The contribution of meta-analytical methodology to the investigation of primary central nervous system tumors

Doctoral Dissertation uoadl:2881580 301 Read counter

Unit:
Τομέας Κοινωνικής Ιατρικής - Ψυχιατρικής και Νευρολογίας
Library of the School of Health Sciences
Deposit date:
2019-09-26
Year:
2019
Author:
Georgakis Marios
Dissertation committee:
Ελένη Πετρίδου, Καθηγήτρια, Ιατρική, ΕΚΠΑ
Γεώργιος Τσιβγούλης, Καθηγητής, Ιατρική, ΕΚΠΑ
Μαρία Καντζανού, Επίκουρη Καθηγήτρια, Ιατρική, ΕΚΠΑ
Παναγιώτα Σουρτζή, Αναπληρώτρια Καθηγήτρια, Νοσηλευτική, ΕΚΠΑ
Γεώργιος Παρασκευάς, Αναπληρωτής Καθηγητής, Ιατρική, ΕΚΠΑ
Σουλτάνα Σιαχανίδου, Επίκουρη Καθηγήτρια, Ιατρική, ΕΚΠΑ
Μαρία Μοσχόβη, Επίκουρη Καθηγήτρια, Ιατρική, ΕΚΠΑ
Original Title:
Η συμβολή της μετα-αναλυτικής μεθοδολογίας στη διερεύνηση των πρωτοπαθών όγκων του κεντρικού νευρικού συστήματος
Languages:
English
Translated title:
The contribution of meta-analytical methodology to the investigation of primary central nervous system tumors
Summary:
Primary central nervous system (CNS) tumors are among the top causes of deaths due to cancer in younger age groups and are associated with poor prognosis. However, effective treatments to halt the progression of the disease are missing. Thus, additional research is required to systematically record and compare the burden of primary CNS tumors worldwide, identify etiological risk factors that would enable the development of preventive and therapeutic strategies, and figure out prognostic biomarkers that would allow optimization of the current management approaches. Primary CNS tumors comprise a highly heterogeneous group of diseases with different etiology, pathology, clinical presentation, and prognosis. Many of the efforts to study the epidemiology of CNS tumors are inherently limited by low sample sizes due to the relatively low incidence of the numerous individual CNS tumor subtypes. To increase analytical power and overcome this limitation, new approaches are required, which would entail pooling of data and collaborative research to maximally exploit available data around the globe.

In the current thesis we leveraged data in different levels of analyses with the objectives to explore features of descriptive, analytical, and clinical epidemiology of primary CNS tumors. Specifically, we pooled data from a collaborative network of population-based cancer registries in 14 countries in Southern and Eastern Europe (SEE) and the US (Surveillance, Epidemiology, and End Results Program, SEER) to explore the incidence, time trends, mortality, and survival patterns of primary CNS tumors and specific subtypes among children (0-14 years), as well as adolescents and young adults (AYAs). We further analyzed data from a Greek nationwide case-control study of CNS tumors recruiting cases from the Nationwide Registry for Childhood Hematological Malignancies and Solid Tumors (NARECHEM-ST) and performed systematic reviews and meta-analyses to explore associations of perinatal and early risk factors with the risk of primary CNS tumors. Finally, we recorded data from all case reports and case series that have to date been published and performed an individual participant data meta-analysis of all described cases of gliomatosis cerebri, a very rare CNS tumor with a widely infiltrating pattern and very poor prognosis.

Within SEE (1990-2014) and SEER registries (1990-2012), diagnoses of 11,438 and 13,573 incident malignant CNS tumors in AYAs were retrieved, respectively. The overall age-adjusted incidence rate of malignant CNS tumors was statistically significantly higher in SEE (28.1/million) compared to SEER (24.7/million). Increasing temporal trends in incidence were documented in 4 SEE registries vs. a rather stable rate in SEER. Mortality rates in SEE (range: 11.8-18.5 deaths/million) were overall higher compared to the overall US population (9.4/million) with rather decreasing trends in both regions. Respectively, survival rates were increasing during a comparable period (2001-2009) in SEE and SEER. Five-year survival was considerably lower in the SEE registries (46%) vs. SEER (67%), a finding consistent across age groups and diagnostic subtypes. Highest 5-year survival was recorded for ependymoma (SEE:76% vs. SEER:92%) and worst for glioblastoma and anaplastic astrocytoma (SEE:28% vs. SEER:37%). Advancing age, male gender and rural residency at diagnosis adversely impacted on outcome in both regions. Childhood pilocytic astrocytomas, comprising the most common CNS tumor in childhood, were also retrieved from SEE registries (N=552) and SEER (N=2,723). The age-adjusted incidence rate of childhood pilocytic astrocytoma during 1990-2012 in SEE was 4.2/million, but much higher in SEER (8.2/million). Increasing trends, more prominent during earlier registration years, were recorded in both regions. Cerebellum comprised the most common location, apart from infants in whom supratentorial locations prevailed. Ten-year survival was 87% in SEE and 96% in SEER. Significant outcome predictors were age<1 year at diagnosis (HR [95%CI]: 3.96, [2.28-6.90]), female gender (HR: 1.38, [1.01-1.88]), residence in SEE (HR: 4.07, [2.95-5.61]) and rural areas (HR: 2.23, [1.53-3.27]), whereas non-cerebellar locations were associated with a 9- to 12-fold increase in risk of death.

In the Greek case-control study (203 cases and 406 age-, and sex-matched controls) instrument-assisted delivery was associated with increased (OR: 7.82, [2.18-28.03]), whereas caesarean delivery with decreased (OR: 0.67, [0.45-0.99]) risk of childhood CNS tumors, as compared to spontaneous vaginal delivery. Maternal alcohol consumption during pregnancy (OR: 2.35, [1.45-3.81]) and history of living in a farm (OR: 4.98, [2.40-10.32]) were associated with higher odds of childhood CNS tumors. Conversely, higher birth order was associated with decreased odds (OR for 2nd vs. 1st child: 0.60, [0.40-0.89] and OR for 3rd vs. 1st: 0.34, [0.18-0.63]). Birth weight did not show a significant association with CNS tumors in this sample (OR per 500 g increment: 1.15, [0.92-1.44]). In a systematic review, after screening >5,000 articles, we identified 41 studies, encompassing 53,167 CNS tumor cases, which explored the association between birth anthropometrics and risk of primary CNS tumors. In the meta-analysis, birth weight >4,000 g was associated with increased risk of childhood CNS tumors (OR: 1.14, [1.08-1.20]). The risk was higher for astrocytomas and embryonal tumors. Increased odds for CNS tumors were also noted among large-for-gestational-age children (OR: 1.12, [1.03-1.22]). In a systematic review, we further explored the association between birth seasonality and risk CNS tumors. Eight out of 10 studies in children vs. 4 out of 8 in adults showed some statistically significant associations between birth seasonality and CNS tumors or tumor subtype occurrence, pointing to a clustering of births mostly in fall and winter months, albeit no consistent pattern was identified by histological subtype. To further explore this question, primary incident CNS tumor cases (N=6014) were retrieved from the SEE cancer registries (1983-2015). Children born during winter were at slightly increased risk of CNS tumors overall and specifically of embryonal histology (IRR: 1.13, [1.01-1.27]). The winter peak of embryonal tumors was higher among boys (IRR: 1.24, [1.05-1.46]), and especially in the course of the first five years of life (IRR: 1.33[ 1.03-1.71]).

We explored the incidence patterns and survival rates of gliomatosis cerebri in a population-based registration sample from the SEER database (176 cases over the period 1973-2012). The annual age-adjusted incidence rate was estimated to 0.1/million. Gliomatosis cerebri was diagnosed in the entire age spectrum (range 1-98 years), but higher incidence (0.43/million) was noted among the elderly (≥65 years). A slight male preponderance was observed. Median overall survival was 9 months with a 5-year survival rate of 18%. Increasing age, primary tumor location not restricted to the cerebral hemispheres and rural residence at diagnosis were identified as negative prognostic factors. We further performed a systematic literature search for published case reports and case series on patients with histologically confirmed gliomatosis cerebri and extracted clinical, diagnostic, neuroimaging, histopathological, molecular, and survival data on individual patient level. A total of 274 studies were identified, including 1,648 patients (59% males, mean age 43.6 years). Seizures (50%) were the most common presenting symptom followed by headache (36%), cognitive decline (32%), and focal motor deficits (32%). There was bilateral hemisphere involvement in 65%, infratentorial infiltration in 30% and a focal contrast-enhanced mass (type II) in 31% of cases. Magnetic resonance imaging (MRI, extensive hyperintensities in T2/FLAIR sequences) and MR spectroscopy (elevated choline, creatinine, and myoinositol levels; decreased NAA levels) showed highly consistent diagnostic findings. Low-grade and anaplastic astrocytoma were the most prevalent diagnostic categories, but features of any histology (astrocytic, oligodendroglial, oligoastrocytic) and grade (II-IV) were reported. Among molecular aberrations, IDH mutation and MGMT promoter methylation were the most commonly reported. Median overall and progression-free survival were 13 and 10 months, with 5-year rates of 18% and 13%, respectively. Age ≥65 years at diagnosis, high-grade tumor, type II gliomatosis cerebri, more CNS regions involved, focal neurological deficits, cerebellar symptoms, higher burden of presenting symptoms, Karnofsky performance scale score <70, MRI contrast enhancement, symmetric bilateral CNS invasion, and high proliferation index (Ki67 >5) were independent predictors of poor outcome. Conversely, seizure occurrence, IDH mutation, and MGMT promoter methylation, were associated with prolonged survival. Chemotherapy and surgical resection were associated with improved outcome, whereas radiotherapy either as monotherapy or combined with chemotherapy was not superior to chemotherapy alone. Among 182 children with gliomatosis cerebri (0-18 years, 63% males), MGMT promoter methylation, IDH mutations, and codeletion of 1p/19q were less common molecular aberrations, as compared to adult gliomatosis cerebri, whereas age at diagnosis >4 years, extended CNS infiltration, coordination abnormalities, and cognitive decline were predictors of poor outcome in children. Exploring the association between seizure occurrence and improved survival, we found IDH mutations, a favorable prognostic marker, to be associated with a higher seizure occurrence at presentation, in accordance with other gliomas.

In conclusion, by exploiting national, European, and international population-based cancer registry data, in-house resources, data from published case-control and cohort studies, as well as individual-level data from case reports and case series, with this thesis we were able to address research questions related to all aspects of the epidemiology of primary CNS tumors. We provided the overview of the incidence and survival of malignant CNS tumors in the age group 15-39 years in Southern Eastern Europe and comparisons with the US, explored the epidemiology of pilocytic astrocytoma, the most common primary CNS tumor in childhood, evaluated the role of a series of perinatal and early-life risk factors in the etiology of childhood and adult primary CNS tumors, and finally documented the diagnostic and prognostic features of gliomatosis cerebri, an extremely rare fatal primary CNS tumor with to-date unknown etiology and features. 
Main subject category:
Health Sciences
Other subject categories:
Nervous system diseases
Keywords:
Central nervous system tumors, Gliomatosis cerebri, Neurology, Epidemiology, Meta-analysis
Index:
No
Number of index pages:
0
Contains images:
Yes
Number of references:
464
Number of pages:
390
Marios Georgakis PhD.pdf (12 MB) Open in new window