Population-based incidence of lymphoid neoplasms_ Twenty years of epidemiological data in the Girona province, Spain

BACKGROUND
The aim of this study was to describe incidence patterns of lymphoid neoplasms in the Girona province (Spain) (1996-2015), and to predict the number of cases in Spain during 2020.


METHODS
Data were extracted from the Girona cancer registry. Incident cases were classified using the ICD-O-3, third revision, and grouped according to the WHO 2008 classification scheme. Age-adjusted incidence rates to the European standard population (ASRE) were estimated and incidence trends were modeled using Joinpoint.


RESULTS
4367 lymphoid neoplasms were diagnosed in the Girona province. The ASRE for overall lymphoma was 37.1 (95% CI: 36.0; 38.2), with a marked male predominance in almost all subtypes. During 1996-2015, incidence trends remained stable for broader lymphoma categories. According to our predictions, 17,950 new cases of LNs will be diagnosed in Spain in 2020.


CONCLUSIONS
This 'real-world' data will provide valuable information to better inform etiological hypotheses and plan future health-care services.


Material and methods
Data were extracted from the population-based Girona cancer registry, covering a population of 738,976 inhabitants in 2015.Incident cases were registered using the International Classification of Diseases for Oncology, third edition (ICD-O-3) and grouped according to the WHO 2008 classification scheme.Age-adjusted incidence rates to the European standard population (ASRE) were estimated and Joinpoint regression modeling was used to examine temporal trends in agestandardized rates.Sex-and age-specific projections of cases in Spain in 2020 were estimated, based on the National Statistics Institute of Spain projections of population.
During 1996-2015, no statistically significant variations in incidence trends were found for LNs [annual percentage change (APC): 0.1 (95% CI: −0.6; 0.8)], nor for broader lymphoma categories (Table 2).Only the LNs NOS subtype category decreased, evidencing the improved diagnostic specificity for these malignancies during the last years.

Discussion
This paper presents epidemiological data of LNs in the province of Girona (Spain) during a 20-year period.Incidence rates for the most frequent subtypes were in accordance with those reported in France (1980-2009) [2], United Kingdom (2004-2014) [3], Europe (2000-2002) [4], and generally lower in comparison to those from the APC, annual percent change; NOS, not otherwise specified.Calculated only for major lymphoma categories and most incident mature B-cell neoplasms.* p-value < 0.05.
United States (projected incidence in 2016) [5] or Australia   [6].Regarding specific entities, we reported higher rates of mantel cell lymphoma (ASRE = 0.92), to the date only linked to few genetic and environmental risk factors (i.e.atopy, allergy or farm life) [7], which merit further research.In the same vein, Burkitt lymphoma/ leukemia incidence (ASRE = 0.47) was particularly high in our region, a pattern largely reported in other Southern European regions, yet unexplained by well-stablished etiological factors [8].Similarly, we reported higher rates of chronic lymphocytic leukemia/small lymphocytic lymphoma (ASRE = 6.62), mainly attributable to a higher completeness of our cancer registry for this entity rather than to a real peak of incidence in our region.Chronic lymphocytic leukemiaas well as Waldenström's macroglobulinemiahave been shown to be underreported by cancer registries given that they are prone to be indolent and their diagnosis is not based on tissue pathology [9].To ensure a complete coverage of all chronic lymphocytic leukemia cases in our region, we recently performed review of alternative sources of information (e.g.flow cytometry and hematologists databases) and evidenced a 18.2% of underreported cases during 1998-2013 [10].Nonspecialized cancer registries thus, should make additional efforts to ensure the surveillance of these malignancies is entirely accurate.Incidence trends are broadly consistent with those reported for other Western countries, which describe a steady increase in the incidence of LNs during the 80′s and a stabilization in the late 90′s [5,6,11].Furthermore, in the United States, even a slight decline in incidence rates by 2000 has been reported [5], which has been widely related to a decrease in AIDS incidence, but also suggested to reflect coding changes in 2001, when the WHO classification system was first published [12].Despite concordance between ICD-O-3, which incorporates the WHO Classification, and cases originally diagnosed and coded under previous systems is generally high [13], several changes can be rarely overcame.For example, the 2008 International Workshop on chronic lymphocytic leukemia, changed the definition of the disease by requiring an absolute B-cell count of 5000 cells/μL rather than the previous absolute lymphocyte count of 5000 cells/μL, causing many former Rai stage 0 cases to be reclassified as monoclonal B-cell lymphocytosis, a pre-malignant disorder not recorded by cancer registries [14].Overall, our results have to be interpreted cautiously since the numerous changes in the classification of LNs over time could result in errors of diagnosis or coding translation, particularly during the earlier period.
The number of expected lymphoma cases depicts accurately the cancer burden of LNs in Spain, which lacks of a cancer registry with national coverage, but data on specific subtypes should be interpreted with caution.Some WHO LNs subtypes are extremely rare, making estimates less robust.However, overall, these results are interesting for clinicians and public health in evaluating the cost of management and new treatments for these pathologies.
In 2016, a new WHO manual was released [15] but the available surveillance data are through 1996-2015 and do not reflect these updates.The capability of registries to recode previous cases to these new schemes will be limited because many require additional molecular or clinical data [5].However, changes to the 2008 classification are relatively small and would not be expected to change our conclusions.The impact of these changes will be unclear until next years, when they become adopted by pathologists and new entities routinely distinguished in clinical practice.

Conclusions
In conclusion, this study describes in detail the incidence of LNs categorized by 2008 WHO subtype in a large population-based cohort.This has not been performed previously for Spain, and complements the subtype-specific analyses published for Europe [2][3][4], the United States [5], and Australia [6].This 'real-world' data will provide valuable information to better inform etiological hypotheses and plan future health-care services.