4. Discussion
The main finding in this study was increasing rates of brain tumours of unknown type in the central nervous system (D43). One might speculate that these tumours represent metastases to the brain of tumours at other sites in the body. This is however unlikely since these are coded as tumours located at the original site. In the IPR the number per 100,000 patients was relatively stable until 2007 when a joinpoint was detected for men and 2008 for women. After that year the rate increased statistically significant yearly with almost 5% in men and more than 4% in women. The tumour may be calculated several times if registered during several years. It is however unlikely that this explains the clear change of the rate in
Figure 1 and of course not the increasing death rate in
Figure 2. Also it must be stressed that the same diagnosis is not included several times in the IPR during the same year if the person moves from one county to another, for example from a county hospital to a university hospital. All counties in Sweden are included in the same register and only unique personal id-numbers are used.
It is striking that the yearly age-standardized death rates per 100,000 inhabitants increased in the CDR from 2008 for brain tumours of unknown type (D43), the joinpoint in both men and women. The yearly statistically significant increase was 24% in men and 20% in women. For the time period 1998–2008 the rate decreased statistically significantly in both men, 9%, and in women, 6%. These results indicate that the increasing rate in the IPR is caused by inpatients with a malignant brain tumour with short survival, since the total (men and women) joinpoint shifted in 2007 in IPR, but one year later, 2008, in CDR.
It should be noted that for both tumours of unknown type (D43) and malignant brain tumours (C71) the same year, 2008, for joinpoint was detected in CDR. Thus, during 2008–2013 the increasing rate for D43 and decreasing rate for C71 might be inversely related, see
Figure 8. However clearly it can be seen in
Figure 9 that the statistically significant increasing rate of D43 in IPR 2007–2013 cannot be explained by the non-statistically significant decreasing rate of C71 in IPR 2006–2013. It might be speculated that the diagnosis is coded both as an unspecified brain tumour (D43) and as malignant brain tumour (C71) at the same time. This is unlikely since at the hospital discharge the diagnosis is usually based on best available data (cytology/histopathology). Furthermore, the results in
Figure 9 do not support that suggestion. A patient might, however, be diagnosed with the code D43 first and at a later hospital discharge the same year with C71. This notion is not supported by the results in
Figure 9; there would have been an increasing rate of C71 also if this was a common case.
In contrast to brain tumours of unknown type only a slight increase of AAPC was found in IPR during 1998–2013 for benign meningeal tumours (D32), the majority would be meningioma. The AAPC for these tumours decreased in the CDR. The reason for that is unclear but might indicate better treatment options for this patient group, other clinical routines or most probably not reported to CDR as the cause of death. The latter is supported by the low number of cases in CDR compared with IPR.
Tumours of unknown type in the meninges (D42) were rather few and AAPC decreased statistically significant in the IPR. In men a joinpoint was detected in 2003, but these results were based on low numbers. The decreasing AAPC in both genders might reflect better diagnosis and in fact improved diagnosis leading to classification as benign meningeal tumours (D32).
Figure 8.
Joinpoint regression analysis of age-standardized death rates per 100,000 inhabitants according to the Swedish Causes of Death Register for both genders combined, all ages during 1998–2013 diagnosed with D43 = tumour of unknown type in the brain or CNS and C71 = malignant brain tumours [
29].
Figure 8.
Joinpoint regression analysis of age-standardized death rates per 100,000 inhabitants according to the Swedish Causes of Death Register for both genders combined, all ages during 1998–2013 diagnosed with D43 = tumour of unknown type in the brain or CNS and C71 = malignant brain tumours [
29].
Figure 9.
Joinpoint regression analysis of number of patients per 100,000 inhabitants according to the Swedish National Inpatient Register for both genders combined, all ages during 1998–2013 diagnosed with D43 = tumour of unknown type in the brain or CNS and C71 = malignant brain tumours [
27].
Figure 9.
Joinpoint regression analysis of number of patients per 100,000 inhabitants according to the Swedish National Inpatient Register for both genders combined, all ages during 1998–2013 diagnosed with D43 = tumour of unknown type in the brain or CNS and C71 = malignant brain tumours [
27].
Benign tumours in the brain or CNS (D33) were a rather large group in the IPR. There was a slight decrease of AAPC but not statistically significant. Due to low numbers analysis of the CDR was less meaningful. This is a heterogeneous group of tumours that do not permit any firm conclusions.
All data in this study were based on official statistics in Sweden. Gender specific APC and AAPC in IPR and CDR were calculated per 100,000 persons. The data were not adjusted for age in the IPR since such information is lacking in that register. However, during a limited time period the age groups would be relatively stable in Sweden. In contrast, data in the CDR and the Swedish Cancer Register permitted calculation of trends in age-standardized rates per 100,000 subjects.
It is postulated that glioma patients would need more inpatient hospital care than patients with benign brain tumours such as meningioma. Only 1%–3% of meningioma cases have an anaplastic tumour, Grade III, with poor survival, one series reported a median survival of 2 years [
35]. Atypical meningioma, Grade II, constitute 5%–7% of all cases whereas 90% are benign Grade I meningioma. Using gamma knife radiosurgery long-term survival has been reported for meningioma with 20-year actuarial rates of freedom from tumour progression of 87% [
36]. Thus most meningioma cases would not need much hospital care as inpatients.
Most of glioma cases are astrocytoma. About 5% of all astrocytoma consist of the pilocytic type (WHO grade I) that occurs mainly in children and young adults. Long-term survival, even decades, is common and malignant transformation is rare.
Diffuse astrocytoma (WHO grade II) represents 10%–15% of all astrocytoma. The peak incidence is among young adults aged 30 to 40 years with a slight predominance in males. The mean survival after surgery is 6–8 years. The prognosis is however mainly influenced by malignant progression to glioblastoma, which has been reported to occur after 4–5 years [
37,
38].
Anaplastic astrocytoma (WHO grade III) is a diffuse infiltrating tumour that primarily affects adults with mean age about 40 years. It may arise from a diffuse astrocytoma, WHO grade II, or without evidence of a malignant precursor. Without surgery it has a strong tendency for progression to glioblastoma (WHO grade IV) with a mean time of about 2 years [
39].
The most common primary brain tumour is glioblastoma (WHO grade IV) that accounts for about 60%–75% of all astrocytic tumours. The peak incidence is between 45 to 75 years of age. Secondary glioblastoma develops from diffuse astrocytoma, WHO grade II, or from anaplastic astrocytoma, WHO grade III. Secondary glioblastoma is less frequent than primary, less than 10%, and typically develops in younger persons, mean age 45 years. The time interval from diffuse astrocytoma grade II to glioblastoma grade IV varies between less than 1 year to more than 10 years, with mean interval 4–5 years [
38,
40]. Survival of patients with secondary glioblastoma has been reported to be longer (median 7.8 months), than for primary glioblastoma (median 4.7 months) [
39,
41].
During 1980–2013 the frequency of autopsy based diagnosis of nervous system tumours in the Swedish Cancer Register decreased from almost 20% in men and about 15% in women to 0 to 2% in both genders. This should have contributed to the statistically significant decreasing APC for brain tumours (ICD-7 code = 193.0) in men during 1980–2013 with −0.36%, 95% CI −0.58, −0.13%. In women there was a non-significant increase, APC +0.06%, 95% CI −0.20, +0.32% for the same time period.
Of all cancer diagnoses 9% in men and 6% in women were incidentally found at autopsy in 1980
versus 0% in both genders in 2013. There has been a general decline of the frequency of all autopsies in Sweden from about 50% in the early 1970’s to about 11% in 2013 for both men and women. In men aged 75+ years the frequency has declined from 80% since 1987 to only 6% in 2013 [
42]. The lower frequency of autopsy may cause registration of the wrong cause of death in the Death Certificate. It is especially alarming that during the last years only 0–2% of CNS tumours are detected during autopsy.
In 1980 83% in men and 87% in women with nervous system tumours (ICD-7 code = 193) reported to the cancer registry were verified by histology and/or cytology. This frequency increased to about 90% or more in both genders from early 1990’s, in 2013 for men 92% and for women 95%. During our study period 1998 to 2013 this frequency has been similar during the whole time period. Using CT and MRI provides good information on type of brain tumour that can be used for clinical decisions on treatment, whereby expectancy (watchful waiting) is one option for e.g., low-grade glioma and meningioma. There may also be aggressive tumours such as glioblastoma that are diagnosed by CT and MRI, but may not be operated if the tumour has a difficult anatomical localisation, the patient is in a severe clinical condition or is old. These patients would contribute to the increasing rate in IPR, but due to the bad prognosis with short survival also the increasing rate in CDR. If all these clinically detected tumours had been reported to the Cancer Register one would expect decreasing frequency of cases with histology and/or cytology verification of the diagnosis, which in fact is not the situation in the Cancer Register.
It should also be noted that there seems to be large regional differences in reporting to the Cancer Register, which the results from Stockholm and Västra Götaland counties show. Furthermore it should be noted that the ICD-7 code 193.0 in the Cancer Register includes both malignant and benign tumours, whereas ICD-10 code C71 in IPR and CDR represents only malignant brain tumours. Thus, these results are not comparable. No firm conclusions can be drawn regarding the decreasing AAPC for ICD-10 code C71 in IPR and CDR and these results are not comparable with ICD-7 code 193.0 in the Cancer Register including both benign and malignant brain tumours.
Quality control has shown that brain tumours are among the tumour types with large deficits in the reporting to the Cancer Register [
43]. In a sample from 1998 of 202 medical records 93 cancer cases were identified that should have been reported to the Cancer Register. Of these 30% had a cancer diagnosis verified by histology (19%) or cytology (11%) whereby the pathology departments around Sweden had missed to report these cases. For 70% the cancer diagnosis was based on clinical examination only and were thus not reported to the Cancer Register by the clinical departments. Of patients with nervous system tumour the ratio of the number of persons with a tumour reported to the IPR to the number of persons reported to the Swedish Cancer Register was 48.2% for university hospitals, 121.1% (more than half not reported) for county hospitals, 4.5% for local hospitals and nursing homes, or 13.9% in total. This was in spite of reporting being mandatory and at least most pathology departments have routines for that.
A study of patients reported to the Swedish register of palliative care in 2009 as deceased due to cancer indicated that 12.5% of patients dying of cancer were not reported to the Cancer Register [
44]. Radiologic investigation without a biopsy was the most common basis for diagnosis and the majority of not reported cancers were tumours in visceral organs that are easily visualized by modern imaging techniques such as CT scan but hard to reach for biopsy. Brain tumours were not part of the study but the same discussion may be applied for these tumours, especially glioblastoma multiforme. These tumours are easily diagnosed with CT/MRI but may be difficult to take a representative biopsy from or operate without a risk for neurological complications. There are limited therapeutic possibilities and short survival for these patients. It should be especially noted that most cases with nervous system tumours (ICD-7 code 193) in the Cancer Register have a diagnosis confirmed with histology or cytology as can be seen in
Figure 4. The frequency has even increased since early 1990’s. That figure clearly indicates that cancer cases with only clinical diagnosis such as CT and/or MRI are underreported to the Cancer Register.
A population-based incidence study on pancreatic and biliary tract cancer for the years 1990–2009 found an overwhelming underreporting of these cancers within the Swedish Cancer Register [
45]. Thus during 1990–1994 63% of pancreatic cancer cases were reported, but declined to 44% in 2005–2009. The corresponding results for biliary tract cancer were 60% and 37%, respectively. These results reflected the lack of tissue samples for histology. It was concluded that these errors may explain the reported decline of these cancer cases in the Swedish Cancer Register during recent years.
Incidence data on brain tumours in the Swedish Cancer Registry are used to dismiss the increased risk associated with use of wireless phones in our studies [
18,
20]; see also SCENIHR and WHO [above]. Interestingly this study showed increasing rates of brain tumours of unknown type with joinpoint 2007 (all). This may be in accordance with our findings of increased glioma risk associated with use of wireless phones. In fact this increase is some years after increasing number of out-going mobile minutes. These data do also fit with our restricted cubic spline plot with a late effect in carcinogenesis [
32], see also Appendix II in the Interphone study [
12]. Tumour promotion by RF-EMF exposure was reported in 2010 in a study on mice [
46]. These findings were recently replicated [
47] and add to the relevance of our results. An even more increasing rate of brain tumours in the future may be predicted based on these data. In contrast the data do not show a statistically significant increase of the rate of meningioma, a tumour type that has no consistent association with use of wireless phones [
8,
48]. Only ionizing radiation is since before an established risk factor for brain tumours. However there are no data indicating increasing such exposure in the Swedish population that may explain our results.