Abstract
Objective
We examined the completeness of TNM staging of small-cell (SCLC) and nonsmall- cell (NSCLC) lung cancer in the national Danish Cancer Registry (DCR) and whether staging varied by year of diagnosis, gender, age, degree of comorbidity, or presence of histopathological diagnosis.
Methods
We identified all patients with SCLCs and NSCLCs registered in the DCR during 2004–2009 and examined the completeness of their TNM registrations. Completeness was defined as the number of recorded individuals with TNM divided by the total number of patients. Completeness was calculated for TNM, T, N, and M individually, overall, and by year of diagnosis, gender, age at diagnosis, and comorbidity. Data regarding comorbidity was obtained from the Danish National Patient Register (DNPR). We performed separate analyses for patients with a histopathologically verified diagnosis of NSCLC. Finally, we designed an algorithm to categorize tumors with missing TNM components as limited, extensive, or distant disease.
Results
Overall TNM staging completeness was 77.5% (95% confidence interval (CI): 76.1%–78.8%) for SCLC and 77.9% (95% CI: 77.3%–78.4%) for NSCLC. Completeness did not vary by gender and increased during the study period. The proportion of staged patients was lower among patients above 80 years of age or with medium to high levels of comorbidity.
Conclusion
Overall TNM completeness for SCLC and NSCLC in the Danish Cancer Registry is high, but decreases with increasing levels of comorbidity and at ages greater than 80 years. Researchers should be aware of these potential sources of bias.
Introduction
Lung cancer is one of the most common malignancies and is the leading cause of cancer-related death in Denmark, as well as worldwide.Citation1–Citation3 In Denmark, incidence rates of lung cancer in 2009 were 83 (men) and 64 (women) per 100,000 person-years. Lung cancer is classified into two subtypes, including small cell (SCLC, 15%–18% of incident cases) and non-small cell lung cancer (NSCLC, 82%–85% of incident cases) based on histopathology and differences in prognosis and treatment. Stages of SCLC (limited and extensive) and NSCLC (I–IV) are assigned according to TNM classification, which addresses tumor size and growth into neighboring organs (T), lymph node involvement (N), and distant metastases (M). Prognosis and treatment choices vary according to cancer stage, but prognosis is also affected by age and level of comorbidity.Citation4–Citation8 The five-year survival rate for SCLC patients is approximately 20% for limited disease and 0% for extensive disease.Citation9 Corresponding rates for NSCLC patients are 50% and approximately 5% for stage IA and IV, respectively.Citation10,Citation11 Only two studies have been conducted regarding the registration of TNM staging for lung cancer in population-based registries.Citation12,Citation13 The proportion of patients with incomplete lung cancer stage in these registries increased with older age, as well as those that lived alone and were of black ethnicity.
The Danish Cancer Registry (DCR)Citation14,Citation15 is a nationwide registry that has recorded cancer incidence in the Danish population since 1943. Cancer diagnoses have been recorded according to the International Classification of Disease, 10th revision (ICD-10), since 2004. Stage at diagnosis has been recorded according to the 6th edition of TNM classificationCitation16 since 2004 and the 7th edition since 2009.
Completeness of TNM staging in the DCR is unknown. We therefore studied TNM completeness of SCLC and NSCLC, as well as conducted stratified analyses by year of diagnosis, gender, age, and level of comorbidity. Additionally, we suggest a method for defining stage categories in the presence of specific missing T, N, or M components.
Material and methods
We performed this study in Denmark, which has a population of approximately 5.4 million. All residents are provided with tax-supported medical care which is free of charge. Since 1968, the Danish Civil Registration System has assigned a unique 10-digit personal identification number to all Danish residents.Citation17 This number is used in all Danish registers and allows unambiguous individual-level data linkage.
Identification and categorization of lung cancer patients
We identified all patients with a first-time lung cancer diagnosis (ICD-10 code C33-C34) recorded between January 1, 2004 and December 31, 2009 in the DCR. Patients were categorized according to histopathological (morphological) diagnosis in the DCR, which is derived from the Danish Pathology Register and recorded according to the International Classification of Diseases for Oncology, 3rd version (ICD-0-3) and combined with a topographic diagnosis code.Citation14,Citation18 We defined SCLC as cases with morphology codes 80413–80459, while NSCLC was defined as any other morphology code including those with no available histopathological diagnosis.
We designed an algorithm allowing assignment of stage categories, despite specific missing T, N, or M values. We categorized SCLC tumors as limited (if tumor was M0), extensive disease (if tumor was M1), or unknown (if tumor was Mx), regardless of the values, known or unknown, of other components (T and N) (Appendix 1). We categorized NSCLC tumors into limited, extensive, distant disease, and unknown (Appendix 1). If TNM stage included component values of T4, N3, or M1, tumor stage was categorized as IIIb or IV, and these tumors were categorized as distant. This was done regardless of the stage, known or unknown, of other components. All other cancers with missing components were categorized as unknown.
Comorbidity
We obtained data regarding comorbidity from the Danish National Patient Register (DNPR).Citation19 This register contains data regarding all admissions to non-psychiatric hospitals in Denmark since 1978 and outpatient settings since 1995. The data includes dates of admission/contact, discharge, and diagnosis codes according to the ICD-10. The presence of comorbidity was defined according to the Charlson Comorbidity IndexCitation20,Citation21 (CCI), and categorized as low (CCI = 0), medium (CCI = 1–2), or high (CCI > 2) level of comorbidity. The CCI was based on all hospital diagnoses recorded in the DNPR within 10 years preceding the date of cancer diagnosis.
Statistical analysis
We computed completeness and corresponding 95% confidence intervals (CI) for the TNM registration overall and for each component (ie, T, N, and M). Completeness was defined as the number of individuals with TNM recordings divided by the total number of patients. We stratified completeness according to year of cancer diagnosis, gender, age at diagnosis (0–39 years, 40–59 years, 60–79 years, or ≥80 years), and level of comorbidity. We computed completeness separately for SCLC and NSCLC and, for NSCLC, for histologically verified disease. SCLC diagnosis had been verified histologically. We repeated the analyses using the algorithm for stage assignment in the presence of specific missing T, N, or M values. Analyses were performed using SAS (v 9.2; SAS Institute, Inc, Cary, NC, USA).
Results
We identified 24,734 patients (50% males) that had lung cancer in Denmark between 2004 and 2009. Of these, 3,658 (15%) had SCLC () and 21,076 (85%) had NSCLC (). Among those with NSCLC, 17,972 (85%) cases had been histologically verified. Median age at diagnosis was 69 years for all patients combined. The proportion of patients with preexisting hospital recorded comorbidity was similar between the two patient categories (47% among patients with SCLC, 50% among patients with NSCLC).
Overall completeness of the TNM staging was 77.5% (95% confidence interval (CI): [76.1%–78.8%]) for SCLC and 77.9% (95% CI: [77.3%–78.4%]) for NSCLC. Overall completeness for NSCLC with histological verification was 83.3% (95% CI: [82.8%–83.9%]) (data not shown). During the study period, completeness increased from 75.9% (95% CI: [72.2%–79.3%]) to 83.4% (95% CI: [80.3%–86.2%]) for patients with SCLC and from 77.2% (95% CI: [75.8%–78.6%]) to 82.4% (95% CI: [81.2%–83.6%]) for patients with NSCLC ( and ).
Completeness did not vary by gender. However, among patients with SCLC, completeness declined from 78.5% (95% CI: [76.9%–80.1%]) among those aged 60–79 years to 67.1% (95% CI: [62.1%–71.9%]) among those aged 80 years or more. For patients with NSCLC, completeness declined from 80.8% (95% CI: [80.2%–81.5%]) among those aged 60–79 years to 59.9% (95% CI: [58.2%–61.5%]) among those aged 80 years or more. Completeness did not vary significantly between age groups 0–39, 40–59 or 60–79 years for SCLC or NSCLC.
Completeness also decreased with comorbidity level. The proportions of complete TNM staging were 81%, 76%, and 69% for low, medium, and high levels of comorbidity among patients with SCLC. The corresponding proportions were 81%, 76%, and 70% for patients with NSCLC.
Finally, we evaluated stage completeness for SCLC and NSCLC, allowing for the inclusion of missing TNM components as described above. When using this algorithm, overall stage completeness was 93.1% for SCLC and 91.1% for NSCLC. However, variations in completeness according to gender, age, year of diagnosis, and comorbidity showed a pattern similar to that of the original data ( and ).
Discussion
We found that the completeness of TNM staging for SCLC and NSCLC in the DCR was lower for elderly patients and/or those with severe comorbidity. Completeness increased during the study period (2004–2009) and did not vary by lung cancer subtype or gender. Among patients with NSCLC, completeness was higher for patients with a histologically verified diagnosis than for those without such diagnosis.
Some SCLC as NSCLC cases without morphological information may have been misclassified. However, this potential misclassification likely had minor impact on the results, since we observed nearly the same completeness for SCLC as for NSCLC.
Completeness of lung cancer staging in the National Cancer Institute’s (US) Surveillance, Epidemiology, and End Results (SEER) ProgramCitation12 and completeness of the Swedish Cancer RegisterCitation13 are higher than that of the DCR. Institutions participating in the SEER are financially rewarded, and SEER promotes education and conducts regular audits for cancer registration,Citation22 which may explain this higher completeness. However, neither report presented overall estimates nor information regarding which components of TNM were missing or the impact of cancer type (SCLC or NSCLC), diagnosis year, or comorbidity on the likelihood of proper TNM staging.
Our findings are consistent with those of previous studies examining unstaged malignant disease which showed that fragile, elderly and/or those requiring nursing care are more likely to have unstaged disease.Citation13,Citation23,Citation24 This is the first study to present TNM staging according to comorbidity levels.
In conclusion, overall completeness of the TNM registration of SCLC and NSCLC in the DCR is high, and the amount of incomplete data is likely to have a minor impact on studies examining prognosis or surveillance that rely on DCR data, particularly when evaluating potential sources of bias such as age and level of comorbidity. We further showed that overall stage completeness can be increased by including patients for which the TNM components provide unambiguous prognostic information, despite missing or misclassified information of individual staging components. Completeness of both SCLC and NSCLC registration decreased with increasing age and comorbidity level, but was not affected by gender or lung cancer subtype. Future studies examining treatment, prognosis, or screening policies that rely on TNM data from the DCR should be conducted to address these potential sources of bias.
Acknowledgments
This study received financial support from the Regional Clinical Epidemiological Monitoring Initiative for Central and North Denmark Regions.
Disclosure
The authors declare no conflicts of interest in this work.
References
- Danish Board of HealthThe Danish Cancer Registry 2010CopenhagenDanish Board of Health2011 http://www.sst.dk/publ/Publ2011/DAF/Cancer/Cancerregisteret2010.pdf
- FerlayJShinHRBrayFFormanDMathersCParkinDMEstimates of worldwide burden of cancer in 2008: GLOBOCAN 2008Int J Cancer20101272893291721351269
- Danish Board of HealthThe Danish Register of Causes of Death 2010CopenhagenDanish Board of Health2011http://www.sst.dk/publ/Publ2011/DAF/Doedsaarsagertal2010.pdf
- KuoYWJerngJSShihJYChenKYYuCJYangPCThe prognostic value of the simplified comorbidity score in the treatment of small cell lung carcinomaJ Thorac Oncol20116237838321173715
- BirimOKappeteinAPBogersAJCharlson comorbidity index as a predictor of long-term outcome after surgery for nonsmall cell lung cancerEur J Cardiothorac Surg200528575976216157485
- Janssen-HeijnenMLSmuldersSLemmensVESmeenkFWvan GeffenHJCoeberghJWEffect of comorbidity on the treatment and prognosis of elderly patients with non-small cell lung cancerThorax200459760260715223870
- BlancoJATosteISAlvarezRFCuadradoGRGonzalvezAMMartínIJAge, comorbidity, treatment decision and prognosis in lung cancerAge Ageing200837671571819004964
- TammemagiCMNeslund-DudasCSimoffMKvalePImpact of comorbidity on lung cancer survivalInt J Cancer2003103679280212516101
- SørensenMPijls-JohannesmaMFelipESmall-cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-upAnn Oncol2010215 Suppl12012519605511
- CrinòLWederWvan MeerbeeckJFelipEEarly stage and locally advanced (non-metastatic) non-small-cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-upAnn Oncol2010215 Suppl103115
- AsamuraHGoyaTKoshiishiYA Japanese Lung Cancer Registry study: prognosis of 13,010 resected lung cancersJ Thorac Oncol200831465218166840
- MerrillRMSloanAAndersonAERykerKUnstaged cancer in the United States: a population-based studyBMC Cancer20111140221936934
- BarlowLWestergrenKHolmbergLTalbäckMThe completeness of the Swedish Cancer Register: a sample survey for year 1998Acta Oncol2009481273318767000
- GjerstorffMLThe Danish Cancer RegistryScand J Public Health2011397 Suppl424521775350
- StormHHMichelsenEVClemmensenIHPihlJThe Danish Cancer Registry – history, content, quality and useDan Med Bull1119974455355399408738
- AJCC Cancer Staging Manual6th edSpringer-Verlag2002
- PedersenCBThe Danish Civil Registration SystemScand J Public Health2011397 Suppl222521775345
- BjerregaardBLarsenOBThe Danish Pathology RegisterScand J Public Health2011397 Suppl727421775357
- LyngeESandegaardJLReboljMThe Danish National Patient RegisterScand J Public Health2011397 Suppl303321775347
- CharlsonMEPompeiPAlesKLMacKenzieCRA new method of classifying prognostic comorbidity in longitudinal studies: development and validationJ Chronic Dis19874053733833558716
- ThygesenSKChristiansenCFChristensenSLashTLSørensenHTThe predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of PatientsBMC Med Res Methodol2011118321619668
- YuJBGrossCPWilsonLDSmithBDNCI SEER public-use data: applications and limitations in oncology researchOncology (Williston Park)200923328829519418830
- KlassenACCurrieroFKulldorffMAlbergAJPlatzEANelomsSTMissing stage and grade in Maryland prostate cancer surveillance data, 1992–1997Am J Prev Med2006302 SupplS77S8716458794
- KoroukianSMXuFBeairdHDiazMMurrayPRoseJHComplexity of care needs and unstaged cancer in elders: a population-based studyCancer Detect Prev200731319920617658225
Appendix 1
Stage categories for SCLC and NSCLC allowing for specific missing T, N, or M values
Small cell lung cancer
Limited disease
T0–4,x N0–3,x M0
Extensive disease
T0–4,x N0–3,x M1
Unknown
T0–4,x N0–N3,x Mx
Non-small cell lung cancer
Limited disease
T0–2 N0–1 M0 = Stadium IA–IIB
Extensive disease
T0–2 N2 M0 = Stadium IIIA
T3 N0–2 M0 = Stadium IIIA (& IIB)
Distant disease
T4 N0–3,x M0,x = Stadium IV
T0–3,x N3 M0,x = Stadium IIIB
T0–4,x N0–3,x M1 = Stadium IV
Unknown
Tx N0–2,x M0,x
T0–3 Nx M0
T0–3 N0–2,x Mx