How Changes in Central Cancer Registries are Impacting Cancer Research

Presented by:

Thomas C. Tucker, PhD, MPHDirector, Kentucky Cancer RegistryUniversity of Kentucky

29th Advanced Cancer Registrars WorkshopLouisville, KY– September 10-11, 2015

Kentucky Cancer Registry

Topics to be covered

• A very short history of the Kentucky Cancer Registry• The importance of population-based cancer research▫ Internal Validity▫ External validity

• How population-based cancer registries can improve cancer research ▫ Providing a Population-Based Sample Frame▫Making Rapid Case Ascertainment Possible▫ Serving as a Virtual Tissue Repository▫Measuring the last Phase of Translational Research

• The Kentucky model for moving evidence-based research into the population

• An example of the potential impact of implementing evidence-based research in the population

The importance of population-based cancer research

Two important concepts

• Internal validity•External validity

Animal Studies

genetically identical mice

Developed Disease

Did not Develop Disease

Exposed Animals

A B

Unexposed Animals

C D

Relative Risk = (A/A+B)/(C/C+D)

Randomized Clinical Trial

Randomized trial (Prospective)

Study Outcome Occurred

Did not Occur

Exposure or Intervention

A B

No Exposure or Intervention

C D

Relative Risk = (A/A+B)/(C/C+D)

Random Allocation

Internal Validity

• When differences between the experimental (exposed) group and the control group are completely accounted for, the study is said to have internal validity and causal inferences can be made.

• In other words, it is possible to determine whether the exposure causes some outcome (disease, etc.).

• Many have argued that “randomization” was the most important scientific advance of the 20th century.

• Why is it that the findings from randomized clinical trials with internal validity almost never have the same effect when they are applied to general populations?

External Validity

• When the findings from a research project or study can be generalized to some defined population, they are said to have external validity.

• Epidemiology (population science) provides the tools to explore external validity and many argue that moving from studies with strong internal validity to studies with strong external validity is the next step in advancing our scientific understanding.

• The continuum from research with strong internal validity to studies with strong external validity is also part of “Translational Research”.

Population-based Central Cancer Registries Can Enhance Cancer Research in Several Important Ways.

Geographic Area Covered by the Population-Based Cancer Registry

Cancer cases occurring each year

The Population-Based Cancer Registry

A scientific sample of data from cancer patients representing the population covered by the registry is provided to the researcher

Cancer research with strong “external validity”

(the ability to generalize the findings to the underlying

population)

Information gathered on each cancer case

1. As a Population-Based Sample Frame

Intimate Partner Violence and Cancer Disparities

• Research Question: Is IPV associated with delayed screening, late stage Dx or poor survival among women diagnosed with breast, colorectal or cervical cancer in Kentucky?

• Design: Population-based prospective cohort study

• Methods: KCR recruitment

• Population-based Sample: n=1,850

• Results: IPV (37% lifetime) Partner Interference (13%)

A novel population-based cohort study

Later Stage at Diagnosis

Suboptimal Treatment

Poorer Quality of Life

Poorer Survival

Direct Pathway

Indirect Pathway

IPV & Partner Interference with Cancer Treatment

Poverty, Risk behaviors, Access to care, Low education,Co-morbidity

Geographic Area Covered by thePopulation-Based Cancer Registry

Pathology Labs

The Population-Based Cancer Registry Receives EPath Reports for 95% of All New Cancer Cases Diagnoses each Year

A population-based sample of cancer cases provided to the researcher

Cancer research with strong “external validity”

(the ability to generalize the findings to the underlying

population)

Electronic Path (EPath) Reports for Cancer Patients Sent Automaticallyat the Time of Diagnosis

2. For Rapid Case Ascertainment

Biospecimens collected

High Arsenic, Chromium and RadonLevels and High Lung Cancer Rates in Appalachian Kentucky

(Top) Arsenic content and coal field locations in Kentucky; (Bottom) Incidence of lung cancer in the Appalachian versus Non-Appalachian region of Kentucky.

Environmental Exposure and Lung Cancer

A population-based lung cancer case-control study By Drs. Arnold, Orren, Mellon, and Huang

Epi, Blood, Urine, Hair,

Toenails,Radon level,

Tap water,Soil,GPS

Hypothesis: Lung cancer incidence is higher than expected from smoking alone and is associated with exposure to environmental carcinogens

57 – 95

96 – 102

103 – 116

117 – 150

Lung Cancer Rate per 100,000

Appalachia0-100

100-250

250-2200

Arsenic PPM

Study Area

Live lymphocytes

Environmental Exposure and Lung CancerA population-based lung cancer case-control study

By Drs. Arnold, Orren, Mellon, and Huang

Geographic Area Covered by the Population-Based Cancer Registry

Pathology Labs

A population-based sample of cancer case data and tissue are provided to the researcher

Cancer research with strong “external validity”

(the ability to generalize the findings to the underlying

population)

Formalin fixed paraffin imbedded tissue samples are collected from the labs by the Registry for a population-based sample of cases

3. As a Population-Based Virtual Tissue Repository

The Population-Based Cancer Registry

1. A retrospective cohort study using the registry as a virtual tissue repository to explore 5 distinct proteins associated with breast cancer recurrence

Determine whether female breast cancer patients who were disease-free following surgery but subsequently had a recurrence of their breast cancer within five years had altered levels or activity of one or any combination of five proteins compared to women who were disease-free and did not have a recurrence.

Purpose of the Study

The Five Proteins

• Par-4: pro-apoptotic tumor suppressor protein downregulated during breast cancer recurrence

• Wnt/-catenin: signaling pathway induces epithelial-mesenchymal transition (EMT)

and promotes metastasis

• SNAIL, TWIST: transcription factors, promote EMT promote cancer progression (metastasis) elevated in metastatic and recurrent tumors

• c-Abl, Arg: tyrosine kinases, oncoproteins promote survival, proliferation, and metastasis activity levels measured by pCrkL activity

A Retrospective Cohort Study

Using the Registry, all female breast cancer patients treated surgically between 2000 and 2007 at UK for their 1st Ca. and determined to be disease free were identified. Tissue blocks from the initial surgery were obtained for all of these patients, TMAs were constructed and stained to determine activity levels for each protein. This cohort was then traced forward in time to determine which patients recurred and which did not. (479 patients met the study criteria).

Comparisons were made between activity levels of each protein among the primary tumors of recurrent patients (62) and the non-recurrent patients (417).

For patients that recurred, comparisons were made between the tumor tissue at first surgery (the primary tumor) and the tissue taken at the time of recurrence. Tissue at the time of recurrence was only available for 22 patients.

2000-2007 2012

TWIST SNAIL PAR4 -catenin pCrkL Intensity HER2+ (n=36) HER2-/ER+/PR+ (n=282) H L HER2-/ER-/PR+ (n=32) HER2-/ER-/PR- (n=75) Allred Score HER2+ HER2-/ER+/PR+ H L HER2-/ER-/PR+ HER2-/ER-/PR- Intensity X Percent HER2+ HER2-/ER+/PR+ H L HER2-/ER-/PR+ HER2-/ER-/PR-

H: Significantly high in the primary tumor of recurrent patientsL: Significantly low in the primary tumor of recurrent patients

PROTEIN EXPRESSION IN PRIMARY TUMORS FROM RECURRENT VS. NON-RECURRENT PATIENTS

Preliminary Results

TWIST SNAIL PAR4 -CATENIN pCrkLIntensity HAllred Score H

Intensity x Percentage H

H: Significantly high at the time of recurrence in the tissue sample of patient who recurred compared to their primary tumors

PROTEIN EXPRESSION IN RECURRENT VS. PRIMARY TUMORS

Preliminary Results

pCrkL EXPRESSION IN RECURRENT VS. PRIMARY TUMORS

Preliminary Results

Conclusions

• Elevated levels of Twist, and low or not activated c-Abl/Arg (pCrkL) in HER2-/ER+/PR+ breast tumors may potentially serve as a novel biomarker for the recurrence of breast cancer.

• c-Abl/Arg was activated and over expressed in the tumors of patients at the time of recurrence. Inhibiting c-Abl/Arg activation may potentially prevent recurrence.

• It is difficult to imagine doing this study without using the Cancer Registry as a virtual tissue repository.

2. Pre-Invasive Cervical Cancer HPV Genotyping Study: Research Questions

• Are the HPV genotypes different for non-Appalachian white women diagnosed with pre-invasive cervical cancer (CIN-3) compared to Appalachian white women?

• Are the HPV genotypes different for non-Appalachian white women diagnosed with pre-invasive cervical cancer (CIN-3) compared to non-Appalachian black women?

• Are the HPV genotypes different for white women diagnosed with CIN-3 compared to white women diagnosed with AIS?

Study Population

• All Kentucky women age 18 or older diagnosed with CIN-3 or AIS between January 1, 2009 and December 31, 2012 were available for this study.

• The cases were divided into four strata (non-Appalachian white women, non-Appalachian black women, Appalachian white women and AIS cases).

• A total of 4903 pre-invasive cervical cancer cases fell into one of these strata.• 3099 were non-Appalachian white women diagnosed with CIN-3.• 290 were non-Appalachian black women diagnosed with CIN-3.• 1360 were Appalachian white women diagnosed with CIN-3.• 154 were white women diagnosed with AIS.• The recruitment goal was to obtain tissue specimens from a random sample

of 150 cases in each strata (for a total of 600 cases). A random sample of 200 cases was drawn from each strata except AIS to allow replacement for cases for which tissue could not be obtained.

Study Procedures• The Kentucky Cancer Registry (KCR) served as the “Honest Broker” for this

study.• KCR solicited the required tissue blocks from each of 48 the labs where the

tissue was stored. • Tissue blocks were sent directly to KCR, the blocks were anonymized and

given a unique code. • The coded blocks were cut by the Markey Cancer Center, Biospecimens and

Tissue Procurement Research Lab and the tissue specimens sent to CDC for genotyping.

• The original blocks were returned to the contributing lab by KCR.• All of the HPV genotyping was done at CDC.• The HPV genotype or types for each specimen and the associated unique

code were then returned to KCR and linked with other data in the pre-invasive cervical cancer surveillance database to create a research data set.

Pre-Invasive Cervical Cancer HPV Genotyping Study: # and % of Cases Currently Available for Analysis

Groupings Frequency + (% of 421)Available for study 421 (100%)

Appalachian white 121 (28.7%)

Non-Appal white 113 (26.8%)

Non-Appal black 105 (24.9%)

AIS white 82(19.5%)

Pre-Invasive Cervical Cancer HPV Genotyping Study:Research Question 1

Are HPV genotypes different for Appalachian white vs non-Appalachian white females?

HPVGenotype

Region % with genotype

p-value

Any oncogenic HPV

Appalachian 22% 0.0588

except 16 & 18 Non-Appalachian 34%

Pre-Invasive Cervical Cancer HPV Genotyping Study:Research Question 2

Are HPV genotypes different for non-Appalachian white vs non-Appalachian black females?

HPVGenotype

RACE % with genotype p-value

16 (alone or White 61.1% 0.0208

with any other HPV) Black 44.8%

16 alone White 46.9% 0.0120Black 29.5%

35 (alone or White 2.7% 0.0043with any other HPV) Black 13.3%

35 alone White 1.8% 0.0158Black 9.5%

Any oncogenic HPV White 33.6% 0.0539except 16 & 18 Black 46.7%

Pre-Invasive Cervical Cancer HPV Genotyping Study:Research Question 3

Are HPV genotypes different for white females with CIN-3 vs white females with AIS?

HPV Genotype CIN-3 Vs. AIS % with genotype p-value18 (alone or CIN-3 3.0% <0.0001

with any other HPV) AIS 39.0%18 alone CIN-3 1.7% <0.0001

AIS 32.9%31 (alone or CIN-3 10.3% 0.001

with any other HPV) AIS 0.0%31 alone CIN-3 7.3% 0.0084

AIS 0.0%52 (alone or CIN-3 6.8% 0.0829

with any other HPV) AIS 1.2%52 alone CIN-3 3.4% 0.1177

AIS 0.0%33 (alone or CIN-3 0.0% 0.0689

with any other HPV) AIS 4.3%

Preliminary Conclusions

• HPV oncogenic types other than 16 and 18 were more common among non-Appalachian white women diagnosed with CIN-3 compared to Appalachian white women

• HPV 16 alone or with any other HPV type was more common among white women diagnosed with CIN-3 compared to black women.

• HPV 35 alone or with any other HPV type was more common among black women diagnosed with CIN-3 compared to white women.

• HPV 18 alone or with any other HPV type was more common among white women diagnosed with AIS compared to white women diagnosed with CIN-3.

The final step in translational research is the broad based implementation of cancer research findings in the genral population and measuring the impact of these interventions.

From the Laboratory to the Population

Genes Cells Animals Humans Populations

Basic Science

ClinicalScience

Epidemiology

Translational Research

EXAMPLE

Quercitrin, a natural product from apple peel, is tested in an animal model to determine if it prevents UV exposure induced skin cancer

Randomized trials in human populations

Broad application of the findings to the general population

From the Population to the Laboratory

Genes Cells Animals Humans Populations

Basic Science

ClinicalScience

Epidemiology

And back again

Translational Research

Example

Epidemiologic studies show that colorectal cancer is excessively high in the Appalachian area of Kentucky and this same population has high exposure to arsenic and chromium.

Cell line and animal studies are conducted to determine if exposure to arsenic and chromium contributes to the onset of colon cancer.

Example

Cell line and animal studies are conducted to determine if exposure to arsenic and chromium contributes to the onset of colon cancer.

Human trials

Epidemiologic studies show that colorectal cancer is excessively high in the Appalachian area of Kentucky and this same population has high exposure to arsenic and chromium.

From the Laboratory or the Population

Genes Cells Animals Humans Populations

Basic Science

ClinicalScience

Epidemiology

Translational Research

The ultimate goal of translational cancer research is the adoption and wide-spread use of evidence-based research findings that significantly reduce the cancer burden in the population.

This includes the wide-spread implementation of evidence-based cancer control interventions.

The Kentucky Model for Moving Evidence-based Cancer Research Findings into to the Population

Kentucky Cancer Registry (KCR)

Kentucky Cancer Program (KCP)Kentucky Cancer Consortium (KCC)

Lung Cancer by Area Development District in KY, 2005-2009

Area Development

District

High School Education 2006-

2010

Current Smokers2001-2005

Age-Adjusted Incidence

Age-Adjusted Mortality

Overall Rank

Percent Rank Percent Rank Rate Rank Rate Rank

Kentucky River 65.6 1 35.7 1 124.7 2 99.8 1 5Big Sandy 69.0 3 35.5 2 131.7 1 96.2 2 8

Cumberland Valley 67.8 2 35.5 3 117.2 3 86.0 3 11Gateway 73.7 6 32.0 6 102.1 6 79.9 4 22

Buffalo Trace 73.3 5 33.0 4 96.9 11 78.3 5 25Barren River 78.6 8 31.8 7 105.8 4 78.0 6 25

Lake Cumberland 70.9 4 31.1 10 101.2 7 77.7 7 28Fivco 78.2 7 32.5 5 99.9 8 71.0 10 30

Green River 83.0 11 30.3 11 105.0 5 76.1 8 35Pennyrile 80.1 9 31.3 8 97.2 10 70.1 11 38

Lincoln Trail 82.7 10 31.1 9 96.3 12 66.4 15 46Purchase 83.0 12 28.5 14 97.7 9 69.4 12 47

Northern Kentucky 86.4 15 29.0 12 96.2 13 71.4 9 49Kipda 86.4 14 28.6 13 94.9 14 66.6 14 55

Bluegrass 84.7 13 28.2 15 92.6 15 68.0 13 56

Source: Kentucky Cancer Registry (KCR)

Source: Kentucky Cancer Registry (KCR)

Demographic Characteristics Contribute to

Risk Factors Contribute to

Incidence and Late Stage DX Contribute to

Cancer Mortality

Combining Data from Multiple Sources

Logic Model

What are the common sources of data that can be used for defining the cancer burden?

• Demographic data (Census U.S)• Risk factor data (BRFSS)• Incidence data (KCR)• Mortality data (State Vital Records)

Lung Cancer by Area Development District in KY, 2006-2010

Area Development

District

High School

Education (%)

2006-2010

Poverty Rate (%)

2006-2010

Smoking Rate (%)

2001-2005

Age-Adjusted Incidence Late Stage

Incidence %

Age-Adjusted Mortality

Number Rate Number Rate

U.S. 87.6 15.1 19.96 292,495 67.0 79.7 229,103 52.5Kentucky 81.0 17.4 30.4 23077 100.5 80.7 16701 73.2

Barren River 78.6 19.1 31.8 1569 105.8 83.1 1148 78.0Big Sandy 69.0 25.2 35.5 1155 131.7 82.9 835 96.2Bluegrass 84.7 16.9 28.2 3449 92.6 81.1 2510 68.0

Buffalo Trace 73.3 22.4 33.0 321 96.9 80.5 256 78.3Cumberland

Valley67.8 28.7 35.5 1590 117.2 81.6 1153 86.0

Fivco 78.2 19.5 32.5 866 99.9 79.1 613 71.0Gateway 73.7 25.2 32.0 442 102.1 79.5 342 79.9

Green River 83.0 15.5 30.3 1284 105.0 80.2 933 76.1Kentucky River 65.6 29.2 35.7 840 124.7 84.4 658 99.8

Kipda 86.4 14.3 28.6 4602 94.9 77.9 3223 66.6Lake

Cumberland70.9 24.3 31.1 1295 101.2 80.2 992 77.7

Lincoln Trail 82.7 14.8 31.1 1291 96.3 79.8 873 66.4Northern Kentucky

86.4 11.4 29.0 1921 96.2 80.8 1413 71.4

Pennyrile 80.1 18.5 31.3 1220 97.2 83.5 873 70.1Purchase 83.0 16.3 28.5 1232 97.7 81.2 879 69.4

Lung Cancer by Area Development District in KY, 2006-2010

Area Development

District

High School Education, 2006-2010

Current Smoker,

2001-2005

Age-Adjusted Incidence

Age-Adjusted Mortality

Overall Rank

Percent Rank Percent Rank Rate Rank Rate Rank

Kentucky River 65.6 1 35.7 1 124.7 2 99.8 1 5

Big Sandy 69.0 3 35.5 2 131.7 1 96.2 2 8Cumberland

Valley67.8 2 35.5 3 117.2 3 86.0 3 11

Gateway 73.7 6 32.0 6 102.1 6 79.9 4 22

Buffalo Trace 73.3 5 33.0 4 96.9 11 78.3 5 25

Barren River 78.6 8 31.8 7 105.8 4 78.0 6 25

Lake Cumberland 70.9 4 31.1 10 101.2 7 77.7 7 28

Fivco 78.2 7 32.5 5 99.9 8 71.0 10 30

Green River 83.0 11 30.3 11 105.0 5 76.1 8 35

Pennyrile 80.1 9 31.3 8 97.2 10 70.1 11 38

Lincoln Trail 82.7 10 31.1 9 96.3 12 66.4 15 46

Purchase 83.0 12 28.5 14 97.7 9 69.4 12 47Northern Kentucky

86.4 15 29.0 12 96.2 13 71.4 9 49

Kipda 86.4 14 28.6 13 94.9 14 66.6 14 55

Bluegrass 84.7 13 28.2 15 92.6 15 68.0 13 56

An Example

In 2001, Kentucky had the highest colorectal cancer incidence rate in the U.S. compared to all of the other states

In 2001, it was also noted that Kentucky was ranked 49th in colorectal cancer screening compared to all other states with the second to the lowest rate (34.7% of the age eligible population).

Using the process previously described, data about the burden of colorectal cancer was assembled and presented to each of the 15 District Cancer Councils. Following these presentations, all 15 of the District Cancer Councils implemented evidence based cancer control intervention programs aimed at increasing colorectal cancer screening for age eligible people living in their District.

What happened following the implementation of these colorectal cancer screening programs?

34.70%

43.90%

47.20%

58.60%

63.70% 63.70%65.70%

30%

35%

40%

45%

50%

55%

60%

65%

70%

2000 2002 2004 2006 2008 2010 2012

Colorectal Cancer Screening in Kentucky

49th 23rd

Source: http://cdc.gov/brfss accessed August 2014

P<.05Source: http://cancer-rates.info/ky, Accessed January 2014

66.768.2

68.8

65.4 65.264.2

61.2

59.3 59.558.1

57.2

54.4

52.5

50

52

54

56

58

60

62

64

66

68

70

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Rat

e

Year

Kentucky Colorectal Cancer Incidence(1999-2011)

P<.05Source: http://cancer-rates.info/ky, Accessed January 2014

22.6

23.6

22.5

24.2

23.0

20.5 20.519.6

20.6

19.1 19.0

17.5 17.4

15

16

17

18

19

20

21

22

23

24

25

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Rat

e

Year

Kentucky Colorectal Cancer Mortality(1999-2011)

A 24% reduction in colorectal cancer incidence anda 28% reduction in colorectal cancer mortality is a significant public health success.

It is important to note that we would not have seen this problem without data from our population-based Cancer Registry and we could not have measured the impact of our interventions without data from our population-based Cancer Registry

Thank You!

Questions?

Kentucky Cancer Registry