Educational and regional differences in population-based incidence and survival of primary central nervous system lymphoma in Finland

Abstract

Background

This is the first nationwide population-based study to assess educational and regional differences in the incidence and survival of primary central nervous system lymphoma (PCNSL) in Finland.

Methods

Data were extracted from the Finnish Cancer Registry based on the histological diagnosis and tumor location for cases of large B-cell lymphoma (LBCL) between 2007 and 2022. Data were divided into five healthcare regions of the Finnish healthcare system, which included 623 PCNSL patients (48% female), 27% of whom had higher education, 34% had secondary education, and 39% had primary or unknown education. The incidence rates of PCNSL and LBCL were calculated and the proportion of PCNSL incidence to LBCL incidence was estimated. The age-standardized relative survival rates were calculated within subgroups defined by educational level and healthcare region.

Results

The annual proportion of PCNSL incidence to LBCL incidence remained stable at 6.5% between periods 2007–2014 and 2015–2022 in Finland (p = 0.324). The age-adjusted incidence rate standardized by the World Health Organization World Standard Population (WHO WSP 2000–2025) was 0.54/100,000 person-years. The incidence rates did not differ between healthcare regions (p = 0.097) or between groups defined by educational level (p = 0.725). From 2007 to 2022, the 2-year and 5-year age-adjusted relative survival rates were 38% and 30%, respectively. In between healthcare regions, the 2-year age-adjusted relative survival rate varied between 25% and 56%, and the 5-year age-adjusted relative survival rate between 16% and 45% (p < 0.001). No statistically significant difference in survival rates was observed according to the educational level (p = 0.282).

Conclusions

The incidence of PCNSL in Finland remains one of the highest observed in population-based studies. However, the proportion of PCNSL incidence to LBCL incidence remained stable between periods 2007–2014 and 2015–2022. The incidence rates did not differ between educational levels or healthcare regions. A minor trend towards improved survival was observed among those with a higher educational level. Remarkable differences were observed in survival after diagnosis of PCNSL according to patients’ healthcare regions.

Background

Primary central nervous system lymphoma (PCNSL) is a rare and aggressive form of non-Hodgkin lymphoma. In approximately 90% of the PCNSL cases, the histological subtype is diffuse large B-cell lymphoma (DLBCL) [1]. DLBCL is a subtype of LBCL, a diverse group of aggressive B-cell lineage neoplasms [2]. PCNSL affects the central nervous system (CNS) in the brain, spinal cord, leptomeninges and eyes [3, 4]. In some cases, the vitreoretinal areas of the eyes are manifested without systemic involvement [5]. The current standard first-line treatments include high-dose methotrexate-based combination chemotherapy (for example, MATRIx) and consolidation therapy with autologous stem cell transplantation (ASCT) [6]. High-dose chemotherapy with ASCT and whole brain radiotherapy are the available consolidation strategies [7]. However, the long-term adverse cognitive effects of irradiating the whole brain, particularly among older patients, have led to most patients being consolidated with high-dose chemotherapy with ASCT.

The incidence of PCNSL is increasing worldwide, especially in Western countries [8, 9], concurrently with the aging population [10–12]. The comparison between studies is challenging due to various reference populations. The age-adjusted incidence of PCNSL in USA is approximately 0.4/100,000 person-years, with USA 2000 as reference population [11, 12]. For example, in Australia, the reported age-adjusted-incidence rate was 0.36/100,000 during 2000–2014, standardized to the World Health Organization World Standard Population 2000–2025 (WHO WSP 2000–2025) [13]. Finland has a high incidence of PCNSL, with a previously reported incidence of 0.68/100,000 person-years (Finnish population 2015) [3, 10]. The established risk factors for PCNSL include advanced age (> 65 years), male sex, and immunosuppressive conditions [14]. Interestingly, a recent study suggested that the allele HLA-A31 in the Finnish population may contribute to the differences in lymphoma incidence between Finnish and Swedish populations [15]. Moreover, the overall incidence of lymphoma is increasing [14, 16].

The 5-year population-based survival rate of PCNSL is estimated to be 30% [17, 18]. Studies have consistently demonstrated poor survival outcomes among older individuals [10, 11]. In a previous population-based study from Finland, the 2-year relative survival rate was 33%, and the corresponding 5-year survival rate was 26% during the period 2007–2017 [10]. Survival rates among older patients in Finland are considerably lower; for individuals aged ≥ 75 years, the 2-year survival rate was only 13% [19].

Some studies have shown that socioeconomic background influences incidence and survival among cancer patients [20–22]. Individuals with lymphomas who are highly educated, have higher incomes, and are married have been shown to experience better survival outcomes [23–25]. Similar findings have been reported regarding socioeconomic factors affecting outcomes and treatment plans among patients with PCNSL [26, 27].

In this study, our aim was to investigate the risk of PCNSL across various socioeconomic and geographic groups in Finland and to determine whether these factors influence survival. To our knowledge, this is the first study to examine geographic and socioeconomic differences in the incidence and survival of PCNSL at the national level in Finland.

Methods

Data were extracted from the Finnish Cancer Registry (FCR) based on tumor location and histology, including cases diagnosed between 2007 and 2022. Isolated ocular lymphomas were excluded. ICD codes for topography were C70.0–C72.9 and 9680/3 for morphology. The received dataset from the FCR included completed calculations of the incidence and survival numbers and information on overall incidence and survival rates in relation to patients’ educational level and regional areas of Finland. The dataset did not include individual raw data, and the research group did not participate in data collection. Finland’s population was approximately 5.6 million as of 2021 https://pxdata.stat.fi/PxWeb/pxweb/en/StatFin/StatFin__vaerak/statfin_vaerak_pxt_11rb.px/. In Finland, every wellbeing service county belongs to a one collaborative area responsible for providing healthcare services. In this study, we use term ‘healthcare region’ to refer to these collaborative areas. There are five healthcare regions, and each has one university hospital. The healthcare regions areas are Northern Finland collaborative area with Oulu University Hospital, Eastern Finland collaborative area with Kuopio University Hospital, Inland Finland collaborative area with Tampere University Hospital, Western Finland collaborative area with Turku University Hospital and Southern Finland collaborative area with HUS Helsinki University hospital [28]. In this study, these healthcare regions are presented as numerical codes (region 1–5) in no specific order and without disclosing information. The healthcare regions mentioned above are illustrated on the map available here.

The information on educational level of the patients was included in the data from the FCR. The original information on the educational level is based on data from Official Statistics of Finland (OSF) [29]. The OSF collects data from individuals performed examinations from various institutions and classifies education according to a model based on UNESCO’s International Standard Classification of Education 2011 (ISCED 2011) [29]. Following this classification, education was analyzed using a three-level model: primary/unknown education, secondary education, and higher education.

Finnish cancer registry

The FCR maintains a nationwide database in Finland, including information on date of diagnosis, histology, and survival for all diagnosed cancer cases [30]. Currently, the FCR receives information on cancer diagnosis from pathology laboratories and departments [30, 31]. Healthcare organizations are legally obligated to report new cancer cases to the FCR. Clinical data is obtained from treating institutions and survival from Finnish death cause registry [31]. The FCR nationwide coverage extends from the 1980s onwards [31]. The data from the FCR has been reported to be high of quality, as demonstrated by the completeness and accuracy of the registered information [31]. The completeness for solid tumors was estimated at 96% and for non-solid tumors at 86% [32]. The database includes information on case numbers and survival; however, it does not contain data on patients’ performance status, immunosuppression, or treatments received.

Statistical analysis

Incidence rates were calculated as the number of new cases per 100,000 person-years for each year. Rates were presented as crude and, as age-adjusted as the number of new cases per 100,000 person-years standardized to the WHO WSP 2000–2025 for the period 2007–2022 [33]. Furthermore, incidence rates were analyzed by healthcare region and patients’ educational level and standardized to the WHO WSP 2000–2025. Additionally, Finnish population 2022 was used as a reference for the PCNSL incidence rate of Finland from 2007 to 2022, as this is the most relevant reference for intra-national comparisons, and as it best reflects Finland’s aging population structure. A Poisson regression model was used to compare incidence trends between healthcare regions and between educational groups. The proportion of PCNSL incidence to LBCL incidence (WHO WSP 2000–2025) was reported as a percentage for periods 2007–2014 and 2015–2022, and as an overall percentage for the period 2007–2022. The proportion was assessed, and statistical significance was determined using cross-tabulation with the chi-squared test.

In relative survival analyses, the follow-up time was calculated from the date of diagnosis to the date of death from any cause or the last follow-up date. Age-standardized relative survival was estimated using the age distribution of patients diagnosed during 2007–2022 as weights, with five age groups: 18–44, 45–54, 55–64, 65–74, and ≥ 75 years, applying the Ederer II method. For analyses of the entire patient population, the expected survival was stratified by age, sex, and period. In education specific analyses, expected survival was additionally stratified by education level, and in region-specific analyses by healthcare region, to account for differences in background mortality between subgroups and over time.

For incidence and survival rates, 95% confidence intervals (95% CI) were estimated. Survival analyses were conducted using the Kaplan-Meier method across the entire cohort and within subgroups defined by education level and healthcare region. Two-year survival rates were calculated for the patients diagnosed until the end of year 2020, and respective 5-year survival rates for the patients diagnosed until the end of year 2017. Statistically significant results were defined as those with p-values < 0.05.

Results

Basic demographics

The gender distribution and educational demographics of the study population are presented in Table 1.

Table 1.

Basic demographics of the study population (n = 623)

Basic demographics	n 623	(%)
Sex
Female	302	48
Male	321	51
Education
Primary or unknown	245	39
Secondary	209	34
High	169	27

Incidence

Crude incidence rates of PCNSL in Finland during 2007–2022 with age-adjusted incidence rates standardized to the WHO WSP 2000–2025 and as a reference to the Finnish population 2022

The age-adjusted population-based overall incidence rate of PCNSL was 0.54/100,000 person-years between 2007 and 2022 in Finland, standardized to the WHO WSP 2000–2025. The rate for females was 0.48/100,000 person-years, and for males it was 0.60/100,000 person-years during the same period and standard population. The crude incidence rate during the same period was 0.89/100,000 person-years. Using Finnish population 2022 as a reference, the age-adjusted incidence rate of PCNSL was 0.97/100,000 person-years. Figure 1 presents the annual crude and age-adjusted incidence rates of PCNSL in Finland from 2007 to 2022, standardized to WHO WSP 2000–2025 and using Finnish population 2022 as a reference.

Fig. 1.

Annual crude and age-adjusted incidence rates of PCNSL in Finland from 2007 to 2022, standardized to the WHO WSP 2000–2025 and using Finnish population 2022 as a reference

Relationship between the PCNSL incidence to LBCL incidence in Finland from 2007 to 2022

The proportion of PCNSL incidence to LBCL incidence rate in Finland during periods 2007–2014, 2015–2022, and 2007–2022 is presented in Table 2, and annual rates of PCNSL incidence and LBCL incidence are presented in Fig. 2. The proportion of PCNSL incidence to LBCL incidence remained stable at 6.5% between periods from 2007 to 2014, and from 2015 to 2022. (p = 0.324).

Table 2.

Distribution of PCNSL incidence to LBCL incidence in Finland during periods 2007–2014, 2015–2022, and for the whole period 2007–2022, standardized to the WHO WSP 2000–2025

Years	Incidence of PCNSL1	Incidence of LBCL1	PCNSL incidence as % of LBCL incidence
2007–2014	0.52	7.91	6.5%
2015–2022	0.56	8.50	6.6%
2007–2022	0.54	8.26	6.5%

¹Age-standardized incidence rate per 100,000 person-years, standardized to WHO WSP 2000–2025

Fig. 2.

Annual rates of PCNSL incidence and LBCL incidence from 2007 to 2022 in Finland, standardized to the WHO WSP 2000–2025

Regional and educational incidence of PCNSL in Finland between 2007 and 2022, standardized to the WHO WSP 2000–2025

Finnish healthcare is organized in five healthcare regions. Regional and educational crude and age-adjusted incidence rates of PCNSL are presented in Table 3. Regional age-adjusted incidence rates varied between 0.47 and 0.64/100,000 person-years during the period 2007–2022. There was no statistically significant difference in the incidence rates between the healthcare regions (p = 0.097). Age-adjusted incidence rate by education level varied between 0.52 and 0.56/100,000 person-years from 2007 to 2022. The difference was not statistically significant (p = 0.725).

Table 3.

Crude and age-adjusted incidence rates (WHO WSP 2000–2025) of PCNSL by healthcare regions of Finland and by educational level during 2007–2022. The five healthcare regions are presented by a number (regions 1–5) and educational groups are presented as primary/unknown, secondary and high educational level

	Incidence per 100,0001	Age-standardized incidence per 100,0002
Region
Region 1	0.74	0.47
Region 2	1.02	0.58
Region 3	0.86	0.52
Region 4	1.07	0.58
Region 5	1.01	0.64
Educational level
Primary/unknown	1.26	0.52
Secondary	0.72	0.56
High	0.79	0.53

¹Crude incidence rate

² Age-standardized to the WHO World Standard Population 2000–2025

Survival

Relative survival rates and age-standardized survival of PCNSL in Finland between 2007 and 2022

The 2-year relative survival rate was 40% (95% CI: 36–44%) and 5-year relative survival rate was 33% (95% CI: 29–37%). With age-adjustments, the 2-year age-standardized survival rate was 38% (95% CI: 32–43%), and the 5-year age-standardized survival rate was 30% (95% CI: 24–36%) between 2007 and 2022. For females, the 2-year age-standardized survival rate was 41% (95% CI: 35–46%), and for males, it was 35% (95% CI: 30–40%). The 5-year age-standardized survival rate was 29% (95% CI: 23–35%) for females and 31% (95% CI: 25–36%) for males. Relative survival and age-standardized relative survival of PCNSL with respect to healthcare regions in Finland during 2007–2022.

The survival rates of PCNSL patients across the five Finnish healthcare regions is presented in Table 4; Fig. 3. Variation between regions was statistically significant (p < 0.001).

Table 4.

Regional 2-year and 5-year relative survival rates with age-standardized rates (95% CI) of PCNSL in Finland by healthcare regions. The five healthcare regions are presented by a number (regions 1–5) with survival rates for the entire country of Finland (all regions)

Survival	Relative	Age-standardized relative	Relative	Age-standardized relative
Years from diagnosis	2	2	5	5
Region (95% CI)
Region 1	30% (24–37)	31% (25–37)	22% (16–28)	22% (16–28)
Region 2	26% (18–35)	25% (17–34)	17% (10–26)	16% (9–25)
Region 3	49% (38–58)	48% (37–57)	45% (34–56)	45% (35–55)
Region 4	39% (30–48)	40% (30–49)	33% (23–42)	34% (24–44)
Region 5	61% (49–70)	56% (44–66)	48% (36–59)	41% (30–52)
All Regions	40% (36–44)	38% (32–43)	33% (29–37)	30% (24–36)

Fig. 3.

Regional age-standardized relative survival of PCNSL by healthcare regions and time from diagnosis in Finland during the period 2007–2022

Age-standardized relative survival of PCNSL by educational level in Finland during 2007–2022

The age-standardized survival rates of patients with PCNSL according to their educational level are presented in Table 5. Patients with higher educational level had slightly better survival rates; however, the difference was not statistically significant (p = 0.282).

Table 5.

2-year and 5-year age-standardized relative survival rates (95% CI) by educational level of PCNSL in Finland from 2007 to 2022. The education levels are presented as primary/unknown, secondary and high educational level

Years from diagnosis	2*	5*
Educational level (95% CI)
Primary/unknown	36% (30–42)	30% (24–36)
Secondary	38% (31–45)	27% (20–35)
High	44% (37–51)	36% (29–44)

^*Age-standardized relative survival rate

Discussion

In this study, covering the years 2007–2022, the incidence rate of PCNSL was 0.54/100,000 person-years in Finland, standardized to the WHO WSP 2000–2025. Incidence among males was higher than females. The age-adjusted incidence rate as a reference to Finnish population 2022 was 0.97/100,000 person-years. However, the population of Finland is somewhat older than the World population. Incidence rates did not differ significantly by educational or regional level, and the proportion of PCNSL incidence to LBCL incidence remained stable. However, significant regional differences in survival were observed across the healthcare regions.

The incidence of PCNSL has been increasing in several high-income countries [13, 34–36], although the underlying reasons remain largely unclear. However, the comparison of incidence rates between studies is difficult due to various reference and standard populations used. For example, the reported incidence rate was 0.47/100,000 in the United States between 1980 and 2008 (USA population 2000 as a reference), 0.41/100,000 in Japan from 1989 to 2004 (reference population not given), 0.51/100,000 in Japan during 2016–2019 (USA 2000) and 0.26/100,000 person-years in Sweden between 2000 and 2013 (Sweden population 2000) [34, 35, 37, 38]. In the Netherlands, the incidence rate was 0.27/100,000 from 1989 to 1994, and in Korea it was 0.17/100,000 person-years during 1999–2009, both standardized to Segi’s World Standard Population [8, 39]. In Australia, the incidence rate was 0.36/100,000 in Australia during 2000–2014 (WHO WSP 2000–2025) [13]. The incidence of PCNSL in Finland appears to be among the highest reported when using World population data, although these populations are not directly comparable.

The factors contributing to the PCNSL incidence rates in Finland are not known. Immunosuppression is a recognized risk factor for PCNSL and may partly explain this trend. However, real-world data indicate that immunosuppression is relatively rare in the Finnish population [40]. Population aging does not account for all of the high incidence observed in our study, as the data were age-adjusted. One potential explanation for the increasing number of cases could be the growing use of imaging and diagnostic biopsies among older individuals and patients with poor performance status compared to previous decades.

Despite similarities in lifestyle and healthcare systems, there may be some difference in PCNSL incidence between Finland and neighboring Sweden. Due to differences in reference populations used for age-adjustments, direct comparison of incidence rates between countries should be interpreted with caution. However, we believe that cautious comparison is feasible, given the demographic similarity between the Finnish and Swedish populations. The populations of Finland and Sweden are relatively similar in terms of age structure (mean age 44.0 vs. 42.2 years) in 2024 and life expectancy (81.2 and 83.1 years) in 2022 [41–43]. Possible differences in incidence rates suggests the possibility of a genetic or biological factors in the Finnish population that increase CNS involvement in lymphoma. Supporting a genetic explanation, a recent study found that the HLA-A31 allele is associated with CNS tropism among patients with aggressive B-cell lymphomas [15]. The HLA-A31 allele has been reported to occur at relatively high frequencies in both the Finnish and Japanese populations [37, 44], and moderately high PCNSL incidence rates are observed in both. In the USA, African Americans who have a lower prevalence of HLA-A31 also show lower PCNSL incidence compared to Caucasians [34].

Despite the rising incidence of PCNSL, we found that the proportion of PCNSL incidence rate relative to incidence of patients with aggressive B-cell lymphomas remained stable. These findings suggest that no specific etiological factors are driving the increase in PCNSL, but rather that the increase reflects broader trends affecting all aggressive B-cell lymphomas.

To our knowledge, this is the first study to report educational- and region-specific incidence rates at the national level. We observed no significant variation in incidence across Finland’s healthcare regions. Schiff et al. previously suggested an association between low socioeconomic status and increased PCNSL risk [45]. However, that report was based on a relatively small case–control study and has not been validated in subsequent research. In our study, we found no association between educational level and PCNSL risk.

Overall, the population-based prognosis of PCNSL in Finland remained poor, although it had improved compared to previous studies [10, 20]. This improvement may reflect the adoption of enhanced treatment protocols over the past decade. However, this remains speculative, as our data did not include information on treatment.

Deng et al., using data from the Surveillance, Epidemiology, and End Results (SEER) database, examined the association between socioeconomic status and PCNSL outcomes [46]. They found improved survival among patients residing in less deprived areas. In our study, we observed a slight trend toward better outcomes among individuals with higher educational levels. It is possible that those with higher education are more likely to seek care earlier. Unfortunately, our data did not include information on pre-diagnostic delays or treatments administered; thus, we were unable to explore this further.

Despite similar incidence rates, we identified substantial differences in outcomes between Finland’s healthcare regions. Korfel et al. previously reported in a prospective multicenter study that treatment center was predictive of PCNSL survival outcomes [47]. Patients treated in centers that enrolled at least four patients had longer median survival (31.5 vs. 9.5 months) compared to those treated in smaller centers. Few studies have examined the regional impact on PCNSL outcomes. However, evidence from other cancer types supports the idea that regional variation can influence survival. Seppä et al. investigated regional disparities in cancer survival and equity in Finland by cancer site [48]. They found statistically significant variation between hospital districts for male skin melanoma, prostate cancer, kidney cancer, and pancreatic cancer in both sexes. The authors suggested that these disparities may reflect differences in early diagnostics as well as regional variability in access to and quality of cancer treatment.

Our study has several limitations. The FCR dataset did not include information on patients’ medical histories, immunosuppression, performance status, or treatment. Therefore, we were unable to investigate the underlying causes of our findings in more detail. For instance, WHO performance status at diagnosis and treatment delays are well-established prognostic factors for PCNSL [49]. The dataset was based on histological biopsies, and cases of isolated ocular lymphoma were excluded. Therefore, not all PCNSL cases were captured, and the actual incidence rate in Finland is likely higher. Regional classification was also complex, as Finnish patients are free to choose their treatment hospital. Thus, regional classification may reflect the location of the treating institution rather than the patient’s place of residence. Furthermore, owing to the rarity of the disease, patients are infrequently referred to other centers for treatment. In this study, we observed a lack of sufficient data on incidence rates across different countries, which led to cautious comparisons. Incidence rates should be standardized to the WHO World Standard Population to enable meaningful global comparisons and to assess the true global incidence in future research.

Conclusions

The incidence rate of PCNSL in Finland is high, although its proportion among aggressive lymphomas has remained stable. This suggests that the overall incidence of aggressive lymphomas is increasing in Finland, rather than PCNSL specifically. No significant association was found between incidence and educational or regional level.

Although the population-based prognosis of PCNSL in Finland has improved compared to earlier studies, it remains poor. Risk of PCNSL associated death was statistically significantly different between healthcare regions. Further research is needed to understand the reasons for these disparities. Such insights may guide the future development of healthcare services and research initiatives, ultimately improving patient care and outcomes.

Abbreviations

PCNSL

Primary central nervous system lymphoma

DLBCL

Diffuse large B-cell lymphoma

LBCL

Large B-cell lymphoma

CNS

Central nervous system

ASCT

Autologous stem cell transplantation

WHO

the World Health Organization

WSP

the World Standard Population

FCR

Finnish Cancer Registry

OSF

Official Statistics of Finland

SEER

Surveillance, Epidemiology and End Results

Authors’ contributions

Conceptualization, EO, IP, and OK; methodology, EO, IP, and OK; validation, EO, IP, and OK; formal analysis, EO, HK, AR, IP, and OK; , EO, IP, and OK; , all authors; visualization, EO and TS; supervision, IP, and OK. All the authors have read and approved the final manuscript.

Funding

This research received no external funding.

Data availability

Data supporting the findings are available from the corresponding author upon reasonable request and with permission from the Finnish Cancer Registry.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The need for ethical review and approval was waived owing to the retrospective nature of the study. Study data were extracted in a fully anonymized form from the Finnish Cancer Registry. Therefore, according to Finnish legislation, no approval from an ethical committee is warranted. Patient consent was waived because of the retrospective nature of this registry-based study, and individual patient consent was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.