Management of primary thalamic low-grade glioma in pediatric patients: results of the multicenter treatment studies HIT-LGG 1996 and SIOP-LGG 2004

Abstract

Low-grade gliomas (LGG; WHO grade I and II) are the most common brain tumors in children and adolescents.¹ Excellent prognosis is seen for patients if total excision is possible, with more than 90% surviving at 10 years following treatment with surgery alone.^2,3 For patients with incomplete resection, different chemotherapy and radiotherapy approaches have been established.^4,5

Primary thalamic LGG are a small and heterogeneous group of tumors. They were previously grouped with tumors arising from other diencephalic regions,^6–9 rendering interpretation of results difficult. Most published studies combine adult and pediatric patients.^10–13 In addition, identification of risk factors is difficult due to lack of uniform histological confirmation, as well as variable use of chemotherapy or radiotherapy.

Outcomes may be poorer in thalamic glioma patients compared with other LGG entities, and bilateral thalamic LGG are less common than monothalamic LGG. Less than 50 cases of bilateral thalamic LGG have been reported in the literature since 1972.^{10,11,13–22} The role of surgery is often limited to biopsy and management of hydrocephalus,²³ and the efficacy of adjuvant therapeutic strategies is not known.

In this study, we present an analysis of 102 pediatric patients with primary thalamic LGG who were followed within the multicenter HIT-LGG 1996⁵ and SIOP-LGG 2004 studies. To our knowledge, this is the largest series prospectively monitoring a pediatric patient group with thalamic LGG. We aimed to identify prognostic factors and to clarify the behavior and response to treatment of this group of tumors.

Patients and Methods

Patients and treatment

Patients under the age of 18 years, with a primary thalamic LGG enrolled in either the HIT-LGG 1996⁵ or SIOP-LGG 2004 [ClinicalTrials.gov Identifier: NCT00276640] studies who were treated in Germany, Switzerland, or Austria from October 1, 1996 until March 31, 2012 were eligible. Patients were either diagnosed during this period or had their tumor diagnosed and observed prior to October 1, 1996, but did not have any nonsurgical treatment and entered the study's therapy arms only at the time of progression. The two studies were approved by institutional review boards.

At diagnosis, the best possible safe resection of the primary tumor was recommended. Patients with completely resected tumors were observed. Similarly, following incomplete resection, biopsy, or radiologic diagnosis, an observation-only strategy was adopted in patients without progression or severe symptoms. Adjuvant radiotherapy or chemotherapy (nonsurgical treatment) was administered in cases with severe or progressive symptoms or radiologic progression. A chemotherapy arm was instituted with the aim of deferring or even obviating radiotherapy in younger patients. Within the HIT-LGG 1996 study, initially only children less than 5 years of age were to receive chemotherapy as their first-line nonsurgical treatment. However, during the study period a growing number of older children were primarily treated with chemotherapy. The lower age limit for radiotherapy as the first nonsurgical treatment was increased to 8 years in the SIOP-LGG 2004 study. Children with neurofibromatosis type 1 received chemotherapy as a nonsurgical therapy irrespective of their age. Central neuroradiologic review was performed as described previously.²⁴

For this analysis, patients were grouped according to the type of primary nonsurgical management, if any, irrespective of the interval between diagnosis and the start of primary treatment, ie, into radiotherapy, chemotherapy, and observation groups.

The definition of the extent of resection considered subsequent surgical interventions removing additional tumor tissue if they occurred within 3 months of the primary surgery without progression or initiation of any nonsurgical therapy. Central histological review was performed in 71 (79%) of 90 surgical cases.

Chemotherapy in the HIT-LGG 1996 study consisted of a 10-week induction phase with weekly vincristine (1.5 mg/m², maximum 2 mg) on day 1 of weeks 1–10 and carboplatin (550 mg/m²) on day 1 of weeks 1, 4, 7, and 10. This was followed by a consolidation phase (weeks 13–53) with concomitant vincristine and carboplatin every 4 weeks. Standard chemotherapy used in the SIOP-LGG 2004 study consisted of an induction phase with vincristine (1.5 mg/m², maximum 2 mg) on day 1 of weeks 1–10, 13, 17, and 21, and carboplatin (550 mg/m²) on day 1 of weeks 1, 4, 7, 10, 13, 17, and 21, followed by a consolidation phase with 6-week cycles of carboplatin on day 1 and vincristine on day 1, 8, and 15 of each cycle from week 25 to week 81. In patients without neurofibromatosis type 1, this was compared in a randomized way to an intensified induction adding etoposide (100 mg/m²) on days 1–3 of weeks 1, 4, 7, and 10.

External beam focal radiation consisted of 54 Gy in fractions of 1.8 Gy with appropriate safety margins. In children younger than 5 years, radiation was limited to 40–45.2 Gy in fractions of 1.6 Gy in the HIT-LGG 1996 study and to 45 Gy in fractions of 1.8 Gy in the SIOP-LGG 2004 study. For patients with small, well-delineated tumors, treatment centers were free to choose brachytherapy.

Evaluation of response

An MRI was performed for each study patient at diagnosis and within 72 hours following surgery. Neuroradiological response assessment in the HIT-LGG 1996 study was performed at weeks 12, 24, 36, 48, and 53 following the start of chemotherapy and thereafter at gradually increasing intervals. In the SIOP-LGG 2004 study, neuroradiological response assessment was done at weeks 24, 54, and 85. Response assessment was similar in the radiotherapy groups. Follow-up exams after the end of nonsurgical treatment or after surgery in the observation group included biannual (year 1 to 5) and annual (year 6 to 10) clinical and MRI exams. In the first 3 years, clinical examinations were performed every 3 months. Assessment by MRI followed the recommended criteria.²⁵ Complete response was defined as no radiological evidence of residual tumor on contrast MRI and no tumor cells in the cerebrospinal fluid, in case of previously disseminated disease. Reduction of solid tumor volume of more than 50% was considered a partial response. Residual tumor manifestation between 50% and 25% with no new lesions or malignant cells in the cerebrospinal fluid was defined as an objective response. Stable disease was defined as a reduction in tumor volume of less than 25% to an increase in tumor volume of less than 25%. Tumor progression was defined as greater than 25% enlargement of solid tumor volume or appearance of new lesions. In both studies, complete response, partial response, objective response, and stable disease were considered positive responses.

Statistical considerations

Univariable distribution of metric variables is described by median and range. For categorical variables absolute and relative frequencies are stated. Chi-square and Fisher's exact tests were used to assess the association between 2 categorical variables (eg, extent of surgery and study). The Mann-Whitney U test was used to assess the association between a metric outcome and a binary predictor variable.

The distribution of overall survival (OS), event-free survival (EFS), and progression-free survival (PFS) was calculated according to the Kaplan–Meier method and was compared between independent groups using log-rank test.²⁶ Survival probabilities are given with standard error (mean ± SE). As defined in this report, OS was calculated from the date of radiologic diagnosis until patient's death from any cause, or the last follow-up for patients alive. EFS was calculated from the date of radiologic diagnosis until an event, defined as occurrence of progression, start of a nonsurgical therapy, death from any cause, or the last follow-up for patients without event. To evaluate the variables extent of resection and histology, or to evaluate survival in subgroups defined by extent of resection and/or histology, OS/EFS were defined from date of resection until death/event or until last follow-up if no event occurred. PFS following nonsurgical treatment (either chemotherapy or radiotherapy) was calculated from start of nonsurgical therapy until an event, defined as tumor progression or death from any cause, or until last follow-up if no event occurred. Univariable Cox regression analysis was used to assess the influence of continuous variables (age at diagnosis) on survival probabilities. Multivariable Cox regression analysis²⁷ with forward stepwise selection (inclusion criterion: score test P ≤ .05, exclusion criterion: likelihood ratio test P > .10) was used to assess the potential impact of multiple factors on overall survival. Potential impact factors were trial (categorical: HIT-LGG 1996 or SIOP-LGG 2004), age at radiological diagnosis (continuous and categorical: 0–4.99, 5.00–7.99, 8.00–10.99, or ≥11.00 years), sex (categorical: male or female), presence of neurofibromatosis type 1 (categorical: yes or no), localization (categorical: monothalamic or bithalamic), presence of disseminated disease at diagnosis (categorical: no or yes), histological subtype (categorical: pilocytic astrocytoma WHO grade I, diffuse astrocytoma WHO grade II, non-pilocytic/non-diffuse low-grade glioma, non-representative histology, or no histology [ie, radiological diagnosis only]), extent of resection (categorical: gross total, subtotal, partial resection, biopsy, or no resection; and also categorized by gross total/subtotal, partial resection/biopsy, or no resection). Analyses for PFS were performed in the radiotherapy and chemotherapy subgroups only, and additionally analyzed the time from diagnosis to start of therapy. Because tumor surgery was not always performed directly after diagnosis, the extent of resection and histology were included as time-dependent variables for OS and EFS in multivariable analysis. For all but 1 patient, tumor surgery occurred before start of nonsurgical treatment. For PFS, we thus analyzed the extent of resection and histology as non-time-dependent variables. One patient had his first tumor surgery after the event for PFS and was considered as being without resections and histology at the beginning of nonsurgical therapy. Interactions of main effects (including time-dependent variables) were tested in a second block. For Cox regression models, the P values of likelihood ratio test and hazard ratios with 95% confidence intervals are indicated. No statistical significance level was fixed, and P values were only considered exploratory. Analyses were performed using SPSS version 20 software (IBM Corporation, Chicago, IL).

Results

Patient and disease characteristics

From October 1, 1996 until March 31, 2012, a combined 2618 patients were registered in either the HIT-LGG 1996 or the SIOP-LGG 2004 study. Diagnosis of a primary thalamic LGG was found in 103 (3.9%) of these 2618 patients. One patient was excluded from this report due to high-grade histology (anaplastic astrocytoma) on retrospective central review of the initial tumor sample. The median follow-up time monitoring surviving patients was 7.6 years (range, 0.3–15.3 years).

Patient and disease characteristics are presented in Table 1. Monothalamic LGG was found in 87 patients (85%), while 15 patients (15%) had a bithalamic LGG. Forty-seven patients (46%) were observed only without any nonsurgical therapy during follow-up (Table 2). Tumor progression was seen in 11 of these patients (23%), all with monothalamic disease, with 5 of them receiving further surgical procedures. Five patients, without radiological progression, received 1 or 2 additional surgeries to remove tumor tissue, 3 of which were within 2 months of the primary procedure.

Radiotherapy was the first nonsurgical treatment in 31 patients using external beam radiotherapy (17), brachytherapy (13), or both (1). Nine patients (29%) eventually were reported to receive a second-line treatment. One patient received a second brachytherapy treatment and 8 patients received between 1 and 3 sequential chemotherapy regimens. One or two additional surgical procedures were performed in 14 patients between 0.7 and 107 months after their first surgery.

Chemotherapy was the first nonsurgical treatment in 24 patients (24%). Upon progression, 8 patients (33%) from this group received radiotherapy as a second-line therapy, and 3 of them subsequently received between 1 and 3 further lines of nonsurgical treatment. The median time from start of the first chemotherapy treatment until start of the first radiotherapy treatment was 1.3 years (range, 0.3–6.3 years), with a median age of 10.0 years (range, 7.0–16.7 years) at start of radiotherapy. Three patients (13%) were treated with an alternative chemotherapy regimen as second-line treatment. One or two additional surgical procedures were performed on 14 patients in this group between 0.5 and 102 months after first surgery.

Survival outcome and response to treatment

The OS probability at 10 years after radiological diagnosis was 88% (±3%), and OS at 5 years was 89% (±3%; Table 3; Fig. 1A). In patients with monothalamic LGG, 10-year OS was 91% (±3%), compared with 65% (±13%) in patients with bithalamic LGG (P = .001; Fig. 2A). Twelve patients died (12%), thereof 7 patients out of 87 with monothalamic LGG (8%) and 5 patients out of 15 with bithalamic (33%) LGG. Death was tumor-related in 11, while 1 patient died from accidental drowning, and occurred within 3 years of diagnosis in all but 1 patient. Patients with diffuse astrocytoma had a lower 10-year OS (68% ±10%) than patients with pilocytic astrocytoma (93% ±4%), and 10-year OS was intermediate (83% ±11%; P = .027)for non-diffuse/non-pilocytic histology, as shown in Supplementary Material, Fig. S1A. Ten-year OS was 100% following complete or subtotal resection compared with 84% (±4%) following partial resection or only a biopsy (P = .149; Supplementary Material, Fig. S2A). For patients with no surgery, 10-year OS measured from radiologic diagnosis was 100% as well. Overall survival probabilities were comparable in both studies, and neither age at diagnosis nor sex showed any prognostic significance. Among the subgroup of patients with either pilocytic or diffuse astrocytoma and only partial resection/biopsy, 10-year OS was found to be significantly different between monothalamic 89% (±5%) and bithalamic tumors 53% (±16%) (P = .001). For patients with diffuse astrocytoma and less than subtotal resection, a trend towards better survival for patients with monothalamic tumors could still be seen with 76% (±12%) OS as opposed to 53% (±18%) OS for bithalamic tumors (P = .161). In multivariable Cox regression analysis, only tumor site was retained as a negative prognostic factor, with a hazard ratio of 5.3 for bithalamic localization (1.7–16.9, P = .009).

Table 3.

Open in new tab

Ten-year overall (OS) and event-free survival (EFS) probabilities after radiological diagnosis in patients with thalamic glioma

.	Overall survivala .			Event-free survivala .
Characteristic .	No. .	10-year OS, % (±SE) .	P value (log-rank test) .	No. .	10-year EFSa, % (±SE) .	P value (log-rank test) .
All patients	102	88 (±3)		102	28 (±5)
Study			.917			.006
HIT-LGG 1996	43	88 (±5)		43	16 (±6)
SIOP-LGG 2004	59	86 (±5)		59	39 (±8)
Sex			.427			.276
Male	45	91 (±4)		45	31 (±9)
Female	57	85 (±5)		57	25 (±6)
Age at diagnosis			.718			.199
0.00 to 4.99 years	27	93 (±5)		27	19 (±8)
5.00 to 7.99 years	23	82 (±8)		23	18 (±9)
8.00 to 10.99 years	22	86 (±8)		22	33 (±11)
≥11.00 years	30	87 (±8)		30	46 (±9)
Localization			.001			.641
Monothalamic	87	91 (±3)		87	29 (±6)
Bithalamic	15	65 (±13)		15	16 (±13)
Histologyb			.027			.367
Pilocytic astrocytoma, WHO grade I	50	93 (±4)		45	22 (±7)
Diffuse astrocytoma, WHO grade II	23	68 (±10)		22	36 (±10)
LGG other than pilocytic or diffuse astrocytoma	12	83 (±11)		11	0 (±0)
Nonrepresentative histology	4	100		4	75 (±22)
No histologyc	1	censored		-	-
Disseminated disease at diagnosis			.132			.871
No	99	88 (±3)		99	28 (±5)
Yes	3	67 (±27)		3	33 (±27)
Extent of resectionb			.149			.028
Complete/subtotal resection	12	100		12	50 (±14)
Partial resection/biopsy	78	84 (±4)		70	20 (±6)

.	Overall survivala .			Event-free survivala .
Characteristic .	No. .	10-year OS, % (±SE) .	P value (log-rank test) .	No. .	10-year EFSa, % (±SE) .	P value (log-rank test) .
All patients	102	88 (±3)		102	28 (±5)
Study			.917			.006
HIT-LGG 1996	43	88 (±5)		43	16 (±6)
SIOP-LGG 2004	59	86 (±5)		59	39 (±8)
Sex			.427			.276
Male	45	91 (±4)		45	31 (±9)
Female	57	85 (±5)		57	25 (±6)
Age at diagnosis			.718			.199
0.00 to 4.99 years	27	93 (±5)		27	19 (±8)
5.00 to 7.99 years	23	82 (±8)		23	18 (±9)
8.00 to 10.99 years	22	86 (±8)		22	33 (±11)
≥11.00 years	30	87 (±8)		30	46 (±9)
Localization			.001			.641
Monothalamic	87	91 (±3)		87	29 (±6)
Bithalamic	15	65 (±13)		15	16 (±13)
Histologyb			.027			.367
Pilocytic astrocytoma, WHO grade I	50	93 (±4)		45	22 (±7)
Diffuse astrocytoma, WHO grade II	23	68 (±10)		22	36 (±10)
LGG other than pilocytic or diffuse astrocytoma	12	83 (±11)		11	0 (±0)
Nonrepresentative histology	4	100		4	75 (±22)
No histologyc	1	censored		-	-
Disseminated disease at diagnosis			.132			.871
No	99	88 (±3)		99	28 (±5)
Yes	3	67 (±27)		3	33 (±27)
Extent of resectionb			.149			.028
Complete/subtotal resection	12	100		12	50 (±14)
Partial resection/biopsy	78	84 (±4)		70	20 (±6)

LGG, low-grade glioma; SE, standard error.

^aEFS/OS calculated from date of radiologic diagnosis until event/death or last follow-up. Event is defined as relapse/progression, start of nonsurgical treatment, or death. For histology and extent of resection, EFS/OS is calculated from date of resection.

^bMissing values to add up to 102: 12 patients without surgery. Further 8 patients with event before/at the date of resection.

^cNo documented histology results (n = 1).

Table 3.

Open in new tab

Ten-year overall (OS) and event-free survival (EFS) probabilities after radiological diagnosis in patients with thalamic glioma

.	Overall survivala .			Event-free survivala .
Characteristic .	No. .	10-year OS, % (±SE) .	P value (log-rank test) .	No. .	10-year EFSa, % (±SE) .	P value (log-rank test) .
All patients	102	88 (±3)		102	28 (±5)
Study			.917			.006
HIT-LGG 1996	43	88 (±5)		43	16 (±6)
SIOP-LGG 2004	59	86 (±5)		59	39 (±8)
Sex			.427			.276
Male	45	91 (±4)		45	31 (±9)
Female	57	85 (±5)		57	25 (±6)
Age at diagnosis			.718			.199
0.00 to 4.99 years	27	93 (±5)		27	19 (±8)
5.00 to 7.99 years	23	82 (±8)		23	18 (±9)
8.00 to 10.99 years	22	86 (±8)		22	33 (±11)
≥11.00 years	30	87 (±8)		30	46 (±9)
Localization			.001			.641
Monothalamic	87	91 (±3)		87	29 (±6)
Bithalamic	15	65 (±13)		15	16 (±13)
Histologyb			.027			.367
Pilocytic astrocytoma, WHO grade I	50	93 (±4)		45	22 (±7)
Diffuse astrocytoma, WHO grade II	23	68 (±10)		22	36 (±10)
LGG other than pilocytic or diffuse astrocytoma	12	83 (±11)		11	0 (±0)
Nonrepresentative histology	4	100		4	75 (±22)
No histologyc	1	censored		-	-
Disseminated disease at diagnosis			.132			.871
No	99	88 (±3)		99	28 (±5)
Yes	3	67 (±27)		3	33 (±27)
Extent of resectionb			.149			.028
Complete/subtotal resection	12	100		12	50 (±14)
Partial resection/biopsy	78	84 (±4)		70	20 (±6)

.	Overall survivala .			Event-free survivala .
Characteristic .	No. .	10-year OS, % (±SE) .	P value (log-rank test) .	No. .	10-year EFSa, % (±SE) .	P value (log-rank test) .
All patients	102	88 (±3)		102	28 (±5)
Study			.917			.006
HIT-LGG 1996	43	88 (±5)		43	16 (±6)
SIOP-LGG 2004	59	86 (±5)		59	39 (±8)
Sex			.427			.276
Male	45	91 (±4)		45	31 (±9)
Female	57	85 (±5)		57	25 (±6)
Age at diagnosis			.718			.199
0.00 to 4.99 years	27	93 (±5)		27	19 (±8)
5.00 to 7.99 years	23	82 (±8)		23	18 (±9)
8.00 to 10.99 years	22	86 (±8)		22	33 (±11)
≥11.00 years	30	87 (±8)		30	46 (±9)
Localization			.001			.641
Monothalamic	87	91 (±3)		87	29 (±6)
Bithalamic	15	65 (±13)		15	16 (±13)
Histologyb			.027			.367
Pilocytic astrocytoma, WHO grade I	50	93 (±4)		45	22 (±7)
Diffuse astrocytoma, WHO grade II	23	68 (±10)		22	36 (±10)
LGG other than pilocytic or diffuse astrocytoma	12	83 (±11)		11	0 (±0)
Nonrepresentative histology	4	100		4	75 (±22)
No histologyc	1	censored		-	-
Disseminated disease at diagnosis			.132			.871
No	99	88 (±3)		99	28 (±5)
Yes	3	67 (±27)		3	33 (±27)
Extent of resectionb			.149			.028
Complete/subtotal resection	12	100		12	50 (±14)
Partial resection/biopsy	78	84 (±4)		70	20 (±6)

LGG, low-grade glioma; SE, standard error.

^aEFS/OS calculated from date of radiologic diagnosis until event/death or last follow-up. Event is defined as relapse/progression, start of nonsurgical treatment, or death. For histology and extent of resection, EFS/OS is calculated from date of resection.

^bMissing values to add up to 102: 12 patients without surgery. Further 8 patients with event before/at the date of resection.

^cNo documented histology results (n = 1).

Fig. 1.

A) Overall survival (OS) and B) event-free survival (EFS) after diagnosis in 102 patients with thalamic glioma.

Open in new tabDownload slide

Fig. 2.

A) Overall survival (OS) and B) event-free survival (EFS) after diagnosis according to tumor location.

Open in new tabDownload slide

Contrasting the high overall survival probability, 10-year EFS probability for the cohort was low at only 28% (±5%), and 5-year EFS was 34% (±5%; Table 3 and Fig. 1B). Sixty-nine patients (68%) had an event as late as 7 years after diagnosis. Patients with monothalamic tumors had a 10-year EFS of 29% (±6%) and those with bithalamic tumors of 16% (±13%; P = .641; Fig. 2B). Pilocytic astrocytoma was associated with a 10-year EFS of 22% (±7%), and diffuse astrocytoma with a 10-year EFS of 36% (±10%; P = .367; Supplementary Material, Fig. S1B). The 10-year EFS was 50% (±14%) after complete or subtotal resection and 20% (±6%) after partial resection or biopsy (P < .028; Supplementary Material, Fig. S2B). For patients without any tumor surgery, 10-year EFS measured from radiologic diagnosis was 74% (±13%).

Progression-free survival after start of first nonsurgical treatment

First-line nonsurgical therapy was able to control tumor growth only in a portion of the patients, leading to a 10-year PFS of 53% (±10%) in the radiotherapy group and 34% (±10%) in the chemotherapy group (Table 4 and Supplementary Material, Fig. S3). No progression from low-grade to high-grade histology was observed.

In the radiotherapy group, no prognostic impact on PFS could be demonstrated for study enrollment, sex, age at diagnosis, tumor location (only 3 patients had a bithalamic tumors in this group), extent of resection (only 1 patient underwent total/subtotal resection and only 1 patient was treated without tumor surgery), histology, or age at the start of radiotherapy.

In the chemotherapy group, being an older patient at the time of diagnosis was an unfavorable prognostic factor for PFS with a hazard ratio of 1.166 (1.010–1.348; P = .039) for a patient x years at diagnosis compared with a patient one year younger. Male patients had a higher 10-year PFS probability with 61% (±18%) compared to female patients with 20% (±10%) (P = .027). No prognostic impact was found for study enrollment, tumor location, extent of resection (however, only 2 patients underwent total/subtotal resection), histology, or age at start of chemotherapy.

Best tumor response after first nonsurgical therapy

Complete response as best response to the first nonsurgical therapy was seen for 4 patients (13%) in the radiotherapy group, while 10 patients (32%) achieved a partial response, and 2 patients (6%) exhibited an objective response. Stable disease after radiotherapy as the first nonsurgical treatment was seen in 13 patients (42%), and 2 patients (6%) showed progressive disease (PD).

In the chemotherapy group, the response to the first nonsurgical treatment was complete response, partial response, stable disease, and PD in 6 (25%) patients each.

Discussion

In this study, we describe the largest published cohort of primary thalamic LGG in pediatric patients, including 87 patients with monothalamic and 15 patients with bithalamic LGG. We found 5-year and 10-year OS rates of 89% and 88%, respectively. These were comparable with 5-year OS of 95% in the entire population (ie, irrespective of tumor location) of the HIT-LGG 1996 study,⁵ and also to the 5-year OS of 92% for 39 patients with supratentorial midline (other than chiasmatic/hypothalamic) LGG, 29 of which were thalamic, who were treated within the large population-based CCLG CNS9702 study.²⁸ Comparison with other series is hampered by differences in patient populations (eg, selection of observation-only patients, chemotherapy-only patients), small patient numbers, retrospective study design, lack of details on tumor location, and the combination of low-grade and high-grade tumors.^29–34

In contrast to the high OS rate, 10-year EFS was only 29%. In this study, tumor progression and start of nonsurgical therapy (eg, due to severe or progressive neurological symptoms) were counted as events as defined by Janss et al.³⁵ Inherent to the use of different definitions of events, our EFS figures are difficult to compare to those of other series. However, the finding that LGG is a disease with high OS and significantly lower EFS rates for several subgroups, is well known from numerous other publications.³⁶

Tumor location had the highest impact on OS rates. Patients with monothalamic tumors showed a 10-year OS rate of 91%, whereas patients with bithalamic tumors only reached 65%. This was also seen to a lesser degree with 10-year EFS (29% vs 16%).

As in other pediatric LGG series,^28,29,37 the histological diagnosis of diffuse astrocytoma was associated with an inferior 10-year OS compared with pilocytic astrocytoma (68% vs 93%). Whether the finding that 10-year EFS was lower in pilocytic astrocytoma (22%) compared with diffuse astrocytoma (36%) represents just a random finding due to limited patient numbers or whether it may be a consequence of a lower threshold for starting postsurgical therapy, which was counted as an event, cannot be determined.

The importance of the extent of resection in LGG has been previously shown by numerous authors.^{20,28,33,34,38,39} We confirm those findings with the results of our series. Patients had a 10-year OS/EFS of 100%/50% following complete or subtotal resection as opposed to 84%/20% after partial resection or biopsy only. Interestingly, patients with only a radiological diagnosis of a thalamic LGG had a 10-year OS/EFS of 100%/74%. These patients probably represent a positive selection with an absence or paucity of neurological symptoms and a more benign radiological pattern based on tumor size or degree of infiltration, all of which possibly correspond with less aggressive disease.

In the current study, tumor location and histology were correlated with a majority of monothalamic tumors showing pilocytic histology, while most patients with bithalamic LGG had diffuse astrocytoma. This is compatible with the findings of Menon et al¹³ in a series of 9 patients with bilateral thalamic lesions, where fibrillary astrocytoma was diagnosed in all 6 biopsied tumors. Also, in the pediatric series of Di Rocco and Iannelli, 3 of 4 bilateral thalamic tumors had diffuse grade II astrocytoma and 1 had grade III astrocytoma.¹⁷ We hypothesize that the inferior long-term prognosis of bithalamic tumors is predominantly determined by histology. The higher rate of diffuse astrocytoma precludes total or subtotal resection, a relation also shown for adults with hemispheric grade I to III glioma and for patients with spinal astrocytoma.^40,41 In the small subgroup of 22 diffuse astrocytoma patients who underwent partial resection or biopsy only, there is an additional, though non-significant, trend for a inferior prognosis for bithalamic compared with monothalamic tumors (10-year OS was 53% vs 76% [P = .161] and 10-year was EFS 22% vs 42% [P = .428]). Thus, respecting the limited significance due to small patient numbers, biology of bithalamic tumors may differ from monothalamic LGG for factors not yet detected. This interpretation is supported by the result of the multivariable Cox regression analysis, yielding bithalamic location as the strongest adverse prognostic factor for OS.

While OS rates were comparable between the SIOP-LGG 2004 and the HIT-LGG 1996 studies, EFS was lower in the SIOP-LGG 2004 study. However, recruitment of older patients was scarce in HIT-LGG 1996, and the indication to start a nonsurgical therapy, which was considered an event in the EFS analysis, was less strictly defined in this study as opposed to SIOP-LGG 2004, rendering a valid comparison difficult.

Within our cohort of thalamic LGG a majority had an indication to start nonsurgical treatment, half of them within 3 months after radiological diagnosis, whereas less than half of the patients remained observed. Thirty percent of patients received primary radiotherapy as opposed to primary chemotherapy (24%). In the radiotherapy group, the 10-year PFS after the start of radiotherapy was 53%, and no prognostic factors affecting progression were identified. Nine patients were reported to receive second-line treatments with a second brachytherapy in 1, and 1 to 3 chemotherapy regimens in the other 8 patients. This is in line with a French study describing 85 patients with progressive optic pathway glioma with more than one subsequent line of chemotherapy being able to postpone or obviate radiotherapy in a number of patients.⁴²

In the chemotherapy group, 10-year PFS was 34%, with an at least transient tumor volume reduction or stable disease achieved in the majority of patients (75%), which is in line with the result for the chemotherapy arm of the HIT-LGG 1996 study and with other series.^5,28,42 Eight patients received radiotherapy as a salvage treatment (3 of them were additionally treated with further chemotherapy), with a median time from start of chemotherapy to subsequent radiotherapy of 1.3 years (range, 0.3–6.3 years). In this small group of children, chemotherapy had fulfilled its role of deferring or even obviating radiotherapy, thus hopefully reducing cognitive sequelae as well as the risk of radiotherapy-induced endocrinopathy, vasculopathy, and secondary malignancies.^43–45

In univariate analyses in the chemotherapy group, the strongest prognostic factor was age, with younger patients having a higher PFS than older patients (hazard ratio 1.166). This mirrors the results of the HIT-LGG 1996 study. We could, however, not assess the impact of age below 1 year with only 1 infant in the chemotherapy treatment group. Sex seemed another prognostic factor, with a 10-year PFS of 61% in male patients and 20% in female patients. We suspect this to be a statistical aberration associated with small patient numbers, and we are not aware of a biological rational or of other large series of LGG patients confirming this finding.^5,29,33,34

This study has several strengths, foremost in that it is the largest series describing primary thalamic LGG in children and adolescents. Due to its prospective design, patients received a uniform diagnostic work-up (including a central review of radiology and neuropathology in most cases), treatment, and follow-up recommendations. A primary limitation of this study is the lack of clinical data including neurological and cognitive outcomes. It would be interesting to correlate clinical outcome with the surgical results, which would allow assessing the cost (if any) of total or near-total resection. A further limitation is the relatively small size of subgroups, rendering difficult a more powerful analysis such as the identification of risk factors for adverse outcome. Another shortcoming is the lack of molecular data within this series, mainly collected prior to the detection of the characteristic molecular aberrations of pediatric LGG. We expect that improved disease characterization will lead to a better understanding of clinical behavior, such as the difference in prognosis between monothalamic and bithalamic tumors, and of potential therapeutic targets.³⁶ Nevertheless, we believe our series significantly contributes to the understanding of thalamic LGG in the pediatric population.

In conclusion, in this largest yet published series of pediatric thalamic LGG, we have shown a high OS rate for monothalamic and an unsatisfactory survival for bithalamic disease. The worse prognosis for bithalamic tumors seems related to the higher portion of diffuse astrocytoma precluding a more extensive resection, while in monothalamic disease total or near-total resection was possible in a substantial number of patients. Nevertheless, additional biologic effects associated with bithalamic localization and negatively affecting prognosis cannot be ruled out.

Among those patients who received nonsurgical treatment, tumor control was achievable in half of the patients treated with radiotherapy and in one-third of the patients receiving chemotherapy as their first-line nonsurgical treatment. The use of chemotherapy allowed for postponement or omission of radiotherapy in a number of patients. Many patients received more than 1 line of nonsurgical therapy. Following progression after first-line treatment, long-term survival is possible with subsequent treatment.

The survival results of thalamic tumors were comparable with the results for the whole population of the HIT-LGG 1996 study .⁵ Thus, it appears that there is no need of separate treatment protocols for thalamic glioma. In future studies, the inclusion of neurological, cognitive, and neuropsychological data will significantly add to the validity of results. Furthermore, the incorporation of molecular markers will most likely contribute to a better understanding of the heterogeneous course of disease and allow for a delineation of the role of targeted therapy in primary thalamic pediatric LGG.

Funding

The German SIOP-LGG 2004 trial center in Augsburg, Germany, the Reference Center for Neuropathology, Institute of Neuropatholgy, University of Bonn, Germany, and the Reference Center for Biostatistics, Institute of Biostatistics and Clinical Research, University of Muenster, Germany, were supported by grants from the German Children′s Cancer Foundation (DKKS). Amedeo A. Azizi received funding from the Dachverband der Österreichischen Kinderkrebshilfe for data management of the SIOP LGG 2004 trial in Austria. He furthermore received support by the Gesellschaft für cerebrale Tumore.

Conflict of interest statement

None declared.

References

Author notes