Description of In Situ Tumors Reported to Cancer Registries, 19952005

1
Description of In Situ Tumors Reported to Cancer
Registries, 1995-2005

• Holly L. Howe, PhD
• Xiao-Cheng Wu, MD, MPH
• NAACCR Annual Conference
• San Diego, CA
• June 2009

2
Background

• In situ tumors are generally reportable to a
  population-based cancer registry for information
  on the full spectrum of tumor progression
• However, these tumors, are not used in cancer
  surveillance nor in cancer incidence statistics.
• Exceptions breast cancer in situ and bladder
  cancer in situ

3
Question?

• Is there value collecting reports that are not
  used?
• Particularly in a time of diminishing resources
  and increasing volume of reports due to the aging
  of the population.

4
Purpose 1 and 2

• Describe the volume of reports and their
  epidemiologic characteristics, i.e., variation by
  cancer site, year of diagnosis, age, sex,
  race/ethnicity, U.S. urban/rural location,
  histology, and affiliation
• For sites with sufficient counts, in situ tumors
  were compared with their invasive counterparts

5
Method

• Used the CINA Deluxe 1995-2005 dataset
• Describe the volume of reports and their
  characteristics
• Examined trend in volume of in situ reports.
• Compared in situ reports with invasive reports
  within cancer types/sites.
• Evaluate data quality of in situ tumor reports.

6
Definitions

• SEER program was defined by five areas Iowa,
  Connecticut, Hawaii, New Mexico, and Utah, as
  they were in the SEER program for the entire
  period, 1995-2005.
• All other U.S. states were defined as part of the
  U.S. NPCR program (including SEER metro in NPCR
  states).
• NHIA v2 was used for Hispanic ethnicity.
• AI/AN was enhanced through I.H.S. linkage for all
  years of data available on the file.

7
U.S. registries included in CINA, 1995-2005
8
Results Data Quality Issues

• Inconsistencies between assignment of in situ
  behavior and in situ stage codes.
• From 1995 to 2000, no inconsistencies
• From 2001-2003, no inconsistencies.
• For 2004-2005 cases, inconsistencies were found
  depending on whether in situ cases were selected
  by Derived Stage 2000 or by behavior, and the
  number of inconsistencies differed based on the
  selection method.
• We omitted all cases from the analysis with
  inconsistent in situ behavior and Derived Stage
  2000 codes.

9
DQ Recommendation Solution

• Feedback to the standard-setters has resulted in
  a modification of these tools EDITS so that the
  errors found in this study will be able to be
  rectified at the reporting source. Future
  investigators should be able to get a clean data
  set for study.
• Complete DQ report is posted on the NAACCR
  website under cancer research.

10
Results Volume of In situ Reports

• 836,298 in situ tumors were identified, or 5.8
  of the 14,425,739 tumors reported to the
  registries.
• The largest numbers and rates of in situ tumors
  occurred for cancer of the breast, melanoma, and
  colo-rectum.

11
Cancer In Situ Sites

• 13 sites where the rate ratio of in
  situ-to-invasive cases exceeded the average of
  6.14 for all sites combined
• floor of the mouth descending colon sigmoid
  colon rectum anus larynx melanoma breast
  vagina vulva penis ureter and eye.
• These accounted for 91 of all in situ cases.

12

Invasive
13
Invasive
RR
14
Grouped these 13 sites

• Screening Sites
• Breast
• Sigmoid colonDescending colon
• Anus
• Melanoma
• HPV Sites
• Anus
• Vagina
• VulvaPenis

• Early Awareness Sites
• Floor of Mouth
• Larynx
• Ureter
• Eye

15
(No Transcript)
16
(No Transcript)
17
Race
18
Trends in In Situ Reports

• Rates were statistically significantly increasing
  from 1995 through 2005 at 3.8 per year.
• Rates of invasive tumors were decreasing,
  although not significantly, at -0.14 per year.

19
Conclusion 1

• The number of in situ cases (836,298) is 5.8 of
  all cancers reported.
• The trend of in situ reports is significantly
  increasing over time.

20
Conclusion 2

• The rate of in situ tumors in cancer sites with
  screening opportunities are higher than the in
  situ rate for all cancers
• ?reflecting that these modalities do detect
  disease at the earliest time of disease
  progression?
• The rate ratio of in situ disease in cancer sites
  associated with HPV-risk was higher than average
• reflecting medical surveillance of high-risk
  populations resulting in earlier detection.

21
Conclusion 3

• The pattern of in situ tumors, and their
  descriptive characteristics, generally follows
  that of invasive tumors higher rates in invasive
  tumors would predict higher rates for in situ
  tumors.
• One notable exception was the higher in situ
  rates in women, primarily attributable to female
  breast cancer in situ reports.

22
Conclusion 4

• Collecting in situ reports as part of the full
  spectrum of disease progression is valuable.
• Benefit most likely outweighs costs since uses
  cannot not always be predicted and the need for
  the data is dynamic experience has shown need is
  not always anticipated or known.

23
Conclusion 5 cont'd

• A census enables us to be immediately responsive
  to
• the advent and adoption of early detection
  modalities
• changes in cancer risk and exposures
• emerging trends or disparities in specific
  population groups.
• Ex breast cancer in situ, HPV-related tumors
• Unforeseen changes can be identified through
  cancer surveillance across the disease progression

24
Recommendation

• Could we expand uses of the in situ data
  reported? Could they be useful in programs such
  as State/Provincial Cancer Profiles?
• Read the complete report of the results and the
  quality assurance report available from NAACCR.
• Check the Epi reports section on the NAACCR web
  site.

25
General Purpose

• The purpose of this study was to assess the
  quality and availability of in situ cancer data
  and explore the use of these data for
  surveillance research.

26
Purpose 3

• We compared VIN III (vulva intra-epithelial
  neoplasia), VAIN III (vagina intra-epithelial
  neoplasia), and AIN III (anus intra-epithelial
  neoplasia) cases by state and national program.
• ACOS/COC cases of ceased reportability to
  hospital registries 01/96.
• May affect completeness when they are reportable
  to the central cancer registry.

27
Method

• Tables for all sites combined include leukemias
  and lymphomas but are not listed separately due
  to the non-existence of in situ tumors.

28
Method

• Rates for all sites combined for both invasive
  and in situ rates (and female genital system) do
  not include them.
• Rates and counts were suppressed when the
  category had fewer than 25 cases.

29
Caveat 1

• Since cervical cancer in situ was not a
  reportable disease for most of the study years,
  1995-2005, they were omitted from all analyses.
• Rates for all sites combined for both invasive
  and in situ rates (and female genital system) did
  not include them.

30
Caveat 2

• Canadian data had to be excluded due to
  variation by province, year, site of in situ
  tumors in the dataset and the overall
  proportional volume of in situ tumors was vastly
  different from reports from the US

31
Conclusion 3

• In situ rates were lower in non-white
  populations children, young adults, and the
  elderly possibly related to specific cancer
  types occurring in these groups.
• That is, the cancer types most common in these
  categories are not the cancers that can occur
  (e.g., leukemia in children) or do occur (e.g.,
  prostate cancer) in an in situ stage.

32
Conclusion 4

• Differences were found between the two U.S.
  surveillance programs, perhaps attributable to
• data quality issues (e.g., the rates of prostate
  cancer in situ or large intestine NOS),
• more thorough case ascertainment (melanoma), or
• in differences in the underlying risks and cancer
  profiles in populations in the two programs.