of imaging technology
 Stage I (Discovery): Establishment of technical parameters and diagnostic criteria.
 Stage II (Introductory): Early quantification of performance in clinical cohorts, usually in single
institution studies.
 Stage III (Mature): Comparison to other modalities in large, prospective, multi-institutional clinical
studies (efficacy ).
 Stage IV (Disseminated): Assessment of performance in the community at large
setting (effectiveness).
Endpoints for diagnostic test evaluation
 Diagnostic performance (sensitivity, specificity, measures of accuracy and predictive value)
 Intermediate process of care:
– Diagnostic thinking/decision making
– Therapeutic thinking/decision making
 Patient outcomes: Quality of life, satisfaction, cost, mortality, morbidity
Studies of diagnostic performance are
 Needed in all phases of diagnostic modality evaluation.
 Should incorporate the assessment of intra- and inter-reader variability.
Facets of the effect on process of care include:
 Effects on choice of further diagnostic workup; impact on utilization and cost.
 Effects on patient management, such as decision to undergo surgery or to administration of
specific types of therapy.
 Techniques relatively easy to carry out.
Assessing impact on process of care
 Has important methodologic limitations:
– Confounding by temporal and institutional factors.
– Reproducibility of results?
– Generalizability of results?
– Need to take into account accuracy of diagnosis and appropriateness of care.
Empirical studies of patient outcomes are more feasible for early occurring outcomes.
Are subject to confounding by treatment for longer-term outcomes.
Magnitude of the effect could be very small, especially for distant outcomes.
5
Randomization is usually needed.
However use of randomization leads to inefficient designs for studying accuracy.
Introduction to Therapy Clinical Trials
Phases of Clinical Trials
Treatment (therapy) trials are usually classified as one of four phases, each designed to answer different
research questions.
Phase1Phase Phase2
Phase3
Phase4
1
Number of
participants
15-30 subjects
Less than 100
subjects
100 or more
Large subject
numbers
Purpose
• To find a safe
dosage
• To decide how the
agent should be
given
• To observe how
the agent affects the
human body
• To determine if the
agent or intervention
has an effect on a
particular disease
• To see how the
agent or intervention
affects the human
body
• To compare the new
agent or intervention
(or new use of a
treatment) with the
current standard
• To further
evaluate the
long-term safety
and
effectiveness
of a new
treatment
Recently, a new special purpose Phase 0 trial structure has been defined: new class of trial has bee
defined ca
Phase 0 trials
Number of participants: about a dozen
Purpose and Method: Using micro-doses to determine if ‘targeted’ therapeutic agents reach the
intended tissue. In early therapeutic development efforts, ‘Phase 0’ human clinical trials often use imageable, nuclear-tagged molecular compounds.
The following outline courtesy of Dr Tony Shields
Most therapeutic (e.g., drug) trials involve treatment of advanced disease.
Are often done with palliative intent, which are the simplest to perform.
Trials done with curative intent are generally adjuvant studies after surgery for solid tumors or
chemotherapy for leukemia/lymphoma and use regimens proven effective in advanced disease, but
require large phase II and III trials.
Prevention trials may include vitamin and hormonal therapies.
– Require extremely large patient populations and years to complete.
Clinical Therapeutic Studies
 Surgery is the backbone of effective treatment for most solid tumors and is usually done with
curative intent.
6


Radiation can be done with curative or palliative intent and these protocols use radiation with
curative intent, either as primary therapy or adjuvant treatment.
Chemotherapy trials generally use new drugs with palliative intent and may include single agents
or complex combinations (usually the latter). Successful use of new agents in the advanced setting
can lead to use earlier in therapy.
Sources of new protocol ideas
Arise from: 1) Extension of pre-clinical cellular or animal laboratory work 2) Inspired by a unique patient
or series of patients 3) A new imaging instrument is available 4) A new imaging probe is available 5) New
software or hardware have improved an older instrument 6) A common problem or disease seen at your
institution 7) A device or pharmaceutical company suggests a protocol
Design basics: you must critically evaluate:
– Feasibility coordinated with other research programs
– Adequate space
– Personnel availability/ commitment
– Competing protocols for patients or protocol complements other studies
– Can include patients enrolled in another protocol
– Will include a similar tumor type as an ongoing protocol with different eligibility
Protocol synopsis should address:
• Study objectives
• A basic description of the study design
• The number of subjects to be enrolled
• Summary of inclusion and exclusion criteria
• The dosage regimen or device utilization plan
• Planned study procedures
• The planned methodology for statistical analysis
Design: Objectives
• The primary objective is key: it will drive your design, statistics, and accrual goals
• Secondary objectives may include related measures of results (e.g. overall vs. progression free survival)
• Secondary objectives may look at completely different issues (e.g. cost analysis)
• Make sure the main objective is one that is clinically or scientifically meaningful
• Correlative studies are often included to expand the base of science knowledge of the disease
Statistical vs. Clinical Significance
• A good study will have validity that is both: Statistical – probability that a finding is true; and Clinical –
the contribution that the finding makes an impact on medical practice
Design: Endpoints
• Should be an identifiable clinical change indicating attainment of the goal
• Needed to define data and assess results in a consistent manner
• Must be objective and measurable
7
Design: Study Population
• Is the total number of subjects needed for clinical and statistical significance taking into account: 1)
Inclusion/ exclusion criteria 2) Demographic criteria 3) Diagnostic methods to determine clinical eligibility
Pitfalls in Protocols: Eligibility
• Write eligibility carefully-- or it will come back to bite you
• Do not include unnecessary conditions that you will be tempted to ignore, such as limits on:
– acceptable labs
– co-morbid conditions
– prior malignancies
– prior treatments
– results from standard imaging techniques (“measurable disease”)
• Conversely, make sure you include limits that would otherwise prevent enrollment of patients who
would be endangered by the study or give uninterpretable results
Screening Procedures Prior to Study Entry
• Make sure the timing of procedures is appropriate
– if screening is done too far in advance of enrollment; lab or imaging may have changed.
– if evaluation must be done just prior to the start you may need to repeat lab and imaging.
– generally, imaging can be done up to 4 weeks before treatment and labs within 7- 14 days.
– insurance will often not cover imaging if too frequent
– screening that is not part of standard of care may need to be paid for by the study.
Consent procedure:
Make sure patient and investigator sign consent before starting protocol or pre-enrollment screening
procedures that are not part of standard of care. The consent should be readable at an eighth-grade level.
It may contain components for opt-in or opt-out sub-consent signatures involving correlative studies or
data-sharing that are not part of the primary endpoint.
Study Structure Determines
• In a diagnostic trial how the test will be verified – type of treatment for each group – duration of
treatment – if more than one treatment, the order in which treatments will be administered
• Phase II Design
– all patients receive same treatment or procedure
– compare to historic controls
– imaging trials may compare to a “gold standard” of pathology or clinical follow up
• Two Group Studies
– Parallel: each group receives a different treatment or procedure from all other groups
– Crossover: each group receives a different treatment order from all other groups
Data Collection
Create a table that defines the clinical test sequence in a study calendar
• Data collection and management of a clinical trial currently accounts for up to 60% of the overall clinical
trial process
• Industry is pressured to test, approve, and market new drugs faster than ever so they are now
automating the documentation process relevant to the clinical trial and can cut the time of the trial by twothirds
8
Electronic Case Report Forms (CRFs) and Remote Data Capture (RDC)
• Streamlines the clinical trials process by capturing, authoring, processing and managing the clinical
documentation
• By standardizing, controlling and securing clinical trials-related content, the companies can now reduce
the costs of data acquisition and clean up during trials, and makes data much more accessible
• Internet-based Clinical Trials (RDC)
Has been recently introduced to replace paper-based processes. Clinical sites are beginning to embrace
this but it often requires them to retrain staff, but leads to more rapid data capture from electronic records
and data sharing, and has potential privacy implications.
Design: Budget
• All protocols cost money, even if it is just your own time but include regulatory and review costs,
physicians and nurses recruiting patients, reimbursement of patient expenses and modest compensation,
physicians and nurses to follow-up results of treatment or imaging, laboratory tests that are not part of the
standard of care, data management costs, imaging costs: machine time, tech support, supplies that are not
part of standard of care. Some of these costs may be hidden and not explicitly charged, but they are all real
• The protocol is the tool to develop the budget– look at all the tests on the study calendar; is
hospitalization needed (just drawing blood can be expensive) cost of data collection, validation, and
analysis. It identifies all study - plan related costs: time: staff spends with subjects & on protocol, varies;
procedures: cost per subject; overhead: dictated by institutional guidelines; divide costs into up front costs
for evaluation and review of the protocol: IRB charges etc., figure per patient costs. You may need to
separately charge for procedures if they are expensive (e.g., image guided therapy, extra scans etc.)
• Develop potential sponsors
– Internal seed money
– Grants from funding institutions, such as NIH program announcements, unsolicited
investigator initiated grants, SBIR grants or Industry such as pharmaceutical manufacturers,
medical device companies, contract research organizations
Pre-study Documents
Laboratory’s accreditation certificate
Current laboratory ranges
Form FDA1572 completed for all investigators (establishes their registration as investigators)
Primary Investigator (PI) and Co-investigator CVs
Financial Conflict of Interest Statements
Radiation Safety Application
IND number, if investigational drug
IDE number, if investigational device
IRB Submission Form
Cancer Center Submission Form
HIPAA compliance documentation
Informed consent
Protocol
Protocol Review
FDA review
• Required for investigational device unless it has “no significant risk”
9
• Always required for investigational drug
Cancer Center Protocol Review Committee (PRC)
Includes: Medical oncologists; Surgeons; Radiation oncologists; Radiologists (not always - but participation
is highly recommended); Statisticians; Nurses; Data managers
Radiology input in the PRC can be variable, so be prepared to have your protocol reviewed by clinical
oncologists.
Protocol Review
Protocol Review Committee (PRC) will assess:
• Scientific and clinical rationale
• Competing protocols and prioritization
• Ability to accrue patients
• Statistical review
• Treatment plan and safety (this group will have the better expertise than the IRB)
• Use of Cancer Center resources (funding)
Institutional Review Board (IRB)
• Safeguards the rights, safety and well-being of all trial subjects
• Reviews protocol, amendments, recruitment procedures, payment & compensation to the subject and
PI’s qualifications
• IRB or Human Investigation Committee includes: Physicians, scientists, statisticians, lay public but is less
likely to include radiologists
• Will assess: Safety -- but may have limited expertise for your specific protocol; Consent form -completeness, readability
Hopefully, your IRB has developed standard wording for common issues such as risks of standard
procedures, confidentiality issues, treatment of side effects or problems.
Design: Adequate Accrual
• Plan for accrual management; prepare remedies for contingencies
• Consider communication of protocol to physicians and practice groups – all communication must be
IRB approved
• Include talking points showing the merits of research
• Consider publicity for potential patients through brochures, posters, newsletters, the web and patient
support groups for specific diseases (IRB approved).
Design: Compliance & Discontinuance
• Ensure special measures are defined to monitor subject compliance
• Include procedures for subjects lost to follow-up
• Define criteria for voluntary and involuntary dismissal
Design: Protocol Revisions
• Despite careful planning they occur – certain aspects of the plan may be unworkable in practice –
circumstances change (adverse event, shortage of subjects)
• Implement when a change will reduce risk to subjects
• One time deviations may be allowed, but must be documented as protocol violations
10
Enrollment Decisions
• Initial intake interview
• Take the medical history
• Alert to medical barriers to participation, conditions listed in the exclusion criteria
• Warning signs that the patient may be unable or unwilling to comply with study procedural requirements
• Identify uncooperative attitudes or problems that may prevent adherence to study schedules - will reduce
instances of poor compliance later in the study
General Considerations
• FDA does not formally recognize protocol “deviations or exceptions.” All departures from the protocol
are considered “protocol violations” even if sanctioned by the sponsor.
• Nothing should be assumed. If it is not documented it theoretically did not happen.
Maintain Trial Files
• Investigator must set up study-specific files per patient in which all appropriate documentation is filed
• Files can be in binders or regular files but they must be coordinated
• All study documents must be stored in a safe, secure and confidential environment
Good Clinical Practice (GCP)
Source documentation should be able to withstand an audit; this includes data obtained from imaging
• First place a clinical observation is recorded
• Supports a study’s findings
• Legally valid raw data
• Establishes an audit trail
• Tells the complete story of the subject
Good Trial Files
• Treat the files as “gold” and ensure that they do not get lost or destroyed prematurely
• Once the study has ended, the PI is required to keep study-related documents until the sponsor tells
them it is no longer necessary
• Documents should be retained at least 2 years after the last approval of a marketing application
Points to Remember
• Source documents include medication charts, patient diaries, traces and printouts from automated
machines and even appointment diaries
• Make sure there is sufficient evidence in a subjects chart to justify patient eligibility
• Data must be recorded on study specific CRFs
• It must be possible to reconstruct CRFs using data recorded in the patient file
Documentation of Adverse Events
• An adverse experience (also called event, AE) is any untoward medical occurrence in a trial subject who
has been administered a pharmaceutical product
• AE may have no relationship to the treatment
• Prompt reporting to the sponsor, the IRB, and the FDA (if required)
• If AE is serious the PI must report it immediately; 21 CFR Part 312.64 (b) (called an SAE, e.g. death)
• In an imaging study one may limit AE to those that occur for a short time after time after the procedure
(e.g. 2 hours)
• For therapeutic trials usually report AE for up to 30 days after treatment
11
Safety Reporting
• Laboratory abnormalities may also be classified as AEs
• Good documentation needed to summarize medical occurrences of serious adverse event (SAE):
results in death; is life threatening; requires in-patient hospitalization; prolongs existing hospitalization;
results in persistent or significant disability or incapacity; is a congenital anomaly/birth defect
When in doubt report it!!!
• Regular Schemes for dose modifications are needed for chemotherapy trials when patients experience
adverse events.
– Dose modifications schemes can be complex when multiple agents are being used.
– They need to take into account the timing, severity, and recovery from the toxicity.
– They need to indicate which patients should be removed from the study, while keeping those
that may benefit from a lower dose.
Imaging’s roles in clinical trials:
Resectable or curable disease imaging may be limited to the detection of recurrence
– in some cases recurrence is documented by biopsy
– in some cases imaging alone is adequate proof of progression.
Imaging may be the primary or secondary endpoint.
The usual primary endpoints are: • Survival • Response (e.g. partial or complete response rate). (See
RECIST and WHO criteria) [Central review may be needed] • Progression-Free Survival (PFS)
When measuring response local reviewers, involved in patient care or unblended to the therapeutic arm,
have a bias to read scans in the patients favor.
Imaging Response to Treatment
• The protocol should define in detail how imaging will be done – include technical details such as use of
contrast or tracer, instrument specifications, training of readers, etc.
• Be careful to limit specifications to those that are essential or you may:: inadvertently disqualify patients–
limit the sites that can participate– interfere with the usual work flow in the Radiology Dept– limit the
generalizability of the study to a few “experienced” academic sites
• Many therapeutic protocols give no direction on how the imaging should be done or even the type (e.g.
“CT or MR measurement is needed”).
Imaging Response to Treatment: Problems
• RECIST (Response Evaluation Criteria in Solid Tumors) criteria are difficult to use as part of routine
care because: – scans are often done away from academic medical centers due to patient travel distance
and/or insurance – readers often don’t measure lesions, especially the 10 allowed by RECIST – measuring
irregular and ill defined lesions is arbitrary – the most recent scan is usually compared to the last scan, not
the pre-treatment or best-response scan – different readers may read scans differently over time – new
targeted therapies may not shrink lesions (e.g. the angiostatins)
• We regularly see reports that: describe the tumor are “slightly better or worse”; give measurements that
are not consistent with the previous reading; making it hard to tell what lesion or where a lesion was
measured
• Objectives: give measurements; look at previous measurement; record the slice used to measure the
lesions (it is always simpler when there are new lesions, since that is automatic progression)
12
Conclusions
Careful planning of clinical trials is essential to their successful conduct.
Make sure that you have thoughtfully defined:
• Eligibility
• Pre-study procedures
• Study calendar
• Treatment
• Dose adjustment
• Data collection forms
13
Investigational Drugs and Devices
“Investigational” means any drug or device not currently marketed in the US in the exact form that you
are using for the exact indication for which you are using it. For example, updated device software is
investigational, as is investigation of an marketed drug for a disease or histology that is not explicitly in the
drug labeling.
Investigational New Drugs (INDs)
21 CFR 312 (CFR = Code of Federal Regulations)
IND is required for all investigational drugs (except for a few radiopharmaceuticals)
Drug is not marketed in US
Drug is marketed, but this is a new indication or route of administration
Drug is a modification of existing drug (even the solution)
IND “Sponsor” can be: Company; Institution (e.g., hospital, NCI); Individual (usually physician)
Application must be made to the appropriate FDA Division (CDER, CBER…)
Specific data are needed for this application;
Pharmacology and toxicology information
Chemistry, manufacturing, and control
Dosimetry for radiopharmaceuticals
Important reporting requirements
o Serious adverse events
o Annual report
FDA Guidance on the IND process with multiple links to other documentation http://www.fda.gov/cder/regulatory/applications/ind_page_1.htm
Comprehensive FDA Guidance Page–http://www.fda.gov/cder/guidance/index.htm
A “how-to” guide from the Biological Development Program at NCI-Frederick with multiple links–
http://wwwbdp.ncifcrf.gov/pdf/GuidetoRegSubs.pdf
Investigational Device Exemptions (IDEs)
21 CFR 812 applies
IDE is required for investigational devices that pose a significant risk
Device is not marketed in US
Device is marketed, but this is a new indication or modification
“Significant risk” as determined by local IRB or by FDA
IND “Sponsor” is responsible which can be: Company, Institution (e.g., hospital, NCI), Individual
Application must be made to the appropriate FDA Division (CDRH, etc)
Important reporting requirements
o Serious adverse events
o Annual and semi-annual report
Information resources: FDA Center for Devices and Radiological Health
http://www.fda.gov/cdrh/index.html
Device Advice: http://www.fda.gov/cdrh/devadvice/
14
Biases and Their Control
Contributor: Nancy Obuchowski, PhD
What is “bias”? bias is any systematic deviation of observation / study result from a true clinical state
Examples of Bias
The (average of readers') sensitivity of MRI for detecting asymptomatic breast lesions is 0.80. But, in a
study by XX the estimated sensitivity is 0.96. The following examples will instruct:
How to control/correct for bias:
Design study to avoid bias
Implement strategies to minimize bias
Use statistical methods that adjust for bias
EXAMPLE 1: Risk factors for Cerebral Aneurysms
Design: Retrospective review of patients who underwent head CT
Sample: 500 patients; 50 with cerebral aneurysms
Goal: Determine if patients with aneurysm and patients without aneurysm differ on symptoms, family
history, demographics
- Selection and referral bias: Composition of sample is influenced by external factors. Here, sample is not
representative of population of patients with and without aneurysms.
Referral bias is a common type of selection bias in radiology studies. Study patients are selected for test
(CT) for various (unknown) reasons. Only “referred” patients included in study.
- Other types of selection bias
- Centripetal bias: Major medical centers with reputations in specialized areas often are referred problem
cases
- Popularity bias: Experts preferentially keep track of difficult cases over less challenging cases
- Referral filter bias: Tertiary care centers generate cases much different than general population.
Correction for referral bias is to prospectively recruit a random sample of headache patients from primary
care.
EXAMPLE 2: MRI detection of early rheumatoid arthritis
Design: Prospective study of patients with rheumatoid arthritis (RA). Recruited patients undergo MRI and
x-ray.
Sample size: 20 RA patients
Goal: Compare sensitivity of MRI and x-ray as a screening tool for early RA
- Generalizability issues and spectrum bias
Study sample does not include complete spectrum of patient characteristics. Here, study does not include
patients with early RA who have not yet been diagnosed. Pathways by which patients entered study and
characteristics of patients should be described so readers able to determine if results apply to patients in
their practice.
15
- Generalizability problems
Results of study have limited value because sample of patients (or readers), techniques used, etc. were
atypical. Here, MRIs and x-rays of only RA patients were interpreted - no patients without RA. No
measure of specificity!
- Correction
Redesign study to avoid or reduce bias (e.g. conduct prospective study of people who present with
symptoms of early RA)
- Verification Bias
Patients with + (or -) test results preferentially referred for gold standard. Bias occurs when accuracy is
estimated only on verified patients. Is most common type of bias in radiology studies.
- Corrections for Verification Bias
Design study so test results don’t influence which patients undergo gold standard
Use different (but equal) gold standards
Statistical corrections are available
EXAMPLE 3: Digital Mammography
Design: Prospective study of patients with lesions found on mammography with a new digital algorithm
applied.
Sample: 30 patients. If digital +, then undergo core biopsy; if -, then needle biopsy.
Goal: Assess accuracy of digital mammo to distinguish benign vs. malignant lesions.
- Imperfect Gold Standard Bias
The gold standard/reference procedure is not nearly 100% accurate. Needle biopsy is not nearly 100%
accurate.
- Work-Up Bias
Gold standards not equally accurate and one is associated with a particular test result. Core biopsy is much
more accurate than needle biopsy. Negative test results will be verified with much less reliability than
positive test results.
- Correction for Imperfect Gold Standard
If possible, re-design study to avoid use of imperfect gold standard Statistical methods exist to correct for
imperfect gold standard bias
EXAMPLE 4: Multi-reader study of breast cancer
Design: Prospective study of high-risk patients - all patients undergo mammo and MRI. Biopsy and/or
clinical follow-up is gold standard. Radiologists interpret all images.
Sample: 150 patients and 8 readers
Goal: Assess and compare accuracy of mammo and MRI
- Biases in studies with subjective test interpretation
Test-Review Bias
Diagnostic-Review Bias
Reading-Order Bias
16
- Test-Review Bias
Diagnostic test is performed or results are interpreted without blinding of gold standard results or
competing test results Need to ask were did readers blinded to the biopsy results and mammo results
when interpreting MR images?
- Diagnostic review Bias
Gold standard is performed or interpreted without blinding from results of test(s) under study. Need to
ask: is clinical FU assessed with blinding of MRI or mammo results? Is clinical FU also used in biopsy
patients?
- Reading Order Bias
When comparing 2 or more tests, readers’ interpretations are affected by memory of results from
competing test. Need to ask: did readers interpret 1/2 cases with mammo first and 1/2 with MRI?
- Corrections
MRI and mammo interpreted without knowledge of biopsy or FU results.
Clinical FU assessed without knowledge of MRI and mammo findings.
FU also used for biopsy patients.
Randomize reading order so 1/2 of cases seen with mammo first and other 1/2 with MRI first.
- Other Biases
Incorporation Bias
Context bias
Publication bias (meta-analyses)
17
Assessing predictive value of tests
Imaging as a biomarker
Identify disease in asymptomatic populations or individuals
Predicting disease course with or without therapy
Biomarkers for disease onset, presence or course of disease
Quantify risk assessment
Prognostic – predict disease outcome at time of diagnosis prior to therapy
Predictive – predict outcome of a particular therapy
Monitoring – measure response to treatment or relapse
Surrogate endpoint: substitutes for definitive endpoint to shorten trial or reduce cost
Prognostic markers do not interact with a particular treatment (so they might not be predictive)
Predictive markers can be used to make individual or group decisions about specific treatments (but a
predictive marker may not be prognostic if it does not predict outcome in untreated patients)
Contributor: Constantine Gatsonis
Imaging as a Measure of Therapeutic Response
Learning Objectives
Understand why and how RECIST imaging measurements are made and used
Describe alternative approaches to quantitative measurement of tumor response
Recognize the role of imaging as a correlative biomarker in drug testing
Report the regulatory meaning of imaging as a ‘surrogate’ marker in therapy trials
Ethics Considerations
Learning Objectives
• Review key historical events regarding participant protection
• Describe how participants are protected through the informed consent process
• Explain how review boards and panels protect participants
• Demonstrate familiarity with government regulations and agencies
• Knowing evolution of participant protection; origins: Nuremberg Code; Tuskegee syphilis study
• National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research
Belmont Report
National Research Act
• Understanding government oversight of safeguards for participants Office for Human Research
Protections (OHRP) – The Common Rule
• Understanding FDA regulations
• Protecting participants before a trial: Scientific review by sponsoring organization; Institutional Review
Board (IRB) approval; Informed consent
• Learning how to protect participants during a clinical trial: Institutional Review Boards (IRBs); Data and
Safety Monitoring Boards (DSMBs) act to minimize risks, ensure integrity of data and can stop study if
necessary
18
History of patient protection
Although we now have strong safeguards for protecting those who participate in research, these
protections resulted from abuses of human rights in the past. The first formal statement of protection for
individuals in research emerged from the Nuremberg trials in Germany, where Nazi scientists and
physicians who had conducted experiments on World War II concentration camp victims were convicted.
The Nuremberg Code outlined broad concepts for the protection of human subjects and forms the basis
of today’s international code of ethics for the conduct of research.
In the United States, several controversial research studies highlighted the critical need for protection
for those participating in clinical trials. None of these studies sought to inform the participants about the
research or gain their consent. From 1932 to 1972, the Tuskegee syphilis study followed low-income
African American men with syphilis but did not treat them. During the study, the men were offered free
medical care and were told that they would be treated for ”bad blood.”
In the 1960s, two other research studies received major public attention. The first was a series of
experiments with mentally retarded children; another involved debilitated elderly participants.
In response to these tragedies, regulations and policies were developed to ensure that people are told
about the benefits, risks, and purpose of research for which they volunteer. In 1976, the National
Commission for the Protection of Human Subjects of Biomedical and Behavioral Research developed
three basic principles governing research involving human subjects that were published in the Belmont
Report. These principles, which today form the basis for human subject protection regulations in the
United States are:
• Respect for persons—recognition of personal dignity and autonomy of individuals, and special
protection of persons with diminished autonomy
• Beneficence—obligation to protect persons from harm by maximizing unanticipated benefits and
minimizing possible risk of harm
• Justice—fairness in the distribution of research benefits and burdens
The Informed Consent Process
Informed consent is a critical part of ensuring participant safety in research. Informed consent is an
ongoing process during which potential participants learn important information about a clinical trial. This
information helps them decide whether to participate. The research team first explains the trial to
potential participants in understandable language. This explanation entails the trial’s:
• Purpose
• Procedures
• Risks and benefits
• Participant rights, including the rights to:
– Make an independent decision about participating
– Leave the study at any time without jeopardizing future treatment
Before agreeing to take part in a trial, people have the right to:
• Learn about all their treatment options
• Learn all that is involved in the trial—including all details about treatment, tests, and possible risks
and benefits
• Discuss the trial with the principal investigator and other members of the research team
19
• Learn if the investigators have any vested interest in the outcome of the study
The participant should hear and read the information in language they can understand.
Informed Consent Form
After discussing all aspects of the study with a potential participant, the team gives the person an informed
consent form to read. The form includes written details about the information that was discussed and also
describes the confidentiality of the participant’s records. If a person agrees to take part in the study, he or
she signs the form. Although informed consent documents can vary in their length and complexity, they
should all contain information on:*
• The clinical trial’s nature, purpose, and duration; the procedures to be followed; and which procedures
are experimental
• Reasonable, foreseeable risks and discomforts
• Benefits to the participants and to others
• Alternative procedures or treatments
• Confidentiality of records
• Procedures if the trial involves more than minimal risk (e.g., compensation, availability of medical
treatment)
• Contact for questions
• Voluntary participation—that there will be no loss of benefits on withdrawal and that participants may
stop participating at any time All Government-funded trials must contain this information by law.
–
* These informed consent requirements are listed in Title 45 CFR Part 46, Sub part A.
Consent Forms:
Make them easy to understand. The informed consent process can be effective only if potential
participants understand the information given to them. In recent years, both participants and investigators
have voiced concerns that informed consent documents for clinical trials were becoming too long,
complicated, and difficult to understand. NCI has issued recommendations designed to help research
institutions and clinical centers write user-friendly informed consent documents. Sample templates can be
found online in both English and Spanish in the clinical trials section of www.cancer.gov.
20
Monitoring of Clinical Trials: IRB, DSM, CTEP
Review Committees
Most clinical trials are subject to different types of review that are designed to protect all participants.
Clinical trials that are sponsored by NCI—whether funded by a grant, run by a Cooperative Group, or
run through a Cancer Center—are reviewed through different types of panels, including experts who
review the scientific and technical merit of the proposed research. Many other clinical trial sponsors, such
as pharmaceutical companies, also seek expert advice on the scientific and technical merit of their trial
protocols. In addition, all federally-funded clinical trials and all trials with investigational drugs or devices
must be reviewed by groups called Institutional Review Boards (IRBs). All trials with investigational drugs
or some investigational devices must be reviewed by the appropriate FDA division, except for highly
limited radiopharmaceutical trials that are reviewed by Radioactive Drugs Research Committees (RDRC),
which are under FDA oversight. Note that a marketed drug or device can be investigational and require
IND or IDE.
Institutional Review Boards (IRBs)
IRBs are made up of people who are qualified to evaluate new and ongoing clinical trials on the basis of
scientific, legal, and ethical merit. The board members determine whether the risks involved in a trial are
reasonable with respect to the potential benefits. IRBs also monitor the ongoing progress of trials from
beginning to end. Federal regulations require that each IRB be made up of at least five people; one
member must be from outside the institution conducting the trial. IRBs are usually made up of a mix of
medical specialists and lay members of the community, and many include members from diverse
occupations and backgrounds. In most cases, IRBs are located where the trial is to take place. Many
institutions that carry out clinical trials have their own IRBs. Federal law requires IRB approval for clinical
trials that are: 1) Federally funded 2) Evaluating a investigational drug, agent, or medical device or a new
indication or method of administration of a marketed drug or device, or 3) subject to FDA regulation
A number of institutions require that all clinical trials, regardless of funding, be reviewed and approved by
local IRBs. Potential participants considering a clinical trial should ask if it has been approved by an IRB.
During the Trial: IRB Monitoring
If the IRB grants approval for a trial, it also must decide how frequently the trial should be reviewed once
it is underway. Frequency is usually determined according to the degree of risk the trial involves. At least
once a year, the IRB must review a progress report provided by the clinical researcher in charge of the
trial. The report features information about how many people are enrolled and how many have
withdrawn, a description of participants’ experiences, including benefits and adverse effects, and the
progress to date. Based on this information, the IRB decides whether the trial should continue as
described in the original research plan, and, if not, what changes need to be made. An IRB can decide to
stop a clinical trial at any time if the researcher is not following requirements or if the trial appears to be
causing unexpected harm to participants.
Data and Safety Monitoring Boards, DSMBs
For phase 3 trials, DSMBs are appointed to help ensure participants’ safety. A DSMB may also be
appropriate and necessary for certain phase 1 and 2 clinical trials. The DSMB is an independent
committee made up of statisticians, physicians, and other expert scientists. The data and safety monitoring
board must:
• Ensure that any risks associated with participation are minimized to the extent practical and possible
• Avoid exposing participants to excessive risk
21
• Ensure the integrity of data
• Stop a trial if safety concerns arise or as soon as its objectives have been met
DSMBs also monitor all trial results. If early results show clear advantages of a new drug, the sponsor of
the study may choose to end the trial early and establish a protocol allowing wider use of the drug before
final approval for marketing. If a drug is shown to have a strongly negative effect, the trial is stopped
immediately. For example, in 1995, a trial of the drug tamoxifen (tamoxifen citrate) showed that the drug
dramatically reduced the short-term risk of breast cancer. The DSMB and the researchers assessed the
data and halted the trial so that the results could be made widely available and all women in the trial could
have the opportunity to take the drug. Researchers submitted a new application to FDA, which expedited
its review status. The new application was the basis for FDA approval of tamoxifen for reducing breast
cancer risk.
Office of Human Research Protection OHRP Regulations
Office of Human Research Protection protects those participating in research and provides leadership for
all Federal agencies that carry out research involving people. The OHRP enforces important regulations
for participant protection in clinical trials called the Common Rule.4 These regulations set standards
regarding the:
• Informed consent process
• Formation and function of IRBs
• Involvement of prisoners, children, and other vulnerable groups in research
FDA Regulations
FDA enforces another set of regulations on participant protection in clinical trials.5 It concerns any clinical
trial that involves an FDA-regulated drug or device, regardless of whether the trial receives Federal
funding. FDA periodically inspects IRB records and operations to certify the adequacy of approvals,
participant safeguards, and conduct of business.
Can a person be put in a clinical trial without his or her knowledge?
No ! The researchers running the trial are required by law to present and explain the study as part of the
informed consent process. This process includes:
• Signing an informed consent document (so that people know they are entering a study)
• Discussing with the research team what the trial entails
• Understanding the potential risks and benefits of participating
Although reputable researchers do not fool people or sign them up against their will, sometimes people
have difficulty understanding the information they need to know about a trial before agreeing to join. For
many people, it is important to ask a friend or family member to come with them to be sure that all
important questions are raised. Taking notes or using a tape recorder can also help.
If someone is in a phase 3 trial and it is found that there is a clear advantage for the
participants in the other group, what happens?
The DSMB would report the information to the study sponsor. If early results show that there is a clear
advantage for one of the groups, the sponsor may choose to end the trial early and establish a protocol
allowing wider use of the drug before final approval for marketing.
22
What happens if someone wants to stop participating in a trial?
Under the informed consent process, a person has the right to discontinue their participation in a trial at
any time. A participant’s decision to leave a clinical trial does not jeopardize future treatment, and the
participant will have the chance to discuss other treatments, or care with a doctor from the trial.
Sponsorship and Economics of Imaging Trials
NCI-Sponsored Clinical Trial Programs
• Clinical Trials Cooperative Group Program
• Clinical Trials Support Unit (CTSU)
• Community Clinical Oncology Program (CCOP)
• Minority-Based Community Clinical Oncology Program
• Cancer Centers Program
• Clinical Grants Program
Organizations That Sponsor Clinical Trials
NCI, pharmaceutical companies, medical institutions, and other organizations sponsor clinical trials. NCI
often partners with pharmaceutical companies to develop new agents. Regardless of sponsor, clinical trials
take place at universities, large medical centers, small hospitals, and doctors’ offices. Individual physicians
at cancer centers and other medical institutions can also sponsor clinical trials themselves.
NCI-Sponsored Clinical Trials
NCI sponsors many clinical trials that are conducted through four different programs:
1. Clinical Trials Cooperative Group Program
2. Community Clinical Oncology Program (CCOP) and the Minority-Based Community Clinical
Oncology Program (MBCCOP)
3. Cancer Centers Program
4. Clinical Grants Program
All NCI-sponsored trials must meet all FDA and Office for Human Research Protections (OHRP)
regulations for participant protection in clinical trials.
Clinical Trials Cooperative Group Program
Clinical trials are often conducted through NCI cooperative clinical trial groups, which are networks of
institutions that jointly carry out large clinical trials following the same protocols.
Members of these groups include: University hospitals, Cancer centers, Community physicians and
community hospitals
Cooperative groups develop and conduct new clinical trials that follow national priorities for cancer
research. They conduct phase 3 trials as well as phase 2 trials. Some of the groups are categorized by type
of cancer, others by type of treatment or other reasons, such as:
 American College of Surgeons Oncology Group (ACOSOG)
 Cancer and Acute Leukemia Group B (CALGB)
 Children’s Oncology Group (COG)
 Eastern Cooperative Oncology Group (ECOG)
 Gynecologic Oncology Group (GOG)
 National Surgical Adjuvant Breast and Bowel Project (NSABP)
 National Wilms Tumor Study Group (NWTSG)
 North Central Cancer Treatment Group (NCCTG)
23



Radiation Therapy Oncology Group (RTOG)
Southwest Oncology Group (SWOG)
American College of Radiology Imaging Network (ACRIN)
For more information about the Cooperative Group program, see
http://ctep.info.nih.gov.
Community Clinical Oncology Programs (CCOPs) and Minority-Based CCOPs
(MBCCOPs)
These programs allow community physicians to work with scientists conducting NCI-supported clinical
trials. Participation in the CCOP benefits lay people and health professionals in the community as well as
scientists in research centers. The MBCCOP provides members of ethnic and racial minorities with access
to state-of-the-art cancer treatment, prevention, and control technology.
Cancer Centers Program
NCI cancer centers conduct clinical trials under an NCI-approved protocol review and surveillance
mechanism. The Cancer Centers Program consists of more than 50 NCI-designated cancer centers
involved in many different cancer research efforts. Cancer centers also participate in at least one
cooperative group. List of Cancer Centers
Clinical Grants Program
Many clinical trial protocols are carried out under the direct support of an NCI peer-reviewed grant (often
labeled R21 grants)
Industry-Sponsored Trials
Pharmaceutical and biotech companies conduct their own trials, both locally nationally and internationally.
These partners may be universities, hospitals, NCI, or local doctors. These trials are subject to the
companies’ own review panels and to an IRB, which may be local or national in scope.
Referrals to Clinical Trials
Once someone is diagnosed with cancer, the health care provider may suggest several possible treatment
options, one of which may be a clinical trial. Similarly, health care providers may offer people at high risk
for cancer several options for prevention, including a clinical trial. If a person finds that his or her
physician does not participate in clinical trials, the person can request a referral to a physician who does.
24
Informatics Tools for Protocol Development
Valuable Information
Tie all your efforts together on your local PC: By using and mastering a combined approach through
integrated plug-ins: Word, Acrobat, PowerPoint, Endnote, PubMed, you will increase your information
access by copying and pasting – and make trial protocol development faster and easier
Learning Objectives:
Identify network resources that provide clinical-trial relevant templates
Use common desktop applications to speed trial document development
Reach time-saving resources that avoid duplicated efforts
Find regulatory guidance and funding sources
Understand how to use internet to interact with colleagues efficiently
There is value in learning in greater depth:
PubMed http://www.ncbi.nlm.nih.gov/entrez/ My NCBI is a central place to customize NCBI Web
services. To use it, you must first register, and your browser must accept cookies. You can use My NCBI
to: Save searches; Set up e-mail alerts for new content; Display links to Web resources (LinkOut); Choose
filters that group search results
What you can do with:
Quosa – http://www.quosa.com/ Connects automatically to PubMed and extracts full article PDFs if
you have an institutional license
Powerful document management and reusable material in:
Acrobat (Reader and Professional) – extract text (Select Tool) and images (Advanced ->export all
images); track changes (Comments & Markup). You can also publish a smaller, unchangeable PPT as a
PDF. You can drag-and-drop text from Acrobat to MSWord.
Learn it
Word – Convert to Adobe PDF using an Acrobat plug-in; Track changes (ctrl+shift+E); email
document as a draft for comment; add images; has hierarchical outline view. Since the patient consent
form must be readable at an eighth grade level, did you know that when you use Spelling & Grammar
check from the Tools menu in MS Word it will provide a readability statistics score? Do you also know
that invaluable plug-ins are available for integrating simultaneous document citation management available
from such programs as EndNote and RefManager?
Comprehensive reference management with Word and PubMed
Endnote – links to Word; does its’ own PubMed searching, constructs a complete bibliography. and
dynamically formats your references while you write
Selling your idea
PowerPoint (PPT) – you can incorporate and edit web pages and images for presentations
25
Templates are readily available – use them!
Find and use Clinical Trial Templates and protocols from NCI:
http://ctep.cancer.gov/guidelines/index.html
Clinical trials nation-wide are listed indetail in:
All government sponsored Cancer Clinical Trials and some privately sponsored trials are listed in
detail in http://www.cancer.gov/clinicaltrials/search and http://www.clinicaltrials.gov/
Regulatory approval of trial results
Find FDA Guidance about clinical trials (remember ClinTrials are mostly about therapies – not
imaging) http://www.fda.gov/oc/gcp/default.htm
Regulatory requirements for Investigational drugs or devices
FDA Guidance on the IND process with multiple links to other documentation:
http://www.fda.gov/cder/regulatory/applications/ind_page_1.htm
FDA Center for Devices and Radiological Health http://www.fda.gov/cdrh/index.html
FDA Device Advice: http://www.fda.gov/cdrh/devadvice/
Comprehensive FDA Drug Guidance Page–http://www.fda.gov/cder/guidance/index.htm•
A “how-to” guide from the Biological Development Program at NCI-Frederick with multiple links–
http://wwwbdp.ncifcrf.gov/pdf/GuidetoRegSubs.pdf
Regulatory guidance for Imaging drugs:
--Part 1: Conducting Safety Assessments:http://www.fda.gov/cder/guidance/5742prt1.pdf
–Part 2: Clinical Indications http://www.fda.gov/cder/guidance/5742prt2.pdf
–Part 3: Design, Analysis, and Interpretation of clinical studies
http://www.fda.gov/cder/guidance/5742prt3.pdf
26
Bibliography
Notes:
a. This bibliography corresponds roughly to the topics of the didactic portion of the workshop.
The list of references included here is meant to provide broad coverage of each topic. It is
not a required reading list for the workshop. The references in each section are listed in
reverse chronological order of publication
b. A small set of papers is highlighted as “required reading” (see below). Students are urged
to familiarize themselves with this material in advance of the workshop.
c. But in addition all students are urged to refresh their command of biostatistics concepts
before coming to the course.. The bibliography lists two handy sources for such material:
The text by Dawson and Trapp (see D1 below): Chapters 4-6 present basic material
you need to know. It will also be useful to have some familiarity with concepts from
correlation and regression analysis (Ch 8) and survival analysis (Ch 9).
The series of articles : Fundamentals of Clinical Research for Radiologists Am J
Roentgenol. 2001-2006 The hyperlinks below will give you the PDF files by just
performing a CTL-click to access the link


















Introduction to Probability Theory and Sampling Distributions Joseph 2003 AJR. 180:917
Clinical Evaluation of Diagnostic Tests Weinstein 2005 Am. J. Roentgenol. 184:14
Data Collection in Radiology Research Crewson 2001 Am. J. Roentgenol. 177:755
Meta-Analysis of Diagnostic and Screening Test Accuracy Evaluations: Gatsonis 2006 AJR.
187:271
Statistical Inference for Proportions Joseph 2005 Am. J. Roentgenol. 184:1057
Randomized Controlled Trials Stolberg 2004 Am. J. Roentgenol. 183:1539
Statistical Inference for Continuous Variables Joseph 2005 Am. J. Roentgenol. 184:1047
Population and Sample Kazerooni 2001 Am. J. Roentgenol. 177:993
Statistically Engineering the Study for Success Beam 2002 Am. J. Roentgenol. 179:47
Observational Studies in Radiology Blackmore 2004 Am. J. Roentgenol. 183:1203
Visualizing Radiologic Data Karlik 2003 Am. J. Roentgenol. 180:607
Reader Agreement Studies Crewson 2005 Am. J. Roentgenol. 184:1391
Radiology Cost and Outcomes Studies: Hollingworth 2005 AJR. 185:833
ROC Analysis Obuchowski 2005 Am. J. Roentgenol. 184:364
Survival Analysis Stolberg 2005 Am. J. Roentgenol. 185:19
Multivariate Statistical Methods Obuchowski 2005 Am. J. Roentgenol. 185:299
Correlation and Regression Dendukuri 2005 Am. J. Roentgenol. 185:3
Screening for Preclinical Disease Herman 2002 Am. J. Roentgenol. 179:825
27
Required reading, in advance of the workshop
1. Bossuyt P, Reitsma J, Bruns D, Gatsonis C, Glasziou P, Irwig L, Lijmer J, Moher D, Rennie
D, de Vet HC. Standards for Reporting of Diagnostic Accuracy steering group. Towards
complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative.
Academic Radiology, 2003;10:664-669.
2. Gatsonis C, McNeil B. Collaborative evaluation of diagnostic tests: Experience of the
Radiologic Diagnostic Oncology Group. Radiology, 1990;175:571-575.
3. Weinstein S, Obuchowski NA, Lieber ML. Clinical evaluation of diagnostic tests. Am J
Roentgenol. 2005 Jan;184(1):14-9
4. Obuchowski, N. ROC analysis. Am J Roentgenol. 2005 ;184:364-72.
5. Black WC, Welch HG. Screening for disease. AJR, 1997;168(1):3-11.
6. Emanuel EJ, Wendler D, Grady C. (2000). What makes clinical research ethical? JAMA.
283(20): 2701-11.
A. Diagnostic imaging evaluation: General
1. The Evidence Base of Clinical Diagnosis. JA Knottnerus (Ed.) London: BMJ Books, 2002.
2. Houn F, Bright RA, Bushar HF, Croft BY, Finder CA, Gohagan JK, Jennings RF, Keegan P,
Kessler LG, Kramer BS, Martynec LO, Robinowitz M, Sacks WM, Schultz DG, Wagner RF.
Study design issues in the evaluation of breast cancer imaging technologies. Academic
Radiology, 2000, Sep; 7(9):684-692.
3. Gatsonis C. Design of evaluations of imaging technologies: Development of a paradigm.
Academic Radiology, 2000;7:681-683.
4. Thornbury JR. Clinical efficacy of diagnostic imaging:. AJR 1994;162:1-8.
B. Introduction to Clinical Trials
nd
1. Piantadosi S. Clinical Trials: A Methodologic Perspective, 2 Ed, New York: Wiley, 2005.
2. Friedman LM, Furberg CD, DeMets DL. Fundamentals of Clinical Trials. 3 ed. St. Louis: Mosby,
1998.
3. Johnson, J. Williams G, Pazdur R. End Points and United States Food and Drug
Administration Approval of Oncology Drugs. J Clin Oncol 2005; 21:1404-1411.
28
4. Perrone F, Di Maio M, DeMaio E, Maione P, Ottaiano A, Pensabene M, Di Lorenzo G,
Lombardi AV, Sinoriello G, Gallo C. Statistical design in phase II clinical trials and its
application in breast cancer. Lancet Oncol, 2003;4:305-311.
5. Eisenhauer EA, O’Dwyer PJ, Christian M, Humphrey JS. Phase I clinical trial design in cancer
drug development. J Clin Oncol, 2000; 18:684-692.
C. Introduction to Biostatistics
1. Dawson B, Trapp RG. Basic & Clinical Biostatistics. 4th ed. Lange Medical Books-McGraw-Hill,
2004
nd
2. Pagano M, Gavreau K . Principles of Biostatistics 2 Ed, Duxbury Press, 2000.
D. Statistical methods for diagnostic test evaluation: General
Textbooks, Chapters, Tutorials
1. Articles in the series: Fundamentals of Clinical Research for Radiologists Am J Roentgenol. 20012006 (see hyperlinked PDFs above)
2. Toledano AY. Cancer diagnostics: Statistical methods. In Biostatistical Applications in Cancer
Research. Beam C (Ed.) Kluwer Norwell: Academic Publishers, 2002, 183-218.
3. Pepe M. The Statistical Evaluation of Medical Tests for Classification and Prediction. New York:
Oxford Press, 2003.
4. Zhou X-H, Obuchowski N, McClish D. Statistical Methods in Diagnostic Medicine. New York:
Wiley, 2002.
E. Statistical methods for diagnostic test evaluation: Special topics
Biases and Their Control
1. Reid MC, Lachs MS, Feinstein AR. Use of methodologic standards in diagnostic test research:
Getting better but still not good. JAMA, 1995;274:645-651.
2. Black WC. How to evaluate the radiology literature. AJR, 1990;154:17-22
3. Begg CB, McNeil BJ. Assessment of radiologic tests, control of bias, and other design
considerations. Radiology, 1988;167:565-569.
4. Ransohoff DJ, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of
diagnostic tests. NEJM, 1978;299:926-930.
29
Predictive Value of Tests
1. Sargent D., Conley B., Allegra C., Collette L. Clinical Trial Designs for Predictive Marker
Validation in Cancer Treatment Trials .JCO Mar 20 2005: 2020-2027.
2. McShane L, Altman DG, Sauerbrei W, et al. Reporting recommendations for tumor marker
prognostic studies (REMARK). J Natl Canc Inst, 2005;97:1180-1184.
3. Heagerty P, Zheng Y. Survival model predictive accuracy and ROC curves. Biometrics,
2005;61:92-105.
4. Fleming, T.R. Surrogate Endpoints and FDA’s Accelerated Approval Process: The challenges
are greater than they seem. Health Affairs 2005 ;24:67-78.
Meta analysis of diagnostic accuracy studies
1. Gatsonis C, Paliwal P. Meta-analysis of diagnostic and screening test accuracy evaluations: A
methodologic primer. AJR 2006; 187:271-281
2. Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-analytic methods for diagnostic test accuracy.
J Clin Epi, 1995;48:119-130.
3. Irwig L, Tosteson AN, Gatsonis CA, Lau J, Colditz G, Chalmers TC, Mosteller F. Guidelines
for meta-analyses evaluating diagnostic tests. Ann Int Med, 1994;120:667-676.
F. Quality of Life, Patient Satisfaction, Health Care Utilization Outcomes
1. Brazier J, Deverill M, Green C. A review of the use of health status measures in economic
evaluation. J Health Serv Res Policy, 1999;4:174-184.
2. Spilker B (Ed.). Quality of Life and Pharmacoeconomics in Clinical Trials. New York: Raven Press,
1996.
3. Torrance G, Furlong W, Feeny D, Boyle M. Multi-attribute preference functions: Health
Utilities Index. Pharmacoeconomics, 1995;7:503-520.
4. Ware JE Jr, Shelbourne CD. The MOS 36-item short-form health survey (SF-36): I.
Conceptual framework and item selection. Medical Care, 1992;30:473-483.
G. Imaging as a Measure of Therapeutic Response
1. Therasse P, Eisenhauer EA, Verweij J.RECIST revisited: a review of validation studies on
tumour assessment. Eur J Cancer. 2006 ;42:1031-9
2. Hallet W Quantification in clinical fluorodeoxyglucose positron emission tomography. Nucl
Med Commun. 2004 ;25:647-50.
30
3. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubinstein L, Verweij J, Van
Glabbeke M, van Oosterom AT, Christian MC, Gwyther SG. New guidelines to evaluate the
response to treatment in solid tumors. European Organization for Research and Treatment
of Cancer, National Cancer Institute of the United States, National Cancer Institute of
Canada. J Natl Cancer Inst. 2000; 92:205-16.
4.
Website for Q&A RECIST problem solving.
H. Imaging as a Predictor of Therapeutic Response
1. Sasaki R, Komaki R, Macapinlac H, Erasmus J, Allen P, Forster K, Putnam JB, Herbst RS,
Moran CA, Podoloff DA, Roth JA, Cox JD. [18F]flurodeoxyglucose uptake by positron
emission tomography predicts outcome of non-small-cell lung cancer. J Clin Oncol,
2005;23:1136-1143.
2. Fernandez FG, Drebin JA, Linehan DC, Dehdashti F, Siegel BA, Strasberg SM. Five-year
survival after resection of hepatic metastases from colorectal cancer in patients screened by
positron emission tomography with F-18 flurodeoxyglucose (FDG-PET). Ann Surg,
2004;240:438-447.
3. Weber WA, Petersen V, Schmidt B, Tyndale-Hines L, Link T, Preschel C, Schwaiger M.
Positron emission tomography in non-small-cell lung cancer: Prediction of response to
chemotherapy by quantitative assessment of glucose use. J Clin Oncol, 2003;2651-2657.
4. Kostakoglu L, Coleman M, Leonard JP, Kuji I, Zoe H, Goldsmith SJ. PET predicts prognosis
after one cycle of chemotherapy in aggressive lymphoma and Hodgkin’s disease. J Nucl
Med, 2002;43:1018-1027.
I. Imaging in clinical studies
1. Schnall M, Rosen M. Primer on imaging technologies for cancer. J Clin Oncol.
2006;24(20):3225-33. Review.
2. Barkhof F, Filippi M, Miller DH, Tofts P, Kappos L, Thompson AJ. Strategies for optimizing
MRI techniques aimed at monitoring disease activity in multiple sclerosis treatment trials. J
Neurol. 1997;244(2):76-84
J. Screening Studies: Methods
1. Hillman BJ, Black WC, Hauser B, Smith R. The appropriateness of employing imaging
screening technologies: Report of the methods committee of the ACR task force on
screening technologies. JACR, 2004;1(11):861-864.
2. Obuchowski NA, Graham RJ, Baker ME, Powell KA. Ten criteria for effective screening:
Their application to multislice CT screening for pulmonary and colorectal cancers. AJR,
2001;176(6)1357-1362.
31
3. Black WC. Overdiagnosis: An underrecognized cause of confusion and harm in cancer
screening. J Natl Cancer Inst, 2000;92(16)1280-1282.
4. Morrison AS. The natural history of disease in relation to measures of disease frequency. In
nd
Screening in Chronic Disease, 2 Ed. New York: Oxford University Press, 1992, 21-42.
K. Modeling for the Assessment of Screening Modalities
1. Claxton K, Cohen JT, Neumann PJ. When is evidence sufficient? Health Affairs,
2005:24(1);93-101.
2. Ramsey S, McIntosh M, Etzioni R, Urban N. Simulation modeling of outcomes and cost
effectiveness [Review]. Hematol Oncol Clin North Am, 2000;14(4):925-938.
3. Russell LB, Gold MB, Siegel JE, Daniels N, Weinstein MC. The role of cost-effectiveness
analysis in health and medicine. JAMA, 1996; 276(14)1172-1177.
L. Medical Ethics
1. Hellman S and Hellman D. (1991). Of mice but not men: problems of the randomized clinical
trial. NEJM 324:1585-9.
2. The National Commission for the Protection of Human Subjects of Biomedical and
Behavioral Research (1979). The Belmont Report: Ethical Principles and Guidelines for the
Protection of Human Subjects of Research. Washington, D.C.: Dept. of Health, Education,
and Welfare
M. Investigational drugs and devices






FDA Guidance on the IND process with multiple links to other documentation:–
http://www.fda.gov/cder/regulatory/applications/ind_page_1.htm
Comprehensive FDA Guidance Page–http://www.fda.gov/cder/guidance/index.htm
A “how-to” guide from the Biological Development Program at NCI-Frederick with
multiple links– http://wwwbdp.ncifcrf.gov/pdf/GuidetoRegSubs.pdf
Regulatory guidance for Imaging drugs:
Part 1: Conducting Safety Assessments www.fda.gov/cder/guidance/5742prt1.pdf
Part 2: Clinical Indications www.fda.gov/cder/guidance/5742prt2.pdf
Part 3: Design, Analysis, and Interpretation of Clinical studies
www.fda.gov/cder/guidance/5742prt3.pdf
FDA Center for Devices and Radiological Health http://www.fda.gov/cdrh/index.html
Device Advice: http://www.fda.gov/cdrh/devadvice/
N. Resources Available on the Web
1. http://www.fda.gov/cder/cancer/docs/overview.html
32