Introduction

Gliomas are the most frequent primary intracerebral tumors in adults. Most patients with gliomas have a fatal prognosis, and the disease has considerable impact on patients and their families’ physical, psychological, and social status [1,2,3].

Clinical epidemiologic data on patients with gliomas are sparse due to the low incidence, high mortality, and heterogeneous diseases with several tumor subtypes. Primary brain tumors including all intracranial tumors account for about 4% of all new tumors in Denmark [4] and gliomas represent <50% of these. Although relatively rare, gliomas contribute disproportionately to cancer morbidity and mortality.

Gliomas occur in the glial tissue of the central nervous system and are classified in subtypes according to the histology of glial cells [5] and graded from I to IV based on morphology and malignant behavior as specified in the World Health Organization (WHO) classification [6]. The most common glioma types in adults include glioblastoma (grade IV), astrocytic tumors (grade I–III), oligodendroglial tumors (grade II–III), and ependymomas (grade I–III). In recent years, molecular information is increasingly used and newly included in the revised WHO classification 2016 of Tumors of the Central Nervous System [7].

Previously, most clinical studies on characteristics of glioma patients have comprised small sample sizes [8, 9]. In the settings of larger cancer therapeutic trials, selection bias is typically seen due to the recruitment, i.e., older patients and patients with lower performance score are less likely to participate. Large-scale population-based studies are necessary in order to examine the natural history of these relatively rare and highly diverse tumors [1, 10].

The Danish Neuro-Oncology Registry (DNOR) was established in 2009 and for the clinical departments it is mandatory by law to report data prospectively on all glioma patients in Denmark. This nation-wide population-based registry provides data on clinical characteristics with a good validity and very high completeness of patient registration [11, 12]. This makes DNOR a reliable data resource for evaluating symptoms in gliomas in general.

The aim of this study was to describe the epidemiology, clinical characteristics, and symptoms of patients with glioma and evaluate the clinical indicators associated with the grade of glioma.

Materials and methods

Study population

The DNOR is a nationwide clinical cancer database comprising prospectively registered data on adult patients (≥18 years) with gliomas since 2009. The establishment, completeness and validity of DNOR have previously been described [11, 12].

The study included adult glioma patients with a primary neurosurgical procedure providing a histologically confirmed glioma (excluding ependymoma) as registered in the DNOR from January 1st 2009 through December 31st 2014.

Data

The following parameters were prospectively registered in the DNOR and reported in this study: age, gender, glioma type, glioma grade and magnetic resonance imaging of the brain (MRI); characteristics prior to surgery (unifocal/multifocal, contrast enhancement, midline cross-over, size, side, and localization), onset symptoms and duration, as well as preoperative Eastern Cooperative Oncology Group performance status [13]. Onset symptoms encompassed focal deficit, epileptic seizure, cognitive change and headache, and symptoms most frequently progressing over weeks or months. Focal deficit comprised motor paresis, motor control deficiency, paresthesia, visual disturbances, and/or speech disturbance. Cognitive change comprised cognitive deterioration or change in personality or behavior.

Statistical analyses

For the descriptive statistics, we used the Χ 2 test for categorical variables and ANOVA for continuous variables, which were all normally distributed. If ANOVA was significant, we compared groups pair-wise by the Bonferroni multiple-comparison test. Five percent was chosen as the level of statistical significance in all analyses.

We used the age and gender distribution of the background population of the first year (2009), as retrieved from StatBank Denmark [14], to standardize the incidence rates (number of patients per 100,000 person-years) for the years 2009–2014.

To assess which of the clinically relevant parameters (gender, age, performance status, and symptoms) that could predict high-grade glioma (III and IV), we performed univariate and multivariate logistic regression analyses with odds ratios (ORs) and 95% confidence intervals (CI). To further evaluate whether a “dose–response correlation”, as related to the duration of symptoms, predicted a high-grade glioma, we incorporated duration categories (no symptom [reference], duration <1 month, 1–3 months, >3 months) for each of the four symptoms in the model [15].

The survival time after diagnosis was illustrated by Kaplan–Meier plots stratified according to the grade of glioma, and differences tested by the log-rank test.

Approvals

The study was approved by the Danish Data Protection Agency (No. 2008-58-0035/14/6151) and the Danish Health and Medicine Authority (No. 3-3013-514/1). According to Danish law, no approval from the Ethics Board or patient’s consent is required as the study did not require intervention with the patients.

Results

Patient population and incidence rates

The study population included 1930 histologically verified adult glioma patients distributed as shown in Table 1. Glioblastomas were most common and present in 71% (n = 1364).

Table 1 Histologic type and grade of glioma in the DNOR database 2009–2014
Full size table

In 2009, the age-standardized incidence rate of histologically verified glioma patients was 7.3 cases per 100,000 person-years. Glioblastomas were the most frequent with 5.1 cases per 100,000 person-years. The incidence rate was higher in men (8.7 cases) than in women (5.9 cases) and increased with age in both sexes up to age 78 years. In patients above 78 years, a decrease was observed. The age-standardized incidence rates in both men and women increased over the 6-year period, from seven cases in 2010, over eight cases in 2012, to nine cases per 100,000 person-years in 2014.

Demographics

The distribution of patients according to gender, age, and grade of glioma is shown in Table 2. The male:female ratio of 3:2 did not differ significantly between grades (p = 0.09). The mean age at the time of diagnosis was 60 years increasing consistently with higher grade (39, 46, 56, and 64 years, for grade I, II, III, and IV, respectively).

Table 2 Clinical characteristics of the study population according to WHO grade I–IV
Full size table

Onset symptoms

Focal deficits was the most common symptom, reported in 64% of all glioma patients, while seizures (31%), cognitive change (43%), and headache (35%) were less frequently reported (Table 2). Among patients with glioma, grade II epileptic seizures were most frequently reported (58%).

Among patients with focal deficits, epilepsy, cognitive change and/or headache at onset, 51–63% experienced symptoms within 1 month before diagnosis and 77–89% within 3 months (Table 2). However, among patients with epileptic seizures at onset, 19% had experienced attacks more than 3 months before their glioma diagnosis, diminishing from 40% in low-grade, over 19% in grade III, to 10% in grade IV glioma.

Figure 1 shows the combination of symptoms only including patients without missing data of any onset symptoms (N = 1813). The most common symptom combination at onset was focal deficit and cognitive change (29%) whereas only 2% had the combined symptom cluster of focal deficits, epileptic seizures, cognitive change, and headache. Only 3% presented with mono-symptomatic headache, while 13% presented with isolated epileptic seizures.

Fig. 1
figure 1

Venn diagram showing the distribution of different combinations of onset symptoms among 1813 patients with complete data on all onset symptoms

Full size image

Preoperative MRI data

The majority of gliomas was unifocal (85%) and primarily localized in the frontotemporal region (65%) (Table 2). Right- and left-sided tumors were equally common disregarding the grade of tumor. We found no statistically significant associations between the presenting symptoms and the site or anatomic location of the tumor.

The majority of grade IV gliomas were contrast enhancing tumors (96%) whereas 34% of gliomas grade II also showed enhancement. Only about 16% of the gliomas crossed the midline at the time of diagnosis. Grade I tumors were significantly smaller than grade II–IV tumors (p = 0.01) whereas the diameters of the latter did not differ from each other (p = 1.0).

Performance status

The preoperative PS was ≤2 in 90% of the patients at the time of the initial diagnostic work-up. However, the percentage of patients with poor performance were increasing with higher grade; PS > 2 was 3, 3, 7, and 11% in WHO grade I, II, III, and IV, respectively (Table 2).

Clinical indicators for grade of glioma

The results from the univariate and multivariate analyses for high-grade gliomas are presented in Table 3. Multivariate analyses showed that higher age, presenting symptoms of focal deficits for up to 3 months, cognitive change for up to 3 months, and short-term headache up to 1 month before diagnosis independently predicted high-grade gliomas. Multivariate analysis showed that younger age and epileptic seizures for more than 3 months independently predicted low-grade gliomas.

Table 3 Predictive clinical features for high-grade gliomas—univariate- and multivariate-analyses for the effect of gender, age, performance status, and duration of different presenting symptoms
Full size table

Survival

Figure 2 shows survival curves for glioma grade I–IV with significantly increasing survival with lower grade of gliomas (p < 0.00001). The 5-year overall survival was 82, 54, 22 and 3% for grade I, II, III, and IV, respectively.

Fig. 2
figure 2

Survival curves with 95% confidence bounds for glioma grade I–IV

Full size image

Discussion

In this large-scale nationwide population-based study of gliomas in Denmark, we report descriptive epidemiologic data, clinical characteristics and associated symptoms of gliomas, grade I–IV. Identifying specific clinical features of gliomas in population-based studies may be important for early detection of the patients, and in the process of optimizing the illness trajectories for brain cancer patients. Such information is important to maximize treatment efforts and alternatively increase focus on early palliative support initiatives.

In the present study, low-grade glioma was present in 15%, grade III glioma in another 14%, and 71% had glioblastoma. The demographic profile regarding age- and sex distribution is consistent with previous studies. Lower age at onset for low-grade gliomas and male predominance in all histologic types corresponds to previous studies [1, 3, 16]. The mean age of 60 years among patients with high-grade gliomas in the present study is significantly higher than reported in clinical studies [8, 9] in which the recruitment of younger patients with better performance for treatment trials probably caused selection bias. The distribution of gliomas according to subtypes and grade is consistent with previous population-based studies [1, 3, 17, 18].

In most cancer registries, CNS tumors also include meningioma, tumors of the spinal cord, cranial nerves and pituitary gland. Incidence data on CNS tumors are very difficult to compare among nations and between studies as different sources are used to track the incidence data, e.g., national cancer registries or various health system records. Some studies include both intracerebral and intracranial tumors and even spinal tumors, and some register malignant versus benign neoplasms or primary versus secondary brain tumors. Some report overall age-adjusted incidence rates, others age-standardized to the World, US or European Standard Populations alternatively total annual crude incidences in the study population [1, 3, 10, 19, 20], which makes comparisons between studies difficult. In the present study, we found an annual age-standardized incidence of adults with gliomas of 7.3 per 100,000 person-years, which is in agreement with previous Danish reports [4, 20, 21]. The present study only includes histologically verified cases, and the real incidence may therefore be even higher. In a regional population in Denmark, 11% of low-grade and 19% of high-grade glioma were not histologically verified [22, 23]. However, it seems that the incidence rate in Denmark is comparable to the other Nordic countries and the rest of the world [1, 17, 24,25,26]. Gliomas were most common among persons aged 58–78 years. The age and gender standardized incidence rates showed male preponderance and increased incidence with age in agreement with previous findings [17, 19,20,21, 26]. The decrease in persons above 78 years may be due to a survival-cohort effect.

We found increasing incidence rates over the 6 year study period in agreement with recent epidemiologic studies [17, 18]. Previous studies have also reported increased brain tumor incidence in the 1980’s and 1990’s, when the use of CT scan and MRI became more common [10]. Our observed increase may, at least partially, be due to more comprehensive registration over time but also to further improvements in diagnostics including better access and more intensive use of neuroimaging techniques in the diagnostic process of neurological disorders. During the last century, the Danish healthcare has implemented national brain cancer pathway programs with an increased focus on improving cancer survival through initiatives for early help-seeking, improved diagnostics, and expedited diagnosis followed by rapid treatment [27,28,29]. Clinical practice may, thus, have evolved contributing to explaining the observed increase in incidence rates.

There is a lack of data from the primary care service on the development of symptoms, and we have to rely on data from patients referred to secondary or tertiary care centers. In these specialized settings the predictive validity of symptoms is not possible to evaluate due to selection bias. However, our data on presenting symptoms in a large national population-based cohort of glioma patients may be essential when planning optimized disease trajectories for these disorders. Our results emphasize the diversity of glioma subtypes. Patients with low-grade glioma present mainly with seizures, whereas patients with high-grade glioma present with focal deficits and frequently in combination with other symptoms. The mean age of glioma grades differs significantly. Knowledge about these clinical characteristics may be useful in the diagnostic process when deciding for additional investigations in patients where brain tumors may be suspected. Furthermore, the diverse symptoms among brain tumor patients call for the physician’s full attention to help patients with multidisciplinary neurorehabilitation to maintain better quality of life for as long as possible.

Most patients with symptoms of brain tumors see a general practitioner. However, patients with new onset epileptic seizures may go directly to emergency departments. Presenting symptoms of brain tumors, including headache, seizure, cognitive change, memory loss, weakness, etc., are all common symptoms in patients with brain tumor but even more common in patients without tumor, and frequently represent more prevalent disorders, e.g., tension-type headache, depression or dementia. Overall, the positive predictive values of single symptoms of brain tumors presenting in primary care are very low, apart from new-onset epilepsy [25, 28].

Headache is a very frequent symptom in the general population [29, 30] and also among patients consulting a general practice [25, 28]. However, in a general practice the positive predictive value of mono-symptomatic headache complaints pertaining to brain tumor is reported to be very low [25, 28]. The present study shows that headache is infrequent as an isolated onset symptom in glioma patients (3%). The majority of glioma patients with headache had clustering of several other additional onset symptoms, which is in agreement with previous studies [31,32,33,34,35]. Patients consulting general practice purely with headache complaints, may require deeper medical history and general physical as well as neurological examination searching for additional symptoms or signs to identify possible red flag suspicions, before supplementing the diagnostic work-up with neuroimaging of the brain [36, 37]. In Denmark we recommend further diagnostic evaluation in patients with recent debut of a new type of headache or marked changes in a former headache pattern progressing over weeks where thorough medical history and examination by a neurologist has not revealed another likely cause [31]. During the course of the glioma illness, headache may be more frequent [32,33,34].

A previous study has reported that patients with seizures may have long diagnostic latency, i.e., epilepsy precedes the tumor diagnosis by years [38]. In the present study, most patients had symptom durations <3 months before the glioma diagnosis, but different onset symptoms influenced the diagnostic latency. The finding that 19% of the glioma patients had experienced epileptic attacks more than 3 months before getting diagnosed, it may justify urgent specialist referral to ensure precise and early diagnosis. However, it is clinically well-known that many patients with newly onset epilepsy may have normal primary diagnostic work-up including neuroimaging without any intracranial pathology. However, the patients require follow-up and repeated scan when attack frequency or attack types are changing. It cannot be excluded that the diagnostic latency among patients with epilepsy, although partly, may also be explained by the clinical practice with expectation (“wait-and-watch”—strategy) before surgical biopsy among patients with magnetic resonance imaging suspicion of low-grade gliomas. In patients consulting with new-onset epileptic seizures without any other probable cause (e.g., substance abuse or sleep deprivation), neuroimaging in the form of MRI is recommended to identify structural abnormalities and, in the course with continuing or changing attack types, new neuroimaging should be performed in order to identify possible up-coming lesions.

The symptom cluster of progressing focal deficits in combination with progressing cognitive change is highly represented among glioblastoma patients, which is in agreement with previous studies [16, 31, 39] and always requires prompt further neuroimaging investigations.

This study is, to our knowledge, the largest study of the symptomatology of glioma patients based on a nationwide population-based sample of adult glioma patients. The DNOR provides a systematic prospective data collection on clinical data and has a very high completeness of patient registration and high validity of clinical data. All glioma patients were histologically confirmed ensuring against misclassification [11, 12]. On the other hand, non-operated cases were not included, and these may be cases with poor performance due to high age, co-existing illnesses, and tumors in eloquent areas or due to the patient’s wish. A previous Danish study has shown that more than 95% of suspected glioma patients, reported to the Danish Cancer Registry, were histologically verified, either by surgical biopsy or at autopsy [16]. Altogether, our results can only be regarded as representative for histologically confirmed gliomas.

The population-based design minimizes selection bias, and the prospective data recording reduces information bias. Under-recording of some symptoms or signs may have resulted in some features associated with brain tumors less likely to be identified in this study. However, this recording bias is estimated to be less of a concern when comparing features in the subtypes as it may pertain to all cases without respect to the grade of glioma. Missing information for some variables is unlikely to be a result of systematic recording bias.

In conclusion, glioma grade I–IV showed high diversity regarding several demographic and clinical characteristics emphasizing the importance of individually tailored disease trajectories. We found that higher age at onset and presenting symptoms of focal deficits and cognitive change for <3 months’ duration and short term headache (<1 month) were independent indicators of high-grade gliomas, while younger age and epileptic seizures for more than 3 months’ duration were indicative for low-grade gliomas. Headache was a rare mono-symptomatic onset symptom. Supplementary studies on possible specific clinical predictors for outcome of glioma grade I–IV are recommended.