WIXLIB

Chapter 1IntroductionLynn A. Gloeckler Ries, Marie-Josephe D. Horner, andJohn L. Young, Jr.INTRODUCTIONCancer Survey), the SEER Program has evolved in response to the mandate of the National Cancer Act of 1971,which requires the collection, analysis, and disseminationof data relevant to the prevention, diagnosis, and treatment of cancer. The SEER Program (http://seer.cancer.gov) collects cancer incidence, treatment, and survivaldata which are used to monitor the burden of cancer onthe population of the United States. The NCI contractswith medically-oriented nonprofit institutions, such asuniversities and state health departments, to obtain dataon all in situ and invasive cancers diagnosed in residentsof the SEER geographic areas, except for basal cell andsquamous cell carcinomas of the skin and in situ cervical cancer.The Surveillance, Epidemiology, and End Results (SEER)Program of the National Cancer Institute (NCI) has devotedthis monograph to examining cancer survival by patientand tumor characteristics for cancers diagnosed during theperiod 1988-2001. The analyses focus on cancer survivalin adults aged 20 years and older, with the exceptions ofacute lymphoblastic leukemia (all ages), placenta (ages15 ), and Hodgkin lymphoma (ages 15 ). This chapterdescribes the sources of the data and the methods used. Italso provides a summary of the results. Each subsequentchapter focuses on a distinct anatomical site and associated histologies.The analyses in this monograph are based on data from12 geographic areas representing approximately 14% ofthe United States population: the States of Connecticut,Iowa, New Mexico, Utah, and Hawaii; the metropolitanareas of Detroit, Atlanta, San Francisco, San Jose, LosAngeles, and Seattle; and ten counties in rural Georgia.Cases were diagnosed during the period 1988-2001 andfollowed through 2002. All registries contributed data fordiagnosis years 1988-2001, except Los Angeles, whichcontributed data for 1992-2001.DATA SOURCESSurveillance, Epidemiology, and End Results(SEER) ProgramThe Surveillance, Epidemiology, and End Results (SEER)Program was established in 1973 as part of the NationalCancer Institute (NCI). A sequel to two earlier NCI initiatives (the End Results Program and the Third NationalTable 1.1: All Cancers: Number of Cases and Exclusions, 12 SEER Areas, 1988-2001Number Selected/RemainingNumber ExcludedReason for Exclusion/Selection2,246,6030Select 1988-2001 diagnosis (Los Angeles for 1992-2001 only)1,925,529321,074Select rst primary only1,901,06724,462Exclude death certi cate only or at autopsy1,874,43226,635Exclude unknown race1,870,2294,203Active follow-up and exclude alive with no survival time1,846,16224,067Exclude children (000-019)1,736,210109,952Exclude in situ cancers for all except breast & bladder cancer1,660,37675,834Exclude no or unknown microscopic con rmation1,629,96430,412Exclude sarcomasNational Cancer Institute1SEER Survival Monograph

Chapter 1IntroductionA total of 1,629,955 primary cancers were used in analyses.Survival rates are calculated on demographic and tumorinformation. Cases of second or later primaries, casesidentified by death certificate or autopsy only, cases ofunknown race, and those alive with no follow-up wereexcluded from the analysis (Table 1.1).extension and lymph node fields are specific to the site ofthe primary tumor. The detail and amount of informationcollected for EOD have varied over time.StageStage of disease is determined from EOD information.In this monograph several different staging systems wereused depending on the extent of disease information available. The American Joint Committee on Cancer’s (AJCC)Staging Manual for the third edition (3), the fifth edition(4), and sixth edition (5) TNM: tumor size/extent (T),node involvement (N), and distant metastases (M) andthen combines TNM into stages. Sometimes additionalinformation is needed such as grade.The SEER data are available for analyses by researchers.See www.seer.cancer.gov for further information.SEER*Stat SoftwareThe SEER*Stat statistical software, a convenient, intuitive mechanism for the analysis of SEER and othercancer-related databases, was used for analyses. It is apowerful PC tool to view individual cancer records andto produce statistics for studying the impact of canceron a population. It is available at the following website:http://seer.cancer.gov/seerstat/Since 1988, the tumor extension information in EOD iscollected utilizing only one variable (except for prostatesince 1995) and is based on the best information availableon the furthest extension of the tumor. For some AJCCschemas, there is both a clinical T and a pathologic T.Therefore, in the conversion from EOD to AJCC, the Tinformation is based on a combination of clinical andpathologic information. If there are distant metastases,the SEER EOD conversion will be TX M1, i.e. the Tinformation is not recorded. Similarly, if distant nodesare involved, the information on regional nodes is notrecorded in SEER. For many primary sites AJCC tumorextension classifications can range from T0 to T4 withsubcategories, node involvement classifications can rangefrom N0 to N3 with subcategories, and metastasis classifications can range from M0 to M1. The AJCC T, N,and M are then combined into stage ranging from Stage0 through Stage IV. There are some primary sites forwhich there is no TNM and/or no AJCC stage. For allcancer sites except bladder and breast, in situ lesionswere excluded from the analyses. For most cancer sites,this means that Stage 0 is excluded, but for breast andcolon/rectum, Stage 0 includes more than in situ alone.For colon/rectum, Stage 0 also includes cases confinedto the lamina propria with no nodes and for breast, Pagetdisease with no underlying tumor.Tumor InformationThe SEER program collects the month and year of diagnosis, primary tumor site, behavior, histology, extent ofdisease at diagnosis, and, starting in 1990, breast cancerreceptor status. The International Classification of Diseases for Oncology, 2nd edition (ICD-O-2) (1) was thestandard reference for classifying primary site, histology,behavior and grade. The ICD-O-2 tumor site and morphology codes allow for precise coding of tumor location(including sub-location within an organ) and histology.For 2001 cases, the third edition of ICD-O (ICD-O-3)was used and all prior histology data were converted toICD-O-3 (2).The histologic grade of malignant tumors is also collected:grade I is well differentiated; grade II is moderately differentiated; grade III is poorly differentiated, and gradeIV is undifferentiated or anaplastic (1, 2). For leukemiasand lymphomas, the grade code can reflect T-cell, B-cell,and N-K cell phenotype.Extent of DiseaseTo perform the analyses in this monograph covering datafrom 1988-2001, it was necessary to achieve consistency ofthe stage variable over time. Changes to EOD were madein 1988 to be compatible with the AJCC third edition.In 1998, some of the EOD schemas were changed to becompatible with the fifth edition of AJCC so that SEEREOD information could be easily converted into the TNMstaging classifications based on the fifth edition of theAJCC Manual for Staging of Cancer. Therefore, dependingSEER has collected extent of disease (EOD) informationon all cancers since the inception of the program. Extentof disease information since 1988, consists of five dataitems: tumor size where applicable, extension (withinthe primary site or contiguous or metastatic), highest involved lymph node chain, number of regional lymph nodesfound positive (with certain exceptions), and number ofregional nodes examined (with certain exceptions). TheNational Cancer Institute2SEER Survival Monograph

Chapter 1Introductionfew places the SEER historic stage is used. The SEERSummary Staging Manual 2000 lists the definitions forSEER Summary Stage 2000 and in the footnotes for eachsite describes how the SEER Summary Stage 1977 andthe SEER historic stage differ from it (7).on the cancer site and the changes between the third andfifth editions of AJCC, some chapters present data according to the AJCC third, AJCC fifth, or a different stagedefinitions (see below). Except for lymphomas, the AJCCstaging criteria were applied to all histologies for eachprimary site. In some chapters, a SEER modified AJCCstage was used. The main difference between the SEERmodified and AJCC versions, is that NX was combinedwith N0 in the conversion of TNM to AJCC stage.SURVIVAL METHODSThe observed survival rate, obtained using the actuarial(life table) method, is the proportion of cancer patientssurviving for a specified time interval after diagnosis.The expected survival rate for a hypothetical cohort ofpersons of the same sex, age, and race as the patientcohort is the proportion, based on the 1990 life table, ofthe given cohort that will survive to the end of the giventime interval. For some sites, median survival times arepresented. The median survival time is based on the observed survival rate and is defined as the point at which50% have died and 50% are alive.SEER has also used a more simplistic stage with five levels:In situ tumors are those that have not yet broken throughthe adjacent basement membrane. For most cancer sitestreated in this monograph, in situ tumors are excludedfrom the analysis; the urinary bladder and the femalebreast are exceptions. The term localized describes tumors,regardless of size, that are confined to the organ of origin.Regional tumors are those that have metastasized to theregional lymph nodes or have extended directly from theorgan of origin. Distant describes a tumor whose metastases have traveled to other parts of the body. (Leukemiaand myeloma are considered distant at diagnosis.) Wheninformation is not sufficient to assign a stage, a cancer issaid to be Unstaged or Unknown. Most of the chapterswhich use stages of localized, regional, and distant arebased on the SEER Summary Stage (1977) (6). Basedon the same principles as Summary Stage 1977, SEERhas used more historical definitions that are more consistent over time for historical trends back to 1973. In aMost of the survival analyses in this monograph is basedon the relative survival rate (8), except in Chapter 31on race and ethnicity, where the cause-specific survivalrate (9) is used.Relative survival is a net survival measure representingcancer survival in the absence of other causes of death.Relative survival is defined as the ratio expressed as apercent, of the proportion of observed survivors in a co-Table 1.2: Ten Most Common Cancer Sites: 1-, 2-, 3-, 5-, 8- & 10-Year Relative Survival Rates by Site, Ages 20 , 12 SEER Areas,1988-2001Relative Survival Rate (%)SiteAll sites (except male andfemale breast in 00.0100.0Breast (female, 8911.583.375.169.963.659.257.7ProstateBreast (female, in ary n Lymphoma65,9324.274.266.362.156.349.947.0Uterine Corpus48,6423.193.589.587.084.783.182.6Leukemia (all ages)42,6782.767.058.053.447.240.738.1Kidney and Renal Pelvis32,5832.180.873.870.465.560.957.9National Cancer Institute3SEER Survival Monograph

Chapter 1Introductionrelative survival rates are presented for some sites conditioned on specific times after diagnosis. For some siteswhere survival is very poor, the eight year survival ratemay obscure that for the small group of patients who havealready survived 3 years, their probability of surviving thenext 5 years may be quite high.hort of cancer patients (the observed survival rate definedabove) to the proportion of expected survivors (the expectedsurvival rate defined above). Thus, a relative survival of100% means that a cancer patient cohort is just as likelyto survive the given interval as a cohort in the generalpopulation of the same sex, age, and race. It does not meanthat everyone will survive their cancer. For example, ina group of screening found cancers, many of the peopleseek medical care on a more routine basis than the generalpopulation and may have better non-cancer survival thanthe general population. In this case the expected life tableis too low which makes the relative rate too high. On theother hand, lung cancer patients who smoke may be atexcess risk of dying of other smoking related causes thanthe general population and the calculated expected ratewould be too high which means that the relative survivalrate may be lower than it would be if life tables based onsmoking could be used.For certain racial and ethnic groups, the life tables thatare typically used for calculating expected survival do notaccurately represent the experience of that specific racial/ethnic population. Since the calculation of relative survivalrates needs accurate life tables, the relative survival ratesare not shown for race/ethnic groups other than white orblack in the individual site chapters. In order to presentinformation for race/ethnic groups other than white patientsor black patients, a cause-specific (c-s) survival rate wasused. Since survival calculated under different methodscan not be compared to one another, the survival rates formore specific racial/ethnic groups were put in a specialchapter on race-and-ethnicity, Chapter 31. The c-s rate isdependent on knowledge not only of the date of death butalso accurate information on the cause of death. The c-srate is similar to the observed survival rate except that onlypatients who died of their cancer are considered as deathsand patients who died of other causes are ‘censored’ at thetime of death. This method avoids problems of finding appropriate expected survival rates which are needed for therelative survival rate, but is dependent on which cause ofdeaths are considered due to the cancer. The cause-specificrate, however, is dependent on accurate cause of death (COD)information. When the population used in calculating theexpected survival is similar to the population of cancerpatients except for the latter’s cancer experience, the relative survival rate and the cause-specific survival rate willWhile many times 5-year relative survival rates are presented, a five year rate may be less informative than asurvival rate over a shorter time frame for a site or groupwith poor survival or over a longer time frame for a site orcharacteristic with good survival. Up to 10-year survivalrates are shown for many sites.The conditional survival rate, while difficult to explain,may be the most clinically informative of the survival rates.Instead of evaluating survival from diagnosis, for examplea 5-year relative survival rate from diagnosis, the conditional survival rate can start anytime after diagnosis, i.e., itis conditioned on the cohort surviving to that point of timeand then a survival rate is calculated for the patients whohave survived to that point. For this monograph, 5-yearTable 1.3: Ten Most Common Cancer Sites: Five-Year Relative Survival Rates by Sex and Race, Ages 20 , 12 SEER maleMaleFemaleMaleFemaleAll sites (except male and female Breast insitu)ProstateBreast (female, in 0.052.998.4n/a93.5n/an/a100.0n/a100.0Breast (female, .354.9Melanoma90.088.292.188.492.470.176.3Urinary Bladder81.984.075.984.877.369.355.4Non-Hodgkin Lymphoma56.352.560.953.461.543.454.8Uterine Corpus84.7n/a84.7n/a86.4n/a61.8Leukemia (ages 0-19 and 20 )47.248.046.249.647.637.237.9Kidney and Renal Pelvis65.565.266.065.966.261.464.8National Cancer Institute4SEER Survival Monograph

Chapter 1IntroductionTable 1.4: Number of Cases by Leading Cancer Site and Stage at Diagnosis, 5-Year Relative Survival Rates, Ages 20 , 12 SEERAreas, 1988-2001Relative SurvivalSiteProstateBreast (female, 70Urinary n Lymphoma19,97169.49,09861.130,46844.36,395Uterine Corpus35,64695.77,23766.23,99326.01,766Kidney and Renal Pelvis17,59190.47,31660.26,5988.21,078@ Local combined with Regional for Prostatebe nearly equal. That is, the relative survival rate closelyindicates the probability that a patient will not die due tocancer-related causes within the given time interval. Whenthe population used for the expected survival is dissimilarto the population of cancer patients, the relative survivalmay differ from the cause-specific survival rate by tumorand patient characteristics. Comparisons of survival ratesshould be based on the same survival method for calculating rates.Blacks seem to fare worse with this disease, where the5-year survival rate is 69% among black males and 55%among black females.Survival by summary stage is presented in Table 1.4 forselect cancers. The differences in 5-year survival by stageare notable. The earlier the stage at diagnosis, the more favorable is the 5-year survival. For screenable cancer sites,survival ranges from 91% at localized stage to 9% at distantstage for colorectal cancer, and 97% at localized stage to24% at distant stage for female invasive breast cancer. Othercancer sites are as extreme in terms of survival by stage ofdiagnosis (urinary bladder, melanoma).RESULTSRelative survival up to 10 years after diagnosis of invasivecancer is shown in Table 1.2 for patients diagnosed in the12 SEER catchment areas during 1988-2001. Survival ratesvary substantially according to the cancer site. Amongthe most frequently diagnosed cancers, the sites with thehighest 10-year relative survival rates are prostate, femalebreast in situ, uterine corpus, and melanoma, which have10-year relative survival rates of 83% (uterine corpus) to100% (female breast in situ). Lung cancer has the leastfavorable survival across the 10-year period following diagnosis (11%).DISCUSSIONMany times in population-based statistics the emphasisis on incidence and mortality statistics. While these areimportant in measuring cancer, they are not as relevant tothe medical community concerned about prognosis. Thefocus of this monograph is to present descriptive analysesof cancer survival by patient and tumor characteristics.Since the emphasis is on the influence of patient and tumorcharacteristics on survival and not on how survival rateshave changed over time, a discussion of biases in survivaltrends

Cancer Survival Among Adults: U.S. SEER Program, 1988-2001 Patient and Tumor Characteristics Edited by: Lynn A. Gloeckler Ries John L. Young, Jr. Gretchen E. Keel Milton P. Eisner Yi Dan Lin Marie-Josephe D. Horner Suggested Citation: Ries LAG, Young JL, Keel GE, Eisner MP, Lin YD, Horner M-J (editors). SEER Survival