INFLUENCE OF METHODOLOGY: Categoric Data

1
INFLUENCE OF METHODOLOGYCategoric Data
Immunohistology

• CASE EXAMPLE
• 68 year old woman with an enlarged right
  supraclavicular lymph node, found during yearly
  physical examination. There were no complaints
  elicited in a review of systems screening
  biochemical testing and chest radiographs were
  likewise unrevealing. An excisional lymph node
  biopsy was undertaken.

2
(No Transcript)
3
INFLUENCE OF METHODOLOGYCategoric Data
Immunohistology

• Immunostaining Results
• Keratin and CD45 stains were reported as
  negative, whereas S100 protein stains were
  interpreted as positive a diagnosis of
  metastatic melanoma was rendered and the patient
  was entered on a protocol study of interferon
  treatment.

4
S100 Protein
Pankeratin
Immunostaining Results Supraclavicular Lymph Node
5
INFLUENCE OF METHODOLOGYCategoric Data
Immunohistology

• Subsequent immunostains done after referral to
  another institution showed keratin-positivity,
  and mammograms then demonstrated a small
  peripheral lesion in the right breast with the
  characteristics of a primary mammary carcinoma.
  Fine needle aspiration biopsy of the latter mass
  supported that impression.

6
Repeated Pankeratin Immunostain
7
Fine Needle Aspirate of Right Breast Mass
8
INFLUENCE OF METHODOLOGYSemiquantitative Data
Immunohistology

• CASE EXAMPLE
• A 29 year old woman found a mass in her right
  breast by monthly self-examination the presence
  of a lesion was confirmed by mammography, which
  suggested a malignancy. Excisional biopsy
  demonstrated an Elston grade II/III ductal
  carcinoma which was immunohistologically
  ERP/PRP-positive and showed no angiolymphatic
  invasion. Margins were uninvolved, and low-grade
  intraductal carcinoma occupied only 5 of the
  specimen.

9
(No Transcript)
10
Estrogen Receptor Protein Immunostain in Invasive
Ductal Breast Carcinoma
11
INFLUENCE OF METHODOLOGYSemiquantitative Data
Immunohistology

• At the insistence of the patients surgeon, an
  immunostain for mutant p53 protein was
  obtained. It showed nuclear reactivity in 75-80
  of the tumor cells, and, on the basis of the
  latter finding, a modified radical mastectomy was
  performed. Although all nodes were negative,
  postoperative chemotherapy was also administered.

12
p53 Protein Immunostain in Invasive Ductal Breast
Carcinoma
13
INFLUENCE OF METHODOLOGYSemiquantitative Data
Immunohistology

• A frozen aliquot of tumor tissue that had been
  placed in an institutional tissue bank was later
  used for Southern blot and PCR analysis of
  putative p53 mutation NONE WAS FOUND

14
Wild Type p53
Patient Sample
15
SOURCES OF CLINICAL BIAS IN REPORTS ON
PROGNOSTIC MARKERS

• 1. Non-discriminatory inclusion of patients in
  protocols who have already failed accepted
  therapies
• 2. Unstratified comparison of patients with
  tumors of dissimilar grades and/or stages
• 3. MOST IMPORTANTLY-- indiscriminate comparison
  of patients who have received non-identical
  therapy or who are in vastly different
  demographic (and co-morbidity) groups

16
McGuire Guidelines for theEvaluation of
Proliferative Index in Invasive Breast Carcinoma
1. Presumptive biologic effect - yes 2.
Definitive study performed - probably
not 3. Control population bias - yes 4.
Methodologic validation - partial 5. Optimized
cut-off values - no 6. Reproducibility - partial
17
DEFINITIONS OF OUTCOMES PRINCIPLES THAT ARE
CRITICAL TO ASSESSMENT OF PROGNOSTIC FACTORS

• Outcome can be defined in a binary fashion (e.g.,
  patients are alive or dead recurrence-free or
  living with recurrence), or in a time-dependent
  fashion (e.g., overall vs. disease-free survival)
• In any comparison of prognostic factors that are
  defined in different patient cohorts, the measure
  of outcome must be the same
• Some factors may be prognostic in regard to one
  outcome measure but not another (e.g., factor X
  may correlate well with disease-free survival but
  not overall survival), OR they may be prognostic
  for one patient group (e.g., those with stage I
  tumors) but not others (e.g., those with stage
  II tumors)
• Surrogate end points for tumor-specific survival
  (the most reliable measure of outcome) are risky,
  but they unfortunately are used commonly because
  of the difficulty of following large numbers of
  patients for many years

18
McGuire Criteria for Evaluation of Putative
Prognostic Markers

• 1. Is there a presumptive biological effect
  caused by the analyte in question?
• 2. Is there a control population bias with
  regard to distribution of the analyte?
• 3. Has there been validation of the methodology
  used to evaluate the analyte?
• 4. Have optimized cutoff values for the
  analyte been determined by rigorous clinical
  testing?
• 5. Has the method used to evaluate the analyte
  demonstrated reproducibility?
• 6. Has the definitive clinical study been
  performed to determine efficacy of the analyte ?

19
McGuire Guidelines for theEvaluation of ERP/PRP
in Invasive Breast Carcinoma
1. Presumptive biologic effect - yes 2.
Definitive studies performed -
probably so 3. Control population bias - yes 4.
Methodologic validation - yes 5. Optimized
cut-off values - yes 6. Reproducibility - yes
20
McGuire Guidelines for theEvaluation of p53
Protein in Invasive Breast Carcinoma
1. Presumptive biologic effect - yes 2.
Definitive study performed - probably
not 3. Control population bias - yes 4.
Methodologic validation - partial 5. Optimized
cut-off values - no 6. Reproducibility - partial
21
McGuire Guidelines for theEvaluation of HER-2
Protein in Invasive Breast Carcinoma
1. Presumptive biologic effect - yes 2.
Definitive study performed - probably
not 3. Control population bias - yes 4.
Methodologic validation - partial 5. Optimized
cut-off values - no 6. Reproducibility -
incomplete
22
C-erbB-2/HER-2/Neu Protein in Invasive Breast
Carcinoma Problems Regarding Prognosis

• Several well-constructed clinicopathologic
  studies have now shown that HER-2 amplification
  is only of potential prognostic import in
  patients with low-grade, small (lt 3 cm), stage I
  carcinomas of the breast (lt10 of all invasive
  breast cancers)
• Detection of true HER-2 amplification is fraught
  with technical difficulties performance of
  antibodies to p185/c-erbB-2 protein may not
  correlate well one with another, or with results
  of FISH or quantitative PCR techniques
• False positivity by immunohistology may be seen
  in as many as 20 of cases

23
68 yo F with disseminated ductal breast
carcinoma failed conventional chemotherapy
gemcitabine radiotherapy
Metastatic tumor is immunoreactive for HER-2
will Herceptin be of benefit?
24
The 26th CAP Consensus Conference on Prognostic
Tumor Markers (Snowbird, UT, June 1994)

• Multidisciplinary meeting (pathologists,
  oncologists, surgeons, radiotherapists)
  convened to decide which prognostic information
  was considered state of the art for carcinoma
  of the breast, prostate, large intestine rectum
• Consensus statements were prepared which divided
  prognostictests for those neoplasms into 3
  groups
• Group I Prognostic factors that are
  well-supported biologically and clinically, with
  adequate outcome information to assure
  reliability
• Group II Prognostic factors that have been
  studied extensively both biologically
  clinically, but which are either technically too
  demanding for general application or unassociated
  with sufficient outcome data to assure
  reliability in general practice
• Group III Prognostic factors that have not yet
  been subjected to rigorous biological and
  clinical testing

25
The 26th CAP Consensus Conference on Prognostic
Tumor Markers (Snowbird, UT, June 1994)

• Grouping of Prognostic Information for Invasive
  Ductal Breast Carcinomas
• Group I (Approved recommended for clinical
  use) TNM stage, histologic tumor type, Elston
  tumor grade, ERP/PRP status (with either
  biochemical or immunohistologic methods)
• Group II (Promising, but not yet recommended for
  clinical use) Markers of tumor S-phase fraction
  (flow cytometry, Ki-67/PCNA, thymidine labeling
  index)
• Group III (Factors needing much more outcome
  analysis to completely assess suitability for
  clinical use) p53, c-erbB-2, angiolymphatic
  invasion by tumor, quantitative measurement of
  intratumoral angiogenesis, nm23 protein, c-myc
  protein, Ha/Ki/N-ras proteins, PS2 protein, EGFR
  protein, Heat shock proteins, int-2/hst/bcl-1
  proteins, Mutant RB-1 protein
• ( Publication Source Prognostic Factors in
  Cancer Hermanek P, et al., Eds, UICC/Springer,
  New York, 1995 pp. 165-176)

26
PREDICTIVE TISSUE-BASED TESTS IN THE EVALUATION
OF CARCINOMA OF THE BREAST

• Estrogen and progesterone receptor proteins (in
  reference to treatment response to tamoxifen and
  other hormonal antagonists)
• c-erbB-2/HER-2/neu gene amplification (in
  reference to potential therapeutic response to
  HerceptinR)

27
The HerceptinR Story and Invasive Breast Carcinoma

• In the early-1990s, Genentech Co. conceived the
  idea that a monoclonal antibody to the c-erbB-2
  growth factor receptor (HerceptinR ) could have a
  beneficial therapeutic effect on metastatic
  breast carcinomas
• After support from in vitro and animal studies,
  phase I II clinical trials were conducted with
  human patients
• Results from those trials showed that as a single
  agent, Herceptin produced a partial or complete
  remission in only 12 of patients, lasting for a
  mean of only 5.1 months subsequent combination
  treatment with chemotherapeutic agents has
  increased these figures to 49 5.3 months,
  prompting the FDA to certify Herceptin as an
  approved therapeutic agent

28
(No Transcript)
29
(No Transcript)
30
The HerceptinR Story, Part 2 Trouble in Paradise

• To establish criteria for patient eligibility for
  Herceptin treatment, the FDA has now been induced
  to certify FISH testing for HER-2 amplification
  (with reagents from Oncor Co. or Visys Labs
  only), as well as an antibody test with a
  heteroantiserum to p185/c-erbB-2 protein (from
  Dako Co.), although correlation between results
  of these techniques may be no better than 77
• It is particularly troubling that the antibody
  used to predict susceptibility to Herceptin is
  NOT Herceptin!!
• Genentech Co. has established a minimum
  threshold of Herceptin eligibility as 2
  positive immunostaining for HER-2 protein
  (without validation by FISH), but many labs have
  already begun to provide this information with
  non-FDA-approved antibody reagents and techniques

31
CONCLUSIONS

• 1. Evaluation of the literature on prognostic
  factors for malignancies should include the
  following requirements
• a. Data types should be defined
• b. Analytical methods should be
  cross-validated, without exception
• c. Journal reviewers (and journal readers)
  should require that all such studies aim to
  eliminate clinical biases, and that results
  have been subjected to rigorous statistical
  analysis using appropriate techniques adequate
  followup!

32
CONCLUSIONS

• 2. It should be anticipated that the ultimately
  advisable approach to this topic will involve
  construction of algorithmic decision-trees, in
  acknowledgment of the reality that there is no
  single transforming agent or prognostic
  factor that will determine the biology of any
  given neoplasm