Population-based epidemiological data of follicular lymphoma in Poland: 15 years of observation

Available epidemiological reports on follicular lymphoma (FL) often highlight a significant discrepancy between its high and low incidence rates in Western and Eastern Europe, respectively. The reasons behind that difference are not fully understood, but underreporting is typically presumed as one of the main factors. This study aimed to assess FL epidemiology in Poland based on 2000–2014 data from the Polish National Cancer Registry, which has 100% population coverage and over 90% completeness of the registration. All cases were coded according to ICD-10 and ICD-O-3 recommendations. The total number of registered FL cases was 3,928 with crude (CR) and standardized (SR) incidence rates of 0.72/105 and 0.87/105, respectively. The median age of FL diagnosis was 61 years, with the male to female incidence ratio of 1.06. The distribution of morphological types of FL: not otherwise specified (NOS), grades 1, 2, or 3 were 72.58, 4.81, 12.88, and 9.73%, respectively. Among all reported mature B-cell non-Hodgkin lymphomas, FL was ranked the fourth in incidence, just after chronic lymphocytic leukemia/small lymphocytic lymphoma (CR 3.62/105, SR 4.99/105), plasma cell neoplasms (CR 3.78/105, SR 4.97/105) and diffuse B-cell lymphoma, NOS (CR 2.13/105, SR 2.65/105). The systematic increase in FL incidence among females was observed. Our study confirms a lower FL incidence rate in Poland as compared to other European countries. Moreover, as our analysis was based on a registry with high data completeness, it provides evidence that reasons other than underreporting are responsible for FL incidence discrepancies between Eastern and Western Europe.

www.nature.com/scientificreports/ codes. The majority of cases are categorized as low-grade (grade 1/2) lymphomas, while high-grade (grade 3) subtypes are being reported in about 10-20% of patients. The frequency of grade 3A versus 3B has not been deeply studied 6 . The pure grade 3B is rare and contains diffuse areas composed of centroblasts 4,7 . Biologically, it is more closely related to DLBCL, NOS, and manifests clinically with higher short-term mortality and intermittent remissions after chemotherapy 8 . The cases without grade specification (not-otherwise specified, NOS) are classified together with pediatric-type FL. In addition, the revised 4 th edition of the World Health Organization (WHO) classification of Tumours of Haematopoietic and Lymphoid Tissues separates in situ follicular neoplasia (ISFN), formerly referred to as follicular lymphoma in situ. In reactive lymph nodes, ISFN is found in about 3% of cases and is associated with a low rate of clinical progression 9 .
There are only a few published epidemiological studies on FL in Poland 3 . However, the current consensus is almost exclusively based on the data published in the HAEMACARE project for Europe, which covered only about 10% of the Poland population (data from 3 local registries) 1 . Here, we analyze the FL incidence and mortality in Poland based on the data from the Polish National Cancer Registry (NCR), which has 100% population coverage and over 90% completeness of the registration. We report the first long-term observational data being under histopathological supervision. The FL trends against other B-cell NHLs and age group-specific rates are also investigated.

Results
Among all of the new cancer cases reported to the NCR in the years 2000-2014, FL accounted for 6.3% of all mature B-cell NHLs, with 5.6% in the male and 6.9% in the female populations, respectively. FL was the fourth most common mature B-cell NHLs in Poland (see Table 1, Supplementary Fig. S1 online). The number of reported FL cases increased from 209 in 2000 to 298 cases in 2014, with the upward tendency more strongly marked in the female population (see Fig. 1A). We observed no significant trend in the standardized overall FL incidence rates (P-val = 0.34), a borderline significant decrease in the standardized FL incidence rate among males (P-val = 0.07), and a significant increase in the standardized incidence among females (P-val < 0.001) (see Fig. 1B). The incidence rate for the studied period increased by about 50% in the female population, from approximately 0.6/10 5 at the beginning of the twenty-first century to 0.9/10 5 in 2014.
In 2014, the standardized death rate in Poland for patients with FL was 0.59/10 5 for men and 0.37/10 5 for women. The higher mortality rate for men was observed in the whole follow-up period; in 2000 and 2014, differences in mortality rates between males and females were over 55% and 46%, respectively. The mortality rates for men and women showed no significant trends, and the values of the standardized mortality rate were 0.43-0.76/10 5 for men and 0.25-0.44/10 5 for women (see Fig. 2).
Distribution of morphological FL subtypes indicated a high proportion of not otherwise specified (NOS) cases concerning the histological grade; FL NOS accounted for 75% in the male and 70% in the female population. Among the data reported to the NCR, low-grade FL was the most frequently indicated (Grade 1 and Grade 2), and represented 48% and 18% of cases for men and women, respectively. More than one third were FLs with a higher grade of histological malignancy, referred to as Grade 3 (see Supplementary Fig. S2 online).
FL was primarily diagnosed in adults, with a median age of 61 years (60 years for men and 61 years for women). According to the NCR data, only about 10% of cases are observed before the age of 40. The highest number of cases in men occurs between 75 and 99 years of age, while in women, between 65 and 74 years of age. The incidence rates for FL are similar in the population of women and men in almost all age groups except in the 65-74 and 75-99 age ranges where the incidence rate is higher in the male population than in the female one by 22% and 41%, respectively. Age group-specific incidence of FL shows almost similar distribution as plasma cell neoplasms; the diffuse B-cell lymphoma, the most frequent mature B-cell NHLs, is observed in patients in more advanced age (> 75 years old) (see Fig. 3).

Discussion
Global patterns and trends in the FL incidence remain poorly understood since FL cases are registered as part of a wide range of NHLs. According to recent population-based studies, i.e. the GLOBOCAN database, the identification of FL among NHLs is not possible 10 . However, our study, based on the comprehensive NCR data, allowed us to calculate the standardized incidence rate for FL in Poland reliably. We found that within the period 2000-2014, FL was ranked fourth among all diagnosed mature B-cell NHLs and had an incidence rate of 0.87/10 5 . Our analysis is in agreement with the results from the Polish histopathological registry of lymphomas-a nonpopulation wide histopathological data collection managed by the Polish Lymphoma Research Group-where FL comprised only 4.89% of NHLs 3 .
It has been previously argued that the low incidence rates for both lymphoid and myeloid malignancies in Eastern Europe compared to Western Europe could be a result of underreporting 1,2 . In our study, however, the population coverage is 100%, and histopathologically confirmed cases together with the completeness of the registration are over 90%.
In HAEMACARE analysis, the standardized incidence rate for patients up to 54 years from Eastern Europe was found similar to other parts of Europe, whereas lower incidence was observed in the 75-to 99 age group 1 . Our data does not support that observation as the highest number of cases in men occurs between 75 and 99 years of age, while in women, between 65 and 74 years of age. We cannot conclude that the FL incidence is lower due to lower life expectancy in Poland or less frequent diagnostic investigation of the elderly patients.
In addition, high pathological and clinical FL heterogeneity may introduce some case registration errors. There is a possibility that the newly diagnosed patients with high-grade FL "skip" the FL registration because they are usually qualified for the DLBCL, NOS treatment regimen. On the other hand, the low-grade FL cases frequently require only the "watch and wait" strategy, and physicians might neglect to report such cases 11 www.nature.com/scientificreports/ Moreover, due to indolent behavior and long follow up periods, there can be multiple registrations of the same patient in various local registries. To prevent the latter, registrations are always identified and double-checked at the level of the NCR. FL incidence differences between populations of Western and Eastern Europe may be associated with variable exposition to known risk factors. The FL prevalence, in contrast to NHLs etiologically associated with EBV, HIV, or HTLV-1 infections, does not reflect any infectious origin 13 . Pesticide exposure measured as lifetime-days of exposure and adjusted risks for NHLs subtypes and FL indicates a significant statistical correlation to lindaneisomer gamma 1,2,3,4,5,6-hexachlorocyclohexane (HCH), a chlorinated hydrocarbon insecticide, which was banned worldwide in 2009-and diazinon-an organophosphate insecticide 14 . After the Second World War, lindan was widely sprayed in Europe and US, and the unwanted by-products (hazardous HCH waste) were discharged at many sites. The release of toxic water-soluble HCH contaminated soil and bioaccumulates via the food chain 15 . According to the latest environmental report, Poland is a country struggling with persistent organic pollutants. However, the scale of contamination was much lower than in other European countries, i.e., Germany, the Netherlands, Spain, France, or Czech Republic 16 . www.nature.com/scientificreports/ The large-scale epidemiological analyses showed complex and multifactorial etiology of FL, with a history of cigarette smoking and alcohol consumption being one of the most powerful factors. Cigarette smoking is associated with an increased risk of FL 17,18 . Smoking habits in Eastern European countries are changing dynamically. Currently, Poland is ranked in the top 10 countries around the world with the steepest annual decline in smoking prevalence in both sexes; however, over 8 million Poles are still everyday smokers 19,20 . In contrast, higher alcohol intake is related to a reduced risk of FL; that correlation was confirmed in several epidemiological studies 21,22 . Poland continuously belongs to one of the world leaders in the consumption of pure alcohol, and the alcohol intake maintains an upward trend 23-26 .
The genetic variations, including the major histocompatibility complex class II, i.e., specific haplotypes of human leukocyte antigen, are thought to be strongly related to FL susceptibility [27][28][29] . Moreover, particular genes polymorphisms might be associated with the elevated risk of lymphoma-specific death, lymphoma progression, www.nature.com/scientificreports/ or overall survival 30 . There are no extensive molecular and genetic studies for the Poland population; the results include only single-center FL studies with a limited number of cases that can not provide more in-depth epidemiological insight into FL etiopathogenesis 31 .
Our study was conducted in the rituximab era. In 1997 this monoclonal anti-CD20 antibody gained the Food and Drug Administration approval in FL treatment, and in early 2000 the European Union and Poland have started its implementation 32,33 . The standardized death rate in Poland in 2014 is similar to the SEER data 34 . In our analysis, the inferior mortality among males as compared to females is described -the differences are reaching 50%. This phenomenon might be better understood if we consider the fact that the highest number of cases in men occurs between 75 and 99 years, while in women a decade earlier (between 65 and 74 years of age). More advanced age may limit the treatment options due to overlapping comorbidities. Despite the improvement in FL overall survival rate, the leading cause of death is lymphoma-related, especially transformation to DLBCL, NOS 35 . Diminishing of FL mortality is still of great interest and encourages the investigation for less-toxic therapies.
In summary, our study confirms that Poland belongs to a group of countries with a low incidence rate of FL. Our results based on NCR data from over 15 years of observation provides a reliable source of epidemiological FL characteristics. The uprising FL incidence trend in females is marked, but still lower than in Central and Western Europe. The FL mortality remains stable in Poland; its comparison to the other European countries is difficult due to different ways of presenting data. In the near future, we may expect an improvement in the standardized death rate as a result of better prognostics and new therapeutic options. A variety of etiological risk factors and genetic susceptibility should still be investigated for better understanding of FL etiopathogenesis in the Eastern European region.

Material and methods
The data source for this study was the Polish National Cancer Registry of Poland (NCR), a nationwide, population-based cancer registry with data collected by registration offices in 16 voivodeships. New cases of lymphoma in Poland are documented by cancer registration forms. A double system of codification was applied according to the 10th Revision of the International Classification of Diseases and Related Health Problems (ICD-10) and the 3rd revision of the Classification of Diseases for Oncology (ICD-O-3). The low-grade FL was reported with the 9,695/3 and 9,691/3 codes which were assigned to grade 1 and grade 2, respectively (ICD-10: C82.0 and C82.1); grade 3 (high grade; ICD-O-3: 9,680/3, ICD-10: C82.2) is further subdivided into A (ICD-10: C82.3) and B (ICD-10: C82.4) 36,37 . The percentage of histopathologically confirmed cases and completeness of the registration exceeded 90%.
The Statistics Poland was the source of cancer mortality data (death certificates) and population size and structure of the Polish population by sex and by 5-years age groups for each observation year.
Number of cases, percentages, crude rates, and age/sex standardized rates were calculated based on the revised European Standard Population from 2014 38 . Confidence intervals for crude and standardized incidence rates were calculated using the method based on the gamma distribution. To analyze the significance of trends, we investigated significance (statistical P-value) of Pearson's product-moment correlation coefficient between year and rate. Statistical analysis was performed using R version 3.6.3 software (R Foundation for Statistical Computing, Vienna, Austria) and dsr package implemented therein.  www.nature.com/scientificreports/