clinical article

Growth patterns of craniopharyngiomas: clinical analysis

of 226 patients
*Jun Pan, MD, PhD,2 Songtao Qi, MD, PhD,1 Yi Liu, MD,1 Yuntao Lu, MD,1 Junxiang Peng, MD,1
XiAn Zhang, MD,1 YiKai Xu, MD,3 Guang-long Huang, MD,1 and Jun Fan, MD1
Departments of 1Neurosurgery, 2Pediatric Neurosurgery, and 3Medical Imaging, Nanfang Hospital, Southern Medical University,
Guangzhou, People’s Republic of China

Object Craniopharyngiomas (CPs) are rare epithelial tumors that are often associated with an enigmatic and unpre-
dictable growth pattern. Understanding the growth patterns of these tumors has a direct impact on surgical planning and
may enhance the safety of radical tumor removal. The aim of this study was to analyze the growth patterns and surgical
treatment of CPs with a focus on the involvement of the hypothalamopituitary axis and the relationship of the tumor to
the arachnoid membrane and surrounding structures.
Methods Clinical data from 226 consecutive patients with primary CP were retrospectively reviewed. Tumor location
and the relationship of the tumor to the third ventricle floor and the pituitary stalk were evaluated using preoperative MRI
and intraoperative findings. A topographic classification scheme was proposed based on the site of tumor origin and
tumor development. The clinical relevance of this classification on patient presentation and outcomes was also analyzed.
Results The growth of CPs can be broadly divided into 3 groups based on the site of tumor origin and on tumor-
meningeal relationships: Group I, infrasellar/infradiaphragmatic CPs (Id-CPs), which mainly occurred in children; Group
II, suprasellar subarachnoid extraventricular CPs (Sa-CPs), which were mainly observed in adults and rarely occurred
in children; and Group III, suprasellar subpial ventricular CPs (Sp-CPs), which commonly occurred in both adults and
children. Tumors in each group may develop complex growth patterns during vertical expansion along the pituitary stalk.
Tumor growth patterns were closely related to both clinical presentation and outcomes. Patients with Sp-CPs had more
prevalent weight gain than patients with Id-CPs or Sa-CPs; the rates of significant weight gain were 41.7% for children
and 16.7% for adults with Sp-CPs, 2.2% and 7.1% for those with Id-CPs, and 12.5% and 2.6% for those with Sa-CPs (p
< 0.001). Moreover, patients with Sp-CPs had increased hypothalamic dysfunction after radical removal; 39% of patients
with Sp-CPs, 14.5% with Id-CPs, and 17.4% with Sa-CPs had high-grade hypothalamic dysfunction in the first 2 postop-
erative years (p < 0.001).
Conclusions The classification of CPs based on growth pattern may elucidate the best course of treatment for this
formidable tumor. More tailored, individualized surgical strategies based on tumor growth patterns are mandatory to
provide long-term tumor control and to minimize damage to hypothalamic structures. Differences in the distribution of
growth patterns between children and adults imply that hierarchical comparison is necessary when investigating out-
comes and survival across treatment paradigms in patients with CP.
http://thejns.org/doi/abs/10.3171/2015.7.PEDS14449
Key Words classification; craniopharyngiomas; growth pattern; pituitary stalk; oncology

C
raniopharyngiomas (CPs) are formidable and com- growth pattern is mandatory to achieve complete resec-
plex benign tumors of the sellar and suprasellar tion with minimal compromise to surrounding vital neu-
regions. Although the optimal treatment for CP re- rovascular structures. For this reason, many topographic
mains controversial, there is consensus that surgical treat- classifications have been proposed based on tumor loca-
ment should be tailored according to tumor localization.53 tion and extension, with relationships of the tumor to the
Selecting an optimal surgical approach based on the tumor sellar diaphragm, optic chiasm, pituitary stalk, and third

Abbreviations ACP = adamantinomatous craniopharyngioma; ASPS = arachnoidal sleeve around the pituitary stalk; BMI = body mass index; CP = craniopharyngioma;
HSS = hypothalamic status score; OS = overall survival; PCP = papillary craniopharyngioma; PFS = progression-free survival; TSH = thyroid-stimulating hormone; TVF =
third ventricle floor.
submitted August 22, 2014. accepted July 13, 2015.
include when citing Published online December 4, 2015; DOI: 10.3171/2015.7.PEDS14449.
* Drs. Pan and Qi contributed equally to this work.

©AANS, 2015 J Neurosurg Pediatr December 4, 2015 1

J. Pan et al.

ventricle floor (TVF) being the main determinants taken ited delayed growth and development and a short stature
into consideration.6,27,32,33,54,55,58,62,63 These classification (height > 3 SDs below the mean for a particular age and
schemes are useful in understanding tumor topography sex). TSH deficiency was diagnosed on the basis of low or
and provide clues to aid in the selection of the optimal “inappropriately normal” TSH with free thyroxine levels
surgical approach for tumor removal. However, a classifi- below the normal reference values. The development of
cation system capable of describing both the details of the secondary sex characteristics was evaluated in adolescent
tumor origin and hypothalamic involvement patterns has patients (girls aged > 12 years, boys aged > 15 years). A
yet to be reported in the literature.46 diagnosis of diabetes insipidus was comprehensively de-
In our preliminary study, we reported the surgical termined according to urine volume (> 3 L/24 hrs) and
relevance of the anatomical relationship of the tumor to urine osmolality (< 300 mOsm/kg).
arachnoidea around the pituitary stalk, and we proposed a Preoperative CT and MRI scans for each patient were
new morphological classification that divided CP growth reviewed and scored by the same radiologist (Y.X.), who
into 4 basic patterns.49 Growth pattern variants with dis- was blinded to the clinical data. Tumor size was calcu-
tinct morphological features were also observed.49 Addi- lated as the maximum measurable dimension on MRI.
tionally, the study concentrated on the relationship of CPs The presence of hydrocephalus was classified as follows:
to surrounding membranous structures, which are always no, without ventricular widening, or yes, CT- and/or MRI-
included in the surgical plane during tumor resection. Al- confirmed ventricular widening with an Evans’ index ≥
though the classification was useful in understanding the 0.30.
diverse growth patterns of CPs, it was complex and theo-
retical to a certain extent. In the current study, we have Surgery and Classification of Tumors
attempted to simplify the classification and enhance its All 226 patients in this series underwent resection with
clinical practicality. the aim of total tumor removal. The procedures were per-
formed under the operating room microscope mainly by 2
Methods surgeons (S.Q. and J.P.) and were video recorded. To ver-
ify tumor location and its topographic relationships to the
Review of Clinical Data sellar diaphragm, pituitary stalk, arachnoid membranes
Patient Details around the stalk, and TVF, we retrospectively reviewed
Patients in this study were recruited from among all pa- and analyzed the preoperative MR images and intrasurgi-
tients with CP who were admitted to the Nanfang Hospital cal findings for each patient (J.P. and Y.L.). A topograph-
of the Southern Medical University between January 1996 ic classification of the tumors was proposed based on a
and December 2011. A total of 311 consecutive patients combination of tumor location and the relationship of the
underwent surgical removal of CP at the hospital during tumor to the membranous structure, stalk, and TVF. The
this period. Eighty-five patients with recurrent tumors extent of tumor removal was determined by intrasurgical
were excluded. All pathological materials and medical re- findings and postoperative contrast-enhanced MRI. Total
cords of the remaining 226 patients with primary tumors tumor removal was declared when operative impression
were reviewed, and the CP diagnosis was histologically and postoperative MRI documented no residual tumor.
confirmed. Of the 226 patients with CP, 96 patients suf-
fered from childhood-onset tumors (age ≤ 16 years at the Follow-Up
time of diagnosis; 58 boys and 38 girls), and 130 patients Follow-up was scheduled at 3 and 9 months after sur-
suffered from adult-onset tumors (87 men and 43 women). gery and annually thereafter. Tumor recurrence was de-
The average age at the time of surgery was 9.0 ± 4.6 years termined according to follow-up imaging, which was per-
for childhood patients (range 1–27 years) and 39.8 ± 13.3 formed at regular intervals based on the clinical pictures
years for adult patients (range 17–72 years). and the physician’s preference, with or without associated
signs and symptoms of recurrence. Each patient’s height
Clinical Data Assessment and weight were recorded at all follow-up visits. The max-
Data points for clinical presentation, preoperative en- imum weight gain in the first 2 postoperative years was
docrine status, imaging findings, operative details, pa- used for analysis. Body mass index (BMI) was calculated
thology reports, and clinical outcomes were reviewed. with the following formula: (weight in kg)/(height in me-
Pituitary function was evaluated by laboratory assess- ters)2. Postoperative weight categories for adult patients
ment of adrenocorticotropic hormone, growth hormone, were classified as normal (BMI < 25), overweight (BMI
insulin-like growth factor, prolactin, thyroid-stimulating 25–30), or obesity (BMI > 30). Weight in children was
hormone (TSH), thyroxin, follicle-stimulating hormone, evaluated by comparing BMI with the reference values of
luteinizing hormone, testosterone, and morning cortisol Chinese children, and weight categories were classified as
levels. A provocation test was applied as needed. Adre- normal, obesity (BMI greater than +2 SD), and extreme
nocorticotropic hormone deficiency was defined as a peak obesity (BMI greater than +4 SDs). Other hypothalamic
cortisol value of less than 500 nmol/L after provocative disturbances, such as hyperphagia, memory deficits, ther-
testing (via an insulin hypoglycemia, glucagon, or short moregulatory abnormalities, emotionally labile behavior,
synacthen test) or a baseline plasma cortisol level of less and sleep-wake cycle disruption, were also evaluated and
than 100 nmol/l. Growth hormone deficiency was defined recorded based on information from the patients and their
as a peak value of less than 7 ng/ml after effective stimu- families as well as records of the follow-up office visits.
lation in an insulin stimulation test or if patients exhib- The ultimate hypothalamic status score (HSS) in children

2 J Neurosurg Pediatr December 4, 2015

 Growth pattern of craniopharyngiomas

was evaluated using the 4-tiered grading scale created by statistical software was used for all analyses. A p value <
De Vile et al.16 For facilitating statistical analysis, the post- 0.05 was considered statistically significant.
operative hypothalamic status in adult patients was evalu-
ated using a 4-tiered grading scale as follows: Grade 1,
normal hypothalamic function; Grade 2, overweight (25 Results
< BMI ≤ 30) and lack of behavior indicative of hypotha- Classification of CPs
lamic dysfunction; Grade 3, obesity (BMI > 30) or weight Theoretically, CPs initially develop along the pituitary
gain without hyperphagia or an associated change in af- stalk in 5 basic growth patterns based on their anatomi-
fective behavior or memory; and Grade 4, obesity (BMI cal location and relationship to the arachnoidal sleeve
> 30) and hyperphagia with cognitive dysfunction, rage around the pituitary stalk (ASPS) (Fig. 1).56 Tumor ana-
behavior, and impaired thirst or disturbances of thermo- tomical location and development combined with tumor
regulation concomitant with sleep-wake disruption or ex- site of origin allowed for precise CP classification. CPs
treme emaciation. were classified into 3 groups: Group I, infradiaphragmatic
CPs (Id-CPs); Group II, suprasellar subarachnoid extra-
Statistical Analysis ventricular CPs (Sa-CPs); and Group III, suprasellar sub-
Differences in tumor growth patterns between age pial ventricular CPs (Sp-CPs) (Table 1). Each tumor group
groups and differences in imaging features among the can develop complex subsets along the hypothalamopitu-
growth pattern groups were analyzed using the chi-square itary axis during tumor development. These growth pat-
test or Fisher’s exact test. Differences in tumor size be- tern variants could be interpreted by the aforementioned 5
tween groups were analyzed using the Student t-test. The basic growth patterns. A detailed description of the 3 CP
Wilcoxon rank-sum test (Mann-Whitney U-test) was used groups is presented below.
to compare nonparametric variables between groups. The
overall survival (OS) and progression-free survival (PFS) Group I: Id-CPs
curves were generated using the Kaplan-Meier method, The Id-CPs usually demonstrated marked enlargement
and differences in survival and recurrence rates in each of the pituitary fossa and a large rounded and symmetrical
group were evaluated using the log-rank test. SPSS 17.0 extension to the suprasellar space (Fig. 2a). Tumor growth

FIG. 1. The diagram shows the anatomical basis of our classification system. Left: The pituitary stalk penetrates the basal
arachnoid membrane (BAM) covering diaphragma sellae where the BAM extends upward along the pituitary stalk, which forms an
arachnoid sleeve (pink dotted line) to envelop the pituitary stalk. The stalk can therefore be divided into 3 segments according to
its relationship with the surrounding arachnoid membrane: infraarachnoid (subdiaphragmatic) segment, intraarachnoid segment
(within the arachnoid sleeve of the pituitary stalk), and subarachnoid segment. CPs arising from the pituitary stalk will have a differ-
ent topographic relationship with these membranous structures. Theoretically, CP developing along the pituitary stalk has 5 basic
growth patterns based on the tumor’s anatomical locations and topographic relationships with the seller meningeal structures: 1)
pattern Id, infradiaphragmatic growth (extraarachnoid); 2) pattern Ts, transstalk growth (intraarachnoid); 3) pattern Sa, suprasellar
subarachnoid extraventricular growth (subarachnoid extrapial); 4) pattern Sp, subarachnoid subpial ventricular growth (subpial);
and rarely, 5) pattern Is, infrasellar growth (extraarachnoid). Right: The growth of CPs can be broadly divided into 3 basic groups:
1) infradiaphragmatic (Id), 2) subarachnoid extraventricular (Sa), and 3) subpial intraventricular (Sp). Each tumor group can
develop complex subsets along the hypothalamopituitary axis during tumor development. These growth pattern variants could be
interpreted by the 5 basic growth patterns. Figure is available in color online only.

J Neurosurg Pediatr December 4, 2015 3

J. Pan et al.

TABLE 1. Growth patterns and tumor-meningeal relationships in 226 patients with primary CP, according to our classification scheme

Possible Site of Patients

Tumor Origin Classification Classic Name of CP Growth Pattern Tumor-Arachnoidal Relationship Juveniles Adults Total
Infradiaphragmatic Group I (Id-CP) Infrasellar CP Purely Is or Is+Id Extraarachnoidal 9 0 62
Infradiaphragmatic CP Id Extraarachnoidal 32 10
Transdiaphragmatic CP Id+Ts Extra-, intraarachnoidal combined 7 0
Transstalk CP Id+Ts+Sp Extra-, intra-, subarachnoidal 0 4
combined
Supradiaphragmatic Group II (Sa-CP) Suprasellar extraventricu- Sa1 (prestalk) Subarachnoidal, extrapial 1 34 46
infundibulum lar CP Sa2 (retrostalk) 7 4
Group III (Sp-CP) Intraventricular CP Purely Sp Subarachnoidal, subpial 7 44 118
Intra-, extraventricular CP Sp+Ts or Sp+Sa Subarachnoidal, intra-, extrapial 33 34
combined
Total 96 130 226

originated within the infrasellar and/or intrasellar cav- (Fig. 4), depending on the orientation of tumor expansion.
ity below the diaphragm and arachnoid. The emergence The main body of the tumor was separated from the sur-
of the tumor caused upward displacement of the sellar rounding structures by the suprasellar arachnoid around
diaphragm and arachnoid. In juveniles, the tumor may the pituitary stalk, except at the proximal portion of the
originate more deeply from the infrasellar region, which pituitary stalk (infundibulum area), where the tumor had a
includes the pharynx, sphenoidal sinus, and ethmoid si- subpial relationship with the stalk. The infundibulum area
nus. In rare cases, the tumor may expand longitudinally may also be a site of tumor origin. Tumors with a suprasel-
up through the diaphragm along the pituitary stalk and lar subarachnoid extraventricular growth pattern tended
dilate the suprasellar portion of the pituitary stalk (Fig. to occupy multiple compartments due to tumor expansion
2b). In addition, suprasellar expansion of the tumor could along the subarachnoid cisterns, especially in pediatric pa-
extrude the covering of the sellar diaphragm, resulting in tients (Fig. 5).
secondary cysts to the tumor capsule dome. In contrast,
Id-CP in adult patients never involved the infrasellar re- Group III: Sp-CPs
gion but rather usually emerged as purely intrasellar or Sp-CPs represented the most complicated growth pat-
intrasellar/suprasellar masses. Few adult cases showed tu- tern. This group of tumors initially grew around the in-
mor involvement along the whole length of the pituitary fundibulotuberal region (the subarachnoid segment of the
stalk (transstalk growth; Fig. 2c). These tumors probably pituitary stalk) and the top portion of the hypothalamopi-
originated in the sella turcica and expanded up through tuitary tract. Growing tumors mainly occupied the third
the diaphragm along the pituitary stalk and were thus as- ventricle cavity and were actually enclosed within the
signed to the Id-CP group. thinning nerve tissue of the third ventricle walls (usually
within the TVF). These tumors corresponded to subpial
Group II: Sa-CPs lesions and were thus enclosed within the thinning ner-
From its point of origin on the proximal stalk and in- vous tissues of the TVF and separated from the suprasel-
fundibulum region, Sa-CPs mainly developed freely in lar cistern by a layer of pia mater. The tumors were in
the suprasellar subarachnoid spaces. Extraventricular tu- direct contact with the neural tissue of the third ventricle
mor development was either prestalk (Fig. 3) or retrostalk wall, making the attachments of the tumor to the hypo-

FIG. 2. The basic (pattern Id) and variant (pattern Is+Id [not shown here]; Id+Ts; and Id+Ts+Sp) growth patterns in CPs with an
infradiaphragmatic origin. The tumor maintains an extraarachnoid location in both pattern Id (a) and/or Is+Id. Rarely, growing tumor
may expand along the pituitary stalk, making the tumor’s anatomical locations either extra- and intraarachnoid combined (pattern
Id+Ts; b) or extra-, intra-, and subarachnoid subpial (pattern Id+Ts+Sp; c).

4 J Neurosurg Pediatr December 4, 2015

 Growth pattern of craniopharyngiomas

FIG. 3. Pre- and postoperative sagittal T1-weighted MR images (b and e) and intraoperative pictures (a, c, d, and f) in a suprasel-
lar subarachnoid extraventricular CP (Group II, Sa-CP) with growth pattern Sa1. The tumor arises from the pituitary stalk and
mainly localizes inside the optical chiasm cistern. Dissection of the tumor is mainly accomplished through the prechiasm space
(a and d). The tumor can be dissected along its own arachnoid dissecting plane from the chiasm, TVF, and the main arteries,
except at the site of the tumor origin where the tumor needs sharp dissecting (c and f). Postoperative MRI showed total removal of
tumor by the right pterion approach (e). Note the site of the tumor origin at the proximal end of the pituitary stalk (asterisk, f). Ara
= arachnoid around the pituitary stalk; ICA = internal carotid artery; L-ICA = left ICA; ON = optic nerve; PS = pituitary stalk; T =
tumor. Figure is available in color online only.

FIG. 4. Pre- and postoperative MR images (c and d) and intraoperative pictures (a, b, e, and f) demonstrating a large subarach-
noid extraventricular CP (Group II, Sa-CP) with retrostalk extension (pattern Sa2). From the origin site at the chiasmatic cistern
around the pituitary stalk, the tumor capsule extended mainly into the retroclivus cistern and finally involved the posterior cranial
fossa and cervical spinal canal (c). The tumor was removed by a right frontotemporal transtentorial approach. Intraoperative
photos show the various aspects of the tumor: the tumor adhered tightly to the TVF and the pituitary stalk (a). Only at the proximal
portion of the pituitary stalk and infundibulum does the tumor have a subpial relationship with the nerve tissue (b). Dissection of
the tumor capsule should trace the arachnoid plane under direct vision (e and f). Postoperative MR image shows the total removal
of tumor (d). 3rd N = oculomotor nerve; ACA = anterior cerebral artery; BA = basal artery; Tent = tentorium; VA = vertebral artery;
Vi N = abducens nerve. See Fig. 3 for additional abbreviations. Figure is available in color online only.

J Neurosurg Pediatr December 4, 2015 5

J. Pan et al.

FIG. 5. Preoperative MR and CT scans (c and d) and intraoperative pictures (a, b, e, and f) demonstrating a huge subarachnoid
extraventricular cisternal CP (Group II, Sa-CP) with retrostalk extension (pattern Sa2). From the origin site at the chiasmatic cis-
tern around the pituitary stalk, the tumor capsule expanded along the subarachnoid space and significantly compressed the TVF
by extending into the adjacent basal cistern and finally involved multiple cisterns. The tumor was removed via a left-sided fron-
totemporal combined supra- and infratentorial approach. Intraoperative photos show the various aspects of the tumor: the tumor
adhered tightly to the TVF but without infiltrating it (a, b, and e). Only at the proximal portion of the pituitary stalk and infundibulum
(star, b) does the tumor have subpial adherences with the nerve tissue. 3rd N = oculomotor nerve (contralateral in this case); 3rd
VF = 3rd ventricle floor. See Fig. 3 for additional abbreviations. Figure is available in color online only.

thalamus and optic tract structures more tenacious. The Correlation Between Tumor Growth Patterns and
anteroinferior portion of the third ventricle usually mani- Pathological Features
fested frog-belly expansion, and the remaining pituitary All pediatric tumors were classified as adamantinoma-
stalk could be seen compressed inferior to the tumor (Fig. tous craniopharyngioma (ACP), while in adults 21.5% of
6). Global extension of the tumor along the pituitary stalk tumors (28/130) were typical squamous papillary cranio-
dilated the distal part of the stalk and severely distended pharyngioma (PCP). The median age of patients with PCP
the stalk and its overlying arachnoid membrane to form was 38 years (range 19–72 years), and the median age of
an external tumor capsule (Fig. 7); this made dissection of patients with ACP was 24 years (range 2–66 years). In-
the tumor from the hypothalamopituitary tract even more fradiaphragmatic tumors were always adamantinomatous,
difficult and dangerous. In rare cases, progressive growth while 28 (17%) of the suprasellar tumors were typical PCP.
caused the low pole of the tumor to penetrate the gray mat- Among these 28 tumors, 23 (82.1%) belonged to Group III.
ter of the infundibulotuberal area; therefore, the tumor was
located both outside and inside the third ventricle cavity Differences in CP Growth Patterns Between Childhood
(Fig. 8). Radical removal of these tumors will inevitability and Adult Cases
result in defects of the anteroinferior portion of the TVF. Classification of the 226 patients with CP is shown in

FIG. 6. Sagittal (a) and coronal (b) contrast-enhanced T1-weighted MR images obtained before surgery and intraoperative photo
(c) obtained in a 6-year-old girl with a typical Sp-CP (Group III, purely Sp growth pattern). Note the inferiorly compressed pituitary
stalk (black arrowhead, a and b). Intraoperative picture showing the frog-belly–like inflated anterior part of the third ventricle (black
arrowhead, c) and a suprasellar cistern that is free of tumor. The tumor was a subpial lesion and was actually embedded within the
thinning nerve tissue of the TVF. Figure is available in color online only.

6 J Neurosurg Pediatr December 4, 2015

 Growth pattern of craniopharyngiomas

FIG. 7. Preoperative MR image and intraoperative photos obtained in a 7-year-old boy with a typical infundibulotuberal CP (Group
III, Sp-CP) with a growth pattern of Sp+Ts. The boy underwent an emergency procedure via the transcorpus callosum transfornix
approach to alleviate the hydrocephalus before tumor removal. Sagittal contrast-enhanced T1-weighted MR image (c) shows a
large cystic-solid mixed CP mainly occupying the third ventricle chamber. The inferiorly compressed pituitary stalk and normal
gland could be identified at the bottom of the tumor. Intraoperative pictures show removal of the tumor. The bulgy lamina terminalis
(LT) has been exposed after dividing the ACoA (a and b). After dissection of the arachnoid enveloping the pituitary stalk, the
bulged pituitary stalk was exposed (f). The tumor was totally removed via the combination of a trans-LT and extraaxial route. The
pituitary stalk was split vertically to expose the tumor (d). Note the attenuated but uninterrupted TVF after tumor removal (e). ACoA
= anterior communicating artery; OC = optic chiasm. See Figs. 3 and 5 for additional abbreviations. Figure is available in color
online only.

Table 1. CP growth patterns were significantly different higher rate in adults than in children (78/130 adults [60%]
between childhood- and adult-onset tumors (p = 0.024). vs 40/96 children [42%]). The trend persisted when only
adamantinomatous Sp-CPs (ACPs) were compared be-
Group I tween children and adults (51% for adults vs 42% for chil-
Id-CPs were predominantly observed in children and dren); however, this difference did not reach statistical sig-
were uncommon in adults (48/96 children [50%] vs 14/130 nificance (p = 0.24). Further analysis of the tumor growth
adults [10.7%]). Juvenile Id-CPs cases differed from adult pattern showed that tumors in adult patients with Sp-CP
cases in terms of tumor size, anatomical location, tumor appeared to exhibit a purely subarachnoid subpial ven-
expansion pattern, and cavernous sinus involvement (Table tricular growth pattern (44/78 [56%]), whereas tumors in
2). The topographic features of juvenile Id-CPs indicated children tended to exhibit a combination of transstalk and
that they were more invasive than adult Id-CPs. In addi- subarachnoid subpial ventricular growth (28/40 [70%]).
tion, juvenile Id-CPs had broad sites of origin, ranging
from the nasopharynx to the junction of the pituitary gland
Clinical Relevance of Tumor Growth Patterns
and stalk, whereas adult Id-CPs had higher and more spe-
cific sites of origin within the pituitary fossa. Patient symptoms at presentation were closely related to
tumor growth patterns (Table 3). Sixty-four percent of pa-
Group II tients sought medical attention because of visual compro-
Sa-CPs were commonly observed in adults but were mise; this correlated to 88.7% of Id-CP patients, 76.1% of
rare in children (38/130 adults [29.2%] vs 8/96 children Sa-CP patients, and 45.8% of Sp-CP patients (p < 0.001).
[8.3%]; Table 1). Furthermore, adult-onset Sa-CPs usually In addition, patients with Id-CPs were more likely to have
emerged as a prestalk growth, whereas childhood-onset pituitary dysfunction; this was particularly true in juve-
tumors tended to expand into the retroclivus and posterior nile patients, as 38.7% of children with Id-CPs had pan-
fossa. hypopituitarism on admission. Diabetes insipidus, which
represents the effect of the tumor on the pituitary stalk,
Group III occurred in 41.9% of patients with Id-CPs, 13% of patients
Sp-CPs involving the third ventricle walls occurred at a with Sa-CPs, and 12.8% of patients with Sp-CPs (p <
J Neurosurg Pediatr December 4, 2015 7
J. Pan et al.

FIG. 8. Pre- and postoperative MRI and intraoperative photos obtained in a 58-year-old man with a typical infundibulotuberal CP
(Group III, Sp-CP) with a growth pattern of Sp+Sa. Preoperative sagittal contrast-enhanced T1-weighted MR image (e) shows a
lobulated cystic-solid mass mainly occupying the third ventricle chamber at the suprasellar region. Intraoperative pictures reveal
the growth pattern of the tumor. Having originated from the infundibulotuberal part of the TVF, the tumor was located partially
outside and partially inside of the third ventricle cavity. The extraventricular part of the tumor has an arachnoid interface to the
chiasm and ACoA (a and b), and the bulgy LT has been exposed after dissection of the tumor’s extraventricular part (star, b). Note
a flying saucer–shaped expansion of the tumor (star, c) through the infundibulotuberal region of the TVF. The posterosuperior
tumor cyst invaginated into the gray matter (tuber cinereum [star], g) of the TVF (d). Involvement pattern of the pituitary stalk (f).
Postoperative sagittal contrast-enhanced T1-weighted MR image (h) and intraoperative photo (i) show total removal of the tumor
and disturbance of the TVF. MB = mammillary body; T-cap = tumor capsule. See previous figures for additional abbreviations.
Figure is available in color online only.

0.001). In contrast, symptoms related to increased intra- pattern of the tumor. Surgical removal of both Group I and
cranial pressure and nonspecific headaches occurred more Group II tumors was accomplished by an extraaxial route,
frequently in patients with Sp-CPs than in those with other which included pterional, frontobasal interhemispheric,
types of tumor growth (p < 0.001). Hydrocephalus rarely or transsphenoidal approaches. A translamina terminalis
occurred in Group I (8%) and Group II (13%) but was com- route was used in all Group III patients, either alone or in
monly observed in Group III (65.3%, including 34% with combination with other extraaxial routes. We favored mid-
severe hydrocephalus and marked ventricular widening line transcranial approaches to the lamina terminalis, par-
and/or periventricular edema; p < 0.0001; Table 3). Apart ticularly the anterior interhemispheric route, which avoids
from a slightly higher percentage of coarse calcification in the disadvantages of anterolateral approaches (e.g., lack of
tumors in Group III, there were no significant differences visualization of the ipsilateral wall of the third ventricle
in tumor size and tumor components among the 3 groups. and hypothalamus, limited oblique visualization superiorly
into the third ventricle, limited visualization inferiorly into
the interpeduncular cistern, and narrowing of the lamina
Surgical Details terminalis working corridor). The anterior communicating
Operative Approaches artery was severed for better tumor exposure in some large
Operative approaches were directly related to the growth tumors.
8 J Neurosurg Pediatr December 4, 2015
 Growth pattern of craniopharyngiomas

TABLE 2. Differences in anatomical locations of Id-CPs between juvenile and adult patients
Parameter Juveniles (48 patients, 50% [48/96] of the pediatric age group) Adults (14 patients, 11% [14/130] of the adult group)
Mean lesion size in cm* 4.39 ± 1.52 2.75 ± 1.09
Tumor’s anatomical location Pattern Is: 1 case (purely infrasellar)
Pattern Is+Id: 8 cases (infra-, intra-, suprasellar, infradia)
Pattern Id: 32 cases (intra-, suprasellar, infradia) Pattern Id: 10 cases (intrasellar, infradia, suprasellar)
Pattern Id+Ts: 7 cases (intra-, suprasellar, transtalk) Pattern Id+Ts+Sp: 4 cases (intra-, suprasellar,
transinfundibulum)
Secondary cysts after break 11 cases (23%) None
through the diaphragm
Involvement of cavernous sinus 2 cases (4%, rare) None
Possible origin site From the nasopharynx to the junction of the pituitary gland & Intrasellar portion of pituitary stalk near foramen
the stalk diaphragmatis
Infradia = infradiaphragmatic.
* There was statistical significant difference between children and adults (p <0.001).

Extent of Tumor Removal measurements of plasma sodium and accurate fluid bal-
The gross-total removal rate was highest in Group II ance charts were essential to assess the need for continued
(95.6%), followed by Group I (82.3%). In contrast, the treatment. Forty patients (17.6%) experienced a medical or
gross-total removal rate in Group III was 67%. This differ- surgical complication. A second surgical intervention was
ence in tumor removal rate between groups was significant necessary in 11 patients (decompression procedures in 4
(p < 0.001). In most patients, any residual tumor was minor patients, evacuation of an epidural hematoma in 2, cere-
on postoperative MRI, and solid tumors were observed in brospinal shunt placement in 6, evacuation of a subdural
only 7 patients. hematoma in 1, and exclusion of an opened frontal sinus
responsible for a cerebrospinal leakage in 1 [some patients
had more than one procedure]). Perioperative mortality
Perioperative Care and Surgical Complications was 3.98% (9 patients) in the entire series, and all cases
During the early postoperative period, hypothalamic- occurred in Group III patients (Table 4). Although the
pituitary axis dysfunction, including water and electro- events associated with operative mortality are multifacto-
lyte disorders, were commonly observed. Therefore, daily rial, a few were found to be major contributors, including

TABLE 3. Clinical presentation and neuroimaging data for 226 patients with primary CPs based on tumor growth
patterns
Parameter Total Id-CPs Sa-CPs Sp-CPs p Value
No. of patients 62 46 118
Age <0.001
   Children (≤16 yrs) 96 48 8 40
   Adults (>16 yrs) 130 14 38 78
Sex
  Male 145 43 25 77
  Female 81 19 21 41
Preop symptoms
  Visual disturbance 63.7% 88.7% 76.1% 45.8% <0.001
  Headache 53.5% 43.7% 30.4% 66.9% <0.001
Pituitary function <0.001
  Normal 15.9% 12.9% 23.9% 14.4%
  Partial hypopituitarism 60% 48.4% 67.4% 63.6%
  Pan-hypopituitarism 24% 38.7% 8.7% 20%
  DI 21.1% 41.9% 13% 12.8% 0.032
Neuroimaging
   Tumor size (cm) 3.83 ± 1.26 4.0 ± 1.55 3.59 ± 1.68 3.84 ± 0.83 0.243
  Hydrocephalus None (61%, 138/226) 57 40 41 <0.001
Yes (39%, 88/226) 5 6 77
DI = diabetes insipidus.

J Neurosurg Pediatr December 4, 2015 9

J. Pan et al.

TABLE 4. Summary of perioperative mortality in 226 patients with primary CP

Age Size (cm) & Death From
Case (yrs), Component of Assessment Time of Op
No. Sex Tumor Type Approach of Removal Events (days)
1 6, F 7 Sp-CP AIH Total Seizures, pan-hypopituitarism, severe hypothalamic dysregulation 12
(hyperpyrexia, hypernatremia, DI), peritonitis of undetermined
cause
2 11, F5, massive Sp-CP AIH Subtotal Frontal contusion, decompression procedure, severe hypotha- 5
calcification lamic dysregulation, acute pulmonary edema
3 27, F 4.5, massive Sp-CP FT Subtotal Intraop brain swelling, hypothalamic infarction, VP shunt, severe 127
calcification hypothalamic dysregulation
4 30, M 4 Sp-CP FT Total Evacuation of epidural hematoma, ischemia in area of perforating 13
vessels, dehydration & cerebral collapse due to hypothalamic
failure
5 9, F 4 Sp-CP FT Total Seizures, ventricle hematoma, external drainage 90
6 34, F 3.5 Sp-CP FT Subtotal Intraop brain swelling, frontal contusion, decompression proce- 5
dure, severe hypothalamic dysregulation
7 19, M 3 Sp-CP FT Total Anorexia & significant weight loss, dehydration & cerebral col- 65
lapse due to hypothalamic failure, pulmonary emboli
8 23, M 5, massive Sp-CP FT Total Ischemia in area of perforating vessels, severe hypothalamic 22
calcification dysregulation (hyperpyrexia, hypernatremia, DI), pulmonary
infection
9 54, M 4, massive Sp-CP AIH Total Tear of left ACA, hypothalamic infarction, severe hypothalamic 8
calcification dysregulation (hyperpyrexia, hypernatremia, DI)
AIH = anterior interhemispheric approach; FT = frontotemporal approach; VP = ventriculoperitoneal.

cerebral vascular injury (both major cerebral artery and tive years, 41.7% of children and 16.7% of adults with Sp-
small perforating vessels), attempted radical resection for CPs had significant weight gain (BMI greater than +4 SDs
massive calcifications, intraoperative brain swelling, and for children or BMI > 30 for adults). The relevant rates of
related postoperative severe hypothalamic dysfunction. weight gain in children and adults were 2.2% and 7.1% for
patients with Id-CPs and 12.5% and 2.6% for patients with
Postoperative Weight Gain and Hypothalamic Status Sa-CPs, respectively.
All but 8 surviving patients were followed up for a The HSS of surviving patients was evaluated at least
minimum of 3 months. The mean follow-up period was 3 months after the initial surgery (Fig. 10). Patients in
75 ± 42.3 months (range 3–184 months). Data on postop- Groups I and II showed a better hypothalamic status than
erative weight gain were available for 60 patients (97%) those in Group III, with a Grade 1 hypothalamic status
with Id-CPs, 46 patients (100%) with Sa-CPs, and 102 (normal hypothalamic function) observed in 57% and 43%
patients (94%) with Sp-CPs. There was a trend toward of patients in Groups I and II, respectively. Patients in
greater mean postoperative weight gain both in children Group III had more severe hypothalamic reactions, with
and in adults within Group III as compared with the other 39% of patients rated as having Grade 3 or 4 hypothalamic
2 groups (p < 0.001; Fig. 9). Over the first 2 postopera- status in the first 2 postoperative years (p < 0.001).

FIG. 9. Histograms summarizing postoperative weight gain in patients with childhood-onset (left) and adult-onset (right) CPs
based on tumor growth patterns.

10 J Neurosurg Pediatr December 4, 2015

 Growth pattern of craniopharyngiomas

FIG. 10. Histograms summarizing postoperative HSS scores based on tumor growth patterns.

Recurrence and Adjuvant Therapies lamic morbidity (such as morbid obesity, adipsic diabetes
Tumor recurrence, defined as evidence of tumor growth insipidus, and related fluid and electrolyte imbalance).
on neuroimaging with or without clinical symptoms, oc- There was no significant difference in the OS and PFS
curred in 52 patients within 1–11 years (median 2 years) rates between patients with childhood- and adult-onset
after initial surgery. The recurrence rate after complete re- disease (p = 0.421 and p = 0.956, respectively; Fig. 11).
section was 16.4%. The PFS rates for all patients after 2, 5, Furthermore, there was no significant difference in the OS
and 10 years were 82%, 72%, and 64%, respectively. In 2 and PFS rates among patients with different pathological
patients with Id-CPs, the recurrent tumors were restricted subtypes (p = 0.14 and p = 0.556, respectively; Fig. 12).
to the pituitary fossa; these 2 patients have not yet been However, there was a significant difference in the OS and
surgically treated because of their stable visual and neu- PFS rates among different growth pattern groups. Patients
rological status. Most patients with recurrent Id-CP were in Group III (Sp-CPs) showed substantially lower OS and
treated with reoperation, while patients with recurrent Sp- PFS rates as compared with patients in the other 2 groups
CPs were more likely to be treated with radiation therapy. (p = 0.053 and p = 0.001, respectively; Fig. 13).
In our institution, we typically use external beam radiation
therapy as salvage radiation therapy for unresectable recur- Discussion
rent CP. Gamma Knife surgery was prescribed for small
and early-detected cases of recurrence at the crucial hy- Growth Patterns of CPs
pothalamic area. Four patients died after their first tumor Classifying CPs based on their anatomical location or
recurrence, 16 of the 46 surviving patients experienced a relationship to neural structures (optic chiasm or TVF)
second recurrence, and 5 ultimately died of uncontrolled may not accurately clarify the growth pattern. Each ba-
disease. Notably, none of the 7 patients who underwent sic growth pattern has a definite topographic relationship
Gamma Knife surgery experienced tumor re-progression to sellar membranous structures and represents tumor
during 1–5 years (median 2 years) of follow-up. growth at its initial stage. In this study, CPs were classified
into 3 groups based on growth pattern, including infra-
Survival diaphragmatic, suprasellar subarachnoid extraventricular,
and suprasellar subpial ventricular growth. The complex
The median follow-up period was 78 months (range
growth of tumors at their final stage could be interpreted
6–184 months) for Group I, 75 months (range 12–176
using these basic growth patterns alone or in combination.
months) for Group II, and 56 months (range 3–168 months)
for Group III. The overall tumor-related mortality rate in
our series was 11.5% (25 of 217 patients). There were 9 Classification and Tumor-Meningeal Relationships in CPs
perioperative deaths, all of which occurred in Group III. Previous CP classification systems can be generally di-
The OS rate for all patients was 92% at 5 years and 87% vided into 2 categories. Classification systems in the first
at 10 years. No death was observed after 11 postoperative category include those based on tumor topographic rela-
years. The 5- and 10-year OS rates were 95% and 89% for tionships to the sellar diaphragm,6,63 optic chiasm,27 pitu-
patients in Group I, 95% and 92% for those in Group II, itary stalk,32 and TVF.58 Classification systems in the sec-
and 87% and 81% for those in Group III, respectively. The ond category include those based on tumor location and
most common causes of follow-up mortalities in Group I vertical extension (e.g., Samii and colleagues54,55). These
were long-term severe endocrinopathy, such as pan-hypo- systems classify CPs as follows: Grade I, intrasellar or
pituitarism; most of the mortalities in Group III patients infradiaphragmatic; Grade II, occupying the cistern with
were attributable to single or multiple recurrences, subse- or without an intrasellar component; Grade III, occupying
quent operations or radiation therapy, and severe hypotha- the lower half of the third ventricle; Grade IV, occupying
J Neurosurg Pediatr December 4, 2015 11
J. Pan et al.

FIG. 11. Survival curves for patients with childhood- and adult-onset tumors. The survival diagrams show that there was no signifi-
cant difference in the OS (left) and PFS (right) rates between patients with childhood- and adult-onset disease (p = 0.421 and p =
0.956, respectively, log-rank test).

the upper half of the third ventricle; and Grade V, reach- to CPs are mainly found in 2 areas: the anterior-superior
ing the septum pellucidum or lateral ventricles. These sys- surface of the junction of the pituitary gland and stalk and
tems are useful in understanding tumor growth and pro- the interface between the pars tuberalis and infundibulum.
vide clues that aid in the selection of the optimal surgical It is reasonable to speculate that CP development from
approach for tumor removal. However, even with the use these nests would give rise to subdiaphragmatic CPs and
of these classification systems, it is not uncommon to find suprasellar CPs, respectively. Our review of MR images
that the relationship and adherence between the tumor and and intraoperative findings for the 226 patients with CP in
surrounding neurovascular structures are quite different, our study was compatible with this hypothesis. We found
even in cases with a similar neuroradiological appearance. that all the tumors could be classified as having either an
This is explained as follows: 1) all of these classification infradiaphragmatic (infrasellar/infradiaphragmatic CPs)
systems are based on neuroradiological appearance, which or an infundibulotuberal (suprasellar infundibular CPs)
represents the final tumor stage at the time of diagnosis; origin. Suprasellar infundibular CPs will push along the
2) although some tumors may exhibit similar growth on path of least resistance, which is dependent on tumor or-
radiological evaluation at the time of diagnosis, their exact igin (superficial near the pia matter [corresponding to a
origin and relationship to surrounding structures may dif- lower site of origin] or deep to the ependymal layer) and, in
fer at onset; and 3) large tumors may severely compress, certain cases, tumor position relative to the optic chiasm.
distort, or involve neural tissue (TVF, pituitary stalk, and This results in 2 kinds of tumors: 1) those with preferen-
optic tract structures), making preoperative evaluation dif- tial subarachnoid growth, and 2) those with preferential
ficult even with excellent-quality neuroradiological stud- ventricular growth. Therefore, we classified CPs into 3
ies.1,2,25,28 Therefore, it is crucial to understand CP growth groups: Group I (Id-CPs), tumors with an infradiaphrag-
and to appreciate its effect on the distortion and displace- matic origin and an extraarachnoid tumor-meningeal re-
ment of vital neurovascular structures intraoperatively. lationship at onset; Group II (Sa-CPs), tumors with supra-
Erdheim21 was the first to theorize that CPs arise em- sellar extraventricular development and a subarachnoid
bryologically from an incompletely involuted craniopha- extrapial tumor-meningeal relationship; and Group III
ryngeal duct. According to his pioneering histological (Sp-CPs), tumors with suprasellar ventricular development
findings, the nests of epithelial cell remnants that give rise and a subarachnoid subpial tumor-meningeal relationship.

FIG. 12. Survival curves for patients with different tumor pathological subtypes. There was no significant difference in the OS (left)
and PFS (right) rates between patients with ACP and squamous PCP (p = 0.14 and p = 0.556, respectively, log-rank test).

12 J Neurosurg Pediatr December 4, 2015

 Growth pattern of craniopharyngiomas

FIG. 13. Survival curves for patients with different tumor growth. Left: OS diagram shows that the survival percentage for Sp-CPs
is lower than that for Id-CPs and Sa-CPs. This difference reached statistical significance (p = 0.053, log-rank test). Right: There
was a significant difference in the recurrence-free survival rates between groups. Patients with Sp-CPs showed substantially lower
PFS rates as compared with patients from the other 2 groups (at 5 years: 63% vs 85% and 86%; at 10 years: 60% vs 81% and
82%; p = 0.001, log-rank test).

Differences in Growth Patterns Between Juvenile and cordance with the purely suprasellar subpial ventricular
Adult Cases growth pattern in our classification scheme.
Variability in CP growth patterns between adults and Another important issue to consider when analyzing
children has not been noted in previously published data. the difference between juvenile and adult CPs are the 2
However, a literature review may provide some clues re- histological variants: ACP, the typical pediatric form, and
garding this issue. Studies focusing on pediatric CPs have PCP, observed almost exclusively in adults. Recently, it
provided detailed data on tumor growth patterns. These was reported that 95% of PCPs contain BRAF p.V600E
studies have shown that infradiaphragmatic tumors always mutations,5 and CTNNB1 and BRAF mutations are mu-
constitute an important subtype of CPs. Using Choux–Ray- tually exclusive and specific to tumor pathological sub-
baud’s6 classification, Lee et al.38 reported that 27 (40.9%) type. Larkin et al.34 examined the relationship between
and 39 (59.1%) of 66 pediatric CPs were purely intrasellar mutations in CTNNB1 and BRAF and the subcellular
and TVF CPs, respectively. Wang et al.63 analyzed the ra- location of b-catenin in a series of 37 patients with CPs.
diological, operative, and pathological findings of 25 con- They found BRAF V600E mutations in 81% (17 of 21) of
secutive pediatric CP patients. They found that the tumors PCPs by targeted Sanger sequencing and in 86% (18 of
in 14 (56%) patients had characteristics of the infradia- 21) of PCPs by immunohistochemistry. In all ACPs, trans-
phragmatic subtype, including prechiasmatic growth and location of b-catenin from the membrane to the cytosol/
a layer of sellar diaphragm covering the tumor top, where- nucleus was observed. Of 16 patients with ACP, 14 (88%)
as the tumors in the other 11 patients had retrochiasmatic were found to have BRAF wild-type tumors by sequencing
growth and a centered TVF. Lena et al.39 provided detailed and immunohistochemistry. These studies suggest that the
topographic CP data from a cohort of 47 children. They genetics of these tumors may be different. We compared
classified tumors into 4 types: purely intrasellar (14/47 the growth patterns of ACPs between children and adults.
[29.8%]), tuberoinfundibular (25/47 [53.2%]), strictly intra- The result showed that the difference persisted, indicating
ventricular (1/47 [2%]), and global or giant (7/47 [14.9%]). that the difference in tumor growth patterns between chil-
In our last series, 42% (34/81) of pediatric CPs were in- dren and adults cannot be simply attributed to pathological
fradiaphragmatic tumors.50 Interestingly, none of these type. Whether this difference is associated with different
studies included an extraventricular subtype in the tumor mechanisms of tumorigenesis between children and adults
classification. Few studies have specifically addressed deserves further study.
Id-CPs in adult patients, which is most likely because of Of note, we found that morphological distortion of hy-
their rarity in the adult population. Based on the literature, pothalamic structures differs between children and adults
purely intrasellar CPs represent the least common variety with Sp-CPs. Few studies have specifically addressed the
in both children and adults.22,31,61,64 Besides differences in degree of hypothalamic involvement in children versus
Id-CP incidence between children and adults, our results adults. However, we can draw some conclusions by analyz-
revealed that tumor size and anatomical location also dif- ing patient outcomes through a literature review. In a study
fered according to age. Juvenile Id-CP usually represents a of 63 childhood CP survivors, all patients with marked
crucial variant and has a wide spectrum of tumor origins obesity after surgery had evidence of significant altera-
(from the nasopharynx to the pituitary chamber), whereas tions in the normal hypothalamic anatomy, with MR im-
adult Id-CP is usually confined within the sellar, intrasel- ages showing either complete deficiency or extensive de-
lar, or suprasellar regions and has a more benign nature. struction of the TVF.15 It is widely accepted that obesity is
However, several studies have shown the predominance the most frequent manifestation of hypothalamic damage
of intraventricular lesions in older patients.4,29,44,45 These following CP surgery. De Vile et al.16 found that a young
findings have been replicated in this study and are in ac- age at presentation was a significant predictor of hypotha-
J Neurosurg Pediatr December 4, 2015 13
J. Pan et al.

lamic morbidity. These findings are compatible with our followed by craniotomy. They believe this approach may
morphological analysis of Group III tumors. Juvenile tu- allow the tumor to descend caudally, facilitating its further
mors had a higher probability of exhibiting a combination resection during the second surgery.18,41 The extraventricu-
of transstalk and subarachnoid subpial ventricular growth lar location of Sa-CPs adjudicates the extraaxial route, as
patterns, whereas adult tumors tended to exhibit a pure the arachnoid membrane surrounding the tumor provides
subarachnoid subpial ventricular growth pattern. Morpho- a plane of cleavage for tumor dissection.1,2,22,26,58,64 This ap-
logical distortion of hypothalamic structures is not neces- proach offers a high probability of preserving the integ-
sarily associated with significant hypothalamic morbidity. rity of the pituitary stalk and hypothalamus-hypophysis
We believe that childhood Sp-CPs more severely violate axis. The surgical removal of Sp-CPs always necessitates
hypothalamic structures as compared with adult Sp-CPs. an intraaxial route (the translamina terminalis route as a
However, it is difficult to determine whether this reflects primary or supplemental corridor) for the dissection of the
the vulnerability of hypothalamic structures in children or tumor from the nervous tissue parenchyma of the third
differences in tumor genetics between children and adults. ventricle walls. Therefore, radical removal of these tumors
Previous studies on the impact of age on prognosis and usually necessitates craniotomy to dissect the tumor under
survival in patients with CP have provided inconsistent re- direct vision. Nevertheless, tumor extension through the
sults. Better outcome in younger patients,7,22,59 better out- TVF toward the subarachnoid space would create an arti-
come in older patients,8,51 and similar outcome between ficial corridor to reach the tumor via an extraaxial route.22
younger and older patients31,42,47,48 have all been reported. Recently, endoscopy has been used in the transsphenoi-
These contradictory data warrant cautious interpretation, dal supradiaphragmatic approach. Several studies have
as a detailed topographic comparison of CPs between demonstrated successful tumor removal using an extended
adults and children was not provided in most of these endoscopic endonasal approach. However, longer follow-
studies. Differences in tumor growth patterns may influ- up studies are needed to confirm its long-term effective-
ence therapeutic options and therefore clinical outcomes. ness.10–14,19,23,24,30,32,35–37,40,41,57,60
Thus, discrepancies in the prognostic significance of age
in CP patients may be due in part to the variability in tu- Implications for CP Treatment Strategy
mor growth patterns between children and adults. There- The optimal therapeutic approach for CP remains con-
fore, we believe that a hierarchical comparison is neces- troversial. Some authors suggest complete removal as the
sary when investigating outcomes and survival across age primary treatment goal to avoid radiation exposure to the
groups in patients with CP. developing brain,3,22,27,61,64 whereas others advocate less ag-
gressive surgery combined with radiotherapy as an effec-
Clinical Relevance of Tumor Growth Patterns tive means of preventing recurrence.9,16,17,20,43,51,52,65 Devel-
Our topographic classification scheme highlights the opment of an intermediate approach using clear objective
original tumor site and the relationship between the tumor criteria and identification of guidelines to determine the
and suprasellar membranous structures. In CP, the point best surgical approach (radical vs conservative) are ongo-
of origin determines the preferential regional expansion of ing challenges for neurosurgeons. For the majority of pa-
the tumor and therefore influences morbidity in terms of tients with Id-CPs and Sa-CPs, radical tumor removal is
both pituitary and hypothalamic damage. Our topographic associated with long-term tumor control and a low likeli-
classification scheme provides more details of tumor-hypo- hood of postoperative hypothalamic dysfunction. On the
thalamus relationships. Each tumor group has a distinct in- contrary, the growth of Sp-CPs results in the gross mor-
volvement pattern toward the hypothalamic-pituitary axis. phological distortion of the hypothalamus, which makes
Id-CPs mainly involve the pituitary gland and the distal surgical manipulation of this area potentially hazardous.
portion of the stalk, while Sp-CPs mainly involve the third In our series, 39% of patients with Sp-CPs experienced
ventricle and hypothalamic structures. The analysis of our significant weight gain after radical tumor removal, indi-
clinical data confirmed that Id-CPs are more likely to be cating that gross-total removal of these tumors requires
associated with pituitary dysfunction and visual symp- caution. In addition, nearly one-third of patients with Sp-
toms, and that Sp-CPs are more likely to be associated CPs did not achieve total removal even after aggressive
with symptoms related to increased intracranial pressure surgery; the higher rate of recurrence may also be related
and mental problems related to hypothalamic dysfunction. to the significantly decreased likelihood of total resection
Furthermore, our classification system might serve as 1 of in this group. All these data indicate that tumor growth
the criterion for surgical planning. It is well known that a patterns and resectability are important factors when de-
transsphenoidal or transcranial extraaxial approach is most fining a treatment plan for patients with CP.
appropriate to treat the majority of patients with Id-CP. Further studies are necessary to more accurately ad-
The diaphragm and overlay arachnoid can serve as the dis- dress CP topography and individual surgical risk associ-
secting plane during operation. An understanding of this ated with radical lesion removal for Sp-CPs. Currently, we
topographic relationship is mandatory to perform radical believe that a large deficient TVF with sacrifice of the in-
resection and minimize damage to hypothalamic struc- fundibulum and stalk should be avoided when performing
tures. However, for large tumors with combination growth gross-total removal during CP surgery, of which children
patterns, an intraaxial route may be required to dissect with Sp-CPs are at increased risk given that childhood tu-
the tumor from the parenchyma around the third ventricle mors more severely violate hypothalamic structures.
walls. For these massive lesions, some authors advocate a In addition to its importance in selecting a surgical strat-
2-stage removal that involves transsphenoidal debulking, egy for tumor removal, tumor classification should also be
14 J Neurosurg Pediatr December 4, 2015
 Growth pattern of craniopharyngiomas

considered in selecting postoperative supplementary radio- transsphenoidal approach for residual or recurrent craniopha-
therapy. For instance, recurrent tumors located in crucial ryngiomas. J Neurosurg 111:578–589, 2009
areas, such as the parenchymal layer of the TVF (Sp-CPs), 13. Cavallo LM, Solari D, Esposito F, Cappabianca P: The en-
doscopic endonasal approach for the management of cranio-
make a second surgery more difficult. Therefore, the indi- pharyngiomas involving the third ventricle. Neurosurg Rev
cation for radiotherapy for these types of tumors should be 36:27–38, 2013
broadened. Nonetheless, Id-CPs and Sa-CPs (Groups I and 14. de Divitiis E, Cappabianca P, Cavallo LM, Esposito F, de
II) may not necessitate radiotherapy and have a high prob- Divitiis O, Messina A: Extended endoscopic transsphenoidal
ability of safe reoperation in cases of recurrence. approach for extrasellar craniopharyngiomas. Neurosurgery
61 (5 Suppl 2):219–228, 2007
Conclusions 15. de Vile CJ, Grant DB, Hayward RD, Kendall BE, Neville
BG, Stanhope R: Obesity in childhood craniopharyngioma:
CP is an extremely formidable tumor that poses a sig- relation to post-operative hypothalamic damage shown by
nificant surgical challenge. We propose a clinical classi- magnetic resonance imaging. J Clin Endocrinol Metab
fication system based on the site of tumor origin and the 81:2734–2737, 1996
relationship of the tumor to the arachnoid membrane and 16. De Vile CJ, Grant DB, Kendall BE, Neville BG, Stanhope R,
surrounding structures. This classification system im- Watkins KE, et al: Management of childhood craniopharyn-
gioma: can the morbidity of radical surgery be predicted? J
proves our understanding of the morphological features Neurosurg 85:73–81, 1996
and growth patterns of CPs. More tailored, individualized 17. Dhellemmes P, Vinchon M: Radical resection for cranio-
surgical strategies based on tumor growth patterns are pharyngiomas in children: surgical technique and clinical
mandatory to provide long-term tumor control and to min- results. J Pediatr Endocrinol Metab 19 (Suppl 1):329–335,
imize damage to hypothalamic structures. Differences in 2006
CP growth patterns between children and adults indicate 18. Duff J, Meyer FB, Ilstrup DM, Laws ER Jr, Schleck CD,
the need for hierarchical comparison in future CP studies. Scheithauer BW: Long-term outcomes for surgically resected
craniopharyngiomas. Neurosurgery 46:291–305, 2000
19. Dusick JR, Esposito F, Kelly DF, Cohan P, DeSalles A, Beck-
References er DP, et al: The extended direct endonasal transsphenoidal
1. Al-Mefty O, Ayoubi S, Kadri PA: The petrosal approach for approach for nonadenomatous suprasellar tumors. J Neuro-
the total removal of giant retrochiasmatic craniopharyngio- surg 102:832–841, 2005
mas in children. J Neurosurg 106 (2 Suppl):87–92, 2007 20. Elliott RE, Hsieh K, Hochm T, Belitskaya-Levy I, Wisoff J,
2. Al-Mefty O, Hassounah M, Weaver P, Sakati N, Jinkins JR, Wisoff JH: Efficacy and safety of radical resection of pri-
Fox JL: Microsurgery for giant craniopharyngiomas in chil- mary and recurrent craniopharyngiomas in 86 children. J
dren. Neurosurgery 17:585–595, 1985 Neurosurg Pediatr 5:30–48, 2010
3. Baskin DS, Wilson CB: Surgical management of craniopha- 21. Erdheim J: Über Hypophysengangsgeschwulste und Hirm-
ryngiomas. A review of 74 cases. J Neurosurg 65:22–27, cholesteatome. Sitzungsb Kais Akad Wissen Math Naturw
1986 Klin 113:537–726, 1904
4. Behari S, Banerji D, Mishra A, Sharma S, Sharma S, Chhab- 22. Fahlbusch R, Honegger J, Paulus W, Huk W, Buchfelder M:
ra DK, et al: Intrinsic third ventricular craniopharyngiomas: Surgical treatment of craniopharyngiomas: experience with
report on six cases and a review of the literature. Surg Neu- 168 patients. J Neurosurg 90:237–250, 1999
rol 60:245–253, 2003 23. Frank G, Pasquini E, Doglietto F, Mazzatenta D, Sciarretta
5. Brastianos PK, Taylor-Weiner A, Manley PE, Jones RT, Dias- V, Farneti G, et al: The endoscopic extended transsphenoi-
Santagata D, Thorner AR, et al: Exome sequencing identi- dal approach for craniopharyngiomas. Neurosurgery 59 (1
fies BRAF mutations in papillary craniopharyngiomas. Nat Suppl 1):ONS75–83, 2006
Genet 46:161–165, 2014 24. Gardner PA, Kassam AB, Snyderman CH, Carrau RL, Mintz
6. Brunel H, Raybaud C, Peretti-Viton P, Lena G, Girard N, Paz- AH, Grahovac S, et al: Outcomes following endoscopic, ex-
Paredes A, et al: [Craniopharyngioma in children: MRI study panded endonasal resection of suprasellar craniopharyngio-
of 43 cases.] Neurochirurgie 48:309–318, 2002 (Fr) mas: a case series. J Neurosurg 109:6–16, 2008
7. Bülow B, Attewell R, Hagmar L, Malmström P, Nordström 25. Hald JK, Eldevik OP, Quint DJ, Chandler WF, Kollevold T:
CH, Erfurth EM: Postoperative prognosis in craniopharyn- Pre- and postoperative MR imaging of craniopharyngiomas.
gioma with respect to cardiovascular mortality, survival, and Acta Radiol 37:806–812, 1996
tumor recurrence. J Clin Endocrinol Metab 83:3897–3904, 26. Hoffman HJ: Surgical management of craniopharyngioma.
1998 Pediatr Neurosurg 21 (Suppl 1):44–49, 1994
8. Bunin GR, Surawicz TS, Witman PA, Preston-Martin S, 27. Hoffman HJ, De Silva M, Humphreys RP, Drake JM, Smith
Davis F, Bruner JM: The descriptive epidemiology of cranio- ML, Blaser SI: Aggressive surgical management of cranio-
pharyngioma. J Neurosurg 89:547–551, 1998 pharyngiomas in children. J Neurosurg 76:47–52, 1992
9. Caldarelli M, Massimi L, Tamburrini G, Cappa M, Di Rocco 28. Hoffman HJ, Hendrick EB, Humphreys RP, Buncic JR, Arm-
C: Long-term results of the surgical treatment of craniopha- strong DL, Jenkin RD: Management of craniopharyngioma
ryngioma: the experience at the Policlinico Gemelli, Catholic in children. J Neurosurg 47:218–227, 1977
University, Rome. Childs Nerv Syst 21:747–757, 2005 29. Ikezaki K, Fujii K, Kishikawa T: Magnetic resonance imag-
10. Campbell PG, McGettigan B, Luginbuhl A, Yadla S, Rosen ing of an intraventricular craniopharyngioma. Neuroradiol-
M, Evans JJ: Endocrinological and ophthalmological conse- ogy 32:247–249, 1990
quences of an initial endonasal endoscopic approach for re- 30. Jane JA Jr, Kiehna E, Payne SC, Early SV, Laws ER Jr: Early
section of craniopharyngiomas. Neurosurg Focus 28(4):E8, outcomes of endoscopic transsphenoidal surgery for adult
2010 craniopharyngiomas. Neurosurg Focus 28(4):E9, 2010
11. Cappabianca P, Cavallo LM: The evolving role of the trans- 31. Karavitaki N, Brufani C, Warner JT, Adams CB, Richards P,
sphenoidal route in the management of craniopharyngiomas. Ansorge O, et al: Craniopharyngiomas in children and adults:
World Neurosurg 77:273–274, 2012 systematic analysis of 121 cases with long-term follow-up.
12. Cavallo LM, Prevedello DM, Solari D, Gardner PA, Esposito Clin Endocrinol (Oxf) 62:397–409, 2005
F, Snyderman CH, et al: Extended endoscopic endonasal 32. Kassam AB, Gardner PA, Snyderman CH, Carrau RL, Mintz

J Neurosurg Pediatr December 4, 2015 15

J. Pan et al.

AH, Prevedello DM: Expanded endonasal approach, a fully 51. Rajan B, Ashley S, Gorman C, Jose CC, Horwich A, Bloom
endoscopic transnasal approach for the resection of midline HJ, et al: Craniopharyngioma--a long-term results following
suprasellar craniopharyngiomas: a new classification based limited surgery and radiotherapy. Radiother Oncol 26:1–10,
on the infundibulum. J Neurosurg 108:715–728, 2008 1993
33. Kitano M, Taneda M: Extended transsphenoidal surgery for 52. Regine WF, Kramer S: Pediatric craniopharyngiomas: long
suprasellar craniopharyngiomas: infrachiasmatic radical term results of combined treatment with surgery and radia-
resection combined with or without a suprachiasmatic trans- tion. Int J Radiat Oncol Biol Phys 24:611–617, 1992
lamina terminalis approach. Surg Neurol 71:290–298, 2009 53. Sainte-Rose C, Puget S, Wray A, Zerah M, Grill J, Brauner
34. Larkin SJ, Preda V, Karavitaki N, Grossman A, Ansorge R, et al: Craniopharyngioma: the pendulum of surgical man-
O: BRAF V600E mutations are characteristic for papillary agement. Childs Nerv Syst 21:691–695, 2005
craniopharyngioma and may coexist with CTNNB1-mutated 54. Samii M, Bini W: Surgical treatment of craniopharyngiomas.
adamantinomatous craniopharyngioma. Acta Neuropathol Zentralbl Neurochir 52:17–23, 1991
127:927–929, 2014 55. Samii M, Tatagiba M: Surgical management of craniopha-
35. Laufer I, Anand VK, Schwartz TH: Endoscopic, endonasal ryngiomas: a review. Neurol Med Chir (Tokyo) 37:141–149,
extended transsphenoidal, transplanum transtuberculum 1997
approach for resection of suprasellar lesions. J Neurosurg 56. Song-tao Q, Xi-an Z, Hao L, Jun F, Jun P, Yun-tao L: The
106:400–406, 2007 arachnoid sleeve enveloping the pituitary stalk: anatomical
36. Laws ER Jr: Surgical outcome in 90 patients with craniopha- and histologic study. Neurosurgery 66:585–589, 2010
ryngiomas: an evaluation of the transsphenoidal approach. 57. Stamm AC, Vellutini E, Harvey RJ, Nogeira JF Jr, Herman
World Neurosurg 74:254–255, 2010 DR: Endoscopic transnasal craniotomy and the resection of
37. Laws ER, Kanter AS, Jane JA Jr, Dumont AS: Extended craniopharyngioma. Laryngoscope 118:1142–1148, 2008
transsphenoidal approach. J Neurosurg 102:825–828, 2005 58. Steno J, Malácek M, Bízik I: Tumor-third ventricular rela-
38. Lee YY, Wong TT, Fang YT, Chang KP, Chen YW, Niu DM: tionships in supradiaphragmatic craniopharyngiomas: cor-
Comparison of hypothalamopituitary axis dysfunction of relation of morphological, magnetic resonance imaging, and
intrasellar and third ventricular craniopharyngiomas in chil- operative findings. Neurosurgery 54:1051–1060, 2004
dren. Brain Dev 30:189–194, 2008 59. Sun HI, Akgun E, Bicer A, Ozkan A, Bozkurt SU, Kurtkaya
39. Lena G, Paz Paredes A, Scavarda D, Giusiano B: Craniopha- O, et al: Expression of angiogenic factors in craniopharyn-
ryngioma in children: Marseille experience. Childs Nerv giomas: implications for tumor recurrence. Neurosurgery
Syst 21:778–784, 2005 66:744–750, 2010
40. Maira G, Anile C, Albanese A, Cabezas D, Pardi F, Vignati 60. Thompson D, Phipps K, Hayward R: Craniopharyngioma
A: The role of transsphenoidal surgery in the treatment of in childhood: our evidence-based approach to management.
craniopharyngiomas. J Neurosurg 100:445–451, 2004 Childs Nerv Syst 21:660–668, 2005
41. Maira G, Anile C, Rossi GF, Colosimo C: Surgical treatment 61. Van Effenterre R, Boch AL: Craniopharyngioma in adults
of craniopharyngiomas: an evaluation of the transsphenoidal and children: a study of 122 surgical cases. J Neurosurg
and pterional approaches. Neurosurgery 36:715–724, 1995 97:3–11, 2002
42. Manaka S, Teramoto A, Takakura K: The efficacy of radio- 62. Wang KC, Hong SH, Kim SK, Cho BK: Origin of craniopha-
therapy for craniopharyngioma. J Neurosurg 62:648–656, ryngiomas: implication on the growth pattern. Childs Nerv
1985 Syst 21:628–634, 2005
43. Merchant TE, Kiehna EN, Kun LE, Mulhern RK, Li C, 63. Wang KC, Kim SK, Choe G, Chi JG, Cho BK: Growth
Xiong X, et al: Phase II trial of conformal radiation therapy patterns of craniopharyngioma in children: role of the dia-
for pediatric patients with craniopharyngioma and correla- phragm sellae and its surgical implication. Surg Neurol
tion of surgical factors and radiation dosimetry with change 57:25–33, 2002
in cognitive function. J Neurosurg 104 (2 Suppl):94–102, 64. Yaşargil MG, Curcic M, Kis M, Siegenthaler G, Teddy PJ,
2006 Roth P: Total removal of craniopharyngiomas. Approaches
44. Pascual JM, Carrasco R, Prieto R, Gonzalez-Llanos F, Alva- and long-term results in 144 patients. J Neurosurg 73:3–11,
rez F, Roda JM: Craniopharyngioma classification. J Neuro- 1990
surg 109:1180–1183, 2008 65. Zuccaro G: Radical resection of craniopharyngioma. Childs
45. Pascual JM, González-Llanos F, Barrios L, Roda JM: Intra- Nerv Syst 21:679–690, 2005
ventricular craniopharyngiomas: topographical classification
and surgical approach selection based on an extensive over-
view. Acta Neurochir (Wien) 146:785–802, 2004
46. Pascual JM, Prieto R, Castro-Dufourny I, Gil-Simoes R, Disclosures
Carrasco R: Classification systems of adult craniopharyngio- The authors report no conflict of interest concerning the materi-
mas: the need for an accurate definition of the hypothalamus- als or methods used in this study or the findings specified in this
tumor relationships. Arch Med Res 43:588–590, author reply paper.
591, 2012
47. Pemberton LS, Dougal M, Magee B, Gattamaneni HR: Ex-
perience of external beam radiotherapy given adjuvantly or at
Author Contributions
relapse following surgery for craniopharyngioma. Radiother Conception and design: Qi, Pan. Acquisition of data: Pan, Liu,
Oncol 77:99–104, 2005 Lu, Peng, Zhang, Xu, Huang, Fan. Analysis and interpretation
48. Petito CK, DeGirolami U, Earle KM: Craniopharyngiomas: a of data: Qi, Pan, Lu, Xu, Huang. Drafting the article: Pan. Criti-
clinical and pathological review. Cancer 37:1944–1952, 1976 cally revising the article: Qi, Pan. Reviewed submitted version
49. Qi S, Lu Y, Pan J, Zhang X, Long H, Fan J: Anatomic rela- of manuscript: Pan. Statistical analysis: Liu, Peng, Zhang, Fan.
tions of the arachnoidea around the pituitary stalk: relevance Administrative/technical/material support: Liu, Fan. Study super-
for surgical removal of craniopharyngiomas. Acta Neuro- vision: Qi.
chir (Wien) 153:785–796, 2011
50. Qi S, Pan J, Lu Y, Gao F, Cao Y, Peng J, et al: The impact of Correspondence
the site of origin and rate of tumour growth on clinical out- SongTao Qi, Department of Neurosurgery, Nanfang Hospital,
come in children with craniopharyngiomas. Clin Endocrinol #1838 North Guangzhou Ave., Guangzhou 510515, China. email:
(Oxf) 76:103–110, 2012 [email protected].

16 J Neurosurg Pediatr December 4, 2015