Natural history and = management of low-grade glioma in NF-1 children

Clinical Study - Patient Study

Natural history and = management of=20 low-grade glioma in NF-1 children

Pablo Hern=E1iz Driever¹,=20 Stephan von Hornstein², = Torsten Pietsch³,=20 Rolf Kortmann⁴, Monika Warmuth-Metz⁵,=20 Angela Emser⁶ and = Astrid K. Gnekow²

(1)

Department of Pediatric Oncology and=20 Hematology, Charit=E9-Universit=E4tsmedizin Berlin, Augustenburger = Platz 1,=20 13353 Berlin, Germany

(2)

Hospital for Children and Adolescents, = I.=20 Klinik f=FCr Kinder und Jugendliche, Klinikum Augsburg, = Stenglinstrasse 2,=20 86156 Augsburg, Germany

(3)

Institute of Neuropathology, = University of=20 Bonn, Sigmund-Freud-Strasse 25, 53105 Bonn,=20 Germany

(4)

Department of Radiotherapy, University = of=20 Leipzig, Stephanstrasse 9a, 04103 Leipzig,=20 Germany

(5)

Department of Radiology, University of = W=FCrzburg, Josef-Schneider-Strasse 11, 97080 W=FCrzburg,=20 Germany

(6)

IMBEI, University of Mainz, Obere = Zahlbacher=20 Stra=DFe 69, 55131 Mainz, = Germany

Astrid K. Gnekow<= /STRONG>
Email:=20 gnekow.hit-lgg@klinik= um-augsburg.de

Received:=20 15 June 2009 Accepted:=20 15 March 2010 Published online:=20 30 March 2010

Abstract

Pediatric neurofibromatosis type 1 (NF-1) patients = are prone=20 to developing low-grade glioma (LGG). The HIT-LGG study 1996 aimed to = observe=20 the natural history of pediatric LGG and to postpone irradiation in = younger=20 children by using carboplatinum and vincristine in case non-surgical = treatment=20 was required. A total of 109 of 1,044 (10.4%) protocol patients had a = genetic=20 NF-1 trait [57 female patients; median age 5.1 years (range=20 1=9615.4 years)]. Eighty-three patients (76%) suffered from an = optic pathway=20 tumor. Neuroimaging only allowed diagnosis in 67 patients. Histology = revealed=20 pilocytic astrocytoma WHO grade I in 38 of 42 biopsied patients. = Sixty-five=20 (60%) patients received non-surgical treatment, either chemotherapy=20 (n =3D 55) or irradiation (n =3D 10). = The overall=20 survival rate of 96% after a median observation time of 5.25 years=20 contrasts with an event free survival rate (EFS) of 0.24 at = 5 years.=20 Progressive LGG were observed even in children older than 11 years. = Chiasmatic/postchiasmatic localization was a univariate risk factor for=20 progressive disease. In the chemotherapy group, we observed a 5-year=20 progression-free survival (PFS) rate of 0.73. Similarly, the PFS rate in = the=20 irradiation group was 0.78. Multivariate analysis revealed surgical = intervention=20 and localization within the optic pathway as factors that increased the = risk of=20 tumor progression. In this large prospective multinational study, LGG in = NF-1=20 patients did progress in 75% of patients. Chemotherapy yielded = acceptable PFS.=20 The biological factors determining progression remain poorly understood. =

Keywords Low grade=20 glioma - Chemotherapy - Radiotherapy - NF-1= - Children=20

This study was conducted = on behalf of=20 the HIT-LGG study group for Children and Adolescents with a Low Grade=20 Glioma.

P. Hern=E1iz Driever and S. von Hornstein = contributed equally to=20 this work.

Introduction

Up to 15% of all neurofibromatosis type 1 (NF-1) = patients will=20 develop a cerebral tumor within the first two decades [1=967],=20 the majority of which are low-grade glioma (LGG) with pilocytic = astrocytoma=20 histology. They are predominantly located along the optic pathway = [2=966].=20 The management challenges [8]=20 are: (1) the natural history of these tumors is not fully understood, = (2)=20 impairment of visual acuity in optic pathway lesions affects only a = portion of=20 NF-1 patients, (3) NF-1 patients may develop serious complications after = irradiation, (4) the diagnosis of a brain tumor may lead to the = diagnosis of=20 NF-1, whereas in other patients the lesion may be diagnosed through = screening,=20 and (5) NF-1 patients often suffer from concurrent disease-related = problems like=20 cognitive impairment, seizures, and the disposition to develop = additional=20 neoplasms.

Several studies focused on the biological differences = between=20 histologically identical tumor lesions in children with and without the = NF-1=20 trait [9=9612].=20 In a systematic review of prognostic factors for progression of = childhood optic=20 pathway glioma, three clinical reports demonstrated an improved = progression free=20 survival (PFS) for NF-1 patients using a multivariate analysis, whereas = in six=20 other studies NF-1 trait was not a significant prognostic factor = [12].=20 Current clinical recommendations remain unclear as to whether the = presence of=20 NF-1 itself should be considered for stratification of therapy if = treatment is=20 necessary [8].=20 In addition, several medical subspecialties care for pediatric NF-1 = patients,=20 each with their own experience and perspective.

The primary aim of the prospective HIT-LGG-1996 study = was to=20 establish a common therapeutic strategy for pediatric LGG patients, = including=20 NF-1 patients with a symptomatic LGG, in the countries of the = German-speaking=20 Society of Pediatric Oncology and Hematology (GPOH). Hence, by reporting = the=20 results on NF-1 patients of the HIT-LGG study, we inform on the largest=20 comprehensively treated cohort so far, and seek to answer questions on = the=20 factors determining clinical biology of NF-1 associated LGG.

Patients and methods

Eligibility

The HIT-LGG 1996 study was designed as a comprehensive = treatment=20 strategy for all participating centers in the German speaking countries = of the=20 Society of Pediatric Oncology and Hematology (GPOH). The eligibility = criteria=20 included children and adolescents younger than 17 years of age = suffering=20 from a LGG. The radiological diagnosis of a LGG was accepted for NF-1 = patients=20 either with optic pathway gliomas or isolated hypothalamic chiasmatic = tumors=20 that demonstrated hypo/isodensity on non-enhanced computed tomography = after=20 central review by one of us (M.W.-M.). Patients were recruited between = October=20 1st, 1996 and March 31st, 2004, and followed till March 2008. = Patients=92 parents=20 or guardians gave their informed consent. Central (Augsburg) as well as=20 participating centers ethics board approval was obtained.

Study strategy

After initial complete surgical resection a =93wait = and see=94=20 strategy was adopted. Adjuvant therapy was recommended in case of severe = neurologic or ophthalmologic symptoms at diagnosis or, if surgery was = not an=20 option, upon clinical, ophthalmologic or radiological progression = following a=20 period of observation (Figs. 1=20 and 2).=20 Adjuvant therapy consisted of either radiotherapy or chemotherapy with = an age=20 limit for stratification at 5 years. At the beginning, only = children=20 younger than 5 years were planned to receive chemotherapy. As the = study=20 continued, an increasing number of older patients received chemotherapy. =

Fig. 1 Study group=20 flow chart

Fig. 2 Observation=20 group flow chart. NF-1 patients received neither chemotherapy nor = radiotherapy=20 within the study period, i.e. observation group; DOD death of = disease,=20 SD stable disease, PD progressive disease, CR = complete=20 remission

Adjuvant treatment

Non-surgical therapy was administered as described in = detail=20 elsewhere [13,=20 14].=20 The chemotherapy regimen consisted of vincristine (1.5 mg/sqm = intravenous=20 bolus) and single dose carboplatin (550 mg/sqm as a 1-h infusion = with=20 appropriate hydration). During the 10 weeks induction, vincristine = was=20 given weekly and carboplatin on day 1 of weeks 1, 4, 7, and 10. Both = drugs were=20 given concomitantly at 4-week intervals during weeks 13 to 53. = Radiotherapy was=20 given as local radiation to the primary tumor region at a total dose of=20 54 Gy. Children younger than 5 years received a reduced dose = of=20 45.2 Gy. No age restriction was given for brachytherapy [15].=20

Methods

Biostatistics

All patients having a genetic NF-1 trait meeting the=20 NIH-criteria for a NF-1 diagnosis were included. The functions of = survival (OS),=20 event-free survival (EFS) and progression-free survival (PFS) were = calculated by=20 the method of Kaplan and Meier. OS was calculated from the date of = diagnosis=20 until the death of the patient regardless of its cause. EFS was = determined from=20 the date of diagnosis until relapse, progression, death of any cause, = and the=20 necessity to start non-surgical treatment [16].=20 PFS was estimated from the initiation of non-surgical treatment to = progression=20 of a residual tumor, relapse following previous complete remission, = evidence of=20 new or progression of existing metastases or death of any cause. We=20 differentiated between subgroups in response to EFS, PFS and OS using = the=20 log-rank test.

The potential influence of resection (as time = dependent=20 variable), age at diagnosis (continuous and age groups), sex, type of = NF-1 trait=20 inheritance, histology [pilocytic astrocytoma (PA) vs clinical diagnosis = vs=20 other] and tumor site (optic pathway glioma vs other supratentorial = midline=20 localization vs other) on EFS was analyzed by multivariate Cox = regression=20 analysis. The final model was built by forward stepwise variable = selection.=20 Estimated hazard ratios of variables of the final model were given.

All analyses were explorative. Therefore no = significance level=20 was fixed. The P values were given for descriptive reasons. =

Response assessment

Assessment of tumor response using neuroimaging was = recommended=20 at certain time intervals, MRI being the preferred mode of investigation = as=20 described in detail elsewhere [13,=20 14].=20 The radiological assessment of tumor response by contrast enhanced MRI = followed=20 recommended criteria [17].=20 Complete, partial, and objective response as well as stable disease were = considered positive responses within the protocol [13,=20 14].=20 Objective response is considered >25% but <50% change of tumor = size=20 radiographically.

Results

Patients

Ninety pediatric treatment centers in Germany, = Switzerland,=20 Austria, and Belgium adhered to the treatment strategy of the HIT-LGG = 1996=20 study. A total of 109 LGG patients with clinically diagnosed NF-1 were=20 registered (10.4% of protocol patients, n =3D 1,044) = and followed=20 prospectively for a median of 5.3 years (1.9=9612.2 years)=20 (Table 1,=20 Fig. 1).=20 The slight female predominance in the cohort was even more prominent in = the=20 adjuvant therapy subgroups. None of the patients enrolled were younger = than=20 1 year of age. Fifty patients were younger than 5 years and 25 = patients (23%) were older than 11 years. Histology was centrally = reviewed=20 by one of us (T.P.). Only one patient had a pleomorphic = xanthoastrocytoma WHO II=20 in the cerebral hemisphere. Sixty-seven patients were diagnosed by = neuroimaging=20 only. Three tumors with astrocytoma WHO II were found in the cerebellum=20 (n =3D 1) and the optic chiasm = (n =3D 2).=20 Seventy-six percent of patients (n =3D 83) had LGG = affecting the=20 optic pathway, the majority comprising Dodge III tumors=20 (n =3D 45) [18].=20 Ninety-three patients had supratentorial midline tumors. None of the = NF-1=20 patients had diencephalic syndrome and neither disseminated nor spinal = disease=20 was reported. Brainstem localization was found in 3 patients. Twelve = patients=20 (11%) developed more than a single primary LGG in the brain. The true = incidence=20 was not assessed systematically. Concerning NF-1 diagnostic NIH = criteria, the=20 patients did not differ from previous characterizations.

Table 1 Patients and=20 treatments included in study

Strategy group	NF-1 cohort	Observation	Chemotherapy	Radiotherapy
Number of patients^a	109	44	55	10
Sex (female:male)	57:52	19:25	32:23	6:4
Age at diagnosis	5.2 y (1.0=9619.8)	7.1 y (1.6=9619.8)	3.7 y (1.7=9612.9)	10.8 y (1.0=9614.7)
Histology
PA WHO I	38	17	13	8
A WHO II	3	0	3	0
PXA WHO II	1	1	0	0
Clinical + imaging diagnosis = only	67	26	39	2
Localization
Cerebral hemispheres	7	6	0	1
SML	93	32	53	7
Optic nerve	16	8	8	0
Chiasm	21	10	9	2
Dodge III	45	8	33	4
Hypothalamus	1	1	0	0
3rd ventricle	1	1	0	0
Thalamus	1	0	1	0
Basal ganglia	3	2	1	0
Tectum	1	0	1	0
Mesencephalon	3	2	0	1
Cerebellum	5	3	1	1
Brainstem	4	2	1	1
Unknown	1	1	0	0
First resection
Complete^b	6	5	0	1
Subtotal	6	4	1	1
Partial	15	9	6	0
Biopsy^c	15 (10)	0	9 (5)	6 (5)
No histology	67	26	39	2
Age at start of therapy			5.2 y (1.7=9614.9)	1.2 y (1.1=9614.8)
Overall interval from diagnosis to = therapy			2.6 m (0.0=9661.3)	3.2 m (0.8-44.4)
<3 months			n =3D 28	n =3D 5
1.0 m (0.0-2.6)	1.1 m (0.8=962.6)
>3 months			n =3D 27	n =3D 5
17.4 m (3.1=9661.3)	32.0 m (3.7-43.8)
Interval until best response			3.4 m (0.8=9624.4)	5.0 m (2.5=9610.3)
Best response
CR			0	1
PR/OR			24	2
SD			29	5
PD			1	1
Unknown			1	1
PD in observation		9
Followed by no surgery		3
Followed by surgery^d		4
Unknown		2
Progression after adjuvant therapy			17	3
Interval until progression after first therapy =			1.7 y (1.6 m =96 7.4 y)	7.4 m/31.1 m/46.1 m
Current status
CR	7	7	0	0
PR/OR	3	0	2	1
SD	90	34	48	8
PD	5	0	4	1
DOD	4	3	1 (OS 4.4 years)	0
Follow-up period	5.3 y (1.9 m =96 12.2 y)	4.6 y (1.9 m =96 9.7 y)	5.7 y (8.8 m =96 12.2 y)	5.2 y (2.3=967.4)

PA Pilocytic astrocytoma, A = astrocytoma,=20 PXA pleomorphic xanthoastrocytoma, CR complete remission,=20 PR partial remission, OR objective response, SD = stable=20 disease, PD progressive disease, DOD dead of disease, = OS=20 Overall survival, y years, m months

^aIn the radiotherapy group, 7 patients = received=20 conventional percutaneous radiotherapy, 3 patients received = brachytherapy=20

^bIn the radiotherapy group, 1 patient had = a=20 complete resection after having had subtotal resection, progressive = disease=20 treated with irradiation and response to the adjuvant therapy =

^cNumbers in parentheses indicate = stereotactic=20 biopsy

^dThree patients had PD before primary = surgery, 1=20 patient was submitted to a second surgery after PD=20

Surgery

Due to the clinical diagnosis of NF-1, 67 patients = (61%) who met=20 the neuroradiological diagnosis did not undergo surgery (Figs. 1,=20 and 2).=20 The tumors of 27 patients (25%), including 5 of 6 patients with = hemispheric=20 tumors, were amenable to surgery. Biopsy only was performed in 15 (14%)=20 patients.

Strategy groups

Eighteen patients remained observed following = surgery, as were=20 26 patients with a neuroradiological diagnosis of LGG (Fig. 2).=20 Nine patients had surgery within 3 months of presentation. In = another 6=20 patients, the decision to subject patients to surgery was taken after=20 3 months of diagnosis. Three patients underwent surgery after = developing=20 progressive disease (Fig. 2).=20 At occurrence of tumor progression or clinical deterioration, 65 = patients (60%)=20 were treated non-surgically (Fig. 1).=20 Seven of the ten irradiated patients received conventional radiotherapy = and 3=20 brachytherapy. Following protocol=92s recommendations, the median age of = patients=20 in the chemotherapy group was expectedly lower than in the radiotherapy = group=20 with 5.2 years and 12.2 years, respectively. In both adjuvant=20 treatment groups, half of all patients started therapy within = 3 months of=20 LGG diagnosis. The 5-year PFS of the adjuvant treatment group was 0.72 = and 0.56=20 at 9.95 years (Fig. 3).=20

Fig. 3 Progression-free-survival for = NF-1 patients=20 with primary chemotherapy (n =3D 55)

Overall and event-free survival

Four patients of the NF-1 cohort died during = follow-up,=20 resulting in an OS of 0.96 at 12 years (Fig. 4).=20 Three of the patients who died of the disease were in the observation = group=20 (0.16, 2.41 and 3.67 years after diagnosis). All had optic pathway = tumors=20 involving either the chiasm alone (n =3D 1) or as a = Dodge III=20 tumor (n =3D 2). They died from an Addison crisis=20 (n =3D 1) and cardiovascular failure = (n =3D 2),=20 i.e. disease-related complications. One patient in the chemotherapy = group died=20 4.34 years after the initial diagnosis of a LGG in the 3rd = ventricle due to=20 progression of disease. Kaplan=96Meier analysis revealed that the OS of = the cohort=20 was not influenced by gender, age (when comparing younger and older = patients,=20 </> 10 years of age), mode of NF-1 trait inheritance = (familiar vs=20 sporadic), histology (pilocytic astrocytomas vs no histology or others), = and=20 localization (optic pathway vs other sites). Seventy-seven events were = recorded,=20 resulting in an EFS of 0.24 at 5 years after median observation = time of=20 5.25 years that was followed up to 9.66 years (Fig. 3).=20

Fig. 4 Event-free-survival and = overall-survival=20 for the whole NF-1 cohort (n =3D 109)

Observation group (n =3D 44) =

After a median follow up time of 4.6 years, 37% = of all=20 patients were successfully treated without adjuvant treatment (CR: 7 and = SD: 34=20 patients). In this group, 9 patients had either a relapse or progressive = disease=20 and were treated with surgery in 5 cases during the observation time. = All=20 eventually experienced a stable disease state (Fig. 2).=20 However, 3 patients died from disease-related complications and not from = disease=20 progression.

Chemotherapy group (n =3D 55) =

Almost all patients (98%) responded to this first = therapy except=20 for two patients (1 PD and 1 not evaluated). Best response to = chemotherapy was=20 recorded after a median treatment time of 3.6 months with the = latest=20 response after 48.3 months. Nine patients had their best response = after the=20 end of chemotherapy. Of 24 patients who received a scan at 24 weeks = after=20 start of chemotherapy, PR/OR and SD were noted in 3 and 21 patients,=20 respectively. Therapy was well tolerated. However, 20 patients developed = an=20 allergy to carboplatin. Substantial vincristine toxicities were not = reported.=20 Sixteen patients progressed after a median time of 3.8 months. = Seven of=20 these patients were treated again with chemotherapy. One of these 7 = patients was=20 eventually treated with radiotherapy 35.2 months after start of = adjuvant=20 chemotherapy. Another 4 patients with PD received radiotherapy as second = line=20 treatment. The primary aim of postponing radiotherapy in this group was = achieved=20 for 3 patients, postponing it by 4, 6.3, and 13.1 months after the = start of=20 chemotherapy. This corresponds to a median delay of radiotherapy after = start of=20 chemotherapy of 9.7 months for the whole chemotherapy group. One = patient of=20 this cohort eventually died of the tumor and 4 patients experienced = progressive=20 disease after a median follow-up of 5.7 years. The 5 years, = PFS rate=20 of the chemotherapy group was 0.73, and 0.57 after 9.95 years. =

Radiotherapy group (n =3D 10) =

Best response to treatment was observed after = 5.1 months in=20 8 of 10 patients. Progressive tumors were observed in 2 patients = following=20 conventional external radiotherapy to the chiasm and the cerebellum, and = in 1=20 patient following brachytherapy of a Dodge III optic pathway tumor. One = of the=20 patients with PD was then treated with chemotherapy and the other two = were=20 observed (one of them had a CR following surgery). The 5-year PFS rate = of the=20 radiotherapy group is 0.78, and 0.59 after 7.36 years. None of the = patients=20 developed any unwanted irradiation-associated acute or late effects. =

Prognostic factors

Univariate analysis of the EFS for the whole NF-1 = cohort by=20 log-rank test revealed significantly worse outcome for tumors located = within the=20 optic pathway especially if these tumors were Dodge stage II/III = (Table 2).=20 However, gender, age, and histological type and mere clinical/imaging = diagnosis=20 did not influence EFS. When analyzing the EFS of the chemotherapy group = for=20 single variables, no influence was found for age, gender, start of = therapy, and=20 extent of surgery, histology, and response to therapy at week 24. Risk = factors=20 were also not found for the EFS of the radiotherapy group patients.

Table 2 Univariate=20 analysis (log-rank test) of the risk factors for event free survival of = the NF-1=20 cohort of the HIT-LGG 1996 protocol

Factor	EFS at 5 years	SE at 5 years	n	P value
Gender				0.3234
Male	0.29	0.07	52
Female	0.20	0.05	57
Localization				0.0334
Optic pathway	0.19	0.04	82
Other	0.34	0.10	26
Optic pathway localization				0.015
Dodge I	0.50	0.12	16
Dodge II/III	0.11	0.05	66
Other	0.34	0.10	26
Histology				0.986
PA	0.18	0.07	38
Clinical diagnosis only	0.27	0.06	67
Other	0.25	0.21	4

EFS Event free survival, SE standard = error,=20 n number of patients, PA pilocytic astrocytomas WHO I=20

Multivariate EFS analysis of the whole NF-1 cohort = revealed=20 that patients who had a biopsy had the highest risk for progressive = disease=20 (Table 3).=20 The risk of PD in the group of patients who underwent a more extensive = surgery=20 was still significantly higher than those whose diagnosis was made by = clinical=20 examination and neuroimaging only. Tumors located outside the optic = pathway had=20 a significantly reduced risk of progression after all histologies were = included.=20 After excluding patients with tumors of PXA WHO II or A WHO II = histology,=20 localization did not play any significant role. Similar results were = obtained=20 with multivariate analysis for the combined adjuvant treatment group and = for the=20 chemotherapy group alone (Table 4).=20 A further multivariate analysis of the combined adjuvant treatment group = and the=20 chemotherapy group revealed that, after exclusion of patients suffering = from=20 tumors with PXA WHO II or A WHO II histology, localization outside the = optic=20 pathway and PA WHO I histology were significantly associated with lower = risk of=20 PD when compared with patients diagnosed by clinical and imaging = findings only.=20 Other risk factors, such as type of NF-1 trait inheritance, gender, and = age,=20 were included but did not display significance in the multivariate = analysis of=20 any group.

Table 3 Multivariate=20 analysis of the risk factors for the EFS of the NF-1 cohort of the = HIT-LGG 1996=20 protocol

Variable	Comparison	HR	95% CI	P value of likelihood ratio test =
Extent of surgery	Biopsy vs no/prior to surgery	6.45	(2.93; 14.20)	P =3D 0.001
Resection vs no/prior to surgery	1.24	(0.63; 2.42)
Localization^a	Other vs optic pathway	0.44	(0.23; 0.84)	P =3D 0.007 =

HR Hazard ratio, CI confidence = interval=20

^aThis variable was not significant when = the 4=20 patients with histologies other than pilocytic astrocytomas or clinical=20 diagnosis only were excluded

Table 4 Multivariate=20 analysis of the risk factors for the EFS of NF-1 patients treated with = adjuvant=20 therapy

Variable	Comparison	HR	95% CI	P-value of likelihood ratio test =
Extent of surgery	Biopsy vs no/prior to surgery	3.66	(1.59; 8.42)	P =3D 0.005
Resection vs no/prior to surgery	2.53	(1.15; 5.58)
Localization^a	Other vs optic pathway	0.27	(0.11; 0.67)	P =3D 0.001
Histology^b	Pilocytic astrocytoma vs clinical diagnosis = only	0.25	(0.10; 0.63)	P =3D 0.001 =

HR Hazard ratio, CI confidence = interval=20

^aThis variable was in the final model = when patients=20 with histologies other than pilocytic astrocytomas and clinical = diagnosis only=20 were excluded

^bThis variable was in the final model = when the 4=20 patients with histologies other than pilocytic astrocytomas or clinical=20 diagnosis only were excluded

Univariate Kaplan=96Meier estimates of PFS for the = chemotherapy=20 group showed no influence of age, gender, mode of NF-1 trait = inheritance,=20 localization, histology, initial time interval from diagnosis to = therapy, or=20 response to therapy. Twenty events in 65 patients with adjuvant therapy = and 17=20 events in 55 patients receiving chemotherapy did not allow multivariate=20 regression analysis.

Discussion

The group of NF-1 patients with LGG in this study is = one of the=20 very few prospectively registered cohorts and was collected within the = first=20 multi-center cooperative trial of the German Speaking Society of = Pediatric=20 Oncology and Hematology (GPOH) to optimize LGG therapy in children and=20 adolescents. In contrast to earlier studies stating that LGG in NF-1 = patients=20 arises at an age of less than 5 years, in our cohort LGG arose = throughout=20 the first two decades of life and became symptomatic [3,=20 5].=20 Nearly 25% of patients diagnosed with a tumor were older than = 11 years of=20 age. This underscores the risk of new symptomatic LGG, especially within = the=20 optic pathway of pediatric NF-1 patients during the second decade of = life=20 [8,=20 19].=20

Female patients dominated all strategy groups, = especially the=20 adjuvant treatment groups. This supports epidemiologic findings that = female NF-1=20 patients are more prone to develop tumors than male NF-1 patients = [20].=20 However, this dominance was not significant in terms of EFS or PFS.

The group is characterized by a majority of optic = pathway=20 lesions (76%) corroborating other studies [2=965].=20 NF-1-associated optic glioma were mainly reported in the optic nerve = itself, yet=20 the LGG of our cohort were predominantly detected in the chiasmatic and=20 postchiasmatic area (80%), and lesions restricted to the optic nerve = were rather=20 rare. Furthermore, our cohort lacked patients with adverse risk factors = such as=20 age under 1 year at diagnosis, disseminated disease pattern and=20 diencephalic syndrome [12,=20 21].=20 Also, the frequency of brain stem involvement was lower (3/109) than = reported by=20 Guillamo et al. [3]=20 (17%), whereas hemispheric localization was comparable (7/109). We found = that=20 the involvement of the chiasm, especially the postchiasmatic area, was=20 associated with worse EFS in the univariate analysis. Two out of eight = studies=20 in the metaanalysis by Opocher et al. [12]=20 support our findings, identifying Dodge II/III tumors versus Dodge I = tumors as=20 an independent risk factor for progression of optic pathway glioma in=20 multivariate analysis, whereas the remainder did not.

In our cohort, one patient suffering from a = hemispheric tumor=20 was diagnosed with a pleomorphic xanthoastrocytoma WHO grade II. To our=20 knowledge, this is the first report of such histology in a pediatric = NF-1=20 patient. Furthermore, despite progression in 3 out of 4 patients, none = in the=20 cohort was diagnosed with a high-grade brain tumor within the follow-up = period=20 [22].=20 Only a quarter of all patients had tumors that were amenable to surgical = therapeutic intervention. Surgery sufficed as sole therapy in only 18 of = 27=20 patients, similar to reports from other groups [3,=20 23=9625].=20 In almost 2 out of 3 patients, diagnosis and decision on therapy were = based upon=20 clinical and imaging findings only, echoing other reports [22].=20

Sixty percent of our patient cohort received adjuvant = treatment,=20 55 of whom were treated with chemotherapy. Until now, many clinicians = have given=20 radiotherapy similar or higher importance than chemotherapy [3,=20 5,=20 6,=20 12].=20 Thus, our cohort is the first to report the results of NF-1 patients = with a LGG=20 treated predominantly with chemotherapy.

Earlier reports on pediatric NF-1 patients indicated = that LGG=20 largely remained asymptomatic [3=966].=20 However, EFS of our patients was low (5-year EFS 0.24). We cannot = exclude a=20 registration bias due to the reporting system as most patients with a = more=20 aggressive course were probably registered. The marked tendency of LGG=20 progression in our cohort may also be due to the fact that hypothalamic = tumors=20 with a more aggressive biology prevail [16].=20 On the other hand, our large observation group also contained one case = of=20 spontaneous regression (signal alteration in the optic tract that did = not=20 fulfill criteria of an unidentified bright object) [26].=20

Response within the NF-1 cohort to vincristine/single = dose=20 carboplatin chemotherapy was promising and comparable to responses = reported by=20 other groups using combinations of 2 drugs [16,=20 21,=20 27,=20 28].=20 Compared with these studies, we were able to achieve high 5-year PFS = rate of=20 0.73 [16,=20 21,=20 27,=20 28].=20

Three of the children of the radiotherapy group = received=20 radioiodine seed implantation. This approach is considered very safe but = has its=20 limitations concerning tumor size and location [15].=20 To date, radiotherapy in our cohort was not associated with risks = reported in=20 other NF-1 patients [29=9632].=20 The small number of patients treated with this approach limits the = assertions=20 that can be made about its safety.

LGG located within the optic pathway, especially = within the=20 chiasmatic and postchiasmatic area, were more likely to be associated = with tumor=20 growth according to univariate analysis. This was further corroborated = by=20 multivariate analysis. These findings are in line with clinical studies=20 reporting hypothalamic and optic pathway LGG to be generally more = aggressive=20 [16].=20 Experimental studies demonstrate that spatiotemporal differences of CXCL = 12=20 expression contribute to the unique pattern of NF-1-associated optic = pathway=20 tumors [33].=20 A risk factor more strongly associated with progressive LGG was biopsy = and=20 surgical intervention versus clinical and imaging diagnosis only. Our=20 preliminary understanding of this observation is that clinicians may = prefer more=20 invasive surgery for tumors that appear abnormal on imaging [22].=20 Another explanation may be that tumors, when submitted to surgery, = responded to=20 inflammatory mediators set free by the surgical trauma with cellular = growth. A=20 third interpretation is that tumors diagnosed with imaging only may be=20 biologically less aggressive.

Despite a countrywide trial and children=92s cancer = registry we=20 cannot exclude that NF-1 children with non-progressive tumors are = followed by=20 medical specialties outside pediatric oncology though we believe that = our study=20 included the relevant patients. Another limitation is that at the = beginning of=20 the study a central review of radiology had yet to be established.

In summary, NF-1 patients have a strong tendency to = develop a=20 progressive LGG throughout the first two decades of life. A high = percentage of=20 our cohort needed therapy. Chemotherapy achieved satisfactory PFS. = Irradiation=20 seems to reach similar PFS rates. Tumor response to chemotherapy was = good and=20 delayed irradiation. The biological risk factors that determine = progression=20 remain unknown. NF-1 patients should follow current treatment = recommendations=20 until separate strategies respecting such biological aspects will have = been=20 developed.

Acknowledgements The study HIT-LGG received a grant from =93Deutsche = Krebshilfe=94. We=20 are gratefully indebted to Sylvia S=F6llner, Marina Geh and Sabine=20 Breitmoser-Greiner for secretarial and documentation = assistance.

References

1.	Arun D, Gutmann DH (2004) Recent advances in=20 neurofibromatosis type 1. Curr Opin Neurol 17:101=96105

2.	Cnossen MH, de Goede-Bolder A, van den Broek = KM et al=20 (1998) A prospective 10 year follow up study of patients with = neurofibromatosis type 1. Arch Dis Child 78:408=96412

3.	Guillamo JS, Creange A, Kalifa C et al (2003)=20 Prognostic factors of CNS tumours in Neurofibromatosis 1 (NF1): a=20 retrospective study of 104 patients. Brain 126:152=96160

4.	Korf BR (2000) Malignancy in neurofibromatosis = type 1.=20 Oncologist 5:477=96485

5.	Listernick R, Charrow J, Gutmann DH (1999)=20 Intracranial gliomas in neurofibromatosis type 1. Am J Med Genet=20 89:38=9644

6.	Rosser T, Packer RJ (2002) Intracranial = neoplasms in=20 children with neurofibromatosis 1. J Child Neurol 17:630=96637; = discussion=20 646-651

7.	Walker L, Thompson D, Easton D et al (2006) A=20 prospective study of neurofibromatosis type 1 cancer incidence in = the UK.=20 Br J Cancer 95:233=96238

8.	Listernick R, Ferner RE, Liu GT et al (2007) = Optic=20 pathway gliomas in neurofibromatosis-1: controversies and = recommendations.=20 Ann Neurol 61:189=96198

9.	Chateil JF, Soussotte C, Pedespan JM et al = (2001) MRI=20 and clinical differences between optic pathway tumours in children = with=20 and without neurofibromatosis. Br J Radiol 74:24=9631

10.	Gutmann DH, James CD, Poyhonen M et al (2003) = Molecular analysis of astrocytomas presenting after age 10 in = individuals=20 with NF1. Neurology 61:1397=961400

11.	Kluwe L, Hagel C, Tatagiba M et al (2001) = Loss of NF1=20 alleles distinguish sporadic from NF1-associated pilocytic = astrocytomas. J=20 Neuropathol Exp Neurol 60:917=96920

12.	Opocher E, Kremer LC, Da Dalt L et al (2006)=20 Prognostic factors for progression of childhood optic pathway = glioma: a=20 systematic review. Eur J Cancer 42:1807=961816

13.	Gnekow AK, Kaatsch P, Kortmann R et al (2000) = HIT-LGG: effectiveness of carboplatin-vincristine in progressive = low-grade=20 gliomas of childhood=97an interim report. Klin Padiatr = 212:177=96184

14.	Gnekow AK, Kortmann RD, Pietsch T et al = (2004) Low=20 grade chiasmatic-hypothalamic glioma-carboplatin and vincristin=20 chemotherapy effectively defers radiotherapy within a = comprehensive=20 treatment strategy=97report from the multicenter treatment study = for=20 children and adolescents with a low grade glioma=97HIT-LGG = 1996=97of the=20 Society of Pediatric Oncology and Hematology (GPOH). Klin Padiatr=20 216:331=96342

15.	Kreth FW, Faist M, Warnke PC et al (1995)=20 Interstitial radiosurgery of low-grade gliomas. J Neurosurg=20 82:418=96429

16.	Janss AJ, Grundy R, Cnaan A et al (1995) = Optic=20 pathway and hypothalamic/chiasmatic gliomas in children younger = than age=20 5 years with a 6-year follow-up. Cancer 75:1051=961059

17.	Gnekow AK (1995) Recommendations of the Brain = Tumor=20 Subcommittee for the reporting of trials. SIOP Brain Tumor = Subcommittee.=20 International Society of Pediatric Oncology. Med Pediatr Oncol=20 24:104=96108

18.	Dodge HW Jr, Love JG, Craig WM et al (1958) = Gliomas=20 of the optic nerves. AMA Arch Neurol Psychiatry 79:607=96621

19.	Listernick R, Ferner RE, Piersall L et al = (2004)=20 Late-onset optic pathway tumors in children with neurofibromatosis = 1.=20 Neurology 63:1944=961946

20.	Airewele GE, Sigurdson AJ, Wiley KJ et al = (2001)=20 Neoplasms in neurofibromatosis 1 are related to gender but not to = family=20 history of cancer. Genet Epidemiol 20:75=9686

21.	Laithier V, Grill J, Le Deley MC et al (2003) = Progression-free survival in children with optic pathway tumors:=20 dependence on age and the quality of the response to = chemotherapy=97results=20 of the first French prospective study for the French Society of = Pediatric=20 Oncology. J Clin Oncol 21:4572=964578

22.	Leonard JR, Perry A, Rubin JB et al (2006) = The role=20 of surgical biopsy in the diagnosis of glioma in individuals with=20 neurofibromatosis-1. Neurology 67:1509=961512

23.	Allen JC (2000) Initial management of = children with=20 hypothalamic and thalamic tumors and the modifying role of=20 neurofibromatosis-1. Pediatr Neurosurg 32:154=96162

24.	Riffaud L, Vinchon M, Ragragui O et al (2002) = Hemispheric cerebral gliomas in children with NF1: arguments for a = long-term follow-up. Childs Nerv Syst 18:43=9647

25.	Vinchon M, Soto-Ares G, Ruchoux MM et al = (2000)=20 Cerebellar gliomas in children with NF1: pathology and surgery. = Childs=20 Nerv Syst 16:417=96420

26.	Piccirilli M, Lenzi J, Delfinis C et al = (2006)=20 Spontaneous regression of optic pathways gliomas in three patients = with=20 neurofibromatosis type I and critical review of the literature. = Childs=20 Nerv Syst 22:1332=961337

27.	Massimino M, Spreafico F, Cefalo G et al = (2002) High=20 response rate to cisplatin/etoposide regimen in childhood = low-grade=20 glioma. J Clin Oncol 20:4209=964216

28.	Packer RJ, Ater J, Allen J et al (1997) = Carboplatin=20 and vincristine chemotherapy for children with newly diagnosed = progressive=20 low-grade gliomas. J Neurosurg 86:747=96754

29.	Grill J, Couanet D, Cappelli C et al (1999)=20 Radiation-induced cerebral vasculopathy in children with = neurofibromatosis=20 and optic pathway glioma. Ann Neurol 45:393=96396

30.	Kestle JR, Hoffman HJ, Mock AR (1993) = Moyamoya=20 phenomenon after radiation for optic glioma. J Neurosurg = 79:32=9635

31.	Rosser TL, Vezina G, Packer RJ (2005) = Cerebrovascular=20 abnormalities in a population of children with neurofibromatosis = type 1.=20 Neurology 64:553=96555

32.	Sharif S, Ferner R, Birch JM et al (2006) = Second=20 primary tumors in neurofibromatosis 1 patients treated for optic = glioma:=20 substantial risks after radiotherapy. J Clin Oncol = 24:2570=962575

33.	Warrington NM, Woerner BM, Daginakatte GC et = al=20 (2007) Spatiotemporal differences in CXCL12 expression and cyclic = AMP=20 underlie the unique pattern of optic glioma growth in = neurofibromatosis=20 type 1. Cancer Res 67:8588=968595