Effect of the staging schema on melanoma cancer reporting, 1999 to 2006

Effect of the staging schema on melanoma cancer reporting, 1999 to 2006 Sue-Min Lai, PhD, MS, MBA,a Jessica B. King, MPH,b Sarma Garimella, MBBS, MPH,a John Keighley, PhD,a and Mary Lewis, CTR, RHITb Kansas City, Kansas, and Atlanta, Georgia CME INSTRUCTIONS Please note this is one article that is part of a 16-article CME supplement. CME credit should only be claimed after reading the entire supplement which can be accessed via the ‘‘Melanoma Supplement’’ tab under the ‘‘Collections By Type’’ pulldown menu on http://www.jaad.org. This journal supplement is a CME activity (enduring material) co-sponsored by the American Academy of Dermatology and the Centers for Disease Control and Prevention and is made up of four phases: 1. Reading of the CME Information (delineated below) 2. Reading all the articles in this supplement 3. Achievement of a 70% or higher on the online Post Test 4. Completion of the CME Evaluation

CME INFORMATION AND DISCLOSURES Statement of Need: Healthcare providers continue to underreport melanoma even though cancer reporting requirements mandate such reporting. Additionally, providers may be unaware of recent trends and descriptive epidemiology regarding melanoma which includes the fact that nonwhites have a higher mortality rate from melanoma than do whites. Target Audience: Dermatologists, dermatopathologists, general physicians, and public health professionals. Accreditation The American Academy of Dermatology is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. AMA PRA Credit Designation The American Academy of Dermatology designates this enduring material for a maximum of 7 AMA PRA Category 1 CreditsÔ. Physicians should claim only the credit commensurate with the extent of their participation in the activity. AAD Recognized Credit This CME activity is recognized by the American Academy of Dermatology for 7 AAD Recognized Credits and may be used toward the American Academy of Dermatology’s Continuing Medical Education Award. Disclaimer: The American Academy of Dermatology is not responsible for statements made by the author(s). Statement or opinions expressed in this activity reflect the views of the author(s) and do not reflect the official policy of the American Academy of Dermatology. The information provided in this CME activity is for continuing education purposes only and is not meant to substitute for independent medical judgment of a health provider relative to the diagnostic, management and treatment options of a specific patient’s medical condition. Disclosures Editors The editors involved with this CME activity and all content validation/peer reviewers of the CME activity have reported no relevant financial relationships with commercial interest(s). Authors The authors of this CME activity have reported no relevant financial relationships with commercial interest(s). Planners The planners involved with this CME activity have reported no relevant financial relationships with commercial interest(s). The editorial and education staff involved with this CME activity have reported no relevant financial relationships with commercial interest(s). Resolutions of Conflicts of Interest In accordance with the ACCME Standards for Commercial Support of CME, the American Academy of Dermatology has implemented mechanisms, prior to the planning and implementation of this CME activity, to identify and mitigate conflits of interest for all individuals in a position to control the content of this CME activity. Learning Objectives After completing this learning activity, participants should be able to describe recent trends in the epidemiologic patterns of melanoma, including ethnic disparities in

melanoma mortality; identify when a private practice dermatologist is required to report melanoma cases to a cancer registry; locate and access central cancer reporting registries (http://apps.nccd.cdc.gov/cancercontacts/npcr/contacts.asp); and recognize and access national and state-based sources on surveillance systems for sun protection behaviors. Date of release: November 2011 Expiration date: November 2014 Ó 2011 by the American Academy of Dermatology, Inc. doi:10.1016/j.jaad.2011.04.034 Technical requirements: American Academy of Dermatology: d Supported browsers: FireFox (3 and higher), Google Chrome (5 and higher), Internet Explorer (7 and higher), Safari (5 and higher), Opera (10 and higher). d JavaScript needs to be enabled. Elsevier: Technical Requirements This website can be viewed on a PC or Mac. We recommend a minimum of: d PC: Windows NT, Windows 2000, Windows ME, or Windows XP d Mac: OS X d 128MB RAM d Processor speed of 500MHz or higher d 800x600 color monitor d Video or graphics card d Sound card and speakers Provider Contact Information: American Academy of Dermatology Phone: Toll-free: (866) 503-SKIN (7546); International: (847) 240-1280 Fax: (847) 240-1859 Mail: P.O. Box 4014; Schaumburg, IL 60168 Confidentiality Statement: American Academy of Dermatology: POLICY ON PRIVACY AND CONFIDENTIALITY Privacy Policy - The American Academy of Dermatology (the Academy) is committed to maintaining the privacy of the personal information of visitors to its sites. Our policies are designed to disclose the information collected and how it will be used. This policy applies solely to the information provided while visiting this website. The terms of the privacy policy do not govern personal information furnished through any means other than this website (such as by telephone or mail). E-mail Addresses and Other Personal Information - Personal information such as postal and e-mail address may be used internally for maintaining member records, marketing purposes, and alerting customers or members of additional services available. Phone numbers may also be used by the Academy when questions about products or services ordered arise. The Academy will not reveal any information about an individual user to third parties except to comply with applicable laws or valid legal processes. Cookies - A cookie is a small file stored on the site user’s computer or Web server and is used to aid Web navigation. Session cookies are temporary files created when a user signs in on the website or uses the personalized features (such as keeping track of items in the shopping cart). Session cookies are removed when a user logs off or when the browser is closed. Persistent cookies are permanent files and must be deleted manually. Tracking or other information collected from persistent cookies or any session cookie is used strictly for the user’s efficient navigation of the site. Links - This site may contain links to other sites. The Academy is not responsible for the privacy practices or the content of such websites. Children - This website is not designed or intended to attract children under the age of 13. The Academy does not collect personal information from anyone it knows is under the age of 13. Elsevier: http://www.elsevier.com/wps/find/privacypolicy.cws_home/privacypolicy

S95.e1

J AM ACAD DERMATOL

S95.e2 Lai et al

NOVEMBER 2011

Background: Staging schemas have changed multiple times over the past 10 years. Objective: We sought to examine the impact of staging schemas on the distribution of stages at diagnosis over time. Methods: We examined the stage at diagnosis for melanoma cancer cases diagnosed between 1999 and 2006 using data provided by the Surveillance, Epidemiology, and End Results (SEER) and National Program of Cancer Registries (NPCR) programs. The staging schemas were summary staging 1977 (SS1977), summary staging 2000 (SS2000), derived SS2000, and SEER historic staging systems. Results: Melanoma was predominantly staged as a localized disease in all schemas. Using SEER data, the proportion of localized melanomas diagnosed in 2001 to 2003 using SS2000 was about 2.5% lower than the proportion diagnosed in 1999 to 2000 using SS1977, whereas the proportion of cases staged as regional was 2.7% higher using the SS2000 than SS1977. The distribution of stages for cases diagnosed in 2001 to 2003 using SS2000 was similar to that for cases diagnosed in 2004 to 2006 using a derived SS2000. Shift in stage distribution among SS1977, SS2000, and SEER historic staging was found to be about 6% (localized to regional) and about 17.5% (unknown to regional stage). The distribution of changes in stage observed for the SEER cases was not evident for cases from NPCR. Limitations: SEER historic staging was not available for NPCR cases. Conclusion: Changes in staging rules resulted in cases being moved from the localized to the regional stage and from unknown to the regional stage. Without staging rules that have been consistently applied to melanomas over many years, surveillance of prevention, treatment, and control of this condition is difficult. ( J Am Acad Dermatol 2011;65:S95.e1-10.) Key words: derived summary staging 2000; malignant melanoma; stages at diagnosis; summary staging 1977; summary staging 2000; Surveillance, Epidemiology, and End Results historic staging.

Cancer staging describes the extent to which a cancer has spread from its point of origin. Cancer staging classification schemas are valuable in that they assist clinicians to understand the extent of the disease, recommend the best therapy, and provide prognostic information to patients and their families.1-3 From the public health perspective, knowing the stage of cancer aids public health practitioners and researchers in assessing the burden of cancer on society, characterizing prognosis including survivorship, and identifying necessary strategies and policies to reduce the cancer burden. State registries report stage distribution of various cancer sites for routine cancer surveillance. In addition, data on staging have been used to target subpopulations

for cancer screening and early detection, and to track and evaluate program effectiveness regarding prevention and early detection over time.4-6 Several staging classification systems are used in clinical practice, such as the TNM (tumor-nodesmetastasis) staging system of the American Joint Committee on Cancer.2,3 The National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) summary staging was intended as a measure for cancer surveillance with longitudinal stability for population-based cancer registries.7 Three staging schemas have been used in the last 2 decades: (1) summary staging 1977 (SS1977) for cases diagnosed in 1995 to 2000, (2) summary staging 2000 (SS2000) for cases diagnosed in 2001

From the Kansas Cancer Registry, University of Kansas Medical Centera; and Division of Cancer Prevention and Control, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta.b Publication of this supplement to the JAAD was supported by the Division of Cancer Prevention and Control, Centers for Disease Control and Prevention (CDC). Conflicts of interest: None declared.

The opinions or views expressed in this supplement are those of the authors and do not necessarily reflect the opinions, recommendations, or official position of the journal editors or the Centers for Disease Control and Prevention. Accepted for publication April 24, 2011. Reprint requests: Sue-Min Lai, PhD, MS, MBA, Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Mail Stop 1008, 3901 Rainbow Blvd, Kansas City, KS 66160. E-mail: [email protected]. 0190-9622/$36.00

J AM ACAD DERMATOL

Lai et al S95.e3

VOLUME 65, NUMBER 5

to 2003, and (3) Collaborative Stage Data Collection Oncology, Third Edition codes 8720 to 8790.13 The System (also called derived summary staging/de33 NPCR that met the US Cancer Statistics publication rived SS2000) for cases diagnosed in 2004 to the standards for the entire time period 1999 to 2006 are present.4,8,9 The most recent change from SS2000 to shown in Fig 1. The coverage from these NPCR derived SS2000 suggests that staging systems will accounted for 63.1% of the US population for those continue to change as oncologists and researchers years. The SEER-11 registries, which covered 13.7% learn more about cancer. of the US population, are also included in this article Although cancer staging to address the effects of remains a critical tool in meachanges in staging schema CAPSULE SUMMARY suring the success of cancer on melanoma cancers (Fig control efforts, there is a need 1). Therefore, the total US Knowledge of melanoma cancer stage to study and evaluate the impopulation coverage in this aids public health practitioners and pact of changes in staging study is 67.2% after removing researchers in assessing the burden on schemas on cancer reporting. population served by the society. In fact, two studies found SEER metropolitan cancer Staging facilitates the identification of inconsistencies in changes registries and was also innecessary strategies and/or policies to from SS1977 to SS2000 across cluded in the corresponding reduce the burden on society. cancer sites.10,11 For melaNPCR state cancer registries. noma, the major shift ocHowever, we could not asStaging rules have changed. curred when cases classified sess the extent of duplicate Understanding the impact of staging as localized stage using case counting because of the rule changes and the associated SS1977 changed to regional nature of deidentified data. limitations of staging data would help stage using SS2000.11 We categorized the stage with interpretation of the data and To reduce longitudinal inat diagnosis reported in the related implications for cancer control. consistencies by producing NPCR using SS1977 for cases uniform staging information diagnosed in 1999 to 2000, for all registered cases over time, the SEER program SS2000 for cases diagnosed in 2001 to 2003, and constructed a variable known as the SEER historic derived SS2000 for cases diagnosed in 2004 to 2006. summary stage (SHSS).12 However, collection of the All cancer staging systems have an underlying simSHSS was not implemented in the National Program ilarity. Each system is based on documentation of the of Cancer Registries (NPCR). Therefore, this article ‘‘extent of disease,’’ a detailed description of the addresses the following: first, using SEER data, we spread of the disease from the site of origin. In show the differences in the distribution of stages of general, summary staging is a single-digit system and melanoma by the 3 staging schemas and the SHSS. has 8 categories: in situ, localized only, regional by Second, using NPCR data, we show the distribution direct extension only, regional lymph node(s) inof stages of melanoma by the 3 staging schemas and volved only, both regional lymph node(s) and then estimate the stage of NPCR-reported cases using regional extension, regional not otherwise specified, the results observed in stage shift among the 3 distant site(s) and/or distant lymph node(s) inschemas and the SHSS. This projection attempts to volved, and unknown. Details of these staging sysdetermine whether the projected SHSS can assist tems can be found elsewhere.13,14 The derived SS2000 system uses a modified extent of disease NPCR in monitoring and evaluating public health format (more detailed than the summary staging efforts designed to increase screening, early detecsystem) to collect information about each case. tion, and control of melanoma cancer over time. Briefly, the data items specific to melanoma are extracted from the medical record and coded using METHODS the derived SS2000 system fields. When data collecWe used data submitted to the NPCR program as tion is complete, the data collector activates the of November 30, 2008, for registries that met the US computer algorithms to derive the values for SS2000, Cancer Statistics publication standards for data qualand the computer algorithms determine the classifiity for all sites combined.13 We also used data from cation stages in a consistent manner. the November 2008 submission to the SEER proThe SHSS was available for all cases reported to gram. To address stage changes in melanoma the SEER registries, although it was not calculated or reporting affected by the 3 staging schemas, we collected by NPCR. The SHSS was never directly used microscopically confirmed invasive melanoma coded by SEER registries; instead, it was derived cases diagnosed between 1999 and 2006 using from the extent of disease, which captured the International Classification of Diseases for d

d

d

J AM ACAD DERMATOL

S95.e4 Lai et al

NOVEMBER 2011

Abbreviations used: derived SS2000: NPCR: SEER: SHSS: SS1977: SS2000:

Collaborative Stage Data Collection System/derived summary staging 2000 National Program of Cancer Registries Surveillance, Epidemiology, and End Results Surveillance, Epidemiology, and End Results historic summary staging summary staging 1977 summary staging 2000

comprehensive information on tumor size, lymph node involvement, and metastases to other tissues.9,15,16 However, the concepts of this schema were used consistently over time so that SEER longterm trends could be assessed.4 Although staging in SHSS is based on the same concepts as SS1977, SS2000, and derived SS2000, the specific definitions of various stages within each differ. All versions of extent-of-disease characterization allow derivation of historic stage categories. In spite of the differences, all schemas categorize cases as in situ, localized, regional (regional by direct extension only, regional lymph node(s) involved only, regional by both direct extension and lymph node involvement, regional not otherwise specified), distant, and unknown. But there are differences among the 3 schemas (Table I). For example, Clark level V (ie, depth of lesion [4.0 mm with or without ulceration), which was considered a localized stage in the SS1977 schema, was categorized as a regional stage in both SS2000 and SHSS. Simple descriptive statistics were used to summarize the distribution of stage at diagnosis by year and staging schema. We used the kappa statistic to derive the level of agreement among the SHSS, the SS1977, the SS2000, and the derived SS2000. We considered a kappa statistic greater than 0.75 to be excellent agreement, a kappa statistic between 0.40 and 0.75 to be fair to good agreement, and a kappa statistic below 0.40 to be very low or no agreement. NPCR used the SS1977, the SS2000, and the derived SS2000 to classify stage of cancers diagnosed in 1999 to 2006. Because NPCR did not stage these cases using the SHSS, the distribution of stages according to SHSS had to be estimated. We used an equation that captures the stage shift of cases to obtain the projected stages for NPCR cases according to the SHSS rules (Table I). The equations shown are for cases diagnosed in 1999 and 2000 using SS1977.

NSHSS; localized ¼ NSS1977; localized  NSS1977; localized  plocalized to regional

ð1Þ

NSHSS; regional ¼ NSS1977; regional 1NSS1977; localized  plocalized to regional 1NSS1977; unk  punk to regional NSHSS; distant ¼ NSS1977; distant

ð2Þ ð3Þ

NSHSS; unk ¼ NSS1977; unk  NSS1977; unk  punk to regional

ð4Þ

Where, NSHSS,i = estimated number of NPCR cases in stage i when SHSS is used, where i = localized, regional, distant, and unknown (unk) NSS1977,i = the number of NPCR cases in stage i when SS1977 is used, where i = localized, regional, distant, and unknown (unk) pij = the proportion of SEER cases that showed a shift from stage i (using SS1977) to stage j (using SHSS). The same formulas can be applied to cases diagnosed from 2001 to 2003 using SS2000.

RESULTS We included 55,665 cases of invasive melanoma cancer of the skin from the SEER registries in the analysis. The proportions of cases categorized as localized, regional, distant, and unknown by diagnosis years are shown in Table II. Cases over the years of diagnosis were collapsed within each staging system because their distribution by year appeared similar. The proportions of cases diagnosed in 1999 to 2000 categorized to each corresponding stage according to SS1977 schema were 87.8% ([5442 1 5652]/[6221 1 6413]), 5.3%, 3.2%, and 3.7%, whereas the proportions of cases diagnosed in 2001 to 2003 categorized to each corresponding stage using SS2000 were 85.3%, 8.0%, 3.4%, and 3.3%. The distribution of stages at diagnosis for cases categorized in 2001 to 2003 using SS2000 was similar to that for cases diagnosed in 2004 to 2006 using derived SS2000. As we looked at the shift in stages among the 3 systems (SS1977, SS2000, and derived SS2000) and the SHSS, we noted that about 6% of localized melanomas using the SS1977 and SS2000 moved to the regional stage using the SHSS. About 17.5% of the unknown cases using the SS1977 and SS2000

J AM ACAD DERMATOL

Lai et al S95.e5

VOLUME 65, NUMBER 5

Fig 1. Thirty-three NPCR and 11 SEER cancer registries that were included in the analysis. NPCR, National Program of Cancer Registries; SEER, Surveillance, Epidemiology, and End Results.

moved to the regional stage using the SHSS. These changes resulted in an increase in regional stage cases. Finally, the distribution of stages for cases diagnosed in 2004 to 2006 was identical in the SHSS and the derived SS2000. The level of agreement characterized by the kappa statistic was 0.77 between the SS1977 and SHSS, 0.82 between the SS2000 and SHSS, and 1.0 between the derived SS2000 and SHSS. NPCR reported a total of 272,074 melanoma cancers in 1999 to 2006, with the number of reported cases increasing each year (Table III). Despite the increasing number of cases reported, the distribution of stages at diagnosis remained quite stable. The distribution of localized, regional, distant, and unknown stages at diagnosis within each staging system showed 78.3%, 7.0%, 3.8%, and 10.9% for 1999 to 2000 diagnosed cases and 78.6%, 8.1%, 3.7%, and 9.6% for 2001 to 2003 diagnosed cases. The proportions of localized and distant stages were almost identical in the two time periods using the SS1977 and the SS2000. The proportion of regional stage cases went up from 7.0% to 8.1% whereas those in the unknown stage went down from 10.9% to 9.6% between 1999 to 2000 and 2001 to 2003 (or SS1977 to SS2000). The proportion of regional staged cancer cases diagnosed in 2004 to

2006 was 9% and the proportion of unknown cases was 8.4%. The projected SHSS stage at diagnosis, based on shifting the stages of cases reported to NPCR, is shown in Table III. For example, there were 22,380 localized melanomas diagnosed in 1999 in the combined NPCR. For 1999 SEER diagnoses, a proportion of 0.064 of localized stage cases (348/5442) (Table II) was shifted from being localized using the SS1977 system to regional using the SHSS system. Using the SHSS proportion as the standard, the projected number of localized cases for NPCR was 20,949 (ie, 22,380 e [22,380 * 0.064]). Similarly, the projected number of regional cases became 3956 (1964 1 1432 [22,380 * 0.064; moved from localized to regional] 1 560 [3126 * 0.179; moved from unknown to regional]). Overall, the projected proportion of localized stage melanomas ranged from 73.3% in 1999 to 78.8% in 2006 whereas the proportion of regional stage melanomas ranged from 13.8% in 1999 to 9.2% in 2006.

DISCUSSION We observed an increase of 2.7% (from 5.3%8.0%) in regional stage cases from 1999 through 2000 (SS1977) to 2001 through 2003 (SS2000) for the SEER registries, and an increase of only 1.1%

J AM ACAD DERMATOL

S95.e6 Lai et al

NOVEMBER 2011

Table I. Known staging criteria differences among summary staging 1977, summary staging 2000, and Surveillance, Epidemiology, and End Results historic summary stage schemas Stage

1 - Localized only

SS1977

Subcutaneous tissue invaded (through entire dermis) Clark level* II-V including NOS with or without ulcerationy

Regional, including Regional by direct extension only d Regional lymph node(s) involved only d Regional by BOTH direct extension and regional lymph node(s) involved d Regional, NOS d

7 - Distant site(s)/lymph node(s) involved

9 - Unknown if extension or metastasis

Lip: Facial, NOS Lip: Facial, NOS: Buccinator (buccal) Lip: Facial, NOS: Nasolabial Lip: Mandibular, NOS: Submental Lip: Parotid, NOS Lip: Parotid, NOS: Infra-auricular Lip: Parotid, NOS: Preauricular Eyelid/canthus: Mandibular, NOS: Submental Face, Other (cheek, chin, forehead, jaw, nose, and temple): Mandibular, NOS: Submental Neck: Mandibular, NOS: Submental Extension unknown

SS2000

SHSS

Clark level II-IV including NOS with or without ulceration

Clark level II-IV including NOS without ulceration

Subcutaneous tissue invaded (through entire dermis) Clark level V with or without ulceration Lip: Facial, NOS Lip: Facial, NOS: Buccinator (buccal) Lip: Facial, NOS: Nasolabial Lip: Mandibular, NOS: Submental Lip: Parotid, NOS Lip: Parotid, NOS: Infraauricular Lip: Parotid, NOS: Preauricular Eyelid/canthus: Mandibular, NOS: Submental Face, Other (cheek, chin, forehead, jaw, nose, and temple): Mandibular, NOS: Submental Neck: Mandibular, NOS: Submental

Subcutaneous tissue invaded (through entire dermis) Clark level V with or without ulceration Clark level II-IV including NOS with ulceration Extension unknown and known lymph node involvement

Extension unknown

Lip: Facial, NOS Lip: Facial, NOS: Buccinator (buccal) Lip: Facial, NOS: Nasolabial Lip: Mandibular, NOS: Submental Lip: Parotid, NOS Lip: Parotid, NOS: Infra-auricular Lip: Parotid, NOS: Preauricular Eyelid/canthus: Mandibular, NOS: Submental Face, Other (cheek, chin, forehead, jaw, nose, and temple): Mandibular, NOS: Submental Neck: Mandibular, NOS: Submental Extension unknown and unknown lymph node involvement

Boldface and italics are used to demonstrate the shifting between the staging schemas. NOS, Not otherwise specified; SS1977, summary stage 1977; SS2000, summary stage 2000; SHSS, Surveillance, Epidemiology, and End Results historic summary stage. *Clark level: measure of thickness of melanoma in subcutaneous skin layers. y Skin ulceration does not alter classification in SS1977 and SS2000.

J AM ACAD DERMATOL

Lai et al S95.e7

VOLUME 65, NUMBER 5

Table II. Stages of melanoma cancer cases: Surveillance, Epidemiology, and End Results registries, 1999 to 2006 SHSS Diagnosis year/staging schema

1999/SS1977 (N = 6221) Localized Regional Distant Unknown 2000/SS1977 (N = 6413) Localized Regional Distant Unknown 2001/SS2000 (N = 6800) Localized Regional Distant Unknown 2002/SS2000 (N = 6692) Localized Regional Distant Unknown 2003/SS2000 (N = 6807) Localized Regional Distant Unknown 2004/Derived SS2000 (N = 7209) Localized Regional Distant Unknown 2005/Derived SS2000 (N = 7810) Localized Regional Distant Unknown 2006/Derived SS2000 (N = 7713) Localized Regional Distant Unknown

n

% Yearly

5442 318 210 251

87.5 5.1 3.4 4.0

5652 353 194 214

88.1 5.5 3.0 3.4

5838 523 236 203

85.8 7.7 3.5 3.0

5700 544 239 209

85.2 8.1 3.6 3.1

5770 551 226 260

84.8 8.1 3.3 3.8

6152 621 260 176

85.3 8.6 3.6 2.5

6716 669 289 136

86.0 8.6 3.7 1.7

6580 681 286 166

85.3 8.8 3.7 2.2

Localized

Regional

Distant

Unknown

5094 5094 0 0 0 5294 5294 0 0 0 5525 5525 0 0 0 5381 5381 0 0 0 5400 5400 0 0 0 6152 6152 0 0 0 6716 6716 0 0 0 6580 6580 0 0 0

711 348 318 0 45 747 358 353 0 36 876 313 523 0 40 900 319 544 0 37 962 370 551 0 41 621 0 621 0 0 669 0 669 0 0 681 0 681 0 0

210 0 0 210 0 194 0 0 194 0 236 0 0 236 0 239 0 0 239 0 226 0 0 226 0 260 0 0 260 0 289 0 0 289 0 286 0 0 286 0

206 0 0 0 206 178 0 0 0 178 163 0 0 0 163 172 0 0 0 172 219 0 0 0 219 176 0 0 0 176 136 0 0 0 136 166 0 0 0 166

Data are from population-based cancer registries that participate in SEER program that meet high-quality data criteria. These registries cover 13.7% of US population for 1999-2006. SS1977, Summary stage 1977; SS2000, Summary stage 2000; SHSS, Surveillance, Epidemiology, and End Results (SEER) historic summary stage.

(7.0%-8.1%) in regional stage cases during the same period for the NPCR. This observed pattern of melanoma cases in SEER registries that were found shifting from one stage to another were mostly a result of changes in staging rules. Clark level V with or without ulceration that was considered a localized stage in SS1977 has been moved to a regional stage in both SS2000 and SHSS. On the other hand, for sites lips, neck, face, and eyelids with certain

histologies the stage moved from distant to regional (Table I). For the NPCR cases, we have seen an increase in counts of localized cases, but we did not observe the expected decrease of about 2.5% in localized cases as was seen for the SEER registries. In fact, there was practically no change in proportion of localized stage for the NPCR from the period 1999 through 2000 to the period 2001 through 2003 (78.3% vs 78.6%). This inconsistency in stage

J AM ACAD DERMATOL

S95.e8 Lai et al

NOVEMBER 2011

Table III. Reported and projected stages of melanoma cancer cases: National Program of Cancer Registries, 1999 to 2006 NPCR-reported stage Year of diagnosis

1999/SS1977 (N = 28,575) Localized Regional Distant Unknown 2000/SS1977 (N = 30,466) Localized Regional Distant Unknown 2001/SS2000 (N = 32,718) Localized Regional Distant Unknown 2002/SS2000 (N = 34,149) Localized Regional Distant Unknown 2003/SS2000 (N = 34,255) Localized Regional Distant Unknown 2004/Derived SS2000 (N = 36,093) Localized Regional Distant Unknown 2005/Derived SS2000 (N = 38,758) Localized Regional Distant Unknown 2006/Derived SS2000 (N = 37,090) Localized Regional Distant Unknown

n

%

Projected SHSS staging* Count

%

22,380 1964 1105 3126

78.3 20,949 73.3 6.9 3956 13.8 3.9 1105 3.9 10.9 2566 9.0

23,879 2155 1114 3318

78.4 22,366 73.4 7.1 4226 13.9 3.6 1114 3.7 10.9 2760 9.0

25,701 2570 1151 3296

78.5 24,323 74.3 7.9 4597 14.1 3.5 1151 3.5 10.1 2647 8.1

26,855 2783 1208 3303

78.6 25,352 74.2 8.2 4871 14.3 3.5 1208 3.5 9.7 2718 8.0

26,904 2855 1338 3128

78.6 25,179 73.6 8.3 5073 14.8 3.9 1338 3.9 9.2 2635 7.7

28,118 3212 1457 3306

77.9 28,118 77.9 8.9 3212 8.9 4.0 1457 4.0 9.2 3306 9.2

30,779 3461 1405 3113

79.4 30,779 79.4 8.9 3461 9.0 3.6 1405 3.6 8.1 3113 8.0

29,234 3413 1423 3020

78.8 29,234 78.8 9.2 3413 9.2 3.8 1423 3.8 8.2 3020 8.2

Data are from population-based cancer registries that participate in NPCR that meet high-quality data criteria. These registries cover 63.1% of population for 1999 to 2006. NPCR, National Program of Cancer Registries; SHSS, Surveillance, Epidemiology, and End Results historic summary stage; SS1977, summary stage 1977; SS2000, summary stage 2000. *Calculation of projected SHSS stage can be referred to equations 1 to 4 in ‘‘Methods’’ section.

pattern shift between the SEER and NPCR is discussed below. The steadily increasing case counts since 1999 give evidence of incomplete reporting of melanoma cancers by some NPCR in the early years of the registries. In fact, underreporting has long been suspected for melanomas.17-20 If underreporting from physicians’ offices is believed to have occurred, particularly in the early years of the registries, we would expect to see a higher proportion of localized disease within each staging schema than is currently shown in Table III. One possible reason for not seeing an increase in the proportion of localized stage could be misclassification because of lack of some staging elements. For example, some of the high percentage of NPCR cases currently classified as unknown stage (;10%) could be a result of incomplete information obtained solely from pathology reports. These pathology-report-only cases have not yet been followed up with physicians’ offices for definitive determination of stages.18,21 This could explain why we did not see a decrease in the proportion of localized NPCR cases between 1999 to 2000 and 2001 to 2003. Because underreporting may have masked the changes in staging distribution, the projected SHSS staging distribution changes, particularly before 2004, may not be as meaningful for the NPCR as they were for the SEER registries. Approximately 10% currently staged as unknown reported to NPCR between 1999 to 2000 and 2001 to 2003 could potentially be reclassified when additional sources of information are obtained. This occurred for the SEER registries when a number of unknown cases were moved from SS1977 and SS2000 into the regional category under the SHSS system. We examined the distribution of stages at diagnosis for cases diagnosed in 2004 to 2006 when most NPCR had better case ascertainment of melanoma cancers. For those cases, the distribution of localized, regional, distant, and unknown stages at diagnosis was 78.8%, 9.0%, 3.8%, and 8.4% for NPCR-reported cases whereas the distribution for the SEER registries was 85.5%, 8.7%, 3.7%, and 2.1%. If about 6% of NPCR unknown cases for 2004 to 2006 were recategorized as localized stage after additional information was obtained, then the distribution of stages at diagnosis would be about the same for cases diagnosed in 2004 to 2006 in both surveillance programs. This would allow the stages projected at diagnosis to be useful for NPCR cases in guiding public health practitioners as they monitor melanoma prevention, screening, and control efforts over time (Table III). On the other hand, if the approximately 10% unknown stage cases that were reported to NPCR

J AM ACAD DERMATOL VOLUME 65, NUMBER 5

between 1999 to 2000 and 2001 to 2003 were not recategorized as localized after follow-up with health care providers, then the projected stages for NPCR-reported cases, even for those diagnosed in 2004 to 2006, may not be as useful for the purpose of trend analysis over time. In that case, further evaluation would be needed to determine why there are differences in stage at diagnosis between the two programs. Limitations This article has some limitations. First, the SEER historic stage was not available for the NPCR cases, making it impossible to examine differences in the distribution of stages of melanoma in the populations served by the two programs. Second, restaging NPCR cases using the SEER historic schema is not readily feasible because the staging elements were not collected previously. Third, our study showed that SEER cases diagnosed from 2004 to 2006 have the same staging distribution when using both the derived SS2000 and the SEER historic staging systems. Unfortunately, we were not able to identify a source of documentation that could potentially explain such a perfect agreement in stages between the two schemas.

CONCLUSION Melanoma is a potentially preventable cancer that has been shown to benefit from screening and early detection.22,23 Public health efforts rely heavily on surveillance data to monitor the burden of cancer across subpopulations and to observe trends over time. These data have also been used to evaluate the effectiveness of various programs that were implemented to reduce this burden. It is important that public health practitioners and other users of these data understand the impact of staging rule changes and the associated limitations of staging data so that inaccurate interpretation will not misinform melanoma control efforts and affect our ability to respond appropriately. It is recommended that public health practitioners and researchers follow up with individual state registries to understand their level of completeness in ascertaining melanoma cases. If individual registries consider their level of ascertainment to be complete even in the years from 1999 to 2004, then the proposed formula for stage conversion from SS1977 and SS2000 to SHSS schema will be meaningful for planning and evaluation of melanoma cancer prevention and control efforts. Otherwise monitoring stages of melanoma cancer over time is not likely feasible and would not be recommended.

Lai et al S95.e9

REFERENCES 1. Kim CJ, Reintgen DS, Balch CM. The new melanoma staging system. Cancer Control 2002;9:9-15. 2. Balch CM, Buzaid AC, Soong SJ, Atkins MB, Cascinelli N, Coit DG, et al. Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol 2001;19:3635-48. 3. Balch CM, Buzaid AC, Atkins MB, Cascinelli N, Coit DG, Fleming ID, et al. A new American Joint Committee on Cancer staging system for cutaneous melanoma. Cancer 2000;88:1484-91. 4. Young JL Jr, Roffers SD, Ries LAG, Fritz AG, Hurlbut AA, editors. SEER summary staging manuale2000: codes and coding instructions. NIH publication No. 01-4969. Bethesda (MD): National Cancer Institute; 2001. 5. Nasseri K. Secular trends in the incidence of female breast cancer in the United States, 1973-1998 [Erratum in: Breast J 2004;10:380]. Breast J 2004;10:129-35. 6. Hu S, Soza-Vento RM, Parker DF, Kirsner RS. Comparison of stage at diagnosis of melanoma among Hispanic, black, and white patients in Miami-Dade County, Florida. Arch Dermatol 2006;142:704-8. 7. NAACCR SEER Summary Stage 2000 Implementation Work Group. Guidelines for implementation of SEER summary stage 2000. Springfield (IL): North American Association of Central Cancer Registries; 2000. 8. Collaborative Stage Work Group of the American Joint Committee on Cancer. Collaborative Stage Data Collection System: site-specific schema and coding instructions. Available from: URL:http://web.facs.org/cstage/schemalist.htm. Accessed March 28, 2010. 9. Shambaugh EM, Weiss MA, editors. Summary staging guide: cancer Surveillance, Epidemiology, and End Results reporting. NIH publication No. 98-2313. Bethesda (MD): National Institutes of Health; 1977. 10. Phillips JL, editor. Summary stage: data effects of the changes in 2000. Springfield (IL): North American Association of Central Cancer Registries; 2003. Available from: URL:http://www.naaccr. org. Accessed March 28, 2010. 11. Howe HL, Jamison M, Havener L, Chen VW, Reis L. Site-specific comparison of summary stage 1977 and summary stage 2000 coding, 2005. Available from: URL:http://www.naaccr.org. Accessed March 28, 2010. 12. End Results Group. End Results Group 1967 code manual. Bethesda (MD): US Department of Health, Education, and Welfare, National Institutes of Health; 1968. 13. Fritz A, Percy C, Jack A, Shanmugaratnam K, Sobin L, Parkin D, Whelan S, editors. International classification of diseases for oncology. 3rd ed. Geneva: World Health Organization; 2000. 14. Johnson CH, Adamo M, editors. SEER program coding and staging manual 2007 (2008 revision). NIH publication No. 07-5581. Bethesda (MD): National Cancer Institute; 2008. 15. Johnson CH, editor. SEER program coding and staging manual 2004, revision 1. NIH publication No. 04-5581. Bethesda (MD): National Cancer Institute; 2004. 16. Fritz A, Ries L, editors. SEER extent of disease 1988: codes and coding instructions. 3rd ed. NIH publication No. 98-1999. Bethesda (MD): National Institutes of Health; 1998. 17. Cockburn M, Swetter SM, Peng D, Keegan TH, Deapen D, Clarke CA. Melanoma underreporting: why does it happen, how big is the problem, and how do we fix it? J Am Acad Dermatol 2008;59:1081-5. 18. Lai SM, Garimella S, Keighley J. Pathology and dermatology reports: key in completeness of melanoma cancer reporting in population-based cancer registries. J Registry Manag 2004;31: 87-91.

S95.e10 Lai et al

19. Hall HI, Jamison P, Fulton J, Clutter G, Roffers S, Parrish P. Reporting cutaneous melanoma to cancer registries in the United States. J Am Acad Dermatol 2003;49:624-30. 20. Karagas MR, Thomas DB, Roth GJ, Johnson LK, Weiss NS. The effects of changes in health care delivery on the reported incidence of cutaneous melanoma in western Washington State. Am J Epidemiol 1991;133:58-62. 21. Merlino LA, Sullivan KJ, Whitaker DC, Lynch CF. The independent pathology laboratory as a reporting source for cutaneous

J AM ACAD DERMATOL

NOVEMBER 2011

melanoma incidence in Iowa, 1977-1994. J Am Acad Dermatol 1997;37:578-85. 22. Guide to community preventive services: preventing skin cancer: education and policy. Available from: URL:http:// www.thecommunityguide.org/cancer/skin/education-policy/ index.html. Accessed November 24, 2010. 23. US Preventive Services Task Force. Clinical guidelines: screening for skin cancer: US Preventive Services Task Force recommendation statement. Ann Intern Med 2009;150:188-93.